Tyrannouroboros: how everything old is new again in recent proposals of Tyrannosaurus taxonomy

Tyrannosaurus engages in some closed-mouth vocalising, but which species of Tyrannosaurus is this? It might seem that, up until recently, there would be only one answer, but this is actually an old question dating back to at least the early 1970s. Hey, writing about this sounds like a fun idea for a blog post. Oh look, someone else thought so too - read on...

Even aliens living on the far side of the Moon are aware that, last month, independent researcher and palaeoartist Gregory S. Paul led a team of authors proposing that our traditional take on Tyrannosaurus rex was wrong. According to Paul et al. (2022), this classic genus does not contain just one species, but actually three: the stocky, geologically older T. imperator, and two descendant, coexisting younger species, T. rex and T. regina. As documented in the New York Times and elsewherein numerousarticles, the response from tyrant dinosaur experts was not enthusiastic; to the contrary, most workers concluded that the study was underwhelming and unlikely to change the status quo. It’s not my intention to provide a detailed breakdown on this paper as relevant experts have already done this in recent weeks on various news and social media sites. It'll suffice to say that, while some of the ideas floated by Paul et al. are certainly interesting, the vague, very short diagnoses of the three species, the failure to provide strong statistical support for “robust” and “gracile” morphotypes within our T. rex inventory, and the inability to assign several well-preserved partial skeletons to the new species will likely prove sufficiently problematic to prevent widespread adoption of this new scheme. I hear a response is already being written.

But one, perhaps surprising, reaction to Paul and colleagues’ (2022) research is that most experts really weren’t shocked by the idea of several species within the T. rex hypodigm, and several even remarked that the idea itself is plausible even if our data, for the time being, do not support it. While the exact geological longevity of Tyrannosaurus rex is open to question (Paul et al. 2022), it was at least around for close to one million years (Brusatte and Carr 2016) and existed in varying habitats across a wide spread of western North America, from southern Alberta to the Texas/Mexico border, and from Utah to the Dakotas (Sampson and Loewen 2005). Compared to other theropods, that’s a broad spatiotemporal spread and one that could conceivably contain hitherto unappreciated variation that requires close analysis to uncover.

And it’s this topic of looking closely at our T. rex sample that I want to discuss today. It might seem that Paul and colleagues’ proposal is a radical dismantling of a fossil icon, and, indeed, this status is emphasised by Paul et al. themselves, who remark in their supplementary data and press work that they are “likely to displease many who are enamoured with the tyrant lizard king Latin moniker”, but that their "mini-revolution" will “force people to face the issue”. But, viewed more broadly, their study is actually only the latest in a long line of research that has expanded or shrunk our inventory of T. rex fossils and toyed with the idea of multiple Tyrannosaurus species. Indeed, even the features said to characterise the T. regina, rex and imperator are familiar from older publications (e.g. Larson 2008a). The result is that, when reviewing older literature on Tyrannosaurus taxonomy in light of these recent discussions, it’s difficult to avoid a sense that these seemingly-new ideas are not only re-treading old ground, but maybe even tripping over it.

Despite the long-held celebrity of Tyrannosaurus and it being known to science for over a century, most of what we’ve learned about this animal has been discovered in the last 30 years. In fact, little of note was published on T. rex during the mid-20th century and only a few significant specimens were uncovered (Larson 2008b). This corresponds with little taxonomic work and means we can skip ahead to start this story around 1970. Tyrannosaurus fans will realise that this omits discussion of the historic taxa “Dynamosaurus” and “Manospondylus” but, given that they were satisfactorily dismissed by Osborn (1906, 1917) and have no chance of being revived today, we can gloss over the earliest history of T. rex studies with only one important note: rushed as it was, Henry Osborn’s (1905) naming and type designation for Tyrannosaurus rex were extremely solid. So remember that, in spite of everything that follows in this article, and whatever else future generations have to say about the number of Tyrannosaurus species, T. rex itself is not going anywhere.

