Mantle Plumes and Continental Tectonics

Author(s): R. I. Hill, I. H. Campbell, G. F. Davies and R. W. Griffiths

Source: Science, New Series, Vol. 256, No. 5054 (Apr. 10, 1992), pp. 186-193
Published by: American Association for the Advancement of Science
Stable URL: https://www.jstor.org/stable/2876982
Accessed: 22-08-2018 17:38 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

American Association for the Advancement of Science is collaborating with JSTOR to

digitize, preserve and extend access to Science

This content downloaded from 168.176.5.118 on Wed, 22 Aug 2018 17:38:25 UTC
All use subject to https://about.jstor.org/terms
 ARTICLES - -

Mantle Plumes and Continental Tectonics

R. 1. Hill, 1. H. Campbell, G. F. Davies, R. W. Griffiths

Mantle plumes and plate tectonics, the result of two distinct modes of convection within the hot-spot tracks that may have formed from
Earth, operate largely independently. Although plumes are secondary in terms of heat plume heads and tails during the past 250
transport, they have probably played an important role in continental geology. A new plume million years. However, extrapolation
starts with a large spherical head that can cause uplift and flood basalt volcanism, and may (14) from the fluid dynamics experiments
be responsible for regional-scale metamorphism or crustal melting and varying amounts on composition-driven plumes showed
of crustal extension. Plume heads are followed by narrow tails that give rise to the familiar that the volume of the head of such a
hot-spot tracks. The cumulative effect of processes associated wfth tail vokcanism may also plume formed in the mantle will be rela-
significantly affect continental crust. tively small [diameter of -300 km for a
plume having a density contrast Ap of 0.1
g cm-3 and rising from the CMB (14)]
compared with the extent of CFB provinces
Plate tectonics, the "quantitative descrip- observations can provide an important test (diameters of -2000 km).
tion of the kinematics of the lithosphere" of the model. Plumes can result also when a layer of
(1), has provided a conceptual framework fluid becomes buoyant upon being heated,
of the processes that operate at the mar- Historical Development and experiments have shown that the heads
gins of the tectonic plates that cover the of thermally driven plumes enlarge consid-
Earth's surface. This framework has, how- Wilson (11) was the first to suggest that erably more during ascent than those of
ever, proven less useful when it comes to lines of ocean islands such as the Hawaii- compositionally driven plumes. Enlarge-
processes operating away from plate Emperor chain formed when the Earth's ment occurs because the plume entrains a
boundaries (1). mobile surface passed over a fixed region of boundary layer heated by its passage (Fig.
Two prominent features that have defied relatively hot mantle (a hot spot) where 1A); the heads of thermal plumes are esti-
a plate tectonic explanation are the oceanic large amounts of magma were produced. mated to reach diameters of about 1000 km
volcanic (or hot-spot) chains and continen- Morgan (6) showed that hot spots on sev- if they start at the CMB (4, 15). These
tal flood basalt (CFB) provinces. Both may eral plates had not moved discernibly rela- results have been used to argue that the
contain basalts derived through the melting tive to each other and suggested that these plumes that give rise to both the CFBs and
of mantle that is unusually hot relative to were regions where plumes of hot material ocean islands arise from a thermal boundary
the mid-ocean ridge source and thus are ascended from the core-mantle boundary layer at the base of the mantle (5). Further
likely to result in large part from tempera- (CMB). He noted also that some hot spots support for this suggestion comes from the
ture rather than compositional variations incould be traced back along volcanic chains correspondence between the measured heat
the mantle (2). The CFB provinces and to often distant flood basalt provinces, a content transported by plumes and the in-
their oceanic equivalents, the basaltic oce- number of which appeared immediately to ferred heat flux across the core-mantle
anic plateaus, which were erupted at rates precede continental rifting. Early fluid dy- boundary (8).
possibly as high as 22 km3/year (3), are namics analog experiments (12) showed
arguably the most spectacular magmatic that a low-viscosity plume initiating in
features on the planet. A model of mantle Earth's mantle would ascend as a spherical A B
plumes (4) appears to explain successfully a pocket of fluid fed by a pipe, that is, that
wide range of observations relating to both the earliest stage of a plume would be the
ocean island chains and flood basalt prov- development and rise of a large-volume
inces (5). Mantle plumes most plausibly head, which would be trailed by a narrow
arise from a hot thermal boundary layer and at possibly long-lived conduit (or tail)
the base of the mantle (6-8), whereas through which buoyant source material
plates are part of the cool thermal boundary could be added continuously to the ascend-
layer at the top of the convecting mantle ing head (see Fig. IA).
(9). Plates and plumes are thus complemen- Morgan (13) suggested that flood ba-
tary, each driving a distinct form of mantle salts formed through melting of the heads
convection (8), and seem to operate largely of new (starting) plumes whereas the hot-
independently (10). spot chains were derived through melting
In this paper we assess the importance of of hot material rising in the long-lived
mantle plumes in initiating thermal and tails that trailed and ultimately superseded Fig. 1. Photographs of a starting plume in
structural reworking of the continents and the starting-plume phase. This suggestion, glucose syrup. (A) effects of entrainment into

use the geological record to infer the extent which was based on the results of experi- the rising vortex immediately before the onset
of near-surface spreading; (B) the plume after
of any interactions between plumes and ments on plumes driven by buoyancy re-
lateral spreading has approximately doubled
plates. We then use an example from con- sulting from compositional (and thus den-
the head diameter. In these experiments hot
tinental geology to illustrate how geological sity) differences between the plume source source material was dyed blue, and appears
layer and overlying fluid (12), was taken black or dark gray in the photographs, while
The authors are with the Research School of Earth
Sciences, Australian National University, Post Office up by Richards et al. (14), who identified entrained surrounding material remains light-
Box 4, Canberra, ACT 2601, Australia. ten couples of flood basalt provinces and colored (4).

186 SCIENCE * VOL. 256 * 10 APRIL 1992

This content downloaded from 168.176.5.118 on Wed, 22 Aug 2018 17:38:25 UTC
All use subject to https://about.jstor.org/terms
-------------------- ............. ----- ---

Dynamical Framework (Tp) of 1300? to 1550?C origin of(2).

