Net transfer reactions result in appearance and disappearance of minerals, as well as their compositional
changes. The pressure and temperature conditions for
these limits and mineral chemistry for a specific bulk-rock
composition can be simulated using thermodynamic
dataset of minerals (e.g., Holland and Powell, 1998). Such
petrogenetic grids, called as P-T pseudosections, have also
become useful approaches to investigate metamorphic
conditions and paths. Thermo-Calc (Powell and Holland,
1998) and PerpleX (Connolly and Petrini, 2002) are examples of the software for these calculations.
A transition from the peak to the retrograde conditions,
with prograde ones if available, draws a pressuretemperature (P-T) path, which we expect to reflect thermal
structures and modes of material flow in subduction
zones. When an HPLT rock returned back to shallower
levels, nearly adiabatic, isothermal decompression
represented by a clockwise P-T path is predicted for the
case of fast ascending, whereas significant cooling with
decompression with a hairpin path for sluggish ascending
(Ernst, 1988). Such contrasting exhumation rates has been
supported by geochronology (Rubatto and Hermann,
2001; Anczkiewicz et al., 2004; Baldwin et al., 2010).
On the other hand, numerical simulations commonly predict characteristic inverted thermal structure in subduction
zones with deeply subducted but still cool slab overlain by
the hot wedge mantle (e.g., Peacock and Wang, 1999;
Figure 2). When an oceanic crustal rock is detached
from the subducted slab, it is expected to ascend via
a higher-temperature conduit than descending one on the
slab, resulting in clockwise P-T paths (Gerya et al.,
2002). Counterclockwise P-T paths with isobaric cooling
known in the Franciscan Complex in California
(Wakabayashi, 1990) are difficult to explain by ordinary
thermal structures in subduction zones. They are attributed
to transient cooling of the subduction zone just after its
inception.
Exhumation tectonics
Exhumation, the processes in which HPLT metamorphic
rocks ascend from depths to the surface, is one of the controversial subjects on subduction zones dynamics. There
are many ideas for the driving forces to lift up the HPLT
rocks. Buoyancy of the subducting edifices (continents,
oceanic plateaus, mid-oceanic ridges, etc.: Maruyama
et al., 1996), of subducted crustal materials (Chemanda,
1995), or of serpentinized wedge mantle (Guillot et al.,
2001) is one of the likely candidates. Isostatic rebound
of the subducted buoyant slab is expected when the dense
precursor parts of the slab is broken off at depths (Davis
and von Blanckenburg, 1995). Traction force of the
subducting slab could induce a counter flow above the
subduction interfaces (Cloos, 1982). Underplating accretion at depths could jack up the HPLT rocks (Platt, 1986).
Because HPLT rocks are exposed not at trenches but
inside the hanging-wall crusts, unroofing mechanisms to
remove the overlying crustal materials are also necessary
for HPLT rocks to reach to the surface. Erosion plays an
important role for unroofing on land (Chemanda et al.,
1995), whereas its effect might be minimal in cases of
marine subduction settings. One of the proposed ideas is
gravity collapse of an accretionary wedge when it is too
steepened by underplating accretion (Platt, 1986). Upper
plate rifting (Hill et al., 1993), and transtension by oblique
subduction (Ave Lallemant and Guth, 1990) also, could
remove the overburden for the HPLT rocks. When HPLT
rocks uprise along with serpentinite diapir, it would penetrate the hanging-wall crust by its buoyancy.
Recently developed numerical simulation techniques
provide useful deductions for subduction zone dynamics.
For example, Gerya and Stöckhert (2006) modeled at least
two kinds of induced flow in a subduction zone depending
on contrasting rheology of crustal and mantle rocks. One
is a set of drag and counter flows in serpentinized parts
of the wedge mantle (serpentinite channel: Guillot et al.,
2001) along the subduction interface, and the counter flow
carries up higher-grade HP and UHP rocks from greater
depths to the base of the hanging-wall crust. The others
are shallower intra-crustal flows, which lift up lowergrade HPLT rocks with fragments of higher-grade rocks.
