Molecular Ecology (2003) 12, 1855 –1867 doi: 10.1046/j.1365-294X.2003.01857.

Historical biogeography of the catfish genus Hypostomus

Blackwell Publishing Ltd.

(Siluriformes: Loricariidae), with implications on the

diversification of Neotropical ichthyofauna
J . I . M O N T O Y A - B U R G O S *†
*Museum of Natural History, 1 rte de Malagnou, CP 6434, 1211 Geneva 6, Switzerland, †Department of Zoology and Animal Biology,
University of Geneva, 154 rte. de Malagnou, 1224 Chêne-Bougeries, Switzerland

Abstract
Tropical South America possesses the largest ichthyofauna of any continental region. To
test whether palaeohydrological changes may have been the causes of such diversification,
the ‘hydrogeological’ hypothesis, the phylogenetic relationships of 51 representatives of
the catfish genus Hypostomus (Siluriformes: Loricariidae) were inferred using mitochon-
drial D-loop haplotype sequences. Specimens were collected in all main tropical South
American rivers systems east to the Andes. The major interrelationships found with the
D-loop data were confirmed with a subset of 21 species using complete internal transcribed
spacer (ITS) region sequences. The phylogenetic analysis indicate that the genus Hyposto-
mus can be divided into four monophyletic clades. The historical biogeographical analysis
of each of these clades allows the identification of seven major cladogenetic events. Using
calibrated D-loop and ITS molecular clocks, date estimations were attributed to each of
these cladogenetic events allowing a linkage between four of them with documented hydro-
geological changes. Comparisons with published distribution patterns of unrelated fish
groups indicate that several of the reconstructed and dated hydrogeological–cladogenetic
events may have acted at a large scale on the diversification of Neotropical freshwater fish
fauna during late Tertiary.
Keywords: catfishes, diversification, D-loop, historical biogeography, Hypostomus, Neotropics

Received 24 October 2002; revision received 21 February 2003; accepted 5 March 2003

ations of the Pleistocene (1.8 million to 11 000 years ago),

Introduction
which fragmented and reconnected habitats several times,
The freshwater biota of tropical Central and South America, promoting allopatric speciation.
or Neotropics, host one-quarter of all fish species of the Several findings, however, challenge the refuge theory,
world (Schaefer 1998). The underlying forces that have at least concerning ichthyofauna diversification. Morpho-
driven this extraordinary diversification are still largely logically based studies of phylogenetic biogeography of
unknown. Until now, few explicit hypotheses have been South American freshwater fishes support an important
put forward trying to explain this phenomenon. Two diversification predating the Pleistocene (Weitzman &
decades ago the most popular hypothesis was the refuge Weitzman 1982; Vari 1988; Vari & Weitzman 1990). The scarce
theory of Haffer (1982), which states that the main diver- Neotropical fossil records also indicate that a high degree
sification of Neotropical organisms occurred recently of diversification was already reached during the Miocene
and that the primary forces were the climatological fluctu- (24–5.2 Ma). In a fossil deposit of late middle Miocene (c.
10 Ma), many fish fossils could be assigned to modern
genera or even species indicating that the fauna was essen-
Correspondence: Juan Ignacio Montoya-Burgos, Department of
tially modern across a wide taxonomic and ecological range
Zoology and Animal Biology, University of Geneva, 154 rte. de (Lundberg 1997, 1998). Recently, new evidence based on
Malagnou, 1224 Chêne-Bougeries, Switzerland. Fax: + 41 22 349 molecular phylogenies and population genetics agree with
86 29; E-mail: [email protected] a pre-Pleistocene diversification of some Neotropical

© 2003 Blackwell Publishing Ltd

1856 J . I . M O N T O Y A - B U R G O S

freshwater fishes (e.g. Lovejoy & De Araújo 2000), as well world, and providing material for repeatedly and inde-
as other vertebrates such as lizards (Glor et al. 2001). These pendently testing hypotheses of historical biogeography,
results point to another possible major force responsible few fruitful morphologically based hypotheses have
for the diversification and this is the particularly event- been proposed (Vari 1988). The reason lies in the poor
ful history of Neotropical river systems, which we will call knowledge of geographical distribution of species and
the ‘hydrogeological’ hypothesis. Models for the evolu- the fact that many taxa do not represent natural assem-
tionary diversification and biogeography of the Neotro- blages (Vari & Weitzman 1990). Despite the extensive use
pical aquatic fauna must indeed consider the long and complex of mitochondrial DNA (mtDNA) based phylogenies for
history of South America’s landscape and river systems. reconstructing phylogeographical patterns (Avise 2000),
The formation of drainage divides, shifting courses of only recently has the issue of Neotropical freshwater fish
rivers and repeated incursions and regressions of marine phylogeography and diversity been addressed using
waters must have produced many vicariant events pro- mtDNA (Hrbek & Larson 1999; Lovejoy & De Araújo 2000;
moting biotic enrichment (Lundberg 1998) (for alternative Martin & Bermingham 2000; Sivasundar et al. 2001).
models concerning nonaquatic vertebrates see Haffer The Neotropical catfish genus Hypostomus belongs to
1997). the family Loricariidae. Members of this family are easily
Freshwater fishes are largely incapable of surviving out- recognizable by their sucker-like mouth and by their
side their aquatic environment due to morphological and body covered with bony plates. The systematics of the
physiological limitations, so that even relatively minor ter- Loricariidae has been re-assessed recently using molecular
restrial barriers isolate populations within their drainage means, leading to an important revision at the subfamily
basins (Vari 1988). This property makes freshwater fishes level (Montoya-Burgos et al. 1997, 1998). In a recent study,
suitable organisms for investigating past geological and Montoya-Burgos et al. (2002) have investigated further the
environmental modifications, because the dispersal of phylogenetic relationships within Hypostomus and related
fishes is linked intimately to direct connections or head- genera using mitochondrial D-loop sequences. Several
waters captures between adjacent basins, which in turn reflect aspects of the systematics of these genera were clarified,
geomorphological or climatic changes (Lundberg 1993; leading to a revision of the genus Hypostomus.
Bermingham & Martin 1998). Thus, phylogeographical With about 110 species, the genus Hypostomus is one of
analysis of fish groups showing wide distribution ranges, the most diversified freshwater catfish group. Hypostomus
inhabiting all major river basins of the studied area, pro- species occur almost everywhere in East Andean tropical
vide natural information for the reconstruction of the his- and south temperate South America and colonize nearly
torical events that originated their diversification. all aquatic habitats, although preferring running waters.
Although Schaefer (1998) estimated 8000 species of They are mainly herbivorous or detritus-eating fishes, with
freshwater fishes in the Neotropical realm, making it the minor commercial importance. Because no marked migra-
largest ichthyofauna of any continental region in the tory behaviour has been reported for Hypostomus species,

Table 1 Documented hydrogeological changes which may have influence freshwater fishes diversification in the Neotropics

Hydrogeological changes Dates in Ma References

1. Isolation of the Lago Maracaibo from 8 Hoorn (1993)

the palaeo Amazon-Orinoco Hoorn et al. (1995)
Diaz de Gamero (1996)
2. Boundary displacements between the 11.8–10 Lundberg et al. (1998)
palaeo Amazon–Orinoco and the Paraná systems
3. Low sea-level period (promoting colonization of 6–5 Vail & Hardenbol (1979)
neighbouring rivers through delta connections) Haq et al. (1987)
Wheeler & Ahron (1991)
4. East coastal Paraíba River disconnected middle Miocene Riccomini et al. (1990)
from the upper Paraná (16–10) Riccomini et al. (1991)
5. Boundary displacements between the Tertiary Beurlen (1970)
upper Paraná and the São Francisco basins (65–1.8)
6. The Uruguay River disconnected from the Miocene Beurlen (1970)
upper Paraná (24 – 5)

© 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 1855 –1867

 H I S T O R I C A L B I O G E O G R A P H Y O F T H E C A T F I S H H Y P O S T O M U S 1857

pared to the reported distribution patterns of other Neo-

tropical freshwater fish groups.

