The Evolution of Continental The high-standing owe their existence to the ’s long history of plate-tectonic activity

by S. Ross Taylor and Scott M. McLennan

xcept perhaps for some remote of Venus compares with that of our own island dwellers, most people have world. Although centuries of telescopic Ea natural tendency to view conti- observations from the earth could give nents as fundamental, permanent and no insight, beginning in 1990 the Ma- even characteristic features of the earth. gellan space probeÕs orbiting radar pen- One easily forgets that the worldÕs con- etrated the thick clouds that enshroud tinental platforms amount only to scat- Venus and revealed its surface with tered and isolated masses on a planet stunning clarity. From the detailed im- that is largely covered by water. But ages of landforms, planetary scientists when viewed from space, the correct can surmise the type of that cov- picture of the earth becomes immedi- ers Venus. ately clear. It is a blue planet. From this perspective it seems quite extraordi- Comparison with the Neighbors nary that over its long history the earth could manage to hold a small fraction ur sister planet appears to be blan- of its surface always above the , en- O keted by rock of basaltic composi- abling, among other things, human evo- tionÑmuch like the dark, Þne-grained lution to proceed on dry land. rocks that line the ocean basins on the Is the persistence of high-standing earth. MagellanÕs mapping, however, continents just fortuitous? How did the failed to Þnd extensive areas analogous earthÕs complicated crust come into ex- to the terrestrial continental crust. Ele- istence? Has it been there all the time, vated regions named Aphrodite Terra like some primeval icing on a planetary and Ishtar Terra appear to be remnants cake, or has it evolved through the of crumpled basaltic lavas. Smaller, ages? Such questions had engendered dome-shaped mounds are found on debates that divided scientists for many Venus, and these forms might indicate decades, but the fascinating story of that a substantially diÝerent bedrock how the terrestrial surface came to take composition does exist in some places; its present form is now essentially re- solved. That understanding shows, re- markably enough, that the conditions required to form the continents of the earth may be unmatched in the rest of the solar system. The earth and Venus, being roughly the same size and distance from the sun, are often regarded as twin planets. So it is natural to wonder how the crust

EARTHÕS CRUST is composed primarily of the basaltic rocks that line the ocean basins. Granitic rocks constitute the high-standing continental platforms. Venus is nearly the same size as the earth, yet radar imagery indicates that it is encrusted al- most entirely by . Only a tiny fraction of that planetÕs surface exhibits pan- cake-shaped plateaus (detail above) that might, like the earthÕs continents, be built of granitic material. The crust of the earthÕs moon is largely covered by white high- lands formed as that body Þrst cooled from a molten state; volcanic eruptions later created dark so-called of basalt.

76 SCIENTIFIC AMERICAN January 1996 Copyright 1995 Scientific American, Inc. it is also possible that these pancake- These Þndings from Venus and simi- mixtures of rock with water, methane like features may be composed merely lar surveys of other solid bodies in the and ammonia ices, may also have aris- of more basalt. solar system show that planetary crusts en from catastrophic melting during After analyzing the wealth of radar can be conveniently divided into three initial . data provided by Magellan, scientists fundamental types. So-called primary In contrast to the product of such );

have concluded that crusts date back to the beginnings of sudden, large-scale episodes of melt- s

(that is, the continual creation, motion the solar system. They emerged after ing, secondary crusts form after heat enu V ) and destruction of parts of the planetÕs large chunks of primordial material from the decay of radioactive elements n

Moo surface) does not seem to operate on came crashing into a growing planet, gradually accumulates within a plane- ( Venus. There are no obvious equiva- releasing enough energy to cause the tary body. Such slow heating causes a ); JPL/NASA (

l ); NASA lents to the extensive mid-ocean ridges original protoplanet to melt. As the small fraction of the planetÕs rocky in- h


