Earth and Planetary Science Letters 393 (2014) 132–145

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

Obduction: Why, how and where. Clues from analog models

P. Agard a,b,∗ , X. Zuo a , F. Funiciello c , N. Bellahsen a , C. Faccenna c , D. Savva a
a
Sorbonne Universités, UPMC Univ Paris 06, UMR 7193 CNRS-UPMC, Institut des Sciences de la Terre Paris (ISTeP), F-75005 Paris, France
b
Institut Universitaire de France, F-75005, Paris, France
c
Dipartimento di Scienze Geologiche, Universita degli Studi “Roma TRE”, Rome, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Obduction is an odd geodynamic process characterized by the emplacement of dense oceanic “ophiolites”
Received 17 October 2013 atop light continental plates in convergent settings. We herein present analog models specifically
Received in revised form 5 February 2014 designed to explore the conditions (i.e., sharp increase of plate velocities — herein coined as ‘acceleration’,
Accepted 6 February 2014
slab interaction with the 660 km discontinuity, ridge subduction) under which obduction may develop as
Available online 15 March 2014
Editor: Y. Ricard
a result of subduction initiation.
The experimental setup comprises an upper mantle modeled as a low-viscosity transparent Newtonian
Keywords: glucose syrup filling a rigid Plexiglas tank and high-viscosity silicone plates. Convergence is simulated by
obduction pushing a piston with plate tectonics like velocities (1–10 cm/yr) onto a model comprising a continental
subduction margin, a weakness zone with variable resistance and dip (W ), an oceanic plate (with or without a
slab dynamics spreading ridge), a preexisting subduction zone (S) dipping away from the piston and an upper active
ophiolites continental margin, below which the oceanic plate is being subducted at the start of the model (as for
mechanical coupling
the Neotethyan natural example).
Several configurations were tested over thirty-five parametric models, with special emphasis on
comparing different types of weakness zone and the degree of mechanical coupling across them.
Measurements of displacements and internal deformation allow for a precise and reproducible tracking
of deformation. Models consistently demonstrate that once conditions to initiate subduction are reached,
obduction may develop further depending on the effective strength of W . Results (1) constrain the
range of physical conditions required for obduction to develop/nucleate and (2) underline the key
role of such perturbations for triggering obduction, particularly plate ‘acceleration’. They provide an
explanation to the short-lived Peri-Arabic obduction, which took place along thousands of km almost
synchronously (within ∼50–10 Myr), from Turkey to Oman, while the subduction zone beneath Eurasia
became temporarily jammed. They also demonstrate that the emplacement of dense, oceanic material
on continental lithosphere is not a mysterious process requiring extraordinary boundary conditions but
results from large-scale, normal (oceanic then continental) subduction processes.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction Why, how and where obduction develops is still poorly con-
strained. Whether obduction nucleates on preexisting discontinu-
Within the frame of plate convergence, obduction (Coleman, ities such as mid-ocean ridges, transform faults or ocean-continent
1971) is an apparent geodynamic anomaly, whereby fragments transition zones is unclear (see below). Whether obduction is
of dense oceanic lithosphere — “ophiolites”, are emplaced onto driven by additional horizontal forces, tied to mantle-scale su-
light, deeply buried continental margins over distances of several perplumes (Vaughan and Scarrow, 2003) or to lithospheric-scale,
hundred kilometers (e.g., Oman, Turkey, Balkans, Newfoundland, abrupt changes of plate velocities (coined as plate ‘acceleration’;
New Caledonia, Papua). Obduction appears as a transient (i.e., < Agard et al., 2007; Fig. 1a) is also an open question. Constrain-
∼10–15 My) yet recurring geodynamic process through time (Ab- ing which physical properties drive obduction is thus important
batte et al., 1985; Nicolas, 1989; Vaughan and Scarrow, 2003), with not only to regional geodynamics but also to our understanding of
recent Mio-Pliocene examples in SE Asia (Linthout et al., 1997; plate tectonics.
Pubellier et al., 2004). Previous interpretations on how obduction forms (see also
Michard et al., 1985; Moores et al., 2000) comprise two main
types: (1) thrusting of ophiolite (Coleman, 1971) via “flake tec-
* Corresponding author. tonics” (Oxburgh, 1972; Vaughan and Scarrow, 2003), or (2) con-

http://dx.doi.org/10.1016/j.epsl.2014.02.021
0012-821X/© 2014 Elsevier B.V. All rights reserved.
 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145 133

Fig. 1. a. Simplified reconstructed geodynamic situation at the inception of intra-oceanic subduction for the Oman (Neotethyan) case study. The high temperature (HT)
metamorphic sole marks the onset of a newly formed subduction at ∼95 Ma, while the preexisting one is jammed, as testified by the exceptional, transient exhumation
of deeply seated blueschist facies (BS) rocks along the subduction plane. This subduction initiation will ultimately lead to the emplacement of the Neotethys ophiolite onto
the Arabian continent (to the left). Subduction initiation coincides with a period of plate reorganization and increased convergence velocities (×2–3) after 115 Ma. b. Model
set-up comprising a preexisting subduction zone (S) and a weak discontinuity (W ). To avoid edge effects the plates are inserted in the middle of the tank. See text for
details. c. Top-view of the model set-up, showing the position of the measurements systematically performed on photographs taken every 30 or 60 s (as well in side-views).
d. Experiments performed are here given as a function of the type of initial weakness introduced. Bold larger font size letters: models detailed in the manuscript. Others:
normal font: runs for which obduction was successfully reproduced; between brackets: unsuccessful obduction.

tinental subduction beneath an oceanic upper plate (Mattauer et is beyond the scope of this study, all of them are characterized by
al., 1981; Boudier et al., 1988; Agard et al., 2007; Fig. 1a). In the ophiolite relative displacements >100’s of km within 10–15 My,
latter interpretation, ophiolite emplacement is mostly viewed as by the presence of an extensive HT metamorphic sole at the base
‘passive’ and is essentially a consequence of subduction initiation, of the metamorphosed, several km thick ophiolite, and of HP con-
one of the frequent outcomes of plate reorganization on a regional- tinental rocks beneath. Major unresolved issues for all of them
scale (i.e., on the order of several hundreds of km; Dewey and Bird, comprise uncertainties as to their exact geochemical affinity and
1971; Gurnis et al., 2004; Stern, 2004). geodynamic setting (i.e., MORB, back-arc or fore-arc type; Dewey,
We herein explore, through analog models, the conditions un- 1976; see below for Oman), as to their thermal structure at the
der which obduction may develop as a result of subduction initia- time of emplacement, or why ophiolite thickness and the pressure
tion, with the following aims: recorded by the metamorphic sole welded at its base generally
(1) study if and how obduction can develop and which pa- do not match (e.g., the ophiolite “conundrum”; Hacker and Gnos,
rameters are most favorable (i.e., lithospheric strength, buoyancy 1997; Moores et al., 2000). We thus focus in the following on the
contrasts, etc.); best known example, namely the Oman obduction, taken here as
(2) understand which perturbations (among which plate ‘accel- a generic representative case study and which has been preserved
eration’, ridge subduction and/or interaction with the 660 discon- from later collision.
tinuity; hereafter referred to as A, R and D, respectively) may
trigger subduction initiation and obduction, and the nature and 2.1. Reference geological setting
range of forces required;
(3) constrain the partitioning and force balance between two Our model configuration is directly inspired from the regional
interacting subduction zones, which has so far never been mod- context of obduction of the Oman ophiolite, which took place
eled. on the southern side of the Neotethyan ocean (Fig. 1a). As for
other Peri-Arabic obducted ophiolites from Turkey to Oman (Ricou,
2. Model set-up 1971), a stage of subduction initiation is documented at 95 Ma by
the formation of extensive metamorphic soles (Thuizat et al., 1981;
Several reasonably well-preserved, large-scale obducted ophio- Hacker, 1994; Wakabayashi and Dilek, 2003). This event coin-
lites worldwide are available for study (e.g., Oman: Coleman 1971, cides with blueschist exhumation on a comparable scale on the
1981; Newfoundland: Dewey and Bird, 1971; Suhr and Cawood, northern side of the Neotethys (Monié and Agard, 2009; Fig. 1a)
1993; Turkey: Çelik et al., 2011; Balkans: Pamić et al., 2002; New and follows a major, regional-scale geodynamic reorganization at
Caledonia: Cluzel et al., 2001, Lagabrielle et al. (2013) Papua: Lus ∼115 Ma marked by a rapid increase in the rate of shortening of
et al., 2004). Although a review of their detailed geological settings the Neotethyan ocean (Agard et al., 2007, 2006).
134 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145

