Spatial-Temporal Evolution of Extensional Faulting and Fluid Circulation in the Amatrice Basin (Central Apennines, Italy) During the Pleistocene

In extensional continental settings, crustal-scale normal faults can accommodate deformation and subsidence at their hanging wall via activation and deactivation of subsidiary tectonic structures. Geological data obtained from subsidiary structures are required to infer the position of the tectonic deformation during the spatial-temporal evolution of the growth-fault system, with significant implications for structures belonging to seismogenic settings. Here, we describe a subsidiary tectonic structure (the Amatrice Fault System) accommodating Quaternary extensional deformation in the Amatrice Basin (central Apennines, Italy), which is an intermountain morpho-structural depression involved by the 2016–2017 seismic sequence. Structurally, the Amatrice Fault System defines a ∼10 km-long tectonic feature running through the Amatrice Basin, and consists of NNW-SSE-striking and E-W-striking fault segments that interact and link over time. Cross-cutting fault relationships are used to reconstruct a kinematic scenario of fault growth and propagation under an ENE-WSW-directed crustal stretching, consistent with the paleostress regime governing the Quaternary activity of the central Apennines. The analysis of stable carbon and oxygen isotopes on syn-kinematic carbonate mineralizations (calcite veins and calcite fibers on fault surfaces) indicates a meteoric water circulation during the development of the growing fault structure, characterized by variable contributions of organic carbon (soil CO2), and suggesting surface rupture and hydrodynamic interconnection with the vadose zone during faulting. Geochronological U-Th dating on the same mineralizations indicates Middle-Late Pleistocene ages for the main phase of tectonic activity of the Amatrice Fault System, with the younger age being 108 ± 10 ka. To date, we cannot exclude minor activations of the Amatrice Fault System during the Holocene. Our results shed light on the Pleistocene tectonics in the Amatrice Basin, in which the Amatrice Fault System records fault growth, hydrodynamic regime and structural permeability network developed under possible coseismic conditions. The evolution of minor tectonic structures, such as the Amatrice Fault System, can provide insights on the localization of tectonic deformation at the hanging wall of a master fault, with implication on the releasing seismogenic potential in active tectonic domains similar to the central Apennines.

In extensional continental settings, crustal-scale normal faults can accommodate deformation and subsidence at their hanging wall via activation and deactivation of subsidiary tectonic structures. Geological data obtained from subsidiary structures are required to infer the position of the tectonic deformation during the spatialtemporal evolution of the growth-fault system, with significant implications for structures belonging to seismogenic settings. Here, we describe a subsidiary tectonic structure (the Amatrice Fault System) accommodating Quaternary extensional deformation in the Amatrice Basin (central Apennines, Italy), which is an intermountain morphostructural depression involved by the 2016-2017 seismic sequence. Structurally, the Amatrice Fault System defines a ∼10 km-long tectonic feature running through the Amatrice Basin, and consists of NNW-SSE-striking and E-W-striking fault segments that interact and link over time. Cross-cutting fault relationships are used to reconstruct a kinematic scenario of fault growth and propagation under an ENE-WSW-directed crustal stretching, consistent with the paleostress regime governing the Quaternary activity of the central Apennines. The analysis of stable carbon and oxygen isotopes on syn-kinematic carbonate mineralizations (calcite veins and calcite fibers on fault surfaces) indicates a meteoric water circulation during the development of the growing fault structure, characterized by variable contributions of organic carbon (soil CO 2 ), and suggesting surface rupture and hydrodynamic interconnection with the vadose zone during faulting. Geochronological U-Th dating on the same mineralizations indicates Middle-Late Pleistocene ages for the main phase of tectonic activity of the Amatrice Fault System, with the younger age being 108 ± 10 ka. To date, we cannot exclude minor activations of the Amatrice Fault System during the Holocene. Our results shed light on the Pleistocene tectonics in the Amatrice Basin, in which the Amatrice Fault System records fault growth, hydrodynamic regime and structural

