review article

220

(3) Not al! intra-plate oceanic volcanoes originate from plumes.Thousands" of small, isolated volcanic seamounts litter the oceaniccrust and are probably formed by local melting anomalies. Inaddition, there are 'hot lines' of volcanoes that show no ageprogression but erupt roughly simultaneously along the en tireextent of the line (for example, the Cameroon Line volcanoes"),Most of these geochemical and melting anomalies probably comefrom the upper mantle.

(4) Continental flood basalts and oceanic plateaus are, accordingto some authors", formed by the surfacing of a large, mushroom-like head of a 'starting' plum e, which has entrained large amounts ofmantle material on the way from its origin to the surface. Often, aprogressively 'younging' line of volcanoes connects an old floodbasalt province (for example, the Deccan Traps of India) with acurrently active hotspot (for example, Reunion Island). This isinterpreted as the surface expression ofthe much narrower conduitor 'stern' of the plume, which may be active for over a hundredmil!ion years.

(5) Sea-floor spreading generates large chemical heterogeneitiesby extracting melt from the mantle, thereby forming a basaltic crustand a refractory residue. Subduction reinjects these heterogeneitiesback into the mantle, where they are gradually rehomogenized byconvection". Subduction also causes melting, resulting in volcanic'island ares', which are ultimately accreted to the continental crust.

The geochemical evidenceTo characterize the mantle source regions of basaltic lavas, thegeochemist uses a variety of geochemical tracers. Such tracers areeither isotope abundance ratios of daughter elements of radio activenuclides, or concentration ratios of incompatible trace elements.The melt 'copies' these tracer ratios from the source and deliversthem to the surface with little or no distortion. In spite of occasionalassertions to the contrary, the geochemical tracing approach haswithstood the test of time quite well (see, for example refs 1 and 25)and will be used here without apologies. Table 1 shows the radio-active decay systems used in this Review, their half-lives, and(radiogenic to non-radiogenic) isotope ratios of the daughterelements, which will vary as a function of age and parent/daughterelement ratio (discussed below). 1 focus here purely on oceanicbasalts (MORB and OlE) rather than mantle xenoliths or volcanicrocks from island ares or continental regions, beca use oceanicbasalts represent relatively large volumes of mantle and carry thesmallest risk of being contaminated during magma transportthrough the crust to the surface.Elemental chemistry of MORB, OIB and continental crust.General chemical characteristics of average continental crust, aver-age oceanic crust (= MORB), and selected OIB are shown in Fig. 2.Concentrations are normalized to those of the primitive mantleand plotted in the order of increasing compatibility, which, exceptfor the 'anornalous' elements Nb, Pb and Ti (see below), coincideswith the order of descending abundances in the continental crust.All types of crustal rock (MORB, OIB and continental crust) areenriched in incompatible elements (relative to the primitive mantle)beca use they originate as melts, which concentrate incompatibleelements from the mande. However, in the oceanic crust(= MORB), the most highly incompatible elements (for example,Rb, Ba and Th) are actually relatively depleted in comparison withthe moderately incompatible elements (for example Sm and Hf),because this crust is formed by melting of a mantle region that haspreviously been depleted by extraction of the continental crust",This effect also explains the mutually complementary positive andnegative anomalies of Nb and Pb.

Ocean island basalts are much more enriched in incompatibleelements than MORB, and they show a large variation in com-position. This variability will be especial!y obvious in the isotopiccompositions discussed below. Here 1 show average concentrationpatterns for basalts from Mauna Loa (Hawaii) and two of the

