Thermochronology of the Adamello Intrusive Suite, N. Italy: Monitoring the Growth and Decay of an Incrementally Assembled Magmatic System

The Adamello intrusive suite is a composite batholith in Northern Italy, with an estimated 2000 km volume, assembled incrementally over a time span of 10 to 12 million years. The history of crystallization has been studied in detail through laser ablation ICP-MS and SIMS U–Pb geochronology of zircon, which records prolonged crystallization of each of the different intrusive units at mid-crustal levels between 43 47 and 33 16 Ma. The magmas were episodically extracted from this storage area and ascended to the final intrusion level at 6 km paleo-depth. Each batch of melt cooled very rapidly down to the ambient temperature of 250 C, evidenced by distinct cooling paths recorded by amphibole, biotite and K-feldspar Ar/Ar dates. The magma source area was moving from SW to NE with time, causing increasing thermal maturity in the mid-crustal reservoir. The resulting temporal trend of higher degrees of crustal assimiliation in the course of the evolution of the magmatic system can be traced through Hf and O isotopes in zircon. Rough estimates of magma emplacement rates (‘magma flux’) yield very low values in the range of 10 km/yr, typical of mid-to-upper crustal plutons and increase with time. Although we cannot discern a decrease of magma flux from our own data, we anticipate that a dramatic decrease of magma flux between 33 and 31 Ma along the northern contact lead to cessation of magma emplacement.


INTRODUCTION
Emplacement and construction of voluminous siliceous plutons has been a controversial issue for many decades, mainly for two reasons: (i) the origin of voluminous granites ('the source problem') and (ii) how they are emplaced ('the space problem').Models range from mostly partial melting of crust (Ducea & Barton, 2007) to mostly fractionation from basaltic sources (e.g.Sisson et al., 2005) and emplacement mechanisms range from kilometre-scale diapiric rise to incremental emplacement of small magma batches (e.g.Paterson & Fowler, 1993;Coleman et al., 2004;Vigneresse, 2004).Increasingly precise and accurate zircon dating, however, has demonstrated that plutons are accreted by addition of small magma batches to the middle to upper crust over 10 4 to 10 5 year timescales (e.g.Coleman et al., 2004;Matzel et al., 2006;Broderick et al., 2015;Samperton et al., 2015).Eventually, a system of nested plutons will form igneous intrusive suites, or accumulate to form batholiths, during a long-lasting process, which may take 10 6 -10 7 years (Coleman et al., 2004).The integrated magma flux in a magmatic system is defined as the product of magma volumes over time and has been shown to be non-monotonic.Pulses of high magma flux can be followed by periods of relative quiescence (de Saint Blanquat et al., 2011) and magmatic flare-up periods have been proposed to explain the non-linear growth of large-scale batholiths (e.g.Ducea & Barton, 2007;Martinez et al., 2018).
The spatial relationships of individual magma batches at the present-day erosional level, however, may not reliably reflect the temporal relations at emplacement level; rapidly following magma batches of differing viscosity (crystal content, chemical composition, or water content) may develop sharp cross-cutting relationships, whereas larger volumes of compositionally very similar magma may have transitional contacts that are not easily recognized in the outcrop.
Applying U-Pb geochronology to zircon is a relatively straightforward way to reconstruct the timescales of magmatic activity on the pluton, igneous intrusive suite, or batholith scale.Overwhelming evidence has been presented that zircon may not necessarily reflect the spatial relationships at outcrop level in the upper crust, but is crystallizing at different times and levels of the plumbing system (e.g.Schoene et al., 2012;Chelle-Michou et al., 2014;Cooper & Kent, 2014;Barboni et al., 2015;Broderick et al., 2015).This implies that previously crystallized zircon is transported as a crystal cargo and mixed between magma batches.These processes operate at timescales of 10 4 to 10 5 years and thus to be temporally resolved, require age determinations using high-precision chemical abrasion-isotope dilution-thermal ionization mass spectrometry (CA-ID-TIMS) (e.g.Schoene et al., 2012;Barboni & Schoene, 2014;Broderick et al., 2015;Tapster et al., 2016).
Here, we present an extensive dataset for zircon U-Pb crystallization ages, measured by laser-ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) and secondary ion mass spectrometry (SIMS), combined with Hf, O isotopic and chemical information on zircon and amphibole, biotite and K-feldspar 40 Ar-39 Ar thermochronological dates from the entire Adamello Igneous Intrusive Suite, northern Italy.
The Adamello igneous intrusive suite offers the possibility to study the evolution of a c.2000 km 3 sized magmatic system over $12 Myr.The main objectives of this study are to: 1. Use an extensive zircon dataset to monitor the temporal intrusion sequence and its isotopic variation and explain the spatial relationships of recognized intrusive units (traditionally called 'super-units'; Callegari & Brack, 2002); 2. document the development of a crustal scale thermal anomaly leading to increased crustal assimilation in the magmas and eventually cessation of magmatic activity; and 3. quantify emplacement rates of Adamello intrusions through time using different modelling techniques.

REGIONAL GEOLOGICAL OUTLINE
The Adamello intrusive suite, located in northern Italy (Fig. 1) , is the largest and oldest Tertiary Periadriatic intrusion in the Alps and reflects the amalgamation of petrographically and isotopically distinctive plutons.The 675 km 2 of rocks exposed are composed 99% of tonalite, quartz-diorite and granodiorite (Callegari & Dal Piaz, 1973;Dupuy et al., 1982;Callegari & Brack, 2002;Brack et al., 2008), the remaining 1% being represented by more mafic rocks such as gabbro and hornblendite.The Southern Adamello (Re di Castello unit, see below) is texturally very diverse, whereas the central and northern parts are formed from compositionally more homogeneous tonalites and leucotonalites.The calc-alkaline magmas intruded into Variscan metamorphic basement that varies between greenschist to amphibolite facies and unmetamorphosed Permian-Triassic cover series (e.g.phyllites, limestones and dolomites) over 10 Myr, from middle Eocene to early Oligocene (Del Moro et al., 1983b).
The melt emplacement generated a contact aureole easily observable in the field along 90% of the igneous intrusive suite border (Callegari & Brack, 2002).The Adamello batholith postdates the subduction beneath and collision of the European plate with the African plate.The Brianc¸onnais exotic terrane and part of the Iberia plate were accreted to the Apulian plate around 45 Ma (Stampfli et al., 1998) and slab breakoff was postulated to have occurred at $30 Ma (Davies & Von Blanckenburg, 1995).The Adamello batholith is pinched between two lineaments (Fig. 1); the dextral Tonale Line to the north (Werling, 1991), which is part of the Periadriatic (or Insubric) lineament and to the southeast, the Giudicarie Line with sinistral transpressive displacement (Viola et al., 2001;Pennacchioni et al., 2006).Post-intrusive shearing along these two structures gave the Adamello igneous intrusive suite its current geometry (Castellarin & Sartori, 1983;John & Blundy, 1993;Werling, 1991).Following Stipp et al., (2004), emplacement of the entire Adamello igneous intrusive suite was unrelated to tectonic movements along this simultaneous fault system, nor to dextral shearing along the Tonale fault; this is in disagreement with Rosenberg (2004)

LITHOLOGICAL CONTENT OF THE ADAMELLO IGNEOUS INTRUSIVE SUITE
The Adamello igneous intrusive suite is an excellent and well-exposed example of a complex igneous rock suite in which distinct petrographic units have been mapped on the basis of their mineralogical composition and texture (Bianchi et al., 1970), traditionally called 'super-units' by Callegari & Brack (2002).Plutonic rocks are exposed over a vertical distance of 2-3 km.The temporal emplacement sequence trending from south to north-east has previously been established based on isotopic dating (Del Moro et al., 1983a, 1983b;Villa, 1983;Hansmann & Oberli, 1991;Kagami et al., 1991;Mayer et al., 2003).In the present study, we have subdivided the Adamello igneous intrusive suite into 7 units (see Fig. 1), i.e. the Corno Alto, Re di Castello south and north, West and Central Adamello, Avio and Presanella super-units, slightly deviating from the originally proposed scheme of 4 units (Callegari & Brack, 2002).New elements in this classification are the Corno Alto unit at the eastern border, previously included as a part of the Re di Castello super-unit and the Central Adamello leucotonalite unit that has been considered as a part of the West Adamello super-unit by Callegari & Brack (2002).They are defined on the basis of the new geochronology data presented below.The fact that some of the super-units are composed of multiple intrusions has already been addressed by Callegari & Dal Piaz (1973) and Brack (1983).Schaltegger et al., 2009).The distinction into super-units is modified from Callegari & Brack (2002).The further subdivision of the Re di Castello S unit is not discussed in this study and refers to the previous geochronological studies of Schaltegger et al. (2009), Schoene et al. (2012) and Broderick et al. (2015).

