Palaeophysiology of pH regulation in tetrapods

Abstract

The involvement of mineralized tissues in acid–base homeostasis was likely important in the evolution of terrestrial vertebrates. Extant reptiles encounter hypercapnia when submerged in water, but early tetrapods may have experienced hypercapnia on land due to their inefficient mode of lung ventilation (likely buccal pumping, as in extant amphibians). Extant amphibians rely on cutaneous carbon dioxide elimination on land, but early tetrapods were considerably larger forms, with an unfavourable surface area to volume ratio for such activity, and evidence of a thick integument. Consequently, they would have been at risk of acidosis on land, while many of them retained internal gills and would not have had a problem eliminating carbon dioxide in water. In extant tetrapods, dermal bone can function to buffer the blood during acidosis by releasing calcium and magnesium carbonates. This review explores the possible mechanisms of acid–base regulation in tetrapod evolution, focusing on heavily armoured, basal tetrapods of the Permo-Carboniferous, especially the physiological challenges associated with the transition to air-breathing, body size and the adoption of active lifestyles. We also consider the possible functions of dermal armour in later tetrapods, such as Triassic archosaurs, inferring palaeophysiology from both fossil record evidence and phylogenetic patterns, and propose a new hypothesis relating the archosaurian origins of the four-chambered heart and high systemic blood pressures to the perfusion of the osteoderms.

This article is part of the theme issue ‘Vertebrate palaeophysiology’.

1. Introduction

The water-to-land transition is one of the most important steps in vertebrate evolution and certainly a pivotal one for our own eventual appearance on the planet. A number of key anatomical and physiological changes were necessitated by this change in physical medium, none of which have been fully accomplished by any semi-terrestrial fish, and even extant amphibians have limitations in some of these aspects of terrestrial life. These changes include respiration (i.e. gaseous exchange, but also mode of ventilation), support (and locomotion), nitrogenous excretion and reproduction.

Popular accounts of this transition often consider it from the perspective of our own experience of going from land into the water. Our mode of nitrogenous excretion works as well underwater as it does on land, and we are not immersed for long enough to be concerned about reproduction. Support is also of little concern, as we find ourselves buoyant in the water, although swimming is energetically expensive compared to walking or running on land. Rather, our main concern is respiration, which we perceive as getting enough oxygen. Note, however, that our modes of obtaining oxygen underwater, whether via a snorkel (of limited length) or an oxygen tank, also reflect the fact that our mode of lung ventilation by expanding our rib cage also does not work well under the pressure of a column of water.

But for a fish coming onto land, obtaining oxygen is potentially not a problem, as oxygen is much more readily available in air (around 15 times more) than in water. Fish suffocate in air only because their ventilatory surfaces, the gills, collapse under gravity, resulting in insufficient area for gas exchange. But many fish have lungs (as did the earliest tetrapods, or land vertebrates, as will be explained below) and even some fish without lungs (e.g. mudskippers) can use other specialized surfaces such as the inside of the opercular chamber, and their skin is also highly vascularized [1]. (The small size of mudskippers also provides a favourable surface area to volume ratio for cutaneous gas exchange.)

The real problem for a fish coming onto land is not obtaining oxygen, but excreting carbon dioxide. We do not perceive carbon dioxide loss as a problem for ourselves, because our lung ventilation rate is sufficiently high that we can eliminate it in the course of obtaining oxygen. But a little appreciated fact is that our rate of lung ventilation is actually determined by our need to remove carbon dioxide from our bodies. In fact, we can decrease our normal ventilation rate by almost half and still have haemoglobin that is 90% saturated [2]. For a terrestrial fish, carbon dioxide elimination would be the critical issue, the build-up of carbon dioxide in the body resulting in the accumulation of protons in the body fluids, also called acidosis. Land vertebrates have an additional problem: not only do they encounter respiratory acidosis from carbon dioxide excretion via regular cellular processes, but they also must exert themselves more to locomote on land and hence also encounter metabolic acidosis from the build-up of lactic acid. Swimming is more expensive than walking for humans, but that is because we are not adapted for aquatic locomotion. For an animal with the appropriate anatomy (streamlined, using axial movements rather than limbs), swimming is the most efficient mode of locomotion [3]. Both body support and locomotion on land would be especially challenging for the earliest tetrapods, which had not yet fully adapted their skeleto-muscular system for such ventures.

It was this consideration of the problem of carbon dioxide elimination in tetrapod evolution, combined with (at the time) new information about the physiology of some extant semi-aquatic reptiles, that led two of us (C.M.J. and D.E.W.) to propose a novel role for the dermal bone cover of many of the early tetrapods: as a means of coping with terrestrially induced acidosis [4]. Here, we review and update this hypothesis, and also extend the issue of the possible metabolic role of dermal bone in tetrapod evolution to extinct archosaurian reptiles (i.e. early members of the crocodile and dinosaur lineages), courtesy of the contribution of J.G.N.

2. The cast of characters of early tetrapods

The phylogeny in figure 1 is a standard representation of the interrelationships of the Sarcopterygii (i.e. fleshy-finned bony vertebrates, or osteichthyians). Coelacanths are not shown here, although they are placed basal to lungfishes (see [5]). Note that the term ‘tetrapod’ is officially reserved for those taxa within the crown group: that is, within the phylogenetic bracket of extant forms. However, we shall use the term ‘tetrapod’ informally to refer to all tetrapodamorphs with limbs (‘tetrapodomorphs’ also include those fish more closely related to tetrapods than to lungfishes) and the term ‘early tetrapod’ to refer to all tetrapods that are not included within either the phylogenetic bracket of extant amphibians (= Lissamphibia) or the phylogenetic bracket of extant amniotes.

Figure 1.

PHYL of selected members of the Sarcopterygii (fleshy-finned bony vertebrates, coelacanth excluded) to show the application of phylogenetic inference. Lissamphibia includes extant amphibians (frogs, salamanders, caecelians); Sauropsida includes reptiles and birds (lizards and snakes would be basal to the shown grouping of crocodiles and turtles, and birds would be the sister group of crocodiles); Synapsida includes mammals and their extinct relatives; x, stem tetrapod Acanthostega gunnari; y, stem amphibian (eryopid) ‘Pelosaurus’ (now Onchiodon labyrinthicus); z, stem amniote Seymouria baylorensis. All images from Wikimedia Commons except Acanthostega, permission from Mike Coates (artist). Grey line, absence of feature; solid black line, presence of feature; dotted grey line, presence of feature possible, but equivocal. (a) All extant taxa possess a feature: feature inferred to be present in all extinct taxa. (b) Only extant amniotes possess a feature: feature inferred to be absent in stem tetrapods and stem lissamphibians; situation in stem amniotes equivocal. (c) Feature present only in lissamphibians; situation in early tetrapods equivocal. (d) Feature present in some extant tetrapods; situation in early tetrapods equivocal.

Taxa that are basal to the phylogenetic bracket of extant tetrapods are referred to as ‘stem tetrapods’ (position ‘x’ in figure 1). Taxa within the crown group that are more closely related to modern amphibians (= Lissamphibia: frogs, salamanders and caecilians) than they are to amniotes, but that are more basal to that grouping, are referred to as ‘stem amphibians’ (position ‘y’ in figure 1). Likewise, taxa that are more closely related to the grouping of extant amniotes (mammals, birds and reptiles) than they are to extant amphibians are referred to as ‘stem amniotes’ (position ‘z’ in figure 1).

