Earth Planet. Sci. Lett.

637 (2024) 118728

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

journal homepage: www.elsevier.com/locate/epsl

Impact of river capture on erosion rates and offshore sedimentation

revealed by geological and in situ 10Be cosmogenic data (Corsica,
western Mediterranean)
Marco G. Malusà a, *, Alberto Resentini a, Hella Wittmann b
a
Department of Earth and Environmental Sciences, University of Milano-Bicocca, Piazza della Scienza 4, Milan 20126, Italy
b
Helmholtz Centre Potsdam, GFZ German Research Centre for Geosciences, Telegrafenberg, Potsdam 14473, Germany

A R T I C L E I N F O A B S T R A C T

Keywords: Quantitative analysis of fluvial topography and sediment yield changes are often independently used to detect
Paleo-ostriconi river catchment major river capture events and episodes of drainage reorganization. Here we use a unique set of geological and in
Pliocene river capture situ 10Be cosmogenic data from Corsica, Western Mediterranean, to provide evidence of major river capture
Provenance
events affecting the former Paleo-Ostriconi river catchment during the Pliocene, and to illustrate how the
Source-to-sink analysis
Long-lived knickpoints
landscape of Corsica is still reacting to the disequilibrium caused by the late Miocene uplift of Alpine Corsica. We
Holocene erosion rates found that ~1280 km2 of basin area originally draining towards the Ligurian Sea were abruptly connected to the
Tyrrhenian Sea by the capturing Tavignano and Golo rivers, which led to the formation of a large Pliocene-
Quaternary submarine fan offshore the Tyrrhenian coast. The increased sediment yield towards the Tyr­
rhenian margin after river capture in the Pliocene was three times greater than the average sediment yield in the
same source-to-sink system during the Holocene (410±100 t⋅km− 2⋅a − 1 vs ~131±8 t⋅km− 2⋅a − 1) and greater
magnitude than any subsequent peaks in sediment yield during late Pleistocene glaciations. 10Be-derived
denudation rates reveal that focused erosion still affects retreating knickpoints near the sites of former river
capture in central Corsica, suggesting persistence of landscape disequilibrium for several millions of years. Our
results demonstrate the potentially large impact of river capture on the stratigraphic record and highlight the
importance of full consideration of landscape response times to onshore disturbances for any reliable interpre­
tation of the offshore sedimentary archive.

1. Introduction whereas Calves et al. (2013) focused their attention on the offshore sink
and highlighted major variations in sediment yield during the late
An increasing number of source-to sink studies exploit the sedi­ Pleistocene and the last glacial cycle.
mentary archive to constrain landscape evolution and its response to In this study, we expand the analysis of the river network of Corsica
tectonic and climatic forcing (e.g., Zheng et al., 2013; Bracciali et al., back to the Miocene, when Alpine Corsica was first exposed above sea
2015; Castillo et al. 2017; Bender et al., 2020). Within this framework, level (e.g., Loÿe-Pilot et al. 2004; Malusà et al., 2015) (Fig. 1). We
sediment yield variations (e.g., Walford et al., 2005) and quantitative provide evidence of major river capture events affecting the former
analysis of fluvial topography (e.g. Gallen, 2018; Loget and Van Den Paleo-Ostriconi River in the late Neogene, and we demonstrate that
Driessche, 2009; Bowman, 2023) are often independently used to detect ~1280 km2 of basin area originally draining towards the Ligurian Sea
major river capture events and episodes of drainage re-organization was abruptly connected to the Tyrrhenian Sea through headward
(Shugar et al., 2017). The source-to-sink system of the Golo River, the erosion by the capturing Tavignano and Golo rivers. We document the
largest catchment of Corsica (Western Mediterranean) provides a impact of river piracy on offshore sedimentation and use a dataset of in
well-established test-case for this kind of studies. For example, Forzoni situ 10Be cosmogenic analysis on river sands to analyze the influence of
et al. (2015) and Molliex et al. (2021) analyzed the onshore part of the disequilibrium inherited from river capture on pattern and rates of
system during the late Quaternary climatic and sea level variations, Holocene erosion. Our results illustrate the complex landscape response

* Corresponding author.
E-mail address: [email protected] (M.G. Malusà).

https://doi.org/10.1016/j.epsl.2024.118728
Received 14 October 2023; Received in revised form 26 March 2024; Accepted 18 April 2024
Available online 26 April 2024
0012-821X/© 2024 Elsevier B.V. All rights reserved.
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

to tectonic uplift and provide a unique look at how a landscape reacts to 2. Geological setting
external forcing on different time scales.
The island of Corsica is located in the Western Mediterranean
(Fig. 1a), at the northern tip of the Corsica-Sardinia block, a continental

Fig. 1. Geologic setting of Corsica. a: Tectonic sketch map of the Western Mediterranean and Cenozoic evolution of the Adria-Europe plate boundary in four steps
(after Malusà et al. 2015). The red box indicates the location of frames (b) to (e). The purple arrows show relative Adria-Europe plate motion (numbers are ages in
Ma). Acronyms: AC, Alpine Corsica; IO, Ionian; LP, Ligurian-Provençal; OF, Ostriconi fault system; TY, Tyrrhenian; SA, Sardinia; VC, Variscan Corsica. b:
Shaded-relief map and associated river network (in blue). Thick red line = main drainage divide. Dashed red line = boundary of the Golo and Tavignano rivers
catchments. Numbers in black = elevation (m a.s.l.) of the main peaks indicated in (c). c: Elevation map showing the major planation surfaces of Variscan Corsica
(white crosses, Danišík et al. 2012) and the distribution of glaciers during the Wurmian maximum (Kuhlemann et al. 2008). Stars indicate the location of pictures
shown in Figs. 2, 4 and 5. White lines = main faults shown in (d). d: Geologic map simplified after Malusà et al. (2016). The red squares indicate the major
knickpoints of Fig. 4 (A=Asco; G=Golo; R=Restonica; T=Tavignano). e: Compilation of apatite fission-track (AFT) and (U-Th)/He (AHe) ages (after Zarki-Jakni et al.,
2004; Fellin et al., 2005; Danišík et al., 2007; 2012). f: E-W, N-S, and elevation distribution of AFT and AHe ages in Variscan and Alpine Corsica.

