ELSEVIER

Earth and Planetary Science Letters 163 (1998) 97108

Episodic continental growth and supercontinents:

a mantle avalanche connection?
Kent C. Condie *
Department of Earth and Environmental Science, New Mexico Institute of Mining and Technology, Socorro, NM 87801, USA
Received 14 May 1998; revised version received 1 August 1998; accepted 12 August 1998

Abstract
Episodic growth of continental crust and supercontinents at 2.7, 1.9, and 1.2 Ga may be caused by superevents in
the mantle as descending slabs pile up at the 660-km seismic discontinuity and then catastrophically sink into the lower
mantle. Superevents, in turn, may comprise three or four events, each of 5080 My duration, and each of which may reflect
slab avalanches at different locations and times along the 660-km discontinuity. Superplume events in the late Paleozoic
and Mid-Cretaceous may have been caused by minor slab avalanches as the 660-km discontinuity became more permeable
to the passage of slabs with time. The total duration of a superevent cycle decreases with time reflecting the cooling of the
mantle. 1998 Elsevier Science B.V. All rights reserved.
Keywords: continental crust; mantle; plate tectonics; models; slabs

1. Introduction
Although the episodic nature of isotopic ages
has been known for nearly 40 years [1], it is only
in the last decade that episodic growth of juvenile continental crust has been recognized [24]
(Fig. 1). Although not as well documented, it appears that episodic crustal growth is also related to
the supercontinent cycle [3,4]. Breuer and Spohn
[5] and Stein and Hofmann [2] propose models by
which layered convection in the earth catastrophically changes to whole-mantle convection during
short-lived episodes as descending plates accumulate
at the 660-km seismic discontinuity and then suddenly sink into the lower mantle. When the slabs
arrive at the D00 layer, they trigger plume produc Tel.: C1 505 835 5531; Fax: C1 505 835 6436; E-mail:
kcondie@mailhost.nmt.edu

tion, and the plumes rapidly rise to the base of the

lithosphere, where they are either directly or indirectly responsible for crustal growth. Condie [4] and
Peltier et al. [6] propose similar models, although
limiting the catastrophic episodes to three: at about
2.7, 1.9 and 1.2 Ga (Fig. 1). Stein and Hofmann
[2] also discuss how this episodic, penetrative convection can explain many of the geochemical and
isotopic characteristics of the crustmantle system.
Although an episodic model involving slab
avalanching is attractive and explains an array of
both geochemical and geophysical data, several outstanding questions remain if the model is to survive.
First, why and how did the first episode of slab
sinking begin in the Late Archean? Was it driven by
the breakup of a previous heretofore unrecognized
supercontinent? Also, we are still facing the problem of why the collapse episodes appear to have
suddenly ended after the Mid-Proterozoic event at

0012-821X/98/$ see front matter 1998 Elsevier Science B.V. All rights reserved.
PII: S 0 0 1 2 - 8 2 1 X ( 9 8 ) 0 0 1 7 8 - 2

98

K.C. Condie / Earth and Planetary Science Letters 163 (1998) 97108

Fig. 2. Supercontinent cycles showing the overlap of breakup

and aggregation times (in gray pattern). Major peaks of juvenile
crust formation are shown as black triangles. The supercontinent
shown forming in the last 100 My is hypothetical. R, Rodinia; G,
Gondwana; P, Pangea. Pn, Pannotia.

Fig. 1. Distribution of U=Pb zircon ages from juvenile crust

(modified after Condie [3,4]). To avoid undue weighting of ages
from specific areas, only one age is used from a given juvenile
crust supracrustal succession or a single pluton or batholith. Because zircon ages from rocks <500 Ma are uncommon, Phanerozoic ages include isotopic ages by other methods as well as rocks
well dated by fossils.

1.2 Ga. Could the superplume events suggested by

Larson [7] in the Mid-Cretaceous and perhaps also
in the late Paleozoic represent still younger examples
of slab avalanches through the 660-km discontinuity? If so, why are isotopic age peaks not recognized at these times? And finally, does enhanced
plume production (caused by slab avalanches) correlate with the formation or with the breakup of
supercontinents? If plume maxima correlate with
supercontinent breakup and crustal growth, as sometimes assumed, why are crustal growth episodes also
episodes of collisional orogeny, which would more
logically correlate with supercontinent formation?
In this contribution, these and other questions regarding the episodic nature of crustal growth are
addressed and a model is presented that incorporates
a combination of episodic crustal ages, supercontinent cycles, and catastrophic events in the mantle.

2. The supercontinent cycle

In the last 1 Gy, the formation and breakup of
three major supercontinents has been documented
(Rodinia, Gondwana and Pangea), with a possible
short-lived supercontinent, Pannotia, in the latest
Proterozoic [810]. Geologic data support the existence of at least two earlier supercontinents, one

