Monday, 14 August 2017

Alternative fieldwork

The summer is here, and it's field season for all palaeontologists (and field scientists) in the Northern Hemisphere. My Facebook and Twitter have been inundated with photos (including articulated dinosaurs, ankylosaur skulls, tyrannosaur teeth, etc etc.). Check the Twitter tag #fieldwork for an idea for some of the things. Sadly I don't have any field plans and am busy with my mysterious experimental work for the rest of the year. However, I was craving my fieldwork fix so organised a trip down to the Jurassic Coast with a few friends from RVC and UCL to go find some fossils.

If you've missed my previous posts, I have a soft spot for the Jurassic Coast. This stretch of coastline spans 95 miles of Dorset and Devon and ranges in age from the Triassic to the Cretaceous (185 million years). It was made famous by Mary Anning finding large numbers of fossils near her home in Lyme Regis, and is today a World Heritage Site. It was also on the coast at Charmouth where I first went fossil hunting when I was a young kid.

Our car consisted of 5 enthusiasts with backpacks, a couple of rock hammers, and a chisel. None of the usual pick-up trucks filled with pick axes, shovels, plaster, tents etc. We even failed to bring the usual supply of beers (although that was rectified from the field site). A 3 hour drive turned into 4 after a decision to drive by Stonehenge (where traffic was awful), but we eventually made the beach at Charmouth in the rain.

This section of coast is famous for it's ammonites and marine reptiles, although many other things are found here including belemnites, crinoids, starfish, fish, sea urchins, and occasionally dinosaur bits. We parked in the Charmouth car park and worked towards Lyme Regis, focussing on the area around Black Venn. Although not as impressive as it used to be, the slide still gets eroded daily and fossils continually wash out. Highlight was finding my first ichthyosaur vertebra (at a massive 1cm), although found some lovely pyrite (fool's gold) ammonites too.

We repeated this on Saturday, when the weather was far nicer, with another ichthyosaur vertebra being found and lots more ammonites across the group. Sunday we decided to mix it up, and walked to Lyme Regis overland (tide was in) so everyone could see the famous little town and pay respects to Mary Anning at her grave.
Mary Anning and her brother share a grave site. Easily identified in the cemetery as it has a fossil alter at the bottom that we duly added to.
We also had a quick walk around the town, one of the fossil shops (a trend seems to be increasingly less of the local stuff, and far more imported fossils), before lunch on the sea wall before heading down onto the beach and walking back towards Charmouth. The offerings to Mary Anning seemed to have helped as two nice ichthyosaur vertebrae were found in quick succession as we walked along the beach admiring all the ammonites in the rocks (and failing to extract many from the rock falls). As we got back to the Charmouth side, we passed a group of people going for the first time with a guided walk from the heritage centre at Charmouth. The group of 30 or so people (mostly kids) had found a bunch of fossils, and one lady found a pair of articulated ichthyosaur vertebrae.

Then we bundled back into the car and drove down to Lulworth Cove, where faulting and folding have created this beautiful natural cove as the softer chalks inland get eroded.

Bright sunshine over Lulworth Cove. Portland just visible off in the distance on the left of the sun.
After that it was a long old drive back up to London and getting some sleep before work began again in earnest in the office on Monday. I did find some time to take some photos of all the fossils I collected on the weekend:
The random little collection. Everything from millimetre sized ammonites/gastropods to bits of bones. There could have been hundreds of more pieces but I leave many bits on the beach now (particularly the belemnites), and try to give away a good number of finds to kids who are struggling to find their own so they aren't paying for them at the shops.
Piece of pyrite ammonite shell showing off sutures
A small pyrite ammonite showing more sutures.
Nicest two ammonites.
Nothing spectacular as far as science is concerned, but always love finding fossils, and some of them are gorgeous (if I do say so myself). I never tire of finding pyrite ammonites, although it can be tedious finding them when the beaches are so heavily picked across the summer. If you want to find your own, go search! I can offer my limited advice on where to find these sort of things (get in touch), but am still needing to go with some of the local pros to see what they do.

