Thoughts on Ornithoscelida … over one year on … (part 1)

March 23rd 2017 saw the publication of an article in the journal Nature in which David Norman, Paul Barrett and I proposed an alternate tree topology within Dinosauria, placing together, for the first time, Theropoda (most meat eating dinosaurs) and Ornithischia.

This new clade was named Ornithoscelida, which roughly translates as ‘bird-limbed’ or ‘bird-flanked’.

For one, this new topology was a little bit unexpected, partly because the ‘traditional’ model always placed Theropoda and Sauropodomorpha closer together and, since time immemorial, Ornithischia has always been considered as that weird ‘other group’ of dinosaurs that don’t really look like their saurischian cousins.

 

Now, more than one year later, I am still finding myself, semi-frequently, the recipient of calls from science journalists who want to cover the research. One of the most common questions I get is ‘how do we solve the current controversy about the base of the dinosaur tree that your work has generated?’

The controversy, while not quite “tearing palaeontology in two“, as some articles may have claimed, did briefly occupy centre stage in dinosaur research. It arose following the publication of a reply by Langer et al. and our rebuttal, both also in Nature.

Not to get into the nitty-gritty too much, but the key points that were raised were largely based on character (anatomical features) interpretation and scoring. What this means is that, between the two research groups, there were discrepancies between what features were said to present in what specimens. As a result, the two groups produced two different trees (albeit, ones not statistically significantly different from one another).

These differences arose for a number of reasons, but the mains ones that I have been able to identify are:

1. The definitions of terms in the characters themselves — anatomical characters are described usually along the lines of “Feature A: 0, absent; 1, present!” and we, as scientists, when observing the absence of the feature A in Specimen X, place a little (0) in the appropriate box in our data matrix: when observing the clear presence of Feature A in Specimen Y, we place a little (1) in the box. Eventually, if we do this for all features/characters for all specimens, we end up with a big excel style spreadsheet of 0s and 1s (but mainly ?s, as the fossil record is not complete). However, here in  lies the first problem with different research teams coding the same matrix – there will be some differences in what one team understand by A and B and what the other team understand by A and B.
2. The expression of the characters in the specimens — if both teams do have the same understanding what exactly Features A and B are, there is then still no guarantee that they agree in which specimens you see it and which you don’t.
3. The validity of characters — while in their reply, Langer et al. 2017 did use the same character list as we did, they made a point of remarking that they felt that some of the characters might not be independent of each other and, therefore, not valid. For example, if having Feature A automatically makes a specimen have Feature B as well, and that no specimen with Feature B does not have Feature A, we cannot say that these features are independent, and that by including both, the data analyses would just be counting the feature twice. While we went to great lengths to try and make sure all of our characters were independent of each other, the fact that Langer et al, disagrees further highlights how there must be some fundamental disagreements over what exactly it is each character is describing.
4. Taxa inclusion — finally, Langer et al. highlighted that not every early dinosaur that could have been included was included in the original analysis. This is true, although it has to also be said that what we produced was still by far the largest and most broadly sampled dataset of early dinosaurs ever assembled. A few extra taxa were thrown in by Langer et al. to expand the data, including the very unusual carnivore Daemonosaurus and the early stegosaur-ancestor Scelidosaurus. It made no difference to the outcome, but it was worth testing.

 

So, how do we fix these?

Well, first of all, there is the option of backing-and-forthing through the literature, as happened with another disagreement between myself and colleagues over a particular species called Chilesaurus.

What I hope happens is that an improved understanding of the specimens in question – new and improved descriptive publications of them, 3D models of them placed online, further prep work on some specimens etc. – will provide a lot of answers to the questions of ‘which group correctly scored Features A and B in Specimens X and Y. I am sure that in some instances we will be shown to have been right, and in others wrong.

Secondly, I think further communication between the two groups about why they each opted to score Features A and B the way they did for Specimens X and Y. More conversing between the research groups might just help clarify a handful of the remaining points of contention.

There is also the possibility of new discoveries being made that illuminate other, currently unknown, aspects of early dinosaur anatomy. Some have been made since our first publication, and these can be seen here and here.

Then, there is something that I can (and should and must) do, and that is, along with my co-authors, publish a comprehensive description of our characters complete with figures illustrating what each character state looks like in the relevant specimens. Its a big task, but I am working on it. Watch this space!

But for now, I will say this. While I am not certain that Ornithoscelida is the correct hypothesis for early dinosaur evolution, I am sure that the ‘traditional’ model, as first proposed by Seeley way back in 1887, in which all dinosaurs are either ornithischians or saurischians cannot be the truth of the matter either.

Whether or not Ornithoscelida is a simple dichotomy between Ornithischia and Theropoda, or something more complex, I cannot say yet. But my hunch, and its a strong one, is that Ornithischia and its place within the dinosaur tree, is perhaps not quite as simple as we have always assumed; they are not weird ‘other group’ outside of  the group of all other dinosaurs.

Why and what evidence I have to support my thinking will be the topic of the net few posts. So stay tuned!

 

by

Dr Matthew Grant Baron