Syn‐Orogenic Exhumation of High‐P Units by Upward Extrusion in an Accretionary Wedge: Insights From the Eastern Elba Nappe Stack (Northern Apennines, Italy)

The E‐vergent Northern Apennines formed by Oligocene‐Miocene convergence and westward subduction of Adria beneath Europe. Extension ensued in the Mid‐Late Miocene reflecting lower plate roll‐back and causing opening of the back‐arc Northern Tyrrhenian Sea. Post‐orogenic extension is commonly advocated as the main driver of the exhumation of the belt's inner domain high‐pressure/low‐temperature (HP‐LT) rock units. The Acquadolce Subunit of the Eastern Elba nappe stack contains HP‐LT rocks recording peak blueschist conditions of 1.5–1.8 GPa at 320°C–370°C loosely dated to the Oligocene‐Early Miocene. It is sandwiched by two Late Miocene, out‐of‐sequence top‐to‐the E thrusts between Jurassic LP serpentinites on top and HT–LP contact metamorphosed marbles at its base. We document widespread W‐verging ductile asymmetries within the Acquadolce Subunit, which correspond to top‐to‐the W extensional shearing for the nappe stack current orientation. This allowed for early syn‐orogenic exhumation from blueschist‐ to greenschist‐facies conditions, wherein coeval W‐directed extension at the top of the exhuming units acted synchronously with E‐directed thrusting at their base causing exhumation by extrusion in an overall contractional setting. The basal, E‐vergent thrusting is, however, challenging to document as the wedge has since been reworked by Late Miocene, E‐verging compressive tectonics, contact metamorphism, and later extension, obliterating much of the evidence supporting exhumation by extrusion during the early stages of wedge build‐up. Syn‐orogenic exhumation by extrusion from deep structural levels within the orogenic wedge is a viable mechanism to account for other exhumed HP‐LT units in the inner part of the belt.

. According to several authors, the Alpine Tethys Ocean closed due to an earlier E-dipping Late Cretaceous-Eocene subduction that culminated with the SW-verging "Alpine" orogeny in Corsica (Figure 1a). The following evolution was characterized by a regional E-NE-ward tectonic vergence, due either to a flip or an along-strike change of the earlier subduction polarity (e.g., Marroni et al., 2017;Molli & Malavieille, 2011;Vignaroli et al., 2008). After the onset of continental collision in the Oligocene, the Northern Apennines orogenic wedge widened and propagated toward the east-northeast, while extension, basin RYAN ET AL.

10.1029/2020TC006348
3 of 30 Figure 1. Regional geological setting of the Northern Apennines: (a) Tectonic sketch map of the Northern Apennines-Alpine Corsica orogenic system (modified after Bonini et al. [2014]) and traces of the CROP M12A and CROP-03 seismic profiles; (b) Interpreted profile along the CROP M12A and part of the CROP-03 seismic lines through the Northern Apennines (simplified after Finetti et al. [2001] and Bonini et al. [2014]). AA, Alpi Apuane; MP, Monti Pisani; MR, Monticiano-Roccastrada; MTR, Mid-Tuscan Ridge. development, and magmatism are believed to have occurred as part of a back-arc evolution in the hinterland of the belt, causing the opening of the Tyrrhenian Sea Malinverno & Ryan, 1986;Serri et al., 1993). Back-arc spreading was continuous in the Southern Tyrrhenian Sea from the early Miocene to present (Malinverno & Ryan, 1986). In the Northern Tyrrhenian Sea, back-arc extension is also generally considered continuous since the early Miocene, accompanied by a coeval migration of the thrust front toward the Adriatic sea Jolivet et al., 1998;Keller & Coward, 1996). In the Corsica Basin, syn-rift sequences suturing the relationships between Alpine Corsica and the Northern Apennines are dated to the Langhian-Messinian (Mauffret et al., 1999;Moeller et al., 2013). Extension in the hinterland of the Apennines east of the Corsica basin (including Elba) is believed to have commenced in the middle/ late-Miocene and has been linked to exhumation of the deep orogenic roots (e.g., Carmignani et al., 1994;Carmignani & Kligfield, 1990;Daniel & Jolivet, 1995;Jolivet et al., 1998;Keller & Coward, 1996). In this conceptual framework, some authors have argued that exhumation was in part syn-orogenic and associated with cold exhumation paths and compressional structures predating back-arc extension in the region (e.g., Jolivet et al., 1998;Rossetti et al., 2002Rossetti et al., , 1999Storti, 1995;Vignaroli et al., 2009). More recently, however, several studies have shown that extension in the Northern Apennines hinterland has been episodically interrupted by compressive pulses in the Late Miocene (e.g., Bonini et al., 2014;Bonini & Sani, 2002;Musumeci et al., 2008;Musumeci et al., 2015;Viola et al., 2018). At present, the Northern Apennines are characterized by active shortening at the easternmost front of the belt (Po Plain and Adriatic domains) coexisting with extension in the Northern Tyrrhenian Sea and internal sectors ( Figure 1a; Boccaletti et al., 2011;Faccenna et al., 2014;Pauselli et al., 2006).
The main thrusts and tectonic lineaments of the belt strike NW-SE to N-S and invariably verge to the east ( Figure 1a). In Figure 1b, we show one of the proposed interpretations of the deep lithospheric architecture of the Northern Apennines, adapted from the interpretation of CROP profiles by Finetti et al. (2001) and Bonini et al. (2014). This CROP profile shows that major reflectors (i.e. tectonic contacts) in the Northern Apennines are predominantly W-dipping and interpreted as regional-scale thrusts, as shown in Figure 1b. This model contrasts, however, with other interpretations of the CROP profiles, which tend to stress the presence of E-dipping low-angle normal fault systems (e.g., Pauselli et al., 2006). We consider the interpretation by Finetti et al. (2001) and Bonini et al. (2014) to be more in line with the actual geology of the Northern Tyrrhenian Sea as also depicted by other recent, high-resolution seismic profiles showing W-dipping reflectors, symmetric rifting, and no evidence of widespread, crustal-scale E-dipping low-angle normal faults (e.g., Moeller et al., 2013).
As mentioned, exhumation of the TMUs was interpreted both as post-orogenic in a scenario of metamorphic core complex exhumation (Carmignani & Kligfield, 1990;Carmignani et al., 1994) and as resulting from a combination of syn-and post-orogenic extension (Jolivet et al., 1998;Molli et al., 2018 et al., 2002). Some authors have stressed the importance of thrusting during the early stages of exhumation leading to inverted tectonic gradients, as in the Alpi Apuane or on the Island of Giglio, where upper greenschist-and blueschist-facies rocks are thrust on top of lower greenschist-facies units (Carosi et al., 2004;Molli et al., 2000;Rossetti et al., 1999;Storti, 1995). Daniel and Jolivet (1995) and Jolivet et al. (1998) proposed low-angle normal faults as key structures for the exhumation of metamorphic units and emplacement of late Miocene plutons within the orogenic wedge. However, most of these structures have only been recognized within the shallowest parts of the wedge, presumably implying only limited throws (in the order of a few km; e.g., Musumeci et al., 2015) along moderately to steeply dipping normal faults, and interpreted either as syn- (Clemenzi et al., 2014;Massa et al., 2017) or post-orogenic extensional structures (Clemenzi et al., 2015;Collettini & Holdsworth, 2004).

