Outside of the mammalian clade, which resolved as expected (V asudevan et al. 2007), only three clades formed monophyletic groups with significant bootstrap support using both the maximum likelihood (ML) and maximum parsimony (MP) methods.
Phylogenetic analysis of odd-skipped protein was performed using Maximum likelihood analysis with 1000-fold bootsupportusingt, using RAxML via the CIPRES portal (Copf et al., 2003 ; Nagy et al., 1994 ).
Phylogenetic analysis of hairy-related protein sequences was performed using Maximum likelihood analysis with 1000-fold bootsupportusingt, using RAxML via the CIPRES portal (Copf et al., 2003 ; Nagy et al., 1994 ).
The three species delimited from the mitochondrial and ITS2 datasets were recovered as distinct clades of ITS1 sequences: the monophyly of species B was very strongly supported (>98% bootstrap support using all three methods), whereas the monophyly of species C received weaker bootstrap support and the monophyly of species A was very weakly supported.
The monophyly of clades A and B in our ITS2 dataset received very strong bootstrap support using all three methods, whereas the monophyly of clade C was only weakly supported using maximum likelihood (<50% bootstrap support) and not at all supported using neighbor-joining and parsimony.
All previously published ITS1 sequences from Australia, Malaysia, the South China Sea, Taiwan and Japan turned out to belong to species C, whereas all published Stylophora sequences from the Red Sea fell in a well supported fourth clade D (>90% bootstrap support using maximum likelihood and neighbor-joining) that can be considered a distinct species following the criterion of reciprocal monophyly.
Bootstrap support using 1000 replicates was calculated using MEGA version 4.0 [54.0
All three clades were very strongly supported (100% bootstrap support using all three methods).
The 96 GRXs were grouped into four distinct groups with high bootstrap support using ClustalX program (Fig. 1A).
Maximum likelihood analyses were performed in PHYML [ 42] with 100 bootstrap replications for node support, using the best-fit models estimated with AIC in Modeltest.