INTRODUCTION
In extensional regime, major faults are promising targets for releasing seismogenic potential during ongoing crustal-scale deformation in active tectonic domains (e.g., Jackson and White, 1989;Lavecchia et al., 1994;Wernicke, 1995;Ofoegbu and Ferrill, 1998). Fault initiation, propagation, and deactivation, as well as the modalities of interaction at fault-segment terminations (e.g., fault linkage, relay ramps, and transverse zones) record the deformation localization during the fault-growth evolution (e.g., Morley et al., 1990;Cartwright et al., 1995;Peacock and Parfitt, 2002;Hus et al., 2006;Fossen and Rotevatn, 2016;Childs et al., 2017;Rotevatn et al., 2019). During crustal stretching, stress concentration and rock-failure recurrence are modulated by overpressure conditions due to fluid circulation within the fault permeability network (e.g., Curewitz and Karson, 1997;Sibson, 2000;Cox et al., 2001;Rowland and Sibson, 2004;Vignaroli et al., 2015;. While master extensional faults localize the largest amount of displacement and subsidence for basin formation at their hanging wall (e.g., Schlische, 1991;Cowie et al., 2000;Khalil and McClay, 2001;Gawthorpe et al., 2003;Mack et al., 2009;Wilson et al., 2009;Egger et al., 2010;Ford et al., 2017), complex patterns of minor fault systems often accommodate the overall tectonic deformation (e.g., Gibbs, 1984;Destro, 1995;Fossen and Hesthammer, 1997;Cifelli et al., 2007;Balsamo et al., 2008;Faccenna et al., 2011). The structural architecture, spatial distribution, and displacement accumulation of these minor structures are strictly correlated with the growth style and the extent of the master fault, contributing to the whole crustal extension. Moreover, the minor structures directly interact with (and modulate) the processes occurring at the hanging wall of the master fault, such as the accommodation space for basin deposition or the hydrodynamic regime characterizing the fluid flow within the structural permeability network. Therefore, defining the structural architecture of the subsidiary faults, as well as their age of activity and the factors triggering their failure (e.g., fluid circulation), provides useful insights into the localization of the tectonic deformation during the spatial-temporal evolution of the growth-fault system.
The present work describes the tectonic pattern of the Amatrice Basin by introducing the Amatrice Fault System (AFS) as a case of subsidiary tectonic structure (∼10-kmlong fault structure) accommodating Quaternary extensional deformation at the hanging wall of the Gorzano-Laga Fault. We describe the AFS in terms of its structural architecture, kinematics, and deformation style. We present geochemical (δ 13 C and δ 18 O stable isotope) and U-Th dating on selected mineralization (calcite-filled veins and calcite fibers on fault surfaces) related to the AFS. The results are used to reconstruct the spatial-temporal evolution of the AFS in a complex mechanism of growth and link of isolated segments attained in the Early-Late Pleistocene time. A kinematic scenario for the AFS, including the hydrodynamic regime feeding fluid flow at the hanging wall of the Gorzano-Laga Fault, is proposed and framed within the Quaternary-to-active tectonics of the central Apennines.
The modes of fault growth include interaction of individual fault segments that evolved both along and across strike in response to a general NE-SW stretching direction for the central Apennines belt, confirmed by geological (e.g., Cowie and Roberts, 2001;Roberts and Michetti, 2004;Iezzi et al., 2019) and geophysical (e.g., Pondrelli et al., 1995;Montone et al., 2012) evidences.
The area hit by the 2016-2017 seismic sequence ( Figure 1B) is characterized by a complex network of NW-SE-striking (mainly SW-dipping) extensional faults arranged in both en-echelon and collinear geometry with deformation accommodated by relay zones and step over zones at the fault tips (e.g., Boncio et al., 2004b;Pizzi et al., 2017;Galli et al., 2019). The Vettore Fault and the Gorzano-Laga Fault are part of this complex network of segmented structures. The Vettore Fault includes dominant SW-dipping and minor NE-dipping fault segments for an overall length of ∼30 km, and it cuts through a Meso-Cenozoic limestone-dolostone succession with a cumulative vertical displacement of 1.2 km (e.g., Galadini and Galli, 2003;Carminati et al., 2020). The Gorzano-Laga Fault, on the other hand, defines a SW-dipping, high angle extensional structure showing up to 2 km of cumulative displacement (e.g., Bachetti et al., 1990;Blumetti et al., 1993;Galadini and Galli, 2003;Boncio et al., 2004b;Bigi et al., 2013). The Gorzano-Laga Fault is ∼30 km-long, including its southeasternmost segment in the area of the Campotosto Lake ( Figure 1B). No major (i.e., Mw > 4) historical earthquakes were associated to the Vettore Fault before the 2016-2017 sequence (Galli et al., 2019). On the other hand, the 1639 Amatrice earthquake (Mw 6.0) has been attributed to the seismogenic activity of the northwestern segment of the Gorzano-Laga Fault, based on the analysis of the regional seismic records (Boncio et al., 2004a), macroseismic surveys and geological evidences (Galli et al., 2016).

