Constructing hypotheses on behaviour in the fossil record

Those keeping up with papers on palaeoethology may well have noticed that a number of papers have gone online in the Journal of Zoology of late with a common theme. Darren Naish has a paper on the behaviour of fossil birds, Andy Farke has one on combat in ornithischians, and Pete Falkingham has a paper on interpretation of trackways. This is not a coincidence, but part of a special issue of the journal out today on behaviour in the fossil record and all of these contributions will eventually be published together with a number of others in a collection I have assembled as a guest editor. The volume has ended up rather dinosaur-biased which is unfortunate as a number of other papers were promised from other fields (including on whales and the Burgess shale) which never appeared and giving the set a more dino-centric appearance than I had planned or hoped for.

Adding to this is in fact my own paper in the volume. This was something I had been working on for a while before being asked to compile the special issue (indeed the fact that I was working on it, and it was intended got the journal may have precipitated the invitation) and in the context was the perfect home for the paper. As with similar cases I had nothing to do with my own manuscript and it was submitted separated and edited and refereed independently by the journal, and only after acceptance could it be added to the list. Most of the papers are reviews of one form or another, and in my case the paper written with my friend Chris Faulkes looks primarily at issues of hypothesis creation on behaviours for fossil taxa.

Our main contention is that in the past palaeontologists have been a bit over zealous in the production of hypotheses and the way in which they have been generated has made them difficult to assess or even simply discuss and in at least a few cases hould probably not have been suggested at all. We don’t think it inappropriate to generate hypotheses that cannot be immediately tested, or those that are difficult generally to assess, but a hypothesis must have at least some support behind it to make it valid in the first place, and poor uses of terms, lack of specificity, or even use of fundamentally flawed concepts have meant that there are problematic ideas in the scientific literature.

Mutual sexual selection is perhaps a good example here. I’ve now penned a number of papers with various authors about the issues surrounding this idea and how it may fit into archosaur evolution. The point is not whether or not we are right about this, but more the fact that this was something hinted at by Darwin, written about by Huxley and extensively studied by numerous ethologists for decades, and yet many palaeontological papers discussed sexual selection purely in terms of dimorphism, or the fact that sexually selected features should feature on only one gender, or indeed that sexual selection should be mutually exclusive of other functions. None of these things are true, so hypotheses that rely on one of these as are starting point are going to be fundamentally flawed, or at least problematic.

Thus the paper sets out to identify some key areas where we feel mistakes have tended to be made (myself drawing on examples from dinosaurs and pterosaurs, Chris from his area of expertise the mammals) and to also then try to find a set of guidelines that might help the better generation of hypotheses allowing for reduced confusion and better testing. Naturally we think this is going to benefit researchers, but given the rampant hypothesising that often accompanies any online discussions of the behaviours and ecology of extinct animals online and in other informal venues, it might just help clean up some of the more egregious suggestions that can be put forwards based on the most tenuous of links. Some of it may sound excessively simple and even obvious, but that doesn’t mean it hasn’t been an issue in the past. I actually had a chat with an ecologist the other day who bemoaned a similar set of problems in her field, and I think the issue is more one of advancement and general improvement that systematic errors or poor science.

Naturally we did try hard not to pick on individual papers (or people) but we did also want to point to some specific examples of the kinds of problems we were discussing and so a few things get the finger pointed at them, but they have mostly had specific rebuttals in the literature already, or were very much generic issues. Hopefully then, we’ve not bent any noses out of joint. I was certainly grateful to Andy Farke for reading an earlier version to check for overall tone and to see if it was working the way we wanted. Anyway, here are a few of the things we looked at.

Terms need to be more specific. Talking about ‘parental care’ say in general terms isn’t very helpful when this can encompass pre- or post-natal care, or both, and differing degrees of commitment from parents over very different timescales. So a statement like ‘X showed parental care and Y didn’t’ may not mean much if the parental care shown was minimal, or two papers might say this where one is referring to all parental care, and the other only post-natal, making them hard to compare.

Overlooking counterexamples or complexity. Descriptions of species or clades as ‘social’ has been creeping into the literature on dinosaurs and yet even if you do somehow have super evidence for sociality in a species, applying that to other taxa, or even other member of the same species is not necessarily a great idea. While we do have highly social species that basically can’t function when not in a group (like some molerats) even famously social animals like lions often spend part or much of their time apart, and some like cheetahs can be incredibly plastic, switching from social to solitary multiple times in their lives, and yet it would be a big mistake to suggest tigers are fundamentally social because their nearest relatives the lions are.

