Geophysical Journal International
Geophys. J. Int. (2015) 201, 1717–1721
doi: 10.1093/gji/ggv120
GJI Geodynamics and tectonics
EXPRESS LETTER
Capture of the Canary mantle plume material by the Gibraltar arc
mantle wedge during slab rollback
C.A. M´eriaux,1 J.C. Duarte,2 S.S. Duarte,1 W.P. Schellart,2 Z. Chen,2 F. Rosas,1,3
J. Mata1,3 and P. Terrinha1,4
1 Instituto
Dom Luiz, Universidade de Lisboa, P-1749-016 Lisboa, Portugal. E-mail: [email protected]
of Earth, Atmosphere and Environment, Monash University, Melbourne, Australia
3 Departamento de Geologia, Faculdade de Ciˆ
encias, Universidade de Lisboa, P-1749-016 Lisboa, Portugal
4 Instituto Portuguˆ
es do Mar e da Atmosfera, P-1749-077 Lisboa, Portugal
2 School
Accepted 2015 March 10. Received 2015 March 9; in original form 2015 February 2
SUMMARY
Recent evidence suggests that a portion of the Canary plume travelled northeastwards below
the lithosphere of the Atlas Mountains in North Africa towards the Alboran domain and was
captured ∼10 Ma ago by the Gibraltar subduction system in the Western Mediterranean. The
capture would have been associated with the mantle return flow induced by the westwardretreating slab that would have dragged and trapped a portion of the plume material in the
mantle wedge of the Gibraltar subduction zone. Such material eventually contaminated the
subduction related volcanism in the Alboran region. In this work, we use scaled analogue
models of slab–plume interaction to investigate the plausibility of the plume capture. An
upper-mantle-scaled model combines a narrow (400 km) edge-fixed subduction plate with a
laterally offset compositional plume. The subduction dominated by slab rollback and toroidal
mantle flow is seen to increasingly impact on the plume dynamics as the area of influence of
the toroidal flow cells at the surface is up to 500 × 1350 km2 . While the plume head initially
spreads axisymmetrically, it starts being distorted parallel to the plate in the direction of the
trench as the slab trench approaches the plume edge at a separation distance of about 500 km,
before getting dragged towards mantle wedge. When applied to the Canary plume–Gibraltar
subduction system, our model supports the observationally based conceptual model that mantle
plume material may have been dragged towards the mantle wedge by slab rollback-induced
toroidal mantle flow. Using a scaling argument for the spreading of a gravity current within a
channel, we also show that more than 1500 km of plume propagation in the sublithospheric
Atlas corridor is dynamically plausible.
Key words: Mantle processes; Subduction zone processes; Dynamics of lithosphere and
mantle; Hotspots; Africa.
1 I N T RO D U C T I O N
Plates and plumes are often seen as two different modes of mantle
convection. According to this view, plates are the consequence of
the top-down cooling of the Earth and plumes are associated with
the bottom-up heat release from the core-mantle boundary (Davies
1999). Notwithstanding, it is possible to envisage situations in which
these two main geodynamic features interact as displayed by some
tomographic studies showing the presence of plumes in the vicinity of subduction zones (Abdelwahed & Zhao 2007; Xue & Allen
2007; Obrebski et al. 2010). Geological/geochemical data have also
pointed out the occurrence of plume/slab interaction. For instance,
Turner & Hawkesworth (1998) have used geochemical tracers to
C
suggest that mantle plume material from the Samoa hotspot has
been dragged into the Lau backarc basin bordering the northern
Tonga subduction zone segment. Examples of modern case studies based on laboratory models include the interaction between the
Yellowstone plume and the Cascades subduction system (Kincaid
et al. 2013), and between the Samoa Plume and the Tonga system
(Druken et al. 2014).
In a recent work, Duggen et al. (2009) proposed a three-stage
conceptual model for the evolution of the Canary plume, Northwest Africa and Western Mediterranean (Fig. 1). The model was
inferred by the authors from geophysical and geochemical data. In
the first stage, ∼70 Ma ago, the Canary plume started to be deflected from the eastern Atlantic towards Northwest Africa. In the
The Authors 2015. Published by Oxford University Press on behalf of The Royal Astronomical Society.
1717
1718
C.A. M´eriaux et al.
Figure 1. Conceptual model of the Canary plume, Northwest Africa and Western Mediterranean following Duggen et al. (2009).
second stage, initiated ∼45 Ma ago, the plume propagated more
than 1500 km through a channel created by delaminated Atlas subcontinental lithosphere. In the third stage, around 10–8 Ma ago,
the plume was trapped in the subducting slab-related return flow
and advected towards the Alboran backarc region in the Western
Mediterranean.