These early tetrapods were formerly referred to as labyrinthodonts (the larger forms) or lepospondyls (the smaller forms). However, the term ‘labyrinthodont’, at least, has been abandoned as it is now known to be a polyphyletic grouping of stem tetrapods, stem amphibians and stem amniotes. Lepospondyls have been variously regarded as a monophyletic or polyphyletic grouping. Witzmann [6] shows them as a monophyletic grouping placed as the sister group of amniotes; more recent work places some of them (the lizard-like microsaurs) within the Amniota and others (the snake-like aïstopods) as stem tetrapods [7]. We consider lepospondyls here only within the context of the distribution of integument types in association with aquatic or terrestrial habitat in table 1.

Table 1.

Integument type in early tetrapods in association with habitat (aquatic or terrestrial). Integument type is from Witzmann [8]. Taxa in bold face have at least some evidence for the presence of internal gills (from [6,9]). Taxa underlined are small forms (similar size to most extant lissamphibians).

integument type	aquatic or semi-aquatic taxa	semi-terrestrial or terrestrial taxa
1. rhombic overlapping bony scales dentine and enamel lost	derived tetrapodomorph fishes [10] e.g. Panderichthys, Elpistostege, Tiktaalik	none
2. thinner scales with the loss of basal isopedine layer	stem tetrapods [11–15] e.g. Acanthostega, Crassigyrinus, Eucritta, Tulerpeton, Whatcheeria	stem tetrapods [16] Pederpes (whatcheerid)
‘standard’ early tetrapod dermal scalation: gastral scales (chevron-shaped), plus dorsal scales (oval); some degree of dermal sculpturing	Palaeozoic stem amphibians [17–20] e.g. Archegosaurus, Australerpeton, Dvinosaurus, Edops, Eryops, Sclerocephalus, Trimemorhachis	Palaeozoic stem amphibians [18,20] e.g. Dendrerpeton, Eoscopus, Zatrachys
	Mesozoic stem amphibians [18] basal stereospondyls: e.g. Mahavisaurus, Tertremoides
	stem amniotes [21–23] e.g. Eogyrinus, Silvanerpeton, Utegenia	stem amniotes [24,25] Discosauriscus, Solenodonsaurus, Gephyrostegus
	lepospondyls [8,18] some microsaurs, nectrideans	lepospondyls [8,18] some microsaurs
3. heavily sculptured dermal bone, including large dorsal osteoderms	stem tetrapods [8,26] baphetids (e.g. Baphetes) and colosteids (e.g. Greererpeton)	Palaeozoic stem amphibians [27,28] Peltobatrachus dissorophids: inc. Cacops, Dissorophus, Platyhystrix
	Mesozoic stem amphibians [17] plagiosaur stereospondyls: inc. Gerrothorax, Plagiosuchus	Mesozoic stem amphibians (all stereospondyls) [29,30] Laidlaeria, Sclerothorax
	stem amniotes [31] some chroniosuchians: inc. Madygenerpeton	stem amniotes some chroniosuchians [17,32]: inc. Chroniosaurus, Bystrowiella
4. dermal scales greatly reduced or lost	Palaeozoic stem amphibians most branchiosaurs [17,33]	Palaeozoic stem amphibians amphibamid dissorophoids [34] inc. Amphibamus, Doleserpeton
	Mesozoic stem amphibians [17,18,21] (all stereospondyls) some capitosaurs and metoposaurs, most trematosaurs	Mesozoic stem amphibians [35] Micropholis (amphibamid)
		stem amniotes (R Reisz 2010, personal communication) [36] Diadectes, Seymouria
	lepospondyls [8,18] aïstopods, some microsaurs	lepospondyls [8,18] some microsaurs

The listing of types of early tetrapods follows Witzmann [6] except when noted, and the order reflects the systematic ordering from more basal to more derived. Stem tetrapods include Late Devonian forms (including Acanthostega, Ichthyostega and Tulerpeton), plus Carboniferous forms: colosteids, adelogyrinids, whatcheerids and baphetids. Stem amphibians are termed temnospondyls: they are mainly known from the Permian and Carboniferous periods, but some derived fully aquatic forms (stereospondyls) are known from the Mesozoic. Temnospondyls include: edopoids, dendrerpetontids, eryopids, stereospondylids, dvinosaurids, zatracheids and dissorophids (containing the ancestry of the Lissamphibia). Stem amniotes are mainly known from the Permian and Carboniferous; they include chroniosuchids (basal according to Witzmann & Schoch [37]; which are also known from the Triassic), anthracosaurs, gephyrostegids, seymouriamorphs and diadectomorphs (possibly synapsid-related amniotes [38]).

3. The physiology of early tetrapods: inferences from phylogeny and the fossil record

During the earlier part of the twentieth century, early tetrapods were referred to as ‘amphibians’ and they were assumed to be similar to extant amphibians in many respects. Nowadays, we appreciate, both from the benefit of phylogenetic thinking and from new fossil evidence, that while lissamphibians may retain basal tetrapod attributes in certain respects (e.g. in their mode of reproduction), in other respects they are highly specialized in their own right (e.g. in the use of a projectile tongue to obtain food). Witzmann [6] reviews earlier ideas about how early tetrapods breathed (i.e. performed gaseous exchange), many of which have been refuted by new evidence. We will here discuss what can be inferred about the biology of early tetrapods from both phylogeny and the fossil record.

It is worth noting much of extinct tetrapod anatomy and physiology can be inferred from the pattern of interrelationships (phylogeny) of extant and extinct tetrapods alone, without any specific information as to the condition in extinct forms. The distribution of the characters of extant tetrapods on that phylogeny informs us whether particular features are basal (generalized) or derived (specialized) for tetrapods as a whole, and from this, the situation in extinct forms can be inferred. Below we address certain aspects of the probable physiology of early tetrapods, first by considering the information from phylogeny alone, and then by considering additional evidence from the fossils.

One particular phylogenetic approach is the application of the extant phylogenetic bracket (EPB [39]). This is a methodology of determining if a feature that does not fossilize (soft anatomy, physiology, behaviour) is likely to be present in extinct taxa by assessing the situation in its extant relatives to determine whether that feature was inherited from their last common ancestor. If the extinct taxon in question is ‘bracketed’ by extant taxa that possess it, providing that the feature in question is likely to be homologous between those extant taxa, then it may be inferred that the extinct taxon also possessed that feature.

A simple example here is the case of the possession of lungs. Figure 1a shows how the EPB can be used to infer the presence of lungs in extinct taxa: we can be confident that lungs are homologous in tetrapods and lungfish because of similarity in patterns of circulatory supply, among anatomical aspects [40]. Because stem tetrapods (e.g. the aquatic forms of the Late Devonian) are ‘bracketed’ between lungfish and tetrapods, we can confidently infer that lungs were present in those extinct taxa (and, of course, in all extinct crown tetrapods that are bracketed by the grouping of extant amphibians and amniotes). From the information in this cladogram alone, we would be unable to deduce whether lungs were present in any fish basal to the lungfish (although further phylogenetic evidence allows us to infer that lungs are a basal bony fish feature, lost in many extant forms).

