JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 116, B11102, doi:10.

1029/2011JB008429, 2011

The South Ecuador subduction channel: Evidence for a dynamic

mega‐shear zone from 2D fine‐scale seismic reflection imaging
and implications for material transfer
J.‐Y. Collot,1 A. Ribodetti,1 W. Agudelo,2 and F. Sage3
Received 8 April 2010; revised 9 August 2011; accepted 24 August 2011; published 10 November 2011.

[1] Tectonic processes that control the transition from poorly consolidated sediment
entering the subduction channel (SC) to the seismogenic zone are documented using
seismic imaging. We applied pre‐stack depth migration and a post‐processing sequence to
a seismic reflection line acquired across the Ecuador convergent margin to obtain a
2D‐quantitative image of the first ∼24 km of the SC. Structural interpretation shows that
the SC consists of a 630–1150‐m‐thick, low‐velocity, continuous sheet of sediment that
dips ∼6° landward and undergoes shear deformation. The long sheet is bounded at top and
bottom by décollement thrusts, and developed over time Riedel shears and basal thrust
faulting and folding downdip, pointing to a dynamic mega‐shear zone. Modeling the
strong uppermost and basal SC reflectors reveals that they are associated with 40–80‐m‐
thick, 50–350 m/s, low‐velocity perturbations layers inferred to be fluid rich and
mechanically weak. A fine‐scale velocity model shows two anomalously low‐Vp areas in
the long sheet, advocating patches of over‐pressured fluids. Evidence for Vp variations
along the upper‐plate foundation suggests either underplated bodies or a fluid‐damage
zone. A simple temporal reconstruction indicates that underthrusting the long sheet
initiated >450 kyr ago and interrupted ∼54 ± 13 kyr ago, when frontal accretion resumed.
During this transient evolution, the SC boundaries revealed highly unstable as most of the
SC was underplated while down going plate material may have been sheared off and
incorporated to the SC.
Citation: Collot, J.‐Y., A. Ribodetti, W. Agudelo, and F. Sage (2011), The South Ecuador subduction channel: Evidence for a
dynamic mega‐shear zone from 2D fine‐scale seismic reflection imaging and implications for material transfer, J. Geophys. Res.,
116, B11102, doi:10.1029/2011JB008429.

1. Introduction considered very weak and aseismic [Byrne et al., 1988], and
is further inferred to have a velocity‐strengthening behavior
[2] The nature of the subduction plate boundary zone has
during the co‐seismic slip [Scholz, 1998; Wang and Hu,
remained a puzzling matter that requires a better under-
2006].
standing of the parameters controlling the transition from [3] Multichannel seismic reflection (MCS) images indi-
soft sediment entering subduction to a strongly coupled fault
cate that the interplate sedimentary layer, frequently inferred
zone capable of generating the greatest earthquakes on earth
to be the SC, consists of a relatively low P wave velocity
[Hyndman et al., 1997; Kanamori, 1986; Pacheco et al., (Vp), up to ∼2 km‐thick, stratified layer bounded at top and
1993; Tichelaar and Ruff, 1993]. Poorly consolidated sedi-
bottom by strong reflectors. The upper reflector commonly
ment, rapidly buried between two strong plates, was pro-
shows a negative polarity, and is usually interpreted as the
posed to act as a viscous, lubricating layer called the
inter‐plate décollement thrust [Aoki et al., 1985; Bangs et al.,
subduction channel (SC), which transmits the velocity and
2004, 2009; Calahorrano et al., 2008; Park et al., 2002; Sage
stress fields between the two plates [England and Holland,
et al., 2006; Shipley et al., 1992; Tsuru et al., 2002; von
1979; Shreve and Cloos, 1986; Cloos and Shreve, 1988a,
Huene et al., 1994].
1988b; Mancktelow, 1995]. The SC updip segment was then
[4] Drilling data collected at the tip of the Nankai [Taira
1
et al., 1992], Barbados [Moore et al., 1998], and Costa Rica
Université de Nice Sophia‐Antipolis, Institut de Recherche pour le
Développement (UR082), Observatoire de la Côté d’Azur, Géoazur,
[Kimura et al., 1997] margins show that shear localizes
Villefranche‐sur‐Mer, France. along a very low strength décollement zone, and that sedi-
2
ECOPETROL, Instituto Colombiano del Petróleo, Piedecuesta, ment below the décollement remains undeformed suggest-
Colombia.
3
ing that the décollement is a sharp discontinuity in stress
Université de Nice Sophia‐Antipolis, Université Pierre et Marie Curie, transmission [Taira et al., 1992] [Ujiie et al., 2003;
Observatoire de la Côte d’Azur, Géoazur, Villefranche‐sur‐Mer, France.
Vannucchi and Tobin, 2000]. In the conventional view, the
Copyright 2011 by the American Geophysical Union. SC is mechanically decoupled from the overlying wedge so
0148‐0227/11/2011JB008429 that, near its inlet, underthrust sediments are subject to a

B11102 1 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

Figure 1. (a) Bathymetric map of the south Ecuador‐Nord Peru margin [after Michaud et al., 2006]
contoured at 100 m intervals with the location of Multichannel Seismic reflection line SIS‐72. Heavy red
line is reprocessed interval of Line SIS‐72 shown in Figure 2. White arrow is the predicted Nazca plate
convergence rate with respect to South America [Kendrick et al., 2003]. Barbed lines show the defor-
mation front and narrow accretionary wedge [Collot et al., 2002; Calahorrano, 2005]. (b) Location of
study area; DGM = Dolores Guayaquil Megashear [Campbell, 1974; Ego et al., 1996]; red stars are
earthquakes epicenters [Swenson and Beck, 1996] (see also SISRA catalog); white dash line is 1906
earthquake rupture zone [Kelleher, 1972].

vertical Maximum Principal Stress. This supposition is along the plate interface. A suite of processes including
supported by the progressive densification and thinning of lithification, diagenesis, metamorphism, fluid overpressure,
lithologic units between drill sites in the trench and those at pressure solution, frictional velocity dependence, and shear
the margin tip [Kimura et al., 1997; Saffer, 2003; Vannucchi localization is typically inferred to govern this transition
and Tobin, 2000]. [Hyndman and Wang, 1995; Moore and Saffer, 2001;
[5] Documenting the strain pattern, fluid circulation, as Saffer, 2003; Saffer and Marone, 2003; Screaton, 2006;
well as sediment transformation with continued under- Vrolijk, 1990].
thrusting toward the seismogenic zone is critical because [6] Direct observations of sparse exhumed SC or
they are thought to control the aseismic‐seismic transition equivalents [Kitamura et al., 2005] bring important observa-
and the décollement dynamics i.e., underplating and tectonic tions on the SC nature, deformation and fluid cycle.
erosion [Shreve and Cloos, 1986]. At modern subduction Vannucchi et al. [2008] showed that the entire material
zones, the detailed structures of the SC as well as the pro- involved in a fossil interplate shear zone studied in the
cesses governing sediment transformation and shear locali- Northern Apennines came from the over‐riding plate,
zation are poorly constrained because of the lack of deep whereas incoming plate sediment was thrust undeformed
drilling and the poor spatial resolution of seismic data. beneath a basal décollement. Therefore, on one hand the SC
Numerical models based on best available measurements, may be viewed as a shear zone from field or drilling data, thus
empirical laws and laboratory experiments were used to defining the SC in the stricter sense of the term (SC s. s.), and
estimate the controls on the aseismic ‐ seismic transition on the other hand, as the interplate low‐velocity sediment

2 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

Figure 2. Prestack depth‐migrated (PSDM) MCS line SIS‐72: (a) PSDM image based on the Kirchhoff
algorithm and a 2D velocity model obtained by focusing analysis and iterative velocity picking; TOC =
Top of Oceanic Crust reflector; TSC = Top of Subduction Channel reflector [after Calahorrano et al.,
2008]. (b) PSDM calibrated image obtained by Iterative ray + Born waveform migration/inversion
[Ribodetti et al., 2011]. Note the difference in thickness and dip of the subduction channel and the finer
resolution in the channel and accretionary wedge.

layer based on MCS data, hence characterizing the SC in the segment of the SC off the Gulf of Guayaquil (Figure 2a)
broader sense (SC b. s.). These SC concepts do not neces- [Calahorrano et al., 2008]. We applied pre‐stack depth
sarily overlap [Vannucchi et al., 2011]. From the combination migration/inversion (PSDM) and a post‐processing sequence
of geophysical imaging and geological data, the SC can be to obtain a 2D quantitative image of the first 24 km of the SC
defined as a dynamic, finite width interplate shear zone that (Figure 2b). We seek to discriminate tectonic deformation
accommodates both viscous deformation and distributed from preserved sedimentary architecture within the SC, dis-
slip [Vannucchi et al., 2011]. Nevertheless, for consistency tinguish whether and where the deformation is distributed or
with previous literature, we keep using the generic term SC localized, identify levels where pressured fluids tend to
b. s. to designate the seismically imaged interplate sediment accumulate, and lastly reconstruct the recent tectonic history
layer, no matter what the sediment origin, but knowing that of the SC along line SIS‐72.
the shear zone, i.e., the SC s. s., may not affect this layer but
only bound it.
[7] Imaging the SC at the best possible resolution, and 2. Geodynamic Context
analyzing the physical properties of its seismic reflector is a [8] In Ecuador, permanent deformation due to the 5.8 cm/yr
pre‐requisite to place new constraints on the structural eastward subduction of the Nazca plate [Kendrick et al., 2003;
evolution of the SC and on the stability of its boundaries. In Trenkamp et al., 2002] causes the opening of the Gulf of
this paper, we reprocessed MCS line SIS‐72 that was col- Guayaquil [Deniaud et al., 1999; Witt et al., 2006] in relation
lected across the Ecuador subduction zone (Figure 1a) to to the ∼6 mm/yr, northeastward motion of North Andean
tentatively resolve structural details of the spectacular updip Block (NAB) along the Dolores‐Guayaquil Megashear

