The limbs are the main locomotory structures in most tetrapods. There are of course exceptions (gymnophionans, snakes, etc.), but most tetrapods walk, run, or fly using their limbs, which probably played an important role when our ancestors ventured onto land. To understand the colonization of land by vertebrates, we must study the origin and evolution of limbs, and especially of their skeleton, which is often the only part to fossilize.
The vertebrate skeleton can be divided into an internal endoskeleton and a dermal skeleton, which is located in a more superficial position (in the deep part of the dermis and immediately below). The dermal skeleton of vertebrates is not an exoskeleton, because it is generally covered by the epidermis (at least). The vertebrate skeleton is composed mostly of bone and cartilage. Cartilage is lighter and more flexible than bone and it is generally not mineralized. However, it wears down and heals more poorly than bone, which explains that old individuals (at least in our species) often suffer from arthrosis caused by extensive cartilage erosion around articular surfaces. In contrast, bone is mineralized (its mineral component is hydroxyapatite) and it heals relatively well if broken. It is also stiffer than cartilage. Thus, it is no surprise that the skeleton of most large vertebrates is composed mainly of bone, cartilage being found around articulations and during early development.
The dermal skeleton was primitively composed of bony scales of variable size that covered the whole body. Among stegocephalians, it is represented by the dermal skull (composed of superficial bones, such as the frontal and parietal), the dermal part of the shoulder girdle, and small dermal scales, which are usually ossified and are most commonly found on the ventral surface of the abdomen (Fig. 4.1). Contrary to the endoskeleton, the dermal skeletal elements do not go through a cartilaginous developmental stage; these bones form directly from mesenchyme condensations (mesenchyme is a loose tissue composed of poorly differentiated cells).
The endoskeleton (Fig. 4.1) includes the axial skeleton (vertebral column and ribs), the neurocranium (the braincase that protects the brain and sense organs like the inner ear, the eye, and the olfactory epithelium), the visceral skeleton (gill arches and the mandibular arch, which represents part of the jaw), and most of the appendicular skeleton (which supports the fins and limbs). The oldest part of the endoskeleton is probably the notochord, a slender rod that stiffens the body of the amphioxus, a close relative of vertebrates. The notochord is a precursor of the vertebral column, around which the vertebrae appeared and around which they develop in embryos. Vertebrae provide additional support and greater stiffness, which were required as vertebrates increased in size. They are composed of a neural arch, which surrounds and protects the spinal chord, and of a centrum, which surrounds and strengthens the notochord. In the tail, there is also a hemal arch, which surrounds a large artery. The endoskeleton of the first vertebrates was mostly cartilaginous, but in stegocephalians, most of the cartilage is replaced in ontogeny (individual development) by endochondral bone (which develops in a cartilage and later replaces it) or perichondral bone (which forms around a cartilage without destroying it).
Figure 4.1. Vertebrate skeleton. Skeleton of Acanthostega, from the Late Devonian. The dermal skeleton consists of the dermal skull, the dermal shoulder girdle, and the caudal fin rays. The endoskeleton includes the vertebrae, ribs, visceral skeleton, neurocranium, pectoral and pelvic girdles, limbs, and caudal fin rays). Modified from Coates (1996). Reproduced with permission from the Royal Society of Edinburgh from the Transactions of the Royal Society of Edinburgh : Earth Sciences, volume 87 (1996), pp. 363–421.
The appendicular skeleton includes the girdles (pectoral and pelvic), the stylopod (first segment of the limbs: humerus in the forelimb, and femur in the hind limb), the zeugopod (second segment of the limb: radius and ulna in the arm, tibia and fibula in the leg), and the autopod (hand and foot, from the wrist and ankle to the tip of the fingers and toes). All these structures are endoskeletal, except part of the shoulder (pectoral) girdle, in which several dermal bones were primitively present (interclavicle, clavicle, cleithrum, and anocleithrum, in a ventral to dorsal order). In many extant tetrapods, the dermal portion of the shoulder girdle is much reduced and only the clavicle is retained (as in humans).