2
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

fragment bounded by the Ligurian-Provençal basin to the west and the 3. Methods
Tyrrhenian basin to the east (Rossi and Cocherie, 1991; Faccenna et al.,
2001). Corsica emerges from the Mediterranean Sea with a typical 3.1. Geological and geomorphological analyses
wedge-shaped profile, showing elevations exceeding 2700 m a.s.l. in NW
Corsica and gradually decreasing towards the SE (Fig. 1b, c). It mainly Field geological and geomorphologic analyses were carried out
consists of Paleozoic rocks classically referred to as Variscan Corsica (e. within the framework of extensive 1/10.000-scale geological mapping
g., Rossi et al., 2009), which are juxtaposed to the NE to the remnants of performed by two of the authors (M.M. and A.R.) in northern Corsica
a Cenozoic subduction wedge formed during Alpine subduction and between 2007 and 2019. We mapped marine terraces, raised beaches
classically referred to as Alpine Corsica (e.g., Caron et al., 1990; Malusà and river profile perturbations (e.g., knickpoints) and their relationships
et al., 2015) (Fig. 1d). with the underlying geology over an area of ~1000 km2. Our field ob­
The Paleozoic magmatic rocks forming most of Variscan Corsica are servations were subsequently extended to the entire northern Corsica.
encased into Panafrican units and Paleozoic high-grade metamorphic We initially focussed our analysis to the beheaded Ostriconi Valley
rocks (P′ and P in Fig. 1d) and belong to three different associations (U1 (Fig. 1b-d), carved along the Ostriconi Fault System and now occupied
to U3 in Fig. 1d) (Rossi and Cocherie, 1991): U1 Mg–K plutonic rocks by a river evidently undersized compared to the size of the valley. We
emplaced around 340 Ma and exclusively exposed in NW Corsica; U2 analyzed the modal composition of pebbles in Cenozoic conglomerates
calc-alkaline plutonic rocks (e.g., monzonite, quartz-monzonite) exposed within the Ostriconi drainage to detect potential changes in
emplaced around 320–290 Ma, and associated U2′ calc-alkaline volca­ eroding sources and past supply of detritus from outside the modern
nic rocks (e.g., andesite); and U3 alkaline and metaluminous magmatic Ostriconi catchment. This allowed us to reveal major river-capture
rocks (e.g., Monte Cinto reddish rhyolite and vesiculated basalt) events and infer the original extension of the newly defined Paleo-
emplaced around 290 Ma. Sparse remnants of Mesozoic sedimentary Ostriconi river catchment, and its progressive seizure through time.
successions are preserved along the Ostriconi Fault System (Fig. 1d), a Our field observations were cast within a more general framework
major left-lateral tectonic structure parallel to the former Alpine sub­ provided by GIS-based geomorphological analysis, including knickpoint
duction trench that remained active even after the late Eocene choking identification and chi-map creation using a digital elevation model with
of Alpine subduction (Fig. 1a). Eocene to lower-Miocene conglomerates 30 m spatial resolution (ASTER GDEM) and the TopoToolbox2 Matlab
and Nummulitic flysch sequences (E2 and E3 in Fig. 1d) are either found software (Schwanghart and Scherler, 2014). Stream network was
unconformably on top of the Paleozoic units of Variscan Corsica, or extracted after DEM carving using a minimum upslope area of 0.45 km2
partly accreted within the Cenozoic subduction wedge. (Fig. S1). Knickpoint identification was performed according to the
The Cenozoic subduction wedge of Alpine Corsica (A1-to-A3 in quantile carving approach of Schwanghart and Scherler (2017). A chi
Fig. 1d) shares the same tectonic structure as the Western Alps, and it map of the study area was created using a concavity m/n = 0.48 for the
was originally part of the same, continuous orogenic segment (Fig. 1a). stream power incision model as derived from the Taravo River in
Its frontal part includes slivers of European continental crust, very-low- southern Corsica, which was assumed to be in equilibrium (e.g., Perron
grade metaophiolites and flysch units (Balagne and Nebbio nappes) (A3 and Royden, 2013). Chi analysis integrates the stream power equation
in Fig. 1d), and greenschist-to-blueschist facies metaophiolites and from the outlet of the river to any given point along the river channel,
continental units (e.g., the Tenda unit, Rossetti et al., 2015) (A2 in originating a parameter of flow length normalized for drainage area that
Fig. 1d). The higher-pressure (>2 GPa) metaophiolites exposed on the can be used as a proxy of the state of equilibrium of the river network.
Tyrrhenian side of Corsica (A1 in Fig. 1d) consist of calcschist, meta­ To constrain the age of detected river capture events, we extended
gabbro and peridotite/serpentinite recording a late Eocene pressure our modal compositional analysis to the Cenozoic conglomerates
peak and subsequent fast exhumation (dark blue in Fig. 1a) (Malusà exposed outside the Ostriconi river catchment, based on an extensive
et al., 2015). compilation of literature data (Jauzein et al., 1976; Caron et al., 1990;
Apenninic subduction started affecting the southern tip of the former Cubells et al., 1994; Ferrandini et al., 1999; Lahondère et al., 1994; Rossi
Alpine subduction wedge by the end of the Oligocene (Fig. 1a), when et al., 1994; Loÿe-Pilot et al., 2004; Serrano et al., 2013; Molliex et al.,
Apenninic slab rollback started inducing extension in the Apenninic 2021) validated by original observations. We finally analyzed the
back arc leading to the opening of the Ligurian-Provençal basin in the offshore sink using seismic reflection lines from the literature (LISA01,
early Miocene (Rollet et al., 2002), and of the Tyrrhenian basin in the Contrucci et al., 2001; LISA10-W, Mauffret et al., 1999; BS97–22, Thi­
late Miocene (Mauffret et al., 1999). In Alpine Corsica, back arc exten­ non et al., 2016; Calcagno et al., 2004) and estimated the volume of
sion was associated with a trend of progressively younger AFT ages from sediment accumulated in the Tyrrhenian Sea after the detected
the west to the east (Fig. 1f). AFT data in Variscan Corsica define instead river-capture events. For time-to-depth conversion, we used a seismic
a trend of northward decreasing ages (Fig. 1e, f) that mainly record the velocity of 2.0 km s− 1 for the Plio-Quaternary strata and 4.4 km s− 1 for
progressive northward translation of the Apenninic slab beneath the the Miocene strata (e.g., Contrucci et al., 2001; Thinon et al., 2016).
European plate during the Oligocene (Fig. 1a) (Malusà et al., 2016). Estimated Plio-Quaternary sediment volumes were converted into
The Neogene uplift of Alpine Corsica above sea level is a first-order sediment yield by assuming a range of sediment densities between 1.4
event in the landscape evolution of the region, probably triggered by t⋅m− 3 (loosely packed sediment) and 2.3 t⋅m− 3 (strongly packed sedi­
the isostatic re-equilibration of the Corsica-Sardinia block after the ment) to consider the impact of sediment porosity (e.g., Manger, 1963).
opening of adjacent backarc basins (Malusà et al., 2016). Its age is
constrained by the first appearance of clasts of high-pressure meta­ 10
3.2. In situ Be cosmogenic analysis
ophiolitic rocks in the shallow-marine deposits of the Aleria Plain (N2 in
Fig. 1d) (Loÿe-Pilot et al. 2004) and by the tilting of the Neogene strata We collected six samples of modern river sands from the Ostriconi,
exposed near St-Florent (N1 in Fig. 1d) (Rossi et al., 1994). Flat surfaces upper Golo and Tartagine (S. Maria tributary) catchments for in situ 10Be
formed by marine abrasion are widespread in the landscape of Variscan cosmogenic analysis to constrain the pattern and rates of Holocene
Corsica (Danišík et al., 2012) (crosses in Fig. 1c). They are still preserved erosion. The sampling strategy was conceived to obtain the highest
between the major incisions of the modern river network (Fig. 1b) and number of nested sub-basins and sub-catchments (Granger et al., 1996)
were not obliterated by the impact of Pleistocene glaciations, which only taking advantage of the samples previously analyzed by Molliex et al.
affected the highest part of the drainage divide (Kuhlemann et al. 2008) (2017). We processed samples for in situ 10Be separation in the HELGES
(Fig. 1c). laboratory, GFZ Potsdam, using the revised methods of von Blancken­
burg et al. (2004) (see Supplementary material for analytical details).
10
Be/9Be ratios were measured at the AMS at the University of Cologne

3
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

(Dewald et al., 2013). To derive denudation rates, we calculated nuclide They are carved in Tenda Unit metagranitoids and metasediments at
production using the CRONUS-Earth online calculator version 2.3 and elevations as high as ~ 400 m a.s.l. in the Agriates Desert (Fig. 2b), and
the time-dependent scaling scheme of Lal/Stone (Lm) (Lal, 1991; Balco as high as 1000 m a.s.l. on the right slope of the Ostriconi Valley, where
et al., 2008). Calculation of basin-wide 10Be production rates (Ptotal, they display an evident stepped configuration (Fig. 2d). Marine terraces
at⋅g− 1⋅a− 1) was carried out for each pixel in a 90 m digital elevation of Alpine Corsica are locally covered by well-sorted sandstones with
model (DEM), including a correction for topographic shielding (Dunne typical swash cross stratification that represent remnants of raised
et al., 1999) that did not result in changes of more than 2% on pro­ beaches. These sandstones are preserved at ~40 m a.s.l. near Punta di
duction rates and was hence considered negligible. The absorption l’Acciolu (lat N 42.6912, long E 9.0681) (Fig. 2c), and at elevations as
depth scale, which is the vertical distance over which the cosmic-ray flux high as 95 m farther south near Monticellacciu (lat N 42.6709, long E
decreases over the e-folding length, divided by the denudation rate gives 9.0848). Marine terraces detected in Alpine Corsica are distributed
the integration timescale of the method (von Blanckenburg, 2005). through an elevation range that is far beyond the range of eustatic sea
Effective denudation rates within nested catchments were computed level oscillations. They cut the regional foliation of Alpine metamorphic
following the approach of Granger et al. (1996). Results were compared rocks and are thus demonstrably younger than Alpine metamorphism.
with published denudation rates from 10Be concentrations in granites Therefore, they are a useful marker to analyze the Neogene evolution of
from high-elevation paleosurfaces (Kuhlemann et al., 2008) and con­ the island.
verted into a 10Be-based sediment yield for an easier comparison with Field evidence reveal that, during Neogene uplift of Alpine and
the sediment yield resulting from the analysis of the sedimentary record Variscan Corsica these marine landforms were progressively overprinted
(e.g., Calves et al., 2013). by fluvial landforms related to a nascent river network (Fig. 1b), which
was in turn conditioned by the presence of major faults in the underlying
4. Results basement. The Ostriconi Valley (Fig. 3a) is carved just along the Ostri­
coni Fault System (Fig. 1d), which juxtaposes the very-low-grade met­
4.1. Marine vs fluvial landforms aophiolites and metasedimentary rocks of the Balagne Nappe to the west
(A3 in Fig. 1d) and the metagranitoids and paragneisses of the Tenda
Our field data reveals that flat surfaces related to marine abrasion (e. Unit to the east (A2 in Fig. 1d). The valley shows a smooth concave
g., Fig. 2a) are widespread not only in Variscan Corsica, as shown by longitudinal profile (brown in Fig. 4a) and a transverse profile that is
previous studies (e.g., Danišík et al., 2012), but also in Alpine Corsica. evidently too large to be related to the modern Ostriconi River (6 in

Fig. 2. Post-Oligocene rock uplift recorded by marine terraces. a: Marine terraces (marked by arrows) in plutonic rocks of Variscan Corsica (U1 in Fig. 1d)
exposed south of Calvi. b: Marine terraces (marked by arrows) in Tenda Unit metasediments, Alpine Corsica (A2 in Fig. 1d) located at elevations between 350 and
400 m a.s.l. (Agriates Desert locality). c: Raised beach with typical swash cross stratification preserved at ~40 m a.s.l. on top of the Tenda Unit paragneisses (Punta di
l’Acciolu locality). Similar deposits on Tenda Unit rocks are documented at elevations as high as 95 m (Monticellacciu locality). d: Series of elevated stepped marine
terraces in Tenda Unit metagranitoids (A2 in Fig. 1d) on the right slope of the Ostriconi Valley, at elevations between 500 and 1000 m a.s.l. (Monte a Lecchia
locality). Picture locations in Fig. 1c.

