Torosaurus latus is not Triceratops sp.

Earlier this year, Scannella & Horner hypothesized that skulls attributed to the large chamosaurine Torosaurus latus represented a “fully adult” ontogenetic stage of Triceratops, thus invalidating Marsh’s pierced lizard as a distinct taxon. This theory was based in part on a growth series for Triceratops published by Horner & Goodwin in 2006. In neither paper do the authors consider Triceratops at the species level. Prior to 1996, as many as sixteen separate species of Triceratops had been considered valid, but in that year Catherine Forster published a revision of the genus which greatly reduced that number to just two: Triceratops horridus and Triceratops prorsus. Actually, in 1986, Ostrom & Wellnhofer went as far as suggesting that only Triceratops horridus is the only valid species. Neither Scannella & Horner nor Horner & Goodwin address the issue of species-level ontogenetic change in Triceratops. Is the Horner & Goodwin growth series for T. horridus or T. prorsus? Is Torosaurus latus the “fully adult” form of T. horridus, T. prorsus, or both? While not stated outright, the Horner & Goodwin growth series implies that features used by Forster to distinguish T. prorsus from T. horridus may simply be ontogenetic markers, including size and shape of epioccipitals, size and shape of the nasal horn, and curvature of the postorbital horns. Certain abstracts that I’ve read but don’t know if I can actually divulge have suggested stratigraphic or anagenetic reasons for the changes between Triceratops species.

In Scannella & Horner’s “Toroceratops” hypothesis, the authors suggest that, very late in life, Triceratops goes through radical morphological change: the length and shape of the frill increases, and the parietal becomes rapidly fenestrated as bone is reabsorbed on either side of the parietal bar. The authors point to a number of Triceratops skulls with extremely thin areas of the parietal where fenestrae would be expected to appear. However, none of the specimens investigated by the authors have even incipient parietal perforations (except Nedoceratops, but we’ll get into that in a minute). All investigated Triceratops skulls retain a solid frill. That is, the authors cannot point to any Triceratops skulls that would form a transition between the "no fenestrae" condition and the "big fenestrae" condition of Torosaurus. With the often-stated abundance of Triceratops material, I believe this is an important consideration.

According to the Horner & Goodwin growth series for Triceratops, specimen MOR 004 is regarded as an “adult” form. It has thick, forwardly-curved postorbital horns; elongate, highly reabsorbed epiparietals; a large, thick, forwardly-directed nasal horn; and a relatively short, deep rostrum below and anterior to the nasal horn. If MOR 004 is the model adult form of Triceratops, then no currently recognized specimen of Torosaurus latus conforms to the Horner & Goodwin growth series. All currently recognized specimens of Torosaurus latus (ANSP 15192, MOR 1122, YPM 1830, and YPM 1831) have short, upwardly-directed nasal horns with an elongate, shallow rostrum anterior to the nasal horn. Additionally, the snout has a distinctive “stepped-up” morphology wherein the dorsal margin of the rostrum anterior to the nasal horn is lower than the dorsal margin of the rostrum behind the nasal horn. This feature is similar to the “large juvenile” and “subadult” examples of Triceratops in Horner & Goodwin’s growth series. However, even those Triceratops specimens lack the elongate, shallow rostrum anterior to the nasal horn.

Diagonally from top to bottom, the proposed Horner & Goodwin growth series for Triceratops showing overall trends in morphology. To the lower left, Torosaurus latus.

The overall shape of the frill is quite distinct in both taxa. In MOR 004, the frill more or less frames the rest of the skull in anterior view: the parietal bar forms the apex of the frill, and the left and right halves of the frill slope down and out from there. This is especially obvious in two other specimens of Triceratops: YPM 1822 (Forster 1996) and YUM 1822 (Hatcher 1903). In lateral view, the frill attains a distinctive upward curve. Additionally, the squamosals of Triceratops are D-shaped, or perhaps axeblade-shaped. In many individuals, the squamosal-parietal suture is not visible for most of its posterior length. In contrast, Torosaurus’ frill is broad and largely flat, and does not frame the face in anterior view. Rather, it grows away from the rest of the skull and does not retain a distinct upward curve. Additionally, the squamosals in all recognized specimens of Torosaurus have a distinct shape reminiscent of a chef’s onion or paring knife, and the suture between the parietal and the squamosals is surprisingly clear along its entire length.



The largely complete frill of Torosaurus latus, after Hatcher 1903.

According to Scannella & Horner, the underlying geometry of the frill of MOR 004 would have been significantly altered—and quickly—to produce a frill attributable to Torosaurus latus. This sort of morphological change is wholly unknown in other ceratopsians where juveniles are known. In centrosaurines, for instance, the adult frill is, by and large, simply a larger version of the juvenile skull with the addition of unique spikes on the parietal margin or parietal bar (Pachyrhinosaurus lakusai and Centrosaurus apertus is particularly illustrative of this). In addition, the morphology of the snout and nasal horn would essentially reverse from the adult condition to a subadult form. These sorts of radical morphological changes have no basis in close relatives of “Toroceratops.”

