The systematic relationships and biogeographic history of ornithischian dinosaurs

Abstract

The systematic relationships of taxa traditionally referred to as 'basal ornithopods' or 'hypsilophodontids' remain poorly resolved since it was discovered that these taxa are not a monophyletic group, but rather a paraphyletic set of neornithischian taxa. Thus, even as the known diversity of these taxa has dramatically increased over the past two decades, our knowledge of their placement relative to each other and the major ornithischian subclades remained incomplete. This study employs the largest phylogenetic dataset yet compiled to assess basal ornithischian relationships (255 characters for 65 species level terminal taxa). The resulting strict consensus tree is the most well-resolved, stratigraphically consistent hypothesis of basal ornithischian relationships yet hypothesized. The only non-iguanodontian ornithopod (=basal ornithopod) recovered in this analysis is Hypsilophodon foxii. The majority of former 'hypsilophodontid' taxa are recovered within a single clade (Parksosauridae) that is situated as the sister-taxon to Cerapoda. The Parksosauridae is divided between two subclades, the Orodrominae and the Thescelosaurinae. This study does not recover a clade consisting of the Asian taxa Changchunsaurus, Haya, and Jeholosaurus (=Jeholosauridae). Rather, the former two taxa are recovered as basal members of Thescelosaurinae, while the latter taxon is recovered in a clade with Yueosaurus near the base of Neornithischia.The endemic South American clade Elasmaria is recovered within the Thescelosaurinae as the sister taxon to Thescelosaurus. This study supports the origination of Dinosauria and the early diversification of Ornithischia within Gondwana. Neornithischia first arose in Africa by the Early Jurassic before dispersing to Asia before the late Middle Jurassic, where much of the diversification among non-cerapodan neornithischians occurred. Under the simplest scenario the Parksosauridae originated in North America, with at least two later dispersals to Asia and one to South America. However, when ghost lineages are considered, an alternate dispersal hypothesis has thescelosaurines dispersing from Asia into South America (via North America) during the Early Cretaceous, then back into North America in the latest Cretaceous. The latter hypothesis may explain the dominance of orodromine taxa prior to the Maastrichtian in North America and the sudden appearance and wide distribution of thescelosaurines in North America beginning in the early Maastrichtian. While the diversity of parksosaurids has greatly increased over the last fifteen years, a ghost lineage of over 40 myr is present between the base of Parksosauridae and Cerapoda, indicating that much of the early history and diversity of this clade is yet to be discovered. This new phylogenetic hypothesis provides a comprehensive framework for testing further hypotheses regarding evolutionary patterns and processes within Ornithischia.

Keywords: Biogeography; Dinosauria; Neornithischia; Ornithischia; Ornithopoda; Phylogeny; Stratigraphic consistency; Systematics.

Figure 1. Recent phylogenetic hypotheses of basal ornithischian relationships.

Tree topology reported by Buchholz (2002) based on analysis of 97 characters for 20 terminal taxa (A), strict consensus of four most parsimonious trees recovered by Spencer (2007) based on analysis of 97 characters for 19 terminal taxa (B), and strict consensus of 756 most parsimonious trees recovered by Butler, Upchurch & Norman (2008) based on analysis of 221 characters for 46 terminal taxa (C). In (B) bootstrap values >50% are listed below nodes. In (C), Bremer support values >1 are to the left of nodes while bootstrap values >50% are to the right of nodes.

Figure 2. Strict consensus of the 36 most parsimonious trees recovered by this study.

Major ornithischian subclades are labeled either along branches (stem-based definitions) or at nodes (node-based definitions). See Table 1 for phylogenetic definitions. Numbers above nodes refer to the list of unambiguous character changes reported for each node in Table S4. Bold numbers beneath nodes are Bremer support numbers >1, while non-bold numbers beneath nodes are bootstrap support values >50%.

Figure 3. Parsimony-based reconstructions of ancestral geographic areas.

Tree topology based on Fig. 2. The pie charts at each node represent the level of support for each ancestral (see Table S5 for precise values). Each color represents a different geographic area (see key). Numbers next to nodes refer to those used in Table S5.

Figure 4. Likelihood-based reconstructions of ancestral geographic areas.

Results obtained when all branch lengths were equal (A) versus results obtained when time calibrated branch lengths were included and set equal to inferred missing fossil records (B). Tree topology based on Fig. 2. The pie charts at each node represent the level of support for each ancestral area (See Table S5 for values). Each color represents a different geographic area (see key). Numbers next to nodes refer to Table S5.

Figure 5. Time-calibrated phylogeny of Ornithischia.

White boxes indicate the uncertainty around the age of first appearance for each terminal taxon (not the known occurrences), while black lines represent implied missing fossil records (i.e., ghost lineages). Note: some branches are necessarily drawn deeper in time due to drawing constraints. Numbers positioned along branches or at nodes indicate the position of major ornithischian subclades. 1, Ornithischia; 2, Heterodontosauridae; 3, Genasauria; 4, Thyreophora; 5, Neornithischia; 6, Parksosauridae; 7, Orodrominae; 8, Thescelosaurinae; 9, Elasmaria; 10, Cerapoda; 11, Marginocephalia; 12, Ornithopoda; 13, Hypsilophodontidae; 14, Iguanodontia; 15, unnamed Gondwanan clade; 16, Dryomorpha; 17, Dryosauridae; 18, Ankylopollexia.