Cladistic Analyses of Coelurosaurian Theropods

Cladistics was developed as a formal analysis only within the past three decades, so the relationships of extant birds to fossil archosaurs were not subjected to cladistic analysis until the mid-1980s (e.g., Thulborn, 1984; Gauthier, 1986). But the groundwork of comparisons was laid by Ostrom (1976), who cataloged the many similarities of Archaeop-teryx to theropod dinosaurs, dromaeosaurids in particular, and explicitly interpreted them in terms of synapomorphies (shared derived characters). The character evidence amassed by Ostrom was persuasive when he presented it anecdotally and became overwhelming when incorporated into cladistic analyses of birds with fossil archosaurs.

The study of Gauthier (1986) is a convenient starting point in discussing the relationships of birds to other dinosaurs because of the detailed discussion of the evidence presented therein. (The analysis by Thulborn [1984] reached similar results but found several theropods to be more closely related to extant birds than was Archaeop-teryx.) Gauthier's study surveyed 84 characters in 18 taxa, with the results shown in Figure 2.1A. Noteworthy was the finding that several theropods with unusually long arms are the closest relatives of birds, together constituting Coelurosauria. One of these, Ornithomimidae (corresponding generally to what is now named Ornithomimosauria), is more distantly related to birds than the others, and Gauthier erected the group Maniraptora for the latter. Beyond this there was little resolution indicating which groups were most closely related to Avialae, as is evident from the lack of resolution in the consensus cladogram (which shows only those groups supported in all the equally most parsimonious cladograms). In the text, Gauthier hypothesized that Dromaeosauridae and Troodontidae formed a group that is the closest relative of Avialae, but in an addendum he pointed out evidence contradicting the placement of the two families together, thus leaving the relationships of non-avialan Maniraptora entirely unresolved.

Several recent cladistic analyses addressed the relationships of birds among theropods, and another analyzed relationships among several nonavialan theropod taxa that are closely related to Avialae. We do not include analyses without published data matrices in this summary. In a study of a new relative of "segnosaurs" from China, Russell and Dong (1994a) presented an analysis of theropods with several novel results (Fig. 2.1B). Segnosaurs are a highly specialized group of dinosaurs known with confidence only from the Late Cretaceous of Asia, and hypotheses of their affinities have ranged throughout the Dinosauria with little consensus (Barsbold and Maryanska, 1990). The study presented by Russell and Dong summarized important evidence for the relationships of segnosaurs and the new taxon, the Early Cretaceous A. elesitaiensis, but unfortunately they apparently restricted the data set, derived mainly from Gauthier (1986), to only those characters that could be determined in the new taxon (59 characters in 11 taxa). Thus, characters shared by, for example, dromaeosaurids and birds are not included unless the anatomical region in which they are expressed is preserved in the Alxasaurus material. The exclusion of this evidence renders the results inconclusive.

Holtz (1994) presented an analysis revising the characters summarized by Gauthier and adding numerous others (a total of 126 characters in 19 taxa), especially many cranial characters. An important aspect of this analysis is that characters considered to be correlated with large size were explicitly excluded; in other words, several characters present only in large taxa were considered to represent a single character—large size—rather than different characters. The results (Fig. 2.1C) differ from those of Gauthier (1986) in at least three significant respects: (1) there is much greater resolution; (2) Ornithomimosauria (including Ornithomimidae) is not as distantly related to birds as Gauthier hypothesized, so that the groups Coelurosauria and Maniraptora are redefined (see Holtz, 1996); and (3) Tyrannosauridae is

Figure 2.1. Cladograms of relationships among coelurosaurian theropods based on published data matrices.A, strict consensus clado-gram of "numerous" equally parsimonious cladograms (84 characters for 18 taxa, CI = .85) presented by Gauthier (1986), who in an addendum to the paper questioned the close relationship of Troodontidae and Dromaeosauridae (treated as a single taxon, Deinony-chosauria, in the analysis). Carnosauria includes Tyrannosauridae and Allosauridae. The original data matrix includes Hulsanpes, a taxon known only from the metatarsals, but Wilkinson (1995) reported that if this taxon is included then relationships of Coelurosauria are unresolved in the strict consensus. B, single most parsimonious cladogram (59 characters for 11 taxa, CI = .551) of Russell and Dong (1993a), who included Microvenator within Oviraptorosauria. C, single most parsimonious cladogram (126 characters for 19 taxa, CI = .470, RI = .684) of Holtz (1994) excluding noncoelurosaurians, with group names (but not relationships) modified following Holtz (1996). D, single most parsimonious cladogram (47 characters for 8 taxa, CI = .597, RI = .587) of Sues (1997), who included Chirostenotes within Oviraptorosauria and synonymized Elmisauridae with Caenagnathidae within this group. E, strict consensus of three equally parsimonious cladograms (95 characters for 14 taxa, CI = .562, RI = .635) of Makovicky and Sues (1998); F, single most parsimonious hypothesized to be more closely related to birds than is Al-losauridae (proposed also in a cladistic analysis by Novas [1992] but without a data matrix), whereas Gauthier's analysis placed the two together, as the Carnosauria, outside Coelurosauria.