Eart or to the great trench systems of the molten rock began to cool, crystals of terior to melt and usually results in the ( earth. Thus, it is unlikely that the crust some types of minerals solidiÞed rela- eruption of basaltic lavas. The surfaces of Venus regularly recycles back into tively early and could separate from the of Mars and Venus and the earthÕs ocean Y/NASA ( OR that planetÕs . Nor would there body of . This process, for exam- ßoors are covered by secondary crusts T seem to be much need to make room ple, probably created the white high- created in this way. The lunar maria (the for new crust: the amount of lava cur- lands of the moon after low-density ÒseasÓ of the ancient astronomers) also The Geosphere Project rently erupting on Venus is roughly grains of the mineral ßoated to formed from basaltic lavas that origi- equivalent to the output of one Hawaii- the top of an early lunar ÒoceanÓ of mol- nated deep in the moonÕs interior. Heat an , KilaueaÑa mere dribble for ten basalt. The crusts of many satellites from radioactivityÑor perhaps from AN SANT OM V

the planet as a whole. of the giant outer planets, composed of the ßexing induced by tidal forcesÑon JET PROPULSION LABORA T



Copyright 1995 Scientific American, Inc. SCIENTIFIC AMERICAN January 1996 77 some icy moons of the outer solar sys- en several billion years to produce its 6 tem may, too, have generated second- tertiary crustÑthe continents. Yet these 4 ary crusts. features amount to just about one half Unlike these comparatively common of 1 percent of the mass of the planet. CONTINENTS 4 2 types, so-called tertiary crust may form if surface layers are returned back into Floating Continents 2 the mantle of a geologically active plan- 0 et. Like a form of continuous distillation, any elements that are otherwise volcanism can then lead to the produc- Mrarely found on the earth are en- VENUS 0 tion of highly diÝerentiated magma of riched in granitic rocks, and this phe- Ð2 a composition that is distinct from ba- nomenon gives the continental crust saltÑcloser to that of the light-colored an importance out of proportion to its Ð2 Ð4 . Because the recy- tiny mass. But geologists have not been EARTH ELEVATION (KILOMETERS) cling necessary to generate granitic mag- able to estimate the overall composi- VENUS ELEVATION (KILOMETERS) Ð4 mas can occur only on a planet where tion of crustÑa necessary starting point Ð6 plate tectonics operates, such a compo- for any investigation of its origin and OCEAN BASINS sition is rare in the solar system. The evolutionÑby direct observation. One formation of continental crust on the conceivable method might be to com- 0 0.5 1.0 RELATIVE AREA earth may be its sole demonstration. pile existing descriptions of rocks that JANA BRENNING Despite the small number of exam- outcrop at the surface. But even this SURFACE ELEVATIONS are distributed ples within each category, one general- large body of information might well quite diÝerently on the earth (blue) and ization about the genesis of planetary prove insuÛcient. A large-scale explora- on Venus (gold ). Most places on the surfaces seems easy to make: there are tion program that could reach deeply earth stand near one of two prevailing clear diÝerences in the rates at which enough into the crust for a meaningful levels. In contrast, a single height char- primary, secondary and tertiary crusts sample would press the limits of mod- acterizes most of the surface of Venus. form. The moon, for instance, generated ern drilling technology and would, in (Elevation on Venus is given with re- its white, feldspar-rich primary crustÑ any event, be prohibitively expensive. spect to the planetÕs mean radius.) about 12 percent of lunar volumeÑin Fortunately, a simpler solution is at only a few million years. Secondary hand. Nature has already accomplished crusts evolve much more slowly. The a widespread sampling through the ero- ly useful in deciphering crustal compo- moonÕs basalt maria (secondary crust) sion and deposition of sediments. Low- sition because their atoms do not Þt are just a few hundred meters thick and ly muds, now turned into solid rock, neatly into the crystal structure of most make up a mere one tenth of 1 percent give a surprisingly good average com- common minerals. They tend instead to of the moonÕs volume, and yet these position for the exposed continental be concentrated in the late-forming gra- so-called seas required more than a bil- crust. These samples are, however, miss- nitic products of a cooling magma that lion years to form. Another example of ing those elements that are soluble in make up most of the continental crust. secondary crust, the basaltic oceanic water, such as sodium and calcium. Because the REE patterns found in a basins of our planet (which constitute Among the insoluble elements that are variety of sediments are so similar, geo- about one tenth of 1 percent of the transferred from the crust into sedi- chemists surmise that weathering, ero- earthÕs mass) formed over a period of ments without distortion in their rela- sion and sedimentation must mix dif- about 200 million years. Slow as these tive abundances are the 14 rare-earth ferent igneous source rocks eÛciently rates are, the creation of tertiary crust elements, known to geochemists as enough to create an overall sample of is even less eÛcient. The earth has tak- REEs. These elemental tags are unique- the continental crust. All the members of the REE group establish a signature of upper crustal composition and pre- serve, in the shapes of the elemental Lanthanum abundance patterns, a record of igneous 58 Cerium events that may have inßuenced the Praseodymium makeup of the crust. 60 Neodymium Using these geochemical tracers, ge- 62 Samarium ologists have, for example, determined that the composition of the upper part Oceanic basalt Europium 64 Gadolinium of the continental crust approximates Terbium that of , an ordinary ig- Continental basalt 66 Dysprosium neous rock that consists largely of light- Granite

