Dinosaur diversity and a super supertree

A major new paper is out today on the vexed subject of dinosaur diversity and it is one that is of special importance for vertebrate palaeontology because I am one of the authors. It is also important for a bunch of other reasons but I won’t trouble you with them here. Oh, OK, if you insist…

The project was a major collaboration between various researchers at the University of Bristol and a couple of recent former members of the palaeo lab, namely Davide Pisani and myself. The intention was to look rather more closely at the issues of dinosaur diversity – could we detect when they were diversifying and expanding, and did the so-called Cretaceous Terrestrial Revolution (KTR) affect them?*

*The KTR is the massive change in terrestrial faunas that supposedly came about as a result of the evolution and subsequent radiation of angiosperms and includes major radiations by insects and mammals amongst others.

Despite my opening, anyone who reads the paper (or even just the full reference) will realise that I am a pretty minor player in all this, and my primary job was collecting and analysing various sauropod trees. At the time this work was started properly (we already had some vague discussions about it while I was based in Bristol more than three years ago now – some things take a long time to get to fruition) I was in Munich working on sauropod phylogenies, so it was an obvious task for me to get stuck into. Full credit must go to Graeme Lloyd who assembled (and a far harder task) marshalled his team to get the work done, it is no mean feat organising such a group of people to do such complex and time consuming tasks and have everyone getting the necessary data or analysis to the right people at the right time. Authorship lists do not always (if ever) reflect exactly who did how much of what work and from where I was sitting Graeme certainly deserves to be at least first and second author.

While the crux of this paper is about diversity, you need something to found your analysis on. What constitutes an increase in diversity? How do you measure that? The simple answer is with a phylogeny, which then immediately begs the question, which phylogeny do you use? How do you pick between the various competing hypotheses and trees of the evolution of various dinosaur groups? And how do you allow for the fact that most analyses focus on relationships within clades and not between them? How can you do this? Which trees do you use?

Well, we used all of them – some 155 cladistic trees by my count.*

* The number we actually checked is quite a lot higher – 240 – as we exclude or combine non-independent trees e.g. where someone has taken an existing dataset and simply added one new taxon and then republished it, plus those trees for which were not actually based on a cladistic analysis.

Subsampling of the data suggests that many if not all of these might simply be a result of collection bias (numbers of exposed areas, amount of time dedicated to digging, size of fossilerous beds etc.) and the curves are significantly flattened. There are many, large fossil beds known for the Late Cretaceous and they have been heavily studied, so it is perhaps no surprise they yield the most species.

You can actually put trees together using a series of methods to build what are called supertrees. Supertrees are not without their problems (pseudoreplication being the most obvious – some trees will feature the same underlying data many times over) but many of them can be minimised or circumvented. If you are careful and clever in what you do, you can assemble a supertree that gives you the best consensus of how science perceives dinosaur species to be related to each other. Some of you may know that this was tried before (waaay back in 2002 by another Bristol team then led by Davide and featuring Adam Yates) which featured some 277 genera, from 130 trees.

This time it has been ramped up significantly to some 420 species and actually the result is surprisingly well resolved (some 99.2% in fact). The early supertree suffered in its resolution as there were still some major contradictions and controversies at the time, notably in the relationships of the maniraptorans (where did alveresaurids go for example) and the sauropodomorphs. With subsequent analyses coming closer and closer to a consensus (and there being more and more of them), naturally the new supertree reflects this – more congruence means greater resolution, in addition to a fair few methodological tweaks and of course new taxa and analyses being included.

The new dino-supertree

The new dino-supertree

(Clicking on this will bring up the whole tree – warning, it’s big).

So now we have a tree and of course we have temporal data from the literature (when each species and thus each clade first appeared, and when they were around) we can turn our attention to diversity. Several analyses were done to account for different issues (e.g. ghost ranges) the primary one being a comparison of the tree shape (and of course branches) and an evolutionary model of diversification to see how they differed (i.e. where high rates of diversification was occurring). Plotted out this shows several large peaks where there was an apparently significant increase in diversity. The results show that actually the majority of diversity increases occurred in the Late Triassic and Early Jurassic and that these tended to coincide with the origins of major clades and not the Cretaceous when the KTR was taking place.

One might expect this, after all an initial radiation can potentially be explosive as a new clade expands to fill various ecological niches, but things do not always occur in this manner and all kinds of patterns are known (continual expansions, an initial radiation, then stasis, low levels of expansion followed by a late radiation etc.). Certainly there is an early increase for dinosaurs, but diversification continues to increase afterwards throughout their history, if less dramatically. Major new radiations (such as that of the derived maniraptoran theropods) do not have the impact one might expect when set against the background increase in diversity.

