Content uploaded by Yanni Gunnell
Author content
All content in this area was uploaded by Yanni Gunnell on May 13, 2014
Content may be subject to copyright.
Geochronologic and stratigraphic constraints on canyon incision and
Miocene uplift of the Central Andes in Peru
J.-C. Thouret
a,
⁎
, G. Wörner
b
, Y. Gunnell
c
, B. Singer
d
, X. Zhang
d
, T. Souriot
a
a
Laboratoire Magmas et Volcans, UMR 6524 CNRS, OPGC et IRD, Université Blaise Pascal, 63038 Clermont-Ferrand cedex, France
b
GZG Abt. Geochemie, Universität Göttingen, Goldschmidtsrasse 1, 37077 Göttingen, Germany
c
Department of Geography, Université Denis Diderot Paris 7 and UMR 8591 CNRS, 2 Place Jussieu, 75251 Paris cedex 05, France
d
Department of Geology and Geophysics, University of Wisconsin, Madison, WI 53706-1692, USA
Received 24 November 2006; received in revised form 14 July 2007; accepted 15 July 2007
Editor: C.P. Jaupart
Available online 21 July 2007
Abstract
The deepest valleys of the Andes have been cut in southern Peru by the Ríos Cotahuasi Ocoña and Colca–Majes. These
canyons are Late Miocene landforms based on a new ignimbrite stratigraphy supported by 42 new
40
Ar/
39
Ar age determinations
obtained on plateau-forming and valley-filling ignimbrites. Between 19 and 13 Ma, a gently sloping surface bevelling the clastic
wedge southeast of the developing mountain front was mantled by widespread ignimbrites. After 13 Ma, this paleosurface was
tilted up from 2.2 km a.s.l. at the mountain front to 4.3 km a.s.l. at the base of the Pliocene and Pleistocene volcanoes that crown
the southwestern edge of the Altiplano. The canyons incised this topography after 9 Ma, while the dated base of younger ignimbrite
valley fills suggests that these canyons had been cut down to near their present depths as early as 3.8 Ma. By 1.4 Ma, however, the
canyons had been almost completely refilled by 1.3 km-thick unwelded pyroclastic deposits, which were subsequently eroded.
Valley incision since 9 Ma at an average rate of 0.2 mm yr
− 1
is the response to topographic uplift after 13 Ma combined with
increasing runoff due to a wetter climate recorded after 7 Ma. Although long-term aridity generated an imbalance between high
long-term uplift rates and low plateau denudation rates, the combination of aridity and volcanism still promoted canyon incision
because episodic volcanic fills maintained a cycle of catastrophic debris avalanches and subsequent dam breakouts.
© 2007 Elsevier B.V. All rights reserved.
Keywords: Andes; Peru; valley incision; tectonic uplift; ignimbrites; argon dating
1. Introduction
Tectonic uplift simultaneously generates topographic
gradients that result in increased erosion (Montgomery
and Brandon, 2003) and orographic barriers that collect
precipitation. Removal of mass by erosion is thus
focussed along narrow belts of relief and can itself induce
further rock uplift, thus creating a positive feedback to
evolving orogenic fronts (e.g. W illett, 1999; Thiede et al.,
2004; Whipple and Meade, 2006). The western margin
of the Central Andes represents an extreme case in this
scenario since several kilometers of uplift has occurred
in a hyperarid region, where long-term denudation rates
remain particularly low but the history of valley incision
documents a long record of topographic uplift and cli-
matic changes.
A
vailable online at www.sciencedirect.com
Earth and Planetary Science Letters 263 (2007) 151 – 166
www.elsevier.com/locate/epsl
⁎
Corresponding author.
E-mail addresses: thouret@opgc.univ-bpclermont.fr
(J.-C. Thouret), gwoerne@dwdg.de (G. Wörner),
gunnell@idf.ext.jussieu.fr (Y. Gunnell), bsinger@geology.wisc.edu
(B. Singer), zhangx@geology.wisc.edu (X. Zhang),
souriot@opgc.univ-bpclermont.fr (T. Souriot).
0012-821X/$ - see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsl.2007.07.023
Fig. 1. Topographic Digital Elevation Model (USGS GTOPO 30 DEM) of the Central Andes (http://www.geo.cornell.edu/geology/cap/CAP_WWW.html) showing the location of the main valleys and
canyons in southern Peru and northern Chile (in addition to 37 sites in Fig. 2, five sites with
40
Ar/
39
Ar age determinations show numbers keyed to Table 1). A: map of southern Peru showing the
principal structural units and the Pliocene and Pleistocene volcanic arc; B: Shuttle image of the Cotahuasi, Ocoña, and Marán confluence ( www.lpi.usra.edu/publications/slidesets/geology/sgeo/
slide_22html; STS-41D. August-September 1984, Picture No.14-34-005). Pla teau-forming Huaylillas and Alpabamba ignimbrites at 4000–4500 m a.s.l. mantle the Miocene paleosurface and support
the Pliocene and Pleistocene v olcanoes (NS: Nevado Solim ana 5888 m a.s.l.; NC: Nevado Coropuna 6379 m a.s.l.; SS: Nevado Sara Sara 5505 m a.s.l.; NF: Nevado Firura 5498 m a.s.l.). Note the Chaucalla–
Las Lomas ridge formed by 1.6 km–thick ignimbrite co ol in g units; the whitish ig ni mb ri te blanketing the high plateau 20 km north of the town of Cotahuasi; and a prominent arcuate scar above the northwest
Cotahuasi canyon edge p ointing to potential sources for the most recent pyroclastic valley fill found in the deepest canyon section.
152 J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
The uplift chronology of the Central Andes is contro-
versial (Gregory-Wodzicki, 2000; Lamb and Davis,
2003; Ghosh et al., 2006 ). It is currently considered that
Andean deformation and uplift began in the Western and
Coastal Cordilleras ∼ 50 Ma ago, developed later and
more slowly in the Eastern Cordillera (40 to 10 Ma), and
accelerated in both Cordilleras and the Altiplano around
25 to 20 My ago (Elger et al., 2005). In the Western
Cordillera, this early uplift led to the deposition of a
thick continental clastic wedge (defined stratigraphi-
cally in southern Peru as the Moquegua Group; Kennan,
2000) and was accentuated by the eruption of large-
volumes of plateau-forming ignimbrites.
We focus on Neogene ignimbrites and lavas, which
provide excellent regional markers for tracing landscape
evolution and the history of valley formation on the
western slope of the Central Andes (Myers, 1979; Noble
et al., 1984; Sébrier et al., 1988; Wörner et al., 2000;
Thouret et al., 2001; Wörner et al., 2002; Thouret et al.,
2003, 2005; Quang et al., 2005; Paquereau-Lebti et al.,
2006; Schildgen et al., 2007). Here, abundant datable
volcanic rocks preserved as valley fills at various levels
in the Cotahuasi and Ocoña canyons of southern Peru
(Figs. 1 and 2) are used for dating phases of valley
incision and orogenic uplift as well as assessing the
geomorphic effects of climate change. Critical to this
Fig. 2. Landsat image showing the two canyon networks of Cotahuasi–Ocoña and Colca–Majes and structural units on the western slope of the
Central Andes. Location of four cross sections (A–A′,B–B′,C–C′, and D–D′) depicted in Electronic Supplement Fig. 1 in the Appendix is
indicated. Faults, localities, and volcanoes mentioned in the text are shown. The sites of the
40
Ar/
39
Ar age determinations are keyed to Table 1 and
further described in the Electronic Supplement table in the Appendix (dated sites outside of the canyon area are indicated in Fig. 1).
