Organizing Melt Flow
through the Crust
Michael Brown1, Fawna J. Korhonen2, and Christine S. Siddoway3
1811-5209/11/0007-0261$2.50
M
DOI: 10.2113/gselements.7.4.261
In compar ison w ith likely
source-rock compositions, many
granulites and migmatites are
chemically depleted with respect
to a granitic component (Solar and
Brown 2001; Guernina and Sawyer
2003; Brown and Korhonen 2009;
Korhonen et al. 2010a). The preservation of peritectic minerals (see
glossary on page 234 for defi nition of terms in italics) in leucosomes (FIG. 1), the lack of extensive
retrogression of peak metamorphic
mineral assemblages in residual
granulites (White and Powell
2002), and the collapse of formerly
melt-fi lled structures (Brown et al.
1999) confi rm that this chemical
KEYWORDS : dike, ductile fracturing, granite, granulite, melt flow, migmatite
signature reflects loss of melt
from these rocks. Thus, most of
the melt generated in the source
INTRODUCTION
was extracted. This melt rose through the crust until the
The transport of melt from sites of generation in the deep
ductile-to-brittle transition zone or other factors arrested
orogenic crust to sites of accumulation at a shallower level
ascent (Brown and Solar 1999; Brown 2007).
is the most important mass-transfer process affecting the
Evidence from thin section studies of residual granulites
continents. As a result the continents became internally
and migmatites shows that initially melt collects at grain
differentiated and stabilized. The volume and composition
boundaries and then migrates either along structural
of melt generated is a function of source-rock composifabrics that are related to a shape-preferred orientation of
tion, pressure (P) and temperature (T) of melting, and
matrix grains or through tensile microcracks (Marchildon
the presence or absence of an aqueous fluid. The range
and Brown 2002), to accumulate in lower-pressure sites, for
of fertility for melt production among common crustal
rocks (greywackes, pelites, amphibolites, and granites example, strain shadows associated with peritectic minerals
sensu lato) at granulite facies P–T conditions requires that
the volume of crust involved in granite production was
much greater—perhaps by an order of magnitude—than
the volume of orogenic granite. The spacing of plutons
in orogens suggests sources of broadly similar volume in
any one setting (Brown 2007, 2010). When combined
with evidence from inversion of gravity anomalies related
to a root of granite extending to depth beneath plutons
(Vigneresse 1995a), which represents the infi lled ascent
conduit, these observations require that lateral flow of melt
must be a common occurrence in the source (Brown 2007,
2010).
elt that crystallizes as granite at shallow crustal levels in orogenic
belts originates from migmatite and residual granulite in the deep
crust; this is the most important mass-transfer process affecting
the continents. Initially melt collects in grain boundaries before migrating
along structural fabrics and through discordant fractures initiated during synanatectic deformation. As this permeable porosity develops, melt flows down
gradients in pressure generated by the imposed tectonic stress, moving from
grain boundaries through outcrop-scale vein networks to ascent conduits.
Gravity then drives melt ascent through the crust, either in dikes that fill
ductile-to-brittle–elastic fractures or by pervasive flow in planar and linear
channels in belts of steep structural fabrics. Melt may be arrested in its ascent
at the ductile-to-brittle transition zone or it may be trapped en route by a
developing tectonic structure.
1 Laboratory for Crustal Petrology, Department of Geology
University of Maryland, College Park, MD 20742, USA
E-mail: [email protected]
2 Department of Applied Geology, Curtin University
Perth, WA6845, Australia
E-mail: [email protected]
3 Department of Geology, Colorado College
Colorado Springs, CO 80703, USA
E-mail: [email protected]
E LEMENTS , V OL . 7,
PP.
261–266
Pristine peritectic garnet (pink) and minor cordierite
(pale green, labelled Crd) in leucosome in stromatic
metatexite migmatite from the Fosdick migmatite–granite complex,
West Antarctica. See Figure 1 in Sawyer et al. (2011 this issue); for
more details see Korhonen et al. (2010).
FIGURE 1
261
A UGUS T 2011
(Brown et al. 1999) and deformation bands (Brown 2004).