NATURE IV0L385 116 )ANUARY 1997

isotopically 'extreme' types of mantle plume (see below and Figs 3-6), namely Tubuai, Austral Islands (HIMU) and Tristan-Inacces-sible Island (EM-l). (See Isotope taxonomy section, below, forexplanation ofHIMU, EM-l and EM-2). Their greater enrichmentsin incompatible elements cause them superficially to resemblecontinental crust, but their Nb and Pb anomalies are similar tothose of the oceanic crust ("= MORB) and opposite to those of thecontinental crust. Thus, except for their grassly different relativelevels of enrichment (which can be explained by differences in themeIting process), all these basaIt types have fundamentally similartrace-element patterns. In contrast, the trace-element patterns ofEM-2 basaIts (not shown for reasons of clarity) differ fram those ofother OIB and resemble that of continental crust (see section ontrace-element ratios).Isotope chemistry. The radio active parent nuclides listed in Table 1are all very long-lived, so the daughter isotope ratios reflect thelong-term history of the basalt-source reservoirs in the mantle.Figures 3-6 show Sr, Nd and Pb isotope data for over 1,100 samplesof MORB and OIB.Strontium, neodymium and hafnium. Isotope data for Sr and Nd inMORB, OIB and continental crust are summarized in Fig. 3. TheMORB points form a relatively tight cluster that defines the upperleft-hand comer of the array. In general, 143Nd/144Nd correlatesnegatively with 87Sr/86Sr, and positively with 176Hf/177Hf (notshown). The isotopic composition of the primitive mande,marked by a square, is normally estimated from this correlationand the Nd value of meteorites", These correlations are consistentwith the relative compatibilities shown in Fig. 2, namely Rb < Sr,Sm> Nd, and Lu > Hf. Thus, mande regions depleted in incom-patible elements have low Rb/Sr and therefore low 87Sr/86Srratios,coupled with hifh Sm/Nd and Lu/Hf and therefore high143Nd/144Nd and 76Hf/177Hf ratios. Continental crust, which isenriched in incompatible elernents, shows the opposite isotopicsignatures. The position of OIB between the depleted MORB sourceand the continental crust suggests that the OIB sources might just bethe result of back-mixing of various types of continental materialinto the mantle. It will be seen below that this is, for the most part,not the case.Lead. The Pb isotope data (Fig. 4) are more complex than those forSr, Nd and Hf, in part because the geochemical behaviour oflead isanomalous (see below and Fig. 2). The U/Pb ratio of the Earthcannot be inferred directly frorn meteorite data beca use the silicateportion ofthe Earth is severelydepleted in lead (either because it hasbeen lost from the Earth along with other volatile elements orbecause much ofthe lead has entered the core)". However, possibleprimitive mande compositions are constrained to lie along a line(the 'geochron') determined by the age of the mantIe. Also shownare estimates for the average upper (UCC) and lower (LCC)continental crust.

~igure 1 Models 01 mantle circulation. a, Old 'standard model' 01 two-Iayer

circulation. The upper layer has been depleted in incompatible eternents by the

lormation 01 continental crust. The tayers are separated by either an endothermic

phase change or by an intrinsic density contrast at a depth 01 660 km. Plumes rise

lrom the base 01 the lower. primitive (or less depleted) layer. The counterflow into

the lower layer is not specified. (MORB, mid-ocean-ridge basalt; OIB, ocean

island basalt.) b, Two-Iayer circulation with nearly complete isolation between

upper and lower layers. Plumes rise Irom the base 01 the upper layer but may

entrain small amounts 01 material Irom the lower layer. Plume sources are created

by recycling 01 oceanic or continentallithosphere. whieh creates reservoirs that

are enriched in incompatible elements. c. Whole-mantle, single-Iayer circulation

with plumes rising lrom the core-mantle boundary. Plume sources are created as

in b. d. Hybrid rnodel, with circulation occurring primarily in two separate layers

with small plumes rising lrom 660 km depth as in b, but with limited exchange by

means 01 occasional loundering 01 subducted lithosphere and rise 01 strong

plumes lrom the core-mantle boundary.

NATUREI V0L38S I 16JANUARY 1997

review article

Two features are noteworthy. (1) The Pb data do not form asimple trend extending from depleted MORB through OIB to thecontinental crust. Thus, OIB sources cannot be explained simply byback-mixing of continental crust (as might have been inferred fromthe Sr-Nd-Hf data). (2) Given the relative compatibilities ofU andPb fram Fig. 2, Pb isotope data in Fig. 4a for MORB should falI wellto the left of the 'geochron' (locus of primitive-mantle Pb composi-tions). In other words, the Pb data are expected to show roughly thesame topology as the Sr-Nd data in Fig. 3, but in fact most MORBdata are raughly centred on the geochron, or even lie on its right-hand side. Moreover, the bulk continental crust (located somewherebetween UCC and LCC) also lies close to the geochran rather than

a MOR8

Okm p

~~660km -----------------

018r¡

Aret> Continent

Continental~~ __ -! lithosphere________;» ~:~~~t~~ntl.

