the transition to a lower metabolic rate,increasing the time between breaths3.

A muscular ridge in the three-chamberedheart can completely separate oxygenatedfrom deoxygenated blood, so mere separa-tion of blood flow does not explain why afour-chambered heart is so beneficial that itevolved twice4. But in the four-chamberedheart, the division of blood flow also allowsthe pressure to vary between the pulmonaryand systemic circulatory systems. The sidethat sends blood to the lungs is much weaker

news and views

NATURE | VOL 434 | 14 APRIL 2005 | www.nature.com/nature 833

than the other side, because high pressures of blood in the lungs could cause fluid to leak across the respiratory membranes. Thestronger side of the heart sends blood to thesystemic arteries, where the higher pressureworks against higher resistance resultingfrom muscle contraction forces, multiplecapillary beds and the extensive filtrationacross the kidneys. The systemic blood pres-sures of modern endotherms are three timesthose of ectotherms.

The crocodilian heart is an interestingmosaic: it is four chambered but it also has ashunting system, although a quite differentone from that of other reptiles (Fig. 1c). Sey-mour et al.1 propose that the ancestors ofmodern alligators and crocodiles wereendothermic and required a four-cham-bered heart for pressure separation, and thatit has since re-evolved a shunting system.The crocodilian ancestors from the Triassic(248 million to 206 million years ago), suchas the long-legged Terrestrisuchus and thebipedal Saltoposuchus, were rather small(less than 2 metres long), completely terres-trial animals that are thought to have had anactive lifestyle. They would also have had adiaphragmaticus, a muscle that inflates thelungs by pulling backwards on the liver5.Thisis analogous to the push that our diaphragmmuscle exerts on the liver for the same pur-pose, and it allows higher ventilation ratesthan contemporary crocodiles require.

In the Jurassic (206 million to 144 millionyears ago), larger, fully aquatic crocodilianforms developed, and presumably adoptedthe same sit-and-wait predatory strategythat characterizes modern crocs. Becauseheat loss in water can be many times higherthan that in air, and because larger animalshave higher metabolic rates, there wouldhave been considerable selective pressure

in the forthcoming Fourth AssessmentReport of the Intergovernmental Panel onClimate Change,due in early 2007.

The evidence for such dramatic changesis still ambiguous, but distributed model-ling14 and coordinated analyses15 will allowus to improve our estimates of the probabil-ity of such events. Results such as those ofWu et al.1 help us on our way. They testify to the ability of today’s models to provide a detailed picture of the environmentalchanges that continued anthropogenicwarming of the planet will cause. ■

Thomas F. Stocker and Christoph C. Raible are atthe Division of Climate and Environmental Physics,University of Bern, Sidlerstrasse 5, CH-3012 Bern,Switzerland.e-mail: stocker@climate.unibe.ch

Evolution

Warm-hearted crocsAdam P. Summers

Our ideas about how crocodiles evolved have just taken a battering. It seems that these cold-blooded creatures, with their limited capacityfor prolonged activity, might have had active, warm-blooded ancestors.

Acrocodile,quietly submerged,waitingto ambush its next meal, is a testamentto the benefits of allowing the envi-

ronment to determine body temperature.Cool water keeps the metabolic rate low,allowing astonishing periods betweenbreaths that can range into hours. Writing in Physiological and Biochemical Zoology,Seymour and colleagues1 propose that insidethese armoured predators beats a heart thatharks back to ancestors that were terrestrial,active and — most surprisingly — ‘warm-blooded’.

Warm-blooded, or ‘endothermic’, ver-tebrates (birds and mammals) have high resting metabolic rates, enabling them tomaintain a constant body temperature that is usually above that of their environment.Endotherms use more oxygen and need more fuel than cold-blooded ‘ectothermic’vertebrates, such as fish, lizards and frogs,with their lower and more variable body temperatures. But, as a consequence, endo-therms have gained a capacity for prolongedstrenuous activity that can’t be matched byectotherms2.

