Reptiles are cold-blooded animals: they get their body heat from the sun. Their legs issue from the sides of their bodies (if they have legs) and their skulls have two openings (‘fenestra’) behind each eye socket. In contrast to mammals, the lower jaw comprises several bones, the hindmost of which, called the angular, articulates with a skull bone called the quadrate. Reptiles hear by detecting vibrations in the ground as well as in the air. From the ground, vibrations reach the middle ear via the quadrate, from the air, via the eardrum. Legless lizards and snakes have lost the eardrum of their ancestors, so they hear only through the quadrate. The middle ear has just one bone, called the stapes, or stirrup, and the teeth are all cone-shaped.

Mammals are warm-blooded animals: they generate their own body heat, controlled primarily by insulation (hair and body fat) and by perspiration (sweat glands). The legs issue from beneath the body and the skull has only one pair of fenestra. The lower jaw is a single bone called the dentary, articulating with a bone in the skull called the squamosal. The middle ear contains three bones, the malleus (hammer), the incus (anvil) and the stapes, and is more sensitive to sound than the ear of reptiles, especially at higher frequencies. On the other hand, there is no ability to detect ground-borne vibrations. The teeth have a variety of shapes – incisors, canines and molars.

Most features that distinguish mammals, such as hair, mammary glands and reproductive organs, do not readily fossilise. Among those that do are teeth (any animal with ‘-don’ or ‘-dont’ at the end of its name is named after its teeth) and skull structure. On the face of it, the common possession of apparently independent features such as hair, mammary glands, a single lower jawbone and mammalian teeth might appear to be evidence of common descent. However, an animal that had mammary glands and, say, a reptilian jaw might also be so interpreted, on the basis that it evolved before the first true mammal, either along the direct line to mammals or along a side branch. An evolutionary scenario can be devised for all eventualities. For example, the echidna and the platypus – mammals in most respects but with the reptilian characteristic that their urinary, defecatory and reproductive systems end in a single duct – are viewed as stemming from an offshoot somewhere between ‘mammalia- forms’ (such as Morganucodon, discussed below) and true mammals. Pterodactyls were reptiles in most respects but had hair. Ultimate common descent is assumed in all cases, but the assumption is untested and might not be true. On the other hand, claims that a particular series of fossils represents an evolutionary succession can be tested, one case being the series making up the transition from mammal-like reptiles to mammaliaforms.

Although the number of temporal fenestra is considered to be among an animal’s most fundamental characters, how they arose is a mystery, for fossils show them appearing abruptly. Their main purpose was to provide anchorage for certain jaw muscles. Thus a change in the number, position, shape or size of the fenestra implies co-ordinated changes in musculature and skull design. Animals with one pair are termed synapsids, those with two, diapsids. Whereas ancient reptiles include all the basic arrangements of these fenestra, modern reptiles are all diapsids (apart from turtles). Rather awkwardly for evolution theory, the reptiles from which mammals are supposed to have evolved were synapsids.

Once a synapsid, it seems, always a synapsid. There are no documented cases of a synapsid evolving into a diapsid. Further back in time, the synapsids are presumed to have evolved from anapsids: tetrapods without temporal fenestra. This transition is also undocumented. The earliest known anapsid is Hylonomus, from the mid Carboniferous site of Joggins in Nova Scotia. The oldest known synapsid is the pelycosaur Archaeothyris, only slightly younger, and it is possible that the fragmentary Protoclepsydrops, a contemporary of Hylonomus, is also a synapsid. The oldest diapsid is Petrolacosaurus, from the Upper Carboniferous. Intermediates linking these markedly different animals are absent. The fossil record therefore suggests independent origins for all three groups: anapsids, synapsids and diapsids.

Towards mammal-like reptiles

Today’s mammals are thought to have evolved from an extinct group of mammal-like reptiles called the therapsids, which in turn are thought to have evolved from an extinct family of pelycosaurs called the sphenacodonts. Pelycosaurs, of which sphenacodonts were among the earliest to be fossilised, first appear in the late Carboniferous and therapsids in the mid Permian. If pelycosaurs diverged from sphenacodonts they must have done so, at the latest, by the end of the Carboniferous (Reisz & Laurin 2004). There is thus a substantial time gap separating the two groups. There is also a substantial morphological gap, amounting to a ‘major remodelling of the global tetrapod fauna’ (Lucas 2004). A full list of differences is given by Hopson and Kitching (2001). In the case of therapsids they include:

• the absence of several skull elements
• modifications of the shoulder and pelvic girdles, associated with a more upright hindlimb posture
• larger temporal fenestra, providing a larger area of origination for jaw adductor muscles
• larger upper and lower canines and a correspondingly expanded lower jaw bone, with reduced palatal teeth

Therapsids possessed heavy skeletons with short stout limbs, broad flat feet, a short tail, massive skull, and almost no neck. In comparison with some pelycosaurs they were lumbering animals, though the carnivorous species were smaller and more lightly built.

