Glyptodonts: Armored sloths, not giant armadillos.

Today three taxa were added
to the large reptile tree (now 720 taxa, subset Fig. 1):

  1. Bradypus (three-toed sloth)
  2. Dasypus (armadillo)
  3. Glyptotherium (glyptodont)

Wikipedia reports, “Glyptodonts are an extinct subfamily of large, heavily armored armadillos. Unfortunately, Glyptotherium the glyptodont nests with Bradypus, the sloth in the large reptile tree – and, as you’ll see… for good reason.

Figure 1. Mammals with a few edentates added. Here glyptodonts nest with sloths, rather than armadillos, contra traditional studies.

Figure 1. Mammals with a few xenarthans added. Here glyptodonts nest with sloths, rather than armadillos, contra traditional studies.

Bradypus is the long-legged arboreal tree-hanging sloth
with fewer toes. Just imagine where the short-legged ground sloth will nest. This hypothesis of edentate relationships runs counter to tradition, but one look at the skulls (Figs. 1, 2) and you’ll wonder why this hasn’t been noticed before.

Figure 2. Bradypus skull from, used with permission, colors added to select bones.

Figure 2. Bradypus skull from, used with permission, colors added to select bones. Compare to the glyptodont skull in figure 3. Darker images bring out lighter details.

Both the sloth and the glyptodont
have deep, narrow, flat-topped skulls with a deep ventral process of the jugal and a very short rostrum. These traits, along with a long list of others, split armadillos + aardvarks from sloths + glyptodonts.

Figure 3. Glyptodon skull with select bones colorized. Compare to the sloth in figure 2 and the armadillo in figure 4.

Figure 3. Glyptodon skull with select bones colorized. Compare to the sloth in figure 2 and the armadillo in figure 4.

Here armadillos nest with
aardvarks, like Orycteropus, a clade that was once accepted under the clade Xenartha. According to Wikipedia, xenarthans have extra vertebral articulations, the lowest metabolic rates among therians, the ischium and sacrum are fused and males have internal testicles. Pangolins were once considered xenarthans, but here nest with basal primates.

Figure 4. Skull of the armadillo Dasypus from, used with permission. and select bones colorized.

Figure 4. Skull of the armadillo Dasypus from, used with permission. and select bones colorized.


Note the nesting of 
whales (Maiacetus) + tenrecs (Hemicentetes), another clade without external testicles. As a scientist, I’m pleased to discover these interrelationships, but a little surprised that no one has seen this before. If you that has happened already, I will be glad to promote those papers here.

Placental mammals
are showing a basal split between carnivores and other insectivore / herbivore placentals (eutherians).


14 thoughts on “Glyptodonts: Armored sloths, not giant armadillos.

  1. You know we have glyptodont DNA (Delsuc et al., 2016), and thus know* not only are they armadillos, they’re members of the modern family Chlamyphoridae? Ditto for cetaceans not being tenrecs and instead being artiodactyls, which have indeed all been analyzed numerous times together for numerous genes. What you’re doing here (as with squamates before) is saying molecular phylogenetics doesn’t work, and thus the thousands of workers and studies in the field are misled. Does that sound likely? Or is it more likely that because you always find a very different topology than the consensus of thousands of experts across the world, no matter what group you look at (pterosaurs, dinosaurs, synapsids, mammals, etc…), that you are the one that is wrong instead of everyone else?

    * “Know” because the support values nesting them there are the highest possible.

  2. Mickey, I did not know that factoid about glypto DNA. To that point, in an ideal world DNA would match up with morphology and turtles would be archosaurs, etc. etc. I challenge you, or any of your [no exaggeration] ‘thousands’ of experts to find the archosaur outgroup for turtles in a morpho analysis. Show me the gradual accumulation of turtle traits in a series of archosaur outgroups. It can’t be done. And so we’re at an impasse. I can only report on morphology. And I show my work. The evidence I show above doesn’t sway you or encourage you to question DNA? Hmmm. DNA is great within a species, or genus. Might be too much to ask it to lump and split larger clades.

    • So you really think that the thousands of molecular phylogenetics workers (who I never specified all worked on turtle relationships) are wasting their time on suprageneric analyses? Just look at the Journal of Molecular Evolution- that’s several thousand pages a year, written by people who all believe what you think is an incorrect premise. Which is why accepting your results over Delsuc et al.’s isn’t merely “questioning DNA”, it’s rejecting the foundational principle of molecular phylogenetics. Is it really likely the entire field is wrong and you’re right?

