The complete mitochondrial DNA sequence of the Mekong giant catfish (Pangasianodon gigas), and the phylogenetic relationships among Siluriformes.

Abstract

The Mekong giant catfish (Pangasianodon gigas) is the largest scale-less freshwater fish of the world, and a critically endangered species. We determined the complete nucleotide sequence (16,533 bp) of the mitochondrial genome of the Mekong giant catfish, and conducted phylogenetic analyses based on mitochondrial protein (the combined amino acid sequences of all 13 mitochondrial protein coding genes) and rRNA (the combined nucleotide sequences of mitochondrial 12S and 16S rRNA genes) data sets in order to further clarify the relative phylogenetic position of P. gigas, and to recover phylogenetic relationships among 15 out of the 33 families of Siluriformes. Phylogenetic analyses (maximum parsimony, minimum evolution, maximum likelihood, Bayesian inference) of the protein data set were congruent with a basal split of the order into Loricarioidei and Siluroidei, and supported a closer relationship of the Mekong giant catfish (family Pangasiidae) to Siluridae than to Bagridae. The rRNA-based Bayesian phylogeny recovered Callichthyidae as the sister group of all other analyzed non-diplomystid catfish families, rendering Loricarioidei paraphyletic. In addition, Loricariidae were recovered as paraphyletic due to the inclusion of Astroblepidae. However, none of the two relationships received bootstrap support in the maximum parsimony, minimum evolution, and maximum likelihood analyses, and should be interpreted with caution. The derived position of Cetopsidae within Siluroidei, the sister group relationship of Pseudopimelodidae and Pimelodidae, and a close relationship of Doradidae and Auchenipteridae to the exclusion of Mochokidae were strongly supported. Pangasiidae was placed as a single lineage without clear affinities.

Full text

Page 1

The complete mitochondrial DNA sequence of the Mekong giant catfish
(Pangasianodon gigas), and the phylogenetic relationships
among Siluriformes
Amnuay Jondeunga,⁎, Pradit Sangthonga, Rafael Zardoyab
aDepartment of Genetics, Kasetsart University, Bangkok 10900, Thailand
bDepartamento de Biodiversidad y Biología Evolutiva, Museo Nacional de Ciencias Naturales, CSIC,
José Gutiérrez Abascal, 2, 28006 Madrid, Spain
Abstract
The Mekong giant catfish (Pangasianodon gigas) is the largest scale-less freshwater fish of the world, and a critically endangered species. We
determined the complete nucleotide sequence (16,533 bp) of the mitochondrial genome of the Mekong giant catfish, and conducted phylogenetic
analyses based on mitochondrial protein (the combined amino acid sequences of all 13 mitochondrial protein coding genes) and rRNA (the
combined nucleotide sequences of mitochondrial 12S and 16S rRNA genes) data sets in order to further clarify the relative phylogenetic position
of P. gigas, and to recover phylogenetic relationships among 15 out of the 33 families of Siluriformes. Phylogenetic analyses (maximum
parsimony, minimum evolution, maximum likelihood, Bayesian inference) of the protein data set were congruent with a basal split of the order
into Loricarioidei and Siluroidei, and supported a closer relationship of the Mekong giant catfish (family Pangasiidae) to Siluridae than to
Bagridae. The rRNA-based Bayesian phylogeny recovered Callichthyidae as the sister group of all other analyzed non-diplomystid catfish
families, rendering Loricarioidei paraphyletic. In addition, Loricariidae were recovered as paraphyletic due to the inclusion of Astroblepidae.
However, none of the two relationships received bootstrap support in the maximum parsimony, minimum evolution, and maximum likelihood
analyses, and should be interpreted with caution. The derived position of Cetopsidae within Siluroidei, the sister group relationship of
Pseudopimelodidae and Pimelodidae, and a close relationship of Doradidae and Auchenipteridae to the exclusion of Mochokidae were strongly
supported. Pangasiidae was placed as a single lineage without clear affinities.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Pangasiidae; Loricarioidei; Siluroidei; Mitochondrial genome; rRNA
1. Introduction
The Mekong giant catfish (Pangasianodon gigas Chevey
1930) is the world's largest scale-less freshwater fish. This
species can reach about 3 m in length, and can weight up to
more than 300 kg (Roberts and Vidthayanon, 1991). The giant
catfish is an endemic species of the main channel and
tributaries of the Mekong River of Thailand, Laos, Cambodia,
and Vietnam (Berra, 2001). For many years, the giant catfish
was heavily fished, but catches have seriously declined in
recent years, and the species is presently considered as
critically endangered (www.redlist.org). Nevertheless, the
giant catfish is successfully bred in captivity, and has shown
Abbreviations: ATPase 6–8, genes encoding for ATP synthase subunits
6 and 8; bp, base pair; BI, Bayesian inference; BP, bootstrap proportion; BPP,
Bayesian posterior probability; CI, consistency index; CSB 1–3, conserved
séquense blocks 1–3; cytb, gene encoding for cytochrome oxidase b; COI–III,
genes encoding for cytochrome oxidase subunits I–III; ETAS, extended
termination-associated sequences; GTR, general time reversible model; ME,
minimum evolution; ML, maximum likelihood; MP, maximum parsimony;
mtDNA, mitochondrial DNA; NAD1–6 and NAD4L, genes encoding for
NADH dehydrogenase subunits 1–6 and 4L; nt, nucleotides; PCR, polymerase
chain reaction; RI, retention index; 16S rRNA and 12S rRNA, genes encoding
for the large and small (12S) subunits of ribosomal RNA.
⁎Corresponding author. Tel.: +66 2 9428716; fax: +66 2 5795528.
E-mail address: fscianj@ku.ac.th (A. Jondeung).