This generally agrees with natural occurrences that
low-grade HPLT accretionary rocks tend to occur as
coherent masses (Kimura et al., 1996) whereas highergrade blocks of eclogites and amphibolites are occasionally hosted in serpentinite mélanges (Coleman and
Lanphere, 1971) in ancient subduction complexes.
Occurrences in the sea
Except for submerged extensions of landmass terrains,
sea-bottom occurrences of HPLT metamorphic rocks have
been so far recognized only in the Philippine Sea Plate.
The first finding (Maekawa et al., 1992, 1993;
Maekawa, 1995) was from ODP Leg125 drill core (Hole
778A), which penetrated the Conical Seamount,
a forearc serpentinite seamount ~50 km arc-ward from
the Mariana Trench. HPLT metamorphic minerals, such
as lawsonite, pumpellyite, sodic pyroxene, aragonite,
and blue amphibole (winchite), were found in pebblesized metabasite clasts hosted in serpentinite. Mineral
chemistry and occurrences suggest the fragments
underwent low-grade blueschist facies, in which lawsonite
does not coexist with sodic amphibole and sodic pyroxene, with the metamorphic conditions of 150–200
C
0.5–0.6 GPa (Maekawa et al., 1993, 1995). These results
first confirmed that HPLT rocks are really formed in modern subduction zone. The serpentinite seamount itself is
build up of highly fractured and sheared serpentinites,
and groundwater vents and serpentine mudflows were also
observed (Fryer et al., 1990). Based on these occurrences,
the HPLT metabasites are considered to have carried up
from the subduction interface of ~20 km depths by diapir
or mud volcanism. Fragments of HPLT rocks with
serpentinites are also found in other serpentinite seamounts and inner trench slopes in the Izu-Bonin-Mariana
314
HIGH-PRESSURE, LOW-TEMPERATURE METAMORPHISM
changes. The pressure and temperature conditions for
these limits and mineral chemistry for a specific bulk-rock
composition can be simulated using thermodynamic
dataset of minerals (e.g., Holland and Powell, 1998). Such
petrogenetic grids, called as P-T pseudosections, have also
become useful approaches to investigate metamorphic
conditions and paths. Thermo-Calc (Powell and Holland,
1998) and PerpleX (Connolly and Petrini, 2002) are examples of the software for these calculations.
A transition from the peak to the retrograde conditions,
with prograde ones if available, draws a pressuretemperature (P-T) path, which we expect to reflect thermal
structures and modes of material flow in subduction
zones. When an HPLT rock returned back to shallower
levels, nearly adiabatic, isothermal decompression
represented by a clockwise P-T path is predicted for the
case of fast ascending, whereas significant cooling with
decompression with a hairpin path for sluggish ascending
(Ernst, 1988). Such contrasting exhumation rates has been
supported by geochronology (Rubatto and Hermann,
2001; Anczkiewicz et al., 2004; Baldwin et al., 2010).
On the other hand, numerical simulations commonly predict characteristic inverted thermal structure in subduction
zones with deeply subducted but still cool slab overlain by
the hot wedge mantle (e.g., Peacock and Wang, 1999;
Figure 2). When an oceanic crustal rock is detached
from the subducted slab, it is expected to ascend via
a higher-temperature conduit than descending one on the
slab, resulting in clockwise P-T paths (Gerya et al.,
2002). Counterclockwise P-T paths with isobaric cooling
known in the Franciscan Complex in California
(Wakabayashi, 1990) are difficult to explain by ordinary
thermal structures in subduction zones. They are attributed
to transient cooling of the subduction zone just after its
inception.