3 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

(Figure 1b) [Campbell, 1974; Ego et al., 1996; Winter et al., 4. Methodology: 2D Fine‐Scale Imaging
1993]. As a consequence, thick Plio‐Quaternary sedimentary by Integrated Iterative PSDM and Simulated
basins controlled by normal faults have formed in the Gulf of Annealing Optimization
Guayaquil, in particular during the past 1.8–1.6 Myr
[Deniaud et al., 1999; Witt et al., 2006]. Offshore the Gulf of [10] To obtain a depth migrated image and small‐scale
Guayaquil, the Grijalva Fracture Zone (GFZ), which sepa- velocities, we used a two‐step approach based on ray + Born
rates the Oligocene Farallon lithosphere to the south, from the waveform migration/inversion, also known as preserved
Neogene and morphologically complex Nazca lithosphere to amplitude pre‐stack depth migration (PSDM) [Thierry et al.,
the north, intersects the Peru trench. The trench is deepest 1999], and very fast simulated annealing (VFSA) optimi-
south of this intersection and favored a ∼500–700 m accu- zation [Ribodetti et al., 2011].
mulation of terrestrial sediment in relation with the activity of [11] 1. Iterative ray + Born waveform migration/inversion,
the Guayaquil canyon. There, an incipient ∼5 km‐wide including an iterative correction of the Vp velocity macro‐
accretionary wedge fronts the margin [Calahorrano et al., model [Al‐Yahya, 1989], is used to obtain a 2‐D quantitative
2008; Collot et al., 2002]. North of the GFZ‐trench inter- and accurate migrated image. Then, we computed several
section, the subduction of the Carnegie Ridge, a prominent iterations of ray + Born waveform modeling and inversion
200‐km wide, 2‐km high volcanic ridge, uplifts the trench, until the convergence was achieved in order to recover the
which is almost bare of sediment [Lonsdale, 1978]. Sub- true amplitude of the velocity perturbations as accurately as
duction of the Nazca plate is also responsible for the mega- possible. In case of multiple arrivals, we noticed that several
thrust earthquakes in1906 (Mw 8.8), 1942 (Mw7.8), 1958 iterations of single‐arrival migration/inversion allowed us to
(Mw7.7) and 1979 (Mw8.2), that ruptured ∼500 km of the estimate velocity perturbations in the areas of the model
Ecuador‐Colombia plate boundary [Kelleher, 1972; Mendoza intersected by caustics as accurate as those inferred from
and Dewey, 1984; Swenson and Beck, 1996]. GPS data single‐iteration multiple‐arrival migration [Operto et al.,
indicate a patchwork of highly coupled asperities in the rup- 2000]. Finally, the migrated image is calibrated by estima-
ture area of the 1906 earthquake (Figure 1b), whereas along tion of the reflection coefficient on the seafloor [Warner,
southern Ecuador‐northern Peru, inter‐plate coupling is very 1990] to obtain the Vp velocity perturbations in the SI
low [Nocquet et al., 2010]. In the Gulf of Guayaquil, seis- system.
micity is low with the exception of the 1953 M7.8 earthquake [12] 2. An automated post‐processing procedure is applied
that occurred offshore Tumbes (Figure 1b) and is reported in to eliminate the source signature from the migrated image,
the South America SISRA catalog and by Silgado [1957]. The and to estimate the correct geometry (i.e., layer thickness)
NE escape of the NAB superimposed on a bow effect due to and the absolute values of the velocity of seismic reflectors,
the convex shape of the northern South America promontory as well to estimate results uncertainties. The post‐processing
[Bonnardot et al., 2008] are considered to release part of the [Ribodetti et al., 2011] is formulated as an inverse problem
inter‐plate pressure, thus possibly accounting for the weak for which the data space is composed of several vertical logs
plate coupling in the Gulf of Guayaquil. Overall, such weak of the migrated image (namely, the limited bandwidth
inter‐plate coupling might facilitate subduction of incoming velocity logs). The model space is composed of the 1D‐
plate sediment. velocity logs (namely, the impulse velocity logs), parame-
terized by a limited number of layers with random velocity
and thickness. The relation between the data and the model is
approximated by a simple time domain convolution with the
3. Geophysical Data and Initial Results limited‐bandwidth source wavelet. For the post‐processing,
[9] The 2D SIS72 data was acquired during the SISTEUR we used the wavelet obtained by linear inversion of the water
cruise on board the R/V Nadir [Collot et al., 2002]. The wave [Ribodetti et al., 2011] based on the linear inversion
streamer was 4.5 km long with 348 groups of geophones at method proposed by Pratt et al. [1998]. The inverse problem
12.5‐m interval, recording 4‐ms samples, with a 15‐s record is solved independently for each log by a random exploration
length. Source was a 48‐L (2869 inch3) air gun array tuned of the model space, using the VSFA algorithm [Sen and
in a single‐bubble mode [Avedik, 1993] and fired at 50‐m Stoffa, 1995]. The uncertainty and the error analysis (cen-
interval. The geometry yielded a 43‐fold MCS data set. The tral and dispersion statistical estimators) are investigated by
line had been initially processed up to PSDM based on the multiple VFSA, which allows to estimate the frequency of
Kirchhoff algorithm using the SIRIUS‐2.0 software package, visits of each accepted specific cell of the model space
and a 2D Vp velocity model had been constructed by focusing parameterized by layer thickness and velocity [Jackson et al.,
analysis and iterative velocity picking [Calahorrano et al., 2004; Sen and Stoffa, 1995].
2008]. The resulting image (Figure 2a) shows the updip
segment of the subduction channel to be a ∼0.8–0.9 km‐thick,
2.7–2.8 km/s low‐Vp layer dipping ∼4° landward on average 5. Application to MCS Line SIS‐72
[Calahorrano et al., 2008]. The SC segment is roofed and [13] Prior to apply the above processing sequence to MCS
floored by the strong and continuous Top of Subduction line SIS 72, we pre‐processed the data to preserve phase
Channel (TSC) and Top of Oceanic Crust (TOC) reflectors amplitudes, using the Geovecteur software. Preprocessing
that Calahorrano et al. [2008] interpreted respectively as the included data sorting to 6.25‐m CDP, first pass velocity
interplate décollement and the roof of the underthrusting analysis, amplitudes attenuation (0.001 factor) of noisy
oceanic crust. traces, a band pass filter (3, 6, 50, 60 Hz), minimum phase
operator, multiple attenuation in the frequency–wave

4 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

number (FK) domain, normal moveout velocity analysis, straightforwardly up to near the seafloor at km 88, ∼3 km
loose external mute, spherical divergence correction, pre- landward of the deformation front. Accordingly, the TSC
dictive deconvolution, second multiple attenuation using reflector and the frontal thrust T0 are structurally dis-
Radon transform, inverse spherical divergence correction connected (Figure 3b). In contrast, although deep seismic
and inverse NMO correction, second band pass filter (3, 6, reflectors are poorly continuous at km 84–85, thrust fault
50, 60 Hz), and sorting of CDP to shot gather. Although, the T1 is interpreted to splay away from near the TOC
work by Calahorrano et al. [2008] provided new constrains reflector, and join seaward active fault T0, pointing to the
on physical and mechanical properties at the scale of the likely structural continuity between the TOC reflector and
SC, applying our new processing sequence allows refining the frontal thrust. This structural interpretation implies
the spatial resolution of the SC velocity structure, enhancing tectonic activity along a basal thrust zone. Hence, the SC
the reflections coherency, and modeling reflectors as thin is inferred bounded by roof and basal décollement thrusts.
layers at the roof and floor of the subduction channel. The
improvement and accuracy of the Vp velocity model, the 6.3. Internal Structures of the SC
migrated image and the validity of the post‐processed model [17] Internal structures of the SC derived from the true
are presented in the auxiliary material.1 Although some amplitude migrated image reveal two main stratified sequences,
reflection patterns may be 2D‐migration artifacts or 3D‐out‐ which are discriminated from their reflector patterns. The
of‐plane effects, in the following we mostly interpret varia- lower sequence (Ls) is 220–500‐m‐thick, and becomes
tions in reflection at a scale relative to the average thickness distinct below T0 fault, whereas the upper sequence (Us) is
of the two major SC internal sequences. 450–800‐m thick. Both sequences show a variable seismic
pattern, which may reflect sedimentary and tectonic struc-
tures. At km 73, Ls sequence and the TOC reflector are
6. Results and Interpretation affected by a ∼250 m vertical offset that we interpret as a
normal fault (F in Figure 4) in the oceanic crust.
6.1. Large‐Scale Structure and Geometry
[18] Sequence Ls divides into three units on the basis of
[14] The SC varies in thickness from 630 m to 1150 m, their reflectivity and structural pattern. Unit Ls1 is the
and shows a large scale, ramp‐flat structure that dips ∼6° shallowest (5.5–5.9 km, at km ∼81–87) and underlies T1
landward (Figure 2b). The channel tapers down landward thrust up to km 87 pointing to the décollement nature of its
from ∼1000 m at km 67 to less than 300 m at km 63, where contact with overlying Us Unit. Unit Ls1 is thinly stratified
TSC reflector outlines a footwall ramp (Figure 2a) and weakly reflective. Unit Ls2 extends at depths ranging
[Calahorrano et al., 2008]. The flat segment of the SC rises from 5.9 to 6.9 km (km 73–81.5). Its lower section is
gently seaward. It is buckled and mimics the topography of strongly reflective where it drapes an irregular segment of
the underlying TOC reflector that reveals topographic the oceanic crust and the hanging wall block of normal
irregularities including small seamounts and horst‐and‐graben fault F. Seaward verging reverse and thrust faults, which
structures. It is remarkable that groups of up to 300‐m‐thick, appear to sole out near the TOC reflector, are interpreted
very strong SC reflectors extending landward from km 83 are to deform Ls2 at depths greater than ∼6.0 km. Unit Ls3 lies
associated with both the TSC and TOC reflectors (Figure 2b). on the footwall block of normal fault F at a depth of 6.9–
[15] In the trench, a package of subhorizontal, well‐ 7.8 km, and shows structural complexity with thickness
stratified turbidite is uplifted by 150 m and deformed by the variations, folding and seaward verging thrust faults. The
seaward‐verging frontal thrust (T0) [Calahorrano et al., contractional deformation affecting Ls unit provides addi-
2008] thus creating the youngest thrust sheet (TS1) tional support for the TOC reflector and likely part of Ls
(Figure 3). Farther landward, our new image allows inter- unit to feature a basal thrust zone.
preting a remarkable structural continuity over the 24 km‐ [19] Sequence Us was divided into four units on the basis
long, studied segment of the SC. Continuity is followed of their reflectivity, boundary types, and structural patterns.
from the landward end of the section at km 67 and depth of Unit Us0 is thinly bedded and moderately reflective in both
∼7.5 km (Figure 4) to the toe of the margin slope near km the trench and thrust sheet TS‐1. In the SC, between km 80–
89, where continuous, well‐bedded and slightly warped up 88, the unit varies in thickness from 180 to 400 m then thins
internal reflectors come close to the seafloor, thus forming a toward km 79, where the unit disappears. Its reflective pattern
long thrust sheet (TS2) (Figure 3). TS2 overthrusts TS1 along varies from weak, to very strong between km 79 and 83.
thrust fault (T1) thus forming a 150 m‐high seafloor relief. At their landward termination, Us0 strata lap onto underlying
6.2. Evidence for Roof and Basal Décollement Thrusts unit Us1 (km 81–82) pointing to a likely stratigraphic contact.
Unit Us1 extends along the SC from km 76.5 to 89. In its
[16] Although TSC reflector, which is strong, continuous, seaward half, Us1 undulates, is 250–300‐m‐thick, well bed-
and sharply truncates seaward dipping reflectors at the ded to discontinuous and poorly reflective (Figure 3). The
bottom of the upper‐plate basement (Figure 4), was inter- unit reaches 580 m in thickness near km 79. Between km
preted as the décollement thrust [Calahorrano et al., 2008], 77–81, the Us1 upper section is strongly reflective and the
the new PSDM image (Figure 3) argues against continuity reflectivity extends across the boundary between units Us0
between the interpreted TSC décollement and the active and Us1. Unit Us1 reflectors are affected by small throw
deformation front. Based on the structural continuity of the vertical offsets between km 78–84 suggesting either artifacts
SC along the line, the TSC reflector can be extended or small‐scale, high‐angle normal faults. Near the landward
termination of Us1, the highly reflective uppermost bedding
1
Auxiliary materials are available in the HTML. doi:10.1029/ is conformable with the TSC reflector, which sharply trun-
2011JB008429.
cates the bedding of the overlying margin wedge rocks, thus

