ARTICLES
Boudinage of a stretching slablet
implicated in earthquakes beneath the
Hindu Kush
GORDON LISTER*, BRIAN KENNETT, SIMON RICHARDS AND MARNIE FORSTER
Research School of Earth Sciences, The Australian National University, Canberra 0200, Australia
* e-mail: [email protected]
Published online: 24 February 2008; doi:10.1038/ngeo132
As the fragments of Gondwana (Africa, Arabia, India and Australia) moved northward, arc-shaped belts with intervening basins
formed in the Alpine–Himalayan mountain chain during and after collision. This was accompanied by subduction (or sinking) of
the ancient Tethyan oceanic plate (or slab) into the underlying mantle. The arc-like shapes could in part be the end result of processes
related to drips forming in the less-viscous mantle layer at the base of the Earth’s rigid outer shell and then falling into the deeper
mantle. Alternatively, the arcs could have formed because slabs constituted of intervening small ocean basins were independently
subducted during convergence, and have now disappeared. The subducting slabs tend to stretch, tear and eventually break off, leaving
behind thin, vertical strips of colder material that can easily be mistaken for mantle drips. Previous work indicates the presence of
such remnant material beneath the Hindu Kush region, close to the collision zone between the Indian and Eurasian continental plates.
Here, we analyse a cluster of intermediate-depth earthquakes beneath this region and suggest the existence of an elongate boudin, a
lens-shaped feature bounded by ductile faults or shear zones. Our data do not support mantle drip and instead offer a snapshot into
the process of break-off, as a thin strip of vertically stretching slab tears free before descending deeper into the underlying mantle.
With the closure or foundering of small ocean basins, the fate of the
subducting lithosphere is a matter of debate: does the slab tear and
roll back, or does the mantle drip? Reconstruction of the tectonic
evolution of the Alpine–Himalayan orogenic belt during closure of
the Tethyan Ocean reveals many instances that suggest the existence
of small ocean basins that have now disappeared1–4 . In other words,
equivalents of the Caspian or the Black Sea may once have been
widespread along the margins of Tethys. These basins were formed
as Tethys opened, as the result of heterogeneous stretching of
the continental lithosphere, with or without seafloor spreading5,6 .
Later, as Tethys closed, fragments of the dispersing Gondwanan
supercontinent (such as Adria, Anatolia and potentially, India)
began to plough into these basins and through adjacent continental
ribbons. What exactly happened thereafter is not clear, although
each ‘collision’ or accretion event seems to be reflected in the age of
a high-pressure metamorphic episode, and by nappe-stacking7,8 .
The disappearance of these small ocean basins during the
closure of Tethys poses a formidable riddle. Nevertheless, this can
readily be explained by foundering of the lithosphere underlying
these basins, which would involve the formation of new subduction
zones within these basins after the impact of the indentor, tearing
of the slab and then the lateral ‘roll-back’ of minor slablets. As
it rolls back, the foundering slablet causes widespread extension
of the overlying crust, with stretching lineations parallel to the
direction of roll-back. This process explains the development of
arcuate orogens adjacent to relatively young basins as observed
throughout the modern Alpine–Himalayan chain today9–11 . In
addition, it provides a mechanism that can explain the existence of
isolated slablets such as that beneath Gibraltar12 , the Vrancea slablet
beneath the Pannonian Basin13 , or the isolated slablet beneath the
Hindu Kush that is the topic of this article (Fig. 1). This hypothesis
is countered by the ‘mantle drippers’14–20 , who in the main ignore
or explicitly deny the large horizontal motions implied by the rollback hypothesis21–23 .