Pelycosaurs were already diverse when they first appeared, indicating that significant evolution had gone on within the group earlier in the Carboniferous. They inhabited what were then equatorial regions (present Europe and North America). Therapsids occurred mostly in the higher-latitude regions of both hemispheres (Russia and South Africa). They too show a high degree of diversity early in their history. Moreover, the similarity of the northern and southern faunas suggests that, despite the absence of a fossil record in the mountainous equatorial belt, there was interchange between the previously separated hemispheres and that therapsids had spread north and south from the tropics (Kemp 2005). Similar movements are recorded for various other kinds of tetrapod (Milner 1993), possibly related to climate change. The ice caps were now melting and the interior of the Pangaea supercontinent was warming. By the late Permian a few pelycosaurs were themselves beginning to migrate to more temperate climes, scratching a living alongside the therapsids. Their dwindling species and individual numbers show they were struggling. Soon afterwards they went extinct, victims of the events that caused mass extinctions in the late Permian across a wide range of groups. Therapsids were among the few that survived.

Was the substantial gap in morphology a direct consequence of the gap in time? The possibility cannot be discounted (cf. Sidor & Hopson 1998), though it is not likely. Thomas Kemp (2005) makes the comment that ‘this may be one of the few cases where the hypothesis of competitive replacement of one taxon by another may be entertained’. This may be fair enough in relation to the simultaneous rise of therapsids and decline of pelycosaurs at the point that therapsids first appear in the record, but it is not a good explanation of their origin at the end of the Carboniferous. If they diverged from early pelycosaurs because they had a competitive advantage and left more offspring, why did they make no mark on the fossil record for 30-35 million years? It was only later still, towards the end of the Permian, that they became the dominant land animals of their day. At the highest taxonomic level, the pattern is one of colonisation rather than evolution.

Therapsids are ranked as an order, with six suborders: biarmosuchians, dinocephalians, anomodonts, gorgonopsians, therocephalians, and cynodonts. Analysis of their similarities and differences reveals, consistent with the progress of time, clear evolutionary patterns within the suborders. By the end of the Permian three of the six were extinct and by the end of the Triassic five were; they failed to evolve further and failed to live on unchanged. Cynodonts continued into the Jurassic. They too are extinct except insofar as one branch evolved into mammals, some of whose descendants make up the 29 orders of modern mammal. Thus the question splits into two parts: (1) did cynodonts evolve into mammals, and (2) can today’s mammals be traced back to the cynodonts?

Did cynodonts evolve into mammals?

The earliest therapsids had a typically reptilian jaw joint, where the articular in the jaw con- nected with the quadrate in the skull. In fossils of the Triassic period these bones decreased in size, as the dentary jaw-bone and squamosal skull-bone became larger, moved closer together and replaced the articular-quadrate joint – thereby approaching the mammalian body plan. Only the cynodonts showed a consistent trend of dentary enlargement; other therapsids retained their ancestral proportions (Sidor 2003). Anomodonts, indeed, showed the opposite trend: decreasing dentary size and increasing size of the postdentary bones.

In some cynodonts the jaw eventually had a double hinge, articulating with the squamosal as well as the quadrate. They also had other features typical of mammals, such as an array of incisor, canines and postcanine teeth, a double occipital condyle (where the skull articulated with the spine), and a secondary palate. A double occipital condyle also occurs in the extinct microsaurs and modern amphibians. The secondary palate, as in mammals, served to separate the airway from the food-processing system, and a series of fossils shows this palate evolving as initially separate plates were sutured together along the midline. Here too the situation is more complicated than can easily fit into a Darwinian world, since a secondary palate arose independently in two other therapsid lineages, the anomodonts and therocephalians (Sidor 2003). They did not evolve into mammals.

