Derek Turner writes . . .
Pisanosaurus
Dinosaur Phylogeny: The State of Play
Last spring, there was big news in dinosaur paleontology. Matthew Baron, David Norman, and Paul Barrett—I’ll refer to them collectively as BNB—published a paper challenging the traditional classification of dinosaurs as saurischian (= “lizard hipped”) vs. ornithischian (= “bird-hipped”).[1] See here for an earlier effort of mine to think through some of the implications. BNB argued that it’s a mistake to fixate overly much on hip morphology. We should look at lots of different characters. And when we do, we get the result that the theropods were more closely related to the ornithischian dinosaurs than they were to the sauropods. So BNB proposed a new monophyletic group, the ornithoscelidans (= “bird-limbed”), that would include all the traditional ornithischians (your ceratopsians, ankylosaurs, stegosaurs, and hadrosaurs) as well as the theropods. Meanwhile the sauropods get stuck off on their own with the herrerasaurs. BNB thus challenged some pretty deeply entrenched assumptions about dinosaur evolution and taxonomy, assumptions reflected in countless dinosaur books and museum exhibits.
Some critique was inevitable. Last week, Nature published a critical response by a group headed up by Max Langer—I’ll call them LG, for “Langer group”.[2] Nature also published BNB’s reply to the critique by LG.[3] (Here is an accessible report on the two papers.)
What were the problems that LG claimed to find with BNB’s paper? LG write:
“Our main concern is that the authors [i.e. BNB] were able to personally study fewer than half of the taxa in their analysis; the others were scored mostly based on published literature, which is problematic, because many characters relate to fine anatomical details, requiring first-hand study to be reliably documented” (p. E1)
LG systematically re-checked all the fossils that BNB had used in their analysis. But LG studied all of them in person, and re-scored all the characters. When they did that, they found some support (though as we'll see, not much) for the traditional tree that placed the theropods together with the sauropodomorphs in the saurischian group. They write: “Character scoring changes explain our different results” (p. E1).
In their reply to LG, also published last week in Nature, BNB say that “we disagree with many of the re-scorings” (p. E4). In particular, BNB zero in on one specimen, Pisanosaurus.
What if it Weren’t About Dinosaurs?
But before we look at Pisanosaurus, it’s worth pausing to consider whether this research would be in the news if it were not about dinosaurs. Would it get published in Nature? Scientists revise phylogenies all the time. And usually, those phylogenetic revisions are merely a matter of technical, professional interest. If researchers proposed a major phylogenetic revision of a different group—salamanders, say—how much attention would that get? How much should it get?
Spotted salamanders have evolutionary history, too.
The broader cultural context here is crucial to the story: countless kids’ books and museum exhibits reinforce the ornithischian/saurischian taxonomy. It’s the classic example of the sort of “sciencey” thing that an amateur—even a kid—can learn about dinosaurs. Understandably, when BNB published their paper last spring, much of the public discussion focused on the question of whether and when we’ll need to rewrite all the dinosaur books.
The Most Important Dinosaur No One Has Ever Heard Of Might Not Really Be a Dinosaur
Pisanosaurus is a modest (only 1m long) animal from the late Triassic of Argentina, around 220 million years ago, but it plays an outsized role in the dinosaur phylogeny debate. Its classification has long been a matter of dispute. Is it a really early ornithischian dinosaur? Or is it not a dinosaur at all, but rather a silesaurid? The silesaurids were small Triassic reptiles thought by some to be a sister group of the early dinosaurs. The kicker is that Pisanosaurus is known from just one skeleton, found in Argentina in the 1960s. And that skeleton is partial and challenging to interpret.
Silesaurus | Not a dinosaur
BNB write:
“Re-scoring of Pisanosaurus alone, based upon our personal observations of the material, results not only in the recovery of Ornithoscelida, but also in the identificiation of this enigmatic taxon as a silesaurid” (p. E4).
