Structural characterization and K-Ar illite dating of reactivated, complex and heterogeneous fault zones: Lessons from the Zuccale Fault, Northern Apennines

. We studied the Zuccale Fault on Elba Island, Northern Apennines, to unravel the complex deformation history that is responsible for the remarkable architectural complexity of the fault. The ZF is characterized by a patchwork of at 20 least six distinct, now tightly juxtaposed Brittle Structural Facies (BSF), that is, volumes of deformed rock characterized by a given fault rock type, texture, color, composition, and age of formation. ZF fault rocks vary from massive cataclasite to foliated ultracataclasite, from clay-rich gouge to highly sheared talc phyllonite. Understanding the current spatial juxtaposition of these BSF’s requires tight constraints on their age of formation during the ZF life span to integrate current fault geometries and characteristics over the time dimension of faulting. We present new K-Ar gouge dates obtained from 25 three samples from two different BSF’s. Two top-to-the E foliated gouge and talc phyllonite samples document faulting in the Aquitanian (c. 22 Ma), constraining E-vergent shearing along the ZF already in the earliest Miocene. A third sample constrains later faulting along the exclusively brittle, flat-lying principal slip surface to < c. 5 Ma. The new structural and geochronological results reveal an unexpectedly long faulting history spanning a ca. 20 Ma long time interval in the framework of the evolution of the Northern Apennines. The current fault architecture is highly heterogeneous as it formed 30 at very different times under different environmental conditions during this prolonged history. We propose that the ZF started as an Aquitanian thrust which then became selectively reactivated by early Pliocene out-of-sequence thrusting during the progressive structuring of the Northern Apennines wedge. These results require the critical analysis of existing geodynamic models and call for alternative scenarios of continuous convergence between the late Oligocene and the early Pliocene with a major intervening phase of extension in the middle Miocene allowing for the isostatic re-equilibration of the Northern Apennines wedge. Extension started again in the Pliocene and is still active in the innermost portion of the Northern Apennines. In general terms, long-lived, mature faults can be architecturally very complex. Their unraveling, including understanding the dynamic evolution of their mechanical properties, requires a multidisciplinary approach combining detailed structural analyses with dating the deformation events recorded by the complex internal architecture, which is a phenomenal archive of faulting and faulting conditions through time and in space. 100 freeze-thaw cycles, and clays were suspended in 100 deionized water. Sedimentation with Stokes law, and a combination of continuous flow and fixed angle rotor centrifuges were used to generate particle sizes of <0.1, 0.1–0.4, 0.4–2, 2–6 and 6– 10 μm. Argon was extracted from clay aliquots packed in Mo foil in a stainless steel ultrahigh vacuum line using a Pond Engineering furnace at 1400 °C. The evolved sample gas was purified in a first stage using a titanium sublimation pump, and in a second stage with one SAES GP50 getter at room temperature and one SAES GP50 getter at 350 °C. Sample gas was spiked with approximately 2*10 − 13 105 moles of pure 38 Ar (Schumacher, 1975) and analyzed in an IsotopX NGX multicollector noble gas mass spectrometer fitted with five faraday cups for 600 integrations of 1 s each. Nominal argon beam intensities were determined by using a degree 2 polynomial regression to gas inlet time zero, with an in‐house Python program. Beam intensities for 38 Ar and 36 Ar were corrected for mass discrimination relative to 40 Ar by a power law (e.g., Renne et al., 2009), using the weighted mean 40 Ar/ 36 Ar ratios of 299.781 ± 0.014 measured from atmospheric argon in an online air pipette, and the atmospheric 110 argon composition of Lee et al. (2006). Radiogenic 40 Ar* concentrations and their uncertainties were calculated using the equations outlined in Hałas and Wójtowicz (2014) . Within this analytical batch, four aliquots of GA‐1550 biotite (98.5 ± 0.5 Ma; Mcdougall and Wellman, 2011) were analyzed and yielded a weighted mean age of 98.53 ± 0.36 Ma. Potassium concentrations were determined by fusing an aliquot of approximately 50 mg in lithium tetraborate at ~1000 °C to form a glass, which was then dissolved in HNO 3 with a rhodium internal standard at 5 ppm prior to analysis with a 115 Perkin Elmer Optima 4300 DV ICP‐OES. K‐Ar dates were calculated using the 40 K abundance and decay constants of Steiger and Jäger (1977). XRD analyses were carried out at the Academic Laboratory of Basin Analysis (ALBA) of Roma Tre University (Italy) with a Scintag X1 X- ray system (CuKα radiation). The tube current and the voltage were 30 mA and 40 kV, respectively. Randomly oriented whole-rock powders of all subfractions were X-rayed with a step size of 0.05° 2θ and a counting time 120 of 3 s/step in the 2- 70° 2θ range. Standard patterns for illite 1M d and 2M 1 polytypes were represented by the pure illites PDU and SG4 (Eberl et al., 1987), respectively. Integrated peak areas of minerals and illite polytype reflections were transformed into mineral concentration by using mineral intensity factors as a calibration constant (for a review, see Moore and Reynolds, 1997). The estimated quantification error is ± 5%. Transmission Electron Microscope (TEM) imaging and energy dispersive spectroscopy analysis were performed at the 125 NORTEM infrastructure at the TEM Gemini Centre of the Norwegian University of Science and Technology (NTNU) in Trondheim, Norway. Samples were dispersed at a concentration of 10 ppm by weight in deionized water, and dried on a carbon-supported TEM grid. Images were obtained on a JEOL JEM-2100 TEM with a LaB6 filament and chemical information was collected using an Oxford X-Max 80 SDD EDX. 3 heterogeneous core; 5°) principal slip surface (PSS). Pervasive subhorizontal slickenlines and crystal fiber lineations and associated asymmetric structures are consistent dominant top-to-the E to NE slip; foliated NW-dipping foliation preferred phyllosilicates clay-rich gouge; gouge cohesive foliated cataclasite in the immediate footwall of ZF striated and subparallel striated planes gently W-plunging slickenlines are associated with slickensides and other kinematic indicators S/C type shear bands in the foliated cataclasite) and are consistent with a top-to-the E sense of shear. The cataclasite centimetric, rounded to subangular a fine-grained foliated matrix containing secondary dolomite/calcite, clay minerals and amorphous silica pockets. Clasts are mainly derived from the hanging wall Cretaceous flysch as well as from late Miocene leucogranite sills hosted in the flysch (Viola et al., 2018). The rare occurrence of quartz- and/or phyllite clasts (Fig. 4d) indicates a contribution also from the footwall metamorphic Triassic Verrucano quartzite and metapelite. The NNE-striking cataclastic foliation dips gently (< 30°) to the west and bears E–W-trending slickenlines. A noteworthy feature of the BSF 2 cataclasite is the 315 abundance in the matrix of Fe-oxides (ilmenite) along with the occurrence of decimetre-thick sulphide (pyrite, galena, sphalerite) and Fe-oxide rich layers at the contact with the underlying BSF 5 cataclasite (Gundlach-Graham et al., 2018). subrounded to angular quartz, phyllite and hornfels clasts are embedded in a massive matrix that consists of very fine-grained quartz, clay minerals and epidote. The cataclasite directly rests on the quartzite and cordierite-biotite-bearing schist of the footwall Verrucano Formation (Fig. 6e), with the local presence of discrete centimetric to decimetric slices of quartzites at the base of the BSF 4 lower massive cataclasite. At the thin section scale, quartz clasts and the matrix are 370 both characterised by the widespread growth of secondary dolomite grains that partially overprint/replace the original cataclastic fabric. Similarly to BSF 5, the locally well-developed clast SPO is at a steep angle (80-85°) to the weak and gently dipping foliation of the overlying BSF 4 cataclasite. Samples ZUC 8 and ZUC 6 are strikingly different from ZUC 1 and, when considering the coarse fractions, also from each other. Their finest fractions, instead, yielded statistically identical dates, both constraining the same deformation 540 event in the Aquitanian, between 21 and 23 Ma ago. As discussed above, K-bearing smectite is the dated mineral in sample ZUC 6, with its content progressively increasing to a maximum of 51% of the total fraction in the finest fraction, corroborating its authigenic origin. ZUC 8, on the other hand, is compositionally more heterogeneous than ZUC 6, with both 2M1 and 1Md illite polytypes contributing to the age of the finest fraction. The “convergence” to the same Aquitanian age is, thus, indeed remarkable, with the ca. 155 Ma age of the coarsest fraction of ZUC 8 more than twice as 545 old as the age from the same grain size of ZUC 6. The meaning of the AAM approach is, therefore, self-evident when of an interesting scientific debate concerning its kinematic interpretation. As mentioned earlier, a very popular school of thought suggests the ZF to be a LANF, which would have accommodated a 625 significant component of crustal extension in the late Miocene and the Pliocene within the geodynamic framework of the in the immediate footwall of the ZF and directly cut across by it. Direct geometric field constraints 645 from those deformation zones and the ZF allowed the authors to conclude that the last slip event recorded by the ZF postdates the youngest dated structure cut across by it in the footwall, that is, the 4.9 Ma old CN-MAT. Modelling of a 7.58 Ma spurious age from sample ZUC 2 (Viola et al., 2018) by removing only ~1% of a 300-Ma-old contaminant contained within the Cretaceous Flysch at the expense of which the gouge of ZUC 2 formed (a reasonable assumption for a siliciclastic rock containing clasts from Paleozoic sources), brought the ZF faulting age to <4.90 Ma (Viola et al., 650 2018). In a regional perspective, those Late Miocene-Early Pliocene ages and the kinematics of the dated structures confirm the existence of a deformation phase affecting the Oligocene to Miocene Northern Apennines wedge accommodating localized and relatively short-lived out-of-sequence thrusting (e.g., Boccaletti and Sani, 1998); Bonini et al., 2014). Out-of-sequence thrusting has been suggested to possibly reflect a discrete shortening episode after an early extension phase constrained to between ca. 17 and 14 Ma in the orogenic wedge in response to initial slab roll-back (e.g., 655 Bonini et al., 2014; Carmignani et al., 1994; Massa et al., 2017). Apennines orogenic exhumed metamorphic Tuscan Nappe Ligurian Miocene-early Pliocene activity of out-of-sequence Northern Apennines. We envisage a scenario wherein the ZF started as an Aquitanian thrust that was then selectively reactivated by early Pliocene out-of-sequence thrusts during the progressive structuring

