How Walruses Got Their Tusks

Titanotaria, a tuskless walrus

Few people would argue that the walrus (Odobenus rosmarus) is not a distinctive animal. Indeed, it is so distinctive that, despite there being only one living species, that species has been placed in a "family" all of its own since 1880, distinct from both the seals and sea lions. 

There are a number of features of walruses that make it clear, from a modern perspective, that this was the right decision to make. But the most obvious is surely that they have huge tusks. These are in many ways a remarkable feature, and while other animals, such as elephants, also have large tusks, none that's alive today has anything quite like those of the walrus. 

Walruses have a long evolutionary history, having been distinct from their closest living relatives for many millions of years. Nonetheless, those relatives are, as we might expect, the seals and sea lions and it's apparent that they entered the water long before growing the tusks. Which, in turn, means that the first walruses must have been tuskless. So when and how did this change?

There are, in fact, rather a lot of fossil walruses, and the group was once wider and more varied than it is today. (One could argue, of course, that only the living species is technically a "walrus", and that the word I really want here is "odobenid". But I'm using a looser sense for clarity.) So, fortunately, we do have quite a bit to go on.

Fossil walruses from the Pliocene and Pleistocene, the two most recent epochs prior to the current one, all have teeth much like those of the living sort, including the tusks. Those from the Middle Miocene, however, do not. Between around 9 and 6 million years ago, during the Late Miocene, both tusked and tuskless walruses lived alongside one another, before the former eventually won out and drove the latter to extinction.

The fossil walruses that have tusks are similar in a number of other respects too, and it's generally agreed that large tusks only evolved once. Thus, we can group all the "tusked walruses" into a single group. The same cannot, however, be said for the tuskless walruses, which represent a number of different evolutionary branches, linked by the absence of a feature rather than the presence of one.

Nonetheless, if we want to trace the origin of tusks in the group, it is to these tuskless species that we should be looking. Somewhere, amongst all of those forms, there must have been, at some point, whatever species the tusked walruses evolved from. The question is, the fossil record being as incomplete as it is, have we found a candidate yet?

The last known tuskless walrus was Titanotaria, which is known to have lived off the coast of California from around 8 to 6 million years ago, towards the end of the Miocene. This does have enlarged canine teeth, but no more so than modern seals and sea lions - which are, after all, large carnivorous mammals. While it might be argued that it's hard to draw a specific line between "large tooth" and "tusk", these teeth were certainly not large enough to have protruded outside the lips when the mouth was closed, which is one definition we could use.

So, it's tuskless. But, not only did it live relatively late, it also, from other features of its skeleton, looks more like living walruses than most of its similarly tuskless relatives. But, since it lived alongside at least some indisputably tusked fossil walruses, such as Gomphotaria, it can hardly be their ancestor.

Recently, however, a review of a number of fossil walruses has been published, including some that have never been described before. Like those I've mentioned above, these were unearthed in California, in a rock formation south of Los Angeles. This has been a fruitful source of fossil walruses, which were evidently common along the Pacific coast of North America during the Miocene to Pliocene, with examples known from the southern tip of Baja California up to Vancouver Island.

Since the world was warmer then than it is now, this means that many of these species would have been living in subtropical conditions, and they were certainly all living much further south than the modern species is inclined to venture. We also, it should be noted, know of a number of similarly ancient walrus species that lived in Japan, and a few from the coasts of the Russian Far East. 

Some of the fossils covered by the review turn out to belong to previously known species of tuskless walrus, but three were apparently completely new. By comparing features of their skeletons with other known fossils, one thing the review was able to achieve was creating a tidier family tree for the group than previously. 

In the longer run, newer studies may contradict this and make it messier again, but this one at least manages to (among other things) define the dusignathine walruses as a single evolutionary line within the modern branch of the family. These are interesting, since they all have a pair of tusks in the lower jaw as well as the ones we'd expect in the upper jaw, but they're not, perhaps, the highlight of the study.

That would be the newly described species Osodobenus eodon. This has a number of features, including the shape of the regular teeth behind the tusks, that place it outside the "tusked walruses" on the family tree. And yet... it clearly had tusks. They certainly aren't huge, but they're still about three times the length that we see in, for example, Titanotaria, and much larger than any of the animal's other teeth. It's also clear that they would have continued to grow throughout life, which is what we'd expect from a modern walrus.

Living around 6 million years ago, it can't be a direct ancestor of the modern species, since earlier close relatives are known. But it may give us some idea of what a transitional form might have looked like, and it also gives us some clue as to how and why walruses bothered to grow tusks in the first place.

That's because its tusks are larger than those of some earlier species that we know really are closer to the direct ancestry of the modern walrus. But it does have some other, superficially unrelated, features that also match later tusked species, suggesting that these tend to go hand-in-hand. And one fossil skull of the species has much larger tusks than the other - if that's right, it could well be that one of them belonged to a male, and the other to a female.

The picture that we can piece together from this is roughly as follows. Early walruses did not have tusks, regardless of whether they were male or female. They didn't need them, any more than sea lions do. But, at some point, males started to develop larger canine teeth, perhaps at least partially because females found such things an attractive sign of a healthy potential partner.

But there's only so far one can grow canine teeth without them being a nuisance. Until that is, walruses changed the way that they fed. Instead of chasing and catching fish, they switched to bottom-feeding, sucking up food that they scraped off the seabed. We can see this in Osodobenus, because not only is its mouth the right shape for suction-feeding, but a specific opening in the skull just below the eye sockets is unusually large.

That's true in modern walruses, too, and that's because the nerves to the whiskers pass through it. And walruses (and, for that matter, manatees) use those whiskers to feel about on the seafloor. So we have an animal that probably fed much as modern walruses do, and suddenly not caring that the enlarged teeth got in the way of catching fish, but instead finding them an advantage, allowing them to scrape things up from the bottom. At which point, they became useful to females, too, and they raced to catch up.

This is, of course, just speculation. We can't know for certain what or how Osodobenus ate, still less whether it originally grew its tusks for a different reason. But it's a plausible theory, backed up by an animal that, whatever else it was, at least represented a halfway stage between walruses that looked like sea lions and walruses that looked like walruses.

[Picture from Magellanes et al. 2018, available from PeerJ under CC BY 4.0.]