      Morphological analyses agree with molecular analyses on some points, but even huge and well done examples (like Zack for mammals, or Gauthier for squamates) don’t recover many of the molecular results. In cases of disagreement, I’ll choose well-supported molecular results every time. Because when they’re strong, the results of different genes in both the nucleus and mitochondria, insertions, replication errors, pseudogenes, etc. reinforce themselves. As taxon and character number increase in molecular analyses, the results are solidified. And these results are often backed up by some morphological characters, biogeography, etc.. That’s science- the convergence of independent data to support a hypothesis.

    • When morphology disagrees, it’s often plausibly explained by convergence or reversals, which everyone agrees happens. In the present case, the fact both glyptodonts and sloths chew plants while armadillos are insectivores explains e.g. the ventral jugal processes. Whereas what makes sloth and glyptodont mitochondrial genetics similar? And why would that covary with all the other molecular variables, many of which don’t even code for anything? There’s no plausible mechanism for this happening in every organism we look at. So faced with backing an entire field of science with convergence and reversals as a plausible explanation for conflicting data, versus backing yours or anyone’s morphological analysis with no plausible explanation for conflicting data, I choose the former.

  3. Morphology trumps genetics. Ultimately you’re going to have to show a gradual accumulation of traits. Snakes still have the gene for legs. Humans still have the gene for gills and a tail.

  4. One other thought… when testing DNA you’re only testing part of the genome, never the whole several dozen chromosomes. It’s still a mystery how they interact and express themselves… or not.

    • “Morphology trumps genetics”

      So I guess you are throwing out an entire field of study. How many scientists do that and turn out to be right?

      And actually no, as sequencing gets easier and cheaper, molecular workers are increasingly analyzing the entire genomes (nuclear, mitochondrial and/or chloroplast depending on the taxa studied). These are called phylogenomic analyses, and a recent mammal example is Tsagkogeorga et al. (2013) which found the same basic molecular tree as everyone else, with e.g. the dolphin closer to the cow than the llama relative.

      Similarly, there is no “gene for legs”, chromosomal interaction is not “a mystery”, and that hole in front of the eye you noted in your whale-tenrec post as a potential grad project is the infraorbital foramen which carries the trigeminal nerve, etc.. There are so many scientists who study and understand these issues, and molecular results are seen as more accurate even by most biologists who work with morphology or fossils. Somebody calling elephants ungulates instead of afrotheres is seen the same way as someone who is against cladistics, uses paraphyletic groups, etc..

      Reference- Tsagkogeorga et al., Phylogenomic Analyses Elucidate the Evolutionary Relationships of Bats, Current Biology (2013),

    • That’s apoor argument David, when “testing” morphology, you are including a number of characters that represent a subset of the whole organism. It’s practically impossible to include every single trait of an organism into a matrix for phyl. analyses. On the other hand IT IS NOT entirely a mistery how genes interact and express themselves. We’ve made gigantic advances in our understanding of molecular biology and evolution, I invite you to research some more about this issue. As such, saying molecular data is not suitable for suprageneric phylogenetics is a somewat random argument, what’s the scientific basis for such a claim? Again, we’ve made huge progress understanding molecukar evolution and we now they’re a useful tool. If they are more widely used than morphology for phylogenetics of extant taxa, it’s for a well established reason, not because we arbitrarily decided to.

  5. Not a random argument. Like Galileo, what I report comes after testing. When I review the recovered taxa and look at the traits I see the gradual accumulation of traits. It’s all very clear. You can do this yourself. Morphology does not, unfortunately, recover the same interrelations as DNA. I don’t know why. This weekend I shall do more reading, but that won’t change the tree topologies. Thank you for your thoughts. I have access to morph data. I leave DNA to others.

  6. Thanks for your reply David. What I was criticizing was your argument AGAINST molecular (DNA) phylogenetics. I wasnt criticizing your methods (although that does not mean I agree with them, but that’s another issue). Morphology SOMETIMES recover similar topologies than DNA actually, and when not, it’s been acknowledged that morphological traits are prone to homoplasies that can easily pass undetected to the human eye, especially at the time of establishing homology and coding the characters for a phylogenetic analysis. That is the reason why they are in conflict sometimes. Yes, there’s a lot of literature about this issue, I hope you enjoy it if you’re going to read about it.
    But again, my previous post was motivated by your claim about the supposedly inferior molecular methods (you said morph trumps genetics), I still havent mentioned anything about your methods.


    • Andres, I agree that DNA works over short phylogenetic distances. It does not seem to do so over larger distances. re: homoplasies: After a taxon nests, I look again at every character for typos (miskeyed scores) and observational errors usually appearing as autapomorphies. At that time I get to read the homoplasies that the scores reveal and it all makes sense. I would be the first person to reexamine the data if it were not so.

  7. The only armadillos in Peters’s dataset are Dasypus and Glyptodon. I have to wonder if taxon exclusion is affecting the results here.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.