So did the dinosaurs take part in the KTR? While there clearly is a serious boost in diversity of dinosaurs in the Late Cretaceous, much of this can be attributed to sampling bias. More recent rocks preserve more fossils and are more easily accessed and their presence in large numbers in continental North America makes them a prime candidate for study. While the ceratopsians and hadrosaurs underwent significant increases in diversity in this period, there was no overall significant increase in diversity that could correspond to the KTR.

In short, the KTR was certainly going on, and dinosaurs diversified many times over the course of their history including during the Late Cretaceous, but the two were not linked – dinosaurs were not part of the terrestrial revolution. While there is evidence from stomach contents, coprolites and some anatomical shifts that at least some dinosaurs fed on angiosperms, this does not seem to have been a significant factor in their evolution and diversification. Hopefully, this study can also put to bed the idea that dinosaurs were already on the way out before the KT extinction, despite the sampling bias, the clade was in rude health and highly diverse when the curtain fell on their stage.

You can find more info here on the tree and the results and download the paper and copies of the tree too. We really are very good you know.

Lloyd, G. T., Davis, K. E., Pisani, D., Tarver, J. E., Ruta, M., Sakamoto, M., Hone, D. W. E., Jennings, R., and Benton, M. J. 2008. Dinosaurs and the Cretaceous Terrestrial Revolution. Proceedings of the Royal Society, Series B, in press. doi:10.1098/rspb.2008.0715

33 Responses to “Dinosaur diversity and a super supertree”

  1. 1 Andrea Cau 23/07/2008 at 6:31 pm

    What is the “Gugyedong maniraptoran” present in the supertree?

  2. 2 Will Baird 24/07/2008 at 1:14 am

    I didn’t see the heterodontosaurs…am I just blind?

  3. 3 Karl Zimmerman 24/07/2008 at 4:01 am

    Neat. I think you must have used an old version of the supertree though, as the phylogeny in detail looks substantially different for several groups when compared to the supertree in the PDF.

  4. 4 Jerry D. Harris 24/07/2008 at 5:55 am

    I didn’t see the heterodontosaurs…am I just blind?

    They’re there — wedged between the ankylosaurs and pachycephalosaurs.

    What is the “Gugyedong maniraptoran” present in the supertree?

    Probably the one described here:

    Kim, H. M., A. D. Gishlick, and T. Tsuihiji. 2005. The first non-avian maniraptoran skeletal remains from the Lower Cretaceous of Korea. Cretaceous Research 26:299-306. doi: 10.1016/j.cretres.2005.01.001″

  5. 5 Dave Godfrey 24/07/2008 at 6:08 am

    The heterodontosaurs are between the Ankylosaurs and the Pachycephalosaurs.

    It looks like in the Sauropods Haplocanthosaurus and Omeisaurus might need to be broken up. Is this something supported by the within-taxa studies, or an artefact of taking the broader view?

  6. 6 Karl Zimmerman 24/07/2008 at 8:31 am

    Not sure what happened to my older comment about how your jpg doesn’t match up with the published tree in the PDF, but I have another comment…

    Why did you guys choose to keep Eshanosaurus in the coding rather than throw it out as a nomen dubium? My understanding is it’s only a partial lower jaw, could be a prosauropod, and as you can see, really throws the maniraptoran tree (along with the nearest outgroups) all wacky with ghost lineages.

    When you take it out, along with Cryolophosaurus (which I believe has since been found to be Dilophosauridae, there is no evidence of Tetanurae before the Middle Jurassic. Of course, there must have been some ghost lineages in the latest part of the Early Jurassic, but this causes the Theropod tree through time to more closely resemble that of the Sauropods or Ornithischians – a LT/EJ burst, followed by a radiation of groups which survive to the latest K

  7. 7 Randy 24/07/2008 at 12:31 pm

    A couple of questions:

    1. I’m a bit surprised to see herrerasaurids and Eoraptor placed as theropods. Perhaps it is an artifact of the method used to make the supertree, but nearly all recent phylogenies place these taxa as basal saurischians outside of Theropoda + Sauropodomorpha (Eusaurischia). e.g., Langer 2004; Langer & Benton 2006; Ezcurra 2006; Yates 2007a,b; Irmis et al. 2007; Smith et al. 2007. The last major analysis to place these taxa as theropods was Rauhut 2003, and its my understanding that Oliver now thinks they are basal saurischians as well.