The Corno Alto superunit (CA)
The Corno Alto super-unit is a rounded body in contact with the Presanella unit in the north and intruded into low-grade Variscan basement.It mainly consists of partially cataclastic and altered granodiorite to trondhjemite, with porphyritic plagioclase and variable amounts of quartz and K-feldspar (Callegari & Dal Piaz, 1973).Biotite and locally muscovite are primary magmatic phases.Based on similar ages and initial 87 Sr/ 86 Sr ratios, the CA unit was thought to be cogenetic with the Re di Castello superunit (RdC) (e.g.Del Moro et al., 1983a;Mayer et al., 2003).

The Re di Castello super-unit (RdC)
The Re di Castello super-unit is subdivided into a southern and a northern part that differ largely in lithology and diversity, as well as age.The southern RdC has received most attention and contains a suite of rocks ranging from ultramafic cumulates, gabbro and diorite, to granodiorite, tonalite and leucotonalite that were intruded between 42Á5 and 41Á7 Ma (Ulmer et al., 1983;Hansmann & Oberli, 1991;Schaltegger et al., 2009;Schoene et al., 2012;Broderick et al., 2015).Primitive high-Mg basaltic to andesitic dykes cut the entire southern Re di Castello at 42Á68 Ma (Hu ¨rlimann et al., 2016).The northern part is comprised of a medium-grained, amphibole and biotite-bearing tonalite to granodiorite and small volumes of granite emplaced at 39Á8 Ma (Hansmann & Oberli, 1991) that include abundant mafic enclaves and are crosscut by dacitic dikes at 38Á62 Ma (Hu ¨rlimann et al., 2016) and by numerous aplitic dykes.

The Western and Central Adamello super-unit
The Western and Central Adamello super-unit contains two different types of tonalite; the Western Adamello tonalite (WAT) and the Central Adamello leucotonalite (CAL).The WAT is a coarse-grained, quite homogeneous, tonalite intrusion with up to 1 cm euhedral amphibole and columnar biotite, which is a distinguishing characteristic used to differentiate the WAT from the northern RdC.From the contact in the southwest to the northeast the unit becomes more leucocratic, with decreasing amphibole and increasing biotite abundance.Sequential emplacement of chemically and isotopically distinct magma batches over 1Á2 Myr has previously been demonstrated by Flo ¨ss (2013).The southern contact of the WAT unit towards the older RdC-N unit is in some places characterized by a tonalitic, schlieren-rich zone with conspicuous enrichment of stubby amphibole crystals (typical of WAT; Callegari & Dal Piaz, 1973).The Central Adamello leucotonalite (CAL) has a very similar texture and grain size compared to the WAT super-unit, but a distinctly lower abundance of amphibole (1 vs 20 vol.%) that serves as a distinctive feature.The contact zone between the two units is transitional over tens of meters, characterizing them as two texturally and chemically different intrusive units that are broadly synchronous or follow each other closely in time.

The Avio superunit (Avio)
The Avio superunit is comprised of fine-grained quartzdiorites and biotite-bearing tonalites that are devoid of amphibole except for a thin contact zone towards the younger Presanella unit.The Avio rocks are progressively foliated towards the northern contact.According to Stipp et al. (2004), a magmatic foliation due to late ballooning of the Avio intrusion, can be distinguished from a syn-to-post-emplacement dextral shearing along the Tonale fault east of Passo Tonale, where the foliation crosscuts the contact and there is shearing and telescoping of the contact aureole (Fig. 1).

The Presanella super-unit (Pres)
The Presanella super-unit contains mainly mediumgrained tonalite, with biotite and both anhedral and euhedral amphibole showing a preferred orientation.The tonalites contain numerous mafic enclaves and are crosscut by muscovite-bearing aplitic dykes.Amphibole-rich schlieren and textural changes within the Presanella tonalite have been observed and described from the contact south of Passo Tonale, where biotite-bearing quartz-diorites of the Avio are in contact with tonalite of the Presanella unit, respectively (Fig. 1; Callegari & Dal Piaz, 1973).The lobate shape of the contacts and the degree of interaction between the two types of tonalite suggest magma mingling between members of the two units and thus imply a small age difference (Bianchi et al., 1970;Callegari & Brack, 2002).The Presanella unit has been affected by syn-to post emplacement deformation along the Insubric line (e.g.Stipp et al., 2004;Grujic et al., 2011).

ANALYTICAL TECHNIQUES Mineral separation and sample preparation
All samples were crushed to a grain size <300 mm and zircon crystals were separated according to shape, density and magnetic susceptibility using a Wilfley wet shaking table, a Frantz magnetic separator and methylene iodide with a density of 3Á32.Some samples were also processed in a bromoform solution with a density of 2. A second magnetic separation with 0 side and 10 alongside slope was performed using a Frantz isodynamic magnetic separator to recover diamagnetic zircon and to avoid grains containing cores or impurities.Zircon grains were selected using a binocular microscope and between 20-23 grains per sample mounted in epoxy resin and polished to expose the crystal interiors, imaged using scanning electron microscopy (SEM based cathodoluminescence; CL) at the universities of Lausanne, Geneva and St. Johns and the Swedish Museum of Natural History in Stockholm.Additional CL images were acquired using a PATCO ELM-3 Luminoscope mounted on an Olympus BX50WI petrographic microscope; digital images were collected using a KAPPA thermal-electrically cooled DX-30C camera.CL images were used for careful selection of the spot locations for all subsequent in situ analyses.

U-Pb age determinations on zircon by LA-ICP-MS
A major portion of the U-Pb analyses were carried out on a sector-field, single-collector Element 2 XR ICP-MS at the University of Lausanne in laser ablation mode.

206
Pb/ 238 U and 207 Pb/ 235 U dates were acquired using a 193-nm excimer ablation system UP-193FX (ESI), using the analytical conditions described in Ulianov et al. (2012).A laser spot diameter of 35 mm, a low energy density of 2Á2-2Á3 J/cm 2 and 5 Hz pulse frequency ensured conditions that minimize isotopic fractionation.The instrument mass bias was controlled through 424 measurements of the GJ-1 reference zircon (CA-ID-TIMS 206 Pb/ 238 U age of 600Á5 6 0Á4 Ma; Boekhout et al., 2012) during the course of the analyses, which yielded a mean 206 Pb/ 238 U age of 600Á17 6 0Á29 Ma (MSWD ¼ 1Á02; Supplementary Data Fig.S1a; supplementary data are available for downloading at http://www.petrology.oxfordjournals.org).The conspicuous drop in the uncertainty of individual analyses was due to a change from a tear-shaped small-volume ablation cell to a standard UP213/193 cell with a higher volume (Supplementary Data Fig.S1a).Secondary reference zircons were repeatedly analysed to assess the accuracy of the mass bias correction: 15 analyses of 91500 zircon (Wiedenbeck et al., 1995)

U-Pb age determinations on zircon by SIMS
Part of the U-Pb age determinations were performed at the Swedish Museum of Natural History, Stockholm, using a Cameca IMS1280 ion microprobe (NordSIM facility).U-Pb analyses followed routine protocols outlined by Whitehouse et al., (1999) and Whitehouse & Kamber (2005).A c.2 nA, -13 kV O À 2 primary beam (imaged aperture of 100 mm corresponding to a spot diameter on the sample of c.10 mm) was used to generate þ10 kV secondary ions, which were admitted to the mass spectrometer and detected in a peak-hopping sequence using a single ion-counting electron multiplier.The mass spectrometer was operated at a mass resolution (M/DM) of 5400, sufficient to separate all species of interest from molecular interferences.Each analysis comprised a 90 second pre-sputter to remove the Aucoating and allow the secondary beam to stabilize, centering of the secondary beam in the field aperture, energy optimization in the 45 eV energy window, mass calibration adjustment using the 90 Zr 16  2 O peak and 16 cycles through the masses of interest.Groups of analyses were performed in fully automated sequences, regularly interspersing standard analyses with those of the sample zircon grains.Data reduction utilized an inhouse developed suite of software.Lead isotope ratios were corrected for common Pb estimated from measured 204 Pb assuming the present-day terrestrial Pb isotope composition estimated from the model of Stacey & Kramers (1975).U/Pb ratios were calibrated using an empirical power-law relationship between 206 Pb/ 238 U and 238 U 16 O 2 / 238 U, assuming the 1065 Ma age of the 91500 zircon (Wiedenbeck et al., 1995).

