High-K andesites as witnesses of a continental arc system in the Western Alps, Italy: constraints from HFSE and Hf–Nd–Sr–Pb–O isotope systematics

Geochemical and isotopic data are presented for ~ 32 Ma-old high-K andesites and dacites from the Alpine Chain. The samples consist of plagioclase, amphibole, titanomagnetite and rare biotite and quartz. Geochemical and isotope data indicate that slab-derived fluids, sediment melts and presumably AFC processes involving continental crust played a key role in the petrogenesis of the high-K rocks. A contribution of fluids is suggested based on the overall enrichment of large-ion lithophile elements and related high Ba/La, Ba/Zr, Ba/Th, Ba/Nb and Pb/Nd, sometimes distinctively higher than average continental crust. Positively correlated Ba/Nb–Th/Nb relationships, low Ce/Pb, low Nb/U and a negative correlation of Pb isotopes with Ce/Pb and Nb/U and positive ∆ 7/4 and ∆ 8/4 values similar to GLOSS imply the additional involvement of a sediment-derived melt. Negatively correlated Nb/Ta–Zr/Hf ratios at overall low Nb/Ta (13–7.5) are best explained by parental magma differentiation involving amphibole and biotite in a continental arc system. The samples have moderately unradiogenic Nd (εNd: – 2.0 to – 6.7) and radiogenic 87Sr/86Sr isotope compositions (0.7085–0.7113), moderately radiogenic Pb isotope compositions (206Pb/204Pb: 18.50–18.72; 207Pb/204Pb: 15.59–15.65; 208Pb/204Pb: 38.30–38.67), and elevated δ18O values (+ 6.5 to + 9.1 ‰). Epsilon Hf isotope values range from + 2.5 to – 4.0. Negative εHf(t) and εNd(t) values and 206Pb/204Pb ratios are correlated with elevated K2O abundances that indicate enrichment in K2O is related to AFC processes. The offset of εHf at a given εNd points to involvement of aged garnet-bearing crustal lithologies. The latter feature is qualitatively consistent with modification of unexposed primary basaltic andesites by AFC processes involving deep crustal material. In conclusion, in an Alpine context, inferred unexposed primitive high-K basaltic to andesitic melts are generated in the mantle wedge through fluid infiltration from the descending slab where fluids may have caused also partial melting of sedimentary rocks that mixed with evolving andesite–dacite compositions towards shallow-level intrusive and extrusive rocks. High-K and related trace element and isotope features thus result from a combination of already elevated values with participation of fluids and melts and probably AFC processes.


Introduction
Igneous rocks associated with subduction-related tectonic settings commonly belong to distinct magmatic series ranging from high-Al basalt to rhyolite, but often differ considerably in their K 2 O content. It is commonly accepted that multiple sources are involved in andesite generation and subsequent interaction between melts derived from these sources are required to explain the geochemical and isotopic diversity of subduction-related volcanic rocks (Arculus and Powell 1986;Ellam and Hawkesworth 1988;Stolz et al. 1990;Hawkesworth et al. 1993). Probably, the volumetrically most important component is the mantle wedge that can have geochemical and isotopic compositions ranging from normal mid-ocean ridge basalts (N-MORB) source mantle (Perfit et al. 1980) or even depleted MORB (D-MORB) source mantle (McCulloch and Gamble 1991;Woodhead et al. 1993;Elliot et al. 1997) to highly enriched ocean island basalt-type (OIB) source mantle (Morris and Hart 1983;Stolz et al. 1990;Edwards et al. 1994;Turner et al. 2017), however, the composition of the down-going slab may also be important. A second critical aspect might be the temperature of the mantle wedge and the down-going slab, its sediment load, subduction angle and velocity. These parameters affect the type and interplay of distinct fluids and melts derived from them (Turner et al. 2016). The elemental and isotopic composition of many subduction-related volcanic complexes requires further additional involvement of other components, including fluids of subducted oceanic crust and associated sedimentary rocks or partial melts thereof (Kay 1980;White and Duprè 1986;Ellam and Hawkesworth 1988;McDermott et al. 1993;Gamble et al. 1996;Price et al. 1999;Vroon et al. 1993Vroon et al. , 2001. In this context, varying contributions of the subduction-related components to the source region of arc magmas, the transfer mechanisms of specific elements from the slab to the sub-arc mantle and the composition and amount of the subducted sediment are important parameters (Ellam and Hawkesworth 1988;Elliot et al. 1997;Plank and Langmuir 1998). Additional modification of arc magmas that usually lead to more enriched isotope compositions may come from interaction with pre-existing continental crust during movement through and stagnation within the crust (Hildreth and Moorbath 1988;Davidson 1987;Davidson and Harmon 1989;Davidson et al. 1990;Ellam and Harmon 1990;Thirlwall et al. 1996). Therefore, based on the numerous important factors that may affect the chemistry of subduction-related volcanic rocks, a universally valid model for the generation of these rocks is precluded.
A series of arc lavas of controversial origin are high-K andesites or shoshonites. Andesites are in general rare in arc settings compared to their basaltic and rhyolitic counterparts (Reubi and Blundy 2009). The enrichment in K (and sometimes Na) is not fully understood as in either scenario of andesite genesis (magma mixing, fractional crystallization with and without assimilation), both, Na and K increase in tandem. However, some andesitic volcanoes develop distinctively high-K contents (Price et al. 1999;Zernack et al. 2012), which require one or more additional processes in the mantle or the crust en route to the surface.
In this contribution, we present mineral data, major and trace element whole-rock compositions and Sr, Nd, Pb, Hf and O isotope data as well as biotite and amphibole Ar-Ar ages for a series of Oligocene high-K subduction-related volcanic rocks from the Southern Alps (Italy). Due to the sparseness of the volcanic centers in the Alps (McCarthy et al. 2021) and pronounced post-orogenic uplift and concomitant erosion, primary volcanic centers are no longer preserved in the Alpine chain, but remnants of subduction-related volcanic rocks occur as large decimeter-sized boulders in molasse-type sedimentary rocks of the Oligocene-Miocene Gonfolite Lombarda Formation (Como, Italy;Gunzenhauser 1985). These boulders may provide the necessary information on the geochemistry and isotope geochemistry of no longer exposed volcanic rocks. Like few subduction-related suites from worldwide occurrences (i.e., lesser Antilles; Nielsen and Marschall 2017), all samples are characterized by radiogenic Sr and Pb isotope compositions, unradiogenic Nd and Hf isotope compositions and moderate to high δ 18 O values. The samples range in composition from large-ion lithophile element (LILE)-and light rare-earth element (LREE)-enriched high-K andesites to dacites with a pronounced depletion in HFSE. Based on geochemical and isotope criteria, we will use these samples to constrain the effects of crustal contamination as well as fluid and melt interaction with mantle wedge peridotite on the generation of geochemical and isotopic variability in andesitic lavas.

