Morphological identification of two sympatric species of Trichiuridae, Aphanopus carbo and A. intermedius, in NE Atlantic

Morphological identification of two sympatric species of Trichiuridae, Aphanopus carbo and A. intermedius, in NE Atlantic by Manuel Biscoito* (1), João DelgaDo (2), José a. gonzález (3), sérgio stefanni (4), Víctor M. tuset (3), eduardo isiDro (4), antonio garcía-MeDeros (3) & Dalila carValho (2) AbsTrAcT. - The black scabbardfish has been subjected to a commercial fishery in the waters of the archipelago of Madeira for more than 150 years, which is probably the oldest deep-sea commercial fishery in the world. Over this period the presence of two sympatric species (Aphanopus carbo Lowe, 1839 and A. intermedius Parin, 1983) in the area has been ignored, mainly due to the difficulty in separating the two species using external morphological characters. The need for a more accurate management of this highly important resource, reinforced by an emergent fishery in Portugal mainland and elsewhere in the North Atlantic, justified a new effort for morphological characterization of the two species, based on the largest genetically validated sample obtained to date. The results presented in this paper demonstrate that it is possible to discriminate the two species on the bases of meristic and morphometric data using discriminant analysis. The outcome of this analysis is supported by the genetic identification based on CR and COI sequences. In addition, a redescription of A. intermedius incorporating the new character ranges found is presented. résuMé. - Identification morphologique de deux espèces sympatriques de Trichiuridae, Aphanopus carbo et A. intermedius de l’Atlantique du Nord-Est. Le sabre noir a fait l’objet de la plus ancienne pêche commerciale en eaux profondes. Son histoire, longue de 150 ans dans les eaux de l’archipel de Madère, n’a cependant pas permis de déceler la présence de deux espèces sympatriques : Aphanopus carbo Lowe, 1839 et A. intermedius Parin, 1983. L’explication de ce hiatus tient à la difficulté de séparer les deux espèces sur la base des seuls caractères morphologiques externes. Or, l’émergence des activités de pêche tant dans les eaux continentales portugaises qu’en d’autres points de l’Atlantique Nord implique une gestion fine de cette ressource halieutique. Cela inclut la caractérisation des deux espèces grâce à l’analyse d’un grand nombre de spécimens dont l’identification est systématiquement validée au niveau génétique. La présente étude démontre ainsi qu’une analyse discriminante appliquée à l’ensemble des données méristiques et morphométriques permet de séparer les deux espèces. D’autre part, cette même étude montre que ces résultats sont validés au niveau génétique par les séquences issues du gène COI. Les travaux réalisés dans le cadre de cet article permettent d’envisager en outre, l’actualisation des caractères descriptifs des deux espèces A. intermedius et A. carbo. Key words. - trichiuridae - Aphanopus - Black scabbardfish - ANE - Discriminant analysis - Distribution - mtDNA - Taxonomy. Cybium 2011, 35(1): 19-32. (1) Museu Municipal do Funchal (História Natural), Rua da Mouraria, 31, 9004-546 Funchal, Madeira, Portugal. (2) Direcção de Serviços de Investigação das Pescas, Estrada da Pontinha, 9004-562 Funchal, Madeira, Portugal. [[email protected]] [[email protected]] (3) Instituto Canario de Ciencias Marinas, Agencia Canaria de Investigación, Innovación y Sociedad de la Información, 35214 Telde, Las Palmas, Canary Islands, sPain. [[email protected]] [[email protected]] [[email protected]] (4) IMAR/DOP, University of the Azores, Cais Sta. Cruz, 9901-862 Horta, Azores, Portugal. [[email protected]] [[email protected]] * Corresponding author [[email protected]] The black scabbardfish Aphanopus carbo (Lowe, 1839), has been subjected to a commercial fishery in the waters of the archipelago of Madeira for more than 150 years, in which is probably the