Wednesday, 10 August 2016

Extraordinary claims need extraordinary evidence

As the post's title says, extraordinary claims need extraordinary evidence and when it comes to palaeontology this couldn't be more true. If you have spent any length of time reading palaeontological literature, you may think that nearly all extinct animals did only one thing all the time. However, to realise how ridiculous that is, all you have to do is observe any living animals. Yet some palaeontologists seem to happily to ignore these facts, or unwilling to correct the media if things are misquoted. Either way the media love a snappy headline so it often makes big news.

Claim 1. T. rex was a scavenger. 
This all started with Jack Horner giving a talk around 1993 where he discussed the idea that T. rex was a very strange build for a predatory dinosaur with its tiny arms. Through the years it keeps re-appearing as a common question, and Jack Horner himself happily will raise it as a debate starter although there is no evidence he truly believes it. *There was an extensive chapter in a book in 2008 by Thomas Holtz that critically appraised this field (and covers many of the same points I do here, although I must admit I hadn't read it at the time of first publishing this article).* Brian Switek has written a great blog post on the whole debate back in 2013, shortly after a paper in PNAS described fossils of two fused hadrosaur vertebrae with a T. rex tooth crown embedded in them. The vertebra in question had regrowth around the damage showing the hadrosaur had survived its hunting and lived for at least a while longer, although the regrowth and trauma may well have led to its ultimate demise.

Figures 1 and 2 from  DePalma et al., 2013. Top showing the fused vertebrae. The horrific change in shape is a result of the trauma from the tooth (probably infection accompanying regrowth). The tooth crown is circled in white. Bottom images show CT scans of the crown embedded within the bone.
This is a case where a cool fossil find gets far more value as a publication than it should do (no offence to the authors as the fossil and science are both cool), because it is putting the nail in the bed of T. rex being a scavenger. But let's think about this in terms of terrestrial animals today. If I told you to think of a large scavenger, you'd almost certainly think of a hyena in Africa (and most likely a spotted one). I will forever remember being taught in a university course that spotted hyenas hunt 60% of their food! Equally shocking was being taught that for lions, they often scavenge more than they hunt! Of course I hear you all being sceptical (as was I), so I've done some research and indeed spotted hyenas across Africa always hunt more of their food than scavenge. The link is to a BBC article I accept, but the scientist interviewed has published many papers on the matter (e.g, Holekamp et al., 1997). As for the lion fact there is far less data out there, however it is widely accepted that lions can, and do, scavenge large percentages of their food. The reason for this is blatantly obvious: lions are the largest carnivores in their environment, if they stumble across a kill from another smaller predator (single hyena, leopard, cheetah etc.) they can easily scare it away and eat the food with no energy expenditure of the hunt (and their hunting success rate is low: 10-20%), so why wouldn't they? So why do we think that dinosaurs would be any different? T. rex was the largest carnivore in its environment so it will almost certainly have done both to varying extents depending on prey and carrion availability.

Claims 2-many. Spinosaurus.
Where to start with this one. I've worked on spinosaurs as part of my MSci project and have published on their snouts and convergences so am potentially biased from that work, but I hope I can walk through my reasoning on many of the issues with publications in the last few years. Just a quick background (if you don't know), spinosaurs are a crazy group of theropod dinosaurs that evolve incredibly elongate snouts. The group has gained its name from the first species found, Spinosaurus aegpytiacus which was discovered in Africa, and remains the largest known of the spinosaurs (and length-wise one of the biggest theropods ever) and possessing the biggest sail on its back. It only really came to global fame with Jurassic Park 3 as the big dinosaur that chased around the actors and killed a T. rex.

Spinosaurs being big and weird, with a global following garner a lot of interest in publications. This is further helped by the fact the original Spinosaurus specimen were destroyed during WW2 in a bombing raid of Munich so new fossils help inform us on this enigmatic group.
The original Spinosaurus specimen found by Stromer in 1912, now lost.
They also have evolved an unusual expansion of the premaxilla (the tip of the snout) called a terminal rosette.
Dal Sasso et al., 2005. The largest known Spinosaurus snout showing the highly elongate snout.
These unusual morphological features got them very quickly compared to lots of crocodiles, particularly the equally unusual and long snouted gharial.