153J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
Table 1
Summary of
40
Ar/
39
Ar ages for the Ocoña–Cotahuasi and Colca canyons, Peru
Sample code no. UTM Easting UTM Northing Elevation Lithology Dated
material
Age
(UTM) (UTM) (m a.s.l.) (Ma)
a
±2σ
Ocoña and Cotahuasi canyons: post-incision volcanic rocks
Late Pleistocene lava flows
1-COTA-05-10 0706959 8299105 2235 Andesitic lava flow Groundmass 0.34± 0.06
2-COTA-05-11 0708072 8301599 2345 Andesitic lava flow Groundmass 0.45± 0.01
3-COTA-05-15 0715085 8302926 4180 Andesitic lava flow Groundmass 0.66± 0.01
4-COTA-05-06 0712734 8300308 3940 Andesitic lava flow Groundmass 0.68± 0.03
Top of pyroclastic valley fill
5-PIG-03-123 0705449 8277851 2300 Andesitic lava flow Feldspar 1.36± 0.27
6-PIG-05-01 0725754 8312325 4060 Andesitic lava flow Feldspar 1.5± 0.32
Upper Sencca-type ignimbrites
7-OCO-05-06 0719720 8273650 3550 Rhyolitic pumice Feldspar 1.95 ± 0.16
Biotite 1.76± 0.17
8-PIG-00-06 0669009 8257827 2550 Rhyolitic pumice Feldspar 1.81 ±0.11
9-PIG-00-25 0750350 8252890 3600 Rhyolitic pumice Feldspar 1.93 ±0.04
10-OCO-04-08 0698800 8243130 1675 Rhyolitic pumice Feldspar 1.96 ± 0.06
11-OCO-05-04 0716172 8251325 1840 Rhyolitic pumice Feldspar 2.09 ± 0.06
Biotite 2.02± 0.04
12-PIG-04-02 0725754 8312325 3570 Rhyolitic pumice Feldspar 2.04 ±0.14
13-OCO-04-05 0704400 8237700 750 Rhyolitic pumice Feldspar 2.05 ± 0.29
Lava flow in Ocoña valley
14-OCO-05-11 0706950 8253980 830 Andesitic lava flow Groundmass 2.27± 0.05
Lower Sencca ignimbrites and base of valley fill
15-PIG-03-122 0705447 8277592 2105 Rhyolitic pumice Feldspar 2.35 ±0.95
16-PIG-00-24 0752577 8262398 3890 Rhyolitic pumice Feldspar 3.16 ±0.04
17-PIG-00-28 0727316 8316532 2800 Vitrophyre Feldspar 3.7 ± 0.1
18-OCO-05-12 0704831 8277228 1665 Rhyolitic pumice Groundmass 4.84 ± 0.07
Lower Barroso lava flows
19-BAR-03-01 0745259 8331693 3510 Andesitic lava flow Groundmass 5.80± 0.10
Caraveli ignimbrites
20-PIG-00-07 0676539 8312021 2420 Rhyolitic pumice Feldspar 8.97 ±0.06
21-CARA-05-07 0683470 8233701 1750 Rhyolitic pumice Feldspar 9.40 ± 0.83
Biotite 8.98± 0.15
22-PIG-00-04 0669009 8257827 2550 Rhyolitic pumice Feldspar 9.02 ±0.11
23-PIG-00-03 0670228 8257664 2350 Rhyolitic pumice Feldspar 9.15 ±0.31
Ocoña and Cotahuasi region: pre-incision volcanic rocks
Huaylillas ignimbrites
24-PIG-03-126 0704629 8229204 2275 Rhyolitic pumice Feldspar 13.21± 0.53
Biotite 13.19± 0.07
25-PIG-00-31 0730308 8308079 4300 Rhyolitic pumice Feldspar 14.23± 0.07
26-PIG-00-33 0806185 8189594 1000 Rhyolitic pumice Feldspar 14.25± 0.08
Tuff intercalated in upper Moquegua Formation
27-PIG-00-32 0772378 8193673 850 Rhyolitic pumice Feldspar 16.25± 0.10
Alpabamba ignimbrites
28-PIG-00-11 0552154 8374953 3200 Rhyolitic pumice Feldspar 18.23± 0.17
29-PIG-00-41 0939492 8115952 2900 Rhyolitic pumice Feldspar 18.90± 0.50
154 J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
study is (1) the presence of an elevated paleosurface,
which bevels the Moquegua formation, and (2) a 25 My-
old Miocene marine layer that is covered by a ∼ 24 Ma
ignimbrites (Fig. 2 and Electronic Supplement Fig. 2 in
the Appendix). Tracking the uplift and inci sion of this
extant paleosurface constitutes the key for understand-
ing the Neogene and Quaternary denudation sequence.
The refined volcaniclastic stratigraphy we document for
southern Peru is based here on 42 new
40
Ar/
39
Ar age
determinations on volcanic rocks from the area between
Nazca and Arequipa (Figs. 1 and 2, Table 1, and the
Appendix). We reconstruct a 25 Ma chronology of
volcanism and repeated valley incision into the initial
seaward-sloping paleosurface (Fig. 1), interrupted by
episodes of temporary valley refilling (Fig. 2), in
relation to tectonic uplift, changes in runoff regimes,
and the type and timing of volcanic activity. The data
indicate a major phase of uplift on the western slope of
the central Peruvian Andes during the Late Miocene.
2. Geological setting and stratigraphy of the Central
Andes in southern Peru
The deepest canyons of South America were cut by the
Ríos Cotahuasi, Ocoña, and Colca at 15–16°S (Figs. 1
and 2). Although their valley heads currently impinge
beyond the Quaternary volcanic arc into the semi-arid
Altiplano (rainfall 250–350 mm yr
− 1
), these canyons
occur in a region where rainfall is b 100–200 mm yr
− 1
.
Another paradox is that the deepest Andean valleys should
exist here while similar strike-perpendicular valleys in
central Peru (north of 15°S) and northern Chile (south of
30°S) are shallower despite present-day precipitation
being higher in those areas (Hoke et al., 2005) and plateau
elevations between canyons being similar.
The Cotahuasi–Ocoña catchment is 4200 km
2
and
extends from the western edge of the Altiplano to the
Pacific across the Western and Coastal Cordilleras
(Fig. 1). Whereas the lowermost reach of the Río Ocoña
cuts a narrow canyon into the Precambrian rocks and
sedimentary cover of the Coastal Cordillera, the upper
and middle courses of the Ríos Ocoña and Cotahuasi
form the deepest canyon section, with maxi mum relief
between the plateau edge and the valley bottom ranging
between 2.5 and 3.5 km (Figs. 2–4). Incision has
affected late Oligocene to middle Pleistocene volcanic
rocks overlying Paleogene and Mesozoic sediments.
Deeper sections of the canyon and valley flanks in the
Coastal Cordillera expose Cretaceous plutons and
Paleozoic to Precambrian metamorphic rocks. The can-
yon's catchment reaches a maximum width of 45 km in
the granitoid plutons.
Table 1 (continued)
Sample code no. UTM Easting UTM Northing Elevation Lithology Dated
material
Age
(UTM) (UTM) (m a.s.l.) (Ma)
a
±2σ
Ocoña and Cotahuasi canyons: post-incision volcanic rocks
Nazca ignimbrites
30-PIG-00-10 0552154 8374953 3500 Rhyolitic pumice Feldspar 22.16± 0.34
31-PIG-02-02 0295064 8091468 1718 Rhyolitic pumice Feldspar 23.92± 0.49
32-PIG-00-38 0324519 8134616 4150 Rhyolitic pumice Feldspar 24.43± 0.17
Colca canyon: post-incision volcanic rocks
Pleistocene lava flows
33-COL-04-17 0794950 8261980 1885 Andesitic lava flow Groundmass 0.13± 0.02
34-BAR-01-62 0221365 8272955 3443 Andesitic lava flow Groundmass 0.15± 0.01
35-COL-04-16 0216104 8273569 2240 Andesitic lava flow Groundmass 0.54± 0.06
36-COL-04-01
b
0209790 8267148 3460 Andesitic lava flow Groundmass 0.61± 0.01
37-BAR-01-61 0219485 8274091 3445 Andesitic lava flow Groundmass 0.65± 0.01
38-COL-04-14 0215841 8267571 3600 Andesitic lava flow Groundmass 0.65± 0.02
39-COL-04-03 0219425 8274042 3770 Andesitic lava flow Groundmass 0.68± 0.01
40-COL-04-07 0803300 8264075 2980 Andesitic lava flow Groundmass 1.06± 0.05
41-COL-04-10 0815528 8270769 3350 Andesitic lava flow Groundmass 1.07± 0.03
Upper Sencca ignimbrites
42-PATA-04-02 0220955 8260826 4675 Rhyolitic pumice Feldspar 2.20 ±0.15
Notes: Numbers (first column) are keyed to the text and to the Electronic Supplement tables in the Appendix and Figs. 1 and 2.
a
Ages are calculated relative to 1.194 ±0.012 Ma Alder Creek rhyolite sanidine (Renne et al., 1998). Ages of feldspar and biotite separates are
isochron ages from laser fusion experiments while ages of groundmass separates are plateau ages from incremental heating experiments. More
details in Electronic Supplement Tables 1 and 2 in the Appendix and text.
b
Isochron age is 0.61± 0.01 Ma; plateau age is 0.61 ±0.01 Ma (M. Fornari, pers. comm.). For age standard age used, see Delacour et al. (2007).