Inferences of palaeophysiological features follow a different pattern, mostly represented by the case of figure 1b. Here, the derived feature in question is limited to amniotes among extant forms: the EPB would include extinct forms within the amniote crown group (none shown in this figure). We can make no inference from the phylogeny alone about the presence or absence of this feature in stem amniotes, but the absence of the feature in extant amphibians and lungfishes would allow us to infer its absence in stem amphibians and stem tetrapods. However, as will be discussed, fossil record evidence does allow us to determine the likely condition in at least some stem amniotes.

(a). Mode of obtaining oxygen

This is via the lungs and the gills in lungfishes, the lungs, gills (in larval or aquatic forms) and the skin in lissamphibians, but by the lungs alone in the amniotes. Extant amniotes also share other derived features in association with lung ventilation, such as internal complication of lung tissue and the presence of a trachea with cartilaginous rings [40]. This pattern of the distribution of features on the phylogeny (figure 1b) informs us that lungs plus gills is the basal (generalized or primitive) condition for sarcopterygians, and that the use of the lungs alone in amniotes is derived. The inference from the phylogeny is that stem tetrapods and stem amphibians retained the basal condition of using relatively unspecialized lungs, plus gills (the latter at least in larval forms: both lungfishes and lissamphibians have larvae with external gills, although these appear to have been convergent acquisitions in the two lineages [41]).

The condition in stem amniotes will be discussed later, in the consideration of the elimination of carbon dioxide. However, new fossil record evidence documents the presence of internal gills in stem tetrapods, many temnospondyls and possibly also in some basal stem amniotes ([6], and references therein; table 1). Thus, we now know that internal gills in some form, even if not as extensive as in the fish condition, were retained in many early tetrapods and may have been of use in carbon dioxide elimination when the animal was in the water.

Extant amphibians employ the skin for cutaneous gaseous exchange while in air, but amniotes do not. Semi-aquatic reptiles, such as turtles, can rely more on cutaneous gas exchange at cold temperatures when metabolic demands for oxygen uptake and carbon dioxide elimination can be met by low diffusion rates afforded by the integument [42]. Several groups of extant turtles (e.g. Trionychidae and Chelidae) possess highly vascularized surfaces (e.g. the pharynx or cloacal bursae) that facilitate aquatic gas exchange, but their utility is limited in air. Lungfishes may lose some carbon dioxide over the body surface, but neither amniotes nor lungfishes possess the thin, specialized and glandular skin of modern amphibians. This physiology cannot be inferred by means of the EPB to be present in any of the extinct forms considered here (figure 1c), although it could be used to infer the condition in extinct crown amphibians (not illustrated). But from the phylogeny alone, the condition in early tetrapods is equivocal: in theory, the extant lissamphibian condition could have been the basal tetrapod one.

The fossil record evidence is of help here. Most early tetrapods were considerably larger than extant lissamphibians. Because cutaneous gaseous exchange is dependent on a favourable surface area to volume ratio, the lissamphibian type of specialized cutaneous respiration is unlikely to have been present in the great majority of early tetrapods, and the relatively large body size would have been especially limiting for animals needing to eliminate carbon dioxide on land. In addition, the pattern of scalation in most early tetrapods (dermal bony scales inherited from their fish ancestors, and discussed later in more detail), with tightly set interlocking scales covering the entire body surface, suggests that the integument was thick (but see discussion below about the views of Bystrow [43]). A notable exception, with respect both to body size and scalation, is some of the dissorophid temnospondyls that contained the ancestry of lissamphibians (see later discussion).

(b). Mode of ventilating the lung

This is via positive-pressure (buccal) pumping in lungfish and lissamphibians, and via aspiration (costal ventilation) in amniotes (different amniote lineages may also acquire additional mechanisms, such as the diaphragm of mammals) [44]. The distribution of features on the phylogeny (figure 1b) shows that costal ventilation is a derived feature of crown amniotes and so, by inference, was absent in stem tetrapods and stem amphibians, which likely retained the basal condition of buccal pumping. Again, the condition in stem amniotes cannot be inferred. All extant tetrapods possess the transversus abdominus layer of hypaxial muscles, which is used to expel air from the lungs, but lungfish rely on the pressure of the water column for this function [44]. From the phylogeny (pattern not shown in figure 1), the use of the transverse abdominus muscles for exhalation can be inferred for stem amphibians and stem amniotes, but not for stem tetrapods (although common sense dictates that if these animals spent time on land, they would likely have used this mechanism).

The fossil record provides clarity for the issue of costal ventilation. The anatomy of the ribs of most early tetrapods is contraindicative of their use in lung ventilation [45]. Some early tetrapods (e.g. the stem tetrapod Ichthyostega and the stem amphibian Eryops) had sturdy rib cages, unlike lissamphibians, prompting earlier suggestions (e.g. [46]) that costal ventilation was a basal tetrapod feature. However, the rib cages of these early tetrapods were likely for support: the ribs of most early tetrapods are flat with little curvature, lacking the appropriate articulations with the vertebrae for movement (i.e. are single-headed rather than double-headed) and their distal ends do not provide evidence of cartilaginous extensions to attach to the sternum. The notable exception is the stem amniotes, which not only have more amniote-like rib cages, but also have smaller and less flattened heads, indicating the abandonment of buccal pumping [42]. Witzmann [6] notes that all stem amniotes appear to have had an amniote-like derived type of rib morphology.

(c). Mode of eliminating carbon dioxide

This follows from the first two palaeophysiological examples, from the phylogenetic pattern of figure 1b. Amniotes use costal ventilation to eliminate carbon dioxide, lungfishes and lissamphibians do not. Lungfishes can eliminate carbon dioxide by means of the gills (and the skin in part) and lissamphibians rely on transcutaneous elimination [47]. By the application of the EPB, stem tetrapods and stem amphibians could not have relied on their lungs for carbon dioxide elimination, but the physiology of stem amniotes cannot be inferred. The fossil record evidence, as for the case with lung ventilation, leads to the inference that stem amniotes, as well as the crown group ones, would have been able to get rid of most or all of their carbon dioxide via the lungs.

(d). Coping with respiratory acidosis

All vertebrates use buffers in the intra- and extracellular fluid as the first line of defence again pH changes. This includes bicarbonate, phosphate and the weakly acidic imidazole residue on the histidine side chain of most proteins. Amniotes adjust their ventilation to either retain or excrete excess carbon dioxide, termed hypo- or hyperventilation, respectively. Finally, they can use their kidney to defend against acidosis by excreting protons or generating new bicarbonate. Extant lungfishes and lissamphibians do not use these latter two mechanisms to the same extent as amniotes. Thus, using these latter mechanisms is likely to be a derived mode within tetrapods in general, confined to amniotes and stem amniotes. As neither lungfish nor extant lissamphibians evidence pH compensation via the kidney, from the pattern of features on the phylogeny (figure 1b), we can infer that this absence was also the condition in stem tetrapods and stem lissamphibians, but we cannot infer the presence or absence of renal pH compensation in stem amniotes.