Materials and methods

The entire mitochondrial control region, or D-loop, was
sequenced in 51 specimens. Hypostomus sampling localities
are given in Fig. 1. Representatives of six genera, deter-
mined previously to be related closely to Hypostomus
(Montoya-Burgos et al. 1998; Montoya-Burgos et al. 2002),
were sequenced additionally and used as an outgroup. The
complete nuclear ITS region (ITS1, 5.8S rRNA, ITS2) was
sequenced for a subset of 21 Hypostomus and five outgroup
species in order to check the validity of the interrelation-
ships obtained using the D-loop sequences.

DNA extraction and sequencing

Ethanol fixed tissue samples were dried, and total DNA
was extracted using a QIAamp DNA Mini Kit (Qiagen). In
order to amplify the complete mitochondrial D-loop region
by polymerase chain reaction (PCR), two primers were
Fig. 1 Map of tropical and south temperate South America designed in the tRNAPro gene, located at the 5′ side of the
presenting the collection localities of the 51 Hypostomus specimens D-loop (DLA-D: 5′ TCCYACCCCTAACTCCCAAAG; DLB-D:
and the seven additional outgroup species. One point may 5′ AGCRYCGGTCTTGTAATCCG), and two primers in
represent more than one locality. the tRNAPhe gene, located at the 3′ side of the D-loop
(DLA-R: 5′ AGTCAGGACCAARCCTTTGTGC; DLB-R: 5′
GGYYCATCTTGACATCTTCAG). Two additional inter-
and because rivers rather than lakes are the primary aquatic nal primers were designed for completing the direct sequen-
habitat in the Neotropics (Vari 1988), species of this genus cing (DLA I: 5′ TTATTRGRTATGTCTGGG; DLA II: 5′
are candidates of choice for investigating the role of past CCATGCYAARGCATTCTTTC).
hydrogeological changes in shaping Neotropical fish Two primers were designed for amplifying the ITS region:
diversity. SBR (5′ GTAGGTGAACCTGCAGAAGG), located at the 3′
This study focuses on two main issues. The first is to end of the 18S rRNA gene, and JM5 (5′ TACCGGCCT-
test the hydrogeological hypothesis by asking whether CACACCGTCC), located at about 200 base pairs (bp)
documented geomorphological or climatic-associated inside the 28S rRNA gene.
changes in river patterns (Table 1) could have acted as major The PCR reactions were conducted in a total volume of
freshwater fish diversifying forces in the Neotropics. The 50 µL, with 2 units of Taq DNA polymerase (Roche) and
second is to assess the temporal dimension of the diver- 1 µL of each primer (10 pmol/µL). Cycle profiles were as
sifying events that have led to the incredible richness of follows: 1 min initial denaturation at 94 °C followed by 40
the Neotropical ichthyofauna. These questions are cycles with 30 s at 94 °C, 30 s at 53–60 °C, and 2 min at
approached by inferring the mtDNA D-loop haplotype 72 °C. PCR products were purified using the High Pure
phylogeny of 51 Hypostomus specimens collected in all PCR Product Purification Kit (Roche). The D-loop PCR
main Neotropical river systems (Fig. 1). To confirm the products were sequenced directly while the ITS amplifica-
major patterns of interrelationships obtained with the tions were cloned before sequencing by ligation into the
D-loop sequences, the entire nuclear ribosomal internal pGEM-T Vector System (Promega) and cloned into XL-2
transcribed spacer (ITS) region was sequenced for a subset Blue Ultracompetent Cells (Stratagene). Cycle sequencing
of 21 Hypostomus species. For both data sets, the rate of was carried out using the BigDye Terminator Cycle
sequence evolution is calibrated using an external geolo- Sequencing Ready Reaction Kit (Applied Biosystems).
gical event for which an accurate age estimate is available. The cycle sequencing reaction profile was set to 25
Combined with distribution data, these phylogenies are cycles with 15 s at 96 °C, 5 s at 52 °C and 4 min at 60 °C.
interpreted in the framework of geological evidences Sequences were scored on a 377 PRISM automated
about the history of Neotropical river systems. In order to sequencer (ABI). The nucleotide sequence data have been
assess the relevance of the findings, the results are com- deposited in GenBank: D-loop sequence Accession nos:

© 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 1855–1867

1858 J . I . M O N T O Y A - B U R G O S

AJ318344–AJ318365; AJ318369–AJ318379; AJ315762–

Results
AJ315764; ITS sequence Accession nos: AJ412849 –AJ412874.