oceanic [(Tp
volcanic plateaus (20),
is defined
by McKenzie and Bickle raises the
(2)question:
as the"What else happens
temper-
Fluid dynamics experiments can be used (4) ature that mantle material when a plume rises beneath a continent?"
would have if
to show that at the time of separation from raised to the Earth's The experiments allow
surface alongseveral inferences
an adia- to
a hot low-viscosity source layer the diame- bat.] On the basis of be made.
the Twomodel,
types of plume average
provinces
ter (D) of a new plume head varies as D - Tp's in plume heads are shouldestimated
be recognizable, depending
to be on of
Q215(v/gaAT) 1/5, where Q is the volume intermediate values, - 1350? to 14000C, whether the crust passes over a plume head
flux from the source layer into the ascend- whereas hot source material rising in the or a plume tail. Plume head provinces
ing plume head, v is the kinematic viscosity plume conduit can have a Tp as high as should be equant and 1500 to 2500 km
of the overlying mantle, g is the accelera- 15500C. Surface uplift due to the positive across, whereas volcanism and tectonism in
tion due to gravity, a is the coefficient of buoyancy of the hot plume material has plume tail provinces should be restricted to
thermal expansion, K iS the thermal con- been quantified [(4, 17, 18) Fig. 21 at as a narrow (<300 km wide) linear belt, that
ductivity of the mantle, and ATS is the much as ~- 1000 m. This amount of uplift is a hot-spot track.
temperature excess of plume source material. can lead to extension of overlying crust Both types of plume province may con-
Similarly, once separated from the source (19) and is predicted to precede initiation tain basaltic magmatism, including high-
layer the diameter of the ascending head of voluminous basaltic volcanism by 3 to 30 Mg varieties (picrites, komatiites) that re-
increases as D(z) - Q5 (v/gaATs) 1"5K2/5z35, million years (5, 17); the time interval is sult from melting of high-Tp material as-
where z is the distance traveled. For condi- particularly dependent on the viscosity of cending within the plume conduit (5, 21).
tions appropriate for the modem Earth, the upper mantle assumed in the model. Both may be associated with extension.
including buoyancy fluxes comparable with Lateral variations in the rheology of the Uplift is expected to predate both extension
those feeding modem hot-spot tracks, and uppermost mantle, especially those associ- and initiation of basaltic magmatism in
for constant material properties throughout ated with different lithospheric thicknesses, both types of plume province. Upward
ascent, plume heads are predicted to attain are predicted to play an important role in transfer of heat from the plume may initiate
diameters of 800 to 1200 km during ascent determining both the minimum depth partial melting of overlying continental crust
through the mantle. Changing the assumed reached by rising plume heads, as well as (22). Conductive heat transfer into overly-
viscosity of the lower mantle from 1022 Pa s many of the details of plume-lithosphere ing mantle and crust may play an important
to 1021 Pa s, a value considered more interaction (4, 17). role in the development of both types of
appropriate for a hotter Late Archean plume province over time scales greater than
Earth, decreases this estimate to 600 to 800 Continental Plume Provinces -10 million years because it can lead to
km (4). Varying the values assumed for weakening (and thus easier extension) of the
source flux (Q) or source temperature (ATe) A number of observations on CFB prov- crust and uppermost mantle.
over likely ranges (103 to 105 N s-1 and inces, including their diameters of -2000 Finally, because of the thickness (100 to
2000 to 8000C, respectively) likewise results km, evidence for uplift before eruption of 200 km) and depth beneath the surface (50
in changes of the estimated head diameter the basalts, and the compositional and iso- to >100 km) of the hot plume layer, the
of only a few hundred kilometers (4). An topic characteristics of erupted basalts, are time scale for cooling is considerable (of
increase in the thermal expansion coeffi- explained satisfactorily by the starting order hundreds of millions of years). The
cient (a) with height in the mantle (16) plume model (5). Acceptance of the plume slow subsidence resulting from the cooling
will result in larger initial plume head vol- model for the origin of ocean island chains of a plume emplaced beneath continental
umes (4) but in less growth by entrainment and CFB provinces (5, 14), and for the crust could result in the formation of long-
during ascent. The final diameter of a lived sedimentary basins (5). We term the
plume head is rather insensitive to the entire process of thermal and structural
precise values of properties such as a, and it ` 600 reworking above mantle plumes "plume
E oO
is difficult to avoid the conclusion that the 5
tectonics."
m.y.
heads of plumes originating at the CMB The most definitive evidence for plume
co ~~~~~~~~-25
will be large features. 0
involvement in continental reworking is
Upon nearing the surface, a plume head W ~~~~~~~~~~~~5 m.y. the presence in the crust of magmas derived
originating at the CMB spreads (Fig. 1B) to through melting of anomalously hot mantle
give a disc of hot material predicted to be (21). Modem-day picrites, which derive
500
1500 to 2500 km across and 100 to 200 km from mantle with a potential temperature
thick (4, 17). This is approximately the size 11 = 1022
up to 250?C hotter than that which yields
of a number of CFB provinces, supporting -3 m.y. mid-ocean ridge basalts (2), and komatiites,
the validity of the model (5). By way of E1000 which were derived from even hotter regions
contrast, a thermally driven plume head in the Archean mantle (21), have been
a.
originating at the 650-km seismic disconti- interpreted as the products of decompression
1500- melting of hot material ascending in plume
nuity and driven by a typical hot-spot buoy- 1500 ~~~~~~~~~-25 m.y.
ancy flux is calculated to reach a diameter of conduits (4, 21). Thus, in searching the
only 600 km after near-surface spreading (4, geological record for examples of continental
17). Slow-moving plume heads fed by fluxes 2000
plume provinces, we have focused initially
much less than the smallest fluxes observed on areas where picrites or komatiites are
(8) are not expected to reach the surface found interbedded with basalts produced
because they become sheared out by the flow from much cooler mantle and where there
associated with plate movement (5). 1000 500 0 500 1000 has been more or less synchronous eruption
For-the modemn Earth, the maximum Horizontal of crust-derived melts. (kin)
distance
temperature variation within basalt sources Fig. 2. Diagram showing the calculated surface There are examples of both types of
appears to be -250?C, equivalent to a uplift above a rising plume head. Modified from plume province. Regions that we infer to
range in mantle potential temperatures (17); m.y., million years. have resulted from impingement of a plume

