molecular phylogeny of thorny catfishes (siluriformes: doradidae)

Molecular Phylogenetics and Evolution 67 (2013) 560–577

Contents lists available at SciVerse ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier .com/locate /ympev

Molecular phylogeny of thorny catfishes (Siluriformes: Doradidae)

Mariangeles Arce H. a,b,⇑, Roberto E. Reis a, Anthony J. Geneva c,d, Mark H. Sabaj Pérez b

a Laboratório de Sistemática de Vertebrados, Pontifícia Universidade Católica do Rio Grande do Sul, Caixa Postal 1429, 90619-900 Porto Alegre, RS, Brazilb Department of Ichthyology, The Academy of Natural Sciences, 1900 Benjamin Franklin Parkway, Philadelphia, PA 19103, USAc Laboratory for Molecular Systematics and Ecology, The Academy of Natural Sciences, 1900 Benjamin Franklin Parkway, Philadelphia, PA 19103, USAd Department of Biology, University of Rochester, Rochester, NY 14627, USA

a r t i c l e i n f o

Article history:Received 10 August 2012Revised 28 January 2013Accepted 17 February 2013Available online 4 March 2013

Keywords:Neotropical fishesSystematicsBiogeographyrag116SCO1

1055-7903/$ - see front matter � 2013 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.ympev.2013.02.021

⇑ Corresponding author. Current address: DeparAcademy of Natural Sciences, 1900 Benjamin Frankl19103, USA. Fax: +1 215 405 5080.

E-mail addresses: [email protected](R.E. Reis), [email protected] (A.J. GenSabaj Pérez).

a b s t r a c t

Doradidae is a monophyletic catfish family endemic to continental South America, and composed of 93valid species here placed in 31 genera. Existing phylogenetic hypotheses for Doradidae are derived fromcomprehensive analyses of morphological data, and a single molecular-based study on a limited subset oftaxa. To provide a robust molecular phylogeny commensurate with those based on morphology, we gath-ered original and published sequence data for 86 species-level taxa (at least 70 valid species plus 16 newor questionably nominal species) and all genera of Doradidae, as well as 10 species (nine genera) ofAuchenipteridae and three species and genera of Aspredinidae as outgroups. 3011 base pairs werealigned for two mitochondrial genes (cytochrome c oxidase subunit 1, and 16S ribosomal RNA) and onenuclear gene (recombination activating gene 1), and analyzed for a total of 143 specimens (130 doradids,10 auchenipterids and three aspredinids). Tree topologies generated by Maximum Parsimony, MaximumLikelihood, and Bayesian analyses were largely congruent, and are compared to existing phylogeniesbased on morphology and molecules. Although many of the relationships supported by our molecularanalyses corroborated those based on morphology, others are newly hypothesized or remain in conflict.

The monotypic Wertheimeria, Franciscodoras and Kalyptodoras, for example, form a newly proposedclade, and the subfamily Astrodoradinae is placed at the base of the doradid tree. The monotypic Doraopsand Centrochir, endemic to Caribbean drainages north and west of the Andes, are sister to Pterodoras andPlatydoras, respectively, two genera that are widely distributed in Atlantic drainages. Additional biogeo-graphic implications are discussed for hypothesized relationships among doradids. Molecular evidencestrongly supports synonymization of monotypic Merodoras with Amblydoras, and transfer of Amblydorasbolivarensis to genus Scorpiodoras. Furthermore, we consider Opsodoras ternetzi to be more properlyplaced in the genus Nemadoras. The genus Opsodoras may warrant synonymization with Hemidoras,and the monophyly of genus Ossancora is not supported; however, we refrain from taxonomic decisionsregarding those taxa until a broader spectrum of doradids can be submitted to further morphological andmolecular phylogenetic analyses.

� 2013 Elsevier Inc. All rights reserved.

1. Introduction the posterior nuchal plate to the tip of the first rib which is sup-

Doradidae is a monophyletic catfish family endemic to conti-nental South America that is often recognized by having a conspic-uous midlateral row of bony scutes, each usually bearing a medialthorn directed posteriorly. Doradidae is promptly diagnosed by atrue synapomorphy that is unique among catfishes: the infranu-chal scute, a superficial, vertically oriented plate-like bone co-formed by fusion between an expanded ossicle of the lateral-linecanal and an ossified infranuchal ligament extending from beneath

ll rights reserved.

tment of Ichthyology, Thein Parkway, Philadelphia, PA

(M. Arce H.), [email protected]), [email protected] (M.H.

ported by the sixth vertebra (Ferraris, 1988; Higuchi, 1992; Birind-elli, 2010; Birindelli and Sabaj Pérez, 2011).

Doradids occur in all of the major river basins of South America:Magdalena, Maracaibo, Orinoco, Essequibo, Amazonas, Tocantins,São Francisco and Paraná-Paraguay (La Plata). They also inhabitAtlantic coastal drainages from the Orinoco Delta to Parnaíba Riveron the northeast coast of Brazil, and Paraguaçu and Jequitinhonharivers on the east-central coast of Brazil. The family is not knownfrom Pacific Coast drainages; nor have they been reported fromcoastal drainages west of the Magdalena River, Colombia, or southof La Plata River, Argentina (Sabaj, 2002). The large-bodied Lithodo-ras dorsalis is an important subsistence and commercial fish in theAmazon estuary, especially in small local markets (Goulding et al.,1996). Most of the other large doradids (e.g., Megalodoras, Pterodo-ras, Oxydoras) that are caught and sold for consumption, however,

http://dx.doi.org/10.1016/j.ympev.2013.02.021

mailto:[email protected]




http://dx.doi.org/10.1016/j.ympev.2013.02.021

http://www.sciencedirect.com/science/journal/10557903

http://www.elsevier.com/locate/ympev

Table 1Primers for amplification and sequencing of the DNA fragments.

Gene Primer name Primer Reference

rag1(50) F74 TTT CGG AAT GGA AGT TTA AGC TsT TTC G Sullivan et al. (2006)rag1(50) R1333 GTC AAA CAC ACA GAC TTC ACA TC Sullivan et al. (2006)rag1(50) F354 CAG AGC ATG AGG TvC AGG GAG ATC T Sullivan et al. (2006)rag1(50) R798 TGA GTT ATA TTC TTT ACC CA Sullivan et al. (2006)rag1(50) F89Ia TTT AAG CTG TTT CGA GTT CGT TCA TTG G Sullivan et al. (2006)CO1 LCO1490 GGT CAA CAA ATC ATA AAG ATA TTG G Folmer et al. (1994)CO1 HCO2198 TAA ACT TCA GGG TGA CCA AAA AAT CA Folmer et al. (1994)16S 16S BR CCG GTC TGA ACT CAG ATC ACG T Palumbi et al. (1991)16S 16sSAR CGC CTG TTT ATC AAA AAC AT Palumbi et al. (1991)

a Alternative amplification primer for the first forward fragment of rag1.

Table 2Support values for each genus recognized in this study, represented by more than one species.

Genus Bremer support Bootstrap Posterior probability Nucleotide synapomorphies

Acanthodoras 52 100 1 61Agamyxis 35 100 1 38Amblydorasa 4 77 1 6Anadoras 25 100 1 25Astrodoras 6 94 1 26Centrodoras 6 95 1 7Doras 14 100 1 15Hassar 21 100 1 25Hemidoras + Opsodoras morei 10 98 1 13Leptodoras 2 95 1 6Megalodoras 2 99 0.96 4Nemadorasb 1 50 0.97 4Oxydoras 22 100 1 29Physopyxis 48 100 1 53Platydoras 8 100 1 12Pterodoras 16 100 1 21Rhinodoras 5 82 1 9Scorpiodorasc 14 99 1 21Trachydoras 16 100 1 24

a Inclusive of Amblydoras nheco, n. comb.b Inclusive of Nemadoras ternetzi, n. comb.c Inclusive of Scorpiodoras bolivarensis, n. comb.

M. Arce H. et al. / Molecular Phylogenetics and Evolution 67 (2013) 560–577 561

are of moderate commercial value (Novoa and Ramos, 1978). A fewof the small and attractive species (in Acanthodoras, Agamyxis,Amblydoras, Platydoras) are routinely exported as ornamentals(Sabaj, 2002).

The past eight years have witnessed a renaissance in doradidtaxonomy and systematics with the addition of 19 newly describedspecies (including one fossil), two new genera, one new subfamily,and the validation of two nominal species previously held in syn-onymy, raising the total diversity of the family to 93 species in32 genera (see Birindelli and Sabaj Pérez, 2011). Higuchi (1992; re-sults published in de Pinna, 1998) used morphology to completethe first phylogenetic analysis of relationships within the family.Subsequent morphological studies, though unpublished, have fur-ther developed our understanding of relationships within the fam-ily down to the species level for most groups (Sabaj, 2002;Birindelli, 2006, 2010; Sousa, 2010). Only one study (Moyer et al.,2004) has used molecular sequence data to investigate doradidrelationships. Moyer et al. (2004) hypothesized relationships for43 doradid species representative of 23 genera using complete se-quences (2.5 kilobases, kb) of mitochondrial 12S and 16S rRNAgenes and partial (1.3 kb) sequences of the nuclear elongation factor1 alpha gene. Some of the relationships proposed in the molecularphylogeny by Moyer et al. (2004) are in significant conflict withthose based on morphology. Their molecular analyses, however,did not include key doradid taxa representative of putatively deep,independent lineages.

The taxonomic coverage of the current molecular study is com-mensurate with the sum of those based on morphology. Using

Maximum Parsimony, Maximum Likelihood, and Bayesian Infer-ence, we hypothesize relationships among 70 valid species and16 species-level taxa representative of all genera of Doradidaebased on sequence data (3011 aligned bases) for one nuclear(recombination activating gene 1) and two mitochondrial genes(cytochrome c oxidase subunit 1 and 16S ribosomal RNA).

2. Materials and methods

2.1. Taxon sampling

Sequence data were gathered for one nuclear gene, recombina-tion activating gene 1 (rag 1) and two mitochondrial genes, cyto-chrome c oxidase subunit 1 (CO1), and 16S ribosomal RNA (16S), ina total of 130 specimens representing all genera, at least 70 of 93valid species, and 16 additional species-level taxa (eight unde-scribed plus eight uncertain as nominal or new) of Doradidae(Appendix A). Thirty of the 86 species-level taxa are representedby more than one specimen. To date, this study represents themost comprehensive sampling of Doradidae for molecular phylo-genetic analysis.

Following previous molecular and morphological studies, weincluded ten species (nine genera) of Auchenipteridae and threespecies (three genera) of Aspredinidae as outgroups. Analysis ofnuclear markers rag1 and rag2 (Sullivan et al., 2006) strongly sup-port Auchenipteridae and Aspredinidae, respectively, as successiveoutgroups to Doradidae. In a comprehensive cladistic analysis

Fig. 1. Phylogeny of Doradidae inferred from Maximum Parsimony analysis of rag1, 16S and CO1 sequences. Strict consensus of 36 trees, each with 6443 steps. Node valuescorrespond to Bremer support.