5 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

Figure 3. Close‐up of the frontal part of PSDM line SIS‐72. Location is shown in Figure 2.
(a) Un‐interpreted data, (b) line drawing; TSC and TOC as in Figure 2. Us0–Us1 = channel Upper
Sequence sedimentary units. Ls1–Ls2 = channel Lower Sequence sedimentary units. TS1 = youngest
thrust sheet; TS2 = long thrust sheet; T0 and T1 = thrust faults; Up2 = possibly underplated duplex. Note
the continuity of TS2 from the seafloor at km 88, downdip, and strongly reflective layers below TSC and
above TOC reflectors.

supporting the roof thrust nature of TSC. Unit Us2 is a normal faults. Unit Us3 has a sigmoid cross section or
rhombic‐shaped block that is present between km 74 and 79, deformed rhomb section, thinning away at km 76 and 70, and
and reaches ∼800 m in thickness, near km 76. It is well at a depth of ∼6.5 km. The interpreted reverse and thrust faults
stratified, poorly reflective, and its bedding shows a sharp dip that cut across underlying unit Ls2 between km 74 and 76
increase with respect to Us1 bedding. Us2 uppermost section deform Us3 implying that deformation postdates Us3 depo-
has a wedge shape thinning updip possibly implying some sition. The poor resolution of Us3 reflections near km 72
tilting of the unit. The nature of its relationship to both Us1 precludes defining its contact nature (R3) with underlying
and Us3 units is however ambiguous. Us1/Us2 boundary, i.e., Unit Us4. Unit Us4 together with Ls3 form the bulk of the
contact R1 could be either structural through a fault, or sed- flat segment of the SC that reaches 1150 m in thickness.
imentary via Us1 downlaps onto Us2 bedding. Us2 beds Us4 seismic facies changes laterally from well‐bedded,
clearly downlap onto Us3 revealing the unconformable highly reflective to poorly reflective and chaotic. Folding and
nature of contact R2. However, the boundary could either be a seaward‐verging thrusting deforms part of Us4 conformably
fault or a stratigraphic contact in as much as the possible with sequence Ls3 indicating synchronous deformation. In
initial boundaries of the sedimentary units were likely conclusion, reprocessed Line SIS72 provides support for two
deformed during subduction. Were these boundaries faults, décollement thrusts bounding a highly deformed zone. This
then the bedding pattern relative to the fault dip direction zone experienced extensional and contractional tectonism
would suggest a minor clockwise rotation along low angle,

6 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

Figure 4. Close‐up of the subduction channel of PSDM line SIS‐72. Location is shown in Figure 2.
(a) Un‐interpreted data, (b) Line drawing; TOC as in Figure 2. Us0–Us4 = channel Upper Sequence
sedimentary units. Ls2–Ls3 = channel Lower Sequence sedimentary units. F = normal fault; Up1 =
possibly underplated duplex. R1–R3 seismic discontinuities between Us units; Note possible small scale
normal faults affecting Us1 unit, and reverse, thrust faults and folding affecting Ls2–3 and Us3–4 units.
Strong reflectivity is associated with layers related to TSC and TOC reflectors.

and shows that a mid‐level décollement may have separated 6.5. A Refined Vp Velocity Model of the SC
Us and Ls units. and Lowermost Margin Wedge
[21] A refined Vp velocity model (Figure 5a) was com-
6.4. Lens‐Shape Structures in the Over‐Riding Margin puted by summation of the velocity macromodel (Figure S1c
[20] The initial PSDM image had revealed a lens‐shape of the auxiliary material) and the mean broadband best fitting
seismic structure that overlies reflector TSC [Calahorrano, perturbation model (Figures 6b (top), 7a (top), and 7b (top)).
2005]. This puzzling structure is asymmetric, 400 m‐thick, This velocity model illustrates a velocity inversion associ-
about 6 km‐long (Up1 in Figure 4). The structure shows ated with the SC where velocities reach values as low as
continuous seismic horizons of moderate amplitude, alike 2.7–3.1 km/s in contrast with 3.2–4.0 km/s velocities within
those of the SC, and form a fold conformable with overlying the overlying margin rocks. Beneath the SC, velocities
basement structures. A similar but smaller lens‐shaped increase to 4.2–4.5 km/s at the top of the oceanic crust.
structure is evident near the margin front at km 86 (Up2 in These findings are in agreement with the larger‐scale
Figure 3). Up2 structure shows, however, a complex struc- velocity model of the SC presented by Calahorrano et al.
tural relationship with overlying rocks possibly forming [2008]. Their velocity model of the SC also revealed two
small imbricates at the margin front. Both Up1 and Up2 areas of subtler, secondary Vp variations. These Vp varia-
structures were truncated at their base along the TSC tions appeared as well in our macro‐velocity model after a
reflector. migration/inversion iterative refinement (See Text S1 and
Figure S1c of the auxiliary material), and were further

7 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

Figure 5. Fine scale Vp velocity model of the deepest segment of the subduction channel as obtained
after post‐processing of the migrated image. (a) Velocity contours in km/s. Note the relative low velocity
of the channel; red arrows point to patches of lowest velocity; Dimmed area indicates zone with poor
velocity control. (b) Superposition of the velocity model over the PSDM image of the subduction channel.
Note two low velocity anomalies LV1 and LV2 in the margin wedge suggesting fluid‐altered rocks or
underplated material; the channel low velocity patches locate respectively at a depth of 6 km ahead of
a seaward verging thrust zone, and at a depth of 7 km sandwiched between fault F, a likely source of
fluids, and compressively deformed Ls3–Us4 unit.

refined by our post‐processing procedure (Figure 5a). First, 6.6. Vp Perturbations of Reflectors Associated With
two patches of remarkably low velocity, reaching 2.7 and the Roof and Floor Thrusts
2.9 km/s (Figure 5a), are located within the Us sequence at km [22] On the calibrated‐migrated image with absolute
72–73 and 78 (Figure 5b). Second, two areas of 3.2–3.5 km/s velocity perturbations, sets of strong positive (Pa/Pb) and
relatively low velocity set astride a 3.6–4.0 km/s, higher negative (Na/Nb) amplitude reflectors are associated with
velocity core in the lower section of the overlying basement both the roof and the basal thrusts (Figure 6a). Pa is the most
(LV1, LV2 in Figure 5b). The subtle SC velocity variations continuous positive amplitude reflector along the roof thrust,
in the macro‐model are validated with a ∼4% velocity var- although it nearly disappears at km 71–72 (Figure 6a). Pa
iation by the sensitivity analysis of the Vp model presented overlies up to three negative amplitude reflectors, Na, Na1
in Text S1 of the auxiliary material (Figures S5–S6). and Na2, which are restricted to the shallowest well‐stratified

8 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

Figure 6

9 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

Figure 7. Results of the post‐processing showing close‐up of (a) middle and (b) lower segment of the
channel; top panels are mean broadband perturbation models with superimposed 5–10% standard devia-
tion; bottom panels are mean broadband perturbation models superimposed on the depth migrated image.
Captions as in Figure 6.

segment of the SC (km 77 to 81 in Figure 6a). Down dip, two however no less than the one‐quarter wavelength of the
negative amplitude reflectors Na and Na3 straddle reflector seismic source, where the wavelength is the media velocity
Pa (km 68–76 in Figure 6a). Reflector Pb at the base of the SC divided by the source wavelet frequency. Because the source
shows very strong positive amplitude of the velocity pertur- frequency is 18 Hz and the average media velocity in the SC is
bation (km 80–81, 78, 73.5–74) associated with strongly 3 km/s, layers less than 42 m‐thick will not be resolved. Pa
negative amplitude reflector Nb. reflector is resolved by a 50 to 110 m‐thick layer with a 50 to
[23] The reflector signatures translate into significant 200 m/s positive velocity perturbation, whereas reflectors Na,
velocity perturbations and thickness variations in the 1D‐ Na1 and Na2 fit with ∼40 to 80 m‐thick layers with ∼50 to
velocity logs that were obtained after post‐processing every 280 m/s negative velocity perturbations. The Pb reflector is
250 m along the line (Figures 6b (top), 7a (top) and 7b (top)). resolved by a 60–100 m thick up to 300 m/s positive velocity
The thinnest layer to be resolved by the post‐processing is

Figure 6. Results of the post‐processing applied to the deepest segment of the SC. (a) Ray + Born depth migrated image of
the SC showing absolute velocity perturbations (m/s). Na and Nb are negative velocity layers; Pa and Pb are positive velocity
layers. (b) Close‐up of upper SC segment: (top) Mean broadband perturbation models with superimposed 5–10% standard
deviation. Blue/Red dots are remarkable negative/positive velocity layers. Layer segments A, B and C are explained in the
text; (bottom) Mean broadband perturbation models superimposed on the depth migrated image. In some areas as near km
80–80.5, when the modeled layers are either slightly dipping or too thin, the post‐processing struggles finding the best model.
A succession of thin layers with alternating high and low velocity was found however to match the data at km 80–80.5, but the
calculated sequence is too deep by ∼50 m. Because the post‐processing is based on a layer stripping approach [Ribodetti et al.,
2011], the vertical shift, which affects successively deeper layers down to the base of the channel, probably increases due to a
stretching effect.

