Earth and Planetary Science Letters 433 (2016) 146–158

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

Seamount subduction at the North-Ecuadorian convergent margin:

Effects on structures, inter-seismic coupling and seismogenesis
Boris Marcaillou a,b,∗ , Jean-Yves Collot c,d,1 , Alessandra Ribodetti c , Elia d’Acremont e ,
Ammy-Adoum Mahamat b , Alexandra Alvarado d
a
Geoazur, IRD-UNSA-OCA-CNRS, 101 Promenade R. Laroque, 98800 Nouméa, New Caledonia
b
UMR 5243 Geosciences–Montpellier Université des Antilles–Université de Montpellier–CNRS, Département de Géologie, Campus de Fouillole, 97759 Pointe a
Pitre, Guadeloupe
c
Geoazur, IRD-UNSA-OCA-CNRS, 250 Rue A. Einstein, 06560 Valbonne, France
d
Instituto Geofísico, Escuela Politécnica Nacional, Ladrón de Guevara, E11-253, Ap. 2759, Quito, Ecuador
e
Université Pierre et Marie Curie (UPMC), Sorbonne Universités Paris 06, Institut des Sciences de la Terre de Paris (iSTeP), CNRS, UMR 7193, F-75005 Paris, France

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

Article history: At the North-Ecuadorian convergent margin (1◦ S–1.5◦ N), the subduction of the rough Nazca oceanic plate
Received 10 July 2015 leads to tectonic erosion of the upper plate and complex seismogenic behavior of the megathrust. We
Received in revised form 23 October 2015 used three selected pre-stack depth migrated, multi-channel seismic reflection lines collected during the
Accepted 25 October 2015
SISTEUR cruise to investigate the margin structure and decipher the impact of the subducted Atacames
Available online 9 November 2015
seamounts on tectonic erosion, interseismic coupling, and seismogenesis in the region of the 1942 Mw7.8
Editor: P. Shearer
earthquake.
Keywords: This dataset highlights a subducted ∼30 × 40 km, double-peak seamount that belongs to the Atacames
subduction zone structure seamount chain and that is associated with a deep morphologic re-entrant containing mass transport
convergent margin tectonic deformation deposits.
subducted seamount The seamount subduction uplifted the margin basement by ∼1.6 km and pervasively broke the margin by
seismogenic zone deep and intense reverse faulting ahead of the seamount, a process that is likely to weaken considerably
subduction earthquakes
the margin. In the seamount wake, the basement reverse fault system rotated counter-clockwise. This
Ecuador margin
faulted basement is overlain with slope sediment sliding along listric normal faults that sole out onto
the BSR. This superposition of deep tectonic contraction within the basement and shallow gravitational
extension deformation within the sediment highlights the key role of gas hydrate on outer slope erosion.
In addition to long-term regional basal erosion, the margin basement has thinned locally by an
extra 0.8–1 km in response to the subduction of the Atacames seamount chain and hydrofracturing
by overpressured fluids at the margin toe. This pervasively and deeply fractured margin segment is
associated with a seismically quiet and GPS-modeled low interseismic coupling corridor that terminates
downdip near the 1942 epicenter and locked zone. We suggest that the deeply buried double-peak
Atacames seamount triggered the 1942 earthquake ahead of its leading flank. This result supports
previous studies proposing that subducted seamounts provide unfavorable conditions for locking the
updip segment of the plate boundary limiting the updip extent of seismogenic zones, but may favor
large subduction earthquakes at greater depths.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction act as seismic asperities capable of triggering great subduction

earthquakes (Cloos, 1992; Scholz and Small, 1997) or as seis-
Seamount subduction impacts the seismic activity at conver- mic barriers that inhibit seismic rupture propagation (Aki, 1979;
gent margins. Models and field studies propose that seamounts Kodaira et al., 2000) by increasing the normal stress to the plate in-
terface. In contrast, other studies argue that subducted seamounts
create a complex stress and structure environment (Wang and
*
Corresponding author at: Université de Nice-Sophia-Antipolis, UMR 7329
Bilek, 2011) by activating or reactivating fracture networks in the
Geoazur, 250 Rue Albert Einstein, 06560 Valbonne, France.
E-mail address: boris.marcaillou@unice.fr (B. Marcaillou). upper plate (Collot et al., 1992; Dominguez et al., 1998) so that
1
Investigator of the Prometeo Project of the Ministry of Higher Education, Sci- seamounts would subduct predominantly aseismically, producing
ence, Technology and Innovation of the Republic of Ecuador. only small and moderate-size earthquakes. Moreover, upward ex-

http://dx.doi.org/10.1016/j.epsl.2015.10.043
0012-821X/© 2015 Elsevier B.V. All rights reserved.
 B. Marcaillou et al. / Earth and Planetary Science Letters 433 (2016) 146–158 147

