Based on Whole Mitogenome Sequences: Implications for Higher-Level Relationships Within the Otocephala

Home , Cromeria, Gonorynchiformes, Gonorynchus, Kneria, Kneriidae, Otocephala

Molecular Phylogenetics and Evolution 37 (2005) 165–177 www.elsevier.com/locate/ympev

Molecular systematics of the gonorynchiform Wshes (Teleostei) based on whole mitogenome sequences: Implications for higher-level relationships within the Otocephala

Sébastien Lavoué a,¤, Masaki Miya b, Jun G. Inoue a, Kenji Saitoh c, Naoya B. Ishiguro a, Mutsumi Nishida a

a Department of Marine Bioscience, Ocean Research Institute, The University of Tokyo, 1-15-1 Minamidai, Nakano-ku, Tokyo 164-8639, Japan b Department of Zoology, Natural History Museum and Institute, Chiba, 955-2 Aoba-cho, Chuo-ku, Chiba 260-8682, Japan c Tohoku National Fisheries Research Institute, Shinhama, Shiogama, Miyagi 985-0001, Japan

Received 8 November 2004; revised 15 March 2005 Available online 10 May 2005

Abstract

Although the order Gonorynchiformes includes only 31 species assigned to seven genera and four families, it exhibits a large vari- ety of anatomical structures, making diYcult the reconstruction of phylogenetic relationships among its representatives. Within the basal teleosts, the Gonorynchiformes belong to the Otocephala where they have been alternatively placed as the sister group of the Otophysi and of the Clupeiformes. In this context, we investigated the phylogeny of the Gonorynchiformes using whole mitogenome sequences from 40 species (six being newly determined for this study). Our taxonomic sampling included at least one species of each gonorynchiform genus and of each other major otocephalan lineage. Unambiguously aligned, concatenated mitogenomic sequences (excluding the ND6 gene and control region) were divided into Wve partitions (1st, 2nd, and 3rd codon positions, tRNA genes, and rRNA genes) and partitioned Bayesian analyses were conducted. The resultant phylogenetic trees were fully resolved, with most of the nodes well supported by the high posterior probabilities. As expected, the Otocephala were recovered as monophyletic. Within this group, the mitogenome data supported the monophyly of Alepocephaloidei, Gonorynchiformes, Otophysi, and Clupeiformes. The Gonorynchiformes and the Otophysi formed a sister group, rending the Ostariophysi monophyletic. This result conXicts with previous mitogenomic phylogenetic studies, in which a sister relationship was found between Clupeiformes and Gonorynchiformes. We discussed the possible causes of this incongruence. Within the Gonorynchiformes, the following original topology was found: (Gonorynchus (Chanos (Phractolaemus (Cromeria (Grasseichthys (Kneria, Parakneria)))))). We conWrmed that the paedomorphic species Cromeria nilotica and Grasseichthys gabonensis belong to the family Kneriidae; however, the two species together did not form a monophyletic group. This result challenges the value of reductive or absent characters as synapomorphies in this group.  2005 Elsevier Inc. All rights reserved.

Keywords: Mitogenomics; Otocephala; Gonorynchiformes; Denticeps; Paedomorphosis; Long PCR

1. Introduction ynchiformes, and Otophysi (Inoue et al., 2001d, 2004; Lê et al., 1993; Lecointre and Nelson, 1996; Saitoh et al., Recent molecular phylogenetic studies provided 2003; Zaragüeta-Bagils et al., 2002). Ishiguro et al. (2003) strong support for a novel clade among basal teleosts gave additional molecular evidence for the unexpected named Otocephala, comprising Clupeiformes, Gonor- inclusion in this group of the Alepocephaloidei, a small suborder of deep-sea Wshes, assigned by Greenwood * Corresponding author. Fax: +81 3 5351 6822. and Rosen (1971) to the suborder Argentinoidei E-mail address: [email protected] (S. Lavoué). (Protacanthopterygii) (Fig. 1G). Altogether, these four

Fig. 1. Eight alternative phylogenetic hypotheses for basal euteleosts based on morphology (A–F), and molecule (G and H). Otocephalan lineages are indicated by boldface. Although Saitoh et al. (2003) and Inoue et al. (2004) did not include any representative of Alepocephaloidei in their studies, they obtained a similar intra-otocephalan phylogenetic hypotheses as Ishiguro et al. (2003) (hypothesis G). Tree H is a strict consensus of the two 50% majority rule consensus trees from our datasets #1 and #2 (see text for details). In trees G and F, numbers above internal branches indicate jack- knife values and Bayesian posterior probabilities, respectively. S. Lavoué et al. / Molecular Phylogenetics and Evolution 37 (2005) 165–177 167 otocephalan groups were never considered closely related 1996; Gayet, 1986), while the phylogenetic relationships based on morphology (Fink, 1984; Johnson and Patter- among its taxa are still discussed (Fink and Fink, 1996; son, 1996; Lauder and Liem, 1983; Patterson, 1994; Gayet, 1993; Grande, 1994; Grande and Poyato-Ariza, Rosen, 1982; Sanford, 1990) (Figs. 1A–F). As currently 1999; Greenwood et al., 1966; Howes, 1985; Lenglet, documented, the Otocephala encompasses about one 1974). Recurrent debates concern the relative position of quarter of the total Wsh species diversity, with more than the paedomorphic Wshes Cromeria and Grasseichthys 7000 species placed in 67 families and 1081 genera (some of the smallest vertebrates) (Grande, 1994; (Nelson, 1994). Howes, 1985; Lenglet, 1974), and the relative positions Within this group, the traditional view among researchers of Chanos and Gonorynchus as the most basal group supports the respective monophyly of Gonorynchiformes (Fink and Fink, 1981, 1996; Gayet, 1993; Grande and (DAnotophysi), Otophysi (DCharaciformes, Cypriniformes, Poyato-Ariza, 1999; Greenwood et al., 1966; Patterson, Siluriformes, and Gymnotiformes), Clupeiformes (D Clupeo- 1984) (Figs. 2A–E). morpha), and Alepocephaloidei (Begle, 1992; Fink and Fink, Recently, systematic studies based on whole mitoge- 1981, 1996; Grande, 1985), while recurrent discussions deal nome sequences have been shown to be appropriate for with their relative positions and, especially, the question of the resolving long-standing controversies and/or providing monophyly of the Ostariophysi, which includes Otophysi original phylogenetic hypotheses within Actinopterygii. and Gonorynchiformes (Greenwood et al., 1966; Gwo As part of a large-scale project to investigate the phylo- et al., 1995; Nelson, 1994; Roberts, 1973). Thus, the genetic relationships of Actinopterygii based on the order Gonorynchiformes has been considered as the comparison of whole mitogenome sequences (Inoue sister group of the Otophysi (making the Ostario- et al., 2001b,c, 2003a,b, 2004; Ishiguro et al., 2003; physi, sensu Rosen and Greenwood (1970), monophy- Miya et al., 2001, 2003; Saitoh et al., 2003), we address letic), or as the sister group of the Clupeiformes. The in this study the phylogenetic inter- and intrarelation- Wrst of these two hypotheses is more commonly ships of the order Gonorynchiformes, in the context of accepted. However, recent molecular studies based on the Otocephala. For this purpose, Wve new whole mitog- whole mitogenome sequences supported the second enome sequences of African gonorynchiforms have been hypothesis: the Gonorynchiformes are more closely determined, as well as that of Denticeps clupeoides, the related to Clupeiformes than to Otophysi (Inoue et al., single extant species from the clupeiform suborder Den- 2003a, 2004; Ishiguro et al., 2003; Saitoh et al., 2003). ticipitoidei (Table 1). Combining these data with the 34 To date, no other published molecular study using a previously published mitogenome sequences, we Wrst test diVerent set of characters focused on the question of the monophyly of the order Gonorynchiformes as pro- higher relationships among the Otocephala. In short, posed by Gosline (1960), and investigate its position we considered that the phylogenetic position of Gon- within the basal teleostean phylogeny. Second, we test orynchiformes relative to Otophysi, Clupeiformes, the currently accepted phylogenetic relationships among and Alepocephaloidei is still uncertain and deserves fur- representatives of Gonorynchiformes and we propose an ther studies. alternative phylogenetic view. Lastly, we discuss the evo- Gosline (1960) was the Wrst author to propose the lutionary origin of the paedomorphic nature of Crome- order Gonorynchiformes, but did not provide any for- ria nilotica and Grasseichthys gabonensis. mal diagnosis. Later, Greenwood et al. (1966) listed Wve anatomical characters shared by all extant representatives of this group. This group is represented by only 31 2. Materials and methods extant species (FishBase ver. September 2004: http:// www.Wshbase.org/search.cfm), belonging to four families 2.1. Taxonomic and character sampling (Nelson, 1994): Chanidae (one species, Chanos chanos), Gonorynchidae (Wve species belonging to one genus, A list of taxa examined in this study is provided in Gonorynchus), Phractolaemidae (one species, Phractolae- Table 1, along with DDBJ/EMBL/GenBank accession mus ansorgii), and Kneriidae (24 species belonging to numbers of the corresponding mitogenomic sequences. four genera: Cromeria, Grasseichthys, Kneria and Para- We determined the nucleotide sequences of six new kneria). The phractolaemid and the kneriids are endemic mitogenomes from Wve African gonorynchiform taxa to African freshwaters, while the chanids and the gono- (C. nilotica, G. gabonensis, Kneria sp., Parakneria camer- rynchids live in the Indian and PaciWc oceans. The fossil onensis, and P. ansorgii) and the only extant representa- record of the Gonorynchiformes goes back to the early tive of the clupeiform suborder Denticipitoidei Cretaceous and indicates that this group was once much (D. clupeoides). We combined these new mitogenomic more widespread than it is today. Although the order data with the 34 previously published data that we have Gonorynchiformes shows considerable morphological purposely selected from the basal teleosts: (1) All avail- diversity among the families, its monophyly is well able taxa from Otocephala. Thus, considering both new accepted (de Pinna and Grande, 2003; Fink and Fink, and published mitogenome sequences, our study 168 S. Lavoué et al. / Molecular Phylogenetics and Evolution 37 (2005) 165–177

