Horns, Frills, Knobs and Domes – Morphological Ontogeny

Recently I’ve been embroiled in a discussion with various professional paleontologists, interested scientists, and lay paleophiles on the topic of the synonymy proposed by Scannella and Horner [1] for the Hell Creek chasmosaurine ceratopsians Triceratops and Torosaurus. I feel it necessary to place my thoughts here, as they are long and doing so in blog entries or on forums, etc., is getting tedious. I should have done this long ago, but attempts have failed because of the scattered nature of the individual parses I’d like to say about it. So, what I will do it summarize their hypotheses, then attempt a synthesis with my thoughts on the range of topics they argue about. It should be further notes that Horner and Goodwin [2] also recently proposed a similar synonymy of the Hell Creek pachycephalosaurids Pachycephalosaurus, Stygimoloch and Dracorex. I handled this latter paper in part in various places, most notably here and here, although I hope to endeavor better here

Scannella and Horner [1] synonymize Triceratops and Torosaurus, and draw on some extensive data in order to do so, primarily based on a series of skull material found over the last 100 years but especially through a project in which any and all skeletal material that could be recovered was. They studied various morphological and deep-tissue analyses, including the shape of the squamosal, the orientation of the supraorbital horns, the shape and size of the epioccipital nodes (episquamosals and epiparietals), presence of thinned or “sunken” regions of the squamosal and parietal as well as the appearance of fenestrae within them, and inferences of the ontogenetic series of chasmosaurine skulls from the Hell Creek [3].

This follows Horner and Goodwin [2], which argued that Hell Creek pachycephalosaurids formed an ontogenetic series and that, accordingly, it should be reduced to the oldest name available: Pachycephalosaurus wyomingensis, which would then contain Stygimoloch spinifer and Dracorex hogwartsi. The authors perform this in two steps, first synonymizing the latter taxon into the middle one, then the middle one into the former. This is based on inferences of the histology of the domes and of the peripheral exoccipital nodes (virtually only the episquamosals), relative proportions, and the inference of an ontogenetic series of a related taxon, Stegoceras validum (of which, see [4,5]).

Both of these analyses are drastic, and they evoke emotional responses in many. I cannot say that I was not somehow affected by reading about an effort to reduce taxonomy of several forms that had withstood decades of diagnosis, re-diagnoses and analysis affirming or contradicting, but generally confirming, a mult-taxic arrangement of these forms. Horner et al. (the collective term for the works of Jack Horner and colleagues) have done some smash up work on the relative osteohistology of both groups (a forté of Horner’s), but it seems they may have gone too far, especially in the sense of the inference of some of their data across taxa, or so some of those emotional response [not mine, however] would have us think. I disagree with Horner et al. on several points here, but as I alluded to in one of the links above, it is my purpose not to disagree with their conclusions (the synonymy) but on the process by which the authors arrived at this conclusion. In this disagreement, I will argue that the authors have made not one but several leaps in logic.

Ontogenetic Series Inference

Previous work of Horner and Goodwin [3], Sullivan [4], and Williamson and Carr [5] have recognized well-preserved ontogenetic sequences of either chasmosaurines or of pachycephalosaurids. These analyses make a basic assumption, which is common in the inference of an ontogenetic series, that stages of growth will look different from one another, but present with common trends through ontogeny. In many cases, however, influence such as peramorphosis may arise where, at an early point in development, derived features of a “mature” morphology will appear; but also paedomorphosis, its opposite, where “young” features will persist into later mature stages of life. These aspects can only be discriminated when one has a reasonable certainty that one is only dealing with a single taxon. It is difficult to address a series when one has cobbled it together from different taxa, some of which may alter ontogenetic trajectories (developing pera- or paedomorphic qualities at different points from one another) as well as preserving distinct portions of the body, with slight or minimal overlap of material. The stegoceratine series in [4,5] is based on the same essential portions of the skull, the frontoparietal, which is fused in early ontogeny in virtually all pachycephalosaurids, a feature that is considered diagnostic of the group [4,5]; even in the smallest pachycephalosaurid domes, there is a pronounced, “heart-shaped” doming structure, broadest caudally.