oldest deep-sea commercial fishery in the world (Maul, 1950). This long-line fishery has yielded in the last seven years an average of 3450 tons per year (data obtained from Direcção Regional de Pesca, Madeira), which is mainly consumed on the island where it is used as emblematic dish. until recently, A. carbo was the only recognized species in this genus, although several other junior synonyms were described, including one from Madeira, A. acus Maul, 1948 (Parin, 1983; Nakamura and Parin, 1993). In 1983, a description of A. intermedius, partially sympatric with A. carbo, was published and now the genus Aphanopus comprises seven species distributed throughout all oceans except in the polar regions and the Mediterranean Sea (Parin, 1983, 1995). For fisheries purposes these two species have been treated as one (A. carbo) in Madeira and consequently landing statistics and fisheries research refer to a mixture of these two similar species. It is worthwhile pointing out that the main differences by which A. carbo may be separated from A. intermedius (dorsal fin and vertebral counts) (Nakamura and Parin, 1993; Parin, 1995) are not easy to use in the field and are totally unsuitable for large scale fisheries-purpose identification, on board or at landing sites. Aphanopus carbo and A. intermedius from the northeastern Atlantic Biscoito e t a l . 20 Cybium 2011, 35(1) Several projects and initiatives aimed for the search of new deep-sea species of potential commercial interest in the waters of the Macaronesian archipelagos of the Azores, Madeira and the Canaries have been carried out along the last decade. This gave the opportunity to acquire specimens of Aphanopus in the three regions involved, either from the commercial fishery or from experimental fishing. The need for a more accurate management of this highly important resource, reinforced by an emergent fishery in Portugal mainland (Figueiredo et al., 2003), the Azores and elsewhere in the North Atlantic (Lorance and Dupouy, 2001), also including the Canary region and nearby seamounts, and the findings by Stefanni and Knutsen (2007), raised again the question of the species identification. However, the issue of discriminating the A. carbo from A. intermedius based on their morphology was not resolved yet. With this contribution the authors aim to characterize morphologically a large sample of specimens of Aphanopus collected in different fishing grounds of the Northeast Atlantic. To have comparable data to previous work, molecular sequences of two mtDNA regions (Control Region and COI) were amplified in these specimens, thus allowing a correct identification of A. carbo and A. intermedius. MATeriAl And MeThods A series of 145 specimens of Aphanopus spp. were collected in the waters off Sesimbra (mainland Portugal), the islands of the Azores, Madeira, and Canaries; off the coasts of Morocco and Western Sahara (Appendix I, Fig. 1). The specimens from Sesimbra and the Azores were taken randomly from the commercial fishery. In the other localities, specimens were obtained from experimental fishing. All specimens were measured, weighed and dissected for determination of sex, maturity stage and vertebral counts. A tissue sample was also extracted and preserved in 70% ethanol. Eighteen measurements were made point to point to the nearest millimetre and follow Nakamura and Parin (1993) with modifications introduced in the present paper (Figs 2, 3) and nine counts were made directly. Due to damage to the dorsal fin, it was not always possible to count spines and soft rays separately, although it was possible to count the total number of dorsal fin elements in those specimens. Vertebral counts were divided in total, pre-caudal and caudal vertebrae. Fused vertebrae forming the hypural plate were counted as one. The position of anus and first anal spine in relation to dorsal-fin elements was also noted. Methodology used for genetic analysis All specimens were screened