The Devonian period is traditionally known as the "Age of Fish", marking the diversification of numerous extinct and modern major fish groups.^[7] Among them were the early bony fishes, who diversified and spread in freshwater and brackish environments at the beginning of the period. The early types resembled their cartilaginous ancestors in many features of their anatomy, including a shark-like tailfin, spiral gut, large pectoral fins stiffened in front by skeletal elements and a largely unossified axial skeleton.^[8]

They did, however, have certain traits separating them from cartilaginous fishes, traits that would become pivotal in the evolution of terrestrial forms. With the exception of a pair of spiracles, the gills did not open singly to the exterior as they do in sharks; rather, they were encased in a gill chamber stiffened by membrane bones and covered by a bony operculum, with a single opening to the exterior. The cleithrum bone, forming the posterior margin of the gill chamber, also functioned as anchoring for the pectoral fins. The cartilaginous fishes do not have such an anchoring for the pectoral fins. This allowed for a movable joint at the base of the fins in the early bony fishes, and would later function in a weight bearing structure in tetrapods. As part of the overall armour of rhomboid cosmin scales, the skull had a full cover of dermal bone, constituting a skull roof over the otherwise shark-like cartilaginous inner cranium. Importantly, they also had a pair of ventral paired lungs,^[9] a feature lacking in sharks and rays.

It was assumed that fishes to a large degree evolved around reefs, but since their origin about 480 million years ago, they lived in near-shore environments like intertidal areas or permanently shallow lagoons and didn't start to proliferate into other biotopes before 60 million years later. A few adapted to deeper water, while solid and heavily built forms stayed where they were or migrated into freshwater.^[10][11] The increase of primary productivity on land during the late Devonian changed the freshwater ecosystems. When nutrients from plants were released into lakes and rivers, they were absorbed by microorganisms which in turn were eaten by invertebrates, which served as food for vertebrates. Some fish also became detritivores.^[12] Early tetrapods evolved a tolerance to environments which varied in salinity, such as estuaries or deltas.^[13]

The lung/swim bladder originated as an outgrowth of the gut, forming a gas-filled bladder above the digestive system. In its primitive form, the air bladder was open to the alimentary canal, a condition called physostome and still found in many fish.^[14] The primary function of swim bladder is not entirely certain. One consideration is buoyancy. The heavy scale armour of the early bony fishes would certainly weigh the animals down. In cartilaginous fishes, lacking a swim bladder, the open sea sharks need to swim constantly to avoid sinking into the depths, the pectoral fins providing lift.^[15] Another factor is oxygen consumption. Ambient oxygen was relatively low in the early Devonian, possibly about half of modern values.^[16] Per unit volume, there is much more oxygen in air than in water, and vertebrates (especially nektonic ones) are active animals with a higher energy requirement compared to invertebrates of similar sizes.^[17][18] The Devonian saw increasing oxygen levels which opened up new ecological niches by allowing groups able to exploit the additional oxygen to develop into active, large-bodied animals.^[16] Particularly in tropical swampland habitats, atmospheric oxygen is much more stable, and may have prompted a reliance of proto-lungs (performing essentially an evolved type of enteral respiration) rather than gills for primary oxygen uptake.^[19][20] In the end, both buoyancy and breathing may have been important, and some modern physostome fishes do indeed use their bladders for both.

To function in gas exchange, lungs require a blood supply. In cartilaginous fishes and teleosts, the heart lies low in the body and pumps blood forward through the ventral aorta, which splits up in a series of paired aortic arches, each corresponding to a gill arch.^[21] The aortic arches then merge above the gills to form a dorsal aorta supplying the body with oxygenated blood. In lungfishes, bowfin and bichirs, the swim bladder is supplied with blood by paired pulmonary arteries branching off from the hindmost (6th) aortic arch.^[22] The same basic pattern is found in the lungfish Protopterus and in terrestrial salamanders, and was probably the pattern found in the tetrapods' immediate ancestors as well as the first tetrapods.^[23] In most other bony fishes the swim bladder is supplied with blood by the dorsal aorta.^[22]