elements of a fault architecture is due to the temporal and spatial evolution of the fault such that many brittle faults are best interpreted as the summation of multiple deformation episodes.
To improve our understanding of faulting and produce time-constrained models firmly based on physical and chemical constraints, a deep knowledge of the structural, mechanical, hydrogeological and petrophysical properties of BSF is required and geologists need to pay particular attention to their characterization when unravelling the deformation history 80 preserved in the rock record of fault zones.
To contribute to this issue and further develop our understanding of fault zones and faulting, we document here how conclusions based on the study of fault rocks that are presently tightly juxtaposed in an architecturally complex fault may potentially be misleading or may be revealing only part of long and complex geological histories unless they are based upon the direct dating of all or some of the juxtaposed BSF. We use the Zuccale Fault (ZF) of Elba Island in the Northern

85
Apennines of Italy as an example of a multiply reactivated complex fault. Structural characterization of its constituent BSF and their direct dating by illite K-Ar geochronology allows us to constrain a geological history that is remarkably longer than previously assumed and the unravelling of which suggests deformation in the inner portion of the northern Apennines to be more articulated than generally reported. Lastly, we discuss the potential implications of our new results on the tectonic history of of the Northern Apennines orogenic wedge from the early Miocene to the early Pliocene.

Methods
Our study relies on an integrated approach combining the multiscalar structural analysis and characterization of the Zuccale Fault (ZF) according to classic field and petrographic approaches and the K-Ar dating of selected, representative fault rocks. These were sampled from well characterized BSF of the ZF. The dated material was additionally qualitatively 95 and quantitatively characterized by X-ray diffraction analysis and TEM imaging of selected grain size fractions to characterize the finest dated mineral phases.
The analytical procedures for separating, characterizing, and dating the samples are described in detail in Viola et al. (2018) and represent the workflow at the laboratory for clay characterization and K-Ar dating at the Geological Survey of Norway. Fault rocks were disintegrated using approximately 100 freeze-thaw cycles, and clays were suspended in 100 deionized water. Sedimentation with Stokes law, and a combination of continuous flow and fixed angle rotor centrifuges were used to generate particle sizes of <0.1, 0.1-0.4, 0.4-2, 2-6 and 6-10 μm. Argon was extracted from clay aliquots packed in Mo foil in a stainless steel ultrahigh vacuum line using a Pond Engineering furnace at 1400 °C. The evolved sample gas was purified in a first stage using a titanium sublimation pump, and in a second stage with one SAES GP50 getter at room temperature and one SAES GP50 getter at 350 °C. Sample gas was spiked with approximately 2*10 −13 105 moles of pure 38 Ar (Schumacher, 1975) and analyzed in an IsotopX NGX multicollector noble gas mass spectrometer fitted with five faraday cups for 600 integrations of 1 s each. Nominal argon beam intensities were determined by using a degree 2 polynomial regression to gas inlet time zero, with an in-house Python program. Beam intensities for 38 Ar and 36 Ar were corrected for mass discrimination relative to 40 Ar by a power law (e.g., Renne et al., 2009), using the weighted mean 40 Ar/ 36 Ar ratios of 299.781 ± 0.014 measured from atmospheric argon in an online air pipette, and the atmospheric 110 argon composition of Lee et al. (2006). Radiogenic 40 Ar* concentrations and their uncertainties were calculated using the equations outlined in Hałas and Wójtowicz (2014). Within this analytical batch, four aliquots of GA-1550 biotite (98.5 ± 0.5 Ma; Mcdougall and Wellman, 2011) were analyzed and yielded a weighted mean age of 98.53 ± 0.36 Ma.
Potassium concentrations were determined by fusing an aliquot of approximately 50 mg in lithium tetraborate at ~1000 https://doi.org/10.5194/egusphere-2022-229 Preprint. Discussion started: 13 May 2022 c Author(s) 2022. CC BY 4.0 License. °C to form a glass, which was then dissolved in HNO3 with a rhodium internal standard at 5 ppm prior to analysis with a 115 Perkin Elmer Optima 4300 DV ICP-OES. K-Ar dates were calculated using the 40 K abundance and decay constants of Steiger and Jäger (1977).
XRD analyses were carried out at the Academic Laboratory of Basin Analysis (ALBA) of Roma Tre University (Italy) with a Scintag X1 X-ray system (CuKα radiation). The tube current and the voltage were 30 mA and 40 kV, respectively.
Randomly oriented whole-rock powders of all subfractions were X-rayed with a step size of 0.05° 2θ and a counting time 120 of 3 s/step in the 2-70° 2θ range. Standard patterns for illite 1Md and 2M1 polytypes were represented by the pure illites PDU and SG4 (Eberl et al., 1987), respectively. Integrated peak areas of minerals and illite polytype reflections were transformed into mineral concentration by using mineral intensity factors as a calibration constant (for a review, see Moore and Reynolds, 1997). The estimated quantification error is ± 5%.
Transmission Electron Microscope (TEM) imaging and energy dispersive spectroscopy analysis were performed at the 125 NORTEM infrastructure at the TEM Gemini Centre of the Norwegian University of Science and Technology (NTNU) in Trondheim, Norway. Samples were dispersed at a concentration of 10 ppm by weight in deionized water, and dried on a carbon-supported TEM grid. Images were obtained on a JEOL JEM-2100 TEM with a LaB6 filament and chemical information was collected using an Oxford X-Max 80 SDD EDX.