155J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
To the east, the Colca catchment covers 5500 km
2
across the Western Cordillera and north of the volcanic
arc. Between 130 and 170 km from the ocean, the 1.5 to
2.5 km-deep Colca valley cuts through folded Jurassic
and Lower Cretaceous metamorphic and sedimentary
rocks and dioritic plutons of Late Cretaceous to Early
Paleogene age (Fig. 2). Seaward of the orogeni c front,
the river changes its name to Majes–Camaná and cuts
into folded and overthrust Mesozoic sediments covered
to the south by slightly folded Paleogene to Early
Miocene Moquegua conglomerates, distal sand and
sandstones. Near the coast, this clastic piedmont is
dissected and consists of non-folded Paleogene sedi-
ments and conglomerates of the Moquegua Group
unconformably overlying the metamorphic basement.
Syn-uplift reverse and overthrust faults reflect only
limited Neogene crustal shortening in the Western
Cordillera and fore-arc regions (Fig. 2). The lower
Moquegua Group has only been slightly folded and
faulted (M ajes–Camaná valley), and only minor
erosional discontinuities are observed between the
Moquegua sub-groups (Fig. 3).
3. Methods
Extensive field surveys consisted of mapping the
Neogene ignimbrites, other pyroclastic deposits and lava
flows, as well as determining the paleo-morphologies
preserved at the valley flanks (canyon bluffs, pediments,
hanging fans) and in the valley channel (fill terraces,
Fig. 3. Composite stratigraphic scheme collated for southernmost Peru after Quang et al. (2005), Roperch et al. (2006), southern Peru (Tosdal et al.,
1981; Noble et al., 1984; Sébrier et al., 1988; Paquereau-Lebti et al., 2006), and the area of the Cotahuasi, Ocoña, and Colca canyons (this study and
Paquereau-Lebti et al., 2006; Delacour et al., 2007) highlighting the significance of syn-uplift ignimbrites (dark gray, third column) and syn-incision
ignimbrites (light grey, third column). Wavy lines indicate periods of erosion and/or tectonic phases, serrated lines are angular unconformities. The
stratigraphy has been established on data collated from literature (e.g. Roperch et al., 2006) and from our survey in the canyon area as well as in the
area of Arequipa (Paquereau-Lebti et al., 2006; Delacour et al., 2007).
156 J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
Fig. 4. Longitudinal profiles and incision history of the Cotahuasi and Ocoña canyons. The topography of the paleosurface is illustrated along the plateau edge where schematic sections show local
stratigraphy and reference ignimbrites. K: knickpoint; S1: reference paleosurface, S2: pre-Caraveli paleotopography; S3: pre-Sencca paleotopography.
157J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
debris-avalanche and lahar deposi ts). A 20-m digital
elevation model (DEM with an error of c.50 m) of the
Cotahuasi–Ocoña valleys was generated from two pairs
of SPOT 4 satel lite images using computerized stereo-
photogrammetric techniques available in the ENVI™
software package. The DEM defines the uppermost
contour of the ignimbrite plateau as the initial pre-
incision downwarped paleosurface (Fig. 4). It was thus
used to quantify volumes of eroded material and incision
rates based on elevation differences between key dated
ignimbrite contours and paleovalley floors defined by
the base of dated volcanic valley fills. Ignimbrites and
other pyroclastic deposits we re distinguished a nd
classified in terms of stratigraphy, facies, and petrogra-
phy. In order to reconstruct the position of paleothal-
wegs, ignimbrite and lava flow samples were collected
vertically down canyon walls and throughout the
drainage basins for lateral correlation. In addition to
non-welded pyroclastic valley-fills, datable basal pum-
ice-rich layers and vitrophyres of ignimbrites occurring
near the present thalweg but also on the canyon edge
were favoured as sampling targets. At outcrop scale, the
bases and tops of ignimbrite sheets were sampled for age
determination in key sections.
Forty two age determinations were made at the Univer-
sity of Wisconsin in addition to samples dated earlier, using
the same age standard, at UMR GéoAzur (CNRS-IRD) in
Nice (see Thouret et al., 2003; Delacour et al., 2007). From
15 lava flows and 27 ignimbrites sanidine, biotite, plagio-
clase, or groundmass samples were prepared, irradiated and
aliquots of ∼ 200-μm grains were measured using either
laser fusion (single or up to 10 crystals from ignimbrites) or
furnace incremental heatin g (50–100 mg of lava ground-
mass) and following protocols described in Singer et al.
(2004) and Smith et al. (2006) and in the Appendix. The
40
Ar/
39
Ar age determinations in Table 1 (keyed to locations
in Figs. 1, 2, 6 and the Appen dix) are based on isochrons
defined by concordant single-crystal or plateau-age data
and are calculated relative to 1.194 Ma Alder Creek
Rhyolite Sanidine (Renne et al., 1998), using ArArCALC
software (Koppers, 2002), with analytical uncertainties
reported at the 2σ level (see also Delacour et al., 2007).
4. Results
4.1. A new chronostratigraphy of Neogene volcanic
rocks in southern Peru
The existing stratigraphy of Neogene rocks in southern
Peru (Tosdal et al., 1981, 1984; Noble et al., 1984; Sébrier
et al., 1988; Quang et al., 2005; Paquereau-Lebti et al.,
2006; Roperch et al., 2006) can be compared with our
chronostratigraphy based on the new
40
Ar/
39
Ar age deter-
minations (Table 1, Fig. 3). The correlation with the
broader-scale stratigraphy of southern Peru (Fig. 3)is
basedonagecorrelation(e.g.Roperch et al., 2006)and
also on our surveys and mapping between this area and
Arequipa (Thouret et al., 2003; Paquereau-Lebti et al.,
2006; Delacour et al., 2007). These ages, which range
between 24.43 ± 0.17 and 0.34± 0.05 Ma (Table 1, keyed
to Figs.1,2, and tables in Appendix), help to distinguish
six major pyroclastic deposits, including five Neogene
rhyolitic ignimbrite sheets totalling N 1500 km
2
in area and
about 300 km
3
in volume, and six groups of lava flows.
These results call for the following update of volcanic
stratigraphy in the central Andes of Peru (see Fig. 3).
The thick 24.6–21.8 Ma-old Nazca ignimbrite sheets
(No. 30 to 32, Table 1) cap extensive elevated plateaus
towards the WNW of the study area (Fig. 1) as well as in
southernmost Peru (Moquegua area) where they corre-
late with the Chilean ‘Oxaya’ ignimbrites of latest Oli-
gocene to Early Miocene age (Wörner et al., 2000, 2002;
Elger et al., 2005). They are overlain, and therefore
older than, the coarse upper Moquegua conglomerates
(Moquegua ‘C’ and ‘D’ intervals; Tosdal et al., 1984).
