Although it is certainly not obvious from their present diversity—just 6 species in two genera—the tree sloths of Central and South America represent the last vestiges of a major experiment in placental evolution. The living two- and three-toed sloths are exclusively tropical, but during favourable intervals in the Pleistocene their extinct relatives were distributed from Alaska to southernmost South America. In earlier times sloths reached the West Indies, and possibly even Antarctica. 

These early sloths came in many shapes and sizes, most of them very different from the tree sloths of today. Some were giant, like Megatherium; the terminal species of this genus were the size of elephants. At least one lineage, the thalassocnines, became thoroughly aquatic and foraged in the seas off the west coast of South America. Most groups progressively declined during the late Cenozoic, eventually leaving only the tree sloths and—surprisingly—a few species in the Greater Antilles that finally disappeared only five or six thousand years ago.

Systematic biologists have pondered the relationships of living and extinct sloths for almost 250 years. But the story is complicated: although palaeontologists have discovered a startling array of fossil species, a number of finds are difficult to fit into existing taxonomic schemes. Until recently, all such efforts were based on morphology; now, however, ancient biomolecules have begun to play an important role in testing old ideas and proposing new ones. 

As a palaeomammalogist, I have long been interested in the use of ancient DNA in systematic studies of late Quaternary mammals. However, because efforts to gain genomic information from sloth fossils had been largely unsuccessful, I wondered whether we could make progress using a quite different source of molecular information, namely the structural proteins that are normal components of bones and teeth. If well preserved, the amino acids comprising such proteins can be sequenced, providing information in a form that can be easily adapted for use in phylogenetic studies. 

With the help of palaeoproteomics expert Matthew Collins (Universities of Copenhagen & Cambridge), an analytics team was put together consisting of Samantha Presslee (University of York), Graham Slater (University of Chicago), Francois Pujos (CONICET, Mendoza), and Brian Chait (Rockefeller University), with the active support of many palaeontologists who supplied both specimens and ideas. Our protein of choice was “bone” collagen, which makes up almost all of the organic component of vertebrate bones and is often present in quantity in well preserved fossils.

Representatives of major sloth groups. In the tree, going clockwise from left: extinct West Indian sloth (Acratocnus), extant three-toed sloth or ai (Bradypus), and extant two-toed sloth or unau (Choloepus). From left to right below are representative giant ground sloths: Mylodon (South America), Megalonyx (North America), and Megatherium (South America). According to new molecular studies, Bradypus is most closely related to a group that includes Megalonyx and Megatherium. Choloepus is related to Mylodon, but Acratocnus and its West Indian kin represent a separate branch that is distinct from that of all other sloths. Human for scale: 1.8.5 m (6 ft 1 in). Credit: Jorge Blanco

Getting suitable results was time consuming, but by 2018 we had managed to cover all the groups that had survived into the last part of the Cenozoic. And what we found from our phylogenetic analyses of their sequence data was quite remarkable. The details are presented in our Nature Ecology and Evolution paper, but there is one discovery that I would like to highlight here. I was fascinated by a proposal made many years ago by an AMNH colleague, the ichthyologist Donn Rosen, who argued that some of the larger islands in the West Indies had been briefly connected with northern South America as a consequence of the tectonic motion of the Caribbean Plate. Manuel Iturralde-Vinent, a geologist from the Cuban Academy of Sciences, and I put this idea into a testable form as the GAARlandia hypothesis. In both Cuba and Puerto Rico we discovered remains of early sloths that seemed consistent with the idea that their ancestors reached the Greater Antilles over a landbridge some 33-34 million years ago. 

At the time this conclusion was controversial because our supporting fossil evidence was slim. Imagine our delight when analysis of the palaeoproteomic evidence indicated that the Antillean sloth radiation had diverged from all related lineages about 30 million years ago, making it the sister-group of other sloths. Much remains to be done to confirm this finding and its relevance to paleobiogeography, but it has already found support in an ancient DNA study led by Frederic Delsuc (University of Montpellier) and Hendrik Poinar (McMaster University), in which I also participated. This study relied on mitogenomic evidence, whereas the genes specifying collagen are nuclear. 

That the two sets of results agreed in detail, despite varying substitution rates in different parts of the total genome, is certainly meaningful, but it is not definitive. Morphology will always be critically important in palaeontological studies, and molecular evidence is unlikely ever to be available for the vast majority of species. Even so, given that there is only one history of life, ancient biomolecules will surely have an important role to play in bringing that history to light in future.