130
The Northern Apennines (Fig. 1) form a stack of NE-to E-vergent thrust sheets scraped off from oceanic and continental lithosphere during the Upper Cretaceous-Eocene convergence between the European and the Adriatic plates.
Located in the northern Tyrrhenian Sea, Elba Island belongs to the Northern Apennines inner sector and consists of five 135 metamorphic and non-metamorphic tectonic units derived from both the (i) Adria continental (e.g. Tuscan Nappe) and (ii) Ligurian oceanic domain (e.g. Ligurian Unit), stacked with a general transport direction toward the northeast during orogenic shortening (Boccaletti et al., 1971;Keller and Coward, 1996;Bortolotti et al., 2001).   Figure 2. Modified after Viola et al. (2018).
The tectonic units within this nappe stack are organized into two major thrust complexes (Musumeci and Vaselli, 2012;145 Massa et al., 2017;Papeschi et al., 2021). The Upper Complex is an imbricate fan formed of three thrust sheets of sedimentary and low-grade metamorphic rocks ascribable to Ligurian units and Tuscan Nappe. The Lower Complex, instead, consists of two metamorphic units, namely the Calamita Unit overlain by the Ortano Unit, both derived from the Adriatic continental domain (Fig. 1a).
The Upper and Lower Complex are juxtaposed along a major N-S-striking and W-dipping thrust fault, the Capo Norsi-

150
Mt. Arco thrust (CN-MAT of Viola et al., 2018;Musumeci and Vaselli, 2012), which is outlined by an intensely deformed slice of Ligurian serpentinized peridotites sandwiched between the metamorphic Ortano Unit below and the overlying Tuscan nappe sedimentary sequence. As a whole, the system forms a W-dipping, 10 km wide and 15 km long monocline that defines the first-order geological structure of eastern and central Elba (Barberi et al., 1967;Pertusati et al., 1993;Keller and Coward, 1996;Bortolotti et al., 2001;Papeschi et al., 2021;Ryan et al., 2021).
The emplacement-related thermal anomaly caused diffuse, late Miocene LP/HT metamorphism with amphibolite facies mineral assemblages that are the main metamorphic record in the Calamita and Ortano units (Duranti et al., 1992;Musumeci and Vaselli, 2012;Papeschi et al., 2017;Papeschi and Musumeci, 2019). Early Miocene blueschist facies 160 metamorphism (Papeschi et al., 2020 and reference therein) related to the late Oligocene-early Miocene Apennine subduction (Ryan et al., 2021) is only preserved in the northernmost Ortano unit, which escaped the static thermal overprint.
According to existing reconstructions, the tectonic evolution of the nappe stack in easternmost Elba is interpreted as resulting from late Oligocene-early Miocene SW-NE shortening, which caused folding and nappe stacking. Compression 165 was followed by middle to late Miocene extension broadly coeval with magmatism and the formation of sedimentary basins in the mainland immediately to the east of Elba Island, and related to either fundamental geodynamic processes causing the opening of the northern Thyrrenian sea (e.g., Keller and Coward (1996)) and/or more local factors associated with magma emplacement and ballooning (Pertusati et al., 1993;Bortolotti et al., 2001;Westerman et al., 2004).
Recently, however, mapping and structural observations aided by geochronology have led some authors to a partial 170 reinterpretation of the nappe stack tectonic evolution that can be summarized by the following four evolutionary stages: (i) Oligocene to early Miocene folding and nappe stacking under very low metamorphic grade (i.e., anchizone) in the Upper Complex and greenschist to blueschist metamorphism in the Lower Complex. The early Miocene blueschist facies metamorphism is recorded and well preserved in the northernmost Acquadolce subunit of the Ortano Unit (Bianco et al., 2015;Papeschi et al., 2020;Papeschi et al., 2021;Ryan et al., 2021);

175
(ii) Middle Miocene extension of the stack upper portion due to the gravitational readjustment of the orogenic wedge causing the tectonic elision of part of the Tuscan nappe; https://doi.org/10.5194/egusphere-2022-229 Preprint. Discussion started: 13 May 2022 c Author(s) 2022. CC BY 4.0 License.
(iii) Middle-late Miocene folding and out of sequence thrusting that reshaped the nappe stack (Musumeci and Vaselli, 2012;Papeschi et al., 2017;Viola et al., 2018;Papeschi et al., 2021). The slice of Ligurian peridotites tectonically sandwiched between the upper and lower complex along the CN-MAT supports out of sequence thrusting associated 180 with the formation of east-facing thrust anticlines (e.g., Massa et al., 2017). Geochronology of Neogene illite in fault gouge from the CN-MAT constrains this pulse of renewed shortening to ca. 5 Ma (Viola et al., 2018), which is distinctly younger than the emplacement of the Miocene Porto Azzurro pluton dated to between 7 and 6 Ma (Gagnevin et al., 2011;Musumeci et al., 2015).
(iv) Renewed extension associated with the present-day extensional seismotectonic features of the entire inner part of the 185 Northern Apennines (e.g., Faccenna et al., 2014).
In summary, recent mapping and structural data integrated by K-Ar geochronology of synkinematic illite from key fault gouge samples argue for a complex tectonic evolution wherein shortening pulses have continued to the early Pliocene, with a discrete, significant phase of extension in the middle Miocene (Viola et al., 2018). In this different scenario, also the Zuccale Fault (see below) has been recently reinterpreted as having partly accommodated one late increment of out-190 of-sequence thrusting on the Elba Island (Musumeci et al., 2015;Viola et al., 2018).

The Zuccale Fault
The Zuccale Fault (Fig. 2) was first described and interpreted as a low-angle detachment fault (LANF) that cuts downsection the central-eastern Elba nappe stack with a top-to-the E displacement of ca. 6 km (Keller and Pialli, 1990;Keller 195 and Coward, 1996). Its first-order main structural domains are: -a purely brittle fault zone composed of breccia, cataclasite and gouge forming a 3 to 5 m thick heterogeneous fault core; -a general E-dipping low-angle (< 5°) discrete principal slip surface (PSS). Pervasive subhorizontal slickenlines and crystal fiber lineations and associated asymmetric structures are consistent with dominant top-to-the E to NE slip; -a foliated domain comprising: (1) a W-to NW-dipping foliation defined by the preferred orientation of phyllosilicates   (in black) and in Viola et al. (2018;in red). b) Fence diagram illustrating fault rock distribution along the NS and EW natural cross-sections that are directly accessible along the shore line of Punta Zuccale.
Following the initial description of this remarkable fault structure and the structural characterization of the 210 abovementioned domains and features, over the last decade numerous studies have dealt with the ZF mechanic and dynamic significance as well as with its kinematic relevance within the framework of the overall evolution of the Northern Apennines orogenic wedge. Collettini and Holdsworth (2004), for example, first reported a spatial zonation of fault rocks in the ZF core, which they interpreted as reflecting an evolution from initial frictional deformation by cataclasis to a more distributed deformation style via the formation of a thoroughly overprinting phyllonitic fabric in the core. This progressive 215 evolution was interpreted as suggesting that initial pervasive cataclasis enhanced the overall system permeability, which https://doi.org/10.5194/egusphere-2022-229 Preprint. Discussion started: 13 May 2022 c Author(s) 2022. CC BY 4.0 License.
would have enhanced the influx of CO2-rich hydrous fluids that triggered low-grade alteration and the onset of pressure solution leading to overall weakening, shear localization and formation of talc-rich mylonitic layers in the fault core (this model has been further developed in Collettini et al., 2011;Collettini, 2011;Smith and Faulkner, 2010;Smith et al., 2011a;Smith et al., 2011b). This, in turn, would have promoted rock weakening and a switch from frictional to an overall 220 frictional-viscous fault behaviour (e.g., Collettini et al., 2009). Smith et al. (2011b) subsequently reported an alternative deformation history wherein the early development of mylonitic fabrics (talc phyllonite) was followed by the formation of cataclasite and breccia. The spatial zonation of fault rocks proposed by Smith and Faulkner (2010) and Smith et al. (2011b) is shown in Table 1.  Collettini and Holdsworth (2004) also stressed the role of the ZF as the main structure accommodating the late Neogene 230 upper crustal extension in the Northern Apennines and proposed a possible mechanical explanation for its development as a LANF. Although the ZF is traditionally described and interpreted as a subhorizontal to gently dipping top-to-the-E LANF, Musumeci et al. (2015) proposed an alternative model wherein the exposed ZF represents the flat of a much larger, out-of-sequence top-to-the-E thrust that cuts through the Eastern Elba nappe stack. The same authors bracketed the https://doi.org/10.5194/egusphere-2022-229 Preprint. Discussion started: 13 May 2022 c Author(s) 2022. CC BY 4.0 License. activity of the ZF to between 6.23 ± 0.06 and 5.39 ± 0.46 Ma, based on the youngest available muscovite 40Ar/39Ar age 235 from the footwall and a U-Th-He age on adularia and hematite mineralizations in the hanging wall (e.g., Lippolt et al. 1995). Viola et al. (2018) provided the first absolute dating of part of the fault activity, constraining the age of the gouge associated with the ZF PSS to < 4.9 Ma, consistent with the fundamental observation that the fault gouge and cataclasite contain clasts of the monzogranitic Porto Azzurro pluton dated to ca. 6.5 Ma (e.g., Gagnevin et al., 2011;Musumeci et al., 2015).