Note here that two issues influence an ontogenetic series’ self-contained authenticity: 1) the specimens must generally possess the necessary morphology diagnostic of the taxon to which they are being refered, and 2) that taxic diversity in the formation or locality from which the material is located is narrow enough to permit referral in the absence of or alongside issue 1. Issue 1 is prone to preservation complications, such that a diagnostic ridge or sulcus is simply not preserved, but also that some diagnostic features will not preserve at certain stages of growth; and the latter complication overlaps with a problem in issue 2, where some taxa overlap with others, are closelely related, and will share ontogenetic morphology at various stages in growth.

1. Pachycephalosaurian Ontogeny

When it comes to pachycephalosaurids, Stegoceras validum is a hefty contender for “over-fluffed” specimen referral, due in large part to the extensive sampling of the Dinosaur Park Formation (while it is also apparently found in the Oldman and Fruitland Formations [4,6]) as well as the presence of other small, stegoceratine-like pachycephalosaurids, such as ?Prenocephale brevis and Hanssuesia sternbergi [6][n1], in which portions of the dome are identical and are differentiated on small portions of the periphery or overall shape (which changes plastically in growth).

Asserting the growth series is intact, the authors above demonstrate a series from which to compare pachycephalosaurids in general, but as noted, it is fairly specific to a taxon, and no such series is known other than it; there are, in fact, very few skulls of Stegoceras validum more complete than a frontoparietal complex, and certainly very few with more than squamosals or other peripheral bones attached. What is present, scaling up to the only complete skull (UALV 2), shows an ontogeny of the dome from nearly flat (and likely flat in hatchlings), to a small posterior dome-like structure that is broader than long, and very low; expands rostrally as it increases in height; and eventually develops triangular or “heart-shaped” aspect in dorsal view; expands laterally to incorporate portions of the peripheral bones; then expands caudally to encompass the supratemporal  fenestrae until they are rendered slit-like (it is unknown if they are complete closed off, as in Prenocephale prenes). This series appears to be consistent among some stegoceratines, although others (such as Sphaerotholus [5] or Prenocephale [4,6]) are known from skulls possessing domes that expand to the posterior and lateral margins of the skull, and are steeply sloped, semispherical in aspect in all views, as in Pachycephalosaurus wyomingensis.

Hell Creek Pachycephalosauridae, after Horner and Goodwin, 2009. A-B, Pachycephalosaurus wyomnensis, AMNH 1696; C-D, the Sandy specimen; E-F, Stygimoloch spinifer, MPM 8111; G-H, Dracorex hogwartsi, holotype TCMI 2004.17.1.

The domes of Stygimoloch spinifer and an undescribed Hell Creek skeleton familiarly called Sandy are narrow, low, and long, and do not correspond to the ontogenetic series of other pachycephalosaurids; this differs from some skulls referred to Pachycephalosaurus wyomingensis, which are broad, long, and high, but also from some domes that have been named Pachycephalosaurus reinheimeri [7], which are very shallow, but still long and broad. However, Horner and Goodwin [2] argue that despite this, these skulls should form a series, and it should be of a single “genus” and species, Pachycephalosaurus wyomingensis.

Pachycephalosaurid variation in the Hell Creek can be relegated to several specific features: Dome presence, dome shape, dome extent; squamosal node shape and relative size; facial node extent. Speculating on the series, Horner and Goodwin [2] propose that the reduction in node size and the formation of a lumpy, squat node in larger skulls is due to their resorption, and this is backed up by histological analysis showing zones of remodelled tissue in progressively larger specimens. It is implied that juveniles should not have remodelled nodes, although the authors did not sample the smallest skull (TCMI 2004.28.1, holotype of Dracorex hogwartsi [8]) which limits their comparative data. Morphologically, however, the squamosal nodes in Stygimoloch spinifer [9] and Dracorex hogwartsi [8] are circular in cross section, and closely appressed at the bases, and in Pachycephalosaurus wyomingensis the nodes are spaced from one another as well as circular in section; the Sandy specimen, contrarily, has pyramidal nodes that are appressed at the base. Pachycephalosaurus wyomingensis, unlike the other specimens, lacks extensive facial, jugal or postorbital nodes as seen in the Sandy specimen or Dracorex hogwartsi. And finally, the dome shape in Stygimoloch spinifer is larger than the comparatively sized Sandy specimen, very narrow and high; it is lower and rostrocaudally shorter than in Pachycephalosaurus wyomingensis and severely narrower.