Some extant tetrapods can perform pH compensation by means of using their bone tissue as a source of buffering protons and also to sequester lactate. This physiology is discussed in detail below, but in brief, among lissamphibians, some frogs can use mineralized tissues including their skeleton and endolymphatic lime deposits to buffer exercise-induced lactate (metabolic acidosis) [48] and carbon dioxide (respiratory acidosis) [49]. Among amniotes, some sauropsids can similarly use dermal bone to buffer carbon dioxide in the blood when submerged [50]. This physiology is discussed further in a later section. The use of bone in this fashion is not known in any fish.

The phylogenetic pattern for bone buffering is shown in figure 1d. Because the use of bone to compensate for acidosis is seen among both lissamphibians (frogs) and amniotes (some sauropsids), although in a different fashion in the two groups, some use of bone as a buffer may be a basal tetrapod feature and thus may have been present in many or all early tetrapods. Note that the basal lineages of both sauropsids and synapsids are shown here as ‘equivocal’: from this pattern of character distribution in the phylogeny, the condition in extinct synapsids basal to extant mammals, or extant and extinct sauropsids basal to the grouping of turtles and crocodiles (e.g. lizards and snakes) cannot be determined. Within the EPB of crocodiles and turtles, the condition is equivocal in stem avians (including dinosaurs).

In summary: from the phylogenetic distribution of characters alone, we can infer that all early tetrapods had lungs, but that many or most of them used buccal pumping for lung ventilation. The addition of fossil record evidence allows us to infer that some or all stem amniotes used costal aspiration and were better able to get rid of carbon dioxide on land than other early tetrapods. Most or all early tetrapods used the transverse abdominus muscles for exhalation. From the phylogeny alone, we might infer that it was possible that early tetrapods used cutaneous respiration in the fashion of extant amphibians, but considerations of body size and likely integument type render that unlikely. We can infer that most early tetrapods could not cope with acidosis via renal pH compensation (although the condition for stem amniotes cannot be determined). The overall inference for early tetrapods is that most of them would have had considerable difficulty managing carbon dioxide elimination on land. Finally, while the use of bone as a buffer for acidosis in early tetrapods cannot be determined from this phylogeny, it remains a possibility.

4. The integument of early tetrapods

As mentioned above, the skin of early tetrapods contained dermal scales. These were bony scales embedded in the dermis, and were basically a holdover from the ancestral fish condition. Such scales are sometimes also referred to as ‘osteoderms’, but this term is now usually reserved for scales that are thicker and more irregular in shape, with a pitted outer surface [8]. This type of integument is in contrast with that seen in extant tetrapods, where the skin is naked (lissamphibians—although caecilians have fish-like dermal scales, which may be a secondary acquisition [51]), or has a covering of material made from keratin, an epidermal tissue (scales in reptiles, hair in mammals, feathers in birds). Cranial dermal bones (a basal jawed fish feature) are retained in all tetrapods, and various extant tetrapods have dermal ossifications in the form of osteoderms, even some large species of frogs [52]. The carapace of turtles and the bony coverings of crocodiles and some extant mammals (e.g. armadillos and pangolins) are also dermal ossifications. We will discuss an example of the possession of osteoderms in amniotes, and implications for palaeophysiology, in a later section. But meanwhile, we note that while all tetrapods retain the capacity for dermal ossification, the integument of many early tetrapods was more fish-like than like that of extant tetrapods, even though some degree of dermal keratinization may have been a relatively basal tetrapod feature (at least at the level of the common ancestor of lissamphibians and amniotes), retained in the keratinized claws of certain frogs [53]. Stem amniotes may have been more like amniotes in their integument, but some early amniotes retained at least gastral scales [8].

Basal tetrapodomorph fishes had scales that comprised basal (dermal) bone with outer layers of dentine and enamel, but these outer layers were reduced in more derived forms and lost in the most derived tetrapodomorph fishes (the elpistostegids). Elpistostegid scales were overlapping, arranged in dorsal and ventral rows, with an intervening middle row, and the bone was somewhat sculptured ([8]; see below for discussion of sculptured dermal bone). The scales of the earliest tetrapods were different in certain respects: they were thinner, having lost the basal layer of isopedine (lamellar bone), the median row of scales was also absent and there was a greater extent of overlapping in the gastral scales. This different pattern of scalation may have allowed for greater axial flexibility in early tetrapods, in correlation with limb-based terrestrial locomotion, and the loss of the isopedine layer would have aided weight reduction on land [8].

The dermal bone of the skull and pectoral girdle in early tetrapods was sculptured with pits and ridges [8,17,48,50,54]; in some forms, the dorsal scales were similarly sculptured and these scales could be considered to be osteoderms [17]. This type of integument has been referred to as ‘dermal armour’ (figure 2).

Figure 2.

Dorsal view of a partial skull of Eryops megacephalus (AMNH FARB (American Museum of Natural History Fossil Amphibians, Reptiles, and Birds) 23529), showing portions of the jugal, prefrontal, frontal, postfrontal and postorbital bones (all dermal bones of the skull roof). The orbit is visible to the left and anterior is to the top of the image. The dermal bone shows extensive sculptured ornamentation.

There have been numerous speculations about the possible function of this ‘armour’ (e.g. mechanical and/or osmoregulatory protection, thermoregulation, axial strengthening, see [4]); but an important feature of these bones was that they were heavily vascularized [43], implying a physiological, rather than merely protective, function. (Note that keratinized dermal cover, being epidermal, cannot be vascularized.)

Bystrow ([43]; see discussion in [4,17]) proposed that the complex structure of this dermal armour was to support a vascular network related to cutaneous respiration: he envisaged ‘hydrophilous’ early tetrapods using this system of respiration in the water and ‘xerophilous’ forms relying on lungs on land. However, contra to Bystrow's views, we show here (see also [4]) that sculptured dermal armour is usually more prevalent in the terrestrial early tetrapods. It may be the case that vascularized dermal bone was important for underwater cutaneous gas exchange in some highly specialized aquatic early tetrapods (e.g. plagiosaurs, see later description), but the loss of carbon dioxide over the body surface is relatively easy for aquatic animals, due to the high solubility of carbon dioxide in water (see discussion below). The issue that we address here is the loss of carbon dioxide on land and in animals that were too large to have a favourable surface area to volume ratio for cutaneous respiration to be effective for this activity.

Note, as detailed below, a high degree of vascularization is necessary to perfuse bone that is involved in buffering acidosis, which would be a particular issue for terrestrial early tetrapods. Witzmann et al. [54], in discussing Bystrow's hypothesis, note that the vascularization of the dermal bone is mainly the interior portion, rather than the superficial portion, an observation that does not lend support to the notion of vascularization for cutaneous respiration, but which does support our hypothesis of bone perfusion for buffering.

5. How extant vertebrates manage acidosis

The need for an alternative physiological buffering strategy during the water-to-land transition was detailed in our previous paper [4], where a primary physiological challenge was minimizing acidification of the body fluids caused by excess accumulation of carbon dioxide, a process called respiratory acidosis, or by the accumulation of lactic acid, called lactic or metabolic acidosis. Respiratory acidosis occurs when an animal's ability to lose carbon dioxide to the environment is exceeded by its metabolic production, resulting in its accumulation in the body fluids as carbonic acid, which dissociates into a proton and bicarbonate at physiological pH (6.5–8.2, depending on temperature). Lactic acidosis may also be a problem; this occurs when an animal's ATP demand is exceeded by the capacity of the cardiorespiratory system to deliver oxygen for oxidative phosphorylation, resulting in the production of lactate from pyruvate and protons (normally consumed by oxidative phosphorylation), which accumulates simultaneously, producing what we know to be lactic acidosis [55]. These conditions exist in animals either during exhaustive exercise, when skeletal muscle ATP turnover rates are extremely high, or during hypoxia, when blood oxygen content and, therefore, delivery of oxygen to the tissues, is limited.