Zuccale Fault: Internal Architecture, BSF definition and characterization
The most complete and continuous exposure of the ZF is easily accessible and well exposed along two coastal crosssections oriented N-S and E-W at Punta Zuccale (Figs. 1b and 2). There, the ZF zone is between ca. 1.5 and 3 m thick,

245
and it discordantly cuts across both the footwall (quartzite and andalusite-bearing spotted schist) and the hanging wall (marly Cretaceous flysch). Overall, ZF fault rocks consist of clast and/or matrix-supported breccia and cataclasite exhibiting large lateral variations in thickness, texture and composition (Musumeci et al., 2015). Gouge exists as 2-10 cm thick discontinuous lenses across the fault zone but is particularly common at or close to the upper boundary of the fault zone, that is, immediately below the hanging wall block. Two slivers of mylonitic marble derived from the footwall block 250 (Calanchiole marble; Musumeci and Vaselli, 2012;Viola et al., 2018) are embedded within the fault zone.
Detailed field mapping and multiscalar fault rock characterization has allowed us to refine the detailed characterization of the ZF structural framework and to distinguish and characterize five distinct BSF's, the distribution of which is shown in map view and by means of a fence diagram in Figure 2a and b, respectively, and listed in Table 1. BSF's are numbered according to their geometrical position from the top to the base of the ZF, as observed along the two natural and mutually 255 orthogonal cross-section (see fence diagram in Fig. 2b). Their identification and characterization have steered the sampling strategy to obtain (when possible) absolute illite K-Ar time constraints, as discussed below.

BSF 1 -Upper yellowish cataclasites
A yellowish cataclasite represents the uppermost portion of the ZF core. It crops out continuously below the hanging wall Cretaceous flysch, which is, in turn, itself characterized by a decimetric to metric damage zone defined by increased 260 fracture density and foliated domains with localised S/C fabrics confirming the regional top-to-the E sense of shear of the ZF. BSF 1 rests immediately above the striated Principal Slip Surface (PSS) of the ZF (see below) and is composed of a few discrete, centimetre thick ultracataclasite and gouge bands. It exhibits a subhorizontal contact to the hanging wall block (Figs. 3a,b). Discrete top-to-the E striated fault planes cut across the fault rocks of BSF1. Cataclastic bands are composed of yellowish, cohesive to poorly lithified cataclasite ( Fig. 3c) containing angular to subrounded clasts from 265 centimetre to millimetre in size. Clasts are mainly of limestone, quartzite and igneous rocks (leucogranite sill) commonly replaced/altered by secondary carbonate grains and of pyrite-bearing veins. The fine-to very fine-grained matrix (< 50 μm) of the cataclasite consists of secondary dolomite, calcite, quartz, clay minerals and Fe-oxides. Secondary dolomite/calcite also occur as euhedral grains and /or filling of veins ( Fig. 4a, b). At the mesoscopic scale, a poorly developed and gently (<20°) west-dipping foliation (Fig. 3c) is recognised. It is crosscut by centimetre-spaced and 270 millimetre thick subhorizontal fault planes dipping gently (< 5°) toward the west (Fig. 3c). The basal portion of BSF 1 consists of greenish, cohesive and strongly indurated matrix-supported cataclasite containing foliated to non-foliated https://doi.org/10.5194/egusphere-2022-229 Preprint. Discussion started: 13 May 2022 c Author(s) 2022. CC BY 4.0 License. millimetre to centimetre-thick gouge bands. In the gouge, clasts are represented by rounded quartz grains and/or breccia grains. At the optical microscope, the foliation appears as defined by thin (1 mm) and discontinuous clay mineral-rich layers that outline a centimetre spaced, subhorizontal or gently dipping foliation, while dolomite and calcite occur as 275 recrystallized grains (Fig. 4a). BSF 1 contains clear top-to-the E kinematic indicators such as oblique fabrics, S/C type shear bands and asymmetric imbricated sigmoidal lithons also derived from other BSF's (e.g., Fig. 3b).

BSF 2 -Yellowish foliated cataclasites
This foliated cataclasite occurs in the northernmost exposed ZF, where it is sandwiched between BSF 1 above and BSF 305 5 below (Fig. 3a) and it tapers off progressively towards the south (Fig. 2a, b). BSF 2 fault rocks appear as a whitishyellowish cohesive foliated cataclasite in the immediate footwall of the ZF striated PSS (Fig. 5a, b). The PSS and subparallel striated planes dip gently to the west (Viola et al., 2018, their   This BSF corresponds to a 10 m long and 1-3 meters thick lensoidal body made of calc-mylonite and greenish talcsmectite-tremolite phyllonite (Fig. 3d). The unit is bounded at the top and at the base by yellow discrete gouge layers and 330 greenish cataclasite that can be ascribed to BSF 1 and BSF 4, respectively. The phyllonitic component consists of a serrated alternance of white and greenish, well foliated bands that are centimeter to tens of centimeter thick and that reflect a compositional layering made up of calcite, talc-smectite-tremolite and phyllosilicate rich layers ( Fig. 3d and 5c).
At the thin section scale, the fabric is characterized by fine-grained calcite lithons interspersed within a network of thin (< 1 mm) smectite-and/or talc-rich layers ( Fig. 4e and f) that define a well-developed foliation with millimetric spacing 335 in high strain zones (see Collettini et al., 2009, for a detailed mechanical and microstructural characterization of this fault rock).

BSF 4 -Greenish cataclasites
It is exposed only along the southern, E-W section, as it tapers completely out moving north (Figs. 2a,b ). It varies in thickness from a few cm up to one m. Its basal contact discordantly cuts across the west dipping foliation of the footwall 340 rocks, which are cordierite-biotite-bearing hornfels and quartzite breccia formed at the expense of the underlying footwall Calamita Unit. Fault rocks appear as a cohesive and strongly indurated, matrix-supported cataclasite (Figs. 5e and 4g and h) containing angular clasts of quartzite, amphibole-plagioclase-bearing skarn, epidote-bearing veins and reworked older breccia and cataclasite. Clasts range in diameter from a few millimetres to a few centimetres, with the coarsest clasts mainly localized in up to decimetre thick breccia pockets and/or coarse-grained cataclasite layers. Millimetric clay-rich 345 layers locally define a coarsely spaced subhorizontal or gently dipping foliation. Dark green to rust brown foliated to nonfoliated gouge occurs as millimetre to centimetre-thick discontinuous layers ( Fig. 5f) with sharp contacts to the contiguous breccia and cataclasite (Fig. 4g). In the gouge, the clasts are represented by rounded quartz grains and/or older breccia clasts. The clasts define a weak shape preferred orientation (SPO) that is generally parallel to the gently dipping foliation of the clay-rich layers.