This controversy will be resolved only through a greater depth
of understanding of mantle structures and their evolution. To this
end, therefore, we have directed some effort towards developing a
digital Virtual Earth and to the analysis of the evolving geometry
of subducting lithospheric slabs24 . Intermediate-depth earthquakes
are of interest because, in conjunction with tomographic data sets,
the geometry of these earthquakes may allow recognition of tearing
lithospheric slabs. Brittle and/or ductile faulting suggests a rheology
that is incompatible with the ‘mantle drip’ hypothesis, although
once a slab has torn and dropped into the mantle, diminished
deviatoric stress intensities provide little motivation for further
seismic activity. In contrast, mantle drips involve material falling
from the thermal boundary layer at the base of the lithosphere, and
no seismicity is ever expected, because relatively weak material is
involved from the outset.
We have examined places where seismic tomography suggests
the existence of slablets, and determined whether or not the pattern
of seismicity allows inference of mantle tears more or less ‘in
progress’. The Hindu Kush25–29 (Fig. 1) is of interest because 15
events with magnitude ≥7.0 are documented for the past 111 yr,
and the NEIC database30 records 123 events in the magnitude
range 5 ≤ Mw < 7 and 1,673 events in the range Mw < 5,
since 1973. Regional studies suggest a remnant of a once-larger
subducting slab31–33 , whereas interpretation of a combination of
tomographic and hypocentric data in three dimensions allows
delineation of the geometry of what seems to be a vertically dipping
twisted slablet, ∼120 km across, which tapers and disappears at
∼500 km depth.
196
nature geoscience VOL 1 MARCH 2008 www.nature.com/naturegeoscience
© 2008 Nature Publishing Group
ARTICLES
40° N
N
Tien Shan
Pamir
38° N
I
lith ndian
osp
her
e
Tarim Basin
2
Tadjik
depression
36° N
Surface of the Earth
0 km
100 km
1
du
an ar
Hin
Kun Lun
MKT
sh
Ku
t
Kohis
c
Ka
Tibet
rak
Fau
r
MB
180–210 km
200 km
lt
mi
34° N
m
sh
Ka
ora
T
MM
T
210–280 km
70° E
72° E
MFT
74° E
T
32° N
68° E
MC
India
76° E
37° N
71° E
78° E
300 km
Figure 1 Structural elements map. Locations of the epicentres for the Hindu Kush
(1) and Pamir (2) zones of intermediate-depth seismicity, with major tectonic
elements: MFT = Main Frontal Thrust of the Himalayan ranges; MMT = Main Mantle
Thrust; MCT = Main Central Thrust; MKT = Main Karakorum Thrust. The Pamir
cluster (2) terminates at ∼120 km depth, although with irregularity along strike. It
outlines a south-dipping structure which may lie entirely within the subcontinental
lithosphere, whereas the Hindu Kush cluster (1) penetrates ∼280 km deep, into
the asthenosphere.
The regional structure is complex, related to the ongoing
collision between India and Asia. The two distinct clusters of
intermediate-depth earthquakes (Fig. 1) can be attributed to the
two converging plates34,35 , with the Hindu Kush cluster penetrating
deep into the asthenosphere and marking a thin remnant strip
related to northward subduction, whereas the Pamir cluster can be
attributed to a southward dipping slab that has not penetrated as
deeply, or which has torn off and since disappeared.
FAULT PLANE ORIENTATION AND KINEMATICS
To convincingly argue the existence of a tearing slab, two
circumstances need apply: (1) there should be a spatial alignment
of hypocentres, or centroids and (2) the kinematics of individual
earthquakes should be coincident with these alignments. Here, we
demonstrate spatial alignments by using spherical trigonometry
to map centroid locations onto a north–south cross-section along
longitude 71◦ E, limiting the data plotted to earthquakes that
occurred within ±20 km of the plane of section (Fig. 2). The
kinematics can be constrained using data from the centroid
moment tensor database provided by the Global CMT Project35 .
The alignments of inferred slip lines are plotted on the same crosssection, showing both criteria above can be met.
Orientation groups can be recognized by interactively analysing
concentration maxima on a stereographic projection (Figs 3,4). We
use an open-source computer program (eQuakes36 ) to read data
in ‘ndk’ format as downloaded from the Global CMT Project.