SCIENCE * VOL. 256 * 10 APRIL 1992 187

This content downloaded from 168.176.5.118 on Wed, 22 Aug 2018 17:38:25 UTC
All use subject to https://about.jstor.org/terms
head include the flood basalt provinces of voluminous) basaltic volcanism, whereas preted in terms of the starting plume model,
the Karoo of southern Africa, the Deccan melting of the much hotter material that such as the Deccan and the North Atlantic
of westem India, and the North Atlantic ascended in the axial conduit produced the Tertiary Province (resulting respectively
Tertiary province (5, 13, 14), and the Late komatiites. Conduction of heat upward re- from initiation of the Reunion and Iceland
Archean granite-greenstone terranes of the sulted (after a time delay of -25 million hot spots (5, 13, 14). We infer that this
Yilgarn Block of Westem Australia (21, years) in the production of crustally deriveddifference is a consequence of the earlier
22); plume tail provinces equate with the melts. This first anatectic episode, which extraction of low-melting components from
familiar hot-spot tracks such as that associ- produced both the early (2685 to 2690 Ma) the lower and middle crust of these regions.
ated with the Yellowstone plume or with granites and volcanic rocks, was volumi- The magmatic records of the Yilgam and
lines of volcanic islands such as the Hawaii- nous but apparently of brief duration, per-Karoo differ fundamentally from those of
Emperor chain. haps only several million years. Continued consumptive margins (that is, island and
Plume head provinces. Of the plume head conduction of heat into the crust resulted in continental arcs) (Fig. 3). The key features
provinces, the Yilgarn and Karoo, in par- the production of later anatectic melts, of the Yilgarn craton are its equant form;
ticular, have extended magmatic histories including a second major episode having scale (>800 km); presence of komatiites;
consistent with conductive heat transfer high SiO2, K20, and Na2O contents, low magmatic sequence of basalt, granite, and
upward from the top of plume heads that mafic mineral abundances, and negative Sr later diverse magmatism; the occurrence of
had presumably ascended to near the base and Eu anomalies, all interpreted as indi- brief but voluminous episodes of magma-
of the crust. Both are clearly plume-related cating derivation from originally structural- tism separated by periods of quiescence; and
because of the high mantle temperatures ly higher material that underwent partial the evidence that the felsic rocks were
implied by the presence of either komatiitesmelting at lower temperatures and pressures derived through partial melting of conti-
(Yilgam) or picrites (Karoo). than those prevailing during the first epi- nental crust rather than from subducted
In the southeastern Yilgarn province, sode. The youngest granites have character- oceanic crust.
eruption of a thick (to 8 km) sequence of istics (they are alkali-rich, water-poor, and These features contrast with the situa-
basalts that contains komatiite and komati- may contain fluorite or alkali pyroxenes) tion in magmatic arcs clearly associated
itic components began at -2715 million often interpreted as indicating a high-tem- with subduction, which usually have an
years ago (Ma). The basalts are underlain perature origin (25). The end result of Late elongate form; lengths of thousands but
by felsic volcanic rocks that were erupted at Archean reworking has been the produc- widths of only a few hundreds of kilometers;
2940 Ma. These rocks provide direct evi- tion of typical internally differentiated, lay- no clear evidence for the presence of high-
dence that the basalts were erupted onto ered stable crust (26). Later episodes of temperature mafic magmas; a magmatic
continental crust (22). Further evidence for basaltic magmatism (for example, during continuum from gabbro through granite,
the presence of continental basement emplacement of a widespread suite of gab- with more or less simultaneous eruption of
comes from the presence of xenocrystic broic dykes at -2400 Ma) are not associat- compositionally diverse rocks for tens of
zircon in felsic volcanic rocks, granites and ed with important crustally derived magma- millions of years; and an inferred origin for
basalts; and from lead isotope compositions tism, presumably because components with the more mafic rocks through dehydration
of granites, and from ores from the basaltic low melting points had been removed from melting of amphibolite [Fig. 3 (31, 32)].
sequences (22, 23). The basaltic rocks are the crust in Late Archean time (22, 27). Although some rocks that may have formed
overlain by felsic volcanic rocks, which The presence of Late Archean granites through dehydration melting of amphibo-
began erupting at 2688 ? 8 Ma, and vol- throughout the Yilgarn craton shows that lite (the so-called tonalite-trondjemite-
canogenic sedimentary rocks. Both basaltic this crustal reworking event (22) affected granite suite) do occur in Yilgarn craton,
and felsic volcanic sequences are intruded profoundly an approximately equant area they are of only minor importance and may
by granites of similar (2685 to 2690 Ma) age -800 km across. If this was the result of a have formed through partial melting of
to the felsic volcanic rocks; felsic magma- plume event, and if plumes commonly play buried greenstones rather than through
tism was followed by minor mafic magma- an important part in initiating continental melting of subducted oceanic crust (22).
tism. This history provides direct evidence breakup (13, 28), then this is likely to be a Other continental magmatic terrains
that there was a thermal anomaly in the fragment of an originally larger area. such as the Mid-Proterozoic magmatism of
mantle that overlapped in both time and The much younger Karoo sequence of the United States midcontinent and of
space with the production of large volumes southern Africa has many similarities. A northern Australia (33) and the Late Cale-
of felsic melts. More granite was emplaced brief period of basaltic volcanism at 193 ? donian granites of Scotland and Australia
during a second major period of felsic mag- 5 Ma (29, 30) produced thick sequences of (34) (Fig. 3), which are difficult to interpret
matic activity 2660 to 2665 Ma, and small, picrites near the inferred position of the in plate tectonic terms, have similarities
isolated granitic intrusions have ages as plume track (5). In the Lebombo area, the with the Yilgarn (in particular, geographic
young as 2600 ? 10 Ma. The overall time basalts are overlain by felsic volcanic rocks form, scale, and derivation through partial
scale for magmatism is thus -100 million of crustal derivation; these have ages of melting of older continental crust). We
years. Derivation of the granitic rocks -177 Ma (29) and are interbedded with, suggest that reworking above plume heads
through partial melting of preexisting con- and overlain by, basaltic rocks. Small syen- deserves consideration as a potentially im-
tinental crust now appears to be accepted ite intrusions are as young as 130 to 135 Ma portant process in the evolution of the
by most workers (22, 24). Evidence for this (29, 30). Felsic volcanism is only important observed continental crust.
includes lead isotope compositions, the dif- off the craton, presumably because the low- Plume tail provinces. The most obvious
ficulty of producing large volumes of K-rich er part of the relatively young crust that feature of plume tail provinces is the pres-
granite directly from the mantle, and the encircled the craton had not been stripped ence of a time-progressive chain of volcanic
presence of old xenocrystic zircons in a of anatectic components and also possibly centers that may be many thousands of
number of samples (22). because the thinner lithosphere beneath kilometers long, corresponding to plume
These observations have been interpret- this younger crust permitted easier ascent of tails that may have persisted for as long as
ed as follows (21, 22). Partial melting in the hot plume head (28). 200 million years (13). The longevity of the
the bulk of a rising and spreading plume Felsic magmatism is much less important main volcanic interval at any given site on
head resulted in an early period of (often in many other areas that have been inter- the chain is predicted to be related to the

188 SCIENCE * VOL. 256 * 10 APRIL 1992

This content downloaded from 168.176.5.118 on Wed, 22 Aug 2018 17:38:25 UTC
All use subject to https://about.jstor.org/terms
 XARTICLES

rate at which the overlying plate passes inces (Yellowstone, Afar, Iceland, Reunion, years [see (40) for details]. For comparison,
across the hot-spot conduit. The province Tristan, Tasmania-Ferrar, Karoo, Siberian inclusion of an extra 16 small (R = 800 km)
resulting from the Yellowstone plume (the traps) have risen beneath predominantly plumes decreases the estimated repeat time
Snake River Plain) provides one on-land continental crust during the past 250 millionto 500 million years. Sixteen new plumes
example. The Plain is a volcanic-filled gra- years; another eight, probably smaller, ex- rising beneath continents in 250 million
ben atop a broad arch (35). Early rhyolitic amples (East Africa, Raton, Eastern Austra- years implies a global total of 43 (Fig. 4) if
volcanism is followed by basaltic volca- lia, Trindade or Martin, New England, Me- we assume that plume initiation is indepen-
nism, a main phase of crust-derived mag- teor-Discovery-Bouvet, Australian North- dent of near-surface processes (10). Al-
matism, and a final episode of plain-flood- west Shelf, Fernando) have also been sug- though the values for plume head radius used
ing basaltic volcanism. The entire volcanic gested (39). The actual number of new in these calculations derive from the fluid
history takes -8 million years at any one plumes could be considerably greater thandynamics experiments, the close correspon-
locality (36), consistent with passage of this. Our assessment is meant to be indica- dence between predictions and the observed
North America at 25 km per million years tive rather than definitive; we use these bestdimensions of one of the inferred surface
over a plume conduit 200 km across. examples to show the likely bounds that manifestations of plume heads (the CFB
Although lateral spreading of material might be placed on the repeat time (40). provinces) is taken as indicating that the
from the top of the conduit may result in the Eight large (radius R = 1200 km following major uncertainties in estimating the repeat
production of a strip of hot plume material near-surface spreading) plume heads in 250 time result not from poor estimates of plume-
-1000 km wide (8, 37), a striking charac- million years implies a repeat time for the head radii but from incomplete knowledge of
teristic of continental plume tail provinces rise of a plume head beneath a particular the rate of plume initiation and of the
(as well as of oceanic hot-spot chains) is the spot on the Earth's surface of 1300 million size-frequency distribution of plume heads.
restriction of major tectonism and magma- years; addition of a further eight medium- A reasonable estimate is thus that a piece
tism to a narrow (commonly <200 km) sized (R = 1000 km) plume heads brings the of continental crust will be affected by a
volcanic- and sediment-filled trough atop a estimated repeat time down to 800 million plume head event on average every 500 to
broad arch. Thermal subsidence of the arch
may result in later and much more wide-
Plume head Continental margin
spread sedimentation. Formation of the gra-
ben associated with the Snake River Plain Peninsular Sierra
Eastern Ranges Nevada Lachlan
has been suggested to result from removal of Karoo Yilgam batholith batholith fold belt

