Carrier (1987) connected styles of terrestrial locomotion with air breathing, metabolic level, and evolution in terrestrial vertebrates. In my view this is one of the most significant paleontological papers of the 1980's, but it and its implications do not seem to have found their way into paleontological textbooks yet, apart from my own (Cowen 1990, 1994)! :-)
However, early tetrapods moving about on land faced a much more serious problem. They inherited the fish-like locomotory flexing of the body as they evolved feet and limbs from fins, but their limbs were sprawling. The sprawling gait of living amphibians and living reptiles requires that the trunk is flexed first to the left and then to the right as the animal walks and runs. As the animal steps forward with its left front foot, the right side of the thorax and the right lung inside it are compressed, while the left side and left lung are expanded. Then the cycle reverses with the next step. This distortion of the thorax interferes with and essentially prevents normal breathing, in which both lungs expand uniformly and simultaneously, and then contract. If the animal is walking, it may be able to breathe between steps, while the thorax is momentarily not distorted, but if it tries to run, the cycling is so fast that breathing becomes impossible. Living amphibians and living reptiles cannot run and breathe at the same time (Carrier 1987, 1991). I have called this "Carrier's Constraint" (Cowen 1990).
Animals can run for a while without breathing: for example, Olympic sprinters normally don't breathe during a 100-meter race. But running soon builds up an oxygen debt and a dangerously high level of lactic acid in the blood. Living amphibians and reptiles, then, can hop or run fast for a short time, first using up the oxygen stored in their lungs and blood, then switching to anaerobic glycolysis. They cannot run for long, however. It is an everyday observation that lizards run in short rushes, with frequent stops. By attaching recorders to the body, Carrier (1987) showed experimentally that the stops are for breathing, and that lizards do not breathe as they run. Later, Carrier (1991) refined his techniques to show that lizards could breathe without coming to a complete stop, but only by interpolating brief pauses in the locomotor movements.
Ecologically, all living amphibian and reptilian carnivores use ambush tactics to capture agile prey: chameleons and toads flip their tongues at passing insects, for example, and crocodiles lurk close to shore under the water. The largest living varanid lizard, the ora or Komodo dragon, which attacks deer, pigs, and humans, has a sprinting range of ten meters at most, but it prefers to ambush prey from one meter away.
Amphibians and most living reptiles have a three-chambered heart, which has often been regarded as inferior to the four-chambered heart of living mammals and birds. But the three-chambered heart is useful to a lizard. Lizards run to catch food or to get away from danger, so they must use their resources most efficiently at this time. In a run, it is useless and perhaps dangerous for the lizard to waste energy pumping blood to lungs that cannot work. The lizard thus uses all the heart and blood capacity it has to circulate its store of oxygen around the whole body. The price it pays is a longer recharging time when it has to resupply oxygen to the blood, but it is usually able to do this at a less critical moment.
Early tetrapods all had sprawling gaits and faced a great problem in terrestrial locomotion. Their respiration and locomotion used much the same sets of muscles, and both systems could not operate efficiently at the same time. Imagine the laborious journey of Ichthyostega from the water to its breeding pools, with a few steps and a few gasps repeated for the whole journey (Cowen 1990, 1994).
Of course, land vertebrates have evolved effective solutions to Carrier's Constraint. Mammals have freed the mechanics of respiration from the mechanics of locomotion by evolving erect stance, with all four limbs supporting the body vertically, and moving in a vertical plane during locomotion. This movement does not twist the thorax much, and the mammal can breathe even during rapid locomotion. In fact, locomotion in quadrupedal mammals encourages breathing on the run. The backbone flexes and straightens up and down with each stride, alternately expanding and compressing the rib cage evenly. So horses, dogs, and jackrabbits running at full speed take one breath per stride.
