The NAFB extends for ∼1,000 km from Switzerland to Austria (Figure 1). Its Cenozoic evolution is connected to the approaching Adriatic plate, which is responsible for the loading and flexural bending of the distal southern European continental margin (Sinclair, 1997; Kuhlemann and Kempf, 2002). From 28 Ma onward, continental Molasse-type sedimentation dominated the Swiss and German parts of the basin (Kuhlemann and Kempf, 2002). These sediments were shed northward from the evolving Alps and were redirected eastward in the basin to a delta east of Munich (Füchtbauer, 1964). In contrast, the NAFB east of Munich remained deep marine (Rögl et al., 1979) from ∼28 to 18.3 Ma (Grunert et al., 2013), notwithstanding, an eastward sediment routing also prevailed there (De Ruig and Hubbard, 2006).

In the Upper Austrian part of the basin, depositional processes were controlled from ∼28 Ma onward by a 3–6 km wide, deep-marine, gravity-flow dominated channel system that transported sediment eastward for >100 km (Figures 2, 4). Submarine channel sedimentation lasted for more than 8 Myrs (Hülscher et al., 2019). The Zupfing Formation (ZFM), the Puchkirchen Group (Lower (LPF) and Upper Puchkirchen Formation (UPF)) and the basal Hall Formation (Figure 3) are characterized by structured and structureless sandstones, clast- and matrix-supported conglomerates and silty marls along the basin axis (Bernhardt et al., 2012; De Ruig and Hubbard, 2006; Hubbard et al., 2009). These channelized deposits of sediment-density flows (mostly turbidity currents and debris flows) are bounded towards the north (Figure 4) by wide (≤15 km) overbanks where sedimentation was controlled by hemipelagic settling and northward-directed overspill of sediment by density flows in the channel, resulting in a dominance of silty marls and less abundant turbiditic sandstones (Hubbard et al., 2009; Masalimova et al., 2015). From the tectonically active southern basin margin (Figure 4), submarine fans prograde into the basin (Covault et al., 2009; Hinsch, 2008). Eastward channelized sediment transport ended in the Hall Formation (HFM, Figures 3, 4) during a sea-level highstand at 19.0 Ma and dominantly northward sediment routing was established in the entire NAFB (Hinsch, 2008; Grunert et al., 2013; Garefalakis and Schlunegger, 2019; Hülscher et al., 2019). The Upper Austrian NAFB shallowed from deep-marine conditions to water depth <200 m within the next 0.7 Myrs (Grunert et al., 2013).

The Northern Slope Unconformity (NSU) and the Base Hall Unconformity (BHU, Figure 3) form two major basin-wide unconformities (De Ruig and Hubbard, 2006; Masalimova et al., 2015). Whereas an oversteepening of the northern slope caused the NSU during the initial deepening of the basin between 28.1 and 26.9 Ma (Masalimova et al., 2015; Hülscher et al., 2019), the BHU (∼19.6 Ma) represents a non-depositional period on the overbanks and probably in the channel (Hülscher et al., 2019).