All earthquakes in the database from 1976–2006 were considered.
Each stereoplot shows fault plane poles and corresponding slip
lines, although which point is a pole and which point is a slip
line cannot be distinguished because of the ambiguity inherent in
CMT data. The centroid moment tensor allows definition of two
possible fault planes for each earthquake, each of which can be
represented on a stereonet by the combination of a slip line and
–200
km S
–100
0
50
km N
Figure 2 Cross-section of the Hindu Kush boudin. A cross-section produced using
the program eQuakes, showing earthquake centroids from the Global CMT Project
database35 . The centroids selected lie within 20 km of a vertical section through
longitude 71◦ E. The inferred Hindu Kush slablet is shown shaded. Fault surfaces
and slip-line trends inferred in this study cluster around lines parallel to the dotted
lines shown. Zero on the horizontal axis is located at 37◦ N. There is no
vertical exaggeration.
a fault plane pole (shown on a conventional earthquake beachball
in Fig. 3a). The difficulty is that the pole to one fault plane is
parallel to the slip line of the conjugate solution, and vice versa.
Without further information, it is not possible to determine which
of the two points represents a fault plane pole, and which represents
the slip line. This inherent ambiguity reduces the usefulness of
CMT solutions, particularly when large numbers of earthquakes
are examined.
There was no obvious orientation clustering evident in
earthquakes in the depth range 75–180 km (Fig. 3b), but patterns
emerged at deeper levels. On the basis of interactive analysis, two
depth groups can be delineated: (1) from 180–210 km (Fig. 3c) and
(2) from 210–280 km (Fig. 3d). Of the 82 earthquakes in the depth
range 180–280 km, 33 had a magnitude greater than 5.5, and all
except three exhibited the geometry of thrusts or reverse faults.
The deeper earthquakes show obvious orientation clusters,
here classified by assuming that the almost vertical maximum
comprises slip lines (shown blue in Fig. 4a), with eQuakes used
to identify corresponding fault poles (shown green in Fig. 4a).
A subclassification of the slip lines can be identified (red
solutions in Fig. 4b, with corresponding fault plane poles mauve).
Restricting the analysis to earthquakes with magnitude >5.5
shows that this pattern becomes more precisely defined (Fig. 4c).
Interpreted in this way, the slip lines of earthquake-generating
faults are kinematically coordinated (close to parallel), although
on curvilinear faults that dip steeply southeast through to south
through to southwest. The slip-line directions plunge steeply
southwards, and are scattered about an orientation that plunges
∼80◦ –>180◦ (the red and blue points on Fig. 4c).
The orientation groups in the depth range 180–210 km (Fig. 3c)
are less obvious, but classification of the stronger maximum as
nature geoscience VOL 1 MARCH 2008 www.nature.com/naturegeoscience
© 2008 Nature Publishing Group
197
ARTICLES
a
b
Compressional field
P
Slip line or fault pole
T
Tensional field
Slip line or fault pole
c
d
Figure 3 Stereoplots of fault poles and slip lines. a, ‘Beachball’ representation of orientation data from a centroid moment tensor, showing compression (P) and tension
( T ) axes. The slip line and the corresponding fault plane pole lie in either of two (conjugate) locations. The great circle through these axes is the mirror symmetry plane of the
tensor. b–d, Fault plane poles and slip lines of earthquakes from the Global CMT Project database35 , plotted on the upper hemisphere of a stereographic projection. Data is in
the depth range 75–180 km (b), 180–210 km (c) and 210–280 km (d). Only centroids with reverse fault geometry are shown. For c and d only three solutions are thereby
excluded, and these are oblique strike–slip in their geometry.