Bipedal posture also relieves Carrier's Constraint. In a biped, the thorax is lifted off the ground, the forelimbs are no longer so involved in support and locomotion, and the rib-cage does not flex laterally during motion. A biped can breathe while it runs. Human runners usually take a breath every other stride, and wallabies take one breath per hop. (David Jablonski pointed out to me that frogs and toads have also evolved an analogous evasion of Carrier's Constraint.)
Some archosaurs evolved bipedal posture as a solution to Carrier's Constraint. However, a biped is intrinsically unstable, and one can imagine that permanent bipedal locomotion in archosaurs evolved gradually from a stage in which bipedality was used on the run, avoiding Carrier's Constraint during emergency locomotion, while normal walking could be done in a sprawling or semi-erect quadrupedal gait. A few lizards operate this way. A mixed style of locomotion is a reasonable intermediate stage in which selection for erect bipedal posture could work effectively. Early archosaurs include animals that can be seen as stages in the process of evolving "semi-erect" limbs. Euparkeria is perhaps the earliest possible example, while routine biped erect gait evolved somewhere among early archosaurs. Very early dinosaurs like Eoraptor had certainly become fully erect, bipedal runners, shedding Carrier's Constraint completely.
There is a corollary to Carrier's Constraint. The ability to sustain fast locomotion automatically implies the ability to sustain long periods of high energy output. Respiratory and circulatory systems must both work at high levels. Once a high metabolic rate is possible, even intermittently, it may become desirable for metabolic rate to remain high all the time, resting as well as active-in other words, for the animal to be warm-blooded. Erect stance, then-a solution to Carrier's Constraint-allows endothermy in vertebrates. It does not require it, and this point is often missed. In this paper I do not wish to argue for or against endothermic inferences in the vertebrates that have solved Carrier's Constraint.
Ruben has documented several instances of surprisingly high levels of metabolic performance in living reptiles (see his 1995 review), but these are all emergency or short-term, largely anaerobic activities, and Ruben explicitly emphasizes this point. There is no question that among living vertebrates, all birds and mammals perform at much higher sustained locomotory levels than all reptiles and amphibians. The superior performers are endothermic, and they have solved Carrier's Constraint, and it is arguable which of these derived characters lies at the root of superior sustained locomotory performance.
The only point I wish to argue is that among living terrestrial vertebrates, sustained high locomotory performance is associated with, and probably requires, a solution to Carrier's Constraint. Ruben has expressed doubts on this point (1995, pp. 83-84), but in reference to unpublished work, so that I cannot discuss it further at this time. It would be an astounding coincidence if a solution to Carrier's Constraint happened to coincide by chance with high locomotory performance in all living terrestrial vertebrates.
If Carrier's Constraint is real, it should be true for all air-breathing vertebrates that have paired lungs. In this study, which is explicitly an "idea paper," I shall argue that it must apply also to aquatic air-breathing tetrapods, and I shall explore some of the implications of this argument. Fortunately, the mechanics, anatomy, and physiology of swimming are now well studied (see Maddock et al. 1994 for recent advances), and there are good analyses of swimming among extinct vertebrates (see Webb and Buffrénil 1990, Riess and Frey 1991, and Massare 1994 for recent overviews).
Tetrapods primitively have no solution to Carrier's Constraint, and many groups of living and extinct aquatic air-breathers have no apparent adaptations for avoiding Carrier's Constraint: these include quadrupedal amphibians, advanced crocodilians, mesosaurs, and the diapsids in Carroll's (1985) "functional category 3": askeptosaurids, pleurosaurs, aigialosaurs, and dolichosaurs.