Recent works in developmental molecular biology have yielded new insights into old evolutionary problems, especially through data on gene expression patterns in developing embryos. The classical problem of the origin of digits has thus been tackled using Hox gene expression patterns in limb or fin buds. Hox genes, present in several copies in vertebrates, are involved in body plan organization (fate mapping) in animals (see the box titled “Hox genes”).
The Hox genes (short form of Homeobox) are expressed in embryonic development; their expression influences development by determining (along with other genes) the destiny of cells located in various positions in the embryo. Among vertebrates, they are particularly important in determining the identity of cells along the anteroposterior axis (through which the vertebral column extends), but they also play an important role in secondary axes, such as the fin and limb axes. They are also involved in the development of other metazoans. Even plants have similar homeotic genes, although they have a different origin and are not homologous. They were discovered independently in 1983 by two teams (Scott and Weiner, 1984).
In early metazoans, there was a single Hox gene complex comprising 13 genes (Hox 1 to 13). This single complex persists in some chordates, such as the amphioxus Branchiostoma floridae. Among vertebrates, this number increased quickly through duplication. Hox gene evolution is complex (because some genes, or even whole complexes, that had appeared through duplication were subsequently lost) and still under study, but it seems that the first craniates had at least two complexes, and the first gnathostomes, four. This number persists in sarcopterygians and in some actinopterygians (such as Polypterus senegalus), but another duplication took place in a teleostean ancestor, because three distantly related teleost species possess seven such complexes (Amores et al., 2004).
Since the 19th century, anatomists and paleontologists have studied the origin of the autopod (the distal part of the limbs, from the wrist or ankle to finger and toe tips). The oldest studies suggested that the phalanges were homologous with the distal radials of our ancestor’s fins, but this idea was rejected in 1941.
Early works on Hox gene expression patterns in developing fin buds of the teleost Danio rerio and in developing limb buds of the chick (Gallus gallus) and mouse (Mus musculus) suggested that the hands and feet were neomorphs, new structures without homologues in the vertebrate fins. These data showed that in mice and chicks, a few Hox genes (D-10 to 13) were expressed in three phases in early limb development (Fig. 4.2). The first phase seems to delimit the territory of the stylopod (proximal segment of the limb, in which the humerus and the femur, the arm and thigh bones, are located); it consists of a distal expression pattern (Fig. 4.2A, B). The second phase, which is expressed in the posterior half of the limb bud (Fig. 4.2C, D), seems to delimit the zeugopod (second segment, which includes the radius and ulna of the upper arm, and the tibia and fibla of the leg). In mice, the third expression phase is restricted to the posterior part and the distal portion of the limb bud (Fig. 4.2E); it seems to correspond largely to the territory of the autopod (hand and foot). In the teleost Danio, there is no third phase; in the developmental stage at which it could be expected to occur, we still find the expression pattern of the second phase (Fig. 4.2F). This suggests, according to some authors, that there is no homologue between the autopod (hand and foot) and the fin of Danio. The Hox A-11 expression pattern leads to the same conclusion because it is expressed in the distal portion of the zeugopod, far from the limb bud apex in mice (Fig. 4.2G), whereas it is expressed in the fin bud apex in Danio (Fig. 4.2H).
The problem with this argument is that Danio, like all teleosts, has a reduced fin endoskeleton, as shown by a comparison with the sturgeon (Fig. 4.3, Acipenser), and it lacks a metapterygial axis (which appears in fin development, and along which the structures homologous with the limb differentiate); thus, no homologue of the autopod is expected in the Danio fin. It would have been much more interesting (although admittedly much more difficult) to perform similar studies on sarcopterygians, such as dipnoans, in which the metapterygial fin axis is well developed (Fig. 4.3, Neoceratodus). A similar study on an extant basal actinopterygian that retains a metapterygial axis (Polyodon spathula) subsequently showed a distinct third phase of Hox gene expression in fin buds, which is similar to the tetrapod third phase. The authors thus concluded that the third phase is an osteichthyan synapomorphy that was lost in teleosts, along with the metapterygial axis (Davis et al., 2007). To sum up, the few available data on Hox gene expression patterns in fin and limb buds do not settle the debate on the origin of the autopod (neomorph, or homologue of distal radials), even though several recent studies raise the possibility that the autopod is homologous with the distal part of the fin (Fig. 4.3).