Lu-Hf isotope composition of zircon grains
Zircon grains from 21 samples were analysed for Lu-Hf isotopes using a Thermo-Scientific Neptune multicollector ICP-MS at Goethe-University Frankfurt (GUF; Germany) coupled to a RESOlution M50 193 nm ArF Excimer (Resonetics) laser system (Gerdes & Zeh, 2006, 2009).For one Hf isotope analysis, zircons were ablated for 60 seconds using a spot diameter of 40 mm and energy density of 5 J/cm 2 .The isotopes 172 Yb, 173 Yb and 175 Lu were simultaneously monitored during each analysis step to allow the correction of isobaric interferences 176 Yb and 176 Lu, using 176 Yb/ 173 Yb ¼ 0Á796218 (Chu et al., 2002) and 176 Lu/ 175 Lu of 0Á02658 (GUF inhouse value).Instrumental mass bias for Hf isotopes was corrected using a 179 Hf/ 177 Hf of 0Á7325 and an exponential law.The mass bias behaviour of Lu was assumed to follow that of Yb.The mass bias for the Yb isotope was corrected using the one for Hf for an individual integration step multiplied by a daily bHf/bYb offset factor.The reported uncertainties (2r) for each analysis were estimated by quadratic addition of the reproducibility of JMC 475 ( 176 Hf/ 177 Hf ratio of 0Á282160; 2 SD ¼ 0Á0028%, n ¼ 8) to the within-run precision of each analysis (2 SE).Standard zircons GJ-1 and Ple sovice were analysed to check the accuracy and external reproducibility of the method, which yielded 176 Hf/ 177 Hf of 0Á282008 6 0Á000022 (2 SD, n ¼ 31) and 0Á282479 6 0Á000016 (n ¼ 20), respectively.These results are in agreement with the LA-MC-ICPMS long-term average of GJ-1 (0Á282010 6 0Á000024; n > 1000) and Ple sovice (0Á282478 6 0Á000023, n > 530) reference zircons at GUF.
In situ analyses of Lu-Hf isotopes in zircons from six samples were carried out at the Memorial University of Newfoundland (MUN) Micro Analysis Facility (MAF-IIC; St. John's, Canada) using a ThermoFinnigan Neptune MC-ICP-MS interfaced to a Geolas Pro 193 nm Ar-F excimer laser.The diameter was 49 lm, the repetition rate 10 Hz and the energy density 5 J/cm 2 .The first 30 s of each analysis were used to determine instrumental background counts, followed by 60 s of laser ablation sampling.The data reduction scheme is described in detail by Fisher et al. (2011) ¼ 121), 0Á282171 6 0Á000040 (2 SD, n ¼ 41) and  0Á282302 6 0Á000052 (2 SD, n ¼ 229), respectively.All of the Hf analyses were done on the same locations or close to the previously measured U-Pb LA-ICP-MS/SIMS analysis spots whenever possible.Initial ratios are calculated via a decay constant value of 1Á867 Â 10 -11 a -1 (Scherer et al., 2001), 176 Hf/ 177 Hf CHUR, today and 176 Lu/ 177 Hf CHUR, today of 0Á282785 and 0Á0336, respectively (Bouvier et al., 2008).

Hf isotopic composition of whole-rock samples
Whole-rock Lu-Hf isotope analyses were done for the same six samples that were analysed at MUN for their zircon Lu-Hf isotope compositions at the Radiogenic Isotope and Geochronology Laboratory (RIGL) at Washington State University using a ThermoFinnigan Neptune MC-ICP-MS.These analyses follow the procedures described in detail by Vervoort et al. (2004).

Ar-39 Ar age determinations
Fresh and unaltered amphibole and biotite were handpicked using a binocular microscope.Amphibole grains were leached with 5% HNO 3 (aq) in an ultra-sonic bath for several minutes to remove other mineral phases and alteration from the rims.K-feldspar was separated from plagioclase using a density separation in sodium polytungstate (SPT) solution with a density of 2.58 g/ cm 3 and a centrifuge.Between 0.006-0.09g of handpicked, clean mineral separates were packed in a copper foil and placed in a silica tube that was irradiated for 9 hours in the CLICIT facility of the TRIGA reactor at the Oregon State University.J-values were calculated by the irradiation of Fish Canyon Tuff sanidine, assuming an age of 28.201 6 0.046 Ma (Kuiper et al., 2008).All flux monitors were separated by a distance of <1cm in a silica tube and J-values for samples were calculated by linear interpolation.All sample data including the J-values are reported in Supplementary Data Table S5.Argon isotope measurements were carried out at the University of Geneva.Samples were degassed by CO 2 -IR laser step-heating and the extracted gas was gettered (3 x ST101 getters) in a stainless steel UHV line, after passing through a cold trap chilled to $150 Kelvin.Argon isotopes were analysed using a GV Instruments Argus V mass spectrometer equipped with four highgain (10 12 X) Faraday detectors and a single 10 11 X Faraday detector ( 40 Ar).Ages were calculated using the ArArCalc software of Koppers (2002).Corrections were made for mass discrimination, isotopic decay of 39 Ar and 37 Ar and for interfering nucleogenic Ca-, K-and Clderived isotopes.Errors on J-value and mass discrimination are propagated into the final results.Blanks were calculated before every new sample and after every three heating steps.Plateaus must satisfy the definition of Dalrymple et al. (1974) and include >50% of the released 39 Ar in 3 or more contiguous steps, yielding an MSWD < 2.5.A weighted mean age is reported for specific regions of an age spectrum where the plateau definition is not met.

Oxygen stable isotopes analysis
Ten zircon fractions of 1.5-2.5 mg weight were analysed using laser fluorination procedures at the University of Oregon at Eugene, USA (Bindeman, 2008).After laser extraction, the oxygen was purified by a series of cryogenic traps and then passed through a boiling mercury diffusion pump, then converted to CO2 gas prior to analysis with a Thermo MAT 253 mass spectrometer and normalized to the Gore Mountain Garnet (d 18 O ¼ 5.8%) and San Carlos Olivine (SCO, d 18 O ¼ 5.25%) standards.Based on repeated analyses of standards, the analytical uncertainties on d 18 O measurements are <0.1&.

Trace elements in zircons
The same Element XR single-collector (LA-SF-ICP-MS) and 193-nm excimer ablation system UP-193FX (ESI) that was used for the U-Pb dating at the University of Lausanne was also used to analyse trace elements in zircon, using an identical spot diameter of 35 mm A SRM 612 glass was used for standardization.The crystallization temperature of zircon was estimated by applying the Ti-in-zircon thermometer (Ferry & Watson, 2007), using Ti and SiO 2 activities of 0.5 and 1, respectively.The fO 2 was calculated using the Ti-in zircon temperature and the Ce anomaly in the zircons (Trail et al., 2011).

U-Pb age determinations
Mean 206 Pb/ 238 U ages of zircon from 27 samples from all seven super-units were calculated spanning the entire Adamello intrusive suite (Table 1; Fig. 2), representing 10 million years of magmatic activity.The full data set is presented in Supplementary Data Tables S1 (LA-ICP-MS analyses) and S2 (SIMS analyses).

Corno Alto (CA)
Zircon from two samples were dated by LA-ICP-MS; sample CA-AS4 returned a mean age of 43Á47 6 0Á16 Ma from 10 analyses (MSWD ¼ 0Á75) after rejection of five older analyses between 44 and 45 Ma.The CL images in Supplementary Data Fig.S2

Re di Castello South (RDC-S)
The RdC3A tonalite of the Lago della Vacca suite was dated by SIMS and yielded an age of 42Á26 6 0Á21 Ma (N ¼ 43; MSWD ¼ 1Á02), in good agreement with the CA-ID-TIMS age from the same sample at 41Á89 6 0Á02 Ma (youngest zircon date from Schoene et al., 2012).Published CA-ID-TIMS data of Schaltegger et al. (2009) and Schoene et al. (2012) from the Lago della Vacca complex and from Broderick et al. (2015) from the Val Fredda complex (Fig. 2), overlap with this age and, together with the new data presented here, provide evidence for c.2Á3 Myr of magmatic activity in the southermost part of the Adamello intrusive suite.A second sample (Bruff2) was collected from the southeastern tip of the Adamello intrusive suite and has a slightly younger age of 40Á4 6 0Á2 Ma (SIMS, N ¼ 45; MSWD ¼ 1Á20).Both samples contain some zircon with cores (see Supplementary Data Fig.S3 for examples of inherited cores in sample RdC3).