Rushing through that earliest period of Tyrannosaurus research puts us in the mid-20th century when, following a world war-induced hiatus in North American palaeontology, collectors and researchers returned to the field during the 1960s and ‘70s to resume exploration of American fossil resources. Among many important discoveries were new specimens of Tyrannosaurus and, at this point in history, each one was significant because T. rex was still only represented by a few examples. As our sample size grew from a few to a handful, and then to a dozen and more, it became apparent that no two T. rex individuals were exactly alike: T. rex evidently held a large amount of intraspecific variation within its population (Carpenter 1990). With this observation came the question of how much difference could be permitted in these fossils before specimens could no longer be identified as T. rex. The first rumblings on this never made it to print but were included in an unpublished 1972 thesis by then-student Douglas “Quetzalcoatlus” Lawson, who named Tyrannosaurus“vannus” for a relatively small, seemingly unusual tyrannosaurid maxilla from the Texan Javelina Formation. T. “vannus” never became “official” in the eyes of zoological nomenclature because it was never formally published and, in any case, Lawson revised his interpretation soon after, referring the same bone to T. rex itself (Lawson 1976). But in a later, 1990 review of T. rex variation, Kenneth Carpenter independently opined that this maxilla might represent a distinct southern Tyrannosaurus species. Others (Brochu 2003; Carr 2020) have been less convinced that it represents a new species, with Brochu (2003) suggesting that if the Texan maxilla was not T. rex, it must still represent a very close relative.

This was only the start of the buzz circulating around the concept of Tyrannosaurus having multiple species, which gained further traction in the 1980s. Again, the amount of variation in T. rex specimens was generating conversation, and both Horner and Lessom (1993) and Larson (2008) give Robert Bakker credit for observing features in Tyrannosaurus skeletons that might distinguish certain morphotypes from others. It’s reported that Bakker was so confident in this that only a slightly increased T. rex sample size might have seen him formally propose more Tyrannosaurus species before the turn of the millennium. It was perhaps these rumours and conversations that prompted Paul (1988) to muse on multiple Tyrannosaurus species in his influential book Predatory Dinosaurs of the World. Despite acknowledging variation in T. rex dentition and limb structure, Paul concluded that no, the Tyrannosaurus specimens known in the late 1980s were not diverse enough to represent more than one species. But other researchers weren’t so sure. Molnar (1991) responded to Paul’s comments by questioning whether we had enough T. rex specimens to draw any firm conclusions about the number of species they really represented, and suggested that the question would remain open until we acquired more data. Others were more confident that Paul was wrong. Donald Glut (1997) recorded that researcher Stephen Pickering privately issued manuscripts in 1996 proposing that several T. rex specimens, including the famous “Sue” skeleton, represented a new species that he named Tyrannosaurus "stanwinstonorous" after the special effects pioneer responsible for the animatronic dinosaurs in Jurassic Park and its sequels. Pickering argued that multiple features of the skull and the trait of adults being 6-7 % larger than Tarbosaurus or T. rex distinguished "stanwinstonorous" from other tyrannosaurs, but — as with Lawson’s T. “vannus” — lack of widespread publication prevented this name from entering scientific consideration.