While geochemical and geophysical data appear to support such
model, the feasibility of the above-described processes has never
been tested with physical models. This work primarily focusses on
the third stage of this geodynamic conceptual model by testing the
hypothesis of capturing Canary plume material by the Gibraltar subduction system. To do so, we present a novel model set up of scaled
dynamical experiments of plume material spreading in the vicinity
of a retreating slab. Our results support the feasibility of the plume
material advection mechanism for a system alike Canary/Gibraltar.
We also assess the dynamic conditions for the second stage of
Duggen et al. (2009) conceptual model using scaling arguments.
2 A NA LO G U E M O D E L L I N G
period after the slab tip arrived at the bottom (steady-state period).
The plume was simultaneously started with slab initiation with the
supply of a constant volume flux Q = 9.36 × 10−8 m3 s−1 . Photos
were taken every 10 s simultaneously from two sides and from the
top.
The buoyancy flux of the subducting plate Bs relative to the plume
Bp was estimated by the buoyancy flux ratio Bs /Bp = [(ρ s /ρ a −
1)gUWd]/[(1 − ρ a /ρ p )gQ], where g is the gravitational acceleration. This ratio equal to 7.85 is comparable to the one for the
Gibraltar–Canary system for which Bs /Bp is in the range of 6.67–
13.12, using a subduction rate U of 5 cm yr−1 , a slab width and
thickness of 400 km and 75 km respectively, a density contrast
(ρ s − ρ a ) = 80 kg m−3 and a Canary plume buoyancy Bp /g in the
range of 290–570 kg s−1 as estimated by King & Adam (2014).
The models were scaled with 1 cm in the model corresponding to
50 km in nature and 1 s corresponding to 7.9 kyr. The scaling of time
was estimated from the timescale ratio tNature /tModel = [μa H/(ρ s −
ρ a )gd2 ]Nature /[μa H/(ρ s − ρ a )gd2 ]Model using a sublithospheric upper
mantle viscosity of 4 × 1020 Pa s.
2.1 Methodology
2.2 Results
Experiments were performed in a tank of dimensions 100 (length) ×
62 (width) × 60 (height) cm3 that had been filled up to a model depth
H = 14 cm with glucose syrup of Newtonian viscosity (Schellart
2011). A dense plate made of a Newtonian viscosity silicone putty
(Weijermars 1986) mixed with fine iron powder and of dimensions
L × W × d equal to 80 × 8 × 1.5 cm3 was placed on the glucose
surface 3 cm off centre of the tank. A buoyant plume was introduced
through a nozzle at the base of the tank that was 12 cm away from
the plate edge and 32.5 cm away from the wall of the fixed slab
edge (see Fig. 2). Table 1 details all the material properties at 20◦ C.
Subduction of the plate was initiated by downward bending 3 cm
length of its free edge at an angle of 30◦ . The subduction rate
U of 0.0245 cm s−1 was determined by linearly fitting the slab
trench positions as a function of time during trench retreat for the
First, a transient phase took place during which the plate sunk to
the base of the upper mantle and the plume rose up to the surface
without interaction, as plate and plume were 2375 km (47.5 cm)
apart. During the second phase, the slab started retreating towards
the plume whose head began to grow and spread axisymmetrically.
The influence of the subducting plate on the spreading plume head
was seen with the onset of asymmetry in the plume head in a direction of the trench and parallel to the plate, as shown in Fig. 3. The
asymmetry started when the trench was 902 km (18 cm) away from
the plume head centre, or 497 km (9.93 cm) away from the plume
edge in the direction perpendicular to the trench, while the plume
head edge was 138 km (2.76 cm) far from the plate edge in the direction parallel to the trench. With the passing of the trench at the apex
of the plume head centre, capture of the plume head towards the
Plume capture
1719
d
xe g
Fi ilin
a
e
tr d g
1.5 c
m e
Subducting P
late
80 c
m
Plume
32.5
27
cm
cm
3
15
cm
e
Hing m
8
c
cm
14 cm
Box
Leng
th (1
00 c
m)
ox
id
W
th
(
62
cm
)
B
Figure 2. Analogue modelling setting.
Table 1. Material properties.