Fig. 2. Six phylogenetic hypotheses for the Gonorynchiformes based on osteological data (A, B, D, and E), myological data (C), and molecule (F). Noted that Greenwood et al. (1966) (hypothesis A) did not examine the generic relationships within the family Kneriidae. Gayet’s (1993) phylogenetic hypothesis is similar to the hypotheses D and E. However, she did not examine generic relationships within the family Kneriidae. “*” indicates that we have excluded fossils from the original published phylogenetic hypotheses. included at least one representative of each major line- 2.2. Total DNA extraction age known within Otocephala. (2) We chose six taxa within Euteleostei which represent the sister group of the A portion of the epaxial musculature (ca. 0.25 g) was Otocephala, and six taxa within Elopomorpha which rep- excised from fresh specimens of each species and imme- resent the sister group of the clade (Otocephala+ Euteleo- diately preserved in 99.5% ethanol. Because of their stei). (3) Based on recent information (Inoue et al., 2003a, small size, the whole body of G. gabonensis and C. nilot- 2004), we chose three osteoglossomorpha taxa as collective ica were directly preserved in 99.5% ethanol. Total geno- outgroup to root our trees. This taxonomic sampling com- mic DNA was extracted using the Qiagen DNeasy tissue bines information from Saitoh et al. (2003), Ishiguro et al. kit following the manufacturer’s protocol. (2003), and Inoue et al. (2004), which allows direct comparisons with these studies and illustrates the beneWt of 2.3. Mitochondrial DNA puriWcation by long PCR increasing taxonomic sampling. The voucher specimens for tissues used in this study First, the mitogenomes of the Wve gonorynchiforms are deposited in the Museum National d’Histoire Natu- and D. clupeoides were ampliWed in their entirety using a relle, Paris (G. gabonensis, voucher accession number long PCR technique (Cheng et al., 1994). The following MNHN 2004-1691), the Cornell University Museum of Wve Wsh-versatile long PCR primers were used in various Vertebrates (P. cameronensis CU 90377), and the combinations to amplify the entire mitogenome in two National Science Museum, Tokyo (D. clupeoides NSMT- or three reactions: L3074-16S (5Ј-CGA TTA AAG TCC P68224 and P. ansorgii NSMT-P68225). Voucher speci- TAC GTG ATC TGA GTT CAG-3Ј), L12321-leu (5Ј- mens of C. nilotica and Kneria sp. are under the care of GGT CTT AGG AAC CAA AAA CTC TTG GTG Joachim Schwahn (private/personal collection) and Guil- CAA-3Ј), L2508-16S (5Ј-CTC GGC AAA CAT AAG lermo Ortí (University of Nebraska, voucher number CCT CGC CTG TTT ACC AAA AAC-3Ј), H2990-16S G.O. 194), respectively. (5Ј-TGC ACC ATT RGG ATG TCC TGA TCC AAC S. Lavoué et al. / Molecular Phylogenetics and Evolution 37 (2005) 165–177 169