BSF 5 -Massive cataclasites
BSF 5 crops out in the northernmost portion of the ZF (Fig. 2a, Fig. 3a and Fig. 6a and b). It is characterized by significant thickness variations, lobate/digitate contacts with the footwall Triassic quartzites and an overall 3D wedge shape that tapers towards the south. It is a matrix-supported cohesive and strongly indurated greyish to whitish cataclasite referred to as "honeycomb breccia" in the literature (Smith et al., 2008;Fig. 6c). Coarse to medium-grained subrounded to angular 355 quartz, phyllite and hornfels clasts are embedded in an isotropic matrix devoid of any internal fabric and composed of very fine-grained quartz, clay minerals and epidote (Fig. 6c). Locally, a greenish colour is due to the great abundance of epidote in the matrix. The massive cataclasite rests directly on Verrucano quartzites (Fig. 2) and centimetre to decimetre thick slices of quartzites are locally embedded as clasts within the basal massive greyish cataclasite. At the thin section scale, the original rock fabric (quartz clasts and matrix) is partially replaced by the growth of secondary dolomite. Clasts 360 form a very weak SPO in this BSF, which is generally roughly perpendicular to the gently dipping foliation of the overlying BSF 2 and principal slip surface.

BSF 6 -Lower quarzitic breccia
BSF 6 represents the structurally lowermost BSF of the complex ZF internal architecture. It occurs as discontinuous decametric bands of white breccias at the top of Triassic Verrucano Fm. quartzite footwall rocks (Fig. 2). It is formed by 365 a matrix-supported, cohesive greyish to whitish breccia and cataclasite ( Fig. 6d and e). Coarse-to medium-grained https://doi.org/10.5194/egusphere-2022-229 Preprint. Discussion started: 13 May 2022 c Author(s) 2022. CC BY 4.0 License. subrounded to angular quartz, phyllite and hornfels clasts are embedded in a massive matrix that consists of very finegrained quartz, clay minerals and epidote. The cataclasite directly rests on the quartzite and cordierite-biotite-bearing schist of the footwall Verrucano Formation (Fig. 6e), with the local presence of discrete centimetric to decimetric slices of quartzites at the base of the BSF 4 lower massive cataclasite. At the thin section scale, quartz clasts and the matrix are

Footwall of the ZF
Footwall rocks are well exposed along the N-S section and mainly consist of white quartzite of the Verrucano Fm. (Fig.   6f) and biotite-cordierite-bearing schists that become dominant north of the investigated Zuccale section (Musumeci and 385 Vaselli, 2012). Above the cordierite-bearing schist, white calcite mylonitic marble with greenish amphibole-epidotebearing calcsilicate are exposed for a length of several meters and discordantly capped by BSF 1 (Fig. 6). They have been assigned to the core of the Calanchiole shear zone (Musumeci and Vaselli, 2012;Viola et al., 2018) and are characterised by a penetrative N-S-striking mylonitic fabric that dips at ca. 40° toward the west. The calcite mylonitic marble has been dated to 6 Ma (Viola et al., 2018), which attests to lower amphibolite-upper greenschist facies ductile deformation with 390 dynamic recrystallization of calcite grains together with synkinematic growth of fine-grained tremolite and talc within the thermal aureole of the Porto Azzurro pluton (Musumeci et al., 2015).

ZF samples for K-Ar dating and K-Ar dating results
Aiming to further constrain the development of the ZF fault architecture and to add absolute time constraints to it, thus 395 strengthening the initial results reported by Viola et al. (2018), we sampled and dated all the BSF described above.
Unfortunately, only the following three BSF's yielded datable material:

ZUC 1 (BSF 1)
This sample is from the N-S trending section at Punta Zuccale and is representative of the yellowish cataclasites of BSF 1. The dated material rests directly upon the striated ZF PSS that bears E-W trending and W-plunging slickenlines and is 400 composed of a matrix-supported yellowish clay-rich gouge formed at the expense of the hanging wall Cretaceous flysch ( Figure 7). The sampled gouge is internally chaotic, devoid of any internal fabric and does not exhibit any consistent indication of a specific sense of shear (Figure 7). The kinematics, however, are clearly constrained by a plethora of topto-the E kinematic indicators associated with the slip recorded along the discrete ZF PSS immediately at the base of ZUC 1, such as slickensided surfaces or the dragging of foliated cataclasites of BSF 2 into the slip plane.  Concerning the mineralogical composition of the investigated sample, the 2-6 and 6-10 µm grain size fractions of gouge ZUC 1 are mainly composed of quartz (38-41%), carbonate minerals (dolomite and aragonite) between 15 and 27%, illite-2M1 (14-17%) and subordinate amounts of illite-1Md (6-7%), kaolinite (6-9%) and mixed layers illite-smectite (I-S; 5-7%). Low halite (4%) contents have also been detected in the coarse fractions.
Importantly, this sample is from the same sampling site of sample ZUC 2 reported by Viola et al. (2018), which constrained faulting at this locality to < 4.9 Ma.

425
Sample ZUC 8 is also from BSF1, but from the easternmost exposed tract of the E-W trending section of the ZF (Fig. 2 and 7). It represents chestnut brown gouge lenses, which are internally devoid of any pervasive fabric, interfingered with yellowish cataclasite and gouge. The latter is pervasively foliated and contains abundant kinematic indicators such as asymmetric and imbricated clasts and lenses and oblique planar fabrics indicating top-to-the E shearing.

ZUC 6 (BSF 3)
It needs to be pointed out that it was not possible to obtain any material from the collected samples of BSF 2, 4, 5 and 6.

455
Our dataset almost invariably exhibits a "grain size -age correlation" for the three samples, wherein the coarser the dated grain size fraction, the older the age (Fig. 8).
While only four size fractions were analyzed by K-Ar geochronology for ZUC 1, as no <0.1 μm grain size fraction could be separated, five fractions were successfully dated for ZUC 6 and ZUC 8 (Table 3 and   TEM imaging integrated the characterization of the dated fractions, particularly of the finest fractions, which are 470 interpreted as providing the tighter constraint upon the last slip episode recorded by the dated fault rock (see below for details). As noted, sample ZUC 1 did not yield any datable material for the finest fraction and, thus, no TEM investigations were carried out. We can rely, however, on observations from sample ZUC 2 by Viola et al. (2018), taken in the same BSF only a couple of m away along strike from ZUC 2 ( Fig. 4c and d by Viola et al., 2018). Two generations of illite crystals were recognized in the finest ZUC 2 fraction, with an acicular and platy shape, respectively.

475
Sample ZUC 8 contains illite platelets with irregular edges (Fig. 8) similar to those documented for ZUC 2 by Viola et al. (2018) and inferred for ZUC 1 in this study.
TEM imaging and EDS of the finest fraction of ZUC 6 confirms the abundant presence of K-bearing smectite crystals, as documented by the XRD analysis ( Fig. 7 and 8).

Discussion
Our results are first discussed and interpreted in general terms, aiming to assess their impact upon a conceptual model for the time-integrated development of the ZF. This is important for understanding the broader context of brittle faulting and deformation localization during long-lived structural evolutions and upon the methodological approach to studying and unravelling complex fault cores. Finally, we discuss the significance of our study upon the tectonic evolution of the 490 Eastern Elba Nappe Stack and in the framework of the Northern Apennines evolution.