for two mtDNA genes to assign to each specimen the correct identification, using the available sequences of the mtDNA Control Region (CR) from Stefanni and Knutsen (2007) and Cytochrome Oxidase subunit I COI from Stefanni et al. (2009). A total of 138 sequences were aligned for the complete Control Region (CR) (GenBank Accession Nos. EU853865-EU854002) and 144 for the partial Cytochrome Oxidase subunit I (COI) (GenBank Accession Nos. EU854003-EU854146). Although the majority of the specimens had amplified and produced good quality sequences for both genes, in some individuals (7 for the CR and 1 for the COI, see Appendix 1) the PCR amplification was very weak and of poor quality. However, as the individual that did not amplify for the COI was not one of the 7 that did not amplify for the CR, all 145 fish were screened for correct identification. The thermal cycling profile for the fragment including the CR followed Stefanni and Knutsen (2007), while for COI it followed Stefanni et al. (2009). All sequences were aligned using Seaview (Galtier et al., 1996) and levels of genetic diversity as well as genetic signatures were estimated using Arlequin 3.0 (Excoffier et Figure 1. - Map of the study area in the NE Atlantic Ocean showing the locations where specimens of A. carbo and A. intermedius were collected (shaded areas). Bi s c o i t o e t a l . Aphanopus carbo and A. intermedius from the northeastern Atlantic Cybium 2011, 35(1) 21 al., 2005) implementing the same parameters estimated in Stefanni and Knutsen (2007). All sequences from the CR were aligned with the ones reported by Stefanni and Knutsen (2007) for detection of shared haplotypes and to assign the correct species identification to the specimens used for the morphological work. Methodology used for discriminant analysis Only in 53 specimens of A. carbo (SL 905-1188 mm) and 36 specimens of A. intermedius (SL 852-1345 mm) it was possible to obtain the whole set of measurements and counts (Tab. I), therefore only 89 out of 145 specimens were used for discriminant analyses. All morphometric variables were first examined for normality (Kolmogorov-Smirnov test) and homogeneity of variances (Levene test), and were log-transformed to statistical analysis if these criteria were not satisfied. Analysis of covariance (ANCOVA) was then used to determine the effect of length (standard or head length) on the magnitude of each shape variable. “Species” was treated as the main factor and length was the covariate. Variables for which “species-length” interactions were significant (p < 0.05, samples with unequal slopes) were not included in any further analysis because they could be corrected for length. Those variables found to have samples with equal slopes were corrected using their respective common within-group slope (b) (Bolles and Begg, 2000; Begg et al., 2001; De Vries et al., 2002). The length of reference for body variables was the standard length, whereas for cephalic variables the head length. Multivariate analysis of variance (MANOVA) was used to test the hypothesis of no difference in morphometric and meristic variables among species. This procedure was explored using a canonical discriminate analysis (CDA). This technique allows to evaluate the differences between groups using several discriminant variables and to predict the ownership to a group. The first step was to carry out a single factor ANOVA to find out which variables discriminate between species, using the F statistics to rank the potential predictors. To avoid multicolinearity, a matrix correlation was obtained and eliminated of CDA analysis those variables with a high correlation and small F-score from ANOVA. Stepwise linear discriminant analysis was used to guide selection of variable sets used in each function. This procedure chooses variables to enter or leave the model on the basis of the significance level of an F-test by ANOVA. Homogeneity of the within-group covariance matrices