Figure 4.2. Hox gene expression pattern in actinopterygian and tetrapod appendages. Hox gene expression pattern in mouse limb buds (left) and in fin buds of the teleost Danio rerio (right). The zones of various Hox gene expressions are shaded dark gray; the apical ectodermal ridge, in which fin rays develop, is in light gray. Note that in Danio, the third expression phase is not distinct from the second one (the same expression pattern prevails), contrary to the pattern displayed by the mouse. Redrawn from Sordino et al. (1995).
The origin of the autopod (hand and foot) has also been studied using molecular developmental data, as we saw above, but the morphology of the fins of extant and extinct vertebrates also provides critical data on this topic.
Among actinopterygians, the paired fin endoskeleton is composed of a series of small parallel rods (the radials), sometimes accompanied by smaller bones (Fig. 4.3A). In these fins, only the posterior portion, located along the metapterygial axis, can be homologous with the tetrapod limb, although this implies no homology of the individual skeletal elements. The anterior portions of these fins (Fig. 4.3A, in gray) has no homologue among tetrapods. The sarcopterygian paired fin is monobasal; it articulates with the girdle through a single proximal radial, which is homologous with the humerus or the femur (Fig. 4.3B–F). The distal portion of the fin is much more variable; the second segment (these skeletal segments are called mesomeres) may consist of a single element (homologous with the radius and ulna, for the anterior appendage), as in the coelacanth (Fig. 4.3B) or in dipnoans (Fig. 4.3C), or of two elements, the radius and ulna, as in various stem tetrapods (Fig. 4.3D–F). The third mesomere, homologous with the wrist and ankle, is even more variable: it may consist of one to as many as five radials. From the fourth mesomere upwards, the morphology is still more variable and its homology is more controversial. These radials may be homologous with metacarpals and metatarsals, the bones that extend from the wrist and ankle to the base of the fingers and toes. Similarly, the potential homology between the radials of the fifth mesomere and the proximal phalanges of fingers and toes is highly controversial.
After having initially supported homology between the fourth and fifth fin mesomeres and the metacarpals (or metatarsals) and phalanges (early in the 20th century), morphologists and paleontologists rejected this homology in the 1940s. They proposed that the autopod (hand and foot) was a neomorph without homology with the sarcopterygian fin, as if it had been added to the distal extremity of these fins (Fig. 4.4). That interpretation was initially proposed by Gregory and Raven (1941). It is not entirely convincing, because we generally consider that the radials of the fins of various sarcopterygians are homologous, despite important morphological differences (Fig. 4.3B–F), and numerous transformations must have occurred to explain the diversity of these fins. Why should the stegocephalian autopod (Fig. 4.3G) be treated differently? Could it be because of our anthropocentrism? The great similarities between rhizodontid fins (Fig. 4.3D) and the limb (Fig. 4.3G) could be interpreted as synapomorphies (implying homology between autopod and fin), even though that is not the prevailing interpretation in recent works.
Figure 4.3. Appendicular skeleton of osteichthyans. Right pectoral appendages in dorsal view (the head is to the left). Hatching and shading indicate the hypothesis of maximal homology that can be proposed based on the topology of the elements (Laurin, Girondot, and de Ricqlès, 2000); it is fairly speculative for distal elements (metacarpals and phalanges). Note the great morphological diversity of osteichthyan appendages and the apparent complexity of their evolution. A to C are extant, whereas D to G date from the Late Devonian. The metapterygial axis is shown as a gray line (its distal position in rhizodontids is uncertain; two possibilities are proposed). The fin of Acipenser, the sturgeon, retains a metapterygial axis which is homologous with the axis in the sarcopterygian fin and of the stegocephalian limb.
