Dinosaur dimorphism, cryptic absence

Yes it’s new paper time and this is one I want to talk about in some detail, so here’s a longer than normal post on this. It’s an issue that has been in my brain for years but has taken time to mature with the right set of circumstances and quirks that make up the profile of a research paper possible. This one is returning again to the much investigated area of sexual selection in the extinct Dinosauria. I think it’s fair to say I’ve been a leading researcher in this field with a string of papers on various issues surrounding sexual selection and dimorphism in dinosaurs (and others), how we might detect sexually selected traits and what they may mean for behaviour, ecology and evolution.

The new paper is written with Jordan Mallon, and in it we tackle the issue of the apparent lack of dimorphism in dinosaurs and why there is still no good case for a dimorphic dinosaur. Despite numerous studies suggesting a split between male and female morphs (or similar robust and gracile ones) revisions have generally found the cases to be lacking and Jordan’s own recent paper in this area is relevant for several reasons. The story though starts quite a few years ago.

My own works on sexual selection mostly kicked off with a paper discussing mutual sexual selection and the idea that both sexes in many populations of dinosaurs may have borne ornaments for social and / or sexual dominance. In short, males had big ornaments (claws, horns, frills, crests etc.) to advertise their general good health and status to females for mating and other males in terms of competition, but females likewise advertised their general quality with the same signal. That meant that in something like a typical ceratopsian both males and females had a big frill and horns and hence an inability for us to identify and separate out the two populations.

This general concept had been completely overlooked in the literature in sexual selection and dimorphism in dinosaurs and it’s worth repeating that this can totally confound some ideas and tests for sexual selection and it needs to be borne in mind when discussing these kinds of traits. While later papers have built on this issue and surrounding ones for the function of crests and the like, in my mind it has always been an unsatisfying explanation for things. Sure, it’s a big issue and to ignore it is incorrect, but while we are finding more and more examples of mutually sexually selected species, it seems unlikely that so many dinosaurs (huge numbers of hadrosaurs and ceratopsians, various theropods and other lineages) all had equal ornaments. Sexual dimorphism, be it in body size or crest size and shape (or of course, presence / absence) surely was present in a few of these lineages?

One issue is of course that for a lot of species we have very few specimens (often only one) and certainly don’t have lots of adults in good condition from a single site for many. As I noted in a paper on social behaviour in dinosaurs, lots of large terrestrial mammals at least show different fundamental patterns in group behaviours between the sexes so even if we do have 50 animals from a mass mortality site, there’s no guarantee it is not a group of 50 males or 50 females. Identifying different sexes is also problematic of course but it would help if in a few cases we *knew* we had both present in a sample size.

Even so, where was the dimorphism? Was it really absent or merely for some reason, hard to detect? After quite some thought I realised that what might be a major factor is the growth patterns of dinosaurs. Where large mammals and birds rather race to adult size and then stick there, dinosaurs (at least the larger ones) took a longer and more reptilian growth pattern with an extended growth phase (even if they were sexually mature during much of this). That means that much of the reproducing population isn’t full size and that even if say males were much bigger than females, you’d struggle to tell apart and old female from a young male which might be comparably sized. Right, now I had a hypothesis and a mechanism to test it by getting a dataset on dimorphic animals with differential growth and see how they looked depending on how things were sampled.

And then I got stuck. Datasets like this, (especially with reptiles and birds) simply didn’t seem to exist in the literature. Over perhaps 5 years I sent out dozens of e-mails and spoke to various people about data, including biologists, palaeontologists, conservation and zoo workers – anyone I thought might have or know of a dataset of mass for lots of individuals of known sex and age. I got nowhere. Even trying to compile sets from lots of individual measurements didn’t get me anywhere and I was resigned to having a good idea I couldn’t test, until Jordan got in touch.

He sent me a draft of his now published paper re-examining analyses that had looked for dimorphism and found them wanting. Reading it through I was annoyed to see that he seemed to be leaning the same way and that the elongate growth might be a decisive factor, but while the paper discussed some issues around detection it didn’t go there. I was relieved that my idea seemed to still be mine to work on, but as Jordan had asked for any comments to help improve things, it also seemed a bit mean to withhold an idea that might provide a nice extra aspect of the paper.

Happily, as these things often do, a quick chat via Skype helped revolve things. Jordan liked the idea but agreed we could try and combine forces, as he thought he had a good lead on some data we could finally use and importantly also knew how to run the analyses. So with new impetus, the idea was resurrected and the final output of this collaboration is now out.

So, what did we find? Well it looks like my thoughts were generally correct but far more so that either of us suspected. We used alligators as the reptile model since they form a part of the dinosaur phylogenetic bracket and have large dimorphism (at full size, males are 30% and up longer than females), with rheas on the other side (also show major dimorphism and are large birds). Doing various subsamples of each it is clear that it’s much easier to detect dimorphism in the birds because you are tending to sample animals that are at full size. You can use a much smaller sample size to detect dimorphism in birds than reptiles. You typically need around 30 animals of *each* sex to get a statistically significant difference in alligators, even though that have one of the highest levels of dimorphism recorded.

Given how few dinosaurs even have a dataset of 60 animals (and then the issues of making sure you can measure them all accurately, and of course the fact that you’ll be lucky if even a few are sexed, and you may have all of one sex, and there is often variations between populations) and then it becomes little surprise we have picked up no good signals for dimorphism in dinosaurs to date. This does become a little better when we sample from larger individuals (as there are several biases against juveniles in the fossil record) but still well below what we can do for almost any dinosaur.

One other aspect that we look at is the range of dimorphism appearing in extant reptiles. There’s a surprising (to me) level of variation in populations with major variations in terms of just how dimorphic one population is versus another and these can also change over time. Some populations of single species even show males being larger with others having larger females. That’s also potentially an issue given our tendency to have to lump together specimens from multiple different sites to get to a decent number of animals to measure and while it might not be common, it’s clearly a potentially confounding signal.

This is of course not the final word on any of this. There are other aspects to both growth and dimorphism and how we measure it in both living and extinct animals. Certainly I think it’s possible to make a good case for dimorphism with only a limited sample (as has been looked at for example with oviraptorosaur tails, or indeed for some pterosaurs) but the apparent lack of dimorphism for dinosaurs in the fossil record is not as alarming as it might seem. Yes we might expect numerous species to have been dimorphic but it appears that our sample sizes are simply too small. Through in the unknown age and sex of most specimens, and the potentially confounding effects of mutual sexual selection and it becomes perfectly possible that many species (even those represented by large numbers of good specimens) were strongly dimorphic but we are simply unable to identify it.

For years I’d been puzzled by the apparent lack of dimorphism, and Jordan’s paper confirmed that we really have yet to show it clearly in any dinosaur. Mutual sexual selection is a major issue but it probably doesn’t explain all the cases we know about, but I think this paper adds a pretty substantial concept to our understanding of dinosaur dimorphism. Or rather, that we don’t understand it that well but that the apparent absence could well be a classic absence of evidence problem. As with a number of issues in behaviour and ecology, I rather suspect we don’t know as much as we think we do, but understanding what we do and don’t know with confidence is a major step forwards to getting to grips with the problem, so this hopefully is progress even if we can’t find much right now.

 

Hone, D.W.E., & Mallon, J. [joint first authors]. In press. Protracted growth impedes the detection of sexual dimorphism in non-avian dinosaurs. Palaeontology.