Miocene basement exhumation in the Central Alps recorded by detrital garnet geochemistry in foreland basin deposits

The Neogene evolution of the European Alps was characterized by the exhumation of crystalline basement, the so-called external crystalline massifs. Their exhumation presumably controlled the evolution of relief, distribution of drainage networks, and generation of sediment in the Central Alps. However, due to the absence of suitable proxies, the timing of their surficial exposure and thus the initiation of sediment supply from these areas are poorly constrained. The northern Alpine foreland basin preserves the Oligocene to Miocene sedimentary record of tectonic and climatic adjustments in the hinterland. This contribution analyses the provenance of 25 to 14 Myr old alluvial fan deposits by means of detrital garnet chemistry. Unusually grossularand spessartine-rich garnet is found (1) to be a unique proxy for identifying detritus from the external crystalline massifs and (2) to occur abundantly in ca. 14 Myr old deposits of the foreland basin. In contrast to previous assumptions, we therefore propose that the external massifs were already exposed to the surface ca. 14 Myr ago.


Introduction
Tectonic processes drive the evolution of relief in mountain chains and consequently control the development of the drainage network, sediment supply, and deposition in the foreland basin.The Central European Alps and their northern foreland basin, formed through the collision of the European and the Adriatic continents since the Eocene (Schmid et al., 1996;Handy et al., 2010), are a classic example of such interactions (e.g.Schlunegger et al., 1998;Pfiffner et al., 2002;Vernon et al., 2008Vernon et al., , 2009;;Baran et al., 2014;Fox et al., 2015).The exhumation of large slices of mid-crustal rocks from the European plate, the so-called external crystalline massifs, occurred relatively late in the Alpine evolution, probably during the late Miocene, although the exact timing is not well constrained.The external crystalline massifs are today characterized by high relief, intense glaciation, and some of the highest denudation rates in the Alps (up to 1.4 mm yr −1 ), which all contribute to their relevance as a sediment source (Kühni and Pfiffner, 2001;Wittmann et al., 2007;Stutenbecker et al., 2018).The exhumation is discussed as being related to crustal delamination in response to lithospheric mantle rollback (Herwegh et al., 2017), slab detachment (Fox et al., 2015), or erosional unloading (Champagnac et al., 2009), possibly due to increased precipitation rates in the Pliocene (Cederbom et al., 2004) or enhanced glacial erosion in the Pleistocene (Fox et al., 2015;Herman et al., 2013).
Peak metamorphism of lower to upper greenschist facies conditions occurred between 17 and 22 Ma in all northern external crystalline massifs (Mont Blanc, Aar massifs, and the Gotthard nappe; Challandes et al., 2008;Rolland et al., 2008;Cenki-Tok et al., 2014;Nibourel et al., 2018).Their subsequent exhumation has been investigated using thermochronology (e.g.Schaer et al., 1975;Wagner et al., 1977;Michalski and Soom, 1990;Vernon et al., 2009;Glotzbach et al., 2010).Whereas some studies concluded that exhumation was episodic (e.g.Vernon et al., 2009), others suggest relatively constant exhumation rates of 0.5-0.7 km Myr −1 since 14 Ma (Michalski and Soom, 1990;Glotzbach et al., 2010).A focus in this debate concerns the late Neogene cooling and the onset of glaciation in the Pleistocene and their possible effect on the exhumation, erosion, and sediment accumulation rates (e.g.Kuhlemann et al., 2002;Herman et al., 2013;Schildgen et al., 2018).In contrast, the early Neogene exhumation history received comparably little attention.In particular, the timing of the first surficial exposure of the external massifs has never been constrained because estimates of their total thickness have not been established yet.In most geometric reconstructions (e.g.Pfiffner, 1986Pfiffner, , 2017;;Schmid et al., 2004), the contact between the crystalline basement and the overlying Mesozoic cover is assumed to be relatively flat, and the top of the crystalline basement is hypothesized to have been less than 1 km above the modern topography.Conversely, a new reconstruction of this tectonic contact allows for a substantially greater amount (∼ 8 km) of (now eroded) crystalline rock on top of the present-day topography (Nibourel et al., 2018).
This study aims to constrain the timing of exposure and thus the beginning of sediment supply from the external crystalline massifs, by determining the provenance of the foreland basin deposits.Sedimentary rocks preserved in the northern peripheral foreland basin of the Central Alps, the Swiss part of the Molasse basin, are a well-studied archive recording tectonic and climatic adjustments in the central orogen between ca.32 and 14 Myr ago (Schlunegger et al., 1993(Schlunegger et al., , 1996;;Kempf et al., 1999;Spiegel et al., 2000;Kuhlemann and Kempf, 2002;von Eynatten, 2003;Schlunegger and Kissling, 2015).So far, the provenance of the Molasse deposits has been investigated using optical heavy mineral analysis, framework petrography, and both bulk and single-grain geochemical techniques, including epidote geochemistry and cooling ages derived from zircon fission track analysis and Ar-Ar dating of white mica (Spiegel et al., 2000(Spiegel et al., , 2002;;von Eynatten, 2003;von Eynatten and Wijbrans, 2003).No conclusive evidence for a contribution from the external crystalline massifs, however, has been found thus far, leading to the assumption that their exposure must postdate the youngest preserved (ca.14 Myr old) Molasse deposits (von Eynatten, 2003).
In this study, we use major element geochemistry of detrital garnet in Miocene deposits from the central part of the Swiss foreland basin.The great compositional variability displayed by garnet from different source rocks means that it is a useful provenance tracer in a variety of settings (Spear, 1994;Mange and Morton, 2007).Furthermore, it is a common heavy mineral in orogenic sediments and sedimentary rocks (Garzanti and Andò, 2007) and is relatively stable during transport and diagenesis (Morton and Hallsworth, 2007).In the Central Alps, detrital garnet has recently been shown to be a valuable provenance indicator, especially for distinguishing detritus supplied from the external crystalline massifs (Stutenbecker et al., 2017).We aim (1) to explore if detrital garnet geochemistry can help identifying additional provenance changes in the Miocene Molasse deposits that have gone unnoticed so far and (2) to test whether detritus from the external massifs is present in the younger Molasse deposits in order to give independent constraints on the timing of crystalline basement exhumation.