D-loop and ITS sequence variation

Phylogenetic methods
The D-loop sequence alignment comprises 660 sites, from
Sequences were aligned manually using the Genetic Data which 316 are variable and 243 are parsimony informative.
Environment software, version 2.2 (Larsen et al. 1993). Maxi- Intraspecific sequence divergence (uncorrected) ranges
mum parsimony (MP; Fitch 1971), distance and maximum from 0 to 0.4% among Hypostomus watwata specimens
likelihood (ML; Felsenstein 1981) phylogenetic analyses collected in the Oyapock and Maroni rivers, separated by
were performed using paup* version 4.0b6 (Swofford 300 km (French Guyana). Among all Hypostomus species,
2000). All sites with gaps or of uncertain alignment were maximum divergence is 11.5% (H. hondae vs. Hypostomus
excluded from the analyses. The MP method was performed sp. 1026). Base composition and structural characteristics
using heuristic searches with 100 replicates of random of Hypostomus D-loop sequences are described elsewhere
addition of taxa and tree-bisection–reconnection (TBR) (Montoya-Burgos et al. 2002).
branch swapping. The robustness of the resulting topology The ITS sequences show a minimum of 1.54% uncorrected
was assessed by bootstrap analysis with 500 replicates of divergence (between H. boulengeri and H. punctatus), whereas
heuristic search using stepwise addition of taxa and TBR maximum divergence is 6.62% (H. watwata from Maroni
branch swapping. In order to determine the best model of River vs. H. latifrons). Base composition is biased markedly
sequence evolution for distance and likelihood methods, in favour of G and C (G + C = 67%); mean values are: A =
likelihood scores were determined for three different 0.17, C = 0.35, G = 0.32, T = 0.15. Among the 1011 aligned
models: K2P (Kimura 1980), HKY85 (Hasegawa et al. 1985) sites, 280 are variable and 125 are parsimony informative.
and GTR (Yang 1994). Additionally, the incorporation
into each of the models of (1) among-sites rate hetero-
Hypostomus phylogeny
geneity using the Γ-distribution (Yang 1993) and (2) the
Γ-distribution and the estimated proportion of invariable The likelihood ratio tests determined that the GTR + Γ
sites (Γ + I; Gu et al. 1995) were examined. All model para- model fits the D-loop and the ITS data significantly better
meters were estimated via ML and based on the topology than all other models tested (P < 0.001). Therefore, ML and
obtained under the K2P model. Likelihood ratio tests were distance methods were performed under the GTR + Γ
used to compare likelihood scores obtained for pairs of model. Regarding the D-loop, this result is congruent with
nested models to determine which model fits the sequence previous tests performed on a subset of Hypostomus species
data the best (Yang et al. 1994; Swofford et al. 1996). (Montoya-Burgos et al. 2002).
For the final ML analysis, the parameters of the best fit All methods used for the analysis of the D-loop data
model were estimated on the basis of the maximum likeli- yielded similar topologies. The ML tree is shown in Fig. 2A
hood tree obtained via heuristic search using the NJ start- (lnL = −6726.965). Three monophyletic clades were used
ing tree and TBR branch swapping. The same parameters as outgroups (clades A, B, and C; Fig. 2A). Clade A con-
were used to correct genetic distances for NJ analyses tains two Hypostomus species which were determined pre-
(Saitou & Nei 1987). Bootstrapping was also performed viously to be related much more closely to Aphanotorulus
with 300 replicates for the ML analyses and 1000 replicates than to Hypostomus (Montoya-Burgos et al. 2002). Clade D
for the NJ analyses. contains all remaining Hypostomus species, which can be
In order to check for substitution rate homogeneity organized into four groups, named herein D1, D2, D3 and
among Hypostomus main lineages, relative-rate tests D4. H. gymnorhynchus, the sister species of clade D2, is
were performed using the RRTree software version 1.1 related morphologically to members of this clade. How-
(Robinson-Rechavi & Huchon 2000). The null hypothesis ever, the bootstrap value supporting this relationship is
of constant molecular clock was also tested for the ingroup too weak in the D-loop tree for including this species
taxa using the log likelihood ratio test (Huelsenbeck & confidently within D2. The D2 clade is subdivided further
Crandall 1997). into two monophyletic groups, D2-1 and D2-2, while the
In order to determine if alternative topologies corre- D3 clade is subdivided in turn into D3-1 and D3-2.
sponding to different scenarios of river basin group- The MP analysis produced 201 equally parsimonious
ings were significantly different from the inferred ML trees (length = 1161; CI = 0.383; RI = 0.6). Compared to the
D-loop tree, the Kishino–Hasegawa (KH) test (Kishino & ML tree, the MP strict consensus tree (not shown) differs in
Hasegawa 1989) was used when a single alternative scenario the lack of resolution among basal species of clade D4, the
was tested while the KH and the Shimodaira–Hasegawa lack of resolution among some closely related species of
(SH) (Shimodaira & Hasegawa 1999) tests were used when clade D3, the unresolved position of H. watwata at the base
several alternative scenarios were tested. of clade D2 and the unresolved relationships among H. sp.

© 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 1855 –1867

 H I S T O R I C A L B I O G E O G R A P H Y O F T H E C A T F I S H H Y P O S T O M U S 1859

Fig. 2 (A) Maximum likelihood tree based on the D-loop sequences of Hypostomus species, obtained under the GTR + Γ model (ln L =
– 6726.965). The following model parameters were used: gamma shape parameter = 0.359; base frequency: A = 0.383, C = 0.194, G = 0.113,
T = 0.31; substitution matrix: AC = 0.894, AG = 9.175, AT = 0.845, CG = 0.397, CT = 6.987. Numbers above branches indicate bootstrap
support values above 50% for ML, MP and NJ methods, respectively. Bold letters are abbreviations used for naming clades (see text). (B)
Maximum likelihood tree of a subset of 21 Hypostomus species based on the ITS region and obtained under the GTR + Γ model (ln L =
– 4748.709). The following model parameters were used: gamma shape parameter = 0.345; base frequency: A = 0.169, C = 0.356, G = 0.327,
T = 0.148; substitution matrix: AC = 0.955, AG = 2.779, AT = 0.664, CG = 0.336, CT = 2.964. Numbers above branches indicate bootstrap
© 2003 Blackwell
support Publishing
values above ML,Molecular
50% forLtd, MP, and Ecology, 12, 1855–1867
NJ methods, respectively.
1860 J . I . M O N T O Y A - B U R G O S

Table 2 Relative-rate tests

D-loop ITS

Lineages ∆K SD P ∆K SD P

D1 vs. D2 0.007 0.008 0.428 0.0003 0.006 0.956

D1 vs. D3 0.005 0.010 0.625 0.004 0.006 0.567
D1 vs. D4 0.011 0.008 0.178 0.003 0.006 0.638
D2 vs. D3 0.002 0.008 0.819 0.004 0.007 0.540
D2 vs. D4 0.005 0.008 0.574 0.003 0.007 0.608
D3 vs. D4 0.007 0.008 0.419 0.001 0.004 0.849

1204 + H. sp. 1211, H. sp. Tib13 and the clade composed by example, H. fonchii has been described very recently
H. commersonii + H. ancistroides + H. punctatus + H. affinis + (Weber & Montoya-Burgos 2002) while H. sp. 906, 1013
H. sp. 1161. Compared to the ML tree, the NJ tree differs in and 1134 are currently under description. The discovery,
(i) some interrelationships of basal species of clade D4; (ii) description and knowledge of species distribution is still a
H. luteomaculatus is placed as the sister taxa of H. regani + H. considerable challenge for Neotropical ichthyologists
nigromaculatus; (iii) H. sp. 699 is placed at the base of clade (Vari & Weitzman 1990). In this context, the biogeograph-
D3-1; and (iv) H. commersonii is placed as the sister taxa of ical analyses proposed hereafter have been based on the
H. punctatus + H. affinis + H. sp. 1161 + H. ancistroides. current available data on species distribution and for
Regardless of which method was used to construct the simplicity, species distribution ranges have been replaced
ITS phylogeny, the major groupings within Hypostomus by the point locality of the collected specimens (Figs 3
were identical to the D-loop tree. Figure 2B presents the and 4). In what follows, the discussed phylogenetic
ML tree (ln L = – 4748.709). The MP analysis resulted in relationships concern, in general, deeper than sister-
three equally parsimonious trees (length = 501; CI = 0.68; species relationships in order to reduce the potential
RI = 0.688), differing only in the interrelationships within problem of persistent ancestral polymorphism which may
clade D3-1. The NJ tree is identical to the ML tree. mislead the interpretations (for a relevant view of this
The four main Hypostomus clades, D1 to D4, which are topic, see Edwards & Beerli 2000). It will be assumed that
supported by bootstrap values above 50% with both data reciprocal monophyly of the D-loop haplotypes has been
sets, represent in addition homogeneous morphological reached among the discussed clades.
assemblages.