562 M. Arce H. et al. / Molecular Phylogenetics and Evolution 67 (2013) 560–577

based on morphology, Birindelli (2010) likewise supported Auche-nipteridae as the sister group to Doradidae, together comprisingthe superfamily Doradoidea. Morphological data (e.g., Friel, 1994;

Birindelli, 2010), however, support a sister-group relationship be-tween Doradoidea and the African family Mochokidae, togethercomprising the suborder Doradoidei. Friel’s (1994) study

Fig. 2. Phylogeny of Doradidae inferred from Maximum Likelihood analysis of rag1, 16S, and CO1 sequences. Node values correspond to bootstrap.


supported Aspredinidae as the sister group to Doradoidei, whereasBirindelli’s (2010) study placed Doradoidei sister to the clade

(Amphiliidae (Sisoridae (Aspredinidae + Erethistidae))). Molecularstudies consistently fail to support a close relationship between

Fig. 3. Phylogeny of Doradidae inferred from Bayesian analysis of rag1, 16S, and CO1 sequences. Node values correspond to posterior probability.


Mochokidae and Neotropical catfishes in Doradoidea (i.e., themonophyly of Doradoidei). In fact, Sullivan et al. (2006) placed

Mochokidae sister to Malapteruridae + Amphiliidae in a clade com-posed exclusively of African catfish families (i.e., ‘‘Big Africa’’). For


this reason, the family Mochokidae was not included in the presentmolecular analysis.

2.2. Marker selection

Recombination activating gene 1 (rag1) is a single copy nucleargene that is present in all jawed vertebrates and encodes compo-nents of lymphocyte specific enzymes involved in the initial stepsof V(D)J recombination of T cell receptor and immunoglobulingenes (Oettinger et al., 1990; Schatz et al., 1989). Rag1 gene nucle-otide sequences have been used to reconstruct phylogenetic rela-tionships within vertebrates, including the most comprehensivestudy to date of catfishes (Sullivan et al., 2006). In actinopterygianfishes, one intron is present near the middle of the coding se-quence; in teleost fishes, a second intron is present near the 50-end of the gene (introns ‘‘1b’’ and ‘‘1a’’, respectively, in the termi-nology of Venkatesh et al., 1999). In the zebrafish, Danio rerio(Cyprinidae), rag1 consists of 1057 codons: intron 1a is locatedafter codon 102 and intron 1b is located after codon 475 (Willetet al., 1997). In catfishes, intron 1a in the 50 rag1 fragment rangesin size from 97 bp in Nematogenys inermis to 421 bp in Pimelodusornatus (Sullivan et al., 2006).

Cytochrome c oxidase subunit 1 (CO1) exhibits a greater range ofphylogenetic signal than any other mitochondrial gene (Hebertet al., 2003). As in other protein-coding genes, the third-positionnucleotides of CO1 show a high incidence of base substitution,leading to a rate of molecular evolution that is about three timesgreater than that of 12S or 16S rRNA (Knowlton and Weigt,1998). 16S rRNA (16S) is a mitochondrial gene that codes for thesmall subunit of the ribosome, and has been widely used in spe-cies-level phylogenies due to its relatively fast rate of evolution.

2.3. DNA amplification and sequencing

Tissues (fin clips and/or muscle) were sampled in the field byMAH and MSP, and preserved in 95–100% ethanol; voucher speci-mens were fixed in 10% buffered formalin and transferred to 70–75% ethanol for long-term museum storage. Additional tissue sam-ples were obtained from colleagues (see Acknowledgments).

Total DNA was extracted using the Qiagen DNeasy blood andtissue kit. PCR was carried out in 20 ll reactions; primers foramplification and sequencing are listed in Table 1. For CO1 and16S, PCR reaction mixture consisted of 10 ll of Apex Taq DNA Poly-merase Master Mix, 1.5 mM MgCl2 (Genesee Scientific), 0.5 lM offorward and reverse primer, 5–8 ll of distilled water and 1–4 llof DNA template. Cycles of amplification were programmedaccordingly: 4 min at 95 �C (initial denaturation), 10 cycles of threesteps, 30 s at 50 �C or 55 �C (annealing, temperature decreased by1 �C after each cycle), 2 min at 72 �C (extension) and 1 min at95 �C (denaturation); 30 cycles of three steps, 1 min at 95 �C, 30 sat 40 �C or 44 �C, and 2 min at 72 �C; final extension step for10 min at 72 �C. Amplification of rag1 followed the protocol of Sul-livan et al. (2006): 4 min at 95 �C (initial denaturation), 35 cycles ofthree steps, 30 s at either 50 �C, 55 �C or 59 �C, 2 min at 72 �C, and30 s at 95 �C; final extension step for 4 min at 72 �C.

2.4. Sequence editing and alignment

Sequences were edited and combined into contigs for each frag-ment using Sequencher 4.7 (GeneCodes Corporation, Ann Arbor,MI). Sequences for each gene were aligned using MUSCLE 3.7 (Ed-gar, 2004) with default parameters. The alignments were refinedmanually in MacClade 4.0 (Maddison and Maddison, 2000). Se-quences for all three fragments were concatenated using Mesquite2.74 (Maddison and Maddison, 2011).

2.5. Phylogenetic analyses

We analyzed combined nuclear and mitochondrial data usingMaximum Parsimony (MP), Maximum Likelihood (ML), and Bayes-ian Inference. For MP analysis the trees were generated using the‘‘new technologies search’’ implemented in TNT (Goloboff et al.,2008). The search was performed in two steps: the first step useda combination of sectorial searches (RSS and CSS), 100 iterations ofratchet, 100 cycles of tree fusing, and 100 rounds of Drift. The dri-ven was set to reach the minimum length 50 times. The secondstep used the trees produced in the first step to perform a tradi-tional TBR search. We treated gaps as missing data, and we consid-ered all characters to have equal weights. For nodal support, wecalculated absolute Bremer (Bremer, 1994). All of the support val-ues were plotted on the strict consensus tree.

For ML and Bayesian analyses, the concatenated data matrixwas divided into eight partitions: one for 16S, one for each nucle-otide position per codon of CO1, one for each nucleotide positionper codon of rag1, and one for the rag1 intron. Maximum Likeli-hood (ML) analyses were conducted using RAxML7.2.8 (Stamata-kis, 2006) based on 1000 replicates under the GTR + GAMMAmodel. Nodal support was evaluated with 1000 non-parametricbootstrap pseudoreplicates. Bayesian analyses were conductedusing MrBayes 3.1.2 (Huelsenbeck and Ronquist, 2001; Ronquistand Huelsenbeck, 2003) and the same model GTR + GAMMA. Weran three heated chains and one cold chain from 60 million gener-ations, sampling every 10,000th generation. To ensure samplingfrom a stationary posterior distribution, we examined all analysesfor convergence in two ways. First, by the average standard devia-tion of split frequencies between two MCMC analyses run indepen-dently, with levels below 0.01 being considered indicative ofconvergence. Second, via direct visualization of split frequenciesthroughout the course of each analysis using the cumulative plot-ting feature of the on-line application AWTY (Wilgenbusch et al.,2004) with convergence diagnosed as the point after which splitfrequencies for the 20 most variable nodes achieve stability. Bothmetrics reported convergence around 10–20 million generations,so we discarded the first 30 million generations as burn-in.

3. Results

3.1. Data set characteristics

The rag1 dataset consisted of 1807 bp (566 parsimony-informa-tive) for 131 terminals (i.e., specimens). The 16S dataset consistedof 578 bp (175 parsimony-informative) for 186 terminals. The CO1dataset consisted of 626 bp (254 parsimony-informative) for 144terminals. Analyses were performed on the combined dataset withterminals restricted to those represented by at least two loci(n = 143; Appendix A). In the combined dataset, rag1 is representedby 127 terminals, and 16S and CO1 are both represented by 137terminals. The combined dataset included 3011 bp of which 979were parsimony informative.

3.2. Phylogenetic results

The MP analysis of the combined dataset for restricted termi-nals (n = 143) resulted in 36 most parsimonious trees, each with6443 steps (Fig. 1). Topologies resulting from the ML (Fig. 2) andBayesian (Fig. 3) analyses were largely congruent as expected(see Rindal and Brower, 2011). Discrepancies are associated withnodes that are poorly supported in all three analyses. Examples in-clude the monophyly of Doradidae (supported by ML and Bayesiananalyses only), the sister-group relationship between Ossancorapunctata and the Doras Clade (supported by ML and Bayesian anal-


yses only), the sister-group relationship between the Lithodorasand Platydoras Clades (supported by Bayesian analysis only), andthe placement of Leptodoras hasemani (sister to remaining Leptodo-ras in ML and Bayesian analyses vs. nested within Leptodoras in MPanalysis).

Maximum Parsimony Bremer support (MPBS), Maximum Likeli-hood Bootstrap (MLBS), and Bayesian posterior probability (BPP),are reported for all nodes on their respective topologies. Discussionis limited to clades supported by all three analyses, referred to asweakly supported (absolute Bremer support values 1–3), moder-ately supported (values 4–7), well supported (values 8–19), verywell supported (values 20–39) and extremely well supported (val-ues > 40). Topologies including all specimens analyzed in thisstudy are presented in Appendices B–D.

3.3. Doradid genera and clades

All three analyses (MP, ML and Bayesian) supported the mono-phyly of 18 of the 20 valid genera recognized herein and repre-sented by more than one species-level taxon (see Table 2 forsupport values and nucleotide synapomorphies). The monotypicMerodoras nheco nested within species of Amblydoras, includingthe type species A. affinis. We thereby synonymize Merodoras withAmblydoras (see Section 4.1.2), and recognize Amblydoras nheco, n.comb. Opsodoras ternetzi nested within Nemadoras, a genus repre-sented in our analyses by all valid species (n = 5) plus one unde-scribed. Therefore, we recognize Nemadoras ternetzi, n. comb. Thesister relationship between Amblydoras bolivarensis and Scorpiodo-ras heckelii is well supported (MPBS = 14), and S. heckelii is the typespecies of Scorpiodoras. Therefore, we recognize Scorpiodoras boli-varensis, n. comb.

The monophyly of Ossancora is not supported by our results (seeSection 4.2.4); but, we refrain from taxonomic decisions regardingthis genus. The genus Opsodoras, represented by a single species O.morei, is sister to Hemidoras morrisi, rendering Hemidoras paraphy-letic. The type species of the genus Opsodoras (O. orthacanthus, syn-onym of Opsodoras stuebelii) was not included in our analyses;therefore, we refrain from synonymizing Opsodoras with the oldername Hemidoras. The remaining genera in our analyses are eitherrepresented by a single species (Rhynchodoras) or are monotypic.Well-supported clades (i.e., Bremer support P 8) including morethan one genus are briefly described as follows.

