ridges, oceanic plateaux, and back arc basins (Smithies
et al., 2005; Pearce, 2008). However, these are prone to
overinterpretation, as they are not foolproof on the modern
Earth: for example, continental flood basalts are known to
have low-Ti characteristics normally associated with
island arc volcanism and calc-alkaline rocks that are characteristic of magmatic arcs above subduction zones also
been found in non-subduction settings (e.g., Hooper
et al., 2002; Willbold et al., 2009) are known. There is
a tendency to engrave interpretations from modern settings onto ancient rocks, although most Archean rocks
give mixed signals (Pearce, 2008). But here again, the
oldest arc-like greenstones appear to be between 3.2 and
3.1 Ga (Smithies et al., 2005). An additional option is to
use rock types characteristic for subduction zones, which
include andesites, boninites, shoshonites, and adakites.
Most of these become frequent in the late Archean
(Condie and Kroner, 2008), whereas adakite is a rock type
from modern subduction zones (island arcs and Andeantype subduction beneath continental crust) that is similar
to tonalitic and trondhjemitic gneisses that make up large
tracts of the Archean continental crust. These are thought
to indicate a subduction environment because they do
not form in modern plume-related environments (Martin,
1999).
A different type of plate tectonics or no plate
tectonics at all?
The Archean rock record consists of small continent-like
blocks of high-grade gneisses of broadly tonalitic composition and subordinate greenstone belts that are composed
mostly of volcanic rocks and texturally immature sedimentary rocks. These rocks are quite common from 3 Ga
onwards, but similar rocks are present from the beginning
of the Archean: tonalitic gneisses make up the bulk of the
Itsaq complex of western Greenland, dated at 3,800 Ma
(Nutman et al., 1999), and supracrustal rocks of similar
age occur both here and in northern Quebec
(Mloszewska et al., 2012). As we trace them back into
the Archean, several of the best plate tectonic indicators,
such as paired metamorphic belts, high-pressure,
low-temperature metamorphism, and greenstone volcanics with arc-like geochemistry, seem to “pinch out” at
about 3.1–3.3 Ga. For the beginning of plate tectonics,
no single indicator is as convincing as the testimonial
given by the accumulation of these lines of evidence at this
time. Even so, the lack of eclogites and blueschists leads
some to distinguish between Proterozoic plate tectonics
(approximately 3,300–700 Ma) and modern plate tectonics (after 700 Ma; Brown, 2007). Before 3.3 Ga, we have
only a limited number of indicators, but these tell us that
TTG magmatism occurred from the beginning of the
Archean, although with a tendency for MgO poorer,
SiO 2 -richer compositions earlier on in time. This possibly
indicates a type of shallow subduction, in which the
subducting crustal plate directly underlies the overriding
plate so that reaction of TTG melts with the mantle wedge
could not occur. As time passed and the Earth cooled, subduction to deeper levels occurred and reaction with the
mantle became more common.
However, the production of TTG melts does not necessarily denote subduction. The requirement from highpressure experiments is that these melts must be derived
by melting of basalt, but these experiments say nothing
about how this basalt reaches its melting conditions. The
main competitor hypotheses to subduction for explaining
the origin of TTG melts are oceanic plateaux and stagnant
lid tectonics. The oceanic plateau model uses the analogy
of thickened ocean crust on the modern Earth, which are
caused by extra heating from below by mantle plumes,
resulting in crust up to 35 km thick (Neal et al., 1997).
Thus, formation of some crust as oceanic plateaux could
apply to an Archean Earth with plate tectonics, but
a major uncertainty in this case becomes the behavior of
this thick mafic crust in subduction zones. Can such thick
crust be subducted as a unit, or does just part of it return to
the mantle and the upper parts accumulate at the surface?
In this scenario, the basaltic source rocks for TTG melts
are derived from thick piles of volcanics that simply
become buried to lower crustal levels because of continued volcanism.
The ocean plateau model is generally thought to work
within plate tectonics and does not seem adequate for the
early Archean if complete recycling of crust commonly
occurred. Here a stagnant lid scenario may be more appropriate, in which the thick crust is not subducted, but the
lower levels delaminate and melt back into the mantle.
Only the upper sections of crust are preserved, a scenario
known as “flake tectonics” (Hoffman and Ranalli, 1988).
In the earliest Archean or Hadean, blocks of crust may
have tipped like ice flows, thus returning extensive blocks
of crust to the mantle without the mechanism of subduction as it occurs on the modern Earth.
It is probable that the large degree of crustal recycling
implied by the paucity of surviving crust from the early
Archean is incompatible with plate tectonics in its modern
form. The transition to plate tectonics may have occurred
during the period 3.3–3.1 Ga, followed by an immense
increase in the survival of continental crust (Figure 1).
The Archean-Proterozoic boundary at 2.5 Ga is the much
more publicized major disjuncture in the geological
record, but these changes in geological style could only
have occurred after abundant continental crust had collected for long periods. The middle Archean may have
been a time in which plate tectonics operated on some
parts of the Earth, but not on others, and it may have been
intermittent (O’Neill et al., 2007).
