Those with an interest in dinosaur cranial crests and exaggerated structures (which should really be everyone since they turn up in pretty much every major lineage one way or the other) will probably be aware of the exchanges going on in the literature over these features. Although myself and colleagues have been advocating that sexual selection (and or socio-sexual signaling: the two can be hard to separate) is a likely strong candidate as the prime driver for many of these features, others have been advocating that this is not the case and instead the answer lies in species recognition. The latest to delve into this area is a paper I’ve done with Darren Naish and is the first time we’ve addressed this issue directly. While we have written or contributed to a number of efforts looking at support for sexual selection in dinosaurs, this is the first time we have tackled the other side of the problem.
The paper originally started as a long section that was included in our paper on mutual sexual selection with Innes Cuthill, but as we were later forced to cut down the length of the submission, this was a section that was relatively easy to prune as tangential to the main issue. However, we felt it needed saying and with new data coming out and the discussion ramping up, we revived and revised the work and it is now out. (Well, it has been in press and available for a while but is now properly out).
This is an important area for discussion – after all, the horns, crests, frills, plates, bosses and the rest (not least feathers) are key features and adaptations in various dinosaur lineages and trying to work out how they might have been used and what this means for evolutionary drivers and patterns is going to be a major issue. It’s hard to really understand stegosaurs or ceratopsians say if you can’t say that much with confidence about their ‘bonus’ features. While obviously each clade, or even each genus / species probably needs to be taken on a case-by-case basis when it comes to detailed analyses, some gross patterns can be seen or at least discussed. In the case of species recognition, is it even an actual ‘thing’ when it comes to exaggerated structures, and if it is, how is it supposed to work. The hypothesis has enjoyed some support in the literature for some unusual dinosaur features so it’s well worth examining.
Species recognition (in the context of exaggerated structures) for those who don’t know, is the idea that individuals of a species use these features to help them recognise cospecifics with to ensure they mate with the right species, or to maintain herd coherence. In short, carry round a key feature and you should be able to make it easier to stay in touch with the right animals and avoid the wrong ones. Various lines have been put forward to support this idea (in general and specifically towards dinosaurs) but we feel that none of them actually stack up and some have some serious problems.
First off is a pretty big issue – to our knowledge there is no evidence of any living species using some form of crest or exaggerated structure for species recognition. Individuals of species do recognise each other (not a big shock) but actually things like antlers or casques don’t seem to form part of the pattern that conspsecifics recognise. This may not be a big shock, after all, you can recognise a species by the overall appearance (size, shape, colour), their smell or specific sounds they make, behaviour, and other features. On top of this, some species are very varied in appearance for the big features (antlers of deer look very different as they grow, and are different between males and females and between juveniles and adults etc.) so relying on one feature is a bad idea at best, and a plastic one an especially bad call.
Plus of course, you often get closely related taxa that are sympatric. Is some big set of horns going to help you correctly identify conspecifics if there are half a dozen similarly-looking species also in the area? Look at things like African antelope and gazelle, or more extreme examples like tyrant flycatchers. We have trouble telling them apart sometimes based on their morphology, yet they seem to have no trouble. If this is so critical to dinosaurs, why to the iguanodonts seem relatively free of crests, but the hadrosaurs go nuts with them? And why are they all so similar in general form between species when they are supposed to help separate them out? Surely they should be divergent, not all similar in appearance. And why do we see things like Wuerhosaurus or Spinosaurus running around with all this weight to make sure they don’t mate with the wrong species when there are no other members of their clade to get confused with?
In some cases we see both issues coming together. If we look at the various small protoceratopsians of China / Mongolia, we see disagreement between researchers as to how many species (or genera) there may be. What is notable however, is that the characters being used to separate them out don’t typically involve the frill or bosses of the skull, and where they do, may be things that are not externally visible (e.g. the width of the media bar in the frill). In short – if there are multiple species here, the frills are apparently similar enough that we can’t separate them and so are unlikely to be part of the identity concept of the animals. If however, there is only one species present, then we are back to the paradox of a large frill being carried around but with no other species that could confound any signals.
On top of that, is it really worth it? After all, while you do want to stay in touch and make sure you mate with the right species, bolting on a good few kilos of bone to your head, and then the extra muscle to support it, and then lugging that around for your entire life is a lot of effort. When you can probably already identify conspecifics by their colour, patterns, scent and calls (of simply because nothing else like them at all is on the same continent) surely these would experience strong negative selective pressures if they didn’t have any other support.
