Wednesday, January 26, 2011

Scientific Anachronism and why Biostratigraphy Matters

Image: Stegosaurus confronts Tyrannosaurus, from Walt Disney's Fantasia. Such anachronistic views of paleontology could never form the basis of peer-reviewed literature, could they?

A new study (Carbone, Turvey &Bielby, 2011) suggests T. rex could not have been a pure scavenger.

Yeah, I know. This is already the universally accepted position of modern paleontologists. Of course T. rex scavenged if it could, but there is ample reason to think (and ample fossil support backing this up) that it hunted as well. Even the originator of the scavenging thoery, Jack Horner, has basically admitted that he only came up with it as a way to get young people to think critically about their own preconceptions (obviously, he has never met any young paleontology fans. I'd have given up that strategy after I witnessed my first T. rex vs. Spinosaurus debate).

However, as Denver Fowler pointed out on the DML today, drawing an obvious conclusion is the least of the paper's problems.

I recently participated in at least two paleoart discussion threads in which really awesome artists showed off some mind-blowingly fantastic paintings depicting the Jehol biota. I know first hand that both artists are completely on the ball and know their stuff. But both made common errors in the often neglected field of biostratigraphy.

In one, a Sinornithosaurus watches as two Microraptor glide down from the trees. These are two similar animals from around the same time and place. However, there is no evidence that they ever met. All known fossils of Microraptor come from the Jiufotang formation, dated to 120 million years ago, plus or minus 700k years. The youngest Sinornithosaurus fossils are from the upper Yixian formation, dated to around 122 million years ago. The timespan and environment are grossly similar, but 2 million years is still a long time in a world where most dinosaur species, if not genera, don't span more than a couple million years (and the ones that do are probably egregiously over-lumped, like Iguanodon).

Another painting portrayed a Yixian formation scene with Yixian ornithopods and Yixian insects being fed on by Jeholopterus, a pterosaur which lived in the Daohugou biota, in beds dating to at least 150 million years ago, a full 25 million years before the Yixian faunas existed. The error here was probably based on a confused history of dating the formations (old sources placed the Yixian in the late Jurassic), and many sources, both professional and popular, which tended to conflate the various Chinese feather-preserving formations into one amorphous pseudo-fauna.

Artistic depictions throwing together prehistoric animals from disparate times are obviously nothing new. Walt Disney Pictures has done this at least twice, first and most famously in Fantasia (Stegosaurus meets T. rex meets Pteranodon), and later and more flagrantly in Dinosaur (I can't think of any two animals in that movie that were actually contemporaries, and many didn't even live together on the same continent).

This is somewhat excusable when it's done for the sake of art (as long as that art isn't passed off as being scientifically rigorous). But this kind of disregard for, or generalization of, biostratigraphy can creep into science and completely foul up your results.

In Carbone et al. 2011, the authors attempt to calculate the amount of potential carcasses that would have been available to scavenging Tyrannosaurus rex in its environment to make their case. Their lists of T. rex contemporaries are reproduced in part below:

Species and body masses of carnivorous non-avian theropod dinosaurs of Late Cretaceous North America
Dromaeosaurus albertensis
Richardoestesia gilmorei
Richardoestesia isosceles
Velociraptor sp.
Troodon formosus
Chirostenotes elegans
Chirostenotes pergracilis
Nanotyrannus lancensis
Albertosaurus sarcophagus
Tyrannosaurus rex

Species and body masses of herbivorous dinosaurs of Late Cretaceous North America
Parksosaurus warreni
Prenocephale edmontonensis
Ornithomimus velox
Struthiomimus sp.
Thescelosaurus garbanii
Thescelosaurus neglectus
Leptoceratops gracilis
Montanoceratops sp.
Pachycephalosaurus wyomingensis
Edmontosaurus annectens
Edmontosaurus regalis
Edmontosaurus saskatchewanensis
Lambeosaurus sp.
Parasaurolophus walkeri
Edmontonia rugosidens
Ankylosaurus magniventris
Triceratops horridus
Alamosaurus sanjuanensi

The authors state, "our species list is treated as representing a consistent sympatric faunal unit across this region for the purposes of analysis." But they absolutely don't represent that.

If you have even a little bit of a handle on Late Cretaceous biostratigraphy, or the paleoecology of T. rex, you may notice a few problems with these lists. Namely, the fact that they are complete messes, incorporating erroneous or outdated taxonomic assignments or over-generalizations of the geologic column.

This kind of data crunching would require taxa to be broken down on an environment-by-environment basis. That is, in order to be meaningful, all included taxa have to be demonstrated to be contemporaries. Most of the taxa in those lists were not, or can't be said to have been with any confidence.

