The move from water to land launched a major redesign of every aspect of life, from breathing to reproduction: it was a great trek through biological space. Nevertheless, with what seems almost wanton perversity, a good number of thoroughgoing land animals later turned around, abandoned their hard-earned terrestrial retooling, and trooped back into the water again. Seals and sea lions have only gone part-way back. They show us what the intermediates might have been like, on the way to extreme cases such as whales and dugongs. Whales (including the small whales we call dolphins), and dugongs with their close cousins the manatees, ceased to be land creatures altogether and reverted to the full marine habits of their remote ancestors. They don't even come ashore to breed. They do, however, still breathe air, having never developed anything equivalent to the gills of their earlier marine progenitors. Other animals that have returned from land to water, at least some of the time, are pond snails, water spiders, water beetles, crocodiles, otters, sea snakes, water shrews, Galapagos flightless cormorants, Galapagos marine iguanas, yapoks (aquatic marsupials from South America), platypuses, penguins and turtles.
Whales were long an enigma, but recently our knowledge of whale evolution has become rather rich. Molecular genetic evidence (see Chapter 10 for the nature of this kind of evidence) shows that the closest living cousins of whales are hippos, then pigs, then ruminants. Even more surprisingly, the molecular evidence shows that hippos are more closely related to whales than they are to the cloven-hoofed animals (such as pigs and ruminants) which look much more like them. This is another example of the mismatch that can sometimes arise between closeness of cousinship and degree of physical resemblance. We noted it above in connection with fish that are closer cousins to us than they are to other fish. In that case, the anomaly arose because our lineage left the water for the land, and consequently surged away in evolution, leaving our close fish cousins, the lungfish and coelacanths, resembling our more distant fish cousins because they all stayed in the water. Now we meet the same phenomenon again, but in reverse. Hippos stayed, at least partly, on land, and so still resemble their more distant land-dwelling cousins, the ruminants, while their closer cousins, the whales, took off into the sea and changed so drastically that their affinities with hippos escaped all biologists except molecular geneticists. As when their remote fishy ancestors originally went in the other direction, it was a bit like taking off into space, or at least like launching a balloon, as the ancestors of whales floated free of the constraining burden of gravity and severed their moorings to dry land.
At the same time, the once rather scanty fossil record of whale evolution has been convincingly filled out, mostly by a new trove from Pakistan. However, the story of fossil whales has been so well treated in other recent books, for example Donald Prothero's Evolution: What the Fossils Say and Why it Matters, and, more recently, Jerry Coyne's Why Evolution is True, that I have decided not to cover the same details here. Instead, I have confined myself to one diagram (below), taken from Prothero's book, showing a sequence of fossils ordered in time. Note the careful way the picture is drawn. It is tempting -and older books used to do this - to draw sequences of fossils with arrows from older to younger ones. But nobody can say, for example, that Ambulocetus was descended from Pakicetus. Or that Basilosaurus was descended from Rodhocetus. Instead, the diagram follows the more cautious policy of suggesting that, for example, whales are descended from a contemporary cousin of Ambulocetus which was probably rather like Ambulocetus (and might even have been Ambulocetus). The fossils shown are representative of various stages of whale evolution. The gradual disappearance of the hind limbs, the transformation of the front limbs from walking legs to swimming fins, and the flattening of the tail into flukes, are among the changes that emerged in elegant cascade.
Figure 14.16. Evolution of whales from land creatures, showing the numerous transitional fossils now documented from the Eocene beds of Africa and Pakistan. (Drawing by Carl Buell)
Figure 14.16. Evolution of whales from land creatures, showing the numerous transitional fossils now documented from the Eocene beds of Africa and Pakistan. (Drawing by Carl Buell)
That's all I'm going to say about the fossil history of whales, because it has been so well treated in the books I mentioned. The other, less numerous and diverse but just as thoroughly aquatic group of marine mammals, the sirenians - dugongs and manatees - are not so well documented in the fossil record, but one outstandingly beautiful 'missing link' has recently been discovered. Roughly contemporary with Ambulocetus, the Eocene 'walking whale', is Pezosiren, the 'walking manatee' fossil from Jamaica. It looks pretty much like a manatee or dugong, except that it has proper walking legs both front and rear, where they have flippers in the front and no limbs at all in the rear. The picture opposite shows a modern dugong skeleton above, Pezosiren below.