PARTS: A, the actinopterygian Acipenser sturio (sturgeon). B, the actinistian Latimeria chalumnae (the coelacanth). C, the dipnoan Neoceratodus forsteri. D, a rhizodontid. E, the tristichopterid Eusthenopteron foordi. F, Tiktaalik roseae. G, the stegocephalian Acanthostega gunnari.
ABBREVIATIONS: h, humerus. in, intermedium. r, radius. u, ulna. un, ulnare.
Figure 4.4. First hypothesis of the neomorph nature of the autopod. According to Gregory and Raven (1941), most of the autopod is a neomorph (in gray). The pectoral appendages of three sarcopterygians show the hypothesized evolution of fins into limbs. The metapterygial axis of the appendage is shown (gray line); it is located elsewhere in tetrapods, according to more recent works. The limb of Eryops actually features four digits, rather than six as shown here. Modified from Gregory and Raven (1941).
PARTS: A, Eusthenopteron fin (Late Devonian). B, hypothetical intermediate stage. C, limb of Eryops (Early Permian).
Such an interpretation, suggesting homology between fin radials and phalanges, was proposed a decade ago (Laurin, 2000) as a hypothesis that should be tested in the future, but it was ignored for several years. The discovery of a Devonian sarcopterygian named Tiktaalik, which is probably our closest known finned relative (Fig. 4.3F), prompted its describers (Shubin et al., 2006) to propose the hypothesis of homology between fins and autopod anew, but with more conviction. Shubin et al. (2006) suggest that the origin of limbs involved the proliferation of structures (endoskeletal radials) that were already present in fins such as those of Tiktaalik. However, this idea remains a hypothesis because the morphological gap between the fin of Tiktaalik and the limb of the first stegocephalians remains substantial. For instance, the pre-axial rays of the fin (located cranial to the metapterygial axis), which are present in Eusthenopteron (Fig. 4.5A), are retained in Tiktaalik (Fig. 4.5B). Yet the presence of these preaxial rays had been used as an argument against the hypothesis of homology between autopod and fin (Laurin, 2006).
The origin of digits was an important event because digits are part of the autopod, and this structure enabled stegocephalians to walk on emerged land. It was long thought that the first autopods possessed five digits, which is the maximum number normally present (i.e., in non-pathological cases) in extant tetrapods. However, the oldest known limbs, from the latest Devonian (Famennian, about 360 Ma), all display more than five digits; thus, they are polydactylous. These include the limbs of Acanthostega, whose hand possesses eight digits (Fig. 4.5C), Ichthyostega, whose foot shows seven toes, and Tulerpeton, whose hand has six fingers (Fig. 4.6). From this we can conclude that polydactyly is the primitive condition for stegocephalians, and that pentadactyly, which characterizes tetrapods, appeared later, in the Carboniferous, through a reduction in digit number.
Figure 4.5. Appendicular skeleton of tetrapodomorphs. Right appendages of Devonian sarcopterygians in dorsal view (the head of the animal would be to the left). The main (metapterygial) axis is represented by a thick line; the preaxial rays (A, B) are shown as slightly thinner lines.
PARTS: A, the tristichopterid Eusthenopteron foordi. B, Tiktaalik roseae. C, the stegocephalian Acanthostega gunnari.
Figure 4.6. Tulerpeton forelimb. This fossil was found near Tula, in Russia. It dates from the end of the Devonian (Famennian) and comes from a marine environment. The right forelimb is shown in dorsal view (the head of the animal would be to the left). Modified from Lebedev and Coates (1995).
Figure 4.7. Fragmentary fossils attributed to Elginerpeton. These fossils were found in Scotland, but similar remains were also found in the Baltic countries. They date from the Late Devonian (Frasnian). Right (A) and left (B) mandibles in dorsal view; possible left humerus in dorsal (C) and ventral (D) view; right tibia in ventral (E), dorsal (F), and distal views (G). Modified from Ahlberg (1991).