As observed on the outcrop, pathways by which melt has
migrated are recorded by leucosome, peritectic minerals
± leucosome, and melanosome. Leucosome is commonly
found in layer-parallel veins and in structurally discordant sites such as dilation and shear bands; these leucosomes may link to defi ne a former melt-flow network in
three dimensions (Brown et al. 1999; Brown 2004). When
active, these melt-flow networks allowed redistribution and
accumulation of melt and provided pathways that allowed
melt to flow to conduits in which it was transferred to the
shallow crust.
The inferred depth to the anatectic front in an orogenic belt
varies—it may be as little as 15 km in high-temperature–
low-pressure belts or deeper than 100 km in Phanerozoic
ultrahigh-pressure belts—but is commonly between 20 and
70 km. Thus the length scale of melt ascent varies. Ascent
may be independent of tectonic structures and occur via
dikes (FIG. 2; Brown 2004; Weinberg and Regenauer-Lieb
2010) or it may be controlled by regional deformation and
tectonic structure (Vigneresse 1995a, b; Collins and Sawyer
1996; Brown and Solar 1999). Structurally controlled melt
flow generally occurs at a high angle to the principal
compressive stress, either along the lineation in the case of
constrictional strains (FIG. 3A) or in the plane of the foliation
for flattening strains (FIG. 3B), or associated with crustal-scale
shear zones (Roman Berdiel et al. 1997).
THE RELATIONSHIP BETWEEN RESIDUAL
GRANULITES, MIGMATITE – GRANITE
COMPLEXES, AND OROGENIC GRANITES
Melts that crystallize as granite plutons in the shallow
orogenic crust originate from migmatites and residual granulites in the deep orogenic crust. By contrast, regional-scale
migmatite–granite complexes represent a level in the crust
where some of the migrating melt becomes trapped during
ascent and crystallizes as granite within the formerly suprasolidus crust now represented by host migmatites. These
complexes correspond to an intermediate level—previously
at depths of 15 – 40 km—deeper than the anatectic front
but shallower than the strongly residual deep crust. The
rocks in these complexes record the superimposed effects
of melting, melt migration, melt loss, and melt passage;
B
A
ascending melt may have interacted with the former meltbearing crust at the level now exposed. Thus, regional-scale
migmatite–granite complexes are intrinsically complicated but exceptionally informative about the processes
of melting, melt extraction, and melt ascent. We illustrate
these points with two examples.
Western Maine, USA
In western Maine, Silurian–Devonian turbidites were
deformed within a Devonian crustal-scale shear zone
system (Solar et al. 1998; Brown and Solar 1999) and metamorphosed to high-T–low-P upper amphibolite facies conditions (T > 700 ºC and P > 4.5 kbar (Johnson et al. 2003)
along a counterclockwise P–T path leading to water-fluxed
muscovite-breakdown melting (Solar and Brown 2001).
The shear zone system comprises kilometer-scale zones of
apparent flattening strain that anastomose around zones
of apparent constrictional strain (Solar et al. 1998; Brown
and Solar 1999). The onset of melting is recorded by the
first appearance of leucosomes in the up-temperature direction, which represents the maximum rise in the crust of
the solidus at the peak of metamorphism (FIG. 4). Stromatic
metatexite migmatites occur in the zones of apparent flattening strain whereas diatexite migmatites occur in the zones
of apparent constrictional strain, and contemporaneous
decameter-scale granites occur as either concordant planar
or linear bodies according to the structural zone in which
they occur, forming a regional-scale migmatite–granite
complex. Although the leucosomes and granites exploit
the mechanical anisotropy, pinch-and-swell structures and
the absence of solid-state fabrics in the leucosomes and
granites show that deformation and emplacement were
synchronous (FIG 3B), suggesting that these features record
syntectonic pervasive melt flow through the deforming
crust (Solar et al. 1998; Brown and Solar 1999).
