Posts Tagged 'systematics'

Guest Post: A phylogeny of ankylosaurian dinosaurs

At least a few parts of the dinosaurian family tree are still relatively little studied and are controversial. One of these is the ankylosaurian dinosaurs, though a new paper seeks to delve into this problem and produced a new and comprehensive analysis. Lead author Rick Thompson takes us through it:

In 1842 Richard Owen coined the term “dinosaur”, helping to imbue the world with a fascination for the gigantic ‘terrible lizards’ of the prehistoric world. Owen’s work was based on the fossil remains of three distinct archosaurs; Megalosaurus and Iguanadon have become fixed in our consciousness as archetypal dinosaurs, while the third, Hylaeosaurus, is very much the ‘ugly-duckling’ of the trio. Hylaeosaurus is a member of the Ankylosauria, a group of Middle Jurassic to the Late Cretaceous heavily armoured herbivorous dinosaurs, closely related to the stegosaurs in the clade Thyreophora. The story of Hylaeosaurus somewhat mirrors that of its ankylosaur relatives; fragmentary, difficult to study and lacking the popularity of its theropod, ornithopod and sauropod cousins. The ankylosaurs barely even worked their way into Jurassic Park (apparently Ankylosaurus does appear in Jurassic Park III – if you can stand it…)! The lasting image of the Ankylosauria in the public consciousness is the classic battle between Ankylosaurus and Tyrannosaurus, with the ankylosaur determinedly swinging its tail club towards the leg of its powerful foe. Indeed, it is the presence of the tail club, widely regarded as a defensive weapon that could shatter bone (Arbour et al. 2009), which has helped to raise the profile of this fascinating dinosaur group.

Of course ankylosaurs are more than just a dinosaur with a weaponized tail. The Ankylosauria includes over 50 valid taxa, only a fraction of which possess this idiosyncratic armament. Fragmentary ankylosaur specimens have been recovered from every continent on Earth, though their record in North America and Asia is by far the best. It is their extensive covering of dermal armour, particularly that of the skull, that has proved to be both the blessing and bane of ankylosaur palaeontology. The high level of morphological diversity in the armour is probably responsible for the large number of ankylosaur taxa. The cranial ossifications completely cover the bony sutures of the skull, simultaneously providing a unique pattern to identify new species, while destroying a wealth of cranial characters that could help to resolve the interrelationships of the wider clade. Thus the evolution of the Ankylosauria is very poorly understood, and presents a tough challenge for systematic palaeontologists.

Traditionally the Ankylosauria is split into two families, the Ankylosauridae and Nodosauridae. The Ankylosauridae includes many of the more recognisable taxa including Ankylosaurus and Euoplocephalus. Ankylosaurids are generally identified by their short-broad skulls, which are ornamented with horn-like ossifications above (supraorbital), behind (squamosal) and below (quadratojugal) the eye, along with the presence of a tail club. The Nodosauridae generally have a more elongate skull, with a domed skull roof (parietal region) and boss-like rather than horn-like cranial ossifications. Their trunk armour often incorporates spike-like ossifications along their flanks, though no tail club is present. In recent years, a number of ankylosaurs, such Gastonia and the aptly named Gargoyleosaurus, have been found to possess combinations of these classic traits, thus blurring the distinction between ankylosaurids and nodosaurids. This has led some authors to erect a new group of ankylosaurs, the Polacanthidae, which includes many of these recent discoveries, along with a few older, fragmentary genera like Owen’s Hylaeosaurus.


A mounted skeleton of Gastonia burgei. Note the large shield over the pelvis, a character suggested to unite polacanthid taxa. Photo by Susie Maidment.

The validity of the polacanthid hypothesis has proved hard to test cladistically, as the majority of ankylosaur character sets are either heavily biased towards the skull, or include a limited number of taxa. As many of the potential polacanthid taxa are fragmentary, or predominantly postcranial, the incorporation of these taxa into ankylosaur phylogeny has been problematic. Studies including some polacanthid genera place them either towards the root of the Ankylosauridae, or as a new family (Polacanthidae), sister to the Ankylosauridae. During my Master’s Degree in Biosystematics at Imperial College, London, I was lucky enough to get the opportunity to try to resolve some of these issues in ankylosaur evolution. Paul Barrett and Susie Maidment offered me the chance to revise and update the unpublished PhD thesis of Jolyon Parish on ankylosaur phylogeny. My aim was to expand the taxon sample of Parish’s thesis, as well as modify his extensive character set. In this way I hoped to produce a ‘palaeontological total evidence’ phylogeny of the Ankylosauria, which included all valid taxa regardless of completeness, and a roughly even mix of cranial and postcranial characters.