pulsion of overpressured fluids from within the fractured sub- ing the deformation front (Fig. 1A). The re-entrants formed long
ducting crust and fluid-rich sediments in the subduction channel before 25 kyr, the age of the oldest hemipelagic sediment recov-
around subducted seamounts tend to decrease the interplate seis- ered in the sediment pile deposited in re-entrant R1 (Ratzov et al.,
mic coupling (Moreno et al., 2014). Mochizuki et al. (2008) re- 2010). In contrast, a series of 5-km-wide indentations forms the
ported that a sequence of Mw7.0 earthquakes occurred down-dip 20-km-wide Atacames re-entrant and steep scarp S, incises the de-
of a subducted seamount since the 1920s, while the seamount re- formation front and is associated with a mass transport unit whose
mained aseismic. The authors suggested that the seamount kept youngest deposits are ∼3040 ± 60 yr old (Ratzov et al., 2010).
creeping during this time period to allow renewing stress for the Steep scarp S is bounded landward by shallow spur P between
Mw7.0 earthquakes to produce. In addition, numerical models by re-entrants R1 and R2.
Yang et al. (2012) show that under specific conditions a subducted
seamount may generate or stop megathrust ruptures. These mod- 2.2. Seismicity and subduction earthquakes
els and observations raise the question of the role of subducted
seamounts onto interplate seismic coupling and large earthquake The 1942 Mw7.8 subduction earthquake ruptured an interplate
generation. area of 100 × 200 km that extended northward to the Esmeral-
Subduction of topographic-highs (e.g. Dominguez et al., 1998; das Platform and southward to the area of the Carnegie ridge
Hampel et al., 2004) and hydrofracturing by overpressured fluids crest subduction (Kanamori and McNally, 1982; Swenson and Beck,
(e.g. von Huene et al., 2004) are frequently invoked as major causes
1996) (Fig. 1B). The rupture zone purportedly extended as far west
for subduction erosion and deformation that affect more than 50%
as the trench but no tsunami was reported suggesting a small or
of worldwide convergent margins (e.g. Clift and Vannucchi, 2004).
insignificant seafloor deformation near the trench (Swenson and
At long-term erosive margins, topographic-high subduction leads
Beck, 1996). Most 1942 aftershocks (white circles in Fig. 1B) were
to short-lived indentation, steepening of the outer slope, a tempo-
located in the trench or beneath the continental shelf and very
ral sequence of fore-arc uplift and subsidence (Collot et al., 1992;
few beneath the margin slope (Mendoza and Dewey, 1984). The
Hampel et al., 2004), pervasive fracturing (Dominguez et al., 1998)
CMT Harvard catalog indicates that the earthquakes with thrust
and moderate basal erosion of the margin basement (Bangs et al.,
focal mechanism (red beach-balls in Fig. 1B) are located at depth
2006; Ranero and von Huene, 2000). The results of basal erosion
beneath the continental shelf. 33 additional focal mechanisms
have rarely been imaged with seismic data because of the poor
(blue beach-balls in Fig. 1B) of well-located crustal earthquakes
penetration of seismic energy through highly attenuating acoustic
(Mw > 4) recorded since 1988 by the local Ecuador seismic net-
basement.
work (RENSIG) indicate a more complex distribution of rupture
Using three multi-channel seismic reflection lines from the SIS-
types including normal faulting, oblique thrusting and strike-slip
TEUR cruise (Collot et al., 2002) migrated to depth, we investigate
deformation (Alvarado, 2012).
the structural pattern of a segment of the erosive (Sage et al.,
The micro-seismicity from 1994 to 2007 (colored circles in
2006) and seismogenic North-Ecuadorian (1◦ S–1.5◦ N) subduction
Fig. 1B) is heterogeneously distributed (Font et al., 2013). The shal-
zone (Fig. 1) that is undergoing seamount subduction, adjacent to
low Galera and Jama clusters (Fig. 1C and E) extend trenchward
the epicenter of the 1942 Mw7.8 subduction earthquake. We an-
to 18–20 km from the deformation front and are interpreted as
alyze the influence of this seamount onto the subduction erosion,
interplate or upper plate events (Font et al., 2013). In contrast,
interseismic locking and coseismic slip.
within the study area in between, micro-seismicity is markedly ab-
sent beneath the outer-margin wedge and diffusely distributed at
2. Regional settings
greater depth, near and landward of the 1942 epicenter (Fig. 1D).
Font et al. (2013) indicate that earthquake locations are based on a
2.1. Structure and tectonic pattern
3D georealistic P-velocity model that integrates all published infor-
mation on structures and velocities along the Ecuadorean margin,
The Neogene Nazca plate underthrusts the Ecuadorian fore-
including the results from several OBS experiments. They conclude
arc with a convergence vector of 4.7 cm/yr in a N83◦ E direc-
that the average location uncertainties dx, dy and dz are ±2.1 km,
tion (Nocquet et al., 2014) (Fig. 1A). The Carnegie ridge, up
to 2-km-high, shows a maximum crustal thickness of 14 km
±1.9 km and ±1.4 km. This excellent location accuracy clearly sup-
ports the seismic quiescence of the study area.
(Graindorge et al., 2004) to 19 km (Sallares et al., 2005) at
the ridge apex, 1◦ S–1◦ 20 S, decreasing northward to less-than-
10 km at 0◦ (Sallares et al., 2005). The Atacames seamounts chain, 3. Seismic data and method
0◦ 25 –0◦ 55 N, which stands immediately north of the Carnegie
ridge, is a N–S-trending line of four 1 to 1.5-km-high seamounts. During the SISTEUR cruise (Collot et al., 2002) we recorded
The southernmost Atacames seamount is located in the trench, multi-channel seismic data with a 45-fold coverage, using a
facing the Atacames re-entrant in the margin (Collot et al., 2005). 360-channel, 4.5-km-long hydrophone streamer and a source
N20◦ -trending, landward-facing, bending fault scarps in the trench formed of 45-L air gun tuned in a single bubble mode. Seismic
outer wall bound narrow and thin trench basins, which contain pre-processing performed with Geocluster includes time variant
0.5 km of turbidite/hemipelagite fill overlying ∼0.5 km of Carnegie band-pass filters, minimum phase conversion of the signal, exter-
ridge draping (Ratzov et al., 2010). Multi-kilometer scale depres- nal and internal mutes, multiple attenuation in the frequency-wave
sions, up to 400-m-deep, are widely distributed along the mid- number (FK) domain and deconvolution. The pre-processed data,
slope of the northern and southern flanks of the ridge (Michaud sorted in Common-Depth Point, is stacked and time migrated us-
et al., 2005). These morphologic heterogeneities together with the ing a Kirchoff migration. The pre-processed data are migrated
thin trench fill roughen the surface of the oceanic plate. to depth with a preserved amplitude Pre-Stack Depth Migra-
The 15 to 35-km-wide margin outer slope is indented by gul- tion (PSDM) approach (Lambaré et al., 2003; Operto et al., 2003;
lies, scarps and re-entrants (Collot et al., 2009) and fronted by Thierry et al., 1999) performed in the angle domain. The accuracy
mass transport deposits reworked in frontal prisms indicating a of the migrated image is obtained by iterative correction of the
non-accretionary and/or erosive frontal tectonic regime (Collot et background velocity model (Al-Yahya, 1989) to avoid problems as-
al., 2002). 40-km-wide semi-circular re-entrants R1, R2 and R3 sep- sociated with misfocusing and under/over estimation of amplitude
arated by shallow spurs impinge the upper margin without affect- of seismic reflectors.
148 B. Marcaillou et al. / Earth and Planetary Science Letters 433 (2016) 146–158

Fig. 1. (A) Bathymetry of the North-Ecuadorian margin and the incoming Nazca plate acquired during the AMADEUS cruise (Collot et al., 2005). Plain lines indicate the location
of seismic lines SIS22, SIS54 and SIS55. Dashed lines delineate the deformation front, re-entrant R1, R2, R3 and Scarp S of the Atacames re-entrant. The dotted line associated
with line SIS55 is parallel to the deformation front and shows the slight seismic line obliquity to the margin. NAB: North-Andean Block; S: Scarp of the Atacames re-entrant;
P: Spur. Figure modified from Collot et al. (2009). (B) Location of the epicenters (black stars) with focal mechanism and rupture zones (dashed ellipses) of 1906 (Kanamori
and McNally, 1982), 1942 (Swenson and Beck, 1996) and 1958 (Beck and Ruff, 1984) subduction earthquakes. The open circles are the 1942 three-month aftershocks
(Mendoza and Dewey, 1984). The blue “beach-balls” are focal mechanisms for Mw > 4 and depth <40 km earthquakes from the 1988–2013 RENSIG catalog (Alvarado, 2012;
Beauval et al., 2013). The red “beach-balls” indicate the Centroid-Moment Tensor (CMT) from the Harvard University seismicity catalog (Ekström et al., 2012). The colored
circles show the Mw < 4 micro-seismicity (1994–2007) distribution (Font et al., 2013). The colors represent the focal depth according to the color palette. This seismicity
distribution highlights two shallow clusters of upper plate or interplate events, named Galera and Jama, located to the north (C) and the south (E) of the study area
respectively. In contrast, along the study transect (D) the Mw < 4 micro-seismicity and the earthquakes from the RENSIG and the CMT catalogs are mainly located at greater
depth near and landward of the 1942 epicenter. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
 B. Marcaillou et al. / Earth and Planetary Science Letters 433 (2016) 146–158 149

Fig. 2. Post-stack time-migrated multi-channel seismic reflection line SIS22 (location in Fig. 1). The oceanic Nazca plate that underthrusts the North-Ecuadorian margin
consists of two crustal units oc1 and oc2 overlain with two sediment units os1 and os2 in the trench. Strong reflections near 8 s twt are interpreted as the oceanic Moho.
The fractured margin is fronted by scarp S of the Atacames re-entrant, a mass-wasting unit and an imbricated frontal wedge. Sal-2 indicates the location of the Salieri2 wide
angle model (Sallares et al., 2005).