4
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

Fig. 3. Field geomorphologic evidence of river capture. a: View of the beheaded valley of the Ostriconi River, which has lost the upper part of its catchment
beyond the Pietralba Pass. The valley is carved along the Ostriconi Fault, which juxtaposes to the SW the very-low-grade ophiolites and (meta)sedimentary rocks of
the Balagne Nappe, and to the NE the metagranitoids and paragneisses of the Tenda Unit. On the left side of the photo is the alluvial fan of Urtaca, made up of
boulders of metagranitoids and paragneisses of the Tenda Unit. Behind the Pietralba Pass is Monte S. Petrone, consisting of high-pressure metaophiolites of the
Castagniccia tectonic dome. The transverse profile of the Ostriconi valley is too large to be related to the modern Ostriconi River, which is evidently underfit (cf.
Fig. 4b). b: The wide and relatively flat upper Golo valley seen from the summit of Monte Cinto. The arrow downstream the Calacuccia artificial lake indicates the
major knickpoint marked by a red square in Fig. 4a. Lithologies are the same before and after the knickpoint. In the background are the upper Tavignano and
Restonica valleys, also showing a major knickpoint away from the main divide (Fig. 4a). Alpine Corsica is in the background below the clouds due to its lower
elevation compared to Variscan Corsica. c: Steepened valley walls in the upper Golo valley, downstream the knickpoint shown in (b). d: View of the Porto River
showing steep walls cutting old planation surfaces, but no major knickpoints far from the main divide. Lithologies are the same as in the upper Golo valley. e: Upper
part of the Tavignano drainage interpreted as previously belonging to the Paleo-Ostriconi River catchment. The San Quilico Pass, in the background, was originally
located along the Paleo-Ostriconi valley floor, which was carved in correspondence of the Ostriconi Fault. Here the fault juxtaposes the Variscan basement, to the W,
and the high-pressure metaophiolites of the Castagniccia tectonic dome to the E (Monte S. Petrone). See picture locations in Fig. 1c.

Fig. 4b). The Ostriconi Valley is beheaded (Fig. 3a), and its original 4.2. Modal compositions of paleo-ostriconi deposits
catchment area likely included regions located to the south of the Pie­
tralba Pass (lat N 42.5392, long E 9.1841). The mouth of the modern Ostriconi River (Fig. 5a) is characterized
by fine-grained sediments typical of low-energy depositional environ­
ments, with no input of coarse detritus to the sea even during major

5
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

Fig. 4. River profiles, major knickpoints and sampling sites for cosmogenic 10Be analyses. a: The longitudinal profiles of the Restonica, Tavignano, upper Golo
and Asco rivers, flowing into the Tyrrhenian Sea (on the right), reveal major knickpoints away the main divide (dark red squares) that are not observed along rivers
that flow into the Ligurian Sea (on the left). Black dots in the map indicate sampling sites for cosmogenic 10Be analyses (S1 to S6, this work), smaller green dots
indicate samples analyzed by Molliex et al. (2017). b: Transverse profiles (see (a) for locations) highlight wider valleys for the Ostriconi and upper and intermediate
Golo compared to the narrower valleys of the lower Golo and Tavignano.

floods. Nevertheless, we detected outcrops of ancient coarse-grained found east of Punta di l’Acciolu, where the pebbly beach of Cala di a
fluvial and fan-delta deposits near the Ostriconi river mouth. One of Recisa (lat N 42.6917, long E 9.0685), near the eastern boundary of the
these outcrops is found in the Punta di Paraghiola locality (lat N Ostriconi Fault System, includes cobbles of serpentinite, greenschist,
42.6673, long E 9.0590), where former fan-delta deposits are partly reddish rhyolite and metagranite.
reworked to form a pebbly beach (Fig. 5e) (lat N 42.6717, long E Another relevant outcrop of ancient coarse-grained fluvial deposits
9.0605). Cobbles include metagranitoids of the Tenda Unit (Fig. 5f) and preserved within the modern Ostriconi river drainage is found in the
very-low-grade metasedimentary rocks of the Balagne Nappe, which Ogliastro locality (lat N 42.6512, long E 9.0810) (Fig. 5c). It includes not
form the metamorphic substratum of the river drainage, as well as rocks only cobbles of metagranitoids and paragneisses of the Tenda Unit and
not exposed today in the modern Ostriconi river catchment, such as metasediments of the Balagne Nappe, which form the metamorphic
high-pressure metaophiolites (metagabbro, greenschist, serpentinite substratum of the modern Ostriconi river drainage, but also cobbles of
and other ultramafic rocks; Fig. 5g-k) and U2-U3 magmatic rocks quartz-monzonite (Fig. 5d), andesite with plagioclase phenocrysts, and
(vesiculated basalt, vesiculated andesite with plagioclase phenocrysts, Monte Cinto rhyolite (marked by arrows in Fig. 5c) that are not exposed
monzonite with K-feldspar phenocrysts, quartz-monzonite and Monte today in the modern Ostriconi river catchment. These deposits will be
Cinto reddish rhyolite) (Fig. 5l-o). These deposits will be referred to as referred to as “Paleo-Ostriconi 2″ stage hereafter (Fig. 6a). Unlike the
“Paleo-Ostriconi 1″ stage hereafter (Fig. 6a). They attest to that the Anse de Peraiola deposits, they do not include cobbles of high-pressure
Ostriconi river drainage was much larger than today. The pie chart in metaophiolites. Cobbles of metagranitoids and paragneisses of the
Fig. 6a shows that cobbles derived from high-pressure metaophiolites (a Tenda Unit are dominant (b = 88%) compared to cobbles derived from
= 56.2%) and Alpine continental metamorphic rocks (b = 28.5%) are Variscan Corsica (c = 10%) and the Balagne Nappe (d = 2%). Cobbles of
dominant with respect to cobbles derived from Variscan rocks (c = Monte Cinto rhyolite are systematically smaller in size than the associ­
14.6%) and the Balagne Nappe (d = 0.7%) (see Supplementary ated cobbles of plutonic rocks (Fig. 5c), as also observed in the modern
Tables 1–2 for further details on modal composition of Paleo-Ostriconi fluvial deposits of the upper Golo valley, which are exclusively derived
detritus). Other remnants of reworked Paleo-Ostriconi 1 deposits are from erosion of Variscan rocks (Fig. 5b). Smaller remnants of Paleo-

6
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

Fig. 5. Field sedimentology evidence of river capture. a: Fine-grained sediments and associated low-energy depositional environments at the mouth of the
modern Ostriconi River. b: Modern sediments of the upper Golo River including cobbles of monzonite, quartz-monzonite, andesite (U2 magmatic suite sensu Rossi
and Cocherie 1991) and smaller cobbles of reddish rhyolite (marked by arrows) of the ancient Monte Cinto caldera (U3 magmatic suite sensu Rossi and Cocherie
1991). c, d: Outcrop of ancient fluvial sediments of the “Paleo-Ostriconi 2″ stage (Ogliastro locality) composed of metagranitoids and paragneisses of the Tenda Unit
but also including cobbles of quartz-monzonite (see close-up view in (d)), andesite with plagioclase phenocrysts, and Monte Cinto rhyolite (marked by arrows in (c))
not exposed today in the modern Ostriconi river catchment. e: Pebbly beach of reworked deposits of the “Paleo-Ostriconi 1″ stage (Anse de Peraiola locality) including
abundant pebbles derived from high-pressure metaophiolitic units of Alpine Corsica (see frames (g) to (k)) and pebbles derived from U2-U3 magmatic suites of
Variscan Corsica (see frames (l) to (o)), which are not exposed today in the modern Ostriconi river catchment. Close-ups: f: metagranitoid of the Tenda Unit. g:
high-pressure calcschist. h: metagabbro; i: greenschist. j: ultramafics (peridotite/pyroxenite). k: serpentinite. l: vesiculated basalt (U3 suite). m: vesiculated andesite
with plagioclase phenocrysts (U2 suite); n: monzonite with K-feldspar phenocrysts (U2 suite) and Monte Cinto reddish rhyolite (U3 suite). o: quartz-monzonite (U2
suite). Remnants of coarse-grained Paleo-Ostriconi fluvial deposits are also found on the mountains in the background of picture (a), at elevations as high as 110 m a.
s.l., and include boulders of U2 quartz-monzonite and U3 rhyolite (Monticellacciu and Cima Forca localities). See picture locations in Fig. 1c.