The Island of Elba
Zooming into the geological framework of the Island of Elba reveals a stack of W-dipping and E/NE-verging thrust nappes intruded by Late Miocene plutonic rocks and cut by high-and low-angle normal faults (Figures 2a and 2b;Bortolotti et al., 2001;Collettini & Holdsworth, 2004;Keller & Coward, 1996;Massa et al., 2017;Pertusati et al., 1993). The nappe edifice is defined by thrust sheets stacked along N-S striking, top-to-the E/NE thrusts and is folded by E/NE-verging folds with N-S to NNE-SSW trending axes. E-W trending lineations are the main linear fabric of the stack, consistent with the overall E-vergence of the belt (Figures 2a and 2b;Keller & Coward, 1996;Massa et al., 2017). The nappe stack can be subdivided internally into an Upper and a Lower Complex (Musumeci & Vaselli, 2012), separated by the out-of-sequence Capo Norsi-Monte Arco Thrust (CN-MAT by Viola et al. [2018]), which is marked by a slice of tectonized serpentinite (Serpentinite Unit in Figure 2a) sandwiched between continental units (Figure 2a).
The Upper Complex, in the hanging wall of the Serpentinite Unit, is composed of a sequence of nonmetamorphic to anchizone-facies Ligurian and Tuscan Units, intruded by Late Miocene plutonic rocks of the Monte Capanne Pluton and the Central Elba Laccolith Complex in the west, and the Porto Azzurro Pluton in the east (Figure 2a; Barboni et al., 2015;Bouillin, 1983;Dini et al., 2002). Regional metamorphism in the Upper Complex is only locally observed in the lower greenschist-facies Rio Marina Unit (Figure 2a; Elter & Pandeli, 2001;Franceschelli et al., 1986).
The Middle-Late Miocene was marked by the emplacement of large volumes of magma (Dini et al., 2002;Rocchi et al., 2010) up to shallow crustal levels in the nappe stack (P < 0.2 GPa; Duranti et al., 1992;Rossetti et al., 2007). Magma emplacement is generally interpreted as coeval with extensional tectonics, due to either gravitational collapse during intrusion (Daniel & Jolivet, 1995;Pertusati et al., 1993;Trevisan, 1950;Westerman et al., 2004) or regional crustal extension connected to the opening of the Northern Tyrrhenian Sea (Keller et al., 1994;Keller & Coward, 1996;Jolivet et al., 1998). Evidence in favor of this phase of extension includes high-and low-angle fault zones that crosscut the nappe pile. For instance, the Zuccale Fault (Figures 2a and 2b) is a subhorizontal fault zone with an eastward displacement of c. 6 km that has RYAN ET AL.
10.1029/2020TC006348 5 of 30 been almost invariably interpreted as a low-angle normal fault, accommodating significant post-orogenic extension and exhumation of the Porto Azzurro Pluton and the metamorphic Calamita Unit (Collettini & Holdsworth, 2004).
More recently, however, the extensional models have been questioned on Elba by the recognition of Late Miocene compressional structures shown to be coeval with magmatism and the associated LP-HT metamorphism. Mazzarini et al. (2011), Musumeci and Vaselli (2012), and Papeschi et al. (2017Papeschi et al. ( , 2018, for example, documented E-verging folds and top-to-the E shear zones with thrust kinematics that accompanied the development of the contact aureole of the Porto Azzurro Pluton. The CN-MAT itself was recognized as RYAN ET AL.   Massa et al. [2017]). The frame denotes the study area. (b) Cross-section (roughly oriented E-W) across the island. Note that the vertical scale is exaggerated by 2.5×. an out-of-sequence thrust that superimposes the nonmetamorphic Upper Complex over amphibolite-facies hornfelses of the Lower Complex (Massa et al., 2017;Musumeci & Vaselli, 2012;Viola et al., 2018). Consequently, the interpretation of the Zuccale Fault as an extensional fault has been questioned in favor of its interpretation as a flat segment of a thrust (Musumeci et al., 2015;Viola et al., 2018). K/Ar dating of authigenic illite on the CN-MAT, the Zuccale Fault, and similar thrusts in the Lower Complex showed that a phase of thrusting was active until the early Pliocene between 6.14 ± 0.64 Ma and 4.90 ± 0.27 Ma (Viola et al., 2018) and that the last recorded phase of extension of the nappe stack started in the Pliocene and was accommodated by high-angle normal faults (Mazzarini et al., 2019;Musumeci et al., 2015).