Sample locality, petrography and mineral composition
In the alpine chain, Eocene to Oligocene igneous rocks are exposed along the several 100 km-long Periadriatic Fault System that strikes E-W. Subduction-related magmatism occurred from ~ 43 Ma to ~ 29 Ma with input of mafic material at ~ 32-29 and ~ 43-40 Ma (Müntener et al. 2021). The first period ceased prior to the collision of Europe and Africa and the later period represents the phase of continental arc magmatism. The latter age range agrees well with the 40 Ar/ 39 Ar ages of 30.1-32.6 Ma obtained in this study (Table A1), which also fall in the range of ages obtained on the Bergell intrusion (30-32 Ma; von Blankenburg 1992; Oberli et al. 2004;Gianola et al. 2014); an intrusion that is characterized by high-K magmatic rocks. Numerous authors (Nievergelt and Dietrich 1977;Gautschi and Montrasio 1978;Beccaluva et al. 1979Beccaluva et al. , 1983Dal Piaz et al. 1979Venturelli et al. 1984;Diethelm 1990;von Blanckenburg et al. 1992;Ruffini et al. 1997) reported on the occurrence of andesitic rocks including rare shoshonites from the Alpine chain that occur mostly as dykes and as fragments within post-orogenic sedimentary successions. The Oligocene to Miocene succession of the Gonfolite Lombarda Formation crops out in western Lombardy (Fig. 1). It can be considered as a typical example of a post-orogenic molasse deposit although subsequent work indicates that it more likely represents a clastic wedge infilling a deep foreland basin flanking a zone of continental collision (Gelati et al. 1988). The succession covers a time span from late early Oligocene to early Miocene and consists of clastic sedimentary rocks that were deposited, based on bentic foraminifera and macrofossil evidence, in a deep-sea environment. The Gonfolite Group as part of the Gonfolite Lombarda succession consists at the base of the Como Conglomerate. Within this conglomerate, numerous boulders of granite-granodiorite of the Val Masino-Val Bregglia massif occur as well as numerous boulders of high-grade metamorphic rocks (eclogites, blue schists, mica schists, felsic granulites, mafic granulites, amphibolites) from the crystalline basement of the Alps (Gelati et al. 1988). This formation also yields numerous boulders of volcanic rocks that form the sample basis of this study. These boulders which are elliptical in shape have  Como long axes between 15 and 30 cm and a length to width ratio of 1.5 to 2 and were collected at Rio della Maiocca south of Chiasso, at San Fermo Est and at Ponte Chiasso ( Fig. 1). Typical mineral assemblages of the high-K andesites and dacites consist of plagioclase, amphibole, biotite and titanomagnetite. Mineral compositions are given in Table A2 and element profiles for amphibole are shown in Fig. A1. All samples are highly porphyritic and phenocrysts of plagioclase and amphibole may reach grain sizes up to 2 mm. Amphibole occur as phenocrysts or forms mineral aggregates and can be classified as edenite. Two samples contain large flakes of biotite. Apatite is common and occurs as microphenocrysts. Phenocrysts, microphenocrysts and mineral aggregates are enclosed in a fine-grained, slightly altered groundmass in which primary glass is absent. Euhedral to subhedral plagioclase is usually twinned and its composition ranges from An39 to An58 (Table A2). Zoning is commonly normal, reverse or oscillatory.

Methods
40 Ar/ 39 Ar dating was carried out at Argonlabor Freiberg (ALF), TU Freiberg, Germany. Raw-data reduction and time-zero intercept calculations were done using an in-house developed Matlab software package, isochron, inverse isochron and plateau ages have been calculated with the Excel Add-In ISOPLOT 3.7 (Ludwig 2008). Fish Canyon Tuff sanidine was used as flux monitor (28.305 ± 0.036 Ma; Renne et al. 2010) and the decay constants of Renne et al. (2010) served as base for all calculations. A detailed description of the method is given in the appendix and data are given in the online resources. Minerals were analyzed with a JEOL microprobe at the University of Münster, Germany. Operating condition were 15 kV and 5 nA, with a peak counting time of 5 s and a background counting time of 3 s with a beam size of 5-20 μm. The ZAF correction procedure was applied to the data; errors in the major oxides are estimated to be about 1-2%. Representative mineral data are given in Table A2. Whole rock powders were prepared from rock chips free from surface alteration using a jaw crusher, a ball mill and an agate mortar. Major and some trace elements (except for rare-earth elements (REE) and Nb, Ta, Hf, Th and U) were determined on fused lithium-tetraborate glass beads using standard XRF techniques. A more detailed description of the method is given in the appendix and data for certified reference materials are given in Table A3. For some samples, trace element concentrations of Nb, Hf, Ta, Th, and U were determined by isotope dilution inductively coupled mass spectrometry (ID-ICPMS) using a Thermo-Electron ELEMENT 2 high-resolution ICPMS at Max-Planck Institut für Chemie, Mainz. A detailed description of the method can be found in the appendix and in Willbold and Jochum (2005) and data for certified reference materials are given in Table A3. Sr, Nd and Pb isotope analyses were carried out at the Max-Planck-Institut für Chemie at Mainz with a Finnigan MAT 261 multicollector thermal ionization mass spectrometer operating in static mode. A more detailed description of the method can be found in the appendix and data for certified reference materials are given in Table A4. Isotope analyses of Lu and Hf were carried out on a Thermo Fisher ® Neptune MC-ICP-MS at U Amsterdam. A more detailed description of the method can be found in the appendix and data for certified reference materials are given in Table A4. Oxygen isotope analyses for samples CS 42,CS 56,CS 66,CS 88,CS 22,CS 39,CS 53,CS 59 and CS 89 were performed at the University of Bonn on ~ 10 mg aliquots of powdered whole-rock samples. Oxygen isotope analyses for samples CS 69, CS 32, CS 54, CS 57, CS 91 and CS 97 were analyzed at Universität Göttingen using infrared (IR) laser fluorination in combination with gas chromatography isotope ratio monitoring gas mass spectrometry. Analytical uncertainties are < 0.2‰. All samples were measured twice and the average of these measurements is given in Table 2. All values are reported as per mil deviations relative to SMOW (Standard Mean Ocean Water) and refer to a certified value of 9.6 ± 0.2‰ for NBS 28. A more detailed description of the method can be found in the appendix and data are given in Table 2.