at the end of the Archean and one in the Early

Proterozoic [1113]. As more precise ages for the
breakup and aggregation of supercontinents have become available, it seems clear that in the supercontinent cycle, one supercontinent is accompanied in
part or entirely by the formation of another supercontinent (Fig. 2). For instance, Gondwana was forming
at about the same time (650 and 550 Ma) as Rodinia
broke up (700530 Ma) [9,10,14]. Also, the breakup
of Pangea in the last 160 My is accompanied by
several major terrane and microcontinent collisions
(India, North American Cordillera, SE Asia), which
may represent the beginnings of a new supercontinent. Perhaps this is not surprising or unexpected, if
we consider that supercontinents fragment over large
mantle upwellings (geoid highs) and the fragmented
blocks travel to mantle downwellings (geoid lows),
where they collide with other blocks to form a new
supercontinent [6,15]. Thus, collisions occur over the
geoid lows while supercontinents are still fragmenting over geoid highs. Because there is always an
external ocean as supercontinents come and go, the
supercontinent cycle can be viewed as the expansion
and contraction of this ocean.
Another aspect of the supercontinent cycle, which
is not well understood, is the question of just which
forces are most important in fragmenting supercontinents. Most modeling of mantle plumes has concluded either that they cannot fragment large supercontinents, or they can fragment them only when the
lithosphere is already under stress [1618]. Perhaps
the most convincing energy source for the breakup
of supercontinents is large mantle upwellings [6,19].
It is known, for instance, that Pangea was centered
over such an upwelling when it began to fragment
160 Ma [15], and this upwelling, perhaps reduced in

K.C. Condie / Earth and Planetary Science Letters 163 (1998) 97108

size, is still present beneath Africa [19]. If indeed,

mantle upwellings are a necessity for the breakup of
supercontinents, how are they produced? Some models have suggested that the continental lithosphere
acts a thermal blanket, and the mantle beneath such
lithosphere heats up, eventually producing a mantle
upwelling that breaks the supercontinent [20]. However, it is known that the thermal insulation effect of
lithosphere is quite small, and probably lags behind
a plate shielding effect, which can result in heating the mantle beneath both continental and oceanic
plates if they are large [21]. Shielding by large plates
can result in the production of a large mantle upwelling beneath the plates, simply because the plates
shield the underlying mantle from the cooling effects of subduction [19]. The shielding mechanism is
also appealing in that today large mantle upwellings
are recognized beneath both the African and Pacific
plates, the first dominated by continental lithosphere
and the latter by oceanic lithosphere.

3. Production of juvenile continental crust

Juvenile continental crust is produced at two tectonic settings: subduction zones and mantle plumes.
The former is most important for the upper continental crust and the latter perhaps, for the lower continental crust [22]. The energy for crust production in
both tectonic settings is tied to the earths thermal
budget at any given time. In slab avalanching models, the increased production rate of juvenile crust
associated with each avalanche is generally related
to mantle plumes triggered when the sinking slabs
arrive at the D00 layer above the core. The plumes
give rise to juvenile crust either directly, by partial
melting as they arrive at the base of the lithosphere,
or indirectly by heating the upper mantle. Heating of
the upper mantle increases the rate of production of
ocean-ridge lithosphere, and hence also the rate of
subduction of oceanic lithosphere, which in turn increases the rate of production of subduction-related
magmas.
The importance of large mantle upwellings in producing juvenile crust is not well understood. Clearly
if the upwellings penetrate the mantle solidus, melting will begin, and the widespread magmatism in
Africa in the last 135 My [23] may, in part, re-

99

flect melting of the large mantle upwelling beneath Africa. It is more likely, however, that mantle
plumes, which are concentrated within the African
upwelling [24], are responsible for most of the
African magmatism during this time interval. Although the volume of juvenile magmas added to the
African crust during this time is unknown, the fact
that most of western and southern Africa is underlain
by thick Archean lithosphere [25,26] suggests it may
be rather minor.

4. The 660-km seismic discontinuity

Most models for the episodic growth of continents involve catastrophic sinking of slabs through
the 660-km seismic discontinuity in the upper mantle [2,6,27]. Although seismic tomographic results
clearly suggest that descending slabs sink into the
lower mantle today [28], this may not have been the
case in the geologic past when the earth was hotter.
Christensen and Yuen [29] have shown that in a hotter mantle with a larger Rayleigh number, such as
probably existed in the Archean [30,31], the amount
of leakage across the 660-km discontinuity is considerably reduced, resulting in layered convection.
Computer models of mantle evolution also suggest
that increased internal heating of the mantle strongly
favors layering in the mantle [32], and this would be
the case during the Archean when heat-production
by radiogenic isotopes was more pronounced than
it is today. Layered convection in the Archean is
important in that slab avalanches would not occur.
It may have been in the Late Archean when the
660-km discontinuity became less robust that slabs
occasionally fell through to the lower mantle [2,4,6].
Cooling of the earth may also have been responsible
for the shutdown or decrease in intensity of slab
avalanches after 1.2 Ga. As the mantle temperature
and Rayleigh number decreased, slabs would more
easily penetrate the 660-km discontinuity, leading
eventually to whole-mantle convection.