The PSDM approach includes the following steps. 1) A smooth At depth greater than 5 km below sea level (bsl), PSDM-derived
2-D reference medium is computed from the velocity macro-model velocities are poorly constrained. The 1-centered gamma function
estimated by velocity analysis, transformed in depth by the Dix and flat reflectors in Iso-X panels (Fig. 3) indicate that errors in Vp
equation, then interpolated and smoothed. 2) A dynamic ray trac- estimates are likely lower than 5% up to 8 km depth and greater
ing method (e.g. Operto et al., 2000) is used to estimate travel than 10% at greater depth.
time, amplitude and other ray related parameters. 3) The short In line SIS22 (Figs. 2 and 3) between 0 and −7 km, high-
wavelength components of the velocity distribution, alias depth amplitude, high-frequency, bedded reflectors of unit os1 (up to
migrated image, is computed in only one iteration by using the 500-m-thick) lap westward onto a westward-rising basal uncon-
method developed by Thierry et al. (1999). 4) An accurate migrated formity. Unit os2 is a 600 to 800-m-thick unit of low-amplitude,
image (correct geometry and amplitude of velocity perturbations low-frequency, bedded reflectors, extending over the Nazca plate
of seismic reflectors) is obtained by using the “migration-velocity-
abyssal plain. PSDM-derived Vp velocities for these units range
analysis” approach (Al-Yahya, 1989). Iso-X panels, or Common Im-
from 1.5 to 2.5 km/s. A high-amplitude prominent reflector To
age Gathers, are stored during migration and semblance panels are
separates unit os2 from underlying unit oc1, which, on the PSDM
estimated in the depth domain to obtain a local correction func-
section, reveals remarkably high-amplitude bedded reflectors. Unit
tion for the velocity macro-model. The velocity macro-model is
oc1, which is 1.5 to 2-km-thick and locally thickens to 2.5 km be-
iteratively corrected during migration until the semblance panels
tween 8 and 12 km westward from the trench, is bounded down-
are around 1 for all the main reflectors. When this condition is
satisfied, we stack all the Common Image Gathers to obtain the fi- ward by diffracting unit oc2. PSDM-derived Vp velocities for unit
nal and accurate migrated image. In post-processing, an Amplitude oc1 range from 3.0 to 5.1 km/s. A prominent high-amplitude re-
Gain Control is applied to the final migrated image. flector, which we interpret as the Moho discontinuity, floors layer
oc2 and deepens landward. The seafloor and reflectors of os1, os2,
4. Data description To and oc1 step down landward by 50 to 100 m along steeply
landward-dipping normal faults F1 to F5 that outcrops 15, 11, 8,
Lines SIS22 (Figs. 2, 3), SIS54 (Fig. 4), and SIS55 (Fig. 5) intended 7 and 4.5 km to the west from the trench.
to image a subducted seamount in order to highlight related mar- In line SIS54 (Fig. 4), between −2 and −10 km, a diffracting
gin specific structures and quantify tectonic basal erosion associ- topographic-high is the image of the southernmost seamount of
ated with seamount subduction.
the Atacames chain (Fig. 1A). To the east, in the trench, 500-m-
thick unit os1 overlays 1-km-thick unit os2, while to the west of
4.1. Nazca plate seismic units
the seamount, 800-m-thick unit os2 blankets the Nazca plate. Units
4.1.1. In the trench oc1 and oc2 show a contrasting reflectivity. A high-amplitude re-
Time and depth sections of seismic lines image four units la- flector outlines the Moho. PSDM-derived Vp velocities for seismic
beled os1, os2, oc1 and oc2 from top to bottom (Figs. 2 and 3). units os1, os2 and oc1 are similar to those in line SIS22. Normal
150 B. Marcaillou et al. / Earth and Planetary Science Letters 433 (2016) 146–158

Fig. 3. Pre-stack depth migrated multi-channel seismic reflection line SIS22 with velocity VP model (location in Fig. 1). os1, os2, oc1, oc2, BSR, ls, ms, us, Er, To, D, sc and smt
are as in Fig. 2. The oceanic upper crust oc1 is strongly reflective. The interpreted subduction channel is up to 1–1.5-km-thick. The flat to slightly landward-rising section
of reflector To at 20–30 km distance possibly images the southern flank of the subducted Atacames seamount. The top of the margin basement is outlined by strongly
reflective layer ls and is overlain by as much as 2.5 km of Cenozoic sediment. Semblance and Iso-X panels at −15 km and −5 km highlight the less-than-5% uncertainty in
Vp velocities that are reliable until 5 km depth below sea level. The vertical exaggeration is 1.25.

faults affect the Nazca plate at 23.5, 20, 16 km to the west from plate contact. Unit sc, between D and To, is thus interpreted as
the trench. the subduction channel, which is commonly defined as a layer of
Thus, across the incoming Nazca plate, seismic lines SIS22 and trench sediments and possibly rocks eroded from the margin toe
SIS54 image two crustal units interpreted as upper and lower mafic (e.g. Cloos and Shreve, 1988). Although Vp is poorly constrained
oceanic crust and a landward-dipping Moho. Normal faults, likely deeper than 5 km, sc is interpreted at greater depth near the land-
related to the downgoing plate bending, affect the upper crust. In ward termination of line SIS22.
spite of poorly constrained velocities at depth greater than 5 km, In line SIS54, instead of bending downward into the subduc-
the oceanic crust thickness is possibly of ∼7.5 km. tion zone, reflector To rises landward from 5.6-km-deep in the
trench to 2.7-km-deep at 12 km distance, and then dips down
4.1.2. Beneath the margin to a 4.5 km depth at 20 km distance (Fig. 4). Therefore, To de-
In line SIS22, strongly reflective unit oc1, topped by To is iden- lineates a ∼2.9-km-high summit that culminates beneath the sea-
tified as far landward as 40 km distance (Fig. 2). Reflector To dips ward tip of shallow spur P (Fig. 1A). The diffracting, poorly lay-
landward from 0 to 18.5 km distance with a mean 4 to 5◦ an- ered acoustic facies beneath the summit is similar to that of the
gle. Farther east, To flattens and slightly rises through trenchward- Atacames seamount in the trench and contrasts with the lay-
dipping faults to a 6-km depth at 22 and 28 km distance, where ered high-amplitude reflectors of unit oc1. The velocity struc-
1.4 to 1.8-km-thick unit oc1 thickens locally to ∼2.9 km (Fig. 3). ture within the seamounts, from 3.2–3.4 km/s at the summit to
At 31 km distance, To steps down by ∼700 m through an inter- 4.2–4.6 km/s at the root, is also similar (Fig. 4). Landward of
preted landward-dipping fault, and dips landward with a ∼10◦ the seamount leading flank, reflector To flattens at a depth shal-
angle. Above To, discontinuous reflector D, with high amplitude lower than that of the trailing flank in the trench, and then rises
from 10 to 35 km, dips landward. D tops a 0.5 to 1-km-thick unit landward suggesting the existence of a second peak farther land-
sc that is poorly reflective at 0–10 km and 17–20 km distance and ward. Thus, this ∼3 km-high western peak WP that climaxes at
strongly reflective at 32–42 km distance. Numerous margin faults 11–13 km from the deformation front is interpreted as belonging
sole out downward onto D, particularly beneath the frontal wedge, to a double-peak seamount. No subduction channel is evidenced in
where D shows low amplitude. This low acoustic impedance con- line SIS54.
trast suggests similar velocities above and beneath reflector D that Strike line SIS55 crosses line SIS22 at 40.2 km distance and
we interpret as the décollement thrust and therefore the inter- terminates close to the eastern end of line SIS54 (Fig. 5). Strong re-
 B. Marcaillou et al. / Earth and Planetary Science Letters 433 (2016) 146–158 151

Fig. 4. Pre-stack depth migrated multi-channel seismic line SIS54 with velocity VP model (location in Fig. 1). os1, os2, oc1, oc2, BSR, ls, ms, us, Er, To, D, sc and smt are as
in Fig. 2. The western peak WP of the subducted Atacames seamount is located beneath the margin slope and the shallow spur P of the continental shelf. The margin is
pervasively fractured over the leading flank and collapses in the wake of the seamount. The flat to landward-rising top of the oceanic crust landward of the leading flank of
WP suggests an eastern peak located landward from the eastern end of line SIS54. Semblance and Iso-X panels at −25 km and −15 km highlight the less-than-5% uncertainty
in Vp velocities that are reliable until 5 km depth below sea level. The vertical exaggeration is 1.25.

flectors To and D, that bounds unit sc, are interpreted at 40.2 km 4.2. Margin seismic units
distance at 7.6 and 6.2 km depths respectively and at 52 km dis-
tance at 4.5 and 4.2 km depths respectively. The consistency of Shallow units ls, ms and us consists of bedded reflectors sug-
these reflectors’ depths at the crossings between lines SIS55, SIS22 gesting a sedimentary nature. ls overlays a diffracting acoustic
and SIS54 supports the interpretation of D and To in line SIS55 as basement (Figs. 3, 4, 5).
being the décollement and the top of the subducting oceanic crust. The basement is bounded downward by reflectors D or To and
Thus, along line SIS55, the top of the subducting oceanic crust rises upward by strongly reflective unit ls. In line SIS22, the acoustic
northward by ∼3 km with a mean 15◦ angle between lines SIS22 basement thickness decreases from 5 km at 44 km distance to
2.7 km at 24 km and 1.3 km at 16 km (Fig. 3). In line SIS54,
and SIS54, imaging the southern flank of the eastern peak EP of
the basement is 2.2-km-thick at 24 km distance and as thin as
the subducted seamount that likely climaxes 30 km landward from
200 to 800 m above the western peak at 8–16 km (Fig. 4). Thus,
the western peak interpreted in line SIS54. The velocity structure
at similar distance from the trench, 24 km for instance, the base-
of this seamount is similar to that of the seamounts observed in
ment is 500 m thinner in line SIS54 than in line SIS22. Line SIS55
line SIS54 (Fig. 5). The 1.4 km-thick subduction channel smeared
confirms this difference with a 2.9 km-thick basement at the inter-
over the seamount flank pinches out at the northern end of line section with line SIS22 and 2.1-km-thick basement near line SIS54
SIS55, near EP summit. (Fig. 5). However, Line SIS55 is not strictly parallel to the margin
This set of interpreted seismic lines thus depicts a ∼30 × 40 km front and a slight and gradual thinning is possibly related to this
double-peak subducted seamount with a western peak imaged obliquity (Fig. 1A). Unit ls is up to 1.2-km-thick and differs from
on line SIS54, and an eastern peak likely located 30 km further overlying units by internal reflectors of higher amplitude (Fig. 2).
landward, its southern flank being imaged on lines SIS22 and Unit ms is up to 2.4-km-thick and consists of low-amplitude and
SIS55. disrupted reflectors topped by reflector Er. In line SIS22, dipping
152 B. Marcaillou et al. / Earth and Planetary Science Letters 433 (2016) 146–158