Birds and mammals have independentlyevolved a four-chambered heart divided intotwo sides, so that the oxygenated blood com-ing from the lungs is separated from thedeoxygenated blood arriving from the rest ofthe body (Fig. 1a). The three-chamberedheart of most reptiles and amphibians ismore flexible, allowing controlled mixing ofpulmonary and systemic blood (Fig. 1b).This is especially important in diving, whereshunting blood from the lungs accelerates

Figure 1 Pump rooms. a, The four-chambered heart of mammals and birds separates oxygenated (red)from deoxygenated (blue) blood, and applies different pressures to the blood that goes to the lungsthrough the pulmonary artery (PA) and the blood that is pumped through the aorta to the rest of the body. b, The three-chambered heart of most reptiles allows oxygenated and deoxygenated blood to mix and does not permit the large pressure difference between the pulmonary and aorticsystems. c, Modern crocodiles have a four-chambered heart, but can ‘shunt’ blood between the two systems through the foramen of Panizza, and are hypothesized1 to have ancestors that were able to maintain large blood-pressure differences. RAt, right atrium; LAt, left atrium, RV, right ventricle;LV, left ventricle; RA, right aorta; LA, left aorta.

a Mammals and birds c Crocodilesb Turtles, snakes and lizards

PA PA

RAtLAt

LVRV

Aorta RALA

PARA

LA

1. Wu, P., Wood, R. & Stott, P. Geophys. Res. Lett. 32, L02703

(2005).

2. Stott, P. A. et al. Science 290, 2133–2137 (2000).

3. Peterson, B. J. et al. Science 298, 2171–2173 (2002).

4. Simmons, A. J. & Gibson, J. K. The ERA40 Project Plan

(European Centre for Medium-Range Weather Forecasts,

Reading, UK, 2000).

5. Curry, R., Dickson, B. & Yashayaev, I. Nature 426, 826–829

(2003).

6. Delworth, T. L. & Mann, M. E. Clim. Dyn. 16, 661–676 (2000).

7. Houghton, J. T. et al. (eds) Climate Change 2001: The Scientific

Basis (Cambridge Univ. Press, 2001).

8. Corell, R. et al. Impacts of a Warming Arctic — Arctic Climate

Impact Assessment (Cambridge Univ. Press, 2004).

9. Stocker, T. F., Wright, D. G. & Broecker, W. S. Paleoceanography

7, 529–541 (1992).

10.Bryan, F. O. et al. J. Clim. (submitted).

11.Stocker, T. F. & Schmittner, A. Nature 388, 862–865 (1997).

12.Knutti, R. & Stocker, T. F. J. Clim. 15, 179–186 (2002).

13.Huybrechts, P., Gregory, J., Janssens, I. & Wild, M. Earth Planet.

Sci. Lett. 42, 83–105 (2004).

14.www.climateprediction.net

15.www-pcmdi.llnl.gov

14.4 n&v 829 8/4/05 5:30 pm Page 833

Nature Publishing Group© 2005

news and views

834 NATURE | VOL 434 | 14 APRIL 2005 | www.nature.com/nature

Geophysics

Double-crossed againMichael E. Wysession and Viatcheslav S. Solomatov

An idea that a mineral phase transition may occur not once, but twice,close to the core–mantle boundary has been tested with seismic data.The resulting picture of the deep Earth is sure to provoke debate.

Thermal boundary layers within planetsact like thermal bottlenecks, control-ling the rate at which heat escapes from

the planet and regulating much of its internalbehaviour. In the solid Earth, there are twomajor thermal boundary layers: the litho-sphere at the top of Earth’s mantle, and theD�� layer at the bottom of the mantle. If weknew the temperature increase across the D��layer, we would have a sense of the amount of heat flowing out of the core, and thus thepower that drives the core geodynamo andthe generation of plumes in the mantle.

The temperature gradient across D�� hasremained largely unconstrained. However,the model of Hernlund et al. (page 882 of thisissue)1 provides a constraint. If the model is correct, the characteristics of the newlydiscovered mineral phase change from perovskite (Pv, the dominant mineral in thelower mantle) to a post-perovskite (pPv)phase2 may have fortuitously given us ameans of seismically measuring both thestrength and lateral variation of the thermalboundary layer at the base of the mantle.

How does the model work? Phase tran-sitions in Earth’s minerals are natural

thermometers, because the pressure at which they happen depends on temperature(this dependence is characterized by theClapeyron slope). So if we know the pressure, or the depth, at which the phase transition occurs we know the temperature.Hernlund et al. push this principle a step further and argue that the Pv–pPv phasetransition may happen not just once buttwice — perovskite transforms to post-perov-skite and then post-perovskite transformsback to perovskite. Earth’s temperature pro-file (geotherm) may therefore ‘double-cross’the Pv–pPv phase boundary (Fig. 1a). Such a fortunate situation allows us to estimatenot only the temperature, but also the temperature gradient and thus the heat flowout of the core.