Exhumation tectonics
Exhumation, the processes in which HPLT metamorphic
rocks ascend from depths to the surface, is one of the controversial subjects on subduction zones dynamics. There
are many ideas for the driving forces to lift up the HPLT
rocks. Buoyancy of the subducting edifices (continents,
oceanic plateaus, mid-oceanic ridges, etc.: Maruyama
et al., 1996), of subducted crustal materials (Chemanda,
1995), or of serpentinized wedge mantle (Guillot et al.,
2001) is one of the likely candidates. Isostatic rebound
of the subducted buoyant slab is expected when the dense
precursor parts of the slab is broken off at depths (Davis
and von Blanckenburg, 1995). Traction force of the
subducting slab could induce a counter flow above the
subduction interfaces (Cloos, 1982). Underplating accretion at depths could jack up the HPLT rocks (Platt, 1986).
Because HPLT rocks are exposed not at trenches but
inside the hanging-wall crusts, unroofing mechanisms to
remove the overlying crustal materials are also necessary
for HPLT rocks to reach to the surface. Erosion plays an
important role for unroofing on land (Chemanda et al.,
1995), whereas its effect might be minimal in cases of
marine subduction settings. One of the proposed ideas is
gravity collapse of an accretionary wedge when it is too
steepened by underplating accretion (Platt, 1986). Upper
plate rifting (Hill et al., 1993), and transtension by oblique
subduction (Ave Lallemant and Guth, 1990) also, could
remove the overburden for the HPLT rocks. When HPLT
rocks uprise along with serpentinite diapir, it would penetrate the hanging-wall crust by its buoyancy.
Recently developed numerical simulation techniques
provide useful deductions for subduction zone dynamics.
For example, Gerya and Stöckhert (2006) modeled at least
two kinds of induced flow in a subduction zone depending
on contrasting rheology of crustal and mantle rocks. One
is a set of drag and counter flows in serpentinized parts
of the wedge mantle (serpentinite channel: Guillot et al.,
2001) along the subduction interface, and the counter flow
carries up higher-grade HP and UHP rocks from greater
depths to the base of the hanging-wall crust. The others
are shallower intra-crustal flows, which lift up lowergrade HPLT rocks with fragments of higher-grade rocks.
This generally agrees with natural occurrences that
low-grade HPLT accretionary rocks tend to occur as
coherent masses (Kimura et al., 1996) whereas highergrade blocks of eclogites and amphibolites are occasionally hosted in serpentinite mélanges (Coleman and
Lanphere, 1971) in ancient subduction complexes.
Occurrences in the sea
Except for submerged extensions of landmass terrains,
sea-bottom occurrences of HPLT metamorphic rocks have
been so far recognized only in the Philippine Sea Plate.
The first finding (Maekawa et al., 1992, 1993;
Maekawa, 1995) was from ODP Leg125 drill core (Hole
778A), which penetrated the Conical Seamount,
a forearc serpentinite seamount ~50 km arc-ward from
the Mariana Trench. HPLT metamorphic minerals, such
as lawsonite, pumpellyite, sodic pyroxene, aragonite,
and blue amphibole (winchite), were found in pebblesized metabasite clasts hosted in serpentinite. Mineral
chemistry and occurrences suggest the fragments
underwent low-grade blueschist facies, in which lawsonite
does not coexist with sodic amphibole and sodic pyroxene, with the metamorphic conditions of 150–200
C
0.5–0.6 GPa (Maekawa et al., 1993, 1995). These results
first confirmed that HPLT rocks are really formed in modern subduction zone. The serpentinite seamount itself is
build up of highly fractured and sheared serpentinites,
and groundwater vents and serpentine mudflows were also
observed (Fryer et al., 1990). Based on these occurrences,
the HPLT metabasites are considered to have carried up
from the subduction interface of ~20 km depths by diapir
or mud volcanism. Fragments of HPLT rocks with
serpentinites are also found in other serpentinite seamounts and inner trench slopes in the Izu-Bonin-Mariana
314
HIGH-PRESSURE, LOW-TEMPERATURE METAMORPHISM