288
Wei-Jen Chen et al.
DNA extraction was performed following Lecointre et al. [8]. DNA
amplification was performed by polymerase chain reaction (PCR) [9,10].
Double-stranded PCR products were sequenced in two ways: direct
sequencing or sequenced after a cloning step. Primer sequences, PCR
protocols and sequencing protocols are available on request (Chen [7]).
Obtained sequences were read and entered twice using the computer
package MUST [11], and aligned using the facilities of ED, within MUST.
MUST includes the neighbor-joining (NJ) method [12] and allows very fast
bootstrap analyses with this tree-construction method using the NJBOOT
program. The main phylogenetic analyses were performed using P AUP
3.1.1. [13]. Branch and Bound searches were performed with the whole
species set (15 species). Relative and intrinsic saturation were evaluated for
transitions and transversions separately using the methods described in
Lecointre et al. [8]. No weighting schemes were used because of absence
of transitional saturation. Bootstrap analyses [14] were performed with
PAUP, using Branch and Bound searches and 500 iterations, and with NJ
using 1000 iterations. Morphological data were evaluated using P AUP and
MacClade [15].
Results
Sequence Analysis
Sequences are available m GenBank under accession numbers from
AF037109 to AF037138.
Both the alignment of 16S mtDNA sequences (500 bp) and D2 (611 bp)
and D8 (443 bp) of the 28S rDNA showed that these genes were not
variable enough to allow the inference of channichthyid phylogeny. For the
cytochrome b data, DNA sequences from 15 species (including one outgroup taxon) were aligned according to comparison with Trachurus
trachurus (Teleostei, Perciformes). Of the total 752 sites, 161 (21%) are
variable within the ingroup and 89 (12%) are informative for parsimony (at
least two types of nucleotides each represented at least twice). When codon
positions are considered, the third codon position provides 88% of the
variability. A test of absolute saturation of transitional substitution in the
third codon position showed no saturation (not shown, the regression
coefficient of the plot is 0.91). For the D-Ioop data, the size of the D-loop
sequences are not the same through taxa because of numbers of
deletion/insertion events. After alignment, 383 positions were obtained for
all species studied including 187 (49%) variable sites in which 99 (26%)
sites are informative for parsimony. Chionodraco rastrospinosus has the
longest sequence owing to a significant insertion for 12 cytosines in the 5'
Wei-Jen Chen et al.
DNA extraction was performed following Lecointre et al. [8]. DNA
amplification was performed by polymerase chain reaction (PCR) [9,10].
Double-stranded PCR products were sequenced in two ways: direct
sequencing or sequenced after a cloning step. Primer sequences, PCR
protocols and sequencing protocols are available on request (Chen [7]).
Obtained sequences were read and entered twice using the computer
package MUST [11], and aligned using the facilities of ED, within MUST.
MUST includes the neighbor-joining (NJ) method [12] and allows very fast
bootstrap analyses with this tree-construction method using the NJBOOT
program. The main phylogenetic analyses were performed using P AUP
3.1.1. [13]. Branch and Bound searches were performed with the whole
species set (15 species). Relative and intrinsic saturation were evaluated for
transitions and transversions separately using the methods described in
Lecointre et al. [8]. No weighting schemes were used because of absence
of transitional saturation. Bootstrap analyses [14] were performed with
PAUP, using Branch and Bound searches and 500 iterations, and with NJ
using 1000 iterations. Morphological data were evaluated using P AUP and
MacClade [15].
Results
Sequence Analysis
Sequences are available m GenBank under accession numbers from
AF037109 to AF037138.
Both the alignment of 16S mtDNA sequences (500 bp) and D2 (611 bp)
and D8 (443 bp) of the 28S rDNA showed that these genes were not
variable enough to allow the inference of channichthyid phylogeny. For the
cytochrome b data, DNA sequences from 15 species (including one outgroup taxon) were aligned according to comparison with Trachurus
trachurus (Teleostei, Perciformes). Of the total 752 sites, 161 (21%) are
variable within the ingroup and 89 (12%) are informative for parsimony (at
least two types of nucleotides each represented at least twice). When codon
positions are considered, the third codon position provides 88% of the
variability. A test of absolute saturation of transitional substitution in the
third codon position showed no saturation (not shown, the regression
coefficient of the plot is 0.91). For the D-Ioop data, the size of the D-loop
sequences are not the same through taxa because of numbers of
deletion/insertion events. After alignment, 383 positions were obtained for
all species studied including 187 (49%) variable sites in which 99 (26%)
sites are informative for parsimony. Chionodraco rastrospinosus has the
longest sequence owing to a significant insertion for 12 cytosines in the 5'