Summary
Taking multiple indicators into account together, there is
good evidence for the operation of plate tectonics on the
Earth as far back as 3.3 Ga, but it may have been
18
ANCIENT PLATE TECTONICS
et al., 2005; Pearce, 2008). However, these are prone to
overinterpretation, as they are not foolproof on the modern
Earth: for example, continental flood basalts are known to
have low-Ti characteristics normally associated with
island arc volcanism and calc-alkaline rocks that are characteristic of magmatic arcs above subduction zones also
been found in non-subduction settings (e.g., Hooper
et al., 2002; Willbold et al., 2009) are known. There is
a tendency to engrave interpretations from modern settings onto ancient rocks, although most Archean rocks
give mixed signals (Pearce, 2008). But here again, the
oldest arc-like greenstones appear to be between 3.2 and
3.1 Ga (Smithies et al., 2005). An additional option is to
use rock types characteristic for subduction zones, which
include andesites, boninites, shoshonites, and adakites.
Most of these become frequent in the late Archean
(Condie and Kroner, 2008), whereas adakite is a rock type
from modern subduction zones (island arcs and Andeantype subduction beneath continental crust) that is similar
to tonalitic and trondhjemitic gneisses that make up large
tracts of the Archean continental crust. These are thought
to indicate a subduction environment because they do
not form in modern plume-related environments (Martin,
1999).
A different type of plate tectonics or no plate
tectonics at all?
The Archean rock record consists of small continent-like
blocks of high-grade gneisses of broadly tonalitic composition and subordinate greenstone belts that are composed
mostly of volcanic rocks and texturally immature sedimentary rocks. These rocks are quite common from 3 Ga
onwards, but similar rocks are present from the beginning
of the Archean: tonalitic gneisses make up the bulk of the
Itsaq complex of western Greenland, dated at 3,800 Ma
(Nutman et al., 1999), and supracrustal rocks of similar
age occur both here and in northern Quebec
(Mloszewska et al., 2012). As we trace them back into
the Archean, several of the best plate tectonic indicators,
such as paired metamorphic belts, high-pressure,
low-temperature metamorphism, and greenstone volcanics with arc-like geochemistry, seem to “pinch out” at
about 3.1–3.3 Ga. For the beginning of plate tectonics,
no single indicator is as convincing as the testimonial
given by the accumulation of these lines of evidence at this
time. Even so, the lack of eclogites and blueschists leads
some to distinguish between Proterozoic plate tectonics
(approximately 3,300–700 Ma) and modern plate tectonics (after 700 Ma; Brown, 2007). Before 3.3 Ga, we have
only a limited number of indicators, but these tell us that
TTG magmatism occurred from the beginning of the
Archean, although with a tendency for MgO poorer,
SiO 2 -richer compositions earlier on in time. This possibly
indicates a type of shallow subduction, in which the
subducting crustal plate directly underlies the overriding
plate so that reaction of TTG melts with the mantle wedge
could not occur. As time passed and the Earth cooled, subduction to deeper levels occurred and reaction with the
mantle became more common.
However, the production of TTG melts does not necessarily denote subduction. The requirement from highpressure experiments is that these melts must be derived
by melting of basalt, but these experiments say nothing
about how this basalt reaches its melting conditions. The
main competitor hypotheses to subduction for explaining
the origin of TTG melts are oceanic plateaux and stagnant
lid tectonics. The oceanic plateau model uses the analogy
of thickened ocean crust on the modern Earth, which are
caused by extra heating from below by mantle plumes,
resulting in crust up to 35 km thick (Neal et al., 1997).
Thus, formation of some crust as oceanic plateaux could
apply to an Archean Earth with plate tectonics, but
a major uncertainty in this case becomes the behavior of
this thick mafic crust in subduction zones. Can such thick
crust be subducted as a unit, or does just part of it return to
the mantle and the upper parts accumulate at the surface?
In this scenario, the basaltic source rocks for TTG melts
are derived from thick piles of volcanics that simply
become buried to lower crustal levels because of continued volcanism.
The ocean plateau model is generally thought to work
within plate tectonics and does not seem adequate for the
early Archean if complete recycling of crust commonly
occurred. Here a stagnant lid scenario may be more appropriate, in which the thick crust is not subducted, but the
lower levels delaminate and melt back into the mantle.
Only the upper sections of crust are preserved, a scenario
known as “flake tectonics” (Hoffman and Ranalli, 1988).
In the earliest Archean or Hadean, blocks of crust may
have tipped like ice flows, thus returning extensive blocks
of crust to the mantle without the mechanism of subduction as it occurs on the modern Earth.
It is probable that the large degree of crustal recycling
implied by the paucity of surviving crust from the early
Archean is incompatible with plate tectonics in its modern
form. The transition to plate tectonics may have occurred
during the period 3.3–3.1 Ga, followed by an immense
increase in the survival of continental crust (Figure 1).
The Archean-Proterozoic boundary at 2.5 Ga is the much
more publicized major disjuncture in the geological
record, but these changes in geological style could only
have occurred after abundant continental crust had collected for long periods. The middle Archean may have
been a time in which plate tectonics operated on some
parts of the Earth, but not on others, and it may have been
intermittent (O’Neill et al., 2007).
Summary
Taking multiple indicators into account together, there is
good evidence for the operation of plate tectonics on the
Earth as far back as 3.3 Ga, but it may have been
18
ANCIENT PLATE TECTONICS