Content uploaded by Leo G. Nico
Author content
All content in this area was uploaded by Leo G. Nico on Jul 31, 2014
Content may be subject to copyright.
Accepted by Gilbert: 31 Mar. 2007; published: 28 May 2007 1
ZOOTAXA
ISSN 1175-5326 (print edition)
ISSN 1175-5334 (online edition)
Copyright © 2007 · Magnolia Press
Zootaxa 1484: 1–38 (2007)
www.mapress.com/zootaxa/
Molecular systematics of Serrasalmidae: Deciphering the identities of piranha
species and unraveling their evolutionary histories
BARBIE FREEMAN1, LEO G. NICO2, MATTHEW OSENTOSKI1, HOWARD L. JELKS2 &
TIMOTHY M. COLLINS1,3
1Dept. of Biological Sciences, Florida International University, University Park, Miami, FL 33199, USA.
2United States Geological Survey, 7920 NW 71st St., Gainesville, FL 32653, USA. E-mail: Leo_Nico@usgs.gov
3Corresponding author. E-mail: CollinsT@fiu.edu
Table of contents
Abstract ...............................................................................................................................................................................1
Introduction......................................................................................................................................................................... 2
Overview of piranha diversity and systematics .................................................................................................................. 3
Material and methods........................................................................................................................................................ 10
Results............................................................................................................................................................................... 16
Discussion .........................................................................................................................................................................16
Conclusions....................................................................................................................................................................... 26
Acknowledgments............................................................................................................................................................. 26
References .........................................................................................................................................................................26
Appendix 1. Photographs of living and preserved specimens .......................................................................................... 30
Appendix 2. GenBank sequences used in this study. ........................................................................................................38
Abstract
Piranhas and their relatives have proven to be a challenging group from a systematic perspective, with difficulties in
identification of species, linking of juveniles to adults, diagnosis of genera, and recognition of higher-level clades. In this
study we add new molecular data consisting of three mitochondrial regions for museum vouchered and photo-docu-
mented representatives of the Serrasalmidae. These are combined with existing serrasalmid sequences in GenBank to
address species and higher-level questions within the piranhas using parsimony and Bayesian methods. We found robust
support for the monophyly of Serrasalmus manueli, but not for Serrasalmus gouldingi when GenBank specimens identi-
fied as S. gouldingi were included in the analysis. “Serrasalmus gouldingi” sequences in GenBank may, however, be
misidentified. Linking of juveniles to adults of the same species was greatly facilitated by the addition of sequence data.
Based on our sampling and identifications, our data robustly reject the monophyly of the genera Serrasalmus and Pristo-
brycon. We found evidence for a well-supported clade comprised of Serrasalmus, Pygocentrus, and Pristobrycon (in
part). This clade was robustly supported in separate and combined analyses of gene regions, and was also supported by a
unique molecular character, the loss of a tandem repeat in the control region. Analysis of specimens and a literature
review suggest this clade is also characterized by the presence of a pre-anal spine and ectopterygoid teeth. A persistent
polytomy at the base of this clade was dated using an independent calibration as 1.8 million years old, corresponding to
the beginning of the Pleistocene Epoch, and suggesting an origin for this clade more recent than dates cited in the recent
literature. The sister group to this clade is also robustly supported, and consists of Catoprion, Pygopristis, and Pristobry-
con striolatus. If the term piranha is to refer to a monophyletic clade, it should be restricted to Serrasalmus, Pygocentrus,
and Pristobrycon (in part), or expanded to include these taxa plus Pygopristis, Catoprion, and Pristobrycon striolatus.
FREEMAN ET AL.
2 · Zootaxa 1484 © 2007 Magnolia Press
Key words: Serrasalmidae, piranhas, Characidae, Characiformes, Serrasalmus, Pristobrycon, phylogeny, polytomy, bio-
geography
Introduction
Piranhas are neotropical freshwater fishes belonging to the order Characiformes and infamous for their formi-
dable dentition and predatory habits. According to Machado-Allison (1983, 1985) they comprise a monophyl-
etic group within the Serrasalminae, a subfamily of the Characidae that also includes the pacus. Some recent
researchers still maintain the subfamily (Jégu 2003), while others recognize piranhas and their relatives as a
distinct family, the Serrasalmidae (Géry 1972, 1977; Reis 1998; Calcagnotto et al. 2005). In this manuscript,
we treat piranhas and pacus as a separate family, Serrasalmidae, realizing the exact relationship of this group
to other characiformes is yet to be determined.
Serrasalmids are endemic to South America, with species distributed in all major and some minor Atlantic
river systems from about 10° N latitude south to about 35° S latitude. Many members of the family are in
demand as aquarium ornamentals, and several pacus (e.g., Piaractus and Colossoma) are economically impor-
tant to commercial fisheries and aquaculture (Araujo-Lima & Goulding 1997; Castagnoli 2000). Piranhas are
generally less valued, although they are commonly consumed by subsistence fishers and frequently sold for
food in local markets. A few piranha species occasionally appear in the aquarium trade, and, in recent
decades, dried specimens have been marketed as tourist souvenirs (L. G. Nico, pers. obs.). Piranhas occasion-
ally bite and sometimes injure bathers and swimmers, but truly serious attacks are rare and the threat to
humans has been largely exaggerated (Braga 1975; Goulding 1980; Nico & Taphorn 1986; Sazima &
Andrade-Guimaraes 1987; Haddad & Sazima 2003). However, piranhas are a considerable nuisance to com-
mercial and sport fishers because they steal bait, mutilate catch, damage nets and other gear, and may bite
when handled (Agostinho et al. 1997; L. G. Nico, pers. obs.).
Ecologically, piranhas are important components of their native environments. Although largely restricted
to lowland drainages, these fishes are widespread and inhabit diverse habitats within both lotic and lentic
environments. Some piranha species are abundant locally and multiple species often occur together. For
example, as many as seven different piranha species have been recorded from a single reach of Caño Maporal,
a small savannah stream in Venezuela (Nico & Taphorn 1988; L. G. Nico, pers. obs.). As both predators and
scavengers, piranhas influence the local distribution and composition of fish assemblages (Nico & Taphorn
1988; Winemiller 1989; Sazima & Machado 1990). In spite of the group’s reputation as carnivores, certain
piranha species consume large quantities of seeds, but unlike Colossoma and Piaractus, herbivorous piranhas
thoroughly masticate and destroy all seeds eaten and consequently do not function as dispersers (Goulding
1980; Nico 1991).
The taxonomy and systematics of piranhas and their relatives are complicated and much remains unset-
tled. Consequently, both species identification and phylogenetic placement of many taxa are problematic.
Both nomenclatural uncertainty and difficulties involving positive identification have long frustrated scien-
tists and hampered systematic as well as non-taxonomic research (e.g., ecology and physiology) on piranhas.
For instance, investigators have misidentified study animals as a single species when two or more species
were actually represented, or have erroneously recognized several species in a sample that actually consisted
of a single species. As a result, names used for piranhas in the literature, field survey reports, museum collec-
tions, and other sources (e.g., GenBank) are often suspect. Sensitive to this issue, a number of ecologists have
been reluctant to assign names, using codes rather than providing species or even generic names of question-
able validity (e.g., Goulding et al. 1988).
Some reasons for the continued confusion include: 1) shortage of comparative material; 2) lack of distinct
or reliable external features useful in distinguishing most genera as well as many species; 3) wide intra-spe-
cific variation in morphology (generally related to ontogenetic changes); 4) wide intra-specific variation in
Zootaxa 1484 © 2007 Magnolia Press · 3
PIRANHA MOLECULAR SYSTEMATICS
color pattern in both preserved specimens and live individuals within members of a population, and between
geographic localities (often associated with ontogeny, reproductive condition, or environmental influences);
5) marked overlap in color patterns and morphologies among different species; 6) poorly known geographic
ranges; 7) a number of poorly-defined nominal species (in some cases coupled with the absence or loss of type
material); 8) probable existence of species complexes (e.g., Serrasalmus rhombeus); 9) uncertainty about the
generic placement of certain species in Serrasalmus versus Pristobrycon; and 10) the existence of numerous
synonymies, some speculative (e.g., Fink 1993; Machado-Allison & Fink 1996; Machado-Allison 2002; Jégu
2003). Over the past few decades, several new species have been described. Although some are markedly dis-
tinct (e.g., Pristobrycon maculipinnis Fink & Machado-Allison 1992) most of these recent additions closely
resemble previously described nominal species (e.g., S. altuvei Ramirez 1965, followed by S. compressus
Jégu et al. 1991, and S. hastatus Fink & Machado-Allison 2001).
The goal of our research was to use mitochondrial DNA sequences to elucidate serrasalmid relationships
and to aid in species-level identification. To resolve systematic questions, we analyzed three mitochondrial
gene regions with varying rates of evolution: the relatively slowly evolving small (12S) and large (16S) ribo-
somal RNA (rRNA) genes, along with the more rapidly-evolving control region. We focused on the species S.
manueli, S. gouldingi, and an unidentified species (Serrasalmus sp. “A”). Because of the difficulty in identify-
ing juvenile and adult specimens that were either fresh or preserved, genetic analyses were necessary to con-
firm field identifications, to determine whether the three piranhas were distinct, and to clarify their systematic
relationships. We have included photographs to document similarities and differences in fresh and preserved
vouchered specimens. An additional objective was to investigate the issue of monophyly among certain pira-
nha genera, in particular, Serrasalmus and Pristobrycon. To test hypotheses about a persistent polytomy
within the piranhas, we expanded our sampling to include 12S, 16S, and control region sequences available
for Serrasalmidae in GenBank. Finally, because of the confusion surrounding serrasalmid systematics and tax-
onomy, we discuss the diversity of the piranhas and provide a brief overview of past research on their system-
atics.
Overview of piranha diversity and systematics
Diversity within the Serrasalmidae. Compared to many neotropical fish groups, the Serrasalmidae is well
defined, and there is wide agreement concerning which genera and species should be included (Machado-Alli-
son 1983; Jégu 2003). Serrasalmids are medium to large-sized characids (up to about 1-m long) generally
characterized by a deep, laterally compressed body with a series of mid-ventral abdominal spines or scutes,
and a long dorsal fin (>16 rays). Most species also possess an anteriorly-directed spine just before the dorsal
fin extending from a supraneural bone; exceptions include members of the genera Colossoma, Piaractus, and
Mylossoma (Jégu 2003). Jégu (2003) reported the family as comprised of 15 genera and 80 valid species,
although he noted that the status of as many as eight of these “valid” forms was uncertain (incertae sedis).
According to Jégu, in addition to the four piranha genera, the family currently includes Acnodon (3 species),
Catoprion (1), Colossoma (1), Metynnis (11), Mylesinus (3), Myleus (14 or 15), Mylossoma (3), Ossubtus (1),
Piaractus (2), Tometes (2), and Utiaritichthys (2).
Traditionally, “piranhas” or “true piranhas” are a group that includes only the four genera Serrasalmus,
Pristobrycon, Pygocentrus, and Pygopristis. These genera possess a number of morphological traits that sepa-
rate them from other serrasalmids (Machado-Allison 1985; Machado-Allison & Fink 1996; Jégu 2003). In
particular, they are most easily distinguished by their unusual dentition. All piranhas have a single row of
sharp teeth in both jaws; the teeth are tightly packed and interlocking (via small cusps) and used for rapid
puncture and shearing. Individual teeth are typically broadly triangular, pointed, and blade-like (i.e., flat in
profile). There is minor variation in the number of cusps; in most species the teeth are tricuspid with a larger
FREEMAN ET AL.
4 · Zootaxa 1484 © 2007 Magnolia Press
middle cusp that makes the individual teeth appear markedly triangular. The exception is Pygopristis which
has pentacuspid teeth and a middle cusp that is usually only slightly larger than the other cusps (Machado-
Allison 1982, 1985). A few authors apply the term “piranha” or “true piranhas” more broadly, grouping them
with what is considered to be their closest relatives, the scale-eating, monotypic genus Catoprion and the
plant-eating genus Metynnis (Machado-Allison 1982, 1985; Ortí et al. 1996; Nakayama et al. 2002). How-
ever, the shape of the teeth of Catoprion and Metynnis are markedly different from that of the four traditional
piranha genera, and, similar to other serrasalmids, their premaxillary teeth are in two rows, not one (Machado-
Allison & Fink 1995, 1996).
The number of piranha species is not known, and new species continue to be described. Fink (1988) stated
that fewer than half of the approximately 60 nominal species of piranhas were valid. In a more recent treat-
ment, Jégu (2003) recognized a total of 38 or 39 species, although the validity of some taxa remains question-
able. Based on the annotated list of Jégu, the most species rich genus is Serrasalmus (24 species, perhaps as
many as 28), followed by Pristobrycon (5), Pygocentrus (3, possibly 4), and Pygopristis (1). The natural dis-
tribution of many piranha species is poorly known. According to Jégu (2003), 25 species are distributed in the
Amazon basin, 16 in the Orinoco, 9 in rivers of the Guyanas, 3 in the Paraguay-Paraná, and only 2 in the São
Francisco. Some species have extremely broad geographic ranges, occurring in more than one of the major
basins mentioned above, whereas others appear to have much more limited distributions.
Systematic relationships. Investigators interested in the systematics of piranhas and their relatives have used
a variety of methods to evaluate and compare a broad range of characters. Early studies focused on morphol-
ogy (e.g., Eigenmann 1915; Norman 1929; Gosline 1951; Machado-Allison 1982, 1985) while a few of the
more recent investigations have dealt with karyological (Nakayama et al. 2002, and references therein) and
parasitological (Van Every & Kritsky 1992) characters. Some researchers made minor attempts to compare
piranha behavioral ecology (e.g., Nico 1991). During the past decade, investigations have included analysis of
molecular data (Ortí et al. 1996; Calcagnotto et al. 2005; Hubert et al. 2006). While most of the studies agree
the family Serrasalmidae is monophyletic, there is disagreement concerning relationships within this family.
As reviewed in detail by Machado-Allison (1985, 2002), early morphological studies involved numerous
changes in both the number and hierarchy of serrasalmid genera. For example, Eigenmann (1915) recognized
six piranha genera (Pygopristis, Gastropristis, Rooseveltiella, Pristobrycon, and Serrasalmo) whereas Nor-
man (1929) and Gosline (1951) relegated all species to two genera (Serrasalmus and Pygopristis). In contrast,
Géry (1972, 1976) grouped all piranhas into a single genus, Serrasalmus, dividing the species among five sub-
genera. The first cladistic analysis of the Serrasalminae was conducted by Machado-Allison (1982, 1983,
1985). Based on a comparative study of 65 anatomical characters, mostly osteological and myological, he
divided 13 serrasalmid genera into two major clades, with the piranha genera monophyletic (Fig. 1A). Largely
limiting himself to generic-level analyses, Machado-Allison’s phylogenetic hypothesis endured unchallenged
for many years, consequently, his tree of generic relationships has been widely cited and often reprinted (e.g.,
Machado-Allison 1983, 1985, 2002; Lundberg et al. 1986; Machado-Allison & Fink 1996). As part of a later
review focusing on the genus Pristobrycon, Machado-Allison et al. (1989) offered a slightly modified hypoth-
esis, revising their piranha clade so as to reflect possible evolution of piranhas based, in part, on the presence
versus absence of the pre-anal spine (Fig. 1B).