5. The Mid-Cretaceous superplume event

In two important papers, Larson [7,33] suggested
that one or more superplumes beneath the Pacific

100

K.C. Condie / Earth and Planetary Science Letters 163 (1998) 97108

basin could explain a large number of geological and

geophysical features of the earth during the Mid-Cretaceous. In this model, the superplume(s) arose from
the D00 layer enhancing the production rate of juvenile crust in the form of oceanic plateaus, ocean
ridges, and arcs [33,34]. Deformation and orogenesis are also widespread in the continents around
the Pacific at this time [35]. Perhaps the geoid high
[19] and the slow S-wave velocities beneath the
Pacific [36] reflect the remains of the Cretaceous
superplumes. Another interesting feature of the superplume event is that it rapidly cooled the outer
core causing a shutdown in magnetic reversals for
about 40 My, giving rise to the Cretaceous superchron [7]. Although there was no supercontinent to
break up in the Pacific, the Cretaceous superplume
event has some characteristics of the three major
juvenile crust-forming events at 2.7, 1.9 and 1.2 Ga.
For instance, it occurred soon after the beginning
of fragmentation of a supercontinent (Pangea) and
correlates with relatively high rates of juvenile crust
production around the margins of the Pacific basin
[7,35]. Could a minor slab avalanche in the mantle
have caused it, perhaps the last avalanche as the mantle continued to cool? The reason we do not observe
a spike in isotopic ages at 130100 Ma (Fig. 1) may
be that the spike is relatively small and it is hidden
in background noise. Also, there is a question as to
whether the distribution of ages from juvenile crustal
rocks <500 Ma are representative of their volumetric
abundance.
Although not as well documented, another superchron occurs in the late Paleozoic (320250 Ma
[33]), and if superchrons reflect superplume events,
which in turn reflect slab avalanches in the mantle, perhaps there was another episode of slightly
enhanced juvenile crustal growth in the late Paleozoic [37]. In the model developed below, both the
Mid-Cretaceous and late Paleozoic events are assumed to represent juvenile-crust formation events,
similar to, but less intense than their three counterparts in the Precambrian. An interesting test of this
idea is to check for magnetic reversals in rock successions that formed at the three Precambrian age
peaks of 2.7, 1.9 and 1.2 Ga. If the model is feasible,
superchrons should occur near each of these peaks.
Although magnetic reversals have been identified in
igneous rocks formed during the 1.9-Ga peak [38

40], multiple reversals in single stratigraphic successions at or near any of the three peaks have not been
documented.

6. Analysis of the three Precambrian superevents

As high precision zircon ages have become available in the last 10 years, it is possible to analyze
in greater detail each of the three major periods
of juvenile crust formation (Fig. 1). There is some
suggestion in the isotopic ages that each peak comprises several subpeaks, although it is not possible
to clearly resolve such subpeaks. If the subpeaks are
real, they are not worldwide, but appear to characterize certain geographic regions. To further evaluate
the nature of possible subpeaks, zircon ages from juvenile greenstonegranite terranes within each of the
three major events (2.7, 1.9 and 1.2 Ga) have been
divided into pre- and syncollisional groups. Pre-collisional ages come from volcanic sequences erupted
prior to deformation in a given granitegreenstone
cycle, whereas syncollisional ages come from syntectonic granitoids, and are interpreted to represent
collisional ages of blocks of juvenile crust [3,4].
Examples of results for each of the three major
juvenile crust events, which hereafter are referred to
as superevents, are given in Fig. 3. To approach
the maximum time interval for a given greenstone
granite cycle, only the oldest pre-collisional juvenile crustal ages are plotted in the figure. The Late
Archean superevent appears to comprise three, nonoverlapping events (Fig. 3a). Only the event centered
at about 2.7 Ga is recorded on all continents. An
older, less well defined event is centered at 2.95 Ga,
and it also may be worldwide. A third event, which
has been well documented so far only in India and
the North China craton, occurs at 2.52.6 Ga. Note
that the two latter events are just above background
level on the age histogram plot (Fig. 1). With the
three events, the total duration of the Late Archean
superevent is in the order of 450500 My.
Of the three superevents, the Early Proterozoic
superevent is most complex and perhaps the most
long-lived (Fig. 3b). The most widespread event
within this superevent is centered at about 1.9 Ga and
has been well documented on all continents except
Antarctica and South America. The oldest Early Pro-

K.C. Condie / Earth and Planetary Science Letters 163 (1998) 97108

Fig. 3. Examples of pre- and syntectonic juvenile crust ages

from specific granitegreenstone cycles that comprise each of
the three Precambrian superevents. Solid symbols are maximum
pre-tectonic greenstone eruption ages and open symbols are
syntectonic ages from collisional granitoids within accurately
dated granitegreenstone cycles. References are given in Condie
[3]. Symbols in each diagram refer to samples from the different
geographic regions. Only the 2.7 and 1.9 Ga events appear to
be worldwide, and thus the triangular symbols represent many
different geographic localities.

101

terozoic juvenile crust event, centered at about 2.15

Ga, is well documented only in the West African
and Guiana shields (the Birimian event) [41,42].
Two juvenile crust events, one at about 1.751.8 Ga
and another at about 1.651.7 Ga, are widespread
in a belt extending from southwestern North America into southern Scandinavia, and probably into the
Russian platform [4345]. The 1.651.7 Ga event
(Labradorian=Mazatzal event) is just above background level on the age histogram (Fig. 1). Although
1.751.8 and 1.651.7-Ga events are also found
on other continents, whether they represent juvenile-crust-forming events on these continents is not
yet clear. As with the Late Archean superevent, the
Early Proterozoic superevent appears to span approximately 500 My, and it may comprise at least
four subevents. Subevents may more easily be resolved in the stratigraphic record of cratonic basins
and passive margins [13,46,47].
The Mid-Proterozoic superevent at 1.2 Ga is less
well defined than the two earlier superevents, due
to a lack of zircon ages from juvenile crust in MidProterozoic orogens. From the available data base,
it may comprise three events, although these events
overlap in time (Fig. 3c). An event at 1.251.32
Ga is recorded in Antarctica, South Africa, and
LaurentiaBaltica; a second event, recorded chiefly
in LaurentiaBaltica and perhaps in India and Australia occurs at about 1.2 Ga and contains the largest
number of ages (Fig. 1); and a final event, near background level at 1.01.1 Ga, is well documented as a
juvenile crust event only in LaurentiaBaltica. If the
zircon ages are representative of the Mid-Proterozoic
superevent, it would appear to be about 320 My in
duration, less than the duration of the earlier two
superevents.
Because of unequal sampling on the continents,
there may be many more subevents within each
major event, and in fact, there could be a continuum
of subevents within each event. The significance of
the time interval between the pre- and syncollisional
ages of specific subevents in Fig. 3 is not yet clear,
and in part depends on identifying and dating the
oldest volcanics within a given granitegreenstone
cycle. It is interesting that the maximum values for
this interval in all three juvenile crust episodes is of
the order of 5080 My. This age appears to represent
the time interval between the first greenstone crust