To be fair, some of the mistakes are due to very new research, some of which has only appeared in abstracts or mentioned briefly in papers. For example, while Edmontosaurus regalis is widely reported from the late Maastrichtian Hell Creek and Lance formations, this is mainly by default, skeletons that are not identifiable to the species level. Ongoing stratigraphic and taxonomic work by Nicolas Campione has shown that E. regalis was actually not a contemporary of E. annectens, and specimens assignable to that species are only known from lower strata. The validity of E. saskatchewanensis, which is from the same stratographic level as T. rex, is pretty dubious. E. annectens is its likely synonym.

Parasaurolophus is known exclusively from the Campanian-age Dinosaur Park Formation, over five million years before the earliest known T. rex fossils. Same goes for Lambeosaurus. While the former was tentatively identified in the Hell Creek by Sullivan & Williamson (1999), this was based on very fragmentary remains that almost certainly belong to Edmontosaurus instead. Montanaceratops is from the St. Mary River Formation, dated to the early Maastrictian and also pre-dating T. rex.

Some cases are even more nuanced. Alamosaurus did coexist with T. rex, but not with many of the other listed species. Current indications are that the southern part of North America during the late Maastrichtian supported a different fauna from the north, comprised many of species which are related to, but distinct from, their northern counterparts. Alamosaurs, for example, did not coexist with Triceratops, but Ojoceratops (assuming they're distinct). It didn't coexist with Torosaurus latus (which the authors apparently lump with Triceratops), but with "Torosaurus" utahensis. Indications are that these beds are a bit earlier than the late Maastrichtian as well, so while Alamosaurus lived alongside Edmontosaurus, it was E. regalis rather than E. annectens.

The carnivores don't fare much better. Most are tooth taxa, like Troodon (another Dinosaur Park critter from the Campanian). While "Troodon" teeth are known from the same beds as T. rex, they're almost certainly not Troodon formosus. The same goes for Dromaeosaurus. However, these are taxonomic issues, not biostratigraphic ones, and don't really affect species count--whatever we name them, there were at least one troodontid and at least two dromaeosaurids present (though the authors erroneously list both Velociraptor and Saurornitholestes, based on the same specimens, first referred to the former and then the later, both incorreclty). Not so for the inexplicable inclusion of Albertosaurus. I can figure out where these other misplaced species came from, but I don't know of any albertosaur remains having been reported from Lancian-age deposits. Anybody? Either way, it's almost certainly an error (as is making Nanotyrannus a distinct taxon, but that's another story).

So you can see why failing to understand which dinosaurs lived together, specifically, can have major implications for actual science. This kind of paper also illustrates why it's a bad idea to keep non-diagnostic genera around as nomina dubia and not sink them into their better known, probably-synonymous counterparts or simply designate neotypes from the good material. These authors avoided pitfalls like including Thescelosaurus infernalis (=T. sp.), Manospondylus gigas (=Tyrannosaurus rex), Aublysodon molnari (=Tyrannosaurus rex), Thespesius occidentalis (=Edmontosaurus annectens), Trachodon mirabilis (=Edmontosaurus annectens), or Agathaumas sylvestris (=Triceratops horridus), but those taxa aren't doing science any favors by cluttering the playing field.

Here's my preliminary attempt to clean up their faunal lists, based on a Lancian-age, northern ecosystem: (updated thanks to additional information provided by Mickey Mortimer in the comments. note that I'm following the authors in not including avialans)
Dromaeosaurinae sp.
Zapsalis abradens
Richardoestesia gilmorei
Richardoestesia isosceles
Troodontidae indet. spp. (multiple species)
Pectinodon bakkeri
Paronychodon sp.
Avimimidae sp.
Chirostenotes elegans
Chirostenotes? sp.
Tyrannosaurus rex (=Manospondylus gigas)
Struthiomimus sedens
Ornithomimus velox
Ornithomimidae sp. (="Orcomimus")
Dromeiceiomimus sp.
Alvarezsauridae sp.
Therizinosauridae sp.
Thescelosaurus garbanii
Thescelosaurus neglectus
Leptoceratops gracilis
Pachycephalosaurus wyomingensis
Edmontosaurus annectens (=Thespesius occidentalis)
Edmontonia schlessmani (=Denversaurus schlessmani)
Ankylosaurus magniventris
Torosaurus latus?
Triceratops horridus (=Agathaumas sylvestrus?)

* Campione, N.E. (2009). "Cranial variation in Edmontosaurus (Hadrosauridae) from the Late Cretaceous of North America." North American Paleontological Convention (NAPC 2009): Abstracts, p. 95a.


  1. Knew something was off on this one as soon as I saw Albertosaurus and Dromaeosaurus in the list.

  2. Hit the post button too quickly.

    I concur there needs to be more awareness on the smaller stratigraphic divisions. Too many reconstructions stick anything from North America in the Late Cretaceous together.