Testing the hydrogeological hypothesis

Evolutionary rates
To test the hydrogeological hypothesis of Neotropical
Relative-rate tests were performed to compare substitution freshwater fish diversification and to evaluate its temporal
rates among the four monophyletic Hypostomus lineages dimension, the D-loop and ITS rates of evolution are
(D1–D4). For the D-loop and the ITS data, pairwise lineage calibrated in Hypostomus based on the amount of sequence
comparisons were performed using clade A as outgroup. divergence between species separated by a dated geo-
The results (Table 2) indicate that substitution rates are logical event. The biogeographical analysis of clade D1
homogeneous among all Hypostomus lineages for both (Fig. 3) indicates that H. hondae, distributed only in the
genetic markers (P > 0.05). When analysing only the Lago Maracaibo and Magdalena basins, is the closest
ingroup taxa (clade D, Fig. 2) the log-likelihood ratio test relative to H. plecostomoides, which is known only from the
showed no significant differences between the likelihood Orinoco basin and some localities in the upper Amazon.
scores obtained when enforcing or not the molecular clock, Because these distribution patterns match with the
either for the D-loop (χ2 = 60.3; d.f. = 47; P > 0.05) or the ITS vicariant episode that separated Lago Maracaibo system
data (χ2 = 26.2; d.f. = 18; P > 0.05). from Amazon and Orinoco basins 8 Ma ago (Table 1, event
1), it is reasonable to attribute this age to the speciation
event that gave rise to H. hondae and H. plecostomoides. This
Discussion
same geological event has been used previously for
Several specimens analysed could not be identified at the calibrating other Neotropical fish phylogenies (Lovejoy
species level, despite meticulous expertise, and were et al. 1998; Sivasundar et al. 2001).
labelled therefore with their field numbers. Many if not all Because all Hypostomus lineages were found to have sim-
these unidentified specimens may be new species. For ilar substitution rates in the D-loop and the ITS regions, the

© 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 1855 –1867

 H I S T O R I C A L B I O G E O G R A P H Y O F T H E C A T F I S H H Y P O S T O M U S 1861

Fig. 3 Biogeographical scheme of Hypostomus clades D1 and D2. The cladograms show the resolved relationships found in both the D-loop
and ITS phylogenetic trees and which are supported by bootstrap values above 50%. In clade D1, the splitting between H. hondae
and H. plecostomoides can be dated to 8 Ma on the basis of geological evidence (see text). In clade D2, the two main boxes indicates lineages
found exclusively in the Amazon plus northeastern South America coastal rivers (NE coastal) and in the Paraná system plus eastern
South America coastal rivers (E coastal). Divergence date estimations were inferred using the D-loop and, when possible, the ITS molecular
clocks.

H. hondae–H. plecostomoides separation date can be used for hydrogeographical changes and their temporal estimates
calibrating the molecular clock and inferring age estimates (Table 1).
throughout the ingroup subtree. For the D-loop data, The biogeographical analysis of clade D2 (Fig. 3) indi-
node-to-tip distances taken from the ML tree result in 7.4% cates an old splitting between species from the Amazon
sequence divergence between H. hondae and H. plecosto- plus northeastern South American coastal rivers (NE coastal
moides (for an 8-Ma period), that is 0.93% per Ma. Using rivers: French Guyana, Gurupí, Itapicurú and Parnaíba rivers)
this same procedure, the evolutionary rate for the ITS on one side and species inhabiting the Paraná system plus
region was estimated to be 0.31% sequence divergence per eastern South American coastal rivers (E coastal rivers:
Ma; that is, three times slower than the D-loop. The rate of Ribeira, Ubatiba and Paraíba rivers) on another side. The
sequence divergence obtained for the D-loop is 1.7–3.6 Paraná system refers to the Paraná, Paraguay and Uruguay
times less than previously reported estimations based on rivers. This splitting event can be dated between 11.4 and
other fish groups (Donaldson & Wilson 1999; Sivasundar 10.5 Ma (using the D-loop and ITS molecular clocks, respect-
et al. 2001). This difference can be explained by the exclu- ively). This major division and its inferred age match
sion from our analyses of the variable regions located at the almost perfectly with the documented boundary displace-
3′ and 5′ ends of the D-loop which could not be aligned ments between the northern palaeo Amazon–Orinoco
unambiguously, in particular with the distant outgroup system and the southern Paraná system which occurred
taxa (of the 1025 D-loop positions as a mean, only 660 could 11.8–10 Ma. (Table 1, event 2). These boundary changes must
be used). have implied water (and fish) exchanges between northern
In what follows, the phylogenetic data coupled with and southern river systems. Once interrupted (about
the distribution of Hypostomus species of clades D2, D3 10 Ma), separated populations might have speciate by allo-
and D4 will be analysed and contrasted with the known patry, giving rise to the two major lineages of clade D2.

© 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 1855–1867

1862 J . I . M O N T O Y A - B U R G O S

Fig. 4 Biogeographical scheme of Hypostomus clades D3 and D4. The cladograms show only the congruent relationships found in both the
D-loop and ITS phylogenetic trees and supported by bootstrap values above 50%. In clade D3, the main boxes separate lineages
found exclusively in the Tocantins River (a major tributary of lower Amazon) and in the Paraná system. Within the latter a monophyletic
group emerges occurring in the São Francisco River. Date estimations were inferred using the D-loop and, when possible, the ITS molecular
clocks.

In order to test the reliability of the phylogenetic divi- poral dimension of the boundary displacements between
sion between the northern system (NS) and the southern the palaeo Amazon–Orinoco system and the Paraná system
system (SS) lineages of clade D2, alternative topologies (11.8–10 Ma).
corresponding to different scenarios of river basin group- Within clade D2, the well-supported lineage inhabiting
ings were tested, based on the D-loop data set (Table 3, hypo- Amazon + NE coastal rivers diversified between 5.5 Ma
theses 1, 2, 3 and 4). Of the four alternative groupings tested, (separation between H. plecostomus and the remaining
the only grouping which was not significantly different species) and 5.3 Ma (separation among H. sp. 177, H. sp.
(P < 0.05) from the groupings of the ML tree was the sister 49 + 36, and H. sp. 219 + 270), according to the D-loop
group relationship of the Paraguay River species (H. bou- molecular clock (Fig. 3). These speciation events may have
lengeri) with the northern system species (Table 2, been promoted by the contemporaneous low sea-level
hypothesis 1). This means that at least all the Paraná, Uru- period reported between 6 and 5 Ma (Table 1, event 3). Due
guay and E coastal river species form a distinct lineage to the very shallow saltwater found along the coast of
while the Paraguayan species could eventually be the northeast South America (the Amazon fan, which formed
southern system species related most closely to the during late Miocene (Damuth & Kumar 1975)), a decrease
northern system lineage of clade D2. This scenario is of the sea level may have allowed fish dispersal among
not unfounded knowing that the headwater exchanges neighbouring rivers via delta connections or coastal marshes
between the northern and southern systems involved the (as seen today in the Guianas countries and in the Amapa
upper Paraguay and southern tributaries of the Amazon region of Brazil). Once these connections were interrupted
(Lundberg et al. 1998). Nevertheless, the divergence time by the subsequent sea level, increase (5 Ma), fish popula-
between the Uruguayan species and the northern system tions would have diverged by allopatric speciation.
lineage estimated on the basis of this plausible alternative Regarding the clade D2 lineage inhabiting Paraná + E
scenario is 9.7 Ma, which is still compatible with the tem- coastal rivers (Fig. 3), species from the E coastal rivers form