102

K.C. Condie / Earth and Planetary Science Letters 163 (1998) 97108

in a given cycle and the collision of this crust with

another crustal block? If so, could this represent the
mean age of oceanic plates at the time they subduct?
This is an intriguing possibility, and if supported by
later data, it may indicate faster spreading rates of
oceanic plates in the Precambrian, since the mean
age of oceanic plates when they subduct at present is
about 100 My [48].

7. Towards a model of mantlecrust evolution:

the superevent cycle
7.1. Introduction
The following major assumptions are made in
developing a model that ties together catastrophic
slab avalanches in the mantle, the supercontinent
cycle, and episodic production of continental crust.
(1) There are three main episodes of production
of juvenile continental crust centered at 2.7, 1.9 and
1.2 Ga, and two minor episodes, one in the late
Paleozoic (350 Ma) and one in the Mid-Cretaceous
(110120 Ma).
(2) Supercontinents form over geoid lows (mantle
downwellings) and breakup over geoid highs (mantle
upwellings). Mantle upwellings are chiefly responsible for the breakup of supercontinents.
(3) Mantle temperature has decreased exponentially with time as Th, U and K isotopes have decayed, and the Rayleigh number of the mantle has
decreased in a similar manner.
(4) Convection in the mantle has changed from
layered before about 1 Ga to chiefly whole-mantle thereafter. Three episodes of penetrative, wholemantle convection occur prior to 1 Ga.
(5) The 660-km seismic discontinuity has become
a progressively less effective barrier to the descent of
lithospheric slabs with time, as mantle temperature
decreased.
(6) During earth history, there have been three
major and perhaps two minor superevents. Each superevent involves avalanching of slabs through the
660-km discontinuity, production of mantle plumes
in the D00 layer [27,49], formation of a supercontinent, and enhanced production of juvenile crust.
(7) Each Precambrian superevent may comprise
three or more subevents lasting 5080 My each.

7.2. The model

As shown by Condie [4], earth history can be
broadly divided into three stages (Fig. 1): Stage
I (>3.0 Ga), characterized by the absence of superevents and rapid recycling of juvenile crust into
the upper mantle; Stage II (3.01.0 Ga), during
which three major superevents occurred in the mantle and led to enhanced production of juvenile continental crust; and Stage III (<1.0 Ga), during which
two rather minor superevents occurred in the mantle.
7.2.1. Stage I
The probable occurrence of coexisting depleted
and enriched upper-mantle reservoirs in the same
geographic area prior to 3.0 Ga as shown by Nd isotopic studies suggests rapid recycling of continental
crust into the mantle [50,51]. The changes in the Nd
isotopic composition of clastic sediments with time
are also consistent with such recycling [52]. Before
3 Ga, continental crust survived only as small microcontinents, some (or all) of which are captured
in Late Archean orogens and there is no evidence to
support the existence of supercontinents before 3 Ga.
As the mantle cooled and continental crust became
less easy to subduct, collisions between continental
blocks led to the formation of the first supercontinent beginning about 3 Ga and culminating with
major collisions at 2.7 Ga (Table 1; Fig. 2). Minor
growth of the supercontinent occurred in what are
now North China and India at 2.52.6 Ga.
Just what caused the three Late Archean
subevents at 3.0, 2.7 and 2.6 Ga is not clear. However, because it is likely that the upper mantle convected separately from the lower mantle, perhaps
subducted slabs were not entirely recycled in the
upper mantle, but some collected at the 660-km discontinuity. If so, by 3 Ga this discontinuity may
have destabilized because of the slab load and cooling of the mantle, and perhaps it failed at different
places and at three different times (3.0, 2.7 and 2.6
Ga), leading to three slab avalanches. The first two
avalanches may have occurred over a wide distribution on the surface of the 660-km discontinuity,
whereas the last one at 2.6 Ga, may have been
more localized, since it is represented only in China
and India. It would appear that the avalanche at
2.7 Ga was the most intense. Each slab avalanche