ATOMIC NUMBER Holmium colored and feldspar, along with 68 Archean sodium granite Erbium a peppering of various dark minerals. Thulium Deep within the continental crust, be- 70 Ytterbium low about 10 to 15 kilometers, rock of Lutetium a more basaltic composition is proba- 10 100 500 bly common. The exact nature of this

CONCENTRATION RELATIVE TO METEORITES JANA BRENNING material remains controversial, and ge- ologists are currently testing their ideas RARE-EARTH ELEMENT abundance patterns provide characteristic chemical mark- ers for the types of rock that have formed the earthÕs crust. Although igneous using measurements of the heat pro- rocks (those that solidify from magma) can have highly variable rare-earth ele- duced within the crust by the impor- ment signatures (dotted lines), the pattern for most sedimentary rocks falls within tant radioactive elements potassium, a narrow range (gray band). That uniformity arises because sediments eÝectively uranium and thorium. But it seems rea- record the average composition of the upper continental crust. sonable that at least parts of this inac-

78 SCIENTIFIC AMERICAN January 1996 Copyright 1995 Scientific American, Inc. cessible and enigmatic region may con- found within diamonds, which are have grown by accretion, before Þnally sist of basalt trapped and underplated thought to originate deep in this sub- falling together to form our planet, a beneath the lower-density continents. crustal region. Measurements show that process that required about 50 million It is this physical property of granit- diamonds can be up to three billion to 100 million years. ic rockÑlow densityÑthat explains why years old and thus demonstrate the an- Late in this stage of formation, a most of the continents are not sub- tiquity of the deep continental roots. massive planetesimal, perhaps one the merged. Continental crust rises on av- It is curious to reßect that less than size of Mars, crashed into the nearly erage 125 meters above sea level, and 40 years ago, there was no evidence that fully formed earth. The rocky mantle of some 15 percent of the continental area the rocks lining ocean basins diÝered the impactor was ejected into orbit and extends over two kilometers in eleva- in any fundamental way from those became the moon while the metallic tion. These great heights contrast mar- found on land. The oceans were simply core of the body fell into the earth [see kedly with the depths of ocean ßoors, thought to be ßoored with foundered or ÒThe ScientiÞc Legacy of Apollo,Ó by G. which average about four kilometers sunken continents. This perception grew JeÝrey Taylor; SCIENTIFIC AMERICAN, below sea levelÑa direct consequence naturally enough from the concept that July 1994]. As might be expected, this of their being lined by dense oceanic the continental crust was a world-encir- event proved catastrophic: it totally crust composed mostly of basalt and a cling feature that had arisen as a kind melted the newly formed planet. As the thin veneer of sediment. of scum on an initially molten planet. earth later cooled and solidiÞed, an At the base of the crust lies the so- Although it now appears certain that early basaltic crust probably formed. called Mohorovicic discontinuity (a the earth did in fact melt very early, it It is likely that at this stage the sur- tongue-twisting name geologists invari- seems that a primary granitic crust, of face of the earth resembled the current ably shorten to ÒMohoÓ). This deep sur- the type presumed decades ago, never appearance of Venus. None of this pri- face marks a radical change in compo- actually existed. mary crust has, however, survived. sition to an extremely dense rock rich Whether it sank into the mantle in a in the mineral olivine that everywhere The Evolution of Geodiversity manner similar to to that taking place underlies both oceans and continents. on the earth or piled up in localized Geophysical studies using seismic waves ow was it that two such distinct masses until it was thick enough to have traced the Moho worldwide. Such H kinds of crust, continental and transform into a denser rock and sink research has also indicated that the oceanic, managed to arise on the earth? remains uncertain. In any event, there mantle below the continents may be To answer this question, one needs to is no evidence of substantial granitic permanently attached at the top. These consider the earliest history of the so- crust at this early stage. Telltale evi- relatively cool subcrustal ÒkeelsÓ can be lar system. In the region of the primor- dence of such a crust should have sur- as much as 400 kilometers thick and dial solar nebula occupied by the earthÕs vived in the form of scattered grains of appear to ride with the continents dur- orbit, gas was mostly swept away, and the mineral zircon, which forms within ing their plate-tectonic wanderings. only rocky debris large enough to sur- granite and is enormously resistant to Support for this notion comes from the vive intense early solar activity accumu- erosion. Although a few ancient zircons