slip lines (shown red in Fig. 4d), and corresponding points as fault
plane poles (shown mauve in Fig. 4d), makes obvious a second
broad maximum. This second maximum is selected using eQuakes,
classified as slip lines, and corresponding points as fault plane
poles. Interpreted in this way, the earthquake-generating faults
fall into two groups: (1) faults with slip lines that plunge steeply
north (the blue points on the upper hemisphere projection in
Fig. 4e) and (2) faults with slip lines that plunge steeply south
(the red points in Fig. 4e). The blue points scatter about thrust
vector ∼−70◦ –>∼160◦ and the red points scatter about thrust
vector ∼−80◦ –>0◦ (Fig. 5a). The corresponding fault plane poles
(Fig. 4f) allow definition of curvilinear fault surfaces (shown as
strike lines in Fig. 5b).
A FOLIATION BOUDINAGE ANALOGY
The method whereby eQuakes has been used to define orientation
groups removes ambiguity occasioned by the symmetry of the
centroid moment tensor. In the case at hand, the choice is whether
maxima are due to an alignment of fault plane poles, or to an
alignment of slip lines. Brittle faults in a twisting slab might be
expected to obey the Mohr–Coulomb relation, and form at high
angles to the direction of stretching, which in this case is vertical.
For such circumstances, it might be argued that the maxima
represent well-defined fault plane orientations. The fault planes
after all do lie in Mohr–Coulomb failure orientations as appropriate
to vertical stretching with horizontal north–south compression.
198
nature geoscience VOL 1 MARCH 2008 www.nature.com/naturegeoscience
© 2008 Nature Publishing Group
ARTICLES
a
b
c
d
e
f
Figure 4 Orientation groups in stereoplots. a, A maximum (blue) is selected within a polygon, for centroids in the range 210–280 km, with corresponding solutions (green).
b, A submaximum (red) with corresponding solutions (mauve). c, These orientation groups more sharply defined by Mw > 5.5 earthquakes. d, A maximum (red) is selected
for centroids in the depth range 180–210 km, with corresponding solutions (mauve). e, Classification of slip lines in two orientation clusters (red and blue). f, Corresponding
fault plane poles in three clusters.
Should this be true, however, it would then be difficult to explain:
(1) the vertical extent of the inferred fault planes and (2) the scatter
of slip-line orientations that is contrary to the degree of alignment
that should be evident if horizontal north–south compression is
taking place. Because these factors are inconsistent, this choice can
be rejected.
The alternative choice is that the maxima recognized are due to
an alignment of slip lines. This interpretation is favoured because,
in the ductile regime, faults are often strongly kinematically
coordinated (subparallel to the direction of relative movement).
Mineral stretching lineations in ductile rocks often cluster about
well-defined maxima, although foliation orientations may vary
considerably. This is a natural consequence of the geometry of finite
strains that involve a high degrees of elongation in one direction.
The trends of slip lines in the classified solutions are plotted
(Fig. 5a), and the strike lines of the inferred fault planes, here shown
projected to the Earth’s surface (assumed to be a smooth sphere,
Fig. 5b). The data from the Hindu Kush are consistent with the
geometry to be expected during foliation boudinage (Fig. 6).
Structural geologists are well familiar with geometries as
illustrated (Figs 5,6), but on much smaller scales and only in
continental schist and gneiss terranes37–40 . Foliation boudinage is
typical in cases where an intense (and therefore highly anisotropic)
fabric has been stretched, and failure occurs as the result of
localized ductile shear bands. Because the material is highly
anisotropic (in terms of its rheology), the ductile faults that
define the boundaries of foliation boudins in stretched gneiss
terranes form at relatively low angles (∼30◦ ) to the dominant
foliation and they exhibit a high degree of kinematic coordination
(directions of relative movement, as inferred from stretching
a
b
40° N
40° N
38° N
38° N
36° N
36° N
34° N
34° N
70° E
75° E
70° E
75° E
Figure 5 Map of Hindu Kush earthquakes. a, Slip-line trends of the classified
solutions in the 180–280 km depth range. b, Spherical trigonometry allows a plot of
corresponding fault strike line traces as they would appear if the fault were
projected to the surface of the Earth. The three orientation clusters are once more
evident, with blue faults dipping towards the northwest, north and northeast, and
red faults dipping predominantly southeast and southwest. The geometry is that
expected during foliation boudinage, with conjugate (north and south) steeply
dipping curvilinear fault surfaces.