Table 1 List of species examined in this study ClassiWcation Species Accession Nos. Reference Osteoglossomorpha Family Hiodontidae Hiodon alosoides (RaWnesque) AP004356 Inoue et al. (2003a) Family Osteoglossidae Osteoglossum bicirrhosum (Cuvier) AB043025 Inoue et al. (2001d) Family Pantodontidae Pantodon buchholzi Peters AB043068 Inoue et al. (2001d) Elopomorpha Family Elopidae Elops hawaiensis Regan AB051070 Inoue et al. (2004) Family Megalopidae Megalops atlanticus Valenciennes AP004808 Inoue et al. (2004) Family Notacanthidae Notacanthus chemnitzi Bloch AP002975 Inoue et al. (2003a) Family Anguillidae Anguilla japonica Temminck & Schlegel AB038556 Inoue et al. (2001a) Family Muraenidae Gymnothorax kidako (Temminck & Schlegel) AP002976 Inoue et al. (2003a) Family Congridae Conger myriaster (Brevoort) AB038381 Inoue et al. (2001b) Otocephala Order Clupeiformes ( D Clupeomorpha) Family Denticipitidae Denticeps clupeoides Clausen AF007276 This study Family Clupeidae Sardinops melanostictus (Jenyns) AB032554 Inoue et al. (2000) Family Engraulidae Engraulis Japonicus Temminck & Schlegel AB040676 Inoue et al. (2001c) Order Gonorynchiformes ( D Anotophysi) Suborder Chanoidei Family Chanidae Chanos chanos (Forsskål) AB054133 Saitoh et al. (2003) Suborder Gonorynchoidei Family Gonorynchidae Gonorynchus greyi (Richardson) AB054134 Ishiguro et al. (2003) Family Phractolaemidae Phractolaemus ansorgii Boulenger AF007280 This study Family Kneriidae Kneria sp. AF007278 This study Parakneria cameronensis (Boulenger) AF007279 This study Cromeria nilotica Boulenger AF007275 This study Grasseichthys gabonensis Géry AF007277 This study Order Cypriniformes Family Cyprinidae Cyprinus carpio Linnaeus X61010 Chang et al. (1994) Carassius auratus (Linnaeus) AB006953 Murakami et al. (1998) Sarcocheilichthys variegatus (Temminck & Schlegel) AB054124 Saitoh et al. (2003) Danio rerio (Hamilton) AC024175 Broughton et al. (2001) Family Cobitidae Cobitis striata Ikeda AB054125 Saitoh et al. (2003) Lefua echigonia Jordan & Richardson AB054126 Saitoh et al. (2003) Family Balitoridae Crossostoma lacustre Steindachner M91245 Tzeng et al. (1992) Order Characiformes Family Characidae Phenacogrammus interruptus (Boulenger) AB054129 Saitoh et al. (2003) Chalceus macrolepidotus Cuvier AB054130 Saitoh et al. (2003) Order Gymnotiformes Family Sternopigidae Eigenmania virescens (Valenciennes) AB054131 Saitoh et al. (2003) Family Apteronotidae Apteronotus albifrons (Linnaeus) AB054132 Saitoh et al. (2003) Order Siluriformes Family Bagridae Pseudobagrus tokiensis Döderlein AB054127 Saitoh et al. (2003) Family Callicthyidae Corydoras rabauti La Monte AB054128 Saitoh et al. (2003) Alepocephaloidea Family Alepocephalidae Alepocephalus tenebrosus Gilbert AP004100 Ishiguro et al. (2003) Family Platytroctidae Platytroctes apus Günther AP004107 Ishiguro et al. (2003) Protacanthopterygii Family Salmonidae Oncorhynchus tshawytscha (Walbaum) AF392054 Wilhelm et al. (2003) Coregonus lavaretus (Linnaeus) AB034824 Miya and Nishida (2000) Family Umbridae Dallia pectoralis Bean AP004102 Ishiguro et al. (2003) Family Esocidae Esox lucius Linnaeus AP004103 Ishiguro et al. (2003) Neoteleostei Family Gadidae Gadus morhua Linnaeus X99772 Johansen and Bakke (1996) Family Paralichthyidae Paralichthys olivaceus (Temminck & Schlegel) AB028664 Saitoh et al. (2000) ClassiWcation follows Nelson (1994) with the exception of Otocephala, which groups Clupeiformes, Gonorynchiformes, Otophysi (i.e., Cyprinifor- mes, Characiformes, Siluriformes, and Gymnotiformes) and Alepocephaloidei (Ishiguro et al., 2003). Within the order Gonorynchiformes, the intra- ordinal classiWcation is indicated. 170 S. Lavoué et al. / Molecular Phylogenetics and Evolution 37 (2005) 165–177

ATC-3Ј), H12293-leu (5Ј-TTG CAC CAA GAG TTT 2.5. Phylogenetic analysis TTG GTT CCT AAG ACC-3Ј), and H1065-12S (5Ј- GGC ATA GTG GGG TAT CTA ATC CCA GTT When using a mitogenomic data set and for a particu- TGT-3Ј) (for locations of these primers, see Inoue et al., lar taxonomic sampling, Simmons and Miya (2004) 2001d). The long-PCR products were diluted with TE empirically demonstrated that Bayesian analysis is the buVer (1:15) for subsequent use as templates for the most eYcient method for accurately reconstructing phy- short PCR. logeny. Following their recommendations, we used this A total of 89 Wsh-versatile PCR primers was used in method for two diVerent character matrices. The Wrst various combinations to amplify short (<1500 bp), con- matrix (dataset #1, 10,395 positions) includes concate- tiguous, overlapping segments of the entire mitogenome nated nucleotide sequences from 12 protein-coding for each of the six species (primer locations and genes except the third codon positions (7260 positions), sequences of these primers are available upon request to 22 transfer RNA genes (1104 positions) and the two S.L.). In cases where some versatile primers did not ribosomal RNA genes (2031 positions). The second work, we designed new species-speciWc PCR primers. (dataset #2, 14,025 positions) includes the same set of Long PCR and subsequent short PCR were carried out characters plus the transversions at the third codon as previously described (Inoue et al., 2003a; Miya and positions of protein-coding genes, replacing the purines Nishida, 1999). (C/T) with Y, and the pyrimidines (A/G) with R (Phillips Double-stranded DNA products were Wrst puriWed et al., 2004). using an ExoSap enzyme reaction, before being used as a For the two matrices (dataset #1 and #2), we set four template for direct cycle sequencing with dye-labeled ter- partitions (Wrst and second codon positions of protein- minators (Applied Biosystems). Sequencing primers used coding genes, tRNA genes and rRNA genes), and Wve were the same as those used for PCR. All sequencing partitions (same four as previous plus the transversions reactions were performed according to the manufac- at the third codon positions of protein-coding genes), turer’s instructions. Labeled fragments were run on a respectively. Model 3100 DNA automated sequencer (Applied Bio- Partitioned Bayesian phylogenetic analyses were con- systems). ducted with MrBayes 3.0b4 (Huelsenbeck and Ronquist, 2001; Ronquist and Huelsenbeck, 2003). We used the 2.4. Sequences editing and alignment general time reversible model with some sites assumed to be invariable and with variable sites assumed to follow a The sequence electropherograms were edited with discrete gamma distribution (GTR + I + ; (Yang, EditView version 1.0.1 (Applied Biosystems). Sequen- 1994)), as it was selected as the best-Wtting model with cher software package version 4.1.2 (Gene Codes, Ann Modeltest version 3.06 (Posada and Crandall, 1998). We Arbor, MI) and DNASIS version 3.2 (Hitachi Software set the parameters in MrBayes as follows: “lset nst D 6” Engineering) were used to concatenate the consensus (GTR), “rates D invgamma” (I + ), “unlink” (unlinking mitogenomic sequences. Then, sequences were of model parameters across partitions), and “prset exported to phylogenetic software programs. To build ratepr D variable” (rate multiplier variable across parti- our character matrix, we combined the six newly deter- tions). We used the default settings for the prior on the mined sequences with 34 previously published gamma shape parameter (0.1–50.0) and the proportion sequences. of invariable site (0–1). A Dirichlet distribution was For each individual protein-coding gene, we manually assumed for the base frequency parameters and the rate aligned the sequences for the 40 species, with respect to matrix, and uninformative prior was used for the topol- the translated amino acid sequence. All stop codons ogy. For each matrix, two independent Bayesian analy- were excluded from the subsequent phylogenetic analyses were performed. The Markov chain Monte Carlo ses, as well as ambiguous alignment stretches at the 5Ј (MCMC) process was set so that four chains (three and 3Ј ends for several protein-coding genes. The ND6 heated and one cold) ran simultaneously. gene was also excluded because of its heterogeneous base On the basis of preliminary runs with varying cycles composition and consistently poor phylogenetic perfor- (0.3–6.0 £ 106), we estimated average log likelihood mance (Miya and Nishida, 2000). The 12S and 16S scores at stationarity (dataset #1 D¡118,398.37; dataset rRNA sequences, as well as the 22 tRNA genes were #2 D¡180,837.43). After reaching stationarity in the two aligned using the software Proalign ver. 0.5 (Löytynoja independent runs, we continued the runs for 2.5 £ 106 and Milinkovitch, 2003), and default setting parameters. cycles (25,000 trees) to conWrm lack of improvement in Regions with posterior probabilities of 650% were the likelihood scores. Parameter values and trees were excluded from the subsequent phylogenetic analyses. sampled every 100 generations. For each dataset, 50% Our complete data matrix includes 14,025 positions, of majority-rule consensus trees were calculated using which 7388 were variable and 6151 informative under PAUP* (SwoVord, 2002), from the 50,000 trees pooled the parsimony criterion. from the two runs. S. Lavoué et al. / Molecular Phylogenetics and Evolution 37 (2005) 165–177 171

2.6. Testing alternative topologies Maddison, 2000). We estimated maximum-likelihood (ML) tree with those constraints using a GTR + I + Alternative tree topologies were individually com- model of sequence evolution as implemented in PAUP. pared to the resulting Bayesian tree (Fig. 3) using the Then, we estimated the variance in likelihood diVerence likelihood-based SH test (Shimodaira and Hasegawa, between two topologies using the resampling estimated 1999) as implemented in PAUP. We Wrst created the con- log-likelihood (RELL) method from 10,000 bootstrap straint topologies corresponding to previous phyloge- replications and the diVerence was statistically netic hypotheses using MacClade (Maddison and evaluated.