10 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

perturbations, when Nb reflector reaches the highest negative lengths of TS2 between km 88 and 79. As a result, the time
velocity perturbations >350 m/s near km 74 (Figure 7a). required to underthrust TS‐2 from km 88 to km ∼79 was
[24] According to [Ribodetti et al., 2011] three types of estimated 180 ± 9 kyr. Hence, Us0 youngest sediment at km
layer segments can be identified on the basis of their 79 would be 234 ± 22 kyr old (Figure 8). Similarly, Us4
velocity perturbations. A‐layer segments have strong nega- sediment near km 68 would include ∼450 kyr‐old material.
tive velocity perturbation (>150 m/s). Examples are given This approximate ages line points to an apparent change in
for Nb at km 73.5–74, 75.75 in Figure 7a, and km 80.75–81 in trench sedimentation at 234 ± 22 kyr. Prior to this age, thick
Figure 6b; Na at km 77.5, 78 and Na1 at km 80.75 (Figure 6b); packets of sediment may have filled the trench, whereas
and Na3 at km 70 (Figure 7b). These segments are fre- ensuing Us0 sedimentation progressively overlaid Us1
quently overlain or rest above a high velocity layer. B‐layer unconformably.
segments show smaller (<150m/s) negative velocity pertur-
bation. Na examples are shown at km 76.5; 78.25–78.5 in 7.2. Is the SC Structure a ∼450 kyr Time Window
Figure 6b). A high velocity layer locally underlies these on Trench Sedimentation?
segments. Na examples are given at km 70.4 (Figure 7b) and [28] Conceptually, the sedimentary architecture of the SC
km 75 (Figure 7a). C‐layer segments are not associated with should reflect the lithologic changes and sedimentary
detectable amplitude variations and therefore do not show dynamics that prevail in the trench as subduction proceeds.
noticeable velocity perturbations. Na examples are shown at Along line SIS‐72 the Nazca plate is of Oligocene age
km 72.25; 73.5; 75.25 (Figure 7a) and km 81.25 (Figure 6b). [Lonsdale, 2005] i.e., older and deeper than the southern
flank of the Carnegie Ridge, where ODP Site 1238 recov-
ered the complete Neogene sequence down to the Miocene
7. Discussion oceanic crust [Mix et al., 2003]. Extrapolating the lithologies
[25] The new constraints put on the sedimentary and from this site suggests however that sequence Ls, which
tectonic structures of the updip segment of the SC along line blankets the oceanic crust in the south Ecuador trench,
SIS72 permit interpretation of the tectonic evolution over consists of diatom and nannofossil ooze with discrete ash
the last ∼450 kyr, and provide a novel insight over the layers. The lithologies of the well‐stratified reflective trench
mechanisms that controlled its structural fabric and fluid fill (sequences Us) can be approximated by fine grain tur-
flows. bidites and mud clasts flows inter bedded with mud and
hemipelagic silty clays in accordance with sediment cored
7.1. Time Frame for Trench Sedimentation and SC from the nearby central Ecuador trench [Lonsdale, 1978;
Tectonics Ratzov et al., 2010].
[26] Our interpretation of line SIS72 suggests that as [29] Considering steady sediment subduction in a non‐
trench sedimentation continued and subduction proceeds, accretionary regime, architectural variations of the trench
sediment kept underthrusting the margin as a long sheet sediment might be preserved along the SC as a function of
(TS‐2) prior to initiating frontal accretion (TS‐1). The roof the sediment progressive aging along the roof thrust. Pre-
thrust is not viewed as an out‐of‐sequence fault because, in served characteristics, among which apparent unconformities
contrast with T0 and T1 thrusts, no seafloor relief is asso- between the sedimentary units, (Figure 4) might unveil
ciated with it (Figure 3). We then consider that the roof remnants of the original sedimentary architecture. For
thrust was abandoned when T1 thrust initiated. The timing example, several turbidite and debris flow systems could
of this tectonic change can be appraised by restoring the have successively formed in the trench over the ∼234 to
fault‐related deformation in the first 4 km of the wedge tip. 450 kyr time period, thus possibly accounting for the
In addition, an age line can be constructed along the roof apparent unconformities and changes in layering dip
thrust to help constrain the temporal evolution of SC tec- between Us1–Us4 units. However, the 13 to 18° dip angle
tonic deformation. between the apparent unconformities and the roof thrust
[27] Cumulated tectonic shortening along T0 and T1 appears too large to be of stratigraphic origin, in comparison
thrusts is estimated to range between 2.3 km and 2.9 km with the shallow 0.5–2° seaward dip of the trench seafloor at
taking into account a 10% error on fault finite displacements. the Astoria and Esmeraldas deep sea fans [Nelson, 1983;
We added diffuse tectonic shortening of the deformed sedi- Collot et al., 2008]. Then, the hypothesis of several dia-
mentary section with values ranging between 10% [Adam chronic turbidite systems is unlikely to account for the SC
et al., 2004] and a 50% upper bound, considering that internal structure. Alternatively, large blocks transported
strain distribution along the SC is poorly known. Under these with debris avalanches [e.g., Collot et al., 2001; McAdoo
considerations, a simple timing reconstruction based on a et al., 2000] could have contributed to the sedimentary
constant 5.8 cm/yr plate convergence rate [Kendrick et al., architecture of the SC. The landward‐tilted section of the
2003] indicates that fault T1 initiated 46 ± 4 kyr to 63 ± ∼800 m thick Us2 unit that rests unconformably against Us3
6 kyr ago, thus implicitly dating the roof thrust abandonment could suggest that a ∼5 km‐wide coherent block of sediment
at 54 ± 13 kyr. Accordingly, this age also dates the last and slid into the trench. Nevertheless, slump blocks are expected
youngest Us0 sediment to have underthrust the margin tip at to be more compact than surrounding trench sediment
km 88. Therefore, the youngest age of both Us0 at the land- implying locally higher Vp. Our detailed velocity model of
ward tip of the unit (km 79), and Us4 sediment at the end of the SC (Figure 5) does not reveal high‐velocity patches
the line segment (km 68) could be roughly estimated. By within the SC, thus discarding this hypothesis. In conclusion,
considering that underthrusting of long sheet TS‐2 was steady the structural changes imaged in the SC are rather a conse-
state, we added an up to 20% diffuse tectonic shortening to quence of tectonic deformation than the preserved signature
the estimated minimum (8.3 km) and maximum (9.1 km) of temporal variations of the trench sedimentary architecture.

11 of 20
 B11102

12 of 20
Figure 8. Seismic units and structural interpretation of the SC. Labels are as in Figures 3 and 4. Red numbers along the
roof thrust are ages for TS2 underthrusting according to a 5.8 cm/yr plate convergence [Kendrick et al., 2003]. Finite
displacement along T1 thrust was measured between the blue and the pink and white stars for error assessment. Under-
thrusting of TS2 initiated earlier than ∼450 kyr and terminated 54 ± 13 kr ago, when TS1 frontal accretion commenced (see
text for explanation).
COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL
B11102
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

7.3. The SC Structure: A ∼450 kyr Record metrically compatible with en echelon faults that formed
of Interplate Strain during simple shear deformation. Us2 and Us3 units likely
[30] A remarkable result of this study is to show that the acquired their rhomb like shape during this tectonic shear-
entire ∼<1 km‐thick sediment pile of the Nazca plate was ing. The interpreted en echelon faults formed in a similar
underthrust over a distance >23‐km during a ∼400 kyr period, manner as synthetic Riedel shears in a strike‐slip environ-
prior to initiating frontal tectonic accretion 54 ± 13 kyr ago. ment, according to the scheme by [Logan et al., 1992].
During these events, a specific shear fabric was acquired Therefore, the sense of slip along faults R1 to R3 owes to be
under increasing strain. normal, an inference that is compatible with our structural
7.3.1. The SC Internal Deformation interpretation of line SIS‐72. Synthetic R shears form at a
[31] According to Shreve and Cloos [1986] tectonic ∼15° angle relative to the main line of faulting and they
deformation along the SC may reflect viscous shear as further achieve a closer angle as shearing along the main
material flows downward. On the long‐term, the flow of SC fault proceeds [Davis and Reynolds, 1996]. Accordingly, the
material can be broken up into a steady state Poiseuille flow ∼13° ± 1° angle of fault R1 to the roof thrust is in good
and a Couette flow [England and Holland, 1979] (Figure 9a). agreement with a mature roof thrust (Figure 9e). Although
Under the effect of the Poiseuille flow, sediment tend to limited, the vertical displacement of R1 to R3 low‐angle
flow seaward i.e., counter‐subduction as the low‐density normal faults implies motion along the roof of unit Ls,
viscous flow is driven by the downdip increasing pressure which is thus interpreted as a mid‐level décollement. The
gradient exerted by the hanging wall on the channel. Under décollement may have formed along the Us/Ls boundary,
the Couette flow effect, shear develops within the SC which separates the interpreted sandy to hemipelagic trench
material because subduction forces the flow downdip as fill from an underlying pelagic unit. This décollement level
material viscosity makes it stick to the SC boundaries. The is substantiated by thrust fault T1, which presumably soled
resulting composite deformation (Figure 9a) suggests that a out on Us1/Ls1 boundary (Figure 3). Faults R2 and R3 were
mid‐level shear zone could theoretically form, depending on later deformed by folding and thrust faulting associated with
subduction velocity and viscosity of the media. Observa- the downdip increasing strain. Similarly, in the exhumed SC
tions of exhumed SC and accretionary wedges that were of the Northern Apennines, Vannucchi et al. [2008] docu-
buried at depths <7 km [Moore et al., 2007; Vannucchi et al., ment discrete normal faults, which become overprinted by
2008] reveal, however that their rocks mainly deformed in the contraction in the SC deeper part (>5 km). In contrast with
brittle field, under 50°–150°C temperatures. normal faulting, Riedel shears and thrust faults hardly ever
[32] Along line SIS‐72, the few discrete high‐angle, form in the down going plate and trench sediment, hence
small‐throw normal faults that are interpreted to cut the indicating that SC Riedel shear and contractional strain was
uppermost SC layers at a depth of 5.2–5.5 km reflect both acquired during subduction. An important implication of the
compaction and extensional brittle deformation. Such an contractional strain identified within the SC is that, a few km
extension regime was documented in the 500 m‐thick fossil downdip from the margin tip, tectonic pressure (Ptec)
SC of a former erosive margin preserved in the Northern (maximum horizontal principal stress) cannot be neglected
Apennines of Italy [Vannucchi et al., 2008]. This study reported when calculating the total pressure (Pt) supported by the SC
evidence for pervasive extensional shear fractures and gently because Pt equals the lithostatic pressure (Plit) augmented by
dipping normal faults affecting yet poorly consolidated Ptec.
sediment in the shallow part (<∼3 km) of the SC. [35] In conclusion, our structural interpretation of line
[33] Interpreted thrust and reverse faults indicate that SIS‐72 does not provide evidence for Poiseuille flow within
contractional strain becomes distributed across both Ls and the studied portion of the SC, but strongly supports brittle
Us units in the downdip part of the SC at a depth of ∼6.5– and slip behavior within a mega‐shear zone. This behavior
7.5 km, supporting a complex and highly deformed zone. requires the rheology to vary from elastic to plastic depending
This contraction contributes to thicken the SC. Near oceanic on material strength, temperature, stresses and shear rates,
crust fault F, thrusting distributes broadly within units Ls2– and fluid pressure.
3 and Us3–4, with the exception of a limited area on the [36] SC excess pore pressure, is shown to reach up to
down throw side of fault F, which could form a shadow ∼40 MPa, 25 km from the trench [Calahorrano et al., 2008],
zone to deformation (Figure 9e). The fault likely renders the a value that is close to the 32 MPa calculated 20 km from the
basal thrust inactive immediately ahead of the fault. It is trench in the Nankai SC [Tobin and Saffer, 2009]. This
worth noting that contraction begins penetrating deeply into elevated pore pressure considerably weakens SC material and
units Ls2 and Us3 near km 76.5, which is ∼3 km behind facilitates internal deformation. According to the Vp‐Porosity
fault F location. Considering that the basal thrust is active at empirical relationship by Erickson and Jarrard [1998], the
the average plate convergence rate, Fault F traveled this two patches of remarkably low velocity identified within the
distance perhaps over the last ∼50 kyr, an age that is close to SC (Figure 5) are interpreted as zones of slightly higher
our estimated age for basal thrust inception. Motion transfer porosities, and hence higher fluid content, that were preserved
across fault F was then either accommodated by a ramp fault within an otherwise more compacted or finer grain sediment.
connected to the basal thrust near km 71, or distributed across Although sediment permeability and compaction by loading
the greater part of the SC further landward (Figure 9e). This are key parameters for fluid flow, faulting squeezes sediment
analysis brings to light the driving influence of a subducting and concurrently affects porosity and fluid flow [Knipe,
oceanic fault scarp on the shear distribution in the SC. 1993]. Interestingly, the low velocity patches are located up
[34] The gently seaward‐dipping contacts R1 to R3 dip of thrusting zones (Figure 9e) from which, fluids were
between units Us1–Us2, Us2–Us3 and Us3–Us4 are geo- potentially expelled to accumulate in the interpreted fluid
sinks. Alternatively, the deepest fluid sink is associated with

