Serpentinization pulse in the actively deforming Central Indian Basin

Abstract

Heat flow in the actively deforming Central Indian Basin is on average 30 mW/m2 higher than the theoretical 55 mW/m2 heat flow expected from plate cooling of a Cretaceous oceanic lithosphere. Strong spatial correlation between the anomaly and the active thrust fault network at local (faults) and regional scales suggests two potential tectonically driven mechanisms activated at the time of initiation of deformation: friction-to-heat conversion or exothermic serpentinization. We quantitatively examine both processes using an updated geometry of the thrust fault network and simple thermal models. Friction generated heat is limited in all cases: at shallow levels, shear stresses remain small, while heat generated at deeper levels does not contribute significantly to the surface heat flow since permanent regime is not reached. In the exothermic serpentinization model, a maximum anomaly of 20 to 30 mW/m2 is reached 2 to 6 Myr after the onset of widespread serpentinization, depending on the efficiency of the water circulation. The amount and timing of heat release can fully explain the present-day surface heat flow of the Central Indian Basin, provided vigorous hydrothermal circulation closely followed the onset of deformation. Based on a reprocessed multichannel seismic line, we suggest that faults cutting through the entire crust and across the Moho discontinuity drive water at mantle levels and trigger the exothermic serpentinization reaction. We interpret sub-Moho reflectors imaged at depths of 8 to 15 km below the top of the crust – and coinciding with the location of the maximum reaction rate coefficient of serpentinization – as serpentinization fronts. We discuss the significance of this pulse of serpentinization in terms of timing of deformation, weakening and transient rheology of the oceanic lithosphere.

Full text

Serpentinization pulse in the actively deforming Central Indian Basin
Matthias Delescluse
a,c,
⁎, Nicolas Chamot-Rooke
a,b
a
Ecole normale supérieure, Laboratoire de Géologie, 24 rue Lhomond, 75005, Paris, France
b
CNRS, UMR 8538, France
c
Université Paris XI, Orsay, France
abstractarticle info
Article history:
Received 27 November 2007
Received in revised form 8 September 2008
Accepted 14 September 2008
Editor: C.P. Jaupart
PACS:
91.45.G
91.50.Ln
91.55.Jk
91.55.Ln
91.60.Hg
93.30.Nk
Keywords:
heat flow
oceanic lithosphere
deformation
active faults
friction
serpentinization
multichannel seismic
Heat flow in the actively deforming Central Indian Basin is on average 30 mW/m
2
higher than the theoretical
55 mW/m
2
heat flow expected from plate cooling of a Cretaceous oceanic lithosphere. Strong spatial
correlation between the anomaly and the active thrust fault network at local (faults) and regional scales
suggests two potential tectonically driven mechanisms activated at the time of initiation of deformation:
friction-to-heat conversion or exothermic serpentinization. We quantitatively examine both processes using
an updated geometry of the thrust fault network and simple thermal models. Friction generated heat is
limited in all cases: at shallow levels, shear stresses remain small, while heat generated at deeper levels does
not contribute significantly to the surface heat flow since permanent regime is not reached. In the exothermic
serpentinization model, a maximum anomaly of 20 to 30 mW/m
2
is reached 2 to 6 Myr after the onset of
widespread serpentinization, depending on the efficiency of the water circulation. The amount and timing of
heat release can fully explain the present-day surface heat flow of the Central Indian Basin, provided vigorous
hydrothermal circulation closely followed the onset of deformation. Based on a reprocessed multichannel
seismic line, we suggest that faults cutting through the entire crust and across the Moho discontinuity drive
water at mantle levels and trigger the exothermic serpentinization reaction. We interpret sub-Moho
reflectors imaged at depths of 8 to 15 km below the top of the crust –and coinciding with the location of the
maximum reaction rate coefficient of serpentinization –as serpentinization fronts. We discuss the
significance of this pulse of serpentinization in terms of timing of deformation, weakening and transient
rheology of the oceanic lithosphere.
© 2008 Elsevier B.V. All rights reserved.
1. Introduction
Anomalously high heat-flow measurements have been acquired in
the northeastern Indian Ocean fordecades (Pollack et al., 1993)andwere
promptly related to the abnormal level of intraplate seismicity recorded
between the India and Australia tectonic plates (Weis sel et al., 1980;
Wiens et al.,1986; Stein et al., 1988). The heat-flow anomaly was shown
to coincide with an area of widespread intraplate deformation that
started in the late Miocene (Cochran et al., 1989). Seismic profiles (Bull
and Scrutton, 1992; Chamot-Rooke et al.,1993; VanOrman et al.,1995)
further imaged the highly faulted sediments, crust and mantle of the
Central Indian Ocean, confirming the spatial correlation between high
heat-flow and active deformation. The thermal anomaly was thus seen
as a consequence of friction-to-heat conversion along deeply rooted
reverse faults (Weissel et al., 1980; Geller et al., 1983; Stein et al., 1988;
Gordon et al., 1990). An alternative view was immediately raised:
deformation may have concentrated in areas of high thermal state and
weaker rheology, the high heat flow being a cause rather than a
consequence of the localization of deformation.Stein and Weissel (1990)
ruled out large-scale reheating at the base of the lithosphere showing
that there was no associated bathymetric swell. They further inferred
that the source of additional heat had necessarily to be shallow (Geller
et al., 1983; Stein et al., 1988) in order to satisfy simultaneously the
present-day depth to basement, the deep seismicity (Okal, 1983)
(suggesting low temperatures at depth of 30–40 km) and the high
surface heat flow. The topic remained closed and rather unsolved until
Verzhbitsky and Lobkovsky (1993) proposed that the required shallow
heat source may relate to the exothermic serpentinization of mantle
peridotites, as first mentioned in Fyfe (1974). They provided a crude
estimate of the produced heat and concluded that serpentinization may
be as efficient as other mechanisms to increase significantly the surface
heat flow in this part of the Indian Ocean.
Substantially more data are available today: (1) Using deep seismic
refraction, Louden (1995) detected a low velocity zone thathe attributed
to partially serpentinized clasts of peridotites in the gabbros layer at the
base of the oceanic crust; (2) At the scale of the thrusts network, one
Earth and Planetary Science Letters 276 (2008) 140–151
⁎Corresponding author. Ecole normale supérieure, Laboratoire de Géologie, 24 rue
Lhomond, 75005, Paris, France.
E-mail address: delesclu@geologie.ens.fr (M. Delescluse).
0012-821X/$ –see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsl.2008.09.017
Contents lists available at ScienceDirect
Earth and Planetary Science Letters
journal homepage: www.elsevier.com/locate/epsl