Morphometric techniques have also been used to address serrasalmid relationships and to quantify differ-
ences among taxa more precisely (Géry 1972; Fink 1989; Machado-Allison et al. 1989; Fink & Zelditch 1995,
1997). These methods have proven valuable for quantifying variation in shape but have been less useful for
clarifying serrasalmid relationships. For example, Fink (1993) determined that head width dimensions sepa-
rated Pygocentrus from other piranhas but was unable to detect any significant shape differences among the
three recognized species of Pygocentrus.
Zootaxa 1484 © 2007 Magnolia Press · 5
PIRANHA MOLECULAR SYSTEMATICS
FIGURE 1. Alternative hypotheses of serrasalmid relationships: (A) Machado-Allison (1983, 1985), based on morphol-
ogy, divides family into two major clades; (B) Machado-Allison et al. (1989) revised piranha clade showing the position
of Pristobrycon striolatus if absence of pre-anal spine is considered to be primitive character (arrow indicates occurrence
of this trait); (C) Ortí et al. (1996), based on mitochondrial ribosomal RNA sequence data, defines three major clades.
Upper tree includes 13 of the currently 15 recognized genera, lower tree includes 11 genera. Note: the genus Tometes was
presented in original tree of Ortí et al. (1996) as "N. gen. A" (P. Petry, pers. comm. 2005). The authors also stated that
specimens assigned by Machado-Allison (1982, 1983) to Utiaritichthys do not belong to that genus, apparently suggest-
ing the specimens are Tometes.
The first published work on serrasalmid relationships using sequence data is Ortí et al. (1996) who per-
formed a cladistic analysis using mitochondrial rRNA genes to test Machado-Allison’s 1982 phylogeny. The
results from the molecular data differed from that based on morphological data, suggesting the family was
divided into three groups rather than the two proposed by Machado-Allison (Fig. 1C). Moreover, their analy-
sis indicated that Serrasalmus, Pristobrycon, and Myleus may be para- or polyphyletic. Ortí et al. (1996) also
FREEMAN ET AL.
6 · Zootaxa 1484 © 2007 Magnolia Press
reported that mitochondrial data showed Pristobrycon striolatus to be very different from the other Pristobry-
con analyzed in their study. An additional unpublished molecular study by Ortí et al. (2000) based either on
12S and 16S, or control region sequence data, included trees that contain several species that are not mono-
phyletic. They also reported that the genera Serrasalmus and Pristobrycon were paraphyletic and fell within a
clade that included Pygocentrus, with a sister clade of Catoprion, Pygopristis and Pristobrycon. Recently,
Calcagnatto et al. (2005) investigated characiform phylogeny by parsimony analysis of four nuclear and two
mitochondrial genes. Focusing on generic and higher-level relationships, their results were inconsistent with
those of both Machado-Allison (1983) and Ortí et al. (1996), especially in terms of placement of Colossoma,
Mylossoma, and Piaractus. Partly because their study included only eight of the 15 serrasalmid genera, Cal-
cagnatto et al. (2005) noted that their scheme of relationships within the family should be interpreted with
caution. Hubert et al. (2006) examined nuclear DNA of sympatric Serrasalmus from the Bolivian Amazon,
finding evidence of reproductive isolation in some species.
A few recent authors have reported on chromosome numbers and position of nucleolar organizing regions
(NORs) of various serrasalmid fishes (e.g., Gaviria et al. 2005; Nirchio et al. 2005). Ortí et al. (1996) mapped
the available information (Oliveira et al. 1988; Porto et al. 1991, 1992; and Cestari & Galetti 1992) onto their
phylogenetic tree and concluded there is an obvious trend for chromosome numbers to increase during the
evolution of serrasalmids, with the 2n number ranging anywhere from 54 to 62 among genera and species
sampled. Most piranhas and closely related genera were characterized by having 2n = 60, except for Metynnis
which had 2n = 62, and according to Nakayama et al. (2001), Pristobrycon striolatus, Catoprion, and Pygo-
pristis, which also possessed 2n = 62. In addition, they reported the autapomorphic reduction of chromosome
number (to 2n = 58) of one Serrasalmus species. A recent study by Gaviria et al. (2005) on the karyotype and
NORs of Pygocentrus cariba, however, does provide additional support for the hypothesized close relation-
ship between Pygocentrus and Serrasalmus.
Parasitological research intended to obtain a better understanding of serrasalmid relationships is still in its
early stages. Van Every and Kritsky (1992) described 13 new species of helminth gill-parasites of the genus
Anacanthorus taken from central Amazonian piranhas. Included were 10 parasite species from 3 piranha gen-
era (Pygocentrus, Pristobrycon, and Serrasalmus). Using the parasites as indicators of host evolution, the two
researchers proposed their own phylogenetic hypothesis (Fig. 2). According to Van Every and Kritsky, the
relationships among the three genera evident in their parasite-host cladogram generally supported those
offered by Machado-Allison (1983). Similar to the mitochondrial DNA data of Ortí et al. (1996), their para-
site-host hypothesis suggested that Pristobrycon and Serrasalmus were paraphyletic. Subsequently, Ortí et al.
(1996), reanalyzed Van Every and Kritsky’s (1992) data set and were able to match the parasite information
with their own molecular-based cladogram. They concluded that more information was needed to resolve
relationships confidently among the parasites studied. In a subsequent paper, Nakayama et al. (2001) noted
that differences in parasite species supported recognition of a cryptic species of piranha within Serrasalmus
rhombeus. Unfortunately, use of parasites to generate phylogenies is problematic (Lovejoy 1997). Moreover,
this type of research requires exhaustive sampling and considerable amounts of data to avoid the false conclu-
sion that certain parasites are absent from particular host taxa.
Researchers have long been interested in the diets of serrasalmid fishes and several investigators have
attempted to infer or refine serrasalmid systematic relationships by comparing trophic variation (e.g., Géry
1977; Fink 1989; Nico 1991). As a group, the diets of serrasalmid fishes are extremely broad and include
seeds, fruits, leaves, various invertebrate and vertebrate prey, as well as fish flesh, scales, and fins. To empha-
size the polarity of diets, authors commonly highlight the fruit- and leaf-eating pacus such as Piaractus brac-
hypomus and the highly carnivorous piranhas such as Pygocentrus nattereri. Most non-piranhas in the family
are primarily herbivorous. In contrast, it was long believed that piranhas were strict carnivores. Consequently,
Géry (1977) argued that feeding specializations could be used to divide “Serrasalmidae” into three supposedly
natural groups or subfamilies: 1) Myleinae (pacus and their allies) composed of vegetarians, 2) Serrasalminae
(piranhas) consisting of carnivores, and 3) Catoprioninae composed of the scale eating Catoprion mento.
Zootaxa 1484 © 2007 Magnolia Press · 7
PIRANHA MOLECULAR SYSTEMATICS
FIGURE 2. Van Every and Kritsky (1992) hypothesis of the evolutionary relationships of 10 piranha species from the
central Amazon based on their helminth (Anacanthorus) parasite fauna.
Subsequent diet studies of more piranha species exposed flaws in Géry’s somewhat simplistic classifica-
tion scheme, revealing that the diets of most genera and many serrasalmid species are much more complex
than previously believed, and not easily divided into trophic specialties (Goulding 1980; Nico & Taphorn
1988; Nico 1991). For instance, many species are known to adopt multiple diets depending on age and
resource availability. Nico (1991) attempted to integrate the trophic ecology and piranha phylogenies by using
diet and intestine length data for nine serrasalmid genera and superimposing this information onto Machado-
Allison’s (1985) suggested phylogeny (Fig. 3). Results were mixed, largely because of the variation in trophic
characters found for some upper Orinoco species. Based on the variation and complexity of feeding behaviors
among serrasalmids, diets alone appear to be of limited use in predicting the phylogenetic relationships among
serrasalmid fishes.
The fossil record available for serrasalmid fishes, particularly for piranhas, is relatively sparse. Most
known serrasalmid fossils are from the Miocene, although a few unidentified forms are considered Paleocene
and two reportedly date to as early as the Late Cretaceous (Arratia & Cione 1996). Miocene remains include
those of an unidentified Serrasalmus from Peru (Arratia & Cione 1996), a Colossoma-like fish unearthed in
Colombia (Lundberg et al. 1986), and an unidentified non-piranha serrasalmid found in Chile (Rubilar 1994).
Reis (1998:359) applied the known fossil evidence to the cladogram of Machado-Allison (1983) and con-
cluded that all serrasalmid genera had originated by the middle Miocene, with the possible exception of three
of the four piranha genera (Pygocentrus, Pristobrycon, and Serrasalmus).
FREEMAN ET AL.
8 · Zootaxa 1484 © 2007 Magnolia Press
FIGURE 3. Diet and intestinal length data mapped onto Machado-Allison’s (1985) proposed phylogeny (modified from
Nico 1991). Diet data based on 18 serrasalmid species from the Orinoco River basin (Venezuela); number in parentheses
following generic name represents numbers of species in each genus included in study; Jv = juvenile trait; ad = adult
trait; long intestine defined as mean intestine length >1.2 X standard length.
Piranhas of southern Venezuela. The piranha fauna of Venezuela is relatively well known, with 16 spe-
cies appearing in the identification guide by Machado-Allison and Fink (1996, also see Machado-Allison
2002). Nevertheless, many parts of the country are poorly sampled and a few undescribed forms are not
included. Little or no information (e.g., diagnostic characters) on juveniles is provided for most piranhas.
Consequently, even in Venezuela, there are still taxonomic and identification issues that need to be resolved.
Least is known about the piranhas occurring in southern Venezuela, a biogeographically important region that
includes the upper Orinoco, upper Negro, and Casiquiare river systems.
The two most diverse piranha genera, Serrasalmus and Pristobrycon, are also the most problematic taxo-
nomically and diagnostically. No single morphological feature has been found that completely diagnoses
either, although combinations of characters have been presented to distinguish members of one piranha genus
from the other and from related species (Machado-Allison 1985; Machado-Allison & Fink 1996). Some of
these difficulties are exemplified by two Serrasalmus species, Serrasalmus manueli (Fernández-Yepez &
Ramírez 1967) and Serrasalmus gouldingi (Fink & Machado-Allison 1992), found in southern Venezuela.
These two large piranhas are found in Venezuela and Brazil and commonly co-occur in the Casiquiare River
drainage (Fig. 4). Depending on age and environment, individual S. gouldingi and S. manueli specimens can
be difficult to distinguish (Fig. 5). Most meristic and morphological characters provided to separate S. gould-
ingi from S. manueli show extensive overlap (Fink & Machado-Allison 1992; Machado-Allison & Fink
1996). Although intermediate-sized individuals of the two species are generally distinct in appearance, juve-
niles and large adults of S. gouldingi and S. manueli resemble each other closely, especially darkly-pigmented
Zootaxa 1484 © 2007 Magnolia Press · 9
PIRANHA MOLECULAR SYSTEMATICS
adults (over about 250 mm TL) found in tannin-stained, blackwater habitats. Identification is occasionally, but
not always, resolved when pigmentation patterns are revealed following preservation.
FIGURE 4. Map of northern South America showing collection sites of Serrasalmus manueli (triangles), S. gouldingi
(circles), and Serrasalmus sp. “A” (diamond). Symbols may represent more than one collecting locality. Solid red sym-
bols represent capture sites for material used in present genetic study (Maps A and B); numbers pertain to individual
specimens, S. manueli (1-10), S. gouldingi (11-16), and Serrasalmus sp. “A” (17) (see Table 1). Hollow symbols on Map
A are based on museum records and published information (specimen identities and capture localities were not verified
for all records). Stars represent type localities for S. manueli (Pr, Rio Parguaza) and S. gouldingi (lower Rio Negro). Prin-
cipal rivers: A, Amazon; B, Branco; C, Casiquiare; G-N, Guainia-Negro; J, Japurá; N, Negro; R, Orinoco; and S-A,
Solimões. Other rivers: Ar, Arirará; Ca, Capanaparo; Ci, Cinaruco; Cu, Cunucunuma; Cv, Cuchiverio; D, Daraá; G,
Guaypo-Sipapo; P, Pasimoni; Pr, Parguaza; Sb, San Bartolo (Guariquito system); Si, Siapa; T, Atabapo-Atacavi; and V,
Ventuari.
FREEMAN ET AL.
10 · Zootaxa 1484 © 2007 Magnolia Press
FIGURE 5. Adult and juvenile specimens of Serrasalmus gouldingi (A and B) and S. manueli (C and D) from southern
Venezuela. Adult specimens (upper frame) are 195 and 240 mm SL; juvenile specimens (lower frame) are both 65 mm
SL. Museum catalogue numbers for A-D: UF 148231, UF 120211, UF 121513, and UF 81180.
In addition to S. manueli and S. gouldingi, there is a Serrasalmus species inhabiting the upper Orinoco
that we have been unable to identify. Nico (1991) referred to this piranha as Serrasalmus cf. eigenmanni, indi-
cating its uncertain taxonomic status. We are unaware of any confirmed records of S. gouldingi in the upper
Orinoco, but have speculated that Nico’s S. cf. eigenmanni, referred to in this paper as Serrasalmus sp. “A”,
might be the whitewater form of S. gouldingi. Juveniles of species “A” also are similar in appearance to young
S. manueli. (Appendix 1, Plate 3)
Material and methods
Specimen collection and vouchers. Original material, the source of tissue samples used in this study, con-
sisted of 33 serrasalmid specimens. Of these, 31 (representing all four piranha genera and seven piranha spe-
cies) were collected in the wild from the middle and upper Orinoco and upper Negro river systems during the
years 1991–1992 and 1999 (Table 1; Fig. 4). Many of the source localities were in remote areas rarely or never
sampled previously. Juvenile and adult specimens ranging from 28 to 275 mm SL were collected by seine,
cast net, gill net, or angling. Capture localities for S. gouldingi, S. manueli, and Serrasalmus sp. “A” used in
genetic analyses are shown in Figure 4. In addition, Piaractus tissues were obtained, one from a Venezuelan
market fish and another from an introduced fish captured in a California lake. Piranha tissues (liver, muscle,
caudal fin, or entire fish) for genetic analysis were either frozen, preserved in 95% ethanol, or preserved in
SED (salt [NaCl], EDTA [ethylene-diamine-tetra-acetic acid], and DMSO [dimethyl sulfoxide]) buffer.