However, two points were debated recently. The first is the date of the appearance of digits. The debate originated with the description of fragmentary fossils slightly older (Frasnian, Late Devonian, about 380 Ma) than Ichthyostega. These fossils were called “tetrapods,” even though the distal portion of the appendage is not preserved. The best known of these very old stegocephalians is Elginerpeton, a taxon represented by jaw fragments, a bone that might be a humerus (its identity has been disputed), and a tibia (Fig. 4.7). Another contemporary genus, Obruchevichthys, is represented only by a fragmentary lower jaw. Other Famennian stegocephalians are represented by similarly fragmentary remains. In the absence of preserved digits, we may wonder why several authors have stated that these animals were “tetrapods.”
In fact, an optimization of appendage type on a consensual sarcopterygian phylogeny shows that these taxa may not have had digits, except Hynerpeton, which is known from a shoulder girdle (Fig. 4.8). The position of Hynerpeton in the clade of stegocephalians with digits suggests that it had a true autopod. Recently described stegocephalian trackways from the Eifelian (Middle Devonian) of Poland suggest that digits appeared much earlier than previously thought, about 395 Ma ago, rather than about 365 Ma ago (Niedzwiedzki et al., 2010). The authors also suggested that this implied very long ghost lineages at the base of the stegocephalian tree, but this is only one of several possible interpretations. Niedzwiedzki et al. (2010: fig. 5b) placed the Middle Devonian trackway in a polytomy with Acanthostega, Ichthyostega, and the clade that includes Tulerpeton and post-Devonian stegocephalians (at the base of the gray portion, in Figure 4.8). This is the most crownward plausible position for this trackway, and it implies the greatest total duration of ghost ranges. However, the trackway could also represent the sister group of the more inclusive clade that includes Elginerpeton, Obruchevichthys, Ventastega, Metaxygnathus, and more crownward stegocephalians (thus appearing in the most basal position in stegocephalians). This second possible position of the trackway would imply the presence of limbs in the basalmost stegocephalians (Elginerpeton and the three others mentioned above), and it would require a much shorter ghost range of early stegocephalians because a single lineage leading to the Late Devonian stegocephalians needs to be assumed. In any case, these trackways were made in a fully marine, intertidal environment (Niedzwiedzki et al., 2010), and this supports a marine origin of stegocephalians, as discussed in Chapter Three.
Another controversial topic is the number of appearances of pentadactyly. A popular hypothesis in the 1980s and 1990s suggested that the reduction to five digits per autopod occurred independently in amphibians and in reptiliomorphs (the group that includes amniotes and extinct related taxa). The most recent analyses suggest instead that pentadactyly appeared only once, in stem tetrapods, before the appearance of amphibians and reptiliomorphs (Fig. 4.9). This implies that our last common ancestor with lissamphibians (frogs, salamanders, and gymnophionans) already possessed pentadactyl limbs.
Figure 4.8. Appendage evolution in sarcopterygians. This tree emphasizes stegocephalians. Note that the numerous Devonian taxa (from Elginerpeton to Metaxygnathus) represented by fragmentary remains, often described as “tetrapods,” may not have had digits. Modified from Laurin, Girondot, and de Ricqlès (2000).
Figure 4.9. Evolution of digit number in the stegocephalian hand. This tree shows the unique appearance of pentadactyly (in gray) from the ancestral polydactylous condition (in white) in the stegocephalian forelimb. Additional losses of digits yielded tetradactyl hands (in black) in temnospondyls and in amphibians. Modified from Laurin (1998a).
The limb with digits was initially considered a terrestrial adaptation. This hypothesis has recently been rejected because it is likely that the first vertebrates with such limbs were fully aquatic. The limb thus appears to be an exaptation (sometimes called pre-adaptation) to terrestrial life because its initial function was not to enable stegocephalians to walk on dry land, even though it fulfilled that function later when stegocephalians invaded the land. This new hypothesis rests on the inference that the oldest stegocephalians were aquatic.