Fig. 5. Pre-stack depth migrated multi-channel seismic line SIS55 with velocity VP model (location in Fig. 1). ls, ms, us, To, D and sc are as in Fig. 2. The top of the oceanic
crust To rises northward along the southern flank of the eastern peak EP of the Atacames seamount. EP likely climaxes near the eastern limit of line SIS54, ∼30 km landward
from the western peak. Blue arrows indicate the northward thinning of the margin basement. The vertical exaggeration is 1.25. (For interpretation of the references to color
in this figure legend, the reader is referred to the web version of this article.)

reflectors of ms toplap under Er at 32–37 km and 39–45 km 4.3.2. Above the subducted seamount
(Figs. 2 and 6). Similar toplaps are also observed in line SIS54 In line SIS54, above the subducted seamount summit and lead-
between 14 and 21 km. These truncations indicate an erosion sur- ing flank, at 11 to 20 km, the seafloor shows a very uneven
face, Er, restricted to the continental shelf, extending trenchward morphology (Fig. 4). A network of dominantly landward-dipping
to 30 km in line SIS22 and 11 km in line SIS54 (Figs. 6 and 7). high-angle faults cut through units us, ms and locally ls with a
Unit us is up to ∼500-m-thick and consists of moderate-amplitude dominant reverse throw (Fig. 7). Over the trailing flank of the sub-
discontinuous reflectors (Fig. 6). These reflectors are horizontal to ducted seamount, unit us is divided into thin reflective layers that
slightly trenchward-dipping along the flat continental shelf and dip trenchward generally steeper than the overlying seafloor sug-
dips trenchward along the upper outer slope. gesting either prograding deposits or slides. Some of these layers
are vertically displaced along bounding slide surfaces parallel to
their internal reflectors. The trenchward displacement along these
4.3. Margin tectonic pattern
surfaces is attested by the updip correlation between the slid-
ing surfaces and small seafloor scarps at 4.4, 7, 9.3 and 10 km
4.3.1. At the margin front from the trench (Figs. 4 and 7). These sliding surfaces sole out
In line SIS22, a 1-km-thick unit of low-amplitude shattered hor- downward onto a reflector that mimics the seafloor with inverse
izontal reflectors rests over the high-amplitude reflectors of unit polarity and cuts through the stratigraphic layering. Thus, this
ls, between 7 and 14 km, separating the 1.3-km-high scarp S from Bottom Simulating Reflector (BSR) is thought to act as a master
the frontal wedge. This slope basin unit consists of hemipelagic surface promoting slope destabilization. BSRs are widely extended
deposits and debris flows, of which the most recent is dated along the Ecuadorean–Colombian margin (Marcaillou et al., 2006b,
3040 ± 60 yr (Ratzov et al., 2010) (Figs. 2 and 3). At the defor- 2008). The thickness of underlying sediments and basement units,
mation front, 600-m-thick trench unit os1 is shifted upward by when taken collectively, is ∼1 km over the seamount trailing flank
∼200 m and deformed through a steep upward-concave fault ft1 whereas this thickness reaches ∼3 km at the landward end of line
that is interpreted to outcrop at the trench axis, and sole out SIS54 (Fig. 4). Strongly reflective unit ls lies at 2.5 to 3.0 km depth
downward onto reflector D, thus delimiting the frontal thrust slice. at 17 km distance along line SIS54, rises to 2.0–2.2 km above the
To the east, other tectonic slices are possibly related to imbricate western peak and deepens trenchward to 4 km depth at 7 km dis-
sediment as well, forming a small 4-km-wide frontal prism likely tance, implying that the layer subsided by ∼2 km in response to
associated with reworked mass-wasting deposits. In contrast, no the seamount subduction. Between 7 and 12 km from the trench,
imbricate prism fronts the steep outer margin slope in line SIS54 unit ls is deformed by sub-horizontal to low-angle thrust faults
(Fig. 4). (Fig. 7). These shallow-angle thrusts, which deform both unit ls
 B. Marcaillou et al. / Earth and Planetary Science Letters 433 (2016) 146–158 153

Fig. 6. Close-up of post-stack time migrated seismic line SIS22 highlighting the stratigraphy and deformation pattern of the margin shelf-edge area. BSR, ls, ms, us, Er, To, D,
sc and smt are as in Fig. 2. Erosion surface Er separates pervasively fractured sedimentary units ls and ms from overlying flat reflector of unit us. The current depth of Er
suggests long-term subsidence of the continental shelf.

and the margin basement, likely connect with the interplate dé- the base of oc1 and the Moho are 3, 5.8 and 12-km-deep bsl re-
collement shear zone located along the seamount trailing flank. spectively. Therefore, immediately north of the Carnegie ridge, the
Nazca plate oceanic crust is ∼9-km-thick, including an 8.2-km-
5. Discussion thick mafic crust and a 0.8-km-thick sediment layer. The crust is
thus thickened by ∼20% in comparison with the ∼7.5-km-thick
5.1. Thickness and velocity of the oceanic crust oceanic crust measured ∼200 km further north (Agudelo et al.,
2009) suggesting a moderate influence of the Galapagos hotspot
The N–S-trending wide-angle seismic velocity model Sal-2 within the study area. This thickening appears to result directly
(Sallares et al., 2005) (Fig. 8A) that intersects line SIS22 near its from thickening of the lower crust, oc2 (6.3 km in Sal-2, 4.0 km in
western end (Fig. 2) indicates mantle-like velocities (>7.5 km/s) Agudelo et al., 2009), while the thickness of the upper crust, oc1,
at depths close to 10 km bsl. Seismic lines SIS22 and SIS54 mi- is similar.
grated to depth image a strong reflector interpreted as the Moho at Where constrained, at depth lower than 5 km bsl, Vp velocities
∼7.5-km depth beneath the seafloor, but Vp velocities are uncon- derived from PSDM for line SIS22 in the trench are significantly
strained at depth greater than 5 km and underestimated compared slower than in generic 15-Myr-old oceanic crust (Fig. 8B) (White
to velocities calculated from wide-angle data (Fig. 8B). A better es- et al., 1992). The 4 km/s velocity for instance is shifted downward
timate of the Nazca plate oceanic crust thickness is calculated at by 0.8–1 km compared with this generic oceanic crust. Moreover,
the intersection of line SIS22 and model Sal 2, by using PSDM- Vp velocities at the top of the oceanic crust, 3.0–3.2 km/s, is 40%
derived velocities Vp for sediment and oc1 layers, and wide-angle- slower in comparison with the 4.9 km/s estimated with a simi-
derived Vp for oc2 which depth is greater than 5 km bsl. The lar PSDM method ∼200 km further north (Table 3 in Agudelo et
thickness for sediments and oc1 and oc2 layers is 0.8, 1 and 1.9 s al., 2009) or with the 4.8 km/s estimated from wide angle data at
two-way-travel time respectively (Fig. 2). PSDM-derived Vp veloc- the Carnegie ridge crest ∼50 km further south (Graindorge et al.,
ities range from 1.5 to 2.7 km/s within the sediments and from 2004). Vp velocities derived from wide-angle seismic model Sal-2
3.0 to 4.9 km/s within oc1. Wide-angle velocity ranges from 6.0 to (Sallares et al., 2005) (Fig. 8), although possibly uncertain near the
7.2 km/s within oc2. Thus, 37 km west of the trench, the seafloor, northern end of the line, range from 3.0 to 5.5 km/s within the up-
154 B. Marcaillou et al. / Earth and Planetary Science Letters 433 (2016) 146–158

