Molecular Biology and Evolution

MBE Advance Access originally published online on February 7, 2008
Molecular Biology and Evolution 2008 25(5):874-891; doi:10.1093/molbev/msn031

This Article Abstract FREE Full Text (PDF) Supplementary Data All Versions of this Article: 25/5/874    most recent msn031v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Request Permissions Google Scholar Articles by Kurabayashi, A. Articles by Hasegawa, M. Search for Related Content PubMed PubMed Citation Articles by Kurabayashi, A. Articles by Hasegawa, M. Social Bookmarking          What's this?

© The Authors 2008. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org

Research Articles

Phylogeny, Recombination, and Mechanisms of Stepwise Mitochondrial Genome Reorganization in Mantellid Frogs from Madagascar

Atsushi Kurabayashi^*, Masayuki Sumida^*, Hiromichi Yonekawa, Frank Glaw, Miguel Vences and Masami Hasegawa^||,¶

^* Institute for Amphibian Biology, Graduate School of Science, Hiroshima University, Hiroshima, Japan
Department of Laboratory Animal Science, The Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan
Zoologische Staatssammlung München, München, Germany
Zoological Institute, Technical University of Braunschweig, Spielmannstrasse 8, Braunschweig, Germany
^|| Institute of Statistical Mathematics, Tokyo, Japan
^¶ School of Life Sciences, Fudan University, Shanghai, China

E-mail: kuraba{at}hiroshima-u.ac.jp.

	Abstract

TOP Abstract Introduction Materials and Methods Results and Discussion Supplementary Material Acknowledgements References

 
In Malagasy frogs of the family Mantellidae, the genus Mantella is known to possess highly reorganized mitochondrial (mt) genomes with the following characteristics: 1) some rearranged gene positions, 2) 2 distinct genes and a pseudogene corresponding to the transfer RNA gene for methionine (trnM), and 3) 2 control regions (CRs) with almost identical nucleotide sequences. These unique genomic features were observed concentrated between the duplicated CRs surrounding cytochrome b (cob) and nicotinamide adenine dinucleotide dehydrogenase subunit 2 (cnad2) genes. To elucidate the mechanisms and evolutionary pathway that yielded the derived genome condition, we surveyed the reorganized genomic portion for all 12 mantellid genera. Our results show that the mt genomes of 7 genera retain the ancestral condition. In contrast, adding to Mantella, 4 genera of the subfamily Mantellinae, Blommersia, Guibemantis, Wakea, and Spinomantis, share several derived genomic characters. Furthermore, mt genomes of these mantellines showed additional structural divergences, resulting in different genome conditions between them. The high frequency of genomic reorganization does not correlate with nucleotide substitution rate. The encountered mt genomic conditions also suggest the occurrences of stepwise gene duplication and deletion events during the evolution of mantellines. Simultaneously, the majority of duplication events seems to be mediated by general (homologous) or illegitimate recombination, and general recombination also plays a role in concerted sequence evolution between multiple CRs. Considering our observations and recent conditional evidences, the following outlines can be expected for recombination processes in mt genome reorganization. 1) The CR is the "hot spot" of recombination; 2) highly frequent recombination between CRs may be mediated by a replication fork barrier lying in the CR; 3) general recombination has a potential to cause gene rearrangement in upstream regions of multiple CRs as the results of gene conversion and unequal crossing over processes. Our results also suggest that recombination activity is not a direct cause of convergent gene rearrangement; rather, homoplasious gene rearrangement seems to be mediated by persistence of a copied genomic condition through several lineage splits and subsequent parallel deletions.

Key Words: mitochondrial genome • Mantellidae • gene rearrangement • concerted evolution • recombination

	Introduction

TOP Abstract Introduction Materials and Methods Results and Discussion Supplementary Material Acknowledgements References

 
Animal mitochondrial (mt) DNA is a closed circular molecule, typically 16 kbp in size but ranging from 14 to 42 kbp (Wolstenholme 1992). The mt gene content is nearly identical across metazoans, with 13 protein-coding genes, 2 ribosomal RNA genes (rrns), 22 transfer RNA genes (trns), and 1 long noncoding region, the control region (CR; also referred to as D-loop region) that includes the signals necessary for regulation of mtDNA replication and transcription (Wolstenholme 1992; Boore 1999). Animal mt genomes generally lack introns and have very few noncoding regions and intergenic spacers (ISs), except for the CR (e.g., Wolstenholme 1992).

Genomic characters of animal mtDNA, that is, codon usage, gene content, secondary structures of trns, and especially gene arrangement, have been considered to constitute effective phylogenetic markers for metazoans (see e.g., Brown 1985; Wolstenholme 1992; Boore and Brown 1998; Boore et al. 2005). This is partly supported by the apparent complete absence of recombination activity in the animal mt genomic system, which would suppress gene rearrangements and thereby lead to a low incidence of homoplasious rearrangement events (Brown 1985; Boore and Brown 1994). Based on the assumption of absence of recombination in mt genomes, rearrangements of mt genes have usually been interpreted as a result of tandem duplication caused by replication errors (the tandem duplication and random loss [TDRL] model; Moritz et al. 1987; Boore 2000). However, recent evidence for recombination in the animal mt genome (Thyagarajan et al. 1996; Lunt and Hyman 1997; Kajander et al. 2000, 2001; Kraytsberg et al. 2004; Rawson 2005; Sato et al. 2005; Tsaousis et al. 2005) urges to reconsider other duplication modes mediated by recombination (Dowton and Campbell 2001; Endo et al. 2005; Mueller and Boore 2005).

For metazoan mt genomes, it is often difficult to trace the evolutionary pathway of genome reorganization because of a generally low number of reorganization events, too drastic reorganizations in some animal lineages (e.g., nematodes, bivalves, snails, brachiopods, echinoderms, and ascidians; reviewed by Boore 1999 and see Noguchi et al. 2000; Yokobori et al. 2005), and, especially, very few species with intermediate genomic conditions (but see Inoue et al. 2003; Sano et al. 2005; San Mauro et al. 2006). One of the factors that is correlated with and may influence mt genome rearrangements are nucleotide substitution rate (Shao et al. 2003; Xu et al. 2006). Also, it is known that specific genomic regions (i.e., initiation and termination points of replication, and trn clusters) and potential secondary structures of trns and noncoding sequences are involved in gene rearrangements (e.g., Macey et al. 1997; Kraytsberg et al. 2004; San Mauro et al. 2006). The understanding of mechanisms, patterns, and phylogenetic significance of mt genome rearrangements could be further fostered by analyzing an animal group with 1) a high frequency of genome rearrangements in one lineage and a low such frequency in other lineages and 2) intermediate conditions from a stepwise rearrangement process being conserved in some taxa.

Among vertebrate lineages the arrangement of mt genes tends to be conserved, with all 37 genes and the CR organized in a similar order. However, rearranged mt genomes have been reported from representatives of numerous vertebrate taxa (e.g., Macey et al. 1997; Boore 1999). This also applies to neobatrachian frogs, a phylogenetically nested clade comprising the vast majority of amphibian species (Hoegg et al. 2004; Frost et al. 2006). Basal lineages of frogs (a paraphyletic group sometimes referred to as archaeobatrachians) conserve the plesiomorphic vertebrate gene order (e.g., Roe et al. 1985; San Mauro et al. 2004; Gissi et al. 2006). In contrast, in most neobatrachians studied so far the genomic positions of 4 tRNA genes are rearranged and the reduced form of nicotinamide adenine dinucleotideNADH dehydrogenase subunit 5 gene (nad5) is commonly translocated from upstream nad6 to downstream of CR in 2 neobatrachian families, Rhacophoridae and Mantellidae (see fig. 1). Furthermore, in the family Mantellidae, the members of the genus Mantella are known to possess a highly reorganized mt genome with the following characteristics: 1) some rearranged gene positions, 2) two distinct genes and a pseudogene corresponding to transfer RNA gene for methionine (trnM), and 3) duplicated CRs with almost identical nucleotide sequence (Kurabayashi et al. 2006). These unique genomic features were observed to concentrate between the duplicated CRs, surrounding the genes for cytochrome b (cob) and nad2. Because mt genomic organization is generally conserved within closely related taxa, the mantellid mt genomes with such a diversity in their organization seem to be a good model to elucidate molecular evolutionary and phylogenetic aspects of mt genomic rearrangements.

View larger version (54K): [in this window] [in a new window] [Download PowerPoint slide]   FIG. 1.— Gene organizationof mantellid mtDNAs (color version of this figure is available as Supplementary Material online). The transcriptional direction of H-strand–encoding gene is shown by open arrows, and open arrowheads indicate upstream and downstream notations used in this paper. The H-strand–encoded and L-strand–encoded genes are denoted above and below each gene box, respectively. The sizes of the gene boxes do not reflect actual gene length. Closed arrows show the rearranged genes from the typical vertebrate arrangement and direction of rearrangement. The hypothesized gene rearrangement pathway within mantellids is not indicated in this figure (supplementary fig. 4, Supplementary Material online), but gray boxes represent the rearranged genes. Transfer-RNA genes (trns) are designated by single-letter amino acid codes. L1 and L2 indicate trns for Leu(UUR) and Leu(CUN), respectively. M1 and M2 denote 2 distinct trnMs. indicates a pseudogene of its corresponding gene; Mori and Mmos (psMori and psMos in the text) show the pseudogene of original single copied trnM and the mosaic pseudogene having both trnM1 and M2 characteristics (see supplementary fig. 1, Supplementary Material online), respectively. Other genes are abbreviated as follows: rrS and rrL, 12S and 16S ribosomal RNA genes; Cob, cytochrome b; nad1, 2, and 5, NADH dehydrogenase subunits 1, 2, and 5. Multiple CRs within a species are named CR1–3 (and 3'), and rCR3 indicates a CR3 remnant (see text). CRs and long (>30 bp) ISs are shown by black boxes with their lengths. The homogenized regions between multiple CRs are shown by "*** (CR1 vs. CR2)," "+++ (CR1 vs. CR3)", and "(CR2 vs. CR3)" (see text and table 2). The CRs containing 3' side repeats are denoted by "+3rep." The CRs, trnA, and trnN within parentheses indicate that these regions were partially sequenced. a–c Data from previous studies; aSumida et al. (2001), bSano et al. (2004), and cKurabayashi et al. (2006).