Ambient Glucose
Density (ρ a )
Viscosity (μa )
Subducting plate
Density (ρ s )
Viscosity (μs )
Plume
Density (ρ p )
Viscosity (μp )
Value
Units
1416
100
kg m−3
Pa s
1516
64 000
kg m−3
Pa s
1376
5
kg m−3
Pa s
mantle wedge began, and after 6 Myr, 9.5 per cent of surface area of
the plume head had been captured. This sequence of events is shown
in Fig. 3, and in Movie S1 (Movie_Gibraltar_19112014_GJI.mp4,
Supporting Information) showing the full dynamics of the
interaction.
3 DISCUSSION
The Gibraltar slab (also known as the Betic-Rif slab) and the Calabrian slab are thought to be the narrowest actively subducting slabs
on Earth (Schellart et al. 2007). Their widths range between 200
and 400 km (Gutscher et al. 2002). Although narrow laboratory
slab models have been run in the past (but without a plume; e.g.
Bellahsen et al. 2005), here we demonstrate for the first time that a
narrow slab with a width of only 400 km can generate toroidal return
flow cells that affect the mantle at considerable distance and over
a very large surface area. Our experiments show that the toroidal
return mantle flow could cause the motion of the black and buoyant
passive tracers laid on the glucose surface at a distance of about
27 cm (i.e. 1350 km or 3.4 times the slab width) from the trench
edge in the direction parallel to the trench. In the direction perpendicular to the trench, the influence of the mantle return flow on the
plume head can be observed when the edge of the plume head is
about 500 km apart from the retreating trench, which is a length
scale that is 1.25 times the plate width. Such 500 × 1350 km2 area
of influence of the 400 km wide slab is not negligible.
Our model very much supports the third stage of Duggen et al.’s
(2009) model, that is the suction of mantle plume material from below the Atlas towards the mantle wedge of the Gibraltar subduction
system by slab rollback induced toroidal mantle flow. Yet the second stage of this model, that is the migration of the Canary plume
across the Moroccan Atlas Mountains in the Northwest Africa, has
been controversial (Berger et al. 2010; Duggen et al. 2010), and
alternative mechanisms such as small-scale convection beneath the
High-Middle Atlas and Anti Atlas commencing in the mid Eocene
have been proposed (e.g. Kaislaniemi & van Hunen 2014).
It is well known that plumes are often captured into mid-ocean
ridges by large-scale plate-driven mantle flow (Weinstein & Olson
Figure 3. Plume capture by trench in rollback motion. The four top, middle and bottom frames represent the along, top and across slab views, respectively.
From left to right, elapsed times are 1280 s (10 Ma), 1960 s (15 Ma), 2580 s (20 Ma) and 3190 s (25 Ma) since experiment initiation. Note that the black dots
best seen in the top views are buoyant markers that had been laid on the glucose surface before initiation of the experiment.
1720
C.A. M´eriaux et al.
1989; Jellinek et al. 2003). Off-axis plumes in the proximity of
mid-ocean ridges may also migrate towards the ridge (Kincaid et al.
1995, 1996). However, the occurrence of plume migration/capture
under a continental lithosphere is not so common. The Yellowstone
plume however is an example (Camp 1995). The Canary ‘plume’
migration under the Atlas can be another example but with a difference. In this case, the plume migration would have been facilitated
by an abnormally thin lithosphere as shown by gravity and geoid
studies (e.g. Missenard et al. 2006; Fullea et al. 2010). The presence
of the Canary plume material below such a thinned lithosphere is
inferred from plume-related volcanism and geochemical signatures
of the lavas in the region together with tomographic low-velocity
anomalies. The geochemical data on basic-ultrabasic Middle Atlas lavas such as the Guilliz and Gourougou volcanic fields have
similarities with those of primitive lavas from the Canary Islands
(Duggen et al. 2009; Bosch et al. 2014). Tomographic studies show
low-velocity anomalies beneath an SW–NE trending corridor under
the Atlas and the Western Alboran Sea (Seber et al. 1996; Bezada
et al. 2013).
Up until now it has never been assessed if such a propagation
would dynamically be plausible. Here we use a force balance between the gravity and viscous forces to propose a simple scaling for
the propagation length L of a gravity current with time t in a viscous
medium along a channel (see eq. A4). The full details are given in
Appendix. When applied to the spreading of the Canary plume in an
Atlas channel, the plume could have spread ∼1870 and ∼2940 km
in 45 Ma along a corridor of 150 km width with a plume buoyancy
of 290 and 570 kg s−1 , respectively, assuming a density contrast of
40 kg m−3 , and a viscosity of the sublithospheric upper mantle of
4 × 1020 Pa s. These estimates are consistent with more than
1500 km of plume propagation proposed by Duggen et al. (2009),
and so the scenario should not be ruled out from a dynamical point
of view.