Fig. 7. Close-up of pre-stack depth migrated seismic line SIS54 showing the western peak WP of the subducted Atacames seamount and related margin deformation. BSR,
ls, ms, us, Er, To, D and sc are as in Fig. 2 and the vertical exaggeration is 1.4. The close-up highlight the role of the BSR on shallow sediment sliding in the wake of the
subducted seamount.

per crust oc1 that is consistent with our estimates at depth lower and SIS54 normal faults extend from topographical scarps at the
than 5 km and also significantly lower than wide angle model to seafloor through the oceanic sediment and upper crust, possibly
the north and the south of the study area (Agudelo et al., 2009; triggering seawater percolation and rock alteration. Fluid-rich sub-
Graindorge et al., 2004). ducting oceanic crust is thought to be one of the major mecha-
This lower velocity is possibly related to the Galapagos hotspot nisms for fluid supply into subduction zones (Ranero and Sallares,
influence during the oceanic crust formation. However, compila- 2004; von Huene et al., 2004). At depth, expelled fluids migrate
tions of studies of oceanic crust structures reveal that hotspots upward into the subduction channel where they are trapped due
mainly affect the thickness and velocity of oceanic layer 3 (Mutter to low permeability of the upper plate, possibly increasing pore-
and Mutter, 1993). Consistently, wide angle models across Carnegie fluid pressure and lowering the degree of the interplate locking
(Graindorge et al., 2004), Cocos (Sallares et al., 2003) and Malpelo (Moreno et al., 2014). Moreover, expelled overpressured fluids pos-
ridges (Marcaillou et al., 2006a) compared with velocity models sibly invade fractures in the upper plate, drag fragments into the
through generic crust further north and south (Agudelo et al., subduction channel, triggering basal hydrofracturing at the margin
2009; Gailler et al., 2007) show a reduced velocity at the base of toe (von Huene et al., 2004).
the thickened layer 3 at ridges axis but insignificant variations in
layer 2 velocity and thickness from the ridge crest to the adjacent 5.2. Structural effects of the Atacames seamounts subduction on
abyssal plain. The Sal-2 Vp model in the study area (Sallares et al., inter-seismic coupling and seismogenesis
2005), with velocities depressed within oceanic layer 2 but similar
to generic crust within layer 3 (Fig. 8B), is not typical of hotspot In the study area, a ∼30 × 40 km subducted double-peak
influence onto oceanic crust velocity structure. seamount extends as far landward as ∼45 km from the defor-
Alternatively, Ranero and Sallares (2004) attributed depressed mation front and affects margin tectonic deformation, basal ero-
velocities within the mafic crust for the 48-Myr-old Pacific plate sion and inter-seismic coupling. Analogue modeling (Dominguez
offshore off Chile (Fig. 8B) to pervasive rock alteration by seawa- et al., 1998), on land observations (Vannucchi et al., 2006) and
ter percolation through extensional bending-faults. In lines SIS22 seismic reflection data (Ranero and von Huene, 2000) attest that
 B. Marcaillou et al. / Earth and Planetary Science Letters 433 (2016) 146–158 155

Fig. 8. (A) Wide-angle model Sal-2 (Sallares et al., 2005) showing Vp velocity distribution and the northward shallowing of the Moho of the oceanic plate along the northern
flank of the Carnegie ridge. Black and white circles on the seafloor are Ocean Bottom Seismometer (OBS) location. Line SIS22 intersects wide-angle model Sal-2 at 18 km.
(B) Velocity–depth structure within the sediment and oceanic upper crust, oc1, of the incoming Nazca plate derived from PSDM line SIS22 at −20 km (thick blue line). At
depth greater than 5 km bsl, velocities from PSDM (thin dashed blue line) are unconstrained. This velocity–depth structure is compared to velocities from wide angle model
Sal-2 at 18 km (green line) (Sallares et al., 2005), to velocities estimated with a similar PSDM method ∼200 km further north (purple line) (Table 3 in Agudelo et al., 2009),
to velocities of generic 15-Myr-old Pacific oceanic crust (black lines) (White et al., 1992) and to velocities within the hydrated crust of 48-Myr-old Pacific plate offshore of
Chili (red line) (Ranero and Sallares, 2004). Sed, To, oc1, oc2 and M stand for sediment, top of the oceanic crust, oceanic upper and lower crust and Moho respectively, as
interpreted in line SIS22 (Fig. 2). Velocities within the upper crust oc1 are depressed possibly due to pervasive rock alteration by seawater percolation through extensional
bending-faults. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

subducted seamounts generate compressive and strike-slip defor- while at an equivalent distance from the trench (22–25 km) in line
mation in the outer margin resulting in a complex network of SIS22, the basement thickness is 2.9 km. Moreover along line SIS54
arcuate backthrusts, sub-vertical transtensive fractures and normal the basement is ∼1 km thinner than in line SIS22 at an equiv-
faults. In the wake of seamounts, masses sliding result in steep alent distance from the deformation front between the seamount
bathymetric scarps associated with trench-facing semi-circular re- peaks (18 to 24 km). Thus, in addition to regional basal erosion,
entrants. Basal erosion related to seamount subduction has been the subduction of the seamount has tectonically eroded the base
inferred from seismic data (Bangs et al., 2006; Ranero and von of the margin basement by at least ∼0.8–1 km. This measurement
Huene, 2000). The seismic data presented here provides evidence is likely to be underestimated as the margin basement in line SIS22
for prominent basal erosion and documents the superposition of has also undergone the erosional impact of the subduction of the
two strain fields contributing to tectonic deformation and frontal seamount southern flank (Fig. 3). The interpreted subduction chan-
erosion related to seamount subduction. nel along the southern flank of the eastern peak is locally thicker
(∼1.4 km) than further south along line SIS55 and along line SIS22.
5.2.1. Tectonic processes for basal erosion Thus, the products of basal erosion possibly contribute to thicken
The Atacames seamount subduction uplifted the inner trench the subduction channel in the vicinity of the subducted seamount.
slope by ∼1.6 km as attested by the depth difference between the High fluid content in the descending upper crust, suspected
1.4-km-deep seafloor of promontory P above the western peak and from the low velocity structure of mafic layer 2 in the trench and
the 2 km bathymetric contour extending both north and south of in the subduction channel possibly favors margin basal hydrofrac-
the promontory at equivalent distance from the trench (Fig. 1A, turing by upward expulsion of overpressured fluid (von Huene
Fig. 4). Moreover, at 12 km from the trench, along line SIS54, et al., 2004). Moreover, basement outcrops along seamount flanks
the 2.5 km depth of the base of unit ls above the western peak (Fig. 4) are known to be high-permeability regions of hydrothermal
is similarly ∼1.7 km shallower than its 4.2 km depth along line recharge and discharge (e.g. Fisher et al., 2003). The tectonic role
SIS22 at equivalent distance from the trench. Substantial tectonic of overpressured fluids is thus expected to be particularly strong
shortening accompanies the margin uplift as shown by the numer- in the vicinity of subducted seamounts. Thus, margin tectonic frac-
ous reverse faults affecting the margin wedge between the peaks turing due to the compressive stress generated by the seamount
(Fig. 7), consistent with sandbox models (Dominguez et al., 1998). subduction and hydrofracturing by overpressured fluids expelled
During seamount subduction, the margin outer wedge under- from the subducted plate likely contribute to the margin basal ero-
went significant thinning by basal tectonic erosion as evidenced sion and thinning (Fig. 9).
from the analysis of the MCS lines. Line SIS55 reveals a spectacu-
lar ∼3-km-thinning of the margin basement that is ∼5.5-km-thick 5.2.2. Tectonic processes for frontal erosion
on average at 0–40 km and 2.5-km-thick at 52 km. This thinning is In the wake of the subducted Atacames seamount, structures of
related to a 3-km-shallowing of the top of the oceanic crusts, along the margin slope reveal the overlap of two tectonic patterns. First,
the southern flank of the eastern peak, which is confirmed by the shallow sediment sheets sliding along trenchward-dipping arcu-
consistent To depth in perpendicular lines SIS22 and SIS54. A small ate faults that sole out downward onto the BSR, highlight the key
part of this thickness variation is possibly related to underplat- role of the base of the gas hydrate on the over-steepened outer
ing at 0–35 km distance in line SIS55, although unsubstantiated in margin destabilization. Just below the lower stability limit of gas
the data. Moreover, line SIS55 is not strictly parallel to the trench hydrates, methane and water exists as low-density fluids and are
(Fig. 1) which implies a slight and gradual basement thinning. probably under high pressure, causing significant weakening of the
Thus, a small part of the abrupt and local ∼3-km-thinning is possi- sediment and generating a rise of pore pressure (Cochonat et al.,
bly apparent and related to this trenchward normal basement thin- 2002). This favors the basal shear plane to propagate within the
ning. This normal thinning can be measured on seismic lines and slope sedimentary cover and thus cause slope instabilities. Second,
subtracted from the observed thinning at the northern end of line in depth, contraction is shown by low-angle to flat thrust faults
SIS55. At 52 km in line SIS55, the basement thickness is 2.1 km (6–10 km in Fig. 7) that deform both the basement and unit ls
156 B. Marcaillou et al. / Earth and Planetary Science Letters 433 (2016) 146–158