Granite in coeval syntectonic plutons, which have ages in
the range 408 – 404 Ma (Solar et al. 1998), was sourced from
crustal rocks with chemical characteristics similar to those
of the host Silurian–Devonian succession (Pressley and
Brown 1999; Tomascak et al. 2005). The migmatites have
residual compositions relative to the associated subsolidus
metapelites, consistent with loss of a granite component,
and the leucosomes represent the cumulate products of
fractional crystallization of the melt, with variable melt
loss (Solar and Brown 2001). Decameter-scale granites
have a range of chemistries that reflect variable entrainment of residual plagioclase and biotite, the accumulation
A
B
(A) Lineation-parallel irregularly shaped granite layer
that is inferred to record magma migration through
Paleoproterozoic lower crust via structurally controlled pathways
parallel to the moderately to steeply plunging regional elongation
direction (for more details see Collins and Sawyer 1996). Pen
parallel to lineation for scale (B) Subhorizontal and subvertical
outcrop surfaces exposing concordant planar granite layers in
residual stromatic metatexite migmatite; the granite represents
melt trapped during ascent within a regional-scale zone of apparent
flattening strain, western Maine, USA (for more details see Brown
and Solar 1999). Lens cap in upper center of image for scale
FIGURE 3
(A) Sketch showing ductile-fracture dikes feeding
brittle-elastic dikes. FROM WEINBERG AND REGENAUER- LIEB
(2010), WITH PERMISSION FROM THE GEOLOGICAL SOCIETY OF AMERICA . (B)
Centimeter-scale dikes of granite propagate in an irregular (zigzag)
fashion from shear bands in the lower part of the image; they are
interpreted as opening-mode ductile fractures (for more details see
Brown 2004). Lens cap in upper center of image for scale
FIGURE 2
E LEMENTS
262
A UGUS T 2011
of products of fractional crystallization, and the loss of
most of the evolved liquid (Solar and Brown 2001). Granite
from hectometer-scale bodies that occur in the migmatite
domains have compositions consistent with crystallization
of evolved liquids produced by fractional crystallization of
primary muscovite-breakdown melts derived from a source
similar to the exposed migmatites (Solar and Brown 2001).
At the peak of the thermal evolution, as the isotherms
moved up through the core of the actively deforming
orogenic belt, the anatectic front migrated to shallow
levels in the crust, perhaps as shallow as 15 km (Brown
and Solar 1999). Thus, melt transferring through the crust
in lineation- and foliation-parallel conduits was able to
reach shallow crustal levels without freezing. The melt
would have been superheated with respect to the solidus
(Leitch and Weinberg 2002), so it could have migrated
across the migmatite front into the unmelted crust to
ascend some distance before solidification. The level of
exposure of the larger (kilometer-scale), apparently discordant plutons corresponds to the former ductile-to-brittle
transition zone, whereas the smaller (decameter- to hectometer-scale), apparently concordant plutons correspond
to a deeper level in the crust (Brown and Solar 1999). For
a geotherm consistent with these inferences, the plutons
would likely extend downward into increasingly more
residual migmatite at deeper crustal levels. As deformation waned and the isotherms moved down through the
core of the orogenic belt, the height to which any batch of
melt could ascend in the crust from the source decreased
and the intervals between arrivals of melt batches into the
root zone of plutons became longer. In this circumstance,
individual batches of melt may have crystallized before the
arrival of subsequent batches, constructing a downwardgrowing composite root zone in which the top of the active
system moved progressively downwards.
Fosdick Mountains, West Antarctica
The Fosdick migmatite–granite complex is exposed as a
gneiss dome in the Fosdick Mountains of West Antarctica.
It comprises interlayered migmatitic gneisses intruded
by multiple phases of anatectic granite (FIG. 5). A Lower
Paleozoic metasedimentary succession and a suite of
Devonian granodiorites exposed outside the dome are
the primary source rocks of the migmatitic gneisses
and anatectic granites inside the dome (Korhonen et al.
2010a, b; Siddoway and Fanning 2009). Two episodes of
crustal anatexis are recorded in the gneisses; both involve
biotite-breakdown melting at granulite facies conditions
(T = 820–870 °C, P = 7.5–11.5 kbar, and T = 830–870 °C, P =
6–7.5 kbar, respectively; Korhonen et al. 2010a). The fi rst
episode was related to Andean-style convergence during
the Upper Paleozoic, whereas the second was related to
a change from wrench deformation to oblique divergence and gneiss dome formation during the Cretaceous
(McFadden et al. 2010).
