5. The origin of turtles

The five-part BBC program, Journey of Life, opens with the sun setting on a remote beach in the Pacific Ocean. Leatherback hatchlings emerge from the sand while the voice-over tells of an epic struggle:
It’s a battle fought by all living things. Only the individuals that are best suited to the world in which they live have a chance of surviving: what we know as survival of the fittest. And at no point in these babies’ lives is this more critical than when they leave the nest. Hungry predators are waiting. Although they may look the same, each baby is different from the next. The tiniest variation in size and shape can determine who lives and who dies. Some are weak, others strong. Anything that boosts the baby’s chances of surviving, such as a sturdier shell or longer flippers, will be passed on to its young in its genes and over time these inherited changes can lead to the evolution of a new species. These babies are the genetic veterans of a battle for survival that has been going on since the first turtles evolved. Here only a few make it to the water’s edge. This is natural selection, the way life evolves, and it has been shaping new kinds of creatures throughout the whole of life’s great journey.

So runs the undoubting, modern account of how living things come to be. But as an illustration of the message that life evolved by natural selection, turtles are an unfortunate choice. The oldest fossils go back to the Late Triassic, about the time of the first dinosaurs, and they are as much turtles as any modern species is. Although hundreds of new species have arisen since, no lineage has lost the essential turtle form, and modern turtles are no “fitter” than their predecessors were. Survival rates, indeed, are extremely low, as the footage of hungry pelicans swooping down on the toddlers brutally emphasises.

Like the extinct ichthyosaurs, turtles (including tortoises and terrapins) are reptiles, and they have terrestrial ancestors. Consequently sea turtles must draw their oxygen from the air and come ashore to lay their eggs. Whatever its powers, natural selection has not delivered them from these handicaps.

Perhaps the best known sea turtle is the leatherback. Guided by an unfathomable instinct, the pregnant animal emerges from the surf after dark, close to high tide. Laboriously she crawls up the beach, selects a site just beyond the high-water mark and starts to dig. Her front flippers scoop out a cavity for her body. The rear flippers then excavate a smaller, deeper hole. Typically she deposits around 70 fertile eggs and 40 smaller infertile ones. Her duty done, she fills the nest with sand to conceal its location and returns to the sea. She will never know her infants.

She will repeat the burial ritual several times in the course of the nesting season. Meanwhile, in about two months, that first clutch will hatch, and over a few days the newborn turtles will dig their way to the surface. Like their unseen mothers, they know what to do by instinct. Predators may be lurking, and the babies sense that they must wait for the top sand layers to cool – a sign of nightfall – before making their break for the sea.

Turtles in the Triassic

Proganochelys (reconstructed)Among the Triassic turtles the best known is Proganochelys, found in Thailand and in several parts of Germany, where sandstone and shale suggest a brackish or marginal marine environment. The turtle was probably washed down to its resting place by coastal rivers. As well as the characteristic carapace and plastron (underside), Proganochelys had toed feet, palatal teeth, a spiky neck and a club-like tail, and both its forelimb dimensions and heavy armour show that it lived on land (Joyce & Gauthier 2003). Some of its features, for example its teeth and long tail, hint at an ancestry among non-turtles. Many more were unique.

Equally old, but somewhat more advanced in the family tree, are Proterochersis, also from Germany, and Palaeochersis, from Argentina. We know less about Proterochersis because, although more than two dozen shells were recovered, all from stream deposits, the skulls were not preserved. Palaeochersis was buried in a mud layer within a wadi, along with other land animals. It too shared several features with non-turtle amniotes.

Another key fossil is Odontochelys, from China. Odontochelys is both slightly older and more basal in form than Proganochelys – for example, it had teeth on both upper and lower jaws, and its overall shape was distinctly elongate. Most significantly, it lacked horny scutes, suggesting that the outer carapace consisted of non-bony tissue, as with modern leatherbacks, and was not preserved. Broadened ribs and vertebrae formed a shell-like underbelly.

Odontochelys – artistic reconstruction by Marlene Donnelly

Other Late Triassic specimens have been recovered from Skye (British Isles) and North America. From their first appearance turtles were already global.