Museum vouchers have been reposited for 31 of the 33 specimens (Table 1). In general, specimens were fixed
in 10% formalin and later transferred to 70% ethyl alcohol. Most specimens from which tissues were removed
were photographed. Depending on the specimen, images in our possession include photographs of live fish
taken in the field shortly after capture or held in aquaria, preserved specimens, or combinations thereof (spec-
imen photographs appear in Appendix 1). Radiographs of selected Serrasalmus and Pristobrycon specimens
were created to examine osteological characters and verify the presence or absence of pre-anal spines.
Zootaxa 1484 © 2007 Magnolia Press · 11
PIRANHA MOLECULAR SYSTEMATICS
TABLE 1. Collection localities and voucher information for original specimens used as tissue sources in this study.
Refer to Appendix 1 for photographs of most specimens listed.
Specimen IdentificationaCapture Locality InformationbMuseum No. & size,
SL [Specimen Code]c
1) Serrasalmus manueli Orinoco–Cinaruco. Field # LN99–38. Venezuela, Apure: Río
Cinaruco, vicinity of Laguna Larga, 06°33’03”N, 67°24’32”W,
collected by LGN, HLJ, AA, and JA, 20 Jan 1999.
UF 120067, 252 mm
[LN99–38–C]
2) Serrasalmus manueli Orinoco–Cinaruco. Field # LN99–38 (see above) UF 120067, 185 mm
[LN99–38–F]
3) Serrasalmus manueli Orinoco–Cinaruco. Field # LN99–38 (see above) no voucher [LN 99–
38–G]
4) Serrasalmus manueli Amazon–Negro –Casiquiare. Field # HLJ99–1. Venezuela,
Amazonas: lagoon of Río Yatua, Río Pasimoni drainage,
01°29’00”N, 66°28’32”W, collected by HLJ, KOW et al., 10
Jan 1999.
UF 136342, 265 mm
[HLJ99–1–A]
5) Serrasalmus manueli Orinoco–Cinaruco. Field # LN99–38 (see above) UF 120067, 215 mm
[LN99–38–D]
6) Serrasalmus manueli Orinoco–Cinaruco. Field # LN99–38 (see above) UF 120067, 208 mm
[LN99–38–E]
7) Serrasalmus manueli Amazon–Negro–Casiquiare. Field # KOW99–7. Venezuela,
Amazonas: Río Yatua about 5 km above confluence with ríos
Pasimoni and Baria, 01°29’12”N, 66°29’25”W, collected by
KOW, FP, TT, and JC, 9 Jan 1999.
UF 121513, 240 mm
[KOW99–7–A]
8) Serrasalmus manueli Amazon–Negro–Casiquiare. Field # KOW99–7 (see above) NLU 78895, 269 mm
[KOW99–7–E]
9) Serrasalmus manueli Amazon–Negro–Casiquiare. Field # KOW99–7 (see above) UF 121513, 205 mm
[KOW99–7–C]
10) Serrasalmus manueli Amazon–Negro–Casiquiare. Field # LN99–05. Venezuela,
Amazonas: Río Casiquiare at Raudal Yacmin, 02°00’47”N,
66°48’23”W, collected by LGN, HLJ, KOW, ABD, and HLF, 7
Jan 1999.
UF 120063, 88 mm
[LN99–05–A]
11) Serrasalmus gouldingi Amazon–Negro–Casiquiare. Field # KOW99–7 (see above) UF 148231, 195 mm
[KOW99-7–D]
12) Serrasalmus gouldingi Amazon–Negro–Casiquiare. Field # LN99–08. Venezuela,
Amazonas: lower Río Pasimoni at Laguna Arapacoa,
01°50’49”N, 66°35’07”W, collected by LGN, KOW, HLJ,
ABD, HLF, FP, JC, and TT, 7 Jan 1999.
MCNG 41975, 168
mm [LN99–08–H]
13) Serrasalmus gouldingi Amazon–Negro–Casiquiare. Field # LN99–31. Venezuela,
Amazonas: Caño Cachiapo, left bank affluent of Río Casiquiare,
01°56’08”N, 66°41’48”W, collected by LGN, HLJ, KOW, et al.,
16 Jan 1999.
UF 120211, 96 mm
[LN99–31–B]
14) Serrasalmus gouldingi Amazon–Negro–Casiquiare. Field # LN99–24. Venezuela,
Amazonas: Río Yatua at Piedra Catipán, Río Pasimoni drainage.
01°30’58”N, 66°24’54”W, collected by LGN, HLJ, FP, KOW,
and HLF, 13 Jan 1999.
MCNG 48008, 162
mm [LN99–24B]
15) Serrasalmus gouldingi Amazon–Negro–Casiquiare. Field # HLJ99–1 (see above). UF 121512, 225 mm
[HLJ99-1–B]
16) Serrasalmus gouldingi Amazon–Negro–Casiquiare. Field # LN99–31 (see above) UF 120211, 65 mm
[LN99–31–A]
FREEMAN ET AL.
12 · Zootaxa 1484 © 2007 Magnolia Press
aSpecimen identification includes scientific name and codes used in Figures and codes associated with individual pre-
served vouchered specimens.
bCapture locality information includes river drainage, original field number, country, state, and water body, coordinates,
names of collectors, and date of collection. Names of collectors are as follows: AA = Albrey Arrington, ABD = Aniello
Barbarino-Duque, FP = Frank Pezold, FM = Fabian Morillo, HLF = Hernan López-Fernández, HLJ = Howard Jelks, JA
17) Serrasalmus sp. “A” Orinoco–upper Orinoco. Field #LN91-50. Venezuela, Amazo-
nas: upper Río Orinoco at Isla El Tigre, approx. 03°55’N,
67°00’W, collected by LGN and FM, 20 Feb 1991.
UF 162488, 142 mm
[LN91-50].
18) Serrasalmus medinai Orinoco–Apure. Field # LN99–39. Venezuela, Apure: Caño
Caicara, at bridge between Mantecal and Bruzual, 07°33’31”N,
69°15’31”W, collected by LGN, HLJ, AA, and JA, 21 Jan 1999.
UF 120212, 76 mm
[LN99–39–A]
19) Serrasalmus medinai Orinoco–Apure. Field # LN99–39 (see above). UF 120212, 56 mm
[LN99–39–H]
20) Serrasalmus medinai Orinoco–Apure. Field # LN92–13A. Venezuela, Apure: Caño
Caicara west of Mantecal, approx. 07°33’30”N, 69°20’W, col-
lected by KOW and ABD, 23 Jan 1992.
UF 147757, 95 mm
[LN92–13A–Smd]
21) Serrasalmus irritans Orinoco–Apure. Field # LN99–39 (see above). UF 120070, 51 mm
[LN99–39–C]
22) Serrasalmus irritans Orinoco–Apure. Field # LN99–39 (see above) UF 120070, 49 mm
[LN99–39–D]
23) Serrasalmus irritans Orinoco–Apure. Field # LN99–39 (see above) UF 120070, 55 mm
[LN99–39–E]
24) Serrasalmus irritans Orinoco–Apure. Field # LN99–39 (see above) UF 120070, 80 mm
[LN99–39–F]
25) Pygocentrus cariba Orinoco–Apure. Field # LN99–39 (see above) UF 120069, 95 mm
[LN99–39–B]
26)Pygopristis denticulatus Orinoco–Apure. Field # LN92–12. Venezuela, Apure: lagoon of
Río Cinaruco, 6° 33’06”N, 67°30’43”W, collected by LGN,
LMP, PC, and JL , 23 Jan 1992.
UF 147756, ~75 mm
[LN92–12–Pd–01]
27) Pygopristis denticulatus Orinoco–Apure. Field # LN92–12 (see above). UF 147756, 122 mm
[LN92–12–Pd–03]
28) Pristobrycon striolatus Orinoco–Apure. Field # LN99–39 (see above). UF 120068, 46 mm
[LN99–39–G]
29) Pristobrycon striolatus Amazon–Negro–Casiquiare. Field # LN99–12. Venezuela,
Amazonas: Río Pasimoni at Laguna de Candela, 01°31’25”N,
66°33’31”W, collected by LGN, KOW, HLJ, ABD, HLF, FP,
JC, and TT, 7 Jan 1999.
UF 120064, 47 mm
[LN99–12–A]
30) Pristobrycon striolatus Amazon–Negro–Casiquiare. Field # LN99–12 (see above) UF 120064, 46 mm
[LN99–12–B]
31) Pristobrycon striolatus Amazon–Negro–Casiquiare. Field # LN99–19. Venezuela,
Amazonas: Caño Cuca, left affluent of Río Pasimoni,
01°33’56”N, 66°35’18”W, collected by LGN, KOW, HLJ,
ABD, HLF, FP, JC, and TT, 7 Jan 1999.
UF 120065, 24 mm
[LN99–19–A]
32) Piaractus brachypomus Orinoco–Apure. Field # LN93–01. Venezuela, Portuguesa: Gua-
nare fish market, approx. 09° 05'N, 69°45'W, collected by LGN,
7 Dec 1992.
no voucher, >400 mm
[LN93–1–Pb]
33) Piaractus cf. brachypo-
mus USA, California (introduction): San Francisco Bay basin, Sandy
Wool Lake, Santa Clara County, approx. 37° 27'N, 121°52'W,
collected by angler, 6 Aug 2002.
CAS 217648, 245 mm
Zootaxa 1484 © 2007 Magnolia Press · 13
PIRANHA MOLECULAR SYSTEMATICS
= J. Arrington, JC = James Cotner, JL = John Lyons, KOW = Kirk Winemiller, LGN = Leo Nico, LMP = Larry Page, PC
= Pat Ceas, and TT = Thomas Turner.
cInstitutional abbreviations for voucher specimens are as follows: UF = Florida Museum of Natural History (Gaines-
ville), MCNG = Museo de Ciencias Naturales (Guanare, Venezuela), NLU = Museum of Natural History, Louisiana Uni-
versity at Monroe, CAS = California Academy of Sciences. Size is standard length (SL). Specimen code (field number
plus additional information) represents unique alpha-numeric assigned to individual fish, especially important in distin-
guishing fish that are part of museum lots containing more than one specimen.
DNA extraction and PCR amplification. DNA was isolated from approximately 50 mg of tissue using stan-
dard phenol/chloroform extraction methods (Saghai-Maroof et al. 1984). The polymerase chain reaction
(PCR) was used to amplify three portions of the mitochondrial genome: approximately 390 base pairs (bp) of
the 12S rRNA, 580 bp of the 16S rRNA, and over 1200 bp of the control region and adjacent tRNAs. The 12S
and 16S rRNA genes were sequenced for 33 serrasalmids in this study, including two from the genus Piarac-
tus, which were used as outgroups, following Machado-Allison (1982). These sequences are deposited in
GenBank under the accession numbers EF543653-EF543685 and EF543686-EF543718. In addition, complete
control region sequences were generated for 25 of the above serrasalmids within our ingroup, and are depos-
ited in GenBank under the accession numbers EF543719-EF543743. Finally, 52 serrasalmid sequences from
GenBank were added for phylogenetic analyses (Appendix 2). With the exception of assigning serrulatus to
the genus Serrasalmus rather than Pristobrycon (following Jégu 2003), we use the scientific names as they
appear in GenBank (including those designating individual specimens not identified to species, e.g., “Serras-
almus sp. 218”). We recognize the possibility that GenBank taxonomy may be imperfect, but in the absence of
voucher information, it is not possible to verify identifications and we felt it was more informative to include
these sequences in spite of possible inaccuracies.
The primers used for amplification were 12SAL and 12SBH for the 12S rRNA, and 16SARL and
16SBRH for the 16S rRNA (Kocher et al. 1989; Palumbi 1996). The 12S BH primer is slightly modified from
the original (5'- GAGAGCGACGGGCGATGTGT- 3'). Control region primers were F-TTF and F-12R
(Sivasundar et al. 2001). Amplification of the control region included a portion of tRNA Threonine, the com-
plete tRNA Proline, the entire control region, and a portion of tRNA Phenylalanine. Reactions were per-
formed in a MJ Research PTC-200 thermal cycler in 50ul volumes with 1.5 mM MgCl2, each dNTP at 200
micromolar, 10–100 nanograms of genomic DNA, and 1x Promega buffer B, and 1 U of Taq polymerase
(Promega, Madison, WI) with a hot start and annealing temperature ranging from 49°C to 56°C. PCR prod-
ucts were purified with a GeneClean III Kit (Bio 101, Carlsbad, CA) or QIAquick PCR Purification Kit
(QIAGEN Inc., Valencia, CA). PCR products were cycle-sequenced with Big Dye version 3.1 chemistry fol-
lowing the manufacturer’s protocol (PE-ABI) using the amplification primers for the ribosomal genes and
both amplification and internal sequencing primers for the control region (Fig. 6). Sequencing reactions were
analyzed on an ABI 377 automated DNA sequencer or an ABI Prism 3100 Genetic Analyzer. All samples
were sequenced on both strands.
Alignment. Sequences were aligned using CLUSTAL X (Thompson et al. 1997) with an initial gap opening
cost of 10, a gap extension cost of 2.5, and a transition weight of 0.50. Insertions deletions (indels) were
uncommon in the ribosomal gene alignments. For example, an alignment of the 55 individuals for which we
had both12S and 16S sequences, using the parameters listed above, generated an alignment of 988 positions.
This alignment contained only 17 indel regions, only 3 of which had a length greater than 1. Indels were more
common in the control region sequences, with one region in particular exhibiting great length variation among
individuals that affected the overall alignment. When we analyzed this region with the program Tandem
Repeats Finder 3.21, designed to find variable number tandem repeats (VNTRs, Benson 1999) it became
apparent that there were between 0 and 30 copies of a complex tandem repeat in this region (Table 2). In sub-
sequent alignments we deleted all but one copy of the repeat per individual, thereby dramatically improving
FREEMAN ET AL.
14 · Zootaxa 1484 © 2007 Magnolia Press
the overall alignment, although length variation remained in the region immediately surrounding the repeats.
We tested the sensitivity of results by repeating all phylogenetic analyses both with and without this smaller
variable region. In addition we carried out profile alignments of the control region sequences, adding taxa
based on clades robustly supported in the 12S and 16S trees. No major differences were present in the various
analyses. The alignment without both the tandem repeats and the associated adjacent variable length region
was used for final control region analyses.