Gharial skull showing the elongate narrow snout, with terminal rosette expansion (photo from Glasgow Museum Collections)
The gharial is a crocodile from India which uses its unusual snout to slash through water and to stun/capture fish. As such it follows that because things look the same, they eat same things and in pretty much all Spinosaurus papers discuss it eating fish (e.g. Ibrahim et al., 2014). Some of this is backed up by findings of acid-etched fish scales in the gut region of Baryonyx (Charig and Milner, 1997), and the skull shape of Spinosaurus performing similarly to that of the Gharial in finite element models (Rayfield et al., 2007).

However, everyone fails to take into account size.

Whilst the biggest gharials are 970kgs, the biggest spinosaurs are amongst the largest theropods to have ever lived with mass estimates between 7-20 tonnes. This is where my research came in. Spinosaurs do have incredibly rubbish snout morphology for relative strength as they are effectively solid cylinders. Alligators flatten their snouts so have big, flat, wide ones which are much stronger than similar sized cylinders. However, when size is taken into account, all crocodiles snouts were less strong than the Spinosaurus and Baryonyx snouts. Again I know people may argue that it means they are just able to take bigger fish, and there were indeed 6ft long saw-fish in the regions where Spinosaurus lived. However, that argument comes under the Tyrannosaurus only scavenging. Why would a super-massive predatory dinosaur with a strong skull (due to size over shape) only eat fish in an environment that would have had loads of other prey items, whether to hunt or scavenge? Many crocodiles are known to change diets as they get bigger shifting from mainly fish and amphibians to increasing amounts of terrestrial vertebrates. In the case of alligators, they shift from invertebrates and amphibians, to fish, then turtles and ultimately large terrestrial prey as they grow up. Whilst impossible to prove for Spinosaurus at present, a tooth from a South American spinosaur was found embedded in a pterosaur vertebra and an Iguanodon bone was found in proximity to the Baryonyx remains which may have also been consumed.

Due to spinosaurs consuming at least some fish in their diets, it is assumed they must have lived at least partially in the water to catch them. The first major paper attempting to scientifically prove this was Amoit et al. (2010) which used isotopes from their teeth (particularly the apatite - the stuff that makes bone and enamel hard). Some important things to know about the isotopes in question (in this case δ18Op): 1) they vary with body size; 2) they vary with diet; 3) they vary location you are living (whether in actual location or water vs land). That being said, the authors test isotopes of teeth for spinosaurs, theropods and crocodiles (and their osteoderms) as well as isotopes of turtle shells from the same rock layers.

To compensate for the first issue, they get teeth of the same size for the spinosaurs and the theropods. They do not say though how they ensured the teeth all come from the same sized individuals and teeth in spinosaurs are notoriously different sizes in one individual, with big teeth in the terminal rosette, followed by some small ones, followed by some more big ones. The small sample numbers also worry me (most 3-4, but up to 9 teeth) were tested for each site/time as this size issue may come into play more. In addition they do not specify that they compensate for the size differences between the turtles and crocodiles compared to the spinosaur teeth, which when there conclusions are that the spinosaur isotope numbers are similar to that of turtles and crocodiles thus they are at least semi-aquatic, worries me greatly.

The second issue, with diets influencing isotopes, is tricky. Spinosaurs, as a group, are known to eat fish and terrestrial prey but we only have this data for one individual from England, and one from South America. Species and individuals in Thailand and Africa may have different preferences and proportions of each in their diets. For the other theropods, the diets are unclear as not all species are listed, but for the specimens that can be identified to taxa like Carcharodontosaurus presumably most prey was terrestrial. For crocodiles, presumably diet varies with body size like in modern crocodilians, but with no data even on the size of the crocodile teeth, this is tough to even guess at. So again let's assume their assumptions are fine.