Turtles are the only living tetrapods without temporal openings in their skull, and they have sufficient features in common to substantiate the view that they have a common ancestor. Within the order, Proganochelys stands apart from the other species, showing that, early on, they split into two distinct branches. The non-Proganochelys branch then split into two further sub-groups, the ‘pleurodires’ (Proterochersis being the earliest representative) and the ‘cryptodires’. This also happened before the Late Triassic. Although initially neither sub-group had the neck retraction mechanisms that would become their most obvious characteristic, pleurodires (‘side-necks’) later folded their necks to the side of the shell, while cryptodires (‘hidden-necks’) retracted the head by pulling their necks up and back. Both mechanisms require complex, co-ordinated modifications of the neck vertebrae and muscles. Living turtles are either pleurodires or cryptodires. Triassic turtles protected their necks by other means: Proganochelys by a collar of horny spikes, Palaeochersis by an extension of the carapace.

Later turtles

Manchurochelys from the Liaoning beds of ChinaPleurodires are now the less common group, restricted to the continents of the southern hemisphere, but once they were very widespread, both on land and in the estuaries and shallow seas round the coasts. Sea turtles are all cryptodires. The first true cryptodire was Kayentachelys, from the Early Jurassic, the only turtle apart from Proganochelys known to have had teeth.

The oldest fossil sea turtle is Santanachelys, from the Early Cretaceous. Like its modern counterparts, it had huge salt glands under the eyes, essential for excreting the salts that accumulated as a result of living off seafood. However, the metacarpals and short digits of the feet were still moveable, as in land turtles: only later did the toes lengthen and become encased in flesh, and the feet turn into flippers. As the period progressed, sea-turtle diversity exploded. The largest animal grew to a width of 4 metres and a length of 6 metres.

Today sea turtles are much less diverse than in prehistoric times: just two families and six genera. The leatherback is a family all by itself. As its name implies, it differs from the others in having a soft, cartilaginous shell rather than a hard bony one. It also differs in having no scutes, a reduced skeleton, a throat and upper digestive tract lined with spines (to stop swallowed jellyfish from escaping) and an ability to generate some of its body heat from within. The other marine family (Chelonioidea) comprises the Kemp’s and olive ridleys, the green and black sea turtles, the loggerhead, the hawksbill, and the flatback. Most of these genera are capable of hybridising with others in the family (Bowen and Karl 1997).

Highly simplified cladogram of habitat preferences of major turtle groups (Joyce & Gauthier)Towards the end of the Cretaceous, amazingly, some of the cryptodires began an evolutionary journey back to the land. Present-day tortoises descend from turtles that swam in the sea, not directly from older land-going turtles. The anatomical changes involved are not well documented, but many turtles again became toe-walkers, and while the number of scutes remained constant, shell ornamentation and shape varied greatly. The oldest tortoise fossils come from the late Palaeocene of Mongolia, so by then the transition back to the land, at least in one lineage, was already complete. In the Eocene, tortoises colonised North America, Europe and Africa and adapted to an immense range of environments, from forests to deserts. Some were capable of subsisting both in water and on land.

They also colonised some islands. The Galapagos islands are named after these animals, their carapaces reminding Spanish explorers of a kind of saddle they called a ‘galapago’. Their closest relatives are a species on Chile, 1,000 kilometres to the east. Since the archipelago is volcanic in origin and has never been connected to the mainland, they must have reached the islands by rafting, transported by currents that pass up the coast of Chile and Peru before circulating westward. Giant tortoises used to exist on all continents, but except on the remote oceanic islands of the Galapagos and the Seychelles, they are now extinct. The largest, Colossochelys atlas, attained the size of a Volkswagen Beetle.