Materials and Methods
To propose a self-consistent model capable of incorporating the geological evidence presented below, our study has implemented a multidisciplinary approach combining field work and structural analysis, petrography, microstructures, fabric analysis by Electron Back-Scattered Diffraction (EBSD), and 40 Ar/ 39 Ar thermochronology. We conducted detailed structural-geologic mapping over the ∼2 km 2 area shown in Figures 3 and 4, selecting representative structural stations on well-exposed coastal outcrops of the Acquadolce Subunit. Thin sections cut parallel to the stretching lineation and shear direction (i.e., L p lineation) and perpendicular to the regional and mylonitic foliation (S p /S m ) were prepared from the main lithologies of the Acquadolce Subunit. Sample details are available in Table S1. Our EBSD analysis focused on calc-mylonitic material, as it best preserves the W-directed mylonitic shearing and kinematic indicators related to the deformation of the Acquadolce Subunit.
For the 40 Ar/ 39 Ar analysis, we selected a white mica-bearing sample (ELB5) from the Rio Marina outcrop (Area A in Figure 4). We chose this area because (1) it is located far away from the contact-metamorphosed rocks to the south and was less affected by thermal metamorphism, (2) the chemistry of the dated white mica is well known from companion samples from the same outcrop that was investigated by Papeschi et al. (2020), and (3) independent white mica 40 Ar/ 39 Ar dating was performed in the same area by Deino et al. (1992) and on glaucophane by Bianco et al. (2019). We also chose this specific sample because the white mica was relatively coarse-grained compared to other samples.
The analytical details of the EBSD analysis and 40 Ar/ 39 Ar dating are available in dedicated Appendixes A and B.
The Acquadolce Subunit is structurally bounded at its base by the Felciaio Shear Zone (FSZ; Musumeci & Vaselli, 2012) and at its top by the CN-MAT (Viola et al., 2018). The FSZ and CN-MAT are Late Miocene E-verging thrusts that accommodated shortening during and shortly after the emplacement of the Porto Azzurro Pluton at shallow crustal levels ( Figure 3). These thrusts sandwiched the Acquadolce Subunit between serpentinites at its top and the diopside-bearing LP-HT Ortano Marble at its base ( Figure 3). The contact-metamorphic overprint of the Porto Azzurro Pluton is spatially confined to the southernmost part of the Acquadolce Subunit (mostly south of Mt. Fico-Il Porticciolo; Figure 3). This thermal overprint is defined by distinctive cordierite and/or andalusite (Papeschi et al., 2020) statically overgrowing the lower grade schistose complex.

Structural Framework of the Acquadolce Subunit
The principal foliation (S p ) and internal lithological contacts in the Acquadolce Subunit strike NNE-SSW and dip between 10° and 50° to the W/SW (Figure 4), with only very limited along-strike variations causing local northward and southward dips (e.g., Figures 4c-4e). In the metasediments of the Acquadolce Subunit, the S p is defined by the metamorphic layering and the preferred orientation of greenschist-and blueschist-facies mineral grains (white mica, chlorite, quartz, calcite, albite, and Na-amphibole). Mineral lineations (L p ) on the foliation planes are defined by oriented aggregates and grains of quartz, calcite, mica, albite, Na-amphibole that trend from ENE-WSW to NE-SW with a general plunge between 10° and 40° ( Figure 4) to the SW and only locally to the NE (e.g., Figures 4a, 4b, 4d, and 4e). Metabasite lenses, hosted in metacarbonates, are weakly foliated and contain strongly retrogressed lawsonite + Na-amphibole + omphacite surrounded by chlorite + epidote + Ca-amphibole + titanite.
The northern part of the study area, north of the B-B′ cross-section trace in Figure 3, contains well-preserved blueschist-and greenschist-facies parageneses where the LP-HT contact metamorphic overprint is weak to absent. North of the Mt. Fico-Il Porticciolo alignment (Figure 3), top-to-the W deformation is common and well-preserved ( Figure 4)

Top-To-The W Deformation in the Acquadolce Subunit
To clarify the role of the mapped W-verging structures within the Acquadolce Subunit, we describe the mesoscopic structural framework of the Acquadolce Subunit and its microstructural characteristics in more detail below.

Mesostructures
The Acquadolce Subunit is a generally highly strained unit that preserves hitherto ignored, yet significant, top-to-the W deformation associated with penetrative, generally N-S striking and W-dipping fabrics (Figures 4 and 5). Higher strain domains occur therein as centimetric to metric, W/SW-dipping mylonitic structures, which are particularly well-developed and preserved in the metacarbonatic lithologies (marble and calcschist; Figures 5a-5c). Metapelitic lithologies, characterized by lower competence compared to the metacarbonates, preserve well-foliated structures with local, recumbent W-verging, noncylindrical isoclinal folds in metapsammitic and metacarbonatic interlayers (supporting information). Competent metabasite lenses within the metacarbonates form boudins and pinch-and-swell structures and sometimes exhibit fractured necks infilled by calcite + quartz + dolomite ± chlorite ± white mica ± epidote ± titanite veins ( Figure 5c). Asymmetric metabasite boudins display W-vergent asymmetries and are themselves segmented along W-SW-dipping, top-to-W shear planes ( Figure 6a).
Within the metacarbonatic lithologies, a gradual and continuous transition from intensely foliated and fine-grained mylonite (with a mylonitic foliation that we refer to as S m ) to relatively less foliated and coarser calcschist and marble is common (Figure 5b). We interpret the S m as the high-strain equivalent of the S p , as both foliations are parallel and defined by the preferred orientation of the same mineral assemblage, namely calcite, quartz, white mica, chlorite, biotite, albite, and Na-amphibole. Lineations throughout the RYAN ET AL.

10.1029/2020TC006348
10 of 30 Acquadolce Subunit are invariably defined by the same mineral phases and monotonously trend from E-W to SW-NE.
The S m in the mylonitic calcschist is defined by tightly spaced, sub-millimetric, subparallel to anastomosing layers of chlorite + Na-amphibole and carbonate minerals (Figure 6b). Foliation boudinage localizes in quartz and mica-rich layers that form neck structures filled by coarse-grained calcite (yellow arrow in Figure 6b). In places (e.g., area A-Rio Marina lighthouse, Figures 4 and 5b), the well-foliated mylonite contains centimetric to decimetric, highly transposed tight-to-isoclinal SW to NW-verging asymmetric folds with W-dipping axial planes, thickened hinges, and stretched limbs (Figure 6c). These folds exhibit strongly noncylindrical to sheath-like geometry and closed, elliptical fold surfaces (eye structures) are visible on N-S sections (Figure 6d   Top-to-the W kinematics in mylonite zones are reliably constrained by asymmetric, cm-scale intrafolial features including imbricated duplex-like stacks (Figure 6e), sigmoidal quartzitic, and carbonatic clasts (Figures 6e and 6f), and asymmetric mm-sized albite grains (see Section 6.2 below). C′-type shear bands (following Berthé et al., 1979) developed at low to very low angles to the S m and deform the mylonitic metacarbonates with displacements ranging from a few millimeters to some centimeters, as measured on displaced layers of metabasite, metachert, and metapsammite (Figures 6g and 6h).