3.3.1. Subfamily Astrodoradinae (sensu Higuchi et al., 2007)A clade composed of six genera: Amblydoras, Anadoras, Astrodo-

ras, Hypodoras, Physopyxis, and Scorpiodoras. Amblydoras is repre-sented by five specimens of five species-level taxa consideredherein (status as nominal or undescribed uncertain for three taxa):Amblydoras affinis (n = 1), Amblydoras nheco, n. comb. (1), andAmblydoras spp. 1–3 (1, 1, 1). Anadoras is represented by the twovalid species, A. grypus (1), and A. weddellii (1) and one specimenof an undescribed species (sensu Sousa, 2010). Astrodoras is cur-rently monotypic, but there appears to be one or two undescribedspecies (MHSP pers. obs.); the two specimens represented in theanalysis are tentatively considered Astrodoras asterifrons and Astro-doras sp. (sensu Roa-Fuentes et al., 2010). Hypodoras is monotypicand represented by one individual. Physopyxis includes two speci-mens representing two, P. ananas and P. lyra, of three valid species(Sousa and Rapp Py-Daniel, 2005). Scorpiodoras includes threespecimens representing two, S. bolivarensis, n. comb., and S. heck-elii, of four valid species considered herein.

The monophyly of Astrodoradinae (sensu Higuchi et al., 2007) isvery well supported in all three analyses: MPBS = 30, BPP = 1, andMLBS = 100; nucleotide synapomorphies = 36. Within Astrodoradi-nae, the clade composed of Amblydoras affinis, A. nheco and Amblydo-ras sp. 3 is also very well supported: MPBS = 20, BPP = 1, and

MLBS = 100; nucleotide synapomorphies = 25. The sister group rela-tionship of Hypodoras and Astrodoras is well supported in all threeanalyses: MPBS = 13, BPP = 1, and MLBS = 100; nucleotide synapo-morphies = 20. Also well supported is the sister group relationshipof Scorpiodoras bolivarensis and S. heckelii: MPBS = 14, BPP = 1, andMLBS = 99; nucleotide synapomorphies = 21. The monophyly ofastrodoradin taxa minus Anadoras is moderately supported:MPBS = 7, BPP = 1, and MLBS = 98; nucleotide synapomorphies = 17.

3.3.2. Clade 1A clade composed of all doradids except subfamily Astrodoradinae

and Acanthodoras is well supported in all three analyses: MPBS = 13,BPP = 1, and MLBS = 99, nucleotide synapomorphies = 10.

3.3.3. Wertheimeria cladeThe Wertheimera clade includes three monotypic genera repre-

sented by seven total specimens: Franciscodoras (n = 2), Kalyptodo-ras (3), and Werthemeria (2). The monophyly of this group is verywell supported in all three analyses: MPBS = 31, BPP = 1,MLBS = 100, nucleotide synapomorphies = 32.

3.3.4. Clade 2Clade 2 is composed of Agamyxis, Rhinodoras clade (n = 3 gen-

era), Pterodoras clade (2), Oxydoras, Lithodoras clade (3), Platydorasclade (2) and Doras clade (9); all genera are represented in theanalyses. The clade is well supported in all three analyses:MPBS = 15, BPP = 1, MLBS = 100, nucleotide synapomorphies = 22.

3.3.5. Rhinodoras cladeThe Rhinodoras clade is composed of three genera: Rhinodoras

(n = 5 valid species, one specimen of each analyzed), Rhynchodoras(three valid species, two specimens of R. woodsi analyzed), and themonotypic Orinocodoras (one specimen). The monophyly of thisgroup is well supported in all three analyses: MPBS = 17, BPP = 1,and MLBS = 100, nucleotide synapomorphies = 22. Within the Rhi-nodoras clade, the sister group relationship of Rhinodoras and Ori-nocodoras is very well supported: MPBS = 26, BPP = 1, andMLBS = 100, nucleotide synapomorphies = 29.

3.3.6. Pterodoras cladeThe Pterodoras clade is composed of Pterodoras and the mono-

typic Doraops. In our analyses Doraops was represented by oneindividual and Pterodoras included both valid species, P. granulosus(two specimens, one from the Amazon, the other from the Paraná)and P. rivasi (one specimen from the Orinoco), plus one possiblyundescribed species, Pterodoras sp. ‘‘spotted’’ (one specimen fromthe Amazon). The monophyly of the Pterodoras clade is well sup-ported in all three analyses: MPBS = 13, BPP = 1, MLBS = 100, nucle-otide synapomorphies = 20.

3.3.7. Lithodoras cladeThe Lithodoras clade is composed of three genera: the mono-

typic Lithodoras, Centrodoras with two valid species (C. brachiatusand C. hasemani) and Megalodoras with two valid species (M.guayoensis and M. uranoscopus). Our analysis included one speci-men each of Lithodoras, C. brachiatus, M. guayoensis, and M.uranoscopus, and two specimens of C. hasemani. The monophylyof the Lithodoras clade is very well supported in all three analyses:MPBS = 22, BPP = 1, and MLBS = 100, nucleotide synapomor-phies = 24. Within this clade, a sister group relationship betweenLithodoras and Megalodoras is well supported: MPBS = 8, BPP = 1,and MLBS = 98, nucleotide synapomorphies = 11.

3.3.8. Doras cladeThe Doras clade is composed of all doradid species with fimbri-

ate barbels (uniquely derived in Doradidae) and represented by at


least 33 valid species (plus up to six undescribed) in nine genera(67 specimens total): monotypic Anduzedoras, Doras (3 of 5 validspecies represented), Hassar (3 of 4), Hemidoras (2 of 2), Leptodoras(10 of 12 plus two undescribed), Nemadoras (6 of 6 plus one unde-scribed), Opsodoras (1 of 3), Ossancora (3 of 4), and Trachydoras (4of 5 plus at least one undescribed). The Doras clade is very wellsupported in all three analyses: MPBS = 23, BPP = 1, MLBS = 100,nucleotide synapomorphies = 26.

3.3.9. Hemidoras cladeThe Hemidoras clade is composed of five genera: Hassar, Hemi-

doras, Nemadoras, Opsodoras, and Ossancora (minus O. punctata).The monophyly of the Hemidoras clade (minus O. punctata) is wellsupported in all three analyses: MPBS = 9, BPP = 1, MLBS = 100,nucleotide synapomorphies = 12. Within this clade, a group com-posed of two species of Ossancora (O. fimbriata + O. asterophysa),two species of Hemidoras (H. morrisi and H. stenopeltis), and Opso-doras (O. morei) is well supported: MPBS = 11, BPP = 1, MLBS = 100,nucleotide synapomorphies = 15. Opsodoras ternetzi is nestedamong species of Nemadoras, and is transferred to that genus.The sister relationship of Nemadoras ternetzi, n. comb., and N. lepo-rhinus is moderately supported in the three analysis (MPBS = 7,BPP = 1, MLBS = 96, nucleotide synapomorphies = 10).

4. Discussion

The doradid clades resulting from this study are compared tothose of other phylogenetic studies, both morphological andmolecular, to highlight examples of congruence, conflict anddiscovery.

4.1. Clade congruence

4.1.1. Monophyly of DoradidaeThe family Doradidae has long been recognized as a distinct

group of catfishes (Bleeker, 1863; Miranda Ribeiro, 1911; Eigen-mann, 1925). Higuchi (1992) completed the first comprehensivecladistic analysis of Doradidae, and proposed three morphologicalsynapomorphies for the family: tympanic area connected to Mülle-rian ramus of parapophysis of fourth vertebra by a strong ligamentand delimited by supracleithrum, postoccipital process, infranu-chal plate and humeral (=posterior) process of cleithrum; presenceof at least two ossified lateral scutes in the postcranial region, onearticulating with posterior nuchal plate and first pleural rib; andreduction or absence of middle posterodorsal process of cleithrum,between the articular and humeral (posterior cleithral) processes.Birindelli (2010, in press) reevaluated Higuchi’s synapomorphieswithin the broader context of the superfamily Doradoidea (Auche-nipteridae + Doradidae) and representatives of 16 additional cat-fish families. Birindelli’s cladistic analysis of morphologicalcharacters identified three synapomorphies for Doradidae: pres-ence of bony scutes formed from lateral-line tubules, presence ofligament between Müllerian ramus and lateral line, and presenceof ossified ligament (infranuchal scute) between first rib and pos-terior nuchal plate.

In the present study, the monophyly of Doradidae is supportedby two of the three separate analyses: Maximum Likelihood(MLBS = 62) and Bayesian (BPP = 0.94). The Maximum Parsimony(MP) analysis, however, placed the doradid subfamily Astrodorad-inae as sister to Auchenipteridae + clade of remaining doradids.Bremer support values are high for the monophyly of Astrodorad-inae (MPBS = 30), the monophyly of Auchenipteridae (MPBS = 11)and the superfamily Doradoidea (MPBS = 19). Bremer support islow, however, for the sister group relationship of

Auchenipteridae + non-astrodoradin doradids (MPBS = 2), and themonophyly of non-astrodoradin doradids (MPBS = 2).

Given the strength of the morphological evidence, we do notdispute the monophyly of Doradidae based on our MP analysis.Long-branch attraction can seriously affect Maximum Parsimony(Kück et al., 2012, and references therein). For example, there isa short internal branch between the common ancestor of Doradi-dae and that of Doradidae + Auchenipteridae (Appendix A). Thebranch leading from the former node to the common ancestor ofAstrodoradins is relatively long, whereas the one leading to thecommon ancestor of remaining doradids (non-astrodoradins) is ex-tremely short. The internal branch below the latter node (i.e., com-mon ancestor of Doradidae + Auchenipteridae) is extremely long.In MP analysis, lineages adjacent to a short internal branch (i.e.,Auchenipteridae and non-astrodoradins doradids) may be mis-placed as sister groups in the class I effect of Kück et al. (2012).Alternatively, signal erosion may cause a single long branch (i.e.,the one leading to astrodoradins) to slip towards the base of thetree (outgroup Aspredinidae in this case) in what has been coinedthe class II effect (Kück et al., 2012). A broader molecular samplingof outgroup taxa in Auchenipteridae and Aspredinidae may over-come the putatively long-branch effect confounding the mono-phyly of Doradidae in our MP analysis.

4.1.2. Monophyly of AstrodoradinaeHiguchi’s (1992) cladistic analysis of morphology supported a

clade composed of seven doradid genera: Amblydoras, Anadoras,Astrodoras, Hypodoras, Merodoras, Physopyxis, and Scorpiodoras.Higuchi et al. (2007) formally named this clade Astrodoradinae,and diagnosed the group on the basis of several morphologicalcharacters: lacrimal (infraorbital one) serrated (except in Anadoras)and participating in the orbital margin (except in Physopyxis); fourto seven pleural ribs; spines on the posterior cleithral process (ex-cept in Anadoras); and postero-inferior portion of the coracoid ex-posed (lattermost mentioned in Abstract only, p. 31). Quagio-Grassiotto et al. (2011) identified spermatic characteristics in twospecies of Astrodoradinae that corroborate the distinctiveness ofthis subfamily. Specifically, semi-cystic spermatogenesis and mod-ified Type III spermiogenesis (both confirmed for Anadoras weddel-ii), and biflagellate sperm (confirmed for A. weddellii andAmblydoras sp.) may be diagnostic characteristics unique withinDoradidae to Astrodoradinae.