Fig. 9. Interpretative cross-section of the margin structure, tectonic pattern and basal erosion due to the subduction of the Atacames double-peak seamount. Line SIS54 images
the section from the oceanic plate to the leading flank of the western peak and Line SIS55 images the southern flank of the eastern peak (double arrows). Seamount-induced
compressive stress and possible overpressured fluids upward expulsion from the hydrated subducting upper crust are the main drivers for tectonic deformation and local
subduction erosion. This subducted seamount is associated with a seismically quiet and low inter-seismic coupling area down to ∼8 km depth (Fig. 10). The 1942 Mw7.8
earthquake (Swenson and Beck, 1996) likely initiated ahead of the seamount leading flank in the transition zone between the low ISC and the locked seismogenic zone,
substantiating the model of Mochizuki et al. (2008). D, To, sc, BSR, Er, ls, ms and us are as in Fig. 2.

and that likely sole out against the seamount trailing flank. The and Dewey, 1984), which were mainly located in the trench and
thrusts separate tectonic sheets that appear to slightly overlap each around the subducted seamount. Besides, the noticeable absence
other. The low angle thrusts and associated sheets are thought to of a reported tsunami during the 1942 earthquake (Swenson and
result from a counter-clockwise rotation of the high-angle reverse Beck, 1996) implies a small seafloor acceleration along the updip
faults and margin tectonic blocks that were created landward of segment of the interplate contact consistent with low ISC. Events
the seamount (11–16 km in Fig. 7) and are thus considered as with focal mechanisms from the CMT Harvard catalog (Ekström
remnant imprints of the seamount collision with the margin. The et al., 2012) and the RENSIG catalog earthquakes (Alvarado, 2012;
rotation occurred while the margin subsided and tilted seaward in Beauval et al., 2013) (Fig. 10B) appear to be distributed on ei-
the wake of the seamount. Ultimately, the imbricated sheets might ther side of the low ISC corridor and reveal a dominant eastward
be reworked and governed by gravity tectonics as they take place slip with a slight dextral component across the plate boundary
along the steeply dipping outer-margin slope. Thus, two tectonic associated with a more complex distribution of rupture types. Be-
regimes separated vertically are controlling the margin deforma- cause of depth uncertainties (Beauval et al., 2013), we cannot
tion above the trailing flank of the seamount, although in the end rule out that some events located near the trench are intraslab.
they may join to reflect the single expression of gravity tectonics. However, at least for shallow earthquakes located near the coast-
line, these focal mechanisms indicate reactivation of multiple faults
5.2.3. Seamount subduction, inter-seismic coupling and seismogenesis with normal faulting and oblique thrusting within the margin.
In the study area, an interseismic coupling (ISC) model compat- Thus, the seismically quiet, low ISC corridor supports an outer-
ible with onshore GPS measurements (Chlieh et al., 2014; Nocquet margin wedge incapable of storing sufficient elastic strain to pro-
et al., 2014) supports a low ISC (ISC < 0.3) indentation within a duce large earthquakes and tsunamis. This incapacity may result
coupled (ISC = 0.5–0.6) asperity (Fig. 10A). This low ISC corridor is from highly fractured rocks in response to the Atacames seamount
∼70-km-long along-dip, in a NW–SE direction and ∼60-km-wide subduction. This interpretation concurs with our structural obser-
along line SIS55 from latitudes 0.5◦ S to 1◦ N. Because of the short vation of an extremely disrupted outer margin wedge near the
distance between the GPS network and the trench, 60–80 km in seamount (Fig. 9) and with the complex rupture pattern revealed
the study area, this network can resolve any ISC patch as small as by the RENSIG catalog, at least near the coastline. The 40 × 30 km
60 × 60 km, except near the trench (Chlieh et al., 2014). The shal- deepest low ISC indentation located landward of the double-peak
low part of the corridor spatially coincides with the trajectory of Atacames seamount remains unexplained. This indentation size is
the double-peak Atacames seamount while subducting beneath the smaller than the spatial resolution. If it exists it may be indicative
margin since 1 Myr toward its current location (Fig. 10). Moreover, of a deeply buried seamount or of the imprint left by past sub-
this low ISC corridor remarkably coincides with the seismically ducted seamounts of the Atacames chain.
quiet area over the 1994–2007 recording period (Fig. 10B) (Font These results substantiate the model of Wang and Bilek (2011)
et al., 2013) and during the 1942 aftershocks (Fig. 10B) (Mendoza that predicts high margin fracturing and low inter-seismic cou-
 B. Marcaillou et al. / Earth and Planetary Science Letters 433 (2016) 146–158 157

Fig. 10. Spatial distribution of (A) interseismic coupling along the interplate contact modified from Chlieh et al. (2014), (B) shallow earthquake and micro-earthquake
hypocenters modified from Font et al. (2013) and Alvarado (2012) (see Fig. 1B for details). Yellow stars define the western and eastern peaks, WP and EP, of the subducted
Atacames seamount. Plain yellow lines highlight the seamount flanks extent as observed in seismic lines. Grey stars and lines indicate the location of the seamount summit
and flanks 1 Myr ago and their trajectory while subducting beneath the margin. Black star and dashed ellipse represent the 1942 earthquake epicenter and rupture zone.
The pervasively fractured margin zone associated with the subducted double-peak Atacames seamount (DPAS) corresponds with low interplate coupling coefficient (Chlieh
et al., 2014) and a gap in the distribution of the aftershocks for the 1942 earthquake (white circle) (Mendoza and Dewey, 1984), earthquakes with thrust focal mechanisms
from the CMT Harvard catalog (red “beach-balls”) (Ekström et al., 2012), the Mw > 4 and depth < 40 km earthquakes from the 1988–2013 RENSIG catalog (blue beach-balls)
(Alvarado, 2012; Beauval et al., 2013) and the micro-seismicity (colored circle, legend in Fig. 1B) (Font et al., 2013). The 1942 great subduction earthquake nucleated landward
from this weakened margin corridor. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