7
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

Fig. 6. Changes in modal compositions and overincision during progressive river capture. a: Inferred Paleo-Ostriconi 1 (t1) and Paleo-Ostriconi 2 (t2)
paleodrainages, and modal compositions of associated pebbly deposits (pie charts) compared to the modern Ostriconi River (t3). Pebble derivation: a, dark green =
Castagniccia metaophiolites; b, pink = Alpine continental metamorphic rocks (e.g., Tenda Unit); c, violet = Paleozoic units of Variscan Corsica; d, light green =
Alpine very-low-grade metasediments (Balagne Nappe). The black lines in the map indicate drainage boundaries, the thick brown line indicates the position of the
main divide. b: Transverse profiles of modern rivers rearranged according to their position in the inferred Paleo-Ostriconi 1 drainage network. The Pietralba and S.
Quilico passes, located along the original Paleo-Ostriconi valley floor, provide lower-bound constraints to the amount of overincision after capture by Paleo-
Tavignano (> 400 m overincision, in green) and Paleo-Golo (> 250 m overincision, in red). Profiles acronyms: A1-A3 = Asco; B = Bistuglio; G1-G7 = Golo; L1-
L2 = Lagani; O1-O3 = Ostriconi; R = Restonica; T1-T4 = Tavignano; Tt1-Tt2 = Tartagnine; V = Vecchio.

Ostriconi 2 deposits, with boulders of quartz-monzonite and reddish north and the Restonica and Tavignano rivers in the south (Fig. 4a).
rhyolite, are recognized in the modern Ostriconi river catchment at el­ Rivers flowing towards the Ligurian Sea (e.g., Porto and Fango) are
evations as high as 110 m a.s.l. near Monticellacciu (lat N 42.6703, long carved in the same lithologies as the upper Golo, upper Tavignano, Asco
E 9.0843) and Cima Forca (lat N 42.6624, long E 9.0810), respectively. and Restonica, and often show steep walls cutting old planation surfaces
Notably, deposits with mixed Alpine-Variscan composition characterize (Fig. 3d). However, they do not show any major knickpoint away from
not only the coarse-grained deposits of the Paleo-Ostriconi stages 1 and the main divide, as instead observed in some of the rivers flowing to­
2, but also the Pleistocene-Holocene successions exposed on the eastern wards the Tyrrhenian Sea (Fig. 4a). A systematic analysis of transverse
coast of the island, as will be described in Section 4.4. (Fig. 7). valley profiles reveals that the Ostriconi and the upper and intermediate
Golo valleys are much wider than the lower segments of the Golo and
4.3. Distribution of river knickpoints Tavignano valleys (Fig. 4b). Chi analysis shows that the upper and in­
termediate segments of the Golo and Tavignano catchments have higher
The upper Golo valley is wide and relatively flat (Figs. 3b and 4b). It chi values (>5⋅103) than the nearby catchments (Fig. 8c), supportive of a
shows a major knickpoint and steepened valley walls downstream of the transient landscape with the drainage divide moving toward higher-chi
Calacuccia artificial lake (Fig. 3b-c). The knickpoint (lat N 42.3438, long headwaters.
E 9.0480) is indicated by a red square in Fig. 4a and is far from the main
divide. Lithologies are the same before and after the knickpoint, con­ 4.4. Modal composition of conglomerates outside the Ostriconi river
sisting of Paleozoic magmatic rocks (Fig. 1d). Notably, there are other drainage
three rivers flowing towards the Tyrrhenian Sea that also show major
knickpoints away from the main divide. They are the Asco River in the The sparse Eocene to mid-Miocene conglomerates preserved in

8
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

Fig. 7. Time constraints provided by sediment modal composition and low-temperature thermochronology. Biostratigraphic ages and provenance infor­
mation for the Miocene-to-Holocene formations of the Aleria Plain, Costa Verde and Marana Plain (see inset for locations) indicate an evolution of the Paleo-Ostriconi
river drainage bracketed between the late Tortonian and the Pliocene (data after Jauzein et al. 1976; Caron et al. 1990; Lahondère et al., 1994; Loÿe-Pilot et al. 2004;
Serrano et al. 2013). Provenance data for the Pleistocene-Holocene formations of the Marana Plain and cosmogenic data for the borehole GBEC5–2 conglomerate are
after Molliex et al. (2021). Apatite fission track and (U-Th)/He age constraints same as in Fig. 1. Neogene time scale, composite marine δ18O isotope sequence (in
blue), and planktonic foraminifera zonation after Gradstein et al. (2004). MSC= Messinian Salinity Crisis. Numbers in the map indicate successions described in Sect.
4.4 (1 = Palasca; 2 = Lozari; 3 = Vazzio; 4 = Punta di l’Acciolu; 5 = Francardo-Ponte Leccia; 6 = St-Florent; 7 = Aleria).

northern and central Corsica have modal compositions invariably Serravallian marls and calcarenites (Farinole Formation) with channel­
pointing to clastic sources located in Variscan Corsica or in the frontal ized conglomerate bodies mainly including Monte Cinto rhyolite cobbles
part of the Cenozoic subduction wedge (e.g., Rossi et al., 1994; Fer­ (Rossi et al., 1994).
randini et al., 1999; Loye-Pilot et al., 2004). The Eocene conglomeratic A similar picture is provided by the ~2 km thick stratigraphic suc­
successions of Palasca and Lozari and the Oligocene Vazzio conglom­ cession exposed in the Aleria Plain, which is almost complete from the
erate preserved on top of the Variscan basement (1 to 3 in Fig. 7) chiefly Burdigalian to the Messinian (Fig. 7). Until the early Tortonian, detritus
consist of cobbles of Paleozoic volcanic and plutonic rocks. The Eocene in the Aleria Plain was exclusively supplied from erosion of Variscan
succession of Punta di l’Acciolu (4 in Fig. 7), lying on top of the Tenda Corsica (Loye-Pilot et al., 2004; Serrano et al., 2013): (i) the
Unit, includes cobbles derived from the Tenda Unit, the Balagne Nappe mid-Burdigalian Saint-Antoine Formation includes (sandy) marls and
and Variscan Corsica, but no clasts of high-pressure metaophiolites massive breccia with blocks of greenish granite; (ii) the Langhian marls
(Rossi et al., 1994). The ~600 m thick Burdigalian - lower Tortonian and conglomerates of the Aghione Formation include cobbles of Varis­
succession exposed near Francardo-Ponte Leccia (5 in Fig. 7) consists of can rhyolite and granite; (iii) the Serravallian Alzitone Formation con­
continental conglomerates chiefly including Variscan rock cobbles sists of pebbly sandstones with cobbles of greenish granite; (iv) the lower
(Ortone and Francardo formations) and, in between, Burdigalian Tortonian Vadina Formation consists of marine sedimentary rocks with
lagoonal to shallow marine deposits with cobbles of Variscan rhyolite interleaved sandstone and conglomerate containing clasts of Variscan
(Taverna Formation) (Cubells et al. 1994). The ~400 m thick Miocene rhyolite and granite.
succession exposed near St-Florent (6 in Fig. 7) consists of: (i) basal On the eastern coast of the island, the first appearance of detritus
continental conglomerates (Fium Albino Formation) with cobbles of derived from high-pressure metaophiolites is recorded in the upper
Tenda Unit metagranitoids; (ii) mid-Burdigalian to lower Langhian Tortonian Casabianda Formation (Fig. 7) (Loÿe-Pilot et al. 2004). Later
calcarenites (Sant’Angelo Formation) with rhodoliths grown around on, high-pressure metaophiolites provided most of the detritus supplied
clasts derived from the underlying Nebbio Nappe, the Tenda Unit and to the lower Messinian Aleria Formation, including cobbles of calcschist,
Variscan Corsica, and Monte Cinto rhyolite clasts that become dominant greenschist, metachert, metagabbro, peridotite, and to a lesser extent
up section in the St Florent conglomerate; (iii) upper Langhian - lower Variscan granites and rhyolites. A mixed Alpine-Variscan provenance

9
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

Fig. 8. Holocene erosion pattern by cosmogenic 10Be data. a: Elevation map of the Paleo-Ostriconi river catchment and locations of detrital samples analyzed for
cosmogenic 10Be (the smaller grey dots indicate samples analyzed by Molliex et al. 2017). b: Map of 10Be-derived erosion rates (mm/ka, numbers in bold) with rates
for nested catchments computed according to the approach of Granger et al. (1996) (see Supplementary Table S3). Values in italics assume similar erosion rates in the
lower Golo valley (measured) and the lower Tavignano valley (inferred). Numbers in brown are denudation rates from 10Be concentrations in granites of
high-elevation paleosurfaces (Kuhlemann et al. 2008). c: Chi-map of northern Corsica. The higher chi-values in the captured Paleo-Ostriconi catchment, as compared
to the nearby catchments, indicate a transient landscape with the drainage divide moving toward high-chi headwaters. d: Reconstruction of the Paleo-Ostriconi 2
river profile and associated 10Be-derived erosion rates. Note the highest erosion rates still observed in correspondence of the capture site. River color code as in Fig. 4.