So, even as palaeontologists circled Tyrannosaurus looking for specimens to split from the T. rex hypodigm in the late 20th century, no formal proposals were published. But taxonomic acts relevant to T. rex were being carried out as smaller, often fragmentary North American tyrannosaurs that we’d eventually recognise as Tyrannosaurus were receiving names or reclassification. With some of these specimens having been found decades prior, fairly convoluted taxonomic histories are associated with them (see Carr and Williamson 2004 for a summary). Here, let’s just cut to the chase and list them with their final, accepted binomials: Dinotyrannus megagracilis (Paul 1988; Olshevsky and Ford 1995), Stygivenator molnari (Paul 1988; Olshevsky and Ford 1995) and — most famous of all — Nanotyrannus lancensis (Gilmore 1946; Bakker et al. 1988). Despite recognition that at least some of these species (D. megagracilis and S. molnari) were based on juvenile or subadult material, the idea that they may represent subadults of T. rex was not considered, and the validity of N. lancensis was hinged, to some extent, on it being a subadult or adult that reached maturity at a smaller size and lighter frame than T. rex itself (Paul 1988; Bakker et al. 1988). This thinking reflects, in part, a period in vertebrate palaeontology dominated by typological approaches where specimens were assigned to species based on morphology alone without factoring in the influences of growth or individual variation. Under such schemes, it was rare, though not unheard of (Rozhdestvensky 1965; Russel 1970), to ally juvenile theropod specimens with adult specimens.

A more holistic philosophy toward the classification of fossil species has, in the last few decades, replaced such thinking and allows us to realise that, for example, a young juvenile specimen might look very different to an old adult. This can make characterising fossil species more difficult, but the results are more rewarding as we get more detailed and realistic insights into the life histories and ecologies of ancient animals. Such practices have typified most recent taxonomic work on Tyrannosaurus, emphasising the need to investigate features of specimen growth stage (e.g. bone fusions, histological evidence of specific age etc.) alongside morphology when considering what is, and what isn’t, Tyrannosaurus rex (e.g. Carr 1999, 2020; Carr and Williamson 2004; Woodward et al. 2020).

These works have shown that Tyrannosaurus was even more variable than initially realised, not only being differentiated in build and size as an adult but also looking entirely different as a juvenile. As is now well-known to Tyrannophiles, T. rex grew from a dog-sized, slender-snouted hatchling to a supersized, gnarly-skulled adult with unprecedented robustness among theropods. Carr (2020) characterises T. rex growth as undergoing “secondary metamorphosis” at the onset of sexual maturity, in which it transformed from a slender, “bird-meets-horse” juvenile body plan to a robust, deep-snouted, massive-bodied adult form. The result is an interpretation of Tyrannosaurus rex that occupied a range of disparate anatomies throughout its lifespan, within which small tyrants like “D. megagracilis” and “S. molnari” can be absorbed into a well-substantiated growth sequence (Carr and Williamson 2004). And yes, while a minority favour Nanotyrannus as a distinct taxon (e.g. Larson 2013), it also plots neatly into this system without issue (Carr 1999, 2020; Currie 2003; Woodward et al. 2020).

This hypothesis does more than just tidy up T. rex taxonomy. It also establishes that Tyrannosaurus was probably not a “normal” theropod in terms of its ecological role, and that there may be a good reason why this one species occupied such an expanse of time and space in Maastrichtian North America. The 30-odd year lifespan of one Tyrannosaurus encapsulated the ecological potential of several grades of predatory dinosaurs (Holtz 2021), and we might expect such an adaptable animal to have a long evolutionary history and wide geographic range. We might also predict an unusual amount of variation in our T. rex samples because, if this one species was undergoing such a transformation across three decades of growth, it would have transitioned through a large number of “morphs”. Coupled with distortion caused by fossilisation and the dusting of individual variation we’d expect among a reasonable sample of biological entities and we're going to find a lot of variation among our Tyrannosaurus fossils. It’s an interesting idea that explains a lot of weirdness around our T. rex sample, perhaps more parsimoniously than greater taxonomic granularity or sexual dimorphism. But on those topics…

Developing in parallel to this “everything is Tyrannosaurus” interpretation were research projects that argued to split T. rex apart. The drip of new Tyrannosaurus specimens that started in the 1960s and ‘70s had turned into a relative torrent in the 1990s, such that researchers of the 2000s had dozens of T. rex specimens to examine. This was a large enough sample that we had some hope of teasing more obvious trends or patterns out from the statistical noise of individual variation and we might, finally, get some perspective on whether T. rex was one species or several.

