Paleobiology, 29(1), 2003, pp. 105–122

Dinosaur body temperatures: the occurrence of endothermy and ectothermy

Frank Seebacher

Abstract.—Despite numerous studies, the thermal of dinosaurs remains unresolved. Thus, perhaps the commonly asked question whether dinosaurs were ectotherms or endotherms is inappropriate, and it is more constructive to ask which dinosaurs were likely to have been en- dothermic and which ones ectothermic. Field data from crocodiles over a large size range show that body temperature fluctuations decrease with increasing body mass, and that average daily body temperatures increase with increasing mass. A biophysical model, the biological relevance of which was tested against field data, was used to predict body temperatures of dinosaurs. However, rather than predicting thermal relations of a hypothetical dinosaur, the model considered correct paleogeographical distribution and climate to predict the thermal relations of a large number of dinosaurs known from the record (Ͼ700). Many dinosaurs could have had ‘‘high’’ (Ն30ЊC) and stable (daily amplitude Յ2ЊC) body temperatures without metabolic heat production even in winter, so it is unlikely that selection pressure would have favored the evolution of elevated resting metabolic rates in those . Recent evidence of ontogenetic growth rates indicates that even the juveniles of large species (3000–4000 kg) could have had biologically functional body temper- ature ranges during early development. Smaller dinosaurs (Ͻ100 kg) at mid to high latitudes (Ͼ45Њ) could not have had high and stable body temperatures without metabolic heat production. How- ever, elevated metabolic rates were unlikely to have provided selective advantage in the absence of some form of insulation, so probably insulation was present before endothermy evolved, or else it coevolved with elevated metabolic rates. Superimposing these findings onto a phylogeny of the Dinosauria suggests that endothermy most likely evolved among the Coelurosauria and, to a lesser extent, among the Hypsilophodontidae, but not among the Stegosauridae, Nodosauridae, Anky- losauridae, Hadrosauridae, Ceratopsidae, Prosauropoda, and Sauropoda.

Frank Seebacher. School of Biological Sciences, A08, The University of Sydney, New South Wales 2006, Australia. E-mail: [email protected]

Accepted: 20 June 2002

Introduction species (Paladino et al. 1990, 1997) to learn about dinosaur body temperatures (T ). Heat The thermal physiology of dinosaurs has b transfer relations between animals and their been the subject of much discussion for de- environment are extremely complex, and the cades (Colbert et al. 1946; Bakker 1972; Farlow exact value of some parameters, such as con- 1990; Seebacher et al. 1999; O’Connor and Dod- son 1999). Much of the debate has focused on vection coefficients, can only be determined the question of whether dinosaurs were endo- empirically, as their complexity precludes the- therms or ectotherms. This, however, is not the oretical derivation (Mitchell 1976; Incropera appropriate question to ask. Given the enor- and DeWitt 1996). Hence, biophysical model- mous diversity of the Dinosauria, both in space ing of Tb in relation to environmental condi- and time, all dinosaurs are unlikely to have tions requires a number of simplifying as- possessed the same physiological make-up. sumptions and approximations (O’Connor Hence, it is more constructive to ask which di- and Spotila 1992). Predictions from theoretical nosaurs would have gained a selective advan- models, therefore, are informative only if their tage from elevated metabolic rate and its atten- underlying assumptions are valid in the bio- dant heat production, and which dinosaurs logical context under consideration, and the would have benefited more from ectothermy? only way to test the validity of theoretical Past research on dinosaur thermal physiol- models is to compare their predictions against ogy has used biophysical modeling (Spotila et empirical data. A major limitation of most bio- al. 1973; Dunham et al. 1989; O’Connor and physical studies of dinosaur Tb is that the rel- Dodson 1999) and comparisons with extant evance of their predictions has not been tested,

᭧ 2003 The Paleontological Society. All rights reserved. 0094-8373/03/2901-0016/$1.00 106 FRANK SEEBACHER so that the results must be treated as specu- edge gained from other animals, such as tur- lative. tles and elephants, will be the best approach Comparisons with extant species are poten- to advance knowledge in this field. tially useful, but the value of such compari- It was the aim of this study to explore body sons depends to a large extent on the choice of temperatures (Tb) of a large sample of dino- the model species, because no modern ani- saurs represented in the fossil record under the mals are quite like dinosaurs. Elephants, conditions prevalent at the time and place which have been considered an endothermic where they lived, in order to identify those model for dinosaurs (Paladino et al. 1997), can groups in which selection pressures favoring reach body mass comparable to medium- endothermy would have been greatest or least. sized dinosaurs, so studying elephants can re- The study uses crocodiles as a ‘‘null-model,’’ veal some of the challenges that large terres- and crocodiles are assumed to represent the trial animals must have faced. However, ele- simplest physiological and behavioral config- phants do not represent ‘‘typical’’ endo- uration possible for dinosaurs. The hypothesis therms. Their great mass places elephants at tested here is that as crocodile-like ectotherms one extreme of endothermic radiation, and that thermoregulated behaviorally by moving their morphological and behavioral features between heating and cooling environments that have evolved in order to dissipate heat (Seebacher and Grigg 1997; Grigg et al. 1998; (Wright 1984; Williams 1990) bear testimony Seebacher 1999) dinosaurs could have had high that endothermy in large terrestrial animals is and stable Tb (see below for definitions). Test- the exception rather than the rule. Because ing this hypothesis will reveal which dino- they reveal that it was of selective advantage saurs, living at a particular place and time, may for large endotherms to evolve heat-dissipat- have had high and stable Tb with an ectother- ing mechanisms, elephants are interesting, mic physiological make-up, so that selection but it would be illogical to use an endothermic pressures favoring endothermy would have exception to establish a rule for dinosaurs. been comparatively weak in those species, and, Similarly, marine turtles (Paladino et al. vice versa, which dinosaurs were unlikely to 1990) are atypical reptiles because they are have been simple ectotherms. wholly aquatic and the selection pressures Endothermy may not have evolved as a re- that acted to produce modern turtle species sult of selection pressures acting directly on must have been quite different from those that thermoregulatory ability. Rather, it has been acted on terrestrial species. Nonetheless, the suggested that selection pressures favored fact that leatherback turtles are able to main- high levels of activity (Bennett and Ruben tain Tb elevated above water temperatures (Pa- 1979; Hayes and Garland 1995) or increased ladino et al. 1990) is of interest and bears tes- capability for parental care (Farmer 2000; Ko- timony to the diversity of thermal relations teja 2000). In either scenario, high maximal among reptiles. aerobic metabolic rates would have been se- Several authors have used crocodilians as lected for and the increase in resting metabolic models to learn about dinosaur thermal rela- rate, which characterizes endothermy, would tions (Colbert et al. 1946; Spotila et al. 1973; have been a correlated response. However, Seebacher et al. 1999). Crocodiles are probably aerobic metabolic rates are dependent on the the most appropriate models for dinosaurs be- thermal sensitivity of underlying biochemical cause they are the last living non-avian archo- processes (Crawford et al. 1999; St. Pierre et al. saurs and are likely to be similar to the type 1998), so and high aerobic of basal archosaur from which dinosaurs metabolic rates cannot be viewed in isolation. evolved (Parrish 1997). In addition, crocodil- Moreover, animals that are ‘‘warm’’ without ians are the largest extant reptiles with an possessing metabolic rates typical of modern enormous ontogenetic size range. However, endotherms could also maximize aerobic met- this is not to say that crocodilians are the only abolic rates, because their Tb ranges would be useful model. Crocodilians may be the single optimal for related biochemical processes. most useful model, but a synthesis of knowl- Hence, it would be wrong to view endothermy DINOSAUR BODY TEMPERATURES 107 as ‘‘superior’’ to ectothermy, because both Materials and Methods have their advantages and disadvantages Crocodile Field Data. Field data and sam- (Pough 1980); in addition, selection pressures pling methods were reported previously (See- favoring endothermy would be relatively bacher and Grigg 1997; Grigg et al. 1998). Brief- weak in animals that could be ‘‘warm’’ with- ly, 11 free-ranging crocodiles (Crocodylus po- out incurring the high energetic cost of ele- rosus; 32–1010 kg) were studied in both winter vated, endothermic metabolic rates. and summer under seminatural conditions at In many respects endothermy and ectother- Edward River Crocodile Farm on Cape York my differ in degree rather than in kind, because Peninsula, Australia (14Њ55ЈS,141Њ35ЈE). Tb all animals produce metabolic heat, and the was measured with calibrated temperature- mechanisms responsible for metabolic heat pro- sensitive radio transmitters (Sirtrack, Have- duction are similar in all animals (except for lock North, New Zealand) that crocodiles brown fat in some hibernating mammals). En- swallowed and retained as pseudogastroliths dotherms, however, have an increased capacity in their stomachs. Each animal was sampled for metabolic heat production as a result of, for for 4–41 days. example, increased relative organ mass, in- In addition, data from two freshwater croc- creased density and complexity of mitochon- odiles (Crocodylus johnstoni; 2.5–3.6 kg) were dria, and increased activity of ion pumps in used to validate predictions of the theoretical their cell membranes (Else and Hulbert 1981, model (Seebacher and Grigg 1997; Seebacher 1987). Moreover, mechanisms of heat transfer et al. 1999). Wild Crocodylus johnstoni were between animals and their environment are sampled in the Lynd River, Cape York Penin- identical regardless of metabolic rates, and the sula, Australia (17Њ07ЈS,144Њ03ЈE), and Tb was importance of metabolic heat in thermoregula- recorded with temperature-sensitive trans- tion is meaningful only when interpreted in mitters that were surgically implanted into context of heat lost or gained from the environ- the abdominal body cavity. ment. Hence, ‘‘endothermy’’ and ‘‘ectothermy’’ Average daily Tb was calculated as the in- describe extreme expressions of the same mech- tegral of continuous Tb measurements during the day divided by the period of integration. anisms rather than different evolutionary ‘‘in- This is conceptually similar to an arithmetic ventions.’’ It is conceivable, therefore, that dur- mean, but it does not suffer from the non-in- ing the evolution of endothermy there existed a dependence of individual measurements. continuum of physiological constitutions that The behavior of all study animals was re- were intermediate to those known from typical corded by scan sampling and direct observa- ectotherms and endotherms today (Reid 1997a; tion during daylight hours. O’Connor and Dodson 1999). In the text below, Operative environmental temperatures (Bak- ‘‘ectothermy’’ and ‘‘endothermy’’ are used to ken and Gates 1976; Tracy 1982; Seebacher 1999) denote metabolic of modern rep- were calculated from measurements of environ- tiles, and of birds and mammals, respectively. mental parameters in the field. Shaded air tem- Dinosaur Tb were calculated with a heat peratures, shallow and deep water tempera- transfer model the predictive validity of tures, and ground temperatures were measured which was tested extensively against a large with temperature sensors (National Semicon- set of field data from crocodiles (Crocodylus po- ductors LM335) connected to a data-logger rosus and C. johnstoni) ranging in mass from (Data Electronics, Melbourne, Australia). Solar 2.5 to 1010 kg (Seebacher 1999; Seebacher et al. radiation was measured with a tube solarimeter 1999). This approach differs from previous (Irricrop Technologies, Narrabri, Australia) con- studies, because the theory and its underlying nected to the same data logger. Operative tem- assumptions have been validated against field peratures change with relative exposures of data and because an attempt is made to con- body surface area to different heat transfer sider ‘‘real’’ dinosaurs in their appropriate pa- mechanisms, so that different behavioral pos- leoenvironment. tures have different operative temperatures as- 108 FRANK SEEBACHER sociated with them. Behavioral postures were perature distributions (Incropera and DeWitt incorporated into operative temperature calcu- 1996) of each layer gives the following con- lations by changing relative contributions of duction equations: heat transfer by conduction, convection in air C dT /dt ϭϪK(T Ϫ T ) (1) and water and radiation as described by Grigg cc cb s et al. (1998) and Seebacher (1999). Css dT /dt ϭ K(T cbϪ T) sϪ K(T ssϪ T e ) (2) Calculations of Dinosaur T . Dinosaurs in- b for the core (1) and outer layer (2), where C cluded in this study and their paleogeograph- c and C are the heat capacities (cM, where c ϭ ic distributions were taken from the catalog s specific heat, and M ϭ mass) for the core and published by Weishampel (1990), as well as outer layer, respectively; K and K represent from the sources listed in the Appendix. In to- c s kA/l, where k ϭ conductivity, A ϭ surface tal, 701 dinosaurs from 321 genera were con- area, and l ϭ thickness through the core and sidered. Body mass of dinosaurs was calculat- outer layer, respectively. T and T are the core ed according to methods described by See- c s and outer layer temperatures, respectively, bacher (2001). Note that this method of mass and T is the operative environmental temper- calculation takes the different shapes of di- e ature. Substituting for T gives nosaur groups into account, but possible s 22 shape-related modifications in heat transfer (Csc C /K cs K )d T b /dt were not considered beyond the calculations of body mass. When a fossil did not allow an ϩ (Ccs /K ϩ C cc /K )dT b /dt ϩ T b accurate determination of the animal’s length ϭ Te (3) (from which mass could be calculated [See- bacher 2001]), the mass of a closely related The particular solution chosen for this second- species was used instead. For example, Gra- order differential equation was a Fourier series ham et al. (1997) describe the fossil of the right to describe the slight asymmetry of the daily foot of a sauropod that was identified as be- Tb and Te fluctuations observed in the field longing to the genus Mamenchisaurus,sothe (see ‘‘Results,’’ below; also see Grigg et al. mass of this specimen was assumed to have 1998; Seebacher et al. 1999): been similar to that of Mamenchisaurus hochu- Tb ϭ T Ϫ A sin((2Pt) ϩ␸) anensis.