Sunday, 9 July 2017

A day in the life of a palaeontologist

I regularly get asked what I do on a day to day basis, and the honest answer is there is no standard day. So I came up with a lot of things that I reasonably could be be doing on any given day (minus paperwork).

If I was to average all of my days from the beginning of my PhD to now, the majority of days would be sat in front of a computer doing CT scan segmentation. I love (and hate) CT scan segmentation. It's basically high tech colouring in, and who doesn't love that (although I admit some geographers I've teased for their degrees being sponsored by Crayola now get the last laugh).
Fig. 1 from Cuff et al., 2017. See last post for all the details.
Whilst it may not sound particularly fun, and it is incredibly time consuming where the bones are hidden amongst the rock, it is often the first time a lot of the material has ever been seen by anyone and those digital bones can be used for so many more things, like reconstructing skeletons, estimating body masses etc.
Fig. 2 from Cuff et al., 2017. Skeletal reconstruction showing the original bones from Panthera atrox.
But what about those days I escape the computer and get to do other things? It really varies a lot. Most recently there has been nearly all of my time out of the office as we are preparing a new experimental setup in the lab for XROMM (X-ray Reconstruction Of Moving Motion). We will have some animals moving through two X-ray beams to be able to look at their bone movement in real time, and because we have two aligned sources, we are able to get 3D models (see the website by the brilliant people at Brown University which explains it all in greater detail if you are interested). As such we are building runways, aligning and testing settings of the X-rays, and spending far too many hours in X-ray protective lead vests in a surprisingly warm English summer:
Rocking the lead protective gear, with the X-ray setups to the left of the figure with a "bent" runway between them. Also a dog treadmill at the right which we will be using too.
However, due to the fact the experimental stuff is all in the early stages I won't be saying more about it now, but I promise you there will be a lot of posts on it when it is all done. Suffice it to say we've just gotten some preliminary data and it looks incredible! I've also done experimental work on skulls for my PhD using strain gauges which was a whole different set of issues. Linked to this sort of work is a lot of coding too, but my coding is generally rubbish compared to people who spend a lot of time on it, so I won't linger.

Besides that I've also spent a lot of time learning anatomy and carrying out dissections (please note, all of the animals have been humanely euthanised, and were not put down just for the research). This is perhaps the area that has changed the least since Richard Owen's times (1800s) when comparisons with modern animals helped him to realise that fossils of dinosaurs belonged to a completely different and now extinct (ignore birds) group. Whilst for some it seems incredibly disgusting, there is something indescribably fascinating about actually getting to see how the inside of animals works:
Tiger dissection at the early stages of identifying the muscles.
It would appear that a lot of people agree, as we ran a public dissection of a cheetah after hours in the vet college last year and we got hundreds of people attending over the sessions (more photos here).
The cheetah dissection being carried out for the audience, showing off various bits of the anatomy of the animal.
Which brings us nicely to something I spent a lot of time doing during my PhD, and still am actively involved in, which is outreach. Teaching the public (of all ages) about science is incredibly fun and rewarding. During my PhD, I taught over 1000 kids in various small classes in and around Bristol about the Bristol dinosaur (Thecodontosaurus) as well as interacting with thousands of people at various science festivals and events.
The junior school children are often the most easily enthused about dinosaurs. Who doesn't love dinosaurs when they are young? Why do so many grow out of it?
Since then I've co-curated museum exhibits, done talks, and recently have been back in schools as part of the DawnDinos project where we've been working through science and art to teach about evolution (see updates here).

We also do a lot of travels for work as conferences take us all over the world. Since my PhD I have been to Las Vegas NV, San Francisco CA, Barcelona, Raleigh NC, Berlin, Los Angeles CA, Dallas TX. Whilst conferences are a lot of work (attending talks, meeting people, presenting your research), there is always at least some time to go have an explore of the cities.

