The 'Pteranodon complex' and dismantling our understanding of the most famous flying reptile

Pteranodon longiceps, Pteranodon sp. or something else entirely? In recent years one of our most famous and abundant pterosaurs has been carved up into multiple species, but is this overzealous taxonomic splitting, or is there more to it than that?

Writing about pterosaurs can be difficult because so much of their classification is disputed. The number of pterosaur species, their assignment to different groups, appropriate clade nomenclature and the arrangement of branches in the pterosaur tree are all contested, sometimes to polarising extents.

A bastion of taxonomic stability in all this is Pteranodon, everyone's favourite giant, toothless Late Cretceous ornithocheiroid (or pteranodontoid) from interior regions of the United States. Known since the late 1860s, Pteranodon is one of the most substantially sampled of all pterosaurs and we now have well over 1100 specimens in museums around the world. This record stems from a relatively limited geographical area and is constrained stratigraphically to the Smoky Hill Chalk Member of the Niobrara Formation, with a smattering of fossils from the overlying Pierre Shale Group.

A series of papers documenting Pteranodon anatomy, variation and stratigraphy, all penned by pterosaur expert S. Christopher Bennett during the 1980s-2000s, have made this pterosaur one of the best understood of all flying reptiles (perhaps the most important entries in this series are Bennett 1992, 1993, 1994, 2001a, 2001b). These publications are the result of examining several hundred Pteranodon specimens and are among the most significant and comprehensive contributions to pterosaur literature in modern times. I recommend them to any students of vertebrate palaeontology: even if you don't agree with their conclusions, they're great examples of clear writing, of hypotheses being established and tested, and of large amounts of data being presented clearly and logically.

Skeletal restorations of P. longiceps male (the larger animal) and female morphs, based on Bennett (1992). Illustration from Witton (2013).

For pterosaur workers, one of the most important outcomes of Bennett's work was a robust taxonomy for Pteranodon. This genus was once a polyspecific monster composed of 13 species, but Bennett (1994) whittled it down to two stratigraphically segregated forms: the geologically older Pteranodon sternbergi and its direct descendent, Pteranodon longiceps (Bennett 1994). Measurements and observations of hundreds of Pteranodon fossils and detailed analysis of its growth regime suggested that most variation seen in Pteranodon samples resulted from sexual dimorphism (above), where (presumed) males are identical to females except for being 50% larger, bearing bigger headcrests and narrower pelves (Bennett 1992). We can recognise osteologically mature Pteranodon by details of skeletal fusion, bone texture and histological structure (Bennett 1993), thus allowing us to determine that the small, 'female' individuals were not just juveniles but, in fact, relatively small adults. Although sexual dimorphism had been proposed for pterosaurs previously, few studies went to such detail in making their case and Bennett's 1992 work stands as one of the better cases made for sexual dimorphism in a fossil reptile. This complex consideration of Pteranodon diversity can be viewed as a milestone in our modernisation of pterosaur research, it being a clear sign that pterosaur studies were maturing to the level attained by dinosaur or mammal vertebrate palaeontology in the 1980s and 1990s. This work has been uncontested for over a decade and subsequent studies have since found evidence for similar morphological trends in other pterosaur species. Hurrah, hooray and huzzah for Pteranodon, then, the pterosaur worker's faithful friend and our securest mast in a taxonomic storm.