The obducted Oman ophiolite, most likely the best preserved velocities and can be viewed as corresponding to ridge push or
on earth (it was selected as the classic ophiolite type by the additional horizontal forces (Bott, 1993). Such forces, in the case
Penrose conference, in 1972), has been the target of numerous of Oman, were likely provided by the late Cretaceous Pacific super-
structural, petrological and geochemical studies, largely devoted swell and opening of the Southern-Central Atlantic (Vaughan and
to processes of oceanic lithosphere formation (e.g., Nicolas, 1989; Scarrow, 2003; Agard et al., 2007).
Godard et al., 2000, 2003; Nicolas et al., 2000; Shervais, 2001; The choice of placing an initial weakness W in the model (as
Python et al., 2008). imposed by Chemenda et al., 1996, 1995) is dictated by the lack of
It is generally agreed that the Oman ophiolite once formed possible localization due to the quasi-Newtonian behavior of sili-
in the Neotethyan realm (which formed from the Permo-Triassic cone material over laboratory timescales (see below; as opposed
onwards; Lippard et al., 1986; Chauvet et al., 2009) and was em- to Faccenna et al., 1999). This is also justified by the existence of
placed onto Arabia as a result of a NE-dipping subduction zone ubiquitous, inherited weak tectonic discontinuities in nature (i.e.,
initiated at 95 Ma. Stable sedimentation along the north Arabian ancient suture zones, rift-related normal faults, subsiding basins,
platform indeed abruptly ended at the end of the Cenomanian, transform faults, etc.; Hacker et al., 1996; Chauvet et al., 2009;
90 Ma ago (Béchennec, 1987), when the passive margin was flexed Lévy and Jaupart, 2012). Such discontinuities are in fact systemat-
down, turned into a foreland basin (Glennie et al., 1973, 1974) and ically introduced for nucleating subduction initiation in numerical
subducted shortly afterwards as shown by HP metamorphic con- models (e.g., Toth and Gurnis, 1998; Funiciello et al., 2003; Gurnis
ditions (i.e., between ∼90–85 Ma and ∼75 Ma; Searle and Cox, et al., 2004; Yamato et al., 2007a; Lévy and Jaupart, 2012; Duretz
1999; El-Shazly et al., 2001; Warren et al., 2003, 2005; Yamato et and Gerya, 2013). However, by contrast to Chemenda et al. (1995,
al., 2007b). 1996) or Boutelier et al. (2003, 2004), there is no predefined sub-
However, despite numerous petrological and geochemical stud- duction in our models (at W ) where obduction may nucleate.
ies, the exact nature and original setting of the Oman ophio- The absence of net extension in the lithospheric plates of our
lite is still debated, with evidence for both a MORB-type sig- models means that the amount of slab-pull driven oceanic sub-
nature (Boudier et al., 1988; Ceuleneer et al., 1988; Nicolas et duction never exceeds the amount of overall shortening imposed
al., 2000; Godard et al., 2000, 2003; Le Mée et al., 2004) and a by the piston. The geodynamic situation modeled here thus cor-
supra-subduction, back-arc geochemical imprint (Shervais, 2001; responds to a case such as in the Indian Ocean, where intrao-
Searle et al., 2004; McLeod et al., 2013). ceanic compressive deformation has long been documented (Bull
Irrespective of its exact geodynamic setting, age constraints and Scrutton, 1990).
for the ophiolite crustal sequence (∼110–95 Ma, with mainly
96.5–95.5 Ma ages; Ceuleneer, 1986, p. 149 and references therein; 3. Methods
Rioux et al., 2012, 2013), for syn-obduction cross-cutting pla-
giogranite dykes derived from the metamorphic sole 3.1. Materials and similarity criteria
(∼95 Ma: Tilton et al., 1981) and for the metamorphic sole it-
self (∼95–94 Ma; Hacker, 1994), indicate that the Oman ophio- Lithospheric plates are treated as thin viscous sheets on top of
lite corresponds to a young portion of oceanic lithosphere (i.e., glucose syrup and arranged in a rectangular Plexiglas tank (110 cm
<1–10 My) formed close to the Arabian continental margin. long, 60 cm wide and 20 cm high; Fig. 1b) much larger than the
Given the short time span between subduction initiation plate so as to minimize lateral friction. Following previous analog
(95 Ma) and peak burial of the edge of the Arabian continental studies (Bellahsen et al., 2003 and references therein), the rheology
lithosphere (∼80 Ma; Warren et al., 2003, 2005), any hetero- of the lithosphere is approximated by a visco-elastic material (sil-
geneity localizing deformation and thereby subduction initiation icone putty; Table 1) behaving viscously at laboratory strain rates,
would indeed have had to be close to Arabia (i.e., <200–300 km; as the experimental time scale is higher than the Maxwell relax-
Searle et al., 2004). However, whether subduction initiated near ation time (∼1 s). This type of material has been shown to repro-
the continent-ocean transition (COT; e.g., Ishikawa et al., 2005; duce the most essential features of subduction processes (Becker et
Lévy and Jaupart, 2012) or at some other tectonic discontinu- al., 1999; Funiciello et al., 2003; Schmeling et al., 2008). The upper
ity, such as a ridge (e.g., Nicolas and Le Pichon, 1980; Boudier mantle is simulated by a low-viscosity and high-density Newtonian
et al., 1988) or transform fault (Hacker et al., 1996; Rioux et al., fluid (glucose syrup).
2013), remains an open question. Whether an active Neotethyan Physical parameters used to scale our laboratory models, in-
ridge even existed at the time is also unknown (Ricou, 1994; cluding densities, were set through similarity criteria by Bellahsen
Dercourt et al., 2000). et al. (2003; Table 1). This experimental setting is properly scaled
for normal gravity field to simulate the competition between act-
2.2. Experimental set-up ing gravitational and viscous resistive forces stored within the
mantle and the lithosphere adopting adimensional number such
The model set-up (Fig. 1b) corresponds to the Neotethyan case as buoyancy number (B, Table 1; e.g., Weijermars and Schmeling,
study during the late Cretaceous (Fig. 1a): convergence is simu- 1986; Davy and Cobbold, 1991; Becker et al., 1999). The viscosity
lated by a piston pushing onto a continental passive margin (with and density ratio between the lithosphere and the mantle is set
variable continent to ocean transition structure; COT), an oceanic to 1.02 and 400, respectively to properly simulate the range ob-
plate and an active upper plate continental margin, under which served in nature (Mitrovica and Forte, 1997). The scaling factor is
a pre-existing subduction zone is set. An accretion ridge (R), char- 1.6 × 10−7 : 1 cm in the model corresponds to 60 km, and the char-
acterized by a lighter and weaker silicone putty, is inserted with acteristic time is such that 1 min equals 1 Myr. The convergence
the oceanic plate in some models. In Fig. 1b, labels S (subduction rate, which is controlled by the velocity of the piston, varies ac-
zone) and W (Weakness zone; see Fig. 1d) respectively denotate cordingly between 3.5 and 13.5 cm/yr. Intertia is negligible in the
plate boundaries where subduction takes place and where subduc- experiment as in nature. In some experiment, the velocity of con-
tion initiation and later obduction are expected to take place. vergence (i.e., piston velocity) increase sharply by a factor of 2.5 or
The S boundary behaves as a proxy for existing subduction 1.75 (Table 3).
zones, either adjacent (as was the case in New Caledonia or SE The bottom of the Plexiglas tank corresponds to the upper–
Asia; Cluzel et al., 2001; Stern, 2004) or more distant. The push lower mantle boundary at ∼660 km, treated here as an imper-
from the piston was set so as to produce adequate convergence meable barrier as direct penetration of the slab through this layer
 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145 135

Table 1
List of physical parameters of the model and their corresponding natural values. Piston velocities were varied between 0.58 and
2.26 cm/mn (i.e., 13 and 50 Hz), which scales back to kinematic velocities of 3.48 to 13.56 cm/yr.

Parameters Units Model Nature

g Gravitational acceleration m s−2 9.81 9.81

Thickness
ho Oceanic lithosphere m 0.012 70 000
hc Continental lithosphere 0.016 100 000
z Upper mantle 0.11 660 000
Ridge 0.012 100–30 000
Thickness ratio (ho / z) 0.11 0.10
Thickness ratio (hc / z) 0.145 0.15
Length scale factor = 1.6 × 10−7
(model/nature)

Density kg m−3
ρo Oceanic lithosphere 1449 3300
ρc Continental lithosphere 1394 3000
ρm Upper mantle 1420 3220
ρr Ridge 1070 2850–3100
Density ratio (ρo /ρm ) 1.02 1.02
Density contrasts (ρ = ρo − ρm ) 29 80

Viscosity Pa s
μo Oceanic lithosphere 2 × 105 4 × 1023
μc Continental lithosphere 1.4 × 105 1023
μm Upper mantle 460 4 × 1021
μr Ridge
Viscosity ratio (μo /μm ) 435 100–500
Viscosity ratio (μo /μc ) 1.4 4

Characteristic time
t (t nature /t model (μn ρm g m lm /μm ρn g n ln )) s 60 (1 mn) 3 × 1013 (1 Myr)