Geological setting
The Central Alps evolved through convergence between the European continental margin in the north and the Adriatic plate in the south (Schmid et al., 1996).The convergence started during the Late Cretaceous with the subduction of the Alpine Tethys Ocean below the Adriatic microplate (Froitzheim et al., 1996) and ceased during the Paleogene after the European continental lithosphere entered the subduction zone.These Cretaceous to early Neogene orogenic processes are reflected by the syn-orogenic deposition of deep-marine flysch units preserved throughout the Alps (e.g.Wildi, 1985;Winkler, 1996).Around 32 Myr ago, the sedimentation style in the northern foreland basin changed from marine, flysch-like deposition to shallow marine and terrestrial sedimentation (Allen et al., 1991;Sinclair, 1997).This is thought to represent the transition to Molasse-type sedimentation in an overfilled basin and is discussed to be potentially related to a break-off of the European slab around the time of the Eocene-Oligocene boundary (e.g.Sinclair et al., 1991;Sinclair, 1997;Schlunegger and Kissling, 2015).Since this time, the northern foreland basin has become a major sink of orogenic detritus and an important sedimentary archive.
The sedimentary rocks in the Swiss part of the northern foreland basin are divided into four lithostratigraphic units that represent two shallowing-and coarsening-up megacycles (Schlunegger et al., 1998).The first cycle consists of the Rupelian Lower Marine Molasse and the Chattian and Aquitanian Lower Freshwater Molasse.The second megacycle comprises a transgressive facies of Burdigalian age (the Upper Marine Molasse) overlain by Langhian to Serravallian deposits of the Upper Freshwater Molasse.The depositional ages of these units were constrained using mammal biostratigraphy and magnetostratigraphy (Engesser, 1990;Schlunegger et al., 1996).Throughout the Oligocene and the Miocene, the proximal Molasse deposits are thought to have been formed through a series of large alluvial fans (Fig. 1) aligned along the Alpine thrust front (Schlunegger et al., 1993;Kuhlemann and Kempf, 2002).The more distal parts of the basin, on the other hand, were characterized by axial drainage directed towards the Paratethys in the eastnortheast (31-20 Ma) and the western Mediterranean Sea in the southwest (after 20 Ma) (Kuhlemann and Kempf, 2002).Whereas the more distal deposits could be significantly influenced by long-distance transport from the northeast or southwest, the alluvial fans are thought to carry a local provenance signal from the rocks exposed immediately south of each fan system due to their proximal nature.The hinterland of the central Swiss foreland basin comprises, from north to south, potential source rocks derived from the following tectonic units (Figs. 1, 2 -Tok et al., 2014), reaching greenschist facies peak-metamorphic conditions between 17 and 22 Myr ago (Fig. 2a) and were exhumed during the Miocene.The Gotthard nappe, although not a "massif" sensu stricto because of its allochthonous nature, will be included in the term "external crystalline massifs" from here on because the timing and the rates of exhumation are comparable (Fig. 2b, Glotzbach et al., 2010).
4. The Lepontine dome; an allochthonous nappe stack of European Palaeozoic gneiss basement and its Mesozoic metasedimentary cover (Berger et al., 2005).Amphibolite facies peak metamorphism (Frey and Ferreiro Mählmann, 1999;Fig. 2a) in the Lepontine occurred diachronously at around 30-27 Myr ago in the south (Gebauer, 1999) and possibly as late as 19 Myr ago in the north (Janots et al., 2009).Although the onset of exhumation of the Lepontine dome might have been equally diachronous, it is generally assumed to have occurred before 23 Myr ago (Hurford, 1986).
5. The Penninic nappes, containing ophiolite of the Alpine Tethys as well as the continental crust of Briançonnais, a microcontinent located within the Alpine Tethys between the southern Piedmont-Ligurian ocean and the northern Valais trough (Schmid et al., 2004).
6.The Austroalpine nappes, containing the basement and sedimentary cover of the Adriatic plate with a Cretaceous ("Eoalpine", ca.90-110 Ma) metamorphic peak of greenschist facies conditions (Schmid et al., 2004).The Austroalpine nappes were probably part of the nappe stack in the Central Alps prior to their erosion during the Oligocene and Miocene, although they are found exclusively in the Eastern Alps to the east of the Lepontine dome today.