MICHAEL GOODMAN analysis of tiny mineral inclusions lated. These objects themselves must dating from near this time have been





PLATE-TECTONIC ACTIVITY carries oceanic crust deep into buoyantly and forms new continental material near the sur- the earth (left ), burying wet sediments along with the de- face. As this crust matures (right ), heat from radioactivity scending slab. At a depth of 80 kilometers, high tempera- (or from rising plumes of basaltic magma) may subsequently tures drive water from the sediments, inducing the overlying trigger melting at shallow levels. Such episodes create an up- rock to melt. The magma generated in this process rises per layer that is made up largely of granite.






CRUSTAL GROWTH has proceeded in episodic fashion for bil- chean and eons. Widespread melting at this time lions of years. An important growth spurt lasted from about formed the granite bodies that now constitute much of the 3.0 to 2.5 billion years ago, the transition between the Ar- upper layer of the continental crust. found (the oldest examples are from of what were probably true oceans dur- curs. The composition of the upper sedimentary rocks in and are ing this remote epoch. But even these crust before this break contained less about 4.2 billion years old), these grains extraordinarily old rocks from Canada evolved constituents, composed of a are exceedingly scarce. and Greenland date from some 400 mil- mixture of basalt and sodium-rich gran- More information about the early lion to 500 million years after the ini- ites. These rocks make up the so-called crust comes from the most ancient tial accretion of the earth, a gap in the tonalite-trondjemite-granodiorite, or rocks to have survived intact. These geologic record caused, no doubt, by TTG suite. This composition diÝers rocks formed deep within the crust just massive impacts that severely disrupt- considerably from the present upper less than four billion years ago and now ed the earthÕs earliest crust. crust, which is dominated by potas- outcrop at the surface in northwest From the record preserved in sedi- sium-rich granites. Canada. This rock formation is called mentary rocks, geologists know that The reason for the profound change the Acasta Gneiss. Slightly younger ex- the formation of continental crust has in crustal composition 2.5 billion years amples of early crust have been docu- been an ongoing process throughout ago appears to be linked to the earthÕs mented at several locations throughout the earthÕs long history. But the creation plate-tectonic churnings. Before this the world, although the best studied of of crust has not always had the same time, oceanic crust was being recycled these ancient formations is in western character. For example, at the boundary rapidly because higher levels of radio- Greenland. The abundance of sedimen- between the Archean and Proterozoic activity existed and more intense heat- tary rock there attests to the presence eons, around 2.5 billion years ago, a ing tended to drive a faster plate-tecton- of running water and to the existence distinct