© 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 1855 –1867

 H I S T O R I C A L B I O G E O G R A P H Y O F T H E C A T F I S H H Y P O S T O M U S 1863

Table 3 Evaluation of alternative hypotheses of river basin groupings

Tree topologies –ln L ∆-ln L KH-test (P) SH-test (P)

Clade D2
ML tree: (Northern Systemsa) (Southern Systemsb) 6726.96
Hypothesis 1: [(NS, Paraguay), remaining SS] 6734.41 7.44 0.082 0.506
Hypothesis 2: [(NS, Paraná), remaining SS] 6752.29 25.3 0.02* 0.048*
Hypothesis 3: [(NS, Paraguay, Paraná), remaining SS] 6751.41 24.4 0.007* 0.033*
Hypothesis 4: [remaining NS (SS, upper Amazon)] 6765.21 38.2 0.000* 0.003*
Clade D3
ML tree: [NS (SS + São Francisco)] 6726.96
Hypothesis 5: [(NS, São Francisco) SS] 6744.62 17.65 0.04*

aNorthern systems (NS) refers to the Amazon + northeastern coastal rivers.

bSouthern systems (SS) refers to the Paraná system + eastern coastal rivers.
*P < 0.05.

a monophyletic clade emerging from Paraná system spe- respectively. The hydrogeological event which probably
cies. The origin of this coastal clade can be estimated to allowed the colonization of the São Francisco River is
about 4.2 Ma using the D-loop molecular clock. The hydro- the boundary displacement between the upper Paraná–
geological event, which may be geographically linked to São Francisco basins which occurred during the Tertiary
the isolation of the E coastal river species, is the disconnec- (Table 1, event 5). This time interval is, however, large and
tion of the coastal Paraíba River from the upper Paraná, a more refined geological dating is needed to correlate
which occurred in the middle Miocene (Table 1, event 4). this boundary displacement confidently with the origin of
However, if dated correctly, this hydrogeological change the São Francisco lineage.
predates the inferred age of the formation of the E coastal Within Hypostomus clade D4, the D-loop and the ITS data
rivers lineage and therefore cannot be the responsible provide poor phylogenetic resolution (Fig. 2). Neverthe-
vicariant event. More recent headwater captures must be less, an ancient splitting between Paraná system species
invoked to explain the origin of the E coastal lineage of and the NE coastal river species (H. johnii) is found again
clade D2, the members of which are, in turn, distributed in (Fig. 4). This biogeographical scheme, although reduced, is
three coastal rivers (Ribeira, Ubatiba, and Paraíba rivers) similar to the one found in clade D2 (Fig. 3). The estimated
which may have shared tributaries about 4 Ma (Fig. 3). age of this speciation event is 11.8 Ma (based on the branch
Within clade D3, Amazon species and Paraná system lengths of the D-loop tree leading to H. johnii and H. sp. Tib
species form two distinct lineages which originated 1). This is the third compatible and independent age
10.2 or 10.1 Ma according to the D-loop and the ITS estimation of a major biogeographical event separating
molecular clocks, respectively (Fig. 4). These very narrow Amazon plus NE coastal rivers species from Paraná
estimations correspond well with the date estimations system species (Table 1, event 2).
of the Amazon–Paraná dispersal event reported for Within the Paraná system species of clade D4, a mono-
clade D2 (11.4 and 10.5 Ma for the D-loop and ITS data, phyletic group emerges comprising all Uruguay River and
respectively). The concordance of the date estimations Jacuí River representatives (Fig. 4). The origin of this Uru-
obtained independently with two Hypostomus clades guayan clade can be dated to 9.3 Ma (using the D-loop
and two genetic markers suggests strongly that the same molecular clock), and may be correlated with the discon-
major hydrogeological change (Table 1, event 2) has driven nection of the Uruguay River from the upper Paraná which
allopatric speciation on both monophyletic clades. occurred during the Miocene (Table 1, event 6).
Regarding the Paraná system lineage of clade D3, a Under the assumption of an accurate calibration of the
monophyletic assemblage emerges comprising the São D-loop and ITS molecular clocks, the historical biogeo-
Francisco River representatives (Fig. 4). The inclusion of graphical study of Hypostomus has allowed the identifica-
this São Francisco lineage within the Paraná system tion and dating of seven main cladogenetic events linked
species is supported significantly better by the D-loop probably to hydrogeological changes distributed through-
data set than the alternative hypothesis grouping the out East-Andean South America (Fig. 5). Four of these
São Francisco lineage with the Amazon lineage of clade cladogenetic events coincide geographically and tem-
D3 (Table 3, hypothesis 5; KH test P < 0.05). The origin porally with documented geomorphological or climatic-
of the São Francisco clade can be dated to 5.7 Ma and associated changes, which are suggested therefore to be
6.4 Ma using the D-loop and the ITS molecular clocks, the primary vicariant causes.

© 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 1855–1867

1864 J . I . M O N T O Y A - B U R G O S

Fig. 5 Reconstructed vicariant and disper-

sal events in eastern Andean South America
during the late Tertiary. Many of these
hypothetical events are linked to documen-
ted geological changes (see text). Date estima-
tions were inferred using the Hypostomus
D-loop and, when possible, the ITS mole-
cular clocks. Diamonds indicate probable
vicariant events, while double arrows indic-
ate dispersal events followed by allopatric
divergence.

(Weitzman et al. 1988), Paraloricaria and Brochiloricaria

Comparisons with other taxa
(Siluriformes) (Isbrücker 1980). A more recent population
To determine the extent to which the inferred or docu- splitting between Amazon and Paraná systems has,
mented hydrogeological modifications may have acted on however, been reported (between 2.3 and 4.1 Ma), based
Neotropical fish fauna diversification in general, com- on the calibrated phylogeny of the migratory Prochilodus
parisons with other unrelated fish taxa are needed. fishes (Characiformes) (Sivasundar et al. 2001). The recent
Chronologically, the first of these events concerns the basal underlying water exchanges might, however, have been
split observed in clades D2, D3, and D4, separating the of minor importance or might have affected fishes with
northern Amazon + NE coastal species from the southern particular ecology. For example, among the specious
Paraná + E coastal species, and which can be dated between family Curimatidae (Characiformes) only two species are
10 and 12 Ma (Fig. 5). Because historical biogeographical shared between the Amazon and the Paraná systems, one
studies at the South American continental-wide scale are of which is distributed largely throughout the Paraná
scarce, no similar pattern of distribution are documented system while limited to some tributaries of the Madeira
for other fish taxa. However, several biogeographical River in the Amazon basin, close to the present Amazon–
studies report fish faunas not shared between the Amazon– Paraná systems boundary (Vari 1988). The catfish species
Orinoco and Paraná–E coastal systems. Among taxa found Megalechis personata displays a similar but inverted
exclusively in Amazon plus Orinoco basins we can list the distribution pattern (Reis 1997).
Panaque dentex species group (Siluriformes) (Schaefer & The second event is the separation of the upper Paraná
Stewart 1993), the genus Nannoptopoma (Siluriformes) from the Uruguay River, about 9.3 Ma (Fig. 5). Current
(Schaefer 1996) and the genus Glyptoperichthys (Siluriformes) distribution patterns of Oligosarcus species (Characiformes)
(Page et al. 1996), while examples of genera distributed may be the outcome of such event. Indeed, O. pintoi,
exclusively in the Paraná system or Paraná plus E coastal the most basal species of this genus, is restricted to the
rivers are Glandulocauda and Mimagoniates (Characiformes) upper Paraná basin, while two derived species are