The Amatrice Basin
A composite fault pattern occurs at the surface within the Amatrice Basin, where contractional and extensional faults are exposed. In particular, the fault pattern exposed in the central part of the Amatrice Basin shows complexity mainly due to superimposition of deformation features related either to contractional or extensional tectonic regimes (Koopman, 1983;Centamore et al., 1991;Cacciuni et al., 1995). Structures running close to the Amatrice village has been either interpreted as intraformational thrusts (e.g., Centamore et al., 1991;Cacciuni et al., 1995) or as a west-dipping extensional fault dissecting the stratigraphic sequence of the Laga Formation (Festa, 1999(Festa, , 2005Vignaroli et al., 2019), and possibly related to the activity of the Gorzano-Laga Fault.
The Amatrice Basin is filled by up to 60 m-thick of Quaternary continental units, consisting of a multiple stack of alluvial fan deposits and terraced fluvial deposits of active channel environment (e.g., Centamore et al., 1991;Cacciuni et al., 1995;Mancini et al., 2019;Vignaroli et al., 2019). Below, the Messinian siliciclastic foredeep deposits of the Laga Formation show a maximum thickness of ∼1200 m, as inferred by the stratigraphy of the Varoni01 well and interpretation of seismic lines (Bigi et al., 2013;Porreca et al., 2018;Mancinelli et al., 2019; Figure 1C). The Laga Formation is organized to form an asymmetric, N-S-trending, syncline showing an overturned, west-dipping, western limb at the footwall of intraformational thrust (e.g., Centamore et al., 1991;Cacciuni et al., 1995: Festa, 2005Mancini et al., 2019;Vignaroli et al., 2019). Below the Laga Formation, the stratigraphic sequence includes Jurassicto-Miocene pelagic succession (from top to bottom: Marne ad Orbulina Formation, Marne con Cerrogna Formation, Scagliato-Corniola formations, Calcare Massiccio Formation) and Triassic dolostones-anhydrites ( Figure 1C). The Triassic units result to be doubled by the occurrence of SW-dipping regional thrusts that partially involve the Paleozoic basement (Porreca et al., 2018;Mancinelli et al., 2019). The projected hypocenter of the Mw 6.0 seismic event (Chiaraluce et al., 2017) seem to be located in the deepest portion of the thickened dolostonesevaporites sequence (Buttinelli et al., 2018;Porreca et al., 2018), above the top of the basement placed at a depth of 9-10 km (Mancinelli et al., 2019).

METHODS AND SAMPLES
The geological-structural investigations were carried out along the trace of the AFS, from Cornillo Vecchio, to the north, to Arafranca-Pinaco, to the south (Figure 2A), with the aim to constrain its brittle structural architecture in terms of spatial distribution, geometry, kinematics, and crosscutting relationships of the deformation structures (fault and fracture systems). Collectively, 87 structural data of faults were collected from 11 sites (Table 1 and Figure 3A). Classical field criteria (e.g., Petit, 1987) such as fault offset, growth fibers, and synthetic shears were used to identify the kinematic shear sense of fault segments. The collected structural data are presented in stereographic projections and in statistical analyses of fault populations (Figures 2B, 3B). Statistical analysis of fault populations ( Figure 2B) was done by using the Daisy v.4.1 software (Salvini, 2004) 1 . An in-house Monte Carlo convergent routine was used to determine the fault kinematic pattern and the orientation of the principal paleostress tensors (e.g., Angelier, 1984Angelier, , 1990) responsible for generating the fault population under consideration.
The secondary mineralization related to the tectonic features was systematically sampled for both geochemical and geochronological purposes. Collected samples are represented by (i) the calcite fibers growing on fault surfaces, (ii) the calcitefilled veins that are structurally related to the fault segments, and (iii) the country rock of veins (mostly represented by sandstone-dominated facies of the Laga Formation).
Isotopic analyses (δ 13 C and δ 18 O) of selected mineralizations were done to characterize the origin and properties of the parental fluids that circulated throughout the active structural features. We performed carbon and oxygen stable isotope analyses on 75 samples (23 fault fibers, 31 calcite veins, and 21 sandstone host rock) from the AFS (Supplementary Table A1). 1 http://host.uniroma3.it/progetti/fralab/ Oxygen and carbon isotopes are reported with respect to the Vienna Pee Dee Belemnite standard (V-PDB). Analytical details are reported in Appendix A1.
Geochronological investigations using U-Th methods were done to constrain the age of sealing mineralization within the deformation structures and, therefore, to provide the age for the activity of the structures themselves. We dated two slickenfibers from fault surfaces and six calcite-filled veins by using U-Th techniques ( Table 2). U-Th isotopic compositions and contents were determined by using two different methods: (1) α spectrometry done at the Laboratorio di Geochimica Ambientale of the IGAG-CNR, Italy, and (2) mass spectrometry done with the Thermo Electron Neptune multi-collector inductively coupled mass spectrometer (MC-ICP-MS) (Shen et al., 2012) at the High-Precision Mass Spectrometry and Environment Change Laboratory of the National Taiwan University, Taiwan. The analytical details are reported in Appendix A2.

STRUCTURAL DATA Overview
The AFS runs 4-5 km west to the Gorzano-Laga Fault (Figure 2A) in an area encompassing a continuous exposure of the sandstone-dominated lithofacies of the Laga Formation (hereafter called LAGa, as opposed to the siltstone-dominated lithofacies named LAGp). While the LAGa is composed of fineto-medium-sized sandstone arranged in up to 6-m-thick tabular bedsets, the LAGp consists of centimeter-to-decimeter thick, well-layered planar beds of gray to beige siltstone, interlayered with decimeter-thick siltstone beds. Several generations of Quaternary fluvial and alluvial-fan deposits (Amatrice-Sommati Unit, Retrosi Unit, Terraced Alluvial Deposits, and Recent Alluvial Deposits; Vignaroli et al., 2019) are distributed in the area between the Gorzano-Laga Fault and the AFS. These Quaternary deposits, which record alternate phases of syn-uplift and climatedriven sedimentation and erosion, cover the Miocene substratum above a gently W-dipping angular unconformity, Pliocene in age .
In the following, we describe geometry, kinematics, and occasional filling mineralization for both NNW-SSE and E-W striking structures.