Table 1
In situ 10Be concentrations and denudation rates.
10
Sample River Quartz weight In situ Be concentration (×104at/ P_mu (z = 0) P_sp (z = Denudation rate (mm/ Apparent age (×103
Lab label Lat/Long (g) gQTZ) (at⋅g− 1⋅yr− 1) 0) ka) a)
(grain size)

S1 - MM1 Golo 43.43 13.09±0.54 0.1221 13.62 67.6 ± 4.1 12.2

(500–800) N
42.307370
E 8.946067
S2 - MM2A Golo 31.75 17.23±0.90 0.1226 13.90 52.1 ± 3.6 16.1
(250–500) N
42.318729
E 8.983192
S2 - MM2B Golo 41.77 17.75±0.78 0.1226 13.91 50.6 ± 3.2 16.1
(500–800) N
42.318729
E 8.983192
S3 - MM3 Golo 28.77 16.29±0.67 0.1187 12.95 51.6 ± 3.2 16.4
(500–800) N
42.373785
E 9.119820
S4 - MM4 S. Maria 37.62 10.24±0.49 0.0967 7.79 51.8 ± 3.4 16.0
(500–800) N
42.518805
E 9.171593
S5 - MM5 Ostriconi 32.45 11.98±0.59 0.0951 7.53 42.8 ± 2.9 21.3
(500–800) N
42.602333
E 9.138909
S6 - MM6 Ostriconi 25.45 12.56±0.60 0.0913 6.82 37.2 ± 2.4 23.8
(500–800) N
42.625068
E 9.100545

10
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

characterizes all the Quaternary deposits of the Tavignano river terraces (2017), the effective denudation rates computed for nested catchments
(Fz2-Fv in Fig. 7), which contain cobbles of Variscan granite and rhyolite provide a finer image of the Holocene erosion pattern in northern Cor­
beside cobbles of high-pressure calcschist, metagabbro and peridotite. sica (Fig. 8b and Supplementary Table 3). In the Ostriconi River catch­
By contrast, no Variscan rock cobble is observed in the Plio-Quaternary ment, 10Be-derived denudation rates decrease downstream, as normally
formations of the Costa Verde (Fig. 7), where the upper Pliocene Peri expected for a river in equilibrium, from ~43 to ~32 mm⋅ka− 1.
10
Formation exclusively contains cobbles of high-pressure metaophiolitic Be-derived denudation rates are ~30–33 mm⋅ka− 1 in the Lagani and
rocks, and the Quaternary deposits of the Bravona river terraces show a Tartagine catchments, and ~49–52 mm⋅ka− 1 in most of the Asco and
pure Alpine provenance with cobbles of high-pressure calcschist, met­ upper Golo catchments, where denudation rates are as low as ~15
agabbro, greenschist and glaucophanite (Fz2-Fv in Fig. 7) (Jauzein et al., mm⋅ka− 1 near the main drainage divide. Rates ~47 mm⋅ka− 1 are
1976; Caron et al., 1990). observed in the Restonica Valley, and rates ~74 mm⋅ka− 1 are found in
Farther north, the uppermost Miocene – Pliocene Casatora Forma­ the upper Tavignano Valley. These values are systematically higher than
tion exclusively contains clasts of high-pressure calcschist, greenschist the 10Be-derived denudation rates, between ~9 and ~24 mm⋅ka− 1,
and quartz (Lahondère et al., 1994). In the Marana Plain, the first evi­ measured in granites from high-elevation paleosurfaces (Kuhlemann
dence of mixed Alpine-Variscan composition is provided by the ~2.6 Ma et al., 2008) (Fig. 8b).
old conglomerate from the GBEC5 borehole (Molliex et al., 2021). Mixed A downstream increase in denudation rates, from ~62 to ~95
Alpine-Variscan compositions are also invariably observed in the Qua­ mm⋅ka− 1, is observed along the Casaluna catchment. The highest
ternary deposits of the lower Golo river terraces (Fz to Fv in Fig. 7). denudation rates (>200 mm⋅ka− 1) are found in central Corsica, in river
segments located along the Ostriconi Fault System (Fig. 8b), such as
10 along the intermediate Golo Valley to the south of the Pietralba Pass
4.5. In situ Be cosmogenic data
(Fig. 3a). Denudation rates drop sharply in the lower Golo Valley, where
they are much lower (<4 mm⋅ka− 1) than in the nearby catchments
In situ 10Be concentrations measured in samples S1 to S6 range from
draining towards the Tyrrhenian Sea (Bevinco, Fium Alto and Bravona
10.24±0.49 to 17.75±0.78 × 104 at⋅g−QTZ 1
(Table 1). 10Be-derived
in Fig. 8b). If similar denudation rates <4 mm⋅ka− 1 are considered for
denudation rates are between 67.6 ± 4.1 mm⋅ka− 1 in the uppermost
the lower Tavignano river segment, denudation rates would exceed 200
Golo (sample S1) and 37.2 ± 2.4 mm⋅ka− 1 near the mouth of the
mm⋅ka− 1 also in the intermediate Tavignano Valley, to the south of the
Ostriconi River (sample S6). Associated apparent ages range from 12.2
San Quilico Pass (Fig. 3e). In situ 10Be concentrations measured in the
ka (sample S1) to 23.8 ka (sample S6).
Golo and Tavignano catchments can be converted into an average
When these data are combined with previous data by Molliex et al.

Fig. 9. Impact of river capture on offshore sedimentation. a: Post-Oligocene drainage evolution and offshore sedimentation in four steps. The thick brown line
indicates the location of the main drainage divide. Offshore deposits related to the Messinian salinity crisis (MSC, yellow area) after Thinon et al. (2016).
Pliocene-Quaternary isopachs (km, in red) after Contrucci et al. (2001) for the Ligurian side (as inferred from the LISA01 profile) and Calcagno et al. (2004) for the
Tyrrhenian side (as reported in Thinon et al. 2016) using a seismic velocity of 2.0 km s − 1 for time-to-depth conversion. Base Langhian to base MSC sediment
thickness (km, in grey) after Contrucci et al. (2001) (LISA01) for the Ligurian side and Mauffret et al. (1999) (LISA10-W) for the Tyrrhenian side using a seismic
velocity of 4.4 km s − 1 for time-to-depth conversion. Onshore Holocene erosion foci (in red) as in Fig. 8b. Numbers in the early-middle Miocene map as in Fig. 7. b:
Sediment yield from the Golo and Tavignano catchments as constrained by: (i) 10Be data for the Holocene (131±8 t⋅km− 2⋅a − 1, this work); (ii) analysis of the
sedimentary record for the last 130 ka (blue empty boxes, after Calves et al. 2013); (iii) seismic stratigraphy for the Pliocene-Pleistocene (410±100 t⋅km− 2⋅a − 1, red
bar, this work; the error bar includes uncertainties due to time-to-depth conversion and porosity decay). The major increase in sediment yield after the MSC is coeval
with capture of the Paleo-Ostriconi catchment. c: Cross-section of the Plio-Quaternary submarine fan fed by the Tavignano and Golo rivers (seismic profile BS97–22
modified after Thinon et al. 2016), see location in (a). The surfaces bounding the MSC deposits (in yellow) join up at the edges of the basin to form a single surface
(MES). Seismic reflectors are weaker in the lower part of the fan, corresponding to the early stages of drainage reorganization (facies A), and more evident in the
upper part of the fan (facies B).