previously underlying crust through melting Age, Age,

and eruptive dispersal (35). Ma Age, Ma Age,
Age, 260-_77~ Ma 40Ma
Ma2604
Plumes and Crustal Reworking 20 360
100 2620 60
How important is the reworking of conti-
40 380
nental crust above a plume head, and how 120 2640 80
frequently might such reworking occur? EAST
6040
Both questions can be addressed with the
140 2660 8 100
geological record. The first question has
80 420
been answered above. The model developed
for the Yilgam craton carries the implication
that thermal and structural reworking of 100 >- 440
180 2700 k140
continental crust above a plume head can
12046
play an important part in the development
of that crust. Answering the second question 200 2720 ~~~~~~~~~~~160
requires an estimate of the frequency with 140 WET480
which a plume head might rise beneath a 220 2740 ~~~~~~~~~~~180
particular piece of continental crust. 160 500
Rate of plume-irutiated crustal reworking. 240 200 --
Estimation of the average time (the repeat 180

time) between plume events affecting a giv-

Fig. 3. Stratigraphic columns summarizing the
en piece of crust is complicated because: (i)
magmatic history of the Yilgarn and Karoo, two
two-thirds of the Earth is covered by oceanic
inferred plume head provinces, and compar- F7X7 Hlgh-Tfelsic
crust, which has a mean age of only 53 Ma ing them with two continental margin arcs, the magmatism
(38); (ii) the effects of plumes (particularly
Peninsular Ranges (PRB) and Sierra Nevada Fei *amts
weak plumes) may be difficult to distinguish (SNB) batholiths of western North America, Fesicmagmatism
in the geological record; (iii) the easily and with the Lachlan Fold Belt of southeastern i?e.?- Combined gabb
observable record (the past 200 million Australia. Extent of gabbroic and intermediate I Intermediate magmatism
years) may not be representative of longer magmatism shown for the PRB and SNB is
time periods; (iv) plume heads may overlap; diagramatic only, and is intended to show that Basalt, gabbro
the proportion of more mafic magmatism de-
and (v) the very processes expected above a
creases through time (54). The magmatic k - Komatlites
plume head (uplift and erosion followed by
record of the plume head provinces, which p Picrites
slow subsidence) tend to obliterate obvious
show the sequence mafic, felsic, mafic, felsic (Yilgarn only), and late high-temperature
plume-related features such as volcanism. felsic magmatism, is in marked contrast to the record of continuous, compositionally
An assessment of likely plume head- variable magmatism characteristic of the continental margin arcs, but is similar to that of
plume tail couples indicates that at least the Lachlan, an area that has proven difficult to interpret in terms of plate tectonics. Data
eight major plumes resulting in CFB prov- from (22, 29, 30, 54).

SCIENCE * VOL. 256 * 10 APRIL 1992 189

This content downloaded from 168.176.5.118 on Wed, 22 Aug 2018 17:38:25 UTC
All use subject to https://about.jstor.org/terms
Fig. 4. Diagram showing the rela- Total number of plumes initiated derived from a thermal boundary layer at
tion between frequency of plume in past 200 million years the CMB (5, 8, 44, 45), although this is
initiation and plume repeat time 100 50 40 30 20 15 also controversial, and to carry through the
(40) for a range of inferred plume mantle -6% of the heat ultimately lost at
head diameters. 200 Ma is cho-
the Earth's surface (8, 37). When the plume
sen for comparison because it is -
head plus tail fluxes are combined (2.8 to 3.3
approximately the age of the m 2000
x 1012 W, or 5.2 to 6.5 mW/M2), the total
longest lived present-day plumes /
(13, 28). Radii (R) are for plume 0 8 large continental amount of heat transported through the
heads before initiation of lateral O plumes in 250 mantle by plumes is equivalent to 7 to 9% of
0
spreading, which approximately million years the mantle-derived surface heat loss [38 x
doubles the plume head diame- E . . 1012 W; 74 mW/M2 (8, 37)].
ter. Details of the calculation are CD The ultimate driving force for plate tec-
given in (40); see text for discus- 1000 tonics is loss of heat from the mantle, where-
sion. The repeat times shown c as plumes appear to be driven by heat loss
here correspond to the time re- /C/
from the core (8, 9, 37, 44, 45). The areally
quired for plumes to affect 63% of 8 large and 8 medium averaged heat fluxes estimated for the two
the area under consideration. continental plumes
boundary layers [18 to 22 mW/M2 for the
3g 2 plumes l CMB; 74 mW/M2 for the mantle component
0 2 4 6 8 10 12 14 of surface heat flow (8, 37)] differ by only a
Average time interval between plume factor of 3 or 4. Their tectonic manifesta-
initiation events (in millions of years)
tions are so different because of the temper-
ature dependence of mantle rheology. The
800 million years (Fig. 4). The geochrono- the upper mantle by plumes. For plume hot boundary layer above the core has re-
logical record of the continents shows appar- heads these are V = 2 x 103 m3/s, M = 6.6 duced viscosity, which gives rise to plume
ent episodicity, with periods 100 to 200 x 106 kg/s, and Q = 0.5 to 1.0 x 1012 W heads and tails, whereas the cool surface
million years containing many dated samples (1.0 to 2.0 mW/M2) for a repeat time of 500 boundary layer forms stiff plates that subduct
separated by much longer periods of mag- million years; and V = 1.4 x 103 m3/s, M as sheets and drive the plate-scale convec-
matic quiescence (41). Although these data = 4.6 x 106 kg/s, and Q = 0.4 to 0.7 x tive flow. The poor spatial correlation be-
are usually interpreted as measuring the 1012 W (0.7 to 1.4 mW/M2) for a repeat tween hot spots and plate boundaries (10)
times of continental growth, we suggest that time of 800 million years. This compares indicates that the two modes of convection
the geochronology of the continents records with estimates of V = 2.2 x 103 m3/s, M = operate largely independently.
the combined effects of crustal growth 7.3 x 106 kg/s and Q = 2.3 x 1012 W (4.5 However, the geological record indi-
(which occurs predominantly at consump- mW/M2) for plume tails (8, 37). The repeat cates that there is at least some interaction
tive plate margins) and crustal reworking, times for heads are less than for tails; total between plumes and plates. The clearest
including that above plume heads. heat flux is also less, and volume and mass example is the observation that continental
Crustal reworking above plume tails is fluxes are comparable. Much of the differ- breakup is often immediately preceded by
likely to be much less important. The ence in estimated heat fluxes between heads the development of flood basalt provinces
methodology of Davies (8) can be used to and tails results from a lower estimated (6, 13, 46). Data from a number of such
estimate that plume-related features are average temperature for plume heads. provinces shows that there is a continuum
produced at a rate of about 106 km2 per If the total mass flux (heads plus tails) for (Fig. 5), from those where initiation of a
million years. The mean age of oceanic plumes estimated on the basis of the record new spreading ridge followed soon after
crust (area: 310 x 106 kM2) is 53 million of the past 250 million years (1.2 to 1.4 x basalt eruption (for example, North Atlan-
years (38), so that --15% of the present 107 kg/s) is representative, then the plume tic Tertiary Province; Deccan; Parana-
ocean floor has been affected by hot-spot flux alone is capable of turning over the Etendeka) through those where a new
activity (or, 1% of the oceans is affected by upper mantle (mass = 1.27 x 1024 kg) on a ocean basin formed only after a significant
hot-spot activity every 3 million years). If time scale of 3000 million years. (15 to 40 million year) time lag (for exam-
the preserved record is representative, and ple, Afar, Karoo, Newark-Palisades) to sit-
if continents are as likely to be affected as Plumes and Plates uations where no new ocean formed (for
are oceans, then the calculated repeat time example, the Siberian traps).
of 300 million years is in broad agreement One further question is the extent and Observations suggest that for the modem
with earlier estimates (42). However, these nature (if any) of the interaction of plumes Earth the rise of a plume by itself cannot
values are calculated for the entire area with the convective system driving. the initiate continental breakup. But in situa-
inferred to be underlain by plume material, tectonic plates. Although there is still tions where the plate-scale motions are suit-
the width of which is much greater [up to much controversy regarding the origin of ably arranged, the extra horizontal devia-
about 1000 km (8, 37)] than the width the forces that move the plates, models of toric stresses generated by uplift above a
(commonly <200 km) of any major geolog- whole-mantle convection driven by the rising plume may at times be of sufficient
ical feature clearly related to the underlying negative buoyancy of a cold, rigid surface magnitude to allow rifting to either proceed
hot spot. Because of this, the repeat time for boundary layer (principally the oceanic more rapidly or result in transfer of the axis
significant reworking above a plume tail is lithosphere) are consistent with a wide of spreading to as near as is possible to the
probably of order 1000 to 2000 million years; variety of observations (44). Regardless of plume center. For example, spreading within
plume tail provinces are thus predicted to be the details of the driving mechanism, this the South Atlantic propagated rapidly north-
a less important feature in the geological plate-scale convective mode is responsible ward following the rise of the plume head
record than plume head provinces. for most of the heat loss from the mantle, which gave rise to the Parana-Etendeka CFB
The estimates of repeat time can also be and >80% of the heat loss from the Earth province. Examples of plume-initiated ridge
used to calculate (43) the flux of volume (9, 44). The plumes that give rise to hot- jumps (28) include the opening of the North
(V), mass (M), and heat (Q) carried into spot chains (plume tails) are inferred to be Atlantic following ascent of the Iceland