Just as whales are related to hippos, so sirenians are related to elephants, as a great deal of evidence, including the all-important molecular evidence, attests. Pezosiren, however, probably lived like a hippo, spending much of its time in water and using its legs to walk on the bottom as well as swim. The skull is unmistakably sirenian. Pezosiren may or may not be the actual ancestor of modern manatees and dugongs, but it is certainly excellent casting for the role.
This book was about to go to the printer when exciting news came in, from the journal Nature, of a new fossil from the Canadian Arctic, plugging a gap in the ancestry of modern seals, sea lions and walruses (collectively 'pinnipeds'). A single skeleton, about 65 per cent complete, Puijila darwinidates from the early Miocene epoch (about 20 million years ago). That's recent enough that the map of the world was almost the same as today. So this early seal/sea lion (they had not diverged yet) was an Arctic animal, a denizen of cold water. Evidence suggests that it lived and fished in fresh water (like most otters except the famous sea otters of California), rather than in the sea (like most modern seals except the famous Lake Baikal seal). Puijila did not have flippers, but webbed feet. It probably ran like a dog on land (very unlike a modern pinniped) but spent much of its time in water, where it swam like a dog, unlike either of the two styles adopted respectively by modern seals and sea lions. Puijila neatly straddles the gap between land and water in the ancestry of pinnipeds. It is yet another delightful addition to our growing list of 'links' that are no longer missing.
I now want to turn to another group of animals that returned from the land to the water: a particularly intriguing example because some of them later reversed the process and returned to the land a second time! Sea turtles are, in one important respect, less fully given back to the water than whales or dugongs, for they still lay their eggs on beaches. Like all vertebrate returners to the water, turtles haven't given up breathing air, but in this department some of them go one better than whales. These turtles extract additional oxygen from the water through a pair of chambers at their rear end that are richly supplied with blood vessels. One Australian river turtle, indeed, gets the majority of its oxygen by breathing (as an Australian would not hesitate to say) through its arse.
Before going any further, I can't escape a tiresome point of terminology, and a regrettable vindication of George Bernard Shaw's observation that 'England and America are two countries divided by a common language.' In Britain, turtles live in the sea, tortoises live on land and terrapins live in fresh or brackish water. In America all these animals are 'turtles', whether they live on land or in water. 'Land turtle' sounds odd to me, but not to an American, for whom tortoises are the subset of turtles that live on land. Some Americans use 'tortoise' in a strict taxonomic sense to refer to the Testudinidae, which is the scientific name for modern land tortoises. In Britain, we'd be inclined to call any land-dwelling chelonian a tortoise, whether it is a member of the Testudinidae or not (as we shall see, there are fossil
'tortoises' that lived on land but are not members of the Testudinidae). In what follows, I'll try to avoid confusion, making allowance for readers in Britain and America (and Australia, where the usage is different again), but it's hard. The terminology is a mess, to put it mildly. Zoologists use 'chelonians' for all these animals, turtles, tortoises and terrapins, whichever version of English we speak.
The most instantly noticeable feature of chelonians is their shell. How did it evolve, and what did the intermediates look like? Where are the missing links? What (a creationist zealot might ask) is the use of half a shell? Well, amazingly, a new fossil has just been described, which eloquently answers that question. It made its debut in the journal Nature in the nick of time before I had to hand this book over to the publishers. It was an aquatic turtle, found in late Triassic sediments in China, and its age is estimated at 220 million years. Its name is Odontochelys semitestacea, from which you may deduce that, unlike a modern turtle or tortoise, it had teeth, and it did indeed have half a shell. It also had a much longer tail than a modern turtle or tortoise. All three of these features mark it out as prime 'missing link' material. The belly was covered by a shell, the so-called plastron, in pretty much the same way as that of a modern sea turtle. But it almost completely lacked the dorsal portion of the shell, known as the carapace. Its back was presumably soft, like a lizard's, although there were some hard, bony bits along the middle above the backbone, as in a crocodile, and the ribs were flattened, as though 'trying' to form the evolutionary beginnings of a carapace.