Compositional trends in the Honegg-Napf fan
The Central Alps are generally regarded as the major sediment source of all proximal Molasse basin deposits, and compositional changes in the foreland are thought to directly reflect tectonic and erosional processes in the immediate Alpine hinterland (Matter, 1964;Schlunegger et al., 1993Schlunegger et al., , 1998)).The compositional evolution in the basin is diachronous and non-uniform between the different fan systems (e.g.Schlunegger et al., 1998;Spiegel et al., 2000  Eynatten , 2003).In this study, we will focus on the Honegg-Napf fan, located in the central part of the basin.It most likely preserves a provenance signal related to external massif exhumation due to its proximity to the large crystalline basement slices of the Aar massif and the Gotthard nappe (Fig. 1).In the Honegg-Napf fan, three major compositional trends have been previously identified (Fig. 3).
-Phase 1. Between ∼ 31 and ∼ 25 Myr ago, the heavy minerals are dominated by the zircon-tourmaline-rutile assemblage and garnet (von Eynatten, 2003).Rock fragments are dominantly of sedimentary origin and zircon fission track ages are Palaeozoic to late Mesozoic (Spiegel et al., 2000).This phase is consistently interpreted to reflect the erosion of Austroalpine flysch-like sedimentary cover nappes, which are structurally the top of the central Alpine nappe stack and probably extended further west during this time (Schlunegger et al., 1998;Spiegel et al., 2000;von Eynatten, 2003).
-Phase 2: 25-21 Myr ago.Around 25 Myr ago, the occurrence of epidote as well as an increase in granitoid rock fragments mark a major compositional change in the foreland.The presence of characteristic colourful granite pebbles suggests an origin from the Austroalpine Bernina nappe (Matter, 1964) -Phase 3: 21-14 Myr ago.At ∼ 21 Ma, metamorphic rock fragments occur in the sediments, whereas the heavy mineral assemblages remain epidote-dominated and overall similar to the second phase.Zircon fission track ages are exclusively Cenozoic (age peaks between ∼ 32 and ∼ 19 Ma).In contrast to the first two phases, the sediment composition allows several, partially contradictory interpretations.Whilst petrographic and mineralogic data might suggest recycling and sediment mixing (von Eynatten, 2003), young 40 Ar/ 39 Ar cooling ages in white mica (von Eynatten, 2003;von Eynatten and Wijbrans, 2003) and a population of zircons with a fission track central age of 19.5 ± 0.9 Ma (Spiegel et al., 2000) point to an additional, newly exhumed source identified as the Lepontine dome (Fig. 2b; von Eynatten, 2003;Spiegel et al., 2000).Based on the abundance of flysch pebbles after ∼ 21 Ma, Schlunegger et al. ( 1998) favour an alternative scenario, in which the erosional front shifted northwards into the flysch nappes of the Romandes Prealps.A mixture of both sources seems possible.Furthermore, the isotopic signature of detrital epidotes suggests a contribution of mantle source rocks between ca.21 and 19 Myr ago, which could point to a contribution by Penninic ophiolites (Spiegel et al., 2002).However, this is not reflected in the heavy mineral spectra (von Eynatten, 2003) that do not contain typical ophiolite minerals such as Cr spinel.
The external crystalline massifs have not been regarded as a possible sediment source.The exact time of their surficial exposure is unknown, but it is believed to post-date the youngest preserved Molasse deposits.This interpretation is based on the lack of granitic pebbles attributable to the external massifs in the Molasse (Trümpy, 1980) and on structural reconstructions (e.g.Pfiffner, 1986) in combination with thermochronological data (e.g.Michalski and Soom, 1990).