detailed heat-flow profile across two major active faults was obtained
during a pre-site survey of the ODP Leg 116 drilling in an area close to the
Afanasy–Nikitin Seamount Chain (Cochran et al., 1989). Distribution of
the observed heat flow along this profile was modeled by Ormond et al.
(1995) as vigorous fluid circulation through the Bengal fan sediments,
redistributing an already anomalous basal heat flow. No heat source
origin is proposed to explain the ∼30 mW/m
2
anomaly, but these data
do confirm the linkwith the faults; (3) Numerous seismic reflectors have
been imaged in the mantle, including deep penetrating faults. Acquired
in 1991, Phèdre multichannel seismic profiles (Chamot-Rooke et al.,
1993) imaged the deep geometry of the thrust faults along a 2100 km-
long profile across the area. Below the ODP Leg 116 site, two fault
reflectors cutacross discontinuous Moho phasesand reach deep into the
mantle, suggesting that fluid paths do exist at sub-Moho depth. Deep
Fig. 1. Location of surface heat-flow measurements in the Indian Ocean and some of the seismic tracks where E–W thrust faults have been imaged. V: Conrad monotrace profile
(VanOrman et al.,1995); P1: Phèdre Leg1 wide angle profile; P2: Phèdre Leg2 wide angle profile (Chamot-Rooke et al.,1993); K: Eastern CIB seismic profile (Krishna et al., 20 01). Red
symbols cover the Central Indian Basin, green symbols the Wharton Basin, blue symbols the remaining areas. The size of the symbol is proportional to the heat-flow value.
Background is a filtered Sandwell 15.1 satellite (Sandwell and Smith, 1997) gravity field showing contrasting structural trends on both sides of NinetyEast Ridge. Contour of the
Afanasy Nikitin Seamount is also shown. The ODP Leg 116 site is represented by a star. (For interpretation of the references to color in this figure legend, the reader is referred to the
web version of this article.)
141M. Delescluse, N. Chamot-Rooke / Earth and Planetary Science Letters 276 (2008) 140–151

mantle reflectors that do not seem to connect to active faults were also
found. Brightness of the fault planes was further interpreted as
evidences for water circulation (Bull and Scrutton, 1990b).
Discarding the possibility of a pre-deformation large-scale reheating
due to external sources, we quantitatively examine in this paper two
heating mechanisms that may explain the strong link between fault
localization and high heat flow: friction on the thrust fault network and
exothermic serpentinization. Our analysis is based on reprocessed
multichannel seismic data showing newevidences for fluid paths down
to the mantle and associated enigmatic deep, bright reflectors. We
successively discuss evidences for serpentinization in the Indian Ocean,
the role of fluid convection, expansion during serpentinization and
finally other sites of serpentinization in the oceans.
2. Data
2.1. Heat flow versus age analysis
The global heat-flow compilation by Pollack et al. (1993) includes
some 300 measurements acquired in the Indian Ocean during various
surveys since the 1960's, to which we have added measurements from
the pre-site survey of ODP Leg 116 drilling (Cochran et al., 1989)(Fig. 1).
Heat-flow values range from 2 mW/m
2
to more than 200 mW/m
2
.We
show in Fig. 2 the distribution of heat flow as a function of age of the
underlying oceanic lithosphere. The main figure is not different from
what waspublished previously (e.g. Geller et al.,1983;Stein and Weissel,
1990) but we choose to split the data into three subsets: the Central
Indian Basin (including the region of the Afanasy Nikitin Seamount
Chain), theWharton Basin, and theremaining data. The first two subsets
cover areas that are presently suffering active deformation. Theoretical
plate cooling models are given for comparison. Quality of data was
Fig. 2. Surface heat flow versus age of the oceanic lithosphere inthe Indian Ocean. Colors
and symbols are similar to Fig. 1. A 10 Myr binning is superimposed, as well as two
conductive plate models (dark envelope) for two different plate thickness (95 km and
125 km). CIB: Central IndianBasin; ANS: Afanasy Nikitin Seamount; WB: Wharton Basin.
Fig. 3. Portion of the reprocessed multichannel Phèdre line across the ODP Leg 116 site. Top inset shows the detailed heat-flow measurements obtained above two active faults during
the ODP Leg 116 pre-site survey (Shipboard Scientific Party, 1989). Open dots underline deep crustal and mantle reflectors.
142 M. Delescluse, N. Chamot-Rooke / Earth and Planetary Science Letters 276 (2008) 140–151

discussed in a number of papers and they all concluded that the heat
flow was abnormally high (Geller et al.,1983; Steinet al., 1988; Stein and
Weissel, 1990; Williams,1990; Ormond et al.,1995).
The observed heat flow is widely scattered above and below the
plate cooling model predictions. This type of scattering may be the
result of transient hydrothermal circulations, and indeed some of the
temperature gradients were demonstrated to be non-linear (Geller
et al., 1983). An over simplistic view is that the lowest heat-flow values
are sites of recharge of cold fluids whereas highest heat-flow values
are hot fluids seeping sites. However, the Central Indian Basin
definitely shows a different pattern, the mean heat flow there being
clearly above the theoretical predictions. An alternative to the high
heat-flow anomaly interpretation is that the measurements system-
atically “missed”the recharge areas, as suggested by apparently
normal values estimated within two ODP holes (Williams, 1990). This
alternative interpretation was dismissed in Stein and Weissel (1990)
and Ormond et al. (1995), as will be discussed in a latter section.
The high heat-flow anomaly appears to be restricted to the Central
Indian Basin, since the Wharton Basin seems to show the right thermal
age (Stein et al., 1988). Actually, the Cretaceous Central Indian Basin
(mean age of 74 Ma at measurements sites) has a higher surface heat
flow (80 mW/m
2
) than the younger Wharton Basin (mean age of 61 Ma
and average heat flow of 65 mW/m
2
). The simplest interpretation is that
in the Central Indian Basin, convective hydrothermal circulation is
superimposed onto an abnormally hot conductive lithosphere, and that
both processes are intimately linked to the active fault network.
Hydrothermal circulation may also be present in the Wharton Basin,
but the background heat flow there is presently normal.
2.2. Spatial correlation between high heat flow, high straining regions
and active faults
Seismic profiling allows us today to draw a quite accurate contour
of the thrust fault network in the Central Indian Basin. It spreads along
a roughly equatorial zone around the Afanasy Nikitin seamount, from
70°E to the NinetyEast Ridge and from 7°S to approximately 5°N. At
large scale, the high heat-flow values cover more or less the thrust
fault network. In our recent kinematic study of India–Australia
intraplate deformation (Delescluse and Chamot-Rooke, 2007), we
showed that the best instantaneous strain rate field, in terms of style
and amplitude of deformation, is obtained when heat flow is used as a
proxy for rheological strength of the Indian Ocean oceanic lithosphere.
In the northeastern Indian Ocean, this is equivalent to considering the
entire fault network area as weak.
At shorter scale, best evidences for correlation between high heat-
flow and active deformation come from one detailed heat-flow profile
across the ODP Leg 116 site (Fig. 3). This ∼15 km -lon g profile was sampled
every half a kilometer or less. The 44 heat-flow measurements average
84 mW/m
2
,withapeakvalueat166mW/m
2
between two active reverse
faults both with large offsets. A reprocessed multichannel line of the
Phèdre Cruise over the same area (Fig. 3) shows that the fault reflectors
plunge deep into the mantle, below Moho seismic phases. At the scale of
these two faults, the heat-flow profile thus reproduces the Central Indian
Basin large-scale anomaly of 25–30 mW/m
2
.Ormond et al. (1995)
successfully modeled the bell shape of the heat-flow anomaly assuming
vigorous fluid circulation within the fault planes above an abnormally hot
basement. Williams (1990) also proposed a set of models in which faults
are included as conduits.
These elements suggest that the spatial correlation –both global
and local –between faulted seafloor and thermal anomaly is not a
coincidence. The question that arises is whether the present-day
deformation is a cause or a consequence of the thermal anomaly. That
phase of deformation began 7.5 (Cochran et al., 1989)to9(Delescluse
et al., 2008) Myr ago as documented by regional late Miocene
unconformities in the Indian Ocean sediments. Stein et al. (1988)
noticed that there was no clear correlation between the exact location
of the earthquakes and the high heat-flow sites, probably because of
the large uncertainties on the position of the epicenters. Nevertheless,
they suggested that the absence of heat-flow anomaly and the low
seismicity in the Wharton Basin was indicating that it is presently
deforming much less than the Central Indian Basin. We know today
that the seismicity is not that low in the Wharton Basin (i.e. Mw= 7.9,
2000; Mw = 6.9, 1995) and that the strike-slip motion accommodated
there is significant (Deplus et al., 1998). Summing moment tensors of
earthquakes, we actually found that the Wharton Basin is presently
the highest straining region of the Indian Ocean (Delescluse and
Chamot-Rooke, 2007). Strain is thus highest in a region where heat
flow is close to normal, which seems paradoxical both if heating is the
cause or the consequence of the deformation. WB and CIB areas are
actively deforming, but with contrasting style of faulting: mainly E–W
reverse faults in the CIB and sinistral shear on N–S vertical faults in the
Wharton Basin (Delescluse and Chamot-Rooke, 2007)(Fig. 1). We will
show in a latter section that the contrasting heat-flow anomaly
between the two regions is indirectly related to this contrasting style
of deformation.
3. Origin of the heat source
We review here two different mechanisms that have been
proposed as potential heat sources for the Central Indian Basin. Both
are tectonically driven: friction-to-heat conversion on the thrust fault
network and exothermic serpentinization along the fault planes at
mantle depth. We do not takeinto account the possibility of a hot-spot
reheating of lithosphere here, as it has already clearly been discarded
in Stein and Weissel (1990).
3.1. Friction on faults
Shear friction on faultshas long been recognized as a potential source
of heat, in particular at large continental strike-slip faults (Lachenbruch
and Sass, 1980; Scholz, 2000; D'Alessio et al., 2006) and at subduction
zones (Wang et al.,1995). Crude estimations have been proposed for the
reversefaults of the Indian Ocean (Geller et al.,1983), butwe now have a
much more detailed picture of the thrust network, including fault
spacing, geometryof the faultsat depth, rate of deformation.134 reverse
faults were recognized along one Phèdre deep seismic profile that span
the entire region of deformation in the Central IndianBasin, 30% of them
associated with faultreflectors withinthe crust itself with an averagedip
angle of 36° to45° (Chamot-Rooke et al.,1993; Jestin,1994). Twoof these
reflectors are clearly imaged in the region of the ODP Leg 116 site (Bull
and Scrutton,1990a), and our reprocessing shows that they cut through
the entire crust, intersect the Moho discontinuity and plunge deep into
the upper mantle.
Our calculation of the friction-related anomaly is based on the
diffusion of heat in a half-space (Carslaw and Jaeger,1959) adapted by
Lachenbruch and Sass (1980, 1992) for the San-Andreas strike-slip
fault. The rationale is to sum the individual contribution of each fault
along the entire profile length. Shear heat flow along a profile normal
to a vertical fault plane is:
qy;tðÞ¼
Qc
2πEi−x2
2þy2
4Kt