The deepest structural level exposed in the central Fosdick
Mountains preserves evidence of an Upper Paleozoic melting
event. This domain comprises a ~5 km thick interlayered
sequence of migmatitic gneisses, with granite distributed
in decimeter- to decameter-scale sills (see McFadden et
al. 2010 for a detailed description of units and architecture). The gneissic layers exhibit centimeter-scale leucosomes deformed into recumbent folds with subhorizontal
axes. Within this domain the granites are dominantly ca
369–336 Ma in age (Siddoway and Fanning 2009; Korhonen
et al. 2010b). Sr–Nd isotope compositions show that the
granites were mainly derived from anatexis of the granodiorites (Korhonen et al. 2010b), but mineral equilibria
modeling indicates that these rocks would have produced
only ~2–3 vol% melt at the level exposed (Korhonen et al.
E LEMENTS
Schematic section through the crustal-scale, dextraltranspressive shear zone system in western Maine,
USA. Dashed lines are boundaries between structural domains. ACZ
= zone of apparent constrictional strain (no pattern), which occurs
within anastomosing zones of apparent flattening strain (dashed
pattern); BHB rocks = Bronson Hill Belt rocks; CMB rocks = Central
Maine Belt rocks; TAD and WAD = Tumbledown and Weld anatectic
domains, respectively; P = Phillips pluton; Bt = biotite; Ms = muscovite. This figure is modifi ed from a similar version published in Solar
and Brown (2001) and is used in accordance with the publications
rights policies of Oxford University Press.
FIGURE 4
2010a), suggesting that the granites were related to more
extensive melting at deeper crustal levels. By contrast,
modeling of several metasedimentary rock compositions
shows that ~4–25 vol% melt could have been produced
from this source. Preservation of the peak mineral assemblages and the absence of retrograde reactions in residual
migmatitic gneisses require a minimum of 70% of this
melt to have been lost to a shallower crustal level that is
no longer preserved (Korhonen et al. 2010a); this scenario
is consistent with the scarcity of Carboniferous granites
derived from a metasedimentary source this deep in the
complex.
At the higher structural levels exposed in the southern and
eastern Fosdick Mountains, residual migmatitic gneisses
host granites from the Cretaceous melting event (FIG. 6A).
Granites associated with subvertical fabrics (ca 115–110 Ma;
FIG. 6 B ) were derived from a dominantly granodioritic
source, whereas granites associated with subhorizontal
fabrics (ca 109–102 Ma; FIG. 6B) were derived from a dominantly metasedimentary source (Korhonen et al. 2010b).
The proportion of leucosome and meter- to decameter-scale
granite sills interlayered with residual gneisses increases
upward, forming a ~2 km thick section of voluminous
granite that is localized in and beneath the South Fosdick
detachment zone (FIG. 5). Beneath the detachment zone the
granite sills are up to 100 meters thick and are separated by
meter-thick layers of residual migmatitic gneiss, resulting
in a sheeted granite complex with dominantly shallowly
dipping fabrics (McFadden et al. 2010). The change in
source and geometry with decreasing age corresponds to
a well-documented transition from wrench deformation to
oblique divergence in the region at ca 100 Ma, the formation of the detachment associated with the emplacement
of the sheeted granite complex, and rapid exhumation of
the Fosdick migmatite–granite complex as a gneiss dome
(McFadden et al. 2010).
EXTRACTION OF MELT
Melt segregation is a consequence of relative motion
between melt and residue. Viscous grain-boundary sliding,
deformation-assisted formation of pore space, and ductile
fracturing create a dynamically evolving permeability
during the production of the fi rst few volume percent
263
A UGUS T 2011
A
B
C
W–E (A) and SW–NE (B, C) sections across the
Fosdick migmatite–granite complex, West Antarctica,
showing the relationship between residual gneiss and the sheeted
granite complex immediately below the South Fosdick detachment
zone. Rock types in the cross section are as follows: blue = migma-
titic gneiss derived from a granodioritic source plus minor granite;
grey = migmatitic gneiss derived from a metasedimentary source
rock plus minor granite; pink = sheeted granite complex.
MODIFIED FROM MCFADDEN ET AL. (2010), USED IN ACCORDANCE WITH AGU
FIGURE 5
PERMISSIONS POLICY
of melt. As a result melt connectivity is achieved in all
common crustal rocks, although not necessarily at a
common temperature for different source compositions
since melt volume is a function of bulk composition at a
given P and T. The development of a permeable porosity
allows melt to flow from grain boundaries to channels,
providing there is a driving force and the matrix of solid
grains compacts to expel the melt.