lineations, are highly collinear). Foliation boudins form over a
wide range of scales, and are observed under the microscope
(∼1,000 µm), in the outcrop (1 cm–100 m) and inferred on the
kilometre scale in map patterns. Where three-dimensional outcrop
is available, it can be seen that foliation boudins are bounded
by intersecting curvilinear faults, and these faults often have
rhomboidal geometries (Fig. 6). This geometry is particularly
evident on the shores of the Lago di Cignana, Val Tournenche,
nature geoscience VOL 1 MARCH 2008 www.nature.com/naturegeoscience
© 2008 Nature Publishing Group
199
ARTICLES
a
b
S
N
Stretching direction
Vertical
Ductil
enic
mog
Seis zone
r
shea
150 km
king
e nec
180 km
210 km
Relatively undeformed
boudin core
Stretching
direction
Cross-section section at ~71° E
Steeply SE to SW dipping seismogenic zone
0.5 m
c
N
Stee
p
seism ly NE dip
pin
ogen
ic zo g
ne
0 km
ing
ipp
W d zone
N
y
nic
epl
W
Ste moge
s
sei
100 km
E
Plane section at ~200 km
Figure 6 Interpretation of the Hindu Kush boudin. a, Boudin in high-pressure
gneisses from the Lago di Cignana, Val Tournenche, Italy. The bounding shear zones
are evident, with quartz grown in the pull apart between the two boudins. The photo
is rotated to the vertical to highlight the similarity with the geometry of the inferred
Hindu Kush boudin, which is 60,000 times larger and stretching vertically.
b, Interpreted vertical section through the vertically extending Hindu Kush slablet
looking horizontally and towards the west. c, Horizontal section at ∼200 km depth,
looking vertically down from the surface of the Earth.
weak zones42,43 even though intermediate-depth earthquakes (for
example, in the Tonga–Kermadec slab, or in the Pacific slab beneath
Japan) define double seismic zones44 with a characteristic alignment
of hypocentres converging with depth. The alignments inferred
in cross-section do not correspond with the alignments that can
be inferred on the stereographic projections of CMT data, and
therefore these spatial alignments of centroids or hypocentres
do not define the operative ‘fault’ planes. The CMT data seem
consistent with more gently dipping planes of failure, as can be
directly determined in some cases using ‘directivity’ studies45 . It
is interesting that this should be so, because otherwise spatial
alignment of centroids or hypocentres might well be accepted
as sufficient for the definition of the geometry of a seismogenic
fault. However, both conditions need to be fulfilled, namely: (1) a
spatial alignment of hypocentres or centroids and (2) kinematic
coordination, as seems to be the case beneath the Hindu Kush.
We conclude that the Hindu Kush slablet records a geometry
observed many times by structural geologists working on smaller
scales in stretched gneiss terranes, but here 2–5 orders of magnitude
larger in scale. There can be little argument that this is a remnant
thin strip of subducted lithosphere in the process of being torn
free. Interestingly, surface relief in the region must be affected
by periodic failure as boudinage progresses and the release of
vertical stress that must accompany such seismic events provides
an example of dynamic loading in reverse46 . The periodic nature
of energy release associated with these earthquakes suggests they
release elastic energy accumulated during periods of solid-state
creep25,47–50 as the boudinaging slablet slowly tears free from the
overlying lithosphere. But our study is limited in that CMT data
provide only indirect evidence as to the geometry of the ‘foliation
boudin’, and in our time we obtain merely a transitory glimpse as
to how parts of subducting slabs may break off, before they drop
into the deeper mantle. Years from now, when active seismicity has
ceased, and only the thermal anomaly remains, will this be seen as
an example of a mantle drip?