−Ei−x2
1þy2
4Kt

ð1Þ
where yis the horizontal distance to the surface fault trace, tis time, x
l
and x
2
are the minimum and maximum depth of the brittle portion of
the fault, Q
c
is the product of the depth-averaged shear stress by the
fault slip-rate, E
i
is the exponential integral function, Kis the thermal
diffusivity.
This formalism is readily adaptable to a thrust fault with a dip
angle θ. This is done by the discrete summation of infinitesimal
vertical planes disposed every dxdepth unit along the fault plane with
length of dx
sin θ
ðÞ to keep the size of the frictional surface identical. This
143M. Delescluse, N. Chamot-Rooke / Earth and Planetary Science Letters 276 (2008) 140–151

also allows for variable shear stress with depth. The heat-flow
anomaly across one thrust fault is then:
q1y;tðÞ¼∑
tx2−x1
dxb
j¼0
Qx
2−jdxðÞ
2πEiRj2

−EiRj1
 ð2Þ
where R
jl
and R
j2
are functions of distance from the calculation point
(x=0,y) to respectively the top and bottom of each infinitesimal
vertical plane. QxðÞ¼ V
cos θ
ðÞτsτs¼μρgx
1−μtan θ
ðÞ

is now the product of the
shear stress –at depth xon the thrust fault plane of dip θ–by the fault
slip-rate. Although these equations are strictly valid for infinitely long
fault, which is obviously not the case in the Indian Ocean, they still
provide a reasonable estimation of the heat input since the fault
network extends for nearly 2000 km in the EW direction, much larger
than the typical depth of the faults.
The final thermal anomaly is obtained by summing the effects of
faults spaced at regular intervals to simplify the natural situation. All
calculations were performed using a shortening rate of 1 cm/yr
integrated along the entire length of the profile crossing the fault
network, which can be regarded as an upper limit (Chamot-Rooke
et al., 1993; Delescluse and Chamot-Rooke, 2007). For simplicity, all
faults dip in the same direction, based on the observation that very
large portion of the deformed oceanic crust (100–200 km) actually
show uniform fault dips (Jestin, 1994). The depth of the brittle
lithosphere is chosen to be 40 km, based on the maximum depth of
earthquakes in the area (Stein and Weissel, 1990). The shear stress
follows a standard Byerlee's law with null pore fluid pressure, which
again will tend to overestimate the frictional heat due to unrealistic
high shear stress at depth. A standard faults spacing of 7 kmwas used,
but larger spacing was also tested since large offset faults are rather
20–30 km spaced. The period of activity is set to 7.5 Myr.
Fig. 4. Main results of the friction heat production model. All graphs are transient states 7.5 Myr after the onset of deformation, except Graph D (gives the thermal evolution through
time) and E (reaches permanent regime). Graphs A to E are 900 km-long profiles across the reverse faults network (Central Indian Basin model, 1 cm/yr of total shortening). Graph F is
a 1000 km crossing vertical strike-slip faults (Wharton Basin model, 2 cm/yr of total shearing). A) Faults spacing variation; B) Brittle depth variation; C) Dip-angle variation; D)
Evolution through time; E) Brittle depth variation (permanent regime); F) Strike-slip faults spacing variation.
144 M. Delescluse, N. Chamot-Rooke / Earth and Planetary Science Letters 276 (2008) 140–151