FIGURE 6. Primers used for amplifying and sequencing the control region and adjacent tRNAs. Internal primers 5’-3’:
662F – ACCATGCCAAGGCGTTCTTT, 662R – AAAGAACGCCTTGGCATGGT, 724F – ACATTTGGTCACTTTCG-
GAGA, 462R – CGGTTGGTGGTCTCTTACTACA, F-TTF2 – CGCCACCAGAAAAGAGAGAT, and F-12R2 -
GCCCGTGGAACTTTCTAGG
Phylogenetic analyses. We carried out phylogenetic analyses of the combined 12S and 16S sequences, the
control region sequences, and all three regions combined. Our phylogenetic analyses included the taxa
sequenced in this study and serrasalmid sequences available in GenBank. For the multi-gene analyses we used
all GenBank entries where we were able to determine that the 12S, 16S, and control region sequences were
obtained from the same individual. The combined 12S and 16S data set comprised 55 individuals, the control
region 63 individuals, and the combined 12S, 16S, and control region 35 individuals. Maximum parsimony
analyses were carried out using PAUP* 4.0b10 (Swofford 2002) with equal weighting, uninformative charac-
ters excluded, and gaps treated as missing data. Heuristic searches with 100 random sequence additions
(RSAs) and TBR branch swapping were performed followed by bootstrap analyses of 300 replicates using the
heuristic option with 10 RSAs. Hierarchical likelihood ratio tests implemented in Modeltest 3.06 (Posada &
Crandall 1998) were used to determine the most appropriate model of sequence evolution for each sequenced
region. Results indicated that the 12S and 16S sequences and the tRNA sequences could be combined in one
partition under the TrN+I+G model, while the best fit for the control region sequences was HKY+G. Bayesian
analyses were performed using MRBAYES 3.0b4 (Ronquist & Huelsenbeck 2003) employing 4 Markov
chains for 1 to 2 million generations and the partition-specific models described. Trees were sampled every
100th generation, and the first 10% of sampled trees were discarded as burn-in. Each run was carried out at
least 4 times with different random starting trees to ensure the chains had converged to the posterior probabil-
ity distribution. Graphs of generation versus log probability were examined visually for a stationary distribu-
tion. Piaractus was the outgroup for the separate ribosomal gene and control region analyses, while Metynnis
was the outgroup for the combined analysis.
Zero-length branches. We used a likelihood ratio test (Goldman & Whelan 2000; Slowinski 2001) to test the
null hypothesis that the branch lengths on the tree with the highest posterior probability were significantly dif-
ferent from zero. We used a power analysis (Walsh et al. 1999; Braun & Kimball 2001; Walsh & Friesen
2001) to test whether enough data had been collected to potentially resolve persistent polytomies.
Zootaxa 1484 © 2007 Magnolia Press · 15
PIRANHA MOLECULAR SYSTEMATICS
TABLE 2. Variable Number Tandem Repeat (VNTR) from the control region. Taxon name is followed by the number of times repeat was present. Position numbers (1-37)
and consensus sequence are presented in the first two rows."Sequences are the strict consensus of intra-individual variation in repeats. Positions that match the consensus are
indicated by “ . ”, while gaps are represented by “ – ” and asterisks (*) denote indels present among repeats within an individual. VNTR variation within individuals is
indicated by IUPAC (International Union of Pure and Applied Chemistry) degeneracy codes. Number preceding selected taxa corresponds to material listed in Table 1 and
Appendix 1.
Taxon #VNTRs 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
CONSENSUS T A T A G T A C A T A A T - G G T T - - T A G T - A C A T A T T A T G T A
*26)Pygopristis denticulatus 6.4 . . . M K . . . . . . . . - . S . Y - - . . . . - . . . . . W . . . . Y .
Pygopristis denticulatus gb 5.8 . . . . . . . . . . . . . - . S . Y - - . . . . - . . . . . . . . . . . .
Catoprion mento gb 18.7 . . . . . . . . . . R . . - . . . . - - . . . . - . . . . - - - . . . . .
28)Pristobrycon striolatus 5.9 . . . . . . . . . . . . . - . . . . - - . . . . - . . . . . . . . . . . .
29)Pristobrycon striolatus 6.9 . . . . . . . . . . . . . - . . . . - - . . . . - . . . . . . . . . . . .
30)Pristobrycon striolatus 8.9 . . . . . . . . . . . . . - . . . . - - . . . . - . . . . . . . . . . . .
31)Pristobrycon striolatus 5.8 . . . . . . . . . . . . . - . . . A - - . . . . - . . . . . . . . . . . .
Pristobrycon striolatus 225 gb 5.9 . . . . . . . . . . . . . - . . . . - - . . . . - . . . . . . . . . . . .
*Metynnis sp p.31 gb 3.5 . . . . . . . . . . . . . - . . . . C - . . R . - . . . . W A . . . . . .
Metynnis sp p.32 gb 4.2 . . . . . . . . . . R . . - . A . C T - A . A . - . . . . T A . . . . . .
*Metynnis hypsauchen gb 2.9 . . . . . . . . . . . . . - . . . C C C . . A . - . . . . . A . . . . . .
*Acnodon normani gb 29.6 . . . . . . . . . . . . . - . . . C C - . . A . - . . . . T A . . . . . .
*Ossubus xinguense gb 2.2 W . Y M K . . . . . . M . - . . . . T - W . A . - G . . . W A W . . S . .
*Tometes sp.246 gb 2.3 W . . M K . . . . . . . . - . . . . T - W . A . - R . . . . A . . . . . .
*Mylesinus paucisquamatus gb 2.0 W . . M K . . . . . . . . - . . . . T - W . A . - R . . . . A . . . . . .
Myleus pacu 69 gb 1.0 . . . . . . . . . . . . . - . . . . T - A . A . - G . . . . A . . . . . .
Myleus pacu 70 gb 1.0 . . . . . . . . . . . . . - . . . . T - A . A . - G . . . . A . . . . . .
*Myleus sp. p49 gb 2.5 W . Y M K . . . . . . . . - . . . . T - W . A . - . . . . W A . . . . . .
*Myleus sp. p51 gb 2.1 W . W M K . . . . . . M . A . . . Y T - W . A . A R . . . W A . . . . . .
*Myleus tiete gb 2.2 W . . M K . . . . . . R . - . . . . T - W . A . - R . . . W A . . . . . .
*Myleus asterias gb 2.0 W . . M K . . . . . . . . - . . . . T - A . A . - . . . . . A . . . . . .
*Myleus rubripinnis gb 2.0 W . . M K . . . . . . . . - . . . . T - A . A . - . . . . W A . . . . . .
*Myleus ternetzi gb 2.2 W . Y M K . . . . . . . . - . . . Y T - W . A . - R . . . M A . . . . . .
Colossoma macropomum gb 1.5 . . . . . . . . . . T . . - . . . A - - . . . . - . . . . T A . . . . Y .
Mylossoma duriventri gb 1.5 . . . . . . . . . . T . . - . . . A - - . . R . - . . . . W A . . . . Y .
Mylossoma paraguayensis gb 1.5 . . . . . . . . . . T . . - . . . A - - . . R . - . . . . W A . . . . Y .
*Piaractus mesopotamicus gb 1.7 . . . . K . . . . . . . . - . A . C C C - . . . - R . . . T A . . . R . .
Piaractus brachipomus 58 gb 1.5 . . . . . . . . . . T . . - . . . A - - . . . . - . . . . W W . . . . . .
*Piaractus brachipomus 60 gb 2.0 . . . M K . . . . . T . . - . . . A - - . . . . - . . . . W W . . . . . .
37
FREEMAN ET AL.
16 · Zootaxa 1484 © 2007 Magnolia Press
Results
Sequence characteristics. The amplified portions of the 12S and 16S genes were approximately 390 and 580
bp in length. The mean base frequencies were A= 0.304, C= 0.256, G= 0.228, and T= 0.212. Base frequencies
did not vary significantly among taxa, based on the chi-square test implemented in PAUP*. Of 988 aligned
positions, 149 were variable and 113 were parsimony informative. Uncorrected pairwise differences for this
data set were between 0 and 7 percent.
The amplified region that included the control region and tRNAs ranged in size from 1216 to 1471 bp in
the taxa we sequenced. As previously discussed, the length variation was due mainly to a single region of
variable number tandem repeats (VNTRs), which are common in vertebrate control region sequences (Ludwig
et al. 2000; Lunt et al. 1998; Ravago et al. 2002; Ray & Densmore 2002). The control region and adjacent
tRNA alignment was 1253 positions, of which 389 were variable, and 292 were parsimony informative. When
GenBank sequences were added, length varied between 1,069 and 1,887 bases with repeats included, and
from 969 to 1230 bases with all but one repeat removed. The mean base frequencies were A= 0.308, C=
0.238, G= 0.166, and T= 0.288, and base frequencies did not vary significantly among taxa based on the chi-
square test.
Phylogenetic analyses. The species Serrasalmus manueli and S. gouldingi were both monophyletic in analy-
ses based on the combined 12S, 16S, and control region sequences we generated in this study. Parsimony and
Bayesian analysis of our combined sequences recovered monophyletic S. manueli and S. gouldingi clades
with 100% bootstrap proportion (BP) and posterior probability (PP). Similarly, analysis of our 12S and 16S
sequences combined with those available from GenBank recovered a well-supported S. manueli clade and an
S. gouldingi clade with 68% bootstrap value and 100% posterior probability (Fig. 7A). Our control region
sequences combined with available GenBank sequences recovered a monophyletic S. manueli, but failed to
support the monophyly of S. gouldingi (Fig. 7B). When we combined our 12S, 16S, and control region
sequences with those serrasalmid individuals in GenBank for which all three genes were available, we once
again recovered a monophyletic S. manueli, but S. gouldingi appeared paraphyletic with respect to Serrasal-
mus sp. “A”, Pristobrycon specimen 224, and Serrasalmus spilopleura (Fig. 8).
There was no support for the monophyly of Serrasalmus in our analyses. Representatives of the genus
Serrasalmus typically formed unresolved polytomies with Pygocentrus cariba, Pygocentrus nattereri, and
Pristobrycon sp. (Fig. 7). In some cases, there was support for the paraphyly of Serrasalmus with respect to
Pygocentrus and Pristobrycon sp. 224 (Fig. 8), “Pristobrycon” serrulatus, or Pristobrycon sp. (Fig. 7B). Pris-
tobrycon was polyphyletic in all analyses that included more than one species of Pristobrycon, although, as
previously noted, Pristobrycon serrulatus is treated as Serrasalmus serrulatus by some recent authors. In the
rRNA data set there was moderate parsimony support and strong support in the Bayesian analysis for a clade
of Serrasalmus compressus, S. rhombeus, S. humeralis, Serrasalmus sp., Serrasalmus sp. 218, and Serrasal-
mus sp. 219 (Fig. 7A). In fact, the three specimens in this clade identified as Serrasalmus sp. are identical over
the 988 positions of the ribosomal RNA alignment with S. compressus, suggesting these may be conspecific.
Finally, several suprageneric clades were also supported within the piranha. A clade of all Serrasalmus,
plus Pygocentrus and part of Pristobrycon (Pristobrycon sp.) was robustly supported in all analyses. All of
the taxa in this clade also lack the control region repeat. In addition, a clade composed of Catoprion and Pygo-
pristis was well supported across analyses, with Pristobrycon striolatus as sister to this clade. The monophyly
of the four piranha genera, with Catoprion, was also found in all analyses.
Discussion
Mitochondrial data presented here support some previous hypotheses of evolutionary relationships and pro-
Zootaxa 1484 © 2007 Magnolia Press · 17
PIRANHA MOLECULAR SYSTEMATICS
vide evidence of unsuspected relationships among serrasalmids. These data also bear on questions about the
timing of divergence events within this clade. Results for specific clades are as follows:
FIGURE 7. Phylogenetic trees of serrasalmids inferred from ribosomal (A) and control region (B) data sets, both includ-
ing original material and GenBank sequences. Parsimony bootstrap percentages are shown above branch nodes, while
proportions (>50%) of trees possessing a given clade in Bayesian posterior distributions are shown below. Taxa shared
by both trees are in bold font. Specimen sequences from original material appear in shadow boxes, preceding numbers
(1-33) correspond to numbered specimens and information presented in Table 1 and elsewhere. GenBank sequences are
followed by gb. Taxa with VNTR in control region marked by an “R” and reconstruction of presence of VNTR shown in
blue. GenBank (gb) species with asterisk (*) indicate taxa of which we question the identification.
FREEMAN ET AL.
18 · Zootaxa 1484 © 2007 Magnolia Press
FIGURE 8. Phylogram of combined ribosomal and control region sequences. Analysis includes specimens appearing in
bold font in Figure 7. Proportion of trees from posterior distribution possessing a given clade below branch, parsimony
bootstrap proportions (>50%) above branch. Specimen sequences from original material appear in shadow boxes, associ-
ated number in parentheses (1–33) correspond to numbered specimens and information presented in Table 1 and else-
where. GenBank sequences are followed by gb. Arrow marks branches with lengths that were not significantly different
from zero. GenBank (gb) species with asterisk (*) indicate taxa of which we question the identification.
Serrasalmus and Pristobrycon. Our mitochondrial data do not support the reciprocal monophyly of Pris-
tobrycon and Serrasalmus as defined by Machado-Allison (1983, 1985), in agreement with Ortí et al. (1996)
and others. This conclusion depends of course, on proper generic placement, which is in doubt for some spe-
cies. For example, one problematic taxon is the piranha species serrulatus, which Ortí et al. (2000) treated as
Pristobrycon serrulatus. Jégu (2003), in contrast, listed serrulatus under the genus Serrasalmus, but without
explanation. Jégu and Dos Santos (1988) reported that the holotype for S. serrulatus (MNHN A9858, 117 mm
Zootaxa 1484 © 2007 Magnolia Press · 19
PIRANHA MOLECULAR SYSTEMATICS
SL) had ectopterygoid teeth. In contrast, Goulding (1980:166) stated that serrulatus lacked palatine (i.e.,
ectopterygoid) teeth. Absence of these teeth suggests the possibility that Goulding's study animals (130-250
mm SL), all from the Madeira River drainage of the Brazilian Amazon, were a species of Pristobrycon. How-
ever, ectopterygoid teeth are often not replaced, so these teeth may be absent in some older Serrasalmus indi-
viduals (for additional discussion see subsequent section on morphological characters). Similar to the
situation with serrulatus, the recent literature is inconsistent concerning generic placement (typically Serras-
almus versus Pristobrycon) of a number of other piranhas, for example, aureus, eigenmanni, and spilopleura.
This uncertainty directly influences testing the monophyly of some nominal genera.
Our molecular results indicate that Serrasalmus compressus, S. rhombeus, S. humeralis, and one or more
unidentified Serrasalmus species form a monophyletic subgroup (Fig. 7) within Serrasalmus. Results from
our combined rRNA and control region sequences provided weak evidence for a clade consisting of Serrasal-
mus rhombeus and Pygocentrus (P. cari ba and P. nattereri) (Fig. 8), but this is contradicted by anatomical evi-
dence (see Machado-Allison 2002).
The genus “Pristobrycon” as currently construed, is not monophyletic. The most studied member of the
genus is P. striolatus, a widespread species found in the Amazon, Orinoco, and Guiana drainages. Mounting
evidence indicates it is distinct both genetically and morphologically from other Pristobrycon species
(Machado-Allison 1985; Ortí et al. 1996, 2000; Nakayama et al. 2002, herein).