Extreme examples or oddities are useful to provide context or even limits on ideas. Some species have incredibly specific requirements or only live in certain environments, while others are much more adaptable. You don’t really find sand cats outside of deserts or dry environments, and while lions show up in quite a few places, you can get puma in everything from high mountains to praries, deserts and rainforests, yet there’s not especially obvious about their osteological anatomy that they could occupy so many more environments.

Make sure the analogy or reasoning behind it is actually correct. Not too long ago it was suggested that azhdarchids had long necks to reach into carcasses of large dinosaurs. However, given that the heads of the biggest azhdarchids (estimated at getting on for 2 m) are longer already than the longest sauropod ribs we know of (2 m) then any kind of neck is redundant in this context, let alone a long one, and vultures do fine with absolutely short necks and heads while feeding on carcasses of animals many times their size. The analogy that the hypothesis is being based on is fundamentally false and if that is the sole support for it as a concept, then it’s really not much of a hypothesis.

The short version of much of this could well be summarised as “look more at the behaviour of extant organisms”. I know Darren bangs this drum a lot on TetZoo and I’ve said it in plenty of talks and to lots of people if less so online. It is confounding when people say that such-and-such behaviour isn’t seen in reptiles when it plainly is, or that only animals with feature Y can do this behaviour when it’s known in numerous species, that are just less specialised towards it (or even show no obvious adaptations – like tree-climbing crocs). True this may not be common or normal, but to assume that it’s impossible, or that there is a perfectly consistent correlation is incorrect.

Part of the difficulty is a lack of good data on many of these things. Ethologists can simply observe behaviours and therefore don’t necessarily go looking for osteological or other correlates that we might be able to detect in the fossil record. That does make things harder, but we need to try and avoid getting trapped by ‘we don’t know if this correlates therefore this hypothesis is valid since we don’t know’. I am actually not against (in principle) hypotheses that are difficult if not currently impossible to test, but as with the azhdarchid neck example, there is a difference between something that can’t be tested, and something which is not even supported at the most basic of level. A hypothesis has to have some support, and some specificity about that will go a long way to making things much more clear and amenable to testing and allow a great fit of existing and future data.

What is most remarkable is how far things have come so quickly. So many modern analyses are using things like FEA and functional morphological analyses, are looking for correlates of behaviour (or aspects of ecology that link to behaviour), and more and better comparisons to extant forms and their anatomy are being used. Such important work or our understanding of the biology of extinct animals should not be let down by poor hypotheses and we do hope that, while things are improving already, this will help better communication and understanding of ideas.


D. W. E. Hone and C. G. Faulkes 2014 A proposed framework for establishing and evaluating hypotheses about the behaviour of extinct organisms (292: 260–267)





6 Responses to “Constructing hypotheses on behaviour in the fossil record”

  1. 1 roberta4949 02/04/2014 at 4:30 pm

    nearly if not impossible to assess behavior of animals long extinct all one can do it use their imagination.

    • 2 David Hone 02/04/2014 at 7:09 pm

      Unsurprisingly I disagree rather strongly. Have you actually read any of the papers or looked at some of the depth and detail put into the analyses? Yes this piece does criticise some previous works, but overall there’s a lot that can be done and determined accurately.

      • 3 roberta4949 03/04/2014 at 1:43 pm

        actually I have been reading a lot of books and articles I can find on dinosaurs and many including abook written by smithosian book for example on how little they know of dino behavior, it is all speculation, but when I see a video of a ancient animal with beaver teeth being called a beaver cat and being a vicious hunter and killer then they lose me, because they are obviously making it up to support their ideas of evolution and dog eat dog world, when you look at the fossil it is basically a very large beaver and there is no way on this green earth that anyone would even think that is even viable explanation. there is nothing now that even comes close. like the terror bird I look at that and think no way a bird built like that can run down another large animal and knock it down with it’s bill, this in my mind is a fabrication, there is no way to tell from fossils how an animal behaved, it is impossible to know for sure, you can guess you can observe animals now, but I wouldn’t put to much credit to humans ideas even if they have a fancy title unless they can present some pretty hard evidence for their ideas. being to college and having access to a lot of writtings by other honored scientists or whatever is not hard evidence in my book. surly come of their ideas are viable, but when they start going into the realm of fantasy and conjecture then I take what they say with a grain of salt.