Page 2

one of the most rapid growth rates of any freshwater fish
(Roberts and Vidthayanon, 1991).
The genus Pangasianodon is monotypic, and belongs to the
family Pangasiidae together with Pangasius (26 valid species)
and Helicophagus (3 valid species) (www.fishbase.org; Roberts
and Vidthayanon, 1991). The monophyly of the Pangasiidae
awaits definition of synapomorphies (Diogo, 2004a; but see
dePinna, 1993). The family is distributed throughout Southern
Asia (Berra, 2001), and exhibits great morphological and
ecological diversity (Pouyaud et al., 2000). In particular,
pangasiid species are characterized by their great differences in
body size. Like P. gigas, Pangasius pangasius and Pangasius
sanitwongsei are also large species that can reach up to 3 m
(www.fishbase.org; Roberts and Vidthayanon, 1991). Medium-
sized species achieve a maximum length of around 80–100 cm
(e.g. Pangasius bocourti or Pangasius larnaudii), and small
species vary between 20 and 40 cm (e.g. Pangasius maha-
kamensis or Pangasius myanmar) (www.fishbase.org; Roberts
and Vidthayanon, 1991). The only phylogenetic study of
Pangasiidae was based on allozyme and partial mitochondrial
cytochrome b (cytb) gene sequence data (Pouyaud et al., 2000).
Although phylogenetic relationships of recently diverged groups
such as e.g. P. gigas and Pangasius hypophthalmus were
supported, deeper phylogenetic relationships remained largely
unresolved (Pouyaud et al., 2000).
The relative phylogenetic position of Pangasiidae within
the highly diversified order Siluriformes is elusive (Mo, 1991;
dePinna, 1998; Diogo, 2004a; Hardman, 2005). The family
Schilbeidae is often suggested as a close relative of Pangasiidae
based on morphological data (Mo, 1991; dePinna, 1998; Diogo,
2004b). However, the monophyly and phylogenetic intrarela-
tionships of the Schilbeidae await a detailed phylogenetic study
(Teugels, 2003). Furthermore, a recent molecular phylogeny
based on complete mitochondrial cytb gene tentatively placed
Pangasiidae as a sister group of Claroteidae without statistical
support (Hardman, 2005). More generally, phylogenetic
relationships among the 33–37 extant families (438 genera,
2753 species; Teugels, 2003) of Siluriformes remain uncertain
(Mo, 1991; dePinna, 1998; Diogo, 2004a,b; Hardman, 2005).
Morphology- and molecular-based phylogenies strongly sup-
port Diplomystidae as a sister group of the remaining catfish
families (Mo, 1991; dePinna, 1998; Diogo, 2004a). However,
the identity of the most basal taxon within non-diplomystid
Siluriformes is contentious (Diogo, 2004a). Several morpho-
logical studies agree on placing Cetopsidae (including Helo-
genes) as sister group of all other extant non-diplomystid
Siluriformes (Mo, 1991; dePinna, 1998; Diogo, 2004a; but see
Diogo, 2004b). Unfortunately, sampled species of the family
Cetopsidae, and in particular Helogenes marmoratus, exhibited
extremely long branches in cyt b-based phylogenies (Hardman,
2005), which apparently hindered the correct placement of the