The external perturbation that we interpret to have caused the environmental signal in the studied succession is related to the beginning exhumation of the greenschist-to amphibolite-facies metamorphic rocks (Schmid et al., 2013) nowadays exposed in the Tauern Window metamorphic dome (Figure 1) (Hülscher et al., 2021), hereafter referred to as Tauern Window exhumation. This exhumation was triggered by the Adriatic indentation into the Eastern Alps (Favaro et al., 2015). After the end of the Barrovian-type metamorphism in the subducted and thickened European crust between 40 and 30 Ma, the ongoing convergence between the upper Adriatic and the lower European plate led to large-scale upright folding of the Penninic units of the future Tauern Window (Cliff et al., 1985; Favaro et al., 2015; Scharf et al., 2013; Schneider et al., 2015). This folding caused uplift and cooling (∼28 Ma, Rb-Sr in white mica) in the Penninic units (Reddy et al., 1993; Favaro et al., 2015) and exhumation and denudation in the overlying Upper Austroalpine nappes, which caused a provenance change in the nearby foreland (Hülscher et al., 2021). Apatite single-grain analysis on northern slope and channel samples of the Puchkirchen Group and HFM samples revealed a change in the trace-element geochemistry and U-Pb data around 23.3 Ma (Hülscher et al., 2021). LREE-enriched apatites with >1,000 ppm (LPF: 6%, HFM: 23%) and Sr/Y values < 0.1 (LPF: 24%, HFM: 38%) increased at 23.3 Ma and reached a maximum in the HFM (Figure 5). Contemporaneously, the number of apatites yielding late Variscan U-Pb ages increased. Based on the discrimination diagram of O’Sullivan et al. (2020), Hülscher et al. (2021) interpreted the change in apatite trace-element geochemistry to mirror a change in source rocks from a dominance of low-grade metamorphic sources before 23.3 Ma to high-grade metamorphic sources thereafter. Combined with the increasing number of late Variscan U-Pb ages, this indicates an increasing input from a high-grade, late Variscan metamorphic source (Figure 5) from 23.3 Ma onward driven by the early exhumation of the Tauern Window (Hülscher et al., 2021). In this period of early (28 ± 1 Ma) and slow (0.3–0.6 mm/a) exhumation (Hülscher et al., 2021), orogen-perpendicular shortening caused the surface response (Favaro et al., 2017). A reorganization to orogen-parallel extension via low-angle normal faults at the eastern and western end of the Tauern Window around 23–21 Ma (Figure 1) (Favaro et al., 2015; Favaro et al., 2017) resulted in a doubling of exhumation rates (1.5–2 mm/a) in the future Tauern Window area (Frisch et al., 2000). The Penninic units in the Tauern Window reached the surface around 13 Ma when Penninic pebbles appeared in the Upper Austrian NAFB (Brügel et al., 2003).

Three SRSs have been suggested to contribute sediment to the Upper Austrian NAFB. 1) The majority of the Eastern Alps drained northeastward from Oligocene onward via the paleo-Inn (Figure 1) (Brügel et al., 2003; Hülscher et al., 2021; Sharman et al., 2018). The river flowed through the Inn valley and built up alluvial and marine deposits of the Inntal-Tertiary in the Inn valley and the Chiemgau, Wachtberg and Munderfinder-Kobernhauser-Hausruck Fan deposits (Figure 1) in the Upper Austrian NAFB (Brügel et al., 2003; Ortner and Stingl, 2001). A direct connection from the Chiemgau Fan to the Puchkirchen channel system existed (De Ruig and Hubbard, 2006) (Figures 1, 2). The southern drainage divide was located along to the Periadriatic Line (Brügel et al., 2000; Brügel et al., 2003) (Figure 1). The western boundary was located in the western Lower Austroalpine nappes in Miocene times (Figure 1) (Skeries and Troll, 1991). The detritus from this SRS was deposited on the southern and northern slope of the Puchkirchen channel system (Hülscher et al., 2021; Sharman et al., 2018). 2) Eastern Alpine material also entered the southern slope of the Upper Austrian NAFB via a southern SRS through the Augenstein Formation (Figure 1) (Frisch et al., 2001; Hülscher et al., 2021; Sharman et al., 2018). The Augenstein Formation accumulated on top of today’s Northern Calcareous Alps, east of the Inn valley from 35−30 Ma onward, mainly sourced from the Austroalpine basement and its Mesozoic cover in the south (Frisch et al., 2001). Around 21 Ma, the initiation of exhumation of the Northern Calcareous Alps caused the erosion and redeposition of the Augenstein Formation into the Upper Austrian NAFB (Frisch et al., 2001; Hülscher et al., 2021). 3) A western SRS supplied material via a large delta system located close to Munich (hereafter Munich SRS) to the Puchkirchen channel system (Figure 1) in Oligocene/early Miocene times (Zweigel, 1998; Sharman et al., 2018). Central Alpine material was transported to the delta (Füchtbauer, 1964) in a meandering river system (Platt and Keller, 1992). The importance of this SRS for sand-sized sediment supply to the Puchkirchen channel system is debated (Sharman et al., 2018; Hülscher et al., 2021). With the reorganization of the SRS during the sea-level highstand around 19 Ma, the Munich SRS was cut off from the Upper Austrian NAFB (Füchtbauer, 1967).