change in the rock record oc- ic engine. There may have been more than 100 separate plates during the Ar- chean, compared with the dozen that currently exist. Unlike the present ocean- Snapshot of the Past ic crustÑwhich travels a long distance and cools signiÞcantly before sinking odium-rich granites back into the mantleÑoceanic crust dur- Sare characteristic of ing this earlier epoch survived for a far 23- TO 35-MILLION- the Archean eon, a time shorter time. It was thus still relatively YEAR-OLD CRUST when the earth’s plate- hot when it entered the mantle and so tectonic engine operat- began to melt at shallower depths than CHILE ed more quickly than it typically occurs now. This diÝerence 6 TO 23 does today. Such tona- accounts for the formation of sodium- MILLION lite-trondjemite-grano- rich igneous rocks of the TTG suite; YEARS 2 TO 6 diorite (TTG) granites such rocks now form only in those few MILLION form only if young (and places where young, hot oceanic crust YEARS hence hot) oceanic crust subducts. enters the mantle, spark- The early tendency for magma to ing melting at shallow form with a TTG composition explains levels. These conditions why crust grew as a mixture of basalt still exist along the and tonalite during the Archean eon. LESS THAN southern coast of Chile, Large amountsÑat least 50 percent and 2 MILLION YEARS where relatively new perhaps as much as 70 percent of the oceanic crust has sub- continental crustÑemerged at this time, ducted underneath the with a major episode of growth between GREATER THAN South American plate. 3.0 and 2.5 billion years ago. Since that 35 MILLION YEARS The region containing time, the relative height of ocean basins TTG granites is marked and continental platforms has remained

MICHAEL GOODMAN in brown. comparatively stable. With the onset of the Proterozoic eon 2.5 billion years

80 SCIENTIFIC AMERICAN January 1996 Copyright 1995 Scientific American, Inc. ago, the crust had already as- 250 million years ago), the ma- sumed much of its present jor continents of the earth con- makeup, and modern plate- verged to create one enormous tectonic cycling began. landmass called Pangea [see Currently oceanic crust forms ÒEarth before Pangea,Ó by Ian by the eruption of basaltic lava W. D. Dalziel; SCIENTIFIC AMER- along a globe-encircling net- ICAN, January 1995]. This con-

work of mid-ocean ridges. More TION Þguration was not unique. The than 18 cubic kilometers of formation of such Òsuperconti- rock are produced every year nentsÓ appears to recur at in- by this process. The slab of tervals of about 600 million

newly formed crust rides on ACE ADMINISTRA years. Major tectonic cycles top of an outer layer of the driving the continents apart mantle, which together make and together have been docu- up the rigid . The mented as far back as the Early oceanic lithosphere sinks back Proterozoic, and there are even into the mantle at so-called suggestions that the Þrst su- subduction zones, which leave percontinent may have formed conspicuous scars on the ocean earlier, during the Archean. TIONAL AERONAUTICS AND SP