A vertebrate subjected to any kind of acidosis can rely on three mechanisms to defend body fluid pH, which vary in complexity and time-course. The most immediate line of defence is to use buffer molecules dissolved in the body fluids, which consist of partially dissociated weak acids whose dissociation constants (pKas) are near physiological pH. These include ${\mathrm{H}}_2\mathrm{C}{\mathrm{O}}_3/\mathrm{HC}{\mathrm{O}}_3^{\hspace{0.17em}\hspace{0.17em}-}$, ${\mathrm{H}}_2\mathrm{P}{\mathrm{O}}_4^{\hspace{0.17em}\hspace{0.17em}-}/\mathrm{HP}{\mathrm{O}}_4^{\hspace{0.17em}\hspace{0.17em}2-}$ and the imidazole side chains in the histidine residues of most proteins. All vertebrate groups possess this buffering ability, though variation exists depending on tissue and animal life history [56]. For example, skeletal muscle has high concentrations of histidine-dipeptides to buffer lactic acid [57] and air breathers have high concentrations of bicarbonate in their body fluids to buffer their higher whole-body carbon dioxide contents [58].

The second fastest line of defence is respiratory compensation, called a compensatory respiratory alkalosis, which requires the animal to increase its rate of ventilation relative to its rate of oxygen consumption in order to eliminate excess carbon dioxide. Among air breathers, this mechanism is best developed in costal ventilators (i.e. amniotes) because ventilation rates in early tetrapods and extant lissamphibians are constrained by the ventilatory mechanics of positive-pressure buccal pumping [44,45].

The third mechanism of compensation is non-respiratory, or metabolic, in origin, involving mainly the gills in water-breathers [59] and mainly the kidneys in amniotes. The response of these organs is to increase bicarbonate concentrations in the extracellular fluid while simultaneously secreting protons. This is the slowest form of compensation and occurs over hours. Extant lungfish and lissamphibians show incomplete compensation in response to acidosis [60–63] and so, by phylogenetic inference, early tetrapods also possessed a limited capacity to compensate acidosis via the kidney, as noted above.

A fourth mechanism of buffering that has come into focus in recent years involves mineralized tissues, especially dermal bone. This mechanism has been shown to be especially important in reptiles [64,65], but has also been demonstrated to be important for lissamphibians [48,49] and possibly in mammals. Much of what we understand comes from various studies in turtles, mostly in response to lactic acidosis that occurs during anoxia [66–71], but also in response to respiratory acidosis [50]. During acidosis, bone releases calcium and magnesium, along with some anions within bone, usually carbonate, and can sequester metabolically produced lactate [72].

The exact mechanism of this buffering is not fully understood, but likely involves passive ion exchange between the bone mineral and the extracellular fluid that bathes the bone [68]. In vitro studies of the buffering properties of turtle bone powders with lactic acid show that the release of calcium and magnesium is pH-dependent; that it is mostly carbonate released, rather than bicarbonate; and that the movement of lactate from the extracellular fluid into the bone is accompanied by a proton, causing alkalinization [72]. However, a recent study of turtle bone that showed a 20% depletion in carbonate, measured in bulk powders by titrating all carbonates to carbon dioxide with a strong acid (nitric acid), showed no measurable change in carbonate using the Raman spot analysis [70]. This suggests that carbonate used in buffering is either bound to the mineral surface or originates from selectively dispersed localities from the bone [70].

Because the buffering mechanism likely involves passive chemical processes, several important physico-chemical limits are also likely to constrain the use of bone as a buffer. The first determinant is the composition of the bone tissue in question. Bone is a composite comprising collagen and mineral, the latter of which comprises hydroxyapatite. Pure hydroxyapatite comprises calcium and phosphate and is relatively insoluble under normal physiological conditions. However, certain elemental substitutions for calcium and phosphate can occur, especially of carbonate for phosphate, and can change the chemical properties of the microcrystalline, changing its vulnerability to acidic dissolution. This is especially true of carbonate substitution for phosphate, which makes bone more vulnerable to acidic dissolution [73]. The effect of composition on ability of bone to buffer acidosis is illustrated by the relative inability of ray-finned fishes to use it as a buffer, most likely because of the low carbonate content of fish bone, approximately 1% of dry weight [74]. Comparatively, species that use bone as a buffer generally have more total bone carbonate content, including lissamphibians (approx. 3% of dry weight bone) [48] and various turtle species (approx. 5–12% of dry weight bone) [66,70]. The causes of this variation in bone composition are uncertain, but is most likely a reflection of the extracellular fluid that bathes it, which itself is dictated by renal function and dietary intake of trace elements.

An additional determinant of the ability of bone to buffer acidosis is likely to be blood flow. Thus, a highly vascularized bone can release more buffer than a less vascularized one. For example, a turtle's shell, which can receive more than half of its cardiac output while incurring a lactic acidosis [75], shows greater lactate uptake than less perfused mineralized tissues such as the skull [69]. As a consequence, it is reasonable to hypothesize that the ability of bone to buffer a given acid load is perfusion-limited; the blood flow (convection) to the bone likely limits its ability to function as a buffer.

6. Challenges in early tetrapod evolution

At the heart of the problem with terrestrial respiratory acidosis in early tetrapods is the differing convection requirements for oxygen of an air-breathing vertebrate compared to a water-breathing one, the latter of which can be 30 times higher [76,77]. A water-breathing vertebrate (e.g. a fish possessing gills) has an extremely high water-convection requirement, due to the relatively low concentration of oxygen in water, even in cold, air-equilibrated freshwater with high oxygen solubility. When these high rates of convection are combined with the relatively high solubility of carbon dioxide in water, the result is extremely low blood PCO₂ in fishes (approx. 2–5 mmHg). In water, oxygen is difficult to obtain, but carbon dioxide is easy to lose; on land, the converse is the case.

An air-breathing vertebrate has a relatively low air-convection requirement to satisfy its oxygen requirements, due entirely to the abundance of oxygen in air [76,77]. Until the advent of costal breathing [45], which allowed for rapid rates of ventilation to satisfy the air-convection requirement to eliminate carbon dioxide, early tetrapods must have relied either on gills or skin for this purpose, both of which were of little use to an early tetrapod producing carbon dioxide on land. Gills could only be used for carbon dioxide elimination while in water and, because of the smaller surface to volume ratio and thick integument of the early tetrapods, Fick's Law dictates that they would have accumulated carbon dioxide in their body fluids that, if unbuffered, would have decreased body fluid pH and disrupted normal physiological functions. An additional acid–base challenge faced by early tetrapods might also have been preventing the severity of non-respiratory acidosis, which occurs when a decrease in body fluid pH is caused by any process other than carbon dioxide accumulation, such as exhaustive exercise.