13 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

Figure 9. Deformation pattern in the SC. (a) A homogeneous viscous flow in the SC is theoretically driven
by Poiseuille flow due to the pressure gradient exerted on the channel by the hanging wall (as a result, weak
material tends to flow counter‐subduction) and Couette flow due to subduction drag forces that shear mate-
rial downdip [England and Holland, 1979]. The resulting viscous flow is the red curve. (b) Considered as a
mega‐shear zone, the SC develop conjugate fault arrays with synthetic (R) and antithetic (R′) Riedel shears;
R shear form as faults with a normal component from the roof thrust (Rt) at an acute angle of ∼15° to the
shear direction, and sole out on a mid‐level décollement (Mld). (c) R shear activation and block rotation tend
to trap thin layers of the overlying margin rocks that are subsequently captured by the SC during activation
of the roof thrust. This process promotes basal erosion, upward migration of the roof thrust and thickening of
the channel. (d) When the roof thrust is abandoned and the basal thrust (Bt) activated, SC units deform under
contraction together with R shear. (e) Strain pattern interpretation in the SC along line SIS‐72. Down to a
depth of 6.5 km, upper stratigraphic unit Us is dominantly affected by small normal faulting and large R
shear faults, whereas lower unit Ls progressively accommodates more contractional strain. Downdip from a
depth of ∼6.5 km, contraction intensifies, folds form and thrust faults tend to distribute across both Us and Ls
units, forming a highly deformed zone in green. This zone may transfer part of the plate convergence from
the updip basal thrust (left of F) to the downdip section of the SC or basal thrust (right of shadow zone) via a
ramp thrust or distributed deformation. Superimposed contours represent areas of lowest Vp velocity (see
Figure 5).

14 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

Fault F, which might be regarded as a conduit for fluids they are hydro geologically isolated from main fluid‐rich
released from the underlying oceanic crust and mantle during layers [Ribodetti et al., 2011]. According to Ranero et al.
dehydration processes [Ranero et al., 2003]. [2008], when the décollement is active at erosive margins,
7.3.2. The Roof Thrust: An Abandoned Megathrust most fluids are inferred to drain from the plate boundary
Acting as a Permeability Barrier? through the fractured upper plate to seep at the seafloor. Hence,
[37] The continuity of the roof thrust along line SIS‐72 the fluid‐depleted state at layer segments C‐type suggests
supports a weak, mature fault zone that has accommodated that, when the roof thrust was active, fluids migrated from the
substantial amount of displacements during complete SC and heterogeneously invaded overlying margin rocks, as
underthrusting of a long sheet of sediment. Total sediment supported by their relatively low velocity (Lv1 in Figure 5b). In
underthrusting requires maintaining a low‐friction at the conclusion, we interpret the inactive roof thrust as an overall
interplate décollement, a process that is related to the fluid permeability barrier for SC fluids that remain dominantly
content and physical properties of trench sediment [Hubbert trapped in the upper most stratigraphic layers of the SC. Based
and Rubey, 1959; Le Pichon et al., 1993; Saffer et al., 2000; on microstructural observations of the décollement zone in
Tobin et al., 2001]. The low friction owes to an elevated Nankai [Ujiie et al., 2003], we speculate that after shearing
pore fluid pressure that results from the rapid tectonic burial died along the roof thrust, pore spaced collapsed and cemen-
in the SC of high‐porosity, low‐permeability, fluid‐rich tation formed binding brecciated fragments to create the
surface trench sediment [Le Pichon et al., 1993; Sage et al., thin, relative 50–100 m/s high velocity modeled layer Pa that
2006]. Consequently, underthrust sub‐surface sediment fits TSC reflector (Figure 6). Consequently, when cementation
remain under‐consolidated relative to overlying wedge is no longer destroyed by cyclic stress conditions, the fault
rocks [Saffer, 2003] thus favoring the propagation of the core, and possibly its associated damage zone act as a barrier
décollement. Structurally, at the margin front the décolle- for fluid flow.
ment zone is a ∼9–40 m thick shear zone that comprises a low 7.3.3. A Possible Mechanism for Basal Erosion
permeability, ductile fault core or gouge overlying unde- [40] The development of Riedel shear faults across the SC
formed fluid‐rich sediment [Tobin et al., 2001; Maltman and may have an implication for a different form of the basal
Vannucchi, 2004]. A highly fractured, high‐permeability erosion proposed by von Huene et al. [2004]. These authors
damaged zone overlies the fault core [Caine et al., 1996; suggested that along a weak plate boundary zone, high‐pore
Saffer, 2003], and forms together with weakly fractured pressure hydrofractured the underside of the margin base-
margin rocks, a localized hydraulic conduit [Tobin et al., ment, dislodging clasts that subducted. In our hypothesis,
2001; Maltman and Vannucchi, 2004; Moore, 1989] along shearing could fully cut up into the overlying margin wedge
which, fluids flow under pressure fluctuation in relation with rocks so that fluids migrate preferentially along the newly
cyclic stress conditions [Morgan and Karig, 1995; Ujiie et created shear zone. This process could be facilitated by the
al., 2003]. Cross‐fault fluid flow is retarded across the duc- formation of R faults oblique to the SC (Figures 9b–9d). In
tile part of the fault, perhaps because of aligned clay minerals, this model, a wedge space opens below the roof thrust as a
so that fluid overpressure can build up as down going sedi- result of the small rotation of the Us unit along the R faults.
ment is rapidly overloaded by the advancing margin wedge Due to lithostatic pressure, overlying margin rocks subside
[Tobin et al., 2001; Sage et al., 2006]. and fill up the wedge space. Subsequent re‐activation of the
[38] Our structural interpretation of line SIS‐72 suggests roof thrust will capture the subsided rock layers, which end
however, that the roof thrust was abandoned, thus raising up being incorporated to the SC. This process could promote
the question of its present role as a long‐term conduit or a basal erosion as well as upward migration of the roof thrust
barrier for fluid flow. The 8 to 40 MPa excess pore pressure and thickening of the SC. In this hypothesis, the wedge‐
calculated within the SC [Calahorrano et al., 2008] together shaped uppermost Us2 layers (Figure 4) may be viewed as
with its low‐velocity patches and the brightest reflectors enclosing slivers of damage margin rocks.
associated with the roof thrust, support that fluids are 7.3.4. Underplating and Basement Damage Processes
trapped in the SC. Layer segments A and B‐type, i.e., layers Associated With the Roof Thrust
modeled with strong to moderate negative velocity pertur- [41] The remarkable lens‐shaped structure Up1 and a
bation (Figures 6 and 7) are inferred to reflect porosity var- heterogeneous, low‐Vp velocity zone (Figure 5b) charac-
iations in fluid‐rich sediment layers potentially characterized terize the bottom of the upper plate. The similar seismic
by excess pore pressure. Accordingly, over‐pressured fluids facies between Up1 and the SC, and the slightly higher Up1
would characterize the uppermost SC stratigraphic layers velocity (Figure 5, 3.2–3.6 km/s versus 2.9–3.2 km/s) sug-
associated with modeled layers Na to Na2 (Figure 6b). The gest that Up1 originated from dewatered subducting sedi-
model of fluid flow by Saffer et al. [2000] indicates that lateral ment, and is therefore interpreted as an underplated duplex.
fluid flow is possible in the SC if intrinsic sediment perme- The iso‐velocity contours of the low velocity zone mimic
ability is strongly anisotropic or if flow is focused along the roof thrust geometry, in particular between km 69 and 74
permeable layers. Moreover vertical fluid flow requires along the line, supporting that the low velocities were
densely spaced, high‐permeability conduits. We believe that acquired during subduction. Therefore, this low Vp velocity
tectonic deformation within the channel has created the zone calls either to under plated sediment less well imaged
conditions for some fluid to migrate upward and accumulate than Up1 or to fluid‐rock interaction in a damage zone (LV1
not only in mid‐level reservoirs, but also in the uppermost in Figure 5).
stratigraphic layers of the SC. 7.3.5. The Basal Thrust Zone: A Recently Activated
[39] Layer segments C‐type, such as Na layer at km 71– Mega‐Shear
72, do not show a velocity contrast with overlying rocks [42] The ∼3‐km finite displacement accumulated along
implying that Na layers are fluid‐depleted, possibly because T0 and T1 thrust faults requires an equivalent amount of