The Late Triassic/Early Jurassic Morganucodon (‘Glamorgan tooth’, named after its find locality in south Wales) was a shrew-sized insectivore. Only the skull is well preserved, but we can hazard some assessment of whether it is related to cynodonts from the skeleton of other members of the morganucodontid family. The groups are not strikingly similar. For example, the axis-atlas complex of morganucodonts was essentially mammalian, they lacked lumbar ribs, the brain was comparatively large and the femur had a ball-like head which fit into the pelvis sideways, indicating a gait that was fully erect. There is also a chronological difficulty in linking the groups, since the cynodonts closest in form to the morganucodonts – opinions differ, either the tritylodonts or the tritheledonts – first appear at the same time as their presumed descendants. They cannot therefore have been directly ancestral to morganucodonts, though it remains possible that their own direct ancestors were. In either case, the cynodonts that continued the trend of increasingly mammal-like features without directly giving rise to mammals provide a strong hint that the evolution was pre-programmed. As with the secondary palate, the same characteristics were developing independently in multiple lineages. The fossils show a pattern of precisely co-ordinated, biologically complementary changes, not a random zigzag.

On balance, it would not be unreasonable to conclude that the morganucodonts did descend from the cynodonts. Their forms are discontinuous, but this has to be evaluated in a context where the putative ancestral group (the cynodonts) itself encompassed a high degree of diversity. The same applies in relation to size, which can vary greatly within a genealogically united group (e.g. within bats). Although morganucodonts were tiny, some tritylodonts were also quite small. Morganucodonts and cynodonts had a similar pectoral girdle. Morganucodon also had the double jaw joint of some advanced cynodonts, but now the dentary-squamosal hinge took most of the stresses associated with biting and chewing, leaving the tiny articular-quadrate hinge free to function as a sound-conductor. It still had a composite jaw, and its cochlea was straight rather than coiled as in modern mammals, but the dentary was much enlarged. Its teeth were also characteristically mammalian.

Can today’s mammals be traced back to the cynodonts?

The most advanced mammaliaform so far known is Hadrocodium (Luo et al 2001), from the Early Jurassic. With an estimated body weight of 2 grams, it is the smallest mammaliaform known from the Mesozoic, though it had an unusually large brain for its size. It is also the earliest animal to have had a single jaw joint, as in triconodonts and present-day mammals: the middle ear was now completely separate from the mandible. This is surprisingly early, for the next oldest animal to show this feature, Triconodon itself, was not to arise for another ‘45 million years’. Differing from its predecessors also in the lack of a postdentary trough, Hadrocodium is an enigma, and it would be rash to assume that it too evolved from the cynodonts.

Morganucodonts were the end of a line (Luo 2007). As with the pelycosaurs and the cynodonts, one has to take several steps back in time in order to identify species close enough to what would be required of an eligible ancestor (yet to be discovered and still not very close). One also has to do this with succeeding groups. Not to prolong what is already a complicated discussion, suffice to say that when today’s mammal orders appear, in the Cenozoic, they appear suddenly, without obvious forbears. As summarised in the diagram, the numerous orders and subclasses that arose, all about the same time, in the Middle Jurassic did not survive to the present day. None seem to have been ancestral to modern mammals.

Moreover, Mesozoic mammals were more disparate then, near the presumed beginning of their evolutionary history, than they are today – certainly in terms of tooth design (regarded, like the temporal fenestra, as one of the most fundamental characters). As can be seen from the terminal width of the spindles representing marsupials and placentals, the number of present-day families and orders is very great; the number of higher-level groups, by contrast, is much smaller. In the early Cretaceous there were at least nine such groups; today there are only three.

In taxonomic terms, mammals are a class, and the three modern groups of mammals are subclasses, comprising around 26 orders. Though they may be less disparate than their Mesozoic counterparts, these three subclasses are still extremely heterogeneous, so much so that, if we assume that they are related, we encounter just the converse problem to the disparity of animals in the Mesozoic. Because the Cenozoic orders include living animals, we are able to analyse their DNA and quantify the extent to which the DNA of one species differs from another. Applying such information, and assuming that all the Cenozoic orders go back to a common ancestor, we can infer the rates at which their DNA diverged over time. These lead to the expectation that the orders must have been preceded by long evolutionary branches. This is not, however, what the fossil record reveals:

Both fossil and extant taxa demonstrate that there are few or no such lineages with a long evolutionary lag time. This discrepancy is so systemic and widespread that it cannot be explained by the difference between minimum age constraint (represented by actual fossils) and the timing of origin that can be hypothetically estimated by molecules [DNA mutation rates] in marsupial and placental evolution.

