Were Dinosaurs Warm-blooded?

Willem J Hillenius. Encyclopedia of Life Sciences: Supplementary Set. Volume 22. Chichester, UK: Wiley, 2007. 

Dinosaurs were most likely ectothermic, with resting and maximal metabolic rates that were lower than those of modern mammals or birds. However, given the favourable Mesozoic climatic conditions, most dinosaurs were probably able to maintain high, constant body temperatures through behavioural or inertial thermoregulation.


There is little doubt that the dinosaurs were one the most successful groups of vertebrates. Not only were they the tallest, longest and heaviest of all land animals, but they also represent one of the broadest adaptive radiations among vertebrates. Dinosaurs encompassed a spectacular array of herbivorous and carnivorous forms, many with remarkable adaptations for feeding, locomotion and defence. With a virtually worldwide distribution, they dominated nearly every terrestrial habitat from the Late Triassic to the Late Cretaceous—a reign that, at almost 160 million years, lasted longer than that of any other group of tetrapods. The fossilized remains of dinosaurs suggest that they were active, dynamic animals. They had an upright, in many cases even bipedal, posture; and their locomotor mechanics were a significant improvement over the awkward, sprawled gait of the earlier amniotes. Many, especially the carnivorous forms, were highly agile, fast-moving creatures very well suited to terrestrial life.

The unparalleled success of the dinosaurs and their dramatic morphological innovations have long fuelled speculations that these animals had also made significant physiological advances over their early amniote ancestors, especially with respect to metabolic capacity. Moreover, the close evolutionary relationship between dinosaurs and birds suggests the possibility that birds may have inherited their endothermic status from their dinosaurian ancestors. The popular view holds that dinosaurs, like modern birds and mammals, may have been endothermic (or ‘warmblooded’), rather than ectothermic (or ‘cold-blooded’), a model that reinforces interpretation of these animals as having led dynamic, competitively successful lives. However, these reconstructions have been based, in large part, on unwarranted assumptions that ectothermic animals are mostly sluggish and inactive, and that only endothermy could account for the dynamic attributes and success of the dinosaurs. In recent decades, ecological and physiological studies have resulted in an increased appreciation of the complex biology of extant ectotherms. This, combined with a more rigorous approach for assessing the palaeophysiology of fossilized organisms, has provided new insights into the biology of dinosaurs.

Dinosaur Thermoregulation

In the debates over the metabolic status of dinosaurs, two separate issues—thermoregulation and aerobic capacity (i.e. the capacity for sustained oxidative metabolism)—have frequently been confused. ‘Warm-bloodedness’ and homeothermy (maintenance of a high, constant body temperature) are often uncritically equated with endothermy and elevated metabolic rates, and ectothermy with poikilothermy (variable body temperature) and ‘cold-bloodedness’. However, the low resting metabolic rates of extant ectotherms do not preclude these animals from achieving high body temperatures or even homeothermy. Under favourable conditions, living reptiles readily maintain body temperatures significantly above ambient. Many lizards, for example, thermoregulate behaviourally when they are active during the day, and routinely have body temperatures that overlap or even exceed those of endotherms. Larger ectotherms, by virtue of their small surface area to volume ratio and correspondingly low heat loss rates (high thermal inertia), can remain virtually homeothermic in their natural environments.

Extrapolating from such observations, a number of increasingly sophisticated mathematical models (e.g. O’Connor and Dodson, 1999; Seebacher, 2003) have examined the biophysical constraints of thermoregulation in large animals such as dinosaurs. First, these models show that thermal inertia increases with increasing body mass. Regardless of metabolic capacity, large-bodied animals tend to become effectively isolated from ambient temperature fluctuations, and reach stable body temperatures. Consequently, it is not necessary to invoke high, endothermic metabolic rates for most dinosaurs to achieve homeothermy. Although small dinosaurs (<75 kg) would still have been affected by daily temperature fluctuations, medium to large dinosaurs (>500 kg) were probably unaffected by such changes, and would respond only to fluctuations on a weekly to monthly scale, even if these animals were reconstructed with low (ectothermic) metabolic rates. In fact, if they did have endothermic metabolism and high rates of internal heat production, large dinosaurs (>1000 kg) would likely have been at risk of overheating, even at moderate ambient temperatures.