In modern tetrapods, three important breathing mechanisms are conserved from early ancestors, the first being a CO₂/H+ detection system. In modern tetrapod breathing, the impulse to take a breath is triggered by a buildup of CO₂ in the bloodstream and not a lack of O₂.^[24] A similar CO₂/H+ detection system is found in all Osteichthyes, which implies that the last common ancestor of all Osteichthyes had a need of this sort of detection system.^[24][25] The second mechanism for a breath is a surfactant system in the lungs to facilitate gas exchange. This is also found in all Osteichthyes, even those that are almost entirely aquatic.^[26][27] The highly conserved nature of this system suggests that even aquatic Osteichthyes have some need for a surfactant system, which may seem strange as there is no gas underwater^{[dubious – discuss]}. The third mechanism for a breath is the actual motion of the breath. This mechanism predates the last common ancestor of Osteichthyes, as it can be observed in lampreys, who belong to Agnatha, the sister clade to all other vertebrates. In lampreys, this mechanism takes the form of a "cough", where the lamprey shakes its body to allow water flow across its gills. When CO₂ levels in the lamprey's blood climb too high, a signal is sent to a central pattern generator that causes the lamprey to "cough" and allow CO₂ to leave its body.^[28][29] This linkage between the CO₂ detection system and the central pattern generator is extremely similar to the linkage between these two systems in tetrapods, which implies homology.

The nostrils in most bony fish differ from those of tetrapods. Normally, bony fish have four nares (nasal openings), one naris behind the other on each side. As the fish swims, water flows into the forward pair, across the olfactory tissue, and out through the posterior openings. This is true not only of ray-finned fish but also of the coelacanth, a fish included in the Sarcopterygii, the group that also includes the tetrapods. In contrast, the tetrapods have only one pair of nares externally but also sport a pair of internal nares, called choanae, allowing them to draw air through the nose. Lungfish are also sarcopterygians with internal nostrils, but these are sufficiently different from tetrapod choanae that they have long been recognized as an independent development.^[30]

The evolution of the tetrapods' internal nares was hotly debated in the 20th century. The internal nares could be one set of the external ones (usually presumed to be the posterior pair) that have migrated into the mouth, or the internal pair could be a newly evolved structure. To make way for a migration, however, the two tooth-bearing bones of the upper jaw, the maxilla and the premaxilla, would have to separate to let the nostril through and then rejoin; until recently, there was no evidence for a transitional stage, with the two bones disconnected. Such evidence is now available: a small lobe-finned fish called Kenichthys, found in China and dated at around 395 million years old, represents evolution "caught in mid-act", with the maxilla and premaxilla separated and an aperture—the incipient choana—on the lip in between the two bones.^[31] Kenichthys is more closely related to tetrapods than is the coelacanth,^[32] which has only external nares; it thus represents an intermediate stage in the evolution of the tetrapod condition. The reason for the evolutionary movement of the posterior nostril from the nose to lip, however, is not well understood.

The relatives of Kenichthys soon established themselves in the waterways and brackish estuaries and became the most numerous of the bony fishes throughout the Devonian and most of the Carboniferous. The basic anatomy of the group is well known thanks to the very detailed work on Eusthenopteron by Erik Jarvik in the second half of the 20th century.^[33] The bones of the skull roof were broadly similar to those of early tetrapods and the teeth had an infolding of the enamel similar to that of labyrinthodonts. The paired fins had a build with bones distinctly homologous to the humerus, ulna, and radius in the fore-fins and to the femur, tibia, and fibula in the pelvic fins.^[34]

There were a number of families: Rhizodontida, Canowindridae, Elpistostegidae, Megalichthyidae, Osteolepidae and Tristichopteridae.^[35] Most were open-water fishes, and some grew to very large sizes; adult specimens are several meters in length.^[36] The Rhizodontid Rhizodus is estimated to have grown to 7 meters (23 feet), making it the largest freshwater fish known.^[37]

While most of these were open-water fishes, one group, the Elpistostegalians, adapted to life in the shallows. They evolved flat bodies for movement in very shallow water, and the pectoral and pelvic fins took over as the main propulsion organs. Most median fins disappeared, leaving only a protocercal tailfin. Since the shallows were subject to occasional oxygen deficiency, the ability to breathe atmospheric air with the swim bladder became increasingly important.^[6] The spiracle became large and prominent, enabling these fishes to draw air.

The tetrapods have their root in the early Devonian tetrapodomorph fish.^[38] Primitive tetrapods developed from an osteolepid tetrapodomorph lobe-finned fish (sarcopterygian-crossopterygian), with a two-lobed brain in a flattened skull. The coelacanth group represents marine sarcopterygians that never acquired these shallow-water adaptations. The sarcopterygians apparently took two different lines of descent and are accordingly separated into two major groups: the Actinistia (including the coelacanths) and the Rhipidistia (which include extinct lines of lobe-finned fishes that evolved into the lungfish and the tetrapodomorphs).

