Sunday, 26 October 2014

Children of the Coal-beasts

The animals of the Pleistocene epoch, of which I've given an overview over the last couple of years, are, on the whole, readily recognisable; they may look different from those today, but it's pretty obvious which modern animals they're related to. Mammoths are clearly elephants, sabretooths are clearly cats, and so on. The further we push back into the Age of Mammals, however, the less clear such things become.

There are two main problems here. One is that the animals simply look less like the ones we have today, especially if they come from a time before the modern groups separated from one another. How do we tell which modern group (if any) an animal belongs to if it lived before that group developed the characteristics that define it today? Secondly, older fossils are both rarer and more fragmentary, so that we're literally missing pieces of the puzzle. As a result, there are a number of early groups of mammal that we know existed, but which it's hard to fit into a family tree, or, indeed, to know much about at all.

One such group are the anthracobunids. The first fossils of animals belonging to this group were identified in 1940 from coal measures in the Punjab. It is from this that they take their name, since "Anthracobune" means "coal mound". Since then we have uncovered several more, all from around the same general area, but quite what they were remained something of a mystery. Did they leave any modern descendants, and, if so, what are they? And what were they like, anyway?

In reality, we can't know for certain whether there are "children of the coal-beasts" alive today. But we can at least try to say what they were most closely related to, which larger group they belonged to, and therefore, which modern animals at least might be descended from them, or something much like them. This is easier said than done (or they wouldn't be such a mystery in the first place), but we have to start with the little that we do know.

They lived during the early to middle Eocene epoch, around 55 to 40 million years ago, which places them early on in the Age of Mammals, and before many modern groups even existed. Until recently, the fossils we had for them were very incomplete: a few teeth, part of a lower jaw, a shoulder bone, a bone from the ankle... and little else. From this, we can at least make the rough guess that they were about the size of a large pig, and so hardly insignificant. We can tell from the teeth that they were likely herbivorous, and even the fact that they lived in modern-day India and Pakistan rules out some possibilities.

So, given the relatively large size, the herbivorous habits, and, indeed, some clues from that ankle bone, we can say that they were most likely related to the hoofed animals and their relatives. Unfortunately, that's not really telling us a lot, because that's a very wide assemblage of creatures, not all of which are that closely related.

The great family tree of the placental mammals (that is, excluding marsupials and monotremes) can be divided, on the basis of genetic and biochemical evidence, into four main branches. Two of them originated in the northern continents: one branch consisting of primates, rodents, and a few minor groups, and the other, known as the laurasiatheres, including, well... almost everything else. The remaining two branches are much smaller, at least today. These are the afrotheres, originating in Africa, and with the elephants as their most familiar representatives, and the xenarthrans, which are basically weird (sloths, armadillos, and anteaters), and which haven't moved far from their origins in South America.

Before we knew this, which means up until about the late 1990s, the large placental herbivores were thought to form a single group, with three divisions. There were the cloven-hoofed animals and their kin, the horses and their kin, and finally, the "subungulates" which were thought to be more primitive than the other two, and included elephants. Those three groups do still exist in modern schemes, but we now know that the so-called subungulates are afrotheres, and the others are all laurasiatheres - in other words, they aren't really related, and first appeared on different continents.

With no genetic data to go on, there's no definitive way of deciding which of those groups the anthracobunids are closest to. We have to go on the little we know of their physical appearance, and the general consensus was that they were, in some way, afrotheres. Whether they were closest to elephants, manatees, or to a now-extinct group of aquatic mammals, however, remained a mystery. There wasn't enough fossil evidence to say, and, honestly, they could have been a primitive group equally related to all three, for all we knew.

Over the last few years, however, a number of new fossils have been discovered, including some parts of skulls, complete lower jaws, and so on. This is sufficient to conduct a new analysis, adding in what we've recently learned about the creatures, and, perhaps to settle the question of what they really were.

The main focus of the paper is the family tree that the researchers have constructed, using a computer to compare over 400 different features among 93 different species, crunching the numbers to see which have the most in common. I've greatly simplified the result in the chart to the right, using English terms and cutting out most of the branches so that you can see the overall pattern, but you can see the full version here if you wish.

Even with computers, the answers you get when trying to create these family trees depend a lot on the assumptions that you feed in - which groups you're already confident are related, for example. The researchers tried a number of different inputs, but no matter what they tried, they couldn't make the coal-beasts shift from their position. The implication is clear: there had previously been a lot of debate about whether these animals were closer to the origin of elephants or the origin of manatees, and they're obviously neither.