Re di Castello North (RDC-N)
U-Pb age determinations were carried out on zircon from four rather homogeneous tonalites and granodiorites collected along a S-N profile spanning the entire unit.The results from three samples dated by LA-ICP-MS yield statistically equivalent 206 Pb/ 238 U ages of 39Á71 6 0Á44 Ma (sample RDC_AS1; N ¼ 6, MSWD ¼ 0Á42), 39Á57 6 0Á19 Ma (RDC_AS6; N ¼ 20; MSWD ¼ 1Á6) and 39Á41 6 0Á27 Ma (RDC_AS9; N ¼ 14; MSWD ¼ 1Á6).An identical age of 39Á1 6 0Á3 Ma (N ¼ 10; MSWD ¼ 1Á15) was obtained by SIMS from sample PB18.Thus, the transect did not reveal any indication of sequential emplacement.A single date from RDC_AS9 and two from PB18 were slightly older and excluded from the mean age calculation.

Western Adamello tonalite (WAT)
Zircons from seven samples were analysed and yielded mean 206 Pb/ 238 U ages scattering over 1Á5 Ma along a cryptic trend, in which younger ages were found towards the north and the west of the intrusive unit (Fig. 1).However, it should be noted that the youngest sample AD_AS6 was collected from the center of the unit.Sample AD_AS2 has a homogeneous zircon population with growth zoning revealed in CL (Supplementary Data Fig.S2 c1 to c3) with a mean age of 38Á23 6 0Á24 Ma (N ¼ 13; MSWD ¼ 1Á6), followed in decreasing order by AD_AS10 (37Á90 6 0Á34 Ma; N ¼ 8; MSWD ¼ 1Á2); AD_AS11 (37Á77 6 0Á20 Ma; N ¼ 18; MSWD ¼ 1Á7), AD_AS15 (37Á65 6 0Á22 Ma; N ¼ 15; MSWD ¼ 0Á90), all measured by LA-ICP-MS.An age of 37Á4 6 0Á3 Ma (N ¼ 13; MSWD ¼ 1Á7) was obtained by SIMS for sample PB25.The youngest ages are represented by PB1002 (37Á20 6 0Á43 Ma; N ¼ 13; MSWD ¼ 1Á8) and AD_AS6 (36Á93 6 0Á19 Ma; N ¼ 19; MSWD ¼ 0Á82).The overall 1Á2 Ma age scatter observed in the LA-ICP-MS and SIMS data is in perfect agreement with high-precision CA-ID-TIMS data that provide evidence for sequential south to north emplacement of chemically and isotopically distinct magma batches between 37Á62 6 0Á04 Ma and 36Á42 6 0Á03 Ma (Flo ¨ss, 2013).Some of the data sets from individual samples are characterised by excess scatter with slightly elevated MSWD values (such as, e.g.PB25 and PB1002) and thus require exclusion of some distinctly older analyses from the calculation of the mean.This scatter is partly due to slightly older, sector-zoned cores (Supplementary Data Fig.S2 c12), possibly inherited from zircon growth at deeper crustal levels and partly analytical.

Central Adamello leucotonalite (CAL)
A single leucotonalite sample was dated from this unit, yielding a LA-ICP-MS U-Pb age of 35Á94 6 0Á15 Ma (N ¼ 28, MSWD ¼ 1Á6).The age gap of 1 Ma between the youngest WAD date (AD_AS6 from the center of the WAD body) and the CAL date suggests a distinct hiatus between the intrusion of these two lithologies.This is, however, in disagreement with the transitional contact observed in the field and possibly due to the fact that both WAD and CAL samples were taken well away from the contact.
Pre-intrusive age components.The extensive U-Pb data set incorporates some analyses that document the assimilation of crustal material with distinctly older age components compared to the timing of magmatic activity.The first detailed geochronological study (multigrain ID-TIMS) by Hansmann & Oberli (1991) was strongly biased by the effect of inheritance (combined with unresolved lead loss), most obvious in samples RdC3 and Bruff2 (tonalites from RdC south).Our data corroborate this finding for the latter sample, but only one analysis with an old component was registered from sample RdC3.Several zircon analyses show the clear presence of an inherited age component from samples of the CA, Avio and Presanella units.Hf isotopic compositions (discussed below) of these inherited cores and xenocrystic grains are not coherent and, therefore, suggest the presence of a multitude of sources.

Hf isotopic analyses of zircon
The Hf isotope compositions of zircon exhibit a distinct trend with age (Fig. 3 and Supplementary Data Fig.S2; Table S3).Although the CA intrusives contain the oldest magmatic zircon crystals of the entire Adamello intrusive suite, their e Hfi values of up to þ7 are not the most juvenile.Tonalites, diorites and gabbros of the Val Fredda and Lago della Vacca units from the southernmost part of the Adamello suite (Schoene et al., 2012;Broderick et al., 2015), as well as a tonalite from the Listino unit (Schaltegger et al., 2009), show higher values between e Hfi þ8 to þ12.The Hf isotopic compositions of zircon from the southern Re di Castello (Vacca and Bruffione intrusions), as well as of the northern Re di Castello tonalites and granodiorites mostly scatter around values of þ3 to -5.Slightly lower e Hfi values are recorded by the WAT and the CAL units; most values vary between -1Á4 and -7Á6.The Avio tonalite reaches the lowest e Hf values of the entire Adamello intrusive suite, with a minimum of -11.Finally, analyses from zircon of the youngest Presanella unit display a poorly defined trend to slightly higher e Hfi values at -4 to -8.
The truly inherited cores, like the wide range in ages outlined above, also span a remarkably wide range in e Hfi from þ7 to -39Á6.In one sample from the Avio unit, PB30, the range is from þ4 to -39Á6, which indicates inheritance from very different sources, in particular an ancient crustal source for the very lowest e Hfi value (Supplementary Data Fig.S5).

Hf isotopic analyses of whole-rocks
Analyses of six whole-rock samples are included in Supplementary Data Table S4 and are plotted in Fig. 3 (inset numbers 1-6 in grey circles), using the mean U-Pb age from the respective sample.All whole-rock Hf isotopic values plot perfectly inside the variation field of LA-ICP-MS Hf analyses from zircon, corroborating the statistically relevant selection of zircon for Hf isotope analysis Fig. 3; Supplementary Data Table S4).The whole-rock value of the Avio unit (number 5 in Fig. 3) is situated at the lower end of zircon eHf values, which means that this unit likely contains a larger component of older material.

Oxygen isotopes in zircon
Oxygen isotope values from bulk zircon fractions of all super-units were determined by laser fluorination alongside with coexisting quartz and amphibole crystals.D 18 O(Qz-Zrc) were in high temperature equilibrium at 2Á5-3Á6&, corresponding to magmatic closure temperatures of 801-973 C using the Trail et al. (2009) Afactor.The composition of the magma was calculated using zircon and quartz data, as described in Bindeman & Valley (2002).The overall d 18 O range measured in zircon varies between 5Á14 and 7Á.67& (Supplementary Data Table S6), leading to calculated d 18 O values for the magma of 7Á3-9Á7& (Fig. 4).This defines the whole Adamello magmatic suite as high-d 18 O; with the d 18 O of silicic differentiates significantly higher than those typical for a silicic differentiate from mantle-derived basalt (around 6&; e.g.Bindeman, 2008) and characteristic of crustal magmas (Valley et al., 1998)

Trace element analysis of zircon
The REE concentrations in zircon from the central and northern super-units from the Adamello (Supplementary Data Table S7) have characteristic positive Ce and negative Eu anomalies (Fig. 5a).The extent of the Eu anomaly (Eu/Eu*) systematically increases with decreasing age, from the CA to the Avio and Presanella super-units (Fig. 5a-c) with very variable Eu/Eu* values between 0Á39 and 0Á05.The Ce anomaly (Ce/Ce*) does not show any systematic trend with age.The biggest variation in the REE patterns occurs in the middle REE and we, therefore, illustrate the Eu/Eu* variation as a function of Dy/Nd (Fig. 5b).The trend defined by the samples of all super-units except for CA indicate decreasing Eu/Eu* with increasing Dy/Nd, suggesting co-precipitation of a LREE-rich phase with zircon.Although the fields for different super-units overlap, the older Re di Castello displays an overall higher Eu/Eu* (0.55-0.3) than the youngest Avio and Presanella units (Eu/Eu*: 0Á38-0Á05), consistent with the more evolved composition of the latter.The Ce/Ce* ratio and the Ti-inzircon temperature are used to calculate the oxygen fugacity fO 2 (Trail et al., 2011); most data fall between the FMQ and the MH buffers within their uncertainties (Fig. 5c).The Ti-in-zircon temperatures calculated according to Ferry & Watson (2007) range between 630 and 829 C (Supplementary Data Table S7), assuming aTiO 2 of 0.5).Large variations of the Ti-in-zircon temperatures within each unit probably point to variable aTiO 2 among different melts.

40
Ar-39 Ar geochronology Amphibole, biotite and K-feldspar in eight samples from five different super-units were analysed for their 40 Ar-39 Ar ages (Table 2 and Supplementary Data Table S5; Supplementary Data Fig.S4).The Avio super-unit does not contain any amphibole, therefore, only biotite and K-feldspar data are reported.

Re di Castello north
Two samples RDC_AS1 and RDC_AS9 were collected from the Val di Daone, sample RDC_AS1 is from the center of the intrusion, whereas RDC_AS9 was taken at the northern contact towards WAT.The entire nine temperature steps of the RDC_AS1 amphibole (Supplementary Data Fig.S4a) yield a plateau with a weighted mean age of 39Á09 6 0Á35 Ma (MSWD ¼ 1Á75); the inverse isochron is identical within uncertainty (39Á59 6 0Á95 Ma) with a 40 Ar/ 36 Ar intercept at 260Á8 6 66Á3, pointing to a mixture of radiogenic argon with an atmospheric component and excluding any excess radiogenic 40 Ar.Nine out of eleven steps of RDC_AS1 biotite define a weighted plateau age of 38Á96 6 0Á18 Ma (MSWD ¼ 1Á86).The two first steps show younger ages, interpreted by small amounts of secondary chlorite, possibly via loss of radiogenic Ar or from 39 Ar recoil implantation (Lo & Onstott, 1989) into K-free chlorite (Heizler et al., 1988;Sanders et al., 2006).The inverse isochron is in agreement with the plateau age and the 40 Ar/ 36 Ar intercept shows no indication of excess Ar in this sample.The K-feldspar analysis yielded a more disturbed extraction pattern than the other mineral phases: The five initial steps released different amounts of 39 Ar, which could be related to the ubiquitous myrmekite textures observed in thin section, or partly due to 39 Ar recoil (Lo & Onstott, 1989).From the remaining four temperature steps, a weighted plateau age of 36Á55 6 0Á18 Ma (MSWD ¼ 2Á33) was obtained.The inverse isochron of this sample is statistically insignificant.
From 98% of released 39 Ar from the amphibole of sample RDC_AS9, a weighted plateau age of 38Á71 6 0Á47 Ma (MSWD ¼ 2Á04) can be calculated (Supplementary Data Fig.S4b).The inverse isochron age of 39Á03 6 2Á58 is imprecise but concordant within error with the weighted plateau age, with a statistically insignificant 40 Ar/ 36 Ar intercept.RDC_AS9 biotite presents a 'hump-shaped' degassing pattern, pointing to some secondary chlorite present.A weighted plateau age of 36Á37 6 0Á20 Ma (MSWD ¼ 1Á76) can be calculated from seven steps, which is in agreement with the inverse isochron, with an age of 36Á30 6 0Á42 Ma and a 40 Ar/ 36 Ar intercept of 312Á2 6 84Á6, excluding significant amounts of excess radiogenic 40 Ar.Due to the slightly more mafic composition of the RDC_AS9 sample compared to RDC_AS1, there was not enough K-feldspar in the rock to be separated and analysed.

Western Adamello tonalite
Sample AD_AS6 is from the Val di Fumo, roughly situated in the center of the intrusion (Fig. 1).The analyses  S4c).

Central Adamello leucotonalite
Sample AD_AS5 was taken from within the leucotonalite, at the upper end of the Val di Fumo, in the center of the CAL super-unit (Fig. 1).Approximately 45% of released 39 Ar by four temperature steps defines a weighted mean age of 36Á53 6 0Á35 Ma (MSWD ¼ 1Á01) for amphibole (Supplementary Data Fig.S4d).The inverse isochron gives an age of 36Á94 6 2Á60 Ma, concordant with the weighted mean age and with an intercept indicating atmospheric argon.The large biotite crystals separated from the CAL yielded a weighted plateau age of 34Á17 6 0Á17 Ma (MSWD ¼ 2Á24) from seven out of ten temperature steps.The inverse isochron age is concordant within the error with the plateau age, its 40 Ar/ 36 Ar intercept is slightly lower than the atmospheric ratio, there is no indication of excess of argon in this sample.The K-feldspar yielded a weighted plateau age of 32Á11 6 0Á15 Ma (MSWD ¼ 1.62), concordant with the inverse isochron age of 31Á74 6 0Á90 Ma within uncertainties, without any indication for excess Ar.

Avio
Sample PB775 sample was taken from the centre of the current surface of the intrusion, whereas PB780 was situated at the western contact of the intrusive rocks, very close to the surrounding basement (Fig. 1).PB775 biotite gives a plateau with an age of 33Á49 6 0Á16 Ma (MWSD ¼ 1Á69) based on eight steps out of ten with 80% 39 Ar released (Supplementary Data Fig.S4e).The two younger steps may be due to traces of recognizable chloritisation.
The inverse isochron age is 33Á45 6 0Á36 Ma (MSWD ¼ 1Á94) and its 40 Ar/ 36 Ar intercept is 309Á0 6 89Á8.The K-feldspar age spectrum is considered as a plateau of 31Á50 6 0Á17 Ma age (six out of nine temperature steps), although the MSWD of 2Á54 is slightly high.The first older step reflects Ar excess, which could be due to 39 Ar recoil.The inverse isochron age of 31Á95 6 0Á63 Ma (MSWD ¼ 2Á13) is in agreement with the plateau age, the 40 Ar/ 36 Ar intercept is 256Á1 6 53Á2.The plateau age from PB780 biotite is 35Á11 6 0Á17 Ma (MSWD ¼ 2Á20), calculated from eight out of ten steps, integrating 85% 39 Ar released.The first two younger steps may indicate chlorite alteration.The inverse isochron has an age of 35Á30 6 0Á48 Ma (MSWD ¼ 2Á37) and has an intercept of 259Á7 6 90Á9.

Presanella
Two different tonalites from the Presanella super-unit were analysed: sample PR_AS1 located in the south of the super-unit and PB782 from the northern part of the batholith (Fig. 1).The PR_AS1 amphibole age spectrum shows an older first step possibly due to the presence of excess Ar and another nine steps with significant excess scatter.The biotite from the same sample has a weighted plateau age of 31Á37 6 0Á20 Ma calculated from six of ten steps and 58% 39 Ar released and with a MWSD of 1Á89.As for the other biotite from all superunits, some degree of chloritization is indicated by the lowest temperature step.The inverse isochron is concordant with the weighted plateau age (31Á64 6 0Á23 Ma; MSWD ¼ 0Á29) with an 40 Ar/ 39 Ar intercept at 262Á6 6 20Á6.Excess scatter of the K-feldspar from PR_AS1 may be explained by 39 Ar recoil.The age spectrum for the PB782 amphibole (Supplementary Data Fig.S4g) yields a weighted plateau age of 35Á42 6 0Á41 Ma (MWSD ¼ 0Á91), with a concordant inverse isochron age of 34Á52 6 0Á91 Ma (MWSD ¼ 0Á13).The biotite for the same sample gives an age of 31Á33 6 0Á14 Ma (MSWD ¼ 1Á41) and is considered as a plateau with almost 50% 39 Ar released over three steps.The inverse isochron age is 31Á49 6 0Á27 Ma (MSWD ¼ 0Á97), which is in agreement with the plateau age within error.Finally, the K-feldspar has a staircase-shaped age spectrum with a plateau age of 27Á52 6 0Á17 Ma and a MSWD of 0Á54 from five out of nine steps integrating 68% 39 Ar released.The inverse isochron age is 27Á70 6 0Á75 Ma (MSWD ¼ 0Á70) and is concordant with the weighted plateau age.

Zircon records 10-12 Myr of melt batch assembly and crystallization
It is generally accepted that large mid-to-upper crustal plutons are incrementally assembled through accretion of magma batches over time (e.g.Coleman et al., 2004;Chelle-Michou et al., 2014;Barboni et al., 2015;Broderick et al., 2015;Samperton et al., 2015).Zircon data from this study document 10 Myr of magmatic activity based on zircon crystallization ages from 43Á47 6 0Á16 Ma to 33Á16 6 0Á68 (Fig. 2).This age range is extended to 30Á73 6 0Á29 Ma by the data of Grujic et al. (2011) from the northernmost rim of the Presanella unit.The CA unit has been identified as the oldest magmatic unit.It is followed by the highly diverse RDC-S unit, which was emplaced between 42Á7 and 41Á8 Ma and includes the juvenile mafic gabbros and diorites of Mattoni and Cadino (Broderick et al., 2015) and the Blumone gabbro (Schoene et al., 2012) and abundant tonalites in both the Val Fredda and Lago della Vacca Complexes (Ulmer et al., 1983).Due to its lithological diversity, abundant outcrops highlighting intrusive relationships and its accessibility, this southernmost part of the Adamello Intrusive Suite has received extensive attention and is over-represented in terms of petrological and geochemical as well as geochronological data (e.g.Hansmann & Oberli, 1991;Schaltegger et al., 2009;Schoene et al., 2012;Broderick et al., 2015).The less accessible central and northern super-units consist mainly of amphibole 6 biotitebearing tonalite and are seemingly more homogenous on the outcrop and map scale.No crystallization ages existed from these latter units prior to the present study, except for two ID-TIMS multigrain U-Pb zircon ages of Stipp et al. (2004) determined for Avio and Presanella samples from the tectonically overprinted, very northernmost rim, which yielded lower intercept ages at 34Á6 6 1 and 32Á0 6 2Á3 Ma, respectively, as well as a SHRIMP age of 30Á5 6 0Á5 Ma of Grujic et al. (2011), on the same samples previously dated by Stipp et al. (2004).Our new age determinations, obtained by both LA-ICP-MS or SIMS techniques (see Fig. 2) indicate that there is significant age scatter within a single superunit, which in turn allows us to test the super-unit concept through quantification of time gaps at clear-cut magmatic contacts between units.
The apparent age dispersion within one unit ranges between zero (0Á29 6 0Á28 Ma for the CA unit) to a maximum of 2Á3 6 0Á2 Ma (south Re di Castello; assuming an uncertainty of 0Á1 Ma for the youngest date from Schoene et al., 2012), compiled in Fig. 2. The other examples of within-unit age dispersions are 0Á6 6 0Á3 Ma (north Re di Castello), 1Á5 6 0Á2 Ma (Western Adamello), 1Á8 6 0Á4 Ma (Avio) and 0Á8 6 0Á4 Ma (Presanella, disregarding the imprecise age of sample PR_AS5).We argue that these values are due to incremental batch assembly of the single units, which is strongly supported by the high-precision CA-ID-TIMS dates from four samples of the WAT scattering over 1Á2 Ma (Flo ¨ss, 2013).Incremental melt addition has been demonstrated as well through high-precision dating for the different units in the south Re di Castello, with zircon age dispersion over 100-200 kyr, that largely exceeds the intrusion and solidification duration of single magma batches (Schoene et al., 2012;Broderick et al., 2015).
The contact between RdC-N and WAT is easily accessible in the Val di Fumo and bracketed by samples RDC_AS9 and AD_AS2 (Fig. 1).The WAT-CAL age difference of 1Á0 6 0Á2 Ma may be taken as confirmation to establish CAL as a separate super-unit; however, this interpretation is based on a single CAL sample only and the two samples are several kilometres apart.The contact zone between the Avio and Presanella units features abundant signs of mingling between different magmas; the apparent age gap is 0Á7 6 0Á4 Ma, probably simply reflecting under-sampling.These new age determinations confirm that the super-unit scheme established by Callegari & Brack (2002) is useful to distinguish plutonic units, even in the paradigm of incremental melt assembly.Major compositional and age differences may, however, exist within the same unit, produced through melt accretion, or rejuvenation of previously emplaced mushes followed by mingling of melts and crystals (e.g.Coleman et al., 2004;Za ´k & Paterson, 2009).The U-Pb age record in Fig. 2 suggests near-continuous crystallization of zircon despite clearcut magmatic contacts visible at the outcrop scale.It should be noted that most data sets collected for individual samples in this study are statistically equivalent within analytical scatter and their age range thus should not be considered as representative of the time interval of zircon crystallization.

Incorporation of crustal components into Adamello magma
The rocks of the Adamello intrusive suite have been interpreted as having evolved from mantle-derived parent magmas (Ulmer et al., 1983;Kagami et al., 1991).Our new Hf and O isotope data from zircon (Figs 3, 4) indicate a general increase of a crustal component in the Adamello magmas over time.This is in line with the conclusion from whole-rock oxygen isotopes (Cortecci et al., 1979) and whole-rock Sr and Nd analyses (Del Moro et al., 1983a;Kagami et al., 1991).The first melts arriving in the upper crust are represented by the tonalites to trondhjemites of the Corno Alto unit and of a juvenile Hf and O isotopic composition in zircon.The trend of decreasing e Hfi with time (Fig. 3) delineates an increasing degree of crustal contamination during the build-up of the intrusive suite (from the Corno Alto to the Avio super-units), possibly with a slight reduction in the amount of assimilation during the last intrusive phase of the Presanella super-unit.Both isotope tracers show the highest proportion of crustal contamination within the Avio super-unit, with an e Hfi around -7 to -4 and d 18 O of 9Á0-9Á8 ( Figs 3,4).A certain relaxation within the Presanella unit is, however, dependent on only few low e Hfi and high d 18 O Avio analyses.
Zircon trace element compositions show a weak but systematic co-variation towards lower fO 2 and lower Eu/Eu* with decreasing age (Fig. 5b, c) that may be compatible with an increasing degree of crustal assimilation of a more reducing composition.The most oxidized and least contaminated magmas are found in the Corno Alto super-unit, whereas some members of the Avio and Presanella units show the most reduced values.
Another indicator of crustal assimilation is the presence of zircon grains with pre-eruptive ages.SIMS and LA-ICP-MS spot analyses yield maxima in age distributions at around 100, 300, 450-550, 850 and 1000 Ma.These age indications correspond to well-known orogenic phases reported from the South-Alpine basement as, e.g.outlined by Schaltegger & Gebauer (1999).Ages younger than the Variscan (300 Ma) may indicate Permian magmatism, which is widespread in the South-Alpine basement or, alternatively, point to assimilation of Triassic sandstones, containing detritus from Triassic volcanism (such as the middle Triassic 'Pietra Verde'; Mundil et al., 1996;Brack & Muttoni, 2000).The number of inherited or xenocrystic zircon grains found during this study is subordinate and not in agreement with the increasing proportion of crustal components identified by Hf and O isotopes over the lifetime of the magmatic system.This indicates that crustal assimilation is largely unrelated to shallow hostrock contamination during magma ascent and emplacement.
Can we reconstruct the lower-to-upper crustal magmatic plumbing system?
The near-continuous temporal record from zircon U-Pb dating, the decreasingly radiogenic evolution of Hf isotope composition through time, as well as the general paucity of preserved crustal zircon xenocrysts, suggests that most processes of assimilation, contamination and fractional crystallization were acting at a deep level of the magmatic system and are responible for the general geochemical signatures.A small number of zircon crystals were possibly formed already at that deep level of the crustal hot zone, when crystallization of pyroxene, amphibole 6 plagioclase 6 Fe-Ti oxides was saturating small volumes of residual liquids in zircon.Melts extracted from these deep levels into the middle crust are likely to be characterized by low crystal percentages; amphiboles carried upwards in mafic melts indicate high Al-in-amphibole pressures of 0Á8-1Á0 GPa, in excess of the ambient pressure at the emplacement level of 0Á3-0Á2 GPa, as recognized in the Mattoni and Blumone gabbros of the southernmost Adamello (Ulmer et al., 1983;Blundy & Sparks, 1992;Nimis & Ulmer, 1998).After ascent to mid-crustal levels, large volumes of magma underwent rapid crystallization of amphibole and plagioclase in the temperature range of zircon saturation at SiO 2 contents of 65-75% (Caricchi & Blundy, 2015), reflected by a dominant mode in zircon age probability curves (Caricchi et al., 2014(Caricchi et al., , 2016)).The last magma batches either carried more deeply crystallized mineral grains to the upper crustal emplacement level and, or, were extracted from a magma mush.This scenario implies that most of the zircons are not forming at the emplacement level and are thus 'antecrystic' (following Miller et al., 2007).This infers the existence of a mid-crustal, intermediate level of magma storage, where the thermal conditions, the size of the magma reservoirs and the influx/extraction rates would define whether stirring and homogenisation would occur.Estimating emplacement rates (often called 'magma flux') at different crustal levels is thus of interest for the reconstruction of physical and chemical parameters of the magmatic system.
We attempt to quantify magmatic volume and flux over the entire history of the Adamello suite using two approaches: (1) measuring the surface area of each super-unit and assuming a constant thickness of 3 km as the most conservative estimate based on the observed topographic relief; and (2) by using the zircon age dispersion within a magmatic unit in combination with the model of Caricchi et al. (2014Caricchi et al. ( , 2016)).
1. Volume estimates based on the surface exposure of each super-unit and assuming a constant 3 km depth (Supplementary Data Table S8; Fig. 6).These results indicate increasing magmatic volumes with decreasing age, with the Avio unit deviating from the trend towards a smaller volume.Calculating magma emplacement rates using these volume estimates and the age span defined by the youngest and oldest U-Pb spot analysis from each super-unit leads to estimates between 0Á8 and 7Á5 x 10 -4 km 3 /y (Fig. 6).Beside the fixed 3 km depth estimate, this approach also assumes age homogeneity in the vertical direction.Due to these gross oversimplifications, the results have to be considered as indicative only.2. Using the statistical parameters of zircon age distribution curves.The LA-ICP-MS and SIMS analyses were grouped to form five populations of zircon ages corresponding to Corno Alto, Re di Castello, Western-Central Adamello, Avio and Presanella.We consider that the analytical uncertainty is distributed normally around the averaged measured value for each spot.Thus, each date is transformed in dates that are normally distributed around the mean date.
From each of these distributions we select one date and we repeat the procedure 500 times.This technique allows us to determine the possible range and variance of the distribution of ages.The resampled dates are converted to the duration of crystallization, in which the oldest zircon age in each repetition is assumed to represent the onset of zircon crystallization within the respective super-unit.The veracity of the conclusion can be confirmed as the distribution of individual spot dates within one super-unit yields  The resulting cumulative distributions of zircon crystallization duration (grey lines in Fig. 7) are compared with the cumulative distribution of zircon ages obtained by thermal modeling.The thermal model has been outlined in Caricchi et al. (2014Caricchi et al. ( , 2016) ) and provides synthetic distributions of zircon dates as a function of the rate of magma injection ('magma flux') and the final volume for mid-to-upper crustal magmatic systems.This approach was initially designed for application to precise CA-ID-TIMS data, where the analytical uncertainty of each individual date is significantly smaller than the duration of zircon growth within one population of dates.This condition is only marginally respected in case of the elevated uncertainties of the LA-ICP-MS and SIMS spot dates collected in this study, therefore, only relative estimates of magma fluxes and volume of the different super-units can be obtained by their comparison to the synthetic populations obtained with thermal modeling.
The modeling results show that the rate at which magma is injected into the magmatic system ('flux') and its final volume; both control the number of zircons crystallizing over the entire super-solidus history of a magma reservoir (Caricchi et al., 2014(Caricchi et al., , 2016) ) and the total duration of zircon crystallization (Fig. 7a).Plotting the cumulative sums of the age distributions for magma injection rates of 0Á003, 0Á001 and 0Á0001 km 3 /yr, (obtained from the thermal model of Caricchi et al., 2014Caricchi et al., , 2016) ) and comparing them with 500 cumulative distributions bootstrapped from the population of individual spot dates of each super-unit (Fig. 7b-f) suggests that for all super-units, the rate of magma input into the system was comparable and equates to 10 -4 km 3 /yr, with final emplacement volumes of 100 to 500 km 3 .The oldest super-unit (CA) shows the shortest duration of zircon crystallization (i.e.zircon ages) and would, therefore, be the volumetrically smallest super-unit (Fig. 7b).The data for the next younger RDC unit (Fig. 7c) are compatible with a larger intrusive volume.Results for the WAT and CAL units indicate volumes similar to those of the RDC super-unit.Moreover, it is the unit with the smallest spread in the duration of zircon crystallization (Fig. 7d) which suggests that the majority of zircon grains crystallized in a restricted amount of time and, therefore, indicating a certain degree of thermal homogeneity within the magmatic system.The dispersion in the duration of zircon crystallization increases again in the Avio (Fig. 7e) and Presanella super-units (Fig. 7f).
Whereas these modeling approaches remain semiquantitative at best, both produce very low emplacement rates in the range of 10 -4 km 3 /yr, typical of mid-toupper crustal plutons (de Saint Blanquat et al., 2011;Caricchi et al., 2014).It should be noted that the two approaches may not reflect the exactly same process.The Caricchi et al. (2014) model quantifies the physical and thermal conditions at mid-crustal 15-20 km depths, where most of the zircon grew in the crystallizing magma, while the map-based volumetric approach would reveal magma emplacement rates at the present outcrop level, ignoring any potentially eroded volcanic edifice.

Cessation of magmatic activity
The simple, map-based approach, with assumed constant thickness, suggests increasing magma volumes with decreasing age, peaking in the 600 km 3 Presanella unit.There is no sign of decreasing magma flux as a precursor to a 'shut-down' of the magmatic system.The reasons for waning magmatism around 30 Ma are poorly understood, possible explanations may be: 1. Despite the increasing volume of arriving magma, the heat content of the entire system is not maintained and the system arrives close to solidification.The model of Caricchi et al. (2016) shows that a 1000 km 3 magmatic system solidifies within 1 Myr timescales at a constant magma flux of 10 -3 km 3 /y (see their Fig. 2), which could be taken as a maximum time estimate for the Presanella unit.This picture may need to be modified, since the 30Á7360Á29 Ma age from the northernmost Presanella rim (Grujic et al., 2011) may point to a rapid decrease of melt volumes during the Presanella magmatic episode towards the very end of magmatic activity.2. The SW-NE younging trend, as well as the systematic change in the Hf isotope composition in the Adamello igneous intrusive suite (with the exception of CA), may suggest a displacement of the magmatic center over the course of time.This effect can be seen at smaller geographic and temporal scales in the southern RDC unit (in the Val Fredda and Lago della Vacca Complexes, respectively; Schoene et al., 2012;Broderick et al., 2015).Magma may have been focused into another crustal segment, laterally or vertically, which is not visible with the exposure accessible today.et al., 2004) and 31Á8-31Á2 Ma (sample BR10-04 of Samperton et al., 2015).This indicates coeval magmatic activity at different centers along the Insubric line that became geographically closer through postmagmatic brittle, dextral shearing along the Insubric Line (Stipp et al., 2004).We thus conclude that the end of magmatic activity at $30 Ma is unrelated to the activity of movements along the Insubric line.
Pressure and temperature conditions during emplacement and post-emplacement cooling With the exception of the north-eastern corner of the Adamello igneous intrusive suite, where the plutonic rocks are bounded by the younger Tonale and Giudicarie faults, more than 80% of its outer borders represent primary intrusive contacts towards pre-Permian basement rocks and Permian-Triassic sedimentary and volcanic cover rocks (Fig. 1).The pressure/ depth of intrusion of some of the Adamello units was estimated to be 0Á25-0Á30 GPa on the basis of contact metamorphic assemblages (Riklin, 1983;Stipp et al., 2002;Pennacchioni et al., 2006), pointing to regional temperatures of c.250 C.This is in line with undisturbed Rb-Sr biotite ages in the basement and in Permian intrusions adjacent to the eastern Adamello border (Martin et al., 1996) and undisturbed Cretaceous zircon fission track ages (Viola et al., 2001), as well as with the illite crystallinity data of Riklin (1983) outside of the contact aureole.Our new 40 Ar-39 Ar age determinations of amphibole, biotite and K-feldspar, together with the published Rb-Sr and 40 Ar-39 Ar data of Del Moro et al. (1983aMoro et al. ( , 1983b)), define progressive post-emplacement cooling of the super-units from SW to NE, along steep magmatic cooling trajectories at rates )100 C/Ma down to at least 300 C, the assumed closure temperature of biotite for Rb-Sr and K-Ar (Fig. 8).This means that every unit was emplaced at a time when the previous intrusions had fully solidified and cooled down to temperatures below the solidus.The cooling trajectories of the seven analysed samples are, however, influenced by several effects: U-Pb dating of zircon, Ar-Ar on amphibole and biotite in the same units, combined with Rb-Sr and Ar-Ar data from the literature, clearly show that rapid cooling to ambient temperatures was mainly controlled by overall small magma fluxes and 'quiet' periods between successive intrusions.This also suggests that prolonged periods of volcanism during the Adamello magmatism are unlikely.We propose that the sedimentary record between 42 and 30 Ma may contain several ash layers that could, eventually, be associated with the major phases of Adamello magmatism, but, so far, the volcanic record is extremely scarce (D'Adda et al., 2011;Martin & Macera, 2014;Lu et al., 2018).

CONCLUSIONS
The results from the present study, combined with previously published studies (e.g.Del Moro et al., 1983aMoro et al., , 1983b;;Mayer et al., 2003;Schaltegger et al., 2009;Grujic et al., 2011;Schoene et al., 2012;Broderick et al., 2015) describe the detailed temporal and Hf, O isotope evolution of a continental magmatic intrusive suite over $12 Myr.Our results have a series of implications specifically for this case study, with some possible generalizations for upper crustal magmatism: 1.The melts of the Adamello intrusive suite very likely saturated in zircon at an intermediate level of a midcrustal magma storage region and a substantial portion of zircon crystals was entrained as 'antecrysts' by rising magmas from deeper levels into the upper crust; the distribution of zircon dates over 12 Ma suggests that each super-unit records a prolonged period of zircon crystallization that was probably semi-continuous; 2. The ascending plutons derived from the lowermiddle crust, digested increasing quantities of crustal material, indicated by systematically decreasing e Hfi and increasing d 18 O values in zircon.This assimilation process occurred at temperatures above zircon saturation and left only a few cases where inherited zircon material could survive.The subordinate proportion of xenocrystic zircon grains and inherited cores indicates that crustal assimilation was largely unrelated to shallow host-rock contamination during ascent and emplacement, as well indicated by fast post-emplacement cooling based on the Ar-Ar dates.3. Extraction of melts from the storage region into the upper crust was episodic and led to cross-cutting emplacement relationships among distinct petrographic units at present-day outcrop level (3-5 km paleo-depth); 4. The final emplacement occurred into c.250C crust, with each magmatic unit cooling rapidly after emplacement to ambient temperatures.This indicates that the overall heat addition was not sufficient to build up a regional thermal anomaly at emplacement level, rather the individual plutons formed a transient thermal signature inducing the formation of a thermal aureole in the immediate vicinity.This implies that contact metamorphic isograds around the Adamello intrusive rocks are diachronous, despite exhibiting similar spatial dimensions and that previous intrusions became thermally overprinted; 5.The source area of melting and/or the focus of magma emplacement was moving towards the NE with time, during the construction of the intrusive complex; 6.Based on O and Hf isotope systematics in zircon grains, we demonstrate non-monotonic isotope variations over time, which reflect variable degrees of crustal contamination in the course of the thermal evolution of this deep magmatic system.The maximum crustal contamination was achieved at the maximal thermal maturity of the magmatic system, after which time the magmatic system was cooling and shutting down magma production despite increasing volumes of magma and increasing magma flux into the upper crust.Rapid cooling from magmatic temperatures down to 300 C is reached presumably through both conductive (?) and convective heat loss from a crystallizing and cooling magma body.The thermal decay is suggested to be the result of insufficient magma flux to maintain the heat content for a large-sized magmatic reservoir ('plutonic death'); migration of the main area of melting and, or, tectonic activity along the Tonale Line are other possible reasons for the fact that no intrusive rocks are known from the Adamello suite following the largest-volume Presanella unit.
, who links shearing and magmatism.Pseudotachylite dates as young as $21 Ma were reported along the Tonale Line by Mu ¨ller et al. (2001), postdating the magmatic rocks by approximately 10 Myr.

Fig. 1 .
Fig. 1.Simplified geological map of the Adamello intrusive suite showing the location of the analysed samples (modified from Schalteggeret al., 2009).The distinction into super-units is modified fromCallegari & Brack (2002).The further subdivision of the Re di Castello S unit is not discussed in this study and refers to the previous geochronological studies ofSchaltegger et al. (2009),Schoene et al. (2012) andBroderick et al. (2015).

Fig. 2 .
Fig. 2. Compilation of 206 Pb/ 238 U age determinations of zircon from all super-units of the Adamello Intrusive Suite by laser-ablation sector-field single-collector ICP-MS and secondary ion mass spectrometry techniques.The mean ages of the Val Fredda complex and the Lago della Vacca complex (both part of the Re di Castello S unit) are from Broderick et al. (2015) and Schoene et al. (2012), respectively.

Fig. 3 .
Fig. 3. Initial eHf values measured by laser-ablation multi-collector ICP-MS on zircon grains from the six different super-units of the Adamello Intrusive suite plotted against the age of each individual zircon grain.Data for the Val Fredda complex (black diamonds) are from Broderick et al. (2015) and from the Lago della Vacca complex (blue triangles) from Schoene et al. (2012), in both cases CA-ID-TIMS U-Pb ages and solution multi-collector ICP-MS Hf data.Numbers 1 to 6 in grey circles represent whole-rock Hf isotopic analyses of six representative samples (Supplementary Data TableS4), plotted against their mean LA-ICP-MS or SIMS age, repectively (Table1): 1, RdC3; 2, Bruff2; 3, PB18; 4, PB25; 5, PB30; 6, P2790.

Fig. 4 .
Fig. 4. d 18 O of the magma in equilibrium with magmatic zircon at a temperature of 760 C calculated according to Bindeman & Valley (2002) from zircon d 18 O analyses.Data from this study are compared to whole-rock data from Cortecci et al. (1979).

Fig. 5 .
Fig. 5. Trace element analyses from zircon: (a) Lanthanide variation diagram normalized to MORB (Sun & McDonough, 1989) for the six super-units from the Adamello batholith, showing large variation in MREE concentrations; (b) Eu/Eu* vs Dy/Nd, defining a hyperbolic trend mimicking a liquid line of descent with decreasing temperature and increasing crystal fraction; (c) Eu/Eu* vs log fO 2 indicating that most of the data points are

Fig. 6 .
Fig. 6.Estimated magma volumes (black squares) and magma flux values (circles) over the entire magmatic evolution of the Adamello igneous suite, derived from measured surface outcrops and an assumed constant 3 km depth.

Fig. 8 .
Fig.8.40 Ar-39 Ar age determinations on amphibole, biotite and K-feldspar from different super-units of the Adamello Intrusive Suite.Single sample trends are marked with individual sample names.A general zircon crystallization temperature of 800 C has been adopted for simplicity; closure temperatures for amphibole, muscovite and biotite are generally accepted estimates.Rb-Sr and biotite40 Ar-39 Ar dates of DelMoro et al. (1983aMoro et al. ( , 1983b) ) for amphibole, muscovite and biotite are shown for comparison.

3 .
The magmatic activity may have been disrupted by tectonic activity along the Insubric Line.The data ofStipp et al. (2004) andGrujic et al. (2011) point to strike-slip along the Tonale line soon after emplacement of the Presanella unit, deforming the contact aureole at temperatures of 250-300 C, restricted to an area east of Passo Tonale.More recent data fromGrujic et al. (2011), however, demonstrate that emplacement of the Presanella unit and deformation were coeval.Our youngest zircon population in the Presanella unit indicates crystallization at 33Á16 60Á68 Ma (sample PR_AS5), whereas the zircon population fromGrujic et al. (2011) extends crystallization to 30Á760Á3 Ma.North of the Insubric Line, approximately 80 km further to the west in the Bergell intrusives, the magmatic activity broadly overlaps with the Presanella data, with earliest zircon crystallization at 32Á97 Ma (zircon 1 of Oberli

Table 1 :
Compilation of geographic coordinates and mean 206 Pb/ 238 U ages (by LA-ICP-MS and SIMS) from all samples . The oxygen isotope values follow a clear trend towards higher d 18 O values with decreasing age.The most primitive values are found in the tonalites of the Corno Alto (d 18 O ¼ 7Á24-7Á45), followed by the Blumone and Mattoni gabbro and tonalites of the southern Re di Castello unit.The trend of increasing d 18 O reaches a maximum with the biotitebearing tonalites and granodiorites of the Avio superunit, whereas the younger Presanella unit tends to go back to slightly lower values.The new d 18 O (magma) values calculated from zircon are closely comparable to whole-rock data from Cortecci et al. (1979), except for the Re di Castello N tonalite, where the Cortecci et al. data plot at lower values than those calculated from the zircon analyses.