Some aquatic air-breathers have evolved special adaptations to optimize performance underwater in lengthy dives, using stored oxygen. Crocodiles have evolved an extraordinary tolerance for blood acidosis, and can maintain vigorous anaerobic activity for a considerable time (Bennett et al. 1985). Penguins are extraordinary underwater swimmers, in terms of both physiology and efficiency (Culik et al. 1995). However, Carrier's Constraint does not apply to diving performance, because divers do not try to breathe and swim fast at the same time: the surface visit after a dive is usually a time for resting and recuperation. Spectacular adaptations for prolonged diving have been evolved several times in living mammals and birds, and at least once along some extinct reptiles, the mosasaurs. Some mosasaurs have skeletal damage diagnostic of exposure to the "bends", so one can infer unambiguously that they were diving deeply-too deep for their own good, in the case of the individuals in question (Rothschild and Martin 1988; Martin and Rothschild 1989). Despite all these examples of prolonged activity under water, we are not discussing a capability for sustained locomotory performance: these animals are adapting ways to maximize performance on stored oxygen.
A few aquatic air-breathers are fishes, and a few are amphibians, and they may have been continuously associated with aquatic habitats during their evolution. But all amniotes are primitively adapted to life in air, with a reproductive stage-the amniotic egg-that must be laid on land. Therefore, those amniotes that are aquatic air-breathers have successfully re-invaded aquatic environments in the face of competition from gill-breathers. We have not yet established the performance limits of the most efficient and powerful gill-breathers: every time new apparatus is introduced to measure performance levels more effectively, higher levels of performance are discovered (see, for example, new information on sharks and tuna in Graham et al. 1994). How can intermittent air-breathers compete successfully with continuous gill-breathers? We should be able to identify the adaptive systems that have allowed them to achieve such a way of life.
Even though diving air-breathers can develop anatomical and physiological adaptations to maximize underwater performance based on one breath, it remains true that sustained high-performance swimming by air-breathers becomes possible only when Carrier's Constraint is avoided. As on land, there are alternative pathways by which this has been achieved by swimming air-breathers. The alternatives correspond approximately with the swimming styles known as axial (trunk and tail dominated) and paraxial (limb dominated) that have been discussed extensively by Riess and Frey (e.g., 1991); with swimming Categories 1-5 defined by Carroll (1985); and with Baupläne I-IV of Massare 1994. As Carroll remarks, functional categories are relatively distinct for good functional reasons: however, every evolutionary paleontologist should expect transitions and intermediates between functional categories, and I shall point to specific examples in the discussion that follows.
I suggest that the ribs of Ichthyostega were adaptations for stiffening the thorax and avoiding Carrier's Constraint, but that the adaptation arose for better swimming rather than running on land. Ichthyostega clearly swam predominantly with its tail, which carried a large fin, and stiffening of the thorax would have allowed ready air-breathing at any time the snout cleared the surface. Ichthyostega may have been capable of more sustained swimming than some of its contemporary amphibians, and in particular it may have been able to out-swim some contemporary rhipidistians. It may not be a coincidence that thoracic stiffening would also have helped the first laborious land locomotion, not to make it faster but to make it more energy-efficient.
Lingham-Soliar (1992) recently inferred that one of the last mosasaurs, Plioplatecarpus, from the late Maastrichtian, was able to "fly" underwater with its forelimbs. He claimed that the forelimbs are modified into hydrofoils, and that Plioplatecarpus is the only mosasaur with a short rigid thorax. The pectoral girdle is massive, especially the scapula, and the humerus and ulna are very strong. It has vertebrae that restrict flexing of the spine in several different ways.
This interpretation has been challenged by Nicholls and Godfrey (1994), but in an interesting way. They agree that the spine of Plioplatecarpus is stiffened, that the pectoral girdle is very strong, and that the thorax has much reduced flexibility. The trunk is probably shorter than it is in most mosasaurs. However, they point out, there is still a long, flexible, laterally compressed tail, and the front limbs have not yet been shown to be truly wing-like.
Put another way, Nicholls and Godfrey are saying that the front end of Plioplatecarpus was stiffened, but not for underwater flying. Why was it stiffened, then? What was it doing that other mosasaurs were not? I suggest that Plioplatecarpus was beginning to solve Carrier's Constraint, by decoupling flexure of the thorax from the swimming propulsion, which increasingly involved only the posterior of the animal. Such an adaptation would only be important for an animal swimming at sustained speed in surface waters. Therefore, I suggest, this animal was the best sustained surface swimmer among mosasaurs, even if it did not fly underwater.