There is no consistent ontogenetic sequence among pachycephalosaurids if this sampling is consistent with a single taxon, which when parsimony is applied provides an either/or option among the taxa. If one of these ontogenetic series is confirmed, then the other should be doubted as being consistent, but which one? Parsimony does not provide an answer, and this makes the premise of the Hell Creek series as a constrained ontogenetic series less viable from a logical standpoint. While the series from an histological perspective is constrained due to a developmental regime, the authors do not regard the option of symplesiopmorphy in development (shared basal features of development among multiple taxa) as parsimonious due to the recognition of other taxa. If, based solely on this concept, this were the case, I would agree; however, there are morphological disparities in the series they propose, including its lack of similarity to the Stegoceras validum series as well as the morphology of the Sandy specimen and Pachycephalosaurus reinheimeri (with special attention to the nodal morphology in both taxa), that suggest taxic diversity in pachycephalosaurids in the Hell Creek.

2a. Ceratopsid Ontogeny

When it comes to ceratopsids, however, there is an abundance of growth series, primarily from centrosaurines. The most recent is that of Pachyrhinosaurus lakustai [10], including adults, subadults, and juveniles, ranging from as large or larger than 180 cm (reconstructed; rostral to midline of parietal margin) to 25 cm, a magnitude of 12:1. Centrosaurine ontogeny tends to follow some concise processes, including a “scalloped” frill margin that differentiates into a smoother margin or with large, elongated or widened epioccipitals and spikes; a generally unelaborate nasal that expands into variously a short or long nasal horn, a low angular ridge, or a thickened, rough “pad.” Supraorbital ornamentation in centrosaurines is limited to low spikes, slightly curving horns, shallow bosses, or even small pits in those bosses, which are evidence for resorption. Frill ornamentation varies drastically, much as pachyrhinosaur nasal ornamentation does, but even the largest specimens do not differ much in frill morphology from younger subadults save in the relative size and position of the various epioccipital spikes.

Even at the smallest size, the frill is simply scalloped, and individual spikes appear to expand and differentiate into spikes, while others may be lost. At the largest sizes, immediate growth appears to occur only in the nasal bosses and the p3 epiparietal (the long laterally projecting and slightly twisted spike); extreme changes do not appear to substantially alter the morphology of the parietal fenestrae, which are present in parietals half the size of the largest parietal fragments (comparing width between p2 positions).

An ontogenetic series proposed by Horner et al. [3,10] for Triceratops (no species given) affords juveniles without nasal horns (which form from secondary ossifications in chasmosaurines, unlike paired nasal prominences in centrosaurines), small posteriorly curving supraorbital horns, and a solid, semicircular fan-shaped frill with scalloped margins and no parietal fenestrae [3]. Ontogenetic progression develops the nasal horn, supraorbital horns (which vary in orientation and length, increasing until adulthood, then reducing in size in some individuals [3]), the shape and length of the squamosal (from a short, fan-shaped element with an L-shaped “step” in its proximal parietal articulation to a long, un-“stepped” and sword-shaped element), an increase in epioccipitals with their shape changing from long and trinagular to broad and nearly flat, to fused and indistinguisbale along the frill margin, a parietal with a posterior end nearly continuous with the end of the squamosals to much longer, and with open parietal fenestrae (which are absent in all but the largest specimens). This last feature was added to the series in [1], due to the inclusion of specimens typically considered Torosaurus latus.

Problematically, of five primary ontogenetic stages (baby, juvenile, subadult, “young” adult, “mature” adult), the last is represented almost solely by specimens that have features that would indicate they’d be referred to Torosaurus latus, and admitted by Scannella and Horner [1]; this includes the holotype of Torosaurus latus (YPM 1830). Of the ontogenetic features listed above, two (elongation of the squamosals into very slender sword-shaped elements and open fenestrae) are noted almost exclusively at a particular size range (midline parietal length at or greater than 100 cm, with one specimen based on features at 85 cm, but still larger than the largest specimen without these features, at 83 cm). This type of size disparity with feature disparity makes it difficult to show a progressive ontogenetic series, when absolute size differences correspond to absolute morphological differences; typically, overlapping sizes and morphology help secure ontogenetic change, and as such, the series presents a “gap.” Moreover, additional analysis of parietal and squamosal fenestrae, in addressing the issue of pathology versus morphology, has considered fenestrae to occur in mechanically unloaded portions of thinning bone, and are present in various frills as an effect of their size and morphology [11]; they may be, quite certainly, likely in a diversity of taxa without any phylogenetic or even ontogenetic significance (in solid-frilled taxa).