Microstructures
To clarify the temporal relationships between blastesis and deformation in the Acquadolce Subunit, we propose a conceptual chart depicting relative mineral growth stages with respect to S m development (Table 1).
This indicative chart is based on numerous consistent microstructural observations from the present work and mineral chemistry data by Papeschi et al. (2020). Metamorphic parageneses of the investigated samples are given in Table S1.
At the microscopic scale, the mylonitic metacarbonates consist largely of a network of stretched and dynamically recrystallized calcite domains surrounding albite porphyroblasts and are interlayered with thin white mica + chlorite + Na-amphibole-bearing layers ( Figure 7a). Calcite-rich domains range in thickness from a few millimeters to some centimeters and in places contain oriented, subparallel grains of white mica, chlorite, and Na-amphibole, as well as lobate quartz and albite grains. The microstructure of calcite consists of larger relic grains (0.1-1 mm in size), mantled by fine-grained calcite and crosscut by top-to-the W shear bands defined by fine-grained (10-50 µm) dynamically recrystallized calcite grains (Figure 7b). Relic calcite grains show ameboid to lobate boundaries with recrystallized bulges (10-50 µm in size), intracrystalline undulose extinction, and bent mechanical e-twins (type III of Ferrill et al., 2004). In the most strained samples, relic calcite grains are strongly elongated parallel to the S m (e.g., Figures 7a and 7b). In others, they display a shape preferred orientation (SPO), defining an oblique foliation with respect to the S m that is invariably consistent with a top-to-the W sense of shear (Figures 7c and 7d).

Table 1
Chart of blastesis/deformation relationships for phases in the calcschists of the Acquadolce Subunit phase that crystallized at blueschist-facies conditions. Its preferred orientation indicates a syn-kinematic origin with respect to the formation of S m (Table 1). Chlorite occurs in gaps and pressure shadows around Na-amphibole and as Mn-rich grains in equilibrium with Na-amphibole rims (see Papeschi et al., 2020), indicating that it was present during greenschist-and part of the blueschist stages (Table 1). Na-amphibole occurs as prismatic and sigmoidal grains and mineral fishes (grain size: 0.1-1.0 mm) oriented and stretched parallel to the S m (Figures 8a-8d). Sigmoidal grains typically display a zoning from a crossite core to an Mn-rich glaucophane-rim (Figures 8a and 8b), which was interpreted by Papeschi et al. (2020) to mark the breakdown of Mn-bearing phases close to the peak metamorphic grade. Zoning is commonly asymmetric, with glaucophane-rich rims forming at the stretched edges of Na-amphibole grains (Figures 8a and 8b). Na-amphibole is stretched and locally boudinaged along its basal cleavage, forming symmetric to asymmetric boudins with necks filled by calcite and quartz, as well as albite, titanite, and chlorite (Figures 8c and 8d). Although boudinage indicates that Na-amphibole is generally a pre-kinematic or relic phase with respect to the S m (Table 1), the presence of sigmoidal glaucophane fish with asymmetric zoning patterns (Figures 8a  and 8b) may suggest that these grains have also partly grown along the S m during shearing.
Albite forms anhedral to euhedral syn-kinematic porphyroblasts up to 1-2 mm in size (Table 1; Figures 9a-9c). The S m foliation is included and curved within albite porphyroblasts, where it is defined by oriented inclusions of calcite, quartz (Figures 9a and 9b), Na-amphibole (Figure 9c), white mica, chlorite, epidote (Figure 9d), and titanite. Its anticlockwise rotation with respect to S m is consistent with a top-to-the W sense of shear. Na-amphibole and lawsonite pseudomorph inclusions in albite are partially to entirely resorbed and preserved as pseudomorphs (Figure 9c  grains indicates that in the surrounding matrix this phase was likely destabilized and obliterated during mylonitic deformation. The phyllosilicate-rich matrix contains euhedral to subhedral albite (grain size: 100-500 µm), epidote (10-200 µm), and titanite (10-100 µm) grains. In some cases, these grains contain gently rotated or deflected internal foliation planes (e.g., Figure 9e) and are, therefore, syn-kinematic with respect to the S m . In others, they overgrow the S m and lack evident pressure shadows and can thus be interpreted as post-kinematic (e.g., Figure 9f; Table 1). As shown by Papeschi et al. (2020), titanite sometimes occurs around relic, partially resorbed rutile grains that we interpret to have been stable at blueschist-facies (Table 1)

EBSD Analysis
We used EBSD to generate spatially oriented data of the crystallographic preferred orientation (CPO) within the Acquadolce Subunit mylonitic calcschists. These data help to further describe the top-to-the W shear sense of these mylonites, but also to document their complex deformation history. The sample discussed herein (IESP3ERY21 A) is representative of the deformation style of all studied calc-mylonites. It is a mylonitic calcschist from the Rio Marina pier outcrop (Figure 3), where the HP-LT parageneses have been extensively documented (Bianco et al., 2015(Bianco et al., , 2019Papeschi et al., 2020). At the outcrop and hand specimen scale, an overall ductile deformation style dominates and is associated with top-to-the W kinematic indicators (Figure 10a).

RYAN ET AL.
10.1029/2020TC006348 16 of 30 The EBSD analysis was performed on a domain of dynamically recrystallized calcite characterized by a strong oblique SPO with respect to the S m (Figure 10b). The resulting EBSD map is shown in Figure 10b, portraying calcite and quartz (dark gray) as indexed phases. We divided calcite grains into twin host (blue), twin lamellae (red), and untwinned (light gray) grain fractions. In total, 2,309 untwinned, 560 twin hosts, and 641 twin lamellae grains were indexed. axis were generated as one-point-per-grain equal area, upper hemisphere projections, with X = lineation and Z = pole to the foliation orientations with respect to the finite strain ellipsoid reference frame (Figure 10d).
The mapped calcite grains have an average equivalent radius of 4.35 µm (p50 value) and exhibit lobate to ameboid grain boundaries (Figure 10b). Low-angle boundaries (misorientation threshold value: <10°) crosscut the calcite grains, dividing them into elongated sub-grains, as shown by varying IPF color within single grains (Figure 10c). The grains define a moderate SPO, inclined at 10°-15° to S m , which is parallel to the bottom border of the frame (Figure 10b). The grains defining the SPO have an average aspect ratio of 2.0-2.5 and constitute most of the grain population. Grains oriented at high angle to S m typically have a lower aspect ratio of c. 1.4-2.0 (Figure 10e). This inclined SPO is consistent with a top-to-the W sense of shear. Twinned grains are distributed throughout the sample, containing e-twins (approximate width 10 µm) oriented at large angles to S m (Figure 10b). The high frequency of misorientation angles related to e-twinning is shown in Figure 10f.
The <c>-axis pole figures of the untwinned, twin host, and twin lamellae grain fractions are shown in Figure 10d. The distribution of the c <0001>-axis for the untwinned grain fraction defines a broad cluster around the pole to the S m foliation (Z; Figures 10d) and 10a weak a <21 10> a-axis girdle distributed parallel to the XY foliation plane (Figure 10d). The <0001> c-axis maximum is only slightly rotated (10°-15°) in a clockwise direction with respect to and is roughly perpendicular to the SPO (Figure 10d). The CPO of the twin grains has been divided into host and lamellae fractions. The twin host grains define highly oblique c-axis girdle, rotated ∼60° clockwise from the pole to the S m . The lamellae also have a strong c-axis maximum, but with a slight anticlockwise rotation of ∼10° from the pole to S m . This is consistent with top-to-the E deformation via e-twinning (e.g., Schmid et al., 1987;Wenk et al., 1986), which occurs at low temperatures (i.e., <250°C-300°C; Burkhard, 1993;de Bresser & Spiers, 1993Wenk et al., 1986). Neither the r-or f-poles of the host or lamellae fractions have a significantly preferred orientation, while the e-poles and a-axes of the lamellae show a weak, girdle-like CPO oriented normal and parallel to S m , respectively (Figure 10d).

