Phylogenetic relationships among the mitochondrial genomes of Anarhichas species and Lycodes toyamensis . The minimum length tree requires 2074 changes, with branch lengths as shown. The tree indicates that A. lupus and A. minor are each other’s closest relatives; this arrangement is supported by 96.9% of bootstrap replications. 

Context in source publication

Context 1

... nodes of the tree. Neighbor-joining phenetic analysis of uncorrected average pairwise distances ( p distances) was also done. To com- pare the extent of intra- and inter-specific mtDNA genome differentiation with other fish species, the four zoarcoid genome sequences were aligned with those of seven gadid codfish ( Gadus sensu lato; Carr et al. 1999) described in Coulson et al. (2006) (NCBI accession Nos. DQ356937–DQ356941 and DQ356946), with the addition of a second sequence from Gadus (= Theragra ) chalcogrammus Pallas, 1814 (NCBI accession No. NC004449) from the western Pacific Ocean. CR locus sequences, which were not available for all gadines, were removed, and PAUP* was used to removed any missing and (or) ambiguous characters. For the consensus data set of 15 082 bp, p distances were calculated and are shown as numbers of changes per kilo base pairs in Table 2. Variability of individual Anarhichas genes within genomes was calculated three ways: first as the average of the three pairwise sequence divergences among species ( p distance) and second as the density of interspecific single nucleotide polymorphisms (SNPs), calculated as the ratio of the number of SNPs observed among all three species to the total gene length, expressed as the number of SNPs / 100 bp. The two indices are closely related: the former will underestimate the actual extent of nucleotide divergence by not correcting for multiple-hit nucleotide substitutions at the same site, whereas the latter will underestimate variability by not counting the homoplasic changes implied by the inferred phylogeny. The latter provides a more intuitive measure of observed variability. The third measure of variability is the difference between the rates of synonymous ( d S ) and nonsynonymous ( d N ) nucleotide substitutions in each gene, according to the algorithm of Nei and Gojobori (1986) as implemented in MEGA version 3.1 (Kumar et al. 2004). This measure detects unusual patterns of amino acid substitution, for example, as a result of selection (cf. Coulson et al. 2006). The expected Poisson distribution of SNPs over tRNA loci was calculated by the recursive algorithm described by Zar (1999: 181). Nonparametric rank-order correlations of SNP density among protein-coding loci between the three Anarhichas species and the three Gadus codfish species ( G. morhua , G. (= T. ) chalcogrammus , and Gadus macrocephalus Tilesius, 1810) (Coulson et al. 2006) (DQ356938, DQ356939, and DQ356946), and comparisons of variability indices among Anarhichas loci made independently of absolute magnitude, were calculated by Kendall’s coefficient of rank correlation ( ) as implemented in BIOMstat version 3.30m (Applied Biostatistics, Inc. 2002; Sokal and Rohlf 1995: 593). The long-range PCR procedure amplified more than 97% of the complete mtDNA genome of A. lupus in three over- lapping fragments of 6835, 4555, and 4276 bp (Fig. 2). The sequence of the remaining 466 bp segment of the genome (including the 3 ’ end of the CYTB locus and the 5 ’ end of the CR locus) had been determined previously. The genomes of the other two wolffish species were amplified with various combination of the Anarhichas and Gadus specific primers in Table 1. The three genomes were sequenced with the same set of primers. The mtDNA genome sequences comprised 16 519 bp in A. lupus and 16 520 bp in the other two species. The sequences have been submitted to GenBank and assigned the accession numbers EF427916, EF427917, and EF427918 for A. lupus , A. minor , and A. denticulatus , respectively. The gene order is identical to that of L. toyamensis , except that L. toyamensis has an extra 169 bp segment 3 ’ to the tRNA– Val locus that is not present in Anarhichas species. This segment has been deleted from the alignment analyzed here. A total of 449 SNP sites were observed among the individuals representing the three species. Anarhichas lupus and A. minor differed by 248 pairwise differences (226 transitions and 22 transversions), A. minor and A. denticulatus by 274 differences (248 transitions and 26 transversions), and A. lupus and A. denticulatus by 286 differences (254 transitions and 32 transversions) (Table 2A). Interspecific p distances among Anarhichas species range from 16.64 to 20.22 substitutions / kbp (subs/kbp), which is approximately one- half t hat a mong G adus s pecies (from 38.46 t o 4 0.25 s ubs/kbp) (Table 2B) and about four to five times as large as the differences among transoceanic individuals within Gadus species ( from 3.25 to 4.77 subs/kbp). With L. toyamensis included as an outgroup, there are 1944 variable sites, of which 146 are phylogenetically infor- mative (sensu Nei 1987). The three possible bifurcating trees for four terminal taxa had lengths of 2074, 2094, and 2100 changes. The shortest tree indicates that A. lupus and A. minor are more closely related to each other than they are to A. denticulatus ; this tree is supported by 96.9% of 10 000 bootstrap replicates (Fig. 3). Random resampling of 1, 2, 4, or 8 kbp of the complete genome supported the same tree in 57.2%, 66.2%, 77.4%, and 89.4% of bootstraps, respectively. This tree is also supported by neighbour-joining distance analysis ( p distances) in 96.3% of bootstraps, as well as by weighted parsimony and various distance models with >95% bootstrap support in all cases (results not shown). The most variable protein-coding loci as measured by SNP density were ND4, CYTB, and ND2, with 4.40, 4.22, and 4.19 SNPs / 100 bp, respectively. The least variable loci were ATP8 (also the shortest region), COX2, and ND3, with 1.19, 1.57, and 1.99 SNPs / 100 bp, respectively (Table 3). The CR locus was less variable than 9 of 13 protein-coding regions (2.45 SNPs / 100 bp) (Table 4). Most tRNA loci were either invariant or had only 1 SNP site, and the 20 tRNA loci collectively showed only 18 SNP sites over 1410 bp (1.28 SNPs / 100 bp), which was lower than all except 1 protein-coding locus. Although 3 tRNA loci had three or more SNP sites and were thus more variable than any protein-coding locus (tRNA–Val, tRNA–Arg, and tRNA– Trp: >4 SNPs / 100 bp), the observed distribution did not de- part significantly from the expected random Poisson distribution of 18 SNPs over 20 tRNA loci ( 2 0 : 05 ; 1⁄2 3 = 4.21, 0.10 < p < 0.25). The 16S and 12S rRNA loci were the least variable gene regions (1.18 and 0.35 SNPs / 100 bp, respectively). The relative ranks of locus variability as measured by SNP density and mean p distance are highly correlated (Kendall’s = 0.9731, p << 0.01), as are the rankings of SNP density and the rate of synonymous substitution ( d S ) (Kendall’s = 0.7871, p << 0.01) and p distance vs. d S (Kendall’s = 0.8387, p << 0.01). The correlations of SNP density and p distance with d N – d S were slightly lower (Kendall’s = 0.7179 and 0.7949, respectively), but still p << 0.01 in both cases. Similarly, there is a significant rank-order correlation of SNP densities between Anarhichas species and Gadus species (Kendall’s = 0.4615, p = 0.0140), but the association is far from exact. ND4 ranked first in Anarhichas and second in Gadus ; however, the second- and third-ranked loci in Anarhichas (CYTB and ND2) ranked seventh and sixth in Gadus . The most highly ranked locus in Gadus (ND6) ranked fourth in Anarhichas , and the ATP6 and ND3 loci also had substantially lower ranks in Anarhichas (ninth and eleventh vs. fourth and fifth, respectively) (Fig. ...