And here we have an interesting controversy. The authors of the paper that introduced Odontochelys to the world, Li, Wu, Rieppel, Wang and Zhao (for brevity, I'll call them the Chinese authors, although Rieppel is not Chinese), think that their animal was indeed halfway towards acquiring a shell. Others dispute Odontochelys s claim to demonstrate that the shell evolved in water. Nature has the admirable custom of commissioning experts other than the authors to write a commentary on the week's more interesting articles, which they publish in a section called 'News and Views'. The 'News and Views' commentary on the Odontochelys paper is by two Canadian biologists, Robert Reisz and Jason Head, and they offer an alternative interpretation. Maybe the whole shell had already evolved on land, before Odontochelys s ancestors went back to the water. And maybe Odontochelys lost its carapace after returning to the water. Reisz and Head point out that some of today's sea turtles, for example the giant leatherback turtle, have lost or greatly reduced the carapace, so their theory is quite plausible.
I need to digress for a brief aside on the question, 'What is the use of half a shell?' In particular, why would Odontochelys be armoured below but not above? Perhaps because danger threatened from below, which would suggest that these creatures spent a lot of their time swimming near the surface - and of course they had to come to the surface to breathe, anyway. Sharks today often attack from below, sharks would have been a menacingly important part of the world of Odontochelys, and there's no reason to suppose that their hunting habits were different in those times. As a parallel example, one of the most surprising achievements of evolution, the extra pair of eyes in the fish Bathylychnops (see over), is probably aimed at detecting predatory attacks from below. The main eye looks outwards, as in any ordinary fish. But each of the two main eyes has an extra little eye, complete with lens and retina, tucked into its lower side. If Bathylychnops can go to the trouble (you know what I mean, don't be pedantic) of growing a whole extra pair of eyes, presumably to look out for attacks coming from below, it seems quite plausible that Odontochelys might grow armour aimed at fending off attacks from the same direction. The plastron makes sense. And if you want to say, yes, but why not have a carapace on top as well, just to be extra safe, the reply is easy. Shells are heavy and cumbersome, they are costly to grow and costly to carry around. There are always trade-offs in evolution. For land tortoises, the trade-off ends up favouring stout, heavy armour above as well as below. For many sea turtles, the tradeoff favours a strong plastron underneath but lighter armour on top. And it is a plausible suggestion that Odontochelys just carried that trend a bit further.
Bathylychnops' extra eye
If, on the other hand, the Chinese authors are right that Odontochelys was on its way to evolving a full shell, and that the shell evolved in water, it would seem to follow that modern land tortoises, which have well-developed shells, are descended from water turtles. This, as we shall see, is probably true. But it is remarkable, because it means that today's land tortoises represent a second migration from water to land. Nobody has ever claimed that any whales, or dugongs, returned to the land after invading the water. The alternative story for land tortoises is that they were on land all along and independently evolved the shell, in parallel to their aquatic cousins. This is by no means impossible; but, as it happens, we have good reason to believe that sea turtles did indeed return to the land for a second go at becoming land tortoises.
KEYbold = landnormal = aquatic
If you draw out the family tree of all modern turtles and tortoises, based on molecular and other comparisons, nearly all the branches are aquatic (normal type). Land tortoises are represented by bold type, and you can see that today's land tortoises constitute a single branch, the Testudinidae, deeply nested within rich branchings of otherwise aquatic chelonians. All their close cousins are aquatic. Modern land tortoises are a single twig on the bush of otherwise aquatic turtles. Their aquatic ancestors turned turtle and trooped back on to the land. This fact is compatible with the hypothesis that the shell evolved in water, in a creature like Odontochelys. But now we have another difficulty. If you look at the family tree, you'll notice that, in addition to the Testudinidae (all modern land tortoises) there are two fossil genera of fully shelled animals called Proganochelys* and Palaeochersis. These are drawn as land-dwellers, for reasons we shall come to in the next paragraph. They lie right outside the branches representing the water turtles. It would seem that these two genera are anciently terrestrial.
Before Odontochelys was discovered, these two fossils were the oldest known chelonians. Like Odontochelys they lived in the late Triassic, but about 15 million years later than Odontochelys. Some authorities have reconstructed them as living in fresh water, but recent evidence does indeed place them on land, as indicated by bold type on the diagram. You might wonder how we tell whether fossil animals, especially if only fragments are found, lived on land or in water. Sometimes it's pretty obvious. Ichthyosaurs were reptilian contemporaries of the dinosaurs, with fins and streamlined bodies. The fossils look like dolphins and they surely lived like dolphins, in the water. With turtles and tortoises it is a little less obvious. As you might expect, the biggest giveaway is their limbs. Paddles really are rather different from walking legs. Walter Joyce and Jacques Gauthier, of Yale University, took this common-sense intuition and provided the numbers to support it. They took three key measurements in the arm and hand bones of seventy-one species of living chelonians. I'll resist the temptation to explain their elegant calculations, but their conclusion was clear. These animals had had walking legs, not paddles. In British English, they were 'tortoises', not 'turtles'. They lived on land. They were only distant cousins, however, of modern land tortoises.