group. Maximum parsimony (MP) analyses recovered Helo-
genes (but not other Cetopsidae) as the most basal non-
diplomystid siluriform whereas Bayesian inference (BI) placed
all Cetopsidae in a highly derived position but without forming
a monophyletic group (Hardman, 2005). Phylogenetic relation-
ships among the remaining catfish families are highly uncertain
both on morphological (Mo, 1991; dePinna, 1998; Diogo,
2004a) and molecular (Hardman, 2005) grounds. Nevertheless,
two suprafamiliar groupings are repeatedly recovered in
morphology-based phylogenetic analyses. The monophyly of
Loricarioidei including Nematogenyidae, Trichomycteridae,
Callichthyidae, Scoloplacidae, Loricariidae, and Astroblepidae
is particularly well supported based on morphological data
(Diogo, 2004a). Within the remaining non-dyplomystid cat-
fishes, there is relative support for a monophyletic clade termed
Doradoidea that would include at least the African family
Mochokidae, and the Neotropical sister groups Doradidae and
Auchenipteridae (dePinna, 1998; Diogo, 2004a).
In this study, we present the complete mitochondrial genome
sequence of the Mekong giant catfish, P. gigas. Phylogenetic
relationships of the Mekong giant catfish and, more generally,
of the family Pangasiidae need to be firmly established. More-
over, both phylogenetic questions are directly related to the
resolution of siluriform systematics. Therefore, we conducted
phylogenetic analyses based on the new mitochondrial se-
quence data with the main aim of tentatively clarifying the
relative phylogenetic position of Pangasiidae, and the phyloge-
netic relationships among siluriform families.
2. Materials and methods
2.1. Mitochondrial DNA extraction and cloning, PCR
amplification, and sequencing
MitochondrialDNA(mtDNA)wasisolatedfromthefreshliver
of a specimen of Mekong giant catfish (P. gigas) as previously
described (Palva and Palva, 1985). Purified mtDNAwas cleaved
with Mbo I, and cloned into pUC18 at the Bam HI site. Three
positive clones were sequenced with universal M13 primers in an
automated DNA sequencer ABI3700, using the BigDye Deoxy
Terminator cycle-sequencing kit (Applied Biosystems; Foster
City, CA), and following manufacturer's instructions.
TheclonedfragmentswerelocatedbetweentheND4andND5
genes, within the WANCY region, and within the 16S rRNA
gene,respectively.Theirnucleotidesequencesallowedthedesign
of three pairs of PCR primers (Table 1), which were used to
amplify three long and contiguous PCR fragments that covered
Table 1
Specific primers used for PCR amplification of the mtDNA of Panganasiodon
gigas
PrimerSequence (5′→3′) Fragment
P5
P6
P7
P10
P8
P12
AB
BA
BC
CB
CA
AC
CCGAACTTAAGTTATAGCTGGTTGC
CCTAGCTGTTAACTAGGAGTTTGT
CCTACAAACTCCTAGTTAACAGC
TGGTTCCTAAGACCAACGGATG
TAGTTCATCCGTTGGTCTTAGG
CCAGCTATAACTAAGTTCGGTAGG
TAGGCACTTAGCGCCCTACT
AGCGCGTGATCGTCACAGG
AGGATAATAGTTCATCCGTTGGTC
TTGGCGGTCGGCTCGACC
GTACCGCAAGGGAACGCTG
TGGCTGCTTTTAGGCCCAC
A
B
C
Overlapping AB
Overlapping BC
Overlapping CA