Received 6 September 2007; accepted 28 January 2008; published 24 February 2008.
References
41
Italy , where foliation boudinage took place during exhumation
of ultrahigh-pressure coesite-bearing eclogites. In this area, the
foliations now are gently dipping, and the stretching lineations have
been reoriented accordingly. A combination of glaciation and later
fluvial erosion lead to an unusually good three-dimensional quality
in many exposures, and this area is used as a type area to study the
processes of boudinage.
There is no reason why similar gneissic foliations should not
develop in stretching mantle rocks, nor any reason to suggest
that foliation boudinage should not take place, even on the very
large scale. The slablet beneath the Hindu Kush may define the
Earth’s largest recorded boudin, with a single steeply south-dipping
curvilinear surface accommodating failure, defined by centroids
in the depth range 180–280 km. The geometry of the boudin
in the depth range 180–210 km is that expected during foliation
boudinage, with conjugate faults defining oblate rhomboidal
geometries. The conjugate faults are defined by steeply dipping
curvilinear surfaces (Figs 5,6).
Should our interpretation as to the origin of intermediatedepth seismicity in the Hindu Kush be correct, a logical implication
is that slab boudinage would be expected elsewhere. However, a
systematic examination of stereographic projections of CMT data
from other terranes has provided no concrete further examples,
although the few data from the Vrancea terrane fit this model.
The geometry of intermediate-depth seismicity in a number of
locations is consistent with reactivation of pre-existing faults or
1. Channell, J. E. T. & Kozur, H. W. How many oceans? Meliata, Vardar and Pindos oceans in Mesozoic
Alpine paleogeography. Geology 25, 183–186 (1997).
2. Robertson, A. H. F. Contrasting modes of ophiolite emplacement in the Eastern Mediterranean
region. Geol. Soc. Lond. Mem. 32, 235–261 (2006).
3. Kovács, I. & Szabó, Cs. Middle Miocene volcanism in the vicinity of the Middle Hungarian zone:
evidence for an inherited enriched mantle source. J. Geodyn. 45, 1–17 (2008).
4. Rosenbaum, G. & Lister, G. S. The Western Alps from the Jurassic to Oligocene: spatio-temporal
constraints and evolutionary reconstructions. Earth-Sci. Rev. 69, 281–306 (2005).
5. Desmurs, L., Manatschal, G. & Bernoulli, D. The Steinmann Trinity revisited: Mantle exhumation
and magmatism along an ocean-continent transition: The Platta nappe, eastern Switzerland. Geol.
Soc. Lond. Special Publ. 187, 235–266 (2001).
6. Weinberg, R. F., Regenauer-Lieb, K. & Rosenbaum, G. Mantle detachment faults and the breakup of
cold continental lithosphere. Geology 35, 1035–1038 (2007).
7. Lister, G. S., Forster, M. A. & Rawling, T. J. Episodicity during orogenesis. Geol. Soc. Lond. Special
Publ. 184, 89–113 (2001).
8. Forster, M. A. & Lister, G. S. Several distinct tectonometamorphic slices in the Cycladic
eclogite-blueschist belt, Greece. Contrib. Mineral. Petrol. 150, 523–545 (2005).
9. Carminati, E., Wortel, M. J. R., Spakman, W. & Sabadini, R. The role of slab detachment processes in
the opening of the western–central Mediterranean basins: some geological and geophysical evidence.
Earth Planet. Sci. Lett. 160, 651–665 (1998).
10. Schellart, W. & Lister, G. S. Tectonic models for the formation of arc-shaped convergent zones and
back-arc basins. Geol. Soc. Am. Special Paper 383, 237–258 (2004).
11. Rosenbaum, G. & Lister, G. S. Formation of arcuate orogenic belts in the western Mediterranean
region. Geol. Soc. Am. Special Paper 383, 41–56 (2004).