In his revision of the nominal subfamily, Sousa (2010), ex-panded Astrodoradinae to include Acanthodoras and Agamyxis.His cladistic analysis of morphological characters supported Ana-doras as sister to a clade composed of Acanthodoras + Agamyxisand the remaining astrodoradins. Sousa (2010) rediagnosed Astr-odoradinae on the basis of seven unambiguous synapomorphies:anterior cranial fontanel oval or circular, reduced number of ribs(68), coracoid not completely covered by muscles ventrally, re-duced number of vertebrae (634), parahypural fused to hypuralone and two, infraorbital one participating in orbit, and third tym-panal scute expanded. The sister group relationship between astr-odoradins (minus Anadoras) and Acanthodoras + Agamyxis wassupported by five synapomorphies: infraorbital one with serratedtransverse crest, infraorbital bones bearing spines, depth of midlat-eral scutes more than half the depth of body (compared anterior toanus), gas bladder short (width greater than length), and midlater-al scutes bearing secondary spines in multiple rows. Only the lastcharacter state was considered non-convergent, and it is reversedfor two taxa (Amblydoras affinis and Physopyxis lyra). Spermaticcharacteristics of Acanthodoras cataphractus (e.g., nucleus sub-spherical, centrioles perpendicular, single flagellum), however, donot corroborate its close relationship with Anadoras and Amblydo-ras (e.g., nucleus bell-shaped, centrioles parallel, two flagella)(Quagio-Grassiotto et al., 2011).


Our study found strong support for the monophyly of Astrodo-radinae as proposed by Higuchi et al. (2007; i.e., minus Acanthodo-ras and Agamyxis): MP Bremer support 30, ML bootstrap 100,Bayesian posterior probability 1. Anadoras is the sister group toother astrodoradins (consistent with Sousa, 2010); the sister grouprelationship between Astrodoras and Hypodoras is well supported(consistent with Higuchi, 1992, and Sousa, 2010); and the mono-typic Merodoras is nested within Amblydoras (consistent with Sou-sa, 2010). Unlike previous studies, Physopyxis is sister toAstrodoras + Hypodoras (MPBS = 4), Scorpiodoras is sister to Ambly-doras (MPBS = 3), and Scorpiodoras bolivarensis is sister to S. heckelii(MPBS = 14). A monophyletic group of several astrodoradin taxa(i.e., Amblydoras, Anadoras, Hypodoras, Physopyxis) also is stronglysupported by Moyer et al.’s (2004: 568, Appendix A) analyses ofcombined 12S, 16S and elongation factor 1 alpha (EF1 alpha) exonsequence data (MP bootstrap = 100, MLBS = 100, BPP = 1). In theiranalyses of EF1 alpha exon sequence data alone (Moyer et al.2004: 565, Appendix A), however, Acanthodoras nested withinthe same clade of astrodoradins with Anadoras sister to that clade(MP bootstrap < 50, MLBS = 50, BPP = 0.98).

4.1.3. Monophyly of Rhinodoras cladeIn a phylogenetic revision of the genus Rhinodoras, Birindelli

(2006) first demonstrated a close relationship between Orinocodo-ras, Rhinodoras, and Rhynchodoras. Higuchi (1992) had previouslyproposed Orinocodoras and Rhinodoras as sister taxa, and suggestedthe latter’s affinity with Rhynchodoras. Birindelli et al. (2007) andBirindelli (2010) upheld the previously proposed relationship:Rhynchodoras (Orinocodoras + Rhinodoras). The results of ourmolecular study are entirely congruent with respect to the mono-phyly of the Rhinodoras clade (MPBS = 17), and the sister grouprelationship of Orinocodoras and Rhinodoras (MPBS = 26).

4.1.4. Monophyly of Doras cladeAmong deeper-level relationships within Doradidae, the Doras

clade is perhaps the most strongly supported by morphology (Hig-uchi, 1992; Birindelli, 2006, 2010). Higuchi (1992) and Birindelli(2010) both identified 16 synapomorphies for this clade. The mostconspicuous of those characteristics is the presence of fimbriae onthe maxillary barbel, an unreversed synapomorphy that unites thenine doradid genera in the clade: Anduzedoras, Doras, Hassar, Hemi-doras, Leptodoras, Nemadoras, Opsodoras, Ossancora, and Trachydo-ras (Higuchi, 1992; Birindelli, 2006, 2010). Likewise, themolecular data strongly support the monophyly of the Doras clade(MPBS = 23). Relationships within the Doras clade, however, differamong Higuchi (1992), Birindelli (2006, 2010) and the presentstudy (see Section 4.2.4).

4.2. Clade conflict

4.2.1. Lithodoras and Pterodoras cladesHiguchi (1992) and Birindelli (2006, 2010) proposed a mono-

phyletic group consisting of Centrodoras, Doraops, Lithodoras, Meg-alodoras and Pterodoras. Higuchi’s (1992) analysis grouped thoselarge-bodied taxa with a relatively small species, the monotypicCentrochir crocodili.

Our molecular study found no support for the monophyly ofthose taxa. There is strong support for both a sister group relation-ship between Doraops and Pterodoras (MPBS = 13), and a separateclade composed of (Centrodoras (Lithodoras + Megalodoras))(MPBS = 22); however, the latter clade appears to be more closelyrelated to the Doras clade. In the MP and ML analyses, the Lithodo-ras clade is sister to the Doras clade (MPBS = 4 and MLBS = 49,respectively); whereas in the Bayesian analysis, the Lithodorasclade is sister to Platydoras clade, but with low support (BPP = 0.8).

Also unlike the present study, Birindelli (2010) placed Doraopssister to Lithodoras + Pterodoras. There is a unique and unreversedmorphological synapomorphy to support Doraops + Pterodoras,the relationship supported by the present study and previouslyby Moyer et al. (2004): anterolateral shoulder of gas bladder withelongated diverticulum curving anteromedially (Birindelli et al.,2009).

4.2.2. Acanthodoras not sister to AgamyxisAcanthodoras and Agamyxis are consistently supported as sister

genera in cladistic analyses of morphological data (Higuchi, 1992;Birindelli, 2010; Sousa, 2010). Species in those two genera sharemany conspicuous features in common such as small size, smallspines on lateral surfaces of dorsal-fin spine and dorsal surface ofpectoral-fin spine, pelvic fin with five branched rays, and darkground color with white markings (Sousa, 2010). Both are foundin small lowland streams, lakes and backwaters, and are some-times syntopic. The present analysis found no evidence for theirclose relationship, results consistent with the previous molecularphylogeny by Moyer et al. (2004). Both genera appear to representrelatively deep solitary lineages. Acanthodoras is placed near thebase of the doradid tree, sister to Clade 1 (all remaining non-astr-odoradins). Although support for that sister group relationship islow (MPBS = 2, MLBS = 56), support for the monophyly of Clade 1is high (MPBS = 13, MLBS = 99, BPP = 1). Agamyxis is placed at thebase of Clade 2, sister to a group composed of the Rhinodoras clade,Pterodoras clade, Oxydoras, Lithodoras clade, Platydoras clade andDoras clade; support for this placement is also high (MPBS = 15,MLBS = 100, BPP = 1).

Although Acanthodoras and Agamyxis share in common severalmorphological traits, each genus is separately diagnosed by fea-tures unique or rare among doradids. Higuchi (1992: 152) consid-ered four characteristics as autapomorphic for Acanthodoras:middle nuchal plate transversely expanded, with straight edges;lateral scutes deep, contiguous but not overlapping; pectoral girdleanteroposteriorly shortened; and posterior cleithral process withwell-developed acicula in adult specimens. Furthermore, Acantho-doras is the only doradid with a rounded caudal fin and hypurals 3,4 and 5 fused into a single plate (Sousa, 2010). Higuchi (1992) alsoproposed five characteristics as autapomorphic for Agamyxis: (1)overall coarse, hispid texture of superficial parts of skeleton, (2)mesethmoid short, flat, quadrangular, without posterior rami, (3)lacrimal (infraorbital one) widely expanded vertically, densely spi-nous, with anterior process connected to mesethmoid and premax-illa, (4) maxilla reduced to tiny splinter, covered by infraorbitalone, and (5) posterior coracoid process with subtriangular spinouspatch fused to basal third. Furthermore, Agamyxis and the mono-typic Hypodoras forficulatus are the only two doradid genera withsix branched rays in the dorsal lobe of the caudal fin (Sousa,2010). The morphological characteristics restricted separately toAcanthodoras and Agamyxis are consistent with their placementas deep, independent lineages within Doradidae.

4.2.3. Oxydoras not sister to Doras clade (fimbriate-barbel doradids)Oxydoras is interesting because it shares several conspicuous

morphological features with members of the Doras clade (e.g., longsnout, compressed body), but has simple barbels. Cladistic analysesof morphological characters consistently support Oxydoras as thesister group to the Doras clade (Higuchi, 1992; Birindelli, 2006,2010). In fact, that relationship is supported by 14 and 13 synapo-morphies identified by Higuchi (1992) and Birindelli (2010),respectively, only about four of which overlap between the twostudies. Moyer et al. (2004) questioned the independence of someof Higuchi’s putative synapomorphies, and disagreed with hisinterpretations of a few character states. Birindelli (2010) reevalu-ated Higuchi’s synapomorphies and corroborated some, but dis-


carded or re-interpreted others. Despite these disagreements, thereis ample morphological evidence for supporting a close relation-ship between Oxydoras and fimbriate-barbel doradids. Alterna-tively, the molecular analysis supports Oxydoras as a somewhatdeep independent lineage within Clade 2 and removed from theDoras clade. Accordingly, our molecular data suggest a remarkablemorphological convergence that may be tied to suctorial bottomfeeding, a habit shared by members of the Doras clade that lackteeth.

4.2.4. Ossancora non-monophyleticBirindelli and Sabaj Pérez (2011) recently proposed Ossancora

for three nominal species and a fourth described as new, O. astero-physa. Higuchi (1992) had previously grouped together two ofthose nominal species (O. punctata and O. eigenmanni) and consid-ered a third (O. fimbriata) to be tentatively placed in this group.Ossancora is diagnosed by having the unique combination of 1)posterior coracoid process approximately as long as posterior clei-thral process in adults, 2) posterior cranial fontanel occluded, 3)long maxillary barbel with smooth elongate fimbriae, and 4) teethpresent on dentary and premaxilla. Within Ossancora, three speciesform a monophyletic group (the O. fimbriata-clade including typespecies O. fimbriata) based on three shared characteristics, one ofwhich is uniquely derived among doradids: smooth elongate fimb-riae inserted in more than one row, one dorsally and one ventrally,along anterior margin of maxillary barbel.