NNW-SSE-Striking Structures
The NNW-SSE system (structural measurement sites 1-5 and 8-11 in Figure 3A and relative stereographic projections in Figure 3B) can be followed continuously along a pronounced morphological ridge running from Cornillo Vecchio (to the north) to Arafranca Pinaco (to the south). In map view (Figure 3A), the NNW-SSE system consists of a complex  Figure 3A. The orientation of the principal paleo-stress tensors after a Monte Carlo-convergent method is reported.
Frontiers in Earth Science | www.frontiersin.org  alignment of subparallel fault segments (up to 2 km long) with changing geometry along the strike (see below). Cumulative vertical displacement of the NNW-SSE fault system cannot easily be estimated because of the lack of unambiguous stratigraphic markers at both footwall and hanging wall rock panels.
The height of the fault-controlled fluvial escarpments merely helps to make an order-of-magnitude estimate (hundreds of meters) for the vertical displacement.
At Cornillo Vecchio (structural measurement sites 1 and 2 in Figure 3A) and at the confluence between the Tronto River and FIGURE 3 | (A) Structural map of the Amatrice Fault System, from Cornillo Vecchio (to the north) to Arafranca-Pinaco (to the south), with structural measurement sites shown with numbers within circles (see Table 1). Location of samples used for U-Th dating is also reported; (B) stereographic projections (Schmidt net, lower hemisphere) of the collected structural data.
Frontiers in Earth Science | www.frontiersin.org the Torrente Castellano (structural measurement sites 3 and 4 in Figure 3A), the fault segments morphologically control the fluvial escarpments within the LAGa. The dominant features are represented by fault surfaces dipping west at relatively high angles (main dip ∼65 • ; Figures 4a,b). In map view, faults are sub-parallel, defining hectometer-long zones of overlap where segments are joined by relay geometries. Fault surfaces have a general planar morphology, having meter-to-decameter persistence along the exposed slopes (Figures 4a-c). Slickenlines occur on the fault surfaces because of the abrasive striation and calcite fibers (see insert in Figure 4b; also Figures 4d,e). Slickenlines show the pitch ranging between 100 • and 110 • , which is indicative of dip-slip kinematics.
Moving southward (i.e., along the western side of the Amatrice village; structural measurement sites 5 and 8 in Figure 3A), a morphological ridge of LAGa is dissected by isolated, NNW-SSE-striking and W-dipping fault surfaces (Figure 5a). Along the strike, these fault segments do not overlap at their tips but are crudely interrupted and dislocated by the E-W-striking faults (see below). At the mesoscale, the NNW-SSE-striking faults are half-meter-scale spaced and have a general planar morphology. As before, fault surfaces host scattered slickenlines provided by abrasion striae and/or calcite fibers (Figures 5b,c). Slickenlines show pitch values generally around 65 • and 115 • . A pervasive foliation, centimeter-to-millimeter-thick, occurs all along the slip zones (Figures 5d-f), distributed across the main fault surface at both the hanging wall and footwall blocks. This foliation is interpreted as a newly formed fault-related discontinuity that developed at high angle with respect to the pristine bedding (Figures 5e,f). At the same outcrops (structural measurement site 5 in Figure 3A), we observed the geometrical relationships between the NNW-SSE fault system and the Quaternary units. Here, roughly 10-m-thick massive conglomerates (lower facies of the Amatrice-Sommati Unit; Figure 3A) cover the LAGa lithofacies. The stratigraphic surface between the LAGa and the conglomerates is not deformed or dislocated (Figure 5g), documenting that, at least for the observed outcrops, the NNW-SSE fault surfaces do not penetrate within the Quaternary units.
To the south, toward Arafranca Pinaco (structural measurement sites 9-11 in Figure 3A), the NNW-SSE system consists of the longest (up to 2 km) fault segments, showing variable strike from NNW-SSE to N-S. In map view, these faults define a collateral geometry of segments that partly join through relay zones with the fault segments near Amatrice, to the north. Principal fault surfaces are exposed along the hectometer-long morphological ridge made of the LAGa lithofacies (Figure 6a). Faults appear either as rectilinear segments aligned parallel to the axis of the ridge (Figures 6b,c) or as a half-meter-spaced set of strands (Figures 6d,e). In both cases, fault surfaces dip west at moderate (about 50 • -60 • ) to high (up to 85 • ) angles (see the stereographic projections in Figure 6). Slickenlines on fault surfaces are generally provided by abrasion striae and/or calcite fibers. Slickenlines have pitch values generally around 80 • and 115 • (insert in Figure 6c). Locally, secondary fault surfaces strike WNW-ESE, making a dihedral angle of ∼40 • with the main surface (Figure 6b). These secondary faults are not pervasive in the rock volume, and they stop on the NNW-SSE-striking fault Kinematic criteria, expressed by stratigraphic offset, synthetic shears, dragging of the bedding at the fault surfaces, and geometrical relationships between the faulting-related foliation and the bedding, indicate normal-sense movement for the NNW-SSE-striking faults.