 
The family Mantellidae is an endemic frog radiation of Madagascar and the Comoro island of Mayotte (Bossuyt and Milinkovitch 2000; Vences, Andreone, et al. 2003; Vences, Vieites, et al. 2003). According to the most recent classification (Glaw and Vences 2006, 2007; Glaw et al. 2006), this family consists of 12 genera and is divided into three subfamilies, Boophinae (Boophis), Laliostominae (Aglyptodactylus and Laliostoma), and Mantellinae (Blommersia, Boehmantis, Gephyromantis, Guibemantis, Mantella, Mantidactylus, Spinomantis, Tsingymantis, and Wakea). Ecologically, mantellines usually breed in either ponds or streams. In both the Boophinae and in the Mantellinae, this reproductive preference seems to be phylogenetically conserved, and at least in boophines, pond breeders may be characterized by lower mt substitution rates (Vences et al. 2002). Despite intensive studies (see Vences et al. 2007), the relationships between various deep lineages in this family so far remain unresolved.

In this study, we newly examined the region of mt genomic reorganization known in Mantella for 12 mantellid species and compared the genome organizations among 17 species covering all known mantellid genera. Based on these results we 1) describe unique genomic features found in various mantellid species, 2) reconstruct the phylogenetic relationships among mantellid genera based on nucleotide sequence data, 3) infer possible mechanisms and pathways of mt genomic reorganizations, 4) search for correlates between rates of nucleotide substitution and genome rearrangement, and 5) discuss the phylogenetic utility of mt gene arrangement.

	Materials and Methods

TOP Abstract Introduction Materials and Methods Results and Discussion Supplementary Material Acknowledgements References

 
Frog Specimens
To survey, for all mantellid genera, the region of mt genomic reorganization (from cob to nad2) known from Mantella, we newly examined mt genomes of 12 species. Previous data of 5 mantellids, a rhacophorid (Buergeria buergeri; Sano et al. 2004), and a ranid (Rana nigromaculata [=Pelophylax nigromaculatus]; Sumida et al. 2001) were also included in the comparisons and analyses, and preliminary data of the length of the genomic reorganization fragment were obtained for further 48 mantellid species (supplementary table 2, Supplementary Material online). Voucher information and accession numbers of the determined DNA sequences are listed in table 1.

View this table: [in this window] [in a new window]   Table 1 Mantellid Specimens Used in This Study

 
Polymerase Chain Reaction, Subcloning, and Sequencing
Total DNA was extracted from muscle tissues using DNeasy Tissue Kits (Qiagen, Hilden, Germany). The amplification and sequence strategy of mt fragments followed similar procedures as in a previous study (Kurabayashi et al. 2006). In brief, to prevent polymerase chain reaction (PCR) jumping (Pääbo et al. 1990) between homogenized CRs, a number of overlapping fragments representing the cob–nad2 region were amplified with degenerated primer sets for each species (the sequences of primers used are available upon request to A.K.). PCR reactions were carried out using LA-Taq (Takara Bio Inc., Shiga), according to the manufacturer's instructions. The primer-walking method was employed for sequencing using an automated DNA sequencer (ABI 3100) with the BigDye Terminator Cycle Sequencing Kit (ABI Japan Ltd., Tokyo, Japan). PCR fragments containing long tandem repeats and/or mononucleotide tracts from which we could not sequence by primer walking were subcloned into Escherichia coli vector pCR-2.1 or pCR-XL using the TOPO TA Cloning Kit (Invitrogen, Carlsbad, CA). Almost all fragments could be subcloned; however, we failed to get suitable clones for the CR of Gephyromantis pseudoasper (ca. 12 kbp) and Mantidactylus grandidieri (ca. 6 kbp). To determine the precise sequences of the long tandem repeats, a series of deleted subclones were made from the resultant subclones using the exonuclease III deletion method (Henikoff 1987). Both DNA strands were sequenced about half of the surveyed regions, but remaining regions, especially long tandem repeats within CRs, were determined for only one strand due to the lack of suitable deletion mutants. From the resulting sequences, genes were identified by comparisons with corresponding gene sequences from other vertebrates.

Molecular Phylogenetic Analyses
A concatenated nucleotide alignment was made from the sequences of 4 protein-coding (cytochrome b [cob]; reduced form of nicotinamide adenine dinucleotide dehydrogenase subunits 1, 2, and 5 [nad1, 2, and 5]), 12S and 16S ribosomal RNA (rrnS and rrnL), and 9 transfer RNA (trnA, F, Lcun, Luur, P, Q, T, V, and W) genes. For protein-coding genes, the elucidated amino acid sequence of each gene was initially aligned using ClustalX 1.83 (Thompson et al. 1997). Gaps and ambiguous areas were excluded using Gblocks 0.91b (Castresana 2000) with default parameters; then the corresponding codon sequences (4 503 sites) were used for further analysis. The rRNA and tRNA alignments (2 178 and 584 sites, respectively) were also constructed using the same programs. The final alignment encompassed 19 taxa and 7 265 nt sites, of which 3 148 sites were parsimony informative.

Phylogenetic analyses based on the concatenated alignment data were conducted using maximum parsimony (MP), maximum likelihood (ML), and Bayesian inference (BI) methods. In all analyses, R. nigromaculata was used as out-group. MP analysis was performed using PAUP*4.0b10 (Swofford 2003). A heuristic search option with 10 random addition replicates and tree bisection-reconnection (TBR) branch swapping was used, and all sites were of equal weighting. Clade support under MP was evaluated using 1 000 replicates of nonparametric bootstrapping. For ML and BI analyses, appropriate substitution models (general time-reversible model with gamma distribution substitution rates and proportion of invariable sites [GTR + G + I]) were chosen using the Akaike information criterion implemented in Modeltest 3.7 (Posada and Crandall 1998). The ML analysis based on the concatenated data was performed using PAUP* with heuristic search and TBR swapping. Nonparametric BPs under ML were calculated using PHYML 2.4.4 (Guindon and Gascuel 2003) with 300 replicates. BI analysis was performed using MrBayes 3.1.2 (Ronquist and Huelsenbeck 2003). To determine a suitable partitioning strategy for BI analysis, we calculated the Bayes factor of 6 different partitioning strategies (see supplementary table 1, Supplementary Material online) and compared the resulting values by the same procedure proposed by Brandley et al. (2005); and the 5-partition strategy (firstst/second/third codons of protein-coding genes + rrns/trns) was selected. The following settings were also used for the BI analysis: number of Markov chain Monte Carlo generations = 15 x 10⁵, sampling frequency = 10. The burn-in size was determined by checking convergences of –log likelihood (–ln L) values, and first 1 x 10⁵ generations were discarded. The statistical support of the BI tree was evaluated by Bayesian posterior probabilities (BPP). Relative rate tests were conducted using Phyltest (Kumar 1996).

Sixteen alternative phylogenetic hypotheses for mantellid frogs were compared using approximately unbiased (AU) and Kishino–Hasegawa (KH) tests implemented in CONSEL 0.1h (Shimodaira and Hasegawa 2001). Sitewise ln L values were calculated using PAUP* and used as input for the program. Resampling of estimated Log-likelihood of sites (Hasegawa and Kishino 1994) was conducted in 1 x 10⁴ resamplings.

Divergence times of mantellid taxa were estimated from the concatenated sequence data by using the Bayesian molecular clock method implemented in the MultiDivtime program package (Thorne and Kishino 2002). We used the root of mantellids (=the last split between Mantellidae and other ranoids) as a single calibration point and 2 alternative constraints, 61 MYA) and 55–72 MYA, at this point were applied. These ages were estimated by Van der Meijden et al. (2007) (61 ± 11); and at least an independent study suggested a similar age (62–67 MYA, Roelants et al. 2007). Several studies indicated different time ranges for the root of mantellids: 53.6–92.6 MYA (Bossuyt and Milinkovitch 2001), 56–86 MYA (Van Bocxlaer et al. 2006), 55–91 MYA (Igawa et al. 2008), and 58–117 MYA (San Mauro et al. 2005). Thus, we also tried an alternative constraint, 53.6–92.6 MYA, at the root of mantellids. These reference point settings did not strongly affect the estimated branching ages of mantellid taxa (though the latter setting resulted in slightly older ages; data not shown); and thus, we only describe the result from the former setting. In this analysis, we excluded Tsingymantis because of its ambiguous phylogenetic position. Although our ML and BI analyses recovered the Spinomantis + stream-breeder clade, we constrained the Spinomantis + pond-breeder clade in branching time estimation. The Spinomantis + pond-breeder clade was recovered in the MP tree in this study and is strongly supported by the mt genomic characters elucidated here (see below) and by ML and BI analyses based on combined data of nuclear and mt genes (Vences et al. 2007).

	Results and Discussion

TOP Abstract Introduction Materials and Methods Results and Discussion Supplementary Material Acknowledgements References

 
Gene Organization of the cob–trnA Region of Mantellid mtDNAs
In this study, the partial mt genomic portion from cob to trnA (or trnN), covering the region of genomic reorganization known for mantellid frogs, has been newly analyzed for 12 mantellid species (table 1 and fig. 1). Adding to our previous results (Kurabayashi et al. 2006), DNA sequences of 17 species of all 12 mantellid genera are now available to compare the mt genome organization. The length of the surveyed portion varies among species from 9 676 bp (Gephyromantis klemmeri) to 17 792 bp (Spinomantis microtis), mainly due to the length difference of CR and/or copy number of CRs. In all analyzed mantellid species we identified, in this genomic portion, one CR and 17 single-copy genes typical for vertebrate mtDNAs (4 protein-coding genes: cob, nad1, nad2, and nad5; 2 ribosomal RNA genes: rrnS and rrnL; 11 transfer RNA genes for A, F, I, Lcun, Luur, M, P, Q, T, V, and W) (fig. 1). In the studied representatives of the genera Boophis, Aglyptodactylus, Laliostoma, Tsingymantis, Boehmantis, Gephyromantis, and Mantidactylus, these genes occur in 1 copy each, with a gene order identical among all these taxa and to the rhacophorid frog B. buergeri (Sano et al. 2004).