That interpretation was first suggested for Acanthostega, based partly on grooves on the hyobranchial skeleton, which supports the larynx in tetrapods, and gills in other vertebrates. These grooves probably housed arteries that carried blood to the gills, where carbon dioxide would have been released into the water and oxygen would have entered the blood. A bony lamella on a bone of the shoulder girdle (the cleithrum) seems to have formed part of the posterior wall of the branchial chamber. Together, these two characters strongly suggest that functional internal gills were present. Furthermore, the tail retained fin rays (lepidotrichia) similar to those of primitively aquatic osteichthyans (Fig. 4.1). The skeleton of Acanthostega was poorly ossified (much of it remained cartilaginous throughout life), as frequently occurs in aquatic tetrapods. All these clues suggest an aquatic lifestyle in Acanthostega.
If all these characters were present only in Acanthostega, we might think that this stegocephalian was atypical and had returned to an aquatic lifestyle, like so many extant tetrapods (seals, whales, marine turtles, etc.). However, all Devonian stegocephalians seem to have been aquatic because the associated fauna is typically aquatic (often marine or coastal), several seem to have retained internal gills (the others are too poorly known to determine if gills were present), and Ichthyostega also retained the fin rays (lepidotrichia) in the tail. Given the aquatic lifestyle of the earliest stegocephalians and their aquatic ancestry, we may conclude that the aquatic lifestyle is primitive for stegocephalians.
Extant Tetrapods
To infer the lifestyle of early stegocephalians, scientists have developed a method based on appendicular long bone microanatomy (especially on the femur, humerus, radius, and tibia) as seen in cross section. These bones initially develop as prechondrogenic condensations (i.e., they form before cartilage appears) that soon start secreting a cartilaginous matrix. Then the membrane that surrounds the cartilage (the perichondrium) transforms into a periosteum that secretes a bony matrix (except at the articulations, which must remain cartilaginous to allow movement between bones). The initial cartilage of vaguely cylindrical shape (it is generally thinner in the midshaft than near the articular ends) is thus encased in a bony cylinder that thickens quickly. This cartilage is often entirely resorbed in ontogeny. Bones elongate mostly through cartilage development at the articular ends, which we call epiphyses, by opposition to the shaft, which we call the diaphysis. We also recognize a transitional zone between epiphyses and diaphysis, which we call the metaphysis. The epiphyses are generally covered in cartilage, but they may have an ossified center, especially in mammals (but not in amphibians). The metaphysis consists of a thin layer of compact bone, called the cortical compacta (or sometimes “cortex”) that surrounds deeper spongy bone (the medullary spongiosa). The diaphysis usually has a thicker cortical compacta and less (if any) medullary spongiosa, at least in most terrestrial tetrapods.
We have long known that the cross-sectional aspect of the long bones reflects (among other things) the habitat. Terrestrial tetrapods generally have a moderately thick, compact cortex (Fig. 4.10A), whereas amphibious tetrapods have more compact bones with a thicker cortex (Fig. 4.10B). Aquatic tetrapods may display either very compact bone with a thick cortex, or spongy bone (Fig. 4.10C). The greatest habitat-related differences occur in the mid-diaphyseal (midshaft) region, because the metaphysis and the epiphyses nearly always have a thin cortex and an extensive spongiosa.
Figure 4.10. Long bones of extant tetrapods of various lifestyles. Mid-diaphyseal cross sections of appendicular long bones. These drawings are not to scale.
PARTS: A, humerus, Cervus elaphus (deer, terrestrial; maximal diameter: 29.2 mm). B, femur, Ornithorhynchus anatinus (platypus, amphibious; maximal diameter: 6.0 mm). C, humerus, Delphinus delphis (dolphin, aquatic; maximal diameter: 33.3 mm).