190 SCIENCE * VOL. 256 * 10 APRIL 1992

This content downloaded from 168.176.5.118 on Wed, 22 Aug 2018 17:38:25 UTC
All use subject to https://about.jstor.org/terms
 ARTICLES

plume, and the rifting of the Seychelles away

Fig. 5. Diagram showing the Time interval between volcanism and rifting
from India following amval of the Reunion
length of the interval between y Beginning of volcanism Y Beginning of rapid oceanic spreading
plume arrival (as shown by the
plume, the head of which is inferred to be the
first record of mafic volcanism) Ridge jump Continental rifting
source of the Deccan basalts (5, 14).
and initiation of oceanic spread- Parania-Etendeka Newark volcanism
When a plume rises beneath a continent
ing for six Mesozoic or Tertiary AT very small AT = 25 Ma
far from an active spreading ridge, spread-
examples. Stipple shows approx- v
ing may be delayed or not occur at all. Theimate extent of volcanic activity. ::I
sequence of events recorded along the Although the beginning of slow 132 120 201 175
southeastern margin of the United States extension between southern Afri- Deccan Ethiopian
AT =3 Ma AT =25 Ma
before the opening of the Central Atlanticca and Antarctica began at 180 +
(28, 47) may provide an example of delayed5 Ma (56), coincident with a peri-
spreading. Here subsidence (and slow ex- od of felsic volcanism (and thus 67 60 30 5
tension) was under way by 230 Ma, presum- rheologically weak lower crust) in North Atlantic Tertiary Karoo
the eastern Karoo at -177 Ma, AT= 6 Ma AT = 40 Ma
ably in response to plate-scale motions.
rapid oceanic spreading was not v v * Slow extension
Cessation of sedimentation in many areas
initiated until much later, 150 to
210 to 215 Ma may be interpreted (28) as 152 Ma. Data from (13,28-30, 46, 63 52 193 -178 151
resulting from uplift presaging the arrival of 47, 55).
the Fernando (Newark-Palisades) plume
head at 201 ? 2 Ma; oceanic crust began
forming about 175 Ma. The present posi- tions of extrusive basalt and intrusive gab- useful also in providing a framework for the
tion of this plume has been suggested to be bro. We attempted to evaluate the possibil- interpretation of complex observations. For
the weak Fernando de Noronha hot spot off ity that basalts derived through melting of example, the sequence of events associated
the northeastern coast of South America ascending plume material may be a signifi- with the opening of the Central Atlantic
(28). The time lag between plume arrival cant contributor to continental growth. It (early subsidence, uplift, basaltic volcanism,
and continental breakup is explicable in has been suggested also that obduction of and formation of a new ocean after a lag of
terms of the numerical experiments of the oceanic equivalent of the CFB prov- 25 million years) is consistent with that
Houseman and England (19), who showed inces, the oceanic plateaus, may be an predicted for the rise of a starting plume
that breakup (or "runaway extension") oc- important mechanism for growth of the beneath a slowly extending continent; this
curs when an appropriate combination of continental crust (20, 48). Sandwell and type of success serves to focus attention on
rheology and elevation is attained (28). Schubert estimated an upper bound to this key predictions of the plume model (such as
Extrapolation from the fluid dynamics mechanism of about 4 km3/yr (48); actual prevolcanism uplift), which are not made by
experiments indicates that the maximum rates most likely are much lower than this other models [such as the passive rifting
elevation likely to be attained above a large and extremely difficult to evaluate. model (46), for which the predicted time
plume head is 500 to 1000 m, unless lateral We suggest that although neither CFB scales are inappropriate (5, 28)].
displacement of lithosphere becomes impor- provinces nor obducted oceanic plateaus An example that illustrates the advan-
tant (5, 17); continued ascent of hot source such as Wrangellia (20) appear to form tages of a predictive model concems the
material up a strong axial conduit may in- more than a minor component of the con- later Tertiary evolution of the western
crease the ultimate amount of uplift associ- tinental mass, their relative areal impor- United States. If much of the magmatic and
ated with plume emplacement. However, it tance may underestimate the real contribu- tectonic complexity of this region results
appears that even in situations where slow tion they make to crustal growth. Both from the rise of two starting plumes (Raton
extension is under way before arrival of the CFBs and obducted oceanic plateaus under- -34 Ma; Yellowstone 17.5 Ma (35), then
plume, the magnitude of the extra forces go rapid weathering and erosion. The dom- the model predicts that the areas now oc-
generated above the plume head are incapa- inant products of weathering are clay min- cupied by, respectively, the southem and
ble of forcing breakup unless the crust is erals, which are light enough to resist sub- central Rocky Mountains should have been
already hot, or unless sufficient heat can be duction and thus will remain as part of the uplifted before eruption of the first plume-
conducted upward from the plume head to continental mass. Crustal addition rates related basaltic rocks. The record of sedi-
alter substantially the rheology of the over- from eruption of CFBs may be 0.1 to 0.3 mentation in the Gulf of Mexico (49)
lying mantle and crust. The time scale for km3/year for short time periods (-5 million provides evidence that this was the case-
this appears to be comparable to the time years); long-term rates are 0.02 to 0.05 uplift of the southem Rocky Mountains was
taken to raise the lower crust to its melting km3/year (or 0.6 to 1.5 x 1011 kg/year) forunder way by 37 Ma, whereas uplift of the
point, that is, 15 to 40 Ma (22). This the eight CFB provinces since 250 Ma. At central and northern Rocky Mountains be-
analysis also accounts for why breakup, this rate, plume volcanism could contribute gan a little later, at 23 to 24 Ma. Other
when it does occur, is often localized along a few percent of the present continental evidence points to uplift of the southem
relatively young orogenic zones: the thinner mass (2.3 x 1022 kg) in 3800 million years. Rocky Mountains commencing at about 38
and weaker lithosphere typically associated Although CFB extrusion and contempora- to 40 Ma (50). The time intervals of -6
with younger crust allows the plume to neous intrusion has probably played no million years between initiation of uplift
ascend farther (28); thus increasing uplift as more than a minor role in the growth of theand the first appearance of basaltic magma-
well as shortening conductive distances. continental crust, it could be an importanttism in both the southem and central
contributor of elements enriched in basalts,Rocky Mountains may be related to the rW$e
Crustal Production such as nickel and chromium. of two new plume heads. If the plume
model is appropriate, for R = 500 km before
Do plumes contribute directly to continen- Looking Ahead spreading and AT = 100?C, a time interval
tal growth? For example, Hill et al. (22) of 6 million years implies that the upper
estimated that as much as 15% of the Late A quantitative dynamically-based model mantle had a viscosity of 1 to 2 x 1020 Pa
Archean crust of the southeastern Yilgarn such as that now being developed for man-s (17). We infer from the presence of
craton resulted from contemporary addi- tle plumes is not only testable, but may be (probably subduction-related) volcanism