They mean that they remeasured the fossils that LG had already remeasured. LG and BNB disagree about how to describe the skeletal morphology of the lone Pisanosaurus specimen, and the two teams are scoring characters differently. Score the characters one way, and you get the old saurischian/ornithischian division with Pisanosaurus as a really early ornithischian. Score them a different way, and Pisanosaurus comes out looking like a Silesaurid, hence not a dinosaur at all. And when you classify it as a non-dinosaur, BNB’s ornithoscelidan group comes back into focus. Alas, poor Pisanosaurus is unable to tell us how to measure its skeleton properly.
We have just one specimen of Pisanosaurus, and the remains are incomplete. For example, though we have a lower jaw, we don't have the complete skull. Also, good luck trying to figure out whether that is a lizard-like hip or a bird-like hip.
It’s impossible for an outsider to tell which researchers are doing the better job scoring characters. And what exactly would count as a better job? Here are BNB again:
“Moreover, we disagree with many of the re-scorings suggested by Langer et al. [LG]. For example, their re-scored Pisanosaurus includes character scores that are impossible to observe, such as ratios between skull length and body length, and skull length and femur length; the only known specimen of Pisanosaurus does not preserve a complete skull, axial column, or femur” (p. E4).
How can you measure the ratio of skull length to femur length when you have neither a complete skull nor an intact femur? Much of LG’s critique of BNB depends on remeasuring specimens and re-scoring characters, but how can you score characters for which you don’t have direct evidence? The answer can only be that you have to estimate. For example, you might be able to estimate skull length indirectly by measuring some part of the skull, such as the lower jaw, and using background information about how that part correlates with skull length in other groups. Perhaps the suggestion is just that LG’s rescoring involves some triangulation which introduces uncertainties.
The general point is that many decisions—about which characters to look at and how to score them—have to get made before the phylogenetic analysis can even get started. And different connoisseurs might not always agree about how to score characters. Statistical techniques give scientists powerful ways of testing ideas about evolutionary relationships, but before you turn the statistical crank, you have to figure out what to feed into the machine.
Yes, We Should Rewrite the Dinosaur Books
Although LG claim that their remeasurement of Pisanosaurus (and re-checking BNB’s character coding more generally) recovers the traditional ornithischian/saurischian tree, the support for the traditional tree is weak:
“We did not recover the Ornithoscelida of Baron et al., but the more traditional saurischian – ornithischian dichotomy that we did recover is weakly supported. It seems that the flood of new discoveries over the past decades has revealed unexpected complexity” (p. E1).
In other words, the would-be defenders of the traditional ornithischian/saurischian classification are freely admitting that the evidence for that traditional view is not very good. The ornithischian/saurischian view enjoys a strong incumbent advantage. But what if that view were not so well established? What if we were starting from scratch, taking our very first look at dinosaur phylogeny? In that case, it’s not clear at all that we would want to enshrine the ornithischian/saurischian dichotomy in books and exhibits.
The lesson I take from the LG vs. BNB debate is just that dinosaur phylogeny is way more confusing than most people ever appreciated. The Triassic fossil record isn’t so great. Some crucial specimens, like Pisanosaurus, are ambiguous and hard to interpret. Back in the Triassic, there were lots of small dinosaurish animals running around, but the morphological differences that we associate with later dinosaurs hadn’t arisen yet. The earliest ceratopsians would have looked quite a bit like the earliest sauropodomorphs. The early dinosaurs also weren’t all that different from closely related groups, like the Silesaurids. The evolutionary relationships among those early dinosaur groups and their relatives are really hard to work out. So, yes, we should redo the dinosaur books and museum exhibits to reflect this complexity, not least of all because the confusingness makes paleontology more fun.
[1] Baron, M.G. Norman, D.B., & Barrett, P.M. (2017), “A new hypothesis of dinosaur relationships and early dinosaur evolution,” Nature 543: 501-506.
[2] Langer, M.C., Ezcurra, M.D., Rauhut, O.W.M.,Benton, M.J., Knoll, F., McPhee, B.W., Novas, F.E., Pol, D., & Brusatte, S.L. (2017), “Untangling the dinosaur family tree,” Nature 551: E1-E3.
[3] Baron, M.G. Norman, D.B., & Barrett, P.M. (2017), “Barrett et al. reply,” Nature 551: E4-E5.