12. Gutscher, M.-A. et al. Evidence for active subduction beneath Gibraltar. Geology 30,
1071–1074 (2002).
13. Wortel, M. J. R. & Spakman, W. Subduction and slab detachment in the Mediterranean–Carpathian
region. Science 290, 1910–1917 (2000).
14. Houseman, G. A. & Molnar, P. Gravitational (Rayleigh–Taylor) instability of a layer with non-linear
viscosity and convective thinning of continental lithosphere. Geophys. J. Int. 128, 125–150 (1997).
15. Kerr, R. A. Making mountains with lithospheric drips. Science 239, 978–979 (1988).
16. Houseman, G. A. & Molnar, P. Mechanisms of lithospheric rejuvenation associated with continental
orogeny. Continental reactivation and reworking. Geol. Soc. Lond. Special Publ. 184, 13–38 (2001).
17. Platt, J. & Houseman, G. Evidence for active subduction beneath Gibraltar: Comment and Reply.
Geology 31, 22 (2003).
18. Platt, J. P., Whitehouse, M. J., Kelley, S. P., Carter, A. & Hollick, L. Simultaneous extensional
exhumation across the Alboran Basin: Implications for the causes of late orogenic extension. Geology
31, 251–254 (2003).
200
nature geoscience VOL 1 MARCH 2008 www.nature.com/naturegeoscience
© 2008 Nature Publishing Group
ARTICLES
19. Platt, J. P. & Vissers, R. Extensional collapse of thickened continental lithosphere: A working
hypothesis for the Alboran Sea and Gibraltar arc. Geology 17, 540–543 (1989).
20. Zandt, G. et al. Active foundering of a continental arc root beneath the southern Sierra Nevada in
California. Nature 431, 41–46 (2004).
21. Royden, L. H. Evolution of retreating subduction boundaries formed during continental collision.
Tectonics 12, 629–638 (1993).
22. Lonergan, L. & White, N. Origin of the Betic-Rif mountain belt. Tectonics 16, 504–522 (1997).
23. Rosenbaum, G., Lister, G. S. & Duboz, C. Reconstruction of the tectonic evolution of the western
Mediterranean since the Oligocene. J. Virtual Explorer 8, 107–130 (2002).
24. Richards, S., Lister, G. S. & Kennett, B. L. N. A slab in depth: Three-dimensional geometry and
evolution of the Indo-Australian plate. Geochem. Geophys. Geosyst. 8
doi:10.1029/2007GC001657 (2007).
25. Chouhan, R. K. S. Seismotectonics of Hindukush. Pure Appl. Geophys. 82, 108–118 (1970).
26. Ram, A. & Yadav, L. Seismotectonics and analysis of earthquakes from the Hindukush region.
Geophys. J. R. Astron. Soc. 68, 559–574 (1982).
27. Roecker, S. W. et al. Seismicity and fault plane solutions of intermediate depth earthquakes in the
Pamir-Hindu Kush region. J. Geophys. Res. 85, 1358–1364 (1980).
28. Roecker, S. W. Velocity structure of the Pamir-Hindu Kush region: Possible evidence of subducted
crust. J. Geophys. Res. 87, 945–959 (1982).
29. Mellors, R. J., Pavlis, G. L., Hamburger, M. W., Al-Shukri, H. J. & Lukk, A. A. Evidence for a
high-velocity slab associated with the Hindu Kush seismic zone. J. Geophys. Res. 100,
4067–4078 (1995).
30. <http://neic.usgs.gov/>.
31. Koulakov, I. & Sobolev, S. V. A tomographic image of Indian lithosphere break-off beneath the
Pamir-Hindukush region. Geophys. J. Int. 164, 425–440 (2006).
32. Khan, P. K. Stress state, seismicity and subduction geometries of the descending lithosphere below the
Hindukush and Pamir. Gondwana Res. 6, 867–877 (2003).