To conclude, this study revisited the model of Duggen et al.
(2009). From a dynamical point view, and using both a laboratory
model and a scaling argument, we show that such a model should
not be ruled out.
AC K N OW L E D G E M E N T S
The present work was supported by the Fundac¸a˜ o para a
Ciˆencia e a Tecnologia under the Project iPLUS-PTDC/CTEGIX/122232/2010. JCD and WPS were supported by a Discovery
Grant and Future Fellowship from the Australian Research Council.
The experiments were carried out in the Geodynamic Laboratory
of the School of Earth, Atmosphere and Environment at Monash
University. All the authors thank an anonymous reviewer and the
editor, Jeannot Trampert.
REFERENCES
Abdelwahed, M.F. & Zhao, D., 2007. Deep structure of the Japan zone, Phys.
Earth planet. Inter., 162, 32–52.
Bellahsen, N., Faccenna, C. & Funiciello, F., 2005. Dynamics of subduction and plate motion in laboratory experiments: insights into the
plate tectonics behavior of the Earth, J. geophys. Res., 110, B01401,
doi:10.1029/2004JB002999.
Berger, J., Li´egeois, J.P., Ennih, N. & Bonin, B., 2010. Flow of Canary
mantle plume material through a subcontinental lithospheric corridor
beneath Africa to the Mediterranean: COMMENT, Geology, 38(1), e202,
doi:10.1130/G30516C.1.
Bezada, M.J., Humphreys, E.D., Toomey, D.R., Harnafi, M., D´avila, J.M. &
Gallart, J., 2013. Evidence for slab rollback in westernmost Mediterranean
from improved upper mantle imaging, Earth planet. Sci. Lett., 368, 51–60.
Bosch, D., Maury, R.C., Bollinger, C., Bellon, H. & Verdoux, P., 2014.
Lithospheric origin for Neogene-Quaternary Middle Atlas lavas (Morocco): clues from trace elements and Sr-Nd-Pb-Hf isotopes, Lithos, 205,
247–265.
Camp, V.E., 1995. Mid-Miocene propagation of the Yellowstone mantle
plume head beneath the Columbia River basalt source region, Geology,
23(5), 435–438.
Davies, G.F., 1999. Dynamic Earth: Plates, Plumes and Mantle Convection,
Cambridge Univ. Press.
Duggen, S., Hoernle, K.A., Hauff, F., Klugel, A., Bouabdellah, M. & Thirlwall, M.F., 2009. Flow of Canary mantle plume material through a subcontinental lithospheric corridor beneath Africa to the Mediterranean,
Geology, 37, 283–286.
Duggen, S., Hoernle, K.A., Hauff, F., Kluegel, A., Bouabdellah, M. & Thirlwall, M.F., 2010. Flow of Canary mantle plume material through a subcontinental lithospheric corridor beneath Africa to the Mediterranean:
REPLY, Geology, 38(1), e203.
Druken, K.A., Kincaid, C., Griffiths, R.W., Stegman, D.R. & Hart, S.R.,
2014. Plume-slab interaction: the Samoa-Tonga system, Phys. Earth
planet. Inter., 232, 1–14.
Fullea, J., Fern`andez, M., Afonso, J.C., Verg´es, J. & Zeyen, H., 2010. The
structure and evolution of the lithosphere-asthenosphere boundary beneath the Atlantic-Mediterranean Transition Region, Lithos, 120(1), 74–
95.
Gutscher, M.A., Malod, J., Rehault, J.P., Contrucci, I., Klingelhoefer, F.,
Mendes-Victor, L. & Spakman, W., 2002. Evidence for active subduction
beneath Gibraltar, Geology, 30(12), 1071–1074.
Jellinek, A.M., Gonnermann, H.M. & Richards, M.A., 2003. Plume capture
by divergent plate motions: implications for the distribution of hotspots,
geochemistry of mid-ocean ridge basalts, and estimates of the heat flux
at the core-mantle boundary, Earth planet. Sci. Lett., 205(3), 361–378.
Kaislaniemi, L. & van Hunen, J., 2014. Dynamics of lithospheric thinning
and mantle melting by edge-driven convection: application to Moroccan
Atlas mountains, Geochem. Geophys. Geosyst., 15(8), 3175–3189.