One specimen (AMNH 5116) is regarded as a “young adult,” has a distinct textural surface in a sunken region where fenestrae “could” be but otherwise lacks fenestrae, and has a parietal midline length of 83 cm (see above). The physical description implies this specimen belongs to Triceratops (again, no species given), but is on display as Triceratops horridus; unlike other Triceratops of a comparable size, however, the squamosal is narrower and shorter than larger Triceratops specimens (for example, MOR 2702); it also lacks a distinct “step” in the squamosal, while the horn-core orientation implies an adult morphology. This may represent one of two things:

1) AMNH 5116 corresponds to the “gap” between “mature adult” (Torosaurus latus) and “young adult” or younger (any species of Triceratops), or
2) it repesents a younger specimen of Torosaurus latus, which would be distinct from Triceratops.

“Triceratops horridus,” AMNH 5116.

A second specimen, ANSP 15192, like AMNH 5116, preserves a squamosal morphology and apparent lack of parietal fenestrae (unconfirmed by [1]) similar to Torosaurus latus, but is as small as AMNH 5116 and preserves otherwise “adult” horn morphology and is classified along with the “mature” adults (unlike AMNH 5116). These specimens are suggestively “variable” from the norm, and in [1:fig.4] are shown to scale with subadult/”young” adult morphology, while are morphologically “mature,” including the presence of extremely derived, low, spindle-shaped episquamosals that have expanded and nearly touch one another along the frill margin.

A potential solution to this argument is that these specimens represent a morph of Triceratops (some species therein), but not a species, and that morphological variation is a result of individual variation, or sexual dimorphism. This is unconfirmed, and the authors did not test dimorphism in the current study. Additionally, one can affirm these specimens represent younger specimens of Torosaurus latus. In response to this proposition, the authors aver that size alone is not representative of age, and base their example on a comparison of YPM 1821 and 1822, the latter of which is larger than the former, and bears open cranial sutures, while the former has these closed. Closure of sutures is commonly accepted as a proxy of relative age, and while it varies within a skeleton based on age (e.g., vertebrae are not always consistently fused along the series), such dismissal warrants an explanation. None is given. The authors also state [1:pg.1166]:

It is telling that no confirmed juvenile ‘Torosaurus’ skulls have been reported.

This is an example of a logical fallacy, specifically a generalization of the particular, or the fallacy of the insufficient sample. Relatedly, the argument from ignorance would phrase this as “Because no Torosaurus juveniles have been found, no Torosaurus juveniles exist.” Carl Sagan, brilliant man that he was, famously responded to such a fallacious argument (although not this specific one) by stating:

Absence of evidence is not evidence of absence.

Even more importantly, an additional fallacy is common to collection: confirmation bias, which argues that a priori assumptions of taxic identification will favor further sampling, and it is common to find in biostratigraphic studies, where workers will assign specimens to a given taxon because it is assumed that taxon is the only one of its group available in a given level. This issue ignores morphological details for the general. Another example of a biased taxon would be Troodon formosus [n2], which is excessively lumped and more than likely to be split on the basis of various cranial and dental parameters (personal observation).