Fig. 4 shows the sensibility of the frictional model to some of the
main parameters. Varying the fault spacing (Graph A) does not change
the overall level of the thermal anomaly (∼5.5 mW/m
2
) since the
integrated amount of shortening remains identical. Given the over-
estimation that we made, it is readily apparent that friction cannot
explain the observed heat-flow anomaly.
Some other straightforward and useful information can be extracted
from the model. Shearing on individual faults could eventually be
detected through surface heat-flow measurements if their spacing
would be greater than 40 km (Graph A). By varying the thickness of the
brittle zone (Graph B), we learn that the heat source must be shallower
than 10 km to clearly individualize 30 km spaced faults: whatever the
strength and mechanism, if the heat source is linked to the faults, heat
conduction implies that we have virtually no chance–given the number
of faults –to measure a non-anomalous heat flow.
The lower the dip angle, the greater the frictional surface. However,
this effect is over compensated (Graph C) by the fact that the slip-rate
grows for larger dip angles, and also that the shear stress grows when
the dip angle is larger than π
4−1
2atan(µ). This is however not sufficient to
reach heat-flow anomalies significantly greater than 5 mW/m
2
. The next
graphs show the evolution through time before permanent regime is
reached.A period of 7.5Myr of heating is far from the permanent regime
if the source is deep (40 km in Graph D). A permanent regime is almost
reached if the source depth is shallower than 10 km (compare Graph E
and Graph B), but then the thermal anomaly is small.
We performed the same calculations for the Wharton Basin to
evaluate frictional heating along a 1000 km-long E–Wprofile normal to
theactiveN–S strike-slip faults (Graph F). The total rate of shear was
assumed to be 2 cm/yr, which is probably maximized (Delescluse and
Chamot-Rooke, 2007). Our calculations lead to a maximum heat-flow
anomaly ranging from a ∼1mW/m
2
100 0 km-long plat eau –in the case
of 10 km spaced strike slip faults –to narrow 14 mW/m
2
spikes if the
shear is distributed onto 4 faults only. If shear heating is present in the
Wharton Basin, the chance to resolve it with surface heat-flow survey is
low unless measurements are performed close to the reactivated
fracture zones.
3.2. Exothermic serpentinization
Some serpentinization reactions are known to be significantly
exothermic (Fyfe, 1974; Lowell and Rona, 2002; Fruh-Green et al.,
2003; Emmanuel and Berkowitz, 2006). Such heating seems to play a
major role in the Lost-City vent field near the Mid Atlantic Ridge,
where mantle peridotites are exhumed at shallow depth (Fruh-Green
et al., 2003). It may be held responsible for sustained convection in
this off-ridge area (Kelley et al., 2001; Lowell and Rona, 2002;
Emmanuel and Berkowitz, 2006). Exothermic serpentinization is thus
an appealing mechanism to generate heat at rather shallow depth, and
we will review pieces of evidence that it may be at work in the Indian
Ocean in Section 4. Beyond the crude estimates found in Verzhbitsky
and Lobkovsky (1993), we describe here the thermal model that we
built to quantitatively calculate the amount of heat that may be
available in the deforming portion of the Indian Ocean.
3.2.1. Thermal model and kinetics of the serpentinization reaction
The hydration reaction is possible up to temperatures of 400 °C
(Emmanuel and Berkowitz, 2006) to 500 °C (Francis,1981). The reaction
is most efficient around 250–300 °C (MacDonald and Fyfe,1985; Fruh-
Green et al., 2003; Emmanuel and Berkowitz, 2006) where lizardite is
the stable mineral while greater temperatures and pressures lead to a
stable antigorite mineral (Christensen, 2004). The reaction becomes
extremely slowaround and below 100 °C. The typical rock temperature
at Moho depth is ∼120 °C for a Cretaceous oceanic lithosphere, which
potentially allows for serpentinization in the lower oceanic crust and in
the upper mantle down to ∼25 km. We consider that the mantle is
virtually anhydrous atthese depths in the Central Indian Basin, sinceany
serpentinization process that would have occur at the very early
formation of the oceanic lithosphere would have been active at shallow
levels only due to the high thermal state at the ridge axis.
We use here the reaction kinetics described in Emmanuel and
Berkowitz (2006), i.e.:
Aρf
At¼−Krρfð3Þ
where ρ
f
is the volumic mass of forsterite in the mantle. K
r
is the rate
coefficient, expressed as K
r
=Ae
−b(T−c)
2
, this last formulation being
empirically constructed from kinetic data (Emmanuel and Berkowitz,
2006). A,band care different kinetic coefficients compiled in Table 1.
Serpentinization produces H=290 kJ/kg of forsterite (MacDonald and
Fyfe, 1985; Emmanuel and Berkowitz, 2006).
We also use the rate of heat generation per unit volume defined in
Emmanuel and Berkowitz (2006) as:
Q¼HAρf
At:ð4Þ
We finally assume that there is 75% of forsterite in the mantle which
leads to ρ
f
(t=0) =2400 kg m
−3
for a mantle density of ρ
m
=3200kgm
−3
.
A2Dfinite difference algorithm is used to solve the transient heat
diffusion equation with spatially and temporally variable heat pro-
duction. We do not take into account volume changes for simplifica-
tion, but we will discuss the matter in Section 4. Simple lithosphere
cooling is first solved in 1D until t
start
=72.5 Ma, which will represent
theapproximativethermalstateattheonsetofdeformation
(∼56 mW/m
2
). The 2D finite difference grid is then chosen to be
30 km wide and 125 km deep with a cell dimension of about
150 m× 150 m. 45° dipping faults are disposed regularly, with periodic
thermal boundary condition, which is a good way to get rid of side
effects, and is appropriate for the fault network that we study. Heat
production then follows serpentinization fronts moving horizontally
away from the fault planes (Fig. 5). The reaction in the heating fronts is
forced to stop when half of the forsterite is serpentinized. At time
t
stop
=80 Ma, we thus obtain an average 50% serpentinization which
leads to a density of ∼2900 kg m
−3
. The rationale is to keep the mantle
density close to the value obtained by Louden (1995) at the base of the
crust.
The amount of heat produced and the temperature field evolution
throughtime are mainlycontrolled by three parameters: (1)the reaction
rate coefficient A, which is a measure of the efficiency of the serpen-
tinization reaction; (2) the serpentinization front velocity S, which will
relate to the mechanical strength around the fault; and (3) the fault
spacing, i.e. the initial surface of serpentinization. Fault spacing was set
to 10 km in most of the runs, which is the distance between the main
faults observed at the ODP Leg116 site. The firsttwo parameters are very
difficult to set in a deep environment. One of the rare studies dealing
with serpentinization in relatively deep (5 km) oceanic environment
(MacDonald and Fyfe,1985) quotes velocity of 1000 m/Myr at 300 °C for
water propagation in t he host rock from an open crack. The same value is
used by Ranero et al. (2003) for serpentinization that may be active
around the bending related normal faults at subduction bulges.
The kinetics of the serpentinization is even less constrained. It is
fast when operated in the laboratory (a few days, A≃10
−6
s
−1
), but it is
most probably much slower in situ because of a smaller reaction
surface (Emmanuel and Berkowitz, 2006). An additional limiting factor
is the water supply (MacDonald and Fyfe,1985) rather than the kinetics
Table 1
Parameters for the kinetics of serpentinization
Rate coefficient magnitude A0.5 to 8 ×10
−12
s
−1
Peak reaction temperature c540 K
Kinetic coefficient b2.5×10
−4
K
−2
Serpentinization front velocity S300–2000 m/Myr
145M. Delescluse, N. Chamot-Rooke / Earth and Planetary Science Letters 276 (2008) 140–151