Kincaid, C., Ito, G. & Gable, C., 1995. Laboratory investigation of the
interaction of off-axis mantle plumes and spreading centres, Nature,
376(6543), 758–761.
Kincaid, C., Schilling, J.G. & Gable, C., 1996. The dynamics of off-axis
plume-ridge interaction in the uppermost mantle, Earth planet. Sci. Lett.,
137(1), 29–43.
Kincaid, C., Druken, K.A., Griffiths, R.W. & Stegman, D.R., 2013. Bifurcation of the Yellowstone plume driven by subduction-induced mantle flow,
Nature Geo., 6, 395–399.
King, S.D. & Adam, C., 2014. Hotspot swells revisited, Phys. Earth planet.
Inter., 235, 66–83.
Lister, J.R. & Kerr, R.C., 1989. The propagation of two-dimensional and
axisymmetric viscous gravity currents at a fluid interface, J. Fluid Mech.,
203, 215–249.
Missenard, Y., Zeyen, H., Frizon de Lamotte, D., Leturmy, P., Petit, C.,
Sbrier, M. & Saddiqi, O., 2006. Crustal versus asthenospheric origin of
relief of the Atlas Mountains of Morocco, J. geophys. Res., 111(B3),
B03401, doi:10.1029/2005JB003708.
Obrebski, M., Allen, R.A., Xue, M. & Hung, S., 2010. Slab-plume interaction beneath the Pacific Northwest, Geophys. Res. Lett., 37, L14305,
doi:10.1029/2010GL043489.
Schellart, W.P., 2011. Rheology and density of glucose syrup and honey:
determining their suitability for usage in analogue and fluid dynamic
models of geological processes, J. Struct. Geol., 33(6), 1079–1088.
Schellart, W.P., Freeman, J., Stegman, D.R., Moresi, L. & May, D., 2007.
Evolution and diversity of subduction zones controlled by slab width,
Nature, 446(7133), 308–311.
Seber, D., Barazangi, M., Tadili, B.A., Ramdani, M., Ibenbrahim, A. &
Ben Sari, D., 1996. Three-dimensional upper mantle structure beneath
the intraplate Atlas and interplate Rif mountains of Morocco, J. geophys.
Res., 101(B2), 3125–3138.
Plume capture
Turner, S. & Hawkesworth, C., 1998. Using geochemistry to map mantle
flow beneath the Lau Basin, Geology, 26(11), 1019–1022.
Weijermars, R., 1986. Flow behaviour and physical chemistry of bouncing
putties and related polymers in view of tectonic laboratory applications,
Tectonophysics, 124(3), 325–358.
Weinstein, S.A. & Olson, P.L., 1989. The proximity of hotspots to convergent
and divergent plate boundaries, Geophys. Res. Lett., 16(5), 433–436.
Xue, M. & Allen, R.M., 2007. The fate of the Juan de Fuca plate: implications
for a Yellowstone plume head, Earth planet. Sci. Lett., 264(1), 266–276.
since the vertical relevant length scale is L (Lister & Kerr 1989).
Eq. (A2) then becomes
μLw
,
(A3)
t
using U ∼ L/t. Equating Fg and Fg gives the asymptotic spreading
relation
1/3
ρgq 2
t.
(A4)
L∼
μ0
Fv ∼
The current length increases linearly with time. Eq. (A4) gives the
order of magnitude of L as a function of time.
A P P E N D I X : G R AV I T Y C U R R E N T
SPREADING IN VISCOUS MEDIUM
WITHIN A CHANNEL
A simple scaling can be obtained for the spreading rate of a gravitydriven current of thickness h, length L, density difference ρ and
viscosity μi μ0 , μ0 being the ambient viscosity, that is fed at a
constant flow rate Q in a channel of width w, so that the flow rate
per unit width is q = Q/w. The analysis of the scales for the gravity
and viscous forces, Fg and Fv , respectively, gives
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article:
2
Movie S1. Movie_Gibraltar_19112014_GJI.mp4 (http://gji.
oxfordjournals.org/lookup/suppl/doi:10.1093/gji/ggv120/-/DC1)
U
Lw,
L
Please note: Oxford University Press not responsible for the content
or functionality of any supporting materials supplied by the authors.
Any queries (other than missing material) should be directed to the
corresponding author for the article.
ρgq 2 t 2 w
,
(A1)
Fg ∼ ρgh w ∼
L2
using volume conservation hL ∼ qt, where t is the time from initiation of the flow, and
Fv ∼ μ0
1721
(A2)