Buoyancy number
B (B = ho2 g ρ /μm v) 0.50 0.54

Convergence rate
v (piston) cm/mn 0.58–2.26 3.48–13.56 (cm/yr)

is inhibited during the timescale of our experiments (a few tens al., 2000), which will depend on ridge density, size and orien-
of millions years). Note that the viscosity increases at the upper– tation,
lower mantle boundary is at least one order of magnitude (Davies, – interaction with the upper–lower mantle discontinuity (D),
1995; Guillou-Frottier et al., 1995; Christensen, 1996; Funiciello which has been shown to be a major barrier for some slabs
et al. 2003). No thermal gradient exists across the model. The (e.g., Houseman and Gubbins, 1997), with time scales of slab
thermal consequence of the aging oceanic lithosphere and phase penetration on the order of a few Myrs.
transformations increasing crustal density are implicitly taken into
account by the choice of a negatively buoyant oceanic plate. The initial strength of W was qualitatively controlled using var-
The influence of these obvious simplifications with respect to ious types of weaknesses (Fig. 1d), both in terms of location and
nature (i.e., viscous rheology, lack of temperature gradient, ab- resistance, from very weak (W 0; throughgoing fault) to very strong
sence of density changes within the slab, exaggerated density (W max). The weakest types (W 0 and W 1) are characterized by a
contrast across the upper/lower mantle interface) are discussed COT shape allowing for an initial overlap and lubrication of the
later. W interface (using Vaseline and glucose syrup, as indicated by the
bold line in Fig. 1c). Models with weakness zones as small putty
bridges, either lubricated or not (W 3, W 3 ), were also tested. Lu-
3.2. Parametric study and measurements
brication was not used for models W 4 to W max. The location of
W was chosen so as to initiate obduction close to the continen-
Ten preliminary tests and twenty-six parametric models were tal margin, either within the ocean or at the COT, and the density
performed using variable combinations of material, strain rates and contrast across W was modified accordingly (e.g., #19).
boundary conditions (see supplementary material). Eleven mod- The experimental procedure is as follows: (1) subduction is ini-
els were performed with an oceanic ridge, twenty-one included tiated by forcing the plate tip inside the syrup down to a depth
a sharp increase in convergence velocity (i.e., ‘acceleration’) and where gravitational forces overcome resisting forces (i.e., 2.5 cm
the slab reached the bottom of the tank in twenty of them. The ∼150 km in nature) and subduction evolves to a self-sustaining
parametric study was not only meant at successfully modeling ob- process (McKenzie, 1977), (2) the upper plate, lubrificated beneath
duction as a result of subduction initiation but also at assessing its forearc to enhance decoupling between the two plates and pro-
the respective influence of the following triggers: mote subduction, is then added and (3) the overall system is short-
ened by means of the piston (i.e., velocity boundary conditions).
– sharp changes in plate velocity (‘acceleration’, A), as hypothe- Experiments lasted ∼30–40 min and were monitored through top-
sized by Agard et al. (2007), and side-view photographs every 30 or 60 s. Eleven characteristic
– entrance of a buoyant oceanic ridge (R), as for the Californian lengths (Fig. 1c; Table 2) were measured at each time step using
or Taitao ridges (Page and Engebretson, 1984; Lagabrielle et regularly-spaced parallel lines drawn atop the plates to calculate
136 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145

Table 2
Variables and units used in the manuscript. Abbreviations: n.d.: dimensionless parameter; n subscript refers to values normalized to the same amount of total model
shortening; i and f to initial and final values; w ◦ : value of w at time t ◦ . Measured values (in cm; see Fig. 1c): w, s1, s2, u, im, io, iu, lm, ir, lo, lu, t, F . Other variables:
c = u − iu; i = im + io + iu; l = (lm + lo + lu )/3; p = w + s2 + i; s = s2 − c; SU = u ∗ lu. Note that %w + %s + %c + %i adds up to 100%.

Variable Type Unit

D w = wi − wt amount of convergence across W cm
%w = D w / Dp fraction of overall convergence taken up by W n.d.
t◦ time at which models are stable (for calculation of w ◦ ) mn
t ∗ = t − t acceleration time referenced with respect to ‘acceleration’ (Fig. 5b) mn
D%w = %w − %w ◦ variations in %w with respect to time t ◦ (i.e., %w ◦ ) n.d.
D w f , %w f , D%w f final values of D w, %w, %D w
eW n = ( D w f / D w ◦ )n deformation ratio across W n.d.
w ∗ = ( w t − w t +dt )/dt instantaneous deformation rate across W cm mn−1
F fraction of effective obduction on overall convergence across W (= S w / D w; Fig. 2b) n.d.
O = %w f . F fraction of obduction with respect to convergence across W and S n.d.
Ds, %s, D%s, e S n , s∗ subduction across S as for w
Dp, e P , p ∗ along strike model shortening as for w
%D P n = ( p i − p )/( p i − p f )n fraction or progress of overall model shortening (in %; normalized for all runs) — see supplementary material n.d.
RW q calibrated resistance of W (Fig. 1d) qualitative
R W = (lot +dt − lot )/( w t − w t +dt ) resistance of W (estim. before perturbation) n.d.
R S = (lut +dt − lut )/(st − st +dt ) resistance of S (estim. after perturbation) n.d.
C = (lut − lut +dt )/(s2t − s2t +dt ) · 100 fraction of fore-arc deformation with respect to convergence across S n.d.

Table 3
Summary of the characteristics and main results for each model. R W q corresponds to the arbitrary resistance value derived from the weakness type
(i.e., W 1 to W 5 for models considered here) which is plotted in Fig. 5d. ‡ superscript denotate fractions including along strike (w, s, i) and out of
plane deformation (l).

# Experiment 11 12 13 15 17 18 18b 19
Weakness type W5 W 1 W2 W 3 W1 W 1 W 1 W 1
t◦ 8 8 8 8 8 5 5 7
t∗ 31 18 12 11 12 9 9 10
‘Acceleration’ ( A) ×2.5 ×2.5 ×2.5 ×2.5 ×2.5 ×1.75 ×1.75 ×1.75
Ridge (R) + + – – – – – –
Slab at 660 (D) 13 12 – 17 – 12 12 11.5

Effective obduction – + + + + + + +
Impact\ A + – ++ (+) ++ + (+) –
Impact\ R (+) ++
Impact\ D – (+) ++ (+) + –

eW n 33.48 59.54 67.79 52.82 65.07 81.10 75.87 47.20

e Sn 76.39 69.07 57.49 62.03 60.82 57.73 60.56 77.93
eP 45.19 51.11 48.53 57.20 51.23 39.74 39.74 45.45
%w f ‡ 14.02 23.03 32.70 22.36 23.98 26.23 26.76 17.27
%sf ‡ 56.03 54.34 46.04 49.19 58.03 51.21 51.57 64.40
%lf ‡ 11.53 8.14 8.04 14.76 7.45 9.14 8.78 7.05
%i f ‡ 18.42 14.48 13.22 13.68 10.54 13.41 12.89 11.29
%w ◦ 9.85 17.76 8.76 10.04 14.71 19.21 14.27 24.96
D%w f 4.86 5.77 23.52 11.42 9.02 18.20 16.46 −5.82
%W = %w f /(%w f + %sf ) 20.01 29.77 41.52 31.25 29.24 41.36 34.16 21.14
%S = %sf /(%w f + %sf ) 79.99 70.23 58.48 68.75 70.76 58.64 65.84 78.86

F 0.00 91.00 71.00 52.00 94.00 70.00 64.00 80.00

%O 0.00 27.09 29.48 16.25 27.49 28.96 21.86 16.91

RW q 10.00 1.50 2.00 3.50 1.00 1.50 1.50 1.50

RW 0.83 0.32 0.32 0.59 0.31 0.29 0.38 0.36
RS 0.47 0.33 0.69 0.87 0.21 1.40 0.75 0.15

finite and cumulative deformation (with Image-J software). These vided through run #18, which was measured across two different
measurements allow defining a number of (mainly dimensionless) transects.
variables such as the respective contributions through time of con-
vergence across W or S (see Table 2). Side-view measurements 4. Results
(though slightly blurred by the Plexiglass wall and less precise than
top-views) are used to evaluate the amount of effective new sub- One reference model (#13; Fig. 2) and six other representative
duction (or “obduction”). Obduction efficiency at W is defined as ones (#11, 12, 15, 17, 18, 19) are presented below. Sketches of
the amount of effective underthrusting (or subduction) over the to- their evolution and systematic measurements are shown in Figs. 3
tal amount of convergence (F ; Table 2; see Fig. 2b), and is used to and 4, respectively.
estimate the integrated contribution of obduction to overall con-
vergence (%O ; Table 2). 4.1. Reference model (#13)
Although models evidence mainly in-plane deformation and
non-cylindrical set-ups have been avoided (except in #25, with the This model (weakness type: W 2; Fig. 1d) is devoid of intra-
presence of a transform fault), minor lateral contrasts exist in most oceanic ridge and the piston was accelerated after 12 My (i.e.,
models. An estimate of the extent of such variations is herein pro- 12 min), before the slab could touch the 660 km discontinuity.
 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145 137

Fig. 2. Detailed evolution of model #13. (a) Top views; (b) side-views (see also Fig. 3). Obduction efficiency (F ; Table 2) is defined as the amount of effective underthrusting
at W , i.e. F = S w / D W ; (c) evolution of characteristic lengths and surfaces (see Table 2); (d) comparison of internal and lateral deformation in the three parts of the model
set-up (M, O , U : Fig. 1c); (e) comparison of deformation velocities across S and W . Inset: evolution of the fraction of convergence taken up by S as a function of the amount
of convergence on W .