Geochronology
To obtain geochronological constraints on the exhumation of the Acquadolce Subunit, we collected one sample of calcschist (ELB5; Figures 11a) from the Rio Marina area for 40 Ar-39 Ar dating (Figures 11b and 11c; see Figure 3 and raw data in the supporting information). The sample is from the same Rio Marina pier outcrop where the EBSD sample was collected. There is no sign of contact metamorphism at this locality, which is also where peak HP-LT conditions were documented by Bianco et al. (2015Bianco et al. ( , 2019 and Papeschi et al. (2020). Those authors reported celadonite-rich white mica with Si p.f.u. contents invariably above 3.00, up to 3.60 p.f.u. per 11 oxygens formula unit, a composition indicating HP metamorphism (Massonne & Schreyer, 1987;Massonne & Szpurka, 1997;Vidal & Parra, 2000). Phase equilibria modeling links the white mica phase to the P > 1.5 GPa mineral assemblage (Papeschi et al., 2020). We here report white mica compositional analyses from sample IESP3SP53 after Papeschi et al. (2020) and integrate them with analyses of a second sample (IESP3SP54) from the same part of the outcrop as ELB5 (Table S1). Analyses constrain an HP white mica with a Si p.f.u. content in the range of 3.30-3.60 and Al p.f.u. ∼1.80-2.00, based on 11 O basis (Figure 11d).
ELB5 is a strongly foliated calc-mylonite with relic calcite grains elongated parallel to S m . Mantles of finer and dynamically recrystallized new grains surround the relic calcite grains (Figure 11a). Chlorite and white mica grains are aligned along the foliation planes and quartz and albite locally form sub-millimetric, subrounded clasts and elongated laths. The sample contains sigmoidal clasts and mica fish indicating top-tothe W kinematics (Figure 11a).
White mica in sample ELB5 yielded a 7-step 40 Ar/ 39 Ar plateau age at 20.99 ± 0.73 Ma (MSWD = 0.221; Figure 11b) corresponding to more than 50% of the cumulative released 39 Ar. The K/Ca ratios (calculated from 39 Ar k / 38 Ar Ca ) over the concordant steps show some variation, but any potential age variations cannot be resolved in this analysis. The plateau age is in good agreement with the inverse isochron analysis (Fig-RYAN   ure 11c), which also precludes excess argon. We use the plateau age as the best age estimate of this sample in the following discussion.

W-Verging Structures and Their Role During Exhumation
The presented structural, microstructural, EBSD, and geochronological data imply a tectonic history that we cannot readily reconcile with the existing post-orogenic exhumation models connected with the opening of the Northern Tyrrhenian Sea. Previous models specific to the Island of Elba suggested that exhumation of its metamorphic units occurred as part of an overall E-verging extensional structural framework, wherein initial top-to-the E thrusting was subsequently overprinted by top-to-the E syn-to post-orogenic low-angle normal faulting (e.g., Brunet et al., 2000;Collettini & Holdsworth, 2004;Jolivet et al., 1998;Keller & Coward, 1996).
While obviously maintaining that the E-vergence is fundamental within the nappe stack, we show that structures with typical Apenninic orientation in the Acquadolce Subunit (N-S to NW-SE striking and W-dipping foliation; Figure 4) diffusely also record W-directed shearing, that is, opposite to the first-order NE-directed thrusting kinematics that characterizes the Northern Apennines (e.g., Boccaletti et al., 1971;Keller & Coward, 1996). For the current orientation of the entire nappe stack within which the Acquadolce Subunit occurs, W-directed deformation corresponds to top-to-the W extensional shearing. We are not aware of any major, orogen-scale structure that may have caused large rotation of the nappe stack or parts thereof and therefore conclude that top-to-the W asymmetries represent overall extension. Although the evidence for this W-vergent kinematic framework is not ubiquitous and is also locally strongly overprinted, it needs to be accounted for in any model aiming to reconstruct the evolution of the area and, in turn, of the Northern Apennines.
We propose that exhumation of the Acquadolce Subunit may have initiated during a deformation phase including syn-orogenic, extensional top-to-the W shearing in a progressively forming E-verging nappe stack in the subduction channel. Indeed, many kinematic indicators in the Acquadolce Subunit are invariably consistent with top-to-the W deformation (Figures 6-9). Greenschist-facies assemblages are syn-kinematic with respect to the top-to-the W shearing and overprint mostly relic blueschist-facies parageneses that formed at peak metamorphic conditions. The general lack of kinematic indicators that can be directly ascribed to the initial HP evolution of the nappe stack inhibits the direct and confident link between blueschist-facies mineral phases and top-to-the W deformation. However, the widespread presence of celadonitic white mica defining the main foliation ( Figure 11d; Papeschi et al., 2020) and independent documentation by Bianco et al. (2015Bianco et al. ( , 2019 and the sigmoidal shape of Na-amphibole with asymmetric zoning patterns (Figures 8a and 8b), indicate that the early development of the exhumation-related S m foliation started under blueschist-facies conditions. The oblique foliation in the calc-mylonites, defined by a pervasive calcite SPO with respect to S m (Figure 10e), is in agreement with top-to-the W kinematics. This sense of shear is possibly supported by the CPO of the untwinned calcite grains as indicated by the slight clockwise rotation of the c-axis maxima (Figure 10d; Rogowitz et al., 2014), although c-axis fabrics in calcite can be difficult to interpret with regards to the sense-of-shear (see the interpretations of Trullenque et al. [2006] vs. Rogowitz et al. [2014]). We interpret the e-twinning as evidence of later deformation, overprinting the earlier top-to-west fabric. This interperation is consistent with the lower temperatures documented for e-twinning in calcite (i.e., <250°C-300°C; Burkhard, 1993;de Bresser & Spiers, 1993Wenk et al., 1986). 40 Ar/ 39 Ar dating of celadonitic white mica from sample ELB5 sets a minimum age for the HP-LT parageneses to 20.99 ± 0.73 Ma (Figure 11b). This age is identical within error to the 19.8 ± 1.4 Ma 40 Ar/ 39 Ar age of syn-kinematic glaucophane from the same locality (Bianco et al., 2019) and the 19.68 ± 0.15 Ma 40 Ar/ 39 Ar age of white mica from the calcschist in the Rio Marina area (Deino et al., 1992). Bianco et al. (2019) associated the dated glaucophane with a retrograde metamorphic mineral assemblage, implying that their age constrains the ongoing exhumation of the HP-LT rocks. We also interpret our age to mark a post-HP metamorphic peak exhumation stage, likely the cooling below the closure temperature for the 40 Ar/ 39 Ar system in white mica. The maximum age for HP metamorphism can be constrained to 31.6 ± 0.5 Ma (e.g., Jacobs et al., 2018), which is the maximum deposition age of the youngest deposits that were subject to blueschist-facies metamorphism.