15 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

Figure 10

16 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

displacement to be taken up by basal shearing along the SC. reduction and cementation as supported by the relative high‐
Seismic imaging suggests that compressive strain essentially velocity layer Pa modeled along the TSC reflector (Figure 6).
localizes near the TOC with evidence for discrete shortening Our observations suggest however, that the abandonment of
within Ls sequence downdip from a depth of ∼6.5 km along the roof thrust might have been first to the benefit of a mid‐
the SC (Figures 4 and 8). The remarkable low‐velocity level décollement prior to down stepping to the basal thrust.
perturbations modeled along layer Nb and other layers 7.3.7. Relative Chronology of SC Deformation
below Nb, in particular at km 73–74 (Figure 7), 77–78 and [44] The scenario in Section 7.3.6 applies well to line SIS‐
80–81 (Figure 6), support that shearing distributes within 72, where step 1 (Figure 10b) would represent the con-
weak, high‐porosity, fluid rich layers of unit Ls and possibly struction of an initial accretionary wedge. The remainder of
fractured rocks of the uppermost section of the oceanic crust this wedge may have been preserved at the margin front
(Figure 9e). Hence, although shear had to localize once at (Figure 10a). During step 2 (Figure 10c), a weak roof thrust
the Us/Ls boundary to account for the growth of the Riedel controlled underthrusting of TS‐2 long sheet. Diffuse nor-
shears, and the initiation of T1 thrust fault (Figure 3), mal faulting deformed subducting sediment as they con-
deformation later stepped down and localized near the TOC, solidated under increasing lithostatic load. According to the
affecting sequence Ls. Fluid concentration near the base of analog model by Gutscher et al. [1996], the long sheet might
the SC is substantiated by drilling results of Costa Rica have cut the initial accretionary wedge along a mid‐level
ODP‐Leg 170 [Kimura et al., 1997], and results of New‐ detachment so that, during steps 2–3 (Figures 10c and 10d),
Hebrides ODP‐Leg134 [Collot et al., 1992] support basal wedge slivers were pushed downdip, and underplated, as
shearing. Consolidation tests conducted on Leg 170 cores Up1 duplexes. During step3, at depths of ∼6 km, Riedel
indicate that sediment near the base of the SC remains shears formed across the channel and initiated a mid‐level
essentially undrained and show pore pressure in excess of décollement. Then, at the end of step 4 (Figure 10e), when
3.1 MPa [Saffer et al., 2000]. Drilling the subduction zone the roof thrust was abandoned, the shear zone shifted
between the North d’Entrecasteaux Ridge and the New downward at the downdip end of the SC thus allowing a
Hebrides Island arc revealed that weathered, volcanic complex contractional pattern to build on. During step 5
breccias of the ridge basement [Coltorti et al., 1994] was (Figure 10f), shear localized in the wake of fault F along the
sheared off and incorporated as 10 to >60 m‐thick tectonic mid‐level décollement that propagated updip. At step 6
slivers into the accretionary wedge [Pelletier and Meschede, (Figure 10g), a shear zone formed within the lower unit of the
1994]. This result demonstrates that the décollement can long sheet, and shear finally localized near the top of the
peel off thin sections of altered volcanic basement that are oceanic crust. Implications of Step 5 and 6 are the initiation
underplated. of the new frontal accretionary wedge and the underplating
7.3.6. What Caused the Shift From the Roof of most of the long sheet. However, pervasive contraction
to the Basal Thrust? may carry on across the whole thickness of the SC near
[43] Scaled sand box experiments have shown that under fault F.
high basal friction at the base of the subducting sediment
unit, the growth of accretionary wedges display cycles
alternating between frontal imbricate thrusting and under- 8. Conclusions
thrusting of long, poorly deformed sheets [Gutscher et al., [45] Reprocessing line SIS‐72 across the southern Ecua-
1996]. Underthrusting of a long sheet is made possible dor margin provides evidence that the low‐Vp velocity SC
when less energy is required to sustain motion along the consists of an Upper Sequence Us interpreted as Plio‐
roof thrust than to initiate a new basal thrust. According to Pleistocene trench fill deposit overlying a lower sequence Ls
the mechanical analysis by [Gutscher et al., 1998], this inferred to be Oligo‐Miocene pelagic sediment. Our struc-
condition is fulfilled when 1) basal friction is high (i.e., 83% tural interpretation reveals that, as trench and Nazca plate
of the internal friction), 2) surface slope is shallow i.e., the sediment was conveyed into the SC, minor extension
arcward increase of the wedge overburden is small, and 3) deformed interplate sediment, prior to the formation of a
because there is less overburden on the roof thrust than on complex shear pattern, which migrated over time downward
the basal thrust. However, even with high basal friction, across the SC, thus forming a dynamic mega‐shear zone.
when the slope is too large, underthrusting is inhibited and a This pattern seemingly overprinted most of the original
new frontal thrust initiates [Gutscher et al., 1998]. Aban- trench sedimentary architecture, and likely controlled fluid
donment of the roof thrust may be favored by water content migration. The ensuing mega‐shear zone appears bounded

Figure 10. Schematic diagram showing the multistage evolution of the SC over a >450 Kyr period of time. (a) Interpreted
tectonic units according to the stages indicated in Figures 10b–10g. Multistage scenario: (b) initial accretion develops to lower
margin slope; (c) underthrusting of a long sheet along a roof thrust (red is active fault), margin slope increases, and a mid‐level
detachment cuts through the initial accretionary wedge [after Gutscher et al., 1996]; (d) underplated duplexes are emplaced
[after Gutscher et al., 1996], Riedel shears develop as a response to shearing along the long sheet; (e) underthrusting the long
sheet continues up to an end when contractional deformation associated to the basal thrust initiate; (f) the roof thrust and the
Riedel shears are abandoned, shear shifts downward and propagates updip from the basal thrust at the top of the oceanic crust to
along a mid‐level décollement; (g) development of a shear zone at the base of the channel, then mid‐level décollement thrust
shifts to top of oceanic crust. Décollement thrust emerges at the seafloor, frontal accretion resumes, and the long sheet is
underplated; horizontal black arrows with vertical dashed lines show the displacement of faulted‐block F relative to the long
sheet from Figure 10e; dotted pattern outlines area of complex internal deformation.

17 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

by a roof and a basal thrust, and locally divided along the Us/ de l’Univers (INSU), and Institut Français pour l’Exploitation de la Mer
(IFREMER), which provided ship time during the SISTEUR‐2000 experi-
Ls boundary by a mid‐level décollement fault that accom- ment. We are grateful to the scientific parties, captains, and crew for their
modates the formation of Riedel shears. support during data acquisition. We acknowledge the Associate Editor
[46] Modeling groups of strong reflectors in the upper SC Greg Moore as well as Casey Moore and Chris Morley for their thorough
and at its bottom demonstrates thin low‐Vp layers inferred reviews and very helpful comments, and we thank Stephane Operto for
wise advice during data processing.
to have a high‐porosity. These layers point to a potential
accumulation of high‐pressured fluid in the uppermost SC,
and highlight in particular the mechanical weakness of the References
basal thrust, which is characterized by the highest negative
Adam, J., D. Klaeschen, N. Kukowski, and E. R. Flueh (2004), Upward
velocity perturbations. Relatively high‐velocity perturba- delamination of Cascadia Basin sediment infill with landward frontal
tions modeled along the continuous roof thrust reflector accretion thrusting caused by rapid glacial age material flux, Tectonics,
support a reduced porosity and probable cementation. 23, TC3009, doi:10.1029/2002TC001475.
Al‐Yahya, K. (1989), Velocity analysis by iterative profile migration,
[47] The formation of Riedel shears across the SC may be Geophysics, 54, 718–729, doi:10.1190/1.1442699.
indicative of a specific form of basal erosion that allows Aoki, Y., H. Kinoshita, and H. Kagami (1985), Evidence of a low‐velocity
capturing thin margin rocks slivers into small‐scale SC layer beneath the accretionary prism of the Nankai Trough: Inferences
from a synthetic sonic log, Deep Sea Drill. Project Initial Rep., 87,
structural wedges below the roof thrust. A fine scale velocity 727–735.
model revealed two anomalously low Vp patches within the Avedik, F. (1993), Single bublle air‐gun array for deep exploration,
SC advocating internal zones of over‐pressured fluids. Geophysics, 58, 366–382.
According to the velocity model, the foundation of the upper Bangs, N., T. H. Shipley, S. P. S. Gulick, G. F. Moore, S. Kuromoto, and
Y. Nakamura (2004), Evolution of the Nankai trough décollement from
plate basement shows low Vp areas that may either indicate the trench into the seismogenic zone: Inferences from three‐dimensional
underplating or a damage zone possibly related to fluid seismic reflection imaging, Geology, 32, 273–276, doi:10.1130/
released from the SC. G20211.2.
Bangs, N., G. F. Moore, S. P. S. Gulick, E. Pangborn, T. Tobin, S. Kuramoto,
[48] A straightforward reconstruction of the temporal and A. Taira (2009), Broad weak regions of the Nankai megathrust and
evolution of the mega‐shear zone structures based on simple implications for shallow coseismic slip, Earth Planet. Sci. Lett., 284,
cinematic considerations support that frontal accretion, 44–49, doi:10.1016/j.epsl.2009.04.026.
underthrusting of long sheets of sediment, down‐stepping of Bonnardot, M.‐A., R. Hassani, E. Tric, E. Ruellan, and M. Régnier (2008),
Effect of margin curvature on plate deformation in a 3‐D numerical
the décollement and underplating are transient processes that model of subduction zones, Geophys. J. Int., 173, 1084–1094,
alternate in a cyclical manner through geological times. doi:10.1111/j.1365-246X.2008.03752.x.
During the ultimate cycle at the south Ecuador margin, total Byrne, D. E., D. M. Davis, and L. R. Sykes (1988), Loci and maximum size
of thrust earthquakes and the mechanics of the shallow region of subduc-
sediment subduction initiated more than 450 kyr ago, inter- tion zones, Tectonics, 7(4), 833–857, doi:10.1029/TC007i004p00833.
rupted ∼54 ± 13 kyr ago and was followed since then by Caine, J. S., J. P. Evans, and C. B. Forster (1996), Fault zone architecture
frontal accretion. Structural consequences can be as follows: and permeability structure, Geology, 24, 1025–1028, doi:10.1130/0091-
[49] 1. Shear strain likely concentrated along a skinny SC 7613(1996)024<1025:FZAAPS>2.3.CO;2.
Calahorrano, A. (2005), Structure de la marge du Golfe de Guayaquil
associated with the roof thrust during underthrusting of the (Equateur) et propriétés physiques du chenal de subduction a partir de
long sheet of sediment, although small throw normal fault données de sismique marine réflexion et réfraction, Ph.D. thesis, 221 pp.,
affected it at shallow burial depths. Subsequently, shear Université Pierre et Marie Curie, Paris.
Calahorrano, A., V. Sallares, J.‐Y. Collot, F. Sage, and C. R. Ranero
strain affected gradually the entire subducting sedimentary (2008), Nonlinear variations of the physical properties along the southern
pile: first through Riedel shears and a mid‐level décollement Ecuador subduction channel: Results from depth‐migrated seismic data,
fault at middle burial depths, and then through folds and Earth Planet. Sci. Lett., 267, 453–467, doi:10.1016/j.epsl.2007.11.061.
Campbell, C. J. (1974), Ecuadorian Andes, in Mesozoic Cenozoic Orogenic
thrust faults, which initiated at greater burial depths, possi- Belts: Data for Orogenic Studies, edited by A. M. Spencer, Geol. Soc.
bly when the roof thrust was about to be abandoned. Con- Spec. Publ., 4, 725–732.
tractional structures overprinted earlier structures in the Cloos, M., and R. L. Shreve (1988a), Subduction‐channel model of prism
deepest arcward segment of the channel and propagated up accretion, melange formation, sediment subduction, and subduction ero-
sion at convergent plate margins: 1. Background and description, Pure
dip in the form of a basal shear zone that thinned out sea- Appl. Geophys., 128(3–4), 455–500, doi:10.1007/BF00874548.
ward. Ultimately, the basal thrust reached the trench seafloor Cloos, M., and R. L. Shreve (1988b), Subduction‐channel model of prism
and frontal accretion resumed. accretion, melange formation, sediment subduction, and subduction ero-
sion at convergent plate margins: 2. Implications and discussion, Pure
[50] 2. The SC boundaries revealed highly unstable Appl. Geophys., 128(3–4), 501–545, doi:10.1007/BF00874549.
through time: (a) Rock slivers were likely ripped of the Collot, J.‐Y., et al. (1992), ODP Leg 134, Ocean Drill. Program, College
foundation of an initial accretionary wedge, pushed ahead of Station, Tex.
the underthrusting long sheet prior to be underplated as Collot, J.‐Y., K. B. Lewis, G. Lamarche, and S. Lallemand (2001), The
giant Ruatoria debris avalanche on the northern Hikurangi margin,
duplexes; (b) Most of the underthrust long sheet underplated New Zealand: Result of oblique seamount subduction, J. Geophys.
when the interplate shear zone down‐stepped from the roof Res., 106, 19,271–19,297, doi:10.1029/2001JB900004.
to the basal thrust, so that the active SC has now been Collot, J.‐Y., P. Charvis, M. A. Gutscher, and S. Operto (2002), Exploring
the Ecuador‐Colombia active margin and inter‐plate seismogenic zone,
reduced to the basal shear zone, although shearing may still Eos Trans. AGU, 83(17), 189–190, doi:10.1029/2002EO000120.
distribute across the bulk of the downdip segment of the SC Collot, J.‐Y., W. Agudelo, A. Ribodetti, and B. Marcaillou (2008), Origin of
landward of fault F. (c) Flakes of the uppermost subducting a crustal splay fault and its relation to the seismogenic zone and underplat-
ing at the erosional N‐Ecuador S‐Colombia oceanic margin, J. Geophys.
oceanic crust are possibly peeled of and added to the SC Res., 113, B12102, doi:10.1029/2008JB005691.
from below, along the active basal thrust. Coltorti, M., P. E. Baker, L. Briqueu, T. Hasenaka, and B. Galassi (1994),
Petrology and geochemistry of volcanic rocks from the New Hebrides
forearc region, Sites 827, 829 and 830, Proc. Ocean Drill. Program
[51] Acknowledgments. This work was funded by the Institut de Sci. Results, 134, 337–352.
Recherche pour le Développement (IRD), the Institut National des Sciences Davis, G. H., and S. J. Reynolds (1996), Structural Geology of Rocks and
Regions, John Wiley, New York.