Other piranhas and related genera. Our combined genetic analysis suggests that Pygopristis is more closely
related to Catoprion than to any other piranha genus, a result also found by Ortí et al. (2000). The monophyly
of the four genera of the traditional “true piranhas” (see Introduction) is not supported by the mtDNA data. If
the terminology is intended to apply to a monophyletic assemblage of genera it should either be expanded to
include Catoprion, or reduced to only three genera (i.e., Serrasalmus, Pygocentrus, and Pristobrycon [without
P. striolatus]) by excluding Pygopristis.
Piranhas of southern Venezuela. Our study focused on three piranha species found in southern Venezuela,
Serrasalmus manueli, S. gouldingi, and an unknown species that we refer to as Serrasalmus sp. “A” (Fig. 5;
Appendix 1). Difficulty in identifying live and preserved specimens at different life stages was the impetus for
us to use molecular data to distinguish and verify identifications of these three taxa. The situation is compli-
cated by the fact that many piranhas coexist in the enormous region that includes the upper Orinoco and
Negro river systems.
Serrasalmus manueli. Our molecular data, based on 11 specimens (10 new and 1 GenBank), indicated
that S. manueli is monophyletic (Figs. 7, 8). The five S. manueli from the Cinaruco River (Orinoco Basin) and
the five S. manueli captured in the Casiquiare (Negro River drainage) were not reciprocally monophyletic.
There does, however, appear to be a geographic structuring based on limited sampling, with the majority of
the Negro River drainage S. manueli being part of a well-supported monophyletic clade, and the Orinoco sam-
ples forming a paraphyletic assemblage at the base of this clade. Because of their morphological similarity
and our preliminary distribution information (see subsequent discussion on biogeography), we originally sus-
pected the sister group of S. manueli was S. gouldingi. Our genetic results provide no support for that assump-
tion, although there is some evidence that S. medinai is more closely related to S. manueli than is S. gouldingi
(Fig. 8). At intermediate sizes, S. manueli is fairly distinct, largely because of the combination of body shape
and pigmentation pattern. Identification of juveniles is problematic because of their similarity to young of sev-
eral other piranha species. In addition, large S. manueli inhabiting blackwater systems are darkly pigmented,
obscuring critical marks for identification. In these cases, our molecular data resolved species identification.
Serrasalmus gouldingi. Our sample of six S. gouldingi specimens, from the Negro-Casiquiare region,
was monophyletic. In contrast, the single S. gouldingi GenBank specimen was more closely related to other
taxa in some analyses, for example, to “Pristobrycon” serrulatus from GenBank for the control region (Fig.
FREEMAN ET AL.
20 · Zootaxa 1484 © 2007 Magnolia Press
7) and to Serrasalmus sp. “A” for 16S and 12S rRNA (Fig. 7) and the combined ribosomal and control region
sequences (Fig. 8). These results suggest that either S. gouldingi is not monophyletic, or that the GenBank
specimen is not S. gouldingi. This specimen has been identified both as S. gouldingi, and S. sp. so this identi-
fication could be considered tentative. We collected a series of small juvenile piranhas, taken in the same
locality and habitat as adult S. gouldingi. Based on their proximity of capture and similarity in general appear-
ance, in the field we tentatively identified the juveniles as young of the adult S. gouldingi. After considering
the molecular data, we reexamined the preserved material and determined the juveniles were Pristobrycon
striolatus.
Serrasalmus sp. “A”. This unidentified species, whose relationship to other piranhas is unclear, was rep-
resented in our analysis by a single specimen taken from the mainstem of the upper Orinoco River. As men-
tioned above, Serrasalmus sp. “A” consistently groups with the GenBank “S. gouldingi” in the control region
analysis (Fig. 7) and in the combined analysis (Fig. 8). One possible explanation is that Serrasalmus sp. “A”
and GenBank “S. gouldingi” represent a single species whose identity is yet to be satisfactorily determined.
Nico (1991) used the name Serrasalmus cf. eigenmanni for specimens that we consider Serrasalmus sp. “A”,
although it is possible that his 48 specimens include other problematic taxa. Prior to our genetic analysis, we
thought that Serrasalmus sp. “A” might simply be the “whitewater form” of S. gouldingi. Our combined ribo-
somal and control region analysis results, however, indicates it is distinct from S. gouldingi (Fig. 8).
Serrasalmus species “A” is morphologically similar to a group of other poorly-defined piranhas, includ-
ing Serrasalmus eigenmanni Norman 1929, S. humeralis Valenciennes 1850, S. nalseni Fernández-Yépez
1969, S. serrulatus (Valenciennes 1850), among others, that have confused systematists and other piranha
researchers. In general, these piranhas are small to moderate in size (< 200 mm SL), have a moderately deep
body, slightly pointed snout, and the caudal fin base is heavily pigmented forming a dark crescent. The pattern
of spotting on the sides is variable with spots often irregular in shape. With age, spots may merge to form
larger spots, some vertically elongate. A humeral blotch, if present, is typically faint and slightly elongate ver-
tically. The humeral blotch and side spotting often are not apparent except in preserved specimens. The body
shape and pigmentation pattern of Serrasalmus sp. “A” varies considerably with growth. Juveniles have a
pointed snout and somewhat elongate body, but in larger individuals the snout is only slightly pointed and the
body much deeper relative to body length. Similar to S. gouldingi, large adult Serrasalmus sp. “A” may have
black pigments extending across nearly the entire tail except for a terminal hyaline band.
We recently reexamined four Serrasalmus sp. “A” specimens collected by Nico (1991), including the
source of our genetic tissue, and determined that all had pre-anal spines. We partially dissected one of these, a
museum specimen (UF 85215, Field Number LN 91-39; 140 mm SL) collected in the Mavaca River and were
unable to detect ectopterygoid teeth. As discussed previously, the absence of these teeth in adults may not be
diagnostic.
A persistent polytomy in the piranha clade. Analysis of the ribosomal RNA data set, with 149 variable and
113 parsimony informative characters, resulted in a large polytomy at the base of the Serrasalmus, Pygocen-
trus, Pristobrycon (excluding P. striolatus) clade (Fig. 7). The retention index of this data set (0.86) did not
indicate excessive character conflict as the cause of the polytomy, so we attempted to increase resolution by
the addition of complete control region sequences. The control region sequences added 560 variable charac-
ters and 385 parsimony informative characters. Despite having more than quadrupled the number of variable
or parsimony informative characters, the polytomy persisted (Figs. 7, 8). The persistence of this polytomy
with increased sampling suggested that it might represent a molecular polytomy. By molecular polytomy we
mean either simultaneous divergence of three or more genes from an ancestral gene (a hard molecular poly-
tomy) or a case in which divergence may not have been simultaneous but in which branch lengths between
divergence events are so short that too few substitutions have accumulated to resolve the order of branching (a
soft molecular polytomy) (Slowinski 2001).
Zootaxa 1484 © 2007 Magnolia Press · 21
PIRANHA MOLECULAR SYSTEMATICS
We used a likelihood ratio test on the tree with the highest Bayesian posterior probability (Fig. 8) to deter-
mine whether we could reject the hypothesis that the branches at the base of this clade were zero-length
(Goldman & Whelan 2000; Slowinski 2001). We used the corrected chi-square for one degree of freedom at
95% from Table 2 in Goldman and Whelan (2000). We were unable to reject the hypothesis that the branches
were zero-length for many terminal nodes, and more importantly for an internal node marked by an arrow in
Figure 8. Tree topologies and branch length tests therefore suggest a hard or soft mitochondrial molecular
polytomy at the base of the Serrasalmus, Pygocentrus, and Pristobrycon (excluding P. striolatus) clade.
There are several possible explanations for this polytomy. One simple explanation is loss of phylogenetic
signal due to saturation. The robust support for clades both above and below this polytomy argues against this
explanation. A second, related argument is that the substitutional dynamics of piranha mitochondrial DNA
might result in an area of poor resolution in the middle of the tree even if branch lengths in the region of the
polytomy were not significantly shorter than in other portions of the tree in terms of absolute time. For exam-
ple, more rapidly accumulating transitional changes might lend resolution in the tips of the tree, and more
slowly accumulating transversions would resolve the deeper branches of the tree, but there might be a zone in
the middle of the tree where transitional changes were beginning to saturate and lose signal, while too few
transversional changes had accumulated to provide adequate resolution. We used MacClade to infer the aver-
age number of unambiguous transitions and transversions in different parts of this clade. The transition-trans-
version ratio was 3.5 above the polytomy, 4.4 in the short branches that make up the polytomy, and 1.3 below
the polytomy, consistent with this scenario. If the molecular dynamics of mitochondrial DNA in piranha are
the cause of this polytomy, we would expect that other unlinked genes with differing substitutional dynamics,
for example nuclear genes, would robustly support a resolved phylogeny in this portion of the tree.
One method for addressing whether a polytomy is soft is to use a power analysis to determine if, based on
the substitution rate of the gene in question, there are enough data to resolve a given polytomy (Walsh et al.
1999; Braun & Kimball 2001; Walsh & Friesen 2001). Donaldson and Wilson (1999) analyzed control region
sequences in sister snook species (Centropomis) separated by the emergence of the Isthmus of Panama 3.5
million years ago to derive an annual rate of 1.8 x 10-8 substitutions. Applying a p-value of 0.05 and a Poisson
algorithm (equation 2, Walsh & Friesen 2001), we estimate a 95% chance of differentiating serrasalmid speci-
ation events between 150,000 and 194,000 years ago using only 986 bp of control region sequence. This sug-
gests that, unless speciation events have happened very recently or our substitution rate is not appropriate, our
data should be sufficient to detect it. Finally, the molecular polytomy might represent a hard or soft species
polytomy. If this were the case, we would expect other unlinked genes to show a polytomy, or poorly sup-
ported and conflicting resolutions in the area of the polytomy. Sampling nuclear genes from these taxa should
allow us to test these competing explanations.
We estimated the time of divergence of this polytomy by applying the control region divergence rate cal-
culated for snook (Centropomis), 3.6 % per million years (Donaldson & Wilson 1999), to our data. The rate
was applied to uncorrected p-values for all pairwise comparisons that passed through the polytomy node
(arrow in Fig. 8). The average divergence time estimated by this method was 1.8 million years. This corre-
sponds to the beginning of the Pleistocene, a time of increasingly seasonal and more cyclic climate. These cli-
matic cycles, which are thought to have resulted in cyclical fragmentation and coalescence of habitats, could
drive simultaneous divergence resulting in true species polytomies. An important caveat to this analysis is that
there is not widespread agreement on the rate of divergence of the control region in characiform fishes. The
calibration chosen is independent of our analysis but other lower rates (e.g., Sivasundar et al. 2001) would
result in a considerably older date for this polytomy. Lundberg (1997, 1998), based on a fossil tooth from the
La Venta fauna assigned to either Pygocentrus, Pristobrycon, or Serrasalmus, and further assuming the mono-
phyly of these genera, concluded the piranha-like serrasalmids had evolved by the late Middle Miocene,
approximately 11 million years ago. Reis (1998), based on a similar analysis, suggested that these genera may
have originated after the Middle Miocene. Our phylogenetic analysis indicates that Pygocentrus, Pristobry-
FREEMAN ET AL.
22 · Zootaxa 1484 © 2007 Magnolia Press
con, and Serrasalmus are not part of a monophyletic clade, and, in addition, the genus Pristobrycon is not
monophyletic. This undermines the phylogenetic rationale for the fossil dating, and therefore the value of this
tooth-type as a synapomorphy for the clade in question. The timing of origin of the piranha-like serrasalmids
must be considered an open question, but may be considerably more recent than the Middle Miocene.
Corroboration from variable number tandem repeats (VNTRs) and morphology. Our sequence-based
molecular analysis provides evidence for a clade formed by the genera Serrasalmus, Pygocentrus, and Pristo-
brycon (excluding P. striolatus) and there is robust support across all analyses for monophyly of this group
(Figs. 7, 8). In addition, all of the members of this clade lack the control region VNTR. In contrast, the VNTR
is present in the majority of the remaining serrasalmid genera sampled (Table 2; Fig. 7, VNTR marked with
the letter R). We randomly resolved the polytomies in this tree 100 times, and reconstructed the gain or loss of
the VNTR. In all reconstructions, the VNTR is lost at the base of Serrasalmus, Pygocentrus, and Pristobrycon
(excluding P. striolatus) clade. This adds a unique complex molecular character further supporting the mono-
phyly of the Serrasalmus, Pygocentrus, and Pristobrycon (excluding P. striolatus) clade. Members of this
clade sampled to date also share certain morphological traits, for example a pre-anal spine and ectopterygoid
teeth (see Fig. 1B).
The genera Serrasalmus and Pristobrycon include most of the currently recognized piranha species (Jégu
2003). However, morphological diagnosis of these two genera is problematic and depends heavily on combi-
nations of characters. Some of these characters exhibit considerable variation within genera and even within
particular species. This unsatisfactory situation was highlighted by Fink and Machado-Allison (1992) when
they noted that some of the features used to diagnose Serrasalmus by Machado-Allison (1985) do not apply
when a larger number of species than he had available are examined. As a result, it is difficult to find obvious
relationships between particular morphological characters and the molecular data. Within the piranha clade,
two main characters considered to be derived are the pre-anal spine and ectopterygoid teeth. These are worth
examining in detail, although determination of the degree of correspondence to our molecular phylogenies
will require further sampling of many species for both morphological and molecular data. As noted above, our
results suggest a possible correspondence between a well-supported clade in our sequence based phylogeny,
presence-absence of the control region VNTR and presence-absence of a pre-anal spine within piranhas and
their close relatives. Piranhas without the VNTR (Pygocentrus + Serrasalmus + Pristobrycon [in part]) have a
pre-anal spine (see Fig. 1B). Others (Pristobrycon [in part] + Pygopristis + Catoprion) have the VNTRs, but
do not have a pre-anal spine. The only serrasalmid taxa that we sampled that lack the VNTRs and are not part
of our Serrasalmus, Pygocentrus, and Pristobrycon (in part) clade are Mylesinus paraschomburgkii, Myleus
schomburgkii, and Myleus rhombeus (Fig. 7), although these genera and species all presumably lack a pre-
anal spine.
The pre-anal spine is a bony element at the anal fin origin. According to Fink and Machado-Allison
(1992), it is not visually apparent but the structure can usually be detected by touch as a sharp process.