The final issue is what is being tested. The authors assumed that they covered issue 1 with same sized teeth, and for sake of argument let's also assume that. They cover issue 2 by saying that spinosaurs eat at least some terrestrial prey so their isotopes should be similar to other theropods on that ground. Thus differences between the other theropods and spinosaurs must be down to where they are living. In this case, it is suggested that it is the water, as the spinosaur isotopic numbers are not statistically different to that of crocodiles and turtles. The problem with that is that spinosaur numbers are expected to be completely different to crocodiles and turtles living in the same environment (e.g. the water at least part of the time), as their body sizes are completely different. Secondly, the results are really heavily biased by the first three locales (all Thailand) where there is an enormous difference between spinosaurs and other theropods whereas in the other locations spinosaurs are not that different to other theropods. In fact, I carried out Mann-Whitney tests on their data and found that six sites are not statistically different (although sample size may well play a role, but can only work with what they've given), and one of the the other three is statistically "more terrestrial" than the other theropods. Brazil, Thailand and one site in Morocco (of the five) are the only places where spinosaurs are statistically different from other theropods. I am not sure how these issues escaped the reviewers, particularly as this completely changes the result of their whole paper. From the isotopes, it would be safer to say there may be a few locations where spinosaurs are doing something different to other theropods (and possibly semi-aquatic) but the rest are indistinguishable from any other theropod. So in this case the exceptional claim does not withstand scrutiny of their own data.

I had fun picking this apart as a first year PhD student, but have only now gotten around to writing it down, and sadly don't have any additional data to put it into a proper rebuttal. Hopefully someone might be able to add to this to either prove or disprove the original hypothesis, but presently it does not withstand statistical testing.

This paper has been used, in addition to the fish feeding spinosaur papers, to help support the new reconstruction of Spinosaurus by Ibrahim et al., 2014. The story of the new fossils has been well publicised by National Geographic and TED talks featuring Ibrahim but involves purchases by museums of some fossil bones that would be identified as Spinosaurus because of the big neural spines. When the original collector was traced, he led the authors to a site where additional Spinosaurus material was collected. They combined their newly collected material, the previously purchased stuff in the museum, and various other specimens including estimates from the original destroyed specimens. The result is a short legged theropod they propose spent a lot of its time in the water. 
Ibrahim et al., 2014 reconstruction of Spinosaurus.
Again an extraordinary claim and why the paper hit the front cover of Science, full size reconstructions were made by National Geographic and there were countless news and popular science articles and talks about it.

Does the evidence back this up? Well, not yet. The problem with reconstruction is they are based on lots of different specimens, so the reconstruction is a chimera. If any part of these comes from a different species the reconstruction will be wrong. Since the new Spinosaurus reconstruction in 2014, Evers et al., 2015, and Hendrickx et al., 2016 suggest there are at least 2 spinosaur species found in the same deposits, and some of the second - Sigilmassasaurus was used in the reconstruction of Spinosaurus. Even worse, we don't even know the specimens in the museum and the newly collected fossils from the field are from the same individual despite the word of the professional fossil collector in Morocco. There is no collection map that would be typical for any fossil collection to show how the specimens were found. The biggest oddity remains the legs. They are so short compared to any other known theropod. An older, African spinosaur for which a single specimen exists showing a lot of the skeleton (Suchomimus, Sereno et al., 1998) shows relatively normal leg sizes.
Sereno et al., 1998 reconstruction of Suchomimus showing the discovered specimen.
There have been suggestions that small legs would be ideal for swimming around in water particularly if it evolved webbed feet (no evidence) so it could catch fish. Funnily one of the original authors presented work at SVP 2015 showing their reconstruction tipping onto its side when put in water, but more recently Don Henderson presented work at ICVM 2016 showing the reconstruction would be stable in water, but then so was a T. rex model. The short legs don't make it better at being in the water in a static test, but I suspect longer legs would be better to counterbalance the large sail to prevent Spinosaurus being top heavy.

The extraordinary claim really is going to have to wait for a more complete publication of all of the original authors' finds. Unfortunately Science only publishes short papers, so the full monograph description has yet to happen. Whether it will, or gets caught up in never ending peer review due to some of the issues that are now being raised in other publications we shall have to wait and see. Ultimately, we really need an exceptionally complete individual fossil to back it up (like the Suchomimus fossil). Undoubtedly there is one out there, quite possibly in the Kem Kem of Morocco, but it needs a proper excavation and not a piecemeal extraction and selling by a commercial dealer who may have ulterior motives.