Interpretation of the age results
The new data depict a complex history "archived" within the ZF fault core at the studied outcrop (Fig. 8). The obtained "K-Ar age vs. grain size" relationship is remarkable, with the three dated samples yielding quite different results and being characterized by heterogeneous "age vs. grain size spectra" (sensu Pevear, 1999), wherein the coarser the fraction, 495 the older the age. In this study, we use the "Age Attractor Model (AAM)" first proposed by Torgersen et al. (2015a) and Viola et al. (2016) to interpret the obtained age data. The AAM suggests that the amount of detrital K-bearing phases inherited from the host rock in a brittle fault rock decreases with the decreasing grain size of the dated fractions, whereas the amount of authigenic and synkinematic clay increases with decreasing grain size. The coarsest dated fractions may thus still contain significant amounts of "old" protolithic K-bearing minerals inherited from the host rock. These grain tectonic and thermal events, many studies have convincingly shown that the age of the finest fractions (< 0.1 µm or finer) is a reliable record of the timing of the last brittle deformation increment recorded by the fault (Viola et al., 2013;Viola et al., 2016;Torgersen et al., 2015a;Torgersen et al., 2015b;Mancktelow et al., 2015;Aldega et al., 2019;Scheiber et al., 2019;Curzi et al., 2020a;Curzi et al., 2020b;Tartaglia et al., 2020). Only in the ideal case where all K-Ar ages from one sample are statistically identical for all dated grain size fractions, it is possible to assume that the age of the finest 510 fraction as well as of all other fractions represents the true age of the last recorded faulting event, without any input from "old" protolithic components (see, for example, ELB 2 and ZUC 4 by Viola et al., 2018;Fig. 8;Torgersen et al., 2015a).
The contribution of potential host rock contamination, particularly in low-temperature fault rocks reworking sedimentary rocks and containing significant amounts of illite derived from the sedimentary history of the faulted lithology, may be evaluated with the Illite Age Analysis (IAA) approach (e.g., Hunziker et al., 1986;Pevear, 1999;Aldega et al., 2019;515 Carboni et al., 2020;Curzi et al., 2020a). The IAA discriminates the mostly detrital 2M1 polytype (which in turn might represent a mixture of authigenic high-temperature illite and cataclastic, synkinematic muscovite) from a truly authigenic phase 1Md formed syn-kinematically during faulting. The IAA, however, did not produce any meaningful dates in this study, yielding negative intercepts for 100% 1Md, suggesting that the 2M1 polytype is also in part authigenic and synkinematic.

520
In accordance with the AAM approach, we thus base the interpretation of the new batch of age results mostly on the dates of the finest fractions as already done by Viola et al. (2018) for other samples from the ZF. We still consider them, however, as maximum ages of the last recorded increment of faulting because varying amounts of inherited K-bearing phases may still be present, thus potentially influencing the obtained apparent age.
Sample ZUC 1, for example, is indeed one such case, as discussed in depth for its companion ZUC 2 sample by Viola et 525 al. (2018). It is a fault gouge formed at the expense of a compositionally heterogeneous flysch rock of Cretaceous age, rich of siliciclastic inputs. In contrast to ZUC 2 in Viola et al. (2018), in the current study it was not possible to separate the < 0.1 µm grain size fraction of ZUC 1, such that the finest dated fraction is the < 0.4 µm. It yielded a 18.8 ± 0.3 Ma date (Fig. 8). The "age vs. grain size" pattern of ZUC 1 is, however, very similar to that of ZUC 2 (Viola et al., 2018), with only minor differences except for the coarsest fraction. XRD analysis of ZUC1 documents a mixture of roughly 530 equal amounts of 1Md and 2M1 illite polytypes in the finest dated fractions (<0.4µm; Table 2). This is consistent with results from detailed SEM and TEM imaging of ZUC 2 ( Fig. 8 and Viola et al., 2018), which disclosed the presence of two different generations of illite within this gouge, with a detrital component coexisting to varying degrees in different size fractions with authigenic and synkinematic illite crystals. Although the age of the < 0.1 µm fraction of ZUC 1 is unknown, it seems reasonable to extrapolate the ZUC 1 "age vs. grain size" trend to an age younger than 10 Ma.

535
Irrespective of the exact date, the extrapolated age is still likely to be spurious as it contains old and protolithic illite, thus making the date artificially older. Viola et al. (2018) thus calculated a modelled age for ZUC 2 of < 4,9 Ma, which we also deem reasonable for ZUC 1.
Samples ZUC 8 and ZUC 6 are strikingly different from ZUC 1 and, when considering the coarse fractions, also from each other. Their finest fractions, instead, yielded statistically identical dates, both constraining the same deformation 540 event in the Aquitanian, between 21 and 23 Ma ago. As discussed above, K-bearing smectite is the dated mineral in sample ZUC 6, with its content progressively increasing to a maximum of 51% of the total fraction in the finest fraction, corroborating its authigenic origin. ZUC 8, on the other hand, is compositionally more heterogeneous than ZUC 6, with both 2M1 and 1Md illite polytypes contributing to the age of the finest fraction. The "convergence" to the same Aquitanian age is, thus, indeed remarkable, with the ca. 155 Ma age of the coarsest fraction of ZUC 8 more than twice as 545 old as the age from the same grain size of ZUC 6. The meaning of the AAM approach is, therefore, self-evident when https://doi.org/10.5194/egusphere-2022-229 Preprint. Discussion started: 13 May 2022 c Author(s) 2022. CC BY 4.0 License. comparing these two samples and this case study is reminiscent of other remarkable cases discussed in the literature, such as those by Torgersen et al. (2015b) and Viola et al. (2016). This strengthens our conclusion that the last recorded faulting increment did indeed "attract" to the same Aquitanian faulting age the ages of the coarsest and remarkably different fractions. The older ages are well preserved in discrete gouge lenses of BSF1 and in BSF 3 along with specific 550 metamorphic mineral assemblages (sample ZUC 6). As noteworthy feature, BSF 3 is devoid of any pervasive brittle deformation and lacks evidence of fluid-driven deposition of secondary calcite/dolomite and Fe-oxides/hydroxide, as instead commonly observed in BSF 1 and BSF2 and to a lesser extent in BSF 4. This suggest that BSF 3 has behaved as an "across-foliation low-permeability barrier" during the ZF younger brittle deformation increments and fluid circulation, thus differently from the other BSF's.

555
As a whole, the new dataset expands on the dataset by Viola et al. (2018), who first documented a significant faulting age cluster in the early Pliocene. ZUC 6 and ZUC 8 now also document the coexistence in the ZF of fault rocks that, in addition to the Pliocene cluster, record a distinct thermotectonic event in the Aquitanian.
The new results allow us to evaluate and refine the available structural models for this major fault zone (e.g., Keller and Pialli, 1990;Keller and Coward, 1996;Collettini and Holdsworth, 2004;Collettini et al., 2009;Smith and Faulkner, 2010;560 Musumeci et al., 2015;Viola et al., 2018). The possibility to analytically demonstrate that the fault core is composed of a range of coexisting BSF's that formed at significantly different times calls, indeed, for a re-evaluation of the existing evolutionary conceptual models.
The new data is of interest in that the kinematic framework of the fault does not appear to have changed from the Aquitanian to the Pliocene, with top-to-the E remaining the regional kinematics along the fault (Collettini and 565 Holdsworth, 2004;Collettini et al., 2009;Smith and Faulkner, 2010;Musumeci et al., 2015). However, our results, indicate that, this consistency notwithstanding, the identified and sampled BSF's formed at very different times, with the phyllonitic fault rocks of BSF 3 and one gouge of BSF 1 both recording Aquitanian top-to-the E faulting under likely warmer conditions, whereas a second gouge from BSF 1 constrains a phase of top-to-the E faulting to the Pliocene. Any static or dynamic model ignoring or overlooking such fault internal heterogeneity is at risk to oversimplify the ZF 570 evolution, potentially leading to erroneous conclusions as to its role in the local tectonic evolution or to unreliable inferences on the system mechanical behavior.