E-W-Striking Structures
The E-W fault system (structural measurement sites 6 and 7 in Figure 3A and relative stereographic projections in Figure 3B) consists of hundred-meter-long fault segments dipping both to the north and south. The E-W fault structures are less mature with respect to the NNW-SSE-striking faults, showing reduced offset (on the order of tens of meters) and persistence. In map view, these E-W-striking structures are recurrent in the central part of the AFS, where they control escarpment morphologies oriented perpendicular to the Torrente Castellano. Structurally, they define transverse faults that interrupt and dislocate the NNW-SSE-striking faults. At the mesoscale, the E-W-striking faults are half-meter spaced and show rough planar slip surfaces cutting the bedding at high angles (Figure 7a). Well-developed E-W fault surfaces hosting striations and kinematic indicators are rarely exposed. When occurring, slickenlines are provided by abrasive striae showing pitch values generally around 80 • and   Figure 7a). Observed shear criteria, such as synthetic shear fractures and stratigraphic offset, indicate normal-sense movement for the E-W-striking faults.

• (see stereographic projection in
One of the most important feature of the E-W fault system is the occurrence of a systematic fracture network (structural measurement sites 4, 5, 9, and 11 in Figure 3A and relative stereographic projections in Figure 3B) that consists of sub-vertical, planar discontinuities showing pluri-decameter persistence. The fracture spacing systematically decreases from decametric to half-meter upon approaching the main slip surfaces (e.g., at structural measurement sites 4 and 5 in Figure 3A). Locally, E-W-striking fractures are filled by calcite agglomerate and are therefore classified as veins (Figures 7b-e). Veins are up to 2 cm thick and have sharp boundaries with the surrounding sandstones. Texturally, the filling calcite forms either a blocky texture, a coarse-grained agglomerate (up to 3-4 mm in size, see Figure 7f), or a fine-grained (submillimeter) matrix (Figure 7g).

CARBON AND OXYGEN ISOTOPES
All analyzed samples are characterized by negative values of both δ 13 C (between −10.26 and −0.29 ) and δ 18 O (between −7.28 and −3.09 ) (Supplementary Table A1 and Figure 8).
Grouping the data according to sampling location produces no evidence of isotopic differentiation, whereas such evidence does appear upon grouping the data according to the different calcite deposits (host rock, veins, and fault fibers). Figure 8 shows samples using this distinction: four samples of calcite  The isotope compositions of the carbonate fraction of the host rock are in the range of marine carbonate rocks, despite the average of the oxygen isotopes (−4.4 , 0.43 StdDev) being about two per mill lower than the typical values for most Meso-Cenozoic marine limestones and dolostones in central Italy (Ghisetti et al., 2001). At the thin section scale, carbonatic cement as pore-filling is almost absent. The isotopic shift to slightly lower values in δ 18 O may be due to submillimeter microfractures filled with calcites that, in some cases, may also account for the episodic shift of carbon isotopes toward lower values (down to −3.9 ).
Overall, C-and O-isotope data of newly formed calcite show that the precipitation fluid along the AFS was likely of meteoric origin for both fibers and veins. Carbon isotopes provide evidence of an important contribution of soil CO 2 to the carbon species in solution during fault-fiber formation (NNW-SSE-striking structures) with respect to the calcite-filled vein (E-W-striking structures) because the remarkable C-isotope decrease can be ascribed to organic carbon incorporation into the calcite fiber crystal lattice. Oxygen isotopes show that meteoric water may have constituted the dominant oxygen reservoir to which both fault fibers and calcite-filled veins re-equilibrated after dissolution of the host-rock carbonate fraction. There are no evident dissolution structures in the host-rock, but also the development of the fault-vein conduit network in the host-rock seems to be extremely limited. The distribution of isotope data in the δ 18 O/δ 13 C diagram seems to indicate that the calcite fibers evolved just from the host-rock isotope composition toward lower values as a function of variable carbon and oxygen molar W/R ratios (Sverjensky, 1981;Rye and Bradbury, 1988). A different evolution may be envisaged for vein calcites whose carbon isotopes do not evince the presence of an organic component, and oxygen-isotope composition exceeds that of the host rock in some instances. The isotope distribution of vein calcites seems to indicate precipitation from a meteoricderived groundwater fluid whose provenance may even be external to the host rock; namely, from isotopically unaltered carbonate rocks.