~¡IC Primitivalower mantle

2,900 km -Liquid outar core

b MOR8 018 Are

Okm ContinentalHthosphereDepletedupper mantle

660 km

~ e Pnmitlvelower mantle

2.900 kmLiquid outer ccre

e MOR8 018 Are

Okm

660 km

~ 1'1", I

""Recycling? I2,900 km

Liquid outer core

d MOR8 018 Are

Okm

Upper mantle660 km

~1Lower rnantle

2,900 km L- -==----==- L- __Liquid outer core

22t

review article

__ EM-1 (Tristan)

_ HtMU (Tubuai)

o Cont. crust___ MORB

__ Hawaii

100

eo~'Es 10e8'O

.~iiiEoz Primitive mantte

0.1 U-~~~~-L~~~~~-L~~~~~-L~~~~ Th U ~ ~ P ~ n ~ 8 ~ ~ ~~ ~ ~ ~ & ~ ~ ~ y ~ ~ &

Compatibility

Figure 2 Concentrations 01 selected trace and major elements, arranged in the

order 01 ascending compatibility and normalized to primitive-mantle concentra-

tions, in average continental crust'", average MORB', average Mauna Loa,

Havvaii'", and three types 01 018: average Tristan and Inaccessible islands

representing EM-l'20·'21,and Tubuai representing HIMU islands". The patterns

lor MORB and the oceanic island basalts differ by their different enrichments (or

depletions) in incompatible elements, but they are similar with respect to their

positive Nb and negative Pb anomalies. The continental crust has opposite Nb

and Pb anomalies. A lourth type 01018, called EM-2, is similarto continental crust

and has been omitted lar clarity. (See text lor explanations 01 EM-l, EM-2 and

HIMU.)

far on its right-hand side. A solution to this puzzle (the 'leadparadox'") will be suggested below.Correlations between Pb and other isotopes. Figure 5 shows relation-ships between 206Pb/204Pb and Sr isotope ratios. AIthough theseisotope ratios correlate positively in MORB from the Atlantic andPacific oceans, they do not correlate in MORB from the IndianOcean and in the overall OIB-MORB array. This indicates thatU-Pb and Th-Pb systems in the different mande reservoirs evolverather differently from the Rb-Sr (and Sm-Nd and Lu-Hf)systems.

In contrast, the parameter 2oapb*/zo6Pb*(the ratio of the radio-genic additions to the initial terrestrial lead, defined as Ie08Pb/204Pb) - e08Pb / 204Pb)ini..ltI e06Pb t 204Pb) - e06Pb / 204Pb)init.}does correlate with 143Nd/144Nd(and with Sr and Hf isotopes); seeFig. 6. This parameter depends primarily on Th/U (and age) but noton U/Pb or Th/Pb. The topology ofFig. 6 is rather similar to that ofFig. 3. This means that Th/U correlates with Sm/Nd (as well as withLu/Hf and Rb/Sr) in mantle evolution, but U/Pb and Th/Pb do not.This identifies Pb (rather than Th or U) as the anomalous elernent".Osmium. Recent improvements in analytical methods have madeosmium iso tope analyses of oceanic basalts feasible. The primitive-mande value o[l870s/1880s (or 1870S/1860S)is identical or similar tothe meteorite value":". This means that the Earth's mantle has thesame Re/Os ratio as meteorites, which is best explained by con-tinued infall of meteoritic material after the main accretion and coreformation of the Earth. Current results indicate the presence ofhigher-than-primitive 1870S/1880S in many OIB32-3S.BasaIts havemuch higher (up to a factor of a hundred or more) Re/Os ratios thantheir mantle sources, so that growth of 1870S/1880Sin the oceaniccrust is quite rapid. Therefore, Os isotopes should furnish an

222

• Paco MORBo Atlan. MORB• Indian MORB+ Other OIB

o FOZO, e• HIMUo EM-2• EM-1

0.5134FOZO

0.5132"OZ 0.5130vv-- 0.5128"OZ

0.5126(V)

v

0.5124

~c!'?() IV;:-o• S}'

0.51220.702 0.704 0.706 0.708

Figure 3 Nd and Sr isotopic compositions. assembled Irom literature sources. 01MORB and OIB, with extreme HIMU. EM-l and EM-2 samples marked in red.