The clay-size fraction from thirty well-cutting samples were analyzed for their Nd isotopic composition. This fraction was separated by mixing the samples with distilled water and centrifuging them for 1 min at 4,000 rpm. The suspended fraction was decanted and dried. The separated clay-sized fractions were treated with 10% acetic acid and washed with Milli-Q water. Because all samples contained significant amounts of organic material, 30 samples were heated up to 830°C in a muffle furnace for combustion of organic materials. Additionally, six samples were selected for replicate analysis (Figure 6). In these cases, the organic material was removed by chemical treatment during digestion using HNO₃ and H₂O₂.

100 mg of sample material was digested in a 3:2 mixture of concentrated HF-HNO₃ at 120°C for 24 h. After dry down, samples were treated three times with 2 ml HNO₃ and dried. This was followed by 12 h calibration in 6 M HCl. After dry down, each sample was taken up in calibrated 2.5 M HCl and centrifuged. The separation of Nd followed the protocol of Richard et al. (1976). The REE were separated using cation exchange resin (AG 50WX8 resin) with HCl. After dry down at 120°C, the REE cuts were taken up in 0.15 M HCl and Nd was separated from other REE on columns using coated Teflon powder cation exchange resin (PTFE in HDEHP). After extraction, Nd cuts were dried at 120°C and dissolved in 1 ml 0.25 M HNO₃. Procedural blanks were <20 pg for the samples where organic matter was combusted. For the unheated samples, the procedural blank was <130 pg. All blanks were negligible, and no correction was necessary.

Samples were analyzed for their ¹⁴³Nd/¹⁴⁴Nd ratios with a Thermo Scientific Neptune Plus multi-collector inductively coupled plasma mass spectrometer at the Institut für Mineralogie (Westfälische Wilhelms-Universität Münster, Germany). JNdi-1 solution was used as reference material during analysis, revealing ¹⁴³Nd/¹⁴⁴Nd = 0.512096 (± 8, 2σ standard deviation, n = 17). The data is reported relative to JNdi-1 (¹⁴³Nd/¹⁴⁴Nd = 0.512099 (Garçon et al., 2018)). Results were normalized to ¹⁴⁶Nd/¹⁴⁴Nd = 0.7219 using the exponential law and reported in εNd notation (DePaolo and Wasserburg, 1976) with 2σ uncertainties.

Isotope analyses were carried out at the Geochronology and Tracers Facility, British Geological Survey, Keyworth, United Kingdom, using a Thermo Scientific Neptune Plus MC-ICP-MS coupled to a New Wave Research UP193UC Excimer laser ablation system. Helium was used as the carrier gas through the ablation cell with Ar make-up gas being connected via a T-piece and sourced from a Cetac Aridus II desolvating nebulizer. The instrument was tuned for low oxide production, and 0.008 l/min of nitrogen were introduced via the nebulizer in addition to Ar to minimize oxide formation.

For neodymium isotope analysis, ¹⁴²Ce+¹⁴²Nd, ¹⁴³Nd, ¹⁴⁴Nd+¹⁴⁴Sm, ¹⁴⁵Nd, ¹⁴⁶Nd, ¹⁴⁷Sm, ¹⁴⁹Sm, ¹⁵⁰Nd and ¹⁵¹Eu were measured simultaneously during static 30 s ablation analyses. Because of the relatively small grain size of the apatites, a 35 µm laser spot was used for most samples. Where grain size allowed, a 50 µm spot was used. The laser fluence was 8–10 J/cm². Correction for ¹⁴⁴Sm on the ¹⁴⁴Nd peak was carried out using the method of Yang et al. (2014). Madagascar apatite was used as the primary reference material for laser-ablation analysis. The Durango apatite reference material was also analyzed thought the analytical session, along with the standard glasses JNd-i, JNd-i LREE, and NIST 610. Data reduction was carried out using the Iolite data reduction package (Paton et al., 2011). Analytical uncertainties for unknowns were propagated by quadratic addition to include the standard error of the mean of the analysis and the reproducibility of the Durango reference material. εNd values were calculated using a¹⁴⁷Sm decay constant of 6.54 × 10⁻¹² y⁻¹ (DePaolo and Wasserburg, 1976), the present-day chondritic ¹⁴⁷Sm/¹⁴⁴Nd value of 0.1967, and ¹⁴³Nd/¹⁴⁴Nd ratio of 0.512638 (Jacobsen and Wasserburg, 1980).