On the contrary, the eight studied species of the genera Spinomantis, Blommersia, Guibemantis, Wakea, and Mantella carry additional copied CRs (CR1–CR3 and/or CR3'; fig. 1). Excluding extra repeat units in particular CRs, the copied CRs have nearly identical sequences within each species (97.8–99.8%; table 2 and see fig. 2), and the homogenized sequence spans from near the 5' end (often including the 5' flanking gene) to downstream of the conserved sequence block (CSB) 3, a characteristic sequence element of vertebrate CR (see section below). A remnant of a third CR (rCR3), 575 bp in length, was also found in S. microtis. Although this rCR3 lacks CSB motifs, a part of it (ca. 280 bp) shows 95% similarity with the other 2 CRs of this species. The eight species with duplicated CRs uniformly contain two distinct trnMs (trnM1 and M2; fig. 1, supplementary fig. 1, Supplementary Material online), although only single trnM is encoded in almost all metazoan mtDNAs. All of these species also contain a pseudo trnM: this pseudogene located just upstream of nad2 corresponds to the original position of this gene (and is thus named psMori) and has been considered to have a function in nad2 mRNA processing (Mabuchi et al. 2004; Kurabayashi et al. 2006). The trnM1 and M2 can be defined by both nucleotide similarity (average identity between two trnMs = 79.5%) and by a characteristic nucleotide of the 5' adjoining anticodon; that is, "T" in trnM1 and "C" in trnM2, respectively (supplementary fig. 1, Supplementary Material online). The single-copy trnM in other mantellids also has a C nucleotide at this position and shows a high nucleotide similarity with trnM2 (90.1%) rather than with trnM1 (83.5%), indicating that trnM1 and trnM2 correspond to the paralogue and homologue of the single copied trnM, respectively. Guibemantis tornieri and Wakea madinika harbor an obvious pseudo trnM1 at the 3' side of nad5, and Guibemantis liber also has an additional pseudo trnM at a similar location (between nad5 and CR2). Notably, the pseudo trnM in G. liber shows sequence similarity with both trnM1 (95% in 5'—22 bp) and trnM2 (100% in 3'—35 bp), and therefore seems to be a "mosaic pseudogene" of these genes (psMmos) (see fig. 2 and supplementary fig. 1, Supplementary Material online). Guibemantis liber furthermore encodes two copied trnIs (only 3 mutations between them; see fig. 2) whereas one pseudo trnI is found in Blommersia blommersae. Spinomantis microtis contains a possible pseudo trnQ (having 50% nucleotide similarity with trnQ in 56 bp) located just upstream of psMori.

View this table: [in this window] [in a new window]   Table 2 Similarities and Boundaries of Homogenized Sequence between Multiple CRs within Species

 

View larger version (49K): [in this window] [in a new window] [Download PowerPoint slide]   FIG. 2.— Structure ofCRs and their flanking regions in Guibemantis liber. (A) Structure of 3 CRs and their flanking region. "rep1–4" indicates 4 distinct tandem-repeat sequences, and the length and copy number of each repeat are also denoted. The 3' side repeats (rep-3 and rep-4) are only found in CR3 in this species. CSB1–3 are shown by open boxes. Grayed boxes denote the homogenized sequence regions between CRs. (B) Nucleotide sequence comparison of 5' regions of CRs. Thin vertical lines indicate the boundaries of gene regions, and grayed boxes show the homogenized sequence regions. Positive and negative numbers indicate downstream and upstream locations from first nucleotide (=+1) of each CR, respectively. (C) Nucleotide sequences comparison of CSB regions. The putative CSBs are boxed. The nucleotide positions are numbered from the 3' adjoining nucleotide (=+1) of CSB3.

 
Summarizing, eight species among the surveyed mantellids, namely the members of pond-breeding mantellines and of Spinomantis, possess divergent mt genomes. Although the mt genomic structures of this group of taxa differ in many respects, there are a number of derived gene arrangements shared by some or all of them (figs. 1 and 4): 1) A derived arrangement, trnM2–CR2–trnQ–psMori located upstream side of nad2, is shared by all these species, excluding pseudo trnQ between CR2 and psMori in S. microtis. 2) A novel trnI–trnM1 arrangement is commonly found. Including their pseudogenes in the count, 7 species possess this arrangement. Although S. microtis lacks this trnI–trnM1 arrangement, an IS (167 bp) is found upstream of its trnM1. The trnI and/or trnM1 is basically positioned downstream side of nad5; all eight species possess either the trnI–trnM1 arrangement, or at least one of these genes, or a pseudogene of trnI or trnM1 in this region. 3) An additional trnI–trnM1 gene block in a unique position between cob and CR1 is found in species of Blommersia and Guibemantis. 4) Four mantellids belonging to Spinomantis, Blommersia, and Guibemantis carry a third CR (CR3) or its remnant (rCR3) on the upstream side of the LTPF trn cluster. 5) Wakea and Mantella species lack this third CR, but these species have an IS of considerable length (43–205 bp) at the same position. Wakea madinika has, in addition, a third CR positioned upstream side of nad5, and the CR–nad5–pseudo trnM1 block is translocated from its original position (upstream side of the LTPF trn cluster) to the upstream side of nad1 (thus, the Wakea CR3' seems to be a paralogue of its first CR and not homologous to the CR3 of the other species). 6) trnT is translocated from its original position in the LTPF trn cluster to the upstream side of nad1 in species of Mantella.

View larger version (33K): [in this window] [in a new window] [Download PowerPoint slide]   FIG. 4.— Divergence times and synapomorphic mt genomic characters of mantellid taxa (a color version of this figure is provided as a Supplementary Material online). Estimated divergence times with standard deviation (gray boxes) and 95% confidence intervals (gray lines) are shown for each node. Inset figures show probable synapomorphic mt characters for each clade. The roman numbers above the nodes indicate the lineages of gene rearrangement events described in supplementary figure 4 (Supplementary Material online).

 
The distribution of these derived gene arrangements among taxa supports the respective synapomorphic origin of these rearrangements at different levels of mantellid phylogeny because alternative scenarios would require assumptions of convergent occurrence of complex molecular evolutionary steps (see section below).

Mantellid Phylogeny and Synapomorphic mt Gene Arrangements
The ML tree (–ln L = 71 527.04; fig. 3) obtained from the concatenated mt gene sequence (7 265 sites) fully agrees in topology with the BI analysis. Various alternative phylogenetic hypotheses have been suggested in previous studies for the Mantellidae (see the references denoted in table 3). Our ML tree supports the most recent taxonomic and phylogenetic proposals of Glaw and Vences (2006) and Glaw et al. (2006), in particular regarding monophyly of the following clades: 1) the family Mantellidae, 2) each of the three mantellid subfamilies (if not considering Tsingymantis), 3) the Boophinae + Laliostominae clade, 4) two major clades corresponding to pond breeders and stream breeders within the subfamily Mantellinae, 5) relationship of Boehmantis + (Gephyromantis + Mantidactylus) within the stream-breeder clade, and 6) two subclades corresponding to Blommersia + Guibemantis and Wakea + Mantella within pond breeders. These clades are supported by very high BPP and/or ML BP values, but the support for the position of Tsingymantis (in a clade with Laliostominae + Boophinae) is low (ML and MP bootstrap values <50 and 69, respectively, and BPP = 90). This relationship differs from the hypothesis of Glaw et al. (2006), where Tsingymantis occupied the most basal position of Mantellinae and was included this subfamily. Also poorly supported is the basal position of Spinomantis within the clade of stream breeders in our ML and BI trees. Although BPP of this clade is very high (=100), BPs of ML and MP are moderate (=77 and 63, respectively). Furthermore, the MP tree did not recover this position of Spinomantis but instead placed Spinomantis with pond breeders (hypothesis 2 in table 3).

View larger version (22K): [in this window] [in a new window] [Download PowerPoint slide]   FIG. 3.— ML phylogram of mantellid frogs based on 7 265 bp of mt gene sequence. The ln L value of this tree is –71527.04. The BI analysis based on the same data set recovered the same topology, but the MP analysis supported a clade containing pond-breeding mantellines and Spinomantis. Upper and lower numbers below branches are percent nonparametric bootstrap probability from ML and MP analyses, respectively. Asterisks above branches indicate Bayesian posterior probabilities (**100).

 

View this table: [in this window] [in a new window]   Table 3 Comparison of Alternative Phylogenetic Hypotheses

 
We compared 16 alternative phylogenetic hypotheses regarding the positions of Spinomantis and Tsingymantis (table 3) using AU and KH tests. For the position of Spinomantis, six alternative hypotheses (hypotheses 3–8) are rejected by both tests at P < 0.05, but the hypothesis placing Spinomantis sister to the pond-breeding mantellines (hypothesis 2) is not rejected. For the position of Tsingymantis, two alternative hypotheses, a basal position of Tsingymantis in the Mantellinae (hypothesis 12) and a basal position in the Mantellidae (hypothesis 13) are not rejected. Alternative hypotheses for other taxa are largely rejected by these tests. For example, the nodal support of Boehmantis + Gephyromantis and Mantidactylus is low in BP of MP (=57), but alternative Boehmantis positions are rejected by the tests (see hypotheses 4, 5, and 8 in table 3).

Interestingly, some aspects of mantellid phylogeny are also supported by mt genomic data. Whereas boophines (Boophis), laliostomines (Aglyptodactylus and Laliostoma), Tsingymantis, and most stream-breeding and direct-developing mantellines (Boehmantis, Gephyromantis, Mantidactylus) share the plesiomorphic gene order, the mainly pond-breeding genera (Blommersia, Guibemantis, Mantella, Wakea) and Spinomantis are characterized by the derived condition, which supports the placement of Spinomantis sister to the pond breeders (hypothesis 2 in table 3 and fig. 4). Indeed, recent analysis including sequence data from various mt and two nuclear genes (Rhodopsin and Rag-2) recovered the clade consisting of Spinomantis plus pond breeders with relatively high statistical supports (BI > 99 and BP of ML = 90; Vences et al. 2007). Furthermore, some genomic characters also support the monophyly of pond breeders and their subgroups (i.e., Blommersia + Guibemantis, Wakea + Mantella) that are also unanimously supported by our ML/BI/MP analyses (fig. 4).