But then things got a complex

Given the established status of Pteranodon taxonomy it came as something of surprise when, in 2010, a counterargument to Bennett's interpretation of Pteranodon was published. Another big name in modern pterosaur research, Alexander Kellner, proposed that Bennett's Pteranodon was in fact a 'complex' of at least four species (perhaps five) in three genera (Kellner 2010). Kellner's alternative scheme suggested that the giant, swollen-crested sternbergi was different enough from longiceps to warrant a separate genus, and resurrected the 'subgenus' Geosternbergia for this purpose (giving us the rather daft name Geosternbergia sternbergi). A second Geosternbergia species was proposed for a partial skull referred to P. longiceps by Bennett (1994), which Kellner named G. maiseyi. Another skull, this one with a broken crest but the best preserved rostrum of any giant Pteranodon specimen, was said to represent a third pteranodontid genus, the deep-snouted Dawndraco kanzai. Finally, although not naming a new taxon, Kellner (2010) singled out another P. longiceps specimen as being distinct from this species, arguing that this long-crested specimen has a crest which is too upright to be referred to longiceps: he referred this simply to Pteranodon sp. for now. You can see these skulls, and how they contrast with Bennett's older scheme, below.

Differing interpretations of some important Pteranodon skulls. Blue text and panelling reflects the Bennett (1994) interpretation of Pteranodon skull taxonomy, green text shows where Kellner (2010) differs. Skull images borrowed from Bennett (1994).

This might not seem like a big deal - after all, famous fossil species are carved up all the time - but this has implications beyond just having to learn a few new binomials. The presence of multiple genera in our 'Pteranodon' sample makes it difficult to classify the majority of Smoky Hill pterosaur material, and thus our thousand-strong Pteranodon catalogue mostly becomes Pteranodontidae incertae sedis, with a few named skulls. With that, the statistical support for our hypotheses of Pteranodon variation, growth and sexual dimorphism require reevaluation, because we've lost our grip on what animals those hundreds of measurements actually pertain to. For pterosaur workers, this is something to pay attention to: one of our 'cornerstone' taxa might not be the dependable, go-to reference pterosaur that we thought it was, and its palaeobiology may not be as well understood as previously considered.

I've been asked about the 'Pteranodon complex' several times and thought it was time to share my thoughts here. I normally avoid talking about detailed taxonomy because I'm aware how dry it can be, but the Pteranodon controversy is pretty interesting. There are lot of strands of data to consider, some philosophising about palaeontology itself, and - if nothing else - the reality about the fossils behind Pteranodon might be of interest. This is only a summary of course - if you're interested, you really need to check out the papers cited below for the full details.

How understanding hundreds of Pteranodon specimens hinges on a handful of important ones

The holotype skull of Pteranodon longiceps, the only Pteranodon specimen which can be objectively referred to the genus. This skull is from a small (presumed female) morph. From Eaton 1910.

Since at least Eaton (1910) it's been recognised that the majority of Pteranodon specimens are not diagnostic to specific level. Most Pteranodon fossils are bits of limb or scraps of bodies that can be identified as Pteranodon (or pteranodontid, if you prefer) but not much further. To know what species we're looking at we need the back of a skull, and ideally, a big one with a good amount of crest. One of the key points to stem from both Bennett's (1994) taxonomic review and Kellner's (2010) paper is that Pteranodon species are best differentiated by the orientation and shape of their headcrests. Bennett (1994) considered this in a fairly simple way: sternbergi has an upright and distally swollen crest, while longiceps has a more posteriorly directed, distally tapering one. These distinctions can be seen in smaller skulls, but are most obvious in the bigger ones. sternbergi and longiceps might also be distinguished by the orientation of the posterior skull margin (sternbergi being more upright than longiceps) and slenderness of the mandible (sternbergi being a touch shallower) but the crest shape and angle is the best way to tell these taxa apart.

Bennett's characterisation may seem quite broad, maybe even simplistic, but there's a reason for that: no two Pteranodon crest specimens are entirely alike and none of our better, larger skull specimens are complete (below). We have some excellent and complete smaller skulls (above), and several incomplete large specimens, but any visage you see of a long skulled, long-crested Pteranodon fossil is an interpretation of fragmentary specimens. Bennett's (1994) taxonomy reflects this, using relatively broad characters to separate the species because the material ultimately offers limited scope for detailed comparison or augmentation with other characters. The fact that the crests differ somewhat within Bennett's species is explained by their likely role in visual communication rather than biomechanics (Bennett 1992; Tomkins et al. 2010): such structures are often far more variable in appearance, and sensitive to factors like ontogeny, than strictly 'functional' anatomies.