Page 3

the entire mitochondrial genome (Fig. 1). Long PCRs were
conducted in a 9600 GeneAmp PCR system (Applied Biosys-
tems, Foster City, CA) using the following cycling conditions: an
initial denaturing step at 94 °C for 1 min; 30 cycles of denaturing
at 94 °C for 20 s, annealing at 50 °C for 30 s, and extending at
68 °C for 7 min. PCR amplifications were performed in 50 μl
reactions containing 30.5 μl distilled H2O, 5 μl 10× LA PCR
buffer II (Takara Bio Inc., Shiga, Japan), 8 μl dNTPs (2.5 mM),
2μlofeachprimer(10μM),0.5μlofLATaqpolymerase(Takara
Bio Inc., Shiga, Japan), and 2 μl DNA template (10–100 ng).
Long PCR products were purified with the AMPure PCR
purification kit (Agencourt, Beverly, MA), and sequenced di-
rectly as described above. Long PCR primers were used for the
initial sequencing of the 5′ and 3′ ends of the PCR products
whereas the remaining nucleotide sequence was obtained by
primerwalkingusing31newlydesignedPCRprimers(available
on request from the authors). The obtained sequences averaged
700 bp in length, and each sequence overlapped the next contig
by about 150 bp. In no case were differences in sequence
observedbetweentheoverlappingregions.Finally,threepairsof
PCR primers (Table 1) were designed to amplify and sequence
the corresponding short fragments, which connected the three
long PCR fragments (Fig. 1). The complete nucleotide sequence
of the mitochondrial genome of the Mekong giant catfish has
beendepositedinGenBankunderaccessionnumberAY762971.
2.2. Sequence editing and phylogenetic analyses
The primary DNA sequence data were characterized using
BLAST searches at NCBI (www.ncbi.nlm.nih.gov). The
different genes were identified by sequence comparison with
catfish (AB054127 and AF482987) homologues. In addition,
the 13 mitochondrial protein-coding genes were delimited by
the presence of initiation and stop codons whereas the 22 tRNA
genes were recognized by their capability to fold into cloverleaf
secondary structures, and the presence of specific anticodons.
Phylogenetic relationships among Siluriformes were based on
twodifferentsequencedatasets.Thefirstone(proteindataset)was
aconcatenateddatasetincludingthededucedaminoacidsequences
ofall13mitochondrialprotein-codinggenesfromthe19speciesof
Ostariophysi (e.g. Saitoh et al., 2003) listed in Appendix 1. Two
Salmoniformes (Oncorhynchus mykiss, L29771 and Galaxias
maculatus, AP004104) and two Clupeiformes (Sardinops mela-
nostictus, AB032554 and Engraulis japonicus, AB040676) were
used as outgroup taxa. The second data set (rRNA data set)
included the combined nucleotide sequences of the mitochondrial
12S and 16S rRNA genes from the 49 species of Siluriformes (e.g.
Moyer et al., 2004) listed in Appendix 1. Two Characiformes
(Chalceus macrolepidotus, AB054130 and Phenacogrammus
interruptus, AB054129) were used as outgroup taxa.
The nucleotide sequences of the mitochondrial 12S and 16S
rRNA genes, and the deduced amino acid sequences of each of
the 13 complete mitochondrial protein-coding genes were
aligned separately, using the default parameters of CLUSTAL
X version 1.81 (Thompson et al., 1997). Alignments (available
from the authors upon request) were subsequently revised by eye
in an effort to maximize positional homology. The construction
of the combined protein data set was finalized after excluding
gapped positions. The rRNA data set included both complete and
partial sequences, and gaps were treated as missing data.
Following Lavoue et al. (2005) and Hardman (2005) that
analyzed similar data sets and phylogenetic questions, we
reconstructed phylogenetic relationships using Bayesian Infer-
ence (BI; Huelsenbeck and Ronquist, 2001). The best-fit models
ofaminoacidandnucleotidesubstitutionwereselectedaccording
to the Akaike Information Criterion (AIC) using Prottest version
1.2.6 (Abascal et al., 2005) and ModelTest version 3.6 (Posada
and Crandall, 1998), respectively. The best evolutionary models
selectedfortheproteinandrRNAdatasetsweremtREV+I+Γand
GTR+I+Γ, respectively. BI was performed using MrBayes
version 3.0b4 (Ronquist and Huelsenbeck, 2003) with four
simultaneous chains, each of 1×106generations, sampled every
100 generations. Trees sampled before the cold chain reached
stationarity (as judged by plots of ML scores) were discarded as
“burn-in”. Runs wererepeatedtwice.Support ofthe recoveredBI
trees was evaluated with Bayesian posterior probabilities (BPPs).
Fig. 1. Gene organization, and PCR amplification strategy for the Mekong giant catfish mitochondrial genome. Localization and direction of the PCR primers are
denoted by arrows (see Table 1 for the primer DNA sequence).