Frequency of Gene Rearrangements among Mantellid Lineages
Our data indicate a multiple occurrence of mt gene rearrangements in pond-breeding mantellines plus Spinomantis. This led to (at least minor) autapomorphic features in the mt genomes of basically every representative of the lineage for which we determined the gene order (fig. 1 and see supplementary fig. 4, Supplementary Material online). In contrast, our data provide evidence that representatives of all other mantellid genera have maintained the ancestral gene order, thus suggesting that rearrangement rates were much lower in these lineages. To test this assumption beyond the analysis of the 17 mantellid species analyzed here, we amplified in 48 additional mantellid species a fragment spanning from the rrns to nad2 and thus encompassing the area where in the pond-breeding species a large part of the duplicated and translocated genes and CRs are situated. We subsequently electrophoretically determined the approximate length of the amplified fragment to identify deviations from the ancestral gene arrangement (which corresponds to a fragment length of ca. 3.2–3.3 kbp). The complete list of taxa examined is found in the supplementary table 2 (Supplementary Material online).

All 21 additionally examined species in the pond-breeding lineage (6 Blommersia, 6 Mantella, 6 Guibemantis, and 3 Spinomantis) showed apomorphic fragment lengths, and fragment length variability among taxa was high (4.9–13 kbp; supplementary table 2, Supplementary Material online). In contrast, 22 other mantellids (1 Aglyptodactylus, 6 Boophis, 3 Gephyromantis, and 12 Mantidactylus) showed typical fragment length, indicating that in this region of these genomes, no major gene translocations or duplications had occurred. However, 2 Boophis, 1 Gephyromantis, and 2 Mantidactylus species showed variable lengths of 5.0–7.7 kbp, suggesting that apomorphic gene rearrangements may have also occurred in a minor proportion of species of these species-rich genera and pointing to the need of further, more detailed studies in these taxa.

Although these data refer to a limited portion of the mt genome only and rearrangements in other parts of the genomes would be ignored by this test, the results are nevertheless in agreement with the conclusions from the species of which gene order was determined in detail (fig. 1) and confirm that these species were indeed representative for their respective lineages.

No Correlation between Rates of Nucleotide Substitution and Gene Rearrangement
A tendency that highly rearranged mt genomes have high nucleotide substitution rate has been observed in some metazoan taxa (mollusks: Hoffmann et al. 1992; Kurabayashi and Ueshima 2000; ascidians: Yokobori et al. 2005; lampshell: Noguchi et al. 2000), and a positive correlation between the rates of nucleotide substitution and genomic rearrangement has been demonstrated in arthropod taxa (Shao et al. 2003; Xu et al. 2006). To assess this correlation in mantellid frogs, we compared the nucleotide substitution rates by relative rate tests among three mantellid groups: 1) the mostly pond-breeding mantellines (Blommersia, Guibemantis, Mantella, Wakea, and Spinomantis, all with highly rearranged mt genomes), 2) the mostly stream-breeding mantellines (Boehmantis, Gephyromantis, and Mantidactylus, without rearrangements), and 3) the remaining lineages (Boophis, Aglyptodactylus, Laliostoma, and Tsingymantis, without rearrangements). These tests did not yield any evidence for accelerated mt nucleotide substitution rate in the pond-breeding mantellines (table 4). Rather, in the full data (rrns, trns, and protein-coding genes) and rrn data sets, the pond-breeding lineage had slightly slower substitution rates than stream breeders (rate constancy rejected; P < 0.05) and almost identical rates as compared with the other mantellids. In protein-coding gene data, all 3 groups compared showed constant nucleotide substitution rates.

View this table: [in this window] [in a new window]   Table 4 Results of Relative Rate Tests among Lineages of Mantellids

 
In the hypothesis of Shao et al. (2003), an acceleration of nucleotide changes leads to many illicit substitutions at both initiation and termination points of mt genome replication; and such illicit initiation and termination points cause frequent occurrence of tandem duplications in the mtDNA molecule, resulting in highly frequent gene rearrangements. However, the clear absence of a correlation between the substitution and rearrangement rates found here indicates that substitution rate is not relevant to the frequent gene rearrangements in pond-breeding mantellids.

Possible Mechanisms and Evolutionary Pathway of mt Genome Reorganization in Mantellids
What has been the process that has led to the reorganized genomic structures seen in mantellids? In general terms, the observed pattern is largely consistent with the "duplication and random loss model," the most general explanation for mt genomic reorganization (e.g., Boore 2000; San Mauro et al. 2006; and see supplementary fig. 2A, Supplementary Material online). In this model, first a portion of the genome is duplicated. Then, one of the copied genes (or CRs) turns into a pseudogene or is excised from the genome by subsequent mutational processes. Which copy is lost is determined randomly (by the first loss-of-function mutation), and therefore, a certain deletion pattern can recover the original gene arrangement but another leads to gene rearrangement. As a possible exception, Lavrov et al. (2002) reported the presence of nonrandom loss events in a specific insect taxon.

On the molecular level, it has been hypothesized that duplication in animal mt genomes is mainly caused by replication errors, such as slipped-strand mispairing or asynchrony in the points of initiation and termination (e.g., Mueller and Boore 2005; San Mauro et al. 2006; and see supplementary fig. 2A, Supplementary Material online). However, such replication errors only generate tandem duplications (Mueller and Boore 2005); and thus, this model (TDRL) has difficulties in explaining the many nontandem duplications observed in mantellids (fig. 1). Therefore, other duplication mechanisms seem to have acted in the mt genomic reorganization process. Two distinct recombination modes, "illegitimate recombination via minicircle" (Holt et al. 1997; Lunt and Hyman 1997; Kajander et al. 2000; Dowton and Campbell 2001; Mueller and Boore 2005) and "general (homologous) recombination" (Thyagarajan et al. 1996; Kajander et al. 2001; Ladoukakis and Zouros 2001; Sato et al. 2005) are the candidates of such an alternative mechanism. In the illegitimate recombination via minicircle, a part of the mt gene region is excised from 1 mt genome, forming a separate minicircle molecule that is then inserted into another genome, and the insertion results in nontandem-duplicated regions within the mtDNA molecule (supplementary fig. 2B, Supplementary Material online). The general recombination process (see e.g., Lewin 2003) exchanges DNA strands of 2 genomic portions with identical or similar nucleotide sequences between chromosomes (or within a DNA molecule) (supplementary fig. 2C, Supplementary Material online). When the exchanged DNA strands contain the same set of genes or regions, this recombination process does not cause gene duplication but can homogenize the nucleotide sequences between the exchanged portions (gene conversion). In contrast, when the exchanged DNA strands carry unequal sets of genes, one of the resultant molecules (or genomic portions) will have an extracopied gene region from the original one (unequal crossing over).

Although any recombination had been considered to be absent in animal mt genome system (e.g., see references in Howell 1997; Shadel and Clayton 1997), many recent studies have provided experimental and circumstantial evidence of at least a limited degree of mt recombination (e.g., Thyagarajan et al. 1996; Lunt and Hyman 1997; Sato et al. 2005). Applying all these possible duplication mechanisms to explain mt genome rearrangements in mantellids results in the conclusion that one tandem duplication, followed by numerous nontandem duplications and deletions, is the most parsimonious scenario (i.e., the one minimizing the duplication and deletion events, and the number of hypothetically duplicated and deleted genes and CRs; in parallel, all nontandem duplications are presumed to be caused by processes related to recombination). The preferred scenario is further elaborated in supplementary figure 4 (Supplementary Material online). According to this hypothesis, gene rearrangements among mantelline taxa were caused mainly by alternative deletions from an ancestral genome with a duplicated condition (a total of eight deletion events assumed in our scenario; i.e., del-2 to del-9 in supplementary fig. 4, Supplementary Material online) rather than by specific duplications (three duplication events; dup-4 to dup-6). The majority of these duplication events would have been nontandem (five events; dup-2 to dup-6) and only one tandem duplication (dup-1) may have occurred, indicating that recombination rather than replication slippage has probably been the main cause of mt genomic reorganization during mantellid evolution. It is also noteworthy that the downstream boundary (in the L-strand) of the three expected duplicated portions seems to fall into the CR in 3 cases. This supports the importance of the CRs as possible "recombination hot spot" which will be further explored in the following sections and indicates that the high frequency of mt genome reorganizations in Spinomantis + pond-breeding mantellines may be explained by the accidental fixation of multiple CRs in their common ancestor.

Multiple Control Regions and Their Flanking Regions with Nearly Identical Nucleotide Sequences
In this study, all taxa with multiple CRs contain putative CSB1–3 elements in each of their CRs (see fig. 2). The presence of CSBs suggests that all multiple CRs within a single mtDNA molecule have a potential to form a D-loop structure because CSBs has been hypothesized to be essential for the synthesis of D-loop DNA (Cairns and Bogenhagen 1986; Shadel and Clayton 1997).

All multiple CRs within mantellid species include portions of nearly identical sequence, except for additional numbers of 5' repeats in certain CRs (table 2 and fig. 2). These homogenized sequences start from the 5' end neighborhood and extend beyond the CSB3 motif. In many cases, the 5' boundaries of the homogenized sequences are located inside CRs. However, the continuous identical sequence across the 5' flanking genes is also observed in Blommersia and Guibemantis species (table 2 and fig. 2B). In B. blommersae, the homogenized sequence between CR1 and CR2 stretches to their 5' flanking trnM1 and trnM2 along 9 bp. Similarly, in G. tornieri, the homogenized sequence continues across the 5' flanking regions of CRs and 35 bp are found between the trnM1 and pstrnM1. But, different from Blommersia, trnM1 and pstrnM1 are located neighboring CR1 and CR3, respectively. Most unique is the 5' flanking region of CR3 in G. liber (fig. 2A and B). In this case, the sequence homogenized between CR3 and CR1 is separated into two partitions; that is, a 103-bp region (homogenized region 1 in fig. 2) spanning from the 3rd nucleotide of trnI to the 22nd nucleotide of psMmos (corresponding to trnM1 in CR1 upstream) and 1.6 kbp starting from the fifth nucleotide of CR3 (=13th nucleotide of CR1) and continuing to downstream of CSB3. Inbetween these homogenized regions are 47 nucleotides of 3' psMmos and 5 nucleotides of 5' CR3 (total 52 nucleotide with 67% similarity to the relative 60 bp of the 3' trnM1–5' CR1 region). Meanwhile, the 5' boundary of the homogenized sequence between CR3 and CR2 of this species falls into the above interval sequence (=34th nucleotide of psMmos and trnM2). Thus, the 5' flanking region of CR3 has a nested homologous sequence with respect to CR1 and CR2.