Such an interpretation of the whole plesiosaur syndrome requires that there must have been concomitant evolution of effective respiration that would provide the oxygen for sustained high metabolic rates. Yet plesiosaurs evolved from a terrestrial ancestor that was undoubtedly a diapsid, possibly a lepidosauromorph (Rieppel 1993, Storrs 1993), that had sprawling gait and suffered from Carrier's Constraint. Plesiosaurs must have solved the problem of Carrier's Constraint as they evolved advanced skeletal anatomy, but how?
In the early stages of sauropterygian evolution, undulatory fish-like swimming must have dominated, in animals that had not solved Carrier's Constraint. This is obvious for most well-known early sauropterygians: the small pachypleurosaurs were still clearly swimming in fish-like mode, with a thorax that had little or no stiffening.
Storrs (1993) argues that an evolutionary sequence through a series of "nothosaurs" (Figure 1B) included a gradual increase in locomotory function for the forelimbs in particular. At the same time, the thorax and spine became stiffer, and undulation of the thorax must have decreased markedly. With further evolution, the pelvic and pectoral girdles became stronger, the thorax in particular was stiffened and strengthened until it became a sturdy box, and at the same time the limbs were gradually modified into rowing appendages. Riess and Frey (1991) cite the Triassic "nothosaur" Ceresiosaurus as an effective rower, with limbs modified into hydrofoils and a ball-and-socket glenoid joint at the shoulder, and Storrs (1993) argues that the "nothosaur" Corosaurus had dominant paraxial propulsion from the forelimbs too.
As the earliest plesiosaurs evolved, massive shoulder and pelvic girdles developed, with a stiffened spine and an impressive array of bony supports along the belly. The hind limbs added a paraxial power stroke, perhaps abandoning any sort of terrestrial locomotion as they did so (Storrs 1993), and the swimming style evolved into underwater flight (Riess and Frey 1991, Storrs 1993).
Storrs (1993, 72-73) was concerned about the evolution of paraxial locomotion in terms of ancestral constraint, and suggested that it was required by a shift to a rigid rib-cage that was in turn required because the body had to carry buoyancy compensation in the form of gastroliths (Storrs 1993, 81-83). This is a rather clumsy scenario, with little supporting evidence. However, the functional shift to paraxial locomotion receives a far simpler explanation in terms of an adaptive innovation that gradually solved Carrier's Constraint. One can now view the evolution of plesiosaurs as the acquisition of an entire adaptive complex that includes vital physiological components as well as anatomical ones. It is probably not an accident that paraxial locomotion was acquired first by the forelimbs, which are attached to the pectoral and thoracic structures intimately connected with respiration.
An alternation of downward power strokes and upward recovery strokes (Riess and Frey 1991) would have led to alternating changes in the body position of plesiosaurs relative to the surface, no matter whether the fore- and hindlimb strokes were synchronous or alternating (for a summary of that controversy, see Massare 1994). Synchronous strokes would have driven the body in a sinusoidal curve through the water, penguin-style. Alternating strokes would have pitched the head up and down relative to the tail. In either case, breathing could have been synchronized with locomotion.
Carroll (1985, 151) argues that the most advanced ichthyosaurs had "nearly rigid" trunks, though he cites no evidence for that inference, except to link it to the general shape of the body. Carroll seems to argue that the spindle-shaped body, and the lunate tail of advanced ichthyosaurs must automatically be linked with a rigid trunk by analogy with tuna. McGowan (1992) argues for a stiff tail in ichthyosaurs, and extrapolates that to the body because the skeleton is bony (p. 567). He does envisage considerable flexibility of the body, however, in his Figure 11C (p. 568).