The 19.4–18.0 Ma-old Alpabamba (No. 28 and 29) and
14.3–12.7 Ma-old Huaylillas ignimbrite sheets (No. 24 to
26) are whitish to brownish welded to strongly welded
cooling units that form extensive plateaus between 4000
and 4500 m a.s.l. with 350 m-high cliffs (Fig. 1C). Plateau-
forming ignimbrites have been collectively termed ‘Huay-
lillas Formation’ (Pecho Gutierrez, 1983; Olchauski and
Dávila, 1994; Quang et al., 2005)and‘Chuntacala For-
mation’ (Sébrier et al., 1988). However, ignimbrites pre-
viously mapped as ‘ Huayl illas ignimbrites’ were dated
here at 14.25–13.19Ma(No.24to26).Theyare
therefore younger than the 22.8–17.6 Ma-old ‘Huay-
lillas Formation’ of Qua ng et al. ( 200 5), Tosdal et al.
(1981, 1984), which we therefore infer represent a
package of several ignimbrite sheets of different ages.
Distal ignimbrites of 16.25 Ma to 14.25 Ma (No. 24 and
27 in Table 1) are interlayered in the fore-arc sediments
towards the top of thick conglomerates of the Early
to Middle Miocene Upper Moquegua Formation in the
Majes, Sihuas, and Victor valleys. The top of these
conglomerate s (and int erca late d tuff layers) are cut by
the erosional surface forming the piedmont between the
Western Cordillera and the Coastal Batholith (Fig. 2).
The coarse conglomerates correspond to the transition
between the ‘Moquegua C and D’
intervals of the Moque-
gua Group (Roperch et al., 2006) and the ignimbrites are
distal equivalents of the Huaylillas ignimbrites. A similar
series of ignimbrites overlying the thick clastic wedge
are observed over an area of 1000 km on the western
158 J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
slope of t he Andes. They are equivalent to the Chilean
‘Oxaya’ and ‘Pica ' ign imbri tes of latest Oligoce ne to
Early Miocene age (Wörner et al., 2000, 2002; Elger
et al., 2005) to the south of our study area. As these
ignimbrites are all largely undeformed and intercalated
with the top of the Moquegua clast ic wedge, they tend
to form large plateaus covering the Western C ordillera
and the piedmont towar ds th e coast. We thus use the
youngest of these plateau-forming ignimbrites as a
reference marker which we refer to as the ignimbrite
plateau of Huaylillas age (16.25 t o 14.25 Ma).
We report here a previously undetected generation of
9 Ma-old ignimbrites (No. 20 to 23) which we term
Caravelí. They fill an irregular paleotopography cut in the
Huaylillas ignimbrites. The 9.4–8.8 Ma welded, 100- to
200-m thick cooling units of the Caravelí ignimbrites
crown smaller and lower plateaus at 2–3kma.s.l.that
slope gently towards the WSW (Fig. 2).
The 5 to 6 Ma-old andesitic lava flows, mapped as
‘lower Barroso Fm.’ of the Barroso Group (Pecho
Gutierrez, 1983; Olchauski and Dávila, 1994), form the
core of deeply eroded shield volcanoes (Figs. 3 and 4)
that once occurred on top of the Huaylillas plateau along
the Western Cordillera and the western edge of the
Altiplano. One of these lava flows (No. 19) has exten-
ded over the eastern edge of the upper Cotahuasi canyon
(near the village of Pajcce, Fig. 2), indicating that a
valley wall existed prior to 5.8 Ma at this site.
The Sencca ignimbrites, previously given as Pliocene
(Sébrier et al., 1988), comprise two distinct sheets (4.9–
3.6 and 2.3–1.4 Ma) in the study area. The 4.9–3.6 Ma-
old lower Sencca ignimbrites (No. 16 to 18) typically
occur in narrow valleys or preserved on ridges and valley
flanks. The slightly welded to non-welded cooling unit
usually exhibits a vitrophyre at its base. The lower Sencca
ignimbrite is similar in age to most of the 6–5Ma-old
lower Barroso lava flows (Figs. 3 and 4).
Lava flows 2.27 Ma old (No. 14 in Table 1 ) overlie a
member of the upper Sencca ignimbrites. Scattered
outcrops are found on the eastern wall and benches of the
lower Cotahuasi canyon upstream of Chaucalla and
along the Ocoña canyon as far as Iquip i and Urrasqui
(Fig. 2). The reconstructed profile of the lava flows
suggests that they were sourced either from below the
Nevado Solimana area and (or) other large stratovolca-
noes of the Pliocene volcanic front. The 2.3–1.6 Ma-old
upper Sencca ignimbrites (No. 7 to 13, and 42 in Table 1)
often consist of non-welded pumice flows, and outcrop
tops form either valley-fill benches, eroded ridges, or
preserved topographic benches above tributary thalwegs.
Ages show that most of the Sencca ignimbrites pre-date
the Pleistocene volcanoes.
A most recent (1.56–1.36 Ma: No. 5 and 6) and thick
succession of non-welded pumice flows and fall deposits
leans against canyon walls or are capped in places by
more recent lava flows. Lava flows ∼ 1 Ma old (No. 40
and 41 in Table 1) form the base on which the Middle
Pleistocene to Holocene volcanoes were constructed.
Dated examples of these widespread lava flow fields
crop out on the western flank of Nevado Solimana, and in
the upper Colca canyon near Huambo (Fig. 2). Lava
flows 0.6 to 0.3 Ma old were produced by parasitic cones
and domes on the western flank of Nevado Solimana
(No. 1 to 4 in Table 1) and mantle the eastern edge of
the Cotahuasi canyon. Channeled lava flows of a similar
age (No. 35 to 39 in Table 1) form high benches in the
upper Colca canyon near Cabanaconde (Fig. 2). Lava
flows 0.13–0.15 Ma old that belong to the Andahua–
Orcopampa–Huambo monogenetic province ( Delacour
et al., 2007) have filled the bottom of Río Ayo (No. 33 in
Table 1), a tributary to Río Colca, and in the upper course
of the Colca itself (No. 34 in Table 1).
Dated lava flows in the area of the canyo ns indicate that
three generations of volcanic edifices have crowned the
Western Cordillera: the eroded Late Miocene (6–4Ma)
shield volcanoes (Barroso Group); the Late Pliocene
(≤2.3 Ma) stratovolcanoes, now already deeply dissected
by glacial erosion; and the Pleistocene composite cones
and domes (b 1.4 Ma), which are either dormant edifices
(Ampato) or active cones (El Misti, Ubinas, Sabancaya)
with bases no older than 0.6 Ma (Thouret et al., 2001;
GerbeandThouret,2004;Thouretetal.,2005).
4.2. The pre-incision paleosurface
Based on the new event chronology provided above,
we now outline the uplift and incision history of the Early
Miocene paleosurface. The thick Moquegua clastic
wedge and its cover of plateau-forming ignimbrites
have their equivalents in similar stratigraphic successions
over a length of ∼ 1500 km along the Andean slope. The
wedge deposits were sourced from the NE, and record the
early phase of uplift and erosion on the western margin of
the Andes at ∼ 25 Ma (Kennan, 2000; Wörner et al.,
2000). The 24.6–12.7 Ma-old ignimbrite sheets are
interlayered with or deposited on Late Oligocene to
Middle Miocene continental conglomerate beds. This 0.4
to 1-km-thick upper Moquegua Formation contains a
marine layer beneath a 24.5±0.8 Ma tuff layer (Cruzado
and Rojas, 2007)situated∼ 100 m below the top of the
sequence; see No. 32, Table 1). This formation lies now at
2.3 km a.s.l. (Cuno Cuno) and 1.8 km a.s.l. (Pampa de
Gramadal), i.e. 1.8 to 1.3 km above the floor of the Ocoña
canyon (Fig. 2). The general 1.6% dip of these Miocene
159J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
ignimbrite sheets towards the SSW parallels the present
gradient of the paleosurface. Supported by the AMS flow
directions (P. Paquereau and P. Roperch, unpubl. data),
this geometry defines a gently sloping surface on the
western Andean slope, which already existed by 19–
13 Ma. This paleosurface is erosional in the higher
Western Cordillera and depositional on the western
Andean slope below 2400 m a.s.l. Its topographic
continuity is remarkable (cf. Figs. 2 and 4) and correlates
with the ‘Puna’ paleosurface as defined elsewhere in Peru
(Tosdal et al., 1981, 1984). The plateau surface has not
been substantially eroded since its formation and today
still dominates the scenery between the Western and
Coastal Cordilleras (Figs. 2–4). The surface of the 19 Ma-
old Oxaya ignimbrites in northernmost Chile, equivalent
to the Huaylillas surface in southern Peru, has been
equally inactive since about 11–12 Ma despite its
westward slope, its erodible volcaniclastic cover, and
elevations exceeding 3 km (Wörner et al., 2000). Such
limited denudation between the times of ignimbrite
deposition and much more recent valley incision may
reflect a period of lower erosion rates. We attribute the
longevity of this low-energy environment to the flatness
of the paleolandscape sealed by thick, hard welded
ignimbrites, and to arid conditions already prevalent in
Miocene times. Canyon incision was thus delayed until a
significant increase in runoff and/or topographic gradient
enhanced erosive power and initiated valley incision.