18 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

Deniaud, Y., P. Baby, C. Basile, M. Ordonez, G. Montenegro, and G. Mascle Mendoza, C., and J. W. Dewey (1984), Seismicity associated with the great
(1999), Ouverture et évolution du Golfe de Guayaquil: Bassin d’avant arc Colombia‐Ecuador earthquakes of 1942, 1958 and 1979: Implications for
néogène et quaternaire du sud des Andes équatoriennes, C. R. Acad. Sci., barrier models of earthquake rupture, Bull. Seismol. Soc. Am., 74(2),
328(3), 181–187. 577–593.
Ego, F., M. Sébrier, A. Lavenu, H. Yepes, and A. Egues (1996), Quater- Michaud, F., J‐Y Collot, A. Alvarado, E. López, and y el personal cientí-
nary state of stress in the Northern Andes and the restraining bend model fico y técnico del INOCAR (2006), República del Ecuador, Batimetría y
for the Ecuadorian Andes, Tectonophysics, 259, 101–116, doi:10.1016/ Relieve Continental, IOA‐CVM‐02‐Post, Inst. Oceanogr. de la Armada
0040-1951(95)00075-5. del Ecuador, Guayaquil.
England, P. C., and T. J. B. Holland (1979), Archimedes and the Tauern Mix, A. C., R. Tiedemann, and P. Blum (2003), Proceedings of the Ocean
eclogites: The role of buoyancy in the preservation of exotic eclogite Drilling Program, Initial Reports [CD‐ROM], vol. 202, Ocean Drill.
blocks, Earth Planet. Sci. Lett., 44, 287–294, doi:10.1016/0012-821X Program, College Station, Tex.
(79)90177-8. Moore, J. C. (1989), Tectonics and hydrogeology of accretionary prisms:
Erickson, S. N., and R. D. Jarrard (1998), Velocity‐porosity relationships Role of the décollement zone, J. Struct. Geol., 11, 95–106,
for water‐saturated siliciclastic sediments, J. Geophys. Res., 103, doi:10.1016/0191-8141(89)90037-0.
30,385–30,406, doi:10.1029/98JB02128. Moore, J. C., and D. Saffer (2001), Up‐dip limit of the seismogneic zone
Gutscher, M.‐A., N. Kukowski, J. Malavieille, and S. Lallemand (1996), beneath the accretionary prism of southwest Japan: An effect of diage-
Cyclical behavior of thrust wedges: Insights from high basal friction netic to low‐grade metamorphic processes and increasing effective stress,
sandbox experiments, Geology, 24, 135–138, doi:10.1130/0091-7613 Geology, 29, 183–186, doi:10.1130/0091-7613(2001)029<0183:
(1996)024<0135:CBOTWI>2.3.CO;2. ULOTSZ>2.0.CO;2.
Gutscher, M.‐A., N. Kukowski, J. Malavieille, and S. Lallemand (1998), Moore, J. C., A. Klaus, N. Bangs, and B. Bekins (1998), Proceedings of the
Episodic imbricate thrusting and underthrusting: Analog experiments Ocean Drilling Program, Initial Reports, vol. 171A, Ocean Drill. Pro-
and mechanical analysis applied to the Alaskan accretionary wedge, gram, College Station, Tex.
J. Geophys. Res., 103, 10,161–10,176, doi:10.1029/97JB03541. Moore, J. C., C. Rowe, and F. Meneghini (2007), How accretionary prisms
Hubbert, M. K., and W. W. Rubey (1959), Role of fluid pressure in elucidate seismogenesis in subduction zones, in The Seismogenic Zone of Sub-
mechanics of overthrust faulting, Geol. Soc. Am. Bull., 70, 115–166, duction Thrust Faults, edited by T. Dixon and J. C. Moore, pp. 288–325,
doi:10.1130/0016-7606(1959)70[115:ROFPIM]2.0.CO;2. Columbia Univ. Press, New York.
Hyndman, R. D., and K. Wang (1995), The rupture zone of the Casca- Morgan, J. K., and D. E. Karig (1995), Décollement processes at the Nankai
dia great earthquakes from current deformation and the thermal accretionary margin, southeast Japan: Propagation, deformation, and
regime, J. Geophys. Res., 100, 22,133–22,154, doi:10.1029/95JB01970. dewatering, J. Geophys. Res., 100, 15,221–15,231.
Hyndman, R. D., M. Yamano, and D. A. Oleskevich (1997), The seismo- Nelson, C. H. (1983), The Astoria fan: An elongate type fan, Geo Mar.
genic zone of subduction thrust faults, Isl. Arc, 6, 244–260, doi:10.1111/ Lett., 3, 65–70, doi:10.1007/BF02462449.
j.1440-1738.1997.tb00175.x. Nocquet, J. M., P. Mothes, J. Villegas Lanza, M. Chlieh, P. Jarrin, M. Vallée,
Jackson, C., M. Sen, and P. Stoffa (2004), An efficient stochastic Bayesian H. Tavera, G. Ruiz, M. Regnier, and F. Rolandone (2010), New GPS
approach to optimal parameter and uncertainty estimation for climate velocity field in the northern Andes (Peru ‐ Ecuador ‐ Colombia): Hetero-
model predictions, J. Clim., 17, 2828–2841, doi:10.1175/1520-0442 geneous locking along the subduction, northeastwards motion of the
(2004)017<2828:AESBAT>2.0.CO;2. Northern Andes, Abstract G41C‐04 presented at 2010 Fall Meeting,
Kanamori, H. (1986), Rupture process of subduction zone earthquakes, AGU, San Francisco, Calif., 13–17 Dec.
Annu. Rev. Earth Planet. Sci., 14, 293–322, doi:10.1146/annurev. Operto, S., S. Xu, and G. Lambaré (2000), Can we quantitatively image
ea.14.050186.001453. complex structures with rays?, Geophysics, 65, 1223–1238, doi:10.1190/
Kelleher, J. (1972), Rupture zones of large South American earthquakes 1.1444814.
and some predictions, J. Geophys. Res., 77, 2087–2103, doi:10.1029/ Pacheco, J. F., L. R. Sykes, and C. H. Scholz (1993), Nature of seismic cou-
JB077i011p02087. pling along simple plate boundaries of the subduction type, J. Geophys.
Kendrick, E., M. Bevis, R. Smalley Jr., B. Brooks, R. B. Vargas, E. Lauria, Res., 98, 14,133–14,159, doi:10.1029/93JB00349.
and L. P. Souto Fortes (2003), The Nazca‐South America Euler vector Park, J.‐O., T. Tsuru, N. Takahashi, T. Hori, S. Kodaira, A. Nakanishi,
and its rate of change, J. S. Am. Earth Sci., 16(2), 125–131, S. Miura, and Y. Kaneda (2002), A deep strong reflector in the Nankai
doi:10.1016/S0895-9811(03)00028-2. accretionary wedge from multichannel seismic data: Implications for
Kimura, G., E. A. Silver, and P. Blum (1997), Leg 170, Ocean Drill. underplating and interseismic shear stress release, J. Geophys. Res.,
Program, College Station, Tex. 107(B4), 2061, doi:10.1029/2001JB000262.
Kitamura, Y., et al. (2005), Melange and its seismogenic roof décollement: Pelletier, B., and M. Meschede (1994), Structural style of the accretionary
A plate boundary fault rock in subduction zone: An example from the wedge in front of the North d’Entrecasteaux Ridge, Proc. Ocean Drill.
Shimanto Belt, Japan, Tectonics, 24, TC5012, doi:10.1029/ Program Sci. Results, 134, 417–429.
2004TC001635. Pratt, G., C. Shin, and G. J. Hicks (1998), Gauss‐Newton and full Newton
Knipe, R. J. (1993), The influence of fault zone processes and diaganesis on methods in frequency‐space seismic waveform inversion, Geophys. J. Int.,
fluid flow, in Diagenesis and Basin Development, Studies in Geology edi- 133, 341–362, doi:10.1046/j.1365-246X.1998.00498.x.
ted by A. D. R. Horbury and A. G. Robinson, pp. 135–154, Am. Assoc. of Ranero, C. R., J. Phipps Morgan, K. D. McIntosh, and C. Reichert (2003),
Pet. Geol., Tulsa, Okla. Bending, faulting, and mantle serpentinization at the Middle America
Le Pichon, X., P. Henry, and S. Lallemant (1993), Accretion and erosion in Trench, Nature, 425, 367–373, doi:10.1038/nature01961.
subduction zones: The role of fluids, Annu. Rev. Earth Planet. Sci., 21, Ranero, C. R., I. Grevemeyer, H. Sahling, U. Barckhausen, C. Hensen,
307–331, doi:10.1146/annurev.ea.21.050193.001515. K. Wallmann, W. Weinrebe, P. Vannucchi, R. von Huene, and K. D.
Logan, J. M., C. A. Dengo, N. G. Higgs, and Z. Z. Wang (1992), Fabrics of McIntosh (2008), Hydrogeological system of erosional convergent mar-
experimental fault zone: Their development and relationship to mechni- gins and its influence on tectonics and interplate seismogenesis, Geochem.
cal behaviour, in Faut Mechanics and Transport Properties of Rocks, Geophys. Geosyst., 9, Q03S04, doi:10.1029/2007GC001679.
edited by B. Evans and T. Wong, pp. 33–67, Academic, San Diego, Ratzov, G., J.‐Y. Collot, M. Sosson, and S. Migeon (2010), Mass‐transport
Calif., doi:10.1016/S0074-6142(08)62814-4. deposits in the northern Ecuador subduction trench: Result of frontal ero-
Lonsdale, P. (1978), Ecuadorian subduction system, Am. Assoc. Pet. Geol. sion over multiple seismic cycles, Earth Planet. Sci. Lett., 296, 89–102,
Bull., 62(12), 2454–2477. doi:10.1016/j.epsl.2010.04.048.
Lonsdale, P. (2005), Creation of the Cocos and Nazca plates by fission of Ribodetti, A., S. Operto, W. Agudelo, J.‐Y. Collot, and J. Virieux (2011),
the Farallon plate, Tectonophysics, 404, 237–264, doi:10.1016/j.tecto. Joint ray + born least‐squares migration and simulated annealing optimi-
2005.05.011. zation for target‐oriented quantitative seismic imaging, Geophysics, 76,
Maltman, A., and P. Vannucchi (2004), Insights from the Ocean Drilling B1–B20, doi:10.1190/1.3554330.
Program on shear and fluid‐flow at the mega‐faults between actively Saffer, D. (2003), Pore pressure development and progressive dewatering
converging plates, in Flow Processes in Faults and Shear Zones, edited in underthrust sediments at the Costa Rica subduction margin: Compar-
by G. I. Alsop et al., pp. 127–140, Geol. Soc., London. ison with northern Barbados and Nankai, J. Geophys. Res., 108(B5),
Mancktelow, N. S. (1995), Nonlithostatic pressure during sediment subduc- 2261, doi:10.1029/2002JB001787.
tion and the developement and exhumation of high pressure metamorphic Saffer, D., and C. Marone (2003), Comparaison of smectite‐ and illite‐rich
rocks, J. Geophys. Res., 100, 571–583, doi:10.1029/94JB02158. gouge frictional properties: Application to the updip limit of the seismo-
McAdoo, B. G., L. F. Pratson, and D. L. Orange (2000), Submarine land- genic zone along subduction megathrust, Earth Planet. Sci. Lett., 215,
slide geomorphology, U.S. continental slope, Mar. Geol., 169, 103–136, 219–235, doi:10.1016/S0012-821X(03)00424-2.
doi:10.1016/S0025-3227(00)00050-5.