RYAN ET AL.
10.1029/2020TC006348 20 of 30 According to Papeschi et al. (2020), the exhumation of the HP-LT blueschists of the Acquadolce Subunit occurred at nearly isothermal conditions from peak conditions at 1.5-1.8 GPa and 320°C-370°C. Isothermal decompression occurred between 21-19 Ma (HP-LT metamorphism) and 6-7 Ma (contact metamorphism at 0.2 GPa; Musumeci et al., 2015) at very fast exhumation rates (2.5-3.7 km/Ma; see Papeschi et al., 2020). Such a cold exhumation path is consistent with syn-orogenic exhumation (see also Rossetti et al., 2002). Therefore, we suggest that during the Oligocene-early Miocene W-verging extensional structures within the east-verging and west-dipping Northern Apennines nappe stack played a significant role in exhuming deep metamorphic units from blueschist-to greenschist-facies conditions.

Extrusion Model of the Acquadolce Subunit: Implications for Exhumation of HP-LT Units in the Northern Apennines
The Northern Apennines contain several arcuate, antiform-to dome-like structures, the cores of which consist of metamorphic rocks currently cropping out below subgreenschist-facies units (Figure 1a). These geological features were initially interpreted as metamorphic core complexes (Carmignani & Kligfield, 1990), and exhumation was proposed to have mainly occurred during syn-to post-orogenic extension in the footwall of major, E-verging low-angle normal faults that bound the complexes (Carmignani et al., 1994(Carmignani et al., , 2001(Carmignani et al., , 2004Daniel & Jolivet, 1995;Jolivet et al., 1998;Rossetti et al., 1999;Pauselli et al., 2006). HP-LT parageneses with peak pressure >1.0 GPa therein (Theye et al., 1997;Giorgetti et al., 1998;Rossetti et al., 1999) have, however, challenged the post-orogenic extension model. Jolivet et al. (1998) and Rossetti et al. (2002), who argued that the exhumation was accommodated by post-orogenic back-arc extension, thus also suggested that HP-LT metamorphic rocks had in part been exhumed along cold geothermal gradients, typical of syn-orogenic exhumation paths. Such a two-stage evolution is supported by data from the Alpi Apuane area (Figure 1a), where "cold," syn-orogenic exhumation occurred between 25 and 13 Ma, while the overlying Tuscan Nappe was still being underthrusted (Fellin et al., 2007;Molli et al., 2018). In the Alpi Apuane, higher-grade rocks of the Massa Unit (P > 0.8 GPa) are structurally above lower-grade units (Apuane Unit; P ∼ 0.4-0.6 GPa), suggesting that the Massa Unit was thrust onto the Apuane Unit at some stage during exhumation (Carosi et al., 2004;Jolivet et al., 1998;Molli et al., 2000). W-verging folds in the area were also interpreted to have formed during exhumation (Molli et al., 2000;Storti, 1995).
Despite this first-order evidence, the mechanisms and structures driving and accommodating syn-orogenic exhumation have proved difficult to document and remain largely unexplored and unaccounted for in the Northern Apennines belt. The data presented here constrain instead widespread W-verging extensional shearing within overall E-vergent thrust nappes in the Tyrrhenian sector of the belt, accommodating exhumation of the Acquadolce Subunit, likely from blueschist-to greenschist-facies conditions. Due to its minor structural thickness of just 500-700 m, the Acquadolce Subunit only represents one slice of HP-LT rocks preserving evidence of early, W-verging structures within the orogenic wedge, which was largely reworked by Late Miocene tectonics and magmatism. We argue that W-directed extension was coeval with E-directed thrusting, thus driving the exhumation of the Acquadolce Subunit by ductile extrusion from blueschist-to greenschist-facies conditions in an overall contractional setting (e.g., Law et al., 2006;Searle, 2010;Lamont et al., 2019). Exhumation by syn-orogenic extrusion is supported by the nearly isothermal P-T path recorded by the Acquadolce Subunit (see Papeschi et al., 2020) and similar units in the region (Jolivet et al., 1998;Rossetti et al., 1999Rossetti et al., , 2001Rossetti et al., , 2002. Extrusion may also have exhumed the Ortano and Calamita Units from HP-LT conditions. However, late Miocene LP-HT contact metamorphism therein has largely obliterated older assemblages and structures and the early Miocene-Oligocene metamorphic peak conditions experienced by the Ortano and Calamita Units remain unknown. The common top-to-the E thrusts and multiscalar asymmetries that pervasively define the current upper and lower boundaries of the Acquadolce Subunit are significantly younger than the peak HP-LT paragenesis (Musumeci et al., 2015;Viola et al., 2018) and are coeval with Porto Azzurro monzogranite-related contact metamorphism at P < 0.2 GPa (Musumeci & Vaselli, 2012). As such, they represent late orogenic structures and cannot be directly associated with the W-vergent geological record presented here.
Based on existing and our new structural and metamorphic data, we propose an evolutionary scheme for the Acquadolce Subunit that has the potential to account for the syn-orogenic exhumation of HP-LT rocks in this portion of the inner Northern Apennines. We implement our constraints in a model that integrates RYAN ET AL.
10.1029/2020TC006348 21 of 30 the geodynamic evolution for the Northern Apennines proposed by Bonini et al. (2014; Figure 12) and use our results to better constrain the processes that took place in the subduction channel at depths between ca. 50 and 20 km from the late Oligocene to the late Miocene ( Figure 12). According to Bonini et al.'s model, the overall late Oligocene-late Miocene evolution shown in Figures 12a-12c can be placed within a framework of progressive Adria slab roll-back, retreat, and delamination. However, in the context of regional upper-plate extension associated with this evolution, late Miocene crustal shortening (Figure 12c) would have ensued in response to a transient stop in the subduction process. Such a stop would have allowed for coupling between the upper and lower plates causing the transfer of compression to the formerly extending upper plate, including the Northern Apennines hinterland and the Tyrrhenian back-arc basin. Viola et al. (2018) have indeed documented compressional pulses in the late Miocene on the Island of Elba.