The oldest known tetrapodomorph is Tungsenia from China, dated at around 409 million years old. Two of the earliest tetrapodomorphs, dating from 380 Ma, were Gogonasus and Panderichthys.^[39] They had choanae and used their fins to move through tidal channels and shallow waters choked with dead branches and rotting plants.^[40] Their fins could have been used to attach themselves to plants or similar while they were lying in ambush for prey. The universal tetrapod characteristics of front limbs that bend forward from the elbow and hind limbs that bend backward from the knee can plausibly be traced to early tetrapods living in shallow water. Pelvic bone fossils from Tiktaalik shows, if representative for early tetrapods in general, that hind appendages and pelvic-propelled locomotion originated in water before terrestrial adaptations.^[41]

Another indication that feet and other tetrapod traits evolved while the animals were still aquatic is how they were feeding. They did not have the modifications of the skull and jaw that allowed them to swallow prey on land. Prey could be caught in the shallows, at the water's edge or on land, but had to be eaten in water where hydrodynamic forces from the expansion of their buccal cavity would force the food into their esophagus.^[42]

It has been suggested that the evolution of the tetrapod limb from fins in lobe-finned fishes is related to expression of the HOXD13 gene or the loss of the proteins actinodin 1 and actinodin 2, which are involved in fish fin development.^[43][44] Robot simulations suggest that the necessary nervous circuitry for walking evolved from the nerves governing swimming, utilizing the sideways oscillation of the body with the limbs primarily functioning as anchoring points and providing limited thrust.^[45] This type of movement, as well as changes to the pectoral girdle are similar to those seen in the fossil record, can be induced in bichirs by raising them out of water.^[46]

A 2012 study using 3D reconstructions of Ichthyostega concluded that it was incapable of typical quadrupedal gaits. The limbs could not move alternately as they lacked the necessary rotary motion range. In addition, the hind limbs lacked the necessary pelvic musculature for hindlimb-driven land movement. Their most likely method of terrestrial locomotion is that of synchronous "crutching motions", similar to modern mudskippers.^[47] (Viewing several videos of mudskipper "walking" shows that they move by pulling themselves forward with both pectoral fins at the same time (left & right pectoral fins move simultaneously, not alternatively). The fins are brought forward and planted; the shoulders then rotate rearward, advancing the body & dragging the tail as a third point of contact. There are no rear "limbs"/fins, and there is no significant flexure of the spine involved.)

The first tetrapods probably evolved in coastal and brackish marine environments, and in shallow and swampy freshwater habitats.^[48] Formerly, researchers thought the timing was towards the end of the Devonian. In 2010, this belief was challenged by the discovery of the oldest known tetrapod tracks named the Zachelmie trackways, preserved in marine sediments of the southern coast of Laurasia, now Świętokrzyskie (Holy Cross) Mountains of Poland. They were made during the Eifelian age, early Middle Devonian. The tracks, some of which show digits, date to about 395 million years ago—18 million years earlier than the oldest known tetrapod body fossils.^[49] Additionally, the tracks show that the animal was capable of thrusting its arms and legs forward, a type of motion that would have been impossible in tetrapodomorph fish like Tiktaalik. The animal that produced the tracks is estimated to have been up to 2.5 metres (8.2 ft) long with footpads up to 26 centimetres (10 in) wide, although most tracks are only 15 centimetres (5.9 in) wide.^[50]

The new finds suggest that the first tetrapods may have lived as opportunists on the tidal flats, feeding on marine animals that were washed up or stranded by the tide.^[49] Currently, however, fish are stranded in significant numbers only at certain times of year, as in alewife spawning season; such strandings could not provide a significant supply of food for predators. There is no reason to suppose that Devonian fish were less prudent than those of today.^[51] According to Melina Hale of University of Chicago, not all ancient trackways are necessarily made by early tetrapods, but could also be created by relatives of the tetrapods who used their fleshy appendages in a similar substrate-based locomotion.^[52]

• Benton, Michael J. (5 February 2009). Vertebrate Palaeontology. John Wiley & Sons. ISBN 978-1-4051-4449-0.
• Clack, Jennifer A. (2012). Gaining Ground: The Origin and Evolution of Tetrapods. Indiana University Press. ISBN 978-0-253-35675-8.