However, if one takes the fossil record at face value, there are few obvious stiffening structures in the ichthyosaur skeleton, especially in the thorax, where the rib-cahe is very lightly built. I infer that there could have been significant lateral undulation in swimming. On the basis of general ichthyosaur anatomy, Riess (1986) has written and drawn interpretations of ichthyosaur locomotion that contradict traditional wisdom. He divides ichthyosaur locomotion into three styles, two of which are comparatively slow, though with good agility, and a third that is still not fast. Some of Riess's arguments about locomotory style are apparently based on an inappropriate comparison with the living Amazon dolphin (see summary of this controversy in Massare 1994), but it remains true that the skeletal anatomy of ichthyosaurs at least permits, and may require, considerable flexibility of the body.
Using the physiological approach used here, this is a critical point. If the body was flexible, ichthyosaurs had not solved Carrier's Constraint. Yet they had excellent streamlining and very low drag. Can these conclusions be reconciled? There are three possibilities.
In this scenario, even advanced ichthyosaurs were ambush predators, with the "high-speed" adaptations of the body shape and the tail giving fast acceleration and maneuverability but not prolonged stamina: they would have been the cheetahs of the Mesozoic oceans. One can envisage the kind of pursuit after prey that is seen in many diving mammals and birds today that must swim down their prey within the confines of one breath.
Aquatic air-breathers face a constraint that I have not yet discussed. Waves and other edge effects mean that drag is greatest at and close under the water surface. Swimming at or near the surface requires several times more power than "deep" swimming (at a depth of 3 body diameters or more) (Blake 1983a), so an air-breather must traverse a zone of increased drag as it comes to the surface. Sea otters swimming purposefully dive relatively deeply between breaths, spending most of their time underwater below the high-drag surface zone (Williams 1989).
In contrast, penguins and dolphins "porpoise", leaping out of the water surface at high speed. Blake (1983a) argued that this behavior reduces overall drag and increases locomotory efficiency, with advantage increasing at higher speeds. The captive penguins studied by Hui (1987) conserved energy by leaping, but not as much as they might have done. Instead, the leaping seemed to be associated more with respiration rate, with the leaps providing a mean of 46 breaths per minute, while optimal energy-saving leaps would have provided much slower breathing. Overall, the experiments suggested that penguins saved as much energy as they could, consistent with the rapid breathing frequency necessary for sustained high-speed swimming.
Ichthyosaurs are not penguins or dolphins, but it is possible that leaping could provide yet another way to avoid Carrier's Constraint. An ichthyosaur could swim at high speed in an undulating path, with the pectoral fins providing upward or downward forces as needed. With appropriate control, the ichthyosaur could "porpoise", lifting clear of the water, pointing straight ahead, and in that position could breathe at high speed, exactly as Hui's penguins did. Propulsion by the laterally flexing body would be resumed during the underwater phase, with the body traversing the high-drag surface zone at a reasonably high angle (Hui's penguins left and re-entered the water at angles close to 30°ree;). In this situation, the performance of an ichthyosaur could certainly approach that of a dolphin of the same body size. Leaping, even in a mode optimized for breathing rather than locomotion, conserves energy compared with subsurface cruising. Living cetaceans use porpoising for more than emergency escape maneuvers-cetaceans as large as orcas have been observed porpoising in a "cruising" mode on their feeding migrations.
In my interpretation, then, leaping is the only way in which ichthyosaurs could have escaped Carrier's Constraint. Either they "porpoised", or they did not use sustained high-speed locomotion. There are rather narrow zones of weight and speed within which leaping is favored energetically over pure swimming (Blake 1983a). At least some ichthyosaurs would have fitted into those zones. David Jablonski suggested to me that size trends in ichthyosaur evolution might repay study in this context: for example, the huge Triassic ichthyosaurs are a long way outside the "leaping" envelope (at least as adults) but many later ones lie within it.
Overall, then, applying Carrier's Constraint to air-breathing aquatic vertebrates gives us new insight into their paleobiology, and raises several important new lines for investigation. As usual, we need more research, and more thought. What more could one ask for?