Fig. 3. The 50% majority rule consensus tree of the 50,000 pooled trees from the two independent Bayesian analyses of the dataset #1 (concatenation of 12 protein-coding genes minus the third codon positions, 22 tRNA genes and 12S and 16S rRNA) from 40 teleosts. Three osteoglossomorphs taxa were chosen to root our tree. We set four partitions (Wrst and second codon positions of protein-coding genes, tRNA gene, and rRNA genes for this data set) and unlinked each partition. Numbers above internal branches indicate Bayesian posterior probabilities (shown as percentages). Bayesian analysis using the dataset #2 (third codon positions converted into purine and pyrimidine; RY-coding) recovered a similar tree, with a topological diVerence being found in one internal branch (arrowheads). 172 S. Lavoué et al. / Molecular Phylogenetics and Evolution 37 (2005) 165–177

3. Results was the sister group of the remaining otocephalans, although the position of the Alepocephaloidei did not 3.1. Mitochondrial genome organisation receive strong support. Within the monophyletic order Gonorynchiformes, Mitogenome size of the six newly determined species Gonorynchus greyi was the sister group to the remaining varied from 16,511 (P. ansorgii) to 17,156 bp (D. clupeo- gonorynchiforms. Then, C. chanos was the sister group ides). The genome content and gene order of the six new of African freshwater gonorynchiforms. Within this lat- are typical as found in most other vertebrate examined. ter group, P. ansorgii (family Phractolaemidae) was the These mitochondrial genomes include 12S and 16S sister group of the family Kneriidae, which includes, in rRNAs, 22 tRNA, 13 protein-coding genes, and the con- addition to Kneria and Parakneria, the two paedomor- trol region. The H-strand codes for most of these genes, phic Wshes, C. nilotica and G. gabonensis. Inferred intra- except the ND6 and eight tRNA genes, which are coded relationships within Kneriidae are original in that on the L-strand. C. nilotica is found as the most basal group, G. gabonensis being the sister group of Kneria and Parakneria. 3.2. Phylogenetic analysis Bayesian analysis of the mitogenomic data set including only the transversions at the third codon positions Bayesian analysis of the complete data set, excluding (dataset #2) recovered a fully resolved tree with a similar the third codon positions (dataset #1), produced a fully topology, in which Otocephala, Ostariophysi, and Gon- resolved and well-supported phylogenetic tree. Fig. 3 orynchiformes (although supported by a BPP of 65%), shows the 50% majority rule consensus of 50,000 com- African freshwater gonorynchiforms, and Otophysi were bined samples from two independent Bayesian analyses all monophyletic (data not shown). The only topological of this data set. In this tree, only three nodes were not diVerence was that Alepocephaloidei is placed as the sis- supported by Bayesian posterior probabilities (BPP) of ter group of the Clupeiformes, supported by a high BPP 100%: Elopomorpha (96%), the basal position of Crome- (100%). ria within the family Kneriidae (75%), and the associa- We used the SH test to statistically compare the topo- tion of Ostariophysi and Alepocephaloidei (83%). All logical congruence between previous phylogenetic well-supported morphology-based groups such as hypotheses with our present one. The diVerences in like- Osteoglossomorpha, Clupeocephala, Euteleostei, Prot- lihood score between our topology and all other topolo- acanthopterygii, and Neoteleostei (sensu Johnson and gies (Figs. 1 and 2) were signiWcant at the threshold of Patterson, 1996) were recovered as monophyletic with 0.05 (Table 2). Thus, from our dataset, we conWdently high support (BPP D 100%). The Otocephala (i.e., Gon- rejected all previous alternative topologies. However, it orynchiformes, Otophysi, Clupeiformes, and Alepoceph- should be noted that, when the alternative topologies are aloidei) also formed a well-supported monophyletic congruent with both the position of the Alepocephaloi- group. Within the Otocephala, the Gonorynchiformes dei within the Otocephala and the monophyly of the were monophyletic and, interestingly, sister group to the family Kneriidae (Fink and Fink, 1996; Grande and Otophysi. The Clupeiformes (Clupeoidei + Denticeps) Poyato-Ariza, 1999; Ishiguro et al., 2003), the diVerences