U-Th GEOCHRONOLOGY
Three samples of calcite-filled vein from the E-W-striking fault collected near the Amatrice village were dated by α spectrometry (Fven1A; Figure 7d) and by the MC-ICP-MS method (Fven1 and Fven2; Figures 7c,d). Sample Fven1A is characterized by a moderate detrital contamination ( 230 Th/ 232 Th activity ratio = 24.2 ± 3.6) due to a moderate acid leaching (see Appendix A2), whereas samples Fven1 and Fven2 are affected by a severe contamination due to dissolution of a non-carbonatic component attributable to the strong acid attack. Both samples are characterized by lower 230 Th/ 232 Th activity ratio (5.82 ± 0.15 and 1.80 ± 0.06, respectively). By assuming a detrital component at secular equilibrium, with the crustal 232 Th/ 238 U value of 3.8, we obtained a corrected age of 118.6 ± 6.4 ka for sample Fven1A and a corrected age of 251 ± 47 ka for sample Fven2, whereas age older than the range of the method (≥800 ka) is obtained for sample Fven1. Eventually, considering these three Fven samples as coeval, if we plot the 230 Th/ 232 Th and 234 U/ 232 Th activity ratios against their respective 238 U/ 232 Th activity ratios (method L/L of Schwarcz and Latham, 1989), we obtain a corrected age of 108 ± 10 ka.
The other samples dated through the MC-ICP-MS method give older ages. In particular, the oldest age obtained is from the calcite fibers from NNW-SSE-striking fault surfaces (355 ± 57 ka; sample Fin8; Figure 4b). A similar age (348 ± 46 ka) is obtained from a calcite-filled vein from the E-W-striking fault collected near the village of Arafranca-Pinaco (sample AR1; Figure 7e). Moreover, if we assume a crustal detritic component in radioactive equilibrium, three samples belonging to both types of calcite mineralization are older than the range of the method (≥800 ka). However, the present δ 234 U values that are significantly greater than zero (at a significance level > 6σ) constrain their mineralization ages to less than about 1250 ka, which is the time required to obtain secular equilibrium in the 234 U/ 238 U activity ratio.