‘‘Stable’’ Tb was defined as one fluctuating ϩ c1 A sin((4Pt) ϩ␸) with an amplitude of 2ЊC or less, a definition Ϫ c2 A sin((6Pt) ϩ␸) (4) drawn from the Tb fluctuations expected from endothermic mammals (Lovegrove et al. where T ϭ mean temperature, A ϭ amplitude,

1987), and one that has been used previously P ϭ period, t ϭ time, ␸ϭphase angle, and c1 in relation to dinosaurs (Barrick and Showers and c2 are constants determining the ‘‘saw-

1994). ‘‘Warm’’ Tb was defined as 30ЊCor tooth’’ of the periodic motion (Halliday and above. This is somewhat cooler than most Resnick 1978), which we determined empiri- modern birds and mammals (but not atypical cally to be 0.305 and 0.109, respectively (See- for some [Lovegrove et al. 1987]), and it is typ- bacher et al. 1999). ical for crocodilians (Seebacher and Grigg For investigations of the effect of insulation

1997; Grigg et al. 1998), which are assumed to on Tb an extra body layer was added to the be a model for ancestral archosaurs. model above. Transient heat conduction Heat conduction between the surface and through this insulation layer was calculated the core of the animal was calculated for an by nondimensional heat transfer analysis animal consisting of two thermally distinct (Carslaw and Jaeger 1995; Incropera and layers (Turner 1987; Seebacher et al. 1999): an DeWitt 1996): outer layer made up of muscle and fat with a (Fo (5ץ/␪ץx* ϭץ/*␪22ץ thickness of 0.15 total radius and an inner core consisting of bone, tissue and fat with 0.85 to- where ␪* ϭ (T Ϫ Te)/(Ti Ϫ Te), and Ti is the tal radius. Applying Fourier’s law to the tem- initial (skin) temperature; x* ϭ x/L where L DINOSAUR BODY TEMPERATURES 109

ϭ half thickness of the insulation; and Fo is the son et al. 1996). Frakes et al. (1994) show a tem- Fourier number. A first-order approximation perature increase (both oceanic and atmo- of the infinite series solution to the above spheric) throughout the Early Cretaceous, equation was used, where the midplane tem- which is followed by a sharp decline in tem- perature perature in the early Late Cretaceous leading to a distinct minimum in the Maastrichtian pe- (␪* ϭ Ce. ٙ (Ϫ␨2.Fo) (6 01 1 riod in the late Late Cretaceous. and C1 is the coefficient for the solution of the Paleoclimate estimates incorporated the

first root (␨1)of␨ntan␨n ϭ Bi where Bi is the patterns described above and were based on Biot number. Conductivity of insulation was values collated from the various sources. Pa- assumed to be the same as that of fur in mod- leoclimate predictions vary in the literature, ern mammals (Bowman et al. 1978). and the estimates presented here represent an In the analysis, dinosaurs were assumed to attempt at a conservative consensus ‘‘model’’ thermoregulate behaviorally by moving be- that is warmer than the coolest prediction tween sun and shade. Differential exposure to (sub-zero temperatures at high latitudes in the sun and shade changes operative environmen- Early Cretaceous [Sloan and Barron 1990]) tal temperatures experienced by animals, and cooler than the warmest prediction (mean which can be calculated (Seebacher 1999) and annual polar temperatures exceeding 14ЊCin incorporated into the thermal model above. the Early Cretaceous [Tarduno et al. 1998]). Paleoclimate. Estimates of paleoclimate were More specifically, equatorial temperatures made from various lines of evidence in the lit- varied little throughout the Mesozoic and, erature (Vakhrameev 1991; Parrish 1993; Golon- judging from paleoclimate modeling maps, ka et al. 1994; Sellwood et al. 1994; Tarduno et were estimated to range from 28.8ЊC in Early al. 1998). It was assumed that solar radiation Jurassic winter to 30.5ЊC in Early Cretaceous throughout the Mesozoic was similar to present- summer (Golonka et al. 1994; Barron 1993; day levels and that there was no sun in winter Frakes et al. 1994). Average latitudinal gradi- at latitudes higher than 70Њ. ents in temperature were estimated to range It is generally agreed that the Mesozoic was from 0.27–0.31ЊC/1Њlat in winter to 0.17– warmer than the present (Barron 1983). The 0.21ЊC/1Њlat in summer (Barron 1983; Yemane climate during the Permian was temperate 1993; Frakes et al. 1994; Francis 1994; Moore (Yemane 1993), and the Pangean megamon- and Ross 1994; Huber et al. 1995; Huber and soon was strongest during the following Tri- Hodell 1996). Equator-to-pole gradients were assic, after which it abated so that the climate uneven, however, with a broader tropical re- became drier during the Jurassic (Parrish gion that peaked in extent in the late Early 1993). This drying was accompanied by cli- Cretaceous (Johnson et al. 1996) and, reflect- matic cooling during the Triassic and Jurassic ing modern climate trends, the greatest de- period (Bardossy 1996), but the climate was crease in temperature occurs at midlatitudes nonetheless ‘‘equable’’ (subtropical to warm- (Golonka et al. 1994). Daily variation in tem- temperate) during the Jurassic (Vakhrameev perature was assumed to be similar to mod- 1991). ern-day variation at the same latitude, and pa- The Early Cretaceous and the early Late Cre- leolatitudes were determined from maps of taceous were very warm periods (Mayer and Smith et al. (1994). Appel 1999; Herman and Spicer 1996; Tarduno et al. 1998), but there also is good evidence that Results and Discussion the climate cooled considerably during the Crocodile Field Data. The Tb of crocodiles Cnemonian–Turonian periods of the early Late larger than 32 kg followed slightly asymmet- Cretaceous (Sellwood et al. 1994; Barron 1995; ric oscillations during the day (Fig. 1). The Kuypers et al. 1999). Massive heat transport asymmetry, i.e., heating phase is shorter than from the superheated Tropics toward the poles the cooling phase and reflects the relatively has been cited as at least part of the reason for short period during the 24-hour day when so- the early Late Cretaceous cooling event (John- lar radiation can be absorbed. The amplitude 110 FRANK SEEBACHER