Four sequential top (Fig. 2a) and lateral views (Fig. 2b) are shown conspicuous 4–5 min after the ‘acceleration’ (Fig. 2c). From this
after 10–15–17–21 min, together with the evolution of some key moment onward, obduction becomes effective (see side-view at
variables (Figs. 2c–e; Table 2). t = 17 min; Fig. 2b) and the fraction of convergence across S de-
Fig. 2a shows that deformation is mainly in-plane: there is creases (Fig. 2c). Subduction gets progressively jammed, as testified
very little rotation of the model and insignificant lateral escape of by the increase in fore-arc deformation (Dc) and the decrease of
the oceanic plate, as for most models. Subduction initiation fol- the upper plate surface (SU ).
lowed by obduction is conspicuous from Fig. 2b. Measurements The progressive evolution of instantaneous velocities best de-
allow for a good precision, as testified by the close agreement be- picts the mechanical behavior of the system and how deformation
tween the overall shortening rate deduced from measurements of is partitioned between W and S (Fig. 2e): shortly after the ‘accel-
the total length of the model and the along strike shortening of eration’ both W and S increase their shortening rates, but after
the model deduced from marker measurements (Fig. 2c: %Sh, %p; 5 min the upsurge in velocity ceases for S, whereas the contrary
see Table 1 for the extensive description of all the adopted pa- is observed for W . The proportion of the overall shortening taken
rameters). Both also fit the calibrated speed of the piston within up by S-subduction, shown as a function of the amount of con-
error. Similarly, the amount of subduction is very consistent for vergence across W (D w: inset, Fig. 2e), points to competing plate
both types of measurements, namely Ds1 and Ds2 (Fig. 2c). Fig. 2c boundaries and W progressively catching up with S. Finally, about
also indicates that forearc deformation (Dc) is only small during 9 min after the ‘acceleration’ (i.e., after 21 min), the contribution
most of the experiment (as also shown by Ds being close to Ds2). of W to model convergence further increases (Figs. 2c, e). Inter-
Fig. 2c gives the cumulative contribution of deformation across W nal deformation of the margin (Dim) no longer increases (Fig. 2d),
to along strike shortening of the model at a given time (%w). Inter- which indicates that all the deformation between the piston and
nal deformation, both in terms of width (Dlm, Dlo, Dlu) and length W is absorbed by obduction.
changes (Dim, Dio, Diu) evolves smoothly and similarly (Fig. 2d;
except Dim after ‘acceleration’: see below). 4.2. Results from the parametric study
The model evolves steadily after 2–3 min, once the resorp-
tion of all the small spaces left by the initial juxtaposition of the Six representative models (#11, 12, 15, 17, 18b, 19) are shown
plates is over. Convergence across W (%w) initially amounts to in Figs. 3 and 4 and compared to the reference one (#13; Figs. 2c,
only ∼10% of the overall convergence. A minor increase across W 4). For the sake of comparison, linedrawings of side-views at
takes place at the time of acceleration, yet a sharper increase is the start, during ‘acceleration’, ridge subduction, encounter of the
138 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145

Fig. 3. Sketches of slab shapes in side-view at critical time intervals: at the start of the experiment, during ‘acceleration’, ridge subduction, slab at 660 km discontinuity
(i.e., potential triggers A, R or D, respectively) and at the end of the experiment. Amounts of shortening on W (eW n) and effective obduction (%O ) are indicated (see also
Table 2). Inset: comparison for slab shapes at the time of ‘acceleration’. The black dot marks the transition between the top, bended part of the slab and the portion beneath.

upper-mantle discontinuity (i.e., potential triggers A, R or D, re- all these cases, once subduction initiates it is followed by obduc-
spectively) and at the end of the experiment are provided in tion;
Fig. 3. The main characteristics of each model, including the over- (b) those for which there is no obduction (e.g., #11). The overall
all amounts of model shortening, shortening across W and S, and shortening (eW n = 33.5%; Fig. 3) and slight increase in the fraction
their respective contributions are given in Table 3. Fig. 4 provides of convergence across W in this run (D%w f = 4.86%; Table 3) is
a continuous record of surface deformation, where the timing of thus not accommodated through obduction. The same holds true
A–R–D events is indicated together with other remarkable obser- for all models with W  3 (except #15).
vations (small arrows). The comparison of slab shapes at the time of ‘acceleration’ (in-
set, Fig. 3) evidences distinct bending curvatures. A black dot high-
lights the inflexion point, which marks the transition between the
4.2.1. Evolution of slab shapes top, bended part of the slab and the portion beneath. The contrast-
Comparison of the final shapes of subducted lithosphere shows ing lengths of the slabs are due to the different durations of these
that all but #13 and #17 rest on the 660 km discontinuity at the experiments prior to ‘acceleration’ (e.g., much longer for #11).
end of the experiment. Amounts of shortening on W (eW n) differ
from 33.5 to 75.9%. The amount of effective obduction, which de- 4.2.2. Evolution of deformation and shortening
pends on the amount of overall W -convergence weighted by the Along strike shortening (Fig. 4) is shown as the contributions
fraction of effective underthrusting (%O = %w f . F ; Table 2), varies from S (%s), W (%w), internal deformation (%i) and forearc defor-
between 16 and 29% (Fig. 3; Table 3; F varies from 0 to 94%). Two mation (%c), the latter reflecting changes in mechanical coupling
main types of behavior are readily observed in the models, which across the subduction interface. After the first few minutes, dur-
are typified by the examples selected here: ing which plates get in contact, oscillations are only minor (#13,
(a) those for which obduction effectively develops (#15, 17, 18, 17) or unnoticeable (#12, 15). Initial partitioning is such that the
18b, 19, as for #13), with a new slab clearly visible in side-views initial contribution of W to shortening (%w ◦ ) ranges between ∼9
(Fig. 3). Obduction is relatively minor for models #15 and #19. In and 25%. Initial amounts >10% are only observed for types W 1
 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145 139

Fig. 4. Detailed evolution of the contributions from S (%s), W (%w), internal deformation (%i) and forearc deformation for some selected models. Arrows indicate the timing
of ‘acceleration’ ( A; long black arrows), ridge subduction (R; thin arrows), interaction with the 660 discontinuity (D; thick grey arrows) or some other event (smallest arrows)
discussed in the text.

or W 1 , but are otherwise only loosely related to weakness types. for #13 (Fig. 5a). By contrast, D%w f  5–6% for runs in which
The eight evolutions display variable final fractions of convergence W -convergence does not develop well (i.e., with low eW n values;
‡
across W , in the range ∼14–33% (%w f , Table 3). #11, 19; Table 3) or only transiently (#12). The greatest differ-
Convergence across W and S are clearly anticorrelated for all ences are only observed for models with rather weak types of
models but #19, for which a sharp increase in internal deformation W (W 1 , W 2: 13, 18, 18b), yet the difference for #17 (W 1 type)
(%i) is observed. Increases in forearc deformation (smallest arrows is less than for #15 (W 3 type). Fig. 5b also emphasizes the rate
in Fig. 4) are observed during the ‘acceleration’ of the piston (all at which W -convergence may or not develop (see supplementary
but #12), the subduction of the ridge (#11; but not #12), the inter- material for a similar plot as a function of normalized shortening).
action with the 660 discontinuity (#15, 18b) or may result from a No systematic relationship is observed between this rate and the
slightly later adjustment (e.g., #18b). Increases in %w (Fig. 4) coin-
‘acceleration’ rate (×1.75 or 2.5; Table 3).
cide with ‘acceleration’ (models #13, 17, 18), and to a minor extent
Models #18 and #18b (Fig. 5a) illustrate some of the lateral,
(#15, 18b), ridge subduction (#11, 12) or the 660 km discontinuity
along strike changes observed in cases. Initially, the contribution
(#15, 18b).
of W -convergence is slightly larger for #18 than for #18b and
The influence of ‘acceleration’ on all models but #12 and #19
is confirmed by the evolution of the respective contribution of develops more rapidly after ‘acceleration’, probably because of a
W -convergence through time (D%w; Table 2) as a function of the somewhat stronger coupling across the S subduction. Subduction
timing of ‘acceleration’ (i.e., t ∗ ; Fig. 5a; see also inset Fig. 3 for then gets choked laterally in #18b, as revealed by the sharp in-
a comparison of slab shapes at that time). The value of D%w f is crease in %c and coeval decrease in SU after 15–16 min (Fig. 4e).
on the order of 10–15% for all runs where W -convergence is at The increase in W -convergence is sharper from then onwards and
least partially triggered by ‘acceleration’ and significantly higher final %w f values are almost similar for #18b and #18.
140 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145

Fig. 5. (a) Net difference of the fraction of convergence taken up by W at t ∗ (time referenced with respect to ‘acceleration’; Table 2) with respect to time that at t ◦ (once
model evolution is stabilized). (b) Evolution of the amount of convergence across S (Ds) and W (D w) as a function of along-strike deformation of their respective hanging
wall (lu, lo; Fig. 1c). These evolutions are used to define the resistance across S (R S) and across W (W ). See small cartoons above figure and text for details. (c) Comparison
of the overall amount of obduction (%O ) with the respective contributions from S, W , internal and lateral deformation. (d) Relationship between efficiency (F ; Table 2) and
the estimated resistance across W (R W ; Fig. 5b).