Line drawings of important Pteranodon/pteranodontid skulls from Witton (2013). A, skull still referred to P. longiceps; B, isolated crest and part of the braincase region referred to either longiceps (Bennett 1994) or Pteranodon sp. (Kellner 2010); C, holotype of longiceps; D, holotype of Pteranodon (or Geosternbergia) sternbergi. Note the twisted posterior skull face in B and how little of the skull remains in D.

Kellner (2010) argues that Bennett's interpretation accommodates too much morphological variation however, picking out several skull characters as sufficiently distinctive to warrant erecting new genera and species. The diagnoses for these new taxa are much more specific than those offered by Bennett, pertaining not only to crest shape and angle, but also size and shapes of skull bones, skull openings and rostrum morphology. Partly because these criteria are quite specific, these novel pteranodontids are currently represented by single specimens. And it's here that I think we hit a bump with the 'Pteranodon complex' hypothesis. The diagnoses are quite specific, and we have good reason to think a lot of the variation apparent in Pteranodon fossils is not taxonomic in origin. For instance, taphonomic damage and the significant crushing that affects all Pteranodon bones (most Pteranodon bones are reduced to thicknesses of mere millimetres) means no two Pteranodon skulls are identical, and many diagnostic characters suggested by Kellner (2010) - specifically those pertaining to bone lengths, fenestra sizes and so on - have yet to be demonstrated through illustrative or quantified means. We've yet to see the measurements, data tables or an illustrated series of Pteranodon skulls which show these features are atypical against a range of specimens, and thus suitable to base new taxa on.

It's not just taphonomic and diagenetic effects which are of concern: there are palaeobiological trends to consider, too. For example, Kellner (2010) uses the breadth of the crest base as a diagnostic feature for both Dawndraco and G. maiseyi, noting that the former has a crest base located largely behind the eye socket, while the latter is expanded to erupt well in front of the orbital region. But Bennett (1994) gives reason to think that crest base size is linked to growth and size, not taxonomy. As can be seen above, there's a steady correlation between crest base size and skull size: larger skulls have much thicker crest bases extending far in front of the orbit, than those of smaller skulls (Bennett 1994, 2001a). Although Kellner (2010) mentions that Dawndraco is a relatively mature specimen, and thus maybe unlikely to change its crest size, there's no discussion of the fact that the Dawndraco skull is quite a bit smaller than some other 'large' Pteranodon skulls (below). The fact this small skull has a smaller crest is, of course, consistent with Bennett's crest scaling hypothesis. Similarly, the wide-crested maiseyi skull meets Bennett's predictions that it should - as a big individual - also have a relatively large crest base.

Dawndraco (red) is a bit of a wimp compared to the largest Pteranodon skulls. Black is the sternbergi holotype, blue is the maiseyi holotype. Note how the crest bases of the black and blue skulls are much broader than that of Dawndraco. Illustrations adapted from Bennett (1994).

Some parts of the 'Pteranodon complex' hypothesis also face issues with specimen comparability. Some allegedly diagnostic features are based on very poorly understood aspects of Pteranodon anatomy, such as the relatively deep jaw of the Dawndraco skull. According to Kellner (2010) this rostrum is diagnostically deep and peculiarly shaped: this is certainly true when compared to complete smaller Pteranodon skulls, but no large Pteranodon has well-preserved jaws and we can't compare like-with-like. The best we can do is look at fragmentary remains, all of which suggest large Pteranodon also had deep, subparallel-sided jaws (below; Bennett 1994, 2001a). However, because none of these are associated with posterior skull remains, we can't gauge their depth in any context. This being the case, the fact that Dawndraco has the deepest rostrum known from a pteranodontid is of questionable significance: similar morphologies clearly existed in other Pteranodon, we just can't tell if they're identical to Dawndraco or not. Similar issues occur when trying to fathom the significance of cranial crest shape and orientation for some unusually crested specimens. Many of these crests are only partly preserved, or not associated with substantial skull remains. As noted above, we have reason to think the context of the wider skull anatomy is important for interpreting crest anatomy, and this suggests a need for caution when it comes to erecting new pteranodontid taxa based on these specimens. Clearly, the issue here is that we have a huge amount of data for Pteranodon, but only a tiny part of it is taxonomically relevant, and only a fraction of that portion can be compared to a meaningful degree across a good number of specimens. Big sample sizes are meant to make things clearer in science, but for Pteranodon they seem to make things more complicated!