K.C. Condie / Earth and Planetary Science Letters 163 (1998) 97108

103

Table 1
Timing of events in the superevent cycle

Late Archean
Early Proterozoic
Rodinia
Gondwana
Pangea
a From

Supercontinent and juvenile

crust formation

Shielding and
mantle upwelling

Breakup of supercontinent

Cycle durationa

3.02.5 Ga (500 My)

2.151.65 Ga (500 My)
1.321.0 Ga (320 My)
650550 Ma (100 My)
450250 Ma (200 My)

2.62.2 Ga (400 My)

1.751.5 Ga (250 My)
1.10.7 Ga (400 My)
600160 Ma (440 My)
350160 Ma (190 My)

2.22.0 Ga (200 My)

1.51.3 Ga (200 My)
700530 Ma (170 My)
1600 Ma (160C My)
1600 Ma (160C My)

850 50 My
790 30 My
650C My
450C My

the beginning of supercontinent aggregation to the end of supercontinent breakup.

produced mantle plumes in the D00 layer, and these

bombarded the base of the lithosphere perhaps in
<10 My [53], where they underwent decompression
melting and produced juvenile crust. Plumes may
have been temporarily stalled at the 660-km discontinuity due to the displaced phase boundary within
the plumes [54]. The plumes also heated the upper mantle resulting in enhanced rates of juvenile
crust production at ocean ridges and arcs. Multiple
slab avalanching is supported by three-dimensional
computer simulations in which unsynchronized and
spatially localized flushing is observed to occur at
different locations on the 660-km discontinuity surface [55]. After the Late Archean slab avalanche, a
return to layered convection occurred as the Rayleigh
number of the boundary layer at 660-km discontinuity returned to its previous, subcritical value, as the
flux of plates crossing the boundary eliminated the
temperature gradient across the boundary [6].
When avalanches occur, they cause super subduction zones above them, which attract plates from
great distance [6], thus aiding in the growth of a
new supercontinent over the avalanche sites. Also
contributing to the growth of the supercontinent is
juvenile crust produced directly or indirectly by mantle plumes generated in D00 as the avalanches reached
the base of the mantle.
7.2.2. Stage II
During Stage II of earth history, the superevent
cycle occurs two more times at 1.9 and 1.2 Ga. After
formation of the Late Archean supercontinent, the
mantle returns to layered convection, and some fragments of descending slabs again begin to accumulate
at the 660-km discontinuity. Due to shielding by the
large Late Archean supercontinental plate, a mantle
upwelling develops beneath the supercontinent lead-

ing to breakup between 2.2 and 2.0 Ga (Table 1;

Fig. 2). As the supercontinent breaks up, fragments
move towards geoid lows, where they collide initiating the growth of a new supercontinent. With
the supercontinent breakup, subduction rates begin
to increase and perhaps the increase in slabs arriving at the 660-km discontinuity causes another slab
avalanche, which in turn initiates mantle plumes,
many of which rise within the mantle upwelling(s).
Again, the slab avalanches are indirectly responsible
for an increase in the production rate of juvenile crust
as well as contributing to the formation of another
supercontinent. The width of the 1.9 age peak may
reflect four slab avalanches at approximately 2.1, 1.9,
1.8 and 1.7 Ga, of which only the 1.9-Ga avalanche
appears to have had worldwide effects. As the Late
Archean supercontinent breaks up over a prolonged
period of about 200 My, fragments travel to geoid
lows and a new supercontinent begins to form, the
oldest portion of which is the BirimianGuiana craton, which formed at about 2.15 Ga. This is followed
by the 1.9, 1.8 and 1.7 Ga collisions, which trap
juvenile crust in orogens between colliding blocks.
The superevent cycle occurs again in the Mid-Proterozoic as shielding beneath the Early Proterozoic
supercontinent results in a mantle upwelling that
eventually breaks the supercontinent between 1.5
and 1.3 Ga. This initiates another slab avalanche resulting in a new episode of juvenile crust production
and formation of a new supercontinent, Rodinia at
1.32 to 1.0 Ga (Table 1; Fig. 2).
7.2.3. Stage III
Rodinia survives for about 600 My, when a new
mantle upwelling fragments it between about 700
and 530 Ma (Table 1; Fig. 2). The formation of
Gondwana (650550 Ma) overlaps the breakup of

104

K.C. Condie / Earth and Planetary Science Letters 163 (1998) 97108

Rodinia, and this is followed by the formation of

Pangea between 450 and 250 Ma. Pangea began
to fragment about 160 Ma, and as previously mentioned, a new supercontinent may have begun to
form in the last 100 My (Fig. 2). Perhaps a small
avalanche of slabs at the 660-km discontinuity initiated each of these events. The events were small because the relatively cool mantle with a low Rayleigh
number made it easier for slabs to penetrate the
660-km discontinuity and only small amounts of
slab material collected at this boundary before collapse occurred. The last event, the Mid-Cretaceous
superplume event, may have resulted from the last
slab avalanche in earth history. The mantle upwelling
beneath the Pacific plate may have developed in the
last 100 My because the large Pacific plate shielded
the underlying mantle from subduction. The fact that
the Pacific plate remained in approximately the same
location between 130 and 95 Ma supports a shielding
origin for the upwelling [56].
7.3. Timing of events in the superevent cycle
The superevent cycle is summarized in Fig. 4.
The timing of various events within this cycle is constrained chiefly by data from two sources: (1) isotopic ages of juvenile crust, and (2) results of computer simulations of mantle processes [3032,55].
Beginning with a supercontinent, computer models
suggest that it takes on average 200400 My for
shielding of a large supercontinental plate to cause
a mantle upwelling beneath it [21] (Table 1). The
upwelling breaks the supercontinent over approximately a 200-My period (160170 My for Rodinia
and Pangea). This breakup may be the trigger for
slab avalanching through the 660-km discontinuity.