While the earliest tetrapods were probably still largely aquatic forms, by the Carboniferous, many were more terrestrial, as indicated by their skeletons (degree of limb bone ossification, absence of lateral line canals in the dermal head bones in adults and general skeletal proportions) [5]. While the stem amniotes tended to be the most terrestrial forms, at least one stem tetrapod (Pederpes) was relatively terrestrial; and terrestriality evolved several times within the stem amphibians (e.g. in the dendrerpetontids, the zatrachids, the dissorophids and in some basal stereospondyls: see [6] and table 1). Some stem amniotes, for example embolomeres such as Eogyrinus, became secondarily aquatic.

Lungfishes (our extant example of the ancestral fish condition) mainly lose carbon dioxide via the gills and the skin in the water. As previously noted, the extremely high ventilatory requirement for gill breathing in fishes is more than adequate to eliminate all metabolically produced carbon dioxide and leads to low steady-state arterial PCO₂ levels at rest. Gill ventilation in lungfish is relatively labile [78], but the mechanisms of control are similar to ray-finned fishes [79].

The inference for early tetrapods is that most would have inhaled via buccal pumping, and would have been unable to rely on the lungs to satisfy the ventilatory requirement for carbon dioxide excretion. Lissamphibians can lose some carbon dioxide via the lungs, but rely on the skin (and/or gills in the water). While in air, the skin of modern amphibians is kept moist to facilitate gas exchange, but this also makes the animals vulnerable to desiccation. The immersion of these animals in water further enhances transcutaneous carbon dioxide excretion due to a redistribution of blood flow to the skin [47]. Most early tetrapods would have been too large for this type of cutaneous respiration, but stem tetrapods and many stem amphibians retained internal gills and would likely have been able to lose carbon dioxide in the water.

In order for early tetrapods to become more terrestrial, they had to either evolve alternative avenues for excreting carbon dioxide, or evolve a mechanism for buffering the resultant respiratory acidosis. Costal ventilation was a key innovation of stem amniotes as it allows for a much higher rate of ventilation than buccal pumping, sufficient to eliminate carbon dioxide on land. But carbon dioxide loss would have remained a problem for stem tetrapods and stem amphibians, unless they spent most or all of their time in the water.

Janis et al. [4] proposed that the ‘dermal armour’ of early tetrapods could have acted as a source of ions for buffering acidosis, in a manner analogous to that seen in extant turtles and crocodylians. Extant amniotes experience hypercapnia, and hence respiratory acidosis, when submerged and are unable to ventilate their lungs; but in the early tetrapods, the situation would have been reversed. Many early tetrapods (stem tetrapods and stem amphibians, at least) would have readily lost carbon dioxide in the water, via the gills or the skin, but would have experienced hypercapnia on land, due to their relatively inefficient means of lung ventilation.

It is not yet possible to determine the physiological function of early tetrapod sculptured dermal bone, although work is in progress by Sophie Sanchez and Francois Clarac at the University of Uppsala. However, as we originally noted [4], the distribution of the possession of sculptured dermal bone among tetrapods of different habitat preferences (terrestrial or aquatic) allows for the crafting of predictions to verify or falsify the hypothesis that such bone was physiologically important for conditions of hypercapnia. These predictive hypotheses are as follows:

• 1.That more extensive and/or more heavily sculptured dermal armour would be more prevalent in taxa that encountered problems with carbon dioxide loss; this would be the case for terrestrial taxa.
• 2.That taxa that had evolved new ways to deal with carbon dioxide loss on land would reduce or lose the dermal scales. This could be due to evolving costal ventilation, as in extant amniotes, or evolving small size and specialized skin for cutaneous respiration, as in extant lissamphibians.

Table 1 shows the distribution of taxa arranged according to integument type and habitat (water or land). At first glance, the distribution of taxa does not appear to support the hypothesis; although the terrestrial taxa fit with the predictions, there are also aquatic taxa that have the types of integument proposed for terrestrial life (i.e. either heavy dermal armour or reduction in scalation). But these were all highly specialized, fully aquatic forms, which can be seen to also fit the pattern of adjustment of dermal cover in correlation with issues concerning carbon dioxide elimination.

The majority of early tetrapods had a moderate extent of dermal armour (integument type 2). This includes forms that are mainly semi-aquatic or semi-terrestrial. Among the ‘aquatic’ forms, only the stem tetrapods Acanthostega and Crassygyrinus were likely fully aquatic (i.e. never or rarely leaving the water), and the ‘terrestrial’ forms were not highly specialized for terrestrial life and were in general rather small sized (although not as small as extant lissamphibians).

Forms with extensive dermal armour (integument type 3) include terrestrial taxa that were more highly specialized for terrestrial life, especially the dissorophid stem amphibians and the chroniosuchian stem amniotes. (The extensive dermal armour of chroniosuchians suggests that these basal stem amniotes had not yet evolved efficient costal ventilation, despite the informal observations of Witzmann [6] concerning their rib morphology.) Several aquatic taxa also had heavy dermal armour, but these taxa can be understood as follows. The chroniosuchian Madygenerpeton was a secondarily aquatic form within a primarily terrestrial radiation [31]. The others, the stem tetrapod colosteids and baphetids, and the stem amphibian plagiosaurs, were flattened, bottom-dwelling forms (all fully aquatic) with heavily ossified skeletons (see [5] and [80]). These tetrapods may have inhabited hypercapnic environments, as do bottom-living fishes today that inhabit swampy, heavily vegetated waters ([58,81]; see further discussion in [4]), and/or the heavy dermal ossification may have been for ballast for a benthic lifestyle.

Terrestrial forms with reduced or absent dermal scales (integument type 4) include terrestrial forms that were either very small (amphibamid dissorophid stem amphibians) or large derived stem amniotes that most likely had costal ventilation (e.g. Diadectes and Seymouria). The amphibamids, which were likely ancestral to lissamphibians, were of similar size to lissamphibians, and several forms (e.g. Amphibamus grandiceps) evidenced calcium-filled endolymphatic sacs that are known to aid with countering acidosis in extant frogs [33,82]. Size considerations are also important for stem amniotes, not simply because smaller forms (e.g. the ones with integument type 2) might have been able to use cutaneous respiration, but because of locomotor considerations. The cost of transport would likely have been lower for the larger forms [83] and larger stem amniotes might have been less likely to encounter metabolic acidosis. The microsaurs that lost their dermal scales were also the smaller-sized ones. Larger microsaurs, like the cat-sized Pantylus, retained scales with some dermal sculpturing (integument type 2) [84].

A number of aquatic forms also greatly reduced or lost their dermal scales. These fall into two groupings. Firstly, miniaturized forms, such as lepospondyls and branchiosaurs. Secondly, the majority of the stereospondyls. Unlike the benthic-adapted plagiosaurs, the skeletal anatomy of these stem amphibians shows them to have been pelagic or ambush hunters in shallow waters (see Schoch & Milner [85]). Without the need to deal with carbon dioxide on land, they would have benefitted from scale loss to promote carbon dioxide loss over the skin while submerged; many of them (e.g. capitosaurs) do not show evidence of retaining internal gills, although such gills may have been present, as in their close relatives.