brown and yellowcolours, respectively. Onlythose samples are shown larwhich

Pb isotope data are also available (see Figs 4-6). HIMU samples are arbitrarily

defined by having 206Pb/20'Pb.;: 20 (Fig. 4). EM-1 and EM-2 samples are alsoarbitrarily defined by their low 143Nd/144Ndand high 87Sr/86Srvalues as shown on

this diagram. The arrow points to the composition 01 average continental crust

("Sr/S6Sr = 0.72, 143Nd/14'Nd= 0.5118). Also marked are the compositions repre-

senting the primitive mantle (PRIMA) and the proposed common mantle compo-

nents 01 most plumes 'FOZO'" and 'C'". See Supplementary Inlormation lar

relerences used in compiling Figs 3-7.

excellent tracer for recycIed basaltic material in the mantle andcurrent results have generaIly been interpreted in this way. MantIexenoliths from ancient subcontinental lithosphere have consistentIylower-than-primitive 1870S/1880Svalues", and this constitutes anexcellent potential tracer for recycIed ancient lithosphere. However,no such low Os values have been found in oceanic basalts as yet.Noble gas isotopes. The radiogenic nucIides 4He and 4°Ar areproduced by the decay of U and Th, and 40K, respectively, andmay be degassed from the Earth's interior into the atmosphere. Theaccumulation of40Ar in the atmosphere will be further discussed inthe section on mass-balance considerations. Oceanic basaIts alsocontain the non-radiogenic isotopes 3He and 36Ar (as well as Ne, Krand Xe). Atmospheric helium is not recycIed into the mantIebecause it is continually lost to space from the atmosphere, andnearly aIl 3He now coming out the mantIe is primordial, thusdemonstrating that the Earth has never been completely degassed.(This nearly universally accepted view has been challenged byAnderson", who suggested that cosmic-dust-derived 3He has beensupplied to the mantIe through subduction of pelagic sediments.However, high 3He/4He ratios do not correlate with other tracers ofrecycIed subducted sediments in OIB38, a fact that seriouslyweakensAnderson's argument.)

Relative to the atmospheric 3He/4He ratio (RA == CHe/4He)atm= 1.4 X 10-6

), continental crust has low ratios eHe/4He= 0.01 RA)

because it is enriched in Th and U, which produce 4He during decay.MORB have rather uniform values of (8 ± l)RA (ref. 39), and OIBrange from 5 to 30RA (ref. 40). The existence ofhigh-3He/4He islands(Hawaii, Iceland, Bouvet, Galápagos, Easter, Juan Femandez,Pitcairn, Samoa, Reunion and Heard islands; see, for example,ref. 40) is consistent with the layered-mantle model: plumes

NATURE I V0L38s116JANUARY 1997

might rise directly from the lower, high-jHerHe reservoir throughthe upper, low-3He/4He layer38,41-43,or they might start from thebase of the upper layer, in which case the helium must migrate intothe plume source from the lower mantle+".

Neon isotope data from basalts are still scarce because of experi-mental difficulties. Recent work has shown that Ne from theHawaiian plume and from MORB have different 21Ne/22Ne and2°Ne/22Ne correlations, indicating the presence of a high-20Nereservoir" in the plume source and a high- 21Ne reservoir in theMORB source". Both components form mixing lines with alow_21Ne/22Ne and 2°Ne/22Ne reservoir of atmospheric origino Thisprovides additional evidence for a layered mantle with restrictedexchange between layers.Isotope taxonomy. The deviations from simple linear correlations inFigs 3-6 are not random but show systematic regularitíes':":",which are illustrated in these figures by colour coding. For example,all samples possessing 206Pbp04Pb~ 20 (Fig. 4) are shown as reddots and labelled 'HIMU' (see key in Fig. 3). They also form well-defined clusters on isotope diagrams not involving the element lead(Figs 3, 5, 6), even though they do not form an obvious endmemberin them. Samples with this'isotopic colour' occur in several oceanicislands in completely different parts of the world. This isotopic-chemical regularity demonstrates the existence of reproducibledifferentiation processes in the mantle and justifies the isotopictaxonomy discussed in the following paragraphs.

