What were Mesozoic mammals eating?

Using jaw shape and biomechanics to understand the diets of our earliest mammalian relatives

Like Comment
Read the paper

Ever since I started my PhD, I have been amazed at how much of the history of mammals took place alongside dinosaurs. We tend to think of the Mesozoic (252-66 mya) as the Age of Reptiles and the Cenozoic (66-0 mya) as the Age of Mammals, but almost two-thirds of the evolutionary history of mammals took place during the Mesozoic (Kielan-Jaworowska et al., 2004; Kemp, 2005). Since their origin (approximately 225 million years ago) mammals have diversified into thousands of species with hugely varying body sizes, behaviours, habitats, diets, morphologies, and modes of locomotion. All the mammals alive today can trace their ancestry back to the Mesozoic (Kielan-Jaworowska et al., 2004; Kemp, 2005). Knowing about Mesozoic mammals is hugely important to understanding where we came from; however, there are dozens of aspects that we could study about these mammals- after all, hundreds of species lived around the world during the Mesozoic. 

That’s why my co-authors and I decided to focus on just one part of the story: inferring their diets based on their jaws. Why use jaws? Well, the reasons are twofold: 1) after teeth, jaws are one of the most commonly preserved fossils of Mesozoic mammals and they’re easily accessible in the literature and in online databases, 2) the main function of jaws is feeding, so we would expect differences in jaw morphology to be indicative of dietary differences. In fact, jaw morphology and function have long been studied in relation to diet in living and extinct animals. So, how do we go about capturing jaw shape and function?

Let’s start with jaw shape. A very popular approach for capturing jaw shape is geometric morphometrics; in our case, we chose 2D geometric morphometrics because of two reasons: 1) the jaws in our sample were relatively flat mediolaterally, and 2) photographs of jaws are widely available at little or no cost. By using this method, we were expecting to quantitatively capture differences in jaw shape among Mesozoic mammals and, by including a comparative sample of modern mammals, we were hoping to gauge what those differences might tell us about fossil mammal diets. 

In 2013, Grossnickle and Polly found differences in jaw shape between small herbivorous and faunivorous (i.e., insectivorous and carnivorous) modern mammals. When Mesozoic mammals were added to the mix, the authors found that most multituberculate (i.e., Jurassic-Eocene mammals with complex jaws and unique chewing movements) jaws resembled modern herbivores, while non-multituberculate species were more similar to faunivorous mammals. 

PCA scatter plot of jaw shape, showing convex hulls of modern herbivores on the left side and modern insectivores and carnivores on the right side. Mesozoic mammals are projected on top, showing multituberculates on the left and other Mesozoic mammals on the right side.

PCA scatter plot of jaw shape, showing convex hulls of modern herbivores on the left side and modern insectivores and carnivores on the right side. Mesozoic mammals are projected on top, showing multituberculates on the left and other Mesozoic mammals on the right side. Modified from Grossnickle and Polly, 2013.

We took Grossnickle and Polly's (2013) paper as a starting point, but chose to focus only on non-multituberculates: as many of these mammals are often described as "generalised insectivores", we wanted to see if we could find any finer-detail differences between faunivorous species that could help us generate better inferences of diet in Mesozoic mammals. But before we could start running any analyses, we needed the data. For this, I got to visit many museum collections around Europe in order to photograph as many Mesozoic mammals as I could! I got the chance to visit the Oxford Museum of Natural History (UK), the Natural History Museum in London (UK), the Institute of Paleobiology (Polish Academy of Sciences, Warsaw, Poland), and the Steinmann Institut of the University of Bonn, Germany. Using these photographs and many others from previously published books, papers and online databases, I set out to capture their jaw shapes.

Photographs from the Oxford University Museum of Natural History, showing the museum interior on the left and two Mesozoic mammals on the right: Palaeoxonodon and Phascolotherium
Oxford University Museum of Natural History, showing main gallery on the left and two Mesozoic mammal jaws held in the museum on the right: Palaeoxonodon and Phascolotherium.

What did we find? Among modern mammals, we found a very good discrimination between herbivores, carnivores and insectivores, largely based on differences in jaw length.

PCA scatter plot of jaw shape, showing shorter jaws on the left, longer jaws on the right, taller ascending rami towards the top and shorter ascending rami towards the bottom. Convex hulls of modern herbivores on the left, carnivores in the middle, and insectivores on the right. The convex hull of omnivores overlaps all other dietary categories.

PCA scatter plot of jaw shape, showing convex hulls of modern herbivores on the left, carnivores in the middle, and insectivores on the right. The convex hull of omnivores overlaps all other dietary categories.

We added Mesozoic mammals next and found that previous hypotheses of their likely diets (based on body size data, dental type, etc.) matched up pretty well with our results, which means that jaw shape is a good indicator of diet in Mesozoic mammals!

PCA scatter plot of jaw shape, showing shorter jaws on the left, longer jaws on the right, taller ascending rami towards the top and shorter ascending rami towards the bottom. Mesozoic mammals projected on top: most plot on insectivore morphospace and some on carnivore morphospace, as previously hypothesised. Some Mesozoic mammals plot near herbivore morphospace.

PCA scatter plot of jaw shape, with Mesozoic mammals projected on top of the convex hulls of modern mammals. Most Mesozoic mammals plot on insectivore morphospace and some on carnivore morphospace, as previously hypothesised. Only a few Mesozoic mammals plot near herbivore morphospace.

