Title: Phylogeny of the Serrasalminae Characiformes based on mitochondrial DNA sequences
1Phylogeny of the Serrasalminae (Characiformes)
based on mitochondrial DNA sequences Guillermo
Ortí1, Jorge I. R. Porto2 and Arjun
Sivasundar1 1School of Biological Sciences,
University of Nebraska Lincoln, Lincoln, NE
68588-0118, USA 2Instituto Nacional de Pesquisas
da Amazonia (INPA/CPBA), CxP 478, CEP 69.
083-000, Manaus, AM, Brazil
12S and 16S rRNA sequences (Fig. 2) Mitochondrial
rRNA sequences from a total of 55 serrasalmin
taxa were analyzed. All 12S and 16S sequences
from other characiform fishes available from
GenBank were used to determine the outgroup for
these analyses. From among all putative
characiform outgroup taxa, the eight with the
shortest branch lengths in the maximum likelihood
(ML) analysis were chosen to root the serrasalmin
phylogeny. The 12S and 16S sequences were
concatenated for analyses the total length of
the alignment was 896 bp. Sequence divergence
ranged from 0 to 8.4 among the ingroup, and from
6.6 to 12 between the serrasalmins and the
outgroup taxa. The ML tree (Fig. 2) strongly
supports the monophyly of the Serrasalminae,
containing three distinct clades (1) the
"piranha" clade (red), (2) the Myleus clade
(blue), and (3) the Colossoma clade (green). The
placement of Acnodon as the sister group to clade
(12), it is not well supported by bootstrap
analysis.
Introduction The subfamily Serrasalminae is
widely distributed in all the major river systems
of South America and contains at least 60 species
in 14 genera. These include the predatory
piranhas, the seed-eating tambaquís and
pacus. Several species are economically
important. Eigenmann (1915) used dental
characters to erect the subfamilies Serrasalminae
(six genera) and Mylinae (nine genera).
Machado-Allisons (1983) proposed phylogeny also
contained these two lineages within the subfamily
(Fig. 1). Ortí et al. (1996) proposed a phylogeny
of the group based on the 12S and 16S rRNA
mitochondrial genes, which contained three
distinct lineages rather than two. In this study,
we extend the taxonomic sampling of the 12S and
16S sequence data set, for a total of 55
serrasalmin taxa. In addition, to obtain
increased resolution we use mtDNA sequence data
within these groups, we employ sequences from the
mitochondrial D-loop region, which display a more
rapid rate of evolution than the previously used
rRNA genes. Our sampling comprises 38 taxa
representing all genera in the subfamily. This
study represents the most comprehensive molecular
systematic treatment of this group to
date. Methods The following primers were used for
PCR and direct sequencing 12S fragment L1091
and H1478 (Kocher et al., 1989), 16S fragment
16Sar-L and 16Sbr-L (Palumbi et al., 1991).
D-loop F-TTF (5-GCCTAAGAGCATCGGTCTTGTAA) and
F-12R (5-GTCAGGACCATGCCTTTGTG). Additional
internal primers were used for sequencing.
Sequences were aligned using ClustalW (Gibson et
al., 1996) Modeltest 3.0 (Posada and Crandall,
1998) was used to determine the optimal model of
nucleotide evolution for each sequence data set.
The model and parameters estimated were used to
perform maximum likelihood and distance (minimum
evolution, ME) analyses. All phylogenetic
analyses were conducted using PAUP v4 (Swofford,
2000).
Control region (D-loop) sequences (Fig.
3) Mitochondrial D-loop sequences from 38
serrasalmin taxa were analyzed. Due to the high
level of sequence divergence (1.7 to 25.5),
alignment was problematic. About 200 bp of
sequence was excluded from the analyses, since
these constituted alignment ambiguous sites.
The same three groups identified from Fig. 2 are
present, with the exception that Acnodon is
placed within the Myleus group. However, shallow
branches within the piranha clade do not receive
strong support. To allow for more rigorous
analyses to resolve within-group relationships,
the D-loop sequences were divided into two data
sets, the piranha group and the Myleus
Colossoma groups.