The applied forces and physical properties of melting source
rocks control the mechanics of melt extraction. The forces
may be gravitationally induced (due to the density difference between matrix and melt), tectonic (due to gradients
in differential stress), reaction induced (due to gradients
in melt pressure generated by volume change in a closed
system), and due to interfacial tension between melt and
matrix grains.
Although the density difference between melt and residue
provides a gravitational potential for separation of melt by
porous flow, the viscosity of the silicate melt may be too
high to allow gravity-driven porous flow to segregate melt
from crustal rocks in geologically reasonable timescales at
low melt volumes (<~10 vol%; Rutter and Mecklenburgh
2006). By contrast, variation in pressure due to deformation
is an intrinsic property of all heterogeneous materials and
is important at the outcrop scale. As a result, gradients in
pressure develop. Thus, gradients occur across lithologic
or structural interfaces, due to differences in composition
from layer to layer or the development of fractures and
shear zones. Gradients also develop due to change in rock
properties during partial melting as melt volume changes,
and due to strain and strain-rate weakening or strain and
dilatancy hardening. The influence of volume change on
gradients in melt pressure depends on the relative rates of
melt generation versus compaction and the reduction in
strength associated with increasing melt volume. At the
grain scale, processes such as the development of underpressured melt pores by dilation during melt-enhanced
grain-boundary sliding drive melt migration from overpressured sites; interfacial tension is likely to be significant
only at low porosity. Overall, the pressure distribution in
deforming suprasolidus crust will be dynamic and will
change in space and time, potentially driving melt from
grain boundaries to channels.
E LEMENTS
Gravity-induced melt-pressure gradients are on the order of
~0.003–0.004 MPa m-1 (Rutter and Mecklenburgh 2006).
By contrast, local gradients in differential stress potentially can generate gradients in melt pressure that may be
hundreds to thousands of times greater than those due to
the gravitational potential. Although such stress differences tend to relax, reducing their effectiveness, and the
flow strength of melting rock decreases significantly with
increasing melt volume, tectonically induced melt-pressure gradients are expected to remain larger than gravityinduced melt-pressure gradients at the outcrop scale. It
is these gradients in pressure that drive local melt flow,
resulting in the formation of a network of channels. The
spacing of the compositional layering, the strength of the
fabric and the compaction length control the scale of these
networks (Rutter and Mecklenburgh 2006; Brown 2010).
Ultimately, it is gravity that drives melt ascent through
the crust, but the link between melt segregation into a
network of channels and melt ascent is problematic. In
some orogenic belts, exhumed residual granulites and
migmatites exhibit a range of grain- to outcrop-scale structures that imply that pervasive melt migration occurred
at various scales (e.g. Collins and Sawyer 1996; Brown et
al. 1999; Weinberg 1999). Weinberg (1999) argued that
pervasive migration gives rise to melt sheets preferentially
emplaced parallel to high-permeability zones, such as foliation or lithologic contacts. On the other hand, Brown
and Solar (1999) made a case for pervasive migration in
strain-controlled planar and linear structures in zones of
apparent flattening and apparent constrictional strain in
western Maine. For western Maine, Brown and Solar (1999)
showed that the granites were emplaced at the ductileto-brittle transition zone and were rooted in the underlying residual migmatites. In both cases, pervasive melt
migration was proposed based on the three-dimensional
form of the leucosomes and decameter- to hectometer-scale
granites and their conformable relationship to the tectonic
structures. The hypothesis proposes that advection of heat
with migrating melt drives expansion of the suprasolidus
domain, allowing melt to ascend to shallower depths
(Brown and Solar 1999; Weinberg 1999). The feedback
relation between pervasive migration of melt and heating
of the country rock allows subsequent batches of melt to
reach increasingly shallower levels. This process has been
264
A UGUS T 2011
modeled numerically by Leitch and Weinberg (2002), who
demonstrated that pervasive migration of superheated melt
into subsolidus crust is an efficient way of moving the
anatectic front to shallower levels in the crust. As melt
reaches the front, a layer composed of about half granite
and half crust forms below it, so that the total rise of the
front corresponds to about half the thickness of granite that
crystallizes below the moving front. The result is a regionalscale migmatite–granite complex, as represented by the
two examples from western Maine and West Antarctica
described above.