Zindler and Hare suggested that the mantle contains the follow-ing isotopically extreme 'mantle components' (Figs 3-5), whichcontribute to the (usually mixed) sources of oceanic basalts. (1)HIMU ('high p.:; p..== 238U1204Pb) has the highest Pb ratios and thelowest 87Sr/86Srof any OIB (almost as low as MORB). Examples areSt Helena, Austral Islands, Balleny Islands and the Azores. (2) EM-l('enriched mantle 1') occupies the lower left-hand comer of the Sr-Nd arra6, (Fig. 3), and is concentrated in the right-hand comer ofthe208Pb*/2 6Pb*-Nd array (Fig. 6), and the upper left-hand comer ofthe Pb-Pb array (Fig. 4b). Representatives are the Pitcairn andTristan hotspots. (3) EM-2 ('enriched mantle 2') has the highest87Sr/86Sr and relatively high 207Pb/204Pbratios. Representatives arethe Societies and Samoa hotspots. (4) DMM ('depleted MORBmantle') is not specifically marked on the figures. It is defined by themost depleted MORB samples with the highest 143Nd/144Ndand thelowest 87Sr/86Sr,206Pb/204Pb,207Pb/ 204Pband 208Pb/204Pbratios.

Many of the isotope data arrays for individual islands or hotspotsare roughly linear, indicating that they represent mixtures of a more

review article

enriched and a more depleted component originally thought to bethe DMM. However, recently published data (for example, refs 38,48, 49) show arrays in which DMM is clearly not a mixing end-member. This has led to the definition of a new component, calledrozo ('focal zone') 16,whose composition is defined bythe point ofconvergence, in three-dimensional isotope diagrams, of linear dataarrays for individual ocean islands. A similar point of convergenceof isotope arrays for MORB defines the component e ('common')40.In the two-dimensional diagrams shown here, the convergence ofHIMU and MORB arrays is most clearly seen in Fig. 4b.

There is much current debate about the meaning of this taxon-omy. Authors disagree as to whether the linear isotopic trends seenin many hotspots are caused by the existence of two components inthe plume source?", whether isotopically distinct but intemallyuniform plume sources are mixed during ascent by entrainingrelatively depleted lower-mantle material":", possibly producing azoned plume", or whether similarly uniform plumes are mixed onlywith upper-mantle asthenosphere and/or lithosphere". In general,individual hotspots contain only a limited range of isotopic com-positions and most hotspots have distinct isotopic 'flavours. On amuch larger geographical scale, EM-l and EM-2 compositions areconcentrated in OIB just south of the Equator. This feature has beencalled the 'DUPAL anomaly'" and has been correlated with large-scale seismic anomalies in the lowermost mantle. AIso, nearly allIndian Ocean MORB have distinctly high 208Pb/204Pb,87Sr/86Srand208Pb*/206Pb*values relative to Pacific and Atlantic MORB (Figs 4-6). All of this provides strong evidence that the different isotopicflavours are not randomly distributed in the mantle but can bemapped. Nevertheless, conclusive interpretations are hard to comeby because we cannot look into the mantle directly but must rely onsimulations" or correlations between geophysical and geochemicalobservations. Until recently, the spatial resolution of both geo-physical observations such as seismic tomography and numericalconvection 'experiments' has been insufficient to 'see' or modelrelatively small features such as plumes. However, this resolution isimproving rapidly and much progress can be expected within thecoming decade.

Quite distinct from the spatial distribution ofplume sources andother heterogeneities within the mantle is 'Darwin's question': whatis the origin of the isotopic mantle species? To answer this, we mustconsider other aspects of mantle chemistry, in particular theconcentration ratios of highly incompatible elements, whichcharacterize the different species and should help to identify the

a 15.9 b

L 4115.81·-

.o l' .c 40a. 15.7 o,.,.. .,..oC\I o 3915.6 C\I-.....o -....

.oa..15.5 o,

38r-,o CXlC\I o

C\I

15.4 37

15.3 I ¡ 1 ____

16 20 21 22 3616 17

206Pb / 204Pb

Figure4 Pb isotopes lar the samples shown in Fig. 3, using the same colour codeo

The solid line labelled "geochron" in a marks the locus 01 possible primitive

mantlevalues assuming an overall age of tne mantle 014.50 Gyr (rels 27,122). uee

NATURE I V0L38s116 JANUARY 1997

18 19 20

206Pb /204Pb

21 22

and Lec mark average eompositions 01 upper and lower continental erust,

respectívely'".

223