19 of 20
B11102 COLLOT ET AL.: ECUADOR DYNAMIC SUBDUCTION CHANNEL B11102

Saffer, D., E. A. Silver, A. T. Fisher, H. Tobin, and K. Moran (2000), margin: Implication of interplate coupling, J. Geophys. Res., 107(B12),
Inferred pore pressures at the Costa Rica subduction zone: Implications 2357, doi:10.1029/2001JB001664.
for dewatering processes, Earth Planet. Sci. Lett., 177, 193–207, Ujiie, K., T. Hisamitsu, and A. Taira (2003), Deformation and fluid pres-
doi:10.1016/S0012-821X(00)00048-0. sure variation during initiation and evolution of the plate boundary décol-
Sage, F., J.‐Y. Collot, and C. R. Ranero (2006), Interplate patchiness and lement zone in the Nankai accrtionary prism, J. Geophys. Res., 108(B8),
subduction‐erosion mechanisms: Evidence from depth migrated seis- 2398, doi:10.1029/2002JB002314.
mic images at the central Ecuador convergent margin, Geology, 34, Vannucchi, P., and H. Tobin (2000), Deformation structures and implica-
997–1000, doi:10.1130/G22790A.1. tions for fluid flow at the Costa Rica convergent margin, ODP Sites
Scholz, C. H. (1998), Earthquakes and friction laws, Nature, 391, 37–42, 1040 and 1043, Leg 170, J. Struct. Geol., 22, 1087–1103, doi:10.1016/
doi:10.1038/34097. S0191-8141(00)00027-4.
Screaton, E. (2006), Excess pore pressure within subducting sediments: Vannucchi, P., F. Remitti, and G. Bettelli (2008), Geological record of fluid
Does the proportion of accreted sediment versus subducted sediment flow and seismogenesis along an erosive subducting plate boundary,
matter?, Geophys. Res. Lett., 33, L10304, doi:10.1029/2006GL025737. Nature, 451, 699–703, doi:10.1038/nature06486.
Sen, M. K., and P. L. Stoffa (1995), Global Optimization Methods in Geo- Vannucchi, P., J. Phipps Morgan, F. Sage, J.‐Y. Collot, and F. Remitti
physical Inversion, Elsevier Sci., New York. (2011), The upper subduction channel from subduction to seismogen-
Shipley, T. H., K. D. McIntosh, E. A. Silver, and P. L. Stoffa (1992), esis, paper presented at Nature of the plate interface in Subduction
Three‐dimensional seismic imaging of the Costa Rica accretionary prism: Zones, W. Alps: Task Force IX Workshop, Int. Lithosphere Program,
Structural diversity in a small volume of the lower slope, J. Geophys. Piemonte, Italy, 2–7 July.
Res., 97, 4439–4459, doi:10.1029/91JB02999. von Huene, R., D. Klaeschen, and B. Cropp (1994), Tectonic structure
Shreve, R. L., and M. Cloos (1986), Dynamics of sediment subduction, across the accretionary and erosional parts of the Japan trench margin,
melange formation, and prism accretion, J. Geophys. Res., 91, J. Geophys. Res., 99, 22,349–22,361, doi:10.1029/94JB01198.
10,229–10,245. von Huene, R., C. R. Ranero, and P. Vannucchi (2004), Generic model of
Silgado, E. (1957), El movimiento sismico del 12 de deciembre de 1953, subduction erosion, Geology, 32, 913–916, doi:10.1130/G20563.1.
Bol. Soc. Geol. Peru, 32, 225–238. Vrolijk, P. (1990), On the mechanical role of smectite in subduction
Swenson, J. L., and S. L. Beck (1996), Historical 1942 Ecuador and 1942 zones, Geology, 18, 703–707, doi:10.1130/0091-7613(1990)018<0703:
Peru subduction earthquakes, and earthquake cycles along Colombia‐ OTMROS>2.3.CO;2.
Ecuador and Peru subduction segments, Pure Appl. Geophys., 146(1), Wang, K., and Y. Hu (2006), Accretionary prisms in subduction earthquake
67–101, doi:10.1007/BF00876670. cycles: The theory of dynamic Coulomb wedge, J. Geophys. Res., 111,
Taira, A., J. Hill, J. Firth, U. Brener, W. Bruckmann, et al. (1992), Sedi- B06410, doi:10.1029/2005JB004094.
ment deformation and hydrology of the Nankai Trough accretionary Warner, M. (1990), Absolute reflection coefficients from deep seismic
prism: Synthesis of shipboard results of ODP Leg 131, Earth Planet. reflections, Tectonophysics, 173, 15–23, doi:10.1016/0040-1951(90)
Sci. Lett., 109, 431–450, doi:10.1016/0012-821X(92)90104-4. 90199-I.
Thierry, P., S. Operto, and G. Lambaré (1999), Fast 2‐D ray + Born inver- Winter, T., J. Avouac, and A. Lavenu (1993), Late Quaternary kinematics
sion/migration in complex media, Geophysics, 64, 162–181, of the Pallatanga strike slip fault (central Ecuador) from topographic
doi:10.1190/1.1444513. measurements of displaced morphological features, Geophys. J. Int.,
Tichelaar, B. W., and L. J. Ruff (1993), Depth of seismic coupling among 115, 905–920, doi:10.1111/j.1365-246X.1993.tb01500.x.
subduction zones, J. Geophys. Res., 98, 2017–2037, doi:10.1029/ Witt, C., J. Bourgois, F. Michaud, M. Ordoñez, N. Jimenez, and M. Sosson
92JB02045. (2006), Development of the Gulf of Guayaquil (Ecuador) during the Qua-
Tobin, H., and D. Saffer (2009), Elevated fluid pressure and extreme ternary as an effect of the North Andean block tectonic escape, Tectonics,
mechanical weakness of a plate boundary thrust, Nankai Trough subduc- 25, TC3017, doi:10.1029/2004TC001723.
tion zone, Geology, 37, 679–682, doi:10.1130/G25752A.1.
Tobin, H., P. Vannucchi, and M. Meschede (2001), Structure, inferred W. Agudelo, ECOPETROL, Instituto Colombiano del Petróleo, Km 7
mechanical properties, and implications for fluid transport in the décolle- Via Piedecuesta, Piedecuesta, Santander 007, Colombia.
ment zone, Costa Rica convergent margin, Geology, 29, 907–910, J.‐Y. Collot and A. Ribodetti, Université de Nice Sophia‐Antipolis,
doi:10.1130/0091-7613(2001)029<0907:SIMPAI>2.0.CO;2.
Institut de Recherche pour le Développement (UR082), Observatoire de la
Trenkamp, R., J. N. Kellogg, J. T. Freymueller, and H. P. Mora (2002), Côté d’Azur, Géoazur, La Darse B.P. 48, F‐06235 Villefranche‐sur‐Mer
Wide plate margin deformation, southern Central America and north- Cedex, France. (collot@geoazur.obs‐vlfr.fr)
western South America, CASA GPS observations, J. S. Am. Earth Sci.,
F. Sage, Université de Nice Sophia‐Antipolis, Université Pierre et Marie
15, 157–171, doi:10.1016/S0895-9811(02)00018-4. Curie, Observatoire de la Côte d’Azur, Géoazur, La Darse B.P. 48 06235
Tsuru, T., J. Park, S. Miura, S. Kodaira, Y. Kido, and T. Hayashi (2002), Villefranche‐sur‐Mer Cedex, France.
Along‐arc structural variation of the plate boundary at the Japan Trench

20 of 20