Results from 30 neodymium isotopic analyses of the clay-sized fraction samples are reported. Six replicates yield similar results within their 2σ error when compared to the original sample (Figure 6). The εNd values show little variation in the ZFM (three samples), the LPF (six samples) and the UPF (16 samples) around a mean of −9.73 (±0.52; Figure 6). In the lower part of the HFM, two samples (1,495 m, 1,475 m) reveal similar values when compared to samples from the UPF. Five samples above 1,420 m in the HFM show higher εNd values (mean: −9.12 ± 0.52; Figure 6) when compared to the earlier 25 samples.

We performed a Monte-Carlo experiment to test the robustness of this shift in εNd values. To provide a robust estimate of the shift in Nd isotope composition in mean and variance between the unbalanced subsamples for >19 Ma [n ₁ = 25 (four replicate measurements)] and for <19 Ma [n ₂ = 5 (+2)] εNd measurements, we performed n_MC = 10,000 random sampling from each measurement assuming a normally distributed measurement error (Figure 6; µ and 2σ-interval). Replicate measurements have been randomly selected once for all n_MC samples. Statistical analyses were performed in R Vers. 4.0.2 using the stats package (R Core Team, 2020). The distribution of the n_MC random sample means appears normally distributed for both samples as expected by the Central Limit Theorem, with a mean µ₁ = −9.674 (σ₁ = 0.018, >19 Ma) and µ₂ = −9.121 (σ₂ = 0.038, <19 Ma). Therefore, we applied a Student’s t-test, suggesting an increase of the mean εNd value of the two subsample sets of 0.5520–0.5536 with 95%-confidence and p-value < 10⁻¹⁵ under the assumption of heteroscedasticity. Therefore, we regard the two subsample sets (>19 Ma, < 19 Ma) as being significantly different.

The detrital apatite trace-element geochemistry and U-Pb data was taken from Hülscher et al. (2021). Based on the discrimination diagram of O’Sullivan et al. (2020), Hülscher et al. (2021) grouped the detrital apatites into four source rocks: mafic igneous rocks (IM); low- and medium-grade metamorphic (<upper amphibolite-facies) (LM); high-grade metamorphic rocks (HM); and ultramafic rocks (UM). The apatites from S-type granitoids and HM sources, as differentiated by O’Sullivan et al. (2020), were grouped with the HM apatites for simplicity.

The ¹⁴⁷Sm/¹⁴⁴Nd and ¹⁴³Nd/¹⁴⁴Nd ratios of 406 sand-sized detrital apatite grains further constrain the differences between the proposed apatite groups based on their trace-element geochemistry. εNd values range between −14.63 and 10.34 (median: −4.09) and are highest in the HM apatite group (median: −1.95) (Figure 5). LM, IM, and UM apatites have similar εNd single-grain distributions (Table 1; Figure 5). The ¹⁴⁷Sm/¹⁴⁴Nd ratios are similar between all different apatite groups (Table 1). A Kolgomorov–Smirnov (K-S) Test (Hodges, 1958) was performed to test whether the εNd single-grain distributions of the different apatites groups resulted from random sampling of the same parent population or was significantly different. p-values of the K-S Test of the HM apatites are <0.01% when compared to the remaining apatite groups, indicating a significantly different εNd distribution of the HM group, in contrast to the remaining apatite groups which show similar εNd distributions (Table 1).

Our dataset is biased towards apatites with high Sm and Nd contents and this bias especially influences the results from the LM apatites. Apatites from low-grade metamorphic sources are usually depleted in REE (Henrichs et al., 2018; O’Sullivan et al., 2020), making their Sm-Nd isotopic analysis difficult. LM apatites are underrepresented in our Nd isotope dataset (31%) compared to the trace-element dataset (47%) of Hülscher et al. (2021).

The clay-sized bulk-rock εNd values remained stable around −9.67 (±0.52) from ∼27 Ma to 19 Ma. In the five samples younger than 19 Ma, a significant increase of 0.55 εNd units to −9.12 (±0.54) (Figure 6) was found. No correlation between the data and the proximity of the channel or the lithology are recognizable (Figure 6). Gier (1998) reported a diagenetic clay mineral transformation from smectite to illite which linearly increases with depth. The described step-change at 19 Ma in the clay-sized bulk-rock εNd values cannot be explained by this linearly increasing clay transformation, as such a trend is not noticeable in the results. In parts of the basin (e.g., in the basal Hall Formation and in a gas field 10 km east of Well H in the lower part of the UPF, Figure 2) methanogenesis has been described as an important diagenetic process (Schulz et al., 2009; Grundtner et al., 2016). The εNd values do not show significant fluctuations in the basal Hall Formation or the lower part of the UPF of Well H (Figure 6). Hence, we interpret our clay-sized εNd values to be unaffected by the diagenetic processes in the basin.