However, the things I enjoy as part of my job above all else are the days in the field. Working with animals is great, although cats are particularly hard work and scared(y).
Setting up forceplates, whilst being closely watched by a tiger.
Nothing compares, for me, to fossil hunting though. Those special few weeks a year (in a very good year) where I get to leave a lot of my usual work behind and just enjoy being in the wilderness, and finding some new fossils. It's a lot of hard work, but I'd do it a lot more if I got the chance to.
Last day of a long season in Dinosaur Provincial Park, we dug up a turtle and then hiked it back a few kilometres. The long sleeves was a mistake for this bit...
So those are the things I do as part of my job. I will also note my list above will be very different to other palaeontologists who have different interests, e.g. people who work on reconstructing phylogenies (family trees) will not do most of what I do and vice versa. I am lucky in that my job is so varied and I love it (or I most certainly wouldn't be doing it). I wonder how many people can say they look forward to going to work more often than not?

Wednesday, 24 May 2017

Reconstructing a fossil lion

I'm overdue for a post but am back with the best reason, a new paper is out! And for the first time in a while it is freely available as it is published in the open access journal Palaeontologia Electronica:

Cuff et al., 2017. Reconstruction of the musculoskeletal system in an extinct lion. Palaeontologia Electronica 20.2.23A: 1-25.

As my last paper post was also on Panthera atrox (or at least its brain) hopefully everyone already knows what I am talking about. If not, P. atrox is an extinct lion from North America. The species evolved from a "cave" lion (P. spalaea) population from Eurasia that crossed the Bering Sea around 340,000 years ago. These lion populations in turn split from the lineage that gives rise to the modern lions about 1.89 million years ago (Barnett et al., 2016).

However, despite this paper being about this cool fossil lion as the specimen of choice, it is more about the methods we are using to help reconstruct extinct animals. When we have only the bones, how do we estimate their masses?

Traditionally we estimate body masses of animals from their bones based on scaling equations (be it lengths, widths, circumferences etc.). Generally, the long/big bones of the limbs are useful as they tend to scale in a predictable manner with increases in body sizes. In mammals, skulls are also pretty good for this too. As such we can estimate masses of similar animals if we know the lengths of these bones. There is also a more modern method for estimating body masses called convex hulling. This method works by wrapping individual bones or regions (where you connect all the edges of each bone so no bone is left uncovered, basically imagine cling film/saran wrap), adding the volumes together and multiplying the volumes by a calculated density. This method has proved to be surprisingly robust, but is sensitive to the density used (much like the bone scaling is sensitive to which bone/metric). Additionally, whilst great for calculating a decent total mass, it will be highly inaccurate if you are trying to get accurate segment masses as it ignores all the soft tissues (e.g. the muscles which are large proportion of the mass of the limbs). So is there a way to estimate those bits more accurately?

So let us get back to our lion. La Brea Tar Pits is the source of "Fluffy" as our lion is affectionately known. Fluffy is the most complete specimen known to date, so we got all of the bones CT scanned. I then segmented the CT scans to have a lot of digital bones. I imported them into Meshmixer (although any software that you can manipulate 3D models would work) and aligned them into a natural pose. However, even the most complete P. atrox individual was not complete (or at least not completely scanned) and the Fluffy reconstruction was lacking the ribs, tail, hands and feet. These missing bits were replaced with equivalents from Asian lion (P. leo persica) that I scaled to match the expected size (based on scaling equations of head and femoral length). This gave us a skeleton:

Fig 2 from Cuff et al., 2017. Skeletal reconstruction showing the original bones from Panthera atrox and those which have been copied from other vertebrae (red), or from P. leo persica (blue). 1, lateral; 2, dorsal; 3, anterior views. Scale bar is 50 cm.
We had a predicted mass from the bones (195-219kg) and wanted to see what convex hull results would produce:
Fig 4. from Cuff et al., 2017. Convex hull model from the reconstructed Panthera atrox skeleton shown in Figure 2. 1, left lateral view; 2, dorsal view. Scale bar is 50 cm.
The volume reconstruction from the convex hull gave mass estimates of 180-219kg satisfyingly close to that predicted from the bones alone, but perhaps not unsurprising  as we scaled some important regions from the Asian lion to match the mass generated from the bones (it is a little circular). However, as discussed before you can see how small the limbs look, particularly the upper region compared to what an expected healthy animal would look like.