11
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

Holocene 10Be-based sediment yield of ~131±8 t⋅km− 2⋅a− 1

(Supple­ catchment to the south of the Pietralba Pass, which likely included the
mentary Table 4). upper and intermediate Golo and Tavignano, Tartagine, Asco, Restonica
and Vecchio rivers, originally forming a dendritic pattern that can be
4.6. Sediment volume in the offshore sinks recognized in map view even today (see Fig. S2). Fig. 6b shows the
modern river profiles rearranged according to their position in the
Seismic lines LISA01 and LISA10-W reveal that the thickness of the inferred Paleo-Ostriconi 1 catchment and shows that transverse river
depositional sequences, from the base Langhian to the base of Messinian profiles from outside the Paleo-Ostriconi 1 catchment (e.g., G5–6 and
salinity crisis (MSC) deposits, is greater in the Ligurian Sea offshore the T3–4 in Fig. 6b) are invariably narrower and likely at a different stage of
Ostriconi river mouth compared to the Tyrrhenian Sea offshore the Golo evolution. During the Paleo-Ostriconi 1 stage, the main watershed in
and Tavignano river mouths (~3.1 vs. ~2.4 km, respectively, in grey in northern Corsica was located much further east than today, i.e., in
Fig. 9a). By contrast, the Pliocene-Quaternary isopachs (in red in Fig. 9a) Alpine Corsica (Fig. 6a).
indicate much thicker Pliocene-Quaternary deposits offshore the Tyr­ Clast lithologies in gravels of the Paleo-Ostriconi 2 stage (t2 in
rhenian side of the island (Calcagno et al., 2004; Thinon et al., 2016) Fig. 6a) reveal a first major change in drainage network. They include
compared to the Ligurian side (>1.0 km vs. ~0.4 km). The metagranitoids and paragneisses of the Tenda Unit (b = 88% in Fig. 6a),
Pliocene-Quaternary deposits offshore the Tyrrhenian side define a very low-grade rocks of the Balagne Nappe (d = 2%), and a suite of
coalescing fan shape geometry, which is evident both in map view quartz-monzonite, andesite with plagioclase phenocrysts, and Monte
(Fig. 9a) and in cross section (Fig. 9c). Seismic reflectors are weaker in Cinto rhyolite already observed during the Paleo-Ostriconi 1 stage (c =
the lower part of the fan and more evident in the upper part, where the 10%). However, no cobble of high-pressure metaophiolites is observed
typical architectures of a small turbidite system are described in detail in the Paleo-Ostriconi 2 gravels. The disappearance of metaophiolitic
by previous work (e.g., Gervais et al., 2004; Sømme et al., 2011; Sweet clasts (56% of the total amount of clasts during the Paleo-Ostriconi 1
et al., 2020; Calves et al., 2013). The apex of the fan is in the region stage) is not attributable only to a sampling bias and indicates that part
between the Golo and Tavignano river mouths, and the of the Paleo-Ostriconi 1 basin area was abruptly connected to the Tyr­
Pliocene-Quaternary depocenter is evidently shifted northwards rhenian Sea by the capturing Tavignano River. Based on the recon­
compared to the underlying and much thinner MSC deposits (Fig. 9a). struction of Fig. 6b, we estimate that ~475 km2 of basin area was
Based on these observations, the volume estimate of these diverted at that time. The elevation of the S. Quilico Pass (1 in Fig. 6a),
Pliocene-Quaternary fan deposits (~2.1 × 1012 m3), which are evidently located along the original Paleo-Ostriconi 1 valley floor, provides a
sourced from Corsica, can be used to evaluate an average lower-bound constraint to the amount of overincision from the time of
Pliocene-Quaternary sediment yield from the Golo and Tavignano river capture. It likely exceeded 400 m (Fig. 6b), pointing to an average
catchments, which is equal to ~410±100 t⋅km− 2⋅a− 1 (Supplementary erosion rate >200 mm⋅ka− 1 in the past 5 Ma, which is consistent with
Table 4). the 10Be-derived denudation rates documented during the Holocene
near the capture site. During the Paleo-Ostriconi 2 stage, the Paleo-
5. Discussion Ostriconi River was still able to convey coarse debris from the Monte
Cinto caldera to the modern Ostriconi river mouth.
5.1. Drainage evolution and river capture events A second major episode of river piracy affecting the Paleo-Ostriconi
basin was governed by the capturing Paleo-Golo River. This capture
The bedrock gorges of central Corsica testify to a rapid base-level abruptly connected to the Tyrrhenian Sea ~805 km2 of basin area
drop in the drainage network of the island (Fig. 3b-d). The presence of chiefly including Variscan magmatic rocks and originally belonging to
major knickpoints along the longitudinal profiles of the Restonica, the Paleo-Ostriconi 2 catchment, leading to the river network configu­
Tavignano, upper Golo and Asco rivers, all flowing towards the Tyr­ ration observed today (Fig. 6a). Since then, no supply of coarse debris
rhenian Sea, and the absence of major knickpoints away from the main from Variscan Corsica (c in Fig. 6a) has ever reached the mouth of the
divide in rivers flowing towards the Ligurian Sea (Fig. 4a), cannot be small (~65 km2) and beheaded Ostriconi Valley (Fig. 5a), because that
explained exclusively with an overincision due to a relative lowering of debris is funneled today into the narrow, V-shaped lower Golo Valley
sea level linked to eustatic causes or to the tectonic uplift of the entire (G5 and G6 in Fig. 6b).
island (e.g., Malusà et al., 2016). Instead, they also require a transient Alternative hypotheses to the proposed two-stage river capture
response of the drainage network to a more localized lowering of the evolution are not consistent with field data. A shorter Paleo-Ostriconi
base level due to river capture, resulting in reorganization of the river would not explain the occurrence of high-pressure metaophiolitic
network. clasts in the Paleo-Ostriconi 1 deposits. A single-stage evolution would
The morphology of the modern Ostriconi Valley, which is carved not explain the disappearance of the high-pressure metaophiolitic clasts
along the NNW-SSE Ostriconi Fault, and the provenance of associated when moving from the Paleo-Ostriconi 1 to the Paleo-Ostriconi 2 de­
river gravels provide additional evidence of a major river capture. The posits. A two-stage evolution starting with the capture by the Paleo-Golo
Ostriconi Valley is clearly beheaded (Fig. 3a), which indicates that the followed by the capture by the Paleo-Tavignano would not explain the
upper part of the Ostriconi catchment was originally extended to the presence of clasts of igneous rocks from Variscan Corsica in the absence
south of the Pietralba Pass, where chi analysis is also supportive of a of clasts of high-pressure metaophiolitic rocks in the Paleo-Ostriconi 2
transient landscape (Fig. 8c). deposits.
Clast lithologies in gravels of the Paleo-Ostriconi 1 stage (t1 in
Fig. 6a) are dominated by rocks not exposed today in the Ostriconi river 5.2. Age of river capture and impact on offshore sedimentation
catchment. They include a distinctive suite of metagabbro, greenschist,
serpentinite and other ultramafics rocks (a = 56.2% in Fig. 6a) that The Paleo-Ostriconi 1 drainage network was necessarily established
indicates sourcing from the high-pressure metaophiolites of the Cas­ after the uplift and subaerial exposure of the high-pressure meta­
tagniccia tectonic dome (Monte S. Petrone), which are exposed farther ophiolites of Alpine Corsica. The age of this major geologic event is
to the SE. A second distinctive suite of vesiculated basalt, vesiculated constrained to the late Tortonian by the first appearance of meta­
andesite with plagioclase phenocrysts, monzonite with K-feldspar phe­ ophiolitic debris in the Casabianda Formation (Aleria Plain, Fig. 7)
nocrysts, quartz-monzonite and Monte Cinto reddish rhyolite (c = (Loye-Pilot et al., 2004). The Tenda Unit and Variscan Corsica have long
14.6%) indicate instead sourcing from U2-U3 magmatic rocks of represented the exclusive source of detritus accumulated around Cor­
Variscan Corsica exposed farther to the SW. Such a range of clast li­ sica, at least between the Eocene and the middle Miocene (Fig. 9),
thologies allow for a detailed reconstruction of the Paleo-Ostriconi 1 whereas in the late Tortonian metaophiolitic debris became