The present molecular analysis supported the monophyly of theO. fimbriata-clade, but did not support the inclusion of O. punctatain Ossancora. Ossancora punctata was either placed at the base ofthe entire Doras clade (Maximum Parsimony analysis) or sister tothe genus Doras (Maximum Likelihood and Bayesian analyses)with Doras + Ossancora as a basal lineage in the Doras clade. Noneof the morphological characteristics used to assign O. punctata toOssancora are unique to the genus; therefore, it is not unreasonableto suspect convergence.

4.3. Doradid clades de novo

4.3.1. Wertheimeria cladeOur molecular analysis is the first to phylogenetically group the

monotypic genera Franciscodoras, Kalyptodoras and Wertheimeria.Wertheimeria was originally described in the family Auchenipteri-dae (for historical treatment, see Vono and Birindelli, 2007), andpreviously placed at the very base of the doradid tree by Higuchi(1992). His analysis supported Franciscodoras as sister to all dorad-ids minus Wertheimeria, and Kalyptodoras at the base of a deepclade containing Acanthodoras, Agamyxis and Platydoras. Birindelli(2010) supported Wertheimeria and Kalyptodoras as sister taxaplaced at the base of the doradid tree, a relationship alluded toin the original description of the latter (Higuchi et al., 1990).Birindelli’s analysis supported Franciscodoras as sister to all dorad-ids minus Wertheimeria + Kalyptodoras. The monophyly of the cladecomposed of all three taxa is strongly supported by molecular data(MPBS = 31, MLBS = 100, BPP = 1). Within this clade there is mod-erate support for a sister group relationship between Franciscodo-ras and Kalyptodoras (MPBS = 6, MLBS = 90, BPP = 1).

4.3.2. Platydoras cladeOur molecular phylogeny is the first to clearly support a sister-

group relationship between Centrochir and Platydoras within thecontext of a comprehensive analysis of Doradidae. Although Max-imum Parsimony Bremer support (=3) and Bayesian PosteriorProbability (0.80) are admittedly weak for this clade, the MaximumLikelihood analysis yielded a more impressive bootstrap value of96. To be fair, Sousa (2010) previously placed Centrochir and Platy-doras as sister taxa in his morphological phylogenetic analysis of a

subset of doradid genera dominated by astrodoradins. And, in hiscomprehensive morphological phylogeny of Doradidae, Birindelli(2010) placed Centrochir in an unresolved trichotomy with Platydo-ras and a large clade of doradids composed of taxa recognized herein the Rhinodoras, Pterodoras, Lithodoras and Doras clades andgenus Oxydoras. Steindachner (1879) was perhaps first to note sim-ilarities between two species, Doras longispinis and Doras dentatus,currently placed in the synonymies of Centrochir crocodili andPlatydoras costatus, respectively (Sabaj and Ferraris, 2003).

4.4. Biogeography

4.4.1. Atlantic Brazilian shield taxaThe three members of the Wertheimeria clade, Franciscodoras

marmoratus, Kalyptodoras bahiensis and W. maculata, are endemicto drainages along the eastern coast of Brazil, from north to south:São Francisco, Paraguaçu, and Pardo + Jequitinhonha, respectively.Higuchi’s (1992) cladistic analysis did not support those taxa asclosely related, but placed Wertheimeria and Franciscodoras as suc-cessive outgroups to all other doradids. Using Higuchi’s topology asa primary example, Stiassny and De Pinna (1994) noted that taxaoccupying basal nodes within their lineages are often species-poorcompared to their sister groups, and have extremely restricted geo-graphical distributions, especially among freshwater fishes. Ribeiro(2006) referred to such as a ‘‘Pattern A’’ distribution, and concludedthat the coastal rivers of eastern Brazil are inhabited by ancient en-demic lineages that trace their origins to the Cretaceous.

Based on our molecular results, the Wertheimeria clade doesrepresent a deep doradid lineage, but its representatives are fur-ther removed from the base of the doradid tree than previouslyhypothesized (Higuchi, 1992; Birindelli, 2010). The fact that theWertheimeria clade is species-poor (three species) and placed atthe base of Clade 1, sister to the much larger and more widely dis-tributed Clade 2 of doradids (ca. 65 species), remains consistentwith the phenomenon identified by Stiassny and De Pinna(1994). Given the proximity of the river systems inhabited by theWertheimeria clade, and the isolation of these basins from thoseof the Amazon, Orinoco and Paraná-Paraguay (La Plata), the mono-phyly of this group is not unexpected. Previous morphologicalstudies have identified a number of plesiomorphic characteristicsshared by members of this clade, but failed to find synapomorphiesuniting all three taxa.

4.4.2. Doradids and the Andean divideThere are only three extant species of doradids in Caribbean

drainages: Centrochir crocodili, endemic to the Magdalena basinnorthwest of the Eastern Andes in Colombia, and Doraops zuloagaiand Rhinodoras thomersoni, both endemic to the Maracaibo basinnorthwest of the Merida Andes. Like members of the Wertheimeriagroup, those taxa are distributed in geographically proximate ba-sins that are currently isolated from the major river systems (Ama-zon, Orinoco and Paraná-Paraguay) draining South America intothe Atlantic Ocean. Unlike members of the Wertheimeria group,the Caribbean taxa are not closely related. The monotypic Centro-chir is sister to Platydoras, the most widely distributed doradidgenus with species inhabiting the Orinoco, Amazon and Paraná-Paraguay (La Plata) basins, as well as smaller coastal drainagesfrom the mouth of the Orinoco to that of the Parnaíba in northeast-ern Brazil (Piorski et al., 2008). Platydoras is not known from theMagdalena basin, and is apparently absent from the Maracaibo aswell. The monotypic Maracaibo endemic Doraops is sister to Ptero-doras, another genus widely distributed in the Orinoco, Amazon,Paraná-Paraguay (La Plata), and Corantijn basins. Maracaibo ende-mic Rhinodoras thomersoni occupies the most basal node within itsgenus, its congeners being widely distributed throughout the Ori-noco, Amazon, Paraná-Paraguay (La Plata), and Essequibo basins.

Appendix ATaxonomic, voucher and sequence data for 143 specimens (130 doradids and 13 outgroup) used in this study. Instances of partial and no sequence data indicated. Single asteriskdenotes likely status as undescribed species; double asterisks denote uncertain status as nominal or undescribed species. Roman numerals used to distinguish multiple specimensof a single species-level taxon.

Family Terminal taxon Voucher museumand catalog number

Voucher tag orother identifier

Country, basin rag1 co1 16s

Doradidae Acanthodoras sp. 1� ‘‘shallow scute’’ ANSP 191096 7602 Venezuela,Orinoco

KC555832 KC555579 KC555836

Acanthodoras sp. 2� ‘‘deep scute’’ I ANSP 182240 V177 Venezuela,Orinoco

KC555711 KC555576 KC555833

Acanthodoras sp. 2� ‘‘deep scute’’ II AUM 43737 V5211 Venezuela,Orinoco

KC555712 KC555577 KC555834

Acanthodoras sp. 2� ‘‘deep scute’’ III AUM 44128 V5306 Venezuela,Orinoco

KC555713 KC555578 KC555835

Acanthodoras sp. 2� ‘‘deep scute’’ IV LBP 4441 24313 Brazil, Negro KC555714 KC555580 KC555837Agamyxis albomaculata I LBP 3036 19176 Venezuela,

OrinocoKC555715 KC555582 KC555839

Agamyxis albomaculata IIa INHS 54563 288 Venezuela,Orinoco

KC555716 KC555581 KC555838

Agamyxis pectinifrons Ia INHS 43281 no tag Peru, Amazon KC555717 KC555583 KC555840Agamyxis pectinifrons II INHS 52017 26733 Peru, Amazon KC555718 KC555584 KC555841Agamyxis pectinifrons III INHS 52017 26732 Peru, Amazon KC555719 KC555585 KC555842Amblydoras afftnis ANSP 179797 2157 Guyana,

EssequiboKC555720 KC555586 KC555844

Amblydoras sp. 1�� ‘‘small eye’’ ANSP 191473 155 Peru, Amazon KC555721 KC555589 KC555847Amblydoras sp. 2�� ‘‘finely reticulateshield’’

ANSP 182525 P6021 Peru, Amazon KC555722 KC555590 KC555848

Amblydoras sp. 3�� ‘‘bold pattern’’ ANSP 191474 078 Peru, Amazon KC555723 KC555588 KC555846Amblydoras nheco, n. comb. ANSP 187416 7328 Brazil, Paraguay KC555724 KC555642 KC555897Anadoras grypus ANSP 179473 T2465 Peru, Amazon KC555725 KC555591 KC555849Anadoras sp.� ‘‘araguaia’’ (sensu Sousa,2010)

MZUSP 89108 CBE8 Brazil, Goias KC555726 no data KC555850

Anadoras weddellii MZUSP 103567 49 Brazil, Para KC555727 KC555592 KC555851Anduzedoras oxyrhynchus Ib ANSP 180293 V111 Venezuela,

OrinocoDQ492547 KC555593 unpublished

Anduzedoras oxyrhynchus II ANSP 191093 no tag Venezuela,Amazon

KC555728 KC555594 KC555852

Astrodoras asterifrons INPA 34100 no tag Brazil, Negro KC555729 KC555597 KC555855Astrodoras sp.� (sensu Roa-Fuenteset al., 2010)

INPA 24660 ARI68 Brazil, Madeira KC555730 KC555596 KC555854

Centrochir crocodili I ANSP 189332 C001 Colombia,Magdalena

KC555731 KC555599 KC555861

Centrochir crocodili II ANSP 189332 C005 Colombia,Magdalena

KC555732 KC555600 KC555857

Centrodoras brachiatusa ANSP 178542 50 Brazil, Amazonas KC555733 KC555601 KC555858Centrodoras hasemani I ANSP 182227 4844 Venzuela,

AmazonKC555734 KC555602 KC555859

Centrodoras hasemani II ANSP 185027 188 Venzuela,Amazon

KC555735 KC555603 KC555860

Doraops zuloagai MCNG uncat.(field no. OC-09-006)

20 Venezuela,Maracaibo

KC555736 KC555604 KC555862

Doras carinatusa INHS 49321 211 Guyana,Essequibo

KC555737 KC555605 KC555863

Doras higuchii MZUSP 96333 7279 Brazil, Amazonas KC555738 KC555606 KC555864Doras micropoeus I ANSP 187110 6932 Suriname,