Partially molten crust represents an overpressured system
with melt-pressure gradients and spatial and temporal
variations in permeability. Permeability has a power law
dependence on melt volume, which may lead to instabilities in the form of waves of melt-fi lled porosity. Connolly
and Podladchikov (2007; see also Connolly 2010) use
numerical simulations to investigate fluid expulsion from
a large volume of elevated porosity obstructed above by
a layer of low porosity; their results show that such a
configuration can induce strongly channelled fluid flow.
In partially molten crust, the channels are propagated by
waves of melt-fi lled porosity that leave behind trails of
incompletely compacted porosity, and these trails act as
preferential pathways for subsequent waves; this scenario
is consistent with the batch model of pluton construction. The waves propagate because compaction expels
melt upward from deeper in the crust to create a region of
melt overpressure at the anatectic front. This overpressure
dilates the solid matrix and propagates the porosity beyond
the anatectic front. At this point the porous domain may
propagate independently as a solitary wave of anomalous
porosity fi lled with superheated melt. Passage of the waves
reduces the background porosity in the suprasolidus crust
and the waves gain melt as they propagate. Connolly and
Podladchikov (2007) infer that the channels will develop
with radial symmetry in three dimensions, but in nature
the effect of far-field stress would tend to flatten the channels. These instabilities may manifest themselves in nature
either as domains of concordant decameter- to hectometerscale granites that mimic the strain state in the host rocks
(e.g. western Maine, USA) or as domains of dikes and sills
representing former melt-fi lled fractures that propagated
from the suprasolidus to the subsolidus crust, enabling
melt to ascend (e.g. Fosdick Mountains, West Antarctica).
How do these fractures form? Although field observations show that leucosome networks may connect with
petrographic continuity to granite in centimeter-scale
dikes (Marchildon and Brown 2003; Brown 2004), the
dikes are commonly discordant to foliation and lineation
in the host, indicating that their formation was related
to stress-controlled fracturing. Blunt fracture tips and
irregular (zigzag) propagation paths (FIG. 2B) demonstrate
that the thin dikes represent ductile opening-mode fractures (Brown 2004). Ductile fractures result from creep
deformation and growth of microscopic melt-fi lled pores
that become interconnected, leading to failure (Eichhubl
2004). Weinberg and Regenaur-Lieb (2010) postulate that
melt-fi lled ductile fractures may reach a critical length,
where melt pressure at the tips overcomes fracture toughness and leads to brittle–elastic diking. Thus, melt ascent
through the suprasolidus crust is postulated to occur via
the formation of ductile fractures, and this process is
expected to switch to brittle–elastic fracturing to allow
migration of melt through the subsolidus crust (FIG. 2;
Brown 2004, 2010).
Theory, field relationships, and petrological results suggest
that melt extraction is strongly self-organized from the
bottom up (Brown 2010). Where ascent occurs in dikes,
the number of dikes decreases as their width increases
E LEMENTS
A
B
(A) Shallowly dipping sheeted Cretaceous granites
(pink-grey color) and intervening layers of migmatitic
gneiss derived from a granodioritic source (medium grey color) and
from a metasedimentary source (dark grey-brown color) make up
the lower part of the sheeted granite complex below the South
Fosdick detachment, Fosdick Mountains, West Antarctica (for more
details see Korhonen et al. 2010b). Geologists at the foot of the cliff
provide scale. (B) An older generation of Cretaceous granites (115–
110 Ma) forms steeply oriented sheets (center right), whereas a
younger generation of Cretaceous granites (109–102 Ma) forms
shallowly oriented sheets (upper center), reflecting the change in
tectonics from wrench deformation to oblique divergence (see FIG.
5B, inset, for a schematic representation). NE side of the Fosdick
migmatite–granite complex, Fosdick Mountains, West Antarctica
FIGURE 6
from the suprasolidus crust through the subsolidus crust
to the site of accumulation as a body of granite (Connolly
and Podladchikov 2007; Hobbs and Ord 2010). Sites of
melt accumulation are controlled by the physical properties of the crust and the stress field, as well as by tectonic
structures. Melt ascent may be arrested at the ductile-tobrittle transition zone (Vigneresse 1995b; Brown 2010),
corresponding to depths of 12–18 km, as proposed for
western Maine, or it may be stopped en route by a developing tectonic structure, such as a major detachment zone,
as occurred in the Fosdick Mountains of West Antarctica.