pling with multiple ruptures occurring on a fracture network in listric normal faults that sole out onto the BSR. This complex slope
the vicinity of subducted seamounts. However, the 1942 Mw7.8 structure highlights the influence of the superposition of deep con-
earthquake initiated at the transition between the low ISC corri- traction and shallow gravitational tectonics, as well as the key role
dor and the seismogenic zone (Fig. 10A). The earthquake ruptured of gas hydrates on outer margin slope instabilities.
the large coupled asperity or several discrete asperities that extend A 0.5–1-km-thick subduction channel thickens to 1.4 km over
from latitudes 0.5◦ S to 1◦ N downdip to a ∼25-km-depth (Chlieh et the southern flank of the subducted seamount. This subduction
al., 2014). According to numerical models by Yang et al. (2012), a channel possibly incorporates products of basal erosion as well as
subducted seamount is able to stop the updip propagation of a co- fluids expelled upward from the hydrated upper crust. This me-
seismic rupture nucleated at greater depth. In Japan, a series of chanically weak layer is thus interpreted to facilitate seamount
Mw7.0 events ruptured the interplate contact in front of the lead- subduction.
ing flank of a subducted seamount, while the ISC is interpreted A seismically quiet, GPS-modeled low-ISC corridor spatially
to remain low along its summit and trailing flank thus favoring overlaps the double-dome Atacames seamount and the SSE-ward
seamount advance by creeping (Mochizuki et al., 2008). We then projection of the Atacames seamount chain in the subduction zone.
conclude that the 1942 Mw7.8 megathrust earthquake occurred The low interplate coupling within this corridor would result from
downdip of the subducted Atacames seamount chain. The perva- both the severe faulting of the margin in response to seamount
sively fractured, tectonically eroded, and creeping margin in the subduction and a fluid-rich subduction channel. This conclusion
wake of the subducted seamounts inhibited rupture propagation supports previous studies (Wang and Bilek, 2011) that proposed
and prevented the generation of a large tsunami. that subducted seamounts provide unfavorable conditions for lock-
ing the updip segment of the plate boundary.
6. Conclusion The 1942 Mw7.8 earthquake initiated in the mechanically un-
stable transition zone between the seismically quiet low ISC cor-
ridor and the locked zone. Consistent with models by Yang et al.
The PSDM seismic reflection images collected across the North-
(2012), we suggest that the subducted Atacames seamounts could
Ecuadorian subduction zone combined with detailed seafloor mor-
have stopped the co-seismic rupture of the large 1942 earthquake
phology, results of GPS modeling and seismicity data reveal the
ahead of its leading flank and left updip a poorly locked margin
following:
corridor.
In addition to the global long-term tectonic erosion of the mar-
gin, an extra >0.8–1.0-km-thinning of the margin basement is Acknowledgements
documented in relation with the subducted ∼30 × 40 km double-
dome Atacames seamount identified beneath the outer-wedge This work was funded by the Institut de Recherche pour le
margin slope and outer continental shelf. Développement (IRD), Institut National des Sciences de l’Univers
During subduction, the seamount uplifted the margin by 1.6 km (INSU), and Institut Français pour l’Exploitation de la Mer (Ifre-
and pervasively broke the basement rocks and sedimentary cover mer), which provided ship time during the SISTEUR-2000 cruise.
by intense faulting that is likely to weaken considerably the margin We are grateful to the scientific parties and crew for their support
framework. during data acquisition, and to IRD for funding the Post-Doctoral
A 20-km-large morphologic re-entrant bounded by a 1.3-km- study of Elia d’Acremont. Finally, we thank Laure Schenini for her
high scarp formed in the wake of the subducted seamount and help during data processing as well as Simon Williams and two
accumulated mass wasting deposits. anonymous reviewers for their very helpful comments.
Reverse faults that had deformed the margin basement ahead of
References
the seamount, rotated counter-clockwise in the seamount wake to
convert into slightly trenchward dipping thrust faults. This base- Agudelo, W., Ribodetti, A., Collot, J.-Y., Operto, S., 2009. Joint inversion of multichan-
ment fault system is overlain with slope sediment sliding along nel seismic reflection and wide-angle seismic data: improved imaging of the
158 B. Marcaillou et al. / Earth and Planetary Science Letters 433 (2016) 146–158