Maronino data KC555607 KC555865

Doras micropoeus II ANSP 187110 6945 Suriname,Maroni

KC555739 KC555609 KC555867

Doras micropoeus III ANSP 187110 6941 (sk) Suriname,Maroni

KC555740 KC555608 KC555866

Franciscodoras marmoratus I MNRJ 23012 MNTI 39018 Brazil, SãoFrancisco

no data KC555611 KC555869

Franciscodoras marmoratus II LBP 272 4193 Brazil, SãoFrancisco

KC555741 KC555610 KC555868

Hassar affmis MCP 45010(1 of 6) P1 Brazil, Mearim KC555742 KC555614 KC555872Hassar orestis ANSP 181090 6154 Peru, Amazon KC555743 KC555615 KC555873Hassar wilderi MZUSP 86216 – Brazil, Araguaia KC555744 KC555616 KC555874Hemidoras morrisi ANSP 182512 6081 Peru, Amazon KC555745 KC555617 KC555875Hemidoras stenopeltis ANSP 182756 6241 Peru, Amazon KC555746 KC555618 KC555876Hypodoras forfwulatus ANSP 179009 1540 Peru, Amazon KC555747 KC555619 KC555877Kalyptodoras bahiensis I MZUSP 87839 – Brazil,

Paraguassuno data KC555620 KC555878

Kalyptodoras bahiensis II MZUSP 87841 – Brazil,Paraguassu


Kalyptodoras bahiensis III MZUSP 100737 – Brazil,Paraguassu

KC555748 KC555622 KC555880

Leptodoras acipenserinus ANSP 182202 P6346 Peru, Amazon KC555749 KC555623 KC555881Leptodoras cataniai ANSP 181043 P6092 Peru, Amazon KC555750 KC555624 KC555882


Appendix A (continued)




Leptodoras cf. copei � ‘‘Amazon’’ ANSP 178540 149 Brazil, Amazonas KC555751 KC555625 KC555883Leptodoras copei ANSP 182225 V073 Venezuela,

OrinocoKC555752 KC555626 KC555884

Leptodoras hasemani I ANSP 180897 4033 Venezuela,Orinoco

KC555753 KC555627 KC555885

Leptodoras hasemani II MZUSP 97363 7258 Brazil, Jamanxin KC555754 KC555628 KC555886Leptodoras juruensis ANSP 181046 P6090 Peru, Amazon KC555755 KC555629 KC555887Leptodoras linnelli Ic ANSP 179631 2093 Guyana,

EssequiboDQ492604 unpublished unpublished

Leptodoras linnelli II ANSP 179177 2433 Guyana,Rupununni

KC555756 KC555630 KC555888

Leptodoras marki MNRJ 33067 MNTI 2439 Brazil, Xingii KC555757 KC555631 KC555889Leptodoras myersi ANSP 181045 P6205 Peru, Amazon KC555758 KC555632 KC555890Leptodoras oyakawai I ANSP 187336 7209 Brazil, Tapajos KC555759 KC555635 no dataLeptodoras oyakawai II MZUSP 97395 7208 Brazil, Tapajos KC555760 KC555634 KC555892Leptodoras praelongus I ANSP 180913 195 Brazil, Negro KC555761 KC555636 no dataLeptodoras praelongus IIa ANSP 178534 54 Brazi, Branco no data KC555637 KC555893Leptodoras cf. oyakawai� ‘‘Teles Pires’’ MZUSP 96597 7079 Brazil, Amazon KC555762 KC555633 KC555891Lithodoras dorsalis ANSP 187376 7332 Brazil, Amazon KC555763 KC555639 KC555895Megalodoras guayoensis MHNLS 20354(1 of 2) no tag Venezuela,

MaracaiboKC555764 KC555640 KC555896

Megalodoras uranoscopus ANSP 178249 1685 Peru, Amazon no data KC555641 no dataNemadoras elongatus I ANSP 182295 P6282 Peru, Amazon KC555765 KC555643 KC555898Nemadoras elongatus II ANSP 182613 P6299 Peru, Amazon KC555766 KC555644 KC555899Nemadoras elongatus III ANSP 182850 198 Brazil, Amazon KC555767 KC555645 KC555900Nemadoras hemipeltis I ANSP 181095 P6111 Peru, Amazon KC555768 KC555646 KC555901Nemadoras hemipeltis II ANSP 182283 P6272 Peru, Amazon KC555769 KC555647 no dataNemadoras humeralis I ANSP 178550 197 Brazil, Amazon KC555770 KC555648 KC555902Nemadoras humeralis II ANSP 182596 P6301 Peru, Amazon KC555771 KC555649 KC555903Nemadoras humeralis III ANSP 182721 P6228 Peru, Amazon KC555772 KC555650 KC555904Nemadoras leporhinus I MZUSP 96596 7082 Brazil, Tapajos KC555773 KC555653 KC555907Nemadoras leporhinus II ANSP 180297 V104 Venezuela,

OrinocoKC555774 KC555651 KC555905

Nemadoras leporhinus III ANSP 182825 P4319 Venezuela,Orinoco

KC555775 KC555652 KC555906

Nemadoras sp. ‘‘ghost’’� I ANSP 180551 4076 Peru, Madeira KC555776 KC555654 KC555908Nemadoras sp. ‘‘ghost’’� II ANSP 182744 P6096 Peru, Amazon KC555777 KC555655 KC555909Nemadoras trimaculatus I ANSP 178252 1679 Peru, Amazon KC555778 KC555656 KC555910Nemadoras trimaculatus II ANSP 182633 P6306 Peru, Madeira KC555779 KC555657 KC555911Nemadoras trimaculatus III ANSP 182824 P4589 Venezuela,

OrinocoKC555780 KC555658 KC555912

Opsodoras morei ANSP 183028 P4570 Brazil, Negro KC555781 KC555659 KC555913Nemadoras ternetzi, n. comb. I ANSP 179203 T2451 Guyana, Takutu KC555782 KC555660 KC555914Nemadoras ternetzi, n. comb. II ANSP 182852 56 Brazil, Negro KC555783 KC555661 KC555915Nemadoras ternetzi, n. comb. III ANSP 180177 2292 Peru, Amazon KC555784 KC555662 KC555916Nemadoras ternetzi, n. comb. IV ANSP 180905 4604 Venezuela,

OrinocoKC555785 KC555663 KC555917

Orinocodoras eigenmannia INHS 54564 – Venzuela,Amazon

KC555786 KC555664 KC555918

Ossancora asterophysa I ANSP 178546 200 Brazil, Negro no data KC555665 KC555919Ossancora asterophysa II ANSP 182516 P6003 Peru, Amazon KC555787 KC555666 KC555920Ossancora fimbriata I ANSP 192457 M353 Brazil, Amazon no data KC555667 KC555921Ossancora fimbriata II ANSP 192457 M354 Brazil, Amazon no data KC555668 KC555922Ossancora punctata I ANSP 181015 A5118 Argentina, Parana KC555788 KC555670 KC555924Ossancora punctata II ANSP 187005 P6333 (sk) Peru, Amazon KC555789 KC555669 KC555923Oxydoras kneri ANSP 182203 A5196(sk) Argentina, Parana KC555790 KC555671 KC555925Oxydoras niger MCNGex. ANSP 181080 P6056 Peru, Amazon KC555791 KC555672 KC555926Oxydoras sifontesi ANSP 189324 P4340 Venezuela,

OrinocoKC555792 KC555673 KC555927

Physopyxis ananas ANSP 190508 7576 Venezuela,Orinoco

KC555793 KC555674 KC555928

Physopyxis lyra ANSP 180176 2300 Peru, Amazon KC555794 KC555675 KC555929Platydoras armatulus ANSP 181008 A5200 Argentina, Parana KC555795 KC555676 KC555930Platydoras brachylecis MCP 45249(1 of 6) no tag Brazil, Maranhao KC555796 KC555677 KC555931Platydoras costatus ANSP 187111 7064 Suriname,

SurinamKC555797 KC555678 KC555932

Platydoras hancockii ANSP 179144 2032 Guyana,Rupununni

KC555798 KC555679 KC555933

Platydoras sp. 1�� ‘‘Maroni’’ ANSP 187377 7050 Guyana, Maroni KC555799 KC555680 no dataPlatydoras sp. 2�� ‘‘Curua’’ MZUSP 96336 7290 Brazil, Amazon KC555800 KC555681 KC555934Pterodoras granulosus I ANSP 178350 1560 Peru, Amazon KC555801 KC555685 KC555938Pterodoras granulosus II ANSP 180883 A5100 Argentina, Parana KC555802 KC555686 KC555939Pterodoras rivasi MHNLS 20353 no tag Venezuela, Apure KC555803 KC555687 KC555940Pterodoras sp.�� ‘‘spotted’’ ANSP 181197 P6203 Peru, Amazon KC555804 KC555684 KC555937

(continued on next page)


Appendix A (continued)




Rhinodoras armbrusteri ANSP 179096 no tag Guyana,Rupununni

KC555805 KC555688 KC555941

Rhinodoras boehlkei ANSP 181044 P6059 Peru, Amazon KC555806 KC555689 KC555942Rhinodoras dorbignyi LBP3218 19423 Brazil, Paraguay KC555807 KC555690 KC555943Rhinodoras gallagheri ANSP 191086 T09020 Venezuela,

GuanareKC555808 KC555691 KC555944

Rhinodoras thomersoni MCNG uncat. (field no. OC-09-006)

21 Venezuela,Maracaibo

KC555809 KC555692 KC555945

Rhynchodoras woodsi I ANSP 181042 P6052 Peru, Amazon KC555810 KC555693 KC555946Rhynchodoras woodsi II ANSP 181042 P6061 Peru, Amazon KC555811 KC555694 KC555947Scorpiodoras bolivarensis, n. comb. ANSP 182267 V183 Venzuela,

AmazonKC555812 KC555587 KC555845

Scorpiodoras heckelii I AUM 42953 V5406 Venezuela,Orinoco


Scorpiodoras heckelii II ANSP 182790 V5404 Venezuela,Orinoco

KC555813 KC555695 KC555948

Trachydoras brevis MHNG 2650.062 GY04-091 Guyana,Rupununni

KC555814 KC555698 KC555951

Trachydoras cf. microstomus �� I ANSP 182619 P6310 Peru, Amazon KC555815 KC555703 KC555956Trachydoras cf. microstomus �� II ANSP 185014 257 Brazil, Negro no data KC555701 KC555954Trachydoras nattereri I ANSP 182593 P6313 Peru, Amazon KC555816 no data KC555957Trachydoras nattereri II ANSP 179853 T2099? Peru, Amazon KC555817 KC555702 KC555955Trachydoras paraguayensis ANSP 181011 A5134 Argentina, Parana KC555818 KC555704 KC555958Trachydoras sp.�� ‘‘Essequibo’’ ANSP 179855 2211 Guyana,

RupununniKC555819 KC555699 KC555952

Trachydoras sp.�

‘‘pseudomicrostomus’’ IANSP 185076 V021 Venezuela,

Orinocono data KC555707 KC555961

Trachydoras sp.�

‘‘pseudomicrostomus’’ IIMZUSP 92813 – Brazil, Tapajos KC555820 KC555700 KC555953

Trachydoras sp.�

‘‘pseudomicrostomus’’ IIIANSP 179866 2475 Peru, Amazon KC555821 KC555705 KC555959