The Dawndraco skull compared to fragmentary Pteranodon sp. jaw tips. Note how the subparallel dorsal and ventral margins and (predicted) Dawndracro overbite are present in other Pteranodon fossils. Note that some small Pteranodon have overbites too. Drawings after Bennett (1994).

Pteranodon stratigraphy and the significance (or not) of geological boundaries

Both Bennett's and Kellner's taxonomies consider Pteranodon distribution through the Niobrara Formation and neighbouring rock units, but there are fundamental differences in how they treat this data. Bennett's (1994) approach sees morphology trump stratigraphy in that the ranges of his species are dictated wholly by specimen anatomy. This is essentially the approach typically taken by biostratigraphers, where it is considered (and relied upon) that species distribution is not linked to our designation of rock units. In this scheme, it doesn't matter where the specimen occurs, but what it looks like that matters. The fact that all the 'sternbergi morphs' occur at the base of the Smoky Hill Chalk Member, and all the 'longiceps morphs' occur at the top (and somewhat beyond - see below) is the basis for Bennett's (1994) idea that our Pteranodon sample is a single, evolving population which entered the fossil record as sternbergi, and left as longiceps. The fact that these species do not overlap can be viewed as helping the verify the Pteranodon chronospecies hypothesis.

Kellner (2010) takes a different approach to stratigraphy, where provenance is a factor in the likelihood of a specimen being assigned distinct taxonomic status. A good chunk of Kellner (2010) is devoted to discussing the role of stratigraphy in taxonomy, it being argued that Pteranodon skulls found several levels away from each other were not contemporaries and thus cannot be reliably assessed for intraspecific variation. When this happens, taxonomic significance takes over as the most likely (or perhaps default) interpretation of morphological differences.

Kellner (2010) makes specific mention of the fact that neither the Dawndraco or maiseyi skulls are from the same horizons as other Pteranodon type material (below). Particular attention is drawn to maiseyi, which comes from the Sharon Springs Formation: a unit two formations above the Niobrara Formation and its glut of Pteranodon material. Of this, Kellner states: "One could argue that the morphological differences of Geosternbergia maiseyi might be due to ontogeny, individual variation or even sexual dimorphism, but there is a considerable time gap between these [pteranodontid] species that never co-existed." (Kellner 2010, p. 1078). The implication here is that there is a stratigraphic limit to when similar-looking animals might be considered conspecific, and that morphological similarity is eventually overruled by provenance.

Pteranodons in time - click to embiggen and see full details. Grey lines show distribution of key Pteranodon specimens, black lines show those associated with skull illustrations. Skull diagrams from Bennett (1994), data from Bennett (1994); Hargrave (2007) and Kellner (2010).