was tested and either a linear (matrices are homogenous and the pooled matrix is used) or a quadratic (matrices are not homogenous and individual within-group matrices are used) discriminant function was computed (Friedland et al., 1994). Classification efficiency (percent correctly classified) estimates were cross-validated according to the methods of Lachenbruch and Mickey (1968). To establish the bias of the analysis Cohen’s kappa (κ) statistic was used, which estimates the improvement over chance of the percent correct classification rates (Titus et al., 1984). The prior probability of classification was equal for both groups. Junquera and Pérez-Gándaras (1993) and Camacho (1995) indicated that if the number of individuals minus the number of variables is greater than 30, then the sample can be considered adequate for analysis and it is only necessary to construct one discriminant function. The misclassification rate was assessed by classifying the same number of fish used to form the discriminant analysis database and summing the number of misclassified fish (Reddin et al., 1988). Two CDA analyses were Figure 2. - Schematic drawing of a trichiurid showing body measurements used in the present study (adapted from Nakamura and Parin, 1993). See table I for abbreviations. Figure 3. - Schematic drawing of head and tail of a trichiurid showing measurements used in the present study (adapted from Nakamura and Parin, 1993). See table I for abbreviations. Aphanopus carbo and A. intermedius from the northeastern Atlantic Biscoito e t a l . 22 Cybium 2011, 35(1) constructed, one using morphometric data only and the other using morphometric and meristic variables together. resulTs Morphology and meristics The morphometric relationships and meristic characters used for identification of the specimens of the two species of Aphanopus studied, which were previously separated based on the genetic results, are given in table I. A comparison with data from the bibliography (Parin, 1983, 1995; Nakamura and Parin, 1993) is also made (Tab. I). Frequency distributions of the nine meristic characters used are also given in table II, in order to show comparatively both range and mode of the different counts. Previously known ranges of most of the characters measured are enlarged for both species. In addition, an overlap of all measurements and counts in both species was found. Genetics On the basis of the sequences of two mtDNA genes, 74 A. carbo and 71 A. intermedius were identified. The complete sequences of the CR were 733 bp long in A. carbo and 732 bp long in A. intermedius. The partial sequences of the COI were 668 bp long for both species. From the alignment with the CR dataset from Stefanni and Knutsen (2007) several common haplotypes were found, either as A.carbo or A. intermedius. Within A. carbo group, all new sequences coded as SHc1, SHc3, SHc4, SHc5, Mad15, Can10, Mor17, SHc7, SHc8 and SHc9 (see column H CR in Appendix 1) correspond to the sequences ShP345, Az22, SN7, FD1, ShM790, ShA374, ShA380, ShP190, ShS130 and SN8 obtained by Stefanni and Knutsen (2007). On the other hand, within A. intermedius group, the common haplotypes between the two datasets are all new sequences coded as SHi1, SHi2, SHi3, Azo23, SHi4, SHi6 and Mad25 (see column H CR in Appendix 1) and correspond to the sequences of the dataset published by Stefanni and Knutsen (2007) as Table I. - Morphometric and meristic characters of Aphanopus carbo and A. intermedius. PESCPROF specimens were genetically identified. Relationships marked with an asterisk mean data from holotype and three paratypes (from Parin, 1993). Aphanopus carbo Aphanopus intermedius PescProf Nakamura & Parin (1993) PescProf Parin (1995) Standard length (SL, mm) 905-1293 (70) -622-1345 (63) 515-1010 (17) Head length (HL, mm) 174-250 (73) -123-270 (70) 95.3-212.1 (17) %SL Pre-anal length (Pal) 58.6-64.4 (55) -57.0-63.8 (46) - Pre-first anal spine length (PASL) 55.6-60.5 (55) -55.2-60.0 (46) 56.1-58.0* Pre-anus length (Panl) 54.2-59.2 (70) 55.6-58.8 52.7-64.0 (63) 52.8-55.0* Pre-pectoral length (PPL) 18.3-20.9 (55) -18.6-20.8 (46) - Pre-dorsal length (PDl) 15.3-18.8 (55) -14.9-18.5 (46) 16.2-17.8* Pre-first dorsal soft ray length (P1DFL) 50.3-60.2 (54) -50.4-59.2 (40) 54.8-56.9* Maximum body depth (Hmax) 7.8-13.2 (55) 7.5-9.3 6.9-12.7 (46) 6.1-8.6 (17) Depth of body at level of first anal spine (H1SFA) 6.0-14.2 (70) -6.0-10.5 (63) - Least depth of caudal peduncle (CPD) 0.4-0.5 (55) -0.3-0.5 (46) 0.3-0.4* Caudal peduncle length (CPL) 1.2-2.9 (55) -2.0-4.2 (46) - head length (hl) 18.4-22.1 (70) 19.2-21.3 17.9-22.5 (63) 18.5-21.0 (17) %HL Pre-opercular length (POL) 77.7-82.8 (55) -77.0-83.9 (46) - snout length (snl) 37.4-49.8 (73) 40.0-43.5 36.7-50.4 (70) 40.4-43.2 (17) Eye diameter (ED) 16.5-26.8 (73) 17.2-20.4 13.8-24.8 (70) 17.8-20.1 (17) Inter-orbital width (IO) 13.6-19.2 (73) -11.6-21.7 (70) 12.3-15.6 (17) Maxillary length (ML) 43.8-51.0 (54) 45.5-47.6 45.6-49.8 (45) 46.9-49.4 (17) head height (hht) 32.3-42.3 (55) -31.4-42.1 (46) 34.5-35.6* Meristic characters Dorsal-fin spines (DS) 38-41 (66) 38-41 39-43 (41) 40-44 (55) Dorsal-fin soft rays (DR) 51-57 (66) 52-56 52-60 (41) 54-59 (55) Total dorsal-fin elements (DT) 89-96 (70) 90-96 92-102 (60) 96-101 (55) Anal-fin rays (without spines) (AF) 42-48 (66) 43-48 45-50 (59) 46-50 (55) Pre-caudal vertebrae (PCV) 40-43 (55) 40-44 43-47 (46) 44-47 (55) Caudal vertebrae (CV) 55-60 (55) 55-60 56-61 (46) 57-61 (55) Total vertebrae (TV) 98-101 (55) 97-100 101-105 (46) 102-107 (55) Bi s c o i t o e t a l . Aphanopus carbo and A. intermedius from the northeastern Atlantic Cybium 2011, 35(1) 23 ShA999, ShA412, Az21, Az6, Az75, ShA129 and ShA334, respectively. The 70 CR sequences of A. carbo defined 46 haplotypes, 37 of which were represented by a single specimen while the remaining sequences were shared among 9 haplotypes (Appendix 1). The nucleotide composition was estimated to be C = 22.9%, T = 31.1%, A = 31.6% and G = 14.4% and the transition/transvertion ratio of 2.23. The 46 haplotypes described an overall haplotypic diversity of 4.7850 ± 2.3662 and nucleotide diversity of 0.0065 ± 0.0036, and they conTable II. - Comparison of meristic characters of Aphanopus carbo and A. intermedius, based on genetic identification. Dorsal-fin spines 38 39 40 42 43 44 n A. carbo 5 18 31 – – – 66 A. intermedius – 1 13 82 – 41 Dorsal-fin rays 51 52 53 55 56 57 58 59 60 n A. carbo 2 12 19 10 41–––66 A. intermedius – 1 – 13 963 1 1 41 Dorsal-fin elements (total) 89 90 91 93 94 95 96 97 98 99 100 101 102 n A. carbo 1 2 2 22 16 123––––––70 A. intermedius ––– ––11 14 14 11 341160 Anal-fin rays 42 43 44 46 47 48 49 50 n A. carbo 1611 10 7 7 – – 66 A. intermedius – – – 12 18 13 11 359 Pre-caudal vertebrae 40 41 42 44 45 46 47 n A. carbo 11 22 15 ––––55 A. intermedius ––– 911 17 246 Caudal vertebrae 55 56 57 59 60 61 n A. carbo 3 4 25 5 1 – 55 A. intermedius –10 11 48146 Total vertebrae 97 98 99 101 102 103 104 105 106 107 108 n A. carbo –29 15 1–––––––55 A. intermedius – – – 1 18 18 8 1–––46 Position of anal fin spines in relation to dorsal-fin spines and soft rays over Last spine 1st ray 2nd ray 3rd ray 4th ray 5th ray 6th ray n A. carbo 5 25 12 11 154 A. intermedius 217 6 13 8138 Position of anus in relation to dorsal-fin spines over Penultimate spine 1st ray 2nd ray 3rd ray 4th ray 5th ray n A. carbo 2633 4 54 Aphanopus carbo and A. intermedius from the northeastern Atlantic Biscoito e t a l . 