The documented tectonic history of the Acquadolce Subunit can be readily framed into this regional picture by referring to the three conceptual stages described by Bonini et al. (2014), as shown in Figure 12. (1) During the late Oligocene-early Miocene, the Acquadolce Subunit was subducted to depths corresponding to 1.5-1.8 GPa by deep underplating beneath the early Apennine orogenic wedge. In our model, this occurred under overall compression during the structuring of the orogenic wedge through nappe stacking (Figure 12a and detail of the subduction channel); (2) during the early middle Miocene (Figures 12a and 12b) the Acquadolce Subunit experienced partial exhumation by syn-orogenic extrusion within the subduction channel to shallower crustal levels, driven by the W-verging structures at the top, documented here, and the inferred E-verging structures at the bottom. This stage likely occurred at c. 19-21 Ma (minimum age of HP-LT metamorphism), and certainly before 6-7 Ma, the age of the Porto Azzurro contact metamorphic overprint at 0.2 GPa. Assuming that exhumation occurred in this age interval, exhumation rates can be constrained to c. 2.5-3.7 km/Ma. We propose that while exhumation at depth was by upward extrusion within an overall contractional setting, extension dominated the uppermost wedge, which was responding to the dynamics of the critical taper (e.g., Massa et al., 2017); (3) in the late Miocene, the hinterland was recompressed and deformation was mostly accommodated by E-verging thrusts within the orogenic wedge leading to final nappe stacking (Figure 12c). This late Miocene event (Massa et al., 2017) is constrained by direct dating of thrusts and syn-kinematic contact metamorphism down to at least 6 Ma (Musumeci et al., 2015;Viola et al., 2018) and occurred at P < 0.2 GPa (Musumeci & Vaselli, 2012).
In this reconstruction, we do not show the Zuccale Fault, whose brittle top-to-the E shearing was dated to < 6 Ma (Viola et al., 2018), after the end of the time window considered in this study. According to our model, the reconstructed structural and metamorphic evolution of the Acquadolce Subunit occurred entirely in an overall contractional scenario, with syn-orogenic extrusion followed by late Miocene thrusting, leading to the incorporation of HP-LT metamorphic units into the shallower portions of the orogenic wedge. Extension occurred in the middle Miocene, but we exclude that it played a major role in the exhumation of the HP units studied here.
In light of similar "cold" exhumation P-T paths recorded by other HP-LT units in the Northern Apennines (e.g., Jolivet et al., 1998;Rossetti et al., 1999Rossetti et al., , 2002Theye et al., 1997), we propose that syn-orogenic extrusion facilitated by distributed W-verging extensional shearing was not only limited to the Island of Elba, but possibly played an important role to facilitate the exhumation of other metamorphic units in the Tyrrhenian sector of the belt (e.g., Gorgona, Giglio and Argentario; e.g., Brogi & Giorgetti, 2012;Giorgetti et al., 1998;Jolivet, 1998;Rossetti et al., 1999Rossetti et al., , 2001Theye et al., 1997). Indeed, a belt of partially preserved HP-LT metamorphic rocks in the Northern Tyrrhenian Sea can be traced from the Island of Gorgona in the north through the Island of Elba to the Island of Giglio and the Argentario Promontory in the south. These HP-LT rocks may testify to the importance of syn-orogenic extrusion in the innermost portion of the Northern Apennines. The effective thickness of the extruding HP-LT unit might therefore be significantly larger than that of the Acquadolce Subunit. The base of a thicker extruding unit may be defined by E-vergent thrusts, coeval to and with a similar metamorphic grade to the extensional W-vergent fabrics described here. Considering the geometry of the HP-LT metamorphic rocks in the Northern Tyrrhenian Sea, the true base of the extruding HP-LT unit is likely located farther east, toward the Italian mainland. Although more accurate radiometric data are needed to better constrain HP-LT events in the Northern Apennines, the data presented here suggest that the Adria continental margin, together with remnants of the Ligurian Ocean, was RYAN ET AL.  involved in the late Oligocene-early Miocene W-dipping subduction, after the Eocene continental collision in the Western Alps-Alpine Corsica.

Conclusions
We have conducted a detailed structural analysis of the Acquadolce Subunit, a tectonic sliver preserving an HP-LT signature within the Eastern Elba nappe stack. Our results show, for the first time, that diffuse top-to-the W extensional shearing affected the subunit as it was being incorporated into the progressively growing orogenic wedge characterized by an overall E-vergence. The documented top-to-the W extensional shearing allowed the overall upward extrusion of the Acquadolce Subunit as a discrete sliver. The subunit's exhumation relative to the surroundings involved its detachment from the subducting slab and upward migration within the subduction channel during continued subduction. Transition from peak blueschist conditions of 1.5-1.8 GPa and 320°C-370°C at least c. 21-19 Ma ago to greenschist-facies conditions reflects localized exhumation through ductile extrusion in the overall contractional setting of the Apennines orogenic wedge, such that the documented exhumation is purely syn-orogenic. Late Miocene, E-verging compressive tectonics, contact metamorphism, and later extension reworked the wedge to its current complex geometric configuration, obliterating much of the evidence supporting exhumation by extrusion during the early stages of wedge build-up.
Future research needs to be directed toward determining whether the mechanism proposed for the local exhumation of the Acquadolce Subunit can be realistically applied to other HP-LT metamorphic units of the inner Apennines belt. It is increasingly accepted that no single exhumation mechanism controls the exhumation style of any particular tectonic domain. Local geological peculiarities may lead to exhumation accommodated by structural styles that change in space and time within the same subduction system. More multidisciplinary studies combining detailed structural, petrological, and thermo-geochronological data are needed to explain the occurrence of HP-LT units in the inner Apennines, where it is clear that post-orogenic extension cannot account for all exhumation.
Many of the complexities on the Island of Elba are due to the Late Miocene intrusions of the Tuscan Magmatic Province, which have partly fully overprinted the geological record of the early stages of exhumation. Studies of HP-LT units elsewhere in the belt may thus be even more telling about the early increments of their exhumation path.