These discussions touch on almost philosophical elements of palaeontological science, and I expect readers will differ as to which approach they think is most useful. Personally, I don't agree with the use of stratigraphy in taxonomic considerations. It's generally accepted that paleontology uses a morphology-based species concept (morphospecies) and, if that's the case, we have to stick by it. This means letting morphology dictate the ranges of fossil species and not deciding a priori that a span of time/extent of rock exceeds an acceptable 'species range'. For abundant, well-documented groups we may be able to bolster such concepts with a sense of their speciation frequency but, with rare fossils like pterosaurs, we know next to nothing about their evolutionary rates. And as unusual as it may seem for a pterosaur to span several formations, there are taxa that seem to do this (Anhanguera, Istiodactylus, Quetzalcoatlus, Rhamphorhynchus are familiar examples). Moreover, plenty of other groups pay little attention to the stratigraphic boundaries that we set. Indeed, the whole science of biostratigraphy is is more or less founded on this fact: we can date the rock record using fossils because so many species do transcend stratigraphic boundaries. Stating that a fossil cannot be conspecific with another just because it occurs in younger or older rocks seems presumptuous and at odds with trying to understand evolutionary history.

More specific concerns with the 'Pteranodon complex' approach to stratigraphy is that its perceived issue with Pteranodon ranges are not mirrored by those who work on other Niobrara Formation vertebrates. From fish to marine reptiles, it's widely thought that many Niobrara species persisted through big chunks of the three million year period recorded by the Smoky Hill Chalk Member and Pierre Shale Group (e.g. Everhart 2005; Carpenter 2008). If large swathes of the Smoky Hill Chalk fauna can survive over long periods of time, why can't Pteranodon species? It is noteworthy here that Hargrave (2007) identified new, potentially diagnostic Pteranodon longiceps bones from the Pierre Shale. If so, this bolsters older suggestions that longiceps occurs above the Niobrara Chalk (Kellner (2010) was unconvinced of their referral to longiceps, however). We might also note that the 'Pteranodon complex' taxa accord less with stratigraphy than alternatives, in that Geosternbergia disappears during the interval represented by the upper Smoky Hill Chalk and Gammon Ferruginous Formation, only to reappear in Sharon Springs beds. This is despite there being a higher number of skulls the upper Smoky Hill than any other Pteranodon bearing interval (Bennett 1994). This isn't an insurmountably complex distribution of course, but in terms of parsimony, Bennett's (1994) scheme must be seen as simpler and more congruent with stratigraphic data.

'Pteranodon complex', or Pteranodon simple?

Tying this all together, I hope it's clear that the 'Pteranodon complex' is quite a complicated issue, and one that will take some work to resolve one way or the other. I've had to skim over many of the details here, so be sure to read the papers cited above if you'd like to read the full story. Many are available online.

It would perhaps be remiss to outline all this without giving my own take on this shake up of Pteranodon taxonomy. In my 2013 book I said I preferred Bennett's (1994) scheme and followed it accordingly and, revisiting this debate several years later has not changed my mind. I stress that I'm not 'against' the idea of more Pteranodon species, just that - in my opinion - the evidence points to Pteranodon containing longiceps and sternbergi, and that these species are each others closest relatives and might as well stay congeneric in Pteranodon. For reasons outlined above I find the stratigraphic arguments about separating these taxa unconvincing, and I don't think the morphological arguments are developed enough yet to overturn those for synonymy.

Concerning the specific taxa, the Dawndraco skull seems to be about right for a small 'male morph' P. sternbergi, and probably mostly seems atypical because of its relative completeness. Most large Pteranodon probably have those big rostra (you'll note that all my paintings of large Pteranodon, like that above and here, have this feature). What I've seen of its postcrania is extremely Pteranodon-like too, right down to its peculiar, highly characteristic tail (see Kellner 2010, p. 1074). I can appreciate why some folks might consider the maiseyi specimen a different taxon because of its seemingly unusual crest. However, the fact the leading crest edge is relatively complete but does not swell forwards means it is not particularly sternbergi-like, despite Kellner's (2010) suggestion that the maiseyi specimen is more closely related to sternbergi than anything else (Kellner 2010). Indeed, as preserved, the maiseyi crest meets the criteria of longiceps provided by Bennett (1994) as well as his predictions that it should have a huge crest base because of its large overall skull size. Moreover, the posterior and dorsal crest margins are broken: there is greater potential for the complete maiseyi crest to be more longiceps-like (longer, posteriorly directed) than sternbergi-like (tall, expanded forwards).