Recent research on a correlation between habitat use and bone microanatomy required thin sections (about 0.05 mm) of long bones in mid-diaphyseal cross sections. These sections were then digitized and analyzed using a custom software application called Bone Profiler that divides the section surface into thousands of small polygons (Fig. 4.11). The program then measures compactness (the ratio between surface covered by bone and total surface) within each polygon. It then fits a mathematical function to the data (Fig. 4.12); that function represents the compactness profile, which describes compactness (Y) as a function of position along the section radius (X), from the center (X = 0) to the periphery (X = 1). The model requires estimating four parameters: Min, the lower asymptote, which generally reflects compactness in the section center; Max, the upper asymptote, which generally reflects compactness in the superficial cortex; P, the position of the inflexion point on the x-axis, which reflects the diameter of the medullary spongiosa; and S, which reflects the width of the transition zone between medullary and cortical regions. This model captures more information from the bone sections than simpler characters that were used in previous studies (Laurin et al., 2006), such as global compactness or the corticodiaphyseal index (which is the ratio between cortical compactness and diameter), or CDI for short. The relationship between CDI and P is simple: CDI = 1 – P, so using the new model allows comparison with results from previous studies.
Figure 4.11. Sampling scheme used by Bone Profiler to model long bone cross sections. The program delimits 51 concentric zones (Z1 to Z51), which measure 2% of the section diameter, except for the two most superficial zones, which measure only 1% of the diameter, to better model sections with a very thin cortex (as in birds and pterosaurs). Bone Profiler also delimits 60 radial sectors of 6° width (360°/60). The intersection between these zones and sectors delimits 3060 polygons (51 × 60) whose compactness is calculated. The parameters of the compactness profile (S, P, Min and Max) can be estimated either on the whole section in a single step (thus yielding a single value for each), or on each radial sector (which yields 60 values per parameter, and allows estimating the variance of the parameters reflecting heterogeneities on the section). Reproduced from Laurin et al. (2004).
The four parameters of the function (S, P, Min, and Max) and the habitat of many species of vertebrates can then be analyzed through statistical tests to determine which combinations of these parameters characterize aquatic, amphibious, and terrestrial species, and more importantly, to show that this relationship is statistically significant. A database on bone microanatomy encompassing more than 200 extant tetrapod species is currently being analyzed (Laurin et al., 2009). These analyses should improve our understanding of bone microanatomy evolution that occurred as a response to habitat shifts in tetrapods. They have already shown statistically significant differences among aquatic, amphibious, and terrestrial species. Flying vertebrates were not studied, because flight imposes different constraints and flight capability can usually be assessed from morphology.
Figure 4.12. Compactness profile model. The x-axis reflects the position on the section (0 is the center, and 1, the periphery). In this hypothetical example, there would be a medullary spongiosa, as in figure 4.10C (Delphinus), but the cortex would be thicker. Modified from Laurin et al. (2004).
Paleozoic Stegocephalians
Bone microanatomical data can be used to infer the habitat of early stegocephalians. To do this, we need only obtain bone sections of the species whose lifestyle we wish to infer. A database of such sections encompassing more than 30 species of early stegocephalians has been compiled and is being analyzed (Kriloff et al., 2008). Once the inferences are available, habitat use can be optimized onto a time-calibrated tree, and the history of the invasion of land by vertebrates can be analyzed. This phylogeny should be based on the ages of all known stegocephalian fossils, not only on the species whose habitat has been inferred, and molecular ages can also be used (Laurin et al., 2009). This method is necessary to avoid underestimating the geological age of the hypothetical ancestors on the tree, and thus to get an accurate age of the habitat shifts.
Figure 4.13. Section of the radius of Ophiacodon. This section comes from an Early Permian specimen (about 290 Ma). Note the thick cortex.
So far, only preliminary inferences have been obtained. The most interesting are briefly discussed here. One concerns Ophiacodon (Fig. 4.13), one of the oldest known amniotes (Late Carboniferous and Early Permian), which from the 1950s to the early 1960s was considered very similar to the ancestor of all amniotes. At that time, the first amniotes were thought to have retained an aquatic or amphibious lifestyle, and Ophiacodon, whose morphology seemed compatible with these lifestyles (Fig. 4.14), was often cited as the best evidence supporting this hypothesis. The description of the earliest known amniote fauna from Joggins, Nova Scotia, led to the rejection of Romer’s (1957) ideas because Hylonomus, the best-preserved reptile from that locality, was plausibly interpreted as a terrestrial animal (Carroll, 1964). However, Joggins may represent an unusual taphonomic assemblage of terrestrial and amphibious taxa, and this does not preclude the existence of aquatic or amphibious amniotes elsewhere (Canoville and Laurin, 2010). Bone microanatomy supports the hypothesis that Ophiacodon retained an amphibious to aquatic lifestyle (Germain and Laurin, 2005; Kriloff et al., 2008), because the long bones of this amniote were relatively compact (Fig. 4.13) or display an extensive medullary spongiosa, as in many aquatic vertebrates (Fig. 4.10B, C). Microanatomical data suggest an amphibious lifestyle for Captorhinus, a taxon usually considered fairly terrestrial, and various paleobiological inferences suggest that the earliest amniotes may have retained the amphibious lifestyle of their distant ancestors (Canoville and Laurin, 2010), thus adding support for Romer’s (1957) views.