Table 2 Results of SH test (Shimodaira and Hasegawa, 1999) for comparisons of alternative phylogenetic hypotheses (Figs. 1 and 2) Alternative topologies tested Authors lnL SH test Within basal Teleostei (see Fig. 1) (Clup., ((Gono., Oto.), (Eso., (Alep., Salm.)), Neot.)) Rosen (1982)* 421.34 P <0.0001 (Clup., (Esoc., ((Gono., Oto.), Alep., (Salm., Neot.)))) Lauder and Liem (1983)* 450.96 P <0.0001 Same constraints as Lauder and Liem (1983) Fink (1984)* 450.96 P <0.0001 (Clup., (Esoc., ((Gono., Oto.), (Alep., Salm.), Neot.))) Sanford (1990)* 515.36 P <0.0001 ((Clup., (Gono., Oto.)), (Esoc., (Alep., Salm.), Neot.)) Patterson (1994)* 354.33 P <0.0001 ((Clup., (Gono., Oto.)), ((Alep., Salm), (Esoc. Neot.))) Johnson and Patterson (1996)* 343.89 P <0.0001 ((Oto., ((Gono., Clup.), Alep.)), ((Esoc., Salm.), Neot.)) Ishiguro et al. (2003)*32.53P D 0.023 Within Gonorynchiformes (see Fig. 2) ((Cro., Gra.), (Phr., Kne., Cha.), Gon.) Lenglet (1974) 1327.13 P <0.0001 ((Phr., Gra., (Cro., Kne.)), Cha., Gon.) Howes (1985) 499.48 P <0.0001 ((((Gra., Cor., Kne.), Phr.), Gon.), Cha.) Fink and Fink (1996) 31.74 P D 0.020 (((((Gra., Cro.), Kne.), Phr.), Gon.), Cha.) Grande and Poyato-Ariza (1999) 28.21 P D 0.038 Abbreviations: Clup., Clupeiformes; Gono., Gonorynchiformes; Oto., Otophysi, Eso., Esociformes; Alep., Alepocephaloidea; Salm., Salmoniformes; Neot., Neoteleostei; Cro., Cromeria; Gra., Grasseichthys; Kne., Kneria + Parakneria; Cha., Chanos; Gon. Gonorynchus. Asterisks indicate that Argen- tinoidea and Osmeroidei are removed from the tested topologies, because they are not included in our study. lnL is the diVerence in likelihood score between our topology (Fig. 3) and the corresponding tested topology. S. Lavoué et al. / Molecular Phylogenetics and Evolution 37 (2005) 165–177 173 in likelihood score were only moderate, with the result- 1929). However, the incongruent position of the Gonor- ing P values approaching the threshold of 0.05 (Table 2). ynchiformes between our study and previous mitogenomic studies requires further comments. It Wrst should be noted that, in two of the previous 4. Discussion mitogenomic studies, the relationship between Clupei- formes and Gonorynchiformes received only weak to As expected, all analyses recovered a well-supported moderate statistical support (bootstrap proportion monophyletic Otocephala, comprising Otophysi, Gonor- (BP) D 59% in Fig. 4 of Saitoh et al. (2003); BPP D 87% ynchiformes, Clupeiformes, and Alepocephaloidei. in Fig. 4 of Inoue et al. (2004)). Moreover, Inoue et al. Inferred phylogenetic relationships among these oto- (2004) noted that, when the quickly saturated third cephalan taxa as well as those within the order Gonor- codon positions are excluded from their analysis, the ynchiformes will be discussed in more detail in Sections Gonorynchiformes are closer to Otophysi than to Clupe- 4.1–4.4. iformes. Saitoh et al. (2003) noted that, when they ana- Outside of Otocephala, our inferred phylogenetic lyzed their dataset by maximum-parsimony (MP), the relationships are in complete agreement with the current Gonorynchiformes was rendered paraphyletic. Only classiWcation. Using the osteoglossomorph taxa as the Ishiguro et al. (2003) reported a stable topology (MP outgroup, the elopomorphs form a clade, which is the and maximum-likelihood (ML) trees), supported by a sister group of Clupeocephala. Within Clupeocephala, strong BP (100%) and a high decay index (26) in the MP the Euteleostei of Johnson and Patterson (1996) (minus tree (see their Fig. 2). Interestingly, Ishiguro et al.’s the Alepocephaloidei) is the sister group of Otocephala. (2003) taxonomic sampling within the Gonorynchifor- Within Euteleostei, the Protacanthopterygii (represented mes is the sparsest, with only Gonorynchus included, by Esociformes and Salmoniformes in our study) are the which exhibited the longest terminal branch within oto- sister group of the Neoteleostei. cephalan taxa (see their Fig. 2). Given that we used a character set derived from com- 4.1. Monophyly and phylogenetic position of the plete mitogenome sequences, we excluded the hypothesis Gonorynchiformes that the diVerent topologies inferred are the result of diVerent gene history. We also excluded any alignment Within Otocephala, our study supports the monophyly artifacts since when we re-analyzed our dataset with the of Gonorynchiformes, as well as those of Alepocephaloi- same otocephalan sampling set as before, by manually dei, Clupeiformes (Denticeps+ Clupeoidei), Otophysi excluding extra-taxa, we obtained a similar topology (Cypriniformes, Characiformes, Siluriformes, and Gym- (data not shown). There remain two major factors that notiformes), and Ostariophysi (Gonorynchiformes + could have an eVect on our results: (1) the taxonomic Otophysi). Although the order Gonorynchiformes has sampling analyzed, and (2) the phylogenetic method been only recently proposed (Gosline, 1960) and diag- used. nosed (Greenwood et al., 1966), little doubt persists about First, in each of these studies, three of the major oto- the monophyly of its living taxa (de Pinna and Grande, cephalan groups were missing: Denticeps, the African 2003; Fink and Fink, 1981, 1996; Grande and Poyato- gonorynchiform clade and Alepocephaloidei in Saitoh Ariza, 1999). Grande and Poyato-Ariza (1999) provided a et al. (2003) and Inoue et al. (2004); Denticeps, Chanos, list of 12 morpho-anatomical synapomorphies to support and African gonorynchiforms in Ishiguro et al. (2003). A the monophyly of the extant gonorynchiforms. series of recent papers examined the advantages of increased taxonomic sampling to the reliability of phylo- 4.2. Taxonomic sampling and phylogenetic reconstruction genetic inference (Hillis, 1998; Hillis et al., 2003; Pollock method et al., 2002; Rannala et al., 1998; Zwickl and Hillis, 2002). One widely recognized beneWt of adding taxa is to break In this analysis, the phylogenetic placement of Gon- up long branches and, therefore, minimize the phenome- orynchiformes diVers from three recent studies using the non of long-branch attraction (LBA) (Graybeal, 1998). same set of characters (complete mitogenome sequence), Miya et al. (2001) already demonstrated that the resolu- in which a close relationship between this group and the tion of higher-level relationships of teleosts based on Clupeiformes was hypothesized (Inoue et al., 2004; mitogenomic data takes advantage of purposefully Ishiguro et al., 2003; Saitoh et al., 2003), making the extensive taxonomic sampling so as to break up possible Ostariophysi (sensu Rosen and Greenwood, 1970) non- long branches. From the previous mitogenomic studies, monophyletic. Such result was, by itself, not surprising we signiWcantly increased the taxonomic sampling because a sister-relationship between Clupeiformes and within Otocephala, including representatives of all of the Gonorynchiformes has been occasionally hypothesized major lineages known within this group. Two of them based on spermatozoid ultrastructure characters (Gwo (Denticeps and the African gonorynchiforms) were never et al., 1995) or anatomical data (Nelson, 1994; Regan, sampled in any of the previous molecular studies. The 174 S. Lavoué et al. / Molecular Phylogenetics and Evolution 37 (2005) 165–177 addition of D. clupeoides seems to eYciently bisect the Gonorynchiformes ( D suborder Gonorynchoidei). The long branch of the Clupeoidei and the African freshwa- monophyly of the Gonorynchoidei was supported by at ter gonorynchiform taxa break up the long branch of the least nine unambiguous morphological synapomorphies. remaining Gonorynchiformes (Chanos and Gonoryn- Gonorynchus was the sister group of the well-deWned chus), reducing the attraction artifacts between the two. clade formed by the African endemic freshwater gono- Second, these previous studies were based on diVerent rynchiforms (Phractolaemus, Cromeria, Kneria, Grass- phylogenetic reconstruction methods: MP and ML for eichthys, and Parakneria). Ishiguro et al. (2003) and Bayesian analysis for Inoue et Our result diVers from these recent hypotheses in al. (2004). The respective advantages of various methods placing Gonorynchus instead of Chanos as the most basal of phylogenetic inference have also been largely debated, group among the Gonorynchiformes. Chanos is the sister with sometimes conXicting conclusions reached. How- group of the African freshwater gonorynchiform clade. ever, a commonly accepted conclusion is that Bayesian Our hypothesis is congruent with the one proposed by analysis is less sensitive to LBA than some other meth- Greenwood et al. (1966) (Fig. 2A), although these ods, including equally weighted parsimony (Beaumont authors did not present any uniquely derived characters and Rannala, 2004). In addition, Simmons and Miya to support the basal position of Gonorynchus. (2004) in an empirical study demonstrated the superior eYciency of Bayesian analysis providing a more reliable 4.4. Intra-relationships among African freshwater phylogenetic estimate from mitogenomic data. It should gonorynchiform Wshes be noted that, although Inoue et al. (2004) used an incomplete taxonomic sampling among otocephalan, Grasseichthys gabonensis is the most reductive species they obtained a similar intra-otocephalan topology of the Gonorynchiformes. It is reported to reach a maxi- based on Bayesian analysis of mitogenomic data set mum standard length of only 20 mm. Its body is gracile, when they excluded the quickly saturated third codon translucent, and scaleless. The eyes are large. The myo- positions (see their legend in Fig. 4). meres are visible and its skeleton is poorly ossiWed In conclusion, reconstruction artefacts such as the (Grande, 1994). In some ways, adults of Grasseichthys phenomenon of LBA are issues that could conceivably look like larval stages of Chanos or Gonorynchus, which bear upon the quality of molecular systematic studies, probably represents a case of extreme progenetic devel- and should explain the previous relationship inferred opmental truncation within Teleostei (Gould, 1977; between Gonorynchiformes and Clupeiformes. In our Johnson and Brothers, 1993). C. nilotica matures at study, to minimize its eVect and provide a more reliable about 27 mm, and shows similar developmental trunca- phylogenetic inference, we estimated the phylogeny tion to Grasseichthys, although little less pronounced. In using Bayesian analysis from a denser taxonomic sam- part because their paedomorphic features makes it diY- pling within Clupeiformes and Gonorynchiformes. Con- cult to establish their phylogenetic aYnities, the system- sequently, the Ostariophysi are found monophyletic. atics of Cromeria and Grasseichthys have been debated Our study did not provide more support to the rela- for a long time, and their phylogenetic placement is still tive phylogenetic positions among Clupeiformes, Alepo- uncertain. Myers (1938) suggested that Cromeria was a cephaloidea, and Ostariophysi than Ishiguro et al. larval stage of some species of Kneria (quoted by Green- (2003). The resolution of the relationships among these wood et al., 1966). In his description of Grasseichthys, three groups will need further investigation. Inclusion of Géry (1964) placed it within Clupeiformes. Lenglet additional basal alepocephaloid taxa, as well as addi- (1974) grouped Cromeria and Grasseichthys together in tional independent molecular characters, will be helpful the suborder Cromeroidei within Gonorynchiformes, to conWdently resolve this question. but they were distantly related to the family Kneriidae. Grande (1994) was unable to provide a single, unambig- 4.3. Intra-relationships within Gonorynchiformes uous phylogenetic hypothesis for these two taxa, but proposed two hypotheses depending on characters opti- Although the relative positions of Chanos, Gonoryn- mization. Fink and Fink (1996) left their position sedis chus, and the African freshwater gonorynchiform taxa mutabilis within the family Kneriidae. were previously debated (Greenwood et al., 1966; Len- Phylogenetic reconstruction of paedomorphic taxa glet, 1974), recent systematic studies based on osteologi- based on morphological data, which exhibit several cal data seemed to reach a consensus (Fink and Fink, reductive or absence of characters poses recurrent prob- 1981, 1996; Grande and Poyato-Ariza, 1999; Patterson, lems in systematics (Begle, 1991). Absent or reductive 1984) (Figs. 2D and E). In the most comprehensive study structures in these taxa make it diYcult to establish to date, based on parsimony analysis of a matrix of 94 homology with other non-reductive taxa, limiting the anatomical characters and 20 taxa (extant and fossil), number of available characters. Thus, for example Grande and Poyato-Ariza (1999) placed Chanos within Teleostei, the phylogenetic positions of extreme ( D suborder Chanoidei) as the sister-group of the rest of paedomorphic Wshes like Schindleria (Gobioidei) S. Lavoué et al. / Molecular Phylogenetics and Evolution 37 (2005) 165–177 175