FIGURE 2. Average daily Tb of crocodiles increased with FIGURE 1. Upper panel, Examples of daily Tb fluctua- body mass. Taking small crocodiles as reference points tions in crocodiles of different mass: 42 kg (open cir- (2.6 kg in winter and 3.5 kg in summer), average daily cles), 520 kg (solid circles), 1010 kg (triangles). Lower Tb increased by several degrees over a size range of 1000 panel, Daily Tb amplitude decreased with mass; Tb am- kg. Redrawn from data of Seebacher et al. (1999). plitudes are dependent on fluctuations in operative tem- perature and are therefore expressed as the ratio of Tb to operative temperature amplitude. The two outliers (open circles) represent crocodiles that were subjected ϭ 0.88) so that a 1010-kg crocodile was on av- to constant aggression by dominant males. Redrawn erage 3–4ЊC warmer than the reference animals from data of Seebacher et al. (1999). (Fig. 2). To test the validity of the mathematical of the daily Tb oscillations decreased with model in predicting Tb, all field data were also body mass, so that in winter Tb varied be- predicted mathematically, given the same en- tween 2Њ and 3ЊC during the day in a 1010-kg vironmental conditions measured in the field crocodile, compared with nearly 10ЊC in a 42- and the crocodiles’ body mass (Fig. 3). Theory kg crocodile (Fig. 1). The Tb of two crocodiles predicted measured Tb extremely well at any did not follow the general pattern (Fig. 1, low- season and for any crocodile mass (Fig. 3). er panel, 233 kg and 865 kg). Both these ani- Overall, there were no significant differences mals were the second largest males in their re- between the predicted and the measured am- spective area and were therefore subjected to plitudes of the daily Tb oscillations (t-test ϭ nearly constant aggression by the dominant Ϫ0.31, df ϭ 8, p ϭ 0.76). males. The resulting altered behavior patterns Paleoclimate. Air temperatures varied by explain the difference in Tb between these and up to 10ЊC between geological time periods, the other crocodiles (Grigg et al. 1998). For and there was a gradient between the equator this reason, data from these two animals have and the poles of nearly 20ЊC in summer and been omitted below. 30ЊC in winter. The Early Jurassic and Late

As well as increasing in stability, Tb in- Cretaceous were the coolest periods, and the creased in magnitude with increasing mass early Cretaceous was the warmest. Seasonal

(Fig. 2). This increase in Tb is the result of the differences in mean daily air temperatures decrease in convective heat transfer (i.e., the in- were most pronounced at high latitudes (up to crease in boundary layer thickness) as croco- 10ЊC difference between summer and winter dile length increases. Taking the average daily at the poles), but differences in air tempera-

Tb of a 2.6-kg (winter) and a 3.5 kg (summer) ture at the equator were negligible (Fig. 4). crocodile as the reference points, Tb magnitude Dinosaur Tb Patterns and Behavioral Thermo- 0.5664 2 increased with mass (Tb ϭ 0.0775Mass , r regulation. Despite fluctuating environmen- DINOSAUR BODY TEMPERATURES 111

FIGURE 3. Theory predicted field data very well. Rep- FIGURE 4. Paleoclimate estimates. Variation in mean resentative examples of measured (solid circles) and daily air temperature in winter and summer during the mathematically predicted (open circles) Tb are shown different time periods of the Mesozoic. for a 32-kg crocodile in summer (top panel) and a 1010- kg crocodile in winter (middle panel). Overall, there were no significant differences between measured and predicted daily Tb amplitudes (bottom panel). Redrawn The Tb of all dinosaurs could have exceeded from data of Seebacher et al. (1999). 30ЊC in summer, and avoiding exposure to so- lar radiation rather than seeking heat would tal temperatures, most dinosaurs could have likely have been the thermoregulatory objec- had average daily Tb in winter of 30ЊCor tive for large dinosaurs at low latitudes in greater at any time in the Mesozoic, even if summer (Fig. 6). A 5000-kg dinosaur (e.g., Tri- they were crocodile-like ectotherms that ther- ceratops, Parasaurolophus,orMuttaburrasaurus), moregulated by shuttling between sun and the same mass as a large elephant, living at 0– shade (Fig. 5). 10Њ latitude in the Early Cretaceous would In addition to being warm, most dinosaurs have had to thermoregulate behaviorally to at low latitudes, and many at higher latitudes, minimize irradiation at all times of the year in could also have had stable Tb (Fig. 5). Daily Tb order to have Tb at or below 30ЊC. This may fluctuations in ectotherms depend upon am- have been achieved by seeking shade during bient thermal conditions, expressed as opera- the day, and maybe by adopting a crepuscular tive environmental temperatures (Bakken and habit. However, at midlatitudes (30–50Њ), the Gates 1976; Tracy 1982) to which the animal is 5000-kg dinosaur would have had to seek the exposed. The range of operative temperatures sun for at least part of the day even in summer, becomes greater with increased exposure to and it would have needed extended basking solar radiation, and it increases with increas- periods in winter at those latitudes to main- ing animal mass (Seebacher et al. 1999). None- tain a seasonally constant Tb (Fig. 6). For com- theless, the analysis showed that winter Tb of parison with data shown in Figure 6, note that medium sized, ectothermic dinosaurs (2000– the sun exposure of a modern, thermoregu- 3000 kg) that thermoregulated behaviorally by lating heliothermic reptile, such as varanid liz- moving in and out of the sun would have been ards (Seebacher and Grigg 2001) or crocodiles high (Ͼ30ЊC) and stable (Ϯ2ЊC or less) during (Seebacher and Grigg 1997; Grigg et al. 1998), any geological time period at latitudes up to would be less than half of the total available 50–55Њ (Fig. 5). during a day. The transient nature of heat 112 FRANK SEEBACHER

FIGURE 5. Body temperature of dinosaurs (703 from 321 genera) in winter during different time periods.

Most dinosaurs were warm (average Tb Ͼ 30ЊC, red circles), but average daily Tb decreased with decreasing mass and increasing paleolatitude (yellow circles ϭ 25–30ЊC, green circles ϭ 20–25ЊC, blue circles ϭϽ20ЊC). Lines in- dicate the combination of mass and paleolatitude at which daily Tb would have fluctuated with an amplitude of 2ЊC or less at different average Tb (colors as above). Note that dinosaurs exposed less to solar radiation would have been cooler, but their Tb would have been more stable. DINOSAUR BODY TEMPERATURES 113 transfer between an animal and its environ- ment means that Tb is determined by the in- tegrated environmental conditions over a giv- en period of time, which increases with mass (Fig. 1) and which can be recognized by the lag in response of Tb relative to operative tem- perature oscillation. Hence, saying that a 5000- kg dinosaur has to avoid the sun means that it has to reduce radiative heat load during a whole solar day. Selective advantages gained from high, en- dothermic metabolic rates would have been reduced in dinosaurs that could have had high and stable Tb without having to incur the high energetic cost of producing heat metabolical- ly. The principal advantage of high and stable

Tb lies in maintaining constant biochemical FIGURE 6. Predicted seasonal behavior (expressed as % and physiological reaction rates near their op- sun exposure) of a 5000-kg dinosaur living at different latitudes. To maintain high and stable T (amplitude ϭ timum. Although an ectothermic physiology b 2ЊC or less, average Tb ϭ 30ЊC) between seasons the di- does not allow the same sustained levels of ac- nosaur would have had to change its thermoregulatory tivity typical of many endotherms, the differ- sun and shade-seeking behavior (0 ϭ full shade, 1 ϭ full ence in physiological performance between sun) between seasons, and behavior would differ be- tween animals living at different latitudes. warm ectotherms with stable Tb, such as many dinosaurs were (Fig. 5), and typical endo- therms diminishes. In contrast, smaller ani- and mammals may not have been feasible for mals at higher latitudes would have had var- most, particularly larger, dinosaurs. The long iable and lower Tb (Fig. 5), so high metabolic necks of large dinosaurs, particularly sauro- heat production may have been of consider- pods, would have required the left ventricle to able advantage; if endothermy evolved among be impossibly large in order to maintain the Dinosauria, it most likely did in these lat- pressures in the head typical of modern en- ter species (see Fig. 5, and below). The conclu- dotherms (Seymour and Lillywhite 2000). In sion that large ectotherms (see Fig. 5 for addition, the extremely long trachea of sau- mass/latitude distributions) could have had ropods would have precluded lung ventilation stable Tb is in agreement with previous theo- rates necessary to sustain endothermic meta- retical case studies (Spotila et al. 1973; Dun- bolic rates (Daniels and Pratt 1992). These re- ham et al. 1989; O’Connor and Dodson 1999). sults are consistent with the present, biophys- However, the crocodile data and the analysis ically based analysis. There is some evidence above clearly show that it would be a mistake that ventilation rates and respiration of even to equate stable Tb with endothermy (Barrick small theropods and early birds may have and Showers 1994). The present analysis com- been inadequate to sustain high, endothermic plements previous work in explicitly showing metabolic rates. Lungs of theropods and early the extent to which dinosaurs with an ecto- enantiornithine birds were simple bellows- thermic metabolic make-up could have had like septate lungs with limited ventilation warm and stable Tb with respect to body mass rates (Ruben et al. 1997), although it appears and paleolatitude (Fig. 5). It is clear that en- that one theropod dinosaur had a more so- dothermy need not be evoked in order to ex- phisticated ventilation apparatus that allowed plain apparently stable temperatures in me- ventilation rates similar to those of mammals dium-sized dinosaurs living at midlatitudes (Ruben et al. 1999). In addition, dinosaurs ap- (Barrick et al. 1998). parently did not possess respiratory turbi- Different lines of evidence indicate that en- nates (Ruben et al. 1996) which are considered dothermy as it is known from modern birds a prerequisite for endothermic metabolic rates 114 FRANK SEEBACHER