In 2008, Pete Larson specifically addressed this issue (Larson 2008a). He noted that some adult Tyrannosaurus skeletons can be much stockier and more heavily built than others and ascribed this to sexual dimorphism for a variety of reasons, most of which are now doubted. But he also highlighted variation in cranial pneumatic features, tooth counts and the number of small "incisiform" teeth at the front of their lower jaw. These, Larson argued, were more likely to be taxonomic differences. Only a few specimens possessed a full suite of distinctive features that separated them from T. rex proper, including the giant skull MOR 008, the privately-owned “Sampson” skeleton, and the iconic AMNH 5027. Larson set these aside from the T. rex hypodigm but erred on the side of caution by not establishing a new taxon: instead, he gave this potential second Tyrannosaurus species a nickname: Tyrannosaurus “x”.

The decade following Larson’s proposal has not provided much support for his suggested split in Tyrannosaurus specimens. In part, this reflects weak statistical support for the alleged taxonomic and sexual morphotypes. Joshua Smith’s (2005) assessment of T. rex dental characteristics and their utility in systematics can be seen as a blind test of size variation in the second dentary tooth, for instance, but failed to find significant characteristics of this in examined specimens. Mallon (2017) specifically looked for evidence of Tyrannosaurus sexual dimorphism along the parameters outlined by Larson but failed to find the sort of bimodal character distribution that would, in theory, result from differentiated sexes. More recently, Thomas Carr’s gigantic 2020 ontogenetic analysis of 44 T. rex specimens also found no taxonomic significance to Larson’s proposed dental and skull characters, nor did it recover evidence for sexual dimorphism. Moreover, it also failed to cluster the specimens ringfenced by Larson as T. "x". The size of Carr's analysis really has to be emphasised; by cataloguing up to 1850 features in 44 specimens, this investigation is, by far, the most in-depth analysis of anatomical variation within Tyrannosaurus (and maybe any dinosaur), and its failure to recover conclusions similar to those of Larson's 2008 study does not look good for the T. "x"hypothesis. Indeed, the scope and depth of Carr's investigation sets a high bar to clear for any different interpretation of T. rex variation, and that brings us to…

As will now be clear, a finger has been hovering over the big red button marked “new Tyrannosaurus species” for decades now, and someone, somewhen, was going to push it eventually. In this context we can see Paul et al. (2022) as simply being the team that took the plunge, offering to give yet another interpretation of Tyrannosaurus ;taxonomy by splitting T. imperator and T. regina from T. rex. And as will also now be clear to anyone familiar with Paul and colleagues’ arguments, there is a certain historic familiarity to some of their ideas: the characters used are very similar to those used by Larson for characterising T. "x"; specifically, regina, rex and imperator are distinguished by features of skeletal robustness and the number of smaller incisiform teeth in the lower jaw. To be fair, how these are quantified and measured is slightly different in Paul et al., but there's undeniable conceptual overlap. I suspect this is one reason the paper has not received the most enthusiastic response among tyrant dinosaur experts, who will not only have experienced a sense of deja vu from the concept of new Tyrannosaurus species but also immediately thought of those analyses which have already investigated and dismissed the taxonomic potential of robustness and dentary incisiforms. And, OK, we need to be careful not to be arrogant or dismissive in these considerations: it’s really easy to say “other analyses didn’t find this result, so you’re probably wrong”, but if a new paper is essentially arguing points that have already been shown as doubtful, it’s fair to point to the same studies that dismantled those ideas the first time around.