In summary: the pattern of the extent of dermal armour in early tetrapods matches the prediction that this bony covering was used as a means of compensating for respiratory acidosis. Extensive, heavily sculptured dermal armour is seen in those forms most likely to have encountered hypercapnia, whether terrestrial forms with inefficient modes of lung ventilation (i.e. the buccal-pumping stem amphibians) or fully aquatic forms (e.g. plagiosaurs) that were likely bottom-dwelling forms in hypercapnic waters. Dermal armour was lost in those forms which would have attained other means of eliminating carbon dioxide on land: either by means of cutaneous gas exchange (in the very small stem amphibians closely related to lissamphibians) or by evolving costal ventilation (stem amniotes). Dermal armour was also lost in fully aquatic stem amphibians (many stereospondyls) that were likely active, pelagic forms.

7. Dermal bone as a buffer for metabolic acidosis in archosaurs

(a). Dermal bone was a feature of archosaurs

Early tetrapods are far from the only tetrapods that possessed extensive, highly ornamented dermal bone: many non-avian archosaurs (crocodylians, dinosaurs, birds and relatives) are or were characterized by ornamented (= sculptured) dermal bone, especially the pseudosuchians (all archosaurs more closely related to modern crocodylians than to birds [86]). Three lineages of the more inclusive clade Archosauriformes (Doswelliidae, Proterochampsia and Phytosauria) also possessed highly ornamented dermal skull bones and osteoderms over the body (figure 3), suggesting that this feature is ancestral for archosaurs in general [87–89]. Aetosaurs in particular possessed extensive osteodermal armour, which allows their osteoderms to be used as species identifiers and as index fossils [90], though osteoderms were present to a lesser degree in many lineages. Modern crocodylians possess a similar morphology, with their osteoderms and dermal skull bones showing a high development of ornamentation [91,92]. Certain dinosaurs (e.g. ankylosaurs) possessed osteoderms, but these were not ornamented and likely had a defensive function rather than a physiological one [93].

Figure 3.

Phylogeny of archosaurs, showing major groups and the distribution of sculptured ornamentation on dermal skull bones and osteoderms. Ornamented osteoderms and dermal skull bones are shared by Doswelliidae, Proterochampsia and Phytosauria, suggesting that they arose early in the history of archosaurs and are a general ancestral trait. Ornamented dermal skull bones were lost three times—once in Avemetatarsalia (the clade that includes dinosaurs (including birds) and pterosaurs), once in Poposauroidea and once in Rauisuchidae. Avemetatarsalia and Poposauroidea also lack ornamented dermal osteoderms, which are retained (albeit reduced) in Rauisuchidae. Avemetatarsalians appear to have been endothermic, and poposauroids and rauisuchids show skeletal adaptations typical of endothermic animals, supporting a link between the loss of ornamented dermal bone and the evolution of endothermic metabolism.

The ornamented dermal bone of many archosaurs is highly reminiscent of that of early tetrapods, exhibiting the same deeply sculptured external anatomy (figure 4). While similarity in morphology does not, in itself, imply similarity in function, it does lend credence to the hypothesis that archosaurs with highly ornamented dermal bone used that bone to buffer against acidosis. As in early tetrapods, the ornamentation would have been associated with increased vascular supply to the tissue, potentially increasing the surface area for buffering reactions to take place. This is especially true for taxa that evolved extensive osteodermal armour; this armour is usually assumed to have had a defensive function, but may have also been an adaptation allowing for a greater capacity to buffer the blood against acidosis. Within crocodylomorphs (modern crocodiles and their close extinct relatives), semi-aquatic forms have a generally high development of dermal ornamentation. This has been suggested as an adaptation to improve basking efficiency [91], but may also reflect an adaptation to buffer against hypercapnia or lactic acidosis during long periods of apnoea associated with diving (in effect a reverse of the hypothesized condition for early tetrapods).

Figure 4.

Dorsolateral view of osteodermal armour of Typothorax coccinarum, an aetosaur (YPM VP (Yale Peabody Museum—Vertebrate Paleontology) 058121#). The osteoderms show the same pattern of sculptured ornamentation found in the dermal skull bones of early tetrapods such as Eryops (figure 2).

Assuming that some extinct archosaurs did use their dermal bone for buffering against acidosis, the question remains: why would so many members of this lineage require this physiological strategy, which was originally hypothesized to have assisted the water to land transition [4]? The answer may be that dermal bone in archosaurs was originally used to buffer metabolic acidosis, specifically lactic acidosis associated with anaerobic locomotion. Ectothermic animals have approximately one-tenth the aerobic metabolic scope of endothermic animals of the same size, requiring them to resort to anaerobic metabolism after a much shorter period of sustained activity [94]. Anaerobic respiration produces lactic acid as a by-product, while regenerating NAD⁺ to continue cellular glycolysis, resulting in less than one-tenth the metabolic energy of aerobic respiration while releasing an acidic by-product [95]. Therefore, ectothermic animals are incapable of the kinds of sustained activity common in endotherms; for instance, there are no vertebrate ectotherms specialized for cursoriality, powered flight or bipedality, and none capable of the long-distance migration observed in endothermic species. It is possible that ectothermic archosaurs used their dermal bone to buffer against lactic (metabolic) acidosis, sequestering lactate to defend body fluid pH and allow for longer periods of sustained, anaerobically driven activity. In essence, such a strategy would be a work-around, enabling endotherm-like locomotor endurance with an ectothermic physiology, assuming that fatigue is determined by muscle pH and not glycogen content or some other physiological factor.

This may explain the early hegemony of archosaur lineages assumed to be ectothermic, or at least not fully endothermic; the metabolic strategies of many archosaurs, especially pseudosuchians, are unresolved, and evidence from bone histology suggests that a trend towards high metabolic rates is basal for the clade [96]. However, it remains unclear what physiological strategy was employed by these animals [97]; many of the studied taxa lack the characteristic upright posture of active endotherms, none had evidence of resting metabolic rates in the ranges of extinct theropods or birds, and their aerobic metabolic scope (which is of principal concern to the question of sustained activity) remains unknown. Therefore, the present evidence indicates that many basal archosaurs were more metabolically active than modern ectotherms, but does not suggest that their aerobic metabolic rates or metabolic scopes approached those of the early mammals, dinosaurs and pterosaurs with which they shared their habitat.