The response of S and W to overall shortening can be mon- and #19 (Fig. 5b; and only moderate for #17). Lesser changes are
itored (at least in first approximation) through the evolution of observed in the evolution of Dw (except for #15).
along-strike footwall deformation at S and W (Ds and D w;
Fig. 5b) versus across-strike hanging-wall deformation (lu and lo, 5. Discussion
which are very precisely measured and very sensitive to defor-
mation increments; e.g., Fig. 2d). Roughly linear correlations are 5.1. Obduction: how and how much
evidenced for both. Ds shows consistently greater values than D w,
indicating that the subduction process across S remains effective Obduction ensuing from subduction initiation is successfully re-
throughout the whole experiment. The evolution of Ds/lu is char- produced in a number of models (15 out of 26) over timescales of
acterized by a marked slope decrease (i.e., an increase in fore-arc a few Myrs, as continental lithosphere gets underthrust beneath
deformation with respect to subduction) for all models but #11 oceanic lithosphere (Fig. 3). In order to better understand what
 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145 141

controls subduction initiation and, in turn, obduction, the amount 5.2. Deformation partitioning and force balance between subduction
of obduction (%O = %w f . F ) is compared, for all runs, to the net and obduction
contributions from W - and S-convergence and from internal and
lateral deformation (Fig. 5c; Table 3). Fig. 5c shows that lateral Fig. 6a shows how the ratio of deformation rates (W over S;
deformation is negligible in first approximation (%lf < 10% for all w ∗ /s∗ ) evolves as a function of the fraction of shortening taken up
models except #15 and #11), suggesting that along-strike defor- by fore-arc deformation (c; Table 2). Deformation rates across W
mation dominates and that plate width was scaled correctly with reach 1.5 to 2 times that of S for models #12, 15, 17, 18b and are
respect to the piston push. Internal deformation (%i f ) is also signif- roughly an order of magnitude larger for #13 and #18. By contrast
icantly smaller than deformation across W and S, except for #11. w ∗ /s∗ ratios remain low for #11 and #19. Obduction starts devel-
This predominance of S and W explains why their respective con- oping for values of c < 5%. Once deformation starts localizing at W
tributions are broadly anticorrelated (Fig. 5c; except for #17). there is very little increase in the amount of fore-arc deformation
Differences in the time evolution of the various models are in- for models #12, #13 and #18b.
terpreted as follows (Figs. 3, 4, 5 and Table 3): These results suggest that once the resistance across S reaches
(i) Models #13 and #18 are the most efficient ones: both were a certain threshold (i.e., when the subduction cannot fully adjust
designed with a fairly low initial W resistance (W 2 and W 1 , to the change in boundary conditions), deformation starts being
respectively) and concentrate much deformation on W (net short- transferred to W and subduction/obduction initiate after deforma-
ening: eW n ∼ 70–80%; Table 3) while showing the smallest short- tion localized there. Convergence then partitions between the two
ening across S of all models (e Sn ∼ 57%). plate boundaries, thereby controlling the amount of obduction.
(ii) Despite a significantly lower %w f value, model #17 is nearly Estimating the new resistance across S (R S) after the pertur-
as efficient as #13 and #18, probably due to its weaker initial bation is not straightforward since the resistance across S and W
resistance (W 1). Time fluctuations in the obduction process are cannot be easily evaluated independently after loading. A reason-
nevertheless noticeable (Fig. 4d). Similar obduction amounts are able proxy for the resistance across S, however, is the dimension-
obtained for #12, for which time fluctuations are strongly corre- less ratio R S = lu / Ds (Fig. 5b), that is the ratio of upper plate
lated with ridge subduction. deformation over effective amount of subduction. As for R W , the
(iii) Obduction amounts are significantly lower for #15 and #19 greater the value of R S the more deformation is transferred to
the adjacent overthrusting plate. Fig. 6b shows that models can
(%O ∼ 16–17%), yet similar despite different weakness types (W 3
be classified satisfactorily as a function of the resistance between
and W 1 , respectively). This low amount is due, for #15, to the re-
S and W . Shortening across W (eW n) increases relatively regu-
sistance across W and the importance of lateral deformation. For
larly with decreasing values of R W and increasing values of R S, as
#19, this due to a particularly efficient subduction (e Sn ∼ 78%) and
expected intuitively. Obduction amounts (%O ) anticorrelate mainly
to the existence of a lighter, continental-type COT (see supplemen-
with R W .
tary material), which partly increased the resistance of the W 1
The force balance at S, prior to subduction initiation at W ,
weakness with respect to other similar runs. As a result model #19
is controlled by (Conrad and Hager, 1999; Funiciello et al., 2003;
preferentially accommodates deformation through the buckling of
Gurnis et al., 2004):
the oceanic lithosphere rather than through W -convergence.
– driving forces: (1) the force applied by the piston (i.e., ridge
(iv) Finally, no subduction initiation (nor obduction) takes place
push) and (2) slab pull forces.
in #11: internal deformation is larger than for other models
– resisting forces: (1) the resistance to the bending of the plate
(Fig. 5c), shortening across W is the smallest of all while short-
(≈ v .h3 .μ/r 3 , where v, h, μ and r respectively denotate veloc-
ening across S is large and similar to #19 (eW n = 33% and e Sn =
ity, plate thickness, viscosity and radius of curvature; Conrad and
76%; Table 3). When subduction gets jammed (as shown by the
Hager, 1999) and (2) viscous resistance forces, which comprise vis-
increase in %c; Fig. 4a), internal deformation takes most of the
cous coupling along the subduction plane down to lithospheric
deformation and W -convergence does not take place through un- depths, and slab repulsion and slab anchor forces beyond (e.g.,
derthrusting but through folding and horizontal shortening. Scholz and Campos, 1995; Heuret and Lallemand, 2005). Note that
These results suggest that obduction efficiency closely reflects viscous mechanical coupling across S is minimum during the early
the initial weakness type, though only qualitatively calibrated stages of model evolution since subduction initially proceeds well.
(Fig. 1c): the resistance of W is consistently greater for #15, which Any increase in driving forces will be balanced by an increase
had a fairly stiff initial interface (W 3 ), and even greater for the in resisting forces, yet with a different partitioning between bend-
stiffest of all models (#11; W 5). This is readily apparent when ing and viscous resistance in each model. Increasing bending forces
using the initial lo/ D w ratio (i.e., before the onset of obduction; will induce a decrease in the radius of slab curvature (i.e., a smaller
Fig. 5b) as an independent, quantitative estimate of the resistance value of r), slab steepening, advancing trench behavior (at least
across W (R W ). The larger this ratio, the more deformation is in- during the first stages), hence more frontal convergence and cou-
deed expected be transferred to the adjacent overthrusting plate. pling at S.
Fig. 5d shows that the efficiency and R W (and initial weakness Comparison of slab shapes at the time of ‘acceleration’ (inset,
type) are indeed largely anticorrelated. Those for which obduction Fig. 3) with the rate of obduction development (Fig. 5a) shows that
effectively develops (all but #11) plot in the area where R W is the runs most sensitive to ‘acceleration’ (#13 and #17) correspond
small, with a threshold around 0.6 (for #15 and all other runs with to those with the shallowest inflexion points (i.e., largest r; inset,
W 3 ) allowing to map out models which failed to reproduce sub- Fig. 3). This is consistent with the fact that, despite the increase
duction initiation and, in turn, obduction. in bending, their dip is still less than in other models: as a conse-
These experiments thus qualitatively but consistently explore a quence, the lithospheric horizontal shear component of the push
variety of situations and of mechanical coupling across W . Their is maximum and most efficiently transferred to the upper plate,
reliability is ensured by the similar behaviors observed for all inducing more mechanical coupling and less efficient subduction
of them: negligible rotation, across-strike deformation or model at S.
thickening, comparable amounts of deformation and shortening Later model evolution depends on the resistance of W (as em-
across W and S, and sensitivity to weakness type (see below). phasized by Fig. 6b): if W is weak enough then a new subduction
Their relevance (in particular as to the set-up, choice of materi- initiates and a significant part of the piston push is absorbed there.
als, . . . ) is addressed in a later Section 5.3. One can anticipate that, as slab pull forces and slab dip increase,
142 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145

Fig. 6. (a) Evolution of the ratio of deformation velocities (W over S) as a function of fore-arc deformation. (b) Values of the amount of shortening and obduction for all
models (Fig. 3; Table 3) as a function of the resistance across W (R W ) and (R S), as estimated in Fig. 5b. (c) Sketch summarizing the feasibility of subduction initiation and
obduction with respect to the resistance at S and W . (d) Idealized 3D configuration illustrating the balance of forces and deformation partitioning between S and W . See
text (5.2 through 5.4) for details.

net viscous coupling will decrease and S will recover partly. Ob- and ridge push forces (i.e., 1013 N m−2 ; Bott, 1993). Deformation
duction will in any case depend on the partitioning of forces at rates are 10−11 s−1 since time scales of subduction initiation
both W and S and may be quite complex (and will also depend are ∼105 years (in any case <1 My, as shown by metamorphic
on toroidal and poloidal flows; e.g., Piromallo et al., 2006). sole clusters). Natural, integrated viscosities at W (which can be
viewed as a proxy to the resistance of W ) are therefore ∼1023 Pa s,
5.3. Scaling and respective influence of key parameters triggering which is the order of magnitude of dry to only slightly hydrated
obduction mantle viscosities and the magnitude chosen for our experimental
lithosphere (Table 1). Although weaknesses in these analog mod-
Analog models demonstrate that obduction can be triggered by els cannot be assessed more precisely than in nature, they are
a modification of the force balance at a preexisting subduction probably reasonable since subduction initiate for some but not all
zone, as captured in Fig. 6c. Prior to extrapolating these results configurations.
to nature, the following key questions need to be addressed: (1) Is Despite inherent limitations, our models provide a means to
the weakness at W realistic? (2) Can we infer which of the three compare the respective influence of ‘acceleration’, ridge subduc-
external perturbations ( A, R, D) has the greatest influence? (3) Is tion and encounter of the 660 km discontinuity on the triggering
the coupling threshold at S applicable to nature? of obduction. Obduction is either triggered by ‘acceleration’ (#13,
Deformation in nature must in any case start localizing at some 17, 18; see also #4, 8, 21), by ‘acceleration’ with a later contribu-
discontinuity such as W in order to nucleate (or initiate) sub- tion from the 660 km discontinuity (#15), by the 660 km rather
duction. The type of this weakness in nature (e.g., existing fault than ‘acceleration’ (#18b) or by ridge subduction (#12). Given the
weaknesses integrated over the COT, presence of a sedimentary similar trends observed for w ∗ /s∗ (ridge only: #12; ‘acceleration’
basin; Shemenda, 1992) and its strength are completely unknown. only: #13; Fig. 6a), ridge subduction and ‘acceleration’ are the most
Qualitative brackets can be put forward: it should not be too weak influential triggers and probably comparable. The impact of the
since S continues and not too strong since subduction initiation 660 km discontinuity is minor (Fig. 6a; though essential in #15
(and obduction) happens repeatedly through time (see also Hall and 18b), particularly when considering that the lower boundary
et al., 2003; Gurnis et al., 2004). One can thus confidently postu- of the model is a (exaggeratedly) stiff barrier compared to the nat-
late that they must be of the same order of magnitude as slab pull ural prototype.
 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145 143