Sampling strategy and methodology
In order to characterize the detrital garnets in the foreland, three samples were taken from 25, 19, and 14 Myr old fine-to medium-grained fluvial sandstones within the Honegg-Napf fan deposits located ca. 40 km to the east and southeast of Bern in the central part of the Swiss Molasse basin.The exact sampling sites were chosen based on the availability of published petrographical, chemical, and mineralogical data (von Eynatten, 2003) as well as magnetostratigraphic calibration (Schlunegger et al., 1996).
It is possible to compare potential source compositions to the detrital ones because the potential source rocks were already narrowed down to particular regions based on other provenance proxies and because many of these rocks are still preserved in the Alpine chain today.For comparison we used detrital data from Stutenbecker et al. (2017) as well as published source rock data from different units across the Central Alps (Steck and Burri, 1971;Chinner and Dixon, 1973;Ernst and Dal Piaz, 1978;Hunziker and Zingg, 1980;Oberhänsli, 1980;Sartori, 1990;Thélin et al., 1990;Reinecke, 1998;von Raumer et al., 1999;Cartwright and Barnicoat, 2002;Bucher and Bousquet, 2007;Angiboust et al., 2009;Bucher and Grapes, 2009;Weber and Bucher, 2015).
In addition, three river sand samples were collected from small monolithological catchments (3-30 km 2 ) draining potentially garnet-bearing source rocks that were previously not, or only partially, considered in the literature.We prefer this "tributary sampling approach" (first-order sampling scale according to Ingersoll, 1990) over in situ sampling of specific source rocks because small monolithological catchments are more likely to comprise all garnet varieties of the targeted source rock and to average out spatial variations of the source rock properties, e.g.mineral size or fertility (Malusà et al., 2016).The targeted plausible source areas are located in the Gurnigel flysch (Romandes Prealps), the Antigorio nappe orthogneisses of the Lepontine dome, and the Lebendun nappe paragneisses of the Lepontine dome (Fig. 1).Sample characteristics are summarized in Tables 1  and 2. For detailed lithological descriptions of the sampling sites in the Honegg-Napf area, see Schlunegger et al. (1993) andvon Eynatten (2003).
The sandstone samples were carefully disintegrated using a jaw breaker and a pestle and mortar.

Alpine amphibolite facies
Paragneiss, Lebendun nappe, Lepontine dome 2.85 g cm −3 to concentrate heavy minerals.The heavy mineral concentrates were dried and, depending on the obtained amounts, split into two to four parts using a microsplitter.All analysed garnet grains were hand-picked from the concentrate of one split part per fraction under a binocular microscope.
The grains were subsequently arranged in lines on sticky tape, embedded in epoxy resin, ground with SiC abrasive paper (grits 400, 800, 1200, 2500, 4000), polished using 3, 1, and 1/4 µm diamond suspensions, and graphite-coated.Major element oxides were analysed using a JEOL JXA-8200 electron probe micro-analyser at the Institute of Geological Science at University of Bern, Switzerland, under standard operating conditions for garnet (see Giuntoli et al., 2018): accelerating voltage of 15 keV; electron beam current of 15 nA; beam diameter of 1 µm; 20 s peak acquisition time for Si, Ti, Al, Fe, Mn, Mg, and Ca and 10 s for both backgrounds.Natural and synthetic standard olivine (SiO 2 , MgO, FeO), anorthite (Al 2 O 3 , CaO), ilmenite (TiO 2 ), and tephroite (MnO) were used for calibration by applying a CITIZAF correction (Armstrong, 1984).Garnet compositions were measured as close as possible to the geometric centres of the grains, unless the area was heavily fractured or showed inclusions of other minerals.In some randomly selected grains, core and rim compositions were measured to identify intra-grain chemical variability; these core-rim pairs are reported separately in Stutenbecker (2019).
Although some individual garnet grains show distinct internal compositional zoning from core to rim, the intragrain chemical variability is generally negligible (see Stutenbecker, 2019).
The major part of garnet in all three samples (>80 %) belong to the B-type garnet of Mange and Morton (2007) and thus point to a dominant contribution by amphibolite facies source rocks (Table 4).Minor amounts are classified as C-type (high-grade metabasic), A-type (granulite facies), and D-type (metasomatic) garnet.The 25 Myr old sandstone contains almost exclusively B-type garnet (92 %; Table 4).

