124
F. Marlétaz and Y. Le Parco
some characters are more difficult to interpret because they could lead to contradictory associations: common segmented bodies have historically been used to cluster
annelids and arthropods into the “Articulés” group (Cuvier 1817) whereas other
prominent characters such as a spiral cleavage pattern and trochophore-type larvae
are shared between annelids and platyhelminthes, whose morphology is far simpler,
but not arthropods (Nielsen 2001).
The cladistic method founded by Willi Hennig in 1950 provided an opportunity
to deal more accurately with character evolution and to improve morphology-based
classification (Hennig 1966). The cladistic method rejects groupings supported by
the lack of a given character and conversely proposes that bona fide groupings (i.e.
clades) should be supported by unambiguous shared derived characters (i.e. synapomorphies). This framework prompted extensive studies of morphological characters
in animal phylogeny, in particular through the development of parsimony algorithms
aimed at inferring a tree from a character matrix (Swofford 1990). For instance, the
previous example of Spiralia versus Articulata was addressed in a landmark study,
which rejected the Articulata hypothesis using a character matrix relying on careful
definition of characters (Eernisse et al. 1992). This study thus led to the proposition
that segmentation was established convergently in both annelids and arthropods, and
uncovered novel new clades reminiscent of the “new view” of animal phylogeny
(cf. infra).
Despite their successes, the studies based on morphological matrices were extensively criticized. Some authors simply proposed that morphology does not provide
a sufficiently large set of unambiguous characters, in particular because of problems
related to homology assessment (Scotland et al. 2003).
One may consider that such opinions are overstated. Methodological advances
can bring generate new lines of morphological evidence and thereby increase the
repertory of available characters. Ultrastructural studies performed using confocal
laser scanning microscopy have recently shed new light on some debates, for example by providing evidence for a close relationship between sipunculids and annelids
despite the apparent lack of segmentation in sipunculids (Wanninger et al. 2005).
Similarly, 4D microscopy technique has offered opportunities to investigate cell lineages and fate maps with greater flexibility, and this has lead to some unexpected
conclusions such as the potentially derived state of the appendicularian Oikopleura
dioica with regard to tunicates (Stach et al. 2008).
However, numerous problems have been detected in the character matrices generally employed (Jenner 2001). In particular, the discrete coding of characters is
rarely compatible with the subtle discussion of homology advocated by original
cladistic thought (Hennig 1966, Jenner 2001). These issues could be especially misleading in the context of so-called total evidence studies that performed inference
using a composite molecular and morphological dataset analyzed in a parcimony
framework (Giribet et al. 2000, Jenner 2001). Nevertheless, this has debate stressed
the importance of dealing with an alternative class of evidence – molecular data – in
order to evaluate the phylogenetic accuracy of commonly discussed morphological
characters.
F. Marlétaz and Y. Le Parco
some characters are more difficult to interpret because they could lead to contradictory associations: common segmented bodies have historically been used to cluster
annelids and arthropods into the “Articulés” group (Cuvier 1817) whereas other
prominent characters such as a spiral cleavage pattern and trochophore-type larvae
are shared between annelids and platyhelminthes, whose morphology is far simpler,
but not arthropods (Nielsen 2001).
The cladistic method founded by Willi Hennig in 1950 provided an opportunity
to deal more accurately with character evolution and to improve morphology-based
classification (Hennig 1966). The cladistic method rejects groupings supported by
the lack of a given character and conversely proposes that bona fide groupings (i.e.
clades) should be supported by unambiguous shared derived characters (i.e. synapomorphies). This framework prompted extensive studies of morphological characters
in animal phylogeny, in particular through the development of parsimony algorithms
aimed at inferring a tree from a character matrix (Swofford 1990). For instance, the
previous example of Spiralia versus Articulata was addressed in a landmark study,
which rejected the Articulata hypothesis using a character matrix relying on careful
definition of characters (Eernisse et al. 1992). This study thus led to the proposition
that segmentation was established convergently in both annelids and arthropods, and
uncovered novel new clades reminiscent of the “new view” of animal phylogeny
(cf. infra).
Despite their successes, the studies based on morphological matrices were extensively criticized. Some authors simply proposed that morphology does not provide
a sufficiently large set of unambiguous characters, in particular because of problems
related to homology assessment (Scotland et al. 2003).
One may consider that such opinions are overstated. Methodological advances
can bring generate new lines of morphological evidence and thereby increase the
repertory of available characters. Ultrastructural studies performed using confocal
laser scanning microscopy have recently shed new light on some debates, for example by providing evidence for a close relationship between sipunculids and annelids
despite the apparent lack of segmentation in sipunculids (Wanninger et al. 2005).
Similarly, 4D microscopy technique has offered opportunities to investigate cell lineages and fate maps with greater flexibility, and this has lead to some unexpected
conclusions such as the potentially derived state of the appendicularian Oikopleura
dioica with regard to tunicates (Stach et al. 2008).
However, numerous problems have been detected in the character matrices generally employed (Jenner 2001). In particular, the discrete coding of characters is
rarely compatible with the subtle discussion of homology advocated by original
cladistic thought (Hennig 1966, Jenner 2001). These issues could be especially misleading in the context of so-called total evidence studies that performed inference
using a composite molecular and morphological dataset analyzed in a parcimony
framework (Giribet et al. 2000, Jenner 2001). Nevertheless, this has debate stressed
the importance of dealing with an alternative class of evidence – molecular data – in
order to evaluate the phylogenetic accuracy of commonly discussed morphological
characters.