Regional climatic reconstructions suggested a stable warm and humid climate during the Oligocene/Early Miocene (Mosbrugger et al., 2005; Grunert et al., 2014; Filek et al., 2021) with mean annual precipitation rates consistently >900 mm in the foreland (Mosbrugger et al., 2005). As climatic records are still sparse from the NAFB, we cannot exclude the possibility that climatic fluctuations on timescales below 1 Myrs have occurred. However, based on the available data, the bulk-rock clay-sized εNd results do not appear to have been significantly affected by climatic changes and represent a reliable provenance proxy, governed by tectonic processes.

Potential source rocks for the clay-sized fraction in the ∼27 to 19 Ma samples are located in the Eastern Alps, as suggested by provenance reconstructions of sand-sized sediment (Hülscher et al., 2021; Sharman et al., 2018). However, Upper Chattian and Aquitanian deposits from the Swiss NAFB show similar bulk rock εNd values of ∼−9.7 (Henry et al., 1997). A connection of the Swiss NAFB to the Munich delta via a meandering river system (Platt and Keller, 1992), which was rich (70%–80%) in silt and clay, has been suggested based on heavy mineral assemblages (Füchtbauer, 1964; Füchtbauer, 1967). During the UPF and basal HFM, the delta was connected to the Puchkirchen channel system via the Halfinger Canyon (Figure 2) (Zweigel, 1998). This Munich SRS had a larger source area than the Eastern Alpine SRSs of the paleo-Inn and the Augenstein Formation (Figure 7) (Winterberg, 2019). We regard the Munich SRS to be a major contributor to the Upper Austrian NAFB clay-sized sediments >19 Ma due to the increasing εNd values after the tectonically induced (Zweigel, 1998) cut-off of the Munich SRS (Füchtbauer, 1967) at 19 Ma (Figure 6). However, the clay-sized sediment provenance before 19 Ma is likely to represent a mixture of Central and Eastern Alpine sources, as an Eastern Alpine source is reported for the sand-sized fraction (Sharman et al., 2018; Hülscher et al., 2021).

In the five <19 Ma clay-sized samples, εNd values increased to −9.12 ± 0.52. An explanation for this changing provenance might be related again to processes in the Swiss NAFB. Henry et al. (1997) reported bulk-rock εNd values of ∼−8.2 from the Swiss NAFB from the beginning of the Burdigalian (∼20.4 Ma); these might have caused the increase in clay-sized bulk-rock εNd values in the Upper Austrian NAFB. Based on detrital zircon U-Pb ages, Sharman et al. (2018) suggested that the Hall Formation may have received sediment from the Bohemian Massif, where granitoids with less negative εNd values (∼−1 to −9) are reported (Finger et al., 1997; Liew and Hofmann, 1988; Teipel et al., 2004). However, both sources seem unlikely to be the driver for the clay-sized provenance change as the major sediment-transport direction became north-directed in the Upper Austrian NAFB (Figure 4, 6) from 19 Ma onward (Hinsch, 2008; Grunert et al., 2013; Hülscher et al., 2019).

This tectonically-driven (Zweigel, 1998) switch in sediment-transport direction from eastward transport before 19 Ma (Platt and Keller, 1992; Füchtbauer, 1964; De Ruig and Hubbard, 2006) to northward transport thereafter in the entire NAFB (Hinsch, 2008; Garefalakis and Schlunegger, 2019; Hülscher et al., 2019) went along with evolution of the Upper Austrian NAFB from a marine transfer zone to a marine terminal sink (Hülscher et al., 2019). A four-fold increase in sediment-accumulation rates (Figure 3), submarine channel sedimentation cessation and prodelta-wedge northward-progradation reflect this evolution (De Ruig and Hubbard, 2006; Hinsch, 2008; Grunert et al., 2013; Hülscher et al., 2019). As the basin evolved from an Alpine-wide transfer zone to an Eastern-Alpine marine terminal sink (Hülscher et al., 2019), the clay-sized sediment provenance switched from a mixed Alpine-wide one to a local Eastern Alpine one, resulting in significantly less negative εNd values in the samples <19 Ma (Figure 6).