itself. We choose a rate coefficient magnitude (A) spanning two orders of
magnitude and somewhat smaller than what is used by Emmanuel and
Berkowitz (2006) (i.e. around 10
−12
s
−1
instead of ∼10
−10
s
−1
), assuming
that the reaction surface at depth is less compared to the shallow
environment of Lost-City. Half-life of the reaction is then around
20,000 yrs at peak temperature, the coefficient bbeing chosen so that
the reaction is extremely slow for temperatures below 100 °C and also
virtually stopped above 400 °C (Emmanuel and Berkowitz, 2006).
3.2.2. Results of the serpentinization model
The goal is to reach at least a 20 mW/m
2
surface heat-flow
anomaly, which would be a significant part of what is observed today.
The second constraint is timing, since a fundamental difference of the
serpentinization model compared to the frictional heating model is
that the source decays with time: by essence, the obtained anomaly is
thus transient and vanishes with time. Fig. 6 shows one typical run
and summarizes the sensitivity of the results to the main parameters.
The test run is for a 10 km faults spacing and a typical front velocity of
1000 m/Myr. The associated temperature anomaly in the upper
mantle does not exceed 80 K and lasts a few Myr only. In the case of a
shorter fault spacing, not shown here, the temperature field is more
homogeneous and does not exceed 50 K. In all cases, the temperature
anomaly is limited to depths between 5 and 25 km. A lowtemperature
anomaly and a thermally quasi-normal lithosphere were actually two
requirements quoted in the early study of Stein and Weissel (1990).
In this simple test run, the peak surface anomaly (22 mW/m
2
)is
reached about 5 Myr after the onset of serpentinization and remains
above 10 mW/m
2
the remaining 2 to 3 Myr. Faster front propagation
(S=200 0 m/Myr) produces a sharper peak anomaly of 30 mW/m
2
which
does not last long enough, while slow propagation (S=500 m/Myr) just
reaches a low maximum of ∼12 mW/m
2
at 80 Ma. Changing the kinetics
results in marginal variations: increasing Ato ∼10
−11
s
−1
does not
produce a higher heat flow for any of the assumed front velocities. The
reason is thatfaster reaction rates result in a saturated heat output, since
the velocity of the serpentinization front is thetrue limiting factor. When
the front propagates to the next discrete 150 m wide cell of the model,
the previously hydrated volume has already released all of its potential
heat. In other words, in our model, the reaction is water limited, and is
equivalent to the successive serpentinization of spatially limited massifs
modeled –or idealized –by the serpentinization front. Finally,
decreasing the fault spacing to 7.5 km or even to 3 km has little effects,
provided that the front propagation velocity is downgraded (respec-
tively to S=750 m/Myr and S= 300 m/Myr). This is not surprising, since a
constant fault spacing to front propagation ratio results in a similar
history of serpentinization (i.e. 50% serpentinization completed after
5Myr).
Several sets of parameters thus lead to the required surface heat
flow. The remaining difficulty is that the peak anomaly generally
comes too early in the thermal history. One possibility is that
serpentinization was delayed with respect to the initiation of de-
formation. In the 3 km spaced faults case, a linear progression of the
serpentinization front velocity from 150 m/Myr to 500 m/Myr would
correspond to an onset of serpentinization at 6.25 Ma. Equivalent
scenarios can be reached with 7.5 km and 10 km faults spacings (Fig. 6,
C-curves 1). New seismic data published in Delescluse et al. (2008)
show massive reactivation of the 3 km spaced oceanic fabric normal
faults at 9 Ma. After a 2 Myr period, faults are selectively abandoned
and only the present-day active faults remain. Our model is thus in
better agreement with the onset of this second phase of deformation
starting at 7 Ma under the effect of a severe weakening of fault planes
Delescluse et al. (2008).
4. Discussion
4.1. Clues to serpentinization in the CIB
Arguments in favor of past and present serpentinization in the
Indian Ocean mainly come from various seismic data. Refraction data
in the CIB southwest of the ODP site revealed a low velocity zone in
the lower crust, interpreted as partially serpentinized (∼35%,
ρ=2950 kg/m
3
) clasts of peridotites in the gabbro Louden (1995).
One seismic reflection profile shows injection into one reverse fault that
has been interpreted as a serpentinite diapir based on gravity modeling
(Krishna et al., 2002). Seismic refraction data from Russian surveys also
Fig. 5. Sketch for the serpentinization model. Serpentinization fronts propagate horizontally from the faults; Serpentinization ends in cells where 400 °C is reached. Periodic thermal
boundary conditions are assumed on both sides. Grey shading indicates the degree of serpentinization.
146 M. Delescluse, N. Chamot-Rooke / Earth and Planetary Science Letters 276 (2008) 140–151

suggest velocities up to 7.3 km/s in the shallow crust, which could be
related to such diapirism (MacDonald and Fyfe, 1985; Levchenko and
Verzhbitsky, 2003; Verzhbitsky and Neprochnov, 2005).
We provide new pieces of evidence from our reprocessing of one
Phèdre seismic profile, where we detect enigmatic, bright and oblique
reflectors deep in the mantle (Fig. 7). These reflectors are disposed at
the termination of a much less visible reflector which connects at the
basement interface with a clear reverse fault. Their approximate depth
(8 to 15 km) coincides with the maximum value of the serpentiniza-
tion reaction rate. We thus interpret these very bright reflectors as
serpentinization fronts –either active or fossil –near an active thrust
fault used as fluid path. Deeper reflectors may be present, since the
serpentinization rate drops 20 km below the top of crust (Figs. 6,7).
Finally, serpentinization of peridotites appears as a quite efficient
heating mechanism: the transient heat-flow anomaly it produces is
much larger than what friction can provide (∼20 mW/m
2
versus
∼5 mW/m
2
), recalling that we maximized the friction effect in our
calculations. However, extreme surface heat-flow values cannot be
modeled without including fluid flow: the exothermic serpentiniza-
tion provides a strong enough basal heat flow at depth, which is then
redistributed by convective fluid circulation in the sediments
(Ormond et al., 1995), but also possibly in the crust (Louden, 1995).
Our thermal model shows that serpentinization is compatible with a
wide range of present-day surface heat flow (including close to
normal values), taking into account the transient nature of the thermal
anomaly. Serpentinization requires a significant fluid input that we
did not discuss nor take into account in our conductive model. The
question of the long term access of fluids to fresh mantle also has to be
addressed.
4.2. Downward flow of water required by serpentinization: amount and
thermal effect
First, the required water input rate is not unrealistic. Serpentiniza-
tion consumes 300 L of water per cubic meter of olivine (MacDonald
and Fyfe, 1985; Fruh-Green et al., 2003). Our model involves half
serpentinization of, say 20 km (depth)× 30 km (width) × 1 km
(arbitrary since our model is 2D)= 600 km
3
of olivine, which means
that we have to bring a total of 0.5×600× 300 × 10
9
=10
14
L of water
during the 5 Myr reaction period. This is equivalent to a flow of
∼0.02 µL s
−1
.m
−2
, or 0.6 L s
−1
through the 30 km×1 km arbitrary
surface at the seafloor. With a conservative value of one single fault
used as fluid path every 30 km, and a fluid path width of 10 m, the
downward fluid velocity is around 6× 10
−8
ms
−1
. This value is similar
to the 7 × 10
−8
ms
−1
upward fluid velocity derived from non-linear
temperature gradients by (Geller et al., 1983).
Fig. 6. A) Snapshots of the anomalous thermal field through time for the exothermic serpentinization model after 1 Myr, 2.5 Myr, 4.5 Myr and 6.5 Myr of deformation; B) Evolution of
the surface heat flow through time since the onset of deformation for three different serpentinization front velocities (500, 1000 and 2000 m/Myr) and variable kinetic coefficients.
Dotted curve is for Wharton Basin with vertical faults having 30 km spacing and a serpentinization front velocityof 1000 m/Myr. C) Evolution of the surface heat flow with the xaxis
chosen so that all models still reach 20 mW/m
2
at the origin: 1) Three overlapping curves show equivalent thermal history for different fault spacings (10 km, 7.5 km, and 3 km) but
scaled serpentinization front speeds (respectively 1000 m/Myr, 750 m/Myr and 300 m/Myr); 2) Test for a linearly increasing serpentinization front velocity with time (from 100 m/
Myr to 500 m/Myr with 3 km fault-spacing); 3) Test for a high serpentinization front velocity 2000 m/Myr for comparison.
147M. Delescluse, N. Chamot-Rooke / Earth and Planetary Science Letters 276 (2008) 140–151