sities may therefore strengthen the argument made above that selection pressures did not favor high metabolic rates for most dinosaurs, at least for those that could have had high and

stable Tb with an ectothermic metabolic make- up (Fig. 5). Ontogeny. The above predictions were made for adult dinosaurs but, as is the case in modern animals, thermal relations would have changed ontogenetically, particularly in species with large adult body sizes. The croc- odile field data shown above provide a good

example of such ontogenetic changes in Tb. Growth rates of dinosaurs are the subject of some debate (Reid 1997b), but recently it has FIGURE 7. Average daily Tb (thick solid line) and Tb been suggested that dinosaurs displayed a ranges (broken lines) of Maiasaura peeblesorum in winter at a latitude of 45Њ in the Late Cretaceous. The growth sigmoidal growth pattern, with rapid growth curve (mass; thin solid line) was based on data of Horner after hatching and slowed growth rates later et al. (2000), who suggested that juvenile M. peeblesorum in life (Padian et al. 2001; Erickson et al. 2001). grew rapidly, reaching late juvenile size (3.5 m, 330.7 kg) within one to two years, and adult body size (7–9 m, This would indicate that growth patterns of 2090.3–4078.8 kg) within six to eight years. If hatching dinosaurs were identical to those observed in occurred in spring, juveniles would have been six to sev- wild American alligators (A. mississippiensis) en months old (232.6 kg) before experiencing their first winter. (Rootes et al. 1991; Elsey et al. 1992, 2001)—a fact not appreciated by Padian et al. (2001). In a particular example based on an ontogenetic (Hillenius 1992) because they dramatically re- series of skeletal elements, Horner et al. (2000) duce respiratory water loss; evidence for this suggested that the hadrosaur Maiasaura peeble- (Ruben et al. 1996, 1997) is based on very few sorum would have reached its adult size of 7–9 specimens, so caution should be exercised in m in six to eight years. The nesting period of applying the conclusions to dinosaurs in gen- M. peeblesorum may have been as short as one eral. If it were true, however, that even small to two months, and the animals would have dinosaurs were unable to sustain endothermic reached a late juvenile size of 3.5 m after one metabolic rates, it would indicate that the to two years (Horner et al. 2000). Assuming ‘‘cool’’ dinosaurs in Figure 5 either underwent that offspring hatched in spring, like modern a period of dormancy during winter (as seen crocodilians for example (Elsey et al. 2001; in many modern reptiles) or displayed re- Webb and Manolis 1989), the juveniles would markable physiological acclimatization to be at least six to seven months old before their function normally at low Tb. Seasonal meta- first winter. Using Horner et al.’s (2000) bolic acclimatization is well known to occur in growth estimate, their body mass (Seebacher fish (e.g., St.-Pierre et al. 1998; Crawford et al. 2001) would have been 120–150 kg at the be- 1999), but little is known from reptiles. ginning of winter and around 200 kg at the Dinosaurs grew much larger than mam- end of winter. Hence, in a Late Cretaceous cli- mals, and large dinosaurs maintained rela- mate at 45Њ latitude, the Tb of juvenile M. pee- tively high population densities compared blesorum would have varied between 15.8ЊC with real or projected population densities of and 35.2ЊC at the beginning of their first win- similar-sized mammals (Farlow 1993; Farlow ter, and between 18.0ЊC and 34.4ЊC at the end et al. 1995). This has been taken to indicate of winter (Fig. 7). By modern reptilian stan- that dinosaurs may have had lower food con- dards, these Tb ranges are very typical (See- sumption rates, i.e., lower metabolic rates than bacher and Grigg 1997, 2001; Grigg et al. 1998; mammals (Farlow et al. 1995). Evidence from Grigg and Seebacher, 1999). Moreover, Tb body size distributions and population den- would have stabilized very quickly as the an- DINOSAUR BODY TEMPERATURES 115

imals grew older, so that average daily Tb of M. peeblesorum at 3 years of age would have been 30ЊC(Ϯ2.7ЊC) even in winter, and winter daily Tb of a 10 year old animal would have varied between 29.0ЊC and 31.0ЊC (Fig. 7). Note that in these predictions, as above, it was assumed that the animal thermoregulated be- haviorally to attain average Tb of 30ЊC when- ever possible. The above example demonstrates that met- abolic heat production is not a pre-requisite for juveniles of large dinosaurs to have Tb within a biologically functional range. It is de- batable, however, whether dinosaurs with an ectothermic physiology could have achieved growth rates similar to those suggested for M. peeblesorum. Using crocodilians as compari- son, Padian et al. (2001) argued that dinosaurs FIGURE 8. Average daily Tb increases with resting met- could not have achieved such fast growth rates abolic rate (averages for modern reptiles and mammals are indicated), but the increase in Tb would be far greater as simple ectotherms. This conclusion must be in animals possessing thermal insulation in the form of treated with caution, however, because croc- fur or feathers, for example (A). Daily Tb amplitudes de- odilians can grow fast enough to form fibro- crease with the thickness and the quality of insulation (thermal diffusivity, ␣, range of 0.035–0.05 ϫ 10Ϫ7 m2 lamellar bone (Reid 1997b), and Ruben (1995) sϪ1), and the biological relevance of these patterns can demonstrated that growth rates of the small be observed in modern birds and mammals that change theropod Troodon were identical to those of al- their fur or feather thickness between seasons and/or latitudes (B). Average daily Tb in winter of a small (3.8 ligators. Moreover, the presence of lines of ar- kg) dinosaur, such as Sinosauropteryx, which lived at 50Њ rested growth, which has been taken to indi- latitude in the Early Cretaceous and had resting meta- bolic rates intermediate between those of modern ecto- cate ectothermic metabolic rates resulting in Ϫ1 Ϫ1 therms and endotherms (0.5 ml O2 g h ). Tb of an an- pronounced seasonal fluctuations in Tb,donot imal with insulation (of 20 mm thickness) would be provide conclusive evidence of metabolic sta- much greater and would fluctuate far less than that of tus; a wide variety of endothermic and ecto- an uninsulated dinosaur. Operative temperatures are also shown. thermic animals including birds, dinosaurs, and crocodilians (Reid 1990; Chinsamy et al. 1994; Farlow et al. 1995; Rasch et al. 2001) verse coelurosaurs living at 45–50Њ paleolati- show lines of arrested growth, and the stable tude in the Early Cretaceous (Chen et al. 1998;

Tb of large ectothermic animals (Fig. 5) may Xu et al. 1999a,b) may represent an example preclude formation of lines of arrested of such an adaptation. growth. Insulation dampens out responses to daily

Insulation and Small Dinosaurs. The Tb of fluctuations in operative temperature and small dinosaurs (Ͻ100 kg) and Avialae (e.g., traps metabolically produced heat (Fig. 8).

Hypsilophodon [Galton 1974], Sinosauropteryx The difference in average daily Tb between an [Chen et al. 1998], Beipiaosaurus [Xu et al. insulated and an uninsulated animal increas- 1999b], Mononykus [Perle et al. 1993]) that es with increasing metabolic heat production, lived at mid latitudes (45–55Њ) or higher would so average daily Tb of an insulated animal havebeenwellbelow30ЊC in winter if they with mammal-like metabolic rates is nearly had been crocodile-like ectotherms (Fig. 5). It twice as high as that of an identical animal may be in this category that selection pres- without insulation (Fig. 8A) (all predictions sures for morphological and/or physiological were made for a 3.8-kg animal living at 50Њ lat- thermoregulatory adaptations would have itude in winter in the Early Cretaceous, and been strongest. The recent discoveries of in- insulation thickness ϭ 20 mm; see below). tegumentary structures on taxonomically di- Moreover, quality and thickness of insulation 116 FRANK SEEBACHER

affect Tb fluctuations considerably (Fig. 8B). fore have been present already or must have For example, an increase in insulation thick- coevolved with elevated metabolic rates. ness from 10% of the body radius to 20% of Incidence of Endothermy and Ectothermy among body radius decreases daily Tb fluctuations by Dinosaurs. If one accepts ectothermy as the nearly one-half. The biological relevance of ancestral condition among archosaurs, endo- these changes in heat transfer is demonstrated thermy must have evolved sometime between in the seasonal changes of fur thickness seen the early Late Triassic, when dinosaurs first ap- in many mammals. peared in the fossil record (Sereno 1999), and

