Hadrosaurs. David A. Eberth
Note: Asterisks indicate taxa whose tooth morphology was estimated using a composite shape from two teeth.
Morphometric Analysis
Many of the discrete characters used in phylogenetic analysis of iguanodontians are crude approximations of shape, so to examine the phylogenetic significance of precise, quantified tooth shape in determining the phylogenetic affinities of “T. cantabrigensis,” we quantified tooth morphology in select iguanodontian taxa (Table 6.1). Unfortunately, the tooth of ‘Iguanodon’ hillii is too incomplete to include in this analysis.
Methods We translated the overall shape of the enamel surface of dentary teeth into 2-D coordinate systems and performed eigenshape analysis, using the resultant shape variables to determine the similarity of tooth shape in “T. cantabrigensis” to other taxa. Eigenshape analysis determines covariance between shape outlines by comparing the net angular deviation (phi function) of coordinate points describing those outlines, between shapes (Lohmann, 1983; MacLeod, 1999). Singular value decomposition of the covariance matrix of phi functions for multiple shapes provides the principal components, eigenvalues, and shape variables used for comparisons (MacLeod, 1999; Krieger, 2010). The method requires a single homologous landmark starting point, and we used the anteroventral margin of the enamel surface as the landmark coordinate after orienting all specimen images to correspond to the left side of the dentary. All images were digitized using tpsDIG2 (Rohlf, 2007) and outlines were resampled to 100 coordinate points. For denticulate enamel margins, outlines points were digitized at the apex of individual denticles. Several taxa did not preserve a complete enamel surface for any individual tooth, and shape was estimated by creating a composite tooth from adjacent elements (Table 6.1).
Table 6.2. Eigenvalues and % Variance Explained for Jolliffe Cut-off of 75%
Eigenshape analysis was performed using PAST v.2.14 (Hammer et al., 2001), and the significant principle components were determined using the Jolliffe cut-off, which retains those components whose eigenvalues are greater than the average of all eigenvalues (Jolliffe, 1986; see also Mutsvangwa and Douglas, 2007). Component scores for all taxa were used in a cluster analysis to create a phenetic tree based on tooth shape for comparison with phylogenetic trees based on larger datasets (e.g., McDonald, Barrett, and Chapman, 2010; McDonald, Kirkland, et al., 2010). We created a pairwise distance matrix by reducing the coordinate distances for the first four components to single dimensional Euclidean distances, and performed a UPGMA cluster analysis (Fig. 6.3C).
Results Analysis resulted in four significant components with eigenvalues accounting for 78.8% for all shape variance within the sample (Table 6.2). Shape space defined by the first three component scores is shown in Figure 6.3. For all significant components, “T. cantabrigensis” is excluded from the space defined by the sample of examined hadrosaurid taxa (Fig. 6.3A, B). Instead, the taxon occupies a position close to Equijubus, Protohadros, and the British Iguanodon-like taxa, despite possessing discrete apomorphies of more derived taxa. Among other taxa, Telmatosaurus occupies an extreme position in coordinate space, reflecting its apomorphically narrow, elongate, and recurved dentary crown surface. Hadrosaurids occupy a narrow range of coordinate space. Although this range represents only a small sample of hadrosaurid species-richness, it includes representatives of most of the primary divisions within the clade and we do not expect that increased taxonomic sampling would result in a greatly expanded shape space.
Comparison of the UPGMA cluster analysis topology with phylogenetic hypotheses of iguanodontians reveals some similarities in the proximal grouping of terminal taxa, but tooth morphometrics shows an overall poor fit to phylogenetic hypotheses. Hadrosaurids cluster together with the exception of Charonosaurus, and hadrosaurines cluster within Hadrosauridae. Levnesovia and Bactrosaurus are closely clustered, which reflects their overall anatomical similarity (e.g., Godefroit et al., 1998; Sues and Averianov, 2009). Other aspects of the cluster topology are not consistent with published phylogenies. Mantellisaurus clusters with Probactrosaurus, Bactrosaurus, and Levnesovia, despite cranial and postcranial data that suggest it is a basal iguanodontian relative to those taxa (McDonald, Barrett, and Chapman, 2010; McDonald, Kirkland, et al., 2010). The highly apomorphic dentition of Telmatosaurus results in it clustering with the group consisting of all other taxa, as opposed to it being recognized as the sister taxon to Hadrosauridae (Weishampel et al., 1993). Eolambia clusters with the majority of hadrosaurids despite being recognized as a more basal iguanodontian (Head, 2001).
“Trachodon cantabrigiensis” clusters with Protohadros, Equijubus, and Owenodon, taxa that constitute a paraphyletic assemblage within Iguanodontia (McDonald, Barrett, Chapman, 2010; McDonald, Kirkland, et al., 2010). This grouping does not strongly support a precise hypothesis of relationships for “T. cantabrigiensis,” but, in combination with the exclusion of the taxon from Hadrosauridae based on morphometry, suggests a position among basal hadrosauroids, consistent with the distribution of discrete characters. The conflict between morphometric tooth data and discrete dental characters, and between dental and osteological characters, may in part be due to dental ecomorphology. Dietary specification is correlated to tooth morphology in herbivorous mammals, but the potential for diet to influence tooth size and shape in ornithopods has yet to be examined.
Validity and Affinities of English “Hadrosaurids”
“Trachodon cantabrigiensis” possesses a unique character combination among basal hadrosauroids and hadrosaurids: the presence of a distally offset carina, the absence of secondary ridges,