Introduction

Spatial and temporal changes in the radiogenic isotopic composition of arc magmatism is a fundamental tool for studying the interactions between the crust and the lithosphere in convergent margins1,2. These changes have been used to interpret diverse tectonic phenomena including continental subduction3, slab-dip4, subduction rate5, subduction erosion6, delamination7, changes in crustal thickness8, and lithospheric extension9 among others. Spatial trends in the radiogenic isotopic composition of arc magmatism have been interpreted as an intrinsic feature of Cordilleran-style orogenic systems2 and likely imply a fundamental change in the dynamics governing the formation of their igneous rocks. A consistent trend has been observed in arc magmas from the Central Andes, U.S. Cordillera and Tibet, whereby more isotopically juvenile compositions are encountered near the trench (i.e. radiogenic or enriched 176Hf/177Hf and 143Nd/144Nd) and increasingly evolved trench-distal (unradiogenic or depleted 176Hf/177Hf and 143Nd/144Nd2). Furthermore, these Cordilleran systems are well-studied examples of both modern and ancient active margins, which suggest that this spatial-compositional trend is long-lived and persists throughout the life of a given continental arc. Additionally, these spatial trends are observed in a broad spectrum of geochemical compositions. These spatial isotopic trends for Cordilleran magmatism have been explained by the mantle lithosphere thinning towards the trench due to sub-lithospheric processes, such as delamination or subduction erosion, allowing the magmas to be sourced from isotopically juvenile asthenospheric mantle with minimal lithospheric interaction2. Conversely, isotopically evolved arc magmas are founded landward, where the absence of lithospheric thinning permits the development of a thicker continental mantle lithosphere, producing isotopically evolved arc magmas.

The Antarctic Peninsula represents the southern terminus of America’s cordilleran arc (i.e. western border of Northern, Central and South America), which is arguably continuous from the western Aleutians10 to the Antarctic Peninsula11, with the exception of the San Andreas fault12 in North America. Although difficult access has partially hindered its understanding and characterisation, a plethora of recent data now permits a more robust characterisation of this Mesozoic-Cenozoic continental margin13,14,15,16,17,18,19,20,21,22. The Antarctic Peninsula thus makes an excellent natural laboratory for both testing and revising the dynamics in Cordilleran convergent margins. We combine these existing datasets with newly presented geochemistry (elemental and isotopic) and geochronology from the Cenozoic arc record, providing key information for a poorly constrained period of the Antarctic Peninsula. We use these combined datasets to test the model proposed for other Cordilleran-active margins and examine the magmatic and tectonic history of the Antarctic Peninsula in comparison with the Central Andes, U.S. Cordillera and Tibet2. These are long-lived and well-studied convergent margins, which have been fundamental to build the knowledge we have on the evolution of arc magmatism2,23,24,25,26.

Results and discussion

The active margin of the Antarctic Peninsula: a Mesozoic-Cenozoic history

The Antarctic Peninsula orogenic system developed above the eastward-subducting oceanic lithosphere of the Phoenix Plate and represents one of the major magmatic arcs of the circum-Pacific rim extending for almost 1500 km south from the Drake Passage (Fig. 1). This convergent margin developed as an autochthonous continental arc on the margin of Gondwana11,14,27,28,29. It is considered to have initiated as early as the Triassic14,17,30 and the subduction system waned during the Cenozoic20,22,31,32. Cessation of subduction progressively occurred from south to north during the Cenozoic as a consequence of ridge-trench collisions of the Antarctic-Phoenix Plate system with the west margin of the Antarctic Peninsula32. Subduction finally ceased concurrently with the end of Antarctic-Phoenix spreading at ~3 Ma33. Continuing sinistral relative motion of South America and Antarctica was accommodated by transtensional opening of the Bransfield Strait as a pull-apart basin34,35. Under this regional transtensional regime without active subduction, low-volume alkaline magmatism commenced at ~15 Ma and is ongoing36,37.

Fig. 1: Arc magmatic ages in the Antarctic Peninsula.
figure 1

Magmatic age compilation used in this work (Supplementary Material: https://doi.org/10.5281/zenodo.10605934) along with estimated ages for subduction cessation by refs. 46,84. The inserted map showing Antarctica and Patagonia was generated by Google Earth™.

Geochemistry: nearly changeless

The whole rock major and trace element compositions of the convergent margin magmatism of the Antarctic Peninsula exhibit little variation for different time windows during the Mesozoic and Cenozoic (Fig. 2), with the exception of a shift from being peraluminous-dominant during the Triassic-Jurassic to metaluminous-dominant during the Cretaceous-Cenozoic (Fig. 2a). Nevertheless, they consistently show a strong affinity to volcanic arc granite compositions (Fig. 2b). Strong similarities are also observed in N-MORB normalised trace element abundances, showing in general an enrichment in Light Ion Lithophile Elements (LILE) and negative Nb, Ta and Ti anomalies, suggesting a subduction-derived component in their source. They also exhibit minor negative Eu anomalies, which suggests that plagioclase has fractionated, and the positive Pb anomaly is likely to have been derived from an upper crustal source38 (Fig. 2c). A lack of temporal trends in the Lan/Ybn and Sr/Y plots (Fig. 2d) throughout the Mesozoic-Cenozoic suggest that the crustal thickness of the arc did not change during this period39,40. However, estimates of continental arc crustal thickness using geochemical indices are problematic due to the multiple petrogenetic processes that occur in this setting41, and these should perhaps only be used as qualitative indicators42.

Fig. 2: Geochemical compilation of the arc magmas of the Antarctic Peninsula.
figure 2

a Aluminium saturation index (ASI) plot85, (Al/Na+K) and (Al/Ca+Na+K) are defined as molecular ratios and take into account the presence of apatite so that rocks with ASI N 1.0 are corundum normative and are termed peraluminous86. b Rb vs Yb + Nb discrimination diagram for tectonic settings from87. WPG within plate granites, VAG volcanic arc granites, ORG ocean ridge granites, syn-COLG syn-collisional granites. c Rare earth element and trace element abundances normalised to the N-MORB values of ref. 88. d Diagrams showing Lan/Ybn and Sr/Y versus age. The full dataset is included in the Supplementary Material (https://doi.org/10.5281/zenodo.10605934).

Isotopic compositions: time sensitive

In situ Lu-Hf isotopic compositions from zircon were compiled and complemented with Sm-Nd and Rb-Sr whole-rock isotopic data from the active margin igneous rocks of the Antarctic Peninsula (Fig. 3). Analysis of inherited zircon cores yield abundant Palaeozoic ages, which have been interpreted as having formed on the active margin of southwestern Gondwana14 and thus autochthonous to the Antarctic Peninsula. Furthermore, zircons formed during the Ordovician-Triassic yield Hf-isotopic compositions that broadly follow the evolution of the average bulk crustal source43 (Fig. 3a). This suggests that during the Palaeozoic there was not a relevant amount of primordial isotopically juvenile sources incorporated in these old arc magmas. During the Jurassic-Cretaceous, zircons yield Hf-isotopic compositions that suggest mixing between juvenile and evolved crustal sources. Isotopically juvenile compositions are dominant during the Palaeogene, which imply greater influence of the asthenospheric mantle in the sources of arc magmatism during this time period. Sm-Nd and Rb-Sr whole-rock isotopic compositions from igneous rocks broadly follow the Hf isotope in zircon trends observed for the Mesozoic-Cenozoic (Fig. 3b). They show a transition from the Triassic to the Cenozoic rocks, whereby they evolve from more isotopically evolved to juvenile magmas.

Fig. 3: Isotopic tracing compilation of the arc magmas of the Antarctic Peninsula.
figure 3

a Comparison of the 206Pb-238U zircon ages and zircon εHfi for crystals from the igneous rocks of the active margin of the Antarctic Peninsula. The 2σ uncertainties are ~±5% for U–Pb zircon ages. b Comparison of the εNdi and 87Sr/86Sri values of the igneous rocks of the active margin of the Antarctic Peninsula. Sources are colour-coded. 176Lu/177Hf = 0.015 for the average continental crust43. The full dataset is included in the Supplementary Material (https://doi.org/10.5281/zenodo.10605934).

The complex spatial relationship

The spatial distribution of igneous rocks with time is a fundamental tool for understanding the evolution of subduction-related magmatism2,22. We have used the geographic position of the igneous arc rocks and calculated the shortest distance to the trench. Although subduction is presently inactive, the trench remains observable as a prominent bathymetric depression along the western margin of the Antarctic Peninsula. The distance of a given arc pluton with respect to the continent-ocean boundary (COB) is plotted in Fig. 4a (locus of arc magmatism with time) and shows that (i) the position of the most proximal arc magmatism relative to the trench has remained broadly static with time, maintaining a consistent distance of ~100–150 km, and (ii) from ~100 Ma onwards the most distal arc magmas progressively migrate to the trench at ~6 km/Myr22 (Fig. 4a). This migration and resultant arc-narrowing results primarily from the arrival of progressively younger oceanic lithosphere at the trench with each ridge-trench collision. Because of the broadly north-south orientation of the Antarctic Peninsula, latitude is a convenient parameter to analyse spatial trends with time along the arc margin (Fig. 4b). This reveals that arc magmatism ceased at ~20–19 Ma along a relevant segment of the margin, from ~67–62°S, and shows the progressive northward cessation from ~100 Ma. Nd isotopic compositions of the arc magmatic rocks of the Antarctic Peninsula do not show a clear relationship with their distance from the trench (Fig. 4c), with both juvenile compositions (i.e. εNdi > 5) and evolved compositions (εNdi < 0) located from ~200 km up to ~450 km from the trench. This argues against a strong correlation between radiogenic isotopic compositions and distance to the trench observed in other Cordilleran orogens2. Nevertheless, the Nd-isotopic compositions of these rocks show a strong correlation with their crystallisation age (Fig. 4d), with a tendency towards more isotopically juvenile compositions with younger ages from ~100 Ma onwards. Juvenile Nd-isotopic compositions become dominant in the Cenozoic, which agrees well with the Hf-isotopic compositions in zircon data (Fig. 3a) and implies coupled behaviour between these two isotopic systems.

Fig. 4: Comparison of distance from the continent-ocean boundary (COB), crystallisation age, latitude, and εNdi of the igneous rocks of the active margin of the Antarctic Peninsula.
figure 4

a Comparison of distance from the COB and crystallisation age, adapted from22. b Comparison of crystallisation age and latitude. c Comparison of distance from the COB and εNdi, with crystallisation age denoted by a colour ramp. d Comparison of crystallisation age and εNdi. Full dataset is presented in the Supplementary Material (https://doi.org/10.5281/zenodo.10605934).

The evolution of the Antarctic Peninsula active margin

From our compilation of geochronological, geochemical and isotopic data, the arc magmatic rocks that formed during the Mesozoic and early Cenozoic along the convergent margin of the Antarctic Peninsula exhibit a wide range in distances to the trench and in their radiogenic isotopic compositions (Figs. 3, 4). During the Mesozoic, relatively diverse magma compositions and variable degrees of crustal assimilation are encountered, coincide with the subduction of the relatively old (i.e. >~40 Myr) oceanic lithosphere of the Phoenix Plate beneath the continental lithosphere of the Antarctic Plate22,44 (Fig. 5a).

Fig. 5: Schematic diagrams showing the sequential evolution of the active margin of the Antarctic Peninsula.
figure 5

a Prior to ~100 Ma, the subduction system was dominated by the subduction of the old oceanic lithosphere of the Phoenix Plate, developing a wide arc with juvenile to evolved radiogenic isotopic compositions. b From ~100 to ~20 Ma, subduction of the progressively younger Phoenix Plate oceanic lithosphere produced a narrower arc with progressively more juvenile radiogenic isotopic compositions. c Post ~20 Ma, subduction ceased along most of the margin following the ridge-trench collision of the Antarctic-Phoenix system.

During the Late Cretaceous and Palaeogene, the progressively younger oceanic lithosphere of the Phoenix Plate was subducted beneath the Antarctic Peninsula22,32 (Fig. 5b). This process is marked by: (i) progressive migration of magmatism at the rear of the arc towards the trench (Fig. 4a) resulting in the narrowing of the belt of arc magmatism; (ii) progressively more juvenile radiogenic isotopic compositions (Figs. 3, 4); and (iii) the northward-progressive collision of ridge-trench segments, causing the wane and eventual cessation of subduction. Our new data indicates that the last magmatic products of the arc date to ~20–19 Ma, which confirms previous studies31,45 along with recent findings22. This not only represents the last phase of arc magmatism in the Antarctic Peninsula, but also the Antarctic continent as a whole. Furthermore, they22 numerically related the slab age, convergence rate and slab dip of the Antarctic-Phoenix system to show that the narrowing of the arc and the cessation of arc magmatism in the Antarctic Peninsula was primarily in response to the subduction of progressively younger oceanic lithosphere along with a slower convergence rate. This suggests that slab age and convergence rate may affect the magmatic arc geometry and compositions in settings that are commonly attributed to slab-dip variation.

While successive ridge-trench collisions of the Antarctic-Phoenix system with the western margin of the Antarctic Peninsula caused the waning and cessation of arc magmatism (culminating at ~20–19 Ma), these collisions effectively occurred ~8–5 Myr after arc magmatism had ceased22 (Fig. 5c). The occurrence of amagmatic subduction for ~8-5 Myr after final arc magmatic cessation at ~20-19 Ma implies subduction effectively ceased by ~15-12 Ma across most of the northern Antarctic Peninsula. Subduction continued in three segments of the Phoenix Plate, located to the north of the Hero Fracture Zone (Fig. 5c) albeit at a much slower rate until ~3 Ma32,33,46. Nevertheless, arc magmatism also waned at ~20-19 Ma in this sector31, as it had along the rest of the Antarctic Peninsula margin.

Comparison with other Cordilleran-style subduction systems

Compilations of Lu-Hf, Sm-Nd, Rb-Sr and Pb isotopic data on magmatic rocks have been used as a tool to understand how active margin orogens evolve with time2,15,16,23,24,26,47,48. We present a compilation of Lu-Hf and Sm-Nd isotopic compositions of magmatic rocks from three long-lived convergent margins2: (i) the Central Andes, an active cordilleran margin; (ii) the western United States (U.S.) Cordillera, an ancient active margin; and (iii) Tibet, an active margin that has transitioned into a continental collisional orogen following India-Asia collision. These other Cordilleran orogens have a distinctive spatial-temporal pattern in the radiogenic isotopic compositions of their igneous rocks, displaying more isotopically juvenile compositions near the (paleo-) trench and increasingly evolved isotopic compositions landward (Fig. 6). This suggests that the isotopic composition of magmatism at any given location in the orogenic system varies within a limited isotopic range throughout the life of the continental arc2, and thus contemporaneous magmatism has different isotopic compositions when emplaced at variable distances from the trench. Furthermore, these magmas exhibit a range of geochemical compositions, suggesting that this spatial trend may be an inherent feature of active margin systems.

Fig. 6: Comparison of the Antarctic Peninsula with other active margins.
figure 6

Compiled Hf and Nd isotopic data from igneous rocks for modern and ancient active margins in the left-hand panel. a Tibet, b U.S. Cordillera, c Central Andes, d Antarctic Peninsula. Magmatic isotopes are compared with their distance to the (palaeo-)trench and their crystallisation age (presented as a colour ramp). Sample locations are presented in the right-hand panels and were prepared with QGIS open-source software. Whilst Tibet, the U.S. Cordillera, and the Central Andes show a clear correlation between the distance to the trench and their isotopic composition, the Antarctic Peninsula active margin lacks such a clear relationship. Data from Tibet, U.S. Cordillera and Central Andes was obtained from ref. 2.

The Tibetan, Central Andes, and U.S. Cordillera yield a strong correlation between the distance to the trench and their isotopic compositions (Fig. 6a,b,c). The distance to the trench in the Tibetan orogen system is measured with respect to the Indus-Yarlung suture (Fig. 6a), along which oceanic lithosphere was subducting northward beneath Asia prior to the subduction of the Indian plate25. The distance to the trench in the U.S. Cordillera is measured by the longitude, which is a useful trench-distance parameter given the approximate east-directed subduction direction of the oceanic lithosphere (Farallon Plate) (Fig. 6b). Similarly, the Central Andes also have east-directed subduction whereby the Nazca Plate oceanic lithosphere subducts under the South American continental lithosphere, and thus longitude is again the chosen trench-distance parameter (Fig. 6c). The colour ramps presented in Fig. 6 show the distribution of the magmatic ages, allowing comparison of their crystallisation age, isotopic composition, and distance to the trench. As it has been shown2, the correlation between the isotopic compositions of the magmas and their distance to the trench is independent of age (Fig. 6a, b, c). However, the arc magmatic rocks from the Antarctic Peninsula do not exhibit this spatial-isotopic trend (Fig. 6d), yielding isotopically juvenile compositions with a wide range of distances from the trench, as do the more isotopically evolved compositions. As shown previously, the igneous rocks of the active margin of the Antarctic Peninsula exhibit a much stronger correlation with their crystallisation age (Figs. 3, 4d).

Implications for active margin dynamics

Melts extracted from the convecting asthenospheric mantle are (continually) depleted in Hf/Lu and Nd/Sm49 and exhibit isotopically juvenile compositions. Conversely, in an isotopically closed system, magmas assimilating older continental crust will be more isotopically evolved than those magmatic products that assimilate more juvenile continental crust. This principle renders radiogenic isotopic systems a powerful tool to understand the evolution of active margin dynamics. Evolved or juvenile isotopic compositions may not be necessarily related to the degree of magmatic differentiation (i.e. the change from mafic to felsic bulk compositions) and thus, bulk geochemistry may be decoupled from the isotopic compositions of a given rock50. This is one of the main advantages of using radiogenic isotopic tracing over bulk rock compositions for assessing mantle-crust dynamics in active margins. If there is crustal assimilation throughout the evolution of a given magmatic system, bulk rock compositions can potentially be linked to the isotopic evolution50. However, the arc magmatic rocks from the Antarctic Peninsula show relatively constant geochemical compositions throughout the life of the margin (Fig. 2), but radiogenic isotopes exhibit a clear shift at ~100 Ma in both the Lu-Hf and the Sm-Nd systems (Figs. 3, 4d).

The spatial relationship between the distance to the trench with the isotopic compositions observed in other active margins2 is not clearly observed in the Antarctic Peninsula (Fig. 6) and therefore, the causative mechanisms for this trend proposed for other active margins may not be applicable to the Antarctic Peninsula. The relationship between the isotopic compositions with distance to the trench in Cordilleran margins has been primarily explained by the relative presence or absence of continental mantle lithosphere2 (which produces evolved or juvenile compositions, respectively). In contrast, the Antarctic Peninsula shows a progressive evolution towards more juvenile compositions (Figs. 3,4d) and a narrowing of the arc (Fig. 4a) from the Late Cretaceous onwards. Furthermore, it is possible to observe a progressive narrowing of the arc, with the arc front maintaining a distance of ~150–100 km and the rear of the arc migrating towards the trench (Fig. 4a). The waning of Phoenix Plate subduction beneath the Antarctic Peninsula has been associated with roll-back45 and slab window processes51. However, recently this process has been modelled22 and showed that the narrowing of the Antarctic Peninsula arc primarily results from the subduction of progressively younger oceanic lithosphere and secondarily with a decrease in convergence rate. This argues against the formation of isotopically juvenile magmas by a slab roll-back mechanism, as in that case the front of the arc should migrate trenchward but instead the arc front maintains a steady position relative to the trench (Fig. 4a). A slab window mechanism is not supported either, because arc magmatism ceased ~5–8 Myr prior to the ridge-trench collisions and thus the slab window event postdates the youngest subduction magmas27. However, we acknowledge that a slab window may have developed after the ridge-trench collisions of the Antarctic-Phoenix system, as it has been suggested51.

We suggest that the progressive narrowing of the arc by the migration of its rear from ~100 Ma onwards played a pivotal role in the generation of isotopically juvenile magmas. This migration occurred as a consequence of the progressively younger age of the subducted slab (Phoenix Plate) and the progressive decrease in convergence rate22. This led to the formation of melts that were derived by differing degrees of differentiation (i.e. from mafic to felsic compositions), but with a common and dominant isotopically juvenile signature. Furthermore, while these magmas differentiated to granitic compositions, they did not assimilate relevant amounts of old and isotopically evolved continental crust. Additionally, as older oceanic lithosphere is colder and as water is stored in the form of serpentine, fast subduction of old oceanic lithosphere allows the slab to remain hydrated until deeper in the mantle52; conversely, the slow subduction of young oceanic lithosphere has a reduced capacity to add volatiles to the lower crust. Therefore, as the rear of the arc migrated towards the trench during the Late Cretaceous–Cenozoic in the Antarctic Peninsula, the subducted slab had a lesser capacity to hydrate the continental plate above. This led to the establishment of narrow arc above a ‘drier’ lithospheric mantle beneath the Antarctic Peninsula lithosphere, which ultimately resulted in reduced capacity to incorporate older continental crustal material to these arc magmas. Additionally, the decrease in convergence rate that occurred throughout the Cenozoic would have reduced the friction between the subducting plate and the forearc53,54. Furthermore, the strength of the subduction interface critically controls the coupling between the subducted slab and the overriding plate54. This relative decoupling in the interface between the slab and the Antarctic Peninsula may also have played a role, in which the reduction of friction provided space that allowed young and juvenile asthenosphere to ascend to the lower continental crust. Therefore, as the arc narrowed throughout the Late Cretaceous and Cenozoic, it incorporated magmas that were dominantly sourced from juvenile asthenospheric mantle and the arc magmas thus become progressively more juvenile with time.

The tectonic plate environment in which the Antarctic Peninsula developed is unique compared to other Cordilleran systems. The Tibetan active margin experienced the subduction of progressively older oceanic lithosphere55 during the late Mesozoic and early Cenozoic, which transitioned into a continental collision25. The western U.S. Cordillera margin experienced a complex history with the subduction of the Farallon Plate56,57 and the Juan de Fuca Plate, along with slab window processes58 and the development of the large strike-slip San Andreas fault system59. The Central Andean margin underwent subduction of the Phoenix and Farallon plates60, followed by the Nazca Plate after their consumption (which is currently subducting beneath South America61). Therefore, the Tibet, U.S. Cordillera and Central Andes active margins experienced a wide diversity of plate configurations and the subduction of different oceanic plates, which contrasts with the simpler subduction history of the Antarctic Peninsula. While isotopic trends can clearly provide information on the architecture and tectonic history of subduction as it has been clearly shown2, the Antarctic Peninsula exhibits differing behaviour, and thus expands our understanding of how Cordilleran orogens evolve.

The role of arc magmatism in forming continental crust

The Hf isotopic compositions pre-Cretaceous zircons from the Antarctic Peninsula exhibit an evolution that mimics that of ancient, enriched lithospheric compositions43 (Fig. 3a), whilst Cretaceous and younger zircons yield compositions suggesting the involvement of relatively more radiogenic sources. The pre-Cretaceous rocks that formed on the active margin of the Antarctic Peninsula were mainly derived from recycled Sunsas-aged crust (1.1-1.0 Ga14,16,62), and that juvenile continental crust mainly formed during the Cretaceous. The dataset presented here suggests that the formation of juvenile crust continued into the Cenozoic. Therefore, while the margin of the Antarctic Peninsula was active from the late Palaeozoic—early Mesozoic14,18,30, it effectively generated juvenile crust from the Cretaceous onwards. This is contemporaneous with the progressively younger age of the subducting slab and its waning subduction rate, and therefore poses a link between the cessation of subduction and the effectiveness of a given active margin to generate juvenile crust. When considering the broader process of continental crustal growth and crustal recycling, this observation suggests that active margins undergoing waning subduction are more efficient at generating juvenile crust.

Methods

We combined new geochronological, geochemical and isotopic analyses on Cenozoic igneous rocks of the Antarctic Peninsula with a comprehensive collection of previous work on this region. This compilation is available in the Zenodo data repository (10.5281/zenodo.10605934). The whole rock geochemistry, whole rock (Nd-Sr) and mineral (Hf in zircon) isotopic tracing along with the zircon geochronology was acquired using the procedure presented in refs. 14,15,16,63,64,65,66. Additionally, we used the dataset presented in ref. 2 to compare our compilation of the Antarctic Peninsula with other active margins (i.e., Tibet, U.S. Cordillera and Central Andes).

Whole rock geochemistry

Agathe mill was used to prepare the whole rock powders, which were analysed using a Philips PW2400 X-Ray Fluorescence (XRF) spectrometer at the University of Lausanne, Switzerland. Standards NIMN, NIMG, BHVO and SY2 were used for quality control. Glass-fused disks were analysed for trace and rare earth elements (REE) by an Agilent 7700x quadrupole ICP-MS, which is also hosted at the University of Lausanne. NIST SRM 610 and 612 were used as external standards. The laser was set to a 10 Hz repetition rate and a spot size between 80 and 120 μm. While blanks were measured for ~90 s, standards and unknowns for 45 s. The Sr or Al2O3 concentrations obtained by XRF were used as an internal standard. Three ablations were performed by sample and their concentrations were calculated offline with LAMTRACE67 and normalised to an anhydrous state in all diagrams. The uncertainties per sample are ±10% for REE and ±5% for other trace elements.

Sr-Nd whole rock isotopes

The powders prepared for the geochemical analyses were also used for the Sr-Nd whole rock isotopes, which were measured at the University of Geneva with a Thermo Neptune PLUS Multi-Collector ICP-MS following the procedure described by refs. 39,68. For this, it was required to dissolve 100 mg of whole rock powder in 4 ml of concentrated HF and 1 ml of 15 M HNO3 in closed Teflon vials at 140 °C for 7 days. These solutions were dried down, to be after re-dissolved in 3 ml of 15 M HNO3 and dried down again. Sr–Nd chemical separation followed the procedures described in refs. 69,70. While the internal fractionation was corrected using 88Sr/86Sr = 8.375209 for the 87Sr/86Sr ratio and 146Nd/144Nd = 0.7219 for the 143Nd/144Nd ratio; the external standards emplyed are SRM987 (87Sr/86Sr = 0.710248, long-term external reproducibility: 10 ppm) and JNdi-1 (143Nd/144Nd = 0.51211571; long-term external reproducibility: 10 ppm). Sr and Nd isotope ratios were further corrected for external fractionation by a value of –0.039 and +0.047 amu, respectively. 83Kr and 85Rb were monitored to correct the mass interferences at 84 (84Kr), 86 (86Kr) and 87 (87Rb). Furthermore, 147Sm was used (144Sm/147Sm = 0.2067) to correct the interference of 144Sm on 144Nd.

Zircon LA-ICP-MS U-Pb geochronology

Zircon U–Pb isotopic composition was collected at the University of Lausanne using an Element XR single-collector sector-field ICP-MS (Thermo Scientific). Ablations were performed with an UP-193FX ArF 193 nm excimer ablation system with a configuration that consisted in: 35 μm beam size, 5 Hz repetition rate, 30 s signal and 2.2–2.5 J/cm2 of beam energy density. While the primary standard used was GJ-1 zircon (CA-ID-TIMS 206Pb–238U age of 600.5 ± 0.4 Ma72,73); the secondary standards were either Harvard 91500 (1065.4 ± 0.3 Ma74) zircon, or Plešovice (337.13 ± 0.37 Ma75) zircon. LAMTRACE67 and IsoplotR76 were used to calculate the dates, for which it was used only zircons yielding concordance greater than 90%.

Zircon in-situ Hf isotopes

The zircons used for U–Pb geochronology were also used for collecting in-situ Hf isotopes at the University of Geneva. These were obtained on a Thermo Neptune Plus MC-ICP-MS coupled to a Teledyne–Photon Machines Analyte G2 ArF excimer laser system equipped with a two-volume HelEx-2 ablation cell77. The configuration of the ablations consisted in a fluence of ~4 J/cm2, a repetition rate of 5 Hz and a spot size of 40 μm or 50 μm. The carrier gas was He mixed with a small amount of N2 before entering the Ar-plasma torch for sensitivity improvement. Low mass resolution was used to collect the measurements. Blanks and standards were analysed at the initiation, finalisation and every 15 analyses of unknown zircons. The standards used are Mud Tank, Plešovice, MUN4 and GJ-1 zircon.

Excel spreadsheets were used off-line to reduce the data, which consisted of the subtraction of (i) blanks and (ii) the isobaric interference of 176Lu and 176Yb on mass 17678. Additionally, the 176Hf/177Hf ratio was also corrected for mass bias79. βHf and βYb mass bias coefficients were calculated using the reference values of 179Hf/177Hf = 0.732580 and 173Yb/171Yb = 1.123481, respectively. Corrections for the isobaric interferences of 176Yb and 176Lu with 176Hf were performed using 176Yb/173Yb = 0.786954 and 176Lu/175Lu = 0.02645, respectively81. The 206Pb–238U date of the respective crystal along with CHUR parameters (176Hf/177Hf = 0.28278582, 176Lu/177Hf = 0.033682 and λ176Lu = 1.87 × 10−11 yr−183) were used to calculate 176Hf/177Hfi, and εHfi.