(Johnson and Brothers, 1993) and Sundasalanx (Suda- ples, S.L. thanks Carl D. Hopkins, John P. Sullivan, and salangidae) (Siebert, 1997) have been longstanding con- John P. Friel (Cornell University), Didier Paugy and troversies. Therefore, to overcome the problem, it is Jean François Agnèse (Institut de Recherche pour le necessary to examine non-reductive characters, identiW- Développement), Melanie Stiassny (American Museum able early in ontogeny, as well as molecular characters. of Natural History), and André Kamdem Toham In all our analyses, the African freshwater gonor- (WWF-International). We also thank Masanori Naka- ynchiform Wshes form a well-supported monophyletic tani, Mitsugu M. Yamauchi, Kohji Mabuchi, and group. This result is completely consistent with the cur- Yusuke Yamanoue for their assistance in the technical rent hypothesis based on osteological data (Fink and work and Shannon De Vaney for English editing. This Fink, 1996; Gayet, 1993; Grande and Poyato-Ariza, study was supported by Research Grants Nos. 1999) and myological data (Howes, 1985). Within this 12NP0201, 15380131, 15570090, and 15-3601 from the clade, P. ansorgii (family Phractolaemidae) is unambigu- Japan Society for the Promotion of Science. S.L. was ously placed as the sister group of the family Kneriidae. supported by a Postdoctoral Fellowship (No. 3601) of So, we reject the hypothesis of the non-monophyly of the Japan Society for the Promotion of Science (JSPS). Kneriidae with respect to Phractolaemidae (Howes, 1985) (Fig. 2C) or Chanidae/Phractolaemidae (Lenglet, 1974) (Fig. 2B): C. nilotica and G. gabonensis being close Appendix A. Supplementary data to Kneria and Parakneria. In accord with the current classiWcation, Parakneria and Kneria compose a mono- Supplementary data associated with this article can phyletic group, with several osteological synapomor- be found, in the online version, at doi:10.1016/ phies supporting this relationship (Grande, 1994). j.ympev.2005.03.024. However, C. nilotica and G. gabonensis did not form a monophyletic group as previously proposed (Grande and Poyato-Ariza, 1999), G. gabonensis and C. nilotica References were regarded as the sequenced sister groups to the clade (Kneria + Parakneria). According to Grande (1994) and Beaumont, M.A., Rannala, B., 2004. The Bayesian revolution in genet- Grande and Poyato-Ariza (1999), C. nilotica and ics. Nat. Rev. Genet. 5, 251–261. Begle, D.P., 1991. Relationships of the osmeroid Wshes and the use of G. gabonensis form a monophyletic group, sharing these reductive characters in phylogenetic analysis. Syst. Zool. 40, 33–53. three derived characters: loss of scales, reduction in cra- Begle, D.P., 1992. Monophyly and relationships of the argentinoid nial ossiWcation and separated frontals. However, as the Wshes. Copeia 1992, 350–366. authors noted, these characters are reductive or absent Broughton, R.E., Milam, J.E., Roe, B.A., 2001. The complete sequence states. Interestingly, our study does not provide more of the zebraWsh (Danio rerio) mitochondrial genome and evolutionary patterns in vertebrate mitochondrial DNA. Genome Res. 11, support for the alternative phylogenetic hypothesis dis- 1958–1967. cussed by Grande (1994). If one considers that these Chang, Y.S., Huang, F.L., Lo, T.B., 1994. The complete nucleotide absent or reductive characters are not independent but sequence and gene organization of carp (Cyprinus carpio) mito- rather the consequence of one unique paedomorphic chondrial genome. J. Mol. Evol. 38, 138–155. event, Grande (1994) suggested that Cromeria and Cheng, S., Higuchi, R., Stoneking, M., 1994. Complete mitochondrial genome ampliWcation. Nat. Genet. 7, 350–351. (Parakneria + Kneria) could form a monophyletic group, de Pinna, M., Grande, T., 2003. Ontogeny of the accessory neural arch which would be supported by at least one synapomor- in pristigasteroid clupeomorphs and its bearing on the homology of phy, the special ventral inclination of the vomer. Instead the otophysan claustrum (Teleostei). Copeia 2003, 838–845. of that, our data supports a third hypothesis, never pro- Fink, S.V., Fink, W.L., 1981. Interrelationships of the ostariophysan posed before, in which Grasseichthys is the sister group Wshes (Teleostei). J. Linn. Soc. (Zool.) 72, 297–353. Fink, S.V., Fink, W.L., 1996. Interrelationships of ostariophysan Wshes (Tele- of (Kneria + Parakneria). Cromeria is the sister group of ostei). In: Stiassny, M.L.J., Parenti, L.R., Johnson, G.D. (Eds.), Interrela- these three taxa. Finally, this result leads us to the ques- tionships of Fishes. Academic Press, San Diego, CA, pp. 209–245. tion of the importance of reductive or absent characters Fink, W.L., 1984. Basal euteleots: relationships. In: Moser, H.G., Rich- as synapomorphies in the systematic of these Wshes as ardson, W.J., Cohen, D.M., Fahay, M.P., Kendall, Jr., A.W., Rich- suggested by Begle (1991). ards, S.L. (Eds.), Ontogeny and Systematics of Fishes. Am. Soc. Ichthyol. Herpetol. Spec. Publ., Lawrence, KS, pp. 202–206. Gayet, M., 1986. Ramallichthys, Gayet du Cémomanien inférieur marin de Ramallah (Judée), une introduction aux relations phylogéné- Acknowledgments tiques des Ostariophysi. Mém. Mus. Nat. d’Hist. Nat., Paris, série C 1, 21–31. For providing tissue samples for Kneria sp. and Gayet, M., 1993. Relations phylogénétiques des Gonorynchiformes (Ostariophysi). Belg. J. Zool. 123, 165–192. Cromeria nilotica, we thank Guillermo Ortí (University Géry, J., 1964. Une nouvelle famille de poissons dulcaquicoles africains; of Nebraska) and Joachim Schawhn (ArCona), respec- les Grasseichthyidae. C.R. Acad. Sci. Ser. III-Sci. 259, 4805–4807. tively. For providing assistance during Weld trips in Gosline, W.A., 1960. Contribution toward a classiWcation of modern Gabon, Congo, and Bénin, and help in collecting sam- isopondylous Wshes. Bull. Br. Mus. (Nat. Hist.) Zool. 6, 325–365. 176 S. Lavoué et al. / Molecular Phylogenetics and Evolution 37 (2005) 165–177