Fig. 4. Flow chart of the Superevent Cycle.

Alternatively, the collapse of slabs may result from

a threshold for total slab mass at a given location
on the 660-km discontinuity, and in this case accumulation could be spread over several hundred My
beginning when a supercontinent fragments. From
the time a slab avalanche begins to the time juvenile
crust is produced is probably quite short, of the order
of 100 My or even less. This is because slabs can
sink to the bottom of the mantle in 100 My or less
[26,57], and in a mantle in which viscosity increases
with depth, mantle plumes can rise to the base of the
lithosphere in a few My [53]. The evidence suggests
a correlation of slab avalanches with supercontinent
formation rather than with supercontinent breakup.
The total duration of a superevent cycle is about
800 My for each of the two Precambrian cycles,
and 450650C My for the Phanerozoic cycles (Table 1; Fig. 2). The existence of three major peaks
in juvenile crust production (2.7, 1.9 and 1.2 Ga)
is consistent with the computer models of Peltier
et al. [6], which predict three major avalanches at
approximately 500 My intervals. The model herein
proposed also suggests that supercontinent breakup
and aggregation overlap, with the proportion of overlap increasing in the Phanerozoic. Except for the
overlap in breakup of the Early Proterozoic supercontinent and the formation of Rodinia, which is
only about 20 My, the overlap between breakup and
formation phases is generally in the order of 100
150 My (Fig. 2). There is nearly complete overlap in
the breakup of Rodinia and the formation of Gondwana, and perhaps in the breakup of Pangea and the
formation of new hypothetical supercontinent.
As shown in Table 1, the duration of both supercontinent formation and superevent cycles decreases
with age. The duration of supercontinent formation
(including enhanced production of juvenile crust) in
the first two superevent cycles is about 500 My,
decreasing to about 300 My in Mid-Proterozoic cycle, and then to 100200 My for each of the two
Phanerozoic cycles. Also, as a whole, the length of
a superevent cycle decreases from 850 My to 450C
My from the ArcheanEarly Proterozoic cycle to
the GondwanaPangea cycle. Paralleling these decreases is a decrease in the volume of juvenile crust
produced at each superevent. Approximately 35
40% of the present continental crust may have been
produced during each of the first two superevent

K.C. Condie / Earth and Planetary Science Letters 163 (1998) 97108

105

Fig. 5. Map of the continents showing the distribution of juvenile continental crust (modified after Condie [52]). About 75% of the
continental crust was produced during the first two superevent cycles.

106

K.C. Condie / Earth and Planetary Science Letters 163 (1998) 97108

cycles at 2.7 and 1.9 Ga, whereas only about 12%

was produced during the Mid-Proterozoic (Rodinia)
cycle (Fig. 5) [58]. The total amount of juvenile
continental crust produced during the Phanerozoic
appears to have been only 1015% of the present
volume.
What may have been responsible for decreases in
the duration of superevent cycles and in the volume
of juvenile crust produced during each cycle? The
major cause may be the decreasing strength of
the 660-km discontinuity with time. As the 660-km
discontinuity became more permeable to descending
slabs with falling mantle temperature, slabs would
begin to steadily sink through the boundary, and
thus fewer slabs would accumulate at the boundary.
Hence, when an avalanche occurred, it would be
relatively small. The small volume of plates in an
avalanche may have two effects: (1) the duration
of the superevent should be shorter, and perhaps
more limited in geographic extent, and (2) because
of the smaller mass of sinking slabs the number of
mantle plumes generated would decrease, and thus
also would the volume of juvenile crust associated
with these plumes. If the model described in this
study is viable, the decreasing duration of both the
time interval of supercontinent formation and of each
superevent cycle probably reflects the cooling of the
mantle as radiogenic isotopes have decayed.

continent (100500 My); and shielding of the mantle

beneath the new supercontinent results in a mantle
upwelling (200400 My) that eventually breaks the
supercontinent and the cycle starts over.
Each of the 2.7, 1.9 and 1.2 Ga superevents
may comprise several subevents each of 5080 My
duration. Such subevents may reflect slab avalanches
at different locations and times along the 660km
discontinuity.
Superplume events in the late Paleozoic and
Mid-Cretaceous may have been caused by minor
slab avalanches as the 660-km discontinuity became
more permeable to the passage of slabs.
The total duration of a superevent cycle is about
800 My for Precambrian cycles and 450650C My
for Phanerozoic cycles. This decrease in duration
probably reflects cooling of the mantle.

Acknowledgements
Discussions with Dallas Abbott and Brad Hager
have been useful in focusing some of the ideas presented in the paper. An earlier version of this paper
was substantially improved from in-depth reviews by
Graham Park, Rob Kerrich and Mark Barley. [RV]

References
8. Conclusions
Maxima in the production of juvenile continental
crust at 2.7, 1.9 and 1.2 Ga correspond to formation
times of supercontinents.
Each of these maxima may reflect a superevent in
the mantle as descending slabs catastrophically sink
through the 660-km seismic discontinuity.
Slab avalanches are correlated with supercontinent (and juvenile crust) formation rather than with
supercontinent breakup.
A typical superevent cycle is as follows (duration
in parenthesis): supercontinent breakup (200 My)
initiating slab avalanches and the beginning of formation of a new supercontinent; arrival of slabs at
the D00 layer at the base of mantle triggers mantle
plumes that rise and bombard the lithosphere producing juvenile crust trapped in the growing super-