The results of this project were recently published in the Journal of Systematic Palaeontology, and give some new insights into ankylosaur evolution. Encouragingly our data did support the traditional split between the nodosaurids and ankylosaurids; however, no polacanthid clade was recovered. Instead, all taxa that have been affiliated with the Polacanthidae by various authors (and Ankylosauridae by others) were resolved within the Nodosauridae. In the strict consensus tree (the tree which summarizes all of the information that each of the shortest possible evolutionary trees agree upon) the nodosaurid clade formed a polytomy – that is, we could say nothing about the internal relationships of the group. This poor resolution of the clade was in fact caused by a number of unstable taxa, whose position on the tree is highly variable. Unsurprisingly three of these unstable taxa have previously been attributed to the Polacanthidae. To counteract their influence we produced a derivative strict reduced consensus tree (shown in the figure). This tree prunes unstable taxa from the strict consensus tree, greatly increasing resolution. In this phylogeny it is clear that the ‘polacanthid’ taxa form a basal grade of nodosaurid dinosaurs, while the traditional nodosaurids (clade D) form a polytomy.

This result simultaneously highlights the strengths and weaknesses of the ‘palaeontological total evidence’ approach. By including all taxa (regardless of completeness) and a character set sampled from the whole skeleton, a large amount of missing data was guaranteed. This lowers the resolution of the tree, as seen in the traditional nodosaurid clade. However, it is the inclusion of such a broad sample of characters and taxa that has allowed completely new relationships to be revealed, placing ‘polacanthid’ taxa in the Nodosauridae. The problems have been further confounded by the extensive cranial ossification of the taxa. The variation in such complex ornamentation is exceptionally hard to capture using traditional discrete characters. This makes it very hard to test for presence of the evolutionary signal in the ornamentation. In the future, a study which incorporates extensive measurements of the cranial armour could better capture this variation, and more clearly reveal the relationships of the group. This is particularly important for the Nodosauridae, many of which are primarily distinguished by their cranial ornamentation.

The phylogeny of the Ankylosauridae was much better resolved, and broadly in agreement with existing studies, if the ‘polacanthid’ taxa are discounted. Again in this clade we see that the basal lineages do not conform to the traditional view of the Ankylosauridae, often having elongate skulls. The tip of the tail has not been recovered in many of these taxa, so the presence of a tail club is uncertain. However, our tree is the first cladistic analysis to place an ankylosaur that clearly lacks a tail club within the Ankylosauridae. Zhongyaunsaurus was originally described as a nodosaurid, though this assignment was subsequently corrected by Ken Carpenter (Carpenter et al. 2008). Our phylogeny confirms that Zhongyaunsaurus was indeed an ankylosaurid, and suggests that the most characteristic trait of the Ankylosauria is only present in the most derived ankylosaurids.

Unfortunately our study has done little to help the plight of the earliest known ankylosaur, Hylaeosaurus. Our analysis suggests that it is a member of the Nodosauridae, but its removal from the reduced consensus tree feels like a sad continuation of its rather anonymous history. Never the less, our study has helped to reveal a new perspective on the Ankylosauria. Although the ankylosaurid-nodosaurid dichotomy has been maintained, it appears that the classical characters of these groups only apply to the most derived of their forms. The earliest lineages of each family show greater levels of diversity in their armour and body form. This has made their classification difficult, and without more fossils, or newer forms of character, it is likely to remain so. There is still much work to do in order to understand the evolution of this extraordinary group of animals, but hopefully our study can serve to trigger a new wave of research in the coming years.



Taxonomic practice and publications

Thanks to some reviews I have been writing, papers I’m working on, conversations with colleagues and a recent blog post, I’ve been thinking much about the practices of myself and my colleagues when it comes to taxonomy. Obviously I’ve written about the how-s and what-s to a degree in the past, but this is a little more specific.

The main point I’d make (as a referee / editor and recipient of referee reports) is one roughly in the line of Voltaire. I disagree with your taxonomy, but I defend your right to publish it.

Now obviously this has limits. You’re specimen must be genuinely diagnostic. You’ll have to use consistent characters. They must be well defined and not vary ontogenetically or be subject to excessive intraspecific variation (or be so massive as to not make this an issue). Or if you are trying to synonymise a bunch of taxa then you’ll need to show that other diagnoses were flawed or didn’t stick to these rules etc.

