Archosauria: Dendrogram

Cladogram

For taxa within Archosauria, this cladogram is based on a single most parsimonious tree generated from 133 characters for 20 taxa derived as follows:

1-73 from Benton (1999).

74-100 from Gower (2002).

101-133 from Olsen et al. (2000)  As revised and corrected by Benton & Walker (2002).

Additional data for Erpetosuchus hand coded from Benton & Walker 2002)

Additional data for Doswellia hand coded from Weems, RE (1980).

The analysis was performed on PHYLIP, using the DNAPARS program. This program treats all characters as unordered and (as applied here) optimizes state changes in a manner similar to the DELTRAN option of PAUP. "Not applicable" characters were coded as such, rather than as unknown. In one or two cases, included taxa were assumed to have the same character states as a higher level taxon in order to promote comparability. Thus, Batrachotomus and Saurosuchus were assumed to have the "prestosuchid" states other than braincase characters, since they were scored separately by Gower, but not in the other sources. Other than this, the hand-coded Doswellia data, and the partially hand-coded Erpetosuchus data, no attempt was made to code characters not scored in the original reports. The data matrix is available on request -- but, frankly, you'd have to be out of your mind to rely on it.

There is some overlap between the characters from Olsen et al. (2000) and Benton 1999). No adjustment was made for the duplication. The objective of the study was largely to see if the close aetosaur - crocodylomorph relationship found by Gower (2002) could be disrupted by diluting the braincase data -- but without the overemphasis on tarsal (ankle) characters which marred studies such as Sereno 1990) and Parrish (1993) concerning which, see Dyke, 1998). Since the inclusion of the duplicate characters effectively weighted cranial characters, other than braincase characters, a little more heavily, the minor duplication promoted the objectives of the test.

Gracilisuchus and Euscolosuchus were not included in this study. They have simply been interpolated for purposes of the cladogram below, as have various other taxa, including Ticinosuchus and Rauisuchidae. Various dinosaur and related taxa from Benton (1999) were included. We are aware of at least one study in which the failure to include a reasonable number of representatives from a deep branch within the study taxon led to arguably spurious results. See, Damiani 2001), discussed at Trematosauroidea. In any event, the topology within Ornithodira not shown here) was conventional. For the same reason, a sphenosuchid was included in the study to help fill out the crocodylomorph branch.

Nodes have mostly been named in accordance with the definitions in Parrish 1993). So far as we know, no one has explicitly defined Suchia as Alligator > Postosuchus or equivalent. However, one common usage of the term is aetosaurs + crocs. This is close to what we need, so we have stretched a point in order to create a good node-stem triad: Rauisuchiformes = Rauisuchia + Suchia.

Despite these extensive efforts to break up the aetosaur-crocodylomorph relationship, only Erpetosuchus was more closely related to crocodylomorphs than aetosaurs. The degree of homoplasy between derived members of the Rauisuchia and Suchia is rather disturbing. Still, there are enough synapomorphies which look like true phylogenetic signal to suggest that this heterodox topology may be worth following, at least for the present. ATW031227.