The search for ancestors

The turtles known from the Triassic must have had evolutionary predecessors of some kind. One suggestion was that turtles derived from a group of Permian anapsids called the pareiasaurs, the smallest of which showed “otherwise uniquely turtle features such as a rigid covering of dermal armour over the entire dorsal region, expanded flattened ribs, cylindrical scapula blade, great reduction of humeral torsion (to 25°), greatly developed trochanter major, offset femoral head, and reduced cnemial crest of the tibia”. Plausible though the hypothesis appeared at the time, it has since been found wanting (Rieppel & Reisz 1999). One reason is that the carapace “dermal armour”) of turtles is a complex structure formed through fusion of the vertebrae and ribs to interlocking plates beneath the skin and through growth of horny scutes above the skin; as such it is without parallel. The dermal plates of pareiasaurs cannot be interpreted as precursors of the plates in the turtle carapace. Furthermore, the limb girdles and shoulder blades lie within the ribcage rather than outside it as in other amniotes, and if the turtle ancestor had been a terrestrial animal, the ribs would somehow have had to lose their function of supporting respiration and locomotion:

In a generalized tetrapod reptile, aspiration of air is effected by an expansion of the body cavity through muscular action exerted on the ribs. Exhalation is effected either by passive recoil of the body walls, and/or by compression of the lungs as a result of active compression of the rib cage. By contrast, respiration in turtles depends on volume changes of the thoracico-peritoneal cavity inside the rigid dermal armor, which is achieved by altering the position of the limb flanks through the activity of anterior and posterior muscles. A comparison of respiration in an aquatic (Chelydra) and a terrestrial (Testudo) turtle resulted in the identification of three forces that influence the relative volume of the thoracico-peritoneal cavity: These forces are muscular activity, gravity (supporting inhalation), and, in aquatic turtles, hydrostatic pressure (supporting exhalation).

A terrestrial ancestor could not could not have relied on water pressure to mitigate its breathing problems. In short, the postulated origin from pareiasaurs ‘greatly oversimplified the evolutionary changes that took place in the origin of the turtle body plan’.

On the basis that turtles have to be placed somewhere in the presumed universal tree of life, the genetic evidence favours crocodiles as the closest living relatives. Crocodiles, however, are diapsids: they have two pairs of temporal openings in the skull, whereas turtles have none. Some time before their appearance in the fossil record turtles must therefore have lost both pairs. Apart from this problem, a turtle + crocodilian grouping is ‘completely unorthodox from a morphological and paleontological perspective’ (Zardoya & Meyer 2001).

Morphological arguments are sometimes adduced for a relationship to other diapsids (e.g. to lizards, or the extinct plesiosaur and pliosaurs), but this is again only on the basis that turtles have to be related to some other group. Turtle morphology is surrounded on all sides by profound discontinuity. “Due to a general lack of shared derived characters with other amniotes and due to conflicting phylogenetic signal in different data sets,” it is not possible to identify any ancestral group with assurance. The presumption of common descent remains unvalidated.

The problem of convergence

Nonetheless, turtles have evolved a great deal in the course of their fossil history. As Darwin perceived while visiting the Galapagos Islands, evolution happens when organisms colonise vacant ecospace: they adapt. Darwin speculated that they adapted through one variation being fortuitously more advantageous in the new environment than another. What specified an organism’s form in the first place was, at that time, a ‘black box’, and within a framework of philosophical materialism it was not unreasonable to suppose that variants came about randomly, that the favouring of one form over another was a hit-and-miss process of ‘mutate or die’. With the discovery of DNA that view became suspect. Once it was known that the growing embryo made proteins and organised cells by following a set of instructions, the possibility opened up that evolutionary change too might be genetically controlled. DNA might determine an organism’s phylogeny as well as its ontogeny.

One of the most compelling reasons for supposing that evolution is non-random is the phenomenon of convergence: the appearance of a novel feature in separate branches of an evolutionary tree, even in different trees. And convergence is ‘rampant’ within the turtle family tree (Rougier et al 1995), so much so that reconstructing their evolutionary history on the basis of unique characteristics can be extremely difficult. Examples include:
  • the same modifications of shoulder girdle and limb extremities in several lines of marine and freshwater turtles (Depecker et al. 2006)
  • the formation of paddles in each of the three chelonioid families (Hirayama 1998)
  • very similar skull shapes, neck and skull colour patterns and shell shapes in the Emydidae, Geoemydidae and Testudinidae families (Claude et al. 2005)
  • the re-appearance of a club-like tail, 180 million years after Proganochelys, in the tortoise Meiolania platyceps
Convergence also complicates matters when one looks beyond turtles and finds, for example, turtle-like armour appearing in another group of obscure origin called the placodonts. Entering the fossil record just before the earliest turtles, placodonts consisted of two main branches: unarmoured and armoured. The unarmoured placodonts had a reptile’s typically elongate body and a relatively high, narrow skull. The armoured placodonts were shorter and, viewed from above, were more rounded, with a dorsal shield composed of interlocking scutes (i.e. a carapace) and in some instances also a ventral shield, or plastron. Henodus was particularly turtle-like, its pectoral girdle being “almost identical to that of Proganochely”. Nonetheless, detailed comparison of their respective armours shows that placodonts and turtles evolved independently of each other. Although they are the most obvious candidates, placodonts cannot have been the ancestors of turtles.
Fossil placodont and artistic reconstruction