And it’s not only in character choice that the past might catch up with the conclusions drawn by Paul et al. As outlined above, T. rex taxonomy has enough history and complications under its skin that it’s hard to create, as Paul et al. attempt, a totally fresh Tyrannosaurus classification without running into leftovers from older proposals. For instance, Paul et al. refer the holotype of Dinotyrannus megagracilis to their new species T. regina, arguing that the megagracilis type is an incomplete juvenile specimen and thus a nomen dubium, and therefore unsuited to be the holotype for their new species. But, unless my understanding of ICZN rules is totally wrong, I’m not sure that’s how nomenclatural rules will shake out. For megagracilis to be identified as regina it must show at least some of the features Paul et al. assert are unique to T. regina; this being the case, it (theoretically) has features distinguishing it from T. rex as well. If a specimen can be identified to species level it cannot be a nomen dubium, and this should would allow megagracilis to re-enter the field as a valid, non-T. rex species. And if megagracilis is the same species as regina, then megagracilis has nomenclatural priority by 34 years: the "Tyrant Queen" becomes… "Tyrant Big Skinny". Note that megagracilis is also a Paul (1988) taxon, so it’s regina’s own senior author complicating the attempted, and admittedly pretty neat, emperor-king-queen naming scheme.

The imperator name is potentially at risk from an older name too, as highlighted by tyrannosaur expert/palaeontology social media machine Thomas Holtz in the New York Times. The holotype of Nanotyrannus lancensis stems from the lower portion of the Hell Creek Formation where imperator, according to Paul et al., is said to have existed independently from rex and regina. Given that the lancensis type specimen is clearly a juvenile animal, and most probably Tyrannosaurus (Carr 1999; Woodward et al. 2020), parsimony suggests it should be the same species as T. imperator. Paul et al. (2022) are non-committal about Nanotyrannus as a separate genus, noting its questionable status but not coming down on either side of the debate around its validity. But whether we regard the generic name as warranted or not, if Nanotyrannus is the juvenile of a stratigraphically older species of Tyrannosaurus, lancensis would have nomenclatural priority over imperator. The Tyrant Emperor would become humdrum T. lancensis.

It’s also interesting to see how Paul and colleagues’ conclusions compare to those of previous workers. Of note are that specimens highlighted by others as potential new Tyrannosaurus taxa — such as AMNH 5027, the Texan maxilla — are considered indeterminable by Paul et al., and that Larson (2008a), despite utilising the same dentary tooth character, retained the holotypes of regina and imperator in T. rex. There’s nothing wrong with different projects drawing contrasting taxonomic conclusions, of course, but this shows that attributes as basic as specimen tooth size are being interpreted very differently by different authors. This demonstrates the difficulty, and perhaps the degree of subjectivity, associated with carving our Tyrannosaurus dataset into multiple species. Perhaps the clusters of specimens we're identifying as distinct "species" are strongly influenced by our own opinions on what features are taxonomically important, and also how we choose to measure and characterise those features? The discussion of Tyrannosaurus dentary incisiform size by Paul et al. is telling in this regard, highlighting the existence of “borderline” specimens that could be interpreted one way or another. Differing takes on such specimens probably account for the varied views proposed by different researchers.

The take-home here is a simple one: T. rex is a tricky animal to split into multiple taxa, but the problem is not because researchers are wedded to the idea of there being one species of this animal. To the contrary, exactly the opposite is true: people have been trying to blow T. rex into smaller taxonomic units for half a century! Rather, it's because what seems like a straightforward job of naming a few new tyrannosaur species is complicated by the enormously variable nature of the T. rex dataset and also the lesser-known, but still important, taxonomic considerations of past workers. I admit to seeing Paul and colleagues’ work as less of a “mini-revolution” than the latest contribution in a decades-long conversation about the variation apparent in Tyrannosaurus specimens. Their work is distinguished, of course, by finally erecting new Tyrannosaurus species after 50 years of researchers tossing this idea around, and in this respect, Paul et al. have moved the conversation on. But in doing this, a greater spotlight is now shone on past attempts to investigate Tyrannosaurus variation, including the legitimacy of species previously sunk into T. rex as well as the value of anatomical characters already of doubtful value in dividing up the T. rex sample. It’s the response to these that will determine how far our understanding of Tyrannosaurus diversity has been moved by these new proposals.