Monday 2 November 2015

Skeletal mass in birds

I've spent a lot of time on this blog rambling about estimating mass in extinct animals, including talking about the "lightweight" skeleton in birds, pterosaur bone mass, and the likelihood of giant pterosaurs weighing as low as 70 kg. Now I'm going to talk a bit more about this problem, specifically looking at the relationship between skeletal mass and total body mass in birds, the topic of my most recent paper.

In 1979, a paper came out looking at the relationship between skeletal mass and total body mass in birds, which was remarkably similar to that same relationship in mammals [1]. Since the two groups had such a similar relationship, the same was to estimate the total body mass in pterosaurs, after estimating the skeletal mass using simple geometric methods [2]. Because mammals and birds are so different and far apart in the evolutionary tree, it was thought that a similar relationship between the two meant that other animals like pterosaurs would share a similar relationship. Of course, I'm interested in pterosaur mass, so these methods are interesting to me.

During my MSc, my supervisor Colin Palmer and I looked at the original 1979 study a bit closer and found some slight problems with it. First of all, while they sampled a large number of bird taxa, each species average skeletal and body mass was determined from just 1-6 individuals, with most of them being just a single individual. How can they know this is a normal average weight? Additionally, the original data were presented in a log-log scale, showing a nice tight relationship with little variability. However, when the data were plotted on a linear scale, significant variability could be seen. This made us more interested in the topic.

When I started my PhD, my supervisor Gareth Dyke told me about a big dataset that his friend Gary Kaiser had meticulously collected on over 700 bird specimens from the Royal British Columbia Museum, which included total body mass and skeletal mass for over 400 individuals from 79 species. This dataset has more individuals than the original, but fewer species, meaning we had 1-30 individuals for each species, giving us a much better picture of average mass and variation within a species. With some help from an undergraduate student Ria McCann, and a lot of stats help from Orsi Vincze, we started to look at this dataset and saw some pretty interesting patterns, which we recently published in PLOS ONE [3]. First of all, our new dataset turned out to result in a pretty similar relationship to the original study, which was good news. We also found that there was even more variability within our dataset than the original study, which isn't surprising with a large number of individuals per species. For example within a single species, the rhinoceros auklet, total body mass varied from 0.4-0.6 kg (approximately 33%), while skeletal mass were varied by almost a factor of two. At a total body mass of about 470-490 g, the measured skeletons weighed in from 26 g to 34 g. This is a large range for a single species. Total body mass ranged from just 256 g, up to 616 g.
Variation in body mass and skeletal mass in the rhinoceros auklet.
Of course, that range could be due to age, and we thought it may be possible that age would affect these relationships. Unfortunately, it can be difficult to determine age of a bird if it is found dead (as most of these specimens were), so the only age classes we could determine was whether the bird was within it's hatchling year, or above that. However, we found no statistical difference between the two groups, suggesting that this feature does not change ontogenetically, which was a bit surprising. We also looked at males vs. females, as it was suggested that this could change things. We know that female birds regulate the amount of calcium in their bones depending on what cycle of egg-laying they are in, as they use the calcium from their eggs to make the hard shells. Again, however, we found no statistical differences between the two groups.
Sexual variation between male (blue) and female (red) birds
One thing I was most looking forward to testing was if there were any differences between different flight modes. Birds that fly in different ways having different body set ups with slightly different morphological adaptations, like longer, more slender wings for birds that soar. But are their skeletons built differently? For example, do birds that spend more time on ground, so-called 'burst-adapted' birds like pheasants and ptarmigans, have more robust skeletons that are maybe less 'light-weight' than traditional bird skeletons? Well at least between the 3 main flight modes we tested (soaring, continuous-flapping, and flap-gliding), they are not different. Unfortunately, we didn't have a large enough sample size of burst-adapted flyers to see if they were statistically different. I'd like to look at this more as my hunch is they will be different. I'd also be interested in seeing how passerines would be different, as passerines use a type of flying that is referred to as intermittent-bounding, where they kind of hop through the air with intermittent periods of flapping and not. Our new data set didn't have any passerines in it, which is fairly uncommon since they make up a large number of modern bird species.
Different flight modes - blue = soaring, red = flap-gliding, green = continuous flapping.
But why is this important? Well going back to the original question about mass estimation, we found that these results were correlated with phylogeny. This means that for a group of modern birds, like Neornithes, this relationship holds true. However, as you move further from this group, the relationship is going to be less and less supported. So moving into groups with no modern representatives like enantiornithine birds, which are significantly different from modern birds, or non-avian theropod dinosaurs, this relationship is going to be less accurate. Finally, moving all the way to pterosaurs, which are the sister group to dinosaurs, this relationship may not yield an accurate result, which is something that we hinted to in my first paper on pterosaur bone mass estimation [4].

I think that using the more traditional methods of volumetric mass estimation is likely to be more accurate for pterosaurs, for this reason, rather than using skeletal correlates as is becoming more common. Unfortunately, that requires more complete skeletons and a lot more work. Pterosaurs have no modern analogues or close relatives, suggesting that skeletal correlates are not going to work. Jon Tennant wrote a great post on this paper as well over at PLOS Paleo if you want to take a look!

Another thing I'd like to point it is the dataset we used. Gary Kaiser and Carl Jonsson collected a large amount of data on these specimens, including measurements of various skeletal elements, most of which we didn't use in this study. We know that a lot more information can be used from this dataset and many more interesting studies can use it, and we have posted the data up on the PLOS One website as supplementary material. We hope that someone can use the data for more in-depth studies like this! Please share if you know someone who could use it!

References:
1. Prange HD et al. 1979. Scaling of skeletal mass to body mass in birds and mammals. American Naturalist 113: 103-122.
2. Witton MP. 2008. A new approach to determining pterosaur body mass an its implications for pterosaur flight. ZittelianaB 28: 143-158.
3. Martin-Silverstone E, Vincze O, McCann R, Jonsson CHW, Palmer C, Kaiser G, Dyke G. 2015. Exploring the relationship between skeletal mass and total body mass in birds. PLOS ONE 10: e0141794.
4. Martin EG and Palmer C. 2014. A novel method of estimating pterosaur skeletal mass using computed tomography scans. Journal of Vertebrate Paleontology 34: 1466-1469.