What is the timescale involved? On the one hand, a pluton
represents a large volume of magma (103 –10 4 km3 or more)
aggregated from many batches of melt (each perhaps
10 -1–102 km3) that crystallized during tens of thousands
to several millions of years (Pressley and Brown 1999;
Brown 2010). On the other hand, the timescale for melt
extraction and ascent from the source is expected to be
short (103 –104 years), based on geochemical considerations
265
A UGUS T 2011
and calculated rates of ascent in dikes (Brown 2010). Given
the batch construction of plutons, these two views are not
incompatible. Furthermore, Hobbs and Ord (2010) have
argued that the maximum melt flux at the anatectic front
is on the order of 10 −10 m s−1 for a maximum thickness
for the anatectic zone of 21 km, which is sufficient to
produce a pluton about 3 km thick over a period of 106
years. Smaller values of the melt flux at the anatectic front
result in longer timescales, up to a limit imposed by the
rate at which thermal conduction can provide heat to drive
crustal melting.
CONCLUSIONS
Granites generated during orogenesis originate from
migmatites and residual granulites in the deep crust.
Outcrop-scale networks of remnant granite within these
rocks record the temporal and spatial relations between
melt flow and syn-anatectic deformation. When active,
these networks allow redistribution and accumulation of
melt, and provide links for melt flow to ascent conduits.
Ascent occurs in dikes, which form in ductile fractures in
suprasolidus crust that switch to brittle–elastic fractures in
subsolidus crust, or by pervasive flow, evidenced by congruence between the form of steeply oriented planar and linear
granites and host-rock state of strain. The process of melt
extraction is self-organized from the bottom up.
ACKNOWLEDGMENTS
We acknowledge Roberto Weinberg and an anonymous
reviewer for helpful reviews, and Bernardo Cesare, Hap
McSween, Ed Sawyer and Richard White for comments.
This material is based upon work supported by the
U.S. National Science Foundation under grant numbers
9705858, 0734505, and 0944615 (to Brown), 0338279 and
0944600 (to Siddoway), and a postdoctoral research fellowship to Korhonen (0631324).
Brown M (2004) Melt extraction from
lower continental crust. Transactions of
the Royal Society of Edinburgh: Earth
Sciences 95: 35-48
Guernina S, Sawyer EW (2003) Largescale melt-depletion in granulite
terranes: an example from the Archean
Ashuanipi Subprovince of Quebec.
Journal of Metamorphic Geology 21:
181-201
Brown M (2007) Crustal melting and
melt extraction, ascent and emplacement in orogens: Mechanisms and
consequences. Journal of the Geological
Society 164: 709-730
Hobbs BE, Ord A (2010) The mechanics
of granitoid systems and maximum
entropy production rates. Philosophical
Transactions of the Royal Society A 368:
53-93
Brown M (2010) The spatial and temporal
patterning of the deep crust and implications for the process of melt extraction. Philosophical Transactions of the
Royal Society A 368: 11-51
Johnson TE, Brown M, Solar GS
(2003) Low-pressure subsolidus and
suprasolidus phase equilibria in the
MnNCKFMASH system: Constraints
on conditions of regional metamorphism in western Maine, northern
Appalachians. American Mineralogist
88: 624-638
REFERENCES
Brown M, Korhonen FJ (2009) Some
remarks on melting and extreme
metamorphism of crustal rocks. In:
Gupta AK, Dasgupta S (eds) Physics
and Chemistry of the Earth's Interior.
Published for the Indian National
Science Academy by Springer, pp 67-88
Brown M, Solar GS (1999) The mechanism of ascent and emplacement
of granite magma during transpression: a syntectonic granite paradigm.