However, I may think you are overly splitting or lumping something, but I still think you should be allowed to publish (as indeed, so should I). There really are no hard or fast rules as to exactly what constitutes a genus or species. Exactly how many characters you need for one or the other, or whether some are better than others. Taxonomy really does work by consensus and the starting point of any discussion will be the publication of a new taxon / synonymy of an old one. Not allowing such a paper actively inhibits discussion / research.

A paper that on balance few people agree with will soon fall by the wayside. But it’s mere existence will allow a greater depth of discussion by getting people to examine and evaluate the characters at hand and compare them more thoroughly. So let these be published. The authors get a paper, the journal gets a paper, and the worst thing to happen is that a few people quibble about it, everyone gets some citations and we move on, but with a better understanding of the issues.

Here is something I wrote in a review of a paper that intended to name a new species (and has now been published):

I actually tend towards ‘splitting’ over ‘lumping’ myself, and I would not wish to prevent the author erecting a new taxon here if he feels it is necessary, but I would say that I do not think it is required and in the same position I would not erect one myself…. The author does effectively concede all of these points in his text, and clearly still feels the erection of a new taxon is necessary and appropriate and I am happy for him to make that call from a superior position of knowledge, but I do offer a contrary position.

This is a position I wish far more people would take. Make your points and give your opinion, but let the author make their decision and allow for the fact that they have been putting more time and effort into this that you have and know the material better.

On a similar note, I notice a tendency by some (admittedly often online rather than necessarily with respect to the literature, either published or at the review level) to second guess people working on specimens. Now for sure, people with detailed anatomical knowledge of specimens can see certain errors that others will miss (or that at least need to be verified more carefully) based on photos or drawings or descriptions alone. There is nothing wrong at all with making that clear. But I have seen people in the past second-guess authors based on things that haven’t seen. It is most frustrating to be told by someone 5 thousand miles away that you have got something wrong when you have the material in your hands and have been looking at it for months and all they have is a black and white photo of it. (And on that note, there’s a huge difference between asking someone to recheck and telling them it’s wrong. Ultimately the message is the same but the first is polite and the second isn’t. And of course you risk looking very foolish if you are wrong).

This is something raised by Larry Witmer in his excellent recent blog post about the tiny Tarbosaurus. His team commented extensively on the potential taxonomic issues of their work and explicitly which specimens / taxa this might effect and even how. But they also pointedly didn’t make any revisions. With none of them having looked in detail at any of the key specimens, they felt it unwise (and I would suggest, even impolite, impertinent, or maybe even unprofessional) to have done so. I wholeheartedly agree (as you probably guessed form the contents of those parentheses).

Yes there are exceptions to every rule, but as with the above point, I think you have to be careful before messing around with taxonomy and however much you disagree, respect the interpretations and work of your colleagues. If there are huge and obvious errors, then point them out do. But I’d avoid any kind of formal synonymy without having seen critical material first hand. There is, after all, nothing quite like seeing a fossil.

An arrangement of archosaurs

As I’m sure many readers are aware that Sterling Nesbitt has just published a monumental chunk of his PhD thesis as a major monograph on archosaur phylogenetics. While obviously new phylogenetic analyses are constantly being published, these tend to rehash and recycle existing work heavily (and that’s not necessarily a bad thing) and tend to be limited in scope. Very big, detailed, comprehensive analyses are incredibly rare since they require a huge amount of time to be applied to the problem. You have to have a really good understanding of anatomy, the historical aspects of taxonomy and phylogeny of the groups at hand, go over all the existing datasets, see tons of material, and then analyse and reanalyse and write it all up. To do all of that for a group as vast and complex as the archosaurs is a truly major achievement and he’s done it in incredible depth.

Few of the results are perhaps really even remotely surprising, but that’s by the by. That gives us great confidence in our other, smaller, studies and tells us we were on the right track. More importantly, Sterling has thrashed out (as far as I can tell, it’s 300 pages long! and I’ve not read a fraction of it in detail yet) some of the more poorly defined characters and redefiend them and their states and then as a result of his work, redefined every major node on the archosaur tree.

This is the real boon for me, making explicit statements about character evolution across the tree and allowing us to say more about what changed when and how and in conjunction with what else. The amount of characters too (over 400) also means that some important nodes now have a lot more support than they used to.

Archosaur phylogeny. From Nesbitt, 2011.

And the rest of it. From Nesbitt, 2011.