Late Oligocene ( ∼ 25 Myr ago)
Although detrital garnet chemistry suggests the presence of only one relatively uniform, amphibolite facies source rock in the hinterland of the Honegg-Napf fan during the late Oligocene, the identification of the exact nature of this source is difficult.This is mostly due to the large compositional overlap of garnet sourced by diverse amphibolite facies metamorphic rocks (e.g.metasedimentary versus meta-igneous; Krippner et al., 2014;Tolosana-Delgado et al., 2018).
Zircon fission track ages from sandstones of the same age are mostly >100 Myr old with a smaller and younger age peak of 41 ± 9 Ma (Fig. 3; Spiegel et al., 2000).This would favour an input from the Austroalpine nappes and/or the Romandes Prealps (Fig. 6a), which yield related cooling ages >50 Ma (Fig. 2b; e.g.Bernet et al., 2009), rather than from the Lepontine dome, which is characterized by zircon fission track ages <30 Ma (Fig. 2b; e.g.Hurford, 1986).The presence of granite pebbles attributable to the Austroalpine Bernina nappe (Matter, 1964;Schlunegger et al., 1998) would further support an Austroalpine rather than a Lepontine provenance.
The drainage divide was probably located close to the Insubric line (e.g.Schlunegger et al., 1998) but north of the Bergell pluton (Fig. 6a), whose detritus is exclusively found in the retro-foreland to the south (Gonfolite Lombarda; Giger and Hurford, 1989;Carrapa and Di Giulio, 2001).

Early Miocene (∼ 19 Myr ago)
The larger spread of garnet compositions in the early Miocene (∼ 19 Ma) sample indicates the presence of several or mixed sources with different metamorphic grades, including amphibolite-, eclogite-, and granulite facies rocks.
The B-type garnet compositions match the range of garnets found in the Lepontine nappes (Fig. 4b, d), which is supported by the occurrence of predominantly young (<30 Ma) zircon fission track ages (Fig. 3) in agreement with the young cooling ages of the Lepontine dome (Fig. 2b; Bernet et al., 2009).Due to the overlap of amphibolite facies garnets, it cannot be excluded that at least some of the garnets were contributed by Austroalpine sources or were recycled from older strata.The Lepontine dome was probably drained both towards the north and the south (Fig. 6b) because old basement detritus with young cooling ages (∼ 30 Ma; derived from K-Ar on white mica) was found in the Gonfolite Lombarda group in the southern retro-foreland (Giger and Hurford, 1989).
Granulite facies metamorphic conditions in the Central Alps were only reached in the Gruf complex located close to the Insubric line between the Lepontine dome and the Bergell intrusion (Fig. 2a).Furthermore, there is evidence for pre-Mesozoic granulite facies metamorphism in some rocks in the southern Alpine Ivrea zone south of the Insubric line (Hunziker and Zingg, 1980), in the Sesia Zone (Fig. 1; Engi et al., 2018;Giuntoli et al., 2018), and in the Dent Blanche nappe (Fig. 1; Angiboust et al., 2009).It is unlikely that erosion reached that far to the south during the Miocene because the Penninic and probably also the exhuming Lepontine nappe stack would have acted as a topographic barrier to the fluvial drainage network (Fig. 6b).However, it has been proposed that the flysch deposits preserved in the Romandes Prealps were partially fed by these units during the Late Cretaceous and the Eocene (Wildi, 1985;Ragusa et al., 2017).This interpretation is supported by the Gurnigel flysch sample (Fig. 4e), which contains garnets of the granulite facies type that are similar to those found in the Ivrea zone (Table 3, Fig. 4h).A recycled flysch origin is supported further by the abundance of flysch sandstone pebbles in Molasse strata of the same age (Schlunegger et al., 1998).
A potential, but minor, contribution from ophiolites, as suggested by Spiegel et al. (2002), could be supported by the two eclogite facies garnet grains found in the 19 Myr old sample (Fig. 4b) that match eclogite facies garnets from Alpine ophiolites (Table 3, Fig. 4g).Eclogite facies garnets occur both in metamorphic rocks of the Penninic Alpine ophiolites (e.g.Bucher and Grapes, 2009;Weber and Bucher, 2015;Fig. 2a) and in Palaeozoic (?) gneisses of the middle Penninic Briançonnais basement (Sartori, 1990;Thélin et al., 1990).Both sources are not distinguishable (Fig. 4g) but would have probably been located in relative close geographic proximity, either in the Penninic hanging wall south of the Simplon fault (Zermatt area) or in the Penninic nappes located between the eastern rim of the Lepontine and the adjacent Austroalpine nappes (Arosa zone; Fig. 6b).