The Tb of Sinosauropteryx (body mass 3.8 the evolution of modern birds whose ancestors kg), for example, living at 50Њ latitude in the probably appeared first in the early Late Juras- Early Cretaceous would have fluctuated daily sic (Dodson 2000; Sumida and Brochu 2000). with an amplitude of 8.1ЊC; average Tb would Moreover, as outlined in the introduction, it is have been 23.8ЊC in winter (Fig. 8C) if the an- conceivable that an evolutionary trend toward imal had no insulation and a resting metabolic endothermy was of selective advantage not rate of two to three times that of a varanid liz- only once but repeatedly, when lineages radi- Ϫ1 Ϫ1 ard (0.5 ml O2 g h [Thompson and Withers ated into new environments, for example. From 1997]). If the integumentary structures (20 the analysis above, it is possible to speculate on mm thickness [Chen et al. 1998]) provided the relative likelihood of an evolutionary trend featherlike insulation, however, winter daily towards endothermy among dinosaurian line-

Tb fluctuations would have been reduced to an ages (Fig. 9). Note, however, that this attempt amplitude of 3.1ЊC, and average Tb would at determining the phylogenetic distribution of have increased to 32.4ЊC (Fig. 7B), assuming endothermy is not meant to represent a final that the thermal diffusivity of the integumen- conclusion; rather, it was designed as a work- tary structures was similar to that of feathers ing hypothesis that will change as new insights of modern birds (0.05. 10Ϫ7 m2 sϪ1 [Bowman et from fossils are gained. al. 1978; Incropera and DeWitt 1996]). Hence, Because the strength of selection pressures even with resting metabolic rates intermediate favoring endothermy decreases with increas- between that of typical reptiles and mammals ing body mass, it is therefore contradictory (Reid 1997a; O’Connor and Dodson 1999), in- that endothermy coevolved with large body sulation has a striking effect on the magnitude size. Thus endothermy is very unlikely to have and stability of Tb. evolved in lineages in which the most recent

Selection for high and stable Tb must have species are predominantly large, such as pro- been a necessary corollary in the evolution of sauropods, sauropods, stegosaurs, nodosaurs, high levels of activity or parental care. This ankylosaurs, and ceratopsids. means that, given the physiological and phys- On the other hand, many coelurosaurs were ical relationships shown in Figure 8, the selec- comparatively small, reaching their greatest tive advantages conferred by elevated meta- diversity in the Late Cretaceous. From this bolic rates on small dinosaurs would have perspective, if endothermy did evolve among been minimal in the absence of insulation; in- dinosaurs, it most likely evolved within this sulation such as may have been afforded by group. Note, however, that this is not to say feathers or featherlike structures must there- that endothermy was obligatory for small gen-

FIGURE 9. Speculated phylogenetic distribution of endothermy among the Dinosauria (cladogram redrawn from Sereno 1999). Endothermy must have evolved sometime in the lineage leading to modern birds (Ornithurae, very dark shading) and is likely to have occurred in coelurosaurs, among which the evolutionary trend was a decrease in body size, and which lived at mid to high paleolatitudes (dark shading). It is less likely that endothermy evolved among other theropods that showed an evolutionary trend toward large body size (light gray shading), or among any other group of dinosaurs in which the most recent members attained large body size (very light gray shading). Hypsilophodontids and heterodontosaurids remained small and occurred at mid to high latitudes, so endothermy may have been of selective advantage in those dinosaurs (gray shading). DINOSAUR BODY TEMPERATURES 117 118 FRANK SEEBACHER era; it was simply feasible. Similarly, among bly have greatly reduced the selective advan- the Ornithopoda, the Heterodontosauridae tage of the latter in uninsulated animals. Most and Hysilophodontidae remained small (Ͻ100 skin impressions from dinosaurs indicate the kg [Brett-Surman 1997; Seebacher 2001]) and presence of naked skin (Czerkas 1997; Sumida relatively undiversified throughout their evo- and Brochu 2000), except for integumentary lutionary history. Moreover, hypsilophodon- structures in coelurosaurs that may have af- tids had a cosmopolitan distribution that forded thermal insulation (Chen et al. 1998). reached into the high latitudes of Gondwana Although other dinosaurs may have possessed (Rich 1996). Hence, at least for hypsilophodon- integumentary structures with insulatory tids living at mid to high latitude (Ͼ45Њ; see qualities, present knowledge indicates that Fig. 5), endothermy may have been of selec- these evolved in the Coelurosauria only (this tive advantage. The case for heterodontosaur- conclusion may change as new fossil evidence ids is more doubtful, as they were a very basal is discovered). The earliest known feathers group of ornithopod that are known primarily stem from the Late Jurassic, so if those feathers from low to mid latitudes in the Early Jurassic, possessed insulating qualities, endothermy and the Late Jurassic fossil finds are fragmen- may have evolved sometime after that. tary and uncertain (Weishampel 1990). Morphological transformation was most Literature Cited sustained in the evolution of the Hadrosauri- Azuma, Y., and P. J. Currie. 1995. A new giant dromaeosaurid from Japan. Journal of Vertebrate Paleontology 15:17A. dae within the Ornithopoda, and in the coe- Bakken, G. S., and D. M. Gates. 1976. Heat transfer analysis of lurosaurs within the Tetanurae (Sereno 1999). animals: some implications for field ecology, physiology, and The evolution of skeletal diversity may indi- evolution. Pp. 255–290 in D. M. Gates and R. B. Schmerl, eds. Perspectives of biophysical ecology. Springer, New York. cate rapidly changing selection pressures as a Bakker, R. T. 1972. Anatomical and ecological evidence of en- result of radiation into new ecological niches. dothermy in dinosaurs. Nature 238:81–85. It is more likely, therefore, that endothermy Bardossy, G. 1996. Carboniferous to Jurassic bauxite deposits as paleoclimate and paleogeographic indicators. Canadian So- evolved in lineages exposed to ‘‘new’’ selec- ciety of Petroleum Geologists Memoir 17:283–293. tion pressures rather than in very conservative Barrett, P. M., Y. Hailu, P. Upchurch, and A. C. Burton. 1998. A lineages, which changed little morphological- new ankylosaurian dinosaur (Ornithischia: Ankylosauria) from the Upper Cretaceous of Shanxi Province, People’s Republic of ly and, presumably, ecologically (Schluter China. Journal of Vertebrate Paleontology 18:376–384. 2001). Hadrosaurs are also distinguished in Barrick, R. E., and W. J. Showers. 1994. Thermophysiology of Ty- being the only other dinosaur lineage that rannosaurus rex: Evidence from oxygen isotopes. Science 265: 222–224. contained members reaching body masses Barrick, R. E., M. K. Stoskopf, J. D. Marcot, D. A. Russell, and comparable to sauropods, so although had- W. J. Showers. 1998. The thermoregulatory functions of the rosaurs reached great diversity, it is doubtful Triceratops frill and horns: heat flow measured with oxygen isotopes. Journal of Vertebrate Paleontology 18:746–750. that they evolved endothermy. Morphological Barron, E. J. 1983. A warm equable Cretaceous: the nature of the diversification among coelurosaurs, on the problem. Earth-Science Reviews 19:305–338. other hand, lends support to the above spec- ———. 1995. Tropical climate stability and implications for the distribution of life. Pp. 108–117 in S. M. Stanley, ed. Effects of ulation that endothermy may have evolved past global change on life. National Academy Press Studies in within that lineage. Geophysics, Washington, D.C. Hypsilophodonts, on the other hand, were Bennett, A. F., and J. A. Ruben. 1979. Endothermy and activity in vertebrates. Science 206:649–654. conservative lineages that changed little in Benton, M. J., E. Cook, D. Grigorescu, E. Popa, and E. Tallo´di. their morphology over geological time, which 1997. Dinosaurs and other tetrapods in an Early Cretaceous may indicate that selection pressures re- bauxite-filled fissure, northwestern Romania. Palaeogeogra- phy, Palaeoclimatology, Palaeoecology 130:275–292. mained relatively stable during their evolu- Benton, M. J., L. Juul, G. W. Storrs, and P. M. Galton. 2000. Anat- tion; endothermy is thus less likely to have omy and systematics of the prosauropod dinosaur Thecodon- evolved in this group than in coelurosaurs. tosaurus antiquus from the Upper Triassic of southwest Eng- land. Journal of Vertebrate Paleontology 20:77–108. The thermoregulatory advantages gained Bowman, H. F., E. G. Cravalho, and M. Woods. 1978. Theory, from elevated resting metabolic rates are measurement, and application of thermal properties of bio- doubtful in animals without insulation. The in- materials. Annual Review of Biophysics and Bioengineering 4:43–80. crease in energy demand associated with an in- Brett-Surman, M. K. 1997. Ornithopods. Pp. 330–346 in Farlow crease in resting metabolic rates would proba- and Brett-Surman 1997. DINOSAUR BODY TEMPERATURES 119