Page 4

Because Bayesian posterior probabilities have been shown to
be significantly higher than corresponding nonparametric boot-
strap frequencies (e.g. Erixon et al., 2003), we also conducted
unweighed maximum parsimony (MP), minimum evolution
(ME), and maximum likelihood (ML) inferences, and evaluated
support for the recovered trees using nonparametric boot-
strapping with 500 pseudoreplicates. MP and ME analyses
were carried out with PAUP⁎ 4.0b10 (Swofford, 2002) using
heuristic searches with 10 random stepwise additions of taxa
and TBR branch swapping. ME analyses of the amino acid and
nucleotide sequence data sets used mean- and GTR+I+G-
corrected distances. ML analyses were conducted with PHYML
version 2.4.4. (Guindon and Gascuel, 2003) using the cor-
responding best-fit evolutionary models.
3. Results and discussion
3.1. Genome organization
The total length of the L-strand of the Mekong giant catfish
mitochondrial genome is 16,533 bp. The overall base com-
position of the L-strand is A: 30.4%, C: 28.4%, G: 15.6% and T:
25.6%. As in other fishes (Miya et al., 2003) and vertebrates
(Boore, 1999), the new mitochondrial genome included two
rRNA, 22 tRNA, and 13 protein-coding genes (Fig. 1). The
organization of the Mekong giant catfish mitochondrial genome
conforms to the consensus gene order of other fish mitochon-
drial genomes (Fig. 1). All protein-coding genes use ATG as
start codon except COI that uses GTG. Five open reading
frames end with TAA (ND1, COI, ATPase 8, ND4L, and ND5),
two end with TAG (ND3 and ND6), and the rest have in-
complete stop codons, either TA (ND2, ATPase 6, and COIII)
or T (COII, ND4, and cytb), which may be likely completed by
post-transcriptional polyadenylation (Ojala et al., 1981). The
complete mitochondrial genome of another catfish, Corydoras
rabauti (Callichthyidae) exhibits a 17-nt insertion sequence
between the ATPase 6 and COIII genes (Saitoh et al., 2003).
This peculiarity is not found in the mitochondrial genome of
P. gigas where these genes are butt jointed to each other as in
most vertebrate mitochondrial genomes.
The control region oftheMekonggiant catfish mitochondrial
genome is 899 nucleotides in length, and it does not contain any
repeat. Positions 1–100 (5′-end) and 380–420 (central region)
were highly conserved when compared with other fish mito-
chondrial control regions. The conserved sequence blocks CSB-
1, CSB2 and CSB-3 (Walberg and Clayton, 1981), and several
extended termination-associated sequences ETAS (Sbisa et al.,
1997)wereunambiguouslyidentifiedinthe3′and5′endsofthe
control region, respectively. The origin of light-strand replica-
tion was identified in a cluster of five tRNA genes (WANCY
region) between tRNA-Asn and tRNA-Cys genes.
3.2. Phylogenetic relationships among Siluriformes
Even though phylogenetic intrarelationships of highly diver-
sified families of the order Siluriformes such as Loricariidae
(Montoya-Burgos et al., 1998), Doradidae (Moyer et al., 2004),
and Callicthyidae (Shimabukuro-Dias et al., 2004) are well
studied at the molecular level, main phylogenetic interrelation-
ships of siluriform families remain largely unresolved (Hardman,
2005). Catfish molecular systematics, and in particular, the
relative phylogenetic position of the family Pangasiidae were
analyzed based on two different (protein and rRNA) data sets.
The deduced amino acid sequences of all 13 mitochondrial
protein-coding genes of all siluriform and several ostariophysan
complete mitochondrial genomes available at NCBI databases
were combined into a single data set that produced an alignment
of 3647 positions. A total of 56 gapped positions were excluded
from further phylogenetic analyses. Of the remaining positions,
2031 were invariable, and 1001 were parsimony informative.
The best BI tree (Burn-in=70 trees; −lnL=41014.79) is shown
in Fig. 2. As expected, the monophyly of Siluriformes, and the
inclusion of the Mekong giant catfish within this order were
highly supported (100% BPP). Pangasianodon (Pangasiidae)
was recovered with maximal support as sister group of Ictalurus
(Ictaluridae) to the exclusion of Pseudobagrus (Bagridae). The
genus Corydoras (Callichthyidae) that is assigned to the sub-
order Loricarioidei was placed as sister group of all other
analyzed siluriforms (Siluroidei) (Fig. 2). MP (one most parsi-
monious tree of 6112 steps; CI=0.59; RI=0.48), ME (one tree
of score=1.39), and ML (one tree, −lnL=41783.47) recovered
identical and highly supported phylogenetic relationships
within Siluriformes (Fig. 2).
At a higher taxonomic level, the monophyly of Otophysi
(Cypriniformes, Gymnotiformes, Characiformes, and Siluri-
formes) (Fink and Fink, 1996; Johnson and Patterson, 1996)
was highly supported by all methods of phylogenetic inference,
whereas the monophyly of Ostariophysi including Gonorynch-
iformes and Otophysi to the exclusion of Clupeiformes (Fink
and Fink, 1996; Johnson and Patterson, 1996; Lavoue et al.,
2005; against Saitoh et al., 2003) was only strongly supported
by BI (98% BPP). The monophyly of the different orders within
Ostariophysi is well supported based on morphological grounds
(Fink and Fink, 1996). However, none of the methods of
phylogenetic inference was able to recover the monophyly of
Characiformes based on the mitochondrial sequence data set. In
addition, MP and ME recovered Gonorynchiformes as para-
phyletic (not shown). These unorthodox results are likely
related with the fact that one of the two species representing
each of the mentioned orders exhibits relatively long branches
(Fig. 2), which may negatively affect phylogenetic reconstruc-
tions. Discrepancies among the different methods regarding the
phylogenetic relationships within Cypriniformes (not shown)
are also likely related with inference artifacts due to the
extremely long branch of Danio rerio.
The rRNA data set had a denser taxon sampling of siluriform
families. This data set included the almost complete mitochon-
drial 12S and 16S rRNA genes of 33 species representing nine
catfish families. In trying to maximize the number of studied
families, partial mitochondrial 12S and 16S rRNA genes (that
were amplified with universal primers; Kocher et al., 1989;
Palumbi et al., 1991) of 16 species representing six additional
families were also included in the phylogenetic analyses. The
rRNA sequences were combined into a single data set that