SCIENCE * VOL. 256 * 10 APRIL 1992 191

This content downloaded from 168.176.5.118 on Wed, 22 Aug 2018 17:38:25 UTC
All use subject to https://about.jstor.org/terms
 . ge

that predates the inferred times of plume in plume-lithosphere interaction, is needed Inter. 33, 304 (1983).
8. G. F. Davies, J. Geophys. Res. 93,10467 (1988).
arrival that this area was already underlain for more precise modeling.
9. _ , ibid., p. 10451.
by unusually hot (and thus low-viscosity) Finally, we have as yet only a poor 10. M. Stefaniac and D. M. Jurdy, ibid. 89, 9919
uppermost mantle. The low value of 1 to 2 understanding of the likely extent of the (1984); D. M. Jurdy and M. Stefaniac, Geophys.
Res. Lett. 17, 1965 (1990).
x 1020 Pa s calculated from the observed interaction between a rising plume and
11. J. T. Wilson, Can. J. Phys. 41, 863 (1963); J. T.
time interval is one commonly associated subduction. The geological record of the Wilson, Philos. Trans. R. Soc. London Ser. A 25,
with unusually hot mantle (51). This ac- continental margin of the northwestern 145 (1965).

cord between the calculated viscosity and margin of the United States does give some 12. J. A. Whitehead and D. S. Luther, J. Geophys.
Res. 80, 705 (1975); J. N. Skilbeck and J. A.
the likely low value inferred from the geo- idea of one possible situation where a plume Whitehead, Nature 272, 499 (1978).
logical observations provides support for the head rose through a subduction zone, one 13. W. J. Morgan, in The Sea, C. Emiliani, Ed. (Wiley,
starting plume hypothesis, and suggests that expression of which is the Cascades mag- New York, 1981), vol. 7, pp. 443-487.
14. M. A. Richards, R. A. Duncan, V. E. Courtillot,
further application of the model in an matic arc, developed for at least the past 50 Science 246, 103 (1989).
attempt to unravel the forces responsible for Ma as the Juan de Fuca Plate has been 15. R. W. Griffiths, J. Fluid Mech. 166, 139 (1986);
the past 40 million years of geological evo- subducted beneath western North America Earth Planet. Sci. Lett. 78, 435 (1986); Phys. Earth
Planet. Inter. 33, 304 (1986).
lution of this region may be productive. (53). A postulated Yellowstone plume head
16. A. Chopelas and R. Boehler, Geophys. Res. Lett.
An alternate mechanism for large-scalenearing the top of its ascent 17 to 18 16,1347 (1989).
continental reworking is that of Hima- million years ago must have reached the 17. R. W. Griffiths and 1. H. Campbell, J. Geophys.
layan-type continent-continent collisions. Res. 96, 18295 (1991).
upper part of the descending slab. At about
18. R. W. Griffiths, M. Gurnis, G. Eitelberg, Geophys.
Although this may result in the structural 18 Ma the rate of magmatic activity within J. 96, 477 (1989).
and thermal reworking of broad areas many this arc slowed considerably, and when it 19. G. Houseman and P. England, J. Geophys. Res.
thousands of kilometers across (1, 52), it increased again 4 million years later, mag- 91, 719 (1986).
20. M. A. Richards, D. L. Jones, R. A. Duncan, D. J.
might not be expected to include the sub- matism had shifted to more mafic composi- DePaolo, Science 254, 263 (1991).
stantial and early basaltic volcanism that is tions (53), consistent with the introduction 21. I. H. Campbell, R. W. Griffiths, R. I. Hill, Nature
a common characteristic of plume head of hot material from the plume head into 339, 697 (1989).

provinces. Other features (52) of Hima- the mantle wedge overlying the downgoing 22. I. H. Campbell and R. I. Hill, Earth Planet. Sci. Lett.
90, 11 (1988); R. I. Hill, I. H. Campbell, W.
lyan-type events (for example, maximum slab. Geometric considerations suggest that Compston, Geochim. Cosmochim. Acta 53,1259
temperature attained within the crust and part of the plume head may have spread (1989); R. I. Hill, I. H. Campbell, R. W. Griffiths,
Explor. Geophys. 22,185 (1991); R. I. Hill and 1. H.
relative timing of major structural and mag-westward as far as the Juan de Fuca ridge.
Campbell, Aust. J. Earth Sci., in press; , B.
matic events) are also likely to differ signif-This example suggests that a better knowl- W. Chappell, in The Archaean: Terranes, Crustal
icantly from those associated with plumes. edge of plume-slab interactions may lead to Processes and Metallogeny, J. Glover and S. E.
In summary, each mechanism for initiating a better understanding of the tectonic and Ho, Eds. (Geology Department and University
Extension, University of Western Australia, in
large-scale thermal reworking of continen- magmatic complexities of some arcs. press); R. I. Hill, B. W. Chappell, I. H. Campbell,
tal crust is likely to have a characteristic The apparently successful application of Trans. Roy. Soc. Edinburgh: Earth Sci., in press.
combination of time scales and magmatic the mantle plume model to the origin of 23. V. M. Oversby, Geochim. Cosmochim. Acta 40,
817 (1976); W. Compston, I. S. Williams, I. H.
evolution that could allow identification. continental flood basalt provinces, oceanic Campbell, J. J. Gresham, Earth Planet. Sci. Lett.
Three areas of uncertainty limit detailed plateaus, and hot-spot tracks carries the 76, 299 (1986); J. C. Claou6-Long, W. Compston,
application (or evaluation) of the plume implication that mantle plumes may play an A. Cowden, ibid. 89, 239 (1988).
24. W. K. Witt and C. P. Swager, Precambr. Res. 44,
model. First, improved resolution of the important role in continental geology. We
323 (1989); K. F. Cassidy, M. E. Barley, D. I.
predicted time interval between uplift and suggest that further candidates for plume Groves, C. S. Perring, J. A. Hallberg, ibid. 51, 51
initial magmatism requires better knowl- head provinces include the Trans-Hudson (1991).