Buffetaut, E., V. Suteethorn, and H. Tong. 1996. The earliest released juveniles. Comparative Biochemistry and Physiology known tyrannosaur from the Lower Cretaceous of Thailand. 103A:667–672. Nature 381:689–691. Elsey, R. M., L. McNease, and T. Joanen. 2001. Louisiana’s alli- Calvo, J. O., and J. F. Bonaparte. 1991. Andesaurus delgadoi gen et gator ranching program: a review and analysis of releases of sp. nov. (Saurischia-Sauropoda), dinosaurio titanosauridae de captive-raised juveniles. Pp. 426–442 in G. C. Grigg, F. See- le formacion Rio Limay (Albiano-Cenomaniano), Neuque´n, bacher, and C. E. Franklin, eds. Crocodilian biology and evo- Argentina. Ameghiniana 28:303–310. lution. Surrey Beatty, Chipping Norton, Australia. Carpenter, K., C. Miles, and K. Cloward. 1998. Skull of a Jurassic Embry, A. F., B. Beauchamp, and D. J. Glass, eds. 1994. Pangea: ankylosaur (Dinosauria). Nature 393:782–783. global environments and resources. Canadian Society of Pe- Carslaw, H. S., and J. C. Jaeger. 1995. Conduction of heat in sol- troleum Geologists Memoir 17. ids. Clarendon, Oxford. Erickson, G. M., K. C. Rogers, and S. A. Yerby. 2001. Dinosaurian Chatterjee, S. 1993. Shuvosaurus, a new theropod. National Geo- growth patterns and rapid avian growth rates. Nature 412: graphic Research and Exploration 9:274–285. 429–433. Chen, P., Z. Dong, and S. Zhen. 1998. An exceptionally well-pre- Farlow, J. O. 1990. Dinosaur energetics and thermal biology. Pp. served theropod dinosaur from the Yixian Formation of Chi- 43–55 in D. B. Weishampel, P. Dodson, and H. Osmolska, eds. na. Nature 391:147–152. The Dinosauria. University of California Press, Berkeley. Chinsamy, A., L. M. Chiappe, and P. Dodson. 1994. Growth ———. 1993. On the rareness of big, fierce animals: specula- rings in Mesozoic birds. Nature 368:196–197. tions about the body sizes, population densities, and geo- Colbert, E. H., R. B. Cowles, and C. M. Bogert. 1946. Tempera- graphic ranges of predatory mammals and large carnivorous ture tolerances in the American alligator and their bearing on dinosaurs. Journal of Science 293A:167–199. the habits, evolution, and extinction of the dinosaurs. Bulletin Farlow, J. O., and M. K. Brett-Surman. 1997. The complete di- of the American Museum of Natural History 86:329–373. nosaur. Indiana University Press, Bloomington. Coria, R. A., and P. J. Currie. 1997. A new theropod from the Rio Farlow, J. O., P. Dodson, and A. Chinsamy. 1995. Dinosaur bi- Limay Formation. Journal of Vertebrate Paleontology 17:40A. ology. Annual Review of Ecology and Systematics 26:445– Coria, R. A., and L. Salgado. 1995. A new giant carnivorous di- 471. nosaur from the Cretaceous of Patagonia. Nature 377:224–226. Farmer, C. G. 2000. Parental care: the key to understanding en- ———. 1996. A basal iguanodontian (Ornithischia: Ornitho- dothermy and other convergent features in birds and mam- poda) from the Late Cretaceous of South America. Journal of mals. American Naturalist 155:326–334. Vertebrate Paleontology 16:445–457. Frakes, L. A., J.-L. Probst, and W. Ludwig. 1994. Latitudinal dis- Crawford, D. L., V. A. Pierce, and J. A. Segal. 1999. Evolutionary tribution of paleotemperature on land and sea from early Cre- physiology of closely related taxa: analyses of enzyme ex- taceous to middle Miocene. Comptes Rendus de l’Acade´mie pression. American Zoologist 39:389–400. des Sciences de Paris, se´rie II, Ge´osciences de Surface 318: Currie, P. J., and X. Zhao. 1993. A new carnosaur (Dinosauria, 1209–1218. Theropoda) from the Jurassic of Xinjiang, People’s Republic of Francis, J. E. 1994. Palaeoclimate of Pangea: geological evidence. China. Canadian Journal of Earth Sciences 30:2037–2081. Pp. 265–274 in Embry et al. 1994. Currie, P. J., S. J. Godfrey, and L. Nessov. 1993. New caenag- Gabunia, L. K., G. Mchedlidze, V. M. Chkhikvadze, and S. G. nathid (Dinosauria: Theropoda) specimens from the Upper Lucas. 1998. Jurassic sauropod dinosaur from the Republic of Cretaceous of North America and Asia. Canadian Journal of Georgia. Journal of Vertebrate Paleontology 18:233–236. Earth Sciences 30:2255–2272. Galton, P. M. 1974. The ornithischian dinosaur Hypsilophodon Czerkas, S. A. 1997. Skin. Pp. 669–675 in P. J. Currie and K. Pa- from the Wealden of the Isle of Wight. Bulletin of the British dian, eds. The encyclopedia of dinosaurs. Academic Press, Museum (Natural History), Geology 25:1–152. San Diego. Gasparini, Z., X. Pereda-Suberbiola, and R. E. Molnar. 1996. Dal Sasso, C., and M. Signore. 1998. Exceptional soft-tissue New data on the ankylosaurian dinosaur from the Late Cre- preservation in a theropod dinosaur from Italy. Nature 392: taceous of the Antarctic peninsula. Memoirs of the Queens- 383–387. land Museum 39:583–594. Daniels, C. B., and J. Pratt. 1992. Breathing in long necked di- Gillette, D. D. 1991. Seismosaurus halli, gen. et sp. nov., a new sau- nosaurs: did the sauropods have bird lungs? Comparative ropod dinosaur from the Morrison formation (Upper Juras- Biochemistry and Physiology 101A:43–46. sic/Lower Cretaceous) of New Mexico, USA. Journal of Ver- Dodson, P. 2000. Origin of birds: the final solution? American tebrate Paleontology 11:417–433. Zoologist 40:504–512. Godefroit, P., Z. Dong, P.Bultynck, H. Li, and L. Feng. 1998. Cre- Dong, Z. 1993. A new species of stegosaur (Dinosauria) from the taceous dinosaurs and mammals from Inner Mongolia 1. New Ordos Basin, Inner Mongolia, People’s Republic of China. Ca- Bactrosaurus (Dinosauria: Hadrosauridae) material from Iren nadian Journal of Earth Sciences 30:2174–2176. Dabasu (Inner Mongolia, P. R. China). Bulletin de l’Institut Dunham, A. E., K. L. Overall, W. P. Porter, and C. A. Forster. Royal des Sciences Naturelles de Belgique (Sciences de la Ter- 1989. Implications of ecological energetics and biophysical re) 68(Suppl):3–70. and developmental constraints for life-history variation in di- Golonka, J., M. I. Ross, and C. R. Scotese. 1994. Phanerozoic, nosaurs. Pp. 1–19 in J. O. Farlow, ed. Paleobiology of the di- paleogeographic and paleoclimatic modeling maps. Pp. 1–47 nosaurs. Geological Society of America Special Paper 238. in Embry et al. 1994. Boulder, Colorado. Graham, S. A., M. C. Hendrix, R. Barsbold, D. Badamgarav, D. Else, P. L., and A. J. Hulbert. 1981. Comparison of the ‘‘mammal Sjostrom, W. Kirschner, and J. S. McIntosh. 1997. Stratigraphic machine’’ and the ‘‘reptile machine’’: energy production. occurrence, paleoenvironment, and description of the oldest American Journal of Physiology 240:R3–R9. known dinosaur (Late Jurassic) from Mongolia. Palaios 12: ———. 1987. Evolution of mammalian endothermic metabo- 292–297. lism: ‘‘leaky’’ membranes as a source of heat. American Jour- Grigg, G. C., and F. Seebacher. 1999. Field test of a paradigm: nal of Physiology 253:R1–R7. hysteresis of heart rate in thermoregulation by a free-ranging Elsey, R. M., T. Joanen, L. McNease, and N. Kinler. 1992. Growth lizard (Pogona barbata). Proceedings of the Royal Society of rates and body condition factors of Alligator mississippiensis in London B 266:1291–1297. coastal Louisiana wetlands: a comparison of wild and farm- Grigg, G. C., F. Seebacher, L. A. Beard, and D. Morris. 1998. 120 FRANK SEEBACHER