5. Incision history of the Cotah uasi and Ocoña
canyons
Based on the age of the Alpabamba and Huaylillas
ignimbrites covering the sediments and their source area
in the Western Cordillera, incision of the Cotahuasi and
Ocoña canyons did not begin before ∼ 13 Ma. Incision
depths during the first phase between ∼ 13 Ma and 9 Ma
were limited to a few hundred meters because the ∼ 9Ma
Caravelí ignimbrites mantle pediments on granodiorites
that grade to shallow valleys cut in Paleogene sediments
only 0.3 to 0.6 km below the Huaylillas ignimbrite
reference surface (Fig. 4A and Electronic Supplement
Fig. 1 in the Appendix). The 13 to 9 Ma pre-Caravelí
valleys were thus wider and much shallower than the
Late Miocene canyons.
The present-day valleys were cut mostly between
9 Ma and ∼ 3.8 Ma (Figs. 3–5; Table 1). Evidence for
this is based on outcr ops of 3.76 ± 0.14 Ma valley-filling
lower Sencca ignimbrites (No. 17, Table 1) situated
∼ 0.3 km above the present valley floor near the tow n of
Cotahuasi and ∼ 2 km below the paleosurface covered
by the 14–12.7 Ma-old igni mbrites (Fig. 4). About
40 km downstream, the paleovalley bottom is ∼ 0.6 km
above the present valley floor at 4.84 ± 0.07 Ma (No. 18,
Table 1). This age is obtained from the base of 1-km-
thick valley-filling ignimbrites that form a high ridge
above the confluence of the Ríos Ocoña, Cotahuasi, and
Marán (insert D, Figs. 1, 2, and Electronic Supplement
Fig. 1 in the Appendix). These depths of incision
suggests 2–2.5 km of uplift for the western Altiplano
between ∼ 13 and 3.8 Ma, and like wise for the Western
Cordillera Front Range between 13 Ma and 4.8 Ma. The
principal phase of uplift and subsequent valley cutting
thus took place before the end of Miocene times (Fig. 5).
The main phase of valley incision started some time
after 9 Ma, when the deposition of alluvial fans in the
northern Chilean fore arc and lake sediments on the
Altiplano indicates a period of increased precipitation at
7Ma(Gaupp et al., 1999 and references therein) and
suggests that canyon incision peaked after 7 Ma when it
became climatically enhanced.
The 3.8 Ma-old valley-filling ignimbrites (No . 17,
Table 1) represent the base (at 2.8 km a.s.l.) of an
extensive valley-filling sequence the top of which lies
4.06 km a.s.l. and was dated at 1.56–1.36 Ma (No. 5 and
6in
Table 1) above the town of Cotahuasi (Fig. 4). This
extensive 1.3 km-thick valley fill also occurs below the
west flank of Nevado Solimana and at the confluence
with the Ríos Ocoña and Marán. Because they occur on
both sides of the deepest canyon sections and their
thicknesses increase towards the north, these pyroclastic
deposits are thought to reflect a series of voluminous
explosive volcanic events from a potential source caldera
situated northwest of Cotahuasi (see insert Fig. 1).
The ages of valley filling volcanics at mid-elevation
(3.2–3.4 km a.s.l.) between the canyon edge and the
valley floors provided intermediate ages for the valley
fill (No . 15 and 16 in Table 1) within the 3.8–1.4 Ma age
bracket. Apparently, this poorly consolidated thick
valley fill was rapidly removed after 1. 4 Ma. Hanging
fans and terraces that are matched on both sides of the
valley refl ect pauses in the re-incision process during the
Pleistocene (El ectronic Supplement Fig. 1 in the
Appendix). Evidence for younger but more local valley
fills is also provided by the staircase morphology
formed by ∼ 0.68 Ma- to 0.45–0.33 Ma-ol d lava flows
(No. 2 to 4, Table 1) on the western flank of Nevado
Solimana (Huachuy, Fig. 2), at elevations of about 1.4 to
0.6 km above the present Cotahuasi channel. Several
breaks in the present (and past) longitudinal profiles
indicate that incision has not been in equilibrium for
much of the younger canyon history. This is due to lava
flows or valley flank collapses that repeatedly filled the
valley at different times and locations.
160 J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
Fig. 5. Synthesis of topographic evolution since 25 Ma from the Pacific piedmont to the edge of the Altiplano based on
40
Ar/
39
Ar age determinations, field mapping, and satellite imagery. Syn-uplift
and syn-incision, dated ignimbrite sheets and/or pyroclastic deposits are shown, and phases of uplift and subsequent valley incision are indicated. Present-day valley floors of the Cotahuasi and Ocoña
canyons are summarized on the basis of observations made near their confluence.
161J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
Valley-filling Sencca ignimbrites were channeled in-
side the deep valleys but upon reaching the western
piedmont, they spread out and mantled the Miocene
ignimbrites and sediments (top of the Cuno Cuno
homocline, Fig. 4) or were deposited in relatively
shallow (0.6 km) valleys cut in the Caravelí palaeosur-
face (e.g. Quebrada Pescadores, Fig. 2). Therefore, the
angle of their deposition (1.6%) is at least half that of the
present slope (3.3%) of the western Andean margin (i.e.
the 14 Ma plateau surface) and also steeper than the
present thalwegs of the canyons (1.1%). This difference
must be due to tectonic tilting. The relatively lower
angle of the top of the Sencca ignimbrites suggest only
limited tilting and uplift since the end o f their deposition
(1.36 Ma).
Respectively since ∼ 0.45 Ma (which is the age of
lava flows 0.6 km above the Río Cotahuasi bed near
Huachuy: Fig. 2; No. 2 in Table 1) and ∼ 0.53 Ma
(which is the age of lava flows forming terraces 0.3 km
above the Colca valley floor near Cabanaconde: Fig. 5,
No. 35 in Table 1), canyon incision rates have de-
creased. In theory, this could reflect diminished uplift
rates, or the rivers approaching the minimum work con-
ditions implied by a regular longitudinal profile (Fig. 4
and Electronic Supplement Fig. 1 in the Appendix).