24 Cybium 2011, 35(1) tained 41 polymorphic sites. On the other hand, the 68 CR sequences of A. intermedius defined 28 haplotypes, 22 of which were unique and the other 6 were shared with the other 36 specimens (Appendix 1). The nucleotide composition was estimated to be C = 23.1%, T = 30.6%, A = 31.4% and G = 14.9% and the transition/transvertion ratio of 7. The 28 haplotypes described an overall haplotypic diversity of 1.5812 ± 0.9526 and nucleotide diversity of 0.0022 ± 0.0014, and they contained 23 the polymorphic sites. The more conservative and shorter fragment of the COI identified 10 haplotypes in A. carbo, 7 of which were represented by a single fish and 3 were shared with the other specimens (Appendix 1). The nucleotide composition was estimated to be C = 29.9%, T = 28.7%, A = 22.5% and G = 18.9% and the transition/transvertion ratio of 1.4. The 10 haplotypes described an overall haplotypic diversity of 0.4527±0.4054 and nucleotide diversity of 0.0007 ± 0.0007, and they contained 12 the polymorphic sites. in A. intermedius, this fragment was characterized by 14 haplotypes, 12 of which were uniquely represented and 2 shared with the other specimens (Appendix 1). The nucleotide composition was estimated to be C = 29.5%, T = 28.9%, A = 22.6% and G = 19.0% and the transition/transvertion ratio of 1.6. The 14 haplotypes described an overall haplotypic diversity of 0.5614 ± 0.4641 and nucleotide diversity of 0.0008 ± 0.0008, and they contained 12 the polymorphic sites. The corrected sequence divergence, the algorithm that compensates for the average number of pairwise differences between and within the two groups, one represented by A. carbo and the other by A. intermedius, was estimated to be 23.40% for CR and 6.86% for COI. Regarding the sequence divergence within each group, the values for CR were 4.78% in A. carbo and 1.58% in A. intermedius, while for the partial COI were 0.45% in A. carbo and 0.56% in A. intermedius. Highly significant (p < 0.05) values of Φst (0.8795 for CR and 0.9314 for COI) put in evidence a strong genetic partitioning between the species. A phylogenetic tree for the COI sequences is shown in figure 4. Figure 4. - Neighbour-joining tree constructed from sequences of the coi using PauP (Swofford, 1999) software and implementing the HKY (Hasegawa et al., 1985) nucleotide substitution model with no invariable sites and equal rate. Numbers above internal branches indicate bootstrap values out of 1000 replicates (only if greater than 50%). AB205442 = Cubiceps paradoxus sequence used as outgroup. Codes for OTU’s are described in appendix 1. Bi s c o i t o e t a l . Aphanopus carbo and A. intermedius from the northeastern Atlantic Cybium 2011, 35(1) 25 discriminant analyses The CPL variable was log-transformed to correct nonnormality, whereas CPD was eliminated of the study due to the impossibility to change the variance heterogeneity. ANCOVA detected significant “species-length” interactions for IOD, SML and HCL being eliminated from posterior analysis. All the remaining variables were significantly correlated with length and therefore were corrected for variable length with their respective common within-group slope (Tab. III). Morphometric and meristic variables, with Table III. - Morphometric variables significantly correlated with length, and the corresponding regression coefficients (b) required to standardizing the variables for length. Standardized with respect to standard length1 or head length2. Table IV. - Results of ANOVA to test morphometric and meristic relationships between species to show variables with highest F statistics. Variables Length x species length b FpFSignificant Pre-anal length (Pal)10.040 0.842 1203.891 0.000 0.624 Pre-first anal spine length (PASL)10.188 0.665 2238.472 0.000 0.601 Pre-anus length (Panl)10.070 0.791 1022.289 0.000 0.559 Pre-pectoral length (PPL)12.757 0.101 533.955 0.000 0.194 Pre-dorsal length (PDl)11.219 0.273 306.477 0.000 0.168 Pre-first dorsal soft ray length (P1DFL)12.138 0.147 548.496 0.000 0.564 Maximum body depth (Hmax)10.381 0.539 66.352 0.000 0.138 Depth of body at level of first anal spine (H1SFA)12.902 0.092 150.996 0.000 0.064 Caudal peduncle length (CPL)10.338 0.563 15.361 0.000 0.020 Pre-opercular length (POL)22.044 0.157 2016.868 0.000 0.780 snout length (snl)21.601 0.209 275.794 0.000 0.387 Eye diameter (ED)20.035 0.852 95.897 0.000 0.259 Inter-orbital width (IO)28.145 0.005 - - - Maxillary length (ML)26.777 0.011 - - - head height (hht)210.400 0.002 - - - Variables Wilks’ lambda Fdf1 df2 P Pre-anal length (Pal) 0.924 7.144 187 0.009 Pre-first anal spine length (PASL) 0.958 3.817 187 0.054 Pre-anus length (Panl) 0.975 2.185 187 0.143 Pre-pectoral length (PPL) 1.000 0.024 187 0.877 Pre-dorsal length (PDl) 0.981 1.674 187 0.199 Pre-first dorsal soft ray length (P1DFL) 0.945 5.103 187 0.026 Maximum body depth (Hmax) 0.990 0.856 187 0.357 Depth of body at level of first anal spine (H1SFA) 0.767 26.470 187 < 0.001 Caudal peduncle length (CPL) 0.925 7.047 187 0.009 Pre-opercular length (POL) 0.999 0.115 187 0.735 snout length (snl) 0.996 0.324 187 0.570 Eye diameter (ED) 0.907 8.946 187 0.004 Inter-orbital width (IO) 0.982 1.551 187 0.216 Maxillary length (ML) 1.000 0.043 187 0.837 head height (hht) 0.966 3.032 187 0.085 Dorsal-fin spines (DS) 0.713 34.966 187 < 0.001 Dorsal-fin rays (DR) 0.662 44.403 187 < 0.001 Dorsal-fin elements (total) (DT) 0.468 98.957 187 < 0.001 Anus in relation to dorsal fin spines (ANDF) 0.946 4.952 187 0.029 Anal-fin rays (AF) 0.578 63.525 187 < 0.001 Anal fin spines in relation to dorsal fin spines and soft rays (ASDF) 0.891 10.620 187 0.002 Pre-caudal vertebrae (PCV) 0.237 280.098 187 < 0.001 Caudal vertebrae (CV) 0.984 1.418 187 0.237 Total vertebrae (TV) 0.136 553.803 187 < 0.001 Aphanopus carbo and A. intermedius from the northeastern Atlantic Biscoito e t a l . 26 Cybium 2011, 35(1) Table V. - Correlation matrix among morphometric and meristic variables to select variables with less relation (< 0.400). Variables Pal Pasl Panl PPl PDl P1Dfl Hmax h1sfa cPl Pol snl eD Ds Df Dt anDf af asDf PcV cV tV Pre-anal length (Pal) 1.000 Pre-first anal spine length (PASL) 0.767 1.000 Pre-anus length (Panl) 0.511 0.587 1.000 Pre-pectoral length (PPL) 0.189 0.397 0.358 1.000 Pre-dorsal length (PDl) 0.390 0.553 0.386 0.424 1.000 Pre-first dorsal soft ray length (P1Dfl) 0.243 0.343 0.249 0.112 0.250 1.000 Maximum body depth (Hmax) 0.274 0.335 0.360 0.311 0.379 0.219 1.000 Depth of body at level of first anal spine (H1SFA) 0.011 0.133 -0.014 -0.028 0.281 0.094 0.216 1.000 Caudal peduncle length (CPL) -0.104 -0.147 -0.021 -0.130 -0.191 0.029 -0.114 0.124 1.000 Pre-opercular length (POL) -0.041 -0.005 0.117 -0.103 0.150 -0.045 -0.007 0.322 0.032 1.000 snout length (snl) 0.101 0.027 0.063 -0.178 0.183 -0.049 -0.062 0.250 0.275 0.269 1.000 Eye diameter (ED) -0.089 0.001 0.029 0.271 -0.028 0.044 0.104 0.023 -0.088 -0.078 -0.289 1.000 Dorsal-fin spines (DS) 0.256 0.294 0.203 -0.104 0.101 0.225 0.083 0.036 -0.034 -0.031 0.006 -0.105 1.000 Dorsal-fin rays (DF) -0.366 -0.319 -0.402 0.055 -0.340 -0.234 -0.173 -0.211 -0.228 -0.034 -0.280 0.048 -0.237 1.000 Dorsal-fin elements (total) (DT) -0.193 -0.126 -0.259 -0.026 -0.263 -0.092 -0.110 -0.176 -0.237 -0.040 -0.259 -0.028 0.372 0.811 1.000 Anus in relation to dorsal fin spines (anDf) 0.262 0.334 0.051 0.029 0.186 0.134 0.210 0.198 -0.173 -0.052 -0.033 -0.058 0.401 -0.179 0.078 1.000 Anal-fin rays (AF) -0.154 -0.051 0.031 0.294 -0.164 -0.050 -0.122 -0.131 -0.217 0.067 -0.269 0.170 -0.126 0.282 0.190 -0.093 1.000 Anal-fin spines in relation to dorsal-fin spines and soft rays (ASDF) -0.123 -0.113 0.059 0.163 -0.009 -0.169 0.022 -0.354 -0.005 0.017 -0.124 0.074 -0.092 0.104 0.036 -0.450 0.037 1.000 Pre-caudal vertebrae (PCV) 0.043 0.116 0.161 -0.055 -0.184 -0.018 -0.077 -0.172 0.140 -0.014 0.050 -0.023 0.193 0.005 0.121 -0.139 -0.126 0.228 1.000 Caudal vertebrae (CV) -0.130 -0.185 -0.196 -0.103 0.035 -0.065 -0.065 0.126 -0.114 0.047 0.024 0.098 -0.127 0.067 -0.008 0.024 0.136 -0.287 -0.760 1.000 Total vertebrae (TV) -0.142 -0.130 -0.090 -0.229 -0.186 -0.122 -0.198 -0.032 0.008 0.053 0.101 0.120 0.058 0.107 0.144 -0.144 0.044 -0.141 0.141 0.536 1.000 Bi s c o i t o e t a l . Aphanopus carbo and A. intermedius from the northeastern Atlantic Cybium 2011, 35(1) 27 the exception of DF and ANDF, presented highest F-score (Tab. IV) and the