  3. Matt.

    Great post and you are absolutely correct that it is an old and ongoing problem. We went through great pains to make sure that our new mural of a Late Triassic scene at Petrified Forest only featured animals not only from a single horizon, but also a restricted geographical area. In the past the park "biota" has been depicted as a single snapshot in time blurring faunal and floral separation. I've been emphasizing for years now that we have 20 million years represented here with actually provides numerous different 'snapshots' through time.

  4. We do not know whether (nor can I scientifically validate based on current data) that "Lancian" and "Judithian" troodonts do not all represent a single "species." I might even break my rule about playing with taxonomy if I ever get a shot at examining NA troodontid teeth and bones and perform the qualitative and quantitative analysis I've wanted to do on them.

    Some additional issues also arise from the paper, in the assumption that the theropods were all carnivores and not potential prey or carcass-producers themselves. I seriously doubt a chance encounter with other predator corpses would have presented a barrier to an opportunistic consumer of any type.

  5. @Bill Parker:
    Glad somebody is taking this kind of thing seriously and doing the work to tease distinct fauna out of previous 'conglomerates'! I think for public exhibits like that, this is very important in helping people realize just how long the time spans we're dealing with are, by showing how ecological communities evolve.

    Given that most Lancian troodontid material is based on teeth, I'm not sure we can know. However, given the temporal ranges usually found for similar forms I sincerely doubt the same species would have persisted for that long (unless you lump everything in an isolated anagenetic lineage into one species). Whether or not they're identical in morphology to T. formosus, I think it would be a leap to include them in that species. Are troodontids known from both teeth and skeletal material diagnosable by teeth alone? I'm not sure anybody has demonstrated this.

  6. Matt,

    No one has demonstrated otherwise, either. Troodon formosus, rather than being a skeletal exemplar, appears more akin to a form genus, much like an ichnite, where the idea of a species lineage segment or reproducible population "bubble" cannot be used to segregate apparent taxa. Authors may claim that formational or horizonal boundaries affect the taxonomy in some meaningful way, but this also fails to be tested, thus leaving us with the current hypothesis in consensus usage:

    Troodon formosus -- "morphospecies" -- known from the Campanian and Maastrichtian of North America.

    There are arguments to be said against this, and surely it is not that hard to make them in print were it not for the vast array of tooth specimens out there that must be examined as well as the large number of skeletal material that sits undescribed at MOR; but so far, what we have is a 5-8myr for a taxon.

  7. Stokosa (2005) discusses a couple teeth from the Lancian that may belong to Albertosaurus or something similar.

    Stokosa, K., 2005. Enamel microstructure variation within the Theropoda. pp. 163-178 in Carpenter, K. (ed.), The Carnivorous Dinosaurs. Indiana University Press, Bloomington.

  8. Treating Nanotyrannus lancensis as a seperate taxon, when it is most probably a jeuvenile T. rex, really sticks out to me as not making any sense. I mean, why even inlcude it?

  9. Dealing with Late Maastrichtian North American taxa, ignoring the geography...

    There are at least two dromaeosaurines- Dromaeosaurus sp. nov. and Zapsalis. Note the "velociraptorine" material is based on an old taxonomy, since most modern analyses (e.g. Longrich and Currie, Senter) don't place any American material in that clade. There are also multiple troodontid morphotypes (Sankey et al., 2002's troodontid, Baszio's 1997 Scollard supposed Troodon teeth, Pectinodon) and Paronychodon. For oviraptorosaurs, there's the large Hell Creek caenagnathid and Avimimus. Also therizinosaurs and alvarezsaurids. Besides sedens, there's Dromiceiomimus (RTMP 93.104.1), "Orcomimus" and Ornithomimus velox itself. And don't forget birds like Avisaurus and Potamornis, which should count if we're counting Avimimus and such.

    I was unaware schlessmani was a valid species now. Reference?

  10. @Mickey Thanks for the update! I wasn't aware of the current status of the velociraptorine/Saurornitholestes teeth, and Avimimus in NA is news to me... How well separated are the various non-S. sedens ornithomimids? I didn't include any since the original authors didn't, for some reason.

  11. Oh, and as for schlessmani, Denver fowler pointed out on the DML that this should be Denversaurus given the younger age compared to the older Campanian/early Maastrichtian Edmontonia species. The type of E. rugosidens is also Judithian so Edmontonia or Denversaurus schlessmani should be the correct name.

  12. The "velociraptorine" teeth should still be on the list though, as Dromaeosauridae sp.. Also, the Paronychodon species is P. caperatus.