Obviously I’m especially interested in pterosaur origins, not least through having looked at some of the characters that tie them to the archosaurs with Sterling just this year. Some of these are disucssed and assessed here, but overall what we see is a much stronger support for an ornithodiran (that is pterosaurs + the dinosaursomorphs) position of pterosaurs than ever before.

Sterling recovers some 12 characters directly the supporting ornithodia, (with perhaps 13 more depending on issues of transformation that I won’t go into here). That is, in the context of something this big, really quite a lot. In fact it’s as many characters as he finds to support the dinosaurs (also 12) and many more than supports the dinosauromorphs (just 5). In short, what has always been a very poorly supported node is, in this analysis, an incredibly well supported node.

There are of course a few little caveats to this in that this is non a dedicated analysis of the pterosaurs and their relationships, and despite the huge size of this work, there are some more detailed and specialised characters that can be added in and there will doubtless be quibbles about exact codings and taxa. I can see a few things I would explore if I was to go and do this. That’s not to denigrate his work one iota (after all this isn’t what he’s trying to do!), I’m merely saying that IF you were to use this as the basis of pterosaur origins, I think there’s more that should be done. However, this is very robust indeed as nodes go, and I think at the very least it adds a lot of support to this hypothesis where previously it had been rather less clear cut.

The only thing I can really say to conclude is congratulations to Sterling and thank you. This is a piece of work which will aid almost every archosaur worker for, quite probably, decades. That is a real achievement.

Traditional taxonomic rankings

Everyone on here is probably familiar with the old taxonomic ranks of kingdom, phylum, class, order, family, genus and species that was a mainstay of school biology classes and of course was a fundamental part of taxonomic work for, well, pretty much centuries. However, with the advent of cladistics more and more specific groupings of taxa were possible based on branching nature of phylogenetic trees. While there were sub-ranks of subfamilies, superorders, tribes and the like, these were rarely invoked and I think people shied away from them because they implied a level of detail that couldn’t really be inferred based on how these groups were generally formalised.

However, many of the intermediate ranks are rapidly falling by the wayside because they simply are no longer useful or don’t make sense. Kingdoms and phyla are still useful and major groups of taxa that represent fundamental divisions, and of course genera and species are the basis of all taxonomy, but that stuff in the middle? Doesn’t fit.

To take an excellent example (floated my way by Mike Taylor, despite the fact that it should have been obvious to me) birds are in Class Aves, we now accept that birds are dinosaurs but more specifically that means they sit within the traditional Superorder Dinosauria, Order Saurischia, Suborder Maniraptora and (depending on your favoured phylogeny / family definition) Family Paraves. So a Class ends up sitting in half a dozen groups that are supposed to be subordinate to it. And all the various orders and families of birds also sit in there. Oh.

We are increasingly getting close to building the Tree of Life – a single vast and unified trees that put whole phyla together at the species level in clades that most experts generally agree on. No longer are new or odd species somewhat arbitrarily assigned to families or have new families erected for them because they are special. We simply don’t think of trees and relationships in those terms anymore – not everything has to sit in a family, and it’s no longer the case that just a couple of researchers look at their clade and split up the taxa with no reference to related clades or other, wider patterns. As such, while at a conversational level, it still makes sense to use and discuss things like the ‘cat family’ or ‘dog family’ increasingly these are being abandoned for their formal names (Felidae and Canidae in this case) though the endings of various names betray their origins as the –idea, -inae, -oidea, and so on demonstrate. These now represent successive ranks rather than ‘families’ and ‘tribes’ as once they would.

Loss and systematics

Having covered both character loss and (supposedly) irreversible characters before, there’s only one obvious issue less to cover about these kinds of characters. And that is the effect such losses have on systematic relationships, or perhaps rather, how people can perceive them. It may seem intuitively obvious to readers here (because most of them here clearly know their biology) but just because whales do not have fur does not make them not mammals and just because penguins can’t fly does not make them not birds. However, if you are not aware of how such relationships are assessed or what these characters mean and how evolution works.

The most obvious point is that at least some of these characters are not really used by biologists. ‘Flight’ is not really something you can use in itself since it’s as huge combination of anatomical and physiological characteristics so it’s a poor choice. Secondly, losing a couple of characters is normal over the course of an organisms’ evolutionary history – snakes have lost their legs, tyrannosaurs lost a few fingers, kiwis got rid of their arms and so on. This doesn’t change the rest of them, and it’s really very common.

