Progress and future directions in archosaur phylogenetics

Journal of Paleontology, Nov 2001 by Brochu, Christopher A

De Braga and Rieppel (1997) and Rieppel and Reisz (1999) included sauropterygians in a phylogenetic analysis of reptiles. They came to the surprising conclusion that sauropterygians were very basal relatives of the lepidosaurs. Sauropterygians were thus derived diapsids, and the euryapsid condition with a single temporal fenestra was derived from the diapsid two-fenestra pattern. An analysis by Merck (1997) went a step further by including ichthyosaurs as well as sauropterygians. His analysis supported euryapsid monophyly and the diapsid affinities of marine reptiles, but Euryapsida was a basal archosauromorph lineage in his analysis, as tentatively shown in Figure 1. Whether inclusion of ichthyosaurs changes the results of Rieppel's analysis depends on the version of the matrix used (Rieppel and Reisz, 1999).

Even more surprising, inclusion of euryapsids drew turtles within Diapsida in both sets of studies. Turtles reflect the ancestral "anapsid" condition and lack temporal fenestration. For this reason, turtles have historically been regarded as "primitive" reptiles lying outside Diapsida, and all previous morphology-based cladistic analyses agreed that turtles were the living sister taxon of Diapsida (Gauthier et al., 1988; Laurin and Reisz, 1996; Lee, 1995, 1997a, 1997b). Turtles were seen as close relatives of sauropterygians (or even members of Sauropterygia) in the analyses including euryapsids; without euryapsids, those matrices support nondiapsid affinities for turtles.

This was not an entirely "new" conclusion, as anatomists of the late 19th and early 20th centuries sometimes made similar suggestions, but the characters used to draw this conclusion were later considered convergent (Gregory, 1946). It remains controversial within the phylogenetics community, and authors of papers supporting this conclusion acknowledge its heterodoxy (Rieppel and Reisz, 1999). The codings used in these analyses have been contested (e.g., Lee, 1997a), and a sauropterygian-chelonian relationship is weakly supported and based on characters that are homoplastic on the tree as a whole (Rieppel and Reisz, 1999). From a morphological perspective, this is a hypothesis clearly in need of further exploration.

Adding to the controversy, several independent molecular analyses also regard turtles as diapsids (Zardoya and Meyer, 2001). Mitochondrial sequence data argue for a clade that includes turtles, crocodylians, and birds to the exclusion of lepidosaurs, regardless of method (Platz and Conlon, 1997; Zardoya and Meyer, 1998; Janke et al., 2001). Nuclear genes have been less precise (e.g., Gorr et al., 1998), and some actually support a crocodylianturtle clade to the exclusion of birds (Hedges and Poling, 1999; Mannen and Li, 1999), making turtles members of Archosauria in the restricted sense.

Although the molecular evidence for diapsid (and possibly archosaurian) affinities of turtles is compelling, it is not problemfree. Most importantly, we are working with ingroup divergences minimally older than the Jurassic, and available outgroups-- mammals and lissamphibians-had diverged from the ingroup by the early Carboniferous. Furthermore, some of these genes show strong rate heterogeneity, which may be influencing the ingroup signal (Gorr et al., 1998). But not all genes show evidence for evolutionary rate heterogeneity, and none of these trees can be rerooted to make a "standard" tree with turtles outside Diapsida.