[1] G. Castil, The distribution of mineral dates in time and

space, Am. J. Sci. 258 (1960) 135.
[2] M. Stein, A.W. Hofmann, Mantle plumes and episodic
crustal growth, Nature 372 (1994) 6368.
[3] K.C. Condie, Greenstones through time, in: K.C. Condie
(Ed.), Archean Crustal Evolution, Ch. 3, Elsevier, Amsterdam, 1994, pp. 85120.
[4] K.C. Condie, Episodic ages of greenstones: a key to mantle
dynamics?, Geophys. Res. Lett. 22 (1995) 22152218.
[5] D. Breuer, T. Spohn, Possible flush instability in the mantle
convection at the ArcheanProterozoic transition, Nature
378 (1995) 608610.
[6] W.R. Peltier, S. Butler, L.P. Solheim, The influence of phase
transformations on mantle mixing and plate tectonics, in:
D.J. Crossley (Ed.), Earths Deep Interior, Gordon and
Breach, Amsterdam, 1997, pp. 405430.
[7] R.L. Larson, Latest pulse of earth: evidence for a mid-Cretaceous superplume, Geology 19 (1991) 547550.
[8] C.R. Schotese, W.S. McKerrow, Revised world maps and
introduction, Geol. Soc. London Mem. 12 (1990) 121.
[9] T.H. Torsvik et al., Continental break-up and collision in

K.C. Condie / Earth and Planetary Science Letters 163 (1998) 97108

[10]

[11]
[12]
[13]

[14]

[15]
[16]
[17]

[18]

[19]
[20]

[21]

[22]

[23]

[24]
[25]

[26]

[27]

[28]

the Neoproterozoic and Paleozoic a tale of Baltica and

Laurentia, Earth-Sci. Rev. 40 (1996) 229258.
R. Unrug, Rodinia to Gondwana: The geodynamic map
of Gondwana supercontinent assembly, GSA Today 7 (1)
1997.
P.F. Hoffman, Speculations on Laurentias first gigayear
(2.0 to 1.0 Ga), Geology 17 (1989) 135138.
J.J.W. Rogers, A history of continents in the past three
billion years, J. Geol. 104 (1996) 91107.
M.E. Barley, D.I. Groves, Supercontinent cycles and the
distribution of metal deposits through time, Geology 20
(1992) 291294.
J.J. Veevers, M.R. Walter, E. Scheibner, Neoproterozoic
tectonics of AustraliaAntarctica and Laurentia and the 560
Ma birth of the Pacific Ocean reflect the 400 My Pangean
supercycle, J. Geol. 105 (1997) 225242.
D.L. Anderson, Hotspots, polar wander, Mesozoic convection and the geoid, Nature 297 (1982) 391393.
R.I. Hill, Starting plumes and continental break-up, Earth
Planet. Sci. Lett. 104 (1991) 398416.
M.F. Coffin, O. Eldholm, Volcanism and continental
break-up: a global compilation of large igneous provinces,
Geol. Soc. London Spec. Publ. 68 (1992) 1730.
B.C. Storey, The role of mantle plumes in continental
breakup: case histories from Gondwanaland, Nature 377
(1995) 301308.
G. Castil, The distribution of mineral dates in time and
space, Am. J. Sci. 258 (1960) 135.
M. Gurnis, Large-scale mantle convection and the aggregation and dispersal of supercontinents, Nature 332 (1988)
695699.
J.P. Lowman, G.T. Jarvis, Continental collisions in wide aspect ratio and high Rayleigh number two-dimensional mantle convection models, J. Geophys. Res. 101 (25) (1996)
485497.
K.C. Condie, Contrasting sources for upper and lower
continental crust: the greenstone connection, J. Geol. 105
(1997) 729736.
D.K. Bailey, Episodic alkaline igneous activity across
Africa: implications for the causes of continental break-up,
Geol. Soc. London Spec. Publ. 68 (1992) 9198.
S.T. Crough, Hotspot swells, Annu. Rev. Earth Planet. Sci.
11 (1983) 165193.
Y.S. Zhang, T. Tanimoto, High-resolution global upper
mantle structure and plate tectonics, J. Geophys. Res. 98
(1993) 97939823.
D.G. Pearson, R.W. Carlson, S.B. Shirey, F.R. Boyd, P.H.
Nixon, Stabilization of Archean lithospheric mantle: a Re
Os isotope study of peridotite xenoliths from the Kaapvaal
craton, Earth Planet. Sci. Lett. 134 (1995) 341357.
D. Brunet, P. Machetel, Large-scale tectonic features induced by mantle avalanches with phase, temperature, and
pressure lateral variations of viscosity, J. Geophys. Res. 103
(1998) 49294945.
S.P. Grand, R.D. van der Hilst, S. Widiyantoro, Global
seismic tomography: a snapshot of convection in the earth,
GSA Today 7 (4) (1997).