However, several problems in the data matrix detract from Holtz's analysis (Clark et al., 1994b; Sereno, 1997), and the abbreviated descriptions of characters (in which only the derived state is described) often obscure the precise feature being specified in each case. Furthermore, it is not clear that removing characters that are correlated is justified, as correlation among characters is expected for those sharing a common evolutionary origin. There may be grounds for dismissing a few of the characters Gauthier cited in support of Carnosauria as redundant in some obvious way, but because ignoring evidence through ad hoc hypotheses is anathema in science, this should not be taken lightly. This study is therefore best considered a "work in progress" that is being refined (e.g., Holtz, 1996, and Padian et al., 1999), although the only published data matrix is that of Holtz (1994).

An important study of an enigmatic theropod dinosaur from the Late Cretaceous of Canada included an analysis of relationships among Coelurosauria, although it did not address the relationships of birds (Fig. 2.1D). Chirostenotes pergracilis had been poorly known until a partial skeleton was discovered in museum collections. Sues (1997) describes this new material and recognizes its affinities with taxa forming Caenagnathidae (indeed, Sues synonymizes Caenagnathus with Chirostenotes, as suggested tentatively by Currie and Russell [1988]). Caenagnathidae forms a group (Oviraptorosauria) with the highly specialized Ovi-raptoridae. A cladistic analysis of 8 taxa and 47 characters finds tyrannosaurs to be outside Coelurosauria, contra Holtz (1994), and provides evidence that Oviraptorosauria are most closely related to Therizinosauroidea. Dromaeo-sauridae and Troodontidae are sister taxa to each other, and together to the Avialae.

A related study by Makovicky and Sues (1998), of an equally enigmatic theropod dinosaur from the Early Cretaceous of Montana, expanded the analysis of Sues (1997) with the addition of vertebral characters surveyed by Makovicky (1995). In their redescription of Microvenator celer Ostrom, 1970, Makovicky and Sues pointed out several features indicative of oviraptorosaurian affinities, including a bone tentatively identified as an edentulous dentary. Their analysis (Fig. 2.1E) found results similar to those of Sues (1997) and with Microvenator within Oviraptorosauria and Therizinosauroidea as the sister group to this group.

In the description of a new species of primitive avialan from the Late Cretaceous of Madagascar, Rahonavis ostromi (Forster et al., 1998a,b), an analysis of selected coelurosauri-ans was presented to explore its relationships (Fig. 2.1F). As in the analysis of Sues, Tyrannosauridae is farther removed from Avialae than indicated by the analysis of Holtz (1994). Noteworthy also is that Troodontidae, rather than Dro-maeosauridae, were found to be the closest relatives of Avialae. Rahonavis and Unenlagia were found to be most closely related to Archaeopteryx, although an alternative placement of Rahonavis within Avialae and Unenlagia outside it was only a single step longer.

One of the most important discoveries in recent years has been new species of nonavialan coelurosaurs with feathers. In their description of Protarchaeopteryx and Caudip-teryx, Ji et al. (1998) presented an analysis supporting the placement of both outside Avialae (Fig. 2.1F). The selection of nonavialan coelurosaurian taxa was limited, however.

Finally, an analysis of relationships among all dinosaurs was published by Sereno (1999). Notable elements of this phylogeny (Figure 2.1G) are the placement of Alvarezsauri-dae with Ornithomimidae (i.e., Ornithomimosauria of some other authors) and Therizinosauridae (contra Sereno, 1997), a clade comprising Caudipteryx and the Oviraptorosauria, and a monophyletic Deinonychosauria (Dromaeo-sauridae and Troodontidae).

In summary, the analyses of Coelurosauria published since the work of Gauthier (1986) have each resulted in more highly resolved relationships among the taxa, but with conflicting results. The problematic analyses of Russell and Dong (1993a) and Holtz (1994) aside, major conflicts are found in (1) the position of tyrannosaurs, either outside Coelurosauria or closer to birds than are ornithomi-mosaurs; (2) the relationships among Troodontidae and Dromaeosauridae relative to Avialae; (3) the relationships of therizinosauroids, either with oviraptorosaurs or or-nithomimosaurs; and (4) the relationships of Alvarezsauri-dae, either with (or within) Avialae or with Ornithomi-mosauria. Given these conflicts, attempts to arrive at a stable nomenclature for coelurosaurians by defining groups impervious to changes in composition (Sereno, 1998; Pa-dian et al., 1999) are more confusing than helpful. A notable shortcoming of all these analyses is that they considered cladogram (113 characters for 14 taxa, CI = .579, RI = .712) of Forster et al. (1998a); four taxa of Ornithothoraces included separately in original data matrix. G, single most parsimonious cladogram (90 characters for 8 taxa, CI = .855, RI = .849) of Ji et al. (1998); two genera of Mononykinae, Shuvuuia and Mononykus, included separately in original data matrix. H, strict consensus of six equally most parsimonious cladograms (204 characters for 17 taxa, CI = .67, RI = .81) of Sereno (1999); Allosauroidea includes Allosauridae and Sin-raptoridae, and four outgroup taxa are not illustrated.

either all higher taxa or a mixture of higher taxa and species, so that the character scores represent inferences of the primitive condition rather than observations.

0 0

Responses

  • ciaran
    Are birds related to coelurosaurian for maniraptorans?
    6 years ago

Post a comment