General implications on workflows for the characterization of faults and on fault mechanics
The workflow presented in here is of general interest to the structural geology and tectonics community as it demonstrates the need to carry out detailed structural and mechanical characterizations of brittle faults, while also keeping in mind the 575 time dimension of faulting and the fact that currently side-by-side BSF's may actually be the product of deformation histories that are possibly very distant in time. Structural models aiming at describing the geometrical, mechanical and potentially seismic evolution of faults should, therefore, at least attempt to account for the absolute timing of deformation responsible for the preserved fault architectures.
In a broader sense, this study confirms the complexity that likely characterizes many faults and helps to better and foliated cataclasites and gouges, the fault system progressively embrittled, leading to foliated and massive cataclasites and, finally, at ca. < 4 Ma to a discrete PSS. During this ca. 18 Myr long, fluid-assisted evolution (Gundlach-Graham et al., 2018), which can be reasonably coupled with progressive retrogression during exhumation, deformation thus likely evolved from aseismic creep to seismic rupturing, although direct evidence of seismogenic activity has not yet been 590 reported.
The documented complex spatial juxtaposition of BSF's highlights a remarkable geological heterogeneity within the ZF.
Geological heterogeneity in fault zones has been long known to exist and, above all, to occur over a broad range of scales (e.g., Faulkner et al., 2003;Collettini et al., 2009;Fagereng et al., 2011;Tesei et al., 2014;Curzi et al., 2020a;Wang et al., 2016;Scheiber et al., 2019 and references therein). Indeed, faults are commonly heterogeneous, anisotropic and 595 discontinuous both along strike and down dip. This complexity is reflected by a broad spectrum of geometrical, mechanical and hydrological properties, at all scales, which is not always easy to resolve. Structural, mineralogical and petrophysical heterogeneities in faults are being increasingly recognized as key players governing the mechanics of faulting (in addition to also affecting reservoir fluid compartmentalization, mineralization formation etc.; Bruhn et al., 1990;Stober and Bucher, 2015;Scuderi et al., 2020;Bedford et al., 2022) and thus attract much attention by the scientific 600 community. The complexity and heterogeneity documented for the ZF are typical of long-lived, mature and regionalscale fault systems Aldega et al., 2019;Vignaroli et al., 2020). Their clear-cut characterization, however, is challenged by (i) coexisting multiple strands formed at different depths, and (ii) the juxtaposition of BSF's resulting from different deformation conditions (Viola et al., 2013;Torgersen and Viola, 2014;Tartaglia et al., 2020).
The net slip behavior of faults (seismic vs. aseismic) is, therefore, the result of the interplay between local and regional 605 stresses, combined with the varying environmental conditions and fault rock evolution.
We conclude by stressing that we are convinced that the illustrated workflow and the BSF approach may be pivotal to well informed studies of faulting in space and through time and of the mechanical implications thereof.

610
Lastly, we consider the implications that the new results have upon the reconstruction of the Miocene-Pliocene evolution of the Northern Apennines as recorded within the nappe stack of Eastern Elba.
It is remarkable that no direct faulting ages have so far been reported for the Oligocene to early Miocene structuring stage of the Eastern Elba nappe complex as part of the Northern Apennines orogenic wedge (e.g., Keller and Pialli, 1990;Vai and Martini, 2001), when nappe imbrication along discrete thrusts and regional NE-vergent folding accommodated most 615 of the regional shortening. Only recently, Curzi et al. (2020) reported a 22.1 Ma K-Ar age (statistically identical to the Aquitanian ages reported here) from a discrete thrust on the Island of Zannone in the Pontian Archipelago (central Italy), and interpreted it as representing the oldest available radiometric constraint on thrusting in the inner part of the central Apennines. By proposing a correlation between the siliciclastic turbidites in the footwall of the Zannone thrust and the Oligocene-Miocene Macigno Fm. farther to the north, Curzi et al. (2020) suggested Zannone to represent the 620 southernmost exposed continuation of the Northern Apennines early Miocene (~22 Ma) foredeep and thrust front, which was advancing toward the east and northeast in both the northern and central Apennines.
However, while the thrust dated to the early Miocene on Zannone has not been debated as to its kinematic significance, the ZF has in recent years become the target of an interesting scientific debate concerning its kinematic interpretation. As mentioned earlier, a very popular school of thought suggests the ZF to be a LANF, which would have accommodated a opening of the Northern Tyrrhenian Sea back-arc basin (e.g., Keller and Pialli, 1990;Daniel and Jolivet, 1995;Collettini and Holdsworth, 2004). Alternatively, we advocate a second school of thought wherein the ZF is the brittle expression of faulting along the flat segment of a late Miocene-early Pliocene out-of-sequence regional-scale thrust (Musumeci et al., 2015;Viola et al., 2018) that crosscuts older regional and local tectonic structures. According to this alternative model, Samples ZUC 6 and ZUC 8 from this study are thus very important in that they document for the first-time Aquitanian 635 top-to-E shearing along the ZF and in the Northern Apennines. The data strengthen the correlation proposed by Curzi et al. (2020) by directly documenting evidence of early Miocene east-vergent tectonic transport in the innermost part of the Northern Apennines. The early Miocene is undoubtedly too early for extension in this sector of the Northern Apennines (Carmignani et al., 1995 and references therein) and a straightforward conclusion is, therefore, that within the ZF there occur BSF's that formed and isotopically equilibrated (for the investigated K-Ar system) at a time when nappes were still 640 being shortened and imbricated in the growing orogenic wedge. Viola et al. (2018) suggested a tectonic compressive event of Late Miocene-Pliocene age following the main phase of early Miocene nappe imbrication, as recorded by their samples ZUC 4 and ELB 2, dated to 6.14 ± 0.64 Ma and 4.9 ± 0.27 Ma, respectively. Those two samples were taken from the Calanchiole Shear Zone and the CN-MAT, two inverse brittleductile to brittle faults in the immediate footwall of the ZF and directly cut across by it. Direct geometric field constraints 645 from those deformation zones and the ZF allowed the authors to conclude that the last slip event recorded by the ZF postdates the youngest dated structure cut across by it in the footwall, that is, the 4.9 Ma old CN-MAT. Modelling of a 7.58 Ma spurious age from sample ZUC 2 (Viola et al., 2018) by removing only ~1% of a 300-Ma-old contaminant contained within the Cretaceous Flysch at the expense of which the gouge of ZUC 2 formed (a reasonable assumption for a siliciclastic rock containing clasts from Paleozoic sources), brought the ZF faulting age to <4.90 Ma (Viola et al., 650 2018). In a regional perspective, those Late Miocene-Early Pliocene ages and the kinematics of the dated structures confirm the existence of a deformation phase affecting the Oligocene to Miocene Northern Apennines wedge accommodating localized and relatively short-lived out-of-sequence thrusting (e.g., Boccaletti and Sani, 1998); Bonini et al., 2014). Out-of-sequence thrusting has been suggested to possibly reflect a discrete shortening episode after an early extension phase constrained to between ca. 17 and 14 Ma in the orogenic wedge in response to initial slab roll-back (e.g.,   Musumeci and Vaselli (2012) and Papeschi et al. (2017)  In summary, we conclude that an Aquitanian thrust was selectively reactivated and exploited by the Late Miocene-Pliocene ZF, with the dated Aquitanian BSF's preserving reworked fragments of such early thrust. During the ZF progressive deformation and fault reactivation, they underwent further transport and transposition and became finally juxtaposed against other BSF's, which formed later on within the core of the ZF, forming the current complex "patchwork" of fault rocks and K-Ar ages.