More models are required to better assess the respective im- Warren et al., 2008; Bialas et al., 2011; Duretz and Gerya, 2013).
pact of A–R–D in nature, in particular to overcome difficulties in Several authors partly circumvented the former problem by fo-
scaling the physical properties of the ridge and choosing adequate cusing on the ability of their models to return some continental
rheologies. Although probably exaggeratedly weak and buoyant in material, by introducing a sharp decollement layer between the
models #1–11, ridge properties are also difficult to assess in na- crust and mantle (Chemenda et al., 1995, 1996; Bialas et al., 2011;
ture, due to the influence of ridge push, accretion type and thermal which is impossible with silicone only) or extensional boundary
structure on both parameters. Convergence velocities are slightly forces (Duretz and Gerya, 2013).
exaggerated with respect to our natural reference case study, but
still of the same magnitude (i.e., 5–10 cm/yr versus ∼2–3 cm/yr 6. Implications and conclusions
for the Neotethys). On the other hand, this sharp increase in con-
vergence velocities (‘acceleration’) would be more influential with Despite inherent rheological simplifications, our analog models
rheologies promoting localization (i.e., unlike the quasi-Newtonian provide useful insights on obduction processes and regional-scale
silicone used here). The influence of these perturbations also de- geodynamics:
pends on the extent of slab pull at the time: for comparable slab (1) Obduction is herein reproduced for the first time at conver-
shapes, slab pull is stronger for #18b than #13 at the time of gence rates (3–13 cm/yr), ‘acceleration’ rates (×2.5) and timescales
‘acceleration’ (which is consistently less influential for the former of ∼10 Myr, which are all comparable to nature.
model). Note that slab pull forces at S are both underestimated, (2) Models show that obduction does not require extraordinary
considering the lack of phase transformations and densification in boundary conditions and provide clues to the following questions:
the model, and overestimated due to the lack of thermal gradient – Why and when? Obduction is a consequence of modifying the
(promoting slab weakening at depth; e.g., Houseman and Gubbins, force balance at preexisting (adjacent) subduction zones. Obduc-
1997). tion proceeds when adjacent subduction zones cannot fully adjust
to changing boundary conditions and get mechanically locked. This
5.4. Implications for regional-scale geodynamics coupling threshold is reached in models when forearc deformation
exceeds ∼5% of the subduction rate.
Keeping in mind the above limitations and necessary approx- – How and where? Obduction may simply proceed as a result
imations, our models demonstrate that subduction may initiate of subduction initiation. The extent of obduction will depend on
without pre-defining a subduction plane (by contrast to Chemenda (1) the subduction force balance between W and S and (2) the
et al., 1996) and may be a viable mechanism for obduction devel- resistance in the weak zones (i.e., obduction may develop if W
opment at realistic timescales (i.e., ∼10 My). We also note that no is sufficiently weak). Further scaling is critical to evaluate natural
slab pull is needed to initiate the obduction-driven subduction of coupling thresholds across S and W .
the light continental lithosphere at W , although negative buoyancy (3) These models support the feasibility of the scenario envi-
(which is lacking in #19) can nevertheless be helpful. sioned by Agard et al. (2007) for the Neotethys, in which regional-
Our models support the scenario envisioned by Agard et al. scale geodynamic reorganization in response to plate ‘accelera-
(2007) for the Neotethys, in which the blocking of the subduc- tion’ increased mechanical coupling in the subduction zone to the
tion zone beneath Eurasia, attested by widespread blueschist ex- north of the Neotethys and triggered the onset of obduction in the
humation coeval with plate ‘acceleration’ and regional-scale geo- south. Although further analog and numerical modeling models
dynamic reorganization (Monié and Agard, 2009), led to defor- are needed, including detailed 3D analysis with tracking of surface
mation partitioning across the Neotethys and to the triggering of evolution, the obduction models presented here represent a useful
obduction. Such long-term changes of mechanical coupling may be scenario for further field studies in well-constrained geodynamic
compared to the short-term changes in coupling observed during settings (i.e., large-scale obducted ophiolites from Newfoundland,
megaearthquakes in present-day subduction zones (e.g., Moreno et Timor, New Caledonia, Balkans, western Turkey).
al., 2010). The models may also apply to natural examples of the (4) We finally underline that the present models represent the
Zenisu ridge (Mazzotti et al., 2002) and to the intraplate deforma- first attempt to model the dynamic interaction between two sub-
tion evidenced by the large-scale folds in the Indian Ocean (Bull duction zones.
and Scrutton, 1990; Chamot-Rooke et al., 1993).
Fig. 6d provides a 3D sketch emphasizing how subduction ini- Acknowledgements
tiation (then obduction) may develop once the S subduction gets
jammed after an increase in mechanical coupling of a few per- This work was essentially funded through the ONLAP project
cent. The effective development of obduction is probably complex (ANR blanche, SIMI6; 2010 BLAN 615 01). Additional support was
in 3D. Since subduction initiation and obduction can develop over provided by IUF, INSU and the ISTeP laboratory.
thousands of km along strike (as shown by the examples of the
Neotethys, during both the Late Jurassic and Late Cretaceous, or
Appendix A. Supplementary material
Papua), one may wonder if the blocking threshold at S needs to
be integrated over the whole length of the incipient subduction
Supplementary material related to this article can be found on-
zone, or if a local instability may be sufficient for the lateral prop-
line at http://dx.doi.org/10.1016/j.epsl.2014.02.021.
agation of subduction initiation.
The applicability of our models is restricted to the onset and
References
early evolution of obduction since the continent, forced by the
piston push, goes in without ultimately blocking the obduction
Abbatte, E., Bortolotti, V., Passerini, P., Principi, G., 1985. The rhythm of Phanerozoic
process, contrary to what happens in nature (e.g., Oman; Searle ophiolites. Ofioliti 10, 109–138.
et al., 2004; Agard et al., 2010). Plate tectonic forces, in addi- Agard, P., Monie, P., Gerber, W., Omrani, J., Molinaro, M., Meyer, B., Labrousse,
tion, are balanced in nature on the whole system of plate bound- L., Vrielynck, B., Jolivet, L., Yamato, P., 2006. Transient, synobduction exhuma-
aries and not solely on two boundaries (S and W ) as modeled tion of Zagros blueschists inferred from P–T, deformation, time and kinematic
constraints: Implications for Neotethyan wedge dynamics. J. Geophys. Res. 11,
here. These shortcomings are common, however, to analog or nu- B11401. http://dx.doi.org/10.1029/2005JB004103.
merical models developed so far for continental subduction (e.g., Agard, P., Jolivet, L., Vrielynck, B., Burov, E., Monie, P., 2007. Plate acceleration: The
Chemenda et al., 1995; Boutelier et al., 2004; Yamato et al., 2008; obduction trigger?. Earth Planet. Sci. Lett. 258, 428–441.
144 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145