Ar-Ar ages
For all three samples dated in this study, the reported plateau and inverse isochron ages ( Fig. A2) are identical within error, and the initial 40 Ar/ 36 Ar compositions are indistinguishable from the atmospheric value of 298.6 ± 0.3 (Lee et al. 2006). Therefore, the plateau ages are taken as best estimates for the time of cooling through the ~ 550 °C (amphibole) and ~ 335 °C (biotite) temperature range (cooling temperatures calculated using Mark Brandon's closure program and diffusion datasets of Harrison (1981) and Grove and Harrison (1996) and assuming a cooling rate of 70 °C/Ma and a uniform grain size of 250 µm). The age spectrum of sample CS22 exhibits an excess Ar component present in the low temperature steps, but provides a plateau age of 32.6 ± 0.3 Ma (2σ) from 78.5% of the released 39 Ar. K/Ca ratios are fairly constant for the steps comprising the plateau and indicate Ar release from a homogeneous unaltered phase. For sample CS66, low signal intensities cause relatively large errors on the individual temperature steps, therefore, all steps have been used to calculate an age of 32.8 ± 0.7 Ma. The biotite separate from sample CS53 provides a slightly disturbed age spectrum and a somewhat lower age than the amphiboles of 30.1 ± 0.1 Ma from 95.5% of the released 39 Ar. Data are given in the online resources.

Geochemistry
Analytical techniques are given in the Appendix A1. Major element compositions of whole-rock samples are given in Table 1. Selected major elements are plotted against SiO 2 in Fig. 3. According to their K 2 O content the samples are classified as high-K andesites and dacites with X Mg (X Mg = molar Mg/molar Mg + molar Fe; all Fe as FeO) ranging from 0.47 to 0.57. SiO 2 ranges from 57.4 to 66.8 wt%. TiO 2 , FeO* (all Fe as FeO), MgO, CaO and Al 2 O 3 decrease with increasing SiO 2 . Na 2 O increases weakly from c. 2.7-3.2 wt % in the least evolved samples to c. 3.7 wt. % in the most evolved samples. P 2 O 5 concentrations are rather low (0.12-0.21 wt. %) and increase with increasing SiO 2 .
Trace element compositions are given in Table 1 and are plotted in Fig. 3a Deutsch (1984), Beccaluva et al. (1979), Diethelm (1990), Gautschi and Montrasio (1978), Dal Piaz et al. (1979 and Ruffini et al. (1997). Reference data for mafic high-K rocks from Hürlimann et al (2016) and Bergomi et al. (2015) are shown for comparison    0.87 to 0.63. The primitive mantle-normalized trace element patterns have a shape that is typical of volcanic rocks from subduction-related tectonic settings. They are enriched in the LILE (Large Ion Lithophile Elements), U, Th, Pb, LREE relative to HREE and are depleted in Nb, Ta, P and Ti (Fig. 5). Isotope compositions are presented in Table 2. Both, the initial 87 Sr/ 86 Sr and the initial 143 Nd/ 144 Nd isotope ratios (expressed as εNd) are variable ranging from 0.7085 to 0.7113 for 87 Sr/ 86 Sr and from -2.0 to -6.7 for εNd ( Fig. 6). Initial Hf isotope compositions (expressed as εHf) vary from -4.0 to + 2.5. δ 18 O values range from + 6.5 ‰ to + 9.1 ‰. 206

Fractional crystallization and continental crustal contamination
The crystal fractionation history of arc magmas involves fractionation of olivine, spinel, and clinopyroxene from high-Mg basalt to basaltic andesite. Further fractionation from basaltic andesite to andesite involves minor clinopyroxene, amphibole and minor plagioclase (Hürlimann et al. 2016). This sequence of fractionation steps is seen in decreasing FeO, MgO, CaO, Ni, Cr, and increasing K 2 O, Rb, Hf, Zr, Th, U, Ta, Pb, Nb and Ba from high-Mg rocks towards the andesites (Figs. 2 and 3). Interestingly, Al 2 O 3 first increase from high-Mg basalt to basaltic andesite. This feature can be interpreted as a sign of high-pressure fractional crystallization due to a lack of plagioclase fractionation. Further, Al 2 O 3 then decreases towards the andesites-dacites implying plagioclase fractionation, compatible with the negative Eu anomaly seen in the REE pattern (Fig. 4). The decrease in TiO 2 together with the development of a negative Ti-anomaly in the multielement diagram ( Fig. 5) implies fractionation of amphibole where Ti becomes compatible in amphibole (Hürlimann et al. 2016). Like in many continental arcs worldwide, rocks that represent primary magmas are absent among the suite investigated here and even the most mafic samples have low abundances of MgO, Ni and Cr suggesting that the parental magmas underwent significant crystal fractionation of likely olivine and pyroxene (Hürlimann et al. 2016). Since isotope compositions vary with elemental abundances (Fig. 8), these isotopic trends are interpreted as indicating assimilation of crustal material (AFC; DePaolo 1981) or sediment melt mediated source enrichment and subsequent partial melting of those sources. The correlation of LILE enrichment, K in particular, and radiogenic isotopic composition supports the inference that the K-rich character is, at least, in part related to continental crustal  andesites-dacites Alpine high-K andesites high-K rocks (Bergomi et al. 2015) high-K rocks (Hürlimann et al. 2016)  andesites-dacites Alpine high-K andesites high-K rocks (Bergomi et al. 2015) high-K rocks (Hürlimann et al. 2016) (a) (b) Fig. 3 Selected trace element diagrams of the andesites and dacites. Grey field denote data for other high-K andesites and dacites from the Alps according to Deutsch (1984), Beccaluva et al. (1979), Diethelm (1990), Gautschi andMontrasio (1978), Dal Piaz et al. (1979), (1988) and Ruffini et al. (1997). Reference data for mafic high-K rocks from Hürlimann et al (2016) and Bergomi et al. (2015) are shown for comparison  Sun and McDonough (1989). Reference data for mafic high-K rocks from Hürlimann et al (2016) and Bergomi et al. (2015) are shown for comparison contamination that caused an increase in K 2 O coupled with an increase in 206 Pb/ 204 Pb and a decrease in εHf and εNd values (Fig. 9). Metasedimentary rocks from the Alps are too unradiogenic to serve as a source (eNd: -11 to -9, Henry et al. 1997) but may be potential contaminants. A lack of data precludes a more detailed evaluation in this case. In a