Secondly, taking into consideration the comparatively warm and equable climates of the Mesozoic, these models further conclude that between behavioural thermoregulation and thermal inertia, ectothermic dinosaurs would have had little problem reaching and maintaining high (>30°C) and stable ( ± 2°C or less) body temperatures in most latitudes. However, small forms at high latitudes, especially during the more seasonal Late Cretaceous, might have had trouble reaching such high diurnal operating temperatures in the winter, and then might have had to resort to dormancy, acclimation or migration. These conclusions do not mean that dinosaurs could not have been endothermic, but emphasize that even if they were ectothermic, dinosaurs were by no means limited to having low body temperatures or sluggish lifestyles.

Dinosaur Metabolic Rates

Although thermal profiles of endotherms and ectotherms are effectively similar under the right circumstances, the biggest physiological difference between ectotherms and endotherms lies in their aerobic capacities. Endotherms have much higher levels of oxygen consumption: even at rest, metabolic rates of modern mammals and birds are 5-15 times greater than those of reptiles of the same mass and temperature; they are as much as 20 times greater if routine activity is included. Such high metabolic rates provide endotherms with levels of stamina well beyond the capacity of ectotherms: mammals and birds can maintain even relatively high levels of activity for extended periods of time. In contrast, ectotherms, such as reptiles, typically rely on nonsustainable, anaerobic metabolism for all activities beyond relatively slow movements. Although capable of spectacular bursts of intense activity, ectotherms generally exhaust soon as they deplete cellular glycogen stores and accumulate lactic acid. The increased stamina that endothermy provides has significant ecological benefits: mammals and birds are able to defend more extensive territories or home ranges and forage more widely than reptiles, and they have the ability to migrate over extensive distances in search of food. With respect to dinosaur palaeobiology, therefore, the most intriguing question concerns not so much their body temperature but the rates of oxygen consumption they sustained at rest and during maximal activity.

How can metabolic rates be deduced from fossils? To be sure, living endotherms and ectotherms are distinguished easily enough. Endotherms have made profound modifications to permit such high levels of aerobiosis: compared to living reptiles, both mammals and birds have highly specialized lungs with greatly enhanced gas-exchange capacities; they have four-chambered hearts, fully divided pulmonary and systemic circulations, expanded cardiac outputs and higher blood pressures. Their blood volume and blood oxygen carrying capacity are greater than those of reptiles, they have more mitochondria, and these in turn have greater enzymatic activities. Unfortunately, these key features of endothermy are not preserved in fossils. Previous discussion of the metabolic status of dinosaurs and other extinct forms have therefore relied primarily on speculative or circumstantial lines of evidence, such as predator/prey ratios or fossilized trackways, or on a variety of supposed correlations of metabolic rate to avian or mammalian morphology such as limb posture, brain size, bone histology, growth rates and so forth. However, closer scrutiny has revealed that virtually all of these arguments are equivocal at best (Farlow et al., 1995; Reid, 1997; Ruben et al., 2003; Chinsamy and Hillenius, 2004). A recent report of a putative fossilized, four-chambered dinosaur heart was subsequently revealed to concern a geological, not a palaeontological structure (Rowe et al., 2001) and moreover, failed to recognize that a fully partitioned, four-chambered heart is also present in ectothermic crocodilians. Similarly, reports of filamentous ‘feather-like’ or ‘down-like’ integumentary structures in Chinese theropods, which have been interpreted as evidence of an insulative covering and thus suggestive of endothermic homeothermy, should also be regarded with considerable caution, as they may comprise partially decayed connective tissues rather than feathers (e.g. Lingham-Soliar, 2003 and Feduccia et al., 2005). Moreover, the presence of even a fully developed set of feathers in dinosaurs, or even in Mesozoic birds, was not necessarily linked to any particular pattern of metabolic physiology. Numerous modern birds are known to utilize behavioural thermoregulation and absorb and use incident solar radiation even while feathered (e.g. Ohmart and Lasiewski, 1971). Feathered dinosaurs, or birds such as Archaeopteryx, whether ectothermic or endothermic, might easily have had a similar behavioural thermoregulatory capacity, and, as explained above, would readily have been able to achieve homeothermy under most Mesozoic climatic conditions.