Agard, P., Searle, M.P., Alsop, G.I., Dubacq, B., 2010. Crustal stacking and expul- Glennie, K.W., Boeuf, M.G., Hughes-Clarke, M.H.W., Moody-Stuart, M., Pilaar, W.F.,
sion tectonics during continental subduction: P–T deformation constraints from Reinhardt, B.M., 1973. Late Cretaceous nappes in the Oman mountains and their
Oman. Tectonics 29. http://dx.doi.org/10.1029/2010TC002669. geologic evolution. Am. Assoc. Pet. Geol. Bull. 57, 5–27.
Béchennec, F., 1987. Géologie des nappes Hawasina dans la partie orientale et cen- Glennie, K.W., Boeuf, M.G., Hughes-Clarke, M.H.W., Moody-Stuart, M., Pilaar, W.F.,
trale des montagnes d’Oman. Thèse. d’Etat Univ. Paris 6, Paris. 412 pp. Reinhardt, B.M., 1974. Geology of the Oman Mountains. Verh. K. Ned. Geol. Mi-
Becker, T.W., Faccenna, C., Connell, R.J.O., 1999. The development of slabs in the jnbouwkd. Genoot. 31, 1–423.
upper mantle: Insights from numerical and laboratory experiments. J. Geophys. Godard, M., Jousselin, D., Bodinier, J.-L., 2000. Relationships between geochemistry
Res. 104, 207–226. and structure beneath a palaeo-spreading centre: a study of the mantle section
Bellahsen, N., Faccenna, C., Funiciello, F., Daniel, J.M., Jolivet, L., 2003. Why did Arabia in the Oman ophiolite. Earth Planet. Sci. Lett. 180, 133–148.
separate from Africa? Insights from 3-D laboratory experiments. Earth Planet. Godard, M., Dautria, J.-M., Perrin, M., 2003. Geochemical variability of the Oman
Sci. Lett. 216, 365–381. ophiolite lavas: Relationship with spatial distribution and paleomagnetic direc-
Bialas, R.W., Funiciello, F., Faccenna, C., 2011. Subduction and exhumation of conti- tions. Geochem. Geophys. Geosyst. 4.
nental crust: insights from laboratory models. Geophys. J. Int. 184, 43–64. Guillou-Frottier, L., Buttles, J., Olson, P., 1995. Laboratory experiments on structure
Bott, M.H.P., 1993. Modelling the plate-driving mechanism. J. Geol. Soc. (Lond.) 150, of subducted lithosphere. Earth Planet. Sci. Lett. 133, 19–34.
941–951. Gurnis, M., Hall, C., Lavier, L., 2004. Evolving force balance during incipient subduc-
tion. Geochem. Geophys. Geosyst. 5.
Boudier, F., Ceuleneer, G., Nicolas, A., 1988. Shear zones, thrusts and related magma-
Hacker, B.R., 1994. Rapid emplacement of young oceanic lithosphere: argon
tism in the Oman ophiolite: initiation of thrusting on an oceanic ridge. Tectono-
geochronology of the Oman ophiolite. Science 80.
physics 151, 275–296.
Hacker, B.R., Mosenfelder, J.L., Gnos, E., 1996. Rapid emplacement of the Oman ophi-
Boutelier, D., Chemenda, A., Burg, J.P., 2003. Subduction versus accretion of intra-
olite: Thermal and geochronologic constraints. Tectonics 15, 1230–1247.
oceanic volcanic arcs: insight from thermo-mechanical analogue experiments.
Hacker, B., Gnos, E., 1997. The conundrum of Samail: explaining the metamorphic
Earth Planet. Sci. Lett. 212, 31–45.
history. Tectonophysics 279, 215–226.
Boutelier, D., Chemenda, A., Jorand, C., 2004. Continental subduction and exhuma-
Hall, C., Gurnis, M., Sdrolias, M., Lavier, L., Müller, R.D., 2003. Catastrophic initiation
tion of high-pressure rocks: insights from thermo-mechanical laboratory mod-
of subduction following forced convergence across fracture zones. Earth Planet.
elling. Earth Planet. Sci. Lett. 222, 209–216.
Sci. Lett. 212, 15–30.
Bull, J.M., Scrutton, R.A., 1990. Fault reactivation in the Central Indian Ocean and the Heuret, A., Lallemand, S., 2005. Plate motions, slab dynamics and back-arc deforma-
rheology of oceanic lithosphere. Nature 344, 855–858. tion. Phys. Earth Planet. Inter. 149, 31–51.
Çelik, Ö.F., Marzoli, A., Marschik, R., Chiaradia, M., Neubauer, F., Öz, I., 2011. Early– Houseman, G.A., Gubbins, D., 1997. Deformation of subducted oceanic lithosphere.
Middle Jurassic intra-oceanic subduction in the İzmir–Ankara–Erzincan Ocean, Geophys. J. Int. 131 (3), 535–551.
Northern Turkey. Tectonophysics 509, 120–134. Ishikawa, T., Fujisawa, S., Nagaishi, K., Masuda, T., 2005. Trace element characteristics
Ceuleneer, G., 1986. Structure des ophiolits d’Oman: flux mantellaire sous un cen- of the fluid liberated from amphibolite-facies slab: Inference from the meta-
tre d’expansion océanique et charriage à la dorsale. PhD Thesis. Nantes Univ. morphic sole beneath the Oman ophiolite and implication for boninite genesis.
349 pp. Earth Planet. Sci. Lett. 240, 355–377.
Ceuleneer, G., Nicolas, A., Boudier, F., 1988. Mantle flow patterns at an oceanic Lagabrielle, Y., Guivel, C., Maury, R.C., Bourgois, J., Fourcade, S., Martin, H., 2000.
spreading centre: the Oman peridotites record. Tectonophysics 151, 1–26. Magmatic–tectonic effects of high thermal regime at the site of active ridge
Chamot-Rooke, N., Jestin, F., de Voogd, B., P.w. group, 1993. Intraplate shortening in subduction: the Chile Triple Junction model. Tectonophysics 326, 255–268.
the Central Indian Ocean determined from a 2100-km-long north-south deep Lagabrielle, Y., Chauvet, A., Ulrich, M., Guillot, S., 2013. Passive obduction and
seismic reflection profile. Geology 21, 1043–1046. gravity-driven emplacement of large ophiolitic sheets: The New Caledonia
Chauvet, F., Dumont, T., Basile, C., 2009. Structures and timing of Permian rifting in ophiolite (SW Pacific) as a case study?. Bull. Soc. Géol. Fr. 184, 545–556.
the central Oman Mountains (Saih Hatat). Tectonophysics 475, 563–574. http://dx.doi.org/10.2113/gssgfbull.184.6.545.
Chemenda, A.I., Mattauer, M., Malavieille, J., Bokun, A.N., 1995. A mechanism for Le Mée, L., Girardeau, J., Monnier, J., 2004. Mantle segmentation along the Oman
syn-collision rock exhumation and associated normal faulting: results from ophiolite fossil mid-ocean ridge. Nature 432, 167–172.
physical modelling. Earth Planet. Sci. Lett. 132, 225–232. Lévy, F., Jaupart, C., 2012. The initiation of subduction by crustal extension at a
Chemenda, A.I., Mattauer, M., Bokun, A.N., 1996. Continental subduction and a continental margin. Geophys. J. Int. 188, 779–797.
mechanism for exhumation of high-pressure metamorphic rocks: new mod- Linthout, K., Helmers, H., Sopaheluwakan, J., 1997. Late Miocene obduction and
elling, field data from Oman. Earth Planet. Sci. Lett. 143, 173–182. microplate migration around the southern Banda Sea and the closure of the
Christensen, U.R., 1996. The influence of trench migration on slab penetration into Indonesian Seaway. Tectonophysics 281, 17–30.
the lower mantle. Earth Planet. Sci. Lett. 140, 27–39. Lippard, S.J., Shelton, A.W., Gass, I.G., 1986. The Ophiolite of Northern Oman. Geol.
Cluzel, D., Aitchison, J.C., Picard, C., 2001. Tectonic accretion and underplating Soc. Lond. Mem., vol. 11. 178 pp. http://dx.doi.org/10.1144/GSL.MEM.1986.011.
of mafic terranes in the Late Eocene intraoceanic fore-arc of New Caledonia Lus, W.Y., McDougall, I., Davies, H.L., 2004. Age of the metamorphic sole of
(Southwest Pacific): geodynamic implications. Tectonophysics 340, 23–59. the Papuan Ultramafic Belt ophiolite, Papua New Guinea. Tectonophysics 392,
Coleman, R.G., 1971. Plate tectonic emplacement of upper mantle peridotites along 85–101.
continental edges. J. Geophys. Res. 76, 1212–1222. Mazzotti, S., Lallemant, S.J., Henry, P., Le Pichon, X., Tokuyama, H., Takahashi, N.,
Coleman, R.G., 1981. Tectonic setting for ophiolite obduction in Oman. J. Geophys. 2002. Intraplate shortening and underthrusting of a large basement ridge in the
Res. 86, 2497–2508. eastern Nankai subduction zone. Mar. Geol. 187, 63–88.
McLeod, C., Lissenberg, C.J., Bibby, L.E., 2013. ‘Moist MORB’ axial magmatism in
Conrad, C.P., Hager, B.H., 1999. Effects of plate bending and fault strength at sub-
the Oman ophiolite: evidence against a mid-ocean ridge origin. Geology 41,
duction zones on plate dynamics. J. Geophys. Res. 104, 17551–17571.
459–462. http://dx.doi.org/10.1130/G33904.1.
Davies, G.F., 1995. Penetration of plates and plumes through the mantle transition
McKenzie, D.P., 1977. The initiation of trenches: A finite amplitude instability. In:
zone. Earth Planet. Sci. Lett. 133, 507–516.
Talwani, M., Pitman, W.C. (Eds.), Island Arcs, Deep Sea Trenches and Back-Arc
Davy, P., Cobbold, P.R., 1991. Experiments on shortening of a 4-layer model of
Basins. In: Maurice Ewing Ser., vol. 1. AGU, Washington D.C., pp. 57–61.
the continental lithosphere. Tectonophysics 188, 1–25. http://dx.doi.org/10.1016/
Mattauer, M., Faure, M., Malavielle, J., 1981. Transverse lineations and large-scale
0040.
structure related to Alpine obduction in Corsica. J. Struct. Geol. 4, 401–409.
Dercourt, J., Gaetani, M., Vrielynck, B., Barrier, E., Biju-Duval, B., Brunet, M.F., Cadet,
Michard, A., Juteau, T., Whitechurch, H., 1985. L’obduction: revue des modèles et
J.P., Crasquin, S., Sandulescu, M., 2000. Atlas Peri-Tethys, Palaeogeographical
confrontation au cas de l’Oman. Bull. Soc. Géol. Fr. 2, 189–198.
Maps: 24 Maps and Explanatory Notes. CCGM/CGMW, Paris. Mitrovica, J.X., Forte, A.M., 1997. Radial profile of mantle viscosity: results from the
Dewey, J.F., Bird, J.M., 1971. Origin and emplacement of the ophiolite suite: Ap- joint inversion of convection and postglacial rebound observables. J. Geophys.
palachian ophiolites in Newfoundland. J. Geophys. Res. 76, 3179–3206. Res. 102 (B2), 2751–2769.
Dewey, J., 1976. Ophiolite obduction. Tectonophysics 31, 93–120. Monié, P., Agard, P., 2009. Coeval blueschist exhumation along thousands of km:
Duretz, T., Gerya, T.V., 2013. Slab detachment during continental collision: Influence implications for subduction channel processes. Geochem. Geophys. Geosyst. 10,
of crustal rheology and interaction with lithospheric delamination. Tectono- Q07002. http://dx.doi.org/10.1029/2009GC002428.
physics 602, 124–140. Moores, E.M., Kellogg, L.H., Dilek, Y., 2000. Tethyan ophiolites, mantle convection,
El-Shazly, A.K., Bröcker, M., Hacker, B., Calvert, A., 2001. Formation and exhumation and tectonic “historical contingency”: a resolution of the “ophiolite conundrum”.
of blueschists and eclogites from NE Oman: new perspectives from Rb–Sr and In: Dilek, Y., Moores, E.M., Elthon, D., Nicolas, A. (Eds.), Ophiolites and Oceanic
40
Ar/39 Ar dating. J. Metamorph. Geol. 19, 233–248. Crust: New Insights from Field Studies and the Ocean Drilling Program. In: Spec.
Faccenna, C., Giardini, D., Davy, P., Argentieri, A., 1999. Initiation of subduction at At- Pap., vol. 349. Geological Society of America, Boulder, Colorado.
lantic-type margins: insights from laboratory experiments. J. Geophys. Res. 104, Moreno, M., Rosenau, M., Oncken, O., 2010. Maule earthquake slip correlates with
2749–2766. pre-seismic locking of Andean subduction zone. Nature 467, 198–202.
Funiciello, F., Faccenna, C., Giardini, D., Regenauer-Lieb, L., 2003. Dynamics of Nicolas, A., Le Pichon, X., 1980. Thrusting of young lithosphere in subduction zones
retreating slabs: 2. Insights from three-dimensional laboratory experiments. with special reference to structures in ophiolitic peridotites. Earth Planet. Sci.
J. Geophys. Res. 108, 2207. Lett. 46, 397–406.
 P. Agard et al. / Earth and Planetary Science Letters 393 (2014) 132–145 145