107

[29] U.R. Christensen, D.A. Yuen, Layered convection induced

by phase transitions, J. Geophys. Res. 90 (1985) 291300.
[30] L.P. Solheim, W.R. Peltier, Avalanche effects in phase transition modulated thermal-convection a model of earths
mantle, J. Geophys. Res. 99 (1994) 69977018.
[31] P.J. Tackley, Effects of phase transitions on three-dimensional mantle convection, in: D.J. Crossley (Ed.), Earths
Deep Interior, Gordon and Breach, Amsterdam, 1997, pp.
273336.
[32] W.L. Zhao, D.A. Yuen, S. Honda, Multiple phase transitions
and the style of mantle convection, Phys. Earth Planet. Inter.
72 (1992) 185210.
[33] R.L. Larson, Geological consequences of superplumes, Geology 19 (1991) 963966.
[34] P. Wessel, S. Lyons, Distribution of large Pacific seamounts
from Geosat=ERS-1: Implications for the history of intraplate volcanism, J. Geophys. Res. 102 (1997) 22259
22475.
[35] A.P.M. Vaughan, CircumPacific mid-Cretaceous deformation and uplift: a superplume-related event?, Geology 23
(1995) 491494.
[36] R.D. Van der Hilst, S. Widiyantoro, E.R. Engdahl, Evidence
for deep mantle circulation from global tomography, Nature
386 (1997) 578584.
[37] A.M.C. Sengor, B.A. Natalin, V.S. Burtman, Evolution of
the Altaid tectonic collage and Paleozoic crustal growth in
Eurasia, Nature 364 (1993) 299307.
[38] G.E. Morgan, The paleomagnetism and cooling history of
metamorphic and igneous rocks from the Limpopo mobile
belt, southern Africa, Geol. Soc. Am. Bull. 96 (1985) 663
675.
[39] Y. Zhai, H.C. Halls, Multiple episodes of dike emplacement
along the NW margin of the Superior province, Manitoba,
J. Geophys. Res. 99 (1994) 2171721732.
[40] K.L. Buchan, H.C. Halls, J.K. Mortensen, Paleomagnetism,
UPb geochronology, and geochemistry of Marathon dykes,
Superior Province, and comparison with the Fort Frances
swarm, Can. J. Earth Sci. 33 (1996) 15831595.
[41] W. Abouchami, M. Bohler, A. Michard, F. Albarede, A
major 2.1 Ga event of mafic magmatism in West Africa: an
early stage of crustal accretion, J. Geophys. Res. 95 (1990)
1760517629.
[42] P.J. Sylvester, K. Attoh, Lithostratigraphy and composition
of 2.1 Ga greenstone belts of the West African craton and
their bearing on crustal evolution and ArcheanProterozoic
boundary, J. Geol. 100 (1992) 377393.
[43] P.F. Hoffman, United plates of America: the birth of a
craton, Annu. Rev. Earth Planet. Sci. 16 (1988) 543603.
[44] K.C. Condie, Proterozoic terranes and continental accretion
in southwestern North America. in: K.C. Condie (Ed.), Proterozoic Crustal Evolution, Ch. 12, Elsevier, Amsterdam,
1992, pp. 447480.
[45] R. Gorbatschev, S. Bogdanova, Frontiers in the Baltic
shield, Precambrian Res. 64 (1993) 322.
[46] B. Krapez, Sequence stratigraphic concepts applied to the
identification of depositional basins and global tectonic
cycles, Aust. J. Earth Sci. 44 (1997) 136.

108

K.C. Condie / Earth and Planetary Science Letters 163 (1998) 97108

[47] M.E. Barley, B. Krapez, D.I. Groves, R. Kerrich, The late

Archean bonanza: metallogenic and environmental consequences of the interaction between mantle plumes, lithospheric tectonics and global cyclicity, Precambrian Res.
(1998) (in press).
[48] B. Parsons, Causes and consequences of the relation between area and age of the sea floor, J. Geophys. Res. 87
(1982) 289302.
[49] C.R. Neal, J.C. Ely, J.J. Mahoney, M.G. Petterson, Noble
metal chemistry of oceanic plateaus, part 2: evidence of a
coremantle boundary origin for the Ontong Java plateau,
Geol. Soc. Am. Abstr. Progr. 29 (6) A245.
[50] S.A. Bowring, T. Housh, The earths early evolution, Nature
269 (1995) 15351540.
[51] M.T. McCulloch, V.C. Bennett, Progressive growth of the
earths continental crust and depleted mantle: geochemical
constraints, Geochim. Cosmochim. Acta 58 (1994) 4717
4738.
[52] R.L. Armstrong, The persistent myth of crustal growth,

Aust. J. Earth Sci. 38 (1991) 613630.

[53] T.B. Larsen, D.A. Yuen, Fast plumeheads: temperaturedependent versus non-Newtonian rheology, Geophys. Res.
Lett. 24 (1997) 19951998.
[54] U.R. Christensen, Effects of phase transitions on mantle
convection, Annu. Rev. Earth Planet. Sci. 23 (1995) 6587.
[55] P.J. Tackley, D.J. Stevenson, G.A. Glatzmaier, G. Schubert,
Effects of an endothermic phase transition at 670 km depth
in a spherical model of convection in the earths mantle, J.
Geophys. Res. 99 (1994) 1587715901.
[56] J.A. Tarduno, W.W. Sager, Polar standstill of the Mid-Cretaceous Pacific plate and its geodynamic implications, Science 269 (1995) 956959.
[57] R.L. Larson, C. Kincaid, Onset of mid-Cretaceous volcanism by elevation of the 670 km thermal boundary layer,
Geology 24 (1996) 551554.
[58] K.C. Condie, Plate Tectonics and Crustal Evolution, 4th
ed., Butterworth-Heinemann, Oxford, 1997, 282 pp.