A problem that has dogged virtually all of these previous arguments is that they are based predominantly on apparent similarities to the mammalian or avian condition, without a clear functional correlation to endothermic processes. Feathers, for example, do not produce heat, nor do they contribute to the uptake, transport or delivery of oxygen. While feathers serve a variety of functions in living birds, it remains very much unclear which of these, if any, is most primitive. The more promising—and more direct—approach to gaining insight into the aerobic capacity of dinosaurs is to look for preservable attributes that have a clear, direct, functional correlation with endothermic processes such as oxygen consumption or lung ventilation rates. Two lines of evidence have permitted an assessment of the ventilatory capacities of dinosaurs and other fossil vertebrates. These include the presence of respiratory turbinates, which are indicative of elevated resting metabolic rates (a key attribute of endothermy), and evidence of lung structure and ventilation, which reflect the maximal aerobic capacity of the animals.

Respiratory Turbinates and the Resting Metabolic Rates of Dinosaurs

Because the lung ventilation rates of living mammals and birds are high even at rest, they are chronically exposed to potentially disastrous rates of dehydration and heat loss, as large volumes of warm, humid air are expelled from the lungs. However, both groups have independently evolved a mechanism—nasal respiratory turbinates—to recapture and recycle the bulk of the heat and moisture contained in respired air. Respiratory turbinates are complex, convoluted bony or cartilaginous structures in the nasal cavity. Covered in moist epithelium, they are situated in the nasal air stream so that air passes over them during both inhalation and exhalation. As the animal inhales cool outside air, the turbinate epithelia give off heat and moisture. This prevents desiccation of the lungs, but also cools the turbinates. During exhalation, therefore, as warm, moist air returns through the nose, the moisture condenses on to the cool surfaces of the turbinates, and can be recycled in the next breath. Over time, this results in a substantial savings of heat and water: as much as an estimated 10% of daily heat production, and up to 30-50% of the daily water budget. All nostril-breathing terrestrial mammals and birds have such turbinates; they are lacking only in certain specialized marine diving animals. In contrast, respiratory turbinates are absent in all extant ectotherms: their lung ventilation rates are low enough that respiratory heat and water loss are not critical problems. The widespread presence of respiratory turbinates in both mammals and birds strongly suggests that these structures evolved, independently in each lineage, in conjunction with the origin of endothermy to support elevated resting metabolic and lung ventilation rates.

The presence or absence of respiratory turbinates among fossil forms is thus a good indicator of their resting ventilation and metabolic rates, especially in those animals generally thought to represent the evolutionary ancestors to modern birds. Unfortunately, turbinates themselves are fragile and not often preserved in fossils. In birds, moreover, they are typically cartilaginous, which further diminishes their chances for preservation. However, in extant mammals and birds a correlation exists between the presence of respiratory turbinates and a marked increase in the diameter of the nasal passage; presumably, this compensates for the increase in resistance to air flow presented by these structures.