Fig. 1 Machado-Allison's (1983) hypothesis of
serrasalmin relationships
Fig.2 12S/16S ML tree TrN model with invariable
sites (I) and among-site rate variation (gamma
shape, G). Bootstrap support is from parsimony.
Fig. 3 D-loop ML tree for all taxa TMV I G
model. Bootstrap support from MP is indicated
above branches.
- The piranha group (Fig. 4)
- Sequence variation within this clade for the
D-loop region ranged from 0.2 to 21.6. - Three divergent groups are seen - Metynnis, the
Catoprion-Pygopristis-Pristobrycon striolatus
group and the Serrasalmus-Pristobrycon-Pygocentrus
group. - Given its basal position within this clade (Figs.
2 and 3), Metynnis was used to root this tree
this is also consistent with the hypothesis shown
in Fig. 1. - Pristobrycon striolatus forms a well-supported
group with Catoprion and Pygopristis. - Serrasalmus gouldingi is more closely related to
the remaining Pristobrycon than to other
Serrasalmus. - The genus Pygocentrus forms a monophyletic unit
within Serrasalmus.
- Conclusions
- This is the most complete molecular systematic
study of this group to date, including taxa from
all genera of the subfamily. - As expected, the higher level of variation in the
D-loop compared to the rRNA genes provides better
resolution of the relationships within each of
the clades. - Pristobrycon, with the exception of P.
striolatus, forms a clade with Serrasalmus
gouldingi that is the sister group to the
Serrasalmus-Pygocentrus clade. - The genera Serrasalmus and Prisobrycon are
paraphyletic. Pristobrycon striolatus, which has
previously been regarded as quite distinct from
its congeners (Machado-Allison et al., 1989),
forms a well-supported group with Catoprion and
Pygopristis. This is consistent with the results
of Ortí et al. (1996). - The monophyly of the subfamily, and the presence
of three lineages within it are well supported,
in contrast to Machado-Allisons (1983)
hypothesis. In particular, the Colossoma-Mylossoma
-Piaractus is the most basal with Colossoma and
Mylossoma as sister taxa. - Future studies including nuclear gene sequences
and morphological characters should help resolve
some of the ambiguous relationships and provide a
strong foundation for taxonomic revision of the
group - Literature cited
- Gibson T et al. ClustalW, EMBL, Heidelberg,
Germany and EMBL/EBI, Hinxton, UK, (1996). - Kocher TD et al. Proceedings of the National
Academy of Science USA, 86, 6196, (1989). - Machado-Allison A. Acta Biológica Venezuélica,
11, 145 (1983). - Machado-Allison A et al. Acta Biológica
Venezuélica, 12, 140, (1989). - Ortí G et al. Journal of Molecular Evolution, 42,
169, (1996). - Palumbi S et al. The simple fool's guide to PCR.
University of Hawaii, Honolulu, (1991). - Posada D and KA Crandall. Bioinformatics 14, 817
(1998). - Swofford DL. PAUP. Version 4, Sinauer
Associates, Sunderland, Massachusetts, (2000). - Acknowledgements
- The Myleus and Colossoma groups (Fig. 5)
- Sequence variation ranged from 1.2 to 25.4.
- The clade consisting of Colossoma, Mylossoma and
Piaractus was used to root this tree based on its
position in the phylogeny from other analyses
(Figs. 2 and 3). - Acnodon was included with these groups based on
Fig. 3, since its placement as shown in Fig. 2 is
not well supported. As seen in the ML tree,
Acnodon falls within the Myleus group, as does
the genus Ossubtus. - The paraphyly of Myleus is evident M. pacu, in
particular, is more closely related to Mylesinus
and Tometes than to other Myleus.
Fig. 4 ML tree for D-loop for the piranha group
HKY G model. Numbers above branches show
bootstrap support from ML (top), MP (bottom) and
below branches, from ME
Fig. 5 ML tree for D-loop for the Myleus and
Colossoma groups TVM I G model. Bootstrap
support from MP (above branches) and ME (below
branches)
Specimen information available from the authors
upon request