Fig. 7. Bottom: Phèdre Leg 1 reprocessed profile showing deep and bright reflectors at sub-Moho depth. Top: Same profile with sediments removed, using trace time-shift and mute to start the display at the top of the oceanic crust. The two-
way travel time is converted to depth using two different values for the mean velocity (6.5 km/s and 7.5 km/s). Right: Reaction rate coefficient for serpentinization as a function of depth, at the time of initiation of deformation and 7.5 Myr later.
Bright enigmatic reflectors coincide with the predicted depth of maximum rate of serpentinization and are interpreted as serpentinization fronts.
148 M. Delescluse, N. Chamot-Rooke / Earth and Planetary Science Letters 276 (2008) 140–151

The nature and extent of recharge zones are however unknown and
the calculated velocity above is likely to be overestimated. Thesearch for
fluid recharge zones was central to the debate on the ODP Leg 116 heat-
flow profile interpretation due to the absence of lower than normal
heat-flow measurements in the CIB. Williams (1990) proposed a fluid
advection process to explain the84 mW m
−2
average value, the modeled
section (15 km)being considered asa discharge zone. Noadditional heat
source is necessary but this approach requires a distant recharge zone
where downward flow of water should cool the sediments. Such a zone
where the heat flow would be lower than normal has never been
observed. The probability that all measurements missed recharge zones
is small, and this leaded previous authors (Geller et al., 1983; Stein and
Weissel, 1990; Ormond et al., 1995) to conclude to an anomalously hot
basement. Ormond et al. (1995) set up a 3D model including anomalous
basal heat flow, and showed that fluids are more likely to convect in the
fault planes themselves rather than between two faults through a deep
conduit. In this case, along-strike variations of heat flow should be
observed and recharge zones are those where heat flow equal the
theoretical values. Although we cannot totally rule out that the linear
temperature gradients at ODP sites 717 and 719 showing normal heat
flow (Fig. 3) do represent the conductive background heat flow, they
may equally be these areas of diffuse recharge.
If exothermic serpentinization is responsible for the high heat-flow
anomaly, then the required deep downward flow of water (i.e. water
needed to feed the serpentinization reaction) must be negligible
compared to the surface fluid convection input, otherwise low heat-
flow values should be observed. Ormond et al. need 2× 10
8
Lyr
−1
.km
−2
of water input in their model while serpentinization only needs
∼2×10
7
Lyr
−1
/(30 km
2
)=7×10
5
Lyr
−1
.km
−2
, which is more than two
orders of magnitude less. We thus conclude that the deep downward
flow of water probably has a limited cooling effect if recharge zones
are diffuse.
4.3. Water progression in the oceanic mantle
Serpentinization tends to reduce rock porosity to zero (MacDonald
and Fyfe, 1985), so that the fluid paths toward the mantle may become
rapidly closed. In a tectonically active setting, the easiest paths for
fluids are then faults and cracks, which need to be permanently
“refreshed”(Ranero et al., 2003). Our model suggests that a more or
less continuous band of serpentinization is needed to reach the
required surface heat flow, since isolated massifs of serpentinites
would not release enough heat. Connection between propagating
serpentinization fronts could be achieved through the dense reverse
faults network. The details of the faults population indicate two
subsets of faults in the Indian Ocean (Delescluse et al., 2008): (1) small
offset faults (b100 m) as close as 2 to 3 km from one another, well
expressed in the sediments, in particular on high resolution seismic
profiles (VanOrman et al., 1995; Delescluse et al., 2008), and (2) large
offset faults, reaching hundreds of meters to a kilometer, generally
associated with a deep bright reflector (Chamot-Rooke et al., 1993;
Jestin,1994), with a much larger spacing (about 20 km). If the presence
of reflectors is related to water, only the second population of large
offset faults could be used as fluid path. Small offset faults are inactive,
but they are numerous and the fact that the lithosphere has been
highly fractured could help the propagation of water between large
faults. Ranero et al. (2003) also notice that some reflectors in the
mantle near trench outer rises are not specifically linked to an
observed fault at the surface. If as they point in their study, deep
reflectors are caused by serpentinization or free fluids, then the 5 to
6 km wide massifs observed in Fig. 7 would suggest that water can
propagate several kilometers away from the main fault. The low
frequency Phèdre seismic (Ziolkowsky et al., 2003) does not reach the
resolution of Ranero et al. seismic, but broadening of faults as they
deepen (Ranero et al., 2003) and/or undetected small faults at depth
may explain the lateral propagation of fluids.
4.4. Expansion during serpentinization
Expansion during the serpentinization reaction may reach 53%
(MacDonald and Fyfe,1985; O'Hanley,1992)in volume. On onehand, the
difficulty is space accommodation of this additional volume, specifically
in a compressional tectonic setting, but on the other hand damages
resulting from thevolume expansion itselfcould allow fluid progression.
At large degree of serpentinization, expansion should show at the
seafloor as uplift, and we should see a shallower than normal seafloor
depth. O'Hanley (1992), while acknowledging that field evidences of an
increase in volume has not been recognized, mentions that realistic
increase in volume is between 25 and 45%. Serpentinization is restricted
to a depth of 20 km due to the temperature threshold. 50% serpen-
tinization affecting a 15 km thick column of mantle (with 75% of
forsterite) should thus produce a 50%× 25%× 75%×15,000=1400 m
uplift. This is a clear issue of the large-scale serpentinization hypothesis.
The bathymetric anomaly is actually opposite to the one that is
predicted by expansion. In details, the unloaded basement depth
(Fig. 8) shows that the 300 km buckling wavelength is superimposed
on a 800 km-long topographic depression which coincides with the
deepest geoid low on earth (Ćadek and Fleitout, 2003; Crosby et al.,
2006). This low was interpreted as the dynamic effect of deep mantle
sources, such as viscosity variations (Ćadek and Fleitout, 2003). The
reference topography in the CIB may thus be significantly deeper than
the age prediction. Apart from partially canceling this very long
wavelength, the extravolume may also participate to the formation of
the buckling wavelength that has grown to a kilometer since the onset
of deformation (Gerbault, 2000).
We finally cannot rule out that other hydration reactions are also
exothermic. Fyfe (1974) cites some very exothermic hydration
reactions in the crust which would double the heat production in
our case. Less than 50% of mantle serpentinization would then be
needed, leading to a lesser expansion. Unfortunately, we know little
about the kinetics of these shallower reactions.
4.5. Serpentinization elsewhere
Fault spacing may explainwhy, while strain rate is at least as large in
the Wharton Basin than in the CIB, no clear heat-flow anomaly is
detected there. The two regions exhibit contrasting style of deformation
(Deplus et al., 1998; Delescluse and Chamot-Rooke, 2007). The CIB is an
area of ubiquitous reactivation of a dense network of normal faults
acquired at the ridge axis (Bull and Scrutton, 1990a, 1992; Jestin, 1994;
Delescluse et al., 2008), while in the Wharton Basin deformation is
Fig. 8. From top to bottom, bathymetry, unloaded basement depth (black), and
basement depth of the entire 1600+ km-long Phèdre Leg 1 seismic profile. The profile
starts at 14°S (55 Ma old oceanic lithosphere) and ends around 1°N (90 Ma old oceanic
lithosphere). Theoretical basement depth obtained from a plate cooling model is shown
by the dotted line (parameters slightly adapted from (Crosby et al., 2006): plate
thickness of 95 km and ridge depth at age zero of 2750 m). The dashed line shows a
hypothetical reference basement depth in the case of a dynamic topography following
the geoidal low of southern India.
149M. Delescluse, N. Chamot-Rooke / Earth and Planetary Science Letters 276 (2008) 140–151