plot of εNd vs. 87 Sr/ 86 Sr (Fig. 7), the samples define part of a curved array that is consistent with simple twocomponent mixing between a hypothetical mantle component and a lower crustal component. Granulite-facies metasedimentary rocks from the alpine basement (Voshage et al. 1990) have radiogenic Sr and less radiogenic Nd isotope compositions and up to 20% assimilation of such rocks by the most primitive andesite may explain the isotope composition of the most evolved andesites and dacites. Although some of these Hercynian gneisses and metasediments have undergone extensive partial melting (Schnetger 1994) and may thus not represent fertile lithologies, they represent a crustal endmember with radiogenic Sr and unradiogenic Nd isotope signatures. Additional presumably more fertile potential contaminants are Permian granitoids that intruded the area (Pinnarelli et al. 2002). In addition, the andesites and dacites have a relatively large range in 207 Pb/ 204 Pb, 208 Pb/ 204 Pb and 206 Pb/ 204 Pb ratios (Fig. 7) supporting that assimilation of ancient continental crust played a role in the evolution of these samples. In Fig. 7, the Pb isotope composition of Hercynian lower crustal mafic, felsic and metasedimentary rocks from the Ivrea zone (Southern Alps), some Alpine upper crustal granites and Permian granites from the Southern Alps as potential assimilants are plotted. Lead isotope data from lower crustal rocks from the southern Alps are limited (Cumming et al. 1987) but alpine granites and Permian granites from the southern Alps have distinctly more radiogenic Pb isotope compositions, and considering Pb isotopes alone, the andesites and dacites may be differentiated mantle wedge derived magmas that underwent substantial assimilation of upper crustal material with a Pb isotope signature similar to that of the Alpine granites or the Permian granites (Fig. 7). Potential mantle-derived subducted material (Cannaò et al. 2015;McCarthy and Müntener 2019) show a large range in Pb isotopes with nearly complete overlap with the andesites-dacites studied here making an evaluation of the significance of such material in the genesis of the andesites-dacites impossible. It is, therefore, assumed that the Pb isotope signature of the andesites and dacites reflects a combination of crustal assimilation and source features, where fluids or partial melts from subducted material (upper mantle rocks, mantle-derived igneous rocks, and metasediments) imprinted an upper continental crustaldominated Pb isotope signature to the mantle wedge prior to melting. Alpine andesite (Callegari et al., 2004) Bergell intrusion at 30 Ma (v. Blanckenburg et al., 1992;Oschidari & Ziegler, 1992) Other felsic periadriatic intrusions at 30 Ma (Juteau et al., 1986;Kagami et al., 1991) Gabbros/Diorites (Ivrea zone) at 30 Ma (Voshage et al., 1990) Felsic and metasedimentary rocks (Ivrea zone) at 30 Ma (Voshage et al., 1990) Permian igneous rocks (Pinarelli et al. 2002) Mafic volcanic rocks (Hürlimann et al. 2016 Rb, Sr, Sm and Nd were measured by N-TIMS and Lu and Hf were measured by MC-ICP-MS. Analytical in-run errors are typically 0.000015 or less. Calculation of εNd values is relative to CHUR according to Jacobsen and Wasserburg (1980). Nd model age (TDM) calculation is according to Michard et al. (1985). Pb isotope ratios were recalculated using U, Th and Pb abundances from The question then arises whether assimilation of crustal rocks occurred as bulk assimilation (DePaolo 1981) or selective assimilation (Glazner 2007). In contrast to bulk assimilation where the entire lower crustal rocks are assimilated, selective assimilation is a process where partial melts consisting of mineral assemblages with a low melting point (feldspars, mica) from the lower crustal rocks are incorporated into the rising magma. The size of the magma chamber(s) that fed the andesites-dacites is not known but their fairly evolved nature suggests that bulk assimilation is unlikely. The lower crust in the southern Alps consists of high-grade metamorphic rocks (kinzigites, stronalites) in which the granulite-facies stronalites are melt-depleted (Schnetger 1994) and may thus represent non-fertile contaminants. More fertile rocks are the amphibolite-facies kinzigites and the Permian granites that intruded the area. Thus, incorporation of partial melts consisting of plagioclase, K-feldspar and biotite from the lower crust seems more reasonable. To constrain the thermal effects of a rising andesite magma incorporating low-melting components from lower crustal metasedimentary or igneous sources, we performed several calculations using the EC-AFC software package (Spera and Bohrson 2001). The results are shown in the inset to Fig. 6 and the EC-AFC parameters used are given in Table A5. It is important to note that we want to show the evolution within the andesite-dacites and not the evolution from more mafic high-K rocks (i.e., Hürlimann et al. 2016) towards the andesites. It is thus not surprising that the results using EC-AFC calculation are broadly similar to those derived by common AFC calculations (i.e., DePaolo 1981). Estimated degrees of assimilation are about 20% using both, a non-depleted kinzingite (Voshage et al. 1987) from the Alpine basement and a Permian granite (Pinarelli et al. 2002) that intruded the area (Fig. 7).
The Nb/Ta ratio is also strongly correlated with indices of fractional crystallization such as SiO 2 and K 2 O (Fig. 8) in which the most fractionated rocks have the lowest Nb/Ta. This implies that during fractionation, minerals with high Nb/Ta were fractionated. These minerals were likely amphibole and biotite  which are important phenocrysts phases in the andesites. In addition, Nb/Ta is also correlated with εHf suggesting some modification of this ratio during AFC processes. This view is compatible with the Nb/Ta of lower crustal rocks which is substantially lower than upper crustal or MORB estimates (Nb/Ta: 8.3; Rudnick and Gao 2004). A full set of HFSE on felsic and mafic lower crust from the alpine basement is not available, therefore, precluding a more detailed evaluation using local endmembers.