12
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

overwhelming all along the eastern coast of the island (Fig. 7). The offshore the Golo during the last glacial cycle (blue boxes in Fig. 9b),
renewed appearance of Variscan rock cobbles in fluvial conglomerates which is anyway smaller than the peak in sediment yield triggered by
exposed on the eastern coast of the island provides an upper bound the river capture events documented in this study. Glaciers in Corsica
constraint to the age of the Paleo-Golo river capture, which marks the were exclusively developed on resistant Variscan rocks on the left slopes
end of the Paleo-Ostriconi 2 stage. These cobbles are found in the ~2.6 of the Paleo-Ostriconi drainage basin, with the Golo and Asco knick­
Ma old conglomerate (26Al/10Be burial ages by Molliex et al., 2021) points located outside the main Wurmian glaciers, and the Tavignano
sampled in the GBEC5 borehole in the Marana Plain. Therefore, the and Restonica knickpoints located just at their boundaries (Fig. 1c).
two-stage river capture evolution illustrated in Fig. 6a can be bracketed Detritus supply per unit area was systematically higher in the eastern
between the late Tortonian and the Pliocene, which is also consistent side of the (Paleo)Ostriconi Valley, carved in less resistant high-pressure
with the lower bound constraints provided by AFT and AHe thermo­ rocks (a and b in Fig. 6a) and hosting the largest alluvial fans (e.g., the
chronologic ages (Fig. 1e-f, Fig. 7). These thermochronologic ages were Urtaca alluvial fan, Fig. 3a), compared to the glaciated western side of
set before the drainage network was established. Overincision during the valley mainly carved in more resistant Variscan rocks (c in Fig. 6a).
and after river capture was not deep enough to reach the partial This suggests a minor influence of the small Corsica glaciers compared to
annealing/retention zones of the AFT and AHe thermochronologic sys­ lithology, at least during the interglacial stages.
tems, which implies that exhumation linked to river capture is not ex­
pected to be recorded by low-temperature thermochronometers in rocks 6. General implications
exposed today in central Corsica.
Starting from the Pliocene, ~1280 km2 of basin area originally The unusually well documented basin history in Corsica provides a
draining towards the Ligurian Sea were thus abruptly connected to the unique look at how a landscape reacts via multi-stage river capture to
Tyrrhenian Sea by the capturing Tavignano and Golo rivers. This event the perturbation caused by a major tectonic uplift event. It demonstrates
determined a shift of depocenter in the Tyrrhenian Sea (Fig. 9a) and the that river captures are drivers of landscape modification with a poten­
formation of a large Pliocene-Quaternary submarine fan offshore the tially large impact on the stratigraphic record. However, a major change
region between the Golo and Tavignano river mouths. The volume of the in sediment flux due to river piracy accommodating tectonic uplift can
fan (~2.1 × 1012 m3) implies an average Pliocene-Quaternary sediment occur millions of years after the triggering tectonic event (3 Ma in
yield from the Golo and Tavignano catchments that is three times Corsica). Although the sediment yield after river capture is generally
greater than the average Holocene sediment yield from the Golo and much greater than the average sediment yield observed in the same
Tavignano based on in situ 10Be concentrations (410±100 t⋅km− 2⋅a− 1 vs source-to-sink system at a later stage of evolution, the concentration of
~131±8 t⋅km− 2⋅a− 1). These findings indicate a higher sediment yield erosion at long-lived knickpoints can be observed at even longer time­
during the early stages of drainage reorganization. scales after tectonic uplift (7–8 Ma in the Corsica case). Using the
sedimentary archive to constrain the tectonic growth of topography thus
5.3. Impact on pattern and rates of holocene erosion requires that the different timescales of landscape response to tectonic
perturbations are duly accounted for, with important implications for
Although the diversion of the Paleo-Ostriconi river catchment to the any source-to-sink study aimed at interpreting the stratigraphic record
Tyrrhenian Sea ended no later than ~ 2.6 Ma ago, the erosion pattern of in terms of tectonic or climate variations.
central Corsica during the Holocene still reflects the Pliocene fluvial
capture events, as illustrated by the reconstruction of the Paleo- 7. Conclusions
Ostriconi river profile and associated 10Be-derived denudation rates
(Fig. 8d). 10Be-derived denudation rates measured in modern river sands Our study illustrates how the landscape of Corsica has reacted and is
are systematically higher than the 10Be-derived denudation rates still adapting to the disequilibrium caused in the late Miocene by the
measured in granites from high-elevation paleosurfaces. This suggests uplift of Alpine Corsica. Based on a unique set of geological and in situ
10
ongoing active erosion along the steep Corsican gorges that formed after Be cosmogenic data, we demonstrate that tectonic uplift pushed the
capture-related base level lowering. drainage divide, originally located on Variscan Corsica, eastward and
Given the age of river capture and the distance of knickpoint retreat generated a linear, fault-controlled valley that hosted the north-flowing
(e.g., Bowman, 2023), we can infer that the upper Tavignano knickpoint Paleo-Ostriconi River. The landscape responded to the uplift via multi-
migrated ~10 km in 5 Ma, at a rate of ~2 mm⋅a− 1. The upper Golo stage river capture, and the divide eventually migrated back to the
knickpoint migrated 8–9 km in 4 Ma, at a similar rate of ~2 mm⋅a− 1, west to again become largely fixed on the resistant Variscan spine. In
with an upper bound of ~3.5 mm⋅a− 1 defined by the 2.6 Ma 26Al/10Be this framework, ~1280 km2 of basin area originally draining towards
burial age of the GBEC5–2 conglomerate. These values are consistent the Ligurian Sea were abruptly connected, since the Pliocene, to the base
with previous studies on bedrock gorges (e.g., Weissel and Seidl, 1997; level of the capturing Tavignano and Golo rivers. This led to the for­
Loget and Van Den Driessche, 2009), and explain the highest denudation mation of a large Pliocene-Quaternary submarine fan offshore the Tyr­
rates (~220 mm⋅ka− 1) provided by cosmogenic data along segments of rhenian coast, associated to an increased sediment yield that was three
the Ostriconi Fault System corresponding to previous capture sites (i.e., times greater than the average sediment yield in the same source-to-sink
south of the Pietralba Pass and south of the S. Quilico Pass) (Fig 8b). system during the Holocene. Concentrated erosion persists today around
These 10Be-derived denudation rates are four-to-five times greater than long-lived knickpoints that are still moving through the drainage system
Holocene average erosion rates measured along the river network (Fig after millions of years. Our results demonstrate that a full consideration
8b), and likely integrate a basin-wide average denudation rate with a of landscape response times to onshore disturbances is a prerequisite for
much faster denudation rate characterizing the regions around the any reliable interpretation of the offshore sedimentary archive.
knickpoints. This demonstrates that the landscape is still responding to
the disequilibrium caused by the uplift of Alpine Corsica, with focused CRediT authorship contribution statement
erosion occurring at rates similar to the peak rates that affected the
system during the late Pleistocene glaciations (~219 mm⋅ka− 1, Calves Marco G. Malusà: Writing – original draft, Visualization, Resources,
et al. 2013). Investigation, Funding acquisition, Formal analysis, Conceptualization.
Alberto Resentini: Visualization, Software, Resources, Investigation,
5.4. Influence of glaciers and lithology Formal analysis. Hella Wittmann: Project administration, Investiga­
tion, Funding acquisition, Formal analysis.
Calves et al. (2013) highlighted an increase in sediment yield