Figure 4.14. Ophiacodon in its habitat. This amniote, from the Late Carboniferous and Early Permian, was long considered amphibious or even aquatic. Drawing by Douglas Henderson initially published by Czerkas and Czerkas (1990). Reproduced with permission.
Another preliminary inference was obtained for Doleserpeton, an Early Permian stegocephalian that was long considered closely related to extant amphibians (Bolt, 1969; Trueb and Cloutier, 1991), although recent research suggests that it is a stem tetrapod (Vallin and Laurin, 2004; Marjanović and Laurin, 2008). It is generally thought to have been terrestrial (Bolt, 1977), like most other dissorophoids. Again, bone microanatomy confirms previous interpretations because the long bones display a moderately thick, compact cortex and a large medullary cavity (Fig. 4.15) typical of terrestrial vertebrates.
Figure 4.15. Cross section of a Doleserpeton femur. This section shows the compact, moderately thick cortex (thinner than in Ophiacodon) and the large medullary cavity (A). The corresponding compactness profile produced by Bone Profiler is typical of terrestrial vertebrates (B). This section is from an Early Permian specimen (about 290 Ma).
An example of the ultimate goal of this research is provided by a recent study that inferred the habitat of early stegocephalians based on several characters (Fig. 2.5). These preliminary results include a time-calibrated tree and trace the complex history of the invasion of land by vertebrates. These results provide an initial hypothesis that can be tested as more reliable results are obtained. More than one acquisition of amphibious and terrestrial lifestyles may have occurred, and several taxa independently returned to a more aquatic lifestyle.
The vertebrate skeleton is composed of a dermal portion, which was superficial in the first vertebrates and is largely restricted to the skull and shoulder girdle in extant tetrapods, and an endoskeletal component, which makes up the vertebral column, most of the appendicular bones, and part of the skull. The autopod (hand and foot) is entirely endoskeletal.
Work on Hox gene expression in vertebrate appendicular development has been claimed to show that the autopod is a neomorph. Unfortunately, since these data come from few species, their interpretation is difficult, but the latest work casts doubt on these conclusions and suggests that the loss of the metapterygial axis in teleosts is responsible for the differences observed between Danio and tetrapods.
Morphologists and paleontologists initially considered that the autotpod was homologous with the fin tip (early in the 20th century), but later (from the 1940s onwards) concluded that it was probably a neomorph. Recently, some authors reverted to the initial hypothesis of homology between autopod and distal fin, but the available data are too scanty to allow a definitive conclusion.
Digits (fingers and toes) appeared between 380 and 360 Ma ago. This date is imprecise because fragmentary fossils from that period show the presence of close relatives of limbed vertebrates, but the presence of digits in these enigmatic species is uncertain. The oldest skeletal remains of digits are from the Fammenian (360 Ma ago), although stegocephalian trackways showing digits were recently described from the Middle Devonian. The first limbs were polydactylous and possessed six to eight digits. Pentadactyly appeared in the Carboniferous, probably only once, in stem tetrapods.
The first stegocephalians were probably primitively aquatic. Thus, the autopod is not an adaptation to life on dry land; instead, it should be considered an exaptation. New methods to infer the lifestyle of early stegocephalians are under development. One of them uses long bone microanatomy and statistical study of vertebrates (mostly extant) of known lifestyle.