    As for the ornithomimids, in my phylogenies sedens is close to Ornithomimus (certainly closer than to Struthiomimus), and "Orcomimus" is somewhere in that same group. Dromiceiomimus is more basal though. But the phylogeny I found is different from Kobayashi's.

  13. Ick. I cringed when I saw the first lists. To list only one example, Alamosaurus and Struthiomimus. The American Southwest and Alberta are worlds apart, both literally and figuratively. Sure, they MIGHT have been widespread---but then they're forgetting about the individual microenvironments that so many genuses or species would live in. A little extra moisture or slightly different vegetation and poof, you might not be suited for the system. Just because they're found on the same continent doesn't mean they lived together.

    ---Crimsonraptor from Wikipedia

  14. Matt, you wrote:

    "Oh, and as for schlessmani, Denver fowler pointed out on the DML that this should be Denversaurus given the younger age compared to the older Campanian/early Maastrichtian Edmontonia species. The type of E. rugosidens is also Judithian so Edmontonia or Denversaurus schlessmani should be the correct name."

    Pardon me, but shouldn't distinguishing taxa be made on a morphological or genetic basis, not whatever formation they may or may not be from? This is the same as the stratigraphic argument (that similar taxa in a single formation should be subsumed) only in reverse.

  15. @qilong:
    In the case of a 10Ma+ difference in age, I think keeping Denversaurus in Edmontonia (or E. schlessmani is E. rugosidens) is just begging to turn Edmontonia into a Euoplocephalus-style wastebin taxon. One of my points in this article is that biostratigraphy is not usually considered, in diagnoses or otherwise, but it probably should be, if we're trying to understand ecology rather than just morphology. I don't know of any dinosaur species that exist over so many multiple formations and tens of millions of years that aren't obviously overlumped. Lumping or splitting genera is fine and subjective--lumping or splitting species should strive to reflect some kind of ecomorphological or chronostratigraphic hypothesis, not just morphological ones. Otherwise, you get nonsense as including Parasaurolophus walkeri in a list of potential prey items for Tyrannosaurus rex.

  16. (Also, I'm not sure stratigraphy was Denver's only rational. I'm not up on ankylosaur literature enough to know what other differences there may be between schlessmani and rugosidens).

  17. Matt, the argument that Edmontonia could become a "wastebasket" taxon like Euoplocephalus could be is not necessarily relevant here. The issue, and the only thing that should matter, is that biostratigraphy or geography is not the only, nor the best, way to determine if two taxa overlap and are therefore synonymous. Likewise, the converse should be true. It is possible that Denversaurus is unique from Edmontonia, or as such that the species are separable if the "genera" are not. But this must be determined through the morphology of the specimens; even thye work separating taxa of Euoplocephalus is being done primarily based on extensive morphological sampling, NOT stratigraphy; it is merely that the stratigraphy shows a segregation of morphology as well, which supports the original hypothesis. We can get a "genus" in both, potentially, if the two species are morphologically identical.

    This is, of course, with the caveat that there is not also other means of distinguishing Bakker's specimen from the Alberta specimens; Bakker's reasons were argued to be erroneous based on Vickaryous' sampling, and further supported in Dinosauria 2nd ed by Vickeryous et al. at least as a range extension of the genus but as a unique species.

    Note that based on the premise of stratgraphic separation of species, we should favor Chasternbergia rugosidens apart from Edmontonia longiceps, and if Denversaurus schlessmanni differs from either or one, then perhaps it can be its own genus (it would actually be just as effective and communicative as a species of either other "genus," and probably my preference -- if unique from either).

  18. @qilong: I'd agree with that, and I don't mean to argue that stratigraphy alone should be used as a diagnostic character, only that it should be used as one valid character along with morphological ones, more often than it is nowadays. The problem I have is that, according to the current state of things and available data from taxa with large sample sizes, species don't usually span more than +/-3Ma chronologically. When taxa last that long, and where samples are good enough, species-level differences can almost always be found in the samples (Psittacosaurus, Iguanodon, Pteranodon, Tyrannosaurids, etc.). A good example of this is Ichthyornis: Clarke 2004 gives a figure of Ichthyornis size relative to time (a taxon spanning 10Ma+ which she assigns to a single species based on morphology), and distinct trends in size over that time are clear. The sample size isn't enormous but it's a fair indicator that over that huge span of time and a wide geographic area, Ichthyornis was almost certainly represented by several distinct species (possibly forming an anagenetic lineage), but they are not split because stratigraphy is not considered appropriate to include in a diagnosis. I think in cases like this where data can support such a thing, there's no reason not to split the taxon based partly on stratigraphy. If later data reverses this, shows that they're all referable to I. dispar and that taxon was among the single most successful, long-lasting species of the Mesozoic, then simply lump them back together.