accommodated by left-lateral, widely-spaced strike slip faults. Serpen-
tinization may be limited to short distances around them in the absence
of a dense active network of secondary faults as found in the CIB. Fig. 6
shows that even if propagation is effective, one vertical fault plane every
30 km leads to a heat-flow anomaly smaller than 10 mW/m
-2
.One
alternative is that heat has already passed in the WB, either because the
serpentinization delay may have been shorter or the deformation
started earlier there. Mantle dipping reflectors and reduced seismic
velocities have been taken as strong evidences for serpentinization of
the oceanic mantle at the outer rise of the Middle America Trench,
possibly down to 20–30 km (Ranero et al., 2003; Grevemeyer et al.,
2007), and at the Central Chile trench outer rise (Grevemeyer et al.,
2005; Contreras-Reyes et al., 2007). In both areas, the surface heat flow
shows a negative anomaly if any (Grevemeyer et al., 2005). The
exothermic nature of the reaction is not discussed and the reduced
heat flow is interpreted as the cooling effect of the downward flow of
water (Ranero et al., 2003; Grevemeyer et al., 2005; Emmanuel and
Berkowitz, 2006) that overcomes the nascent thermal heating at depth.
The contradiction with our results is only apparent. At fast subduction
rate (8 cm/yr in this case), bending faults do not reside for a long time in
the “hydration”window (0.3 to 0.7 Myr Ranero et al., 2003), and heat
will distribute within the cold wedge once the lithosphere has reached
the trench. More data is needed at slow subductions to test this
hypothesis.
5. Conclusion
Among the physical mechanisms that may be responsible for the
high heat flow in the actively deforming Central Indian Basin, our
models indicate that exothermic serpentinization is the only process
that leads to a significant amount of heat release. In contrast to
friction-related release of heat, the exothermic serpentinization
model is by essence transient and delivers a pulse of extra heat that
will rapidly decay with time once serpentinization is completed. We
quantitatively show that a pulse of serpentinization in the Central
Indian Basin that would have started with the onset of the intraplate
deformation there is compatible with the present-day surface heat flow
which is 25–30 mW/m
2
above normal, provided a million year delay for
the establishment of a mature hydrothermal circulation and a high
enough thermal efficiency of the hydration reactions. The moderate
transient temperature anomaly is followed by a definitive weakening of
the CIB fault network due to the low friction coefficient of partially
serpentinized peridotites (Escartin et al., 1997, 2001). In the Wharton
Basin, the thermal pulse has already passed, in relation to a totally
different geometry of the fault network (widely-spaced vertical strike-
slip faultsrather that closely spaced reverse faults). The enigmatic bright
reflectors that we imaged in the CIB on multichannel seismic profiles at
depth of 8 to 15 km below the top of the crust coincide with the location
of the maximum reaction rate coefficient of serpentinization in our
model. They are likely to result from serpentinization fronts propagation
at mantle depth. Our results suggest that pulses of serpentinization
within the oceanic lithosphere result in transient rheology that breaks
the age-dependent mechanicalstrength law: this mechanism may have
been responsible in the past for sudden plate reorganizations at global
scale, andit may also explain theabnormally low rigidity foundtoday for
some of the oldest oceanic plates (Bry and White, 2007; Korenaga,
2007).
Acknowledgements
We thank three anonymous reviewers for their constructive
reviews and the editor Claude Jaupart for his stimulating comments.
Phèdre multichannel lines were reprocessed with the GeoCluster
seismic software of CGGVeritas in the framework of the Enabling Grids
for E-sciencE project, co-funded by the European Commission through
the Sixth Framework Programme (http://www.eu-egee.org).
References
Bry, M., White, N., 2007. Reappraising elastic thickness variation at oceanic trenches.
J. Geophys. Res. 112, B08414. doi:10.1029/2005JB004190.
Bull, J.M., Scrutton, R.A., 1990a. Sediment velocities and deep structure from wide angle
reflection data around Leg 116 sites. Proc. Ocean Drill. Prog. Sci. Results 116, 311–316.
Bull, J.M., Scrutton, R.A., 1990b. Fault reactivation in the central Indian Ocean and the
rheology of oceanic lithosphere. Nature 344, 855–858.
Bull, J.M., Scrutton, R.A., 1992. Seismic reflection images of intraplate deformation,
Central Indian Basin, and their tectonic significance. J. Geol. Soc. 149, 955–966.
Ćadek, O., Fleitout, L., 2003. Effect of lateral viscosity variations in the top 300 km on the
geoid and dynamic topography. Geophys. J. Int. 152, 566–580.
Carslaw, H.S., Jaeger, J.C., 1959. Conduction of Heat in Solids. Oxford University Press,
New York. 510 pp.
Chamot-Rooke, N., Jestin, F., De Voogd, B., the Phèdre Working Group,1993. Intraplate
shortening in the central Indian Ocean determined from 2100-km-long north–
south deep seismic reflection profile. Geology 21, 1043–1046.
Christensen, N.I., 2004. Serpentinites, peridotites, and seismology. Int. Geol. Rev. 46,
795–816.
Cochran, J.R., et al., 1989. Intraplate deformation and Bengal Fan sedimentation:
background and objectives. Proc. Ocean Drill. Program, Initial Rep. 116, 3–11.
Contreras-Reyes, E., Grevemeyer, I., Flueh, E.R., Scherwath, M., 2007. Alteration of the
subducting oceanic lithosphere at the southern central Chile trench-outer rise.
Geochem. Geophys. Geosys. 8, Q07003. doi:10.1029/2007GC001632.
Crosby, A.G., McKenzie, D., Sclater,J.G., 2006. The relationship between depth, age and gravity
in the oceans. Geophys. J. Int. 166, 553–573. do i: 10 .1111/j .13 65 -2 4 6X .2 0 06. 03 015. x.
D'Alessio, M.A., Williams, C.F., Burgmann, R., 2006. Frictional strength heterogeneity
and surface heat flow: implications for the strength of the creeping San Andreas
fault. J. Geophys. Res. 111, B05410. doi:10.1029/2005JB003780.
Delescluse, M., Chamot-Rooke, N., 2007. Instantaneous deformation and kinematics of the
India–Australia plate.Geophys. J. Int. 168,818–842. doi:10.1111/j.1365-246X.2006.03181.x.
Delescluse, M., Montési, L.G.J., Chamot-Rooke, N., 2008. Fault reactivation and selective
abandonment in the oceanic lithosphere. Geophys. Res. Lett. 35, L16312.
doi:10.1029/2008GL035066.
Deplus, C., Diament, M., Hébert, H., et al., 1998. Direct evidence of activedeformation in
the eastern Indian oceanic plate. Geology 26 (2), 131–134.
Emmanuel, S., Berkowitz, B., 2006. Suppression and stimulation of seafloor hydrothermal
convection by exothermic mineral hydration. Earth Planet. Sci. Lett. 243, 657–668.
Escartin, J., Hirth, G., Evans, B., 1997. Effects of serpentinization on the lithospheric
strength and the style of normal faulting at slow spreading ridges. Earth Planet. Sci.
Lett. 151, 181–189.
Escartin, J., Hirth, G., Evans, B., 2001. Strength of slightly serpentinized peridotites:
implication for the tectonics of oceanic lithosphere. Geology 29 (11), 1023–1026.
Francis, T.J.G.,1981. Serpentinization faults and their role in tectonics of slow spreading
ridges. J. Geophys. Res. 86, 11616–11622.
Fruh-Green, G.L., Kelley,D.S., Bernasconi, S.M., Karson, J.A., Ludwig, K.A ., Butterfield, D.A.,
Boschi, C., Proskurowski, G., 2003. 30,000 years of hydrothermal activity at the lost
city vent field. Science 301, 495–498.
Fyfe, W.S., 1974. Heats of chemical reactions and submarine heat production. Geophys.
J. R. Astron. Soc. 37, 213–215.
Geller, C.A., Weissel, J.K., Anderson, R.N.,1983. Heat transfer and intraplate deformation
in the Central Indian Ocean. J. Geophys. Res. 88, 1018–1032.
Gerbault, M., 2000. At what stress level is the central Indian Ocean lithosphere
buckling? Earth Planet. Sci. Lett. 178, 165–181.
Gordon, R.G., DeMets, C., Argus, D.F., 1990. Kinematic constraints on distributed
lithospheric deformation in the equatorial Indian-ocean from present motion
between the Australian and Indian plates. Tectonics 9 (3), 409–422.
Grevemeyer, I., Kaul, N., Diaz-Naveas, J.L., Villinger, H.W., Ranero, C.R., Reichert, C., 2005.
Heat flow and bending-related faulting at subduction trenches: case studies of
Nicaragua and Central Chile. Earth Planet. Sci. Lett. 236, 238–248.
Grevemeyer, I., Ranero, C.R., Flueh, E.R., Klaschen, D., Bialas, J., 2007. Passive and active
seismological study of bending-related faulting and mantle serpentinization at the
Middle America trench. Earth Pla net. Sci. Lett. 258, 528–542. doi:10.1016/j.
epsl.2007.04.013.
Jestin, F., (1994). Cinématique rigide et déformations dans la jonction triple Afar et dans
le Bassin Indien Central. PhD thesis, Université Paris VI.
Kelley, D.S., Karson, J.A., Blackman, D.K., Fruh-Green, G.L., 2001. An off-axis hydrothermal
vent field near the Mid Atlantic Ridge at 30 degrees N. Nature 412,145–149.
Korenaga, J., 2007. Thermal cracking and the deep hydration of oceanic lithosphere: a
key to the generation of plate tectonics? J. Geophys. Res. 112, B05408. doi:10.1029/
2006JB004502.
Krishna, K.S., Bull, J.M., Scrutton, R.A., 2001. Evidence for multiphase folding of the
Central Indian Ocean lithosphere. Geology 29 (8), 715–718.
Krishna, K.S., Gopala Rao, D., Neprochnov, Y.P., 2002. Formation of diapiric structure in
the deformation zone, Central Indian Ocean: a model from gravity and seismic
reflection data. Proc. Indian Acad. Sci. 111, 17–28.
Lachenbruch, A.H., Sass, J.H., 1980. Heat flow and energetics of the San Andreas fault
zone. J. Geophys. Res. 85, 6185–6222.
Lachenbruch, A.H., Sass, J.H., 1992. Heat flow from Cajon Pass, fault strength, and
tectonic implications. J. Geophys. Res. 97, 4995–5015.
Levchenko, 0.V., Verzhbitsky, E.V., 2003. Structural aspects of the intraplate deforma-
tions of the oceanic crust north of the Afanasy Nikitin rise (Indian Ocean.
Oceanology 43 (6), 840–851.
Louden, K.E., 1995. Variation in crustal structure related to intraplate deformation:
evidence from seismic refraction and gravity profiles in the Central Indian Basin.
Geophys. J. Int. 120, 375–392.
150 M. Delescluse, N. Chamot-Rooke / Earth and Planetary Science Letters 276 (2008) 140–151