Homogenized sequences between two copied CRs are often observed in animal mt genomes. But, the presence of 3 CRs found in several mantellines is rare (only known from a Plethodon salamander thus far; Mueller and Boore 2005). It is also relevant that homogenized sequences along 5' flanking genes are found between different CR combinations (i.e., CR1 vs. CR2 in Blommersia, CR1 vs. CR3 in G. tornieri). This strongly suggests that sequence homogenization of multiple CRs occurs not between a specific pair of CRs but between random pairs, also meaning that the process of sequence homogenization is not related to particular genomic portions or to relative distance of CRs.

Possible Roles and Mechanisms of General Recombination in Concerted Sequence Evolution
As in mantellids, duplicated CRs have been reported from a number of animal mt genomes (octopuses: Yokobori et al. 2004; ticks: Black and Roehrdanz 1998, Campbell and Barker 1999; sea fire-fly: Ogoh and Ohmiya 2007; sea cucumbers: Arndt and Smith 1998; fishes: Lee et al. 2001, Inoue et al. 2003; salamanders: Mueller and Boore 2005; rhacophorid frogs: Sano et al. 2005; Snakes: Kumazawa et al. 1996; birds: e.g., Mindell et al. 1998, Eberhard et al. 2001). These multiple CRs within species (and occasionally their upstream regions) have nearly identical nucleotide sequence with few exceptions. To explain such homogenized sequences between CRs, 2 alternative mechanisms of concerted evolution have been postulated: one is the tandem-duplication model by replication slippage (this model is based on the strand-asynchronous replication model) and another is the gene conversion model by general recombination (Kumazawa et al. 1998). Of these alternatives, the gene conversion model is strongly supported by the nested homologous region observed at the 5' flanking region of CR3 found in G. liber. If the homogenized sequences between CRs were maintained by tandem duplication via replication slippage, the sequences of the duplicated region would need to be sequentially homogenized (see fig. 7a of Kumazawa et al. 1998); and thus, this mechanism alone cannot maintain the nested homogenized portions. Furthermore, the tandem-duplication model cannot produce the nested homogenized sequence in Guibemantis. In contrast, the general recombination mechanism can easily create and maintain the observed pattern (see supplementary figs. 3A and 4, Supplementary Material online).

Another remarkable observation from this study is that almost all extracopied genes (with only one exception of two copied trnIs in G. liber) are deleted or become pseudogenes or ISs; at least 15 duplicated genes have apparently been deleted (supplementary fig. 4, Supplementary Material online). In contrast, many of the copied CRs have been maintained with nearly identical sequences, and only two CR deletion events need to be invoked. Similar situations have been reported from other animal mt genomes with multiple CRs (Kumazawa et al. 1996; Mueller and Boore 2005). These observations indicate that multiple CRs are particularly prone to homogenization compared with other genomic portions, suggesting that CRs could be considered as "hot spots" of recombination.

Why do CRs have such a high recombination frequency? To answer this question, a novel scheme for vertebrate mtDNA replication may be suggested here. Recent studies based on the technique of 2-dimensional agarose gel electrophoresis have challenged the strand-asynchronous model, the conventional replication scheme for animal mtDNA (Clayton 1982), and suggested that usual leading-ragging replication is a major mode of vertebrate mtDNA replication (Holt et al. 2000; Yang et al. 2002; Bowmaker et al. 2003; Reyes et al. 2005; Yasukawa et al. 2006): replication of vertebrate mtDNA is initiated from a wide genomic portion without any specific site, in a region of approximately 4 kbp across cob to the 5' region of the CR in mammals (Bowmaker et al. 2003) and throughout the mt genome (only excluding the replication fork barrier [RFB] region in the CR) in birds (Reyes et al. 2005). The 3' side of the CR (D-loop region) corresponds to a RFB rather than the replication origin of H-strand (O_H) so far considered. Furthermore, because replication fork stalls were observed from many stretches within the D-loop region, it has been postulated that the D-loop itself assumes the role of RFB (Bowmaker et al. 2003) rather than specific cis-elements. Based on this replication scheme, the 3' end of the nascent L-strand is suspected to remain free at the RFB region until replication restarts (or replication is completed by the backward directional replication fork; see supplementary fig. 3A, Supplementary Material online). According to the Holliday (1964) model of recombination, this process is initiated by formation of free 3' ends of two homologous DNAs by single-strand nicking; subsequently strand exchange takes place between each free end and the complementary strand in the opposite homologue (also see supplementary fig. 3A, Supplementary Material online). Therefore, the long time exposure of the free 3' end of the nascent strand is thought to lead to a high frequency of recombination. Especially, strand exchange of the nascent strands will take place very easily when replication forks are arrested at RFB regions in CRs of two mtDNA molecules (or within a taxon, if two independent replication forks occur in a single mtDNA) (supplementary fig. 3A, Supplementary Material online). It should be noted that, based on the strand-asynchronous replication model, the same concept for a high recombination potential of the free 3' end of the nascent H-strand in light-strand replication origin in human mtDNA has been proposed by Kraytsberg et al. (2004). These authors also suggested that the high recombination rate at the CR region is caused by recombination of D-loop DNA itself. However, this explanation does not seem to match the expected recombination in mantellid CRs. In fact, the D-loop DNA initiates from upstream neighbor of or within CSB motifs and does not extend to the 5' flanking genes, but nevertheless, the region from the 5' flanking genes to the 3' portion of CSB3 seems to be included in the recombination unit in mantellid CRs (see below).

The frequent recombination mediated by RFBs of distinct CRs also becomes the explanation for the concerted evolution observed between multiple CRs because successive recombination processes initiated from strand exchange of the nascent L-strand will cause gene conversion, resulting in homogenized CRs (and 5' flanking genes with similar nucleotide sequence; see supplementary fig. 3A, Supplementary Material online). Furthermore, in this concerted evolution model, gene conversion only occurs at the upstream region from the 3' end of the nascent L-strand. In mantellid mtDNAs, the 3' boundary of the homogenized region falls into the 30–71st nt portion from CSB3, and only the region upstream from this portion is homogenized. These observations match with predictions of the concerted evolution model and suggest that synthesis of the nascent L-strand often arrests at the CSB3 downstream portion in mantellid mtDNAs.

Consequently, if we accept the novel replication model for animal mtDNA as outlined above, the high frequency of recombination among CRs and the long maintenance of homogenized CRs (and of their 5' flanking genes) can be explained without any contradiction. Until now, for the mechanism of concerted evolution of CRs in animal mt genomes, two alternative hypotheses, recombination or replication, have been discussed (Kumazawa et al. 1998; and see Ogoh and Ohmiya 2007). In our model, however, both recombination and replication contribute to the concerted evolution, because the mechanism of concerted evolution is likely recombination, but recombination occurs during the period of replication fork pausing.

In our data set, there are two short regions having identical sequence with other genomic portions of the same mtDNA molecule: trnI–psMmos (103 bp in length) has identical sequence with trnI–trnM1 in G. liber, and a region of Spinomantis rCR3 (ca. 280 bp) shows a homogenized sequence with those of other CRs of this species (table 2). These short homogenized regions would have been involved in the region of high frequent recombination but they are now discrete from the recombination region (rCR3 lacks CSBs and thus does not appear to be able to form a D-loop). Short recombinant mtDNA fragments occurring at unrelated portion from CR (50- to 100-bp fragments in nad1, nad2, and cox1) have been shown by using mito-mouse, which artificially carries two distinct mtDNA molecules in an individual; but mito-mouse analysis also suggested that such recombination occurs at low frequency (Sato et al. 2005).

Possible Roles of General Recombination in the Process of Gene Rearrangement
Generally, animal mtDNA contains only one set of single copied genes and a single CR; and homoplasmy, mediated by strict maternal inheritance, is a common feature of almost all metazoans (e.g., Wolstenholme 1992). In such a situation, homologous recombination does not seem to occur within a mtDNA molecule, and the possible recombination between molecules will not result in any changes of neither nucleotide sequence nor gene arrangement (Sato et al. 2005). In those mantellid mt genomes that possess several copied CRs and gene regions, however, general recombination seems to have an important influence on gene rearrangement events. At least, general recombination has the potential to cause gene replacement between genes with high nucleotide similarity (ca. 80%) by gene conversion events, as in the cases of trnM1 and M2 of Blommersia and Guibemantis.

General recombination in the mt genome also seems to be responsible for unequal crossing over. First, in mantellids, the copy numbers of 5' side tandem repeats occasionally differ among multiple CRs within species (see fig. 2). The different repeat numbers can be interpreted as a result of the unequal crossing over because unequal crossing over is frequently caused by simple repetitive sequences during the recombination process (e.g., Lewin 2003).

Second, the duplication event of the trnI–trnM1 block into the region upstream of CR1 in the common ancestor of Guibemantis and Blommersia (supplementary fig. 4, Supplementary Material online) seems to be an example of unequal crossing over as well. This nontandem duplication is difficult to explain by illegitimate recombination via minicircle (see supplementary fig. 3B, Supplementary Material online) because a minicircle insertion inevitably changes the nucleotide sequence at the insertion site (i.e., the boundary of newly copied region), and/or the insertion boundary must have an IS sequence corresponding to the remnant of the extra sequence involved in the minicircle. In contrast, in Blommersia and Guibemantis, continuous identical sequences across the boundaries (trnM–CR) of the original and the copied region are observed. These homologous sequences suggest that this duplication event has been caused by an accidental involvement of 5' flanking genes in a recombination process between CRs.

The recombination process mediated by RFB lying in CR might have played a role in the other duplication events during mantellid evolution. If the 3' end of the nascent L-strand arrested at RFB misprimes to another mtDNA molecule and illegitimate replication starts from the mispriming site, one of the resulting mtDNA molecules will contain a duplicated region corresponding to the nascent L-strand (supplementary fig. 3B, Supplementary Material online). The 3' boundaries of two nontandem duplicated regions predicted in mantellid evolution are near to the CSB region of CR (dup-2 and dup-3 in supplementary fig. 4, Supplementary Material online). This might indicate the occurrence of duplication events caused by the nascent strand mispriming process. A similar replication mode, recombination-dependent DNA replication, is known from genome systems of E. coli and other organisms and organelles; and this illicit replication mode is thought to play important roles in DNA repair and replication restart under inhibited condition of replication fork progression (Kogoma 1997).