© 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 1855 –1867

 H I S T O R I C A L B I O G E O G R A P H Y O F T H E C A T F I S H H Y P O S T O M U S 1865

endemic to the Uruguay and Lagoa dos Patos systems logical time span (6.4–3.5 Ma), are congruent with the
(Menezes 1988). Aside from the probable vicariant event distribution pattern of the catfish genus Neoplecostomus.
that separated upper Paraná from the Uruguay basin, This genus comprises six species which are exclusively
it should be noted that the present lower Paraná is restricted to upper Paraná, the headwaters of the São
connected to the lower Uruguay basin via the Rio de la Francisco River and E coastal rivers (Langeani 1990).
Plata estuary. This configuration has been maintained From this general survey it seems likely that the major
almost unchanged for the last 10 Ma (Lundberg et al. 1998). geological–cladogenetic events highlighted in the course
This estuarine connection may explain more recent diver- of this study might have acted at a larger scale, shaping
gence times among Uruguayan species and their closest the distribution patterns of many other Neotropical fish
Paraná system sister clades. For example, in clade D2, the groups.
Uruguayan H. commersoni diverged from E coastal rivers
and upper Paraná species about 4.2 Ma, while in clade D3
Conclusions
the Uruguayan H. luteomaculatus separated from upper
Paraná species about 6 Ma (according to the D-loop The great diversity of the Neotropical freshwater fishes
molecular clock). In addition, several conspecific popu- was suggested to be the result of a few million years of
lations are distributed currently along both systems, evolution (Roberts 1972; Haffer 1982). On the contrary,
dispersing most probably through Rio de la Plata (e.g. palaeontological and biogeographical works suggested
Sivasundar et al. 2001). that fish diversification up to the level of modern species
The third event is the rise of the Mérida Andes, which predated the Pleistocene (Weitzman & Weitzman 1982;
separated Lago Maracaibo from the palaeo Amazon– Lundberg 1997). According to Lundberg et al. (1998), much
Orinoco 8 Ma (Lundberg et al. 1998), and which was used of the diversity of freshwater fishes might be the outcome
here for calibrating the D-loop and ITS molecular clocks of palaeohydrological changes that promoted dispersals
(Fig. 5). The rise of the Mérida Andes probably represents followed by vicariant divergences. This constitutes the
a major vicariant event for Neotropical fish fauna in hydrogeological hypothesis. The historical biogeography
general, as documented for instance in the Characiformes of Hypostomus reveals that the many documented hydro-
genera Potamorhina (Vari 1988) and Prochilodus (Sivasundar geological changes are most probably at the origin of all
et al. 2001) and in the catfish genus Hoplosternum (Reis main cladogenetic events observed in the phylogeny of
1998). this genus. The comparison to unrelated fish taxa suggests
The fourth event regards fish dispersal (followed by that the majority of these geological or climatic changes
allopatric population divergence) among Amazon and NE have probably contributed to the shaping of the tremendous
coastal rivers, probably promoted by a low sea-level ichthyological diversity observed today. Taken together,
period about 6– 5 Ma (Fig. 5). Compatible distribution these results fulfil the prediction of the hydrogeological
patterns are found in Characiformes fishes where ende- hypothesis on the origin of the diversity of Neotropical
mic species from northeastern coastal rivers of Brazil have freshwater fishes. Moreover, the inferred date estimates
their closest relatives in the Amazon basin (Vari 1988). indicate that the main geological–cladogenetic episodes
Regarding Brazilian Shield rivers, headwaters captures in the evolutionary history of Hypostomus occurred be-
occurred from the upper Paraná to the São Francisco River, tween 12 and 4 Ma. These results provide a comple-
probably about 5.7–6.4 Ma (Fig. 5). These water exchanges mentary contribution to geology for the building of a more
reflect perfectly the present distribution patterns of the detailed picture of the landscape evolution in the Neo-
genus Piabina (Characiformes), the members of which are tropics during late Tertiary. However, because the history
restricted to upper Paraná and São Francisco basins, with of Neotropical rivers seems to be so complex and because
one occurrence in the Itapicuru River near the outlet of the its ichthyofauna is still poorly defined phylogenetically
São Francisco River (Vari & Harold 1998). and geographically, a comprehensive view of the tempo
The historical biogeography of Hypostomus also suggests and mode of the Neotropical landscape evolution and
headwater captures between the upper Paraná and E its role as a major biological diversifying force remains a
coastal rivers, about 4.2 Ma (Fig. 5), and water exchanges challenging issue.
among the E coastal rivers (4.3 – 3.5 Ma; Fig. 5). A similar
biogeographical situation was observed by Weitzman et al.
Acknowledgements
(1988) working on Glandulocaudini species (Characi-
I wish to thank M. Burgos, S. Fisch-Muller, V. Mahnert, J. A.
formes). Basal species of their morphological phylogeny
Montoya-Burgos, J. Pawlowski, C. Weber, E. B. Taylor and
inhabit Paraguay and Paraná basins while more derived three anonymous reviewers for their comments and valuable dis-
species are distributed along E coastal rivers. All these cussions. I owe many thanks to J. Armbruster, F. Langeani, P.-Y.
consecutive watershed divide displacements in eastern Le Bail, H. Ortega, P. de Rham, R. E. Reis, E. Pereira, J. Pezzi da
Brazilian Shield rivers, which occurred in a narrow geo- Silva and O. Shibatta, who provided precious material. I