Trachydoras sp.�

‘‘pseudomicrostomus’’ IVANSP 187373 T2293 Peru, Amazon no data KC555706 KC555960

Trachydoras steindachneri ANSP 178256 1673 Peru, Amazon no data KC555708 KC555962Wertheimeria maculata I MCP 43855(1 of 2) – Brazil,

JequitinhonhaKC555822 KC555709 KC555963

Wertheimeria maculata II MZUSP 88614 _ Brazil,Jequitinhonha


Auchenipteridae Ageneiosus inermis ANSP 189090 6996 Sunname, Marom KC555823 no data KC555843Ageneiosus ucayalensisc INHS 52920 26669 Peru, Amazon DQ492540 EU490849 unpublishedAuchenipterus nigripinnis ANSP 182418 A5053 Argentina, Parana KC555824 KC555598 KC555856Centromochlus heckeliib ANSP 182773 – Brazil, Amazon DQ492465 no data unpublishedGelanoglanis sp. � ANSP 180806 1174 Peru, Madeira KC555825 KC555612 KC555870Glanidium leopardum ANSP 189104 2089 Suriname,

MaroniKC555826 KC555613 KC555871

Liosomadoras oncinus ANSP 191102 BO6148 Venezuela,Orinoco

KC555827 KC555638 KC555894

Pseudepapterus hasemani ANSP 178311 1573 Peru, Amazon KC555828 KC555683 KC555936Trachelyopterus galeatusc INHS 49034 – Guyana, Maroni DQ492634 EU490848 JX899742Trachycorystes trachycorystes ANSP 180820 V121 Venezuela,

OrinocoKC555829 KC555697 KC555950

Aspredinidae Aspredo aspredo ANSP 191475 205 Brazil, Amazon no data KC555595 KC555853Pseudobunocephalus rugosus ANSP 185102 A5067 Argentina, Parana KC555830 KC555682 KC555935Xyliphius sp. � ANSP 182322 P6291 Peru, Amazon KC555831 no data KC555965

a Moyer sequenced same specimens for 12s, 16s and EF1 alpha; MAH resequenced 16s.b 16s and rag1 from Sullivanc 16s, co1, rag1 from Sullivan


Fossils from the Miocene Urumaco Formation of Venezuela havebeen identified as Doraops cf. zuloagai and Rhinodoras cf. thomer-soni, establishing the minimum age of divergence from theirrespective clades at about 8 Ma (Sabaj Pérez et al., 2007).

4.4.3. Sub-Andean Foreland DoradidsFrom the Early Eocene to Late Miocene (ca. 59–7 Ma), northern

South America was dominated by a large sub-Andean River systemthat flowed from south to north with outlets either along the Paci-fic (Late Eocene, ca. 43–34 Ma) or Caribbean coasts (Early Eocene,ca. 59–43 Ma, and Oligocene to Middle Miocene, ca. 34–11 Ma)

(see Hoorn and Wesselingh, 2010: Plates 14 and 15). Over thecourse of this time, the uplift of the Andes fragmented the sub-An-dean River system into what are now portions of the upper Para-guay (La Plata), upper Amazonas, Orinoco, and Maracaibo basins.Fragmentation of the foreland system may have began in the southwith the rise of the Michicola Arch ca. 43–34 Ma, isolating its head-waters and redirecting them southward into the present-day Para-guay (La Plata) basin. Deconstruction is nearly complete by theLate Miocene (ca. 11–7 Ma) as the rise of the Vaupés Arch sepa-rated the present-day Orinoco and Amazon basins, and redirectedthe latter eastward to its Atlantic outlet. By the end of the Miocene


(ca. 7 Ma), the uplift of the Eastern Cordillera and Mérida Andes iscomplete, isolating the old outlet of the sub-Andean River system

Appendix B. Phylogeny of Doradidae inferred from Maximum Parsimony analysis of rag16443 steps. Node values correspond to Bremer support.

into the present-day Maracaibo basin, and redirecting the Orinocoto complete its present course and outlet into the Atlantic. During

, 16s, and co1 sequences for all the specimens. Strict consensus of 36 trees, each with


the history of the foreland basin, multiple stream capture eventsbetween the Amazonas and Paraguay (La Plata) likely facilitatedand disrupted gene flow between and within those two river sys-

Appendix C. Phylogeny of Doradidae inferred from Maximum Likelihood analysis of rag1

tems (Lundberg et al., 1998). The Amazonas and Orinoco, on theother hand, have a longer shared history with their initial separa-tion beginning in the Late Miocene. Nowadays, the Amazonas

, 16s, and co1 sequences for all the specimens. Node values correspond to bootstrap.


and Orinoco basins are still connected through the Casiquiare Ca-nal, which drains from the uppermost (southern) Orinoco intothe upper Negro River.

Appendix D. Phylogeny of Doradidae inferred from Bayesian analysis of rag1, 16s, and co

Our molecular phylogeny offers at least three examples forexamining the timing of the fragmentation of the sub-Andean Riv-er system. Three genera (Oxydoras, Pterodoras, Rhinodoras) are

1 sequences for all the specimens. Node values correspond to posterior probability.


presently distributed throughout the Amazon, Orinoco and Paraná-Paraguay (La Plata) river systems. For Oxydoras and Rhinodoras, themolecular phylogeny supports the Paraná-Paraguay species as sis-ter to Amazon + Orinoco species in a topology consistent with bio-geographic analyses of the three basins (e.g., Albert and Carvalho,2011). For Pterodoras, however, the Orinoco species (P. rivasi) ismore basal, sister to the two samples of P. granulosus from theAmazon and Paraná-Paraguay basins, respectively.

5. Conclusions

This study represents the most comprehensive molecular dataset assembled for Doradidae, and has uncovered relationshipsnewly hypothesized for the family with interesting biogeographicimplications. For example, Wertheimeria, Franciscodoras and Kaly-ptodoras, three monotypic genera inhabiting separate Atlanticdrainages along the easternmost coast of Brazil, are supported asa monophyletic group near the base of the doradid tree. Centrochir,a monotypic genus endemic to the Magdalena (Caribbean Dr.), issister to Platydoras, a widespread genus occurring east of the An-dean divide throughout the Amazon, Orinoco, Paraná-Paraguay(La Plata), and Atlantic coastal drainages. Doraops, a monotypicgenus endemic to the Maracaibo basin (Caribbean Dr.), is sisterto Pterodoras, another genus widespread in Atlantic drainages.Other aspects of the molecular phylogeny agree with previous rela-tionships hypothesized on the basis of morphology, such as themonophyly of fimbriate-barbel taxa, subfamily Astrodoradinae(sensu Higuchi et al., 2007), and the clade (Rhynchodoras (Orinoco-doras + Rhinodoras). Some of the phylogenetic relationships sup-ported by the molecular data, however, are not currentlycorroborated by morphological evidence. An expanded data setcoupled to combined analyses of molecular and morphologicalcharacters is needed to resolve these uncertainties.

Acknowledgments

We sincerely thank José L.O. Birindelli and one anonymous re-viewer for their careful read of the manuscript, helpful corrections,and thoughtful suggestions. Special thanks to Katriina L. Ilves forgathering sequence data for Centrochir. For generous donations oftissues we thank: J. Lundberg and J. Sullivan (ANSP), D. Wernekeand L. de Souza (AUM), R. Betancur (GWU), C. Lasso (IAvH), M. Ro-cha (INPA), C. Oliveira and V. Tagliacollo (LBP), R. Covain (MHNG),P. Buckup and J. Maldonado (MNRJ), J. Birindelli, J. Cunha, L. Sousa,and O. Oyakawa (MZUSP), N. Lujan (Texas A&M), N. Piorski (Univ.Fed. Maranhão), G. Moyer (USFWS), F. Andrade, M. P. Sena, A. Clist-enes and M. Hardman. For outstanding help in the field we thank: J.Armbruster, M. Azpelicueta, J. Birindelli, J.D. Bogotá, A. Bullard, B.Burr, T. Carvalho, O. Castillo, C. DoNascimiento, N. Lujan, J. Lund-berg, J.M. Mirande, J. Mol, A. Netto-Ferreira, S.N. Sabino, C. SabajPérez, L. Rapp Py-Daniel, B. Sidlauskas, L. Sousa, J. Stewart, M. Tho-mas, R. Weitzell, D. Werneke and P. Willink. For informative dis-cussions on molecular techniques MAH thanks J. Sullivan and C.Mattoni. The majority of the lab work was conducted in the Labo-ratory for Molecular Systematics and Ecology at the Academy ofNatural Sciences with support to MAH provided by the DeepfinStudent Exchange Program (NSF DEB-0443470), Jessup Award ofthe Academy of Natural Sciences of Philadelphia, and the All Cat-fish Species Inventory (NSF DEB-0315963). MAH also acknowl-edges the Coordenação de Aperfeiçoamento de Pessoal de NivelSuperior (CAPES) for PEC-PG student award. MSP also supportedin part by the All Catfish Species Inventory.

Appendix A.

Appendix B.

Appendix C.

Appendix D.

References

Albert, J.S., Carvalho, T.P., 2011. Neogene assembly of modern faunas. In: Albert, J.S.,Reis, R.E. (Eds.), Historical Biogeography of Neotropical Freshwater Fishes.University of California Press, pp. 119–136.

Birindelli, J.L.O., 2006. Revisão Taxonômica e Filogenia do Gênero RhinodorasBleeker, 1862 (Siluriformes, Doradidae). Unpublished Master Thesis,Universidade de São Paulo.

Birindelli, J.L.O., 2010. Relações Filogenéticas da Superfamília Doradoidea(Ostariophysi Siluriformes). Unpublished Doctoral Thesis, Universidade de SãoPaulo.

Birindelli, J.L.O., in press. Phylogenetic relationships of the south AmericanDoradoidea (Ostariophysi, Siluriformes). Neotropical Ichthyology.

Birindelli, J.L.O., Sabaj Pérez, M.H., 2011. Ossancora, new genus of thorny catfish(Teleostei: Siluriformes: Doradidae) with description of one new species.Proceedings of the Academy of Natural Sciences of Philadelphia 161, 117–152.

Birindelli, J.L.O., Sabaj, M.H., Taphorn, D.C., 2007. New species of Rhynchodoras fromthe Río Orinoco with comments on the genus (Siluriformes: Doradidae). Copeia2007 (3), 672–684.

Birindelli, J.L.O., Sousa, L., Sabaj Pérez, M.H., 2009. Morphology of the gas bladder inthorny catfishes (Siluriformes: Doradidae). Proceedings of the Academy ofNatural Sciences of Philadelphia 158, 261–296.

Bleeker, P., 1863. Systema Silurorum Revisium. Nederlansch Tisjdschrift voor deDierkunde 1, 77–122.