Potential source rocks for the clay-sized bulk-rock εNd values after <19 Ma are the Upper Austroalpine nappes in the Eastern Alps (Figure 1). As bulk-rock εNd values between −5 and −15 are quite common in all Eastern Alpine nappe systems (Supplementary Figure S1) (Thöni and Jagoutz, 1993; Bernhard et al., 1996; Thöni and Miller, 1996; Hoinkes et al., 1997; Thöni and Miller, 2000; Schuster et al., 2001; Thöni, 2003; Tumiati et al., 2003; Schulz et al., 2004; Thöni and Miller, 2004; Habler et al., 2006; Thöni, 2006; Habler et al., 2007; Habler et al., 2009; Thöni and Miller, 2009; Thöni and Miller, 2010; Heinrichs et al., 2012), a source-rock interpretation just from the clay-sized bulk-rock εNd values is difficult, particularly since the values most likely represent a mixture of different Eastern Alpine sources. Provenance information from sand-sized apatites with comparable depositional ages suggests a high-grade metamorphic, late Variscan source with increased εNd values, potentially a part of the Ötztal-Bundschuh Nappe System (see discussion below). As apatites and clay-sized bulk-rock εNd values both evolve towards less negative εNd values, we interpret both to be governed by the erosion of parts of the Ötztal-Bundschuh Nappe System with increased εNd values, driven by the Tauern Window exhumation. However, from the clay-sized bulk-rock εNd values alone we cannot exclude other possible sources in the Eastern Alpine nappe stack.

Contrary to our interpretation of a stable clay-sized sediment provenance before 19 Ma, Hülscher et al. (2021) reported a change in sand-sized single-grain apatite provenance based on their trace-element geochemistry and U-Pb thermochronology dataset at 23.3 ± 0.3 Ma (Figure 5). These authors pointed out that the apatites were derived from the Eastern Alps because of the stable provenance after the cut-off of the Munich SRS at 19 Ma (Figure 5). This changing provenance at 23.3 Ma is assumed to be governed by the Tauern Window exhumation from 28 (±1) Ma onward (Hülscher et al., 2021). Contrary to that interpretation, Sharman et al. (2018) suggested an increasing sediment delivery from the Munich SRS (Figure 1) to explain the decreasing shares of zircon single-grain ages >525 Ma and increasing proportion of grain ages between 250 and 375 Ma from the LPF to UPF found by the authors.

Our new Sm-Nd isotopic dataset supports the idea that the change in apatite provenance at 23.3 (±0.3) Ma is related to processes in the Eastern Alps. The apatites are not only different in their trace-element geochemistry as shown by Hülscher et al. (2021) but also in their Nd isotope geochemistry (Figure 5) with the HM apatites having significantly less negative εNd values than the IM, LM, and UM grains (Table 1). Higher εNd values can be explained by two reasons: either the apatites differ significantly in their formation ages, or the ¹⁴³Nd/¹⁴⁴Nd ratio of the source rock is different (DePaolo and Wasserburg, 1976). Even though the combined U-Pb and Sm-Nd dataset does not contain both types of information for all grains, Figure 5 shows that the U-Pb age distribution of all apatite subgroups are dominated by ages between 250 and 350 Ma. In combination with the slow ¹⁴⁷Sm decay (6.54 × 10⁻¹² y⁻¹) (DePaolo and Wasserburg, 1976), different formation ages seem to play a subordinate role in explaining the different εNd values between the apatite subgroups. Therefore, we interpret the difference to be inherited from the source rocks. The source rocks of the HM apatites must have had significantly higher εNd values compared to the IM, LM, and UM apatites source rocks (Table 1; Figure 5). This source became a significant contributor (∼40% HM apatites) for the sand-sized sediment in the Upper Austrian NAFB after 23.3 Ma (Figure 5) (Hülscher et al., 2021). The Ötztal-Bundschuh nappe system contains gneisses which experienced an upper amphibolite to eclogite facies metamorphic event in Variscan times (Hoinkes et al., 1997; Rode et al., 2012; Schulz et al., 2019) and have enriched ¹⁴³Nd/¹⁴⁴Nd ratios (Supplementary Figure S2) (Hoinkes et al., 1997; Thöni and Miller, 2004) compared to other Eastern Alpine nappe systems with a Variscan metamorphic imprint (Thöni and Miller, 2000; Schuster et al., 2001; Habler et al., 2006). The more negative values from the LM and IM apatites are in line with published data from the metasediments and metamagmatites from the Koralpe-Wölz and Drauzug-Gurktal Nappe systems (Thöni and Miller, 2000; Schuster et al., 2001; Habler et al., 2006; Heinrichs et al., 2012).