Several lines of evidence indicate that dinosaurs had comparatively narrow, ectotherm-like, nasal passages. Computed tomography examinations of several specimens (the theropods Nanotyrannus and Ornithomimus and the hadrosaurHypacrosaurus) revealed no evidence of preserved respiratory turbinates, although remnants of preserved olfactory turbinates were found in the nasal cavity of Nanotyrannus (olfactory turbinates, which enhance the sense of smell, are found in all amniotes, including reptiles). But these scans did show that in all three cases the respiratory nasal passage had a very small diameter, which was proportionally nearly identical to those of extant ectotherms. It thus appears very unlikely that these three dinosaurs possessed respiratory turbinates (Ruben et al., 1996). Moreover, this condition was probably not limited to these three taxa: other theropods apparently lacked respiratory turbinates as well. The CT scans indicate that most of the space in the snout of the two theropods was taken up by large nasal sinuses—blind, air-filled chambers, which do not function in respiration. Such nasal sinuses are also well known from most other advanced theropods, as well as from basal birds, such as Archaeopteryx and the enantiornithine birds. In these taxa, apertures in the rostral portions of the skull indicate the extent of the sinuses, and suggest that the respiratory nasal passage was narrow and most likely lacked respiratory turbinates (cf. Hillenius and Ruben, 2004).

Absence of respiratory turbinates indicates that theropod and other dinosaurs, as well as basal birds, did not have a serious problem with chronic respiratory water loss, and thus that their resting lung ventilation and metabolic rates were probably low compared to those of modern birds. It is possible that the resting ventilation rates of some or perhaps all of these groups were modestly higher than those of modern reptiles, but not yet so high as to warrant a water-recovery mechanism in the nasal air stream. However, at present there is no positive evidence to confirm that the resting ventilation rates of any nonavian dinosaur or basal bird had expanded significantly above those of extant reptiles. Interestingly, in later birds the nasal cavity was drastically redesigned and enlarged. The first birds with nasal cavities large enough to accommodate respiratory turbinates were the mid-Cretaceous ornithurine birds, such as Hesperornis (Hillenius and Ruben, 2004). Thus, it appears that the most significant evolutionary change in metabolic status took place within birds, between basal forms and the ornithurines.

Lung Structure, Ventilation, and the Maximal Metabolic Rates of Dinosaurs

Resting metabolic rates only provide part of the picture of the metabolic status of animals. During intense activity, maximal oxygen assimilation rates determine both the level of activity and the length of time it can be sustained. Both mammals and birds have highly modified lungs that are capable of sustaining very high rates of oxygen consumption. The lungs of reptiles, on the other hand, are much less effcient: reptiles are capable of no more than 15-20% of the maximal performance rates of endotherms, and even in a best-case scenario based on hypothetical improvements of the circulatory system and optimized lung parameters, reptilian lungs attain no more than 50-70% of those rates. For the reconstruction of dinosaur biology, therefore, it is of considerable interest to know what their lung architecture and ventilation pattern may have been. Because of the phylogenetic affnity of dinosaurs with crocodilians and birds, it seems likely that they possessed at least a large, multichambered, crocodilian-like lung. However, the truly interesting question is whether there is enough evidence to support the inference of a more highly specialized, high-performance lung like those of modern birds. The lungs themselves do not preserve in fossils, but some information about their structure and ventilation pattern can be deduced from the postcranial skeleton.

Bird lungs are unique among modern amniotes: they are composed of a highly specialized system of lungs and auxiliary airsacs that allow a continuous, unidirectional stream of air to pass over the respiratory exchange surfaces. Although clearly derived from simpler reptilian lungs, the lung/airsac system of birds is significantly more compartmentalized and complex. Whereas in the lungs of other tetrapods the functions of gas exchange and air movement are mostly integrated, in the avian respiratory system these aspects are completely divorced. The lungs themselves are composed of a series of narrow, parallel tubes, the parabronchi, that are the sites of gas exchange. The lungs are surprisingly compact and rigid, and their volume remains constant during the ventilation cycle. Instead, virtually all volume change for ventilation is provided by the nonvascularized, but highly elastic airsacs, which extend deeply into the abdomen between the visceral organs. The unique pattern of air flow this arrangement permits, combined with a crosscurrent flow of blood through the lungs as well as the large total amount of respiratory surface area in the parabronchial walls, make the avian lung one of the most effcient respiratory organs of air-breathing vertebrates.