    2. Sampling bias may explain the explosion of Cretaceous taxa, but that doesn’t mean the diversity change isn’t real. The amount of exposed outcrop of terrestrial strata correlates with eustasy; has anyone looked at the possibility that sea level changes are driving dinosaur diversity in the Cretaceous?

  8. 8 Richard 24/07/2008 at 2:25 pm

    Congratulations Dave, this is an interesting study, particularly the resampling approach to diversity (which is independent of the supertree). I remain deeply skeptical about supertrees however – just looking at Ornithopoda briefly there are numerous relationships unsupported by any of the source trees (a Rhabdodontidae + Gasparinisaura clade anyone?) and this is a big problem because such results are spurious and may give misleading macroevolutionary signals.

  9. 9 David Hone 24/07/2008 at 8:05 pm

    I’m clearly late coming to my own party, but I’ll *try* and get through all of this:

    1. Andrea, it is a probable undescribed avialian, that featured in a paper by Kim et al. 2005.

    2. Karl, there are several versions of the tree floating around (the one I used is from the supplementary information) as there were various consensus methods used and few taxa got pruned in various versions to try and imporve resolution etc. I used this one simply becuase I have issues with files here and it was the only one I could convert to a jpeg easily.

    3. Jerry, thanks for answering that, rather makes 1 redundant! 🙂

    4. Karl again, comments have to be approved by me unless ou are logged in so it must have been stuck. I happened to be talking over Eshanosaurs the other day with Paul Barrett and Jim Clarke and everyone is happy t really is a therizinosaur. As for the others, direct your comments to, err, someone else. I really was just a data monkey on this and did not make any executive decisions on synonomy, even for the sauropods which I was ‘coding’. Manabu Sakamoto of the Raptor’s Nest can probably help as the theropod guy on this.

    5. Randy, yes it’s a function of tree building (and not using the ‘right’ one here). I take your point about diversity, yes it certianly is incresing, but just not at a big enough rate against background to mark it out above background levels. Diversity tends to increse over time anyway. I know there is at least one more study on the way about diversity, so perhaps the sealevel issues will be taken up there.

    6. Richard, thanks very much, but see above, I really didn’t do much beyond the original discussions and then data entry. I’m not being self decrecating, I didn’t do much, and nothing technical. I myself don’t like supertrees much (as will bcome clear when my big paper on them comes out shortly, TBC) and there are certainly (potentially) big issues with them. I consider them to have only a few very limited uses, within very limited terms, and this is one of them. You really can’t just ‘build a huge tree’ of all dinosaurs so currently this the only feasible method and it does allow us to do these kinds of analyses. True, some glitches appear, but then these are *largely* swamped out by the other signals, so it’s not all doom and gloom.

    Davide and Mark Wilkison (my former tutor and a peerless cladist for anyone who knows their cladistics) have written a bunch of papers on supertrees (primarily negative) and I rather stick with Mark’s tka (if I can paraphrase him correctly from memory) – they would be really good if we knew exacty how they were getting the results they produce. The fundamental algorightms are just not that well understood (though I’ve not read up on them for getting on 5 years now) and as a result we don’t really know why them produce imbalances in certain parts of the tree or why Hennigian combs get favoured over balanced trees. If we can get a handle on that, they could potentially be brilliant.

    One thing in their favour (potentially) is that in the future, our analyses will get better and better (more characters, a better understanding of evolution, more computing power, more independent analyses, more reviews of old analyses, beter descriptions of fossils etc.) and thus the fundamental trees that go into making up a supertree will be more robust and better built. I would be intrigued to see what would have happened if we had just excluded ever tree pre 2000 say, or even just pre 1990. I suspect the result would be a rather closer to the ‘current’ consensus. Even things like alverasaurs have morved significanlty and repreatedly in recent years. A consenus a few years down the line followed by the elemination of early studies might really shore things up. Loo how much the sauropod side of things has resoved since the 2002 supertree with the advent of new analyses and a near consensus from Upchurch, Wilson and Yates.

    Right that’s it for now, there’s nearly a whole post in this one comment!

  10. 10 Richard 24/07/2008 at 8:47 pm

    “You really can’t just ‘build a huge tree’ of all dinosaurs”

    Au contraire. The biogeography study by Upchurch et al. (2002) used an informal supertree created by simply combining what they considered to be the best supported and most comprehensive phylogenies for each major dinosaur clade. I did the same for my 2006 paper on ornithischian biogeography, and Carrano did something similar for his Cope’s rule study. Such an approach is much quicker (it can be built in a few hours, if the builder is very familiar with the phylogeny and taxonomy of the groups in question) and every relationship in the tree is supported by character data. I’m not going to pretend there aren’t problems with this approach, but for my money it offers at least as many (if not more) advantages as a formal supertree, and is much quicker.

