128
F. Marlétaz and Y. Le Parco
the systematic error because evolutionary models generally fit short and homogeneous one-gene alignments better. However, tree combination methods generally
cause loss of information compared to full data phylogenetic estimates (de Queiroz
and Gatesy 2007). The second approach, the supermatrix strategy, involves the
concatenation of all marker genes in a single alignment. The analysis of such a
large number of positions can be difficult in terms of models and computation
but this approach has benefited considerably from recent improvements of inference methods. For instance, efficient maximum-likelihood heuristics (Guindon and
Gascuel 2003, Stamatakis 2006) as well as accurate Bayesian reconstruction algorithms (Huelsenbeck et al. 2001) have recently been developed, advances which
have enabled the implementation of realistic model of evolution. The supermatrix
is thus currently considered as the method of choice for most studies focusing on
animal relationships.
The second, more qualitative, phylogenomic approach mentioned at the start
of this section relies on identification of genome-level features that could constitute molecular signatures valuable as informative characters. Some authors have
coined the term “Rare Genomic Changes” to describe this new class of characters
and they have emphasized their interest as a means of overcoming problems with
large-scale sequence-based phylogenetic inference or at least as a mean of completing such studies (Rokas and Holland 2000a). Several kinds of genome features
can provide valuable phylogenetic information, as depicted by the following examples. The structure of multigenic families and especially the loss and gain of genes
have provided some very interesting clues for the resolution of some problematic
nodes (Copley et al. 2004, Marlétaz et al. 2006). Alternatively, intron positions
within homologous genes are broadly conserved in metazoans, as testified by the
identification of conserved introns in the cnidarian Nematostella vectensis or the
annelid Platynereis dumerii (Putnam et al. 2007, Raible et al. 2005). The pattern
of intron conservation has thus been proposed to be suitable for phylogenetic inference (Roy and Gilbert 2005). Also, gene order and genome rearrangements have
been extensively studied and have yielded very interesting findings, especially as
far as mitochondrial gene arrangements are concerned (Boore 2006).
The growing amount of genomic data available for a large set of metazoans has
made it possible to address the question of metazoan relationships using these different approaches and the contrasting results collected so far have made an important
contribution towards renewing the debate about metazoan phylogeny.
4.4 Phylogenomics Resolves Animal Relationships
Early attempts to improve metazoan relationships using phylogenomic approaches
often generated contrasting evidence and sometimes challenged the results of the
“new view” of animal phylogeny. The settlement of these questions stressed the
importance that should be given to taxonomic sampling.
F. Marlétaz and Y. Le Parco
the systematic error because evolutionary models generally fit short and homogeneous one-gene alignments better. However, tree combination methods generally
cause loss of information compared to full data phylogenetic estimates (de Queiroz
and Gatesy 2007). The second approach, the supermatrix strategy, involves the
concatenation of all marker genes in a single alignment. The analysis of such a
large number of positions can be difficult in terms of models and computation
but this approach has benefited considerably from recent improvements of inference methods. For instance, efficient maximum-likelihood heuristics (Guindon and
Gascuel 2003, Stamatakis 2006) as well as accurate Bayesian reconstruction algorithms (Huelsenbeck et al. 2001) have recently been developed, advances which
have enabled the implementation of realistic model of evolution. The supermatrix
is thus currently considered as the method of choice for most studies focusing on
animal relationships.
The second, more qualitative, phylogenomic approach mentioned at the start
of this section relies on identification of genome-level features that could constitute molecular signatures valuable as informative characters. Some authors have
coined the term “Rare Genomic Changes” to describe this new class of characters
and they have emphasized their interest as a means of overcoming problems with
large-scale sequence-based phylogenetic inference or at least as a mean of completing such studies (Rokas and Holland 2000a). Several kinds of genome features
can provide valuable phylogenetic information, as depicted by the following examples. The structure of multigenic families and especially the loss and gain of genes
have provided some very interesting clues for the resolution of some problematic
nodes (Copley et al. 2004, Marlétaz et al. 2006). Alternatively, intron positions
within homologous genes are broadly conserved in metazoans, as testified by the
identification of conserved introns in the cnidarian Nematostella vectensis or the
annelid Platynereis dumerii (Putnam et al. 2007, Raible et al. 2005). The pattern
of intron conservation has thus been proposed to be suitable for phylogenetic inference (Roy and Gilbert 2005). Also, gene order and genome rearrangements have
been extensively studied and have yielded very interesting findings, especially as
far as mitochondrial gene arrangements are concerned (Boore 2006).
The growing amount of genomic data available for a large set of metazoans has
made it possible to address the question of metazoan relationships using these different approaches and the contrasting results collected so far have made an important
contribution towards renewing the debate about metazoan phylogeny.
4.4 Phylogenomics Resolves Animal Relationships
Early attempts to improve metazoan relationships using phylogenomic approaches
often generated contrasting evidence and sometimes challenged the results of the
“new view” of animal phylogeny. The settlement of these questions stressed the
importance that should be given to taxonomic sampling.