      • 4 David Hone 03/04/2014 at 2:18 pm

        OK well there is a lot we do not know, but there is a huge amount we do know and can test and have tested.

        Case in point, basic pre- and post-hatching parental care in non-avian dinosaurs. That’s something that’s far from a simple behaviour (no one would disagree that things like tyrannosaurs ate meat with those teeth for example), so what is there to support this?

        1. Almost all living birds, and certainly all living crocs carry out some form of prehatching parental care (construct nests, guard eggs, help protect eggs and / or regulate temperature humidity etc.) so based on this we would expect to see something in dinosaurs.

        2. We have nests of dinosaurs that show structure. The eggs have been carefully laid and arrayed by the animal which implies something other than a dump-and-leave strategy.

        3. We have fossils of dinosaurs sitting on nests. The eggs are carefully arrayed and the dinosaur has a posture near identical to that of a bird while on the nest. The eggs contain embryos that can be assigned to the adult on the nest confirming the relationship.

        4. We have hatchlings and animals much older than hatchlings in large nests (that they appear to be unable to leave given their size) where there is food material in the nest. This is most likely brought in from outside, and given that the animals are residing in the nests they would be easy prey if mom and dad were not around.

        So even this complex of fairly special behaviours can be demonstrated and reasonably inferred for at least some dinosaurs. Add to that data from trackways, clusters of adults and juveniles found together, adults with young in a burrow and data on things like brain size and behaviour and it’s all together in a pretty convincing package. It’s not made up.

  2. 5 Herman Diaz 03/04/2014 at 5:27 pm

    I’m surprised more ppl haven’t replied to this blog post. In any case, I’m glad to see such a paper be published, especially by you.

    “None of these things are true, so hypotheses that rely on one of these as are starting point are going to be fundamentally flawed, or at least problematic.”

    I’ve complained about Roach & Brinkman 2007’s many problems elsewhere ( ), but it’s worth mentioning here that the above quote reminds me of said problems. For example, it’s implied that lone adult Komodo dragons can kill prey 10x their size w/”only ser- rated teeth”, the logic being that lone adult Deinonychus would’ve done the same. However, it’s been known since 2005 that the former are venomous ( ), hence why they can kill prey 10x their size. Also, it’s claimed that, among extant archosaurs, “truly coop- erative, coordinated hunting behavior is seen in only a few species of diurnal raptors”, ignoring ground hornbills (Fitz Simmons 1962), corvids (Maser 1975), & shrikes (Frye & Gerhardt 2001).

    “Terms need to be more specific.”

    The above quote is important to me b/c I find social behavior (especially social foraging & parental care) very interesting, yet very hard to learn about when different scientists use different terms for what may or may not be the same things. Specifically, it’s annoying when 1 scientist says that a species hunts “cooperatively”, but doesn’t explain what exactly he/she means by that, or when another scientist says that the same species hunts “communally”. That’s why, when discussing social behavior, both vertebrate paleontologists & wildlife biologists should cite Ellis et al. 1993 ( ), which neatly sorts out the various classes of social foraging. If you can’t get past the paywall, the 1st 2 pages of Orellana & Rojas 2005 ( ) sums them up; There’s non-cooperative hunting (E.g. Mobbing), cooperative searching, pseudo-cooperative hunting, & true cooperative hunting (I.e. Pack-hunting).

    “The short version of much of this could well be summarised as “look more at the behaviour of extant organisms”.”

    That’s why Gardom/Milner’s “The Natural History Museum Book of Dinosaurs” (See “Book Description”: ) & Sampson’s “Dinosaur Odyssey: Fossil Threads in the Web of Life” are some of my favorite popular dino books. The same goes for anything by Bakker. You’re probably tired of hearing this, but that’s also why I hope you write popular dino books too, someday.

  3. 6 George Hancock 08/04/2014 at 9:45 pm

    David you have to check out the latest one show episode. They demonstrate some remarkable goshawk hunting behaviour which can potentially be applied to a parachuting predatory behaviour in Dromaeosaurids.

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 453 other followers

%d bloggers like this: