Download Depleted lithosphere, cold, trapped asthenosphere, and frozen melt

Earth and Planetary Science Letters 245 (2006) 289 – 301
www.elsevier.com/locate/epsl
Depleted lithosphere, cold, trapped asthenosphere, and frozen melt
puddles above the flat slab in central Chile and Argentina
Lara S. Wagner ⁎, Susan Beck, George Zandt, Mihai N. Ducea
Department of Geosciences, University of Arizona, 1040 E. 4th St., Tucson, AZ 85721, USA
Received 22 July 2005; received in revised form 2 February 2006; accepted 6 February 2006
Available online 19 April 2006
Editor: K. Farley
Abstract
Recent studies of the upper mantle above the flat slab in central Chile and Argentina indicate the seismic velocity structures in this
area are very different from those found in subduction zones with “normal,” steeper slab geometries. The mantle above the horizontal
section of the flat slab is characterized by low P-wave velocities, high S-wave velocities, and low Vp / Vs ratios. As the slab begins to
transition to a more normal dip to the south, the mantle above it changes as well. Above this “transition zone”, the mantle is
characterized by high P-wave velocities, high S-wave velocities, and generally high Vp / Vs ratios. Above this high velocity anomaly, in
the corner between the south-dipping slab of the transition zone and the east-dipping slab of the normal subduction zone, lies a small
shallow anomaly distinguished by its very low P- and S-wave velocities and its moderately high Vp / Vs ratio. We interpret these
anomalies to represent very depleted cold lithosphere, cooled trapped asthenosphere, and frozen pooled melt, respectively. Recent
research looking at the elastic properties of cratonic xenoliths indicate that a low Vp, high Vs, and low Vp / Vs signature is indicative of
cratonic lithosphere. This suggests that the material above the flat slab may be ancient Laurentian lithosphere, or other melt-depleted,
dry lithospheric material. The presence of frozen asthenosphere to the south of the flat slab may indicate that large volumes of
asthenosphere were displaced in advance of the eastward progression of the flattening slab between 10 and 2 Ma, forcing a change in
flow patterns from trench normal to SSW around the south eastern corner of the flattening slab. As flattening progressed this
asthenosphere became trapped above the transition zone. Eventually, temperatures dropped sufficiently to freeze the last remaining
pocket of extracted melt which had collected in the corner of the descending slab.
© 2006 Elsevier B.V. All rights reserved.
Keywords: tomography; flat-slab; harzburgite; Chile; Argentina
1. Introduction
Beneath central Chile and western Argentina, the
subducting Nazca slab has an unusual geometry. Between ∼29°S and 33°S, the oceanic plate descends to
100 km depth and then flattens, traveling horizontally for
⁎ Corresponding author. Fax: +1 520 621 2672.
E-mail address: [email protected] (L.S. Wagner).
0012-821X/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsl.2006.02.014
hundreds of kilometers before resuming its descent into
the mantle (Fig. 1). This segment of flat slab subduction
is spatially correlated with the cessation of arc volcanism
and the inland basement cored uplifts of the Sierras
Pampeanas. Jordan and Allmendinger [1] and others
have suggested that the central Chilean flat slab and the
Sierras Pampeanas are a modern analogue of a period of
flat subduction of the Farallon plate beneath the western
United States during Laramide deformation. While this
comparison has its limitations [2], some insights into the
290
L.S. Wagner et al. / Earth and Planetary Science Letters 245 (2006) 289–301
Fig. 1. Map of study area. Diamonds show the locations of the CHARGE seismic stations used in the tomographic inversion and stars show the
locations of regional array stations reporting to the ISC that are also included in the inversion. Triangles show the locations of active volcanos.
Contour lines show the depth to the top of the subducted Nazca plate according to Cahill and Isacks [4]. Grey shaded lines show the projected location
of the subducted Juan Fernandez Ridge at specified times (0, 10, and 12Ma).
correlation between flat or shallow subduction and inland basement cored uplifts can still be made.
In order to study the mantle above this unusual slab
geometry, we performed an inversion of P, S, and S–P
seismic travel times from regional events recorded at
stations from the Chile Argentina Geophysical Experiment (CHARGE). The experiment consisted of 22
broadband seismic stations deployed for 18months beginning in late 2000. This data was complemented by
data collected from regional arrays that report data to the
International Seismic Center (ISC). The additional data
improves resolution, but does not change any of the
major results from the inversions using only CHARGE
stations. The tomographic results and the locations of the
CHARGE stations are shown in Fig. 2. Further information on this inversion and associated resolution tests
can be found in Wagner, et al. [3].
Above the flat slab and below the crust in central Chile
and Argentina we found a sliver of material with low Pwave velocities, high S-wave velocities, and anomalously
low Vp /Vs ratios (anomaly A, Figs. 2 and 3) [3]. The
northern and eastern boundaries of this anomaly are poorly
constrained, but the anomaly appears to be continuous
across the top of the flattened portion of the slab. South of
the flat slab, the downgoing plate is either bent or torn to
accommodate the change in dip from horizontal to nearly
30°. There is no direct evidence for a tear in this area, and
previous authors have modeled this transition as a steep
bend (though the lack of deep seismicity in this area makes
the determination of exact structure impossible) (e.g.,
Cahill and Isacks [4]). The southern edge of the low P, high
S, low Vp /Vs anomaly appears to lie directly along the
edge of this bend (Fig. 3). As soon as the slab is no longer
completely horizontal to the south, the low Vp /Vs anomaly
is no longer visible directly above the slab. In its place,
above this “transition zone” from flat to normally dipping
slab, we find high P-wave velocities and generally high Swave velocities and Vp /Vs ratios (anomaly B, Figs. 2 and
3). Above the westernmost portion of this high velocity
anomaly at 65km depth, in the corner between the east
verging dip and the south verging dip of the Nazca plate,
lies a small region with very low P- and S-wave velocities
L.S. Wagner et al. / Earth and Planetary Science Letters 245 (2006) 289–301
291
Fig. 2. Tomography results for Vp, Vs, and Vp / Vs at 65 and 85km depth from Wagner et al. [3]. Green stars show CHARGE seismic station locations.
Red triangles indicate active volcanos. Dashed lines are contours for slab depth from Cahill and Isacks [4]. Boxes show areas used in forward
modeling. A: low P, high S, low Vp / Vs area; B: high P, high S, high Vp / Vs area; C: low P, low S, high Vp / Vs bulls-eye. Black line indicates the location
of the north–south cross-section in Fig. 3. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version
of this article.)
and relatively high Vp /Vs ratios (anomaly C, Figs. 2 and 3).
While this area is just south of the actual flat slab, it is still
characterized by a lack of surface volcanism. Normal arc
volcanism resumes just to the south of this low Vp, low Vs,
high Vp /Vs bulls-eye (Fig. 2).
Interpreting these anomalies would ideally help decipher the compositions, states, and temperatures that are
likely to give rise to the observed seismic velocities, and
also help explain how they have been produced. In this
paper, we discuss a range of possibilities, explore their
tectonic implications, and propose that the most likely
explanation involves the presence of ancient continental
lithosphere and a complicated subduction history.
2. Tectonic history
While there is ongoing debate about many of the
details of the regional geology in our study area, a general
tectonic history seems to be agreed upon. This area of
Chile and Argentina is a conglomeration of accreted
terranes with generally north south trending boundaries
[5] (Fig. 4). The first terrane to accrete onto the western
margin of Gondwana was the Pampian terrane after the
complete subduction of the Puncoviscan ocean under the
Rio de la Plata craton in the Early Cambrian [6]. By the
Early Ordovician, subduction of the Iapetus Ocean began
beneath the Pampian terrane, giving rise to Famatinian
arc volcanism. There has some debate whether the Famatina terrane was accreted or was simply a product of
this volcanism [6,7], but most studies indicate that Famatinian volcanism has continental (rather than island)
arc signatures (e.g., Saavedra et al. [8]). By the Late
Ordovician, the subduction of the Iapetus Ocean was
complete, resulting in the accretion of the Cuyania/Precodilleran terrane. This terrane is believed to have originated from Laurentia, with numerous data showing
292
L.S. Wagner et al. / Earth and Planetary Science Letters 245 (2006) 289–301
Fig. 3. North–south cross-section of the tomography results. Colors for Vp and Vs are in percent deviation. Contours for Vp are every 0.2km/s.
Contours for Vs are every 0.1 km/s. Vp / Vs colors show absolute Vp / Vs values. Ovals indicate the locations of the anomalies indicated in map view in
Fig. 2. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Grenvillian ages for basement rocks, both from the Precordillera and from the Pie de Palo range [5,9–14]. The
final terrane to accrete was the Chilenia in the Devonian.
Little is known about the origin of the Chilenia terrane, as
most of its original basement rocks have been affected by
extensive volcanism and metamorphism. It is also unclear
whether the subduction that resulted in the accretion of
Chilenia was west-verging or east-verging [5,7], but more
recent works [7,12] seem to prefer the former. This would
imply that no subduction has occurred beneath the Cuyania/Precordilleran terrane at least since the Ordovician.
The locations of the terrane boundaries are also
somewhat uncertain. Perhaps the least controversial
boundary lies between the Chilenia and Cuyania terranes, along the Calingasta–Iglesia valley west of the
Precordillera [5] (Fig. 4). The boundary between the
Cuyania terrane and the Pampian terrane is now believed to lie along the SSE trending Sierras de Valle
Fertil [7,15], though some sort of ancient suture is
visible for a short distance along the western edge of
the Pie de Palo range extending SSW towards the
Precordillera. Originally believed to be the boundary
between the Precordillera and the Pampian terrane, this
suture is now suggested to be evidence of Middle
Proterozoic subduction that resulted in the accretion of
the Pie de Palo range to the Precordillera prior to rifting
from Laurentia [16,17]. These terrane boundaries, between Chilenia and Cuyania, and Cuyania and Pampea,
are the most relevant to this paper, as the tomography
results do not sample far enough east to study the
boundary between the Pampean terrane and the Rio de
la Plata craton.
Following these accretionary events, the Gondwanian
cycle (Early Carboniferous to Early Cretaceous) was
characterized by extension and relative stability within
the continent [18]. Subduction of the proto-Pacific plate
beneath the western margin of Gondwana resulted in the
formation of cordilleran type batholiths during the Carboniferous [19]. This era ended with the rifting of Gondwana and the opening of the Atlantic, which heralded the
beginning of the Andean cycle (Early Cretaceous to
present). This cycle was characterized by several periods
L.S. Wagner et al. / Earth and Planetary Science Letters 245 (2006) 289–301
293
Fig. 4. Geological features in the study area. The white diamonds show CHARGE seismic stations. Red triangles show active volcanos. Purple lines
indicate terrane boundaries. Shaded areas indicate major geologic features. Black contour lines show the depth to the subducted Nazca plate according
to Cahill and Isacks [4]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
of compression, including the most recent beginning in
the Tertiary [19].
Development of the current unusual slab geometry in
this area began during the Miocene. Based on the timing
of the eastward progression and eventual cessation of arc
volcanism and the arrival of the subducting Juan Fernandez ridge track to these latitudes, flattening in central
Chile is likely to have begun between 20 and 18Ma, with
the most dramatic eastward progression occurring between 10 and 2Ma [11,20]. Most volcanism ended between 6 and 4.8Ma, with only a small amount of
volcanism from the Sierra de San Luis occurring later
around 1.9Ma [21,22] indicating the slab's current configuration has likely been stable for at least the past
∼2–5Ma.
3. Compositions, states, and temperatures
Determining the composition, state, and temperature
of a material with specific seismic characteristics can be
difficult, but some constraints can be made. Accurate
analysis requires a good understanding of error bounds
on both the tomographic results and the thermodynamic
and experimental results that link petrology and temperature to elastic properties.
In order to investigate the error bounds of the tomographic results, we have employed a forward modeling
approach to test the range of P and S velocities that can
explain the travel times for the three anomalies discussed
here. This method involves creating a 3-D velocity model
through which travel times are calculated for rays used in
the tomographic inversion. Event locations are fixed at
the final locations determined during the tomography. We
then identify an area of roughly the size and shape of the
anomaly of interest, and vary the velocities within that
anomaly, keeping the rest of the model constant. Travel
times and errors between observed and predicted travel
times are calculated for each velocity model and are
plotted in Figs. 5, 6, 7 and 8.
For the low Vp / Vs region (anomaly A) above the flat
slab, we created a 40km thick layer in a rectangular
shape above the flat slab (see Fig. 2), using the starting
model from the tomography for the rest of the volume.
Results of this modeling are shown together with possible
compositions (to be discussed later) in Figs. 5 and 6. We
find that S-wave velocities in this area range from 4.6–
294
L.S. Wagner et al. / Earth and Planetary Science Letters 245 (2006) 289–301
Fig. 5. Results for anomalies A, B, and C (white stars), including innermost contours from forward modeling results (see Figs. 6, 7, and 8).
Endmember peridotite and other compositions plotted are from Hacker and Abers [25], except where explicitly specified. Size indicates temperature,
and color indicates composition.
4.7km/s and the Vp / Vs ratios range from 1.64–1.74, with
the best fit located at Vs = 4.65km/s and Vp / Vs = 1.70.
Decreasing the thickness of this layer results in higher S
velocities and lower Vp / Vs ratios, which would make the
result more unusual, and the interpretation even more
difficult.
Modeling the anomalies further to the south is more
complicated because rays sampling these regions are
more likely to traverse other anomalies either before or
after crossing the feature of interest. In order to account
for this, we use the final tomography results as a starting
model, and then vary the anomaly of interest within that
model. The results for the high velocity anomaly (anomaly B) and the low velocity bulls-eye (anomaly C) are
shown in Figs. 7 and 8, respectively. For the high velocity
anomaly above the transition zone, we find that the Swave velocities are likely between 4.6 and 4.7km/s, and
that the Vp / Vs ratios are most likely between 1.72 and
1.84 with the best fit at 4.65km/s and 1.775 for Vs and
Vp / Vs, respectively. For the low velocity bulls-eye, we
find S-wave velocities between 4.0 and 4.2km/s, and Vp /
Vs ratios between 1.72 and 1.80. The best fitting result for
this anomaly was 4.1km/s for Vs and 1.76 for Vp / Vs.
Errors on the petrologic and thermodynamic end are
difficult to quantify. Discrepancies exist between studies
[23,24] (Figs. 5, 6 and 7), and in general, error bars are
not available. We present mostly results from Hacker
and Abers [25], except where other results are available
(Figs. 5, 6 and 7).
While a definitive answer to the question of composition in each of these regions is difficult, some constraints from Figs. 5, 6, 7 and 8 can be made. We begin with
a discussion of anomalies B and C because their
explanations are more straightforward. To study the
high velocity anomaly B in the slab dip transition zone,
we assumed an average pressure of 3GPa and an average
temperature of 500°C (since this material has likely been
cooling during the past 2Ma of flat slab subduction). We
L.S. Wagner et al. / Earth and Planetary Science Letters 245 (2006) 289–301
295
Fig. 6. Results of the forward modeling for the low Vp, high Vs, low Vp / Vs anomaly A. The white star indicates the lowest residual. The yellow
symbols show velocities from Hacker and Abers [25] for specified compositions at 500°C. Green circles are for pure forsterite, blue circles are for
pure enstatite, also from Hacker and Abers [25]. Stars indicate velocities from Ji and Wang [23] for a range of compositions between pure enstatite
and pure forsterite. All values are calculated for 3GPa at specified temperatures. (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of this article.)
follow the Hacker, et al. [24] approach of 4 possible
starting compositions, MORB, lherzolite, depleted
lherzolite, and harzburgite. All of these compositions
are assumed to be hydrated, but the percent hydration
depends on the phases present at a given pressure/
temperature condition. At 500°C and 3GPa, these compositions would manifest themselves as zoisite eclogite
(see Table 3 in Hacker et al. [24] for exact composition),
chlorite serpentinite wehrlite (I) (chl16at g2ol57cpx26),
chlorite serpentinite wehrlite (II) (ol89at g2chl5cpx3),
and serpentine chlorite dunite (ol94at g2chl4), respectively. Increasing the temperature 100°C would not
change these compositions, and most of these reactions
are relatively pressure insensitive. Thermodynamic
calculations from the spreadsheet associated with Hacker
and Abers [25] provide the velocities for these compositions shown in yellow in Fig. 7. These results indicate that
the high velocity anomaly could be modeled by any cool
hydrated peridotite composition, with the marginally best
fit going to depleted lherzolite.
We repeat this process for the low velocity bulls-eye at
65km depth (anomaly C). We assume a pressure of 2GPa
and a temperature of 350°C, which would result in
jadeite–lawsonite blueschist (see Table 3 in Hacker, et al.
[24] for exact composition), serpentinite (I) (atg4chl20cpx33br43), serpentinite (II) (atg7br80chl11cpx5), and
serpentinite chlorite brucite (atg7br87chl6) for MORB,
lherzolite, depleted lherzolite, and harzburgite, respectively. The velocities for these compositions are shown in
red in Fig. 8. The previously calculated velocities for the
296
L.S. Wagner et al. / Earth and Planetary Science Letters 245 (2006) 289–301
Fig. 7. Results of the forward modeling for the high Vp, high Vs, high Vp / Vs anomaly B. The white star indicates the lowest residual. The yellow
symbols show velocities from Hacker and Abers [25] for specified compositions at 500 °C. Green circles are for pure forsterite, blue circles are for
pure enstatite, also from Hacker and Abers [25]. Stars indicate velocities from Ji and Wang [23] for a range of compositions between pure enstatite
and pure forsterite. All values are calculated for 3 GPa at specified temperatures. (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of this article.)
phases at 500°C and 3GPa are also plotted in Fig. 8, in
yellow as in Fig. 7. In this case, the notably best fit goes
to the MORB composition, with the depleted end-members having distinctly lower S-wave velocities than observed in the tomographic results. The higher pressure
and temperature results do not match the observed velocity, reaffirming the assertion that this material must be
especially cold.
The low P velocities, high S velocities, and low Vp / Vs
ratios above the flat slab in anomaly A, on the other hand,
cannot be modeled using these hydrated compositions
(Figs. 5 and 6). Instead, we look at the velocities of pure
end-member compositions (Fig. 5) to determine what the
range of possible compositions might be for these high
shear velocities and low Vp / Vs ratios. Looking at the
trends shown in Figs. 5 and 6, we infer that the material
above the flat slab is probably cold (because S-wave
velocities increase and Vp / Vs ratios decrease with
decreasing temperature), dry (because the addition of
even a small amount of water dramatically increases the
Vp / Vs ratio), and depleted (because of the low S velocity
and high Vp / Vs ratio of clinopyroxene and the iron endmembers of all three peridotite compositions).
Recent work by Rossi and Abers [26] indicates a
similar low Vp / Vs (1.6–1.7) structure above the subducting plate beneath Alaska. They argue that this Vp / Vs ratio
L.S. Wagner et al. / Earth and Planetary Science Letters 245 (2006) 289–301
297
Fig. 8. Results of the forward modeling for the low Vp, low Vs, high Vp / Vs anomaly C. The white star indicates the lowest residual. The red and yellow
symbols show velocities from Hacker and Abers [25] for specified compositions. Red shows results at 350 °C and 2GPa, and yellow shows results at
500°C and 3GPa. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
is too low to be explained by any peridotitic composition
(based on Hacker and Abers [25]), and suggest that the
only mineralogy that has a Vp / Vs ratio b 1.7 at mantle
conditions is quartz. They propose that this addition of
silica could be achieved by metasomatism, or by the
subduction of felsic material. For reference, we plot the
three phases of quartz that might be found at upper
mantle conditions in Fig. 5. Alpha and β quartz are stable
at lower pressures (below 2–3GPa), with β quartz being
the high temperature phase. Coesite is stable at higher
pressures (between 2–3GPa, depending on temperature)
[27]. Only α quartz has a Vp / Vs ratio significantly below
1.7 at 3GPa, but the low Vp / Vs ratios are accompanied by
very low S-wave velocities. While Rossi and Abers do
not specify absolute S-wave velocities in Alaska, we find
high S-wave velocities associated with our low Vp / Vs
ratios, limiting the amount of α quartz that can be present
in our mantle material if the absolute velocities are to be
accounted for. It is also possible that at depths closer to
the flat slab, α quartz may be transitioning to coesite [27],
which has a Vp / Vs ratio and S-wave velocity comparable
to harzburgitic compositions such as forsterite and
enstatite, making its presence hard to discriminate.
Given the elastic properties of mineral compositions
provided by Hacker and Abers [25], between 20% and
30% pure α quartz mixed with pure forsterite would
match our observed velocities. However, there are many
potential issues with how to introduce pure α quartz into
a harzburgitic, cold, depleted mantle composition.
Subducted continental crustal material would also
introduce Fe, Ca and K rich compositions, which tend
to have higher Vp / Vs ratios. More importantly, any
crustal material would bring significant quantities of
water into the mantle. At lower temperatures, this water
would likely react with the peridotites in the mantle to
form serpentinites (which have high Vp / Vs ratios), and at
higher temperatures, this water would likely induce
melting. It is unclear how it would be possible to isolate
298
L.S. Wagner et al. / Earth and Planetary Science Letters 245 (2006) 289–301
large quantities of α quartz out of crustal material before
or during subduction in order to lower the mantle Vp / Vs
to match the observations of this study and that of Rossi
and Abers.
Whether or not α quartz is present in any significant
quantity in our low Vp / Vs layer, at least 70% of this
material must still be cold, dry, and depleted, as explained
earlier. Looking at the seismic velocities presented in
Hacker and Abers [25] (Fig. 6), it appears, assuming
quartz is not present, that the material responsible for the
low Vp / Vs anomaly would have to be dominated by pure
enstatite in order to approach the observed low Vp / Vs
ratios. However, work by Ji and Wang, [23] done at room
temperature (0°C) and 3 GPa indicates that the Vp / Vs
ratios for both enstatite and forsterite may be lower than
reported by Hacker and Abers [25], suggesting that any
harzburgitic mixture of forsterite and enstatite could
produce the observed seismic anomalies. This finding is
consistent with the work of Lee [28] who showed that
Mg# (defined as 100 * Mg / Mg + Fe) is directly proportional to Vs, and inversely proportional to Vp / Vs. He finds
little correlation between Mg# and Vp, although Vp does
appear related to olivine mode. Lee [28] points out that
while both temperature and composition could be
responsible for an observed change in Vs, Vp / Vs is far
less sensitive to temperature than it is to composition,
making it an ideal indicator for mantle composition. Low
Vp / Vs ratios, together with high Vs, would indicate depleted, pyroxene rich, Mg rich, mantle [28].
4. Tectonic implications
The low temperatures required by both the observed
velocity structures and the absence of volcanism indicate
that normal asthenospheric corner flow is not currently
present in any of these areas, probably because this flow
has been pinched out by slab flattening [29]. Since the
cessation of volcanism is believed to indicate the timing
of slab flattening, the material in this area has probably
been trapped there at least since volcanism ceased some
2–5Ma. Once trapped, the material is continuously
cooled by the subducted cold oceanic lithosphere.
Normally at these depths, the downgoing oceanic
crust would be dehydrating, releasing water into the
overlying asthenosphere, producing melt. In the completely flat portion of the slab below the low Vp / Vs ratios,
the slab is probably not currently dehydrating, nor is it
likely to have dehydrated since volcanism ended, or we
would see evidence of water in the seismic velocities of
the overlying material. The only way to remove water
from a hydrated mineralogy (in the absence of melt) is to
increase the temperature, but because the temperatures
above the slab must be decreasing, any water released by
the slab would stay in the overlying rocks in the form of
serpentinite and manifest itself in the tomographic results
as zones with higher Vp / Vs ratios.
While a non-dehydrating slab may help explain the
cold and dry character of the trapped mantle material, it
does not explain the depleted signature of this material.
The seismic anomaly above the flat slab suggests that this
material contains little if any clinopyroxene, and virtually
none of the iron end-members of olivine and orthopyroxene. Most likely, this zone is composed of some
mixture of enstatite and forsterite, though the exact
composition is uncertain. Such a high level of depletion
suggests that the material may be ancient continental
lithosphere or depleted oceanic lithosphere (residual
oceanic mantle column). Kopylova, et al., [30] found that
the Archaen cratonic lithosphere beneath the Canadian
Slave province is characterized in some areas by low Pwave velocities, high S-wave velocities, and low Vp / Vs
ratios, based on both calculated and measured elastic
properties of local mantle xenoliths. They suggest that
low Vp, high Vs, and low Vp / Vs are the expected elastic
properties of any highly depleted cratonic lithosphere.
This is consistent with the work of Lee [28] who finds
increased Mg# and corresponding increased Vs and
decreased Vp / Vs in cratonic peridotite xenoliths.
Seismic studies of cratons are less conclusive on this
issue. Many look at only S-wave velocities, using surface
wave tomography (e.g., Morelli and Danesi [31] in Antarctica, Fishwick, et al. [32] in Australia, Feng, et al. [33]
in South America/Brazil, and Nettles and Dziewonski
[34] in North America). These studies show increases in
S-wave velocities beneath these cratons, but the Vp / Vs
ratios are unknown. Because temperature can mimic the
effect of composition on Vs, Vp / Vs is an important discriminant in areas where the thermal gradient is
uncertain. The one paper with results for both Vp and
Vs is by Fouch, et al. [35] who investigate the lithosphere
beneath the Kaapvaal craton using relative teleseismic
travel times. They find increases in both P and S velocities, but cannot explicitly calculate Vp / Vs due to the
relative nature of the data. The study of Niu et al. [36]
infers relative Vp / Vs ratios beneath the Kaapvaal craton
by looking at S–P times, and finds evidence for comparatively low Vp / Vs in the mantle. The low Vp, high Vs,
and low Vp / Vs above the flat slab in central Chile and
Argentina correlate well with the values predicted by
both Kopylova, et al. [30] and Lee [28] for Archaen
cratonic lithosphere.
The origin of the depleted lithosphere cannot be uniquely determined. It may be oceanic or continental in
origin, though because of difficulties explaining how
L.S. Wagner et al. / Earth and Planetary Science Letters 245 (2006) 289–301
Fig. 9. Vp / Vs results at 85km depth from the tomographic inversion in
Wagner et al. [3] overlain by the projected current location of the Juan
Fernandez Ridge (grey line), and the location of the terrane boundaries
(red lines). Green stars indicate CHARGE seismic station. (For
interpretation of the references to colour in this figure legend, the
reader is referred to the web version of this article.)
oceanic lithosphere could become associated with the
overriding plate, we are currently hypothesizing this
material is more likely to be continental. There are a
number of terrane boundaries transecting this study area,
all of which trend roughly north south (Fig. 4). These
terranes are believed to have accreted by the Devonian,
though the direction of subduction remains unclear in
some cases. If the observed anomalies are indicative of
original lithosphere, we might expect to see evidence of
these terrane boundaries in the tomography, especially in
areas under which subduction has occurred. We see no
strong evidence of east–west changes in the velocity
anomalies (Fig. 9), which may indicate that the terrane
boundaries at the surface do not accurately represent
these terrane boundaries at depth, or that all of the terranes were underlain by similar cratonic lithosphere.
Because the amount of recent volcanism above this area
is insufficient to create a new depleted lithosphere, ancient lithosphere is the most likely explanation for the
observed anomalies. Nd isotopic data available on Miocene–Quaternary mafic volcanic rocks from the area
[37,38] are consistent with the presence of relatively old
(500Ma) continental mantle lithosphere beneath the
studied area; however, the interpretation of the isotopic
data is non-unique (e.g., the volcanic rocks may have
been modified by crustal contamination). The only terrane in the area whose origin is constrained is the Cu-
299
yania terrane, which may have been part of Laurentia
until the Ordovician. It is possible, then, that the observed
seismic anomalies above the flat slab are indicating the
presence of ancient Laurentian lithosphere. We recognize
that Laurentian lithosphere would likely be of Grenvillian
age (1.1Ga), which is significantly younger than the age
of the Archean cratons mentioned earlier. However, it is
uncertain whether there would be a significant difference
in seismic velocities between Archean cratons and Grenville-age cratons. Evidence from Raleigh wave velocities
indicates the presence of a high S-wave velocity Grenvillian keel beneath the eastern United States [39] which
could be consistent with our observed seismic velocities.
Interpreting the velocity anomalies above the transition zone is less complicated. Unlike the completely flat
portion of the Nazca plate, the trapped mantle above this
area (anomaly B) need not be depleted, and the slab in the
transition zone may still be dehydrating. The subduction
zone compositions proposed by Hacker, et al. [24] all
contain between 4% and 15% water with the exception of
zoisite eclogite, the MORB composition at 3GPa and
500°C. The simplest explanation for the high velocity
anomaly is a body of cooled asthenosphere of uncertain
composition. The fate of any ancient lithosphere that may
have been present in this area cannot be uniquely
determined. The low velocity bulls-eye above the high
velocity anomaly (anomaly C) can be explained as a
small region of extracted cooled melt with a MORB
composition. It is interesting that this anomaly lies directly beneath the Cortaderas ophiolite complex [12].
These Precambrian, Ordovician, and Silurian mafic rocks
have generally EMORB affinities, but it is difficult to
explain how our comparatively large anomaly at 60km
depth might be associated with the sliver of ophiolites
observed at the surface. We discuss our preferred explanation for how these compositions may have evolved in
the following section.
5. Proposed regional history
We propose the following progression for the formation of these seismic velocity anomalies in the upper
mantle. Originally, the mantle directly below the crust in
the area currently characterized by a cessation of arc
volcanism was made of a relatively uniform ancient
depleted lithosphere. As the flattening of the Nazca plate
progressed eastward between 10 and 2Ma, a large volume
of asthenosphere was displaced from the area east of the
flattening slab. This asthenosphere would most likely
escape to the south, because the transition to normal dip to
the north is much more gradual, and would accommodate
much less material. We propose that the displaced
300
L.S. Wagner et al. / Earth and Planetary Science Letters 245 (2006) 289–301
asthenosphere flowed around the southeastern corner of
the flat slab, below the ancient depleted lithosphere, and
into the corner now characterized by the low P, low S, high
Vp / Vs bulls-eye (anomaly C). There, it would join with
the normal asthenospheric corner flow responsible for arc
volcanism, and would eventually be moved out of the area
along the top of the normally subducting plate. During this
initial displacement phase, as the slab was still in the
process of flattening, the movement of this additional
material, added to the pre-existing corner flow, contributed to a delamination of the ancient lithosphere up to the
base of the crust beneath the main cordillera. The level of
this delamination appears to become shallower west of
San Juan, as the depleted lithosphere appears to be intact
east of San Juan at 85km depth.
As flattening progressed, the slab beneath what is now
the transition zone became shallow enough to prevent
new asthenosphere from entering. The asthenosphere
above this part of the slab became trapped and started to
cool. Remnant hydration and melt collected and pooled
directly below the crust in the corner, resulting in the low
velocity bulls-eye. Flow patterns of the asthenosphere's
movement around the southeastern corner of the flat slab
became frozen into the mantle fabric, and are visible
today in the form of shear wave splitting directions [40].
As flattening ceased to progress further east, additional
asthenosphere was no longer displaced to the south, and
normal corner flow continued south of the transition
zone, resulting in the stable magmatic arc that is still
active today. The final result is preserved lithosphere
above the horizontal portion of the flat slab, injected
frozen asthenosphere in the transition zone, pooled frozen
melt in the corner between the south dipping and east
dipping portions of the slab, and normal asthenospheric
corner flow to the south below the modern active arc.
6. Conclusions
Evaluation of errors on both the seismic tomographic
results and the petrologic and thermodynamic constraints
leads us to the following major conclusions:
1) Forward modeling tests of seismic velocity structures
in the region of the central Chile/Argentina flat slab
confirm the tomographic results of Wagner, et al. [3],
finding low Vp, high Vs, and low Vp / Vs (anomaly A)
above the horizontal portion of the flat slab, high Vp,
high Vs, and high Vp / Vs (anomaly B) in the transition
zone immediately to the south of this portion, and low
Vp, low Vs, and high Vp / Vs (anomaly C) in the corner
between the south and east dipping portions of the
downgoing Nazca plate.
2) The seismic signature of low Vp, high Vs, and low Vp /
Vs (anomaly A) found above the horizontal portion of
the central Chilean flat slab is consistent with the
highly depleted, cold, Mg rich compositions found in
cratonic lithosphere. While the precise composition
and origin of the material cannot be uniquely determined, it is possible that this material observed here
is ancient Laurentian lithosphere, though the possibility remains that this lithosphere is oceanic.
3) The high velocity anomaly (anomaly B) to the south
of the flattest portion of the slab is most likely cooled
asthenosphere that was displaced in advance of the
eastward progression of the flattening slab between 10
and 2Ma, and eventually trapped by the increased
shallowing of the slab in the transition zone.
4) The low velocity, high Vp / Vs anomaly (anomaly C)
at 65km depth in the corner between the east and
south dipping portions of the Nazca plate is probably
caused by a frozen pool of extracted melt with a
MORB-like composition.
Acknowledgements
This material is based on work supported by the
National Science Foundation under Grant No.
EAR9811870 at the University of Arizona. Seismic
data was acquired with instrumentation provided
by the IRIS/PASSCAL program. We acknowledge
INPRES (Argentina), the University of Chile, and the
University of San Juan for their logistical support and
assistance in the field. We thank the entire CHARGE
working group for their many efforts and contributions.
We would also like to thank our anonymous reviewer
and Sue Kay for their insightful and helpful comments.
This material is based upon work supported under a
National Science Foundation Graduate Research Fellowship to L. Wagner.
References
[1] T.E. Jordan, R.W. Allmendinger, The Sierras Pampeanas of
Argentina — a modern analog of rocky-mountain foreland
deformation, Am. J. Sci. 286 (1986) 737–764.
[2] J. Saleeby, Segmentation of the Laramide slab: evidence from
the southern Sierra Nevada region, Geol. Soc. Amer. Bull. 115
(2003) 655–668.
[3] L.S. Wagner, S. Beck, G. Zandt, Upper mantle structure in the
south central Chilean subduction zone (30° to 36°S), J. Geophys.
Res. 110 (2005), doi: 10.1029/2004JB003238.
[4] T. Cahill, B.L. Isacks, Seismicity and shape of the subducted Nazca
plate, J. Geophys. Res.—Solid Earth 97 (1992) 17503–17529.
[5] V.A. Ramos, T.E. Jordan, R.W. Allmendinger, C. Mpodozis, S.M.
Kay, J.M. Cortes, M. Palma, Paleozoic terranes of the central
Argentine–Chilean Andes, Tectonics 5 (1986) 855–880.
L.S. Wagner et al. / Earth and Planetary Science Letters 245 (2006) 289–301
[6] C.W. Rapela, R.J. Pankhurst, C. Casquete, E. Baldo, J. Saavedra,
C. Galindo, Early evolution of the Proto-Andean margin of South
America, Geology 26 (1998) 707–710.
[7] R.A. Astini, The Early Paleozoic evolution of the Argentine
Precordillera as a Laurentian rifted, drifted, and collided terrane: a
geodynamic model, Geol. Soc. Amer. Bull. 107 (1995) 253–273.
[8] J. Saavedra, A. Toselli, J. Rossi, I. Pellitero, F. Durand, The Early
Paleozoic magmatic record of the Famatina System: a review, in:
R.J. Pankhurst, C.W. Rapela (Eds.), The Proto-Andean Margin of
Gondwana, The Geological Society, London, 1998, pp. 283–296.
[9] W.A. Thomas, R.A. Astini, The Argentine Precordillera: a
traveler from the Ouachita embayment of North American
Laurentia, Science 273 (1996) 752–756.
[10] A.E. Rapalini, R.A. Astini, Paleomagnetic confirmation of the
Laurentian origin of the Argentine Precordillera, Earth Planet.
Sci. Lett. 155 (1998) 1–14.
[11] S.M. Kay, J.M. Abbruzzi, Magmatic evidence for Neogene
lithospheric evolution of the central Andean gflat-slabh between
30°S and 32°S, Tectonophysics 259 (1996) 15–28.
[12] J.S. Davis, S.M. Roeske, W.c. McClelland, S.M. Kay, Mafic and
ultramafic crustal fragments of the southwestern Precordillera
terrane and their bearing on tectonic models of the Early
Paleozoic in western Argentina, Geology 28 (2000) 171–174.
[13] L. Dalla Salda, C. Cingolani, R. Varela, Early Paleozoic orogenic
belt of the Andes in southwestern South America: result of
Laurentia–Gondwana collision? Geology 20 (1992) 617–620.
[14] B. Baldis, M. Beresi, O. Bordonaro, A. Vaca, The Argentine
Precordillera as a key to Andean structure, Episodes 7 (1984) 14–19.
[15] S.M. Kay, S. Orrell, J.M. Abbruzzi, Zircon and whole rock Nd–
Pb isotopic evidence for a Grenville age and Laurentian origin for
the basement of the Precordillera in Argentina, J. Geol. 104
(1996) 637–648.
[16] V.A. Ramos, M. Escayola, D.I. Mutti, G.I. Vujovich, Proterozoic–
Early Paleozoic ophiolites of the Andean basement of southern South
America, in: Y. Dilek, E.M. Moores, D. Elthon, A. Nicolas (Eds.),
Ophiolites and Oceanic Crust: New Insights from Field Studies and
the Ocean Drillilng Program, Geological Society of America Special
Paper, Boulder, Colorado, 2000, pp. 331–349.
[17] G.I. Vujovich, S. Kay, A Laurentian? Grenville-age oceanic arc/
backarc terrane in the Sierra de Pie de Palo, western Sierras
Pampeanas, Argentina, in: R.J. Pankhurst, C.W. Rapela (Eds.),
The Proto-Andean Margin of Gondwana, The Geological
Society, London, 1998, pp. 159–179.
[18] C. Mpodozis, S.M. Kay, Late Paleozoic to Triassic evolution of
the Gondwana margin: evidence from Chilean Frontal Cordilleran batholiths (28°S–31°S), Geol. Soc. Amer. Bull. 104 (1992)
999–1014.
[19] R.J. Pankhurst, C.W. Rapela, The proto-Andean margin of
Gondwana: an introduction, in: R.J. Pankhurst, C.W. Rapela
(Eds.), The Proto-Andean Margin of Gondwana, The Geological
Society, London, 1998, pp. 1–9.
[20] S. Kay, Probing the evolving Andean lithosphere: Mid–Late
Tertiary magmatism in Chile (29–30 30′S) over the modern zone
of subhorizontal subduction, J. Geophys. Res. 92 (1987)
6173–6189.
[21] S.M. Kay, C. Mpodozis, Magmatism as a probe to the Neogene
shallowing of the Nazca plate beneath the modern Chilean flatslab, J. South Am. Earth Sci. 15 (2002) 39–57.
[22] V.A. Ramos, E.O. Cristallini, D.J. Perez, The Pampean flat-slab
of the Central Andes, J. South Am. Earth Sci. 15 (2002) 59–78.
[23] S.C. Ji, Z.C. Wang, Elastic properties of forsterite–enstatite
composites up to 3.0 GPa, J. Geodyn. 28 (1999) 147–174.
301
[24] B.R. Hacker, G.A. Abers, S.M. Peacock, Subduction factory —
1. Theoretical mineralogy, densities, seismic wave speeds, and
H2O contents, J. Geophys. Res.—Solid Earth 108 (2003),
doi:10.1029/2001JB001127.
[25] B. Hacker, G.A. Abers, Subduction Factory 3. An Excel
worksheet and macro for calculating the densities, seismic
wave speeds, and H2O contents of minerals and rocks at pressure
and temperature, Geochem. Geophys. Geodyn. (G3) 5 (2004),
doi:10.1029/02003GC000614.
[26] G. Rossi, G.A. Abers, submitted for publication. Unusual
mantle Poisson's ratio, subduction and crustal structure in
Alaska, J. Geophys. Res.
[27] M. Akaogi, H. Yusa, K. Shiraishi, T. Suzuki, Thermodynamic
properties of alpha quartz, coesite, and stishovite and equilibrium
phase relations at high pressures and high temperatures, J.
Geophys. Res. 100 (1995) 22337–22347.
[28] C.-T.A. Lee, Compositional variation of density and seismic
velocities in natural peridotites at STP conditions: implications for seismic imaging of compositional heterogeneities in
the upper mantle, J. Geophys. Res. 108 (2003), doi:10.1029/
2003JB002413.
[29] M. Barazangi, B. Isacks, Spatial distribution of earthquakes and
subduction of the Nazca plate beneath South America,
Tectonophysics 4 (1976) 686–692.
[30] M.G. Kopylova, J. Lo, N.I. Christensen, Petrological constraints
on seismic properties of the Slave upper mantle (Northern
Canada), Lithos 77 (2004) 493–510.
[31] A. Morelli, S. Danesi, Seismological imaging of the Antarctic
continental lithosphere: a review, Glob. Planet. Change 42 (2004)
155–165.
[32] S. Fishwick, B.L.N. Kennett, A.M. Reading, Contrasts in
lithospheric structure within the Australian craton–insights
from surface wave tomography, Earth Planet. Sci. Lett. 231
(2005) 163–176.
[33] M. Feng, M. Assumpcao, S. Van der Lee, Group-velocity
tomography and lithospheric S-velocity structure of the South
American continent, Phys. Earth Planet. Inter. 147 (2004)
315–331.
[34] M. Nettles, A.M. Dziewonski, Advances in global and regional
tomography using the GSN, Eos Trans. AGU, 2004 Spring
Meeting 2004, abstract # S44-02.
[35] M.J. Fouch, D.E. James, J.C. VanDecar, S. Van der Lee, K.S.
Group, Mantle seismic structure beneath the Kaapvaal and
Zimbabwe cratons, S. Afr. J. Geol. 107 (2004) 33–44.
[36] F. Niu, A. Levander, C. Cooper, C.-T.A. Lee, A. Lenardic, D.E.
James, Seismic constraints on the depth and composition of the
mantle keel beneath the Kaapvaal craton, Earth Planet. Sci. Lett.
224 (2004) 337–346.
[37] K. Futa, C.R. Stern, Sr and Nd isotopic and trace element
compositions of Quaternary volcanic centers of the southern
Andes, Earth Planet. Sci. Lett. 88 (1988) 253–262.
[38] S. Kay, C.E. Gordillo, Pocho volcanic rocks and the melting of
depleted continental lithosphere above a shallowly dipping
subduction zone in the central Andes, Contrib. Mineral. Petrol.
117 (1994) 25–44.
[39] A. Li, D.W. Forsyth, K.M. Fischer, Shear velocity structure
and azimuthal anisotropy beneath eastern North America
from Raleigh wave inverson, J. Geophys. Res. 108 (2003),
doi:10.1029/2002JB002259.
[40] M. Anderson, P. Alvarado, G. Zandt, S. Beck, Geometry and
brittle deformation of the subducting Nazca plate, central Chile
and Argentina, Geophys. J. Int. (submitted for publication).