This is where we did something clever. Previous work we have done on how muscles scale in living cat species (see papers 1 and 2, or blog) gives us scaling equations for expected lengths and masses. The Asian lion which we borrowed the various missing skeletal elements from again came in useful. The CT scans of that specimen showed the muscles even without special staining so I was able to segment them out individually.

Fig. 1 from Cuff et al., 2017. CT scan slice showing an approximately mediolateral view (i.e., longitudinal section) of an Asian lion’s forelimb. 1, Dark grey is adipose and connective tissues, lighter grey is muscles, white is bone. Bottom right corner white is a density calibration phantom (1.69 g cm-3; “cortical bone”). 2, Segmentation of the lion forelimb with select muscles highlighted. Abbreviations: FCU - flexor carpi ulnaris; DDF - deep digital flexors; ECR - m. extensor carpi radialis; Pro Quad - m. pronator quadratus; Abd1 - m. abductor digiti I.
I then scaled them to the predicted mass of  P. atrox (we were working with 207kg as an average from the bone measures):

Mass P. atrox = Mass P. leo x length scale factor x width scale factor 2

Not all of the muscles could be isolated from the CT scans (particularly the vertebral ones) and neither could most of the tendons of the hands and feet, in both cases due to a lack of contrast on the CT scans. The tendons were reconstructed by creating a tube of material off of the muscle and extending it to the insertion point whilst maintaining the volume to match the predicted mass as best as possible. The result:
Fig. 3 from Cuff et al., 2017. Muscled reconstruction of Panthera atrox showing the major muscle groups in lateral view. Abbreviations: FCU - m. flexor carpi ulnaris; ECU - m. extensor carpi ulnaris; ECR - m. extensor carpi radialis; EDL - m. extensor digitorum longus. Scale bar is 50 cm.
If you overlay the muscled reconstruction on the convex hull you can see the differences, with muscles visible wherever they extend beyond the convex hull:

Fig. 5a from Cuff et al., 2017. Reconstructed muscles overlaid on the convex hull of just the bones. Any muscles that are visible extend beyond the range of the convex hull, thereby demonstrating the underestimation of size by convex hulls based solely on bone
These massive differences between the segmental masses are not a huge deal for overall mass calculations as the densities of the hulls have correction factors built in, but if we wanted to do more complex modelling this matters a lot (e.g. musculoskeletal modelling). Thus we needed to see if we could create an accurate segmental mass calculation.

We compared the mass estimates for models of the bones, muscles and bones, and convex hulls over the bones, and convex hulls over the muscles on an Asian lion to see which produced the most realistic results. (Un?)surprsingly, convex hull muscle models produce results that are really close to those for the actual lion data (whether that is segmental masses, moments of inertia and segmental centres of mass (See tables 4-6 in the paper)). This gave us confidence in the results for the P. atrox reconstruction and segmental masses which we will be using (hopefully) one day for some musculoskeletal models. The overall body centre of mass doesn't move much, but this is likely due to the fact that the body hull grossly overestimates actual mass as it does not deal with air spaces or soft tissues.
Fig. 5b from Cuff et al., 2017. Reconstructions showing the posteroventral movement of the centre of mass (COM) between the bone convex hull and the muscled convex hull models of Panthera atrox. Scale bar is 50 cm.
It's a cool method that works, and the results can be achieved using only freely available software (although in our paper we used a licensed segmentation software). It does have the important caveat that the muscle scaling is easily doable as the species in question falls within the phylogenetic bracket of those we have worked on. For species that fall outside of that range (e.g. Smilodon, the other large La Brea felid, which is far more distantly related) this scaling pattern is likely to be more uncertain, although a similar method would probably still produce reasonable results.