Because all of the known specimens of Torosaurus latus share certain morphological features regardless of size or, presumably, age, we are forced to conclude that they are taxonomically valid characters that differentiate it from its closest relatives. For reference, its closest relatives have been consistently shown to be Triceratops and Nedoceratops. Whereas skulls attributed to Triceratops are far more numerous and individualized, those recognized as Torosaurus are much more uniform (ANSP 15192 does differ significantly in the length of the frill and size of the postorbital horns, but this may be an age-related character). While there is obviously a sampling bias at work here, it is instructive to list characters shared by the best-known specimens of Torosaurus latus—ANSP 15192, MOR 1122, YPM 1830, and YPM 1831:

1) Relatively small, upwardly-directed nasal horn that is triangular and develops about halfway up the snout. The horn never grows into the impressive forward-pointing thick horn you see in many Triceratops skulls.

2) An elongate, shallow rostrum anterior to the nasal horn. There is a clear height differentiation between this region and the dorsal margin of the skull behind the nasal horn in the adult stage. This may be termed a “stepped-up” condition.

3) An elongate nasal passage retained into the adult stage. In Torosaurus, the extent of the nasal system resembles the juvenile condition in Triceratops. In that taxon, however, the nasal passage compresses and rounds out as the subadult and adult stages are reached. Torosaurus, by contrast, retains an elongate nasal passage.

4) Blade-like squamosals which remain distinct from the parietal even in the adult stage.

5) An elongate frill that is “swept back” rather than “swept up,” is relatively flat and broad, and contains large parietal fenestrae. In adults, the margins of the frill may be almost completely smooth due to the absorption of the epioccipitals.

Before we go too much farther into Torosaurus, let’s turn our attention to Triceratops. In fact, let’s go back to Forster’s paper regarding species diversity in that genus. She references two individual specimens as models for the two species of Triceratops: YPM 1822 for T. prorsus and SDSM 2760 for T. horridus. I note that the latter displays several features in common with Torosaurus, including the structure of the snout and nasal horn (though not to the same extent), and possibly the structure of the squamosals. It also seems to lack distinct epioccipitals, and the frill is broader than its sister species, T. prorsus. That species, represented by YPM 1822, shows very different features: the nasal horn is large and directed forward, the rostrum in front of the nasal horn is short and deep, and the nasal area is rounded. The epoccipitals are fairly large and triangular.

Neither skull displays a singular suite of ontologic characters consistent with any one growth stage according to Horner & Goodwin. In SDSM 2760, the morphology of the snout and nasal horn are juvenile characters according to the suggested growth series, whereas the thickness and orientation of the brow horns, as well as the loss of epioccipitals, are indicative of adult status. By contrast, YPM 1822 shows a clearly adult snout and nasal horn, but the ends of the brow horns point a bit upwards (a subadult trait) and the frill retains distinct, fairly large epioccipitals (a juvenile or subadult trait). Are these isolated incidents? I’m afraid not. Just ask John Hatcher. In 1903, he included several beautiful illustrations of skulls attributed to Triceratops in his wonderful monograph on the horned dinosaurs. Exactly none of them conform to the Horner & Goodwin growth series to a "T." Going forward, where I reference "juvenile," "subadult," and "adult" features, I'm talking about those ontogenetic stages as defined in Horner & Goodwin 2006.




This is USNM 2100, identified by Hatcher as Triceratops prorsus(?). Although the anterior portion of the snout is missing, one may notice the underdeveloped nasal horn (juvenile-subadult), low-angled brow horns (adult) with sloped tips (subadult), and small but distinctive epioccipitals (subadult). The parietal bar is interesting in that it’s quite bumpy. Where would USNM 2100 fit on Horner & Goodwin’s growth series?




This is Triceratops brevicornus, ”YPM” 1834. It also shows a curious mix of characters: the nasal horn is distinct and forwardly-directed (adult), but the anterior portion of the snout is quite long (subadult). The brow horns are directed forward (adult) and are unusually short. The parietal-squamosal suture appears to be lost (adult), but the epoccipitals are reasonably distinct and not entirely rounded (subadult). Where would YPM 1834 fit on Horner & Goodwin’s growth series?




This is Triceratops elatus, USNM 1201. It also displays some curious features. The nasal horn’s growth seems to have stalled: the underlying nasal is indeed reaching forward (subadult), but the epinasal is still apparent (juvenile). The brow horns are large and directed forward (adult). The frill’s margins are bumpy—the epoccipitals have largely been absorbed (adult). Notice the distinct upward bend to the squamosals. While it gives the bones a superficially onion-knife appearance, the parietal curves distinctly upward rather than being directed back. Where would USNM 1201 fit on Horner & Goodwin’s growth curve?