Finally, as an extension of the second point, single characters don’t really count for much when it comes to assessing relationships. As I have stressed before (and indeed as others have before me) details matter and you have to take into account all of the available evidence. Whales are still mammals because despite the loss of the hindlimbs and hair, they still have mammary glands, a mammalian inner ear and various other characters. Penguins can’t fly, but they do have air sacs, beaks, feathers and more. To get fixated on a single character is to assume that the others are either not important or less important and to ignore the other evidence.

Things change over time and some animals have become modified to quite a profound degree from the ancestral condition, but this is ultimately what is important. Ancestrally mammals had hair and this remains a useful character to define the clade even if some don’t have it anymore. Snakes, though they may not look it, are still tetrapods.

Relative relatives

The subject of relatives has been bouncing around my brain of late. The statements, ‘the gorilla, a relative of modern humans’ and ‘the salmon, a relative of modern humans’ are both true, but one is perhaps more true than the other. Obviously all things alive today are relatives at some very deep and very ancient level. As far as we know and can tell, life only originated once on Earth and if you go back far enough fungi, trees, people, dinosaurs, bacteria and viruses have some form of common ancestor somewhere, not to mention all the things in between. However, there are of course, relatives and then there are relatives – relativity is all relative.

It’s clear that humans are more closely related to gorillas than they are to deer, closer to deer than to salmon, and closer to salmon than spiders, but they are still all related. Move in though, or move to an unfamiliar group and that becomes trickier – are beetles closer to flies or mantids, are spiders closer to weevils or centipedes? Unless you have a pretty good grasp of the detailed phylogeny to hand or some clear qualifiers any statements about relationships rapidly, or even instantly, become fuzzy to the point of being irrelevant.

You do see this at its worst in the media (big surprise) but I am sympathetic here. When trying to communicate the idea that some groups or species are closely related to others, aimed at an uninformed audience and avoiding technicalities and illustrations it becomes very tricky very fast. =I’ve also seen a fair amount of tired internet discussions along the lines of ‘Velociraptor isn’t related to Tyrannosaurs at all’. OK the ‘at all’ we can allow as conversational hyperbole, and of course tyrannosaurs and dromaeosaurs (within the theropods) are quite well separated, but they are also related. Compared to say, Dilophosaurus, they are quite close, compared to a crocodile, very close, compared to Troodon, not close. That statement is as true as it is false.

But is it really so hard to use a qualifier or two? Something like ‘Linheraptor is not a particularly close relative of Velociraptor’ is potentially confusing and in a sense no more accurate than ‘Linheraptor is a particularly close relative of Velociraptor’ – just how ‘not close’ or ‘close’ is it? But add just a few extra words and the distinction is obvious, accurate, succinct and informative: ‘Linheraptor belongs to a small group of dinosaurs called dromaeosaurs that includes Velociraptor, but within the group the two are not especially close relatives”. There’s really quite a lot of information in that 20-odd word sentence. It puts the two in context with each other, and the wider group of dinosaurs or theropods as a whole (since this sentence should obviously not stand in isolation in an article). So stick in a few qualifiers and make things more clear, it really will help and it might stop a few discussions were both people are right but arguing pointlessly from different perspectives.

How to read a phylogenetic tree

Nowadays even the media seem quite happy to occasionally put up a phylogenetic tree as part of their scientific coverage, and they are proliferating on the internet on websites, research papers and blogs, in addition to books and magazines. However, while it is hardly difficult to get the gist of a tree, there is a certain skill and amount of knowledge that needs to go into pulling out all of the information correctly from a tree. It is easy to make mistakes about what a tree actually tells you so hopefully I can clear up a few misconceptions about tree creation and how trees should be read.

Continue reading ‘How to read a phylogenetic tree’


Time for another ‘science basics‘ post and this time out I’m tackling the often complex and misunderstood field of taxonomy, quite simply the naming of things. It is quite possibly the most fundamental and important part of biology as a field and yet seems to be glossed over in the few undergraduate courses that even bother to mention it, which is both a shame and a worry. Why is it so fundamental? Well for the quite simple reason that if you do not know what species any given organism belongs to, then it becomes very hard to say anything meaningful about it. How do you protect a given species if you don’t even know what is and what is not a member of that species? Want to treat a snake bite? What species was it? Got a new drug from a rare frog, great! What frog was it? Are you working on a single species of a bunch of them in your lab that have just not been revised properly? You can see the issues – there is a basic one of clarity (all scientists want to be able to communicate clearly about what they are dealing with) but the consequences of making sure that is the case go far beyond that and touch every branch of biology.

Continue reading ‘Taxonomy’

@Dave_Hone on Twitter


Enter your email address to follow this blog and receive notifications of new posts by email.

Join 574 other followers