However, the apatite Sm-Nd isotopic dataset also highlights limitations of the apatite trace-element discrimination diagram used here. Eight out of the ten UM apatites have εNd below −5 (Figure 5), which is an unusual low value for Eastern Alpine ultramafic rocks (Thöni and Jagoutz, 1993; Schulz et al., 2004; Habler et al., 2006). The most likely explanation for this mismatch between the geochemical classification and the isotopic composition is a misclassification of the grains by the discrimination diagram. However, as these grains only make up a small share in the apatites population (10 of 406), this misclassification does not interfere with the interpretation.

In summary, the two analyzed grain-size fractions reveal significant differences when provenance changes are recorded, implying that the fractions are recording different information (Figure 5, 6). The sand-sized apatite single-grain distributions record the exposure of the new Upper Austroalpine source with higher εNd values as a changing provenance at 23.3 Ma. However, the clay-sized bulk-rock εNd values reveal stable conditions at that time. Vice versa, the clay-sized bulk-rock analysis indicates a provenance change at 19 Ma, when the sand-sized apatites suggest stable conditions. Between 23.3 and 19 Ma the sand-sized apatites and the clay-sized bulk-rock record significantly different provenance information. As both εNd provenance records evolve into less negative values, both records may be governed by the same process: the exposure of the new Upper Austroalpine source with higher εNd values driven by exhumation of the Tauern Window.

A difference in the timing of signal arrival between grain-size fractions has been described above. However, contrary to the hypothesis posed by Tofelde et al. (2021), where an environmental signal may be first recorded in the fine-grained sediment fraction due to rapid sediment transport as suspended load, the provenance change is first recorded in the sand-sized apatites; the clay-sized fraction potentially does not show the same provenance shift until 4–5.3 Myrs later. The total signal-lag time [sensu Tofelde et al. (2021)], the time span from the onset of the early Tauern Window exhumation to the recording of the provenance change in the sand-sized apatite single-grain distributions, amounts to 3.4–6.0 Myrs (Figure 3). The total signal-lag time of the clay-sized bulk-rock εNd values in response to the Tauern Window exhumation is 8.0–10.3 Myrs (Figure 3). Hence, the exposure of the new Upper Austroalpine source with higher εNd values was recorded 4–5.3 Myrs later in the clay-sized fraction than in the sand-sized apatites (Figure 3), only after a major reconfiguration of the drainage system reduced the drainage area that provided sediment to the study area by ∼60% (Figure 7).

The described difference in total signal-lag times (sensu Tofelde et al. (2021)) of provenance changes between different grain-size fractions creates challenges and opportunities for our understanding of SRS evolution. Often the sand-sized and/or coarse silt-sized fraction is analyzed for provenance reconstructions of a SRS (e.g., Huber et al., 2018; Morton, 1985; Van Andel, 1950). As highlighted here, their provenance information might significantly differ from other grain-size fractions at least temporarily (Figure 3), posing uncertainties on the reliability or completeness of such reconstructions and related sediment-budget calculations (Kuhlemann, 2000).

We regard methodological characteristics in combination with the reconfiguration of the drainage system at 19 Ma to be responsible for the delayed recording of provenance changes. As single-grain analyses provide specific source-rock information from the investigated grains, this single-grain information in the sink can be (in a best-case scenario) directly linked to a specific source (Mange-Rajetzky and Oberhänsli, 1982; Morton, 1985; von Eynatten and Dunkl, 2012). Furthermore, single-grain distributions are often biased towards sources with high mineral fertility and erosion rates which leads to a dominance of these sources in the sink samples, even though those sources are geographically small (Gemignani et al., 2017; Malusà et al., 2017). For a reliable source-to-sink reconstruction, a sufficiently large single-grain population is required (Vermeesch, 2004). In contrast, bulk-rock analytical methods represent a mixture of the entire SRS by integrating over the catchment upstream of the sampled location; information about specific sources can only be gained if results are unmixed (Dickinson, 1985; Roser and Korsch, 1986; Ingersoll, 1990). The resulting provenance information is more comprehensive but less specific than single-grain information. The mixture leads to a dilution of extreme values in e.g., geochemistry, erosion rates, or isotopic ratios from small source areas in the course of the SRS and this extreme endmember information does not reach the sink (Wittmann et al., 2016; Awasthi et al., 2018; Malkowski et al., 2019). Only large-scale changes (e.g., large-scale drainage reorganizations or climatic change) in an orogen-wide SRSs influence the bulk-rock composition of the sedimentary archive in the sink (Clift and Giosan, 2014; Fildani et al., 2018).

In this study, exhumation of the geographically small (∼3,125 km², Favaro et al. (2017)) but proximal sediment source above the future Tauern Window, with two to six times higher exhumation rates compared to the remaining Eastern Alps (Frisch et al., 2000; Hülscher et al., 2021; Kuhlemann, 2007), changed the apatite single-grain distribution in the sedimentary sink at 23.3 Ma (Figure 5). Coarser grain-size fractions tend to be biased towards more proximal sources (Gemignani et al., 2019; Hülscher et al., 2018). However, the Upper Austroalpine source area above the future Tauern Window, which had higher εNd values, was too small to change the clay-sized bulk-rock εNd values in the sink at 23.3 Ma (Figure 6). The signal was diluted by the overwhelming contributions from the remaining sources, especially by the Munich SRS, which was rich in silt and clay (Füchtbauer, 1964; Füchtbauer, 1967). The relative proportion of the area above the future Tauern Window represents ∼3% of the entire Chattian/Aquitanian Upper Austrian NAFB source area (Figure 7). The tectonically induced (Zweigel, 1998) cut-off of the Munich SRS (Füchtbauer, 1967) at 19 Ma reduced the Upper Austrian Molasse catchment area (Figure 7); thereby increasing the relative share of the source area above the future Tauern Window to ∼8%. Only this reduction had a large-enough effect to allow the provenance change to be recognizable in the bulk-rock clay-sized εNd values (Figure 6). However, as revealed by the apatite single-grain distribution, the provenance change was recorded the Upper Austrian NAFB 4–5.3 Myrs earlier (Figure 3).

The differences between signals obtained using single-grain and bulk-rock methods can emphasize information about the extent of a perturbated area in an SRS. As described above, it requires spatially extensive environmental perturbations in a SRS to change the bulk-rock provenance of the sink, but only small-scale environmental perturbations to change the single-grain distribution of the sedimentary sink. In this study, the small-scale exhumation of a late Variscan high-grade metamorphic source rock with higher εNd values above the future Tauern Window was recognizable in the Upper Austrian apatite distribution, but only the large-scale reorganization of the SRSs made it recognizable in the bulk-rock clay-sized εNd values. However, processes that affect larger parts of an SRS have been reported to change bulk-rock and single-grain methods simultaneously (Clift and Giosan, 2014; Bracciali et al., 2015; Fildani et al., 2018). These processes include climatic perturbations such as a changing erosional focus (Clift and Giosan, 2014) or drainage reorganization (Fildani et al., 2018) but also tectonic perturbations triggering large-scale drainage reorganizations (Bracciali et al., 2015). However, tectonic exhumation processes interfere only with segments of an orogen-wide SRS and related provenance changes are driven by the exposure of a new source rock (Brügel et al., 2003; Anfinson et al., 2020) or a drainage-divide migration caused by surface uplift (Mark et al., 2016; Lu et al., 2019). Taking these ideas further, by combining single-grain and bulk-rock approaches, provenance changes driven by tectonic exhumation processes from those driven by large-scale processes may be distinguished just from their imprint in the sedimentary sink. This enables a previously unattained understanding of the underlying environmental change, especially for ancient SRSs where independent constrains are missing.