Thermal relations of large crocodiles, Crocodylus porosus, free- Moore, G. T., and C. A. Ross. 1994. Kimmeridgian-Tithonian ranging in a naturalistic situation. Proceedings of the Royal (Late Jurassic) dinosaur and ammonoid paleoecology from a Society of London B 265:1793–1799. paleoclimate simulation. Pp. 345–361 in Embry et al. 1994. Halliday, D., and R. Resnick. 1978. Physics. Wiley, New York. Norell, M. A., and J. M. Clark. 1992. New dromaeosaur material Hammer, W. R., and W. J. Hickerson. 1994. A crested theropod from the Late Cretaceous of Mongolia. Journal of Vertebrate dinosaur from Antarctica. Science 264:828–830. Paleontology 12:45A. Hayes, J. P., and T. Garland Jr. 1995. The evolution of endother- Novas, F. E. 1997. Anatomy of Patagonykus puertai (Theropoda, my: testing the aerobic capacity model. Evolution 49:836–847. Avialae, Alvarezsauridae), from the Late Cretaceous of Pata- Heckert, A. B., S. G. Lucas, and A. P. Hunt. 1994. A late Carnian gonia. Journal of Vertebrate Paleontology 17:137–166. theropod from New Mexico: implications for the early evolution ———. 1998. Megaraptor namunhuaiqii, gen. et sp. nov., a large of Theropoda. Journal of Vertebrate Paleontology 14:28A. clawed, Late Cretaceous theropod from Patagonia. Journal of Herman, A. B., and R. A. Spicer. 1996. Palaeobotanical evidence Vertebrate Paleontology 18:4–9. for a warm Cretaceous Arctic Ocean. Nature 380:330–333. Novas, F. E., and P. F. Puerta. 1997. New evidence concerning Hillenius, W. J. 1992. The evolution of nasal turbinates and avian origins from the Late Cretaceous of Patagonia. Nature mammalian endothermy. Paleobiology 18:17–29. 387:390–392. Hilton, R. P., F. L. Decourten, M. A. Murphy, P. U. Rodda, and P. O’Connor, M. P., and P. Dodson. 1999. Biophysical constraints on G. Embree. 1997. An Early Cretaceous ornithopod dinosaur the thermal ecology of dinosaurs. Paleobiology 25:341–368. from California. Journal of Vertebrate Paleontology 17:557–560. O’Connor, M. P., and J. R. Spotila. 1992. Consider a spherical Horner, J. R., A. de Ricqle`s, and K. Padian. 2000. Long bone his- animal: when do short cuts matter in biophysical models? tology of the hadrosaurid dinosaur Maiasaura peeblesorum: American Zoologist 32:179–193. growth dynamics and physiology based on an ontogenetic se- Padian, K., J. A. de Ricqle`s, and J. R. Horner. 2001. Dinosaur ries of skeletal elements. Journal of Vertebrate Physiology 20: growth rates and bird origins. Nature 412:405–408. 115–129. Paladino, F. V., M. P. O’Connor, and J. R. Spotila. 1990. Metab- Howse, S. C. B., and A. R. Milner. 1993. Ornithodesmus: a man- olism of leatherback turtles, gigantothermy, and thermoreg- iraptorian theropod dinosaur from the Lower Cretaceous of ulation of dinosaurs. Nature 344:858–860. the Isle of Wight, England. Palaeontology 36:425–437. Paladino, F. V., J. R. Spotila, and P. Dodson. 1997. A blueprint for Huber, B. T., and D. A. Hodell. 1996. Reply. Geological Society giants: modeling the physiology of large dinosaurs. Pp. 491– of America Bulletin 108:1192–1196. 504 in Farlow and Brett-Surman 1997. Huber, B. T., D. A. Hodell, and C. P.Hamilton. 1995. Middle-Late Parrish, J. M. 1997. Evolution of the archosaurs. Pp. 191–203 in Cretaceous climate of the southern high latitudes: stable iso- Farlow and Brett-Surman 1997. topic evidence for minimal equator-to-pole thermal gradients. Parrish, J. T. 1993. Climate of the supercontinent Pangea. Journal Geological Society of America Bulletin 107:1164–1191. of Geology 101:215–233. Incropera, F. P., and D. P. DeWitt. 1996. Fundamentals of heat Perez-Moreno, B. P., J. L. Sanz, A. D. Buscalioni, J. J. Moratalla, and mass transfer. Wiley, New York. F. Ortega, and D. Rasskin-Gutman. 1994. A unique multi- Jacobs, L. L., D. A. Winkler, W. R. Downs, and E. M. Gomani. toothed ornithomimosaur dinosaur from the Lower Creta- 1993. New material of an Early Cretaceous titanosaurid sau- ceous of Spain. Nature 370:363–367. ropod dinosaur from Malawi. Palaeontology 36:523–534. Perle, A., M. A. Norell, L. M. Chiappe, and J. M. Clark. 1993. Jain, S. L., and S. Bandyopadhyay. 1997. New titanosaurid (Di- Flightless bird from the Cretaceous of Mongolia. Nature 362: nosauria: Sauropoda) material from the Late Cretaceous of 623–626. Central India. Journal of Vertebrate Paleontology 17:114–136. Pough, F. H. 1980. The advantages of ectothermy for tetrapods. Johnson, C. C., E. J. Barron, E. G. Kauffman, M. A. Arthur, P. J. American Naturalist 115:92–112. Fawcett, and M. K. Yasuda. 1996. Middle Cretaceous reef col- Qiang, J., P. J. Currie, M. A. Norell, and J. Shu-An. 1998. Two lapse linked to ocean heat transport. Geology 24:376–380. feathered dinosaurs from northeastern China. Nature 393: Kellner, A. W. A., and D. de A. Campos. 1996. First early the- 753–761. ropod dinosaur from Brazil with comments on Spinosauridae. Rasch, R., A. D. Tucker, L. Daddow, and E. Wentrup-Byrne. Neues Jahrbuch fu¨ r Geologische und Pala¨ontologische, Ab- 2001. Electron microprobe investigation of growth marks in handlungen 199:151–166. crocodile osteoderms. Pp. 144–155 in G. C. Grigg, F. Seebach- Koteja, P. 2000. Energy assimilation, parental care and the evo- er, and C. E. Franklin, eds. Crocodilian biology and evolution. lution of endothermy. Proceedings of the Royal Society of Surrey Beatty, Chipping Norton, Australia. London B 267:479–484. Rauhut, O. W. M., and C. Werner. 1997. First record of a Maas- Kuypers, M. M. M., R. D. Pancost, and J. S. Sinninghe Damste´. trichtian sauropod dinosaur from Egypt. Paleontologica Af- 1999. A large and abrupt fall in atmospheric CO2 concentra- tion during Cretaceous times. Nature 399:342–345. ricana 34:63–67. Lovegrove, B. G., G. Heldmaier, and T. Ruf. 1987. Perspectives Reid, R. E. H. 1990. Zonal ‘‘growth rings’’ in dinosaurs. Modern of endothermy revisited: the endothermic temperature range. Geology 15:19–48. Journal of thermal Biology 15:185–197. ———. 1997a. Dinosaurian physiology: the case for ‘‘intermedi- Mayer, H., and E. Appel. 1999. Milankovitch cyclicity and rock- ate’’ dinosaurs. Pp. 449–473 in Farlow and Brett-Surman 1997. magnetic signatures of palaeoclimatic change in the Early ———. 1997b. How dinosaurs grew. Pp. 403–413 in Farlow and Cretaceous Biancone Formation of the southern Alps, Italy. Brett-Surman 1997. Cretaceous Research 20:189–214. Rich, T. H. 1996. Significance of polar dinosaurs in Gondwana. Milner, A. C., and J. J. Hooker. 1992. An ornithopod dinosaur Memoirs of the Queensland Museum 39:711–717. from the Upper Cretaceous of the Antarctic peninsula. Jour- Rich, T. H., and P. V. Rich. 1989. Polar dinosaurs and biotas of nal of Vertebrate Paleontology 12:44A. the early Cretaceous of southeastern Australia. National Geo- Mitchell, J. W. 1976. Heat transfer from spheres and other ani- graphic Research 5:15–53. mal forms. Biophysics Journal 16:561–569. Rootes, W. L., R. H. Chabreck, V. L. Wright, B. W. Brown, and T. Molnar, R. E., and J. Wiffen. 1994. A Late Cretaceous polar di- J. Hess. 1991. Growth rates of American alligators in estuarine nosaur fauna from New Zealand. Cretaceous Research 15: and palustrine wetlands in Louisiana. Estuaries 14:489–494. 689–706. Ruben, J. A. 1995. The evolution of endothermy in mammals and DINOSAUR BODY TEMPERATURES 121