© 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 1855–1867

1866 J . I . M O N T O Y A - B U R G O S

acknowledge J. Fahrni for technical assistance. This project was Huelsenbeck JP, Crandall KA (1997) Phylogeny estimation and
supported by the Swiss National Science Foundation (project hypothesis testing using maximum likelihood. Annual Review of
n°3100–045885.95). Ecology and Systematics, 28, 437–466.
Isbrücker IJH (1980) Classification and catalogue of the mailed
Loricariidae (Pisces, Siluriformes). Verslagen En Technische
References Genevens, Universiteit van Amsterdam, 22, 188.
Kimura M (1980) A simple method for estimating evolutionary
Avise JC (2000) Phylogeography. The History and Formation of Species. rate of base substitutions through comparative studies of
Harvard University Press, Cambridge, MA. nucleotide sequences. Journal of Molecular Evolution, 16, 111–120.
Bermingham E, Martin AP (1998) Comparative mtDNA phylo- Kishino H, Hasegawa M (1989) Evaluation of the maximum like-
geography of neotropical freshwater fishes: testing shared history lihood estimate of the evolutionary tree topologies from DNA
to infer the evolutionary landscape of lower Central America. sequence data, and the branching order in Hominoidea. Journal
Molecular Ecology, 7, 499 – 517. of Molecular Evolution, 29, 170–179.
Beurlen K (1970) Geologie Von Brasilien. Gebrüder Borntraeger, Langeani F (1990) Revisão do género Neoplecostomus Eigenmann
Berlin/Stuttgart. and Eigenmann, 1888, com a descriçao de quatro novas espécies
Damuth JE, Kumar N (1975) Amazon Cone: morphology, sedi- do sudeste Brasileiro (Ostariophysi, Siluriformes, Loricariidae).
ments, age, and growth patterns. Geological Society of America Comunicação Do Museo de Ciências Da PUCRS, Série Zoológica,
Bulletin, 86, 863–878. Porto Alegre, 3, 3–31.
Diaz de Gamero ML (1996) The changing course of the Orinoco Larsen N, Olsen GJ, Maidak BL et al. (1993) The ribosomal
River during the Neogene: a review. Palaeogeography, Palaeo- database project. Nucleic Acids Research, 21, 3021–3023.
climatology, Palaeoecology, 123, 385 – 402. Lovejoy NR, Bermingham E, Martin AP (1998) Marine incursion
Donaldson KA, Wilson RR (1999) Amphi-panamic geminates of into South America. Nature, 396, 421–422.
snook (Percoidei: Centropomidae) provide a calibration of the Lovejoy NR, De Araújo MLG (2000) Molecular systematics, bio-
divergence rate in the mitochondrial DNA control region of geography and population structure of Neotropical freshwater
fishes. Molecular Phylogenetics and Evolution, 13, 208 – 213. needlefishes of the genus Potamorrhaphis. Molecular Ecology, 9,
Edwards SV, Beerli P (2000) Perspective: gene divergence, 259–268.
population divergence, and the variance in coalescence time in Lundberg JG (1993) African–South American freshwater fish
phylogeographic studies. Evolution, 54, 1839 – 1854. clades and continental drift: problems with a paradigm. In:
Felsenstein J (1981) Evolutionary trees from DNA sequences: a Biological Relationships Between Africa and South America (ed.
maximum likelihood approach. Journal of Molecular Evolution, Goldblatt P), pp. 156–199. Yale University Press, New Haven,
17, 368–376. CT.
Fitch WM (1971) Toward defining the course of evolution: mini- Lundberg JG (1997) Fishes of the La Venta fauna: additional
mum change for a specific tree topology. Systematic Zoology, 20, taxa, biotic and paleoenvironmental implications. In: Vertebrate
406–416. Paleontology in the Neotropics: the Miocene Fauna of la Venta Colom-
Glor RE, Vitt LJ, Larson A (2001) A molecular phylogenetic ana- bia (eds Kay RF, Madden RH, Cifelli RL, Flynn JJ), pp. 67 – 91.
lysisi of diversification in Amazonian Anolis lizards. Molecular Smithsonian Institution Press, Washington, DC.
Ecology, 10, 2661– 2668. Lundberg JG (1998) The temporal context of the diversification of
Gu X, Fu Y-X, Li W-H (1995) Maximum likelihood estimation of Neotropical fishes. In: Phylogeny and Classification of Neotropical
the heterogeneity of substitution rate among nucleotide sites. Fishes (eds Malabarba LR, Reis RE, Vari RP, Lucena ZMS,
Molecular Biology and Evolution, 12, 546 – 557. Lucena CAS), pp. 49–68. Edipucrs, Porto Alegre, Brazil.
Haffer J (1982) General aspects of the refuge theory. In: Biological Lundberg JG, Marshall LG, Guerrero J et al. (1998) The stage for
Diversification in the Tropics (ed. Prance GT), pp. 6 – 24. Columbia Neotropical fish diversification: a history of Tropical South
University Press, New York. American rivers. In: Phylogeny and Classification of Neotropical
Haffer J (1997) Alternative models of vertebrate speciation in Fishes (eds Malabarba LR, Reis RE, Vari RP, Lucena ZMS,
Amazonia: an overview. Biodiversity and Conservation, 6, 451– Lucena CAS), pp. 13–48. Edipucrs, Porto Alegre, Brazil.
476. Martin AP, Bermingham E (2000) Regional endemism and cryptic
Haq BU, Hardenbol J, Vail PR (1987) Chronology of fluctuating species revealed by molecular and morphological analysis of a
sea levels since the Triassic. Science, 235, 1156 – 1167. widespread species of Neotropical catfish. Proceedings of the
Hasegawa M, Kishino H, Yano T (1985) Dating the human–ape Royal Society of London, B Biological Sciences, 267, 1135–1141.
splitting by a molecular clock of mitochondrial DNA. Journal of Menezes NA (1988) Implications of the distribution patterns of the
Molecular Evolution, 22, 672 – 677. species of Oligosarcus (Teleostei, Characidae) from Central and
Hoorn C (1993) Marine incursions and the influence of Andean Southern South America. In: Proceedings of a Workshop on Neo-
tectonics on the Miocene depositional history of northwestern tropical Distribution Patterns (eds Vanzolini PE, Ronald Heyer
Amazonia: results of a palynostratigraphic study. Palaeogeo- W), pp. 295–304. Academia Brasileira de Ciências, Rio de
graphy, Palaeoclimatology, Palaeoecology, 105, 267 – 309. Janeiro.
Hoorn C, Guerrero J, Sarmiento GA, Lorente MA (1995) Andean Montoya-Burgos JI, Muller S, Weber C, Pawlowski J (1997) Phylo-
tectonics as a cause for changing drainage patterns in Miocene genetic relationships between Hypostominae and Ancistrinae
northern South America. Geology, 23, 237 – 240. (Siluroidei: Loricariidae): first results from mitochondrial 12S
Hrbek T, Larson A (1999) The evolution of diapause in the killifish and 16S rRNA gene sequences. Revue Suisse de Zoologie, 104,
family Rivulidae (Atherinomorpha, Cyprinodontiformes): a mole- 185–198.
cular phylogenetic and biogeographic perpective. Evolution, 53, Montoya-Burgos JI, Muller S, Weber C, Pawlowski J (1998) Phylo-
1200–1216. genetic relationships of the Loricariidae (Siluriformes) based on