Present-day channel gradients, however, exhibit sever-
al knickzones close to the orogenic front (Fig. 4)and
two di stinc t reaches : the upper course has a grad ie nt
of 2.1% upstream of a knickpoint where the river cuts
into 0.4 km-thick debris-avalanche deposits (e.g. Río
Cotahuasi at Cotahuasi town, and at the Ríos Arma and
Ocoña confluence). Downvalley, the gentler sloping (1%)
lower Ocoña and Majes valleys, with braided chan-
nels 1 km wide, have been choked by large volumes of
sediment. Evidence for raised base levels and excess
bedload in the Ocoña section (and in the Majes section of
the Colca valley) is based on two observations: (1) given
that the gradient of the 2.27 Ma-old lava flow just above
the b ed is steepe r (2.8%) than the present-day river
profile (Fig. 4), the lower Ocoña canyon reach was
steeper during late Pliocene times than today and
possibly related to a depressed sea level; (2) alluvial,
debris-avalanche, lahar and lacustrine deposits, pre-
served 0.6 km above the valley floor on both sides of
Río Ocoña (Urrasqui, Fig. 2), show that two thirds of
the canyon had been filled to this level at some time
during the early Pleisto cen e. Knickp oi nts along the
channel (Fig. 4) and debris-avalanche deposits that
have not been completely removed by the river suggest
that the river still has a slightly disturbed thalweg
profile today. Howe ver, while evide nce for uplift and
incision is obvious, significant tectonic displacements
along discrete faults, for example at the present orog-
enic front, as expressed by prominent knickpoints, are
lacking. This is also true for the older valley gradients
as reconstructed here from the late Miocene valley
fills. Absence of significant tectonic shortening in the
Andean f ore-arc region during the b 9Myphaseof
uplift has been noted before and has supported a model
of lower crusta l flow from E to W as th e main process
of crustal thickening and uplift in this region (Isacks,
1988; Wörner et al., 2000; Roperch et al., 2006; Schildgen
et al., 2007).
6. Recent phases of incision in the Colca canyon
The older incision history of the Río Colca is less
well constrained than for the canyons of the Ríos
Cotahuasi and Ocoña rivers because the pre-incision
plateau is less well defined and extensively covered by
Miocene volcanics of the Barroso Group. However, the
more recent valley history is well documented by our
new
40
Ar/
39
Ar-ages because the Río Colca, east of Rio
Ocoña, flows from east to west on the northern edge of
an elevated plateau capped by 2.2 Ma-old Sencca
ignimbrites (No. 42, Table 1, Electronic Supplement
Fig. 2 in the Appendix; Klinck et al., 1986). This young
age, however, does not imply that the Colca canyon is
younger than 2.2 Ma. Instead, we suggest that, as in the
Ocoña canyon, the Sencca ignimbrites represent refills
of an older valley, later rein cised in three stages
(Electronic Supplement Fig. 2 in the Appendix). The
first was rapid and occurred before 1.06 Ma, as inferred
from the age of lava flows now hanging 1.5 km above
the valley bed near Huambo, and at mid canyon
elevation (No. 11, Table 1). The second phase of re-
incision was brief, and occurred when a dammed lake
broke out after 0.61 Ma (No. 36, Table 1), which is the
age of a lava flow capping the uppermost lake deposits
near Achoma (Electronic Supplement Fig. 2 in the
Appendix). Lacustrine deposits 0.3 km thick have filled
the upper reach of the valley (see Klinck et al., 1986).
They are related to natural dams that formed as a
consequence of valley-flank collapse and lava flow fills.
Thick, boulder-rich terraces down valley testify to
catastrophic paleolake breakouts into the Río Majes.
As in the case of the Ocoña canyon, most of the incision
was achieved before the middle Pleistocene because
topographic benches formed by lava flows and situated
0.3 to 0.6 km above the valley floor near Cabanaconde
and Huambo, are 0.53 Ma old (No. 35, Table 1). The
third and minor re-incision phase thus occurred between
b 0.53 Ma and 0.2 Ma, which is the age of lava flows
being currently cut by the Rio Colca near the town of
162 J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
Chivay. Below the preserved paleolake level, a series of
terraces and hanging fans reflect repeated valley cut-
and-fill episodes during the b 0.4 My-long period. The
left tributar ies of the Colca such as the Río Ayo are
entrenched by 150 m into two 0.15 –0.13 Ma lava flows
(No. 33 and 34 in Table 1).
7. Discussion
7.1. Timing of incision and uplift of the Altiplano and
Western Cordillera
Early uplift and increased erosion have been inferred
from the late Oligocene to middle Miocene “Upper
Moquegua” conglomerates, and on the basis of interca-
lated ≥ 24.5 Ma-old marine strata in this Formation, now
occurring at 1.8 and 2.3 km a.s.l. (Fig. 3 and Electronic
Supplement Fig. 1 in the Appendix). The depositional
record of this early uplift phase was punctuated by the
voluminous plateau-forming ignimbrites, which sealed
the land surface and appears to have remained mostly
uneroded for several million years. Whereas uplift may
have been more or less continuous, erosion was not. We
have argued above that the Huaylillas surface was dis-
sected between 9 and 3.8 Ma (see also Schildgen et al.,
2007). This finding supports other data indicative of rapid
uplift obtained for the Bolivian Altiplano between 10.3
and 6.4 Ma (Marshall et al., 1992; Ghosh et al., 2006).
However, in southern Peru, as outlined earlier, significant
uplift occurred prior to the eruption of the plateau-forming
ignimbrites between 25 and 14 Ma and continued until
Pliocene times. The evidence of a protracted uplift history
weakens the case for rapid Andean uplift being restricted
to a peak between 10.3 and 6.4 Ma (cf. Marshall et al.,
1992; Ghosh et al., 2006), and thus casts doubt on mantle
delamination as the prime cause of vertical uplift (Hartley
et al., 2007).
The most recent 1.3 km-thick “Sencca” pyroclastic
valley-fill (3.8–1.36 Ma) has been largely removed (Fig. 4,
Table 1). This renewed phase of incision was relatively
rapid (∼ 1.05–1.2 mm yr
− 1
) because pumice-flow depos-
its of that valley fill were capped by 0.45 Ma lava flows at
3.2 km a.s.l. on the eastern edge of the Cotahuasi canyon
near Huachuy (No. 2, Table 1). This cannot be attributed to
some local effect because most of the incision was
achieved before the middle Pleistocene in the middle reach
of the Colca canyon, as suggested by 0.53 Ma-old lava
flows forming topographic benches 0.3 to 0.6 km above
the valley floor. However, accelerated erosion is detected
during early to middle Pleistocene times when the vertical
gradient of the canyon wall has increased due to the thick
lava flows, which were added to the pre-existing canyon
edges. There is no direct evidence for either tectonic uplift,
increased runoff, or isostatic rebound following the deep
Neogene incision phase. However, given the maximum
computed rate of incision (1.7 km in 0.45 Ma, i.e.
∼ 3.8 mm yr
− 1
), there is also no reason to assume a decline
in either mean uplift rate or mean glacial meltwater supply
exactly during this brief period of the middle to late
Pleistocene.
7.2. Variable incision rates along canyon transects
Whereas our observations broadly coincide with recent
findings (Ghosh et al., 2006; Garzione et al., 2006;
Schildgen et al., 2007), with this study we provide finer
resolution to show that incision (and inferred uplift) was
neither uniformly fast between 13 and 3.8 Ma, nor spatially
consistent across the western Central Andes. Based on the
20-mDEM(withanerrorofc.50m),avolumeof
∼ 1960 km
3
has been removed from the 160 km-long
Cotahuasi–Ocoña canyons (excluding recent volcaniclas-
tic deposits). At least 75% of that volume was removed
before 3.8 Ma in the Cotahuasi valley, and before 4.8 Ma at
the Ocoña–Marán confluence. The mean vertical incision
rate (0.19 mm yr
− 1
in the area of Cotahuasi, Fig. 5) is
inferred by computing the elevation difference between the
edge of the Huaylillas plateau ignimbrite and the current
river bed. The mean bedrock incision rate was moderate
(0.21 mm yr
− 1
) between 13 and 9 Ma (before the Caravelí
ignimbrite pulse), but it probably doubled between 9 and
3.8 Ma. This is sugg ested by a 1.2 km drop in elevation in
the ‘Pliocene’ paleo longitudina l profile (Fig. 4)overa
30 km distance between Cotahuasi and Chaucalla (insert
Fig. 1, and the Appendix).
The re-incision rate through the upper Sencca pyroclas-
tic valley fill was ten times faster after 1.4 Ma (1.2 mm yr
− 1
near the town of Cotahuasi, Electronic Supplement Fig. 1 in
the Appendix). Further downstream and since 1.36 Ma, the
Río Ocoña and its Arma tributary incised the volcanic fill
by as much as 1.4 km, i.e. at ∼1.0 mm yr
− 1
. However ,
these rapid incision rates only represent valley-fill re-
incision rate into young non-welded pyroclastic deposits,
and not a bedrock incision rate directly correlated to
tectonic uplift or increased runoff.