Because alcohol preservation typically shrinks soft tissues, the pre-anal spine is often more exposed in pre-
served specimens, particularly in juveniles and some piranhas with highly compressed bodies (e.g., Serrasal-
mus irritans). However, the structure may still be difficult to locate in large piranha specimens and
confirmation of its presence often requires radiographs (see Fink & Machado-Allison 1992). Machado-Alli-
son (1985, 2002) considered the pre-anal spine to be a derived character. Among the “true piranhas” it is
present in all members of the genera Serrasalmus and Pygocentrus but absent in Pygopristis. Moreover, the
pre-anal spine is present in some, but not all, species of the genus Pristobrycon (Géry 1972; Machado-Allison
2002). It has been suggested that P. striolatus is the only species within the genus lacking a pre-anal spine
(Ortí et al. 1996), but this is not the case. For example, at least three of the four Pristobrycon species known to
occur in Venezuela do not have a pre-anal spine (i.e., P. striolatus, P. m ac uli pi nni s, and P. careospinus) (Fink
& Machado-Allison 1992; Machado-Allison & Fink 1996). In contrast, Pristobrycon calmoni and reportedly
Zootaxa 1484 © 2007 Magnolia Press · 23
PIRANHA MOLECULAR SYSTEMATICS
a few other members of the genus have a pre-anal spine (Jégu & Dos Santos 1988; Machado-Allison & Fink
1995). Based on the available information, the pre-anal spine is apparently absent in all of the remaining ser-
rasalmid genera, for example, Catoprion and Metynnis (Machado-Allison et al. 1989); Mylossoma, Piaractus,
and Colossoma (Ortí et al. 1996); and Myleus (based on our own examination of M. torquatus specimens).
Ectopterygoid teeth (referred to as palatine teeth in some piranha literature) are small teeth situated on the
roof of the inner mouth. The presence versus absence of these teeth is often used in identification keys
because the ectopterygoid teeth are present in all Serrasalmus and some Pristobrycon. In contrast, according
to Fink (1993), Pygopristis and Pygocentrus of all ages lack ectopterygoid teeth (but see Machado-Allison
1985, 2002). Unfortunately the situation is complicated. In many Serrasalmus and Pristobrycon species
replacement of ectopterygoid teeth is not continuous throughout ontogeny and these teeth may be absent in
large individuals (Fink 1993). Loss of ectopterygoid teeth with age has been documented for most Serrasal-
mus species studied in detail: S. altuvei, S. compressus, and S. geryi (Jégu et al. 1991); S. gouldingi and S.
manueli (Fink & Machado-Allison 1992); S. altispinis (Merckx et al. 2000); S. maculatus (Jégu & Dos Santos
2001), and S. rhombeus (Machado-Allison 2002), among others. The size or age at which individuals no
longer have ectopterygoid teeth varies considerably among species. For example, in their original description
of S. gouldingi Fink and Machado (1992) remarked that replacement of these teeth ceases early in life so most
adults lack them. However, in some species a few of the larger adults still retain at least a few of these teeth
(e.g., S. compressus, Jégu et al. 1991). Although certain Pristobrycon have ectopterygoid teeth, these teeth are
fewer and differently shaped than those in Serrasalmus (Machado-Allison 2002). In Serrasalmus their shape
is similar to that of the jaw teeth (i.e., triangular and often with cusps) whereas Pristobrycon ectopterygoid
teeth are relatively wide, square, and blunt.
There are additional complications concerning this character among other serrasalmid taxa. Fink (1993)
stated that Pygocentrus lacked ectopterygoid teeth at all ages, but Machado-Allison (2002:56) noted that early
juveniles (>10 mm SL) do possess 6 or more minuscule, unicuspid teeth on the ectopteryogoid bone. Previ-
ously, Machado-Allison (1985:33) reported the adults of some other serrasalmid genera have edenticulate
ectopterygoides and the juveniles of Pygocentrus, Pygopristis, Mylossoma, Colossoma, and Piaractus have
small conical teeth on the ectopterygoid that temporarily form during development, a situation that Machado-
Allison remarked as apparently representing a plesiomorphic character among members of the family Charac-
idae. Machado-Allison (1985) suspected the strongly tricuspid ectopterygoid teeth found in Pristobrycon and
Serrasalmus were a specialization that distinguished them from the unicuspid (primitive) condition present in
other serrasalmids. Based on this information, the presence of ectopterygoid teeth, even if only temporarily in
Pygocentrus, combined with pre-anal spines and absence of the VNTR, are indicative of all Pygocentrus and
Serrasalmus and some Pristobrycon.
Machado-Allison (1982, 1983) considered Pristobrycon to be more closely related to Serrasalmus than
any other group in the family (Fig. 1A), although Machado-Allison (1985) later concluded that an exhaustive
revision of the genus was necessary to establish precise limits of the taxa. In a subsequent review, Machado-
Allison (2002) addressed some anatomical characters helpful in delimiting Pristobrycon and Serrasalmus and
commented that the reanalysis of some characters suggests that Pristobrycon should be divided into two sub-
groups, one that includes species with a pre-anal spine and the other without this character. Although this view
is a slight modification to the cladogram of Machado-Allison (1985), the idea that existing Pristobrycon can
simply be considered two “subgroups” within the genus clashes with the alternative cladogram of Machado-
Allison et al. (1989) (compare Figs. 1A and 1B). The revised hypothesis of Machado-Allison et al. (1989)
recognized the absence of the pre-anal spine as a plesiomorphic “primitive” character, and, in particular, sug-
gested paraphyly of Pristobrycon. Based on our review of literature, those Pristobrycon species without
ectopterygoid teeth reportedly also lack pre-anal spines (Jégu & Dos Santos 2001; Machado-Allison 2002).
However, the published information is incomplete, highlighting the need to examine a wider range of juvenile
and adult specimens and confirm whether ectopterygoid teeth are absent at all sizes (and not just in adults). If
this anatomical dichotomy holds, then it lends additional support to the hypothesis that the species now gener-
FREEMAN ET AL.
24 · Zootaxa 1484 © 2007 Magnolia Press
ally considered to make up the genus Pristobrycon are actually two distinct groups that should be recognized
as separate genera that are not sister taxa.
A conceivable remedy is to simply transfer any “Pristobrycon” species with pre-anal spines and ectop-
terygoid teeth into the genus Serrasalmus. Unfortunately when Eigenmann (1915) erected the genus Pristo-
brycon, he unwittingly created problems for later systematists by designating P. calmoni as the type for the
genus, a species with a pre-anal spine and, at least in juveniles, ectopterygoid teeth (Jégu & Dos Santos 1988;
but see Machado-Allison 2002). Given the above, there appears justification in creating a new genus for stri-
olatus (with inclusion of the closely-related “Pristobrycon” forms that also lack pre-anal spine and ectoptery-
goid teeth). Older names used for P. striolatus include Serrasalmus and Pygocentrus, neither of which is
available. Differences in dentition and other anatomical characters argue against lumping striolatus into a sin-
gle genus together with one or both of the monotypic genera Pygopristis Müller & Troschel 1844, and Catop-
rion Müller & Troschel 1844. Ultimately, a study of all Pristobrycon species that combines both
morphological and genetic analyses will be required to provide final determination on their relationships and
true generic identity.
Biogeography and ecophenotypic variation. Because precise information on piranha distributions is lack-
ing, our understanding of piranha biogeography is imperfect. Jégu (1992) presented data on the distribution
patterns of certain Serrasalmid taxa and attempted to relate current distributions of selected serrasalmids with
glacial and interglacial events of the Quaternary. However, his most detailed analyses involved certain non-
piranha taxa (e.g., Acnodon and Mylesinus). Based on the fossil record and the proposed phylogenetic rela-
tionships among serrasalmid genera (Lundberg et al. 1986; Fig. 1A), Nico (1991) speculated that no less than
a proto-piranha existed before the start of the Pleistocene. If this “first piranha” entered the Quaternary
unchanged, subsequent radiation might have been in the form of adaptive responses to the dramatic changes
wrought by glacial events (Nico 1991). Our estimate of the average time of the initial radiation based on the
control region sequence (Table 3) is consistent with a Plio-Pleistocene origin of this group and may help
explain that lack of differentiation among some terminal groups.
The Orinoco and Amazon basins contain the majority of piranha species. Moreover, there is geologic and
biologic evidence that the two basins periodically have had very close associations in the past (including as
recently as the Late Pleistocene-Holocene), and it has been hypothesized that tectonic events shifted the pri-
mary outlet of the central Amazon region a number of times between the north and east (Frailey et al. 1988).
In more recent times the Orinoco and Amazon were distinct basins and, conceivably, their separate piranha
assemblages may have radiated independently to some extent. Today, the two basins are connected by the
Casiquiare, a natural waterway flowing southward from the upper Orinoco into the upper Negro.
The role of the Casiquiare in South American fish biogeography is intriguing because the channel is per-
manent and large, thereby permitting exchange of fishes between the Orinoco and Negro-Amazon. However,
the significance of this natural waterway in dispersing piranhas and other fishes is unknown. For example, it is
not known which, if any, of the piranha species currently widespread (i.e., occurring in both the Orinoco and
Amazon) originated in the Orinoco as opposed to the Amazon. Our data for S. manueli, with a paraphyletic
Orinoco population giving rise to a primarily Amazonian crown group suggest the direction was Orinoco to
Amazon for this species. In any case, any possible reshuffling of piranha distributions by the Casiquiare is
likely to have occurred quite recently because the Casiquiare connection to the Orinoco is reportedly very
recent (see Stern 1970). Hydrologically, it represents a stream capture in progress that, if gone unchecked, will
ultimately lead to the takeover of a large portion of the upper Orinoco by the Negro (Stern 1970; Sternberg
1975). Exactly when the initial connection was formed is uncertain. Some have hypothesized that native
Amerindians began the process by cutting a small and fairly short channel over the low area for their canoes to
avoid having to portage when crossing from the Orinoco to the Negro system (Raffles & Winkler-Prins 2003;
but see Sternberg 1975). In any case, over time, the force of the Orinoco current has increasingly enlarged the
uppermost end of the Casiquiare.
Zootaxa 1484 © 2007 Magnolia Press · 25
PIRANHA MOLECULAR SYSTEMATICS
Similar to most other piranha species, the precise boundaries of the native distributions of S. gouldingi, S.
manueli, and Serrasalmus sp. “A” are not fully known. Consequently, any biogeographic discussion is highly
speculative. Based on limited collecting by us and others, we know that both S. gouldingi and S. manueli are
common to the Casiquiare drainage. We initially believed their overall ranges differed significantly, with S.
manueli in the Orinoco (i.e., a northern species) and S. gouldingi in the Negro (i.e., a southern species).
Although distribution information is still incomplete, recent literature and museum records (some uncon-
firmed) indicate S. manueli has a broader distribution than previously thought, and includes various major
tributaries in the middle and upper Orinoco and the Casiquiare system, as well as a large portion of the Negro
mainstem (Fig. 3). There are museum records suggesting its occurrence as far north as the San Bartolo and
Aguaro rivers in the Venezuelan Llanos, but two specimens from the region that we examined were found to
be incorrectly identified and clearly not S. manueli. The Aguaro “S. manueli” record is listed by Machado-
Allison and Fink (1996:147) although the site is not included in their distribution map for the species. Consid-
ering its southern distribution, in addition to the Negro, there is evidence that S. manueli occurs in other parts
of the Amazon. During a 1986 visit to the Museu de Zoologia of the Universidade de Sao Paulo, one of us
(LGN) photographed preserved piranhas (many labeled simply as Serrasalmus sp.). We recently re-examined
the photographs and realized three adult specimens (MZUSP 15771, 20290, and 25587) from the Tapajos and
Trombetas river drainages (Amazon Basin) likely represent S. manueli or a closely-related form.
Serrasalmus gouldingi ranges widely in the Negro River, from above its confluence with the Casiquiare
downstream to at least as far as the Archipelago das Anavilhanas in the lower Negro (Fig. 3) and probably to
its mouth. The species is widely distributed in the mainstem Casiquiare and certain tributaries (e.g., Pasimoni
or Pacimoni River). In the Amazon Basin outside the Negro, the only record known to us is that of a single S.
gouldingi taken from Lago Amana in the lower Japurá River. According to W.G.R. Crampton (pers. comm.),
the Amana is a blackwater lake with typical blackwater fish fauna. In their original description of S. gouldingi,
Fink and Machado-Allison (1992) did not include any records from the Orinoco basin. Similarly, a follow-up
publication on Venezuelan piranhas indicated S. gouldingi is absent from the Orinoco (Machado-Allison &
Fink 1996). However, recent publications list the species as occurring in both the Amazon and Orinoco basins
(Jégu 2003; Lasso et al. 2004). The authors do not provide details and the information may be based on incor-
rect identifications. We are unaware of any confirmed records of S. gouldingi in the Orinoco, although its
presence in the Casiquiare suggests the species could freely move north into the basin.
Serrasalmus sp. “A” is relatively widespread in the upper Orinoco. Nico (1991) reported that it (under the
name Serrasalmus cf. eigenmanni) was the third most common piranha in his upper-Orinoco samples. Based
largely on the collections of LGN, it appears Serrasalmus sp. “A” inhabits primarily clearwater systems or
whitewater systems with relatively low sediment loads, including the Orinoco tributaries Mavaca, Ocamo,
Padamo-Matacuni, and Ventuari rivers. Its presence in the Casiquiare drainage is uncertain, but would seem
likely given the species’ occurrence in nearby Orinoco sites. S. gouldingi, which is similar in appearance,
seems to be restricted, or nearly so, to blackwater habitats.
Ecophenotypic variation in piranha colors. A wide range of neotropical fishes occurring in black- or tan-
nin-stained waters tend to be very darkly colored whereas in white- or muddy waters individuals are much
lighter (Araujo-Lima & Goulding 1997; L. G. Nico, pers. obs.). Such differences are often much greater than
the subtle color differences used by some ichthyologists to differentiate purported new species. Consequently,
we suspect the influence of water type on phenotype (i.e., intensity and pattern of pigmentation) has contrib-
uted, on occasion, to erroneous new species descriptions. Tropical South American rivers are generally classi-
fied into one of three main types, according to their color and clarity: clear, white, or black. The scheme is by
no means perfect because individual rivers may change seasonally or appear to be a mix of more than one
type. Serrasalmus manueli and S. gouldingi are most commonly reported from blackwater systems, although
there are exceptions. Fernández-Yépez and Ramirez (1967) based their description of S. manueli solely on
specimens captured in the Parguaza River (Orinoco Basin, Venezuela; Fig. 4), reporting that the species
FREEMAN ET AL.
26 · Zootaxa 1484 © 2007 Magnolia Press
occurred in clear water. Serrasalmus manueli also is relatively abundant in the nearby Cinaruco River, an
Orinoco tributary whose waters, depending on season, may appear as a mix of black and white water types, or
sometimes clear (but tinted green from algae) (LGN, pers. obs.). The mainstem Orinoco in its upper reaches
has little or no tannin and a relatively low sediment load, possibly best described as clear water.