edge of the values and variability of upper orogen of the North American Proterozoic, 25. W. J. Collins, S. D. Beams, A. J. R. White, B. W.
Chappell, Contrib. Mineral. Petrol. 80,189 (1982);
mantle viscosity, as well as a better under- the Proterozoic anorogenic magmatism of J. B. Whalen, K. L. Currie, B. W. Chappell, ibid.
standing of the process by which plumes the American midcontinent and of north- 95, 407 (1987).
penetrate the lithosphere [see (17) for a ern Australia, and the 'Late Caledonian' 26. B. J. Drummond and C. D. N. Collins, Earth
Planet. Sci. Lett. 79, 361 (1986); B. J. Drummond,
discussion]. The latter point in particular is granite terrains of Scotland and of south- Precambr. Res. 40/41, 101 (1988).
at present poorly understood. Plume pene- eastern Australia. The Midcontinent Rift 27. I. H. Campbell and G. T. Jarvis, Precambr. Res.
tration may involve thermomechanical ero- of North America may be a candidate for a 26,15 (1984).
28. R. I. Hill, Earth Planet. Sci. Lett. 104, 398 (1991).
sion, heat transport by magma, and the Proterozoic hot-spot track. Mantle plumes
29. F. J. Fitch and J. A. Miller, Geol. Soc. S. Africa
development of gravitational instabilities in may also exercise considerable control over Spec. Publ. 13, 247 (1984); H. L. Allsopp, W. I.
the lower part of the lithosphere. Indeed, the properties of the top half of the subcon- Manton, J. W. Bristow, A. J. Erlank, ibid. 13, 273
although both the compositions of inferred tinental upper mantle. (1984).
30. H. L. Allsopp, J. W. Bristow, C. T. Logan, H. V.
plume-derived basalts and the observed Eales, A. J. Erlank, ibid., p. 281.
time intervals between initial basaltic mag- REFERENCES AND NOTES 31. T. H. Green and A. E. Ringwood, Earth Planet. Sci.
matism and the beginning of voluminous Lett. 1, 307 (1966); Contrib. Mineral. Petrol. 18,
1. P. Molnar, Nature 335, 131 (1988). 105 (1968); C. W. Burnham, in Geochemistry of
crust-derived magmatism appear to require 2. J.-G. Schilling, ibid. 242, 565 (1973); D. McKen- Hydrothermal Ore Deposits, H. L. Barnes, Ed.
that the top of the plume head ascends to zie, J. PetroL. 25, 713 (1984); and M. J. (Wiley, New York, ed. 3, 1979), pp. 71-136.
within a short distance (-10 km) of the Bickle, ibid. 29, 625 (1988); C. Langmuir, paper 32. A. R. McBirney and C. M. White, in Andesites:
presented at the Caltech Conference on Mantle Orogenic Andesites and Related Rocks, R. S.
base of the crust (22), the mechanism by Plumes, Pasadena, CA, 2 to 4 May 1991. Thorpe, Ed. (WNiley, New York, 1982), pp. 115-
which this is accomplished is unknown. 3. J. A. Tarduno et al., Science 254, 399 (1991); M. 135; L. T. Silver and B. W. Chappell, Trans. Roy.
Second, calculation of the uplift and F. Coffin, paper presented at the Caltech Confer- Soc. Edinburgh Earth Sci. 79, 105 (1988).
ence on Mantle Plumes, Pasadena, CA, 2 to 4 33. L. 7. Silver et al., Geol. Soc. Am. Abstr. Prog. 9,
extension history of crust above a rising
May 1991. 1176 (1977); M. E. Bickford, W. B. Van Schmus, I.
plume head is limited by knowledge of the 4. R. W. Griffiths and 1. H. Campbell, Earth Planet. Zietz, Geology 14, 492 (1986); M. E. Bickford,
temperature-dependence of rheology (19). Sci. Lett. 99, 66 (1990). Geol. Soc. Am. Bull. 100, 1375 (1988); L. A. I.
5. I. H. Campbell and R. W. Griffiths, ibid., p. 79. Wyborn, Precambr. Res. 40141, 37 (1988).
Improvement in our knowledge of the ma-
6. W. J. Morgan, Nature 230, 42 (1971); Mem. Geol. 34. Writing about the Late Caledonian (430 to 390
terial properties of upper mantle and lower Soc. Am. 132, 7 (1972). Ma) granites of Scotland, W. E. Stephens and A.
crustal rocks, and of the processes involved 7. D. E. Loper and F. D. Stacey, Phys. Earth Planet. N. Halliday [Trans. Roy. Soc. Edinburgh Earth

192 SCIENCE * VOL. 256 * 10 APRIL 1992

This content downloaded from 168.176.5.118 on Wed, 22 Aug 2018 17:38:25 UTC
All use subject to https://about.jstor.org/terms
 ARTICLES

Sci. 75, 259 (1988)] commented that "The rela- 46. R. White and D. McKenzie, J. Geophys. Res. 94, 54. L. T. Silver, H. P. Taylor, Jr., B. W. Chappell, in
tionship of these late granitoids to the collision 7685 (1989). Mesozoic Crystalline Rocks, P. L. Abbott and V. R.
process is obscure, and hitherto their ages and 47. A. J. Froelich and G. R. Robinson, Jr., Eds., U.S. Todd, Eds. (Guidebook, Geological Society of
compositional variations have not accorded with Geol. Surv. Bull. 1776 (1988). America, Denver, 1979), pp. 83-1 10; J. H. Chen
those predicted by various plate models." B. W. 48. G. Schubert and D. Sandwell, Earth Planet. Sci. and J. G. Moore, J. Geophys. Res. 87, 4761
Chappell and W. E. Stephens (ibid., p.71), com- Lett. 92, 234 (1989). (1982); M. D. Barton et a/., in Metamorphism and
paring granites of similar (Late Caledonian) age in 49. W. E. Galloway, Am. Assoc. Pet. Geol. Bull. 73, Crustal Evolution of the Western United States, W.
Scotland and southeastern Australia with the 143 (1989). G. Ernst, Ed. (Prentice-Hall, Englewood, NJ,
Mesozoic continental margin batholiths of western 50. D. P. Elston and R. A. Young, J. Geophys. Res. 96, 1988), pp. 110-178; P. G. Stuart-Smith, R. I. Hill,
North America, concluded: "The Paleozoic gran- 12389 (1991); K. M. Campion, S. R. Morgan, J. M. M. J. Rickard, M. A. Etheridge, Tectonophysics, in
ites show little evidence of such a direct relation- Lohmar, Geol. Soc. Am. Abstr. Prog. 23,11 (1991); press.
ship to subduction." See also A. N. Halliday and J. M. Lohmar and S. R. Morgan, ibid., p. 74. 55. R. G. Bohannon, C. W. Naeser, D. L. Schmidt, R.
W. E. Stephens, Phys. Earth Planet. Inter. 35, 89 51. Recent estimates of average upper mantle vis- A. Zimmerman, J. Geophys. Res. 94,1683 (1989).
(1984); M. Owen and D. Wyborn, Austral. Bur. cosity range from -5 x 1019 to 5 x 1020 Pa s; the 56. D. Rowley, personal communication.
Min. Res. Bull. No. 204 (1979). lower values come from tectonically younger re- 57. Our geochronological work in Western Australia
35. J. Suppe, C. Powell, R. Berry, Am. J. Sci. 275A, gions such as the Pacific Ocean, and southern has received substantive support from Australian
397 (1975). and eastern Australia. This latter region, the site of Consolidated Minerals, Aztec Exploration, BP
36. R. L. Armstrong, W. P. Leeman, H. E. Malde, ibid. intermittent and scattered basaltic volcanism for Minerals, Central Norseman Gold Corp. Ltd., Hill
275, 225 (1975). the past 70 million years, yields an estimate of 2 to Minerals, Pancontinental, Placer Exploration Ltd.,
37. N. H. Sleep, J. Geophys. Res. 95, 6715 (1990). 3 x 1020 Pa s. M. Nakada and K. Lambeck, in West Coast Holdings, and Western Mining Corpo-
38. B. Parsons, ibid. 87, 289 (1982). Glacial Isostasy, Sea-Level and Mantle Rheology, ration. R.I.H. acknowledges support from an Aus-
39. Evidence for many of the inferred plume heads is R. Sabadini, K. Lambeck, E. Boschi, Eds. (Kluwer, tralian Research Council Fellowship. We are
Dordrecht, 1991), pp. 79-94. grateful to W. Compston for access to the
from [5, 6, 13, 14, 20, 28, 461, R. A. Duncan [J.
52. P. C. England and A. B. Thompson, J. Petrol. 25, SHRIMP ion microprobe. We thank N. T. Arndt, B.
Geophys. Res. 89, 9980 (1984)], and N. H. Sleep
[J. Geophys. Res. 95, 21983 (1990)]. We infer that 894 (1984); S. M. Peacock, Geol. Soc. Am. Bull. W. Chappell, W. Compston, M. A. Richards, A. E.
magmatism -124 Ma in the White Mountains and 101, 476 (1989). Ringwood, S.-S. Sun, J. S. Turner, and D. Wyborn
Monteregian Hills resulted from the plume head 53. A. R. McBirney, J. F. Sutter, H. R. Naslund, K. G. for discussions, as well as D. L. Anderson for
phase of this hot spot. Tracks for Discovery- Sutton, C. M. White, Geology 12, 585 (1974); E. D. organizing the Caltech Mantle Plume Conference
Bouvet-Meteor are presented by R. A. Duncan; Verplanck and R. A. Duncan, Tectonics 6, 197 at which a number of the ideas incorporated in
White and McKenzie (46) link basalts in the Cape (1987); G. R. Priest, J. Geophys. Res. 95, 19583 this paper were first presented. We thank S. B.
Basin to Bouvet, and we infer that initiation of this (1 990). Shirey and N. H. Sleep for thorough reviews.
plume family resulted in the extended period of
kimberlite eruption in southern Africa that began
about 128 Ma [see also S. T. Crough, W. J.