Tectonic Synthesis
Our multidisciplinary dataset is used to address the spatialtemporal evolution of the AFS within the tectono-stratigraphic setting of the Amatrice Basin.
(1) The internal structural architecture of the AFS includes kilometer-long extensional fault segments oriented NNW-SSE and accompanied by subsidiary extensional fault segments oriented E-W. Structural complexities reside in the style and modalities of link and interaction at the fault terminations, including fault overlaps linked by relay zones, collateral faults, and fault segmentation by transverse structures. The recognized fault array for the AFS can be framed within a kinematic scenario of fault growth and propagation under an ENE-WSW-directed crustal stretching ( Figure 9A). The AFS developed over time by mechanisms that are commonly related to fault population, including radial propagation of isolated fault segments, fault overlapping, and fault linkage (Morley et al., 1990;Cartwright et al., 1995;Peacock and Parfitt, 2002;Hus et al., 2006;Fossen and Rotevatn, 2016; Figure 9B). In this scenario, the E-W-striking system represents the last structural step of the AFS, suggesting conditions of local stress perturbation that induced a change in deformation style (from longitudinal to transverse faults) during the final stage of the AFS growth. This kinematic scenario offers a comparison term for the growth of the Quaternary-toactive faults of the central Apennines (see below). (2) The analysis of the isotopic data suggests a meteoric water ingress within tectonic discontinuities to crystallize calcite as fibers on fault surfaces. According to the negative values of C-isotopic data, a significant contribution of organic carbon (soil CO 2 ) was incorporated into the calcite fibers on fault surfaces, suggesting an open circulation connected to the vadose zone. Conversely, the dominant carbon reservoir for calcite in veins is represented by the host rocks or an adjacent carbonate rock, whose dissolution provided the carbonate for calcite-vein precipitation. The variability presented by the oxygen isotopes in the veins results from the variable oxygen molar W/R ratios experienced by each carbon species in solution during its path to the deposition sites. The calcite closest to the host-rock values likely precipitated near the place where dissolution occurred. In any case, carbon from the soil zone did not enter the vein system, as if the vein openings were sealed against the vertical infiltration that characterizes the circulation through fault surfaces. Therefore, we propose that the change in deformation style (from longitudinal to transverse faults) corresponds to a change in the fluid pattern. First, a hydraulic system open to vadose-water influx percolated along the NNW-SSE-striking system. Next, a hydraulic system of fluid circulation, closed to vertical influx, precipitated calcite within the E-Wstriking system after dissolution from the surrounding matrix or from adjacent carbonate rocks ( Figure 9B). These propositions require the assumption of nearly constant temperature (close to the surface temperature) during calcite deposition, which seems to be a reasonable assumption if we consider the recent development of the fault system and the remarkable contribution of soil CO 2 as a possible source of organic carbon for the fibers on the fault surfaces.
(3) The U-Th ages allow dating of the syn-tectonic mineralization along the AFS and this represents the first geochronological record provided for the Amatrice Basin. Apart from the samples that are older than the range of the U-Th method (>800 ka), both fault fibers and calcite-filled veins gave ages spanning from ∼350 to 108 ka. These ages constrain the main phases of the tectonic activity of the AFS to a period spanning from the late Early Pleistocene to the early Late Pleistocene. Therefore, we infer the AFS accumulated deformation and most of its displacement during the Pleistocene, dissecting the stratigraphic sequence of the Laga Formation. The cross-cutting relationships between the dated structures and the overlying continental deposits show that NNW-SSE-striking faults do not cross through the conglomerates and cobbles (Figure 5g). Similar observations along other sections of the AFS Vignaroli et al., 2019) suggest that the deposition of the continental deposits filling in the Amatrice Basin post-dated the structures we dated at the early Late Pleistocene. Anyway, we cannot exclude the possibility that the activation of younger structures within the AFS (i.e., the veins dated to the Late Pleistocene) may have involved the overlying sediments by producing not appreciable deformation at the mesoscale. In addition, the diachronic deposition of the continental deposits cannot be excluded over the Amatrice Basin, being influenced by the hydrodynamics of local alluvial drainages (e.g., Cacciuni et al., 1995;Mancini et al., 2019). Finally, we aware that our ages do not preclude the fact the AFS (or its segments) reactivated during the Holocene, in consequence to the seismogenic activity of the adjacent Gorzano-Laga Fault (e.g., Galli et al., 2016). Further geochronological constraints and/or geomorphic analyses could be addressed to elucidate possible minor activations of the AFS connected to the seismic activity in the Amatrice Basin.
To summarize, our results document that tectonic deformation and surficial fluid circulation were active in the Amatrice Basin during the Pleistocene. The hydrodynamic interconnection with the vadose zone suggests the AFS acted as a growing fault structure that produced ruptures at the surface. Meteoric waters infiltrated downward into the tectonic cracks and were channelized through structural conduits within the rock volume, compatible with a scenario of tectonically maintained vertical permeability (e.g., Sibson, 2000;Cox et al., 2001;Rowland and Sibson, 2004). Similar scenarios of fluidassisted tectonic deformation in the central Apennines have been documented in faulted carbonate rocks (e.g., Maiorani et al., 1992;Conti et al., 2001;Ghisetti et al., 2001;Agosta et al., 2008;Smeraglia et al., , 2018, suggesting the interplay between transient permeability at the fault damage zones and mixing of fluids from different reservoirs (meteoric, groundwater, subcrustal). Noteworthy, the AFS provide a case history of fluid-rock interaction in siliciclastic units affecting by extensional faulting. Fluids permeating the AFS preserve the geochemical signature of cold water circulation at shallow depths, with the dominant meteoric water that progressively changes to fluids in isotopic equilibrium with the surrounding siliciclastic host rock. Overall, two options are here proposed to model the fluid-rock interaction along the AFS in the Early-Late Pleistocene. The first model invokes a dominant infiltration of meteoric fluids (the open hydraulic system) and progressive (over time) mixing with groundwater that has exchanged isotopically with carbonates, which plays a dominant role in the development of the veins ( Figure 9C). In this scenario, the isotopic transition correlates with the fluid flow that moved throughout the host rock adjacent to the fault zones (e.g., Pili et al., 2002). This scenario is in line with the general considerations regarding the calcite deposits during fluid flow in diverse fault systems from the central Apennines (Ghisetti et al., 2001;Agosta et al., 2008). Given the surficial conditions under which the AFS developed, the lithological control on the fluid composition could be more important than the advective transport of meteoric waters in the final stage of the Pleistocene fault activity. Possibly, NNW-SSE-striking faults of the AFS led to a structural permeability to azimuthal infiltration of meteoric water (soil CO 2 contributes to C species in solution) greater than that produced by the E-Wstriking structures. The second model invokes a discontinuous fluid history involving a two-step development of newly formed calcites: initially, an open system meteoric circulation developed, mainly driven by a vertical infiltration from the vadose soil zone through the fault surfaces; this was a localized flow along the extensional fault segments oriented NNW-SSE. The subsequent formation of the E-W subsidiary fault segments promoted a horizontal fluid flow, likely originating from a lateral water reservoir; it was closed to the former meteoric circulation and, consequently, to contributions of carbonatic matter from the soil zone.