[1] Scannella, J. B. & Horner, J. R. 2010. Torosaurus Marsh, 1891, is Triceratops Marsh, 1889 (Ceratopsidae: Chasmosaurinae): Synonymy and ontogeny. Journal of Vertebrate Paleontology 30(4):1157-1168.
[2] Horner, J. R., & Goodwin, M. B. 2009. Extreme cranial ontogeny in the Upper Cretaceous dinosaur Pachycephalosaurus. PLoS ONE 4:e7626.
[3] Horner, J. R., & Goodwin, M. B. 2006. Major cranial changes during Triceratops ontogeny. Proceedings of the Royal Society of London, B 273:2757-2761.
[4] Sullivan, R. M. 2003. Revision of the dinosaur Stegoceras Lambe (Ornithischia, Pachycephalosauridae). Journal of Vertebrate Paleontology 23(1)181-207.
[5] Williamson, T. E. & Carr, T. D. 2003. A new genus of derived pachycephalosaurian from western North America. Journal of Vertebrate Paleontology 22(4):779-801.
[6] Sullivan, R. M. 2006. A taxonomic review of the Pachycephalosauridae (Dinosauria; Ornithischia). In Lucas & Sullivan (eds.) Late Cretaceous Vertebrates from the Western Interior. New Mexico Museum of Natural History and Science Bulletin 35:347-365.
[7] Brown, B. & Schlaikjer, E. M. 1943. A study of the troödont dinosaurs with the description of a new genus and four new species. Bulletin of the American Museum of Natural History 82(5):115-150.
[8] Bakker, R. T., Sullivan, R. M., Porter, V., Larson, P. & Saulsbury, S. J. 2006. Dracorex hogwartsia, n. gen., n. sp., a spiked, flat-headed pachycephalosaurid dinosaur from the Upper Cretaceous Hell Creek Formation of South Dakota. In Lucas & Sullivan (eds.) Late Cretaceous vertebrates from the Western Interior. New Mexico Museum of Natural History and Science Bulletin 35:331-345.
[9] Giffin, E. B. 1989. Notes on pachycephalosaurs (Ornithischia). Journal of Paleontology 63:525-529.
[10] Currie, P. J., Langston, W. & Tanke, D. H. 2008. A new species of Pachyrhinosaurus (Dinosauria, Ceratopsidae) from the Upper Cretaceous of Alberta, Canada. p1-108 in Currie, Langston & Tanke (eds.) A New Horned Dinosaur from an Upper Cretaceous Bone Bed in Alberta. (NRC Research Press, Ottawa.)
[11] Tanke, D. H. & Farke, A. A.. 2007. Bone resorption, bone lesions and extra cranial fenestrae in ceratopsid dinosaurs: a preliminary assessment. p.319-347 in Carpenter (ed.), Horns and Beaks: Ceratopsian and Ornithopod Dinosaurs. (Indiana University Press, Bloomington and Indianapolis)

[n1] Stegoceras is a “wastebasket” taxon of old, originally contamining a large number of species that have now been virtually completely split off. Some of these are considered valid, but are rendered to different “genera:” Stegoceras brevis, Stegoceras edmontonense, Stegoceras sternbergi and Stegoceras lambei. Stegoceras brevis and edmontonense have both been referred to Prenocephale (the latter as Prenocephale edmontonensis [n3]), while sternbergi was placed in Hanssuesia and lambei into Colepiocephale as type species of new genera [7].
[n2] Troodon formosus, based on a single tooth from the Judith River Formation of Montana, but which has been “located” (in decreasing age) in the Two Medicine Formation of Montana, the Oldman and Dinosaur Park Formations of Alberta, the Horseshoe Canyon Formation of Alberta and Prince Creek Formation of Alaska, the Lance and Hell Creek Formations of Montana, and have been suggested or explicitly identified from the Javelina Formation of Texas and Kirtland Formation of New Mexico.
[n3] Subsequent to the 4th edition of the ICZN, the Code does not permit respelling of a species name when it is considered a noun in apposition that disagrees in gender with the genus name (Art. 34.2.1). In this case, if I am correct in arguing this, Stegoceras (fem) and edmontonense (fem) agree in gender and operate as appositive nouns, but when edmontonense is combined with Prenocephale (masc), it was changed to edmontonensis (masc). As this was considered mandatory prior to enactment of the 4th edition of the Code, it is easy to see how this mistake may have occurred. Thus Sullivan’s use of Prenocephale edmontonensis may be considered an unnecessary emendation.

This entry was posted in Paleontology, Science Reporting, Taxonomy and tagged , , , , . Bookmark the permalink.

One Response to Horns, Frills, Knobs and Domes – Morphological Ontogeny

  1. Pingback: Nano Keeps on Slipping, Slipping… | The Bite Stuff

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s