crustal structure of the N Ecuador–SW Colombia convergent margin. J. Geophys. Kodaira, S., Takahashi, N., Nakanishi, A., Miura, S., Kaneda, Y., 2000. Subducted
Res. 114, B02306. seamount imaged in the rupture zone of the 1946 Nankaido earthquake. Sci-
Aki, K., 1979. Characterization of barriers on an earthquake fault. J. Geophys. Res. 84, ence 28, 104–106.
6140–6148. Lambaré, G., Operto, S., Podvin, P., Thierry, P., 2003. 3D ray + born migration/in-
Alvarado, A., 2012. Néotectonique et cinématique de la déformation continentale en version – part 1: theory. Geophysics 68, 1348–1356. http://dx.doi.org/10.1190/
Equateur. Université de Grenoble. 1.1598128.
Al-Yahya, K.M., 1989. Velocity analysis by iterative profile migration. Geophysics 54, Marcaillou, B., Charvis, P., Collot, J.-Y., 2006a. Structure of the Malpelo Ridge (Colom-
718–729. bia) from seismic and gravity modelling. Mar. Geophys. Res. 27, 289–300. http://
Bangs, N.L.B., Gulick, S.P.S., Shipley, T.H., 2006. Seamount subduction erosion in the dx.doi.org/10.1007/s11001-006-9009-y.
Nankai Trough and its potential impact on the seismogenic zone. Geology 34, Marcaillou, B., Spence, G.D., Collot, J.-Y., Wang, K., 2006b. Thermal regime from bot-
701. http://dx.doi.org/10.1130/G22451.1. tom simulating reflectors along the N Ecuador–S Colombia margin: relation to
margin segmentation and the twentieth century great subduction earthquakes.
Beauval, C., Yepes, H., Palacios, P., Segovia, M., Alvarado, A., 2013. An earthquake
J. Geophys. Res. 111.
catalog for seismic hazard assessment in Ecuador. Bull. Seismol. Soc. Am. 103,
Marcaillou, B., Spence, G., Wang, K., Collot, J.-Y., Ribodetti, A., 2008. Thermal segmen-
773–786. http://dx.doi.org/10.1785/0120120270.
tation along the N. Ecuador–S. Colombia margin (1–4◦ N): prominent influence
Beck, S.L., Ruff, L.J., 1984. The rupture process of the great 1979 Colombia earth-
of sedimentation rate in the trench. Earth Planet. Sci. Lett. 272, 296–308. http://
quake: evidence for the asperity model. J. Geophys. Res. 89, 9281–9291.
dx.doi.org/10.1016/j.epsl.2008.04.049.
Chlieh, M., Mothes, P.A., Nocquet, J.-M., Jarrin, P., Charvis, P., Cisneros, D., Font, Y.,
Mendoza, C., Dewey, J.W., 1984. Seismicity associated with the great Colombia–
Collot, J.-Y., Villegas-Lanza, J.C., Rolandone, F., Vallée, M., Regnier, M., Segovia, M.,
Ecuador earthquakes of 1942, 1958 and 1979: implications for barrier models
Martin, X., Yepes, H., 2014. Distribution of discrete seismic asperities and aseis-
of earthquake rupture. Bull. Seismol. Soc. Am. 74, 577–593.
mic slip along the Ecuadorian megathrust. Earth Planet. Sci. Lett. 400, 292–301.
Michaud, F., Chabert, A., Collot, J.-Y., Sallares, V., Flueh, E.R., Charvis, P., Grain-
http://dx.doi.org/10.1016/j.epsl.2014.05.027.
dorge, D., Gutscher, M.-A., Bialas, J., 2005. Fields of multi-kilometer scale sub-
Clift, P.D., Vannucchi, P., 2004. Controls on tectonic accretion versus erosion in sub- circular depressions in the Carnegie Ridge sedimentary blanket: effect of un-
duction zones: implications for the origin and recycling of the continental crust. derwater carbonate dissolution? Mar. Geol. 216, 205–219. http://dx.doi.org/
Rev. Geophys. 42, RG2001. http://dx.doi.org/10.1029/2003RG000127. 10.1016/j.margeo.2005.01.003.
Cloos, M., 1992. Thrust-type subduction-zone earthquakes and seamount asperities: Mochizuki, K., Yamada, T., Shinohara, M., Yamanaka, Y., Kanazawa, T., 2008. Weak in-
a physical model for seismic rupture. Geology 20, 601–604. terplate coupling by seamounts and repeating M ∼ 7 earthquakes. Science 321,
Cloos, M., Shreve, R.L., 1988. Subduction-channel model of prism accretion, melange 1194–1197. http://dx.doi.org/10.1126/science.1160250.
formation, sediment subduction, and subduction erosion at convergent plate Moreno, M., Haberland, C., Oncken, O., Rietbrock, A., Angiboust, S., Heidbach,
margins. 1. Background and description. Pure Appl. Geophys. 128, 455–500. O., 2014. Locking of the Chile subduction zone controlled by fluid pres-
Cochonat, P., Cadet, J.-P., Lallemant, S.J., Mazzotti, S., Nouzé, H., Fouchet, C., Foucher, sure before the 2010 earthquake. Nat. Geosci. 7, 292–296. http://dx.doi.org/
J.-P., 2002. Slope instabilities and gravities processes in fluid migration and tec- 10.1038/NGEO2102.
tonically active environment in the eastern Nankai accretionary wedge (KAIKO– Mutter, C.Z., Mutter, J.C., 1993. Variations in thickness of layer 3 dominate oceanic
Tokai’96 cruise). Mar. Geol. 187, 193–202. crustal structure. Earth Planet. Sci. Lett. 117, 295–317.
Collot, J.-Y., Charvis, P., Gutscher, M.-A., Operto, S., 2002. Exploring the Ecuador– Nocquet, J.-M., Chlieh, M., Mothes, P.A., Rolandone, F., Jarrin, P., Cisneros, D., Al-
Colombia active margin and interplate seismogenic zone. Eos Trans. AGU 83, varado, A., Audin, L., Bondoux, F., Martin, X., Font, Y., Régnier, M., Vallée, M.,
189–190. Tran, T., Beauval, C., Maguiña Mendoza, J.M., Martinez, W., Tavera, H., Yepes, H.,
Collot, J.-Y., Lallemand, S.E., Pelletier, B., Eissen, J.-P., Glaçon, G., Fisher, M.A., Greene, 2014. Motion of continental slivers and creeping subduction in the Northern
H.G., Boulin, J., Daniel, J., Monzier, M., 1992. Geology of the d’Entrecasteaux– Andes. Nat. Geosci. http://dx.doi.org/10.1038/NGEO2099.
New Hebrides Arc collision zone: results from a deep submersible survey. Operto, S., Lambaré, G., Podvin, P., Thierry, P., Noble, M., 2003. 3D ray + born mi-
Tectonophysics 212, 213–241. gration/inversion – part 2: application to the SEG/EADGE overthrust experiment.
Collot, J.-Y., Michaud, F., Alvarado, A., Marcaillou, B., Sosson, M., Ratzov, G., Migeon, Geophysics 68, 1357–1370. http://dx.doi.org/10.1190/1.1598129.
S., Calahorrano, A., Pazmino, A., 2009. Visión general de la morfología submarina Operto, S., Xu, S., Lambaré, G., 2000. Can we quantitatively image complex structures
del margen convergente de Ecuador–Sur de Colombia: implicacionese sobre la with rays. Geophysics 65, 1223–1238. http://dx.doi.org/10.1190/1.1444814.
transferencia de masa y la edad de la subducción de la Cordillera de Carnegie. Ranero, C.R., Sallares, V., 2004. Geophysical evidence for hydration of the crust and
In: Collot, J.-Y., Sallares, V., Pazmino, N. (Eds.), Geología y Geofísica Marina y mantle of the Nazca plate during bending at the north Chile trench. Geology 32,
Terrestre del Ecuador: Desde La Costa Continental Hasta Las Islas Galápagos. 549–552.
CNDM/INOCAR/IRD, Guayaquil, pp. 47–74. Ranero, C.R., von Huene, R., 2000. Subduction erosion along the Middle America
Collot, J.-Y., Migeon, S., Spence, G.D., Legonidec, Y., Marcaillou, B., Schneider, J.-L., convergent margin. Nature 404, 748–755.
Michaud, F., Alvarado, A., Lebrun, J.-F., Sosson, M., 2005. Mapping the seafloor Ratzov, G., Collot, J.-Y., Sosson, M., Migeon, S., 2010. Mass-transport deposits in the
of the Ecuador–SW Colombia margin helps understand great subduction earth- northern Ecuador subduction trench: result of frontal erosion over multiple seis-
quakes. Eos Trans. AGU 86, 463–465. mic cycles. Earth Planet. Sci. Lett. 296, 89–102.
Sage, F., Collot, J.-Y., Ranero, C.R., 2006. Interplate patchiness and subduction erosion.
Dominguez, S., Lallemand, S.E., Malavieille, J., von Huene, R., 1998. Upper plate de-
Geology 34, 997–1000.
formation associated with seamount subduction. Tectonophysics 293, 207–224.
Sallares, V., Charvis, P., Flueh, E.R., Bialas, J., 2003. Seismic structure of Cocos and
Ekström, G., Nettles, M., Dziewoński, A.M., 2012. The global CMT project 2004–2010:
Malpelo volcanic ridges and implications for hotspot–ridge interaction. J. Geo-
centroid-moment tensors for 13,017 earthquakes. Phys. Earth Planet. In-
phys. Res. 108, 2564.
ter. 200–201, 1–9. http://dx.doi.org/10.1016/j.pepi.2012.04.002.
Sallares, V., Charvis, P., Flueh, E.R., Bialas, J., the SALIERI Scientific Party, 2005.
Fisher, A.T., Davis, E.E., Hutnak, M., Spiess, V., Zühlsdorff, L., Cherkaoui, A., Chris-
Seismic structure of the Carnegie ridge and the nature of the Galapagos
tiansen, L., Edwards, K.M., Macdonald, R., Villinger, H., Motti, M., Wheat, C.G.,
hotspot. Geophys. J. Int. 161, 763–788. http://dx.doi.org/10.1111/j.1365-246X.
Becker, K., 2003. Hydrothermal circulation across 50 km on a young ridge flank:
2005.02592.x.
the role of seamounts in guiding recharge and discharge at a crustal scale. Na-
Scholz, C.H., Small, C., 1997. The effect of seamount subduction on seismic coupling.
ture 421, 618–621.
Geology 25, 487–490.
Font, Y., Segovia, M., Vaca, S., Theunissen, T., 2013. Seismicity patterns along the Swenson, J.L., Beck, S.L., 1996. Historical 1942 Ecuador and 1942 Peru subduction
Ecuadorian subduction zone: new constraints from earthquake location in a earthquakes, and earthquake cycles along Colombia–Ecuador and Peru subduc-
3-D a priori velocity model. Geophys. J. Int. 193, 263–286. http://dx.doi.org/ tion segments. Pure Appl. Geophys. 146, 67–101.
10.1093/gji/ggs083. Thierry, P., Lambaré, G., Podvin, P., Noble, M., 1999. 3D preserved amplitude prestack
Gailler, A., Charvis, P., Flueh, E.R., 2007. Segmentation of the Nazca and depth migration on a workstation. Geophysics 64, 222–229.
South American plates along the Ecuador subduction zone from wide an- Vannucchi, P., Fisher, A.T., Bier, S., Gardner, T.W., 2006. From seamount accretion
gle seismic profiles. Earth Planet. Sci. Lett. 260, 444–464. http://dx.doi.org/ to tectonic erosion: formation of Osa Mélange and the effects of Cocos Ridge
10.1016/j.epsl.2007.05.045. subduction in southern Costa Rica. Tectonics 25, TC2004.
Graindorge, D., Calahorrano, A., Charvis, P., Collot, J.-Y., Béthoux, N., 2004. Deep Von Huene, R., Ranero, C.R., Vannucchi, P., 2004. Generic model of subduction ero-
structures of the Ecuador convergent margin and the Carnegie Ridge, possible sion. Geology 32, 913–916.
consequence on great earthquakes recurrence interval. Geophys. Res. Lett. 31, Wang, K., Bilek, S.L., 2011. Do subducting seamounts generate or stop large earth-
L04603. quakes? Geology 39, 819–822. http://dx.doi.org/10.1130/G31856.1.
Hampel, A., Adam, J., Kukowski, N., 2004. Response of the tectonically erosive White, R.S., McKenzie, D., O’Nions, R.K., 1992. Oceanic crustal thickness from seis-
south Peruvian forearc to subduction of the Nazca Ridge: analysis of three- mic measurements and rare earth element inversions. J. Geophys. Res. 97,
dimensional analogue experiments. Tectonics 23, TC5003. http://dx.doi.org/ 19683–19715.
10.1029/2003tc001585. Yang, H., Liu, Y., Lin, J., 2012. Effects of subducted seamounts on megathrust earth-
Kanamori, H., McNally, K.C., 1982. Variable rupture mode of the subduction zone quake nucleation and rupture propagation. Geophys. Res. Lett. 39, 24302. http://
along the Ecuador–Colombia coast. Bull. Seismol. Soc. Am. 72, 1241–1253. dx.doi.org/10.1029/2012GL053892.