Finally, a model of minicircle formation via two RFB (Ter) sites has been postulated for the E. coli genome (Bierne et al. 1997). Because the region between two RFBs forms a minicircle in this model, the upstream boundaries of the nontandem duplications expected in mantellid mtDNAs do not match for this model. However, this minicircle formation probably would lead to highly frequent duplication of the region between RFBs and therefore may be one possible explanation for the highly frequent gene rearrangements that are often found between duplicated CRs (Kumazawa and Endo 2004).

It is generally recognized that the CR neighborhood is one of gene rearrangement hotspot in metazoan mt genomes (e.g., Mindell et al. 1998; Boore 1999); and, as in mantellids, divergent genomic structures have been commonly observed from many animal lineages with multiple CRs (e.g., Kumazawa et al. 1996, 1998; Inoue et al. 2003). The recombination mediated gene rearrangement modes proposed here could be a general explanation for this phenomenon. On the other hand, other rearrangement hot spots, that is, WANCY and IQM trn clusters, are also known from vertebrates (see Macey et al. 1997; Inoue et al. 2003; San Mauro et al. 2006). For these CR extraneous regions, our rearrangement models seem not to be relevant directly. Indeed, even in pond-breeding mantellids, the WANCY trn cluster retains the typical vertebrate condition (Kurabayashi et al. 2006; and A.K. unpublished data). Furthermore, almost all rearrangements of these trn clusters in vertebrates can be explained by only a single tandem duplication (e.g., San Mauro et al. 2006). Thus, at present, the TDRL model would be a suitable explanation for gene rearrangements in these regions. It should be also noted that the IQM trn cluster appears to be a particularity of a duplicated CR formation. Adding to pond-breeding mantellids, several vertebrate lineages possess a second CR in this region (snakes, Kumazawa et al. 1996; a gulper eel, Inoue et al. 2003; a hyperoliid frog, A.K. unpublished data). Furthermore, several species of these taxa retain duplicated copies or remnants (=pseudogenes) of these trns around the second CR (see fig. 1 and Kumazawa et al. 1996). These may suggest that the first step of 2nd CR formation is duplication of the trn cluster by tandem duplication; and then, a CR is inserted in the previously duplicated region by recombination (see dup-1 to dup-2 in supplementary fig. 4, Supplementary Material online).

Does Recombination Discount the Phylogenetic Utility of mt Gene Arrangement?
Gene rearrangements of animal mtDNA are considered to be effective phylogenetic markers (Brown 1985; Wolstenholme 1992; Smith et al. 1993; Boore and Brown 1994, 1998; Boore et al. 2005). In fact, rearrangements supposedly occur only at low frequency and homoplasy is rare in animal mt genomes (e.g., Boore and Brown 1998; Kurabayashi et al. 2006); and the lack of recombination has been considered as one of the reasons for this (Brown 1985; Boore and Brown 1994). In this study, however, we found evidence for gene duplication events that can best be explained by general or illegitimate recombination (dup-2 to dup-6 in supplementary fig. 4, Supplementary Material online), and the resultant duplicated condition seems to be a main cause of the high frequency of subsequent gene rearrangements in pond-breeding mantellines. But at the same time, we show the good congruence of the molecular phylogenetic reconstructions with gene order synapomorphies, indicating that recombination by itself did not lead to homoplasious gene rearrangements.

To explain this pattern, several hypotheses can be drawn. 1) Recombination in mtDNA mainly begins inside CRs (RFB region), proceeds to the upstream side, and stops at a region with less sequence similarity (see supplementary fig. 3A, Supplementary Material online). Thus, gene replacement (gene conversion) occurs only exceptionally and is limited to cases where two genes with high nucleotide similarity are located at 5' sides of duplicated CRs. 2) Occurrence of unequal crossing over as an illegitimate result of homologous recombination is also limited to 5' sides of duplicated CRs and occurs at only low frequency. 3) Nontandem duplication can emerge from processes of illegitimate recombination or from mispriming on the nascent strand. However, because almost all boundaries of animal mt genes have very few intergenic nucleotides, integration of a minicircle or mispriming of a nascent DNA strand, will cause destruction of a certain single copied gene in the resulting mtDNA molecule (Boore and Brown 1994; and see supplementary fig. 2C, Supplementary Material online). Thus, such integration and mispriming events occur rarely and only at the noncoding regions with significant length (i.e., CR), previously copied regions (like dup-2 in supplementary fig. 4, Supplementary Material online), or rarely at the precise gene boundary between single-copy genes.

Although recombination by itself will thus not lead to a high degree of gene rearrangements, it can possibly become a remote cause of parallel occurrence of the same rearrangement pattern. Boore and Brown (1998) and Boore et al. (2005) pointed out that the occurrence of identical duplications followed by identical deletions is unlikely, but homoplasious rearrangement would occur if a duplicated genomic condition persisted through several lineage splits. In this context, several descendent lineages would inherit the identically duplicated genes; subsequently, independent events of identical deletions may cause the parallel occurrence of the same derived gene arrangement with relative ease in remotely related groups. Several genomic conditions reported here almost certainly correspond to this situation. For example, two copied trnI–trnM1–CR blocks seem to have been maintained through the splits of Spinomantis, Blommersia, and Guibemantis, and the trnI upstream of CR3 was deleted independently in each lineage leading to S. microtis and G. tornieri (supplementary fig. 4, Supplementary Material online). Thus, except for the ISs between nad5 and trnM1, these two species have the same derived nad5–trnM1–CR3 (rCR3 in Spinomantis) arrangement (fig. 1) despite their appartenances to rather distant phylogenetic lineages. An important conclusion from our study is that such processes can extend over very long time spans. Boore et al. (2005) mentioned that duplicated gene conditions in animal mt genomes are usually short lived. In mantellids, the copied CRs and their 5' flanking genes have been maintained by recombination for a long span through lineage splits (>33 MYA; fig. 3). There is no doubt that mt genomic rearrangements are in most cases appropriate markers to resolve ancient and fast divergence processes, but a careful estimation of the rearrangement pathway is especially required where derived arrangements consisting of CRs and their 5' genes are to be interpreted as synapomorphic characters.

	Supplementary Material

TOP Abstract Introduction Materials and Methods Results and Discussion Supplementary Material Acknowledgements References

 
Supplementary tables 1 and 2, figures 1–4, and color versions of figures 1 and 4 are available at Molecular Biology and Evolution online (http://www.mbe.oxfordjournals.org/).

	Acknowledgements

TOP Abstract Introduction Materials and Methods Results and Discussion Supplementary Material Acknowledgements References

 
Sampling and fieldwork in Madagascar has been carried out in collaboration with the Département de Biologie Animale, Université d'Antananarivo, and the Association Nationale pour la Gestion des Aires Protegées and has been supported by the German Science Foundation (grant VE247/1 to M.V.) and the Volkswagen Foundation. We are grateful to the Malagasy authorities for research and export permits. We also thank T. Nakajima for providing specimens, Y. Kumazawa, T. Yasukawa, and T. Horiuchi for valuable comments and advices on this research, and S. Schiemann and M. Kondermann for their help in the lab. The present study was supported by Grants-in-aid for Scientific research from the Japan Society for the Promotion of Science (#15255007 to M.H.) and the Ministry of Education, Culture, Sports, Science, and Technology, Japan (#16770066 and #19770064 to A.K.).

	Footnotes

 
Norihiro Okada, Associate Editor

	References

TOP Abstract Introduction Materials and Methods Results and Discussion Supplementary Material Acknowledgements References

 

Arndt A, Smith MJ. Mitochondrial gene rearrangement in the sea cucumber genus Cucumaria. Mol Biol Evol (1998) 15:1009–1016.[Abstract]

Bierne H, Ehrlich SD, Michel B. Deletions at stalled replication forks occur by two different pathways. EMBO J (1997) 2:3332–3340.

Black WCIV, Roehrdanz RL. Mitochondrial gene order is not conserved in arthropods: prostriate and metastriate tick mitochondrial genomes. Mol Biol Evol (1998) 15:1772–1785.[Abstract]

Boore JL. Survey and summary: animal mitochondrial genomes. Nucleic Acids Res (1999) 27:1767–1780.[Abstract/Free Full Text]

Boore JL. The duplication/random loss model for gene rearrangement exemplified by mitochondrial genomes of deuterostome animals. In: Computational biology series. Vol. 1—Sankoff D, Nadeau J, eds. (2000) Dordrecht (The Netherlands): Kluwer Academic Publishers. 133–147.

Boore JL, Brown WM. Mitochondrial genomes and the phylogeny of mollusks. Nautilus (1994) 108(Suppl 2):61–78.

Boore JL, Brown WM. Big trees from little genomes. Mitochondrial gene order as a phylogenetic tool. Curr Opin Genet Dev (1998) 8:668–674.[CrossRef][Web of Science][Medline]

Boore JL, Macey JR, Medina M. Sequencing and comparing whole mitochondrial genomes of animals. Methods Enzymol (2005) 395:311–348.[CrossRef][Web of Science][Medline]

Bossuyt F, Milinkovitch MC. Convergent adaptive radiations in Madagascan and Asian ranid frogs reveal covariation between larval and adult traits. Proc Natl Acad Sci USA (2000) 97:6585–6590.[Abstract/Free Full Text]

Bossuyt F, Milinkovitch MC. Amphibians as indicators of early tertiary "out-of-India" dispersal of vertebrates. Science (2001) 292:93–95.[Abstract/Free Full Text]

Bowmaker M, Yang MY, Yasukawa T, Reyes A, Jacobs HT, Huberman JA, Holt IJ. Mammalian mitochondrial DNA replicates bidirectionally from an initiation zone. J Biol Chem (2003) 278:50961–50969.[Abstract/Free Full Text]

Brandley MC, Schmitz A, Reeder TW. Partitioned Bayesian analyses, partition choice, and the phylogenetic relationships of scincid lizards. Syst Biol (2005) 54:373–390.[CrossRef][Web of Science][Medline]

Brown WM. Evolution of the animal mitochondrial DNA genome. In: Molecular evolutionary genetics—MacIntyre R, ed. (1985) New York: Plenum Press. 95–130.

Cairns SS, Bogenhagen DF. Mapping of the displacement loop within the nucleotide sequence of Xenopus laevis mitochondrial DNA. J Biol Chem (1986) 25:8481–8487.