Lowell, R.P., Rona, P.A., 2002. Seafloor hydrothermal systems driven by the serpenti-
nization of peridotite. Geophys. Res. Lett. 29 (26), 1–4.
MacDonald, A.H., Fyfe, W.S., 1985. Rate of serpentinization in seafloor environments.
Tectonophysics 116, 123–135.
O'Hanley, D.S., 1992. Solution to the volume problem in serpentinization. Geology 20,
705–708.
Okal, E.A., 1983. Oceanic intraplate seismicity. Annu. Rev. Earth Planet. Sci.11,195–214.
Ormond, A., Boulègue, J., Genthon, P., 1995. A thermoconvective interpretation of heat
flow data in the area of Ocean Drilling Program Leg 116 in a distal part of the Bengal
Fan. J. Geophys. Res. 100, 8083–8095.
Pollack, H.N., Hurter, S.J., Johnson, J.R.,1993. Heat flow from the earth's interior: analysis
of the global data set. Rev. Geophys. 31 (3), 267–280.
Ranero, C.R., Phipps Morgan, J., McIntosh, K., Relchert, C., 2003. Bending-related faulting
and mantle serpentinization at the Middle America trench. Nature 425, 367–373.
Sandwell, D.T., Smith, W.H.F., 1997. Marine Gravity from Geosat and ERS-1 Altimetry. J.
Geophys. Res. 102, 10039–10054.
Scholz, C.H., 2000. Evidence for a strong San Andreas fault. Geology 28 (2), 163–166.
Shipboard Scientific Party,1989. Site 718: Bengal Fan.In: Cochran, J.R., Stow, D.A.V., et al.
(Eds.), 1988 Proc. ODP, Init. Repts., College Station, TX (Ocean Drilling Program),
vol. 116, pp. 91–154.
Stein, C.A., Weissel, J.K., 1990. Constraints on the Central Indian Basin thermal structure
from heat flow, seismicity and bathymetry. Tectonophysics 176, 315–332.
Stein, C.A., Hobart, M.A., Abbott, D.A., 1988. Has the Wharton Basin's heat flow been
perturbed by the formation of a diffuse plate boundary in the Indian Ocean?
Geophys. Res. Lett. 15, 455–458.
VanOrman, J., Cochran, J.R., Weissel, J.K., et al., 1995. Distribution of shortening between
the Indian and Australian plates in the Central Indian Ocean. Earth Planet. Sci. Lett.
133, 35–46.
Verzhbitsky, E.V., Lobkovsky, L.I., 1993. On the mechanism of heating-up of the Indo-
Australian plate. J. Geodyn. 17, 27–38.
Verzhbitsky, E.V., Neprochnov, Y.P., 2005. Deep structure of the central Indian Ocean
inferred from geophysical data. Geotectonics 39 (3), 213–223.
Wang, K., Mulder,T., Rogers, G.C., Hyndman, R.D., 1995. Case for verylow coupling stress
on the Cascadia subduction fault. J. Geophys. Res. 100, 12907–12918.
Weissel, J.K., Anderson, R.N., Geller, C.A.,1980. Deformation of the Indo-Australian plate.
Nature 287, 284–291.
Williams, C.F., 1990. Hydrothermal circulation and intraplate deformation: constraints
and predictions from in-situ measurements and mathematical models. In: Cochran,
J.R., Stow, D.A.V., et al. (Eds.), Proc. ODP, Sci. Results, 116:College Station, TX (Ocean
Drilling Program), pp. 345–359.
Wiens, D.A., Stein, S., DeMets, C., Gordon, R.G., Stein, C., 1986. Plate tectonic models for
Indian-Ocean intraplate deformation. Tectonophysics 132, 37–48.
Ziolkowsky, A., Hanssen, P., Gatliff, R., Jakubowicz,H., Dobson, A., Hampson,G., Li, X.Y., Liu,
E., 2003. Use of low frequencies for sub-basalt imaging. Geophys. Prospect. 51 (3),
169–18 2.
151M. Delescluse, N. Chamot-Rooke / Earth and Planetary Science Letters 276 (2008) 140–151