Hybridization of Bird Species

Morgan, R. B. Hargraves, Earth Planet. Sci. Lett.
50, 260 (1983)]. Data for Eastern Australia are
summarized by R. A. Duncan and 1. McDougall [in
Intraplate Volcanism in Eastern Australia and New
Zealand, R. W. Johnson, Ed. (Cambridge Univ.
Peter R. Grant and B. Rosemary Grant
Press, Cambridge, 1989), pp. 43-54]; three and
possibly four lines of volcanic complexes in east-
Hybridization, the interbreeding of species, provides favorable conditions for major and rap
ern Australia and the adjacent Tasman Sea start-
evolution to occur. In birds it is widespread. Approximately one in ten species is known to
ed at 36 Ma, with the earliest volcanism centrally
located with respect to an oval-shaped topo- hybridize, and the true global incidence is likely to be much higher. A longitudinal study of
graphic uplift that extends across much of central
Darwin's finch populations on a Galapagos island shows that hybrids exhibit higher fitness
Queensland and east to the offshore Queensland
Plateau. Data in support of a possible Raton than the parental species over several years. Hybrids may be at an occasional disadvantage
plume are given by Suppe et al. (35) or are for ecological rather than genetic reasons in this climatically fluctuating environment. Hy-
mentioned in the text.
bridization presents challenges to the reconstruction of phylogenies, formulation of biological
40. The simple repeat time (TR) is calculated from the
relationship TR = S/NA, where S is the area of the species concepts and definitions, and the practice of biological conservation.
continents (1.9 x 108 kM2), N is the number of
plumes per unit time, and A is the cross-sectional
area of the plume head after near-surface spread-
ing. This estimate is used because it is more
readily compared with the observed episodicity in Species of sexually reproducing organisms hybridization have included the crossing of
the geochronologic record of the continents. An are "groups of actually or potentially in- lines in the laboratory or greenhouse for
alternate method is to use a probability calcula-
tion of the time required for plumes to affect a terbreeding natural populations which are genetical analysis, and the estimation of
specified percentage of the surface area; the reproductively isolated from other such frequencies of phenotypic or genotypic class-
definition of TR corresponds to the time for 63% groups"
of (1). Periodically attempts have been es in nature, their mating pattern, and their
the surface to be affected.
made to improve on this definition by deal- reproductive success. By themselves each is
41. R. G. Gastil, Proc. 21st Int. Geol. Congr. 9, 162
(1960); K. C. Condie, Plate Tectonics and Crustal ing inter alia with the awkward fact that for incomplete. In this article, we describe the
Evolution (Pergamon, New York, ed. 2, 1982); B. some populations the criterion of demarca- desired but rarely achieved direct study of
F. Windley, The Evolving Continents (Wiley,
tion is not absolute (2-5). Some populations hybridization in nature through pedigree
Chichester, ed. 2, 1984).
42. S. T. Crough, Tectonophysics 61, 323 (1979); W. occasionally interbreed, and then the ques- analysis. The study populations are birds.
J. Morgan, ibid. 94, 123 (1983); ibid. 61, 321 tion becomes one of determining the fates of We present new information on the conse-
(1979); Annu. Rev. Earth Planet. Sci. 11, 165 the offspring (1, 6). Therefore, hybridiza- quences of hybridization in populations of
(1983).
43. For plume tails, the volume flux (V) = B/paAT,
tion, which strictly is the interbreeding of Darwin's finches over several generations.
where B = differential mass flux [termed "buoy- species, is of pivotal importance in two
ancy flux" by Sleep (37)] = 54.9 x 103 kg/s-1 respects: in framing ideas about the nature of The Broad Patterns
(37); heat flux (Q) = pCAT. The factor of 2
variation in the estimates of Q for plume heads
taxonomic judgments to be made about par-
arises from uncertainty in estimates of the amount ticular populations (7) and more generally Mayr and Short (9) estimated that approx-
of entrainment; 0 is calculated for source per total for understanding biological processes of ev- imately 10% of 516 nonmarine species of
ratios of 0.25 and 0.5. p = 3300 kg/m-3; a = 3 x
olution including speciation (1, 2, 8). birds regularly hybridize. Meise (10) made a
10-5?C-1; C= 1.25 x 103 J/kg-1P/C-1. Increas-
ing a as a function of decreasing pressure (16) Traditional approaches to the study of broader survey and concluded that 2% of all
will have the effect of increasing the source per recent bird species hybridize regularly, and
total ratio. P. R. Grant is Class of 1877 Professor of Zoology and an additional 3% hybridize occasionally. A
44. G. F. Davies and M. A. Richards, J. Geol., in press. B. R. Grant is on the research staff of Princeton
45. F. D. Stacey and D. E. Loper, Phys. Earth Planet. University, Department of Ecology and Evolutionary
more definitive estimate can now be made.
Inter. 36, 99 (1984). Biology, Princeton, NJ 08544. In the last 2 years the total number of

SCIENCE * VOL. 256 * 10 APRIL 1992 193

This content downloaded from 168.176.5.118 on Wed, 22 Aug 2018 17:38:25 UTC
All use subject to https://about.jstor.org/terms