Gould, S.J., 1977. Ontogeny and Phylogeny. The Belknap Press of Har- Johnson, G.D., Brothers, E.B., 1993. Schindleria—a paedomorphic vard University Press, Cambridge. goby (Teleostei, Gobioidei). Bull. Mar. Sci. 52, 441–471. Grande, L., 1985. Recent and fossil clupeomorph Wshes with materials Johnson, G.D., Patterson, C., 1996. Relationships of lower euteleostean for revision of the subgroups of clupeoids. Bull. Am. Mus. Nat. Wshes. In: Stiassny, M.L.J., Parenti, L.R., Johnson, G.D. (Eds.), Hist. 181, 231–372. Interrelationships of Fishes. Academic Press, San Diego, CA, pp. Grande, T., 1994. Phylogeny and paedomorphosis in an African family 251–332. of freshwater Wshes (Gonorynchiformes: Kneriidae). Fieldiana 78, Lauder, G.V., Liem, K.F., 1983. The evolution and interrelationships of 1–20. the actinopterygian Wshes. Bull. Mus. Comp. Zool. 150, 95–197. Grande, T., Poyato-Ariza, F., 1999. Phylogenetic relationships of fossil Lê, H.L.V., Lecointre, G., Perasso, R., 1993. A 28S rRNA-based phy- and recent gonorynchiform Wshes (Teleostei: Ostariophysi). Zool. J. logeny of the gnathostomes: Wrst steps in the analysis of conXict Linn. Soc. 125, 197–238. and congruence with morphologically based cladograms. Mol. Graybeal, A., 1998. Is it better to add taxa or characters to a diYcult Phylogenet. Evol. 2, 31–51. phylogenetic problem?. Syst. Biol. 47, 9–17. Lecointre, G., Nelson, G.J., 1996. Clupeiformes, sister-group of Ostar- Greenwood, P.H., Rosen, D.E., 1971. Notes on the structure and rela- iophysi. In: Stiassny, M.L.J., Parenti, L.R., Johnson, G.D. (Eds.), tionships of the alepocephaloid Wshes. Am. Mus. Novit. 2473, 1–41. Interrelationships of Fishes. Academic Press, San Diego, CA, pp. Greenwood, P.H., Rosen, D.E., Weitzman, S.H., Myers, G.S., 1966. 193–207. Phyletic studies of teleostean Wshes, with a provisional classiWcation Lenglet, G., 1974. Contribution à l’étude ostéologique des Kneriidae. of living forms. Bull. Am. Mus. Nat. Hist. 131, 339–456. Ann. Soc. R. Zool. Belgique 22, 52–103. Gwo, J.C., Kao, Y.S., Lin, X.W., Chang, S.L., Su, M.S., 1995. The ultra- Löytynoja, A., Milinkovitch, M.C., 2003. A hidden Markov model for structure of milkWsh, Chanos chanos (Forsskål), spermatozoon progressive multiple alignment. Bioinformatics 19, 1505–1513. (Teleostei, Gonorynchiformes, Chanidae). J. Submicrosc. Cytol. Maddison, D.R., Maddison, W.P., 2000. MacClade 4: Analysis of Phy- Pathol. 27, 99–104. logeny and Character Evolution, Version 4.03. Sinauer, Sunder- Hillis, D.M., 1998. Taxonomic sampling, phylogenetic accuracy, and land, MA. investigator bias. Syst. Biol. 47, 3–8. Miya, M., Nishida, M., 1999. Organization of the mitochondrial Hillis, D.M., Pollock, D.D., McGuire, J.A., Zwickl, D.J., 2003. Is sparse genome of a deep-sea Wsh, Gonostoma gracile (Teleostei : Stomiifor- taxon sampling a problem for phylogenetic inference?. Syst. Biol. mes): Wrst example of transfer RNA gene rearrangements in bony 52, 124–126. Wshes. Mar. Biotechnol. 1, 416–426. Howes, G.J., 1985. Cranial muscles of gonorynchiform Wshes, with Miya, M., Nishida, M., 2000. Use of mitogenomic information in tele- comments on generic relationships. Bull. Br. Mus. (Nat. Hist.) Zool. ostean molecular phylogenetics: a tree-based exploration under the 49, 273–303. maximum-parsimony optimality criterion. Mol. Phylogenet. Evol. Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference 17, 437–455. of phylogenetic trees. Bioinformatics 17, 754–755. Miya, M., Kawaguchi, A., Nishida, M., 2001. Mitogenomic exploration Inoue, J.G., Miya, M., Aoyama, J., Ishikawa, S., Tsukamoto, K., Nish- of higher teleostean phylogenies: a case study for moderate-scale ida, M., 2001a. Complete mitochondrial DNA sequence of the Jap- evolutionary genomics with 38 newly determined complete mito- anese eel Anguilla japonica. Fish. Sci. 67, 118–125. chondrial DNA sequences. Mol. Biol. Evol. 18, 1993–2009. Inoue, J.G., Miya, M., Tsukamoto, K., Nishida, M., 2000. Complete Miya, M., Takeshima, H., Endo, H., et al., 2003. Major patterns of mitochondrial DNA sequence of the Japanese sardine Sardinops higher teleostean phylogenies: a new perspective based on 100 com- melanostictus. Fish. Sci. 66, 924–932. plete mitochondrial DNA sequences. Mol. Phylogenet. Evol. 26, Inoue, J.G., Miya, M., Tsukamoto, K., Nishida, M., 2001b. Complete 121–138. mitochondrial DNA sequence of Conger myriaster (Teleostei: Murakami, M., Yamashita, Y., Fujitani, H., 1998. The complete Anguilliformes): novel gene order for vertebrate mitochondrial sequence of mitochondrial genome from a gynogenetic triploid genomes and the phylogenetic implications for Anguilliform fami- “ginbuna” (Carassius auratus langsdorW). Zool. Sci. 15, 335–337. lies. J. Mol. Evol. 52, 311–320. Myers, G.S., 1938. Fresh-water Wshes and West indian zoogeography. Inoue, J.G., Miya, M., Tsukamoto, K., Nishida, M., 2001c. Complete Ann. Rept. Smith. Inst. 1937, 339–364. mitochondrial DNA sequence of the Japanese anchovy Engraulis Nelson, J.S., 1994. Fishes of the World, third ed., Wiley, New York. japonicus. Fish. Sci. 67, 828–835. Patterson, C., 1984. Family Chanidae and other teleostean Wshes as liv- Inoue, J.G., Miya, M., Tsukamoto, K., Nishida, M., 2001d. A mitoge- ing fossils. In: Eldredge, N., Stanley, S.M. (Eds.), Living Fossils. nomic perspective on the basal teleostean phylogeny: resolving Springer, Berlin, pp. 132–139. higher-level relationships with longer DNA sequences. Mol. Phylo- Patterson, C., 1994. Bony Wshes. In: Prothero, D.R., Schoch, R.M. genet. Evol. 20, 275–285. (Eds.), Major Features of Vertebrate Evolution. Short Courses in Inoue, J.G., Miya, M., Tsukamoto, K., Nishida, M., 2003a. Basal actin- Paleontology, vol. 7. Paleontological Society, Knoxville, pp. 57–84. opterygian relationships: a mitogenomic perspective on the phylog- Phillips, M.J., Delsuc, F.D., Penny, D., 2004. Genome-scale phylogeny eny of the “ancient Wsh”. Mol. Phylogenet. Evol. 26, 110–120. and the detection of systematic biases. Mol. Biol. Evol. 21, 1455– Inoue, J.G., Miya, M., Tsukamoto, K., Nishida, M., 2003b. Evolution 1458. of the deep-sea gulper eel mitochondrial genomes: large-scale gene Pollock, D.D., Zwickl, D.J., McGuire, J.A., Hillis, D.M., 2002. rearrangements originated within the eels. Mol. Biol. Evol. 20, Increased taxon sampling is advantageous for phylogenetic infer- 1917–1924. ence. Syst. Biol. 51, 664–671. Inoue, J.G., Miya, M., Tsukamoto, K., Nishida, M., 2004. Mitogenomic Posada, D., Crandall, K.A., 1998. MODELTEST: testing the model of evidence for the monophyly of elopomorph Wshes (Teleostei) and DNA substitution. Bioinformatics 14, 817–818. the evolutionary origin of the leptocephalus larva. Mol. Phylogenet. Rannala, B., Huelsenbeck, J.P., Yang, Z.H., Nielsen, R., 1998. Taxon Evol. 32, 274–286. sampling and the accuracy of large phylogenies. Syst. Biol. 47, 702– Ishiguro, N.B., Miya, M., Nishida, M., 2003. Basal euteleostean rela- 710. tionships: a mitogenomic perspective on the phylogenetic reality of Regan, C.T., 1929. Fishes. In: Rfsaf, R. (Ed.), Encyclopaedia Britan- the “Protacanthopterygii”. Mol. Phylogenet. Evol. 27, 476–488. nica, 14th ed., vol. 9, London, New York, pp. 305–328. Johansen, S., Bakke, I., 1996. The complete mitochondrial DNA Roberts, T.R., 1973. Interrelationships of ostariophysans. In: Green- sequence of Atlantic cod (Gadus morhua): relevance to taxonomic wood, P.H., Miles, R.S., Patterson, C. (Eds.), Interrelationships of studies among codWshes. Mol. Mar. Biol. Biotechnol. 5, 203–214. Fishes. Academic Press, London, pp. 397–513. S. Lavoué et al. / Molecular Phylogenetics and Evolution 37 (2005) 165–177 177

Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phyloge- Simmons, M.P., Miya, M., 2004. EYciently resolving the basal clades of netic inference under mixed models. Bioinformatics 19, 1572– a phylogenetic tree using Bayesian and parsimony approaches: a 1574. case study using mitogenomic data from 100 higher teleost Wshes. Rosen, D.E., 1982. Teleostean interrelationships, morphological func- Mol. Phylogenet. Evol. 31, 351–362. tion and evolutionary inference. Am. Mus. Novit. 2827, 1–57. SwoVord, D.L., 2002. PAUP*. Phylogenetic Analysis Using Parsimony Rosen, D.E., Greenwood, P.H., 1970. Origin of the Weberian apparatus (* and Other Methods), version 4.0b10. Sinauer Associates, Sunder- and the relationships of the ostariophysan and gonorynchiform land, MA. Wshes. Am. Mus. Nov. 2428, 1–25. Tzeng, C.S., Hui, C.F., Shen, S.C., Huang, P.C., 1992. The complete Saitoh, K., Hayashizaki, K., Yokoyama, Y., Asahida, T., Toyohara, H., nucleotide-sequence of the Crossostoma lacustre mitochondrial Yamashita, Y., 2000. Complete nucleotide sequence of Japanese genome—conservation and variations among vertebrates. Nucleic Xounder (Paralichthys olivaceus) mitochondrial genome: structural Acids Res. 20, 4853–4858. properties and cue for resolving teleostean relationships. J. Hered. Wilhelm, V., Villegas, J., Miquel, A., Engel, E., Bernales, S., Valenzuela, 91, 271–278. P.D., Burzio, L.O., 2003. The complete sequence of the mitochon- Saitoh, K., Miya, M., Inoue, J.G., Ishiguro, N.B., Nishida, M., 2003. drial genome of the Chinook salmon, Oncorhynchus tshawytscha. Mitochondrial genomics of ostariophysan Wshes: perspectives on Biol. Res. 36, 223–231. phylogeny and biogeography. J. Mol. Evol. 56, 464–472. Yang, Z., 1994. Maximum likelihood phylogenetic estimation from Sanford, C.P.J., 1990. The phylogenetic relationships of Salmonoid DNA sequences with variable rates over sites: approximate meth- Wshes. Bull. Br. Mus. (Nat. Hist.) Zool. 56, 145–153. ods. J. Mol. Evol. 39, 306–314. Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of log-like- Zaragüeta-Bagils, R., Lavoué, S., Tillier, A., Bonillo, C., Lecointre, G., lihoods with applications to phylogenetic inference. Mol. Biol. 2002. Assessment of otocephalan and protacanthopterygian con- Evol. 16, 1114–1116. cepts in the light of multiple molecular phylogenies. C.R. Biol. 325, Siebert, D.J., 1997. Notes on the anatomy and relationships of Sundasa- 1191–1207. lanx Roberts (Teleostei, Clupeidae), with descriptions of four new Zwickl, D.J., Hillis, D.M., 2002. Increased taxon sampling greatly species from Borneo. Bull. Br. Mus. (Nat. Hist.) Zool. 63, 13–26. reduces phylogenetic error. Syst. Biol. 51, 588–598.