Implication for the Quaternary-to-Active Extensional Faulting in Central Apennines
The reconstructed structural pattern of the AFS can be framed within the post-orogenic regional extensional tectonic regime described for the axial domain of the central Apennines, where the main NW-SE-striking extensional faults (i.e., oriented parallel to the belt axis) accommodated a crustal-scale extensional regime and the onset of continental sedimentation in intermountain basins (e.g., Malinverno and Ryan, 1986;Cavinato and De Celles, 1999;Cavinato et al., 2002;Giaccio et al., 2012;Mancini et al., 2012;Pucci et al., 2014;Cosentino et al., 2017). In this scenario, the AFS defines a minor tectonic structure resembling (in terms of fault architecture, geometries, and kinematics) the adjacent Gorzano-Laga Fault, which, in turn, represents the basin-bounding master fault that localized up to 2 km of displacement during its activity (e.g., Bachetti et al., 1990;Boncio et al., 2004b). Overall, the evolution of both the AFS and the Gorzano-Laga Fault is compatible with the general NE-SW stretching direction for the central Apennines belt (e.g., Montone et al., 2012). This is inferred from the orientation of the σ3 (minimum compression) obtained by fault-slip analysis for the AFS (Figure 2B) compared with the direction of the active stress tensor (N74 • -trending σ3 axis) computed for the Gorzano-Laga Fault by inversion of the focal mechanisms (Boncio et al., 2004b). Therefore, by linking the Pleistocene tectonic activity of the AFS and the still-ongoing activity of the Gorzano-Laga Fault, it is proposed a self-similar paleostress regime for the area of the Amatrice Basin during the Quaternary.
The structural evolution we illustrated for the AFS reinforces the general scenario of fault growth proposed for the Quaternaryto-active faults of the central Apennines, in which faults nucleated as isolated segments able to modulate their along-strike throws and displacements during propagation and interaction at the fault tips (e.g., Cello et al., 1997;Cowie and Roberts, 2001;Roberts and Michetti, 2004;Faure Walker et al., 2010;Lavecchia et al., 2016;Pizzi et al., 2017;Civico et al., 2018;Iezzi et al., 2018Iezzi et al., , 2019Villani et al., 2018;Galli et al., 2019). In particular, the study of partial and total coseismic ruptures occurring along master extensional faults suggests a propagating fault model (Rotevatn et al., 2019 and references therein) characterized by variable displacement accrual during fault lengthening. The structural evolution we proposed for the AFS (Figures 9A,B) can add some insights on the tectonic activity at the hanging wall of the master extensional faults. The AFS fault architecture suggests a two-stage mechanism of fault growth. A first stage of dominant lengthening and displacement accumulation during formation of longitudinal (NNW-SSE-striking) faults is followed by a second stage of limited longitudinal fault propagation, fault overlapping and local development of transverse (E-W-striking) structures. The switch from longitudinal to transversal fault development, as constrained by both structural field observation and U-Th ages of tectonic-related mineralizations, corroborates the occurrence of complex patterns of minor fault systems accommodating the overall tectonic deformation. Similar patterns of longitudinal and transverse faults accommodating different stretching rates within the extending crust have been already documented elsewhere in central Apennines, in which transverse structures (showing strike-slip to extensional kinematics) seem to develop at the endof-life lengthening of the longitudinal (NW-SE-striking) faults (Faccenna et al., 1994;Aiello et al., 2000;Acocella and Funiciello, 2006;Liotta et al., 2015;Bucci et al., 2016;Vignaroli et al., 2016).

CONCLUSION
The Amatrice Fault System (AFS) provides a promising target to capture snapshots of the Pleistocene, fluid-assisted, tectonic deformation in the Amatrice Basin, which is a seismically active domain of the central Apennines. The AFS can be considered a minor structure that accommodated tectonic deformation at the hanging wall of a master extensional fault, the Gorzano-Laga Fault.
Through new U-Th ages and a reconstructed scenario of fault growth and structurally controlled fluid circulation, we propose the AFS nucleated and evolved through radial propagation of isolated longitudinal faults followed by linkage at the fault terminations (including the activation of transverse faults) in a period spanning from the late Early Pleistocene to the early Late Pleistocene, although minor reactivations during the Holocene cannot be excluded a priori. The fault pattern created a transient structural permeability, possibly developed in coseismic conditions, to the azimuthal infiltration of meteoric waters that progressively equilibrated with the geochemical imprint of the siliciclastic host rock. We conclude that the AFS is an interesting topic to be studied for a better understanding of the processes occurring at the hanging wall of a master fault, such as the spatial-temporal fault growth and the hydrodynamic regime characterizing the fluid flow within the structural permeability network. The study of minor tectonic structures and their evolution may represent an interesting perspective in elucidating the relation between the localized deformation along the master faults and the long-term tectonic structures lying at their hanging wall. This approach is expected to be relevant for the assessment of the style and the modality of crustal-scale deformation accommodated in extensional settings that release seismogenic potential.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
GV planned the structural work and the samples collection. MM, FB, MC, and MB participated at the fieldwork. GV wrote the manuscript, with significant contributions by all the authors. GV drew all the figures with contributions by all the authors. MB and FG performed stable isotope analyses. MV performed geochronological analyses through α spectrometry. T-LY and C-CS performed geochronological analyses through multi-collector inductively coupled mass spectrometer. All the authors contributed to the data processing, result discussion, and final interpretation, reviewed the manuscript, and approved its submission to Frontiers in Earth Sciences -Structural Geology and Tectonics.

ACKNOWLEDGMENTS
The manuscript benefits from the comments provided by the reviewers AM and KG, who are thanked. The editor (AT) is thanked for the editorial management. S. Mukherjee is thanked for the early stage of the editorial management.