So far, this discussion is hypothetical, aswe haven’t left the laboratory. But this iswhere we bring in seismology, which servesfor geophysics the same role that the tele-scope serves for astrophysics: it lets you seewhat’s out (or in) there. Each of the mantle’smineral phase changes increases the speed ofseismic waves and thus alters the way seismicwaves from earthquakes propagate. If the

against endothermy during this return to thewater. In addition, animals employing thelong dive-times characteristic of an ambushpredator would have benefited from a shunt-ing system capable of isolating the lungs orincreasing their relative perfusion. DNAextracted from the mitochondria — the cel-lular structures that use oxygen to convertthe potential energy from food into a formthe cell can use — provides clues to evolutionand energy usage. The mitochondrial DNAfrom the Jurassic crocodilians shows a highrate of evolution6 — roughly equivalent tothat of mitochondrial DNA from modernendotherms. This, together with the dia-phragmaticus muscle and a four-chamberedheart, points to an ancestral croc with an elevated metabolic rate.

The idea of meeting an active, agile, hos-tile croc while hiking is enough to alarm any-one; but the hypothesis of an endothermiccrocodilian ancestor has raised comment on several scientific fronts. There are fewskeletal characteristics that are convincinglyassociated with endothermy7, but nasal‘turbinates’ do fit the bill. These scroll-likebones serve as heat and water exchangers inthe nose, protecting panting endothermsfrom desiccation or from getting chilled bylarge air exchange in the lungs.Crocs have noturbinates,and there is no evidence that theirfossil ancestors did either. It is also possiblethat a four-chambered heart is characteristicof an active lifestyle, rather than of endo-thermy per se 8.

There is little agreement among expertsabout the body temperature of dinosaurs,and without significant new fossils directevidence is unlikely. That leaves the questionin the camp of comparative physiologistswho study the working of extant animals inhopes of deducing the function of animalslong since turned to dust. Evidence frombirds and crocodiles, the two living groupsthat ‘bracket’ the dinosaurs, is particularlyuseful. If ancestral crocs had high restingmetabolic rates, the clear implication is thatdinosaurs did as well. ■

Adam P. Summers is in the Department of Ecologyand Evolutionary Biology, 321 Steinhaus Hall,University of California, Irvine,California 92697-2525, USA.e-mail: asummers@uci.edu1. Seymour, R. S., Bennett-Stamper, C. L., Johnston, S. D.,

Carrier, D. R. & Grigg, G. C. Physiol. Biochem. Zool. 77,

1051–1067 (2004).

2. Bennett, A. F. J. Exp. Biol. 160, 1–23 (1991).

3. Platzack, B. & Hicks, J. W. Am. J. Physiol. 281, R1295–R1301

(2001).

4. Wang, T., Altimiras, J. & Axelsoson, M. J. Exp. Biol. 205,

2717–2723 (2003).

5. Carrier, D. R. & Farmer, C. G. Am. Zool. 40, 87–100 (2000).

6. Janke, A. & Arnason, U. Mol. Biol. Evol. 14, 1266–1272

(1997).

7. Bennett, A. F. & Ruben, J. A. in The Ecology and Biology of

Mammal-like Reptiles (eds Hotton, N., MacLean, P. D.,

Roth, J. J. & Roth, E. C.) (Smithsonian Institution,

Washington DC, 1986).

8. Hillenius, W. J. & Ruben, J. A. Physiol. Biochem. Zool. 77,

1068–1072 (2004).

Pv

Core

pPv

Dep

th

Temperature

Core

Pv

pPv

Mantle

MantleEarthquakeSeismometers

Geotherm

Phaseboundary

Dep

th

a b

Figure 1 Perovskite (Pv) and post-perovskite (pPv) within Earth’s lower mantle. a, Double-crossing ofthe Pv–pPv phase boundary (dashed line) is inferred from the large Clapeyron slope of the Pv–pPvphase boundary (several times bigger than that of other phase transitions in Earth’s mantle) and itslocation inside the thermal boundary layer where the geotherm changes rapidly. Both crossingscreate a change in density and seismic velocity that make them ‘visible’ to seismic waves. b, Ahypothetical picture of the global appearance of the post-perovskite layer, based on the model ofHernlund et al.1. Depending on factors such as the local values of the core heat flow and mantletemperature, the Pv–pPv phase boundary may be crossed once (the post-perovskite layer appears allthe way to the core–mantle boundary), twice (the post-perovskite layer is sandwiched between layersof perovskite phase) or never (the Pv–pPv transition does not occur at all). A hypothetical set ofseismic ray paths is shown sampling both the top and bottom of the Pv–pPv phase boundaries.

14.4 n&v 829 8/4/05 5:30 pm Page 834

Nature Publishing Group© 2005