13
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

Declaration of competing interest Ferrandini, J., Rossi, P., Ferrandini, M., Farjanel, G., Ginsburg, L., Schuler, M.,
Geissert, F., 1999. La formation conglomératique du Vazzio près d’Ajaccio (Corse-
du-Sud), un témoin des dépôts du Chattien supérieur continental synrift en
The authors declare that they have no known competing financial Méditerranée occidentale. Comptes Rendus de l’Académie des Sciences-Series IIA-
interests or personal relationships that could have appeared to influence Earth and Planetary Science 329 (4), 271–278.
the work reported in this paper. Forzoni, A., Storms, J.E.A., Reimann, T., Moreau, J., Jouet, G., 2015. Non-linear response
of the Golo River system, Corsica, France, to Late Quaternary climatic and sea level
variations. Quat. Sci. Rev. 121, 11–27.
Data availability Gallen, S.F., 2018. Lithologic controls on landscape dynamics and aquatic species
evolution in post-orogenic mountains. Earth Planet. Sci. Lett. 493, 150–160.
Gervais, A., Savoye, B., Piper, D.J., Mulder, T., Cremer, M., Pichevin, L., 2004. Present
Data will be made available on request. Morphology and Depositional Architecture of a Sandy Confined Submarine system:
the Golo turbidite System (eastern margin of Corsica), 222. Geological Society,
London, Special Publications, pp. 59–89.
Gradstein, F.M., Ogg, J.G., Smith, A.G, 2004. A Geologic Time Scale 2004, 86. Cambridge
Acknowledgements University Press.
Granger, D.E., Kirchner, J.W., Finkel, R., 1996. Spatially averaged long-term erosion
Cosmogenic analyses funded by DFG Project WI 3874/3–1 to H.W. rates measured from in situ-produced cosmogenic nuclides in alluvial sediment.
J. Geol. 104 (3), 249–257.
and M.M. We thank Stefan Heinze and Steve Binnie from the University Jauzein, A., Pilot, M.-D, Orszag-Sperber, F., 1976. Notice explicative, Carte géologique
of Cologne for performing AMS measurements, Philippe Rossi for the de La France (1: 50 000), Feuille Ghisonaccia. Bureau de Recherches Géologiques et
insightful discussions on the geology of Corsica, and two anonymous Minières Orléans n∘ 1119.
Kuhlemann, J., van der Borg, K., Bons, P.D., Danišík, M., Frisch, W., 2008. Erosion rates
reviewers for their constructive comments.
on subalpine paleosurfaces in the western Mediterranean by in-situ 10Be
concentrations in granites: implications for surface processes and long-term
Supplementary materials landscape evolution in Corsica (France). Int. J. Earth Sci. 97 (3), 549–564.
Lahondère, J.C., Conchon, O., Lahondère, D., Dominici, R., Vautrelle, C., 1994. Notice
explicative, Carte géologique de La France (1: 50 000), Feuille Vescovato. Bureau de
Supplementary material associated with this article can be found, in Recherches Géologiques et Minières Orléans n∘ 1107.
the online version, at doi:10.1016/j.epsl.2024.118728. Lal, D., 1991. Cosmic ray labeling of erosion surfaces: in situ nuclide production rates
and erosion models. Earth Planet. Sci. Lett. 104, 424–439.
Loget, N., Van Den Driessche, J., 2009. Wave train model for knickpoint migration.
References Geomorphology 106 (3–4), 376–382.
Loÿe-Pilot, M.D., Durand-Delga, M., Feinberg, H., Gourinard, Y., Magné, J., 2004. Les
Balco, G., Stone, J.O., Lifton, N.A., Dunai, T.J., 2008. A complete and easily accessible formations burdigaliennes de Corse orientale dans leur cadre géodynamique.
means of calculating surface exposure ages or erosion rates from 10Be and 26Al Comptes Rendus Géosci. 336 (10), 919–930.
measurements. Quat. Geochronol. 3, 174–195. Malusà, M.G., Faccenna, C., Baldwin, S.L., Fitzgerald, P.G., Rossetti, F., Balestrieri, M.L.,
Bender, A.M., Lease, R.O., Corbett, L.B., Bierman, P.R., Caffee, M.W., Rittenour, T.M., Ellero, A., Ottria, G., Piromallo, C., 2015. Contrasting styles of (U) HP rock
2020. Late Cenozoic climate change paces landscape adjustments to Yukon River exhumation along the Cenozoic Adria-Europe plate boundary (Western Alps,
capture. Nat. Geosci. 13 (8), 571–575. Calabria, Corsica). Geochem. Geophys. Geosyst. 16 (6), 1786–1824.
Bowman, D., 2023. Base-level Impact: A Geomorphic Approach. Springer Nature, p. 154. Malusà, M.G., Danišík, M., Kuhlemann, J., 2016. Tracking the Adriatic-slab travel
Bracciali, L., Najman, Y., Parrish, R.R., Akhter, S.H., Millar, I., 2015. The Brahmaputra beneath the Tethyan margin of Corsica–Sardinia by low-temperature
tale of tectonics and erosion: early Miocene river capture in the Eastern Himalaya. thermochronometry. Gondwana Res. 31, 135–149.
Earth Planet. Sci. Lett. 415, 25–37. Manger, G.E., 1963. Porosity and bulk density of sedimentary rocks. Geol. Surv. Bull.
Calcagno, P., Thinon, I., Courrioux, G., Guillen, A., Guenoc, P., 2004. 3D geometric 1144-E.
modelling: a tool for margin and basin interpretation illustrated with the eastern Mauffret, A., Contrucci, I., Brunet, C., 1999. Structural evolution of the Northern
Corse case-study (NW Mediterranean sea). In: Proceedings of the RST, Joint Earth Tyrrhenian Sea from new seismic data. Mar. Pet. Geol. 16 (5), 381–407.
Sciences Meeting, Société Géologique de France–Geologische Vereinigung. Molliex, S., Jouet, G., Freslon, N., Bourles, D.L., Authemayou, C., Moreau, J.,
Strasbourg. Rabineau, M., 2017. Controls on Holocene denudation rates in mountainous
Calves, G., Toucanne, S., Jouet, G., Charrier, S., Thereau, E., Etoubleau, J., Lericolais, G., environments under Mediterranean climate. Earth Surf. Processes Landforms 42 (2),
2013. Inferring denudation variations from the sediment record; an example of the 272–289.
last glacial cycle record of the Golo Basin and watershed, East Corsica, western Molliex, S., Jouet, G., Blard, P.H., Moreau, J., Demartini, J., Storms, J.E., Team, A., 2021.
Mediterranean sea. Basin Res. 25 (2), 197–218. Quaternary evolution of the Golo river alluvial plain (NE Corsica, France). Quat.
Caron, J.M., Loÿe-Pilot, M.D., Dominici, R., Scius, H., Vautrelle, C., 1990. Notice Geochronol. 61, 101115.
explicative, Carte géologique de La France (1: 50 000), Feuille Pietra di Verde. Perron, J.T., Royden, L., 2013. An integral approach to bedrock river profile analysis.
Bureau de Recherches Géologiques et Minières Orléans n∘ 1115. Earth Surf. Processes Landforms 38 (6), 570–576.
Castillo, M., Ferrari, L., Muñoz-Salinas, E., 2017. Knickpoint retreat and landscape Rollet, N., Déverchère, J., Beslier, M.O., Guennoc, P., Réhault, J.P., Sosson, M.,
evolution of the Amatlán de Cañas half-graben (northern sector of Jalisco Block, Truffert, C., 2002. Back arc extension, tectonic inheritance, and volcanism in the
western Mexico). J. South Am. Earth Sci. 77, 108–122. Ligurian Sea, Western Mediterranean. Tectonics 21 (3), 6-1.
Contrucci, I., Nercessian, A., Béthoux, N., Mauffret, A., Pascal, G., 2001. A Ligurian Rossetti, F., Glodny, J., Theye, T., Maggi, M., 2015.
(western Mediterranean Sea) geophysical transect revisited. Geophys. J. Int. 146 (1), Pressure–temperature–deformation–time of the ductile Alpine shearing in Corsica:
74–97. from orogenic construction to collapse. Lithos 218, 99–116.
Cubells, J.F., Ferrandini, J., Maurel-Ferrandini, M., Gaudant, J., Loÿe-Pilot, M.D., 1994. Rossi, P., Cocherie, A., 1991. Genesis of a Variscan batholith: field, petrological and
Présence Du Genre Aphanius Nardo, Famille Des Cyprinodontidae, 21. dans le mineralogical evidence from the Corsica-Sardinia batholith. Tectonophysics 195
Miocène du bassin de Francardo-Ponte Leccia (Corse). Géologie méditerranéenne, (2–4), 319–346.
pp. 19–24. Rossi, P., Lahondère, J.C., Lluch, D., Loye-Pilot, M.D., Jaquet, M., 1994. Notice
Danišík, M., Kuhlemann, J., Dunkl, I., Székely, B., Frisch, W., 2007. Burial and explicative, Carte Géologique de la France (1: 50.000) feuille Saint Florent, n∘ 1103.
exhumation of Corsica (France) in the light of fission track data. Tectonics 26 (1). Bureau De Recherches Géologiques et Minières Orléans. Notice explicative, BRGM,
Danišík, M., Kuhlemann, J., Dunkl, I., Evans, N.J., Székely, B., Frisch, W., 2012. Survival Orléans, p. 93.
of ancient landforms in a collisional setting as revealed by combined fission track Rossi, P., Oggiano, G., Cocherie, A., 2009. A restored section of the “southern Variscan
and (U-Th)/He thermochronometry: a case study from Corsica (France). J. Geol. 120 realm” across the Corsica–Sardinia microcontinent. Comptes rendus géoscience 341
(2), 155–173. (2–3), 224–238.
Dewald, A., Heinze, S., Jolie, J., Zilges, A., Dunai, T., Rethemeyer, J., Melles, M., Shugar, D.H., Clague, J.J., Best, J.L., Schoof, C., Willis, M.J., Copland, L., Roe, G.H.,
Staubwasser, M., Kuczewski, B., Richter, J., Radtke, U., von Blanckenburg, F., 2017. River piracy and drainage basin reorganization led by climate-driven glacier
Klein, M., 2013. CologneAMS, a dedicated center for accelerator mass spectrometry retreat. Nat. Geosci. 10 (5), 370–375.
in Germany. Nucl. Instrum. Methods Phys. Res. Sect. B 294, 18–23. Schwanghart, W., Scherler, D., 2014. Short Communication: topoToolbox 2 – MATLAB-
Dunne, J., Elmore, D., Muzikar, P., 1999. Scaling factors for the rates of production of based software for topographic analysis and modeling in Earth surface sciences.
cosmogenic nuclides for geometric shielding and attenuation at depth on sloped Earth Surf. Dyn. 2, 1–7.
surfaces. Geomorphology 27, 3–11. Schwanghart, W., Scherler, D., 2017. Bumps in river profiles: uncertainty assessment and
Faccenna, C., Becker, T.W., Lucente, F.P., Jolivet, L., Rossetti, F., 2001. History of smoothing using quantile regression techniques. Earth Surface Dyn. 5 (4), 821–839.
subduction and back arc extension in the Central Mediterranean. Geophys. J. Int. Serrano, O., Allanic, C., Magar, M., 2013. Synthèse Géologique du Bassin Tertiaire De La
145 (3), 809–820. Plaine Orientale Corse—Liaison Terre-Mer Entre San Nicolao et Solenzara. BRGM
Fellin, M.G., Zattin, M., Picotti, V., Reiners, P.W., Nicolescu, S., 2005. Relief evolution in report RP62303FR, p. 181.
northern Corsica (western Mediterranean): constraints on uplift and erosion on long- Sømme, T.O., Piper, D.J., Deptuck, M.E., Helland-Hansen, W., 2011. Linking
term and short-term timescales. J. Geophys. Res. Earth Surf. 110 (F1). onshore–offshore sediment dispersal in the Golo source-to-sink system (Corsica,
France) during the late Quaternary. J. Sediment. Res. 81 (2), 118–137.

14
M.G. Malusà et al. Earth and Planetary Science Letters 637 (2024) 118728

Sweet, M.L., Gaillot, G.T., Jouet, G., Rittenour, T.M., Toucanne, S., Marsset, T., Blum, M. von Blanckenburg, F., Hewawasam, T., Kubik, P.W., 2004. Cosmogenic nuclide evidence
D., 2020. Sediment routing from shelf to basin floor in the Quaternary Golo System for low weathering and denudation in the wet, tropical highlands of Sri Lanka.
of Eastern Corsica, France, western Mediterranean Sea. Bulletin 132 (5–6), J. Geophys. Res. Earth Surf. 109.
1217–1234. Walford, H.L., White, N.J., Sydow, J.C., 2005. Solid sediment load history of the Zambezi
Thinon, I., Guennoc, P., Serrano, O., Maillard, A., Lasseur, E., Rehault, J.P., 2016. Delta. Earth Planet. Sci. Lett. 238 (1–2), 49–63.
Seismic markers of the Messinian Salinity Crisis in an intermediate-depth basin: data Weissel, J.K., Seidl, M.A., 1997. Influence of rock strength properties on escarpment
for understanding the Neogene evolution of the Corsica Basin (northern Tyrrhenian retreat across passive continental margins. Geology 25 (7), 631–634.
Sea). Mar. Pet. Geol. 77, 1274–1296. Zarki-Jakni, B., van der Beek, P., Poupeau, G., Sosson, M., Labrin, E., Rossi, P.,
von Blanckenburg, F., 2005. The control mechanisms of erosion and weathering at basin Ferrandini, J., 2004. Cenozoic denudation of Corsica in response to Ligurian and
scale from cosmogenic nuclides in river sediment. Earth Planet. Sci. Lett. 237, Tyrrhenian extension: results from apatite fission track thermochronology. Tectonics
462–479. 23 (1).
Zheng, H., Clift, P.D., Wang, P., Tada, R., Jia, J., He, M., Jourdan, F., 2013. Pre-miocene
birth of the Yangtze River. Proc. Natl. Acad. Sci. 110 (19), 7556–7561.

15