lowest correlation (Tab. V) among them. The first canonical discriminant analysis (CDA) was constructed with morphometric and meristic data, the latter being the most important. Total number of vertebrae was selected as first meristic variable and among morphometric variables only eye diameter was included in the function (Tab. VI). Canonical correlation index was 0.953 with 100% classification success (Tab. VII). The second CDA was calculated using only morphometric variables. Depth of body at level of first anal spine was the variable showing highest differences between species (Tab. VI). The canonical correlation index obtained was 0.607 with 86.4% classification success and Cohen’s κ indicated a classification efficiency of 68% (Tab. VII). discussion And conclusions The genetic structure obtained from the two mtDNA markers supports the findings reported by Stefanni and Knutsen (2007) therefore confirming the validity of both species (A. carbo and A. intermedius). The current work also provides more details on the geographical distribution of the two species. It is confirmed that the only species that reaches mainland Europe is A. carbo and extending southwards to at least 27ºN, off the Western Sahara coast. This southern limit of distribution of A. carbo was until present set with certitude to about 30ºN (Nakamura and Parin, 1993). Concerning A. intermedius, it has been found living in sympatry in the islands of the Azores, Madeira and the Canaries and off the coasts of Morocco and Western Sahara, therefore contributing to the clarification of the northern limit of its distribution, as already proposed by Nakamura and Parin (1993). The values of genetic diversity (at intraand inter-specific levels) are of similar order of magnitude as reported in Stefanni and Knutsen (2007). Pairwise values of Φst for the two species indicate high level of divergence and phylogenetic trees constructed from the sequence alignment of the two mtDNA markers propose only two monophyletic clades. Bootstrap supports are very strong between the two phylogroups but very weak within either of the two (Fig. 4), suggesting the presence of single populations in the NE Atlantic for both species. The calculation of the divergence time between the two species based on the COI sequences are supporting a recent speciation event between A. carbo and A. intermedius as reported in Stefanni and Knutsen (2007). For most morphometric relationships and meristic characters, data obtained have enlarged their previously published ranges for each species (Tab. I). These new ranges contributed in most cases to increase the overlap between the two species and therefore reinforcing their closeness. This enlargement may be due to the size of the sample studied (74 A. carbo and 71 A. intermedius), apparently the largest used Step Variables introduced Wilks’ lambda F exactly statistic df1 df2 df3 statistic df1 df2 Significant Morphometric and meristic data 1Total vertebrae (TV) 0.136 1 1 87 553.8031 187 0.000 2Pre-caudal vertebrae (PCV) 0.106 2 1 87 364.4042 286 0.000 3Anal-fin rays (AF) 0.096 3 1 87 265.5407 385 0.000 4Eye diameter (ED) 0.092 4 1 87 208.0681 484 0.000 Morphometric data 1Depth of body at level of first anal spine (H1SFA) 0.767 1 1 87 26.4696 187 0.000 2Caudal peduncle length (CPL) 0.699 2 1 87 18.5286 286 0.000 3Eye diameter (ED) 0.663 3 1 87 14.3782 385 0.000 4 Pre-anal length (Pal) 0.632 4 1 87 12.2245 484 0.000 Table VI. - Order of variables and value of Wilks’ statistics obtained during the stepwise procedure of discriminant analysis. Table VII. - Results of discriminant functions for identifying species. Actual group Predicted group membership Correct identification (%) Misidentification (%) cohen’s kappa A. carbo A. intermedius Morphometric and meristic data 100.0 0.0 1.000 A. carbo 53 (100%) 0 (0%) A. intermedius 0 (0%) 36 (100%) Morphometric data 84.3 15.7 0.680 A. carbo 46 (86.8%) 7 (13.2%) A. intermedius 7 (19.4%) 29 (80.6%)