Bremer, K., 1994. Branch support and tree stability. Cladistics 10, 295–304.De Pinna, M.C.C., 1998. Phylogenetic relationships of neotropical siluriformes

(Teleostei: Ostariophysi): historical overview and synthesis of hypotheses. In:Malabarba, L., Reis, R., Vari, R.P., Lucena, C., Lucena, L. (Eds.), Phylogeny andClassification of Neotropical Fishes. Edipucrs, Porto Alegre, Brazil, pp. 279–330.

Edgar, R.C., 2004. Muscle: multiple sequence alignment with high accuracy andhigh throughput. Nucleic Acids Research 32 (5), 1792–1797.

Eigenmann, C.H., 1925. A review of the Doradidae, a family of south AmericanNematognathi or catfishes. Transactions of the American Philosophical Society22, 280–365.

Ferraris, C.J., 1988. The Auchenipteridae: putative monophyly and systematics, withclassification of the neotropical doradoid catfishes (Ostariphysi: Siluriformes).Unpublished Doctoral Thesis, City University of New York, New York.

Folmer, O., Black, M., Hoeh, W., Lutz, R., Vrijenhoek, R., 1994. DNA primers foramplification of mitochondrial cytochrome c oxidase subunit I from diversemetazoan invertebrates. Molecular Marine Biology and Biotechnology 3 (5),294–299.

Friel, J.P., 1994. A phylogenetic study of the neotropical banjo catfishes (Teleostei:Siluriformes: Aspredinidae). Unpublished Doctoral Thesis, Duke University,Durham.

Goloboff, P.A., Farris, J.S., Nixon, K.C., 2008. TNT, a free program for phylogeneticanalysis. Cladistics 24, 774–786.

Goulding, M., Smith, N.J.H., Mahar, D.J., 1996. Floods of Fortune: Ecology andEconomy along the Amazon. Columbia University Press, New York, 193p.

Hebert, P.D.N., Cywinska, A., Ball, S.L., deWaard, J.R., 2003. Biological identificationsthrough DNA barcodes. Proceedings of the Royal Society of London Series B –Biological Sciences 270, 313–321.

Higuchi, H., 1992. A phylogeny of the south American thorny catfishes(Osteichthyes; Siluriformes; Doradidae). Unpublished Doctoral Thesis,Harvard University.

Higuchi, H., Britski, H.A., Garavello, J.C., 1990. Kalyptodoras bahiensis, a new genusand species of thorny catfish from northeastern Brazil (Siluriformes:Doradidae). Ichthyological Explorations of Freshwaters 1 (3), 219–225.

Higuchi, H., Birindelli, J.L., Sousa, L., Britski, H., 2007. Merodoras nheco, new genusand species from Rio Paraguay basin, Brazil (Siluriformes, Doradidae), andnomination of the new subfamily Astrodoradinae. Zootaxa 1446, 31–42.

Hoorn, C., Wesselingh, F.P. (Eds.), 2010. Amazonia—Landscape and SpeciesEvolution: A Look into the Past. Wiley-Blackwell, 447p.

Huelsenbeck, J.P., Ronquist, F., 2001. Mrbayes: Bayesian inference of phylogeny.Bioinformatics 17, 754–755.

Knowlton, N., Weigt, L.A., 1998. New dates and new rates for divergence across theisthmus of Panama. Proceedings of the Royal Society of London, Series B:Biological Sciences 265, 2257–2263. http://dx.doi.org/10.1098/rspb.1998.0568.

http://dx.doi.org/10.1098/rspb.1998.0568


Kück, P., Mayer, C., Wägele, J.-W., Misof, B., 2012. Long branch effects distortmaximum likelihood phylogenies in simulations despite selection of thecorrect model. PLoS ONE 7 (5), e36593. http://dx.doi.org/10.1371/journal.pone.0036593.

Lundberg, J.G., Marshall, L.G., Guerrero, J., Horton, B., Malabarba, M.C.S.L.,Wesselingh, F., 1998. The stage for neotropical fish diversification: a historyof tropical south America rivers. In: Malabarba, L., Reis, R., Vari, R.P., Lucena, C.,Lucena, L. (Eds.), Phylogeny and Classification of Neotropical Fishes. Edipucrs,Porto Alegre, Brazil, pp. 13–48.

Maddison, D.R., Maddison, W.P., 2000. MacClade 4: Analysis of Phylogeny andCharacter Evolution. Version 4.0. Sinauer Associates, Sunderland,Massachusetts.

Maddison, W.P., Maddison, D.R., 2011. Mesquite: A Modular System forEvolutionary Analysis. Version 2.74. <http://mesquiteproject.org>.

Miranda Ribeiro, A., 1911. Fauna brasiliensis. Peixes IV(a). EleutherobranchiosAspirophoros. Archivos do Museu Nacional do Rio de Janeiro 16, 1–504.

Moyer, G., Burr, B.M., Krajewski, C., 2004. Phylogenetic relationships of thornycatfishes (Siluriformes: Doradidae) inferred from molecular and morphologicaldata. Zoological Journal of the Linnean Society 140, 551–575.

Novoa, D.F., Ramos, F., 1978. Las pesquerías comerciales del río Orinoco. Corp.Venezolana de Guayana.

Oettinger, M.A., Schatz, D.G., Gorka, C., Baltimore, D., 1990. RAG-1 and RAG-2,adjacent genes that synergistically activate V(D)J recombination. Science 248,1517–1523.

Palumbi, S.R., Martin, A., Romano, S., McMillan, W.O., Stice, L., Grabaswski, G., 1991.The Simple Fool’s guide to PCR. Version 2.0. Privately published, compiled by S.Palumbi, University of Hawaii, Honolulu.

Piorski, N.M., Garavello, J.C., Arce, H.M., Sabaj Pérez, M.H., 2008. Platydorasbrachylecis, a new species of thorny catfish (Siluriformes: Doradidae) fromnortheastern Brazil. Neotropical Ichthyology 6 (3), 481–494.

Quagio-Grassiotto, I., Ortiz, R.J., Sabaj Pérez, M.H., Oliveira, C., 2011. Sperm ofDoradidae (Teleostei: Siluriformes). Tissue and Cell 43 (2011), 8–23.

Ribeiro, A.C., 2006. Tectonic history and the biogeography of the freshwater fishesfrom the coastal drainages of eastern Brazil: an example of faunal evolutionassociated with a divergent continental margin. Neotropical Ichthyology 4 (2),225–246.

Rindal, E., Brower, A.V.Z., 2011. Do model-based phylogenetic analyses performbetter than parsimony? A test with empirical data. Cladistics 27, 331–334.

Roa-Fuentes, C.A., Alonso, J.C., Alfonso, A., Sabaj Pérez, M., 2010. Pisces, Siluriformes,Doradidae, Astrodoras Bleeker, 1862: first record in the Colombian Amazon.Check List, Journal of Species Lists and Distribution 6 (4), 485–487.

Ronquist, F., Huelsenbeck, J.P., 2003. MRBAYES 3: Bayesian phylogenetic inferenceunder mixed models. Bioinformatics 19, 1572–1574.

Sabaj, M.H., 2002. Taxonomy of the neotropical thorny catfishes (Siluriformes:Doradidae) and revision of the genus Leptodoras. Unpublished Doctoral Thesis,University of Illinois at Urbana-Champaign, Illinois, 242p.

Sabaj, M.H., Ferraris Jr., C.J., 2003. Family Doradidae (Thorny catfishes). In: Reis, R.E.,Kullander, S.O., Ferraris Jr., C.J. (organizers), Check List of the Freshwater Fishes ofSouth and Central America, EDIPUCRS, Porto Alegre, 2003, pp. 456–469 (742p).

Sabaj Pérez, M.H., Aguilera, O.A., Lundberg, J.G., 2007. Fossil catfishes of the familiesDoradidae and Pimelodidae (Teleostei: Siluriformes) from the MioceneUrumaco formation of Venezuela. Proceedings of the Academy of NaturalSciences of Philadelphia 156, 157–194.

Schatz, D.G., Oettinger, M.A., Baltimore, D., 1989. The V(D)J recombinationactivating gene, RAG-1. Cell 59, 1035–1048.

Sousa, L., 2010. Revisão Taxonômica e Filogenia de Astrodoradinae (Siluriformes,Doradidae). Unpublished Doctoral Thesis, Universidade de São Paulo.

Sousa, L., Rapp Py-Daniel, L.H., 2005. Description of two new species of physopyxisand redescription of P. lyra (Siluriformes: Doradidae). Neotropical Ichthyology 3(4), 625–636.

Stamatakis, A., 2006. RAxML-VI-HPC: maximum likelihood-based phylogeneticanalyses with thousands of taxa and mixed models. Bioinformatics 22 (21),2688–2690.

Steindachner, F., 1879. Zur Fisch-fauna des Magdalenen-Stromes. Denkschriften derKaiserlichen Akademie der Wissenschaften in Wien, Mathematisch-Naturwissenschaftliche Classe 39, 19–78 (Pls. 1–15).

Stiassny, M.L.J., De Pinna, M.C.C., 1994. Basal taxa and the role of cladistic patternsin the evaluation of conservation priorities: a view from freshwater. In: Forey,P., Humphreys, C.J., Vane-Wright, R.I. (Eds.), Systematics and ConservationEvaluation, Systematics Association Special Volume. Oxford University Press,pp. 235–249.

Sullivan, J.P., Lundberg, J.G., Hardman, M., 2006. A phylogenetic analyis of the majorgroups of catfishes (Teleostei: Siluriformes) using Rag1 and Rag2 nuclear genesequences. Molecular Phylogenetics and Evolution 41, 636–662.

Venkatesh, B., Ning, Y., Brenner, S., 1999. Late changes in spliceosomal intronsdefine clades in vertebrate evolution. Proceedings of the National Academy ofSciences of the United States of America 96, 10267–10271.

Vono, V., Birindelli, J.L.O., 2007. Natural history of Wertheimeria maculata, aprimitive doradid endemic to eastern Brazil (Siluriformes, Doradidae).Ichthyological Explorations of Freshwaters 18 (2), 183–191.

Wilgenbusch, J.C., Warren, D.L., Swofford, D.L., 2004. AWTY: A System for GraphicalExploration of MCMC Convergence in Bayesian Phylogenetic Inference. <http://ceb.csit.fsu.edu/awty>.

Willet, C.E., Cherry, J.J., Steiner, L.A., 1997. Characterization and expression of therecombination activating genes (rag1 and rag2) of Zebra fish. Immunogenetics45, 394–404.

http://dx.doi.org/10.1371/journal.pone.0036593

http://dx.doi.org/10.1371/journal.pone.0036593

http://mesquiteproject.org

http://ceb.csit.fsu.edu/awty

http://ceb.csit.fsu.edu/awty

molecular phylogeny of thorny catfishes (siluriformes: doradidae)

Documents