Incision rates have also fluctuated spatially across the
western slope of the Central Andes as the canyons cut
through three different structural units (Fig. 2). The Wes-
tern Cordillera downstream of the Front Range and the
piedmont and Coastal Cordillera were less incised than the
western edge of the Altiplano in early to middle Pliocene
times. For example, ignimbrites crop out 1–1.3 km above
the Ocoña canyon on the SW flank of Nevado Solimana,
where they form a prominent ridge (inset Figs. 3 and 5B)
163J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
above the confluence at 1.6 km below the reference
Huaylillas plateau ignimbrite surfaces. The pumice sole
layer of the lower member of the ignimbrite pile is dated at
4.84 Ma (No. 18, Table 1)andcanbetraceddownstream
where it crops out at 0.6 km above the present valley floor
at Chaucalla, 1.6 km below the reference Huaylillas
ignimbrite-clad plateau surface. This 4.8 Ma-old thalweg
therefore does not correlate with the ∼ 3.8-Ma-old and
relatively deeper thalweg (at 2.8 km a.s.l.) in the Cotahuasi
canyon above the town of Cotahuasi. This indicates an
increase in gradient of 8 m/km between the 4.8 Ma and the
3.8 Ma river paleoprofiles (Fig. 4). Canyon wall
lithologies within 1 km of the valley floor are similar, so
this difference may be due to renewed or faster incision
between 4.8 and 3.8 Ma. It cannot be ruled out that the
break in slope observed between the two river segments
also reflects differential uplift landward and seaward of the
Front Range. This would be the only indication of
localised tectonic movements at the orogenic front.
North of Lima, deep incision occurring 5.8 Ma ago had
already been inferred by Myers (Kennan, 2000)fromthe
fact that 5.84±0.2 Ma ignimbrites were preserved 0.15 to
0.5 km above the Río Fortaleza canyon bed. Bedrock
incision by drainage networks therefore took place earlier
in the northern Central Andes, suggesting a faster uplift
rate in the north during the Miocene. In theory, this could
also suggest earlier uplift in the north. However, given that
this strike-parallel gradient also reflects the present-day
climatic gradient in which precipitation increases from
south to north, the northward increase in the size and reach
of the valleys is probably a climatic signal, tectonic uplift
being assumed equal along that same gradient. Bedrock
incision by drainage networks therefore took place earlier
in Peru and is younger in northernmost Chile. Ignimbrites
of largely equivalent age (14 to 19 Ma) are at equivalent
elevations from S to N (4200–4500 m). However, the
valleys in southern Peru cut significantly deeper and
further east into the Altiplano than those in N Chile. This
directly implies that the present climate gradient (more
precipitation to the N) must have existed at least for the
past 5 Ma and possibly ever since 9 Ma. Thus, whereas the
incision histories of these valleys are mostly driven by
tectonic uplift, morphological differences between these
canyons may reflect a climatic gradient along the western
margin of South America that could have been in place
since at least ∼ 9Ma.
8. Conclusions
The deepest canyons in the Andes may be unique in
providing important insight into the uplift history of the
Central Andes. A chronostratigraphy of volcaniclastic for-
mations based on 42
40
Ar/
39
Ar age determinations indi-
cates that the deep valleys of southern Peru were incised
into the uplifting bedrock to near their present depths as
early as 3.8 Ma. This occurred on a landscape which, 13 to
14 My ago, consisted of a gently sloping paleosurface
mantled by the widespread Huaylillas ignimbrites. These
also cover a clastic wedge that records the first stage of
Andean uplift in early Miocene times.
Initial valley incision started with the formation of
broad, shallow valleys at ∼ 9 Ma. This is interpreted as a
response to continued uplift since 13 Ma, which is the
age of the ignimbrites capping the plateau. The time lag
between 13 and accelerated incision after 9 Ma could
reflect (1) limited susceptibi lity to erosion of the flat
paleosurface, (2) the hardness of the welded ignimbrite
caprock, combined with (3) low rainfall and runoff at
that time. The onset of increased downcutting after 9 Ma
is attributed to (1) continued uplift, (2) breaching of the
ignimbrite caprock, and (3) increased runoff. Moisture
supply would have been afforded by a relatively wetter
climate that has been documented for the Andes around
7Ma(Gaupp et al., 1999). However, the main driving
force for valley incision was uplift. In the absence of
significant shortening along the western margin of the
Andes (see also Schildgen et al., 2007), it is implied that
uplift was caused by regional tilting and lower crustal
flow rather than along discrete tectonic faults along the
orogenic front. As far as the canyon re-incision history
at the end of the Plioce ne is concerned, there is no direct
evidence for linking it with an increase in uplift rate after
3.8 Ma, or with isostasy due to glacial or canyon erosion
after 2.7 Ma (Marshall et a l., 1992).
As a result of deeper incision in Peru than Chile, the
valleys of southern Peru have also cut more deeply both
down and beyond the belt of the currently active volca-
noes. Erosion has caused repeated catastrophic landslides
and debris flows of volcanic and non-volcanic origin in the
canyons of southern Peru. These pose serious threats to
populated settlements. Additional hazards are also related
to potential dammed lake breakouts that may trigger
devastating debris flows towards the lower populated
valleys and towns of the Majes and Ocoña valleys.
Acknowledgements
This work is a contribution to the research prog-
ramme “Reliefs” funded by the Institut National des
Sciences de l'Univers. We thank Dr P. Soler and R.
Marocco (IRD, Lima), and our colleagues at IGP and
INGEMMET, Lima, for their assistance in Peru. One
age determination used as comparison for the Colca
valley was provided by M. Fornari (GeoAzur-IRD). We
164 J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
thank J. Hora, E. Defive, M. Veber, F. Albino, and P.
Paquereau-Lebti for their help in the field and the
laboratory. Financial support to Wörner was obtained
through grant Wo 362/18-1. Singer is grateful to the
National Science Foundation for partial support of this
work through grants EAR-0516760 and EAR-0337667.
Appendix A. Supplementary data
Supplementary data associated with this article can
be found, in the online version, at doi:10.1016/j.
epsl.2007.07.023.
References
Cruzado, G.H., Rojas, M.C., 2007. Estratigrafía de la Fm. Moquegua
Superior en el area del cerro Cuno Cuno y Pampas del Gramadal,
Tesis de grado, Universidad Mayor de San Marcos, Lima, 95 pp.
Delacour, A., Gerbe, M.-C., Thouret, J.-C., Wörner, G., Paquereau-Lebti,
P., 2007. Magma evolution of Quaternary minor volcanic centres in
Southern Peru, Central Andes. Bull. Volcanol. 69, 581–606.
Elger, K., Oncken, O., Glodny, J., 2005. Plateau-style accumula-
tion of deformation; southern Altiplano. Tectonics 24, TC4020.
doi:10.1029/2004TC001675.
Garzione, C.N., Molnar, P., Libarkin, J.C., MacFadden, B.J., 2006. Rapid
late Miocene rise of the Bolivian Altiplano: evidence for removal of
mantle lithosphere. Earth Planet. Sci. Lett. 241, 543–556.
Gaupp, R., Kött, A., Wörner, G., 1999. Paleoclimatic implications of Mio-
Pliocene sedimentation in the high-altitude intra-arc Lauca Bas in of
northern Chile. Palaeogeogr . Palaeoclimatol. Palaeoecol. 151, 79–100.
Gerbe, M.-C., Thouret, J.-C., 2004. Role of magma mixing in the
petrogenesis of lavas erupted through the 1990–98 explosive activity
of Nevado Sabancaya in south Peru. Bull. Volcanol. 66, 541–561.
Ghosh, P., Garzione, C.N., Eiler, J.M., 2006. Rapid uplift of the
Altiplano revealed through
13
C–
18
O bonds in paleosol carbonates.
Science 311, 511–515.