As for the large longiceps crest referred to Pteranodon sp., the specimen is not only (and obviously) very incomplete but the crest base is badly deformed, and I find it difficult to orientate the specimen against other skulls to determine the crest angle. There are suggestions that the crest base is too tall over the orbit to be longiceps (Kellner 2010) but, again, this region seems to change a lot with size and this specimen seems to have belonged to a big skull (judging by the orbit proportions): this needs to be considered carefully. The crest shape itself is generally longiceps-like, of course, and I suspect this specimen is just a big, mature version of this species.

So cheer up matey, you might not be a 'sp.' after all.

Of course, all this is subject to change should new ideas and data on Pteranodon be published in future. I should close by saying that the 'Pteranodon complex hypothesis' will soon become the 'Pteranodon complex debate': several authors are working on technical follow ups to Kellner's (2010) paper and describing relevant specimens that have bearing on this topic. This matter, then, is far from closed, and it's going to be interesting to see how it pans out. Now that we have a 'primer' article, if and when new papers are published, perhaps we'll cover them here.

This blog post on the 'Pteranodon complex' was made less complex because of support from Patreon

The paintings and words featured here are sponsored by a most excellent group of people, my Patreon backers. Supporting my blog from $1 a month helps me produce researched and detailed articles with paintings to accompany them, and in return you get access to bonus blog content: additional commentary, in-progress sneak-previews of paintings, high-resolution artwork, and even free prints. For this post, we'll be taking a look at one of the most interesting, and barely ever mentioned parts of Pteranodon anatomy. If you want to know what it is, head over to Patreon to get access!

References

• Bennett, S. C. (1992). Sexual dimorphism of Pteranodon and other pterosaurs, with comments on cranial crests. Journal of Vertebrate Paleontology, 12(4), 422-434.
• Bennett, S. C. (1993). The ontogeny of Pteranodon and other pterosaurs. Paleobiology, 19, 92-106.
• Bennett, C. S. (1994). Taxonomy and systematics of the late Cretaceous pterosaur Pteranodon (Pterosauria, Pterodactyloidea). Occasional papers of the Natural History Museum. 169, 1-70
• Bennett, S. C. (2001a). The osteology and functional morphology of the Late Cretaceous pterosaur Pteranodon Part I. General description of osteology. Palaeontographica Abteilung A, 1-112.
• Bennett, S. C. (2001b). The Osteology and Functional Morphology of the Late Cretaceous Pterosaur Pteranodon Part II. Size and Functional Morphology. Palaeontographica Abteilung A, 113-153.
• Carpenter, K. (2008). Vertebrate biostratigraphy of the Smoky Hill Chalk (Niobrara Formation) and the Sharon Springs Member (Pierre Shale). In High-Resolution Approaches in Stratigraphic Paleontology (pp. 421-437). Springer Netherlands.
• Eaton, G. F. (1910). Osteology of Pteranodon. Connecticut Academy of Arts and Sciences, Memoirs.
• Everhart, M. J. (2005). Oceans of Kansas. Indiana University Press.
• Hargrave, J. E. (2007). Pteranodon (Reptilia: Pterosauria): stratigraphic distribution and taphonomy in the lower Pierre Shale Group (Campanian), western South Dakota and eastern Wyoming. Geological Society of America Special Papers, 427, 215-225.
• Kellner, A. W. (2010). Comments on the Pteranodontidae (Pterosauria, Pterodactyloidea) with the description of two new species. Anais da Academia Brasileira de Ciências, 82(4), 1063-1084.
• Tomkins, J. L., LeBas, N. R., Witton, M. P., Martill, D. M., & Humphries, S. (2010). Positive allometry and the prehistory of sexual selection. The American Naturalist, 176(2), 141-148.
• Witton, M. P. (2013). Pterosaurs: natural history, evolution, anatomy. Princeton University Press.