  11. 11 David Hone 24/07/2008 at 9:09 pm

    Sorry, I was thinking in purely cladistic terms (i.e some kind of all dinosaurs matrix) and that is a) incredibly longwinded to the point of almost being impossible, and b) will have all kinds of issues with character constructs and mising data (like who do you code for pnematic characters of saurischians for the ornithischians – you don’t obviously, but it creates huge holes in the matrix).

    I’m not sure either a ‘classic’ supertree like ours, or a hand chosen and assembled one is better, each have their merits and their faults. Most of the reason we went this way was becuase we already had the original data set of Pisani et al., and Rachel Jennings was doing a new maniraptoran tree for her Masters thesis, so it was a logical progression.

  12. 12 Andrea Cau 24/07/2008 at 10:03 pm

    Jarry and David, thank you very much.

    I realized after having written that it was the femur of Korean maniraptoran described in 2005.

  13. 13 Manabu Sakamoto 24/07/2008 at 10:21 pm

    Re: Eshanosaurus
    We followed the Dinosauria II for valid taxa. If they regard it as valid, then we kept it. I don’t think there are any subsequent attempts to invalidate this taxon.

  14. 14 David Hone 24/07/2008 at 10:25 pm

    I believe my old boss Oli Rauhut has muttered that it might be a prosauropod, but having looked at it and talked it over with colleagues (fortunate timing, this was only 2 weeks ago now) it is clearly a therizinosaur. Manabu, did you not chekc it out when you were last in the IVPP?

  15. 15 Manabu Sakamoto 24/07/2008 at 10:38 pm

    Plus, even if we had thrown out Eshanosaurus, it wouldn’t affect the conclusion that dinosaurs weren’t part of the KTR – the presence of Archaeopteryx as an unequivocal maniraptoran in the Late Jurassic indicates a diversification prior to that in this clade…

  16. 16 Manabu Sakamoto 24/07/2008 at 10:39 pm


    No, I was rushed for time…it was almost a daytrip from Japan and back…

  17. 17 David Hone 24/07/2008 at 10:43 pm

    That will do it. And something being in the IVPP and being able to *find* it in the IVPP are not necessarily the same thing…

    Incidentally, for anyone in the know and interested, Gasosaurs has now turned up, current location: my office.

  18. 18 Manabu Sakamoto 24/07/2008 at 11:40 pm

    Sweet! I was looking for that when I was there but they couldn’t find it.

  19. 19 Manabu Sakamoto 24/07/2008 at 11:53 pm

    Re: Cryolophosaurus
    We had already closed the data entry stage when the paper on Cryolophosaurus paper came out. So we still treated it as an allosauroid. Changing its position as a ‘dilophosaurid’ again as in Eshanosaurus would not affect the overall conclusions regarding the KTR.

  20. 20 Thomas R. Holtz, Jr. 25/07/2008 at 1:29 am

    Dave & co.,

    Congrats on the new paper coming out.

    However, I note a misplaced clade in the tree. Jingshanosaurus + Mussaurus wound up within basal lambeosaurines in the big supertree figure (this does NOT show up in the time-calibrated phylogeny, thankfully).

  21. 21 Karl Zimmerman 25/07/2008 at 2:11 am

    David, Manabu

    Thank you both for taking the the time to reply to me. Sorry if I came across as credulous – I’m a fairly experienced layman, but not a professional, and a few things raised some eyebrows.

    I’m still not sold on Eshanosaurus. I heard from David on another page there will be a publication coming out on it soon, but it wasn’t clear Shuvosaurus was not an Ornithomimid until Effigia was discovered, with post-cranial elements clearly showing it was Crurotarsi. My understanding is of the eleven diagnostic features of Eshanosaurus mentioned in the original paper, six are also found in prosauropods, five are debatable, and the teeth have a medial ridge, which *is* found in prosauropods, but not therizinosaurs. Even if it’s not a prosauropod, it might end up being a late-surviving relative of Revueltosaurus or something wilder. I’m not saying it’s impossible, but given no other Averostra (that’s the clade of Ceratosauria + Tetanurae correct?) are known before the middle Jurassic, no Maniraptors are known before the Late Jurassic, and there are basal Therizinosaurs in the Cretaceous, it has to be viewed with some suspicion until we have more corroborating evidence.