© 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 1855 –1867

 H I S T O R I C A L B I O G E O G R A P H Y O F T H E C A T F I S H H Y P O S T O M U S 1867

mitochondrial rRNA gene sequences. In: Phylogeny and Classi- Vail PR, Hardenbol J (1979) Sea-level changes during the Tertiary.
fication of Neotropical Fishes (eds Malabarba LR, Reis RE, Vari RP, Oceanus, 22, 71–80.
Lucena ZMS, Lucena CAS), pp. 363 –374. Edipucrs, Porto Vari RP (1988) The Curimatidae, a lowland Noetropical fish
Alegre, Brazil. family (Pisces: Characiformes); distribution, endemism, and
Montoya-Burgos JI, Weber C, Le Bail P-Y (2002) Phylogenetic phylogenetic biogeography. In: Proceedings of a Workshop on Neo-
relationships within Hypostomus (Siluriformes: Loricariidae) tropical Distribution Patterns (eds Heyer WR, Vanzolini PE),
and related genera based on mitochondrial D-loop sequences. pp. 343–377. Academia Brasileira de Ciências, Rio de Janeiro.
Revue Suisse de Zoologie, 109, 369 – 382. Vari RP, Harold AS (1998) The genus Creagrutus (Teleostei:
Page LM, Armbruster JW, Sabaj MH (1996) Redescription of Glyp- Characiformes: Characidae): monophyly, relationships, and
toperichthys scrophus, a loricariid catfish from Peru. Ichthyological undetected diversity. In: Phylogeny and Classification of Neotro-
Exploration of Freshwaters, 7, 185 – 191. pical Fishes (eds Malabarba LR, Reis RE, Vari RP, Lucena ZMS,
Reis RE (1997) Revision of the Neotropical catfish genus Hoplo- Lucena CAS), pp. 245–260. Edipucrs, Porto Alegre, Brazil.
sternum (Ostariophysi: Siluriformes: Callichthyidae), with the Vari RP, Weitzman SH (1990) A review of the phylogenetic bio-
description of two new genera and three new species. geography of the freshwater fishes of South America. In:
Ichthyological Exploration of Freshwaters, 7, 299 – 326. Vertebrates in the Tropics (eds Peters G, Hutterer R), pp. 381–
Reis RE (1998) Systematics, biogeography, and the fossil record of 394. Alexander Koenig Zoological Research Institute and Zoo-
the Callichthyidae: a review of the vailable data. In: Phylogeny logical Museum, Bonn.
and Classification of Neotropical Fishes (eds Malabarba LR, Reis Weber C, Montoya-Burgos JI (2002) Hypostomus fonchii sp. n.
RE, Vari RP, Lucena ZMS, Lucena CAS), pp. 351 – 362. Edipucrs, (Siluriformes: Loricariidae) from Peru, a key species suggesting
Porto Alegre, Brazil. the synonymy of Cochliodon with Hypostomus. Revue Suisse de
Riccomini C, Coimbra AM, Suguio K, Mihaly P, Maturana EC Zoologie, 109, 355–368.
(1990) Nova unidade litoestratigráfica cenozóica da Bacia Weitzman SH, Menezes NA, Weitzman MJ (1988) Phylogenetic
de Taubaté, SP: Formação Pindamonhagaba. Boltim IGUSP, biogeography of the Glandulocaudini (Teleostei: Characi-
Publicação Especial, 9, 141 – 149. formes, Characidae) with comments on the distributions of
Riccomini C, Suguio K, Alvarenga H, Fittipaldi F (1991) Estrati- other freshwater fishes in Eastern and Southeastern Brazil. In:
grafia e ambientes de Sedimentação da porção central da Bacia Proceedings of a Workshop on Neotropical Distribution Patterns (eds
de Taubaté. Congresso Brasileiro de Paleontologia, 12, Roteiro de Heyer WR, Vanzolini PE), pp. 379–427. Academia Brasileira de
excursões, 12, 1–29. Ciências, Rio de Janeiro.
Roberts TR (1972) Ecology of fishes in the Amazon and Congo Weitzman SH, Weitzman MJ (1982) Biogeography and evolu-
basins. Bulletin of the Museum of Comparative Zoology, 143, 117– tionary diversification in Neotropical freshwater fishes, with
142. comments on the refuge theory. In: Biological Diversification in
Robinson-Rechavi M, Huchon D (2000) RRTree: relative-rate the Tropics (ed. Prance GT), pp. 403–422. Columbia University
tests between groups of sequences on a phylogenetic tree. Press, New York.
Bioinformatics, 16, 296 – 297. Wheeler CW, Ahron P (1991) Mid-oceanic carbonate platforms as
Saitou N, Nei M (1987) The neighbor-joining method: a new oceanic dipsticks: examples from the Pacific. Coral Reefs, 10,
method for reconstructing phylogenetic trees. Molecular Biology 101–114.
and Evolution, 4, 406 – 425. Yang Z (1993) Maximum-likelihood estimation of phylogeny
Schaefer SA (1996) Nannoptopoma, a new genus of loricariid from DNA sequences when substitution rates differ over sites.
catfishes (Siluriformes: Loricariidae) from the Amazon and Molecular Biology and Evolution, 10, 1396–1401.
Orinoco River basins. Copeia, 4, 913 – 926. Yang Z (1994) Estimating the pattern of nucleotide substitution.
Schaefer SA (1998) Conflict and resolution: impact of new taxa on Journal of Molecular Evolution, 39, 105–111.
phylogenetic studies of the neotropical cascudinos (Siluroidei: Yang Z, Goldman N, Friday A (1994) Comparison of models
Loricariidae). In: Phylogeny and Classification of Neotropical Fishes for nucleotide substitution used in maximum-likelihood
(eds Malabarba LR, Reis RE, Vari RP, Lucena ZMS, Lucena phylogenetic estimation. Molecular Biology and Evolution, 11,
CAS), pp. 375–400. Edipucrs, Porto Alegre, Brazil. 316–324.
Schaefer SA, Stewart DJ (1993) Systematics of the Panaque dentex
species group (Siluriformes: Loricariidae), wood-eating armored
catfishes from tropical South America. Ichthyological Exploration Juan I. Montoya-Burgos began his work on systematics and
of Freshwaters, 4, 309 – 342. biogeography of neotropical catfishes during his PhD work held
Shimodaira H, Hasegawa M (1999) Multiple comparisons of at the University and the Museum of Natural History of Geneva.
log-likelihoods with applications to phylogenetic inference. Thereafter, the author collaborated in researches on the processes,
Molecular Biology and Evolution, 16, 1114 – 1116. genes and functions that govern the genetic isolation between
Sivasundar A, Bermingham E, Ortí G (2001) Population structure species, conducted at the University of Montpellier II (France).
and biogeography of migratory freshwater fishes (Prochilodus: This stay was founded by a Postdoctoral Fellowship of the French
Characiformes) in major South American rivers. Molecular Ministry of Scientific Research. He then moved back to the
Ecology, 10, 407–417. University of Geneva to take up a position of professor assistant.
Swofford DL (2000) PAUP*. Phylogenetic Analysis Using Parsimony He is also a member of the staff of the department of Ichthyology
(*and Other Methods), Version 4. Sinauer Associates, Sunderland, of the Natural History Museum, where researches are conducted
MA. on the historical biogeography of Neotropical fishes with the aim
Swofford DL, Olsen GJ, Waddell PJ, Hillis DM (1996) Phylogenetic of understanding the evolutionary processes that have driven the
inference. In: Molecular Systematics (eds Hillis DM, Moritz C, extraordinary ichthyological diversification in the Neotropics.
Mable BK), 2nd edn, pp. 407 – 514. Sinauer, Sunderland, MA.

© 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 1855–1867