Then there’s this famously odd duck: USNM 2412, Nedoceratops hatcheri. The nasal horn isn’t really a horn so much as a bump in the road, giving the snout a very pronounced “stepped-up” profile. The brow horns are quite large and directed almost straight upward. The frill is riddled with accessory fenestrae, some of which are probably the result of pathologic or natural re-absorption. The skull retains distinct epioccipitals, and the squamosal has a bizarre shape. Nedoceratops may be too much of a wildcard to include in this analysis, but Scannella & Horner consider it to be a transitional form that would exist between MOR 004 and Torosaurus latus. But what happened to the nasal and brow horns? Where did all these accessory fenestrae come from? Surely, Nedoceratops is either an incredibly abarrant individual or a distinct taxon, and will not be considered further here.




This is where things start getting interesting. This is Triceratops calicornis, USNM 4928, and it displays a lot of features that I listed for Torosaurus latus, above. In other words, its morphology conforms nicely to known morphologies that are consistent across currently-recognized specimens of Torosaurus. These include: a small, upwardly-directed nasal horn that is roughly halfway down the snout; a stepped-up snout profile with an elongate anterior portion; an elongate nasal passage; and squamosals that are onion-knife shaped. Most of the epoccipitals are completely re-absorbed. There is one distinct epoccipital capping the parietal-squamosal contact, and half of one above it. The parietal was not preserved in USNM 4928—it may have been fenestrated.



AMNH 5116 (top) compared to YPM 1830 (bottom). The overall similarities are striking.

Based on these proposed characters, I could make an argument that the most famous, well-known specimen of Triceratops in the world—AMNH 5116—is actually a Torosaurus. It has a short, upwardly-directed nasal horn, an elongate rostrum anterior to the nasal horn, an elongate nasal passage, onion-knife shaped squamosals, and a lack of epioccipitals. Like USNM 4928, it was also missing portions of its parietal, although major portions were apparently recovered and plastered back on. The beast was restored with a solid frill (maybe it had one), the assumption being that this is Triceratops. However, when you take the parietal out of the equation, AMNH 5116 is strikingly similar to Torosaurus skulls, particularly YPM 1830. Sorry about the lack of fenestrae in the illustration for YPM 1830—I just now noticed that they’re not there. *facepalm*
MNHN 1912.20, housed in Paris, is strikingly similar to USNM 4928, although its “Torosaurus” features are even more obvious: the snout has a more pronounced “step-up,” the nasal horn is small and retains the epinasal. The anterior portion of the snout is elongate, as is the nasal passage. The frill is elongate and broad. Epoccipitals are nearly absent, and the squamosals are onion-knife shaped. Like AMNH 5116, the frill of MNHN 1912.20 has been heavily restored, although exactly how much of the parietal was “touched up” is not specified in this skull’s description (Goussard, 2006). It may very well be that MNHN 1912.20 is a Torosaurus skull.



AMNH 5116 (top) compared to MNHN 1912.20. Again, note the many similarities.

Go back and look at the growth series picture at the top of the post. Diagonally from top to bottom, this is Horner & Goodwin’s proposal for ontogenetic change in Triceratops from small juvenile (MOR 1199) to adult (MOR 004). If Torosaurus latus were to follow MOR 004, the underlying geometry of the frill would have to change radically, the snout would significantly elongate, and the nasal horn would regress to an earlier developmental stage. Additionally, two giant holes would suddenly open up in the parietal! Scannella & Horner expect us to believe that all of these large-scale changes would happen in the dinosaur’s final years, and in stark contrast to the direction of growth that Triceratops had been experiencing up to that point. No other ceratopsian goes through this sort of late-stage transformation, and there is no reason to think that Triceratops is any different. Additionally, as pointed out by many readers, the rarity of “fully adult” individuals of Triceratops (Torosaurus) compared to the incredible abundance of earlier growth stages is a bizarre and unrealistic preservation bias that does not occur in other fossil animals. If anything, the opposite tends to be true: juvenile and subadults are rare while adults and “full adults” are more common. That is certainly the case in other ceratopsids, even among bonebeds. The bottom line is this: assuming the Horner & Goodwin growth series is generally accurate, YPM 1830 and MOR 1122 would NOT follow MOR 004. It really is that simple.

However, Torosaurus material might not be nearly as rare as everyone seems to think. I believe that the genus can be distinguished based on more than just the presence or absence of parietal fenestrae. It’s possible that specimens once referred to as Triceratops are, in fact, Torosaurus. If nothing else, I hope this post has convinced some of you out there in Readerland that the Scannella & Horner paper presents interesting ideas but serious flaws as well, and needs to be given a second look.

I offer an enormous, Pentaceratops-skull-sized thanks to Andrew Farke for taking the time to double-check my claims and correct others. Readers, notice that I don’t really talk about histeology or stratigraphy. This is largely because I am ill-informed to do so intelligently. I do think that Torosaurus is morphologically distinct from Triceratops based on skull anatomy alone, however, so those factors may not necessarily play a part. On the point of stratigraphy, however, I do wonder whether T. prorsus or T. horridus is older, and whether the similarities between the latter and Torosaurus may represent a close relationship. That is, perhaps T. horridus is closer to the common ancestor of Triceratops and Torosaurus, and that Triceratops prorsus is derived, anagenetically or otherwise, from T. horridus?