Campbell NJ, Barker SC. The novel mitochondrial gene arrangement of the cattle tick, Boophilus microplus: fivefold tandem repetition of a coding region. Mol Biol Evol (1999) 16:732–740.[Abstract]

Castresana J. Selection of conserved blocks from multiple alignments for their use in phylogenetic analysis. Mol Biol Evol (2000) 17:540–552.[Abstract/Free Full Text]

Clayton DA. Replication of animal mitochondrial DNA. Cell (1982) 28:693–705.[CrossRef][Web of Science][Medline]

Dowton M, Campbell NJH. Intramitochondrial recombination: is it why some mitochondrial genes sleep around? Trends Ecol Evol (2001) 16:269–271.[CrossRef][Medline]

Eberhard JR, Wright TF, Bermingham E. Duplication and concerted evolution of the mitochondrial control region in the parrot genus Amazona. Mol Biol Evol (2001) 18:1330–1342.[Abstract/Free Full Text]

Endo K, Noguchi Y, Ueshima R, Jacobs HT. Novel repetitive structures, deviant protein-encoding sequences and unidentified ORFs in the mitochondrial genome of the brachiopod Lingula anatina. J Mol Evol (2005) 61:36–53.[CrossRef][Web of Science][Medline]

Frost DR, Grant T, Faivovich J, et al, (19 co-authors). The amphibian tree of life. Bull Am Mus Nat Hist (2006) 297:1–371.[CrossRef]

Gissi C, San Mauro D, Pesole G, Zardoya R. Mitochondrial phylogeny of Anura (Amphibia): a case study of congruent phylogenetic reconstruction using amino acid and nucleotide characters. Gene (2006) 366:228–237.[CrossRef][Web of Science][Medline]

Glaw F, Hoegg S, Vences M. Discovery of a new basal relict lineage of Madagascan frogs and its implications for mantellid evolution. Zootaxa (2006) 1334:27–43.

Glaw F, Vences M. Phylogeny and genus-level classification of mantellid frogs. Org Divers Evol (2006) 6:236–253.[CrossRef]

Glaw F, Vences M. A field guide to the amphibians and reptiles of Madagascar (2007) 3rd ed. Köln (Germany): Vences & Glaw Verlag. 496.

Guindon S, Gascuel O. A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol (2003) 52:696–704.[CrossRef][Web of Science][Medline]

Hasegawa M, Kishino H. Accuracies of the simple methods for estimating the bootstrap probability of a maximum likelihood tree. Mol Biol Evol (1994) 11:142–145.[Web of Science]

Henikoff S. Exonuclease III generated deletions for DNA sequence analysis. Promega Notes No 8: (1987) 1–3.

Hoegg S, Vences M, Brinkmann H, Meyer A. Phylogeny and comparative substitution rates of frogs inferred from sequences of three nuclear genes. Mol Biol Evol (2004) 21:1188–1200.[Abstract/Free Full Text]

Hoffmann RJ, Boore JL, Brown WM. A novel mitochondrial genome organization for the blue mussel, Mytilus edulis. Genetics (1992) 131:397–412.[Abstract]

Holliday R. The induction of mitotic recombination by mitomycin C in Ustilago and Saccharomyces. Genetics (1964) 50:323–335.[Free Full Text]

Holt IJ, Dunbar DR, Jacobs HT. Behaviour of a population of partially duplicated mitochondrial DNA molecules in cell culture: segregation, maintenance and recombination dependent upon nuclear background. Hum Mol Genet (1997) 6:1251–1260.[Abstract/Free Full Text]

Holt IJ, Lorimer HE, Jacobs HT. Coupled leading- and lagging-strand synthesis of mammalian mitochondrial DNA. Cell (2000) 100:515–524.[CrossRef][Web of Science][Medline]

Howell N. mtDNA recombination: what do in vitro data mean? Am J Hum Genet (1997) 61:19–22.[CrossRef][Web of Science][Medline]

Igawa T, Kurabayashi A, Usuki C, Fujii T, Sumida M. Complete mitochondrial genomes of three neobatrachian anurans: a case study of divergence time estimation using different data and calibration settings. Gene (2008) 407:116–129.[CrossRef][Web of Science][Medline]

Inoue JG, Miya M, Tsukamoto K, Nishida M. Evolution of the deep-sea gulper eel mitochondrial genomes: large-scale gene rearrangements originated within the eels. Mol Biol Evol (2003) 20:1917–1924.[Abstract/Free Full Text]

Kajander OA, Karhunen PJ, Holt IJ, Jacobs HT. Prominent mitochondrial DNA recombination intermediates in human heart muscle. EMBO Rep (2001) 2:1007–1012.[CrossRef][Web of Science][Medline]

Kajander OA, Rovio AT, Majamaa K, Poulton J, Spelbrink JN, Holt IJ, Karhunen PJ, Jacobs HT. Human mtDNA sublimons resemble rearranged mitochondrial genomes found in pathological states. Hum Mol Genet (2000) 9:2821–2835.[Abstract/Free Full Text]

Kogoma T. Stable DNA replication: interplay between DNA replication, homologous recombination, and transcription. Microbiol Mol Biol Rev (1997) 61:212–238.[Abstract]

Kraytsberg Y, Schwartz M, Brown TA, Ebralidse K, Kunz WS, Clayton DA, Vissing J, Khrapko K. Recombination of human mitochondrial DNA. Science (2004) 304:981.[Free Full Text]

Kumar S. Phyltest: phylogeny hypothesis testing software (1996) Philadelphia (PA): The Pennsylvania State University.

Kumazawa Y, Endo H. Mitochondrial genome of the Komodo dragon: efficient sequencing method with reptile-oriented primers and novel gene rearrangements. DNA Res (2004) 11:115–125.[Abstract]

Kumazawa Y, Ota H, Nishida M, Ozawa T. Gene rearrangements in snake mitochondrial genomes: highly concerted evolution of control-region-like sequences duplicated and inserted into a tRNA gene cluster. Mol Biol Evol (1996) 13:1242–1254.[Abstract]

Kumazawa Y, Ota H, Nishida M, Ozawa T. The complete nucleotide sequence of a snake (Dinodon semicarinatus) mitochondrial genome with two identical control regions. Genetics (1998) 150:313–329.[Abstract/Free Full Text]

Kurabayashi A, Ueshima R. Complete sequence of the mitochondrial DNA of the primitive opisthobranch gastropod Pupa strigosa: systematic implication of the genome organization. Mol Biol Evol (2000) 17:266–277.[Abstract/Free Full Text]

Kurabayashi A, Usuki C, Mikami N, Fujii T, Yonekawa H, Sumida M, Hasegawa M. Complete nucleotide sequence of the mitochondrial genome of a Malagasy poison frog Mantella madagascariensis: evolutionary implications on mitochondrial genomes of higher anuran groups. Mol Phylogenet Evol (2006) 39:223–236.[CrossRef][Web of Science][Medline]

Ladoukakis ED, Zouros E. Recombination in animal mitochondrial DNA: evidence from published sequences. Mol Biol Evol (2001) 18:2127–2131.[Free Full Text]

Lavrov DV, Boore JL, Brown WM. Complete mtDNA sequences of two millipedes suggest a new model for mitochondrial gene rearrangements: duplication and nonrandom loss. Mol Biol Evol (2002) 19:163–169.[Abstract/Free Full Text]

Lee JS, Miya M, Lee YS, Kim CG, Park EH, Aoki Y, Nishida M. The complete DNA sequence of the mitochondrial genome of the self-fertilizing fish Rivulus marmoratus (Cyprinodontiformes, Rivulidae) and the first description of duplication of a control region in fish. Gene (2001) 280:1–7.[CrossRef][Web of Science][Medline]

Lewin G. Genes VIII (2003) Princeton (NJ): Pearson Prentice Hall.

Lunt DF, Hyman BC. Animal mitochondrial DNA recombination. Nature (1997) 387:247.[CrossRef][Medline]

Mabuchi K, Miya M, Sato TP, Westneat MW, Nishida M. Gene rearrangements and evolution of tRNA pseudogenes in the mitochondrial genome of the parrotfish (Teleostei: Perciformes: Scaridae). J Mol Evol (2004) 59:287–297.[CrossRef][Web of Science][Medline]

Macey JR, Larson A, Ananjeva NB, Fang Z, Papenfuss TJ. Two novel gene orders and the role of light-strand replication in rearrangement of the vertebrate mitochondrial genome. Mol Biol Evol (1997) 14:91–104.[Abstract]

Mindell DP, Sorenson MD, Dimcheff DE. Multiple independent origins of mitochondrial gene order in birds. Proc Natl Acad Sci USA (1998) 95:10693–10697.[Abstract/Free Full Text]

Moritz C, Dowling TE, Brown WM. Tandem duplications in animal mitochondrial DNAs: variation in incidence and gene content among lizards. Proc Natl Acad Sci USA (1987) 84:7183–7187.[Abstract/Free Full Text]

Mueller RL, Boore JL. Molecular mechanisms of extensive mitochondrial gene rearrangement in plethodontid salamanders. Mol Biol Evol (2005) 22:2104–2112.[Abstract/Free Full Text]

Noguchi Y, Endo K, Tajima F, Ueshima R. The mitochondrial genome of the brachiopod Laqueus rubellus. Genetics (2000) 155:245–259.[Abstract/Free Full Text]

Ogoh K, Ohmiya Y. Concerted evolution of duplicated control regions within an ostracod mitochondrial genome. Mol Biol Evol (2007) 24:74–78.[Abstract/Free Full Text]

Pääbo S, Irwin DM, Wilson AC. DNA damage promotes jumping between templates during enzymatic amplification. J Biol Chem (1990) 265:4718–4721.[Abstract/Free Full Text]

Posada D, Crandall KA. Modeltest: testing the model of DNA substitution. Bioinformatics (1998) 14:817–818.[Abstract/Free Full Text]

Rawson PD. Nonhomologous recombination between the large unassigned region of the male and female mitochondrial genomes in the mussel, Mytilus trossulus. J Mol Evol (2005) 61:717–732.[CrossRef][Web of Science][Medline]

Reyes A, Yang MY, Bowmaker M, Holt IJ. Bidirectional replication initiates at sites throughout the mitochondrial genome of birds. J Biol Chem (2005) 280:3242–3250.[Abstract/Free Full Text]

Richards CM, Nussbaum RA, Raxworthy J. Phylogenetic relationships within Madagascan boophids and mantellids as elucidated by mitochondrial ribosomal genes. Afr J Herpetol (2000) 49:23–32.