ßoor in the form of deep NA Such large-scale tectonic cy- trenches. At these sites the de- cles serve to modulate the tem- scending slab of lithosphere VOLCANOES, such as this one erupting on the Kamchatka po of crustal growth. When a Peninsula, mark where new continental material forms carries wet marine sediments supercontinent breaks itself above subducting oceanic crust. In time, such geologically as well as basalt plunging into active will evolve into stable continental crust. apart, oceanic crust is at its old- the mantle. est and hence most likely to At a depth of about 80 kilo- form new continental crust af- meters, heat drives water and other vol- in the mantle and becomes trapped un- ter it subducts. As the individual conti- atile components from the subducted der the granitic lid; the molten rock then nents reconverge, volcanic arcs (curved sediments into the overlying mantle. acts like a burner under a frying pan. chains of volcanoes created near sub- These substances then act as a ßux does duction zones) collide with continental at a foundry, inducing melting in the Crustal Growth Spurts platforms. Such episodes preserve new surrounding material at reduced tem- crust as the arc rocks are added to the perature. The magma produced in this lthough the most dramatic shift in margins of the continents. fashion eventually reaches the surface, A the generation of continental crust For more than four billion years, the where it causes spectacular, explosive happened at the end of the Archean peripatetic continents have assembled eruptions. Pinatubo and Mount St. He- eon, 2.5 billion years ago, the conti- themselves in Þts and starts from many lens are two recent examples of such nents appear to have experienced epi- disparate terranes. Buried in the result- geologic cataclysms. Great chains of vol- sodic changes throughout all of geologic ing amalgam is the last remaining testa- canoesÑsuch as the AndesÑpowered by time. For example, sizable, later addi- ment available for the bulk of the earthÕs boiling volatiles add on average about tions to the continental crust occurred history. That story, assembled from two cubic kilometers of lava and ash to from 2.0 to 1.7, from 1.3 to 1.1 and from rocks that are like so many jumbled the continents every year. 0.5 to 0.3 billion years ago. That the pieces of a puzzle, has taken some time But subduction-induced volcanism is earthÕs continents experienced such a to sort out. But the understanding of not the only source of new granitic rock. punctuated evolution might appear at crustal origin and evolution is now suf- The accumulation of heat deep within Þrst to be counterintuitive. Why, after Þcient to show that of all the planets the the continental crust itself can cause all, should crust form in spurts if the earth appears truly exceptional. By a melting, and the resultant magma will generation of internal heatÑand its lib- fortunate accident of natureÑthe abili- ultimately migrate to the surface. Al- eration through crustal recyclingÑis a ty to maintain plate-tectonic activityÑ though some of this necessary heat continuous process? one planet alone has been able to gen- might come from the decay of radioac- A more detailed understanding of erate the sizable patches of stable con- tive elements, a more likely source is plate tectonics helps to solve this puz- tinental crust that we Þnd so convenient basaltic magma that rises from deeper zle. During the Permian period (about to live on.

The Authors Further Reading

S. ROSS TAYLOR and SCOTT M. MCLENNAN have worked NO WATER, NO GRANITESÑNO OCEANS, NO CONTINENTS. I. H. Campbell together since 1977 examining the earthÕs crustal evolution. and S. R. Taylor in Geophysical Research Letters, Vol. 10, No. 11, pages Taylor has also actively pursued lunar and planetary studies 1061Ð1064; November 1983. and has published four books on planetology. He is a foreign THE CONTINENTAL CRUST: ITS COMPOSITION AND EVOLUTION. S. Ross Tay- associate of the National Academy of Sciences. Taylor is cur- lor and Scott M. McLennan. Blackwell ScientiÞc Publications, 1985. rently with the Research School of Physical Sciences at the OASIS IN SPACE: EARTH HISTORY FROM THE BEGINNING. Preston Cloud. Australian National University. McLennan is a professor in the W. W. Norton, 1988. department of earth and space sciences at the State Universi- THE GEOCHEMICAL EVOLUTION OF THE CONTINENTAL CRUST. S. Ross Taylor ty of New York at Stony Brook. He recently received a Presi- and Scott M. McLennan in Reviews of Geophysics, Vol. 33, No. 2, pages dential Young Investigator Award from the National Science 241Ð265; May 1995. Foundation. His research applies the geochemistry of sedi- NATURE AND COMPOSITION OF THE CONTINENTAL CRUST: A LOWER mentary rocks to studies of crustal evolution, plate tectonics CRUSTAL PERSPECTIVE. Roberta L. Rudnick and David M. Fountain in Re- and ancient climates. views of Geophysics, Vol. 33, No. 3, pages 267Ð309; August 1995.

Copyright 1995 Scientific American, Inc. SCIENTIFIC AMERICAN January 1996 81