As a basis for assessing the phylogenetic performance of cytochrome b and analyzing the molecular evolution of this gene in mantellid frogs, we studied three separate data sets:
Dataset 1: Fragments of 551 bp of the cytochrome b gene as amplified via the polymerase chain reaction (PCR) using the primers Cytb-c and CBJ10933 from Bossuyt and Milinkovitch (2000), from a total of 196 specimens belonging to all five mantellid genera and representing 112 of the 158 mantellid species known (two species of Aglyptodactylus, 34 species of Boophis, 1 species of Laliostoma, 15 species of Mantella, 60 species of Mantidactylus; Glaw and Vences 2003). We included in our analyses sequences of eleven specimens belonging to the genera Paradoxophyla, Platypelis, Plethodontohyla, Rhombophryne, Scaphiophryne and Stumpffia (family
consists of 183 amino acids (from amino acid 144 to amino acid 327). Based on the model of the cytochrome b protein (Degli Esposti et al. 1993) the used fragment spans all the three above-mentioned regions of the mitochondrial membrane.
Dataset 2: Fragments of cytochrome b of 356 specimens of Mantella, largely homologous to the fragment of dataset 1 but slightly shorter (502 bp), obtained in previous studies (Chiari et al. 2004; Chiari et al. in press; Vieites et al. in press). These were collapsed into 122 different unique sequences using Collapse v 1.2 (Posada 2004).
These were used in our study to compare the saturation of the gene among mantellids at the shallower level of closely related species.
Dataset 3: Sequences of fragments of up to 360 bp of the 16S rRNA gene as amplified using the primers 16SAr-L and 16Br-H of Palumbi et al. (1991) were available for a subset of 177 mantellid specimens out of the 196 in dataset 1. Further sequences of five species of microhylids to be used as outgroup were downloaded from GenBank (Accession numbers: Stumpffia AY594124, Scaphiophryne AY594127, Rhombophryne AY594125, Plethodontohyla AY594118, Platypelis AY594103). The complete fragment was used for calculation of pairwise distance whereas hypervariable and gapped sites of the alignment were excluded for phylogenetic analysis (see below).
Molecular protocols
Total genomic DNA was extracted from toe clips fixed in 99% ethanol using a proteinase K digestion (final concentration 1 mg/mL). DNA was isolated by a standard salt extraction protocol (Bruford et al. 1992). Representative voucher specimens are stored in the collections of the Zoological Museum Amsterdam, Netherlands, and the Zoologische Staatssammlung München, Germany.
To amplify the cytochrome b and 16S rRNA fragments, polymerase chain reactions (PCRs) were performed in 25 µL reactions containing 1.0 unit of REDTaq DNA Polymerase (Sigma, Taufkirchen, Germany), 50 ng genomic DNA, 10 pmol of each primer, 15 nmol of each dNTP, 50 nmol additional MgCl2 and the REDTaq PCR reaction buffer (10 mM Tris-HCl, pH 8.3, 50 mM KCl, 1.1 mM MgCl2 and 0.01%
gelatine) using the following conditions: an initial denaturation at 94°C for 1:30 min; 35 cycles at 94°C for 30 seconds, primer-specific annealing temperature (53°C for cytochrome b and 55°C for 16S rRNA) for 45 seconds, extension at 72°C for 1:30 min;
final extension of 10:00 min at 72°C.
PCR products were checked on 1% agarose gels and purified using QIAquick spin columns (Qiagen) prior to cycle sequencing. Sequence data collection and visualisation were performed on an ABI 3100 automated sequencer. Sequencing reactions were prepared according to the manufacturers instructions, using the ABI sequence mix (BigDye® Terminator V3.1 Sequencing Standard, Applied Biosystems).
Newly obtained sequences were deposited in GenBank; accession numbers: ####-####
[to be added upon manuscript acceptance].
Data analysis
Unambiguous alignment of the cytochrome b sequences was possible by eye, as they contained no indels. Sequences were verified and aligned with Sequence Navigator (Applied Biosystems) software. The software TaxI (Steinke et al. 2005) was used to pairwise align the complete 16S rRNA fragments and calculate pairwise divergences.
For phylogenetic analysis the 16S rRNA sequences were aligned using the "Clustal"
option in the program Sequence Navigator (Applied Biosystems). All hypervariable and gapped regions were excluded, leaving fragments of 360 bp for phylogenetic analyses.
We estimated synonymous and nonsynonymous mutation rates in pairwise comparisons using the program PAML version 3.14 (Yang 1997), following the method of Yang and Nielsen (2000).
To evaluate the degree of saturation of the cytochrome b gene in datasets 1 and 2, uncorrected p-distances of each codon positions were plotted against the uncorrected p-distance of the entire fragment. Total uncorrected p-distances of the cytochrome b gene were plotted against both absolute numbers of transition (Ts) and of transversion (Tv). For the 176 specimens of dataset 3, fragments of both the cytochrome b and 16S rRNA genes were available. Pairwise uncorrected p-distances, and corrected K2P distances for the cytochrome b gene among congeneric species, were calculated using the software TaxI (Steinke et al. 2005). Pairwise uncorrected p-distances of each codon position of the cytochrome b gene and of the entire gene were plotted against the pairwise distance of the 16S rRNA fragments, using the software TaxI (Steinke et al.
2005). We largely used uncorrected distances to avoid any bias that could originate by correcting partitions of possibly different molecular evolution (cytochrome b vs. 16S rRNA, and each of the three codon positions) arbitrarily with the same model.
(Swofford 2002), using the heuristic search option with tree-bisection-reconnection (TBR) branch swapping and ten random addition sequence replicates. Maximum likelihood (ML) analyses were carried out using PHYML (Guidon and Gascuel 2003) following substitution model parameter estimation using the AIC criterion implemented in Modeltest version 3.06 (Posada and Crandall 1998). Five hundred bootstrap replicates were computed to evaluate the stability of nodes. Bayesian posterior probabilities were calculated using MrBayes, version 3b4 (Huelsenbeck and Ronquist 2001). 1.000.000 generations were run, every tenth tree collected and the "burn in"
parameter was set at 0.5% based on empirical evaluation.
Phylogenetic trees were separately computed for cytochrome b and 16S rRNA (datasets 1 and 3). For cytochrome b, we calculated trees based on (1) the complete fragment and (2) after removal of third positions. Unpartitioned Bayesian analyses were carried out for datasets 1 and 3, and a partitioned Bayesian analysis (with separate models of evolution for each codon position) was run for dataset 1.
Due to the many closely related taxa in the dataset, MP analyses were completed for 1010 rearrangements for the analysis including first and second codon position and for 109 rearrangement for the one including all the three positions, and a strict consensus tree was calculated from the retained trees.
7.4. RESULTS