Page 5

produced an alignment of 2789 positions. A total of 257 gapped
positions were excluded from further phylogenetic analyses. Of
the remaining positions, 1357 were invariable, and 828 were
parsimony informative. The best BI tree (Burn-in=400 trees;
−lnL=24700.64) is shown in Fig. 3. The Callichthyidae are
strongly supported as the sister group of all other analyzed
catfish families, which are recovered into two well-supported
main clades: one includes Loricariidae, Astroblepidae, and
Trichomycteridae whereas the other (Siluroidei) groups togeth-
er Bagridae, Malapteruridae, Mochokidae, Pangasiidae, Pseu-
dopimelodidae + Pimelodidae, Cetopsidae, Ictaluridae,
Siluridae, and Auchenipteridae + Doradidae (Fig. 3). The rela-
tively high resolution of basal nodes of the siluriform BI
phylogeny may be explained considering that this is a model-
based method of phylogenetic inference, which minimizes
saturation problems and is particularly robust to heterogeneity
of rates among lineages (Huelsenbeck and Ronquist, 2001).
However, it has been shown that BPP are generally significantly
higher than corresponding nonparametric bootstrap frequencies
(e.g. Erixon et al., 2003), and that priors assuming long internal
branches can cause high BPP for apparently incorrect clades
(Yang and Rannala, 2005). Therefore, we also conducted MP,
ME, and ML analyses to reconstruct siluriform phylogenetic
relationships, and tested the robustness of the recovered trees
with bootstrapping (Fig. 3).
The basal position of Callichthyidae is strongly supported in
the BI tree and implies the paraphyly of Loricariodei. ML (one
tree, −lnL=25507.47) also recovered Callichthyidae as the
most basal group of siluriforms, but without bootstrap support
(Fig. 3). Interestingly, neither MP (19 MP trees of 4902 steps)
nor ME (one tree of score=3.26) recovered this relationship.
Callichthyidae is normally grouped within Loricariodei based
on a minimum of 14 morphological synapomorphies (Diogo,
2004a), and the phylogenetic position of Callichthyidae as sister
group of the remaining catfish families has never been proposed
based on morphological evidence. The sampled species of
Callichthyidae show relatively long branches, which could be
affecting phylogenetic reconstruction by pulling Callichthyidae
Fig. 2. Phylogenetic relationships among Ostariophysi. The 50%-majority rule consensus of post-burn-in sampled trees from the Bayesian inference analysis based on
the protein data set under the mtREV24+I+Γ model is shown. Branch lengths are mean estimates. Asterisks indicate nodes with bootstrap values (BP) above 70% and
Bayesian posterior probabilities (BPP) above 95%. Numbers in the nodes are from top to bottom: ML, ME, MP BP, and BI BPP. Dashes indicate either BP below 50%
or BPP below 95%.