Zhe-Xi Luo, Nature 450:1012 (2007)

Nor can the discrepancy be dismissed as due to the incompleteness of the fossil record. The number of Mesozoic mammalian and mammaliaform genera known to science has tripled in the last thirty years, but the most persistent gaps continue to be those around the base of the Cenozoic groups and those surrounding the marsupial and placental subclasses as a whole.

In short, today’s mammals cannot be traced back to the cynodonts. Impressive – indeed wonderful – though the series is that shows reptiles evolving into mammaliaforms, it does not validate the dogma that all organisms are related to each other in a single evolutionary tree.

Evidence of genetic control

The ‘rise of mammals’ is one of the most striking phenomena in the fossil record. No mammals are known from the Palaeozoic. Apart from amphibians, all the tetrapods that flourished at that time sufficiently to make their mark were reptiles. One branch of synapsid reptiles, the cynodonts, became progressively more mammal-like, and possibly one branch of cynodonts evolved into morganucodonts, which were even more mammal-like. The Late Jurassic saw the radiation of several different groups which may reasonably be called mammals, but their origin is obscure, and apart from the monotremes – represented by just the echidna and the platypus – none of them survived to the present day. The Cretaceous saw the radiation of the marsupials and placentals, the latter diversifying spectacularly in the Cenozoic.

So far we can draw the following conclusions:

• Some of the evolution we see is large-scale. Hearing systems change from those designed to pick up both ground and air-borne vibrations to those designed to pick up primarily air-borne vibrations. They become less sensitive to ground vibrations but more sensitive to high-frequency sound. The same animals also acquire a more upright gait and numerous other modifications that are progressively more mammal-like.
• The evolution appears to be ‘directed’. Complex changes to the lower jaw and associated soft body parts (muscles, internal ear organisation) and concomitant changes in the post-cranial skeleton and tissues appear to be orchestrated towards a pre-determined end.
• Not all reptiles evolve in this direction. No modern-day group of reptiles did, nor did most ancient reptiles. Even among the therapsids, all but one of the six suborders retained their ancestral proportions. Anomodonts showed an opposite trend of decreasing dentary size and increasing postdentaries.

Natural selection acting upon chance mutations does not therefore seem to be the power at work here. As, in the vast majority of cases, selection did not result in reptiles becoming progressively more mammal-like, one might question whether selection was the mechanism in the remaining cases. Moreover, ‘There is a paradox when matching an evolutionary mechanism based on single, small changes in discrete characters to a long term, large evolutionary change in very many, fully integrated characters.’ (Kemp p. 133) Darwinian theory postulates single, small, random changes; what we see is synchronous, inter-related and inter-dependent changes, affecting the whole organism. Darwinism cannot account for the evolution, paradoxically, because the scale of it is too great, not too small.

Concerted interrelated changes of the phenotype, resulting from changes in hugely complex genetic systems, are most naturally understood as non-random and therefore as the result of pre-programming of the systems. This is further supported by what is arguably the most devastating of evidences against the Darwinian hypothesis, the phenomenon of convergence. Convergence is where the same feature occurs in two or more related lineages whose common ancestor did not have that feature. How, one might ask, can the same biological structure arise through a process of random mutations more than once?

Here are a number of examples (some of which have already been mentioned). First, taking a broad overview of developments before the therapsids:

• Lathanosuchid reptiles had features characteristic of early anapsids, except that they had a pair of lower temporal fenestra. Thus either their ‘anapsid’ characteristics evolved independently of true anapsids or their fenestra evolved independently of synapsids. Similar anomalies sabotage attempts to fit the early Permian tetrapod Acleistorhinus and the late Permian millerettids into a neat scheme where anapsid tetrapods in the Carboniferous branched into synapsids and diapsids.

• By contrast, captorhinids and protorothyridids had no temporal fenestra but in most other supposedly fundamental respects were similar to diapsids.

• Accompanying the full range of fenestral arrangements that were present soon after the first appearance of terrestrial animals appeared a bewilderingly diverse array of tetrapod body designs. Some tetrapods propelled themselves by lateral undulation of the vertebral column, later forms, increasingly, by a combination of lateral undulation and limb-driven locomotion. The latter arose independently within several groups (Rieppel & Reisz 1999). There were also convergences in respect of skull roof pattern, the mandible, the axial skeleton and limb design (ibid).

As regards the therapsids:

• Phalanges are the bones that form the fingers and toes, with the ‘phalangeal formula’ being the number of such bones counting from the first to the fifth digit. Ideally, reptiles have the formula 2-3-4-5-4, mammals 2-3-3-3-3. In therapsids, however, the formula varied, and in a manner that was ‘extremely complex’ (Rubidge & Sidor 2001). Appearing multiple times in the course of their history, the mammalian phalangeal formula has proved unsuitable for tracing evolutionary relationships.