    You are of course correct that this wouldn’t be determinative, as it still means Maniraptors must have diverged into their main clades prior to the evolution of flowering plants. Anyway, I know I was critical, but it’s still an impressive body of work overall – and as David said, future trees are bound to improve.

  22. 22 Christine 25/07/2008 at 8:08 am

    I am writing to make sure that you are aware of the History television program, Jurassic Fight Club, a new series that illustrates prehistory and depicts how dinosaurs hunted their prey, dissecting these battles and uncovering a predatory world far more calculated and complex than originally thought.

    I think you and your readers would enjoy History.com’s minisite (http://www.history.com/jurassic), and especially the Jurassic Fight Club: Turf Wars game (http://www.history.com/turfwars). The minisite features a program synopsis, an episode guide, a Dinopedia, image gallery, discussion board, a multimedia gallery filled with wallpapers, posters and screensavers, and a three-tiered video player.

  23. 23 Randy 25/07/2008 at 8:25 am


    I can assure you that Eshanosaurus is not a late surviving relative of Revueltosaurus, but your point is very well-taken. No matter how similar Eshanosaurus is to therizinosaurs, I hope people have learned their lesson about the homoplastic nature of archosaur teeth. I wonder if there is any matrix left on the Eshanosaurus specimen – one could use REE signatures to see if it matches other Lower Lufeng fossils.

  24. 24 Manabu Sakamoto 25/07/2008 at 5:36 pm

    Dr Holtz,

    Thank you very much!

    This particular tree our Supplementary Figure 4, a split-fit tree, the method of which I don’t really know about, but it was not well supported so wasn’t used in further analyses. The standard MRP supertree, which had the highest supprt thus chosen to be used in further analyses, though show Jingshanosaurus and Mussaurus as prosauropods.

  25. 25 Manabu Sakamoto 25/07/2008 at 5:41 pm


    Apologies for the grammatical errors in the previous post – I don’t know how to edit posts here so I’ll just add a new one.

    What I meant to say is that the tree Dave chose to present here is the supertree shown in Supplementary Figure 4.

  26. 26 Thomas R. Holtz, Jr. 25/07/2008 at 8:46 pm


    Assuming Berberosaurus is a basal ceratosaur or a basal abelisauroid, than we now have an averostran in the late Early Jurassic.

    That being said, I’m still not convinced Eshanosaurus is a therizinosaur…

  27. 27 David Hone 25/07/2008 at 10:25 pm

    Well I don’t want to dwell on the issue becuase it’s not my paper, but well, having actually looked it over and talked about it with others, I am convinced it’s a theropod. For more than that you wil just have to wait.

    And Christine, in case you do come back and read this, not only do I not have much of a readership, but I rather imagine that any reader of this will only be interested in spotting the errors in the show and criticisng the science. That’s just the way it goes really. Not that I have even see a trailer for the show, but one thing that constantly frustrates me about these things are the attempts to jazz up the science. It doesn’t need it, the damn animals are weird enough as it is and we do know alot of interesting theings about them, you don’t have to pick left field ideas or unsupported nonsense to make the animals or the research cool. Stick to the science, it will be more accurate, more educational and just as fun.

  28. 28 David Peters 14/08/2008 at 6:18 pm

    Just curious… what generic taxa came BEFORE the dinosaurs? IOW What are the outgroup taxa? And how do they line up? I’ll probably get the answer as soon as I download your paper, but I thought I’d get a blog answer, too. BTW, that’s quite an undertaking, that supertree. Congrats on a job well done!

  29. 29 Gothy 14/09/2008 at 3:53 am

    I am curious as to the placement of Deltadromeus agilis as being a basal ornithomimosaur, as all that I have read has placed this species as being abelisaurid in nature.

  30. 30 David Hone 14/09/2008 at 5:17 am

    In answer to both the last comments, it’s just a function of what trees were put into the analysis as this is how supertrees operate. In theory at least, the actual analysis contains no bias, just the results gained by other researchers (those of which could themselves have been biased of course).

  1. 1 Dear Kitty. Some blog :: Dinosaur and other Mesozoic life diversification :: July :: 2008 Trackback on 24/03/2009 at 5:36 am
  2. 2 Guest post: Dinosaurs dying off? « Dave Hone's Archosaur Musings Trackback on 05/05/2011 at 9:39 am
  3. 3 Dinosaur and other Mesozoic life diversification | Dear Kitty. Some blog Trackback on 18/01/2012 at 7:31 pm
Comments are currently closed.

@Dave_Hone on Twitter


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

Join 574 other followers

%d bloggers like this: