Alps to Apennines zircon roller coaster along the Adria microplate margin

We have traced the particle path of high-pressure metasedimentary rocks on Elba Island, Northern Apennines, with the help of a U-Pb-Hf detrital zircon study. One quarter of the analysed zircons are surprisingly young, 41-30 Ma, with a main age peak at ca. 32 Ma, indicating an unexpected early Oligocene maximum deposition age. These Oligocene ages with negative εHf indicate a volcanic source region in the central-southern Alps. Though young by geological means, these zircons record an extraordinary geodynamic history. They originated in a volcanic arc, during the convergence/collision of the the Adria microplate with Europe from ca. 65 to 30 Ma. Thereafter, the Oligocene zircons travelled ca. 400 km southward along the Adria margin and the accretionary prism to present-day Tuscany, where they were subducted to depths of at least 40 km. Shortly thereafter, they were brought to the surface again in the wake of hinge roll back of the Apennine subduction zone and the resulting rapid extensional exhumation. Such a zircon roller coaster requires a microplate that has back-to-back subduction zones with opposing polarities on two sides.

metasiltstones -APM). The APM includes discontinuous layers of calc-schists characterised by scattered lenses of metabasites with high-P paragenesis, i.e. glaucophane formed at epidote blueschist facies conditions, recording maximum pressure of 0.9-1.0 GPa at ca. 330-350 °C 8 . The APM are highly deformed, displaying asymmetric isoclinal folds as well as sheath folds and oblique folds in calc-schists lenses (Supplementary File 1). Muscovite grown on the foliation planes in the calc-schists provided a 40 Ar- 39 Ar age of 19.68 ± 0.15 Ma 14 , which is so far the best age estimate for the stacking-related metamorphic overprint. The deposition age of the APM is still unclear: it is most commonly considered of Cretaceous age 11,13 , although similarities with the Pseudomacigno formation of mainland Tuscany may suggest a much younger age 12 .

Samples and Methods
Three samples of the APM underwent a U-Pb + Lu-Hf detrital zircon characterisation. Two samples, EJ50 and EJ67, were collected to the west of the Capo d´Arco residence in the southern part of the study area, whilst the third sample was collected along the road to Ortano (Fig. 2, Supplementary File 1). All three samples are from silty to sandy parts of the APM and were found to contain abundant zircon grains that were separated using conventional mineral separation techniques. Ca. 300 zircons per sample were handpicked, mounted in epoxy and sectioned into half by grinding and polishing. After cathodoluminescence (SEM-CL) imaging, zircons were first analysed for U-Pb by LA-ICP-MS and/or SHRIMP; in a second step, zircons underwent Lu-Hf characterisation (Supplementary File 2). A SEM-CL zircon typology study was carried out on 55 dated grains. Petrographic and geochemical analyses were also performed to define the nature of the studied metasedimentary rocks.  11,36,37  Petrography and geochemistry. The phyllites and metasiltstones are highly foliated and consist of alternating very fine-and medium-grained layers (Supplementary File 1). The former are dominated by quartz, sericite, chlorite and opaque minerals, the latter contain detrital K-feldspar, plagioclase (An [30][31][32][33][34][35] ) and calcic pyroxene (diopside with Mg# ~0.65-0.70) along with accessory titanite, rutile, zircon, allanite and apatite. Detrital minerals The youngest zircon population is unexpectedly young, Eocene-Oligocene in age, and makes up approximately a quarter of the entire zircon population. This age group has an asymmetric peak of 31.6 Ma and an older age tail up to ca. 41 Ma. These young zircons typically have Th/U ranging from 0.12-1.0 and are characterised by oscillatory zoning, pointing out their igneous nature. A mostly volcanic, rather than plutonic, origin is suggested by the preservation of tiny edges of zircon grains. The magmatic affinity has been investigated by means of a typology analysis 15 using CL images 16 of 55 dated crystals, indicating a most likely origin from a calc-alkaline association, along with minor occurrences of shapes typical of peraluminous crustal products.
With the exception of one age of ca. 70 Ma, our samples lack Mesozoic and Paleocene age components. A significant age peak at ca. 280 Ma makes up 18% of the age population and relates to a late Variscan igneous provenance. The largest age peak with over a quarter of all ages is of Ordovician age, ca. 480 Ma, and most likely relates to voluminous Ordovician felsic magmatism at the active northern Gondwana margin 10,17 . The Neoproterozoic to Archean ages correlates to an African heritage and times when Adria had a peri-Gondwana position. Zircons with Th/U < 0.1, ca. 5% of the entire population, may be of metamorphic origin and are almost entirely related to Cambrian-Neoproterozoic ages.
Lu-Hf measurements for Oligocene zircons show significant scatter (Fig. 2c), from initial εHf of −7 to +1, with a mean of −4, indicating the involvement of old, probably Meso-Paleoproterozoic crust.
Oligocene zircon provenance. The Oligocene zircons are interpreted as igneous zircons because they are euhedral, oscillatory zoned and have typical Th/U from 0.12-1.0. Furthermore, our zircon SEM-CL typology study indicates a calc-alkaline origin; the majority of the zircons are thus most likely related to Alpine-Apennine convergent tectonics. The overall detrital mineralogy and geochemistry indicate an andesitic to mafic source material of likely volcanic nature. The age of deposition must be younger than the youngest detrital zircons that the unit contains. The youngest concordant single zircon ages from the phyllites and metasiltstones are 29 ± 3 Ma (2σ) analysed by LA-ICP-MS, and 31 ± 2 Ma (2σ) analysed by SHRIMP. A more robust estimate of the maximum deposition age is the age of the youngest detrital zircon age component (i.e. group of zircons), which, for both LA-ICP-MS and SHRIMP analyses combined, is 31.6 ± 0.5 Ma (2σ). This is the best estimate for the maximum deposition age of the APM, which is much younger than previously assumed 11,13 , and makes it coeval with the more felsic foredeep deposits of the unmetamorphosed Macigno or the metamorphic Pseudomacigno units on mainland Tuscany 12 . Our new data show that these metasediments were deposited after ca. 32 Ma but before ca. 20 Ma, the age of the greenschist, stacking-related metamorphism 14 . Thus, they reveal very rapid cycling from volcanism to lateral fluvial translation, followed by subduction and extensional exhumation.
The potential source regions for the young Eocene-Oligocene zircons (Supplementary File 4) include the European plate, on the northern/western side of the Alps-Apennine system and/or the Adria Plate on the eastern/ southern side 18 . Potential source regions of the European plate that were exposed by 30 Ma include magmatic active zones in Corsica-Sardinia and Provence-Esterel or rocks from the Alpine foreland basins, such as the Taveyanne Sandstones. The Cenozoic calc-alkaline igneous activity on Corsica-Sardinia ranges from 38 to 12 Ma, but with the climax around 20 Ma, thus postdating the zircon population in our rocks 19 . The same accounts for Provence-Esterel where the igneous activity starts at 41 Ma, but the most represented ages are in the range 33-19 Ma, again younger than APM zircon ages 20 .
Potential sources on the Adria Plate include the Periadriatic igneous complexes associated with, from west to east, the Biella, Bergell and Adamello plutons in the Alps 18,21 . The plutons themselves are no potential sources, since they were not exhumed before 10 Ma 22-24 , nevertheless their ages are important in suggesting ages of potential volcanic activities and/or antecrystic zircons that volcanoes can erupt. The Biella volcanic complex in northern Western Alps lacks ages older than ca. 33 Ma 25 . The Bergell pluton in Central Alps, with ages clustering around ca. 30 Ma, is deeply unroofed and is proven to have supplied volcanic detritus since the beginning of late Oligocene times 23 . The Adamello pluton in the Southern Alps shows emplacement ages of ca. 44-31 Ma 26-28 , which very well overlap with our detrital zircon ages, though no evidence for a volcanic complex above the Adamello pluton is reported so far. Therefore, the Bergell complex appears as the most likely source, although a minor contribution from Adamello cannot be excluded, to explain ages between 33 and 41 Ma found in the APM.
The Hf isotopic composition of our dated zircons also matches the range of isotopic compositions of the felsic rocks of the Adamello massif (Schaltegger et al., 2009;Schoene et al., 2012) very well (Fig. 2D). The zircon data for the Adamello and Bergell mafic samples 27 are more primitive than for the zircons we analysed. For the Bergell, the whole rock Nd isotope data 29 shows that more felsic rocks are isotopically more enriched (i.e. more contaminated; lower εNd and εHf). If we calculate the Hf isotopic composition of zircon in equilibrium with the whole rock Nd isotopes, following the terrestrial array 30 , there is good overlap with the younger group of our zircons. The plutons volcanic roofs are fully eroded now 23 . Their erosional products, however, are seen in the Oligocene Aveto-Petrignacola Formation that includes conglomerates of volcanic origin 31 . Thus, our Oligocene zircons could either have been derived directly from the volcanic roof of Bergell-age volcanic complexes or their respective erosional derivates in Aveto. Their erosional products must have been fed longitudinally from the Alps into the Apennine foredeep basin 32 and reached latitudes equivalent to present-day Elba Island. Paleozoic-Precambrian zircon provenance. Provenance of the pre-Mesozoic zircons (Fig. 3a) also argues for a source region in the Alps. If our zircon pool is compared with the age distribution spectra of various parts of the Alps, Apennines, and Corsica-Sardinia (Fig. 3a), then closest similarities are seen with the proximal and distal succession of the Adria foredeep 22 . In the Oligocene, the Tuscan basement 10,17 was largely covered by sedimentary rocks and was therefore not available to contribute as a source for the sediments. The Eocene-Miocene deposits from Corsica-Sardinia have detrital zircon age distributions that lack a significant 250-300 Ma (Variscan) age component, but instead have a significant Grenville-age age group 33 , the latter of which is not present in our samples. The age distribution spectra (Fig. 3a) show that modern rivers draining the western-central-southern Alps and the northern Apennines have a proportion of Permian-Carboniferous (Variscan) zircons that is commonly larger than found in our samples, with the notable exception of sediments drained from the Ticino subdome of the Lepontine dome, dominated by Ordovician (Caledonian) zircon ages 34 . This suggests that crystalline detritus found in APM most likely derives from an area around the Bergell igneous complex. Our samples most likely represent a mixture of detritus dominated by a Central Alps provenance. Recycling through the Aveto-Petrignacola Oligocene sedimentary basin is also possible, owing to the comparable overall detrital zircon age distributions of Aveto and APM (Fig. 3). Nevertheless, in detail, some differences in the peak distributions and heights can be noted, such as the similar importance of Ordovician and Oligocene peaks in the APM, contrasting with the modest abundance of Ordovician zircons for two Aveto samples, and the strong dominance of Ordovician over Oligocene ages for the third sample 18 . So, recycling of sediments through the Aveto basin is possible, although only partially.
The Oligocene zircon roller coaster. The Oligocene zircons from the APM on Elba Island record a remarkable ca. 12 Ma geodynamic history. They most likely originated from Periadriatic volcanoes of the Central Alps, when Europe collided with and was subducted underneath Adria. These volcanoes were eroded and longitudinally fed into the Apennine foredeep, where they were caught up in the Apennine subduction zone, ca. 400 km away from their volcanic origin, similar to what has been reconstructed for the Aveto basin 18 (Fig. 4). After subduction to depth of at least 40 km, they underwent exhumation and emplacement in a nappe stack under high-pressure greenschist facies conditions by ca. 20 Ma and were subsequently rapidly brought to the surface by extensional exhumation. The rapid subduction-exhumation cycle may be explained by the collision of a microplate with a larger continent, where hinge roll back leads to extensional detachment resulting in rapid exhumation 35 . To take zircons on such a remarkable roller coaster ride thus requires two subsequent subduction zones on two different sides of the microplate (Adria): an older (Alpine) subduction underneath the microplate produces an active plate margin and the associated volcanism, followed by partly coeval (Apennine) subduction of relatively old oceanic lithosphere with an opposite subduction direction that allows for roll back, which proceeds underneath the microplate and thus yields subduction followed by rapid exhumation. In between, the erosional products of volcanoes related to the first subduction zone are transferred several hundred km by longitudinal  transport along the Apennine foredeep, in analogy with the Aveto basin case 18 , before their rapid subduction and exhumation in the second subduction zone system (Fig. 4).