Another example is the innovation of salt glands in groups as diverse as sharks, estuarine crocodiles, sea snakes, marine birds and marine iguanas. Just what is involved in the innovation may be gathered from this description:

Sea turtle salt glands are modified lachrymal glands, each gland consisting of about one hundred lobules separated by blood vessels and connective tissue. The lobules contain many blind-ending secretory tubules which drain into a central canal within each lobule, and the central canals combine to form secondary ducts and a sac-like main duct which opens into the posterior canthus. The central canal and secretory tubules have a rich blood supply and there is extensive vascularization of the duct. The central canals are surrounded by broad sheaths of connective tissue with a rich network of cholinesterase staining nerve fibers around each tubule; a nerve network reactive to monoamine oxidase is also found in the perilobular connective tissue. The lobule is composed of three types of cell.

P. Lutz & J. A. Musick, 1996. The Biology of Sea Turtles, p 347.

If ascribing such intricate mechanisms to the work of chance is not the most plausible of interpretations, ascribing to coincidence the multiple re-occurrence of them only compounds the difficulty. Even so simple a recurrent device as the egg-tooth, the ephemeral spike with which the baby reptile or bird instinctively hacks through the egg, cries out for a design explanation.

Amongst the other wonders that came into operation in the course of turtle evolution two more deserve some mention.

Sea turtle navigation

The long-distance migrations of sea turtles represent some of the most remarkable feats of orientation and navigation in the animal kingdom. Starting from their birth places on the eastern coast of North America, juvenile loggerheads follow the Gulf Stream from one side of the North Atlantic to the other and back again, travelling distances of more than 9,000 miles (15,000 kilometres). In the Pacific the migrations can be even longer. Having deposited her eggs on a beach in Papua New Guinea, one adult loggerhead was tracked all the way to feeding grounds off the coast of Oregon: a distance of 12,774 miles. Nor, very probably, was that the first time she had crossed the ocean. When the time comes to breed, loggerheads return to the beach where they were born. Somehow they remember its location by reference to the Earth’s magnetic field and navigate home with the precision of homing pigeons, which, although not related to loggerheads, have evolved a very similar system.

Sea turtle diving

Nesting leatherback (National Geographic Magazine) title=Both ancestrally and as individuals, sea turtles originate on land. From that starting point, their phylogenetic and ontogenetic journeyings show that a terrestrial origin sets no limit on their world. In this sense, evolution has known no bounds. Like the ichthyosaurs, they have acquired as remarkable an ability to live far out in the ocean as to dive deep into it. Leatherbacks have been known to plumb depths of over 1,000 metres. Rather than holding their breath (which would be fatal), they expel it, collapsing their lungs and drawing the oxygen they need from cells rich in myoglobin and haemoglobin. Green sea turtles can stay under water for as long as five hours. They seem to have retained a tie to the land precisely in order to demonstrate their Creator’s limitless power.

Like all reptiles, turtles need to absorb heat from their environment in order to keep warm and have relatively low rates of metabolism. It would not have been a wise move for land turtles to try to make their living in the ocean, where temperatures were lower. Special equipment would have been needed. But as we have seen, special equipment is just what they acquired, and this included measures to cope with the cold. Leatherbacks expose themselves to colder temperatures than any other turtle, but are also more active than any other, an instinct for constant movement causing them to generate high levels of heat in their muscle tissues. Larger body size, and consequently a greater volume-to-surface ratio, reduces the rate of heat loss, while a thick layer of fat under the carapace provides insulation. The low volume-to-surface ratio of the flippers is counteracted by arranging the arteries and veins side by side and wrapping them in insulating fibre, so that heat from the blood flowing out is transferred to the cooled blood flowing in. (Beavers and ducks also make use of this heat-exchange system, in yet another case of convergence.) By these means, leatherbacks are able to maintain a core-skin temperature difference of up to 30° C. In the form of sea turtles, reptiles have evolved some of the same endothermic abilities as characterise mammals.