Conclusions
Piranha systematics are undergoing major revision and much remains to be done. For example, a functional
key to identify many juvenile and adult piranhas is lacking. In spite of the hurdles remaining, we are confident
that continued and more detailed genetic analyses combined with further scrutiny of morphological characters
will ultimately produce a clear picture of piranha and serrasalmid phylogenetic relationships. Progress will
require inclusion of other Serrasalmus and all or most of the remaining Pristobrycon species. Of particular
importance relative to the piranha clade is the need to determine if genetic analysis consistently separates all
Pristobrycon species with pre-anal spines from those without and if, as has so far been shown with P. striol a-
tus, those without a pre-anal spine are separate from the clade consisting of piranhas with pre-anal spines.
Given the continued confusion and complexity, the work remaining is still substantial. Future studies of ser-
rasalmid phylogeny should include: 1) combined genetic data, including additional unlinked loci and anatom-
ical analysis of new material and reanalysis of older specimens, and 2) careful documentation of specimens
examined, including vouchering of all material and establishment of photographic archives of specimens stud-
ied.
Acknowledgments
This research was carried out as an undergraduate senior project by BF, supported by the NSF Undergraduate
Mentoring in Environmental Biology program. Activities in South America, including field collecting and
museum visits, were supported in part by grants to LGN from the Andrew W. Mellon Foundation (Amazon
Research and Training Program, University of Florida) and various National Geographic Society grants. We
are especially grateful to Kirk O. Winemiller for inviting LGN and HLJ to participate in a 1999 expedition to
the Casiquiare. For laboratory, field, and logistic help in South America, LGN thanks the administrators,
researchers, technicians, and field workers associated with the Universidad Nacional de los Llanos Occiden-
tales “Ezequiel Zamora” and Museo de Zoología (UNELLEZ-MCNG) and Corporación Venezolana Guay-
ana-Técnica Minera (CVG-TECMIN). We are indebted to Anthony B. Falsetti and Nicole DeMers for help
with x-rays. Various museum personnel and university researchers assisted by providing specimens loans,
photographs, or other information. For their help we thank: Carlos Lasso and Oscar M. Lasso-Alcalá (Museo
de Historia Natural La Salle), Robert H. Robins (Florida Museum of Natural History), William F. Fink (Uni-
versity of Michigan), Donald C. Taphorn (UNELLEZ-MCNG), Mike Retzer (Illinois Natural History Sur-
vey), Frank Pezold (NLU), and Flávio C. T. Lima (MZUSP). We are grateful to William Smith-Vaniz and
Timothy Rawlings for critically reviewing various drafts of the manuscript. Lisa Jelks kindly edited the final
draft. Lastly, we thank the Zootaxa editors for their time and assistance.
References
Agostinho, C.S., Agostinho, A.A., Marques, E.E. & Bini, L.M. (1997) Abiotic factors influencing piranha attacks on net-
ted fish in the upper Parana River, Brazil. North American Journal of Fisheries Management, 17, 712–718.
Araujo-Lima, C. & Goulding, M. (1997) So Fruitful a Fish: Ecology, Conservation, and Aquaculture of the Amazon’s
Tambaqui. Columbia University Press, New York, 191 pp.
Arratia, G. & Cione, L.A. (1996) The fossil record of fossil fishes of Southern South America. In: Arratia, G. (Ed.), Con-
Zootaxa 1484 © 2007 Magnolia Press · 27
PIRANHA MOLECULAR SYSTEMATICS
tributions of Southern South America to Vertebrate Paleontology. Special volume of the Münchner Geowissen-
schaftliche Abhandlungen 30(A), pp. 9–72.
Benson, G. (1999) Tandem repeats finder: a program to analyze DNA sequences. Nucleic Acids Research, 27, 573–580.
Braga, R.A. (1975) Ecologia e Etologia de Piranhas no Nordeste do Brasil (Pisces–Serrasalmus Lacépède, 1803).
Banco do Nordeste do Brasil, Fortaleza, Ceará, Brazil, 268 pp.
Braun, E.L. & Kimball, R.T. (2001) Polytomies, the power of phylogenetic inference, and the stochastic nature of molec-
ular evolution: a comment on Walsh et al. (1999). Evolution, 55, 1261–1263.
Calcagnotto, D., Schaefer, S.A. & DeSalle, R. (2005) Relationships among characiform fishes inferred from analysis of
nuclear and mitochondrial gene sequences. Molecular Phylogenetics and Evolution, 36, 135–153.
Castagnoli, N. (2000) Brazil fish culture. In: Stickney, R.R. (Ed.), Encyclopedia of Aquaculture. John Wiley and Sons,
New York, pp. 123–128.
Cestari, M.M. & Galetti Jr., P.M. (1992) Chromosome evolution in the genus Serrasalmus and cytotaxonomic consider-
ations about Serrasalminae (Characidae, Pisces). Revista Brasileira de Genética, 15, 555–567.
Donaldson, K.A. & Wilson, R. R. (1999) Amphi-panamic geminates of snook (Percoidei: Centropomidae) provide a cal-
ibration of the divergence rate in the mitochondrial DNA control region of fishes. Molecular Phylogenetics and Evo-
lution, 13, 208–213.
Eigenmann, C.H. (1915) The Serrasalminae and Mylinae. Annals of the Carnegie Museum, 9(3–4), 226–272.
Fernández-Yepez, A. & Ramírez, M.V. (1967) Los caribes (serrasalmidos) de Venezuela y las pesqueras. Trabajos
anexos a la Comisión Contribuciones al Tema, Ier Foro Internacional sobre Planificación y Desarrollo Pesquero,
Caracas, Venezuela, 25 pp., figures. (mimeo).
Fink, W.L. (1988) Evolutionary biology of piranhas. In: Dalrymple, G.H., Loftus, W.F. & Bernardino, F.S. (Eds.), Wild-
life in the Everglades and Latin American Wetlands. Abstracts on the Proceedings of the First Everglades National
Park Symposium, held February 25–March 1, 1985. Florida International University, Miami, Florida, pp. 13–14.
Fink, W.L. (1989) Ontogeny and phylogeny of shape change and diet in the South American fishes called piranhas. In:
David, B., Dommergues, J.-L., Chaline, J. & Laurin, B. (Eds.), Ontogenese et Évolution. Geobios, mémoire special,
No. 12, 167–172.
Fink, W.L. (1993) Revision of the piranha genus Pygocentrus (Teleostei, Characiformes). Copeia, 1993, 665–687.
Fink, W.L. & Machado-Allison, A. (1992) Three new species of piranhas from Brazil and Venezuela (Teleostei: Charac-
iformes). Ichthyological Exploration Freshwaters, 3, 55–71.
Fink, W.L. & Zelditch, M.L. (1995) Phylogenetic analysis of ontogenetic shape transformations: a reassessment of the
piranha genus Pygocentrus (Teleostei). Systematic Biology, 44, 343–360.
Fink, W.L. & Zelditch, M.L. (1997) Shape analysis and taxonomic status of the Pygocentrus piranhas (Ostariophysi,
Characiformes) from the Paraguay and Paraná river basins of South America. Copeia, 1997, 179–182.
Frailey, C.D., Lavina, E.L., Rancy, A. & de Souza–Filho, J.P. (1988) A proposed Pleistocene/Holocene lake in the Ama-
zon basin and its significance to Amazonian geology and biogeography. Acta Amazonica, 18, 119–143.
Gaviria, J.I., Nirchio, M., Granado, A. & Estrada, A. (2005). Karyotype and nucleolar organizer regions of Pygocentrus
cariba (Serrasalminae) from Caricara del Orinoco, Venezuela. Interciencia, 30, 44–47.
Géry, J. (1972) Poissons characoides des Guyanes II. Famille des Serrasalmidae. Zoologische Verhandelingen Leiden,
122, 134–248.
Géry, J. (1976) Les genres de Serrasalminae (Pisces–Characoidei). Bulletin Zoologisch Museum Universiteit van Amster-
dam, 5, 47–54.
Géry, J. (1977) Characoids of the World. Tropical Fish Hobbyist Publications, Neptune City, New Jersey, 672 pp.
Goldman, N. & Whelan, S. (2000) Statistical tests of gamma-distributed rate heterogeneity in models of sequence evolu-
tion in phylogenetics. Molecular Biology and Evolution, 17, 975–978.
Gosline, W.A. (1951) Notes on the characid fishes of the subfamily Serrasalminae. Proceedings of the California Acad-
emy of Sciences, 27, 17–64.
Goulding, M. (1980) The Fishes and the Forest: Explorations in Amazonian Natural History. University of California
Press, Berkeley, 280 pp.
Goulding, M., Carvalho, M.L. & Ferreira, E.G. (1988) Rio Negro, Rich Life in Poor Water. SPB Academic Publishing,
The Hague, The Netherlands, 200 pp.
Haddad, V., Jr. & Sazima, I. (2003) Piranha attacks on humans in southeast Brazil: epidemiology, natural history, and
clinical treatment, with description of a bite outbreak. Wilderness and Environmental Medicine, 14, 249–254.
Hubert, N., Duponchelle, F., Nuñez, J., Rivera, R. & Renno, J.-F. (2006) Evidence of reproductive isolation among
closely related sympatric species of Serrasalmus (Ostariophysii, Characidae) from the Upper Madeira River, Ama-
zon, Bolivia. Journal of Fish Biology, 69, 31–51.
Jégu, M. (1992) Variations du niveau marin et distribution des poisons d’eau douce en Amazonie Orientale. In: Prost,
M.T. (Ed.), Evolution des littoraux de Guyane et de la zone caraïbe méridionale pendant le quaternaire. ORSTOM,
Paris, pp. 281–297.
Jégu, M. (2003) Subfamily Serrasalminae (pacus and piranhas). In: Reis, R.E., Kullander, S.O. & Ferraris, C.J., Jr.
FREEMAN ET AL.
28 · Zootaxa 1484 © 2007 Magnolia Press
(Eds.), Check list of Freshwater Fishes of South and Central America. Edipucrs, Porto Alegre, Brazil, pp. 182–196.
Jégu, M. & Dos Santos, G.M. (1988) Le genre Serrasalmus (Pisces, Serrasalmidae) dans le bas Tocantins (Brésil, Parà),
avec la description d’une espèce nouvelle, S. geryi, du basin Araguaia–Tocantins. Revue d'Hydrobiologie Tropicale,
21, 239–274.
Jégu, M. & Dos Santos, G.M. (2001) Mise au point à propos de Serrasalmus spilopleura Kner, 1858 et réhabilitation de
S. maculatus Kner, 1858. Cybium, 25, 119–143.
Jégu, M., Leão, E.L.M. & Dos Santos, G.M. (1991) Serrasalmus compressus, une espece nouvelle du Rio Madeira, Ama-
zonie (Pisces: Serrasalmidae). Ichthyological Exploration of Freshwaters, 2, 97–108.
Kocher, T.D., Thomas, W.K., Meyer, A., Edwards, S.V., Paabo, S., Villablanca, F.X. & Wilson, A.C. (1989) Dynamics of
mitochondrial DNA evolution in animals: amplification and sequencing with conserved primers. Proceedings of the
National Academy of Science of the United States of America, 86, 6196–6200.
Lasso, C.A., Mojica, J.I., Usma, J.S., Maldonado-O., J.A., DoNascimiento, C., Taphorn, D.C., Provenzano, F., Lasso-
Alcala, O., Galvis, G., Vasquez, L., Lugo, M., Machado-Allison, A., Royero, R., Suárez, C. & Ortega-Lara, A.
(2004). Peces de la cuenca del río Orinoco. Parte I: lista de especies y distribución por subcuencas. Biota Colombi-
ana 5 (2), 95–158.
Lovejoy, N.R. 1997. Stingrays, parasites, and neotropical biogeography: a closer look at Brooks et al.’s hypotheses con-
cerning the origins of neotropical freshwater rays (Potamotrygonidae). Systematic Biology, 46, 218–230.
Ludwig, A., May, B., Debus, L. & Jenneckens, I. (2000) Heteroplasmy in the mtDNA control region of sturgeon (Aci-
penser, Huso and Scaphirhynchus). Genetics, 156, 1933–1947.
Lundberg, J.G., Machado-Allison, A. & Kay, R.F. (1986) Miocene characid fishes from Colombia: evolutionary stasis
and extirpation. Science, 234, 208–209.
Lundberg, J.G. (1997) Freshwater fishes and their paleobiotic implications. In: Kay, R.F., Madden. R.L., Cifelli, R.L. &
Flynn, J.J (Eds.), Vertebrate Paleontology in the Neotropics: The Miocene Fauna of La Venta, Columbia. Smithso-
nian Institution Press, Washington, D.C., pp. 67–91.
Lundberg, J.G. (1998) The temporal context for the diversification of neotropical fishes. In: Malabarba, L.R., Reis, R.E.,
Vari, R.P., Lucena, Z.M.S. & Lucena, C.A.S. (Eds.), Phylogeny and Classification of Neotropical Fishes. Edipucrs,
Porto Alegre, Brazil, pp. 49–68.
Lunt, D.H., Whipple, L.E. & Hyman, B.C. (1998) Mitochondrial DNA variable number tandem repeats (VNTRs): utility
and problems in molecular ecology. Molecular Ecology, 7, 1441–1455.
Machado-Allison, A. (1982) Studies on the systematics of the subfamily Serrasalminae (Pisces–Characidae). Unpub-
lished Ph.D. dissertation, The George Washington University, Washington D.C.
Machado-Allison, A. (1983) Estudios sobre la sistematica de la subfamilía Serrasalminae (Teleostei, Characidae). Parte
2. Discusión sobre la condición monofilética de la subfamilía. Acta Biologica Venezuelica, 11, 145–195.
Machado-Allison, A. (1985) Estudios sobre la sistemática de la subfamilia Serrasalminae. Parte III. Sobre el estatus
generico y relaciones filogenéticas de los géneros Pygopristis, Pygocentrus, Pristobrycon y Serrasalmus (Teleostei–
Characidae–Serrasalminae). Acta Biologica Venezuelica, 12, 19–42.
Machado-Allison, A. (2002) Los peces caribes de Venezuela: una aproximación a su estudio taxonómico. Boletín de la
Academia de Ciencias Físicas, Matemáticas y Naturales de Venezuela, 62(1), 35–88.
Machado-Allison, A. & Fink, W. (1995) Sinopsis de las especies de la subfamilia Serrasalminae presentes en la cuenca
del Orinoco: claves, diagnosis e ilustraciones. Serie Peces de Venezuela, Universidad Central de Venezuela, Cara-
cas, 90 pp.
Machado-Allison, A. & Fink, W. (1996) Los Peces Caribes de Venezuela: Diagnosis, Claves, Aspectos Ecologicos y
Evolutivos. Collection Monograph 52, Universidad Central de Venezuela, Caracas, 151 pp.
Machado-Allison, A., Fink, W. & Antonio, M. E. (1989) Revisión del género Serrasalmus Lacepede, 1803 y géneros rel-
acionados en Venezuela: I. Notas sobre la morfología y sistematica de Pristobrycon striolatus (Steindachner, 1908).
Acta Biologica Venezuelica, 12, 140–171.
