Patterns of divergence in the morphology of ceratopsian dinosaurs: sympatry is not a driver of ornament evolution

Establishing the origin and function of unusual traits in fossil taxa provides a crucial tool in understanding macroevolutionary patterns over long periods of time. Ceratopsian dinosaurs are known for their exaggerated and often elaborate horns and frills, which vary considerably between species. Many explanations have been proposed for the origin and evolution of these ‘ornamental’ traits, from predator defence to socio-sexual dominance signalling and, more recently, species recognition. A key prediction of the species recognition hypothesis is that two or more species possessing divergent ornamental traits should have been at least partially sympatric. For the first time to our knowledge, we test this hypothesis in ceratopsians by conducting a comparison of the morphological characters of 46 species. A total of 350 ceratopsian cladistic characters were categorized as either ‘internal’, ‘display’ (i.e. ornamental) or ‘non display’. Patterns of diversity of these characters were evaluated across 1035 unique species pairs. Display characters were found to diverge rapidly overall, but sympatric species were not found to differ significantly in their ornamental disparity from non-sympatric species, regardless of phylogenetic distance. The prediction of the species recognition hypothesis, and thus the idea that ornamentation evolved as a species recognition mechanism, has no statistical support among known ceratopsians.

: Time-scaled phylogeny of Ceratopsia Table S1: Temporal dates and location information for all ceratopsians used in study Table S2: Z-scores and associated p values for randomisation test

List of morphological characters used in difference calculation
Supplementary material references Figure S1: Time-scaled phylogeny for all known ceratopsian species, used to calculate relative phylogenetic distances. Estimated species ranges are indicated by thick bars at branch tips. Created using strap package for R (Bell and Lloyd, 2014). Characters range from 0 (all identical) to 1 (all different). In only 6 out of a total of 86 display characters (B) do sympatric species pairs show the highest average, and only two of these have a mean value of 1, indicating consistent difference. These two characters are represented by only one and three sympatric species pairs, and so are likely to be anomalous due to small sample size. The lack of any clear pattern in the display characters plot (B) does not suggest that individual characters are used as distinguishing traits in sympatric species. Internal characters (A) also show universal differences (mean value = 1.00) in three instances. As with display characters, these are associated with small numbers of species pairs sharing known states in each character (Characters 252, 262, and 296, representing two, two and one species pairs for each character respectively).

List of morphological characters used in difference calculation
General Skull 1. Skull, size relative to the body: (0) -small, shorter than the length of the upper hindlimb bones (1) -sub-equal to the length of the femora (2) -large, at least 150% of the length of the upper limb bones (Modified from Makovicky and Norell, 2006:1) 2.

Premaxilla-nasal dorsal contact, length in dorsal view:
(0) -short (1) -premaxilla and nasal contact for at least 30% of the length of the nasal (New Character) 42.
Nasal ornamentation, height compared to the orbit: (0) -short, less than or equal to the height of the orbit (1) -1.5 to 2 times taller than the orbit (2) -elongated, more than 3 times taller than the orbit (New Character) 67.
Squamosal, length relative to parietal on frill: (0) -squamosal as long as the parietal and forms part of the posterior margin of the frill (1) -squamosal shorter than parietal with part of the lateral portion of the frill made up of parietal (Modified from Sampson et al., 2010:60-61)

125.
Squamosal, extent of step "stepped margin": (0) -slight step where posterior expansion of step is less than the dorsoventral thickness of the pocket (1) -posteriorly expanded to greater than twice the dorsoventral thickness of the pocket (New Character) 126.

225.
Palatine, shape and relationship to maxilla: (0) -palatine contacts nearly the entire medial surface of the maxilla, restricting size of choanae, anterodorsal process embraces posterior end of vomer (1) -palatine contacts only the posterior one-third of medial surface of maxilla, contact with vomer lost, choanae enlarged (Osmólska, 1986)

227.
Vomer, relationship to maxillae on secondary palate: (0) -vomers insert between maxillae at the rear of the secondary palate (1) -vomers meet posterior margin of maxillae on secondary palate, do not insert between maxillae, so that the maxillae contact each other anterior to the choanae in palatal view and obscure the tip of the vomer (Modified from Makovicky & Norell, 2006:21) Quadrate 228.