Gregory-Wodzicki, K.M., 2000. Uplift history of the Central and
Northern Andes; a review. Geol. Soc. Amer. Bull. 112, 1091–1105.
Hartley, A.J., Semperé, T., Wörner, G., 2007. A comment on “Rapid
late Miocene rise of the Bolivian Altiplano: evidence for removal
of mantle lithosphere” by Garzione, C.N. et al. [Earth Planet Sci.
Lett. 241 (2006) 543–556]. Earth Planet Sci. Lett. 259, 625–629.
Hoke, G., Isacks, B.L., Jordan, T.E., 2005. Equilibrium landscapes of the
Western Andean mountain front (1 0°S–33°S): long-term responses to
along-strike changes in climate. Sixth Int. Symp. Andean Geody-
namics, Barcelona. IRD Editions, pp. 386–389. Extended Abstracts.
Isacks, B.L., 1988. Uplift of the Central Andean Plateau and Bending
of the Bolivian Orocline. J. Geophys. Res. 93, 3211–3231.
Kennan, L., 2000. Large-scale geomorphology in the Andes: inter-
relationships of tectonics, magmatism, and climate. In: Summerfield,
M.A. (Ed.), Geomorphology and Global Tectonics. J. Wiley,
Chichester, pp. 167–199.
Klinck, B.A., Ellison, R.A., Hawkins, M.P., 1986. The geology of the
Cordillera Occidental and Altiplano, west of Lake Titicaca,
southern Peru. INGEMMET, Preliminary Report Lima.
Koppers, A. A.P., 2002. ArArCALC-software for
40
Ar/
39
Ar age
calculations. Comput. Geosci. 28, 605–619.
Lamb, S., Davis, P., 2003. Cenozoic climate change as a possible cause
for the rise of the Andes. Nature 425, 792–797.
Marshall, L.G., Swisher III, C., Lavenu, A., Hoffstetter, R., Curtis, G.H.,
1992. Geochronology of the mammal-bearing late Cenozoic on the
northern Altiplano, Bolivia. J. S. Am. Earth Sci. 5, 1–19.
Montgomery, D.R., Brandon, M.T., 2003. Topographic controls on
erosion rates in tectonically active mountain ranges. Earth Planet.
Sci. Lett. 201, 481–489.
Myers, J.S., 1979. Erosion surfaces and ignimbrite eruption, measures
of Andean uplift in northern Peru. Geol. J. 11, 29–44.
Noble, D.C., McKee, E.H., Eyzaguirre, V.R., Marocco, R., 1984. Age
and regional tectonic and metallogenic implications of igneous
activity and mineralization in the Andahuaylas–Yauri belt of
Southern Peru. Econ. Geol. 79, 172–176.
Olchauski, L., Dávila, D., 1994. Geología de los cuadrangulos de
Chuquibamba y Cotahuasi, Hojas 32q y 31q, 1:100,000, INGEM-
MET, Bol. 50, Serie A, Carta Geológica Nacional, Lima, 52 p.
Paquereau-Lebti, P., Thouret, J.-C., Wörner, G., Fornari, M., Macedo,
O., 2006. Neogene and Quaternary ignimbrites in the area of
Arequipa, southern Peru: stratigraphical and petrological correla-
tions. J. Volcanol. Geotherm. Res. 154, 251–275.
Pecho Gutierrez, V., 1983. Geología de los cuadrangulos de Pausa y
Caravelí, Hojas 31p y 32p, 1:100,000, INGEMMET, Bol. 37, Serie
A, Carta geológica nacional, Lima, 125 p.
Quang, C.X., Clark, A.H., Lee, J.K.W., Hawkes, N., 2005. Response
of supergene processes to episodic Cenozoic uplift, pediment
erosion, and ignimbrite eruption in the Porphyry Copper Province
of Southern Peru. Econ. Geol. 100, 87–114.
Renne, P.R., Swisher, C.C., Deino, A.L., Karner, D.B., Owens, T.L.,
Depaolo, D.J., 1998. Intercalibration of standards, absolute ages
and uncertainties in
40
Ar/
39
Ar dating. Chem. Geol. 145, 117–152.
Roperch, P., Sempere, T., Macedo, O., Arriagada, C., Fornari, M.,
Tapia, C., Laj, C., 2006. Counterclockwise rotation of lat e
Eocene–Oligocene fore-arc deposits in southern Peru and its
significance for oroclinal bending in the central Andes. Tectonics
25, TC3010.
Schildgen, T.F., Hodges, K.V., Whipple, K.X., Reiners, P.W., Pringle,
M.S., 2007. Uplift of the western margin of the Andean plateau
revealed from canyon incision history, southern Peru. Geology 35
(6), 523–526.
Singer, B.S., Ackert Jr., R.A., Guillou, H., 2004.
40
Ar/
39
Ar geochronol-
ogy of Pleistocene glaciations in Patagonia. Geol. Soc. Amer. Bull.
116, 434–450.
Smith, M.E., Singer, B.S., Carroll, A.R., Fournelle, J.H., 2006. High-
resolution calibration of Eocene strata:
40
Ar/
39
Ar geochronology of
biotite in the Green River Formation. Geology 34, 393–396.
Sébrier, M., Lavenu, A., Fornari, M., Soulas, J., 1988. Tectonics and
uplift in Central Andes (Peru, Bolivia and northern Chile) from
Eocene to present. Géodynamique 3, 85–106.
Thiede, R.C., Bookhage n, B., Arr owsmith, E.R., Sobel, E.R.,
Strecker, M.R., 2004. Climatic control on rapid exhumation
along the southern Himalayan front. Earth Planet. Sci. Lett. 222,
791–806.
Thouret, J.-C., Finizola, A., Fornari, M., Suni, J., Legeley-Padovani,
A., Frechen, M., 2001. Geology of El Misti volcano nearby the city
of Arequipa, Peru. Geol. Soc. Amer. Bull. 113, 1593–1610.
Thouret, J.-C., Wörner, G., Singer, B., Finizola, A., 2003. Valley
evolution, uplift, volcanism and related hazards on the western
slope of the central Andes. EGS-AGU-EUG Joint Assembly, Nice.
abstract and poster EAE03-A-10498 (TS7).
Thouret, J.-C., Rivera, M., Wörner, G., Gerbe, M.-C., Finizola, A.,
Fornari, M., Gonzales, K., 2005. Ubinas: evolution of the histori-
cally most active volcano in Southern Peru. Bull. Volcanol. 67,
557–589.
165J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
Tosdal, R.M., Farrar, E., Clark, A.H., 1981. K–Ar geochronology of
the Late Cenozoic volcanic rocks of the Cordillera occidental,
Southernmost Peru. J. Volcanol. Geotherm. Res. 10, 157–173.
Tosdal, R.M., Clark, A.H., Farrar, E., 1984. Cenozoic polyphase
landscape and tectonic evolution of the Cordillera Occidental,
southernmost Peru. Geol. Soc. Amer. Bull. 95, 1318–1332.
Whipple, R.X., Meade, B.J., 2006. Orogen response to changes in
climatic and tectonic forcing. Earth Planet. Sci. Lett. 243, 218–228.
Willett, S.D., 1999. Orogeny and orography: the effects of erosion on the
structure of mountain belts. J. Geophys. Res. 104, 28957–28965.
Wörner, G., Hammerschmidt, K., Henjes-Kunst, F., Lezaun, J.,
Wilke, H., 2000. Geochronology (
40
Ar–
39
Ar– ,K–Ar–,andHe–
exposure–) ages of Cenozoic magmatic rocks from Northern
Chile (18°–22°S): implications for magmatism and tecton ic
evolution of the central Andes. Rev. Geol. Chile 27, 205–240
(Soc. Geol.).
Wörner, G., Uhlig , D., Kohler, I., Seyfried, H., 2002. Evolution of
the West Andean Escarpment at 1 8°S (N. Chile) during the last
25 Ma: uplift, erosion and collapse through time. Tectonophysics
345, 183–198.
166 J.-C. Thouret et al. / Earth and Planetary Science Letters 263 (2007) 151–166