In Hf-Nd isotope space, the samples deviate from the crustal array (Vervoort and Patchett 1996), a line which marks the co-variation of both isotope systems during magmatic differentiation and crustal evolution (Fig. 9). Assimilation of bulk sediment that diverts from the array through the zircon effect is expected to show a deviation towards radiogenic Nd at unradiogenic Hf (Woodhead et al. 2017), however, the opposite of what is observed. This Hf-Nd isotope decoupling hence implies interaction of an inferred  Juteau et al. 1986) Mafic rocks (lower crust?; Cumming et al. 1987) Felsic and metasedimentary rocks (lower crust?; Cumming et al. 1987) mafic alpine mafic rocks (Hürlimann et al. 2016) Hercynian igneous rocks (Pinarelli et al. 1993, Boriani et al. 1995 Subducted material (Cannaò et al. 2015;McCarthy and Müntener, 2017)

Fig. 7
Plot of a 208 Pb/ 204 Pb and b 207 Pb/ 204 Pb versus 206 Pb/ 204 Pb isotope ratios for the andesites and dacites. Also shown are reference data for alpine granites from Juteau et al. (1986), mafic, felsic and metasedimentary rocks from Cummings et al. (1987) and Permian igneous rocks (light grey-shaded field) from Pinarelli et al. (1993) and Boriani et al. (1995). The dark grey-shaded fields show data from Jurassic mafic rocks that may potentially represent subducted mafic material. Note that mafic and felsic lower crustal rocks are too unradiogenic in 206 Pb/ 204 Pb to serve as possible contaminants and Permian igneous rocks, although most of them are slightly too radiogenic, are more likely candidates for crustal assimilation, at least with respect to the Pb isotope composition primary magma with a reservoir that shows radiogenic Hf at unradiogenic Nd, a feature typical for aged garnet-bearing rocks. Assimilation of Hercynian garnet-bearing lithologies of the lowermost crust would create the observed deviation, and agrees with the correlation observed for Nd-Sr isotope compositions (Fig. 6). Oxygen isotope constraints support a crustal influence during magma differentiation, as oxygen isotopes of the andesites and dacites (6.5-9.1 ‰) are heavier than average MORB (5.4-6.1 ‰; Ito et al. 1987;Eiler 2001). The lack of correlation between oxygen isotopes and radiogenic isotopes as well as trace elements, however, implies that fluid-rock interaction along with crustal contamination likely modified the primary oxygen isotope composition. In conclusion, the most feasible scenario to explain the geochemical features of the andesites and dacites is wall-rock assimilation possibly in the lower crust where heat fluxes are high. Such a scenario fails to explain the high amounts of LILE and K as well as the radiogenic Pb isotope signatures observed in the samples, which require additional assimilation of more evolved silicic upper crustal rocks, or need to be source features that have been imprinted to the depleted mantle wedge by fluid addition or sediment melt from the slab prior to melting.
The results from this study support previous studies on the generation of arc-derived magmatic rocks that also provided clear evidence for the involvement of continental crustal material during their evolution (Davidson et al. 1990;Hildreth and Moorbath 1988;Davidson and Harmon 1989;Rogers and Hawkesworth 1989;Ellam and Harmon 1990;Thirlwall et al. 1996 among many others). Whereas some studies concluded that these crustal signatures are premelting features of the mantle wedge caused by fluid and  (Bergomi et al. 2015) high-K rocks (Hürlimann et al. 2016) sediment or sediment melt addition from the down-going slab (e.g., Keppler 1996;Elliott et al. 1997;Elliott 2003;Kelemen et al. 2004;Chen et al. 2021), other studies provided evidence that these crustal signatures reflect differentiation processes within different levels of the arc crust (e.g., Jagoutz 2014; Jagoutz and Kelemen 2015; Ducea and Saleeby 2015;Tang et al. 2019;Pfänder et al. 2021), or are a combination of both. It is evident, that post-melting assimilation and fractional crystallization (AFC) may also modify the geochemical signatures of arc magmas.

The role of slab-derived components and sediment subduction
Some of the K-rich character may be attributed to combined fractional crystallization-assimilation processes consistent with the correlation between radiogenic isotope compositions and K 2 O (Fig. 8). However, average continental crust has only 1.8 wt. % K 2 O (Rudnick and Gao 2004) and, therefore, assimilation of specific K-rich lithologies (glimmerites, metapelites, K-rich granites) is required. Alternatively, processes other than fractional crystallization-assimilation or specific sources may have contributed to the K-rich nature. An extrapolation back to c. 50 wt. % SiO 2 (Fig. 2) suggests that the hypothetical parental melts also had a high-K character (1.5-2.0 wt. %; Hürlimann et al. 2016) and it is proposed that the infiltration of slab-derived fluids or melts from subducted pelagic sediments produced substantial chemical enrichments in the mantle wedge and hence in the melts derived from it.
Fluid phases produced by slab dehydration effectively concentrate Ba, Pb, Sr and other LILE among further elements such as U (Myers and Marsh 1987;Ben Othman et al. 1989;Kessel et al. 2005). Therefore, incompatible trace element ratios involving Ba relative to fluid-immobile trace elements of similar incompatibility (Nb, La, and Zr) should be high as long as a slab-derived fluid phase is involved ( Fig. 10; McCulloch and Perfit 1981;Tatsumi et al. 1986). Since average continental crust has low Ba/La (22.8), Ba/ Nb (57) and Ba/Zr (3.5) ratios (Rudnick and Gao 2004), the ε Hf (init.) OIB array (Johnson and Beard, 1993) Crustal array (Vervoort and Patchett, 1990)

Fig. 9
Plot of initial εHf vs. initial εNd for the andesites and dacites. Also shown are arrays for continental crust (Vervoort and Patchett 1996) and OIB (Johnson and Beard 1993). The deviation towards radiogenic Hf imply involvement of an ancient high Lu/Hf endmember, most likely garnet-bearing lower crust Compositions of continental crust (CC) and MORB are taken from Rudnick and Gao (2004) and Gale et al. (2013), respectively variable and in part higher ratios of these elements observed in the andesites-dacites (Ba/La: 16-37, Ba/Nb: 32-69 and Ba/Zr: 2.7-8.0) are unlikely to result from crustal contamination alone and follow more likely from fluid infiltration from the slab to the mantle wedge.
All samples have high Pb abundances which are substantially higher than in continental arc andesites which on average have 4-8 ppm Pb (Kelemen et al. 2004). Lead is an aqueous fluid-mobile element (Brenan et al. 1995) and when plotted against Nd because of their similar melt-peridotite partition coefficients and because Nd is less fluid-mobile than Pb, linear relationships between Pb abundances and Pb/ Nd ratios are expected for rock series from aqueous fluidenriched sources, a feature observed for the andesites and dacites (Fig. 11). Moreover, the samples with high Ba/La ratios have also high Pb/Nd ratios further strengthening the argument that an aqueous fluid was involved in magma genesis, along with sediment-derived melts. It is noteworthy that the latter should also have Pb/Nd ratios ≥ 1 (Münker 2000), which thus may have also contributed to the samples with Pb/Nd ≥ 1, compatible with the Ba/Nb-Th/Nb constraints. Other high-K andesites that also show high Ba/Nb and Th/ Nb ratios (Price et al. 1999;Zernack et al. 2012) show low Ce/Pb and low-Nb/U ratios (Zernack et al. 2012: Ce/Pb < 8, Nb/U < 10; Price et al. 1999: Ce/Pb < 6, Nb/U < 10) that are commonly interpreted to indicate the presence of a fluid that was derived from subducted crustal sediments. (Ben Othman et al. 1989;Stern et al. 1993;Klein and Karsten 1996;Jackson et al. 2007). Similarly, Miller et al. (1994) have shown that low Ce/Pb in arc lavas are not generated through common melting processes but require the enrichment of Pb in the sources through incorporation of a sedimentary component in the presence of a fluid phase extracted from the subducted slab (MORB source). For the andesites and dacites from this study, Nb, U, Ce, and Pb increase with decreasing MgO (Fig. 3). Nb/U decreases with increasing Nb and U; the same is observed for Ce/Pb vs. Pb (Fig. 12) and Ce, although Ce is somewhat more scattered. More importantly,  Rudnick and Gao (2004) and Gale et al. (2013), respectively however, Ce/Pb (and Nb/U) ratios decrease with increasing 206 Pb/ 204 Pb (Fig. 12). In addition, ∆7/4 (+ 8.8 to + 12.7) and ∆ 8/4 (+ 38 to + 39) values (∆ 7/4 and ∆ 8/4 after Hart 1984; not shown in detail here) reveal an excess of radiogenic Pb with values that are similar to GLOSS estimates (GLObal Subducting Sediment; Plank and Langmuir 1998;Jackson et al. 2007). These features are consistent with the addition of a fluid phase with high Pb contents and a continental crustal-like Pb isotope signal to the mantle wedge, where the source of the fluid was most likely dehydrating subducting sediment. It is, however, not possible to distinguish clearly between a sedimentary component from the subducting slab and a lower crust-derived metasedimentary component. Ba/Nb and Th/Nb are correlated among the andesites-dacites (Fig. 10). Thorium is considered to be a mostly fluid-immobile element that is strongly depleted in the residual mantle. However, Rustioni et al. (2021) observed some fluid mobility of Th at high salinity at 4 GPa and at low salinity at 6 GPa. This study does not provide Cl data to test this hypothesis and pressures of 4-6 GPa are potentially too high to be relevant for the generation of the andesites studied here. Elevated Th contents in arc-related magmas are, therefore, commonly explained by sediment or sediment melt addition from the down-going slab to the mantle wedge (e.g., Class et al. 2000;Woodhead et al. 2001). Ba is a hydrous fluid-sensitive trace element; hence, the positive correlation between Ba/Nb and Th/Nb implies the presence of hydrous fluid and sediment-or sediment melt in the source of the andesites-dacites. Ba/Th ratios are a proxy to estimate the fluid to sediment/sediment melt ratio of the fluxing component. The andesites-dacites from this study have low Ba/ Th ratios between ~ 40 and 60 that are unrelated to Ba or Th contents (Fig. 10). Ba abundances (247-952 ppm) are similar but Th abundances (4.75-19.6 ppm) are higher relative to other high-K suites (Price et al. 1999: Ba/Th > 100 with Ba: 800-1200 ppm and Th < 10 ppm; Zernack et al. 2012: Ba/ Th > 120 with Ba: 300-700 ppm and Th: < 7 ppm). This suggests a greater contribution of sediment or sediment-derived melt relative to aqueous fluid during mantle metasomatism in the case of alpine subduction. To further constrain the effect of addition of sediment or sediment melt, we evaluated the Th/Yb vs. Ba/La systematics (Fig. 11 Figure 11 shows that the andesites-dacites exhibit a large variation in Th/Yb at low Ba/La which suggests that addition of sediment or sediment melt strongly contributed to the trace element variation in the andesites-dacites. In conclusion, the trace element composition and Pb isotope composition of the andesites and dacites favor the addition of metasomatizing fluids and sediment-derived melts to the mantle wedge, with a dominance of the latter. Both components have been derived from dehydrating subducted sediments or hydrous sediment melts and possibly oceanic crust. To further constrain the significance of oceanic crust, we show data from Jurassic oceanic crust from the Alps (McCarthy and Müntener 2019) in Figs. 10, 11 and 12. The data show that this type of oceanic crust likely played only a minor role in the generation of the andesites-dacites studied here. The involvement of a sediment-derived melt requires comparatively high slab temperatures above the solidus of water-saturated metapelites (> 640 °C at pressures > 2 GPa; Mann and Schmitt 2015). Although such a model best explains the observed geochemical features and in particular the high-K contents, it is emphasized that the elemental and isotopic signals might be in part compromised by postmelting AFC processes within the crust.

Source constraints and intra-crustal differentiation deduced from HFSE systematics
High field strength elements (HFSE) are not transferred significantly to the mantle wedge during slab dehydration because of their low mobility in aqueous fluids (McCulloch and Gamble 1991;Brenan et al. 1995;Keppler 1996;Audétat and Keppler 2005;Antignano and Manning 2008). Therefore, concentrations and ratios of HFSE are insensitive to slab dehydration and mantle wedge metasomatism and mostly constrain the background composition of the mantle wedge (McCulloch and Gamble 1991;Woodhead et al. 1993;Elliott et al. 1997). Depleted HFSE contents relative to similarly incompatible trace elements and low-Nb/Ta and Zr/Hf ratios in arc volcanic rocks have been interpreted as reflecting remelting of a refractory mantle wedge that has been previously enriched in all but the HFSE by metasomatizing fluids, where the HFSE are retained and in part fractionated by residual Fe-Ti phases (mostly rutile) in the slab (the wedge depletion model of Plank and White 1995). The most primitive andesites from this study have low-Nb/ Ta ratios of 11.7-13.0, which are lower than MORB estimates ( Fig. 13)  high-K rocks (Bergomi et al. 2015) andesites-dacites UCC high-K rocks (Hürlimann et al. 2016) Fig. 13 Plot of Nb/Ta vs. Zr/Hf for the andesites-dacites. Figure taken and slightly modified from Pfänder et al. (2007). BSE Bulk Silicate Earth, UCC Upper Continental Crust and LCC Lower Continental Crust from Rudnick and Gao (2004); MORB Mid Ocean Ridge Basalt, OIB Ocean Island Basalt. Reference fields for various intraoceanic and continental arcs are shown as well as reference data of high-K mafic rocks from Hürlimann et al. (2016) and Bergomi et al (2015). Note that both data sets show a trend towards low Nb/Ta similar to the andesites-dacites studied here. The trend can be interpreted as a result of combined AFC processes involving fractionation of amphibole and biotite  combined with uptake of low-Nb/Ta lower crustal rocks (Gertisser and Keller 2003;Kirchenbaur et al. 2022) other hand, Zr/Hf ratios cover the range observed in MORB (Fig. 13) and, therefore, the derivation of the parental magmas of the andesites-dacites from a common mantle peridotite that was depleted by a previous melting event seems not a viable process. This view is consistent with the inferred addition of sediment-derived melt and the negative correlation of Nb/Ta vs. Nb, which is the opposite of what is observed in arc rocks that result from remelting of depleted sources (Münker 1998).
Nb/Ta ratios in medium-to high-K subduction-related basalts and andesites are variable and can range from superchondritic (> 19.9) to subchondritic values as low as ~ 6 (Stolz et al. 1996;Münker 1998;Münker et al. 2004;Fig. 13). The subchrondritic values have been either explained by preferential retention of Nb in rutile (Münker 1998) or by shallow melting in the presence of residual amphibole with high Nb/Ta (König and Schuth 2011). Independent of the primary cause of low Nb/Ta in arc-related primary magmas, the negative correlation of Nb/Ta with SiO 2 observed in the andesites and dacites suggests postmelting fractionation of Nb and Ta, likely in the course of intra-crustal differentiation due to assimilation of low-Nb/ Ta lower crustal rocks (Figs. 8 and 13). Nb, Ta, Zr and Hf concentrations from a coherent set of samples from the Ivrea Zone of the Alps, which were, based on isotope constraints, considered as potential assimilants (Fig. 6) are not available. However, a combination of trace element contents reported by Bea and Monteiro (1999) who do not report Hf concentrations and Schnetger (1994) who do not report Nb concentrations indicate Nb/Ta ratios between 11 and 18 and Zr/Hf ratios of ~ 43 for lower crustal rocks from the Alps. However, due to a lack of appropriate data a more compelling modeling is precluded. Despite the fact that crustal assimilation and magma mixing are considered to be fundamental in the generation or arc magmas (Grove et al. 1983), more recent studies emphasize the role of intra-crustal magma differentiation from basaltic to andesitic and dacitic magmas that are mainly controlled by crystal-liquid segregation (Jagoutz et al. 2009;Lee and Bachmann 2014;Tang et al. 2019;Pfänder et al. 2021). In these models, the trace element signatures of residual magmas depend on the crystallizing assemblages and relevant partition coefficients. The fractionation of ol + cpx + plag has limited effects on Nb/Ta due to the highly incompatible behavior of both elements (Adam and Green 2006), whereas the fractionation of amphibole with low Mg# (less than 70) can impart a lower Nb/Ta ratio to the coexisting melt (Foley et al. 2002;Müntener et al. 2018). Thus, fractionation of amphibole which are phenocrysts in the andesites may have in part contributed to the low Nb/Ta. Müntener et al. (2018) have also shown that fractionation of amphibole alone is not capable of decreasing Nb/Ta ratios significantly and another major mineral phase such as biotite is needed. Since biotite is an additional important mineral phase in the andesites, we concur with Müntener et al. (2018) that additional fractionation of biotite also contributed to the low Nb/Ta of the andesites-dacites. The described features, therefore, provide strong evidence that the andesites and dacites evolved by intra-crustal crystal-liquid differentiation, likely accompanied by assimilation of pre-existing Hercynian lower continental crust.

Conclusions and towards an evolutionary model for arc-related volcanism in the Alps
It is well known that subduction-related magmas originate in the mantle wedge above a subduction zone (e.g., Grove and Kinzler 1986;Crawford et al. 1987;Hawkesworth et al. 1991;McCulloch and Gamble 1991;Tatsumi and Eggins 1995 among many others). Dehydration reactions occurring in the lithosphere of the subducting plate release fluids and melts and interaction with the overlying mantle wedge generate primitive high-Mg basaltic magmas (Kelemen et al. 2004) that later may evolve towards more differentiated rocks by FC or AFC processes (Hawkesworth et al. 1979;Arculus and Powell 1986;Davies and Stevenson 1992;Tatsumi and Eggins 1995;Schmidt and Poli 1998;Ulmer 2001;Grove et al. 2002;Forneris and Holloway 2003).
High-K andesites and dacites analyzed in this study exhibit regular variations in major and trace element abundances indicating that fractional crystallization processes exert a major role in their genesis. In addition, Sr, Nd, Pb and Hf isotopes show significant correlations among each other but also with K 2 O abundances, further implying a contribution of crustal rocks through AFC processes and addition of sediment-derived slab melts to the source of these rocks. Regular variations of fluid-sensitive trace elements such as Ba and Pb show that involvement of a fluid phase also contributed to the geochemical composition. Hf-Nd isotope decoupling suggests assimilation of aged garnetbearing lithologies, likely Hercynian lower crustal rocks. It is obvious that the composition of the alpine andesites and dacites is not the result of a single-stage process and it is, therefore, proposed that after partial melting in a metasomatized mantle wedge, which produced unexposed high-K basalts (i.e., Hürlimann et al. 2016) and basaltic andesites, further evolution of the andesites and dacites took place in the arc crust. Here, they might have crystallized ultramafic cumulates (dunites, wehrlites, pyroxenites, hornblendites, opx-hornblendites; e.g., Müntener et al. 2021;Jagoutz et al. 2009;Hürlimann et al. 2016;Pfänder et al. 2021) and possibly promoted partial melting of the lower crust which, depending on the maturity of the crust, may consist of a mixture of meta-igneous (Hercynian and Permian intrusive rocks; Boriani et al. 1995;Pinarelli et al. 1993Pinarelli et al. , 2002 and metasedimentary rocks (Hercynian amphibolite-to granulite-facies metasedimentary rocks; Schnetger 1994; Voshage et al. 1987Voshage et al. , 1990. This view is in accordance with the AFC model that has been used to explain the coupling of elemental and isotopic composition of the andesites and dacites from this study. Therefore, the chemical diversity of the studied andesites and dacites is, at least in part, acquired in the lower crust through complex AFC processes involving Hercynian or pre-Hercynian lower garnet-bearing crustal lithologies. Subsequent later stage magma differentiation at lower pressures in higher crustal levels involving plagioclase is suggested by decreasing CaO and Al 2 O 3 with increasing SiO 2 (Fig. 2) as well as by slightly negative Eu anomalies (Fig. 4). Presumably, this differentiation was the final process that produced these andesites and dacites, which are thus end products of a process chain that might be representative for the formation of primary continental crust in continental arcs.
Older works (i.e., Beccaluva et al. 1979) have tried to explain the complex geochemical composition of andesites from the Alpine chain as a result of partial melting of quartz eclogite in the down-going slab, followed by reaction with the upper mantle wedge and low-pressure crystal fractionation at crustal levels. These authors also showed the existence of andesites with elevated MgO, Ni and Cr abundances and emphasized a role for distinct primary melts that followed different evolutionary lines. Others (Dal Piaz 1979;Beccaluva et al. 1983) noted distinct parental sources with compositional differences and strengthen a role for a metasomatized subcontinental lithospheric mantle especially for the generation of the LILE-and LREE-enriched high-K andesites and shoshonites. Mantle metasomatism (or mantle contamination) could have been achieved by subduction of continental material accompanied by fluid release during high-pressure transformation (Dal Piaz et al. 1979;von Blanckenburg et al. 1992). Such a process may explain LILE enrichment, elevated 87 Sr/ 86 Sr and radiogenic Pb isotope compositions. However, other evidences (high δ 18 O values, radiogenic 87 Sr/ 86 Sr, negative εNd values, radiogenic Pb isotopes and correlation among these isotope systems) are rather indicative of crustal contamination; a significant process proposed by Beccaluva et al. (1983), Venturelli et al. (1984), Dal Piaz (1988 for the modification of some andesites and shoshonites from the northwestern and southern Alps. The Hf isotope compositions reported here also shed new light on the compositions of andesites in the Alpine chain and on possible crustal constituents in the lower crust. Whole rock and zircon Hf isotope data for alpine igneous rocks (Stille and Steiger 1991) indicate a range of εHf values from + 3.4 to -2.2 which are similar to the εHf values reported from the least contaminated andesites from this study. Pre-alpine igneous rocks have εHf (30 Ma) values ranging from + 1.2 to -6.8 (Schaltegger and Brack 2007), again similar to the andesites and dacites studied here. In view of the apparent AFC history of the andesites and dacites, it is not unreasonable to assume that the andesites were generated by processes similar to those that generated the alpine plutons (Stille and Steiger 1991) and interacted with lower crustal Hercynian garnet-bearing metasedimentary rocks or Hercynian igneous rocks (Schaltegger and Brack 2007) during ascent.
The alpine orogen is distinct from other common subduction-related orogens worldwide. Notable features are the inferred existence of narrow ocean basins of about 400-700 km width, slow-spreading heterogeneous oceanic lithosphere with no preserved volcanic arcs (Agard and Handy 2021) and distinct trench sediments relative to those deposited in common subduction zones. However, the preservation of HP-LT rocks such as blueschists and eclogites and the occurrence of common subductionrelated igneous rocks with negative Nb-Ta anomalies and positive Pb anomalies in mantle-normalized multielement diagrams clearly provide a link to common (Pacific-type) subduction zones. The geochronology of the orogenic events is well constrained and involved early subduction starting around ~ 100 Ma. Continental subduction occurred from ~ 75 to 35 Ma where subducted fragments yielded peak burial dates between ~ 60 and 35 Ma. Collision of Europe and Adria started at ~ 34 Ma (Agard and Handy 2021;Berger and Bousquet 2008). These features let Agard and Handy (2021) to conclude that the preservation of the Alpine subduction record is not atypical but reflects slow closure of a short-lived, slow-spreading ocean. The absence of a magmatic arc was explained as a result of insufficient water transfer from the slab to the infertile asthenospheric mantle or slabs too short to trigger large-scale mantle upwelling. Nevertheless, typical arc magmas are preserved but they are scarce (McCarthy et al. 2020;Müntener et al. 2021) and confined to the period between ~ 43-40 and ~ 34-30 Ma indicating a short-lived igneous history (McCarthy et al. 2018;. The early products have a clear upper mantle signature whereas the late products are more evolved implying melting of ancient crustal material or assimilation of continental crust (Müntener et al. 2021). Medium to high-K rocks dominate the younger spectrum as a result of episodic fluid fluxes late in the orogenic history. Elevated 87 Sr/ 86 Sr and high LILE of these late volcanic products are explained by partial melting of a metapelitic sediment. These high-K magmas were generated during the initial stages of continental collision between ~ 34 and 30 Ma which is in full accordance with the Ar-Ar data presented here. The inferred presence of appreciable amounts of fluids would then come from hydrated subduction mantle rocks (serpentinites; McCarthy et al. 2018;2020) to prime the mantle wedge and this is in accordance with the trace element signatures of the andesites and dacites studied here, notably their moderate high Ba/Nb and Ba/Th ratios. The presence of melts from sedimentary rocks played also a crucial role in shaping the trace element composition, notably elevated Th/Nb, Th/La and Th/Yb and was, at least partly, also responsible for the evolved isotope composition of the andesites and dacites, in particular with respect to radiogenic Pb isotope signatures. Therefore, this study again emphasizes the view that generation of high-K magmas requires a component of partial melt from subducted sediments (Hermann and Spandler 2008;Müntener et al. 2021). In addition, AFC processes contributed to the evolved isotope compositions and contributed to elevated K 2 O and evolved Sr and Nd isotope compositions (Callegari et al. 2004;Owen 2008;Müntener et al. 2021). To solve the apparent problems in explaining the unusual subduction features of the Alps, McCarthy et al. (2020) renewed the process of an A-type (Ampferer-type) subduction. Here, continental subduction involves closure of small continental basins, subduction initiation without magmatism, minor involvement of oceanic crust and subduction of dry lithosphere into the convective mantle. This elegant model can also explain the crustal signature (radiogenic 87 Sr/ 86 Sr, high LILE, low HFSE, high δ 18 O, radiogenic Pb isotopes) of the late orogenic andesites studied here. In summary, to reconcile the apparent contradiction of typical arc lavas with negative Nb-Ta anomalies but crustal-like isotope features, continental arc magmatism in the Alps originated from limited flux-melting of a metasomatized mantle wedge that underwent specific enrichments during aqueous fluid and sediment-derived melt additions during Ampferer-type subduction. Primary magmas, some of them with a high-K nature developed by multiple FC and AFC processes (i.e., Hürlimann et al. 2016) within different levels of an arc crust towards high-K andesites and dacites. Most likely, Hercynian lower crustal rocks or Permian igneous rocks with radiogenic Sr and Pb but unradiogenic Nd and Hf isotope compositions played a key role during late-stage AFC processes.
Acknowledgements W. S. Vogler (Universität Marburg) initiated this study in 1994 and C. Jung (Universität Marburg) performed the major element analyses. Unfortunately, both decided not to work further on this project. A. W. Hofmann (Max-Planck-Institut für Chemie, Mainz) is thanked for hospitality and for giving access to the mass spectrometry facilities. We appreciate the detailed comments of an anonymous reviewer and Anders McCarthy who helped to focus on the paper. We especially thank Anders McCarthy to bring unrecognized literature to our attention. The editor Hans Keppler is thanked for patient editorial handling of the manuscript.
Funding Open Access funding enabled and organized by Projekt DEAL.

Data availability
The dataset generated and analyzed during the current study are available from the corresponding author on request.

Conflict of interest
The authors declare they have no financial interests. The authors have no competing interests to declare that are relevant to the content of this article.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.