Roe BA, Ma DP, Wilson RK, Wong JFH. The complete nucleotide sequence of the Xenopus laevis mitochondrial genome. J Biol Chem (1985) 260:9759–9774.[Abstract/Free Full Text]

Roelants K, Gower DJ, Wilkinson M, Loader SP, Biju SD, Guillaume K, Moriau L, Bossuyt F. Global patterns of diversification in the history of modern amphibians. Proc Natl Acad Sci USA (2007) 104:887–892.[Abstract/Free Full Text]

Roelants K, Jiang J, Bossuyt F. Endemic ranid (Amphibia: Anura) genera in southern mountain ranges of the Indian subcontinent represent ancient frog lineages: evidence from molecular data. Mol Phylogenet Evol (2004) 31:730–740.[CrossRef][Medline]

Ronquist F, Huelsenbeck JP. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics (2003) 19:1572–1574.[Abstract/Free Full Text]

San Mauro D, García-París M, Zardoya R. Phylogenetic relationships of discoglossid frogs (Amphibia: Anura: Discoglossidae) based on complete mitochondrial genomes and nuclear genes. Gene (2004) 343:357–366.[CrossRef][Web of Science][Medline]

San Mauro D, Gower DJ, Zardoya R, Wilkinson M. A hotspot of gene order rearrangement by tandem duplication and random loss in the vertebrate mitochondrial genome. Mol Biol Evol (2006) 23:227–234.[Abstract/Free Full Text]

San Mauro D, Vences M, Alcobendas M, Zardoya R, Meyer A. Initial diversification of living amphibians predated the breakup of Pangaea. Am Nat (2005) 165:590–599.[CrossRef][Web of Science][Medline]

Sano N, Kurabayashi A, Fujii T, Yonekawa H, Sumida M. Complete nucleotide sequence and gene rearrangement of the mitochondrial genome of the bell-ring frog, Buergeria buergeri (family Rhacophoridae). Genes Genet Syst (2004) 79:151–163.[CrossRef][Web of Science][Medline]

Sano N, Kurabayashi A, Fujii T, Yonekawa H, Sumida M. Complete nucleotide sequence of the mitochondrial genome of Schlegel's tree frog Rhacophorus schlegelii (family Rhacophoridae): duplicated control regions and gene rearrangements. Genes Genet Syst (2005) 80:213–224.[CrossRef][Web of Science][Medline]

Sato A, Nakada K, Akimoto M, Ishikawa K, Ono T, Shitara H, Yonekawa H, Hayashi J. Rare creation of recombinant mtDNA haplotypes in mammalian tissues. Proc Natl Acad Sci USA (2005) 102:6057–6062.[Abstract/Free Full Text]

Shadel GS, Clayton DA. Mitochondrial DNA maintenance in vertebrates. Annu Rev Biochem (1997) 66:409–435.[CrossRef][Web of Science][Medline]

Shao R, Dowton M, Murrell A, Barker SC. Rates of gene rearrangement and nucleotide substitution are correlated in the mitochondrial genomes of insects. Mol Biol Evol (2003) 20:1612–1619.[Abstract/Free Full Text]

Shimodaira H, Hasegawa M. CONSEL: a program for assessing the confidence of phylogenetic tree selection. Bioinformatics (2001) 17:1246–1247.[Abstract/Free Full Text]

Smith MJ, Arndt A, Gorski S, Fajber E. The phylogeny of echinoderm classes based on mitochondrial gene arrangements. J Mol Evol (1993) 36:545–554.[CrossRef][Web of Science][Medline]

Sumida M, Kanamori Y, Kaneda H, Kato Y, Nishioka M, Hasegawa M, Yonekawa H. Complete nucleotide sequence and gene rearrangement of the mitochondrial genome of the Japanese pond frog Rana nigromaculata. Genes Genet Syst (2001) 76:311–325.[CrossRef][Web of Science][Medline]

Swofford DL. PAUP*: phylogenetic analysis using parsimony (*and other methods). Version 4 (2003) Sunderland (MA): Sinauer Associates.

Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG. The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res (1997) 24:4876–4882.

Thorne JL, Kishino H. Divergence time and evolutionary rate estimation with multilocus data. Syst Biol (2002) 51:689–702.[CrossRef][Web of Science][Medline]

Thyagarajan B, Padua RA, Campbell C. Mammalian mitochondria possess homologous DNA recombination activity. J Biol Chem (1996) 271:27536–27543.[Abstract/Free Full Text]

Tsaousis AD, Martin DP, Ladoukakis ED, Posada D, Zouros E. Widespread recombination in published animal mtDNA sequences. Mol Biol Evol (2005) 22:925–933.[Abstract/Free Full Text]

Van Bocxlaer I, Roelants K, Biju SD, Nagaraju J, Bossuyt F. Late Cretaceous vicariance in Gondwanan amphibians. PLoS ONE (2006) 1:e74.[CrossRef]

Van der Meijden A, Vences M, Hoegg S, Boistel R, Channing A, Meyer A. Nuclear gene phylogeny of narrow-mouthed toads (family Microhylidae) and a discussion of competing hypotheses concerning their biogeographical origins. Mol Phylogenet Evol (2007) 44:1017–1030.[CrossRef][Medline]

Van der Meijden A, Vences M, Hoegg S, Meyer A. A previously unrecognized radiation of ranid frogs in southern Africa revealed by nuclear and mitochondrial DNA sequences. Mol Phylogenet Evol (2005) 37:674–685.[CrossRef][Web of Science][Medline]

Vences M, Andreone F, Glaw F, Kosuch J, Meyer A, Schaefer H-C, Veith M. Exploring the potential of life-history key innovation: brook breeding in the radiation of the Malagasy treefrog genus Boophis. Mol Ecol (2002) 11:1453–1463.[CrossRef][Medline]

Vences M, Andreone F, Glaw F, Randrianirina JE. Molecular and bioacoustic divergence in Mantidactylus granulatus and M. zavona sp. n. (Anura: Mantellidae): bearings for the biogeography of northern Madagascar. Afr Zool (2003) 38:67–78.

Vences M, Vieites DR, Glaw F, Brinkmann H, Kosuch J, Veith M, Meyer A. Multiple overseas dispersal in amphibians. Proc R Soc Lond Ser B (2003) 270:2435–2442.[Medline]

Vences M, Wahl-Boos G, Hoegg S, Glaw F, Spinelli O, Meyer A, Perry S. Molecular systematics of mantelline frogs from Madagascar and evolution of their femoral glands. Biol J Linn Soc (2007) 92:529–539.[CrossRef][Web of Science]

Wolstenholme DR. Animal mitochondrial DNA: structure and evolution. In: Mitochondrial genomes—Wolstenholme DR, Jeon KW, eds. (1992) New York: Academic Press. 173–216.

Xu W, Jameson D, Tang B, Higgs PG. The relationship between the rate of molecular evolution and the rate of genome rearrangement in animal mitochondrial genomes. J Mol Evol (2006) 63:375–392.[CrossRef][Web of Science][Medline]

Yang MY, Bowmaker M, Reyes A, Vergani L, Angeli P, Gringeri E, Jacobs HT, Holt IJ. Biased incorporation of ribonucleotides on the mitochondrial L-strand accounts for apparent strand-asymmetric DNA replication. Cell (2002) 111:495–505.[CrossRef][Web of Science][Medline]

Yasukawa T, Reyes A, Cluett TJ, Yang MY, Bowmaker M, Jacobs HT, Holt IJ. Replication of vertebrate mitochondrial DNA entails transient ribonucleotide incorporation throughout the lagging strand. EMBO J (2006) 25:5358–5371.[CrossRef][Web of Science][Medline]

Yokobori S, Fukuda N, Nakamura M, Aoyama T, Oshima T. Long-term conservation of six duplicated structural genes in cephalopod mitochondrial genomes. Mol Biol Evol (2004) 21:2034–2046.[Abstract/Free Full Text]

Yokobori S, Oshima T, Wada H. Complete nucleotide sequence of the mitochondrial genome of Doliolum nationalis with implications for evolution of urochordates. Mol Phylogenet Evol (2005) 34:273–283.[CrossRef][Web of Science][Medline]

Accepted for publication January 29, 2008.

CiteULike    Connotea    Del.icio.us    What's this?

This article has been cited by other articles:

				M. Dowton, S. L. Cameron, J. I. Dowavic, A. D. Austin, and M. F. Whiting Characterization of 67 Mitochondrial tRNA Gene Rearrangements in the Hymenoptera Suggests That Mitochondrial tRNA Gene Position Is Selectively Neutral Mol. Biol. Evol., July 1, 2009; 26(7): 1607 - 1617. [Abstract] [Full Text] [PDF]

				O. Verneau, L. H Du Preez, V. Laurent, L. Raharivololoniaina, F. Glaw, and M. Vences The double odyssey of Madagascan polystome flatworms leads to new insights on the origins of their amphibian hosts Proc R Soc B, May 7, 2009; 276(1662): 1575 - 1583. [Abstract] [Full Text] [PDF]

This Article Abstract FREE Full Text (PDF) Supplementary Data All Versions of this Article: 25/5/874    most recent msn031v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Request Permissions Google Scholar Articles by Kurabayashi, A. Articles by Hasegawa, M. Search for Related Content PubMed PubMed Citation Articles by Kurabayashi, A. Articles by Hasegawa, M. Social Bookmarking          What's this?

Online ISSN 1537-1719 - Print ISSN 0737-4038

Copyright © 2009 Society for Molecular Biology and Evolution

Oxford Journals Oxford University Press

• Site Map
• Privacy Policy
• Frequently Asked Questions

Other Oxford University Press sites: Oxford University Press Oxford Journals China Oxford Journals Japan American National Biography Booksellers' Information Service Children's Fiction and Poetry Children's Reference Corporate & Special Sales Dictionaries Dictionary of National Biography Digital Reference English Language Teaching Higher Education Textbooks Humanities International Education Unit Law Medicine Music Online Products Oxford English Dictionary Reference Rights and Permissions Science School Books Social Sciences Very Short Introductions World's Classics