Middle Miocene (∼ 14 Myr ago)
Previous provenance studies have identified metasedimentary detritus in the middle Miocene Molasse and located its source in the unroofing sedimentary cover of the Lepontine dome (e.g.von Eynatten, 2003).This was strongly supported by the young detrital zircon fission track ages (youngest peak at 19.5 ± 0.9 Ma, Fig. 3; Spiegel et al., 2000) that match the zircon fission track ages of the Lepontine dome (Fig. 2b; e.g.Hurford, 1986;Bernet et al., 2009).
However, garnet compositions in the youngest Molasse sandstones are not comparable to Lepontine garnets sampled in this study nor to any detrital garnet found in the main rivers draining the Lepontine dome today (Andò et al., 2014).Instead, the detrital garnet signature of the 14 Myr old sample mirrors almost exactly the compositional range of garnets from the external crystalline massifs (Table 3, Fig. 4c, f).In the external crystalline massifs, these garnets grew in Permo-Carboniferous plutons under Alpine greenschist facies metamorphic conditions (Steck and Burri, 1971, Fig. 2a).They are restricted to the granitoid basement of the external massifs and do not occur anywhere else in the Central Alps, which makes them an excellent provenance proxy (Stutenbecker et al., 2017).A further distinction among garnets supplied by the different plutons (e.g. the Central Aar granite from the Aar massif, the Rotondo granite from the Gotthard nappe and the Mont Blanc granite from the Mont Blanc massif) is not possible based on major element garnet geochemistry alone (Stutenbecker et al., 2017).Until now, the surficial exposure of the external massifs in the Central Alps was thought to post-date Molasse deposition.This interpretation relies principally on the absence of pebbles of external massif origin (e.g.Aare granite) in the foreland basin (Trümpy, 1980).However, many Alpine granite bodies closely resemble each other mineralogically and texturally, especially if present as altered pebbles in the Molasse deposits, and hence it is difficult to discount a specific source only on this basis.Further support of late surficial exposure of the external massifs comes from structural reconstructions (e.g.Pfiffner, 1986Pfiffner, , 2017) ) that have located the top of the crystalline basement at an elevation that is similar to the modern topography, based on a relatively flat-lying contact between the crystalline basement and the overlying Mesozoic sedimentary cover (Fig. 7a).According to this model and the published exhumation rates of 0.5-0.7 km Myr −1 (Michalski and Soom, 1990;Glotzbach et al., 2010), the top of the basement was buried 7-10 km below the surface 14 Myr ago.
However, Nibourel et al. (2018) recently proposed a revised geometry of the contact between crystalline basement and overlying cover, which allows ca. 8 km of additional crystalline basement on top of the present-day topography (Fig. 7b).The presence of external massif-sourced garnets in the youngest Molasse deposits provides independent evidence that parts of the crystalline crust contained in the external massifs were already at the surface at ca. 14 Ma (Fig. 6c).Assuming the aforementioned average exhuma-tion rates, 7-10 km of crystalline basement would have already been exhumed and subsequently eroded during the past 14 Myr, which is in good agreement with the geometric reconstructions by Nibourel et al. (2018).
We suggest that the drainage divide was shifted northwards due to the exhumation of the Gotthard nappe and/or the Aar massif and that it was essentially located at its current position (Fig. 6c, d), but this warrants corroboration from other deposits in the foreland and the retro-foreland.