• A bony secondary palate evolved in therapsids independently three times, in dicynodonts, therocephalians and cynodonts, each time in a different way (Sidor 2003b).

• The specialised ‘leaf-shaped’ teeth of the anomodont Suminia arose independently in at least five lineages of herbivores: iguanid lizards, ornithischian and prosauropod dinosaurs, pareiasaurs and caseid pelycosaurs (Rybczynski & Reisz 2001).

• Propaliny, the ability of the upper jaw to slide backwards and forwards over the lower jaw, arose ‘perhaps as many as seven times’ within non-mammalian synapsids (ibid).

• Trithylodonts and tritheledonts had different combinations of mammalian and non-mammalian characters. Since they cannot both have been ancestral to mammals, some of their mammalian characters must have arisen independently (Kemp p 76).

• The double occipital condyle, while distinctive of mammals, is also found in modern amphibians and the extinct microsaurs.

Even the mammalian middle ear (MME) had more than one origin:

• The middle ear of the recently discovered triconodont Yanoconodon (Luo et al 2007), classified as a true mammal, surprisingly shows the ‘pre-mammalian’ condition where the middle ear bones were still connected to the mandible. Thus, either the MME structure was present in the common ancestor of monotremes, triconodonts, marsupials and placentals, and triconodonts ‘re-evolved’ the more primitive condition, or the MME was absent in the common ancestor and evolved separately in monotremes, marsupials and placentals.

• The same arrangement whereby sound is transmitted to the inner ear via an eardrum, stapes and other cartilaginous and ossified structures arose independently in frogs, squamates, crocodiles, birds and turtles (Müller & Tsuji 2007).

• A different arrangement whereby sound was transmitted to the inner ear via an eardrum, stapes and other cartilaginous structures arose as early as the Permian in the reptile Macroleter (ibid).

And so the story goes on, convergence after convergence:

• Castorocauda was a Middle Jurassic docodont with a broad, scaly tail for swimming, just like that of the modern but unrelated beaver. It is also, incidentally, the oldest fossil mammaliaform with preserved fur.

• Volaticotherium, representing another previously unknown mammal order from the Middle Jurassic, had the membrane and elongate limbs of a glider, convergent to today’s placental ‘flying’ squirrels and marsupial sugar gliders. Indeed, gliding arose no fewer than seven times amongst the mammals, Volaticotherium being an eighth such instance (Meng et al 2006).

• Haldanodon, a docodont from the Late Jurassic, possessed many of the specialisms characteristic of modern, but unrelated, semi-aquatic moles (desmans).

• Fruitafossor, an enigmatic mammal from the Late Jurassic, perfectly imitates two extreme specialisations of teeth and jaw that previously were thought to be unique to the South American xenarthrans (anteaters, armadillos and sloths).

These convergences are examples, not an exhaustive compendium, and any one of them would be grounds for questioning the belief that chance mutations drove macroevolution. In total they may be considered to be totally devastating.

Summary

This contribution is part of a series discussing what is thought to be the prime fossil evidence that fish evolved into people. The evidence discussed here is alleged to document the transition from reptiles to mammals via mammal-like reptiles. A study of the relevant evidence shows:

• The reptiles involved in this alleged transition were synapsids and thus unlike modern reptiles, which are diapsids.

• There are no documented transitions across the anapsid/synapsid/diapsid boundaries.

• It is possible, but not likely on present evidence, that one branch of early reptile, the pelycosaurs, evolved into mammal-like reptiles.

• There was no general trend among reptiles to become more mammal-like (hence, obviously, there are still many kinds of reptile today). Even among the six orders of therapsids, only one – the cynodonts – became progressively more mammal-like. Most were fairly stable in the relevant characters, while the anomodonts showed a progression away from the mammalian state.

• It seems likely that the cynodonts were ancestral to the morganucodonts, representing the next mammal-like grade on from the cynodonts, although it is not clear precisely which group of cynodonts continued in this direction.

• The morganucodonts were not the ancestors of modern mammals.

• No Mesozoic group can be identified as the ancestors of modern mammals.

• Numerous ‘convergences’ confound all attempts to link reptiles and mammals into a single evolutionary tree.

We will consider how the same developments can be understood within a recolonisation scenario in a separate article.

If you would like to explore further the idea that evolutionary change is pre-programmed, see:
Evolution in the genome