Furthermore, how would such features ever evolve? If the populations / species were allopatric then we return to the situation of them not having another group to get confused with and crests are unnecessary for recognition. If they were sympatric though, how would this work? Pretty much the definition of a natural biological population is one that is breeding within itself, but here we’d have to have a population diverging because some don’t recognise each other as conspecifics even though we would expect, pretty much by definition, there not to be too much difference in structure shape between them (e.g. a tiny crest vs no crest). Now some animals might prefer each other, but that’s mate choice, not recognition, and there would have to be enough individuals for this to work – one mutant with a crest when no one else has one is not going to start forming a new species, and if there were a bunch of the with the new crest they’d also have to identify each other as different and avoid mating or hanging around with the others. So how would a large feature that’s for correct recognition allow a population to split in this way? To us at least it appears most unlikely to occur at all, let alone repeatedly.
In addition to this, there is rampant hybridization of closely related species in the natural world (and indeed in captivity). Even extravagantly ornamented species like pheasants with numerous adornments and bright colours and patterns hybridise regularly – clearly no matter how extreme the cue, at least some animals regularly have problems with them or are indiscriminate, but either way they are not that effective.
While some data like the apparent rapid growth of structures late in ontogeny has been used to support the idea that they are characteristics involved in socio-sexual signaling, it’s also a problem for the herd coherency part of the model. After all, lots of juvenile dinosaurs are known from aggregations suggesting they spent a lot of time together, even when the adults did not appear to. If these features were key, we’d expect juveniles to have them, and adult perhaps to shun them when they were no longer needed, but instead the opposite is true. In general the herd coherency argument is a bit odd anyway, again you have lots of ways of identifying and keeping in touch with conspecifics and some are clearly better than visual aids. Scent can have a temporal component, and vocalizations can be interactive beyond line of sight (especially useful in forests, or when things are behind you, or you are foraging and looking down etc.). No matter how big they are, visual structures are not always going to be that useful, even if they are unique.
In the increasingly infamous issue of Torosaurus and Triceratops, if these animals are truly conspecific then for a start we are back to the issue of ‘lone’ taxa (I don’t think Leptoceratops is going to be much of an issue here) and the pointlessness of crests where none are needed. On the other hand, this is also potentially a problem for the mate recognition idea. We know that at least some dinosaurs were sexually mature before they were osteologically mature and this could be the case for these animals too. If so, then the alleged transformation between one morph and the other would create confusion – both the Triceratops morph and the Torosaurus morph (or indeed anything in between) would be viable mates.
In short, we really have no clear evidence for species recognition in any living species, and that alone should make it unlikely to have been a key player across dinosaurs for the whole Mesozoic. Such structures would be costly, and yet not necessarily do the job it is supposed to with other signals being cheaper and just as effective, or more effective in many circumstances. It’s not clear why it should be so important for some clades and not other similar forms (iguanodotids vs hadrosaurs for example) and is clearly either redundant for some taxa, or would not actually reduce confusion. Nor is it clear quite how this would evolve in the first place, or why it would be sustained, and hybridization suggests that crests alone would not even prevent incorrect matings. Put this all together and we feel that there really is no good support for the idea of crests and other structures being primarily used in species recognition. They did of course likely have an effect – it would be odd if Stegosaurus or Corythosaurus didn’t use their respective features as part of how they identified one another. But that does not make them the prime, or only, driving force of all these different features in all these different lineages.
There was a fashion in dinosaur palaeo to write off any odd structure as simply sexual selection and leave it there. This was rightly railed against, but what was often criticised was the fact that sexual selection seemed undiagnosable in the fossil record and so the problem was that it was untestable rather than the fact that such throwaway remarks devoid of context or explanation do little for the subject. Now we are in the odd position where rarely you see very similar comments (in terms of their style) about species recognition popping up in the literature about exaggerated structures despite the lack of support for it, and the now (well, we think), strong cases made for sexual selection, or at least it’s assessment. Although previously the case for sexual selection was pretty weak, it is at least an extremely common phenomenon in living taxa and with obvious powerful effects on anatomy and behaviour. Species recognition has not yet even been shown (in relation to exaggerated structures) in any living clade, and while offhand one-line explanations are not the way to go, it seems odd that one has been replaced with the other.
Hone, D.W.E., & Naish, D. 2013. The ‘species recognition hypothesis’ does not explain the presence and evolution of exaggerated structures in non-avialan dinosaurs. Journal of Zoology, 290: 172-180.
Dave Hone's Archosaur Musings 
Recent Comments