In science, the extraordinary claims make the big headlines, lead to publications in the highest impact journals, and this in turn often helps make researchers' careers. The problem in palaeontology, like all sciences, is this leads to overstatements or overreaches for the data that we have (particularly for small sample sizes). I have been lucky to work in labs that have been very good at helping me tone down my language to accept that there are uncertainties in our work, and really extraordinary claims do require extraordinary evidence. I hope anyone reading considers this when writing their next manuscripts.

*Update. It has been pointed out to me that I missed an important reference in the T. rex scavenging debate which I have now included. My apologies. Please continue to add further reading/comments with regards to the topic.

Amoit R, et al., 2010. Oxygen isotope evidence for semi-aquatic habits among spinosaurid theropods

Charig AJ, Milner AC, 1997. Baryonyx walkeri, a fish-eating dinosaur from the Wealdon of Surrey. Bulletin of the Natural History Museum of London 53, 11-70.

Cuff AR, Rayfield EJ, 2013. Feeding mechanics in spinosaurid theropods and extant crocodilians. PLoS One 8, e65295.

Dal Sasso C, et al., 2005. New information on the skull of the enigmatic theropod Spinosaurus, with remarks on its sizes and affinities. Journal of Vertebrate Palaeontology 25, 888-896.

DePalma et al., 2013. Physical evidence of predatory behaviour in Tyrannosaurus rexProceedings of the National Academy of Sciences USA 110, 12560-12564.

Evers S et al., 2015. A reappraisal of the morphology and systematic position of the theropod dinosaur Sigilmassasaurus from the “middle” Cretaceous of Morocco. PeerJ e1323.

Hendrickx C et al., 2016. Morphofunctional Analysis of the Quadrate of Spinosauridae (Dinosauria: Theropoda) and the Presence of Spinosaurus and a Second Spinosaurine Taxon in the Cenomanian of North Africa. PLoS e0144695.

Holekamp KE, et al., 1997. Hunting rates and hunting success in the spotted hyena (Crocuta crocuta). Journal of Zoology 242, 1-15.

Holtz Jr, T.R., 2008. A critical reappraisal of the obligate scavenging hypothesis for Tyrannosaurus rex and other tyrant dinosaurs. Tyrannosaurus rex the Tyrant King, 371, p.396.

Ibrahim N, et al., 2014. Semiaquatic adaptations in a giant predatory dinosaur. Science 345, 1613-1616.

Rayfield EJ, et al., 2007. Functional morphology of spinosaur 'crocodile-mimic' dinosaurs. Journal of Vertebrate Paleontology 27, 892-901.


  1. Erm..., you seemed to have left out the key reference on the issue of tyrannosaurid obligate scavenging.

    1. My apologies. I should do a more thorough literature review next time, but glad we all agree how nonsensical the idea of T. rex scavenging is.

  2. Hi Andrew -

    I look forward to reading your posting at leisure :-) .

    However you should bear in mind that extraordinary claims would only need extraordinary evidence if every theory we have at the moment to describe all the evidence we have, were Perfect.

    Carl Sagan was not a perfect theoriser, and nor was the Italian sociologist who popularised the slogan in recent times. Using the slogan destroys god knows how many very worthy theories every day, and is not scientific:

    1. I'm glad you are enjoying reading the blog. There should be another at the end of the month.

      I appreciate theories are the best way we have explaining things with the evidence we have at present. But if I can pull apart a hypothesis with the data that is presented in the paper, then either the hypothesis is not good, or the data is insufficient for the claim being made. There is a reason in quantum physics they have such a high sigma level requirement for claiming a new particle exists. That is what the crux of the post is about, and not an attempt at diminishing the many theories for which good data exists.

      Behaviours in nature are incredibly variable and making big theories is difficult as nature is weird and for every animal that does something, there is another that does it differently. Palaeontology is even more difficult as we cannot simply observe what the animals are doing so have to make observations through data. All data has uncertainties, and unless you catch the animal doing something (like having gut contents or teeth embedded in vertebrae) making a definitive claim is unwise. Admitting to these uncertainties is good science, but sadly it doesn't help publishing, making a name for yourself or getting the news to pick up on your research. "New evidence suggests that Spinosaurus may be partially aquatic" isn't as exciting as "New evidence that Spinosaurus was partially aquatic" but we should embrace the shortcomings and admit to our limitations as these are the areas that work will continue to advance in forever more.