Birds also ventilate their lungs differently from other tetrapods: rather than relying on lateral movements of the ribs like most tetrapods, birds use an up-and-down rocking motion of their very large sternum to expand and contract the airsacs. This is made possible by a unique system of hinged ribs with specialized joints between the vertebral and sternal ribs, and especially between the sternal ribs and the sternum. The edges of the sternum, in particular, are very robust with transversely expanded joint facets, and the sternal ribs that articulate with these have a characteristic flattened, tapered shape.

These specialized features are unknown in dinosaurs: instead, the ribcage of most dinosaurs is quite ordinary. Ossified sternal ribs have been documented for a number of taxa, including oviraptorids, ornithomimids, and dromaeosaurs, but in each case they lack the characteristic tapered shape associated with the sternocostal joints of birds. The sternal plates of dinosaurs are generally short, and all lack the thickened lateral edge and transverse hinge joints that characterize avian sterna. The sternum of the dromaeosaur Bambiraptor is unusually long, but its edges, where the sternal ribs articulated, are nevertheless thin and fragile. The size, shape and orientation of the sternocostal joints are critical, because simple articulations between sternum and sternal ribs occur widely among amniotes, including in mammals, crocodilians and lizards (although the sternal ribs are often not ossified). Simple contacts between ribs and sternum are therefore not diagnostic for an avian-like lung ventilation system.

Interestingly, the ribcages of basal birds, such as Archaeopteryx, Confuciusornis and the enatiornithine birds, also lack these specialized sternocostal joints. No ossified sternum is preserved in Archaeopteryx (Wellnhofer and Tischlinger, 2004), but sterna are known from Confuciusornis and at least some enantiornithes. In all cases where the edges are preserved, however, these are thin and fragile and show no sign of specialized, expanded rib joint facets. Ossified sternal ribs are known from only a few of these forms, and in all cases lack the characteristic tapered shape of the sternal ribs of modern birds. The earliest documented appearance of a robust sternum with specialized sternocostal joints is in the mid-Cretaceous ornithurine bird Hesperornis (Hillenius and Ruben, 2004).

The presence of uncinate processes, posterodorsal projections on the thoracic ribs, in several nonavian theropods and basal birds are sometimes mentioned in this context. But the uncinate processes of birds serve no particular role in lung ventilation: instead, they primarily serve to strengthen the ribcage and to stabilize the shoulder. Moreover, cartilaginous or ossified uncinate processes are also known from crocodilians, Sphenodon and some labyrinthodonts. Therefore, it is unlikely that uncinate processes are indicative of a particular breathing adaptation.

Similarly, the presence of pneumatized (air-filled) vertebrae of sauropods, nonavian theropods and Archaeopteryx has been cited as evidence for a bird-like lung/airsac system. However, there is no compelling functional link between pneumatization of the skeleton and the parabronchial lung design. First, hollow bones do not contribute to respiration or ventilation in modern birds: although the air cavities within the bones are connected to the respiratory system, they cannot expand or contract. Instead, they are usually regarded as a means for weight reduction (which might also explain their presence in large animals). Secondly, pneumatized bone is not limited to birds: crocodilians have extensive air cavities (sinuses) in their skulls, and this structural feature may be an ancestral attribute of the archosaurs. Postcranial pneumaticity may thus signify nothing more than the expression of that ability in other regions of the body. Because it plays no role in lung function, skeletal pneumaticity, whether cranial or postcranial, cannot be considered diagnostic for the advanced lung design of modern birds.