birds: from physiology to fossils. Annual Review of Physiol- Sloan, L. C., and E. J. Barron. 1990. ‘‘Equable’’ climates during ogy 57:69–95. Earth history? Geology 18:489–492. Ruben, J. A., W. J. Hillenius, N. R. Geist, A. Leitch, T. D. Jones, P. Smith, A. G., D. G. Smith, and B. M. Funnell. 1994. Atlas of Me- J. Currie, J. R. Horner, and G. Espe III. 1996. The metabolic sta- sozoic and Cenozoic coastlines. Cambridge University Press, tus of some late Cretaceous dinosaurs. Science 273:1204–1207. Cambridge. Ruben, J. A., T. D. Jones, N. R. Geist, and W. J. Hillenius. 1997. Spotila, J. R., P. W.Lommen, G. S. Bakken, and D. M. Gates. 1973. Lung structure and ventilation in theropod dinosaurs and A mathematical model for body temperatures of large rep- early birds. Science 278:1267–1270. tiles: implications for dinosaur ecology. American Naturalist Ruben, J. A., C. Dal Sasso, N. R. Geist, W.J. Hillenius, T. D. Jones, 107:391–404. and M. Signore. 1999. Pulmonary function and metabolic St. Pierre, J., P.-M. Charest, and H. Guderley. 1998. Relative con- physiology of theropod dinosaurs. Science 283:514–516. tribution of quantitative and qualitative changes in mitochon- Russell, D. A., and Z. Dong. 1993a. The affinities of a new the- dria to metabolic compensation during seasonal acclimati- ropod from the Alxa , Inner Mongolia, People’s Repub- zation of rainbow trout Oncorhynchus mykiss. Journal of Ex- lic of China. Canadian Journal of Earth Sciences 30:2107–2127. perimental Biology 201:2961–2970. ———. 1993b. A nearly complete skeleton of a new troodontid Sues, H. 1997. On Chirostenotes, a Late Cretaceous oviraptoro- dinosaur from the Early Cretaceous of the Ordon Basin, Inner saur (Dinosauria: Theropoda) from western North America. Mongolia, People’s Republic of China. Canadian Journal of Journal of Vertebrate Paleontology 17:698–716. Earth Sciences 30:2163–2173. Sumida, S. S., and C. A. Brochu. 2000. Phylogenetic context for Russell, D. A., and Z. Zheng. 1993. A large mamenchisaurid the origin of feathers. American Zoologist 40:486–503. from the Junggar Basin, Xinjiang, People’s Republic of China. Tarduno, J. A., D. B. Brinkman, P. R. Renne, R. D. Cottrell, H. Canadian Journal of Earth Sciences 30:2082–2095. Scher, and P. Castillo. 1998. Evidence for extreme climatic Sampson, S. D., D. W. Krause, C. A. Forster, and P.Dodson. 1996. warmth from Late Cretaceous arctic vertebrates. Science 282: Non-avian theropod dinosaurs from the Late Cretaceous of 2241–2200. Madagascar and their paleobiogeographic implications. Jour- Thompson, G. G., and P. C. Withers. 1997. Standard and maxi- nal of Vertebrate Paleontology 16:62A. mal metabolic rates of goannas (Squamata: Varanidae). Phys- Schluter, D. 2001. Ecology and the origin of species. Trends in Ecology and Evolution 16:372–380. iological Zoology 70:307–323. Seebacher, F. 1999. Behavioural postures and the rate of body Thulborn, R. A. 1994. Ornithopod dinosaur tracks from the temperature change in wild freshwater crocodiles, Crocodylus Lower Jurassic of Queensland. Alcheringa 18:247–258. johnstoni. Physiological and Biochemical Zoology 72:57–63. Tracy, C. R. 1982. Biophysical modeling in reptilian physiology ———. 2001. A new method to calculate allometric length-mass and ecology. Pp. 275–321 in C. Gans and F. H. Pough, eds. Bi- relationships of dinosaurs. Journal of Vertebrate Paleontology ology of the Reptilia, Vol. 12. Academic Press, New York. 21:51–60. Turner, J. S. 1987. On the transient temperatures of ectotherms. Seebacher, F., and G. C. Grigg. 1997. Patterns of body temperature Journal of thermal Biology 12:207–214. in wild freshwater crocodiles, Crocodylus johnstoni: thermoreg- Tykoski, R. S. 1997. A new ceratosaurid theropod from the Early ulation versus thermoconformity, seasonal acclimatization and Jurassic Kayenta formation of northern Arizona. Journal of the effect of social interactions. Copeia 1997:549–557. Vertebrate Paleontology 17:81A. ———. 2001. Changes in heart rate are important for thermo- Vakhrameev, V. A. 1991. Jurassic and Cretaceous floras and cli- regulation in the varanid lizard Varanus varius. Journal of mates of the earth. Cambridge University Press, Cambridge. Comparative Physiology B 171:395–400. Webb, G., and C. Manolis. 1989. Crocodiles of Australia. Reed Seebacher, F., G. C. Grigg, and L. A. Beard. 1999. Crocodiles as Books, Frenchs Forest, N.S.W., Australia. dinosaurs: behavioural thermoregulation in very large ecto- Weishampel, D. B. 1990. Dinosaur distributions. Pp. 63–139 in therms leads to high and stable body temperatures. Journal of D. B. Weishampel, P. Dodson, and H. Osmolska, eds. The Di- Experimental Biology 102:77–86. nosauria. University of California Press, Berkeley. Sellwood, B. W., G. D. Price, and P. J. Valdes. 1994. Cooler esti- Williams, T. M. 1990. Heat transfer in elephants: thermal par- mates of Cretaceous temperatures. Nature 370:453–455. titioning based on skin temperature profiles. Journal of Zo- Sereno, P. C. 1999. The evolution of dinosaurs. Science 284:2137– ology 222:235–245. 2147. Winkler, D. A., P. A. Murry, and L. L. Jacobs. 1997. A new species Sereno, P. C., and F. E. Novas. 1992. The complete skull and skel- of Tenontosaurus (Dinosauria: Ornithopoda) from the early eton of an early dinosaur. Science 258:1137–1140. Cretaceous of Texas. Journal of Vertebrate Paleontology 17: Sereno, P. C., C. A. Forster, R. R. Rogers, and A. M. Monetta. 330–348. 1993. Primitive dinosaur skeleton from Argentina and the Wright, P. G. 1984. Why do elephants flap their ears? South Af- early evolution of Dinosauria. Nature 361:64–66. rican Journal of Zoology 19:266–269. Sereno, P. C., J. A. Wilson, H. C. E. Larsson, D. B. Dutheil, and Xu, X., Z. Tang, and X. Wang. 1999a. A therizinosauroid dino- H. Sues. 1994. Early Cretaceous dinosaurs from the Sahara. saur with integumentary structures from China. Nature 399: Science 266:267–271. 350–354. Sereno, P. C., D. B. Dutheil, M. Iarochene, H. C. E. Larsson, G. H. Lyon, P. M. Magwene, C. A. Sidor, D. J. Varricchio, and J. Xu, X., X. Wang, and X. Wu. 1999b. A dromaeosaurid dinosaur A. Wilson. 1996. Predatory dinosaurs from the Sahara and with a filamentous integument from the Yixian formation of Late Cretaceous faunal differentiation. Science 272:986–991. China. Nature 401:262–266. Sereno, P. C., A. L. Beck, D. B. Dutheil, B. Gado, H. C. E. Larsson, Yemane, K. 1993. Contribution of Late Permian palaeogeogra- G. H. Lyon, J. D. Marcot, O. W. M. Rauhut, R. W. Sadleir, C. A. phy in maintaining a temperate climate in Gondwana. Nature Sidor, D. D. Varricchio, G. P. Wilson, and J. A. Wilson. 1998. A 361:51–54. long-snouted dinosaur from Africa and the evolution of spi- Zhao, X., and P. J. Currie. 1993. A large crested theropod from nosaurids. Science 282:1298–1302. the Jurassic of Xinjiang, People’s Republic of China. Canadian Seymour, R. S., and H. B. Lillywhite. 2000. Hearts, neck posture Journal of Earth Sciences 30:2027–2036. and metabolic intensity of sauropod dinosaurs. Proceedings Zhao, X., and X. Xu. 1998. The oldest known coelurosaurian. Sci- of the Royal Society of London B 267:1883–1887. ence 394:234–235. 122 FRANK SEEBACHER

Appendix Megaraptor namunhuaiquii Novas 1998 Monolophosaurus jiangi Zhao and Currie 1993 Dinosaur taxa used in the analysis in addition to those listed Mononychus olecranus Perle et al. 1993 by Weishampel (1990). Nodosauridae sp. Gasparini et al. 1996 Ornithodesmus cluniculus Howse and Milner 1993 Species Reference Ornithopoda sp. Milner and Hooker 1992 Ornithopoda sp. Thulborn 1994 Afrovenator abakensis Sereno et al. 1994 Patagonykus puertai Novas 1997 Alxasaurus elesitaiensis Russell and Dong 1993a Pelecanimimus polyodon Perez-Moreno et al. 1994 Andesaurus delgadoi Calvo and Bonaparte 1991 Protoarchaeopteryx robusta Qiang et al. 1998 Angaturama limai Kellner and Campos 1996 Rhoetosaurus brownei Rich 1996 Bactrosaurus johnsoni Godefroit et al. 1998 Sauropoda sp. Rauhut and Werner 1997 Beipiaosaurus inexpectus Xu et al. 1999b Sauropoda sp. Molnar and Wiffen 1994 Caenagnathus sp. Currie et al. 1993 Scipionyx samniticus Dal Sasso and Signore 1998 Carcharodontosaurus saharicus Sereno et al. 1996 Seismosaurus halli Gillette 1991 Caudipteryx zoui Qiang et al. 1998 Shanxia tianzhenensis Barrett et al. 1998 Ceratosauridae sp. Tykoski 1997 Shuvosaurus inexpectatus Chatterjee 1993 Chirostenotes pergracilis Sues 1997 Siamotyrannus isanensis Buffetaut et al. 1996 Coelurosauria sp. Zhao and Xu 1998 Sinornithoides youngi Russell and Dong 1993b Cryolophosaurus ellioti Hammer and Hickerson 1994 Sinornithosaurus millenii Xu et al. 1999a Deltadromeus agilis Sereno et al. 1996 Sinosauropteryx prima Chen et al. 1998 Diplodocidae sp. Gabunia et al. 1998 Sinraptor dongi Currie and Zhao 1993 Dromaeosauria sp. Azuma and Currie 1995 Suchomimus tenerensis Sereno et al. 1998 Eoraptor lunensis Sereno et al. 1993 Tenontosaurus dossi Winkler et al. 1997 Gargoyleosaurus parkpini Carpenter et al. 1998 Thecodontosaurus antiqus Benton et al. 2000 Gasparinisaura cincosaltensis Coria and Salgado 1996 Theropoda sp. Molnar and Wiffen 1994 Giganotosaurus carolinii Coria and Salgado 1995 Theropoda sp. Heckert et al. 1994 Herrerasaurus ischigualastensis Sereno and Novas 1992 Theropoda sp. Coria and Currie 1997 Hypsolophodontidae sp. Hilton et al. 1997 Theropoda sp. Benton et al. 1997 Leaellynasaura amicagraphica Rich and Rich 1989 Titanosaurus colberti Jain and Bandyopadhyay 1997 Majungasaurus crenatissimus Sampson et al. 1996 Unenlagia comahuensis Novas and Puerta 1997 Malawisaurus dixeyi Jacobs et al. 1993 Velociraptor mongoliensis Norell and Clark 1992 Mamenchisaurus sp. Russell and Zheng 1993 Wuerhosaurus ordosensis Dong 1993 Mamenchisaurus sp. Graham et al. 1997