Tectonophysics 312: 1-33
Brown MA, Brown M, Carlson WD,
Denison C (1999) Topology of syntectonic melt-flow networks in the deep
crust: inferences from three-dimensional images of leucosome geometry in
migmatites. American Mineralogist 84:
1793-1818
Collins WJ, Sawyer EW (1996) Pervasive
granitoid magma transfer through the
lower–middle crust during non-coaxial
compressional deformation. Journal of
Metamorphic Geology 14: 565-579
Connolly JAD (2010) The mechanics of
metamorphic fluid expulsion. Elements
6: 165-172
Connolly JAD, Podladchikov YY (2007)
Decompaction weakening and channeling instability in ductile porous
media: Implications for asthenospheric melt segregation. Journal of
Geophysical Research 112: B10205,
doi:10.1029/2005JB004213
Eichhubl P (2004) Growth of ductile
opening-mode fractures in geomaterials.
In: Cosgrove JW, Engelder T (eds) The
Initiation, Propagation, and Arrest of
Joints and Other Fractures. Geological
Society of London Special Publication
231, pp 11-24
E LEMENTS
Korhonen FJ, Saito S, Brown M, Siddoway
CS (2010a) Modeling multiple melt
loss events in the evolution of an
active continental margin. Lithos 116:
230-248
Korhonen FJ, Saito S, Brown M, Siddoway
CS, Day JMD (2010b) Multiple
generations of granite in the Fosdick
Mountains, Marie Byrd Land, West
Antarctica: Implications for polyphase intracrustal differentiation in a
continental margin setting. Journal of
Petrology 51: 627-670
Leitch AM, Weinberg RF (2002)
Modelling granite migration by
mesoscale pervasive flow. Earth and
Planetary Science Letters 200: 131-146
Marchildon N, Brown M (2002) Grainscale melt distribution in two contact
aureole rocks: implication for controls
on melt localization and deformation.
Journal of Metamorphic Geology 20:
381-396
implications for granite ascent and
emplacement mechanisms in convergent orogens. Lithos 46: 335-366
Roman Berdiel T, Gapais D, Brun J-P
(1997) Granite intrusion along strikeslip zones in experiment and nature.
American Journal of Science 297:
651-678
Rutter EH, Mecklenburgh J (2006) The
extraction of melt from crustal protoliths and the flow behavior of partially
molten crustal rocks: an experimental
perspective. In: Brown M, Rushmer
T (eds) Evolution and Differentiation
of the Continental Crust. Cambridge
University press, Cambridge, pp
386-429
Siddoway CS, Fanning CM (2009)
Paleozoic tectonism on the East
Gondwana margin: Evidence from
SHRIMP U-Pb zircon geochronology of
a migmatite–granite complex in West
Antarctica. Tectonophysics 477: 262-277
Solar GS, Brown M (2001) Petrogenesis
of migmatites in Maine, USA: Possible
source of peraluminous leucogranite
in plutons. Journal of Petrology 42:
789-823
Solar GS, Pressley RA, Brown M, Tucker
RD (1998) Granite ascent in convergent orogenic belts: Testing a model.
Geology 26: 711-714
Tomascak PB, Brown M, Solar GS, Becker
HJ, Centorbi TL, Tian J (2005) Source
contributions to Devonian granite
magmatism near the Laurentian border,
New Hampshire and Western Maine,
USA. Lithos 80: 75-99
Vigneresse JL (1995a) Control of granite
emplacement by regional deformation.
Tectonophysics 249: 173-186
Marchildon N, Brown M (2003)
Spatial distribution of melt-bearing
structures in anatectic rocks from
Southern Brittany: implications for
melt-transfer at grain- to orogen-scale.
Tectonophysics 364: 215-235
Vigneresse JL (1995b) Crustal regime
of deformation and ascent of granitic
magma. Tectonophysics 249: 187-202
McFadden RR, Siddoway CS, Teyssier C,
Fanning CM (2010) Cretaceous oblique
extensional deformation and magma
accumulation in the Fosdick Mountains
migmatite-cored gneiss dome, West
Antarctica. Tectonics 29: TC4022,
doi:10.1029/2009TC002492
Weinberg RF, Regenauer-Lieb K (2010)
Ductile fractures and magma migration
from source. Geology 38: 363-366
Pressley RA, Brown M (1999) The
Phillips Pluton, Maine, USA: evidence
of heterogeneous crustal sources and
266
Weinberg RF (1999) Mesoscale pervasive
felsic magma migration: alternatives to
dyking. Lithos 46: 393-410
White RW, Powell R (2002) Melt loss
and the preservation of granulite
facies mineral assemblages. Journal of
Metamorphic Geology 20: 621-632
A UGUS T 2011