Conclusions
Garnet geochemistry is a useful tool to further constrain the provenance of sandstones in orogens such as the Central Alps.We have demonstrated that it is possible to distinguish detrital garnets using a combination of garnet classification schemes (Mange and Morton, 2007;Tolosana-Delgado et al., 2018) and case-specific comparison with available Alpine source rock compositions (Stutenbecker et al., 2017).For the Miocene deposits of the Swiss Molasse basin, we were able to (1) confirm the provenance shift possibly related to the exhumation of the Lepontine dome between 25 and 19 Myr ago as suggested previously (e.g.von Eynatten, 2003) and (2) to identify an additional provenance shift between ca.19 and 14 Myr ago that had not been noticed before.This shift is related to the erosion of granites from the external crystalline massifs, which provides a minimum age for their surficial exposure and corroborates their recently revised structural geometry.We conclude that the exposure of the crystalline basement happened already ca.14 Myr ago, which is several million years earlier than previously assumed.
Author contributions.LS designed the project.AM helped during field work and sample collection.PMET and PL gave advice for sample preparation and supported the microprobe measurements and data acquisition at the University of Bern.LS prepared the paper with contributions by all co-authors.
Competing interests.The authors declare that they have no conflict of interest.

Figure 1 .
Figure 1.Simplified tectonic map of the Central Alps after Schmid et al. (2004) highlighting the location of alluvial fan deposits within the northern Alpine foreland basin as well as the most important source rock units in the hinterland.The Honegg-Napf fan, marked by the black rectangle, is located in the central part of the Swiss foreland basin (SFB).For cross section X-X see Fig. 7. Abbreviations used: AR -Aiguilles Rouges massif; BD -Belledonne massif; DB -Dent Blanche nappe; HN -Honegg-Napf fan; MB -Mont Blanc massif; GN -Gotthard nappe; PE -Pèlerin fan; PF -Pfänder fan; RH -Rigi-Höhronen fan; SKH -Speer-Kronberg-Hörnli fan; SZ -Sesia zone.

Figure 2 .
Figure 2. (a) Metamorphic map of the Central Alps (Bousquet et al., 2012) showing the distribution and grade of Alpine metamorphism.Note the increase from north to south from lower greenschist to eclogite facies conditions.(b) In situ bedrock zircon fission track ages according to a compilation of Bernet et al. (2009).Note the predominantly young (<30 Ma) cooling ages in the area around the Lepontine dome and the external massifs in contrast to the predominantly old (>50 Ma) cooling ages in the Austroalpine nappes to the east.The river network (blue) and the thick black outlines of selected geological units (external massifs, Romandes Prealps, and Dent Blanche nappe; cf.Fig. 1) are used to facilitate the orientation and the comparison with Fig. 1.Abbreviations used: PE -Pèlerin fan; HN -Honegg-Napf fan; RH -Rigi-Höhronen fan; SKH -Speer-Kronberg-Hörnli fan; PF -Pfänder fan.

Figure 5 .
Figure 5. Relative frequency of the four most common endmembers almandine, grossular, spessartine, and pyrope in the three detrital samples from the Molasse basin.

Figure 7 .
Figure 7. Cross sections from X to X in Fig. 1 through the Aar massif simplified after Pfiffner (2017) and Nibourel et al. (2018).(a) The reconstructed top of the crystalline basement in the Aar massif is located ca. 1-2 km higher than the present-day topography according to Pfiffner (2017).(b) In a revised version by Nibourelet al. (2018)  the contact between the basement and the overlying Helvetic cover nappes is reconstructed to be steeper, resulting in ca. 8 km of (now eroded) crystalline crust on top of the present-day topography.
Spiegel et al. (2002)rgue that parts of the epidote could originate from Penninic ophiolites as well, thus suggesting that erosion might have reached down into the Penninic nappes already by then.Spiegel et al. (2002)argued against this Penninic contribution based on the 87 Sr/ 86 Sr and 143 Nd/ 144 Nd isotopic signatures of the epidote.

Table 2 .
Sample locations and characteristics of potential sources (tributary sampling approach).

Table 4 .
Results from classification following