680
In this context of long and articulated kinematic evolution, it is currently impossible to know the origin of all BSF's and to constrain the amount of their finite transport within the progressively developing ZF core. Interestingly, preservation of Aquitanian dates within the finest size fractions of samples ZUC 6 and ZUC 8 excludes significant thermal overprinting of the sampled BSF's by the HT-LP metamorphic overprint due to the late Miocene Porto Azzurro pluton, suggesting transport and final juxtaposition of also BSF's from root areas far away from the Porto Azzurro pluton, where they would 685 have escaped any overprinting effects of the thermal anomaly.
The considerations above, together with independent constraints on the tectonic history of the Northern Apennines, are summarized and used to propose the possible evolutionary model of Figure 9, where we illustrate the progressive tectonic evolution that we envisage for the ZF in a broader context. Crustal shortening of Oligocene-Aquitanian age led to stacking 690 of metamorphic units that experienced HP/LT in the deep portion of orogenic wedge ( Fig. 9a; Ryan et al., 2021 and references therein). A large-scale duplex of imbricated horses of metamorphic sheets might be envisaged as being capped by a roof thrust that separates it from the overlying upper portion of wedge (Fig. 9a). The latter consisted of the sedimentary and/or very low-grade metamorphic sequences staked in a nappe structure consisting of the Tuscan Nappe and the overlying Ligurian Units.

695
At the regional scale, an Oligocene-Aquitanian age for orogenic wedge shortening is also constrained by other lines of evidence, including (i) the K-Ar radiometric ages for the first deformation phase (D1) in the metamorphic units of the Apuane Alps, dated to ca. 27 Ma (Kligfield et al., 1986), (ii) the tectonic emplacement of Ligurian units above the Tuscan Nappe dated to the Late Oligocene-Langhian (Montanari and Rossi, 1982) and (iii) the Late Oligocene age of the Macigno Fm. sediment deposition (Dallan-Nardi, 1977). Moreover, late Oligocene ages of 25 Ma and 27-30 Ma have been reported 700 for HP/LT assemblages on the Island of Gorgona and Monte Argentario, respectively (Brunet et al., 2000). Another 20.9 Ma Aquitanian Ar/Ar age has been recently reported for HP metamorphism and tectonic exhumation of the Acquadolce HP/LT unit in the metamorphic complex of eastern Elba Island (Ryan et al., 2021). The still quite loosely constrained age interval for HP/LT blueschist facies metamorphism recorded in several units of the Northern Apennines between 27 and 20 Ma, however, could indicate the diachronous development of subduction and exhumation of units in the deep orogenic 705 wedge.
As shown in Figure 9a, samples ZUC 6 and ZUC 8 may be regarded as representative of tectonic slices belonging to the roof thrust between the Upper and Lower Complex in the orogenic wedge. Thus, the "old" Aquitanian ZF might correspond to a still preserved segment of the original roof thrust that contributed to the tectonic extrusion dynamics of the deeply seated units within the wedge (e.g., Ryan et al., 2021).

710
After a widely acknowledged episode of middle Miocene extension affecting the uppermost orogenic wedge that led to the formation of the highly tectonically excised "Serie Ridotta" of the Tuscan Nappe (Trevisan, 1950;Perrin, 1975;Decandia et al., 1993) in response to the wedge internal dynamics (i.e. gravitational re-equilibration of an overcritical wedge; e.g., Massa et al., 2017 -not reported in Fig. 9), a new shortening phase was recorded in whole Northern Apennines orogenic wedge and affected both the already exhumed metamorphic units and the Tuscan Nappe or Ligurian
As the last out-of-sequence structure that cut across the nappe stack, the brittle ZF partly sampled, reworked and transposed previous structures that now are preserved as "host blocks" in the "cataclastic matrix" of the fault, thus contributing to the complex spatial juxtaposition of the described BSF's.

725
Thus, as shown in Fig. 9c, we interpret the "young" brittle ZF as a late Miocene -early Pliocene ramp-flat-ramp thrust, the western ramp of which nucleated at the base of the upper units reworking the exhumed Aquitanian roof thrust at the top of the lower units that had already been exhumed to shallow crustal levels at ca. 0.2 GPa (Fig. 9b). The flat segment of the thrust now exposed in eastern Elba cuts across the late Miocene nappe stack with a kilometric horizontal displacement. This displacement notwithstanding, it preserves the same sequence of units in both the hanging wall and 730 footwall (Fig. 9c), as clearly seen on geological maps and in the field. Late Miocene-Pliocene out of sequence thrusting in eastern Elba is additionally supported by recent geological mapping of the nappe stack of west dipping tectonic units in easternmost Elba (Papeschi et al., 2021) wherein rocks affected by very low-grade regional metamorphism are thrusted over metamorphic units with late Miocene high-temperature metamorphism.

735
The combined structural and geochronological study of the Zuccale Fault has allowed us to learn lessons of general validity that can be used to define suitable approaches to reconstruct the time-constrained structural and mechanical evolution of architecturally complex fault zones. Additionally, we have gained an in-depth and refined understanding of the complex evolution through time of the Zuccale Fault, which is a major fault of the inner portion of the Northern Apennines that can be used to better constrain the Miocene-Pliocene evolution of the belt.

740
The main highlights of this work can be summarized as follows: (i) Brittle faults are unique archives of the stress state and physical and chemical conditions at the time of both initial deformation localization and subsequent slip(s). Progressive deformation may lead to complex fault architectures, the unravelling of which is at times challenging because, once formed, faults are extremely sensitive to variations in stress field and environmental conditions and are prone to readily slip in a variety of conditions. The detailed, multi-scalar 745 structural analysis of faults and of fault rocks has thus to be the starting point for any study aiming at reconstructing the complex framework of brittle deformation. However, considering that present-day exposures of faults only represent the end result of often protracted and heterogeneous histories, structural and mechanical results need to be integrated over the life span of the studied fault system. K-Ar gouge dating to constrain the time-integrated evolution of faults is, therefore, the natural addition to detailed structural studies. The Brittle Structural Facies concept by Tartaglia et al. (2020) 750 effectively combines these approaches allowing the high-resolution reconstruction of brittle deformation histories and, in turn, multiple constraints to be placed on deformation localization, deformation mechanisms, fluid flow, mineral alteration and authigenesis within actively deforming brittle fault rocks.
(ii) The BSF approach has revealed that the ZF is architecturally complex and composed of at least six different BSF's.
K-Ar gouge dating has constrained an unexpectedly long faulting activity spanning a ca. 20 Ma long time interval in the 755 framework of the evolution of the Northern Apennines. We envisage a scenario wherein the ZF started as an Aquitanian thrust that was then selectively reactivated by early Pliocene out-of-sequence thrusts during the progressive structuring https://doi.org/10.5194/egusphere-2022-229 Preprint. Discussion started: 13 May 2022 c Author(s) 2022. CC BY 4.0 License. of the Northern Apennines wedge. The current fault architecture is a heterogeneous patchwork of BSF's formed at very different times under different environmental conditions in a top-to-the E kinematic framework.
(iii) The structural framework of the Northern Apennines can be elegantly accounted for by a model of continuous 760 convergence between the late Oligocene and the early Pliocene with a major intervening phase of extension in the middle Miocene allowing for the isostatic re-equilibration of the wedge. Extension started again in the Pliocene and is still active in the innermost portion of the Northern Apennines.

Data Availability
All the data produced and used to write the paper are contained in it and in the corresponding Supplement.

Author Contribution
GV conceptualized the study, did fieldwork, acquired the data, collected the samples, elaborated the data, wrote the original draft of the manuscript and prepared the figures. GM and FM did fieldwork, acquired the data, elaborated the data, wrote part of the text and prepared a few figures. LT mapped the BSF's as part of his BSc thesis. ET conceptualized the study, did fieldwork, collected the samples and contributed to text review and editing. RvdL dated 770 the samples and performed TEM imaging. LA run the XRD analyses and contributed to text review, editing and discussions.