In fact, they're not anywhere near those creatures, and instead seem to be related to the "odd-toed ungulates" - horses, tapirs, and rhinos. Any resemblance to primitive elephants is purely coincidental. (Mind you, since this would be before elephants had trunks, or were particularly large, the physical resemblance might well have been stronger at the time than you'd think). While odd-toed ungulates are known from North America quite early on, they appear to have originated in Asia, which, of course, would fit with the coal-beasts being found in India.

The new fossils have helped us in another way, too. Many of the bones in our bodies are composed of a solid outer sheath surrounding a spongier internal structure that encases the bone marrow. Newly discovered limb bones from anthracobunids show that this outer sheath was unusually thick for animals of their size. Among today's ungulates, only hippopotamuses have a similar proportion of outer to inner bone, although rhinos aren't too far behind.

Supported by isotope analysis of the animals' teeth, the authors conclude from this that the anthracobunids may have been closely associated with aquatic habitats. Not quite so much as hippos, to be sure, since they certainly ate land plants, but perhaps at least staying close to lakes and mud-wallows.

At any rate, the anthracobunids are, perhaps surprisingly, odd-toed ungulates, and not at all what we'd thought they were. Whether they have living descendants is harder to say. Probably not, since they seem rather too late to be ancestors of either horses, tapirs, or rhinos. But that doesn't mean that there were no "children of the coal-beasts" at all.

I mentioned above that a third theory, in addition to elephants and manatees was that the anthracobunids might be related to an extinct group of aquatic mammals. These were the desmostylians, vaguely manatee-like animals thought to be afrothere in origin, but actually living round the coasts of the Pacific, from Japan to California. And these, the new study suggests, may be genuinely related to the anthracobunids. Its not quite so definitive as the placing of the coal-beasts themselves, but, if it's right, then it's worth noting that the desmostylians did live much later, first appearing around 31 million years ago.

It's far from definitive, but it's at least possible that there's a genuine pattern here. The anthracobunids may represent an early line of odd-toed ungulates, rhino-like animals that ventured closer to the water, before eventually taking to the seas and living like herbivorous seals. The desmostylians left no descendants of their own, but it could be that they are the true children of the coal-beasts.

[Photo by David Waugh, Tobin Hieronymus. Copyright Cooper Lab, NEOMED, released under CC-BY creative commons licence. Cladogram adapted from Cooper et al 2014]


  1. That's fascinating-- the convergence between Perissodactyls and Proboscideans seems to be quite striking, and the question of which ones the Anthracobunids are closer to has been talked about a lot. If the A'ids are close to basal (or to stem) P'yls, this will also throw light on P'yl origins: another great problem has been that of which, if any, "Condylarths" might be close to the ancestry of the P'yls.

    Another great open question has been that of the affinities of the South American endemic "ungulates." There has apparently been recent progress on that one, too. Molecular evidence (not DNA, which has apparently not yet been extracted from them, but protein stuff) from fossils of late-surviving (Pleistocene) Notoungulates and Litopterns suggests that they may also be close to the Laurasiatherian ungulates (Art'yls and P'yls), and not (despite their Souther Hemisphere residence!) to Xenarthrans or Afrotheres. References in comments to a recent "Tetrapod Zoology" post:

    1. On the subject of the "condylarths" I didn't even have time in the above post to mention that the study also places Radinskya and Cambaytherium as possible stem perissodactyls, along with the rather better knownPhenacodus. I am, of course, eventually going to have to get round to notoungulates and litopterns in the Pliocene series, too...

  2. I've given the paper a first read. I was glad to see that it supports the idea that Phenacodus is on the P'yl stem: Phenacodus has to be about the best-known of all "conylarths," so if this placement holds up, it will give us some idea of what the ancestors of P'yls were like. They also put Radiskya close to the P'yls, which I am relieved to see. (Radinskya was, I think, first billed as a potential ancestor (or late-surviving relative of ancestors) of P'yls, but I think some people have suggested it might be an… Anthracobunid!)

    I'm not a professional, and I don't really know how to interpret the mathematics, but I have a feeling the evidence this paper finds isn't overwhelming: some strange changes in the placement of some taxa when different tree-constraints were tried. So I won't be utterly surprised (though I will be disheartened)if the next study, adding a few more taxa and a few more traits to the data matrix) comes up with a very different topology.