33. Fan, G., Ni, J. F. & Wallace, T. C. Active tectonics of the Pamirs and Karakoram. J. Geophys. Res. 99,
7131–7160 (1994).
34. Negredo, A. M., Replumaz, A., Villaseñor, A. & Guillot, S. Modeling the evolution of continental
subduction processes in the Pamir–Hindu Kush region. Earth Planet. Sci. Lett. 259, 212–225 (2007).
35. <http://www.globalcmt.org/>.
36. <http://rses.anu.edu.au/tectonics/equakes/>.
37. Platt, J. P. & Vissers, R. L. M. Extensional structures in anisotropic rocks. J. Struct. Geol. 2,
397–410 (1980).
38. Lacassin, R. Large-scale foliation boudinage in gneisses. J. Struct. Geol. 10, 643–647 (1988).
39. Goscombe, B. D., Passchier, C. W. & Hand, M. Boudinage classification: End-member boudin types
and modified boudin structures. J. Struct. Geol. 26, 739–763 (2004).
40. Arslan, A., Passchier, C. W. & Koehn, D. Foliation boudinage. J. Struct. Geol. advance online
publication doi:10.1016/j.jsg.2007.11.004 (19 November 2007).
41. Forster, M. A. et al. in Mapping Geology in Italy (eds Pasquaré, G. & Venturini, C.) 279–286 (Servizio
Geologico d’Italia, 2004).
42. Jiao, W., Silver, P. G., Fei, Y. & Prewitt, C. T. Do intermediate- and deep-focus earthquakes occur on
preexisting weak zones? An examination of the Tonga subduction zone. J. Geophys. Res. 105,
28125–28138 (2000).
43. Ranero, C. R., Villaseñor, A., Phipps Morgan, J. & Weinrebe, W. Relationship between bend-faulting
at trenches and intermediate-depth seismicity. Geochem. Geophys. Geosyst. 6
doi:10.1029/2005GC000997 (2005).
44. Kawakatsu, H. Double seismic zone in Tonga. Nature 316, 53–55 (1985).
45. Warren, L. M., Hughes, A. N. & Silver, P. G. Earthquake mechanics and deformation in the
Tonga-Kermadec subduction zone from fault plane orientations of intermediate- and deep-focus
earthquakes. J. Geophys. Res. 112 doi:10.1029/2006JB004677 (2007).
46. Saleeby, J. & Foster, Z. Topographic response to mantle lithosphere removal in the southern Sierra
Nevada region, California. Geology 32, 245–248 (2004).
47. Kelemen, P. B. & Hirth, G. A periodic shear-heating mechanism for intermediate-depth earthquakes
in the mantle. Nature 446, 787–790 (2007).
48. Regenauer-Lieb, K. & Yuen, D. A. Rapid conversion of elastic energy into plastic shear heating during
incipient necking of the lithosphere. Geophys. Res. Lett. 25, 2737–2740 (1998).
49. Kameyama, M., Yuen, D. A. & Karato, S.-I. Thermal-mechanical effects of low-temperature plasticity
(the Peierls mechanism) on the deformation of a viscoelastic shear zone. Earth Planet. Sci. Lett. 168,
159–172 (1999).
50. Regenauer-Lieb, K. & Yuen, D. A. Quasi-adiabatic instabilities associated with necking processes of an
elasto-viscoplastic lithosphere. Phys. Earth Planet. Inter. 118, 89–102 (2000).
Acknowledgements
Research supported by Australian Research Council Discovery Grant DP0343646 ‘Tectonic
Reconstruction of the Evolution of the Alpine–Himalayan Orogenic Chain’. A. Barker is thanked for
his work on the Vrancea slablet using eQuakes. Program eQuakes written by GL, who accepts all
responsibility for errors of interpretation occasioned by its use.
Correspondence and requests for materials should be addressed to G.L.
Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/
nature geoscience VOL 1 MARCH 2008 www.nature.com/naturegeoscience
© 2008 Nature Publishing Group
201