We went a step further and performed a discriminant analysis that took into account the way the mammals in our sample were related to each other (i.e., their phylogenetic relationships), and we used a training dataset of modern mammals. When you perform an analysis like this you are basically telling the software: “This is what a herbivore looks like, this is what a carnivore looks like, and this is what an insectivore looks like. Using this information, what do you think the diet of this fossil mammal was?”. The analysis revealed (with 89% confidence) that most Mesozoic mammals were insectivores, including most early mammals, symmetrodontans, dryolestids, amphitheriids and eutherians (i.e., the group that includes modern placentals, like humans), some Mesozoic mammals were carnivores, including most eutriconodontans and metatherians (i.e., the group that includes modern marsupials, like kangaroos), and only a couple of taxa were classified as herbivores.

Plot showing the results of the phylogenetic discriminant analysis, which classified Mesozoic mammals based on their dietary categories with about 89% confidence. Most stem mammals, symmetrodontans, dryolestids, amphitheriids and eutherians were classified as insectivores. Most eutriconodontans and metatherians were classified as carnivores. Vincelestes and Haldanodon (with weak support) were classified as herbivores.
Plot showing the results of the phylogenetic discriminant analysis, which classified Mesozoic mammals based on their dietary categories with about 89% confidence. Most stem mammals, symmetrodontans, dryolestids, amphitheriids and eutherians were classified as insectivores. Most eutriconodontans and metatherians were classified as carnivores. Vincelestes and Haldanodon (with weak support) were classified as herbivores.

What about jaw function? Well, mammal jaws can be treated as levers, as they are beam-like bones that rotate around a fixed point and transmit energy from one end of the beam to another. Levers transmit a force input into the system (e.g., jaw adductor muscle input force) to provide a greater output force (e.g., bite force) to displace a load (e.g., food). The ratio of these two forces is known as mechanical advantage. In general terms, jaws with high mechanical advantage produce forceful, but slow bites, while jaws with low mechanical advantage produce quick, but weak bites. Mechanical advantage is a functional metric often associated with diet in mammals. 

Biting action of sugar glider (left) and bandicoot (right) shown side by side, with temporalis muscle depicted. Sugar glider has high mechanical advantage and therefore a powerful but slow bite. The bandicoot has low mechanical advantage and therefore a weaker but quicker bite.
Simulated biting action of a sugar glider (left) and a bandicoot (right), demonstrating differences in mechanical advantage between these two mammals. The sugar glider has a shorter jaw with higher mechanical advantage, therefore it has a slow but powerful bite. The bandicoot has a longer jaw with low mechanical advantage, therefore it has a weaker but quicker bite.

We calculated the mechanical advantage of the main two jaw-closing muscles in mammals: the masseter and the temporalis. We found that small modern mammals can be roughly discriminated by these values: insectivores have low mechanical advantage in both muscles, herbivores have high mechanical advantage in both muscles, and carnivores have low masseter mechanical advantage and high temporalis mechanical advantage.

Scatter plot showing masseter mechanical advantage on x axis and temporalis mechanical advantage on y axis. Modern insectivores have low mechanical advantage of both muscles and modern herbivores have high mechanical advantage of both muscles. Carnivores have low mechanical advantage of the masseter and high mechanical advantage of the temporalis.

Scatter plot showing masseter mechanical advantage on x axis and temporalis mechanical advantage on y axis. Modern insectivores have low mechanical advantage of both muscles and modern herbivores have high mechanical advantage of both muscles. Carnivores have low mechanical advantage of the masseter and high mechanical advantage of the temporalis. 

Mesozoic mammals largely play by the same rules and can be largely identified using mechanical advantage values. Stem mammals are an exception: their angular process is anteriorly positioned (because they still retained a group of tiny bones in the back of the jaw, which eventually became the middle ear bones in later mammals) and thus have higher-than-expected masseter mechanical advantage. 

Scatter plot showing masseter mechanical advantage on x axis and temporalis mechanical advantage on y axis. Most Mesozoic mammals have mechanical advantage values as expected from their hypothesised dietary categories. Stem mammals with anteriorly positioned angular processes have higher than expected masseter mechanical advantage.

Scatter plot showing masseter mechanical advantage on x axis and temporalis mechanical advantage on y axis. Most Mesozoic mammals have mechanical advantage values as expected from their hypothesised dietary categories. Stem mammals with anteriorly positioned angular processes have higher than expected masseter mechanical advantage. 

Our study provides a methodological basis for inferring diet in Mesozoic mammals and, in the future, perhaps in small Cenozoic mammals. The results of our study support the idea that most Mesozoic mammals were insectivorous and only a few were carnivorous.

References

  1. Kielan-Jaworowska, Z., Cifelli, R. L., & Luo, Z. X. (2005). Mammals from the Age of Dinosaurs: Origins, Evolution, and Structure. Columbia University Press.
  2. Kemp, T. S. (2005). The origin and evolution of mammals. Oxford University Press on Demand.
  3. Grossnickle, D. M., & Polly, P. D. (2013). Mammal disparity decreases during the Cretaceous angiosperm radiation. Proceedings of the Royal Society B: Biological Sciences280: 20132110.

Nuria Melisa Morales Garcia

PhD student, University of Bristol

Comments

Go to the profile of Connall Darlington
5 months ago

I'm currently writing a report on a similar topic (quantifying the evolution of jaw form/function) in my first year undergrad, and this post plus the paper were incredibly useful! Really nicely laid out and super informative, thank you so much!

Hi Connall! Thanks for your comment, I'm so glad you found this useful!!