Another suggestion is that dinosaurs and basal birds may have used their well-developed gastralia (belly ribs) to ventilate an avian-like abdominal airsac system. According to this proposal, pelvic muscles would pull the gastralia posteriorly and cause the ventrolateral body wall to expand, thus allowing the airsacs to fill. Although this is an intriguing proposal, it has several problems. First, there is no extant example of the proposed model. Although primitively, gastralia occurred widely throughout amniotes, they are absent in modern birds and thus cannot be considered diagnostic for a bird-like lung/airsac system. Moreover, in those extant forms that retain them (crocodilians and Sphenodon), gastralia are not known to contribute to inhalation as proposed. Second, in dinosaurs the gastralia are limited to the ventral body wall, and a large, rather deep and unprotected flank region exists between the dorsal tips of the gastralia and the ribs. Negative pressures generated through the proposed action of the gastralia would likely have resulted in a paradoxical, inward movement of this unreinforced region, and cause significant loss of ventilatory effciency. At present, this reconstruction must be considered highly speculative at best.

There is, therefore, no compelling evidence for avianlike, high-performance lungs in dinosaurs or basal birds. The most reasonable reconstruction is that these animals had large, multichambered lungs broadly similar to those of extant crocodilians, or possibly a somewhat modified ‘protoparabranchial’ lung. But all lacked the hinged ribs and specialized sternocostal joints that are associated with the highly derived, flow-through lung/airsac system lungs seen in modern birds. This suggests that these animals had not attained the levels of maximal aerobic activity and endurance typical of modern birds.

However, there is some evidence to indicate that theropods, at least, may have nevertheless been able to improve lung ventilation, and may have had an aerobic capacity greater than typical for extant reptiles. In two theropod fossils, the compsognathid Sinosauropteryx and the maniraptoran Scipionyx, preserved soft-tissue remains suggest that the abdominal cavity was subdivided into distinct anterior (pulmonary) and posterior (peritoneal) compartments (Ruben et al., 1997, 1999). Such a subdivision of the abdominal cavity is not seen in modern birds (in which the liver is located not behind but below the lungs), but closely resemble the condition of modern mammals and crocodilians. In crocodilians, longitudinal muscles connected to the pelvis pull the transversely oriented liver backwards in a piston-like fashion to augment lung ventilation. In Scipionyx, a patch of muscle tissue with a similar orientation, which is preserved just in front of the pubis, appear to correspond to this so-called diaphragmatic muscle, and it seems likely that these and other theropods had a ‘hepatic piston’ mechanism of lung ventilation similar to that in crocodilians. The large, robust pubis of theropods broadly resembles that of crocodilians, and would have served admirably as a point of origin for these diaphragmatic muscles. Hepatic piston respiration would not only have allowed deeper inhalations than otherwise, but the diaphragmatic muscles would also have been able to control visceral movements during locomotion, and there-by optimize breathing while running.


Questions about when endothermic status was attained in any evolutionary lineage can only be answered with positive evidence of fossilizable attributes unequivocally and causally linked to this derived metabolic condition. The lack of respiratory turbinates in dinosaurs and basal birds suggests that these taxa still had comparatively low resting metabolic rates. Similarly, both of these groups lack compelling evidence for a derived, parabronchial lung/airsac system similar to that of modern birds, and probably had a limited capacity for sustained maximal activity. However, theropods, at least, may have had a hepatic-piston mechanism to supplement lung ventilation, and the aerobic scope and stamina of these cursorial animals may have been somewhat higher than that of extant reptiles. The available evidence suggests that full endothermic status was probably reached first by mid-Cretaceous ornithurine birds.

Reconstructing dinosaurs and basal birds as essentially ectothermic does not mean that they were necessarily sluggish, slow-moving creatures vulnerable to ambient thermal conditions. On the contrary, most forms would have had little trouble achieving effective homeothermy through behavioural or inertial means. Furthermore, the metabolic maintenance costs of ectotherms is considerably lower than those of endotherms, and ectothermy may well have been a more advantageous strategy during the Mesozoic. This reconstruction thus envisions dinosaurs as active, dynamic ectotherms, which opportunistically took advantage of unprecedentedly favourable climatic conditions during the Mesozoic to achieve near-total ecological dominance in the terrestrial habitats.