Extant terrestrial ectotherms do not occupy large herbivore or predator niches unless endotherms with similar ecology are absent; active endotherms appear to outcompete them or competitively exclude them from this niche space. For instance, Komodo island is perhaps the only present ecosystem in which a large ectotherm, Varanus komodoensis, is the apex predator. By contrast, large ectothermic archosaurs appear to have been the dominant terrestrial herbivores and especially carnivores for the majority of the Triassic, with animals more likely to have been endothermic, such as dinosaurs, pterosaurs (both archosaurs) and mammals, a far less significant component of the fauna. In the 1970s, authors such as Bakker and Charig assumed that the presumed endothermic dinosaurs rapidly displaced other archosaurs to become the dominant terrestrial vertebrate clade [98]; but recent studies have found that the pseudosuchian lineage was as or more successful than dinosaurs for the entirety of the Triassic Period [98–101]. The rise of dinosaurs appears to have been a consequence of the end-Triassic Mass Extinction, which disproportionately affected pseudosuchians, leaving Crocodylomorpha as the sole surviving clade. If basal archosaurs and pseudosuchians used their extensive osteodermal armour and sculptured dermal bone to bypass the constraints of ectothermy, it may explain how they sustained high activity and avoided being outcompeted by the emerging endothermic clades—dinosaurs, pterosaurs and mammals—throughout the Triassic Period. This hypothesis requires extensive testing, but is tentatively supported by the fact that ornamented dermal bone is not only absent in dinosaurs and dinosaur-related Triassic archosaurs, but is less prevalent in pseudosuchian archosaurs with skeletal adaptations associated with endothermy. ‘Rauisuchians’ (including Rauisuchidae and Poposauroidea), which had upright postures and may have been facultative bipeds, have only a midline dorsal osteoderm row and lack ornamentation on the dermal skull bones. Likewise, the bipedal shuvosaurs completely lack osteoderms. The metabolic strategy of these animals is unresolved, but their skeletal anatomy and histology imply high metabolic rates [97] and possibly transitional or full endothermy, perhaps freeing them to lose their ornamented dermal bone, which would have incurred a substantial metabolic cost to develop and maintain, as well as being extra weight to carry during a lifetime of locomotion.

Such a hypothesis is difficult to test explicitly for extinct species, as the buffering activity of bone likely leaves no histological signal [4]. However, some lines of evidence supporting this hypothesis can be drawn. All tetrapods are capable of buffering using bone to some capacity. The ability is particularly well developed in turtles [66,102], which are the sister group to the archosaurs [103–105] and is known in crocodylians themselves, which have been shown to use mineral from and sequester lactate in their osteoderms as a buffer [64]. This implies that the ability to buffer the blood is probably ancestral for archosaurs in general. Large quantities of highly vascular dermal bone would increase the animal's capacity for buffering the blood by providing access to the soluble portions of the bone mineral. While inquiries into the bone buffering capacity of turtles and crocodylians have focused on defending blood pH during anoxia-induced acidosis, leopard frogs were shown to buffer lactate in their bones during exhaustive exercise (i.e. in response to metabolic acidosis [48]) and Ruben & Bennett [65] showed that plasma calcium levels increased during exercise in rattlesnake and iguanas at the onset of exhaustive, anaerobic exercise. If dermal bone is employed for buffering, a potential method of testing whether extinct taxa used a similar strategy would be to quantify the extent of the vascularization of the dermal bones. For most archosaurs, the sculpturing is entirely external, meaning that the relief of the bone surface could be used as a proxy for its vascularization, as has been accomplished in several recent studies [91,92]. This hypothesis is consistent with the hypothesis that ornamented osteoderms are and were used for heat exchange by archosaurs [106], as both strategies require highly vascularized bone and would not compete physiologically.

(b). Did the interventricular septum evolve to support perfusion of the osteoderms?

All extant mammals and archosaurs possess an interventricular septum that allows for complete flow and separation of the pulmonary and systemic blood flows. It is unequivocal that pressure separation was essential for the evolution of systemic endothermy, as the high systemic cardiac outputs to meet the high oxygen requirements of endothermic tissues necessitate larger systemic and, therefore, intraventricular pressures. A persistent conundrum of modern archosaurs, however, is that modern crocodylians are not endothermic, yet possess an interventricular septum that allows for higher systemic blood pressures. Seymour et al. [107] proposed that modern crocodylians evolved from endothermic ancestors, and that endothermy was secondarily lost due to the cost associated with maintaining endothermy in a fully aquatic environment. Recent work on basal archosaurs supports the hypothesis that high metabolic rates are ancestral for the clade [96], although these animals lack upright postures and seem to have been considerably less active than modern endotherms. Nevertheless, systemic endothermy is not possible without a four-chambered heart, so such a heart would have had to evolve before the condition of endothermy. What type of selection pressure might have allowed for a four-chambered heart to develop and persist among ectothermic and primarily anaerobic archosaurs?

The primary benefit of a four-chambered heart is the ability to achieve high rates of circulatory convection, which have traditionally been associated with high rates of aerobic performance. Because the capacity to use bone as a buffer is likely perfusion-limited, and most members of the archosaur lineage possessed extensive dermal bone, a high systemic cardiac output would have enabled these anaerobic animals to achieve and maintain high rates of lactic acid production by defending their body fluid pH through an increase in systemic blood flow to mineralized tissues during anaerobic exercise. This hypothesis is testable in modern crocodylians by measuring systemic blood flows while simultaneously measuring blood flow distribution to the osteoderms during exhaustive exercise. An implication of this hypothesis would be that the four-chambered heart of modern archosaurs is an exaptation of a structure initially developed to sustain high blood flow to osteoderms for metabolic buffering, rather than an adaptation acquired during the initial development of endothermy. Therefore, it could also be tested by determining whether the resting metabolic rates inferred for basal archosauromorphs such as Azendohsaurus laarousii [96] are present in any ectotherms with an incomplete ventricular septum (to establish whether the complete septum was likely present in Azendohsaurus), and tracking the acquisition of extensive osteoderm development compared to the acquisition of high growth rates. If our hypothesis of the origin of the interventricular septum in archosaurs is correct, extensive osteoderm coverings should appear before the inferred presence of an interventricular septum.

8. Conclusion

Based on phylogenetic inference and evidence from the fossil record, the ability of early tetrapods to eliminate carbon dioxide was limited by lung ventilation via a positive-pressure buccal pump, large body size and a likely thick integument, increasing the likelihood these animals would have incurred a respiratory acidosis while on land. It is also likely that the ability of these animals to compensate for lactic acidosis by increasing ventilation rate was limited. Many of these early tetrapods had a body cover of highly ornamented dermal bone; based on studies of mineralized tissues in extant tetrapods, such bony tissue could have functioned to help buffer any respiratory or even metabolic acidosis during terrestrial excursions when gills could not be used for gas exchange. Our current understanding indicates that this buffering involves passive chemical interactions between the bone mineral and the extracellular solution, and is most likely limited by the carbonation of the hydroxyapatite as well as its perfusion. We have extended our hypothesis to include a role for mineralized tissue in the buffering of lactic acidosis in heavily ornamented archosaurs (specifically pseudosuchians, including crocodylomorphs), which possessed highly vascularized dermal ossifications that could have been used to buffer exercised-induced lactic acidosis. Additional studies of the bone buffering mechanisms in extant crocodylians are needed to test this hypothesis. However, if such experiments reveal that these animals do increase blood flow to the osteoderms during exhaustive exercise, as occurs in turtles following anoxia, then perfusing this mineralized tissue might have played an important role in shaping cardiovascular design within the archosaurian lineage as a whole, and provide a novel hypothesis for the evolution of their four-chambered heart.

Data accessibility

This article has no additional data.

Authors' contributions

C.M.J. provided expertise on early tetrapod evolution and anatomy, J.G.N. provided expertise on archosaursian evolution and anatomy and D.E.W. provided expertise on animal physiological function and coordinated the project. All authors gave final approval for publication and agree to be held accountable for the work performed therein.

Competing interests

We declare we have no competing interests.

Funding

This work was partially funded by a National Science Foundation CAREER award (no. 1253939) to D.E.W. J.G.N. was supported by a Richard Gilder Graduate School student fellowship and the Newt and Calista Gingrich endowment.