Merckx, A., Jégu, M. & Dos Santos, G.M. (2000). Une nouvelle espèce de Serrasalmus (Teleostei: Characidae: Serras-
alminae), S. altispinis n. sp., décrite du rio Uatumã (Amazonas, Brésil) avec une description complémentaire de S.
rhombeus (Linnaeus, 1766) du Plateau Guyanais. Cybium, 24, 181–121.
Nakayama, C.M., Jégu, M., Porto, J.I.R. & Feldberg, E. (2001) Karyological evidence for a cryptic species of piranha
within Serrasalmus rhombeus (Characidae, Serrasalminae) in the Amazon. Copeia, 2001, 866–869.
Nakayama, C.M., Porto, J.I.R. & Feldberg, E. (2002) A comparative cytogenetic study of five piranha species (Serrasal-
mus, Serrasalminae) from the Amazon basin. Genetica, 114, 231–236.
Nico, L.G. (1991) Trophic ecology of piranhas (Characidae: Serrasalminae) from savanna and forest regions in the
Orinoco River basin of Venezuela. Unpublished Ph.D. dissertation, University of Florida, Gainesville, 210 pp.
Nico, L.G. & Taphorn, D.C. (1986) Those bitin’ fish from South America. Tropical Fish Hobbyist, 34 (Feb.), 24–27, 30–
34, 36, 40–41, 56–57.
Nico, L.G. & Taphorn, D.C. (1988) Food habits of piranhas in the low Llanos of Venezuela. Biotropica, 20, 311–321.
Nirchio, M., Granado, A., Ron, E. & Pérez, J.E. (2005). Karyotype and nucleolar organizer regions in Serrasalmus
Zootaxa 1484 © 2007 Magnolia Press · 29
PIRANHA MOLECULAR SYSTEMATICS
rhombeus (Serrasalminae) from Caricara del Orinoco, Venezuela. Interciencia, 27, 676–678.
Norman, J.R. (1929) The South American characid fishes of the subfamily Serrasalmoninae, with a revision of the genus
Serrasalmus Lacepède. Proceedings of the Zoological Society of London, 52 (1928), 781–829.
Oliveira C., Almeida-Toledo L.F., Foresti F., Britski H. & Toledo-Filho S.A. (1988) Chromosome formulae of Neotropi-
cal freshwater fishes. Revista Brasileira de Genética, 11, 577–624.
Ortí, G., Petry, P., Porto, J.I.R., Jégu, M. & Meyer, A. (1996) Patterns of nucleotide change in mitochondrial ribosomal
RNA genes and the phylogeny of piranhas. Journal of Molecular Evolution, 42, 169–182.
Ortí, G., Porto, J.I.R. & Sivasundar, A. (2000) Phylogeny of the Serrasalminae (Characiformes) based on mitochondrial
DNA sequences. Unpublished poster presented at the 80th Annual Meeting of the American Society of Ichthyolo-
gists and Herpetologists, La Paz, Baja California Sur, Mexico (14–20 June 2000). Available from http://
golab.unl.edu/people/asivasun/index.html (accessed 30 May 2005)
Palumbi, S.R. (1996) Nucleic acids II: The polymerase chain reaction. In: Hillis, D.M., Moritz, C. & Mable, B.K. (Eds.),
Molecular Systematics, 2nd edition. Sinauer Associates, Sunderland, Massachusetts, pp. 205–247.
Porto, J.I.R., Feldberg, E., Nakayama, C.M., Maia, R.O. & Jégu, M. (1991) Cytotaxonomic analysis in the Serrasalmidae
(Ostariophysi, Characiformes). Abstracts of presentations, Seventh International Ichthyology Congress of the Euro-
pean Ichthyological Union, The Hague, The Netherlands. Bulletin Zoologisch Museum, Universiteit van Amsterdam
(special issue), 66.
Porto, J.I.R., Feldberg E., Nakayama, C. & Falcao, J.N. (1992) A checklist of chromosome numbers and karyotypes of
Amazonian freshwater fishes. Revue d'Hydrobiologie Tropicale, 25, 287–299.
Posada D. & Crandall, K.A. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics, 14, 817–818.
Raffles, H. & Winkler-Prins, A.M.G.A. (2003) Further reflections on Amazonian environmental history: transformations
of rivers and streams. Latin American Research Review, 38, 165–187.
Ravago, R.G., Monje, V.D. & Juinio-Menez, M.A. (2002) Length and sequence variability in mitochondrial control
region of the Milkfish, Chanos chanos. Marine Biotechnology, 4, 40–50.
Ray, D.A. & Densmore, L. (2002) The crocodilian mitochondrial control region: General structure, conserved sequences,
and evolutionary implications. Journal of Experimental Zoology, 294, 334–345.
Reis, R.E. (1998) Systematics, biogeography, and the fossil record of the Callichthyidae: a review of the available data.
In: Malabarba, L.R., Reis, R.E., Vari, R.P., Lucena, Z.M.S. & Lucena, C.A.S. (Eds.), Phylogeny and Classification
of Neotropical Fishes. Edipucrs, Porto Alegre, Brazil, pp. 351–374.
Ronquist, F. & Huelsenbeck, J. (2003) MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformat-
ics, 19, 1572–1574.
Rubilar, A. (1994) Diversidad ictiológica en depósitos continentales miocenos de la Formación Cura–Mallín, Chile (37–
39°S): implicancias paleogeográficas. Revista Geológica de Chile, 21, 3–29.
Saghai-Maroof, M.A., Soliman, K.M., Jorgensen, R.A. & Allard, R.W. (1984) Ribosomal DNA spacer length in barley:
Mendelian inheritance, chromosomal location, and population dynamics. Proceedings of the National Academy of
Science of the United States of America, 81, 8014–8018.
Sazima, I. & Andrade–Guimaraes, S. (1987) Scavenging on human corpses as a source for stories about man–eating pira-
nhas. Environmental Biology of Fishes, 20, 75–77.
Sazima, I. & Machado, F.A. (1990) Underwater observations of piranhas in western Brazil. Environmental Biology of
Fishes, 28, 17–31.
Sivasundar, A., Bermingham, E. & Ortí , G. (2001) Population structure of migratory freshwater fishes (Prochilodus:
Characiformes) in major South American rivers. Molecular Ecology, 10, 407–418.
Slowinski, J.B. (2001) Molecular polytomies. Molecular Phylogenetics and Evolution, 19, 114–120.
Stern, K.M. (1970) Der Casiquiare–Kanal, einst und jetzt. Amazoniana, 2, 401–416.
Sternberg, H.O. (1975) The Amazon River of Brazil. Geographische Zeitschrift, 40, 1–74.
Swofford, D.L. (2002) PAUP*: Phylogenetic Analysis Using Parsimony (*and Other Methods). Sinauer Associates, Sun-
derland, Massachusetts.
Thompson, J.D., Gibson T.J., Plewniak F., Jeanmougin F. & Higgins D.G. (1997) The CLUSTAL_X windows interface:
flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Research 25(24),
4876–82.
Van Every, L.R. & Kritsky, D.C. (1992) Neotropical Monogenoidea. 18 Anacanthorus Mizelle and Price, 1965 (Dactylo-
gyridae, Anacanthorinae) of piranha (Characoidea, Serrasalmindae) from the Central Amazon, their phylogeny, and
aspects of host–parasite coevolution. Journal of Helminthological Society of Washington, 59, 52–75.
Walsh, H.E. & Friesen, V.L. (2001) Power and stochasticity in the resolution of soft polytomies: A reply to Braun et al.
Evolution, 55, 1264–1266.
Walsh, H.E., Kidd, M.G., Moum, T. & Friesen, V.L. (1999) Polytomies and the power of phylogenetic inference. Evolu-
tion, 53, 932–937.
Winemiller, K.O. (1989) Ontogenetic diet shifts and resource partitioning among piscivorous fishes in the Venezuelan
llanos. Environmental Biology of Fishes, 26, 177–199.
FREEMAN ET AL.
30 · Zootaxa 1484 © 2007 Magnolia Press
Appendix 1
Photographs of live and preserved specimens of selected serrasalmine taxa used as source material for genetic analyses in
this study. Specimen numbers (1–31) refer to material listed in Table 1 and presented in distribution map (Figure 4)
and cladograms (Figures 7 and 8). Photographs of live or fresh material are indicated by “L”. Photographs in boxes
indicate specimens from same site as the genetic specimens. Hyaline (clear) portions of fins show background color
(typically blue or gray). Specimens are proportional to scale bar in millimeters. All photographs by Howard Jelks
and Leo Nico unless otherwise indicated.
PLATE 1. Serrasalmus manueli (1–9). (No photograph or voucher available for specimen 3.) (Photograph of 8-S. man-
ueli by Frank Pezold.)
Zootaxa 1484 © 2007 Magnolia Press · 31
PIRANHA MOLECULAR SYSTEMATICS
PLATE 2. Serrasalmus manueli (10) and S. gouldingi (11–15). (Photographs of S. gouldingi specimens 12 and 14 by
Donald Taphorn.)
FREEMAN ET AL.
32 · Zootaxa 1484 © 2007 Magnolia Press
PLATE 3. Serrasalmus gouldingi (16), Serrasalmus sp. A (17), and S. medinai (18–19). Images of Serrasalmus sp. “A”
are of specimen 17 (originally captured in 1991) at different ages, including the same fish live in captivity photographed
in 1993 (as juvenile about 2+ years old), 2006, and after preservation in 2007.
Zootaxa 1484 © 2007 Magnolia Press · 33
PIRANHA MOLECULAR SYSTEMATICS
PLATE 4. Serrasalmus medinai (20) and S. irritans (21–23). (Photograph of live 20-S. medinai by Noel Burkhead.)
FREEMAN ET AL.
34 · Zootaxa 1484 © 2007 Magnolia Press
PLATE 5. Serrasalmus irritans (24), Pygocentrus cariba (25), and Pygopristis denticulatus (27). (No photograph or
voucher available for specimen 26.)
Zootaxa 1484 © 2007 Magnolia Press · 35
PIRANHA MOLECULAR SYSTEMATICS
PLATE 6. Pristobrycon striolatus (28–31) and additional small juvenile specimens collected with genetic vouchers
showing life colors.
FREEMAN ET AL.
36 · Zootaxa 1484 © 2007 Magnolia Press
PLATE 7. Piaractus brachypomus (33) (Photograph provided by Robert Lea). (No photograph or voucher available for
specimen 32.)
Zootaxa 1484 © 2007 Magnolia Press · 37
PIRANHA MOLECULAR SYSTEMATICS
Appendix 2. GenBank sequences used in this study.
GenBank Accession Numbers
Species ID Control
region 16S 12S
Locality Citation
Serrasalmus spilopleura AF283948 U33592 U33560 Rio Uruguay, Salto Grande, Argentina Ortí et al. (1996)
Serrasalmus sp. U33593 U33561 Rio Negro-Solimoes, AM, Brazil Ortí et al. (1996)
Serrasalmus compressus U33594 U33562 Rio Solimoes, Ilha da Marchantaria, AM, Brazil Ortí et al. (1996)
Serrasalmus sp. 218 AF283935 AF283914
Serrasalmus sp. 219 AF283936 AF283915
Serrasalmus rhombeus AF283937 AF283920
Serrasalmus rhombeus str. 222 AF283952 AF281941 AF283916
Serrasalmus compressus str. 2241 AF283938 AF283917
Serrasalmus compressus AF283939 AF283918
Serrasalmus humeralis AF283940 AF283919
Serrasalmus sp. 220 AF283951
Serrasalmus manueli str. p18 AF283950 AF283942 AF283921
Serrasalmus gouldingi str. P17 AF283944 AF283943 AF283922
Serrasalmus eigenmanni AF283946
Pygocentrus cariba AF283954
Pygocentrus nattereri str.
INPA10143
AF283953 U33590 U33558 Rio Solimoes, Ilha da Marchantaria, AM, Brazil Ortí et al. (1996)
Pygocentrus nattereristr.
USNM325686 U33591 U33559 Rio Uruguay, Salto Grande, Argentina Ortí et al. (1996)
Pygopristis denticulatus str. p4 AF284464
Pristobrycon sp. 256 AF283949
Pristobrycon serralatus AF283945
Pristobrycon sp. 224 AF283947 U33595 U33563 Rio Solimoes, Ilha da Marchantaria, AM, Brazil Ortí et al. (1996)
Pristobrycon striolatus str. 225 AF284463 U33596 U33597 Rio Pitinga, UHE do Pitinga, AM, Brazil Ortí et al. (1996)
Pristobrycon striolatus str. 226 U33598 U33564 Rio Pitinga, UHE do Pitinga, AM, Brazil Ortí et al. (1996)
Catoprion mento str. 80 AF284462 U33599 U33565 commercial source, locality unknown Ortí et al. (1996)
Metynnis sp. U33600 U33566 commercial source, locality unknown Ortí et al. (1996)
Metynnis cf. Mola U33601 U33567 Rio Miranda, Pantanal Matogrossense, Campo Grande, MS,
Brazil
Ortí et al. (1996)
FREEMAN ET AL.
38 · Zootaxa 1484 © 2007 Magnolia Press
Metynnis hypsauchen AF283957 AF283934 AF283913
Metynnis sp. p31 AF283955
Metynnis sp. p32 AF283956 AF283933 AF283912
Ossubtus xinguense str. 253 AF284461
Acnodon normani str. 243 AF284460
Myleus rhombeus AF283976
Myleus sp. p51 AF283975
Myleus sp. p49 AF283974
Myleus pacu str. p33 AF283970
Myleus pacu str. 238 AF283969 Rio Pitinga, Cachoeira 40 Ilas, AM, Brazil
Myleus schomburgkii AF283968 Rio Pitinga, Cachoeira 40 Ilas, AM, Brazil
Myleus ternetzi AF283967
Myleus tiete AF283966 Rio Miranda, Pantanal Matogrossense, Campo Grande, MS,
Brazil
Myleus rubripinnis AF283965 commercial source, locality unknown
Myleus asterias AF283964 R. Pitinga, UHE do Pitinga, AM, Brazil
Mylesinus paucisquamatus AF283973
Mylesinus paraschomburgkii AF283971 R. Pitinga, Cachoeira 40 Ilas, AM, Brazil
Tometes sp. 246 AF283972
Colossoma macropomum AF283963 R. Solimoes, Ilha da Marchantaria, AM, Brazil
Mylossoma paraguayensis AF283962 R. Miranda, Pantanal Matogrossense, Campo Grande, MS,
Brazil
Mylossoma duriventri AF283961 R. Solimoes, Ilha da Marchantaria, AM, Brazil
Piaractus brachipomus str. 45 AF283960 commercial source, locality unknown
Piaractus brachipomus str. 200 AF283958 R. Solimoes, Ilha da Marchantaria, AM, Brazil
Piaractus mesopotamicus AF283959 R. Miranda, Pantanal Matogrossense, Campo Grande, MS,
Brazil