Nicolas, A., 1989. Structures in Ophiolites and Dynamics of Oceanic Lithosphere Shervais, J.W., 2001. Birth, death, and resurrection: The life cycle of supra-
(Petrology and Structural Geology). Kluwer, Dordrecht. subduction zone ophiolites. Geochem. Geophys. Geosyst. 2. http://dx.doi.org/
Nicolas, A., Boudier, F., Ildefonse, B., Ball, E., 2000. Accretion of Oman and United 10.1029/12000GC000080.
Arab Emirates ophiolite – Discussion of a new structural map. Mar. Geophys. Stern, R.J., 2004. Subduction initiation: spontaneous and induced. Earth Planet. Sci.
Res. 21, 147–179. Lett. 226 (3–4), 275–292.
Oxburgh, E.R., 1972. Flake tectonics and continental collision. Nature 239, 202–204. Suhr, G., Cawood, P.A., 1993. Structural history of ophiolite obduction, Bay of Islands,
Page, B.M., Engebretson, D.C., 1984. Correlation between the geologic record and Newfoundland. Geol. Soc. Am. Bull. 105, 399–410.
computed plate motions for central California. Tectonics 3, 133–155. Thuizat, R., Whitechurch, H., Montigny, R., Juteau, T., 1981. K–Ar dating of some
infra-ophiolitic metamorphic soles from the Eastern Mediterranean: new ev-
Pamić, J., Tomljenović, B., Balen, D., 2002. Geodynamic and petrogenetic evolution
idence for oceanic thrustings before obduction. Earth Planet. Sci. Lett. 52,
of Alpine ophiolites from the central and NW Dinarides: an overview. Lithos 65,
302–310.
113–142.
Tilton, G.R., Hopson, C.A., Wright, J.E., 1981. Uranium-lead isotopic ages of the Se-
Piromallo, C., Becker, T.W., Funiciello, F., Faccenna, C., 2006. Three-dimensional in-
mail ophiolite, Oman, with applications to Tethyan Ridge Tectonics. J. Geophys.
stantaneous mantle flow induced by subduction. Geophys. Res. Lett. 33, L08304.
Res. 86, 2763–2775.
http://dx.doi.org/10.1029/2005GL025390.
Toth, J., Gurnis, M., 1998. Dynamics of subduction initiation at preexisting fault
Pubellier, M., Monnier, C., Maury, R., Tamayo, R., 2004. Plate kinematics, origin
zones. J. Geophys. Res. 103, 18,053–18,067.
and tectonic emplacement of supra-subduction ophiolites in SE Asia. Tectono-
Vaughan, A.P.M., Scarrow, J.H., 2003. Ophiolite obduction pulses as a proxy indicator
physics 392, 9–36.
of superplume events?. Earth Planet. Sci. Lett. 213 (3–4), 407–416.
Python, M., Ceuleneer, G., Arai, S., 2008. Chromian spinels in mafic–ultramafic man- Wakabayashi, J., Dilek, Y., 2003. What constitutes “emplacement” of an ophiolite?:
tle dykes: Evidence for a two-stage melt production during the evolution of the mechanisms and relationship to subduction initiation and formation of meta-
Oman ophiolite. Lithos 106, 137–154. morphic soles. In: Ophiolites in Earth History. In: Geol. Soc. (Lond.) Spec. Publ.,
Ricou, L.E., 1971. Le croissant ophiolitique péri-Arabe, une ceinture de nappes mise vol. 218. Geological Society of London, London, pp. 427–447.
en place au crétacé supérieur. Rev. Géogr. Phys. Géol. Dyn. 13, 327–350. Warren, C.J., Parrish, R.R., Searle, M.P., Waters, D.J., 2003. Dating the subduction of
Ricou, L.E., 1994. Tethys reconstructed: plates, continental fragments and their the Arabian continental margin beneath the Semail ophiolite, Oman. Geology 31,
boundaries since 260 Ma from Central America to South-eastern Asia. Geodin. 889–892.
Acta 7, 169–218. Warren, C.J., Parrish, R.R., Waters, D.J., Searle, M.P., 2005. Dating the geologic history
Rioux, M., Bowring, S., Kelemen, P., Gordon, S., Dudás, F., Miller, R., 2012. Rapid of Oman’s Semail ophiolite: insights from U–Pb geochronology. Contrib. Mineral.
crustal accretion and magma assimilation in the Oman–U.A.E. ophiolite: High Petrol. 150, 403–422.
precision U–Pb zircon geochronology of the gabbroic crust. J. Geophys. Res. 117, Warren, C.J., Beaumont, C., Jamieson, R.A., 2008. Modelling tectonic styles
B07201. http://dx.doi.org/10.1029/2012JB009273. and ultra-high pressure (UHP) rock exhumation during the transition from
oceanic subduction to continental collision. Earth Planet. Sci. Lett. 267,
Rioux, M., Bowring, S., Kelemen, P., Gordon, S., Miller, R., Dudás, F., 2013. Tectonic
129–145.
development of the Samail ophiolite: High-precision U–Pb zircon geochronology
Weijermars, R., Schmeling, H., 1986. Scaling of Newtonian and non-Newtonian fluid
and Sm–Nd isotopic constraints on crustal growth and emplacement. J. Geophys.
dynamics without inertia for quantitative modeling of rock flow due to gravity
Res. 117, B07201. http://dx.doi.org/10.1029/2012JB009273.
(including the concept of rheological similarity). Phys. Earth Planet. Inter. 43,
Schmeling, H., Babeyko, A.Y., Enns, A., Faccenna, C., Funiciello, F., Gerya, T., Golabek,
316–330. http://dx.doi.org/10.1016/0031.
G.J., Grigull, S., Kaus, B.J.P., Morra, G., Schmalholz, S.M., van Hunen, J., 2008.
Yamato, P., Agard, P., Burov, E., Le Pourhiet, L., Jolivet, L., Tibéri, C., 2007a. Burial
A benchmark comparison of spontaneous subduction models—Towards a free
and exhumation in a subduction wedge: mutual constraints from thermo-
surface. Phys. Earth Planet. Inter. 171, 198–223.
mechanical modeling and natural P–T–t data (Sch. Lustrés, W. Alps). J. Geophys.
Scholz, C.H., Campos, J., 1995. On the mechanism of seismic decoupling and back
Res. 112, B07410. http://dx.doi.org/10.1029/2006JB004441.
arc spreading at subduction zones. J. Geophys. Res. 100, 22,103–22,115.
Yamato, P., Agard, P., Goffé, B., de Andrade, V., Vidal, O., Jolivet, L., 2007b.
Searle, M.P., Cox, J., 1999. Tectonic setting, origin and obduction of the Oman ophi- New, high-precision P–T estimates for Oman blueschists: implications for ob-
olite. Geol. Soc. Am. Bull. 111, 104–122. duction, nappe stacking and exhumation processes. J. Metamorph. Geol. 25,
Searle, M.P., Warren, C.J., Waters, D.J., Parrish, R.R., 2004. Structural evolution, meta- 657–682.
morphism and restoration of the Arabian continental margin, Saih Hatat region, Yamato, P., Burov, E., Agard, P., Le Pourhiet, L., Jolivet, L., 2008. HP–UHP exhuma-
Oman Mountains. J. Struct. Geol. 26, 451–473. tion during slow continental subduction: self-consistent thermodynamically
Shemenda, A.I., 1992. Horizontal lithosphere compression and subduction: con- and thermomechanically coupled model with application to the Western Alps.
straints provided by physical modeling. J. Geophys. Res. 97, 11,097–11,116. Earth Planet. Sci. Lett. 271, 63–74.