Now we seem to have a problem. If, as the authors of the paper describing Odontochelys believe, their half-shelled fossil shows that the shell evolved in water, how do we explain two genera of fully shelled 'tortoises' on land, 15 million years later? Until the discovery of Odontochelys, I would not have hesitated to say that Proganochelys and Palaeochersis were representative of the land-dwelling ancestral type before the return to water. The shell evolved on land. Some shelled tortoises returned to the sea, as seals, whales and dugongs were later to do. Others stayed on land, but went extinct. And then some sea turtles returned to the land, to give rise to all modern land tortoises. That's what I would have said -indeed what I did say in the earlier draft of this chapter that preceded the announcement of Odontochelys. But Odontochelys throws speculation back into the melting pot. We now have three possibilities, all equally intriguing.
1 Proganochelys and Palaeochersis might be survivors of the land-dwelling animals that had earlier sent some representatives to sea, including the ancestors of Odontochelys. This hypothesis would suggest that the shell evolved on land early, and Odontochelys lost the carapace in the water, retaining the ventral plastron.
2 The shell might have evolved in water, as the Chinese authors suggest, with the plastron over the belly evolving first, and the carapace over the back evolving later. In this case, what do we make of Proganochelys and Palaeochersis, who lived on land after Odontochelys lived, with its half shell, in water? Proganochelys and Palaeochersis might have evolved the shell independently. But there is another possibility:
3 Proganochelys and Palaeochersis might represent an earlier return from the water to the land. Isn't that a startlingly exciting thought?
We are already pretty confident of the remarkable fact that the turtles accomplished an evolutionary doubling back to the land: an early marque of land 'tortoises' went back to the watery environment of their even earlier fish ancestors, became sea turtles, then returned to the land yet again, as a new incarnation of land tortoises, the Testudinidae. That we know, or are nearly certain of. But now we are facing up to the additional suggestion that this doubling back happened twice! Not just to spawn the modern tortoises, but much longer ago, to give rise to Proganochelys and Palaeochersis in the Triassic.
In another book I described DNA as 'the Genetic Book of the Dead'. Because of the way natural selection works, there is a sense in which the DNA of an animal is a textual description of the worlds in which its ancestors were naturally selected. For a fish, the genetic book of the dead describes ancestral seas. For us and most mammals, the early chapters of the book are all set in the sea and the later ones all out on land. For whales, dugongs, marine iguanas, penguins, seals, sea lions and turtles, there is a third section of the book which recounts their epic return to the proving grounds of their remote past, the sea. But for the land tortoises, perhaps twice independently on two widely separated occasions, there is yet a fourth section of the book devoted to a final - or is it? - re-emergence, yet again to the land. Can there be another animal for which the genetic book of the dead is such a palimpsest of multiple evolutionary U-turns? As a parting shot, I cannot help wondering about those freshwater and brackish water forms ('terrapins'), which are close cousins of the land tortoises. Did their ancestors move directly from the sea into brackish and then fresh water? Do they represent an intermediate stage on the way from the sea back to the land? Or is it possible that they constitute yet another doubling-back to the water from ancestors that were modern land tortoises? Have the chelonians been shuttling back and forth in evolutionary time between water and land? Could the palimpsest be even more densely over-written than I have so far suggested?
On 19 May 2009, as I was correcting the proofs of this book, a 'missing link' between lemur-like and monkey-like primates was announced in the online scientific journal PLOS One. Named Darwinius masillae, it lived 47 million years ago in rain forest in what is now Germany. It is claimed by the authors to be the most complete fossil primate ever found: not just bones but skin, hair, some internal organs and its last meal. Beautiful as Darwinius masillae undoubtedly is (see colour page 9), it comes trailing clouds of hype that obscure clear thinking. According to Sky News it is 'the eighth wonder of the world' which 'finally confirms Charles Darwin's theory of evolution'. Goodness me! The more-or-less nonsensical mystique of the 'missing link' seems to have lost none of its power.
Was this article helpful?