In modern thought, evolutionary change is inextricably associated with the idea that life came into existence from primeval chaos. However, far from the one being evidence of the other, the opposite is true: the greater the change in anatomy, the more difficult it is to understand it as unorchestrated change. This is particularly the case where the end-result is the acquisition of new complex organs, such as salt glands, or new complex systems, such as a sea turtle’s temperature control mechanisms. These innovations clearly evolved with an end in view, namely to equip a formerly terrestrial animal for life at sea. They arose not by natural magic, by some propensity of atoms to assemble themselves into complex structures, but because they were called forth by a pre-existing molecular program. A similar conclusion can be drawn from the development of neural plates but not scutes in the Triassic turtle Odontochelys. Just as plates formed before scutes in the evolutionary history of turtles, so they still do in modern turtle embryos. The evolution of their distinctive carapace was genetically regulated. In not proceeding to the final stage Odontochelys was an example of “paedomorphosis” (Reisz & Head 2008), a common phenomenon in evolution.

In Darwinian philosophy new species arise as a result of nature selecting whatever mutation gives the organism in its environment a competitive advantage. For land animals, however, the sea is a hostile environment: it is cold, its oxygen unusable, and its salts poisonous. Competition only ensures that an animal “adapted” to life on land will remain so adapted; there is no incentive to make the business of survival harder by exchanging a tolerant environment for an intolerant one. If the animal is to be pushed where it will not naturally go, the species will continue to survive only by acquiring the requisite adaptations quickly, not by tiny steps that lead nowhere and achieve nothing until the feature is fully formed.

The record of turtle diversification brings into play questions that have long vexed the palaeontological effort to understand evolution in terms of slow, stepwise change, directed by the vicissitudes of external events. ‘All paleontologists know,’ wrote Stephen Jay Gould, ‘that the fossil record contains precious little in the way of intermediate forms; transitions between major groups are characteristically abrupt’. They also know that science is not, at the highest level of interpretation, about allowing the evidence to speak for itself but about making it conform to a preconceived view of life, what Gould termed ‘the essence of Darwinism’. Anything other than a Darwinian explanation would be ‘apostasy’.

Consensus on the evolutionary position of turtles within the amniote phylogeny has eluded evolutionary biologists for more than a century. This phylogenetic problem has remained unsolved partly because turtles have such a unique morphology that only few characters can be used to link them with any other group of amniotes. (Zardoya & Meyer 2001)

The turtle body plan is evidently highly derived, indeed unique among tetrapods. The problem for an evolutionary biologist is to explain these transformations in the context of a gradualistic process. (Rieppel 2001)

Since the fossil record has yet to reveal what kind of animal gave rise to the earliest known turtles, the speculation continues, but whatever the ancestor was, the apparent absence of intermediates between major groups means the transition is likely to have been abrupt.

Pareiasaurs were not the ancestral group. Nor were placodonts. Does this then mean that the ultimate progenitors were created with the turtle morphology? Again, the answer must be no. In recolonisation theory, as in Darwin’s, the history of the lineage giving rise to turtles goes back way beyond the Palaeozoic, and no one would wish to argue that evolution started only when the first fossils appeared. Indeed, the parallel case of placodonts rather strongly suggest that turtles were not always turtles, for while placodonts also do not present a smooth series of intermediates, armoured placodonts do seem to have descended from unarmoured ones. The turtle shell was just one innovation among many. Like the repositioning of the limb girdles, new breathing muscles, two different neck-retraction mechanisms, salt glands, flippers, the trans-oceanic navigation system and leatherback endothermy, the dermal armour of turtles evolved. In evolution anything is possible, even things that seem impossible. The question is: what, or who, was responsible for its wonders.