Appendix A: EBSD Analysis
Selected samples were chemically polished using an OP-U, 400Å suspended colloidal silica polishing agent and rinsed in 96% alcohol (Moen et al., 2003). Samples were mounted on metal mounting pedestals, at the standard eccentric height, using a tilt angle of 70° with respect to the horizontal (Prior et al., 2009) and a working distance of 21.9 mm. To avoid geometric trapezoidal scan distortions in the collected maps, a correction to the beam scan voltages was implemented using a gridded standard calibration (Sørensen et al., 2020). To acquire optimal EBSD patterns, the SEM vacuum was set on a variable pressure mode, low vacuum setting, and an accelerating voltage of 20 keV, using an extraction voltage of 1.70 kV and open apertures. EBSD patterns were acquired with the NORDIF UF-1000 detector (Chen et al., 2012) using a 240 × 240 pixel resolution and the NORDIF 3.0 software for pattern acquisition. The samples were analyzed using a step size of 2 µm to allow for the mapping of large areas and a significant number of grains. Acquisition and calibration patterns were selected in the corners and at the center of the scan area using the NORDIF interface and electron image, to calibrate the pattern center across the large area scans.
The data were processed using the open-source Matlab toolbox MTEX 5.2.2 (Bachmann et al., 2010(Bachmann et al., , 2011Mainprice et al., 2015). Noise reduction was performed to eliminate mis-indexed points and artifacts. During indexing, grain boundaries between calcite grains are often misinterpreted as dolomite (Sørensen et al., 2020). Therefore, we removed false dolomite grains. During grain reconstruction, unindexed areas less than 20 pixels in area were ignored to avoid tiny holes inside the calculated grains. The threshold for high-angle grain boundaries was set at 10°. RYAN ET AL.

10.1029/2020TC006348 24 of 30
We used the hexagonal unit cell with cell parameters c = 17.064 Å and a = b 4.99 Å to index the calcite EBSD data. E-twin boundaries were identified as grain boundaries, when the angle between either of the two ideal e-twins and observed misorientations was less than 8°. The definition of the e-twin is the reflection across the {−1 0 1 8} plane, which in EBSD translates to the rotation operation of 78.1083° around either the (1 −1 0 4) or the (0 1 −1 4) poles. To differentiate twin lamellae and host grains, we used MTEX grain shape parameters. The widest host grain lamella was selected as the host seed. The other host lamellae were then selected based on their misorientation to the host seed (<15°). The lamella with a misorientation to the host seed of less than 15° from the ideal e-twin misorientation was then classified as e-twin lamella.

Appendix B: 40 Ar/ 39 Ar
The selected sample was crushed and sieved to obtain 180-250 and 80-120 μm grain fractions. Finer particles were decanted in tap water, while the coarser residue was treated with several ultrasonic baths in acetone and deionized water. The optically best grains, void of any coating, were handpicked under a microscope. The grains were packed in aluminum capsules together with a Taylor Creek Rhyolite (TCR) flux monitor standard along with zero age reagent grade K 2 SO 4 and optical grade CaF 2 salts for interference corrections. The sample was irradiated at IFE (Institutt for Energiteknikk) for about 70 h, with a nominal neutron flux density of c. 1.3 × 10 13 n × (cm −2 × s −1 ). The sample was placed in a 3.5 mm pit size aluminum sample disk and step heated using a defocused 3.5 mm CO 2 laser beam from Photon Machine Fusions 10.6 with a flat energy spectrum. The gases extracted from the sample cell were expanded into a Piston Free Stirling Cryocooler to trap potential water into a two-stage low volume extraction line (c. 300 cm 3 ). Both stages were equipped with SAES GP-50 (st101 alloy) getters, the first running hot (c. 350°C) and the second running cold. They were analyzed with a MAP 215-50 mass spectrometer in static mode, installed at the Geological Survey of Norway. The peaks and baseline (AMU = 36.2) were determined during peak hopping for 10 cycles (15 integrations per cycle, 30 integrations on mass 36 Ar) on the different masses ( 40-36 Ar) on a Balzers SEV 217 electron multiplier in analog mode and regressed back to zero inlet time. Blanks were analyzed for every third measurement. After blank correction, a correction for mass fractionation, 37 Ar and 39 Ar decay and neutron-induced interference reactions produced in the reactor was undertaken using in-house software AgeMonster, written by M. Ganerød. The software implements the equations of McDougall and Harrison (1999) and the newly proposed decay constant for 40 K after Renne et al. (2010). A 40 Ar/ 36 Ar ratio of 298.56 ± 0.31 from Lee et al. (2006) was used for the atmospheric argon correction and mass discrimination calculation using a power law distribution of the masses. We calculated J-values relative to an age of 28.619 ± 0.036 Ma for the TCR sanidine flux monitor (Renne et al., 2010). We define a plateau according to the following requirements: at least three consecutive steps overlapping at the 95% confidence level (1.96σ) using the strict test: ≥50% cumulative 39 Ar released, and mean square of weighted deviates (MSWD) less than the two-tailed student T critical test statistics for n − 1. Weighted mean ages are calculated by weighting on the inverse of the analytical variance. The uncertainties are expanded in cases where MSWD > 1 using   MSWD to account for this excess error contribution.

Data Availability Statement
Data sets and Supporting Information can be accessed at Viola, Giulio (2021), " Ryan et al., 2021", Mendeley Data, V2, http://dx.doi.org/10.17632/bj3nwk3wpm.2. Samuele Papeschi during some of his fieldwork in the area and for constructive feedback. The authors are indebted to Gianluca Vignaroli and Francesco Giuntoli (Università di Bologna) for many constructive discussions in the field and in the office on the metamorphic complexes of the Northern Apennines and on the mechanisms steering their exhumation. Critical and constructive reviews by two anonymous reviewers, Michele Zucali, Alberto Vitale Brovarone, Associate Editor Federico Rossetti, and Editor Laurent Jolivet, have contributed to a much-improved final version of the study.