Pre-folding fracturing in a foredeep environment: insights from the Carseolani Mountains (central Apennines, Italy)

Abstract Bedding-perpendicular joints striking parallel (longitudinal) and perpendicular (transverse) to both the axis of the hosting anticline and the trend of the foredeep-belt system are widely recognized in fold-and-thrust belts. Their occurrence has been commonly attributed to folding-related processes, such as syn-folding outer-arc extension, although they can also be consistent with a pre-folding foredeep-related fracturing stage. Here we report the pre-folding fracture pattern affecting the Pietrasecca Anticline, in the central Apennines (Italy), resolved by a detailed field structural analysis. Field observations, scan-lines and interpretation of virtual outcrops were used to study the intensity, distribution and the orientations of fracture pattern along the anticline. The fracture pattern of the Pietrasecca Anticline consists of longitudinal and transverse joints, oriented approximately perpendicular to bedding, and of a pre-folding longitudinal pressure-solution cleavage set, which is oblique to bedding regardless of the bedding dip. Cross-cutting relationships show that joints predated the development of the pressure-solution cleavage. Furthermore, joint intensity does not relate to the structural position along the anticline. Taken together, these observations suggest that jointing occurred in a foredeep environment before the Pietrasecca Anticline growth. Our work further demonstrates that joints striking parallel and orthogonal to the main fold axis do not necessarily represent syn-folding deformation structures.

The simplest criterion to distinguish pre-folding from syn-folding joints is based on the analyses of joint orientation with respect to the orientation of the fold axis. Syn-folding, bending-related structures are oriented parallel and orthogonal to the fold axis, whereas prefolding structures can develop oblique to the fold axis (Lash & Engelder, 2007;Tavani et al. 2015;Labeur et al. 2021), as they reflect the direction of pre-folding layer-parallel shortening or the foreland flexure direction, both of which may be oblique with respect to the direction of compressional strain during folding. Since folds are generally elongated parallel to the trend of the hosting foredeep-belt system, the distinction between pre-folding and syn-folding joints can be challenging. In this sense, the common parallelism between the foredeep-and thrust-related anticlines introduces a possible source of error (both terminological and interpretative), as 'longitudinal' and 'transverse' joints are nearly parallel and perpendicular to both fold axes and the foredeep. In this case, cross-cutting relationships with pressure-solution cleavages developed during layerparallel shortening could help to differentiate pre-folding and syn-folding joints. For example, vertical joints and veins oriented orthogonal to the fold axis, when closely associated with (i.e. perpendicular to and coeval with) tectonic stylolites oriented orthogonal to both joints and bedding, are very likely to have been formed during pre-folding layer-parallel shortening (e.g. Beaudoin et al. 2020;Labeur et al. 2021). Another observation that could allow structural geologists to distinguish pre-folding from syn-folding opening-mode fractures regards their intensity distribution across different structural positions along a fold (i.e. backlimb, hinge, forelimb, periclinal closure). In the case of synfolding fracturing, joint intensities should vary along the fold, reaching their highest values in high-strain regions of the fold (e.g. Watkins et al. 2015). Pre-folding joints and veins are instead characterized by intensity values that are quite homogeneous along a fold (Tavani et al. 2015). The achievement of a complete characterization of joint orientation and intensity along a fold is, however, often complicated by the lack of extensive outcrops and, more importantly, by the severe influence exerted by the mechanical stratigraphy (e.g. Ladeira & Price, 1981;Huang & Angelier, 1989;Gross et al. 1995;Peacock & Mann, 2005;Wu & Pollard, 1995;Laubach et al. 2009;Cawood & Bond, 2018), by the occurrence of faults (e.g. Peacock et al. 2017) or by variations in strain directions due to non-cylindrical folding or flexuring (i.e. Billi, 2005;Petracchini et al. 2015). For these reasons, the investigation of the relationship between folding and fracturing requires the collection of robust and well-distributed (both spatially and stratigraphically) datasets.
A typical case study where the application of the simple criterion of parallelism between joints and fold axis could be risky in terms of tectonic interpretations is represented by the Apennine fold-and-thrust belt (Italy). In this belt, several pioneering studies on fold-related fracturing were carried out during the 1970s and 80s (e.g. Alvarez et al. 1978;Marshak et al. 1982;Engelder & Marshak, 1985;Geiser, 1988). More recent studies demonstrated the occurrence of a pre-to earlyfolding fracture pattern in reservoir-scale anticlines (e.g. Tavarnelli & Peacock, 1999;Tavani et al. 2012Tavani et al. , 2015Francioni et al. 2019).
In the present work, we report on the fracture pattern hosted in the Pietrasecca Anticline, located in the central Apennines. The deformation structures exposed in the study anticline include cm-to m-sized joints, pressure-solution cleavages and the km-scale Pietrasecca Fault (Smeraglia et al. 2016). Our detailed field investigations coupled with digital outcrop model interpretation, allowed us to set robust constraints on the prefolding nature of the fracture pattern in this sector of the central Apennines. This study constitutes a new case history that may help geologists working on the field of fractured reservoirs to better understand the fracture pattern expected within subsurface structures.

2.a. Central Apennines
The central Apennines developed since late Oligocene time on top of the westward subduction of the Adriatic plate under the European plate (Doglioni, 1991;Carminati et al. 2012). The subduction trench retreat produced a progressive northeastward migration of the NE-verging folds and thrusts. The thrusts scraped off and piled up the pre-to syn-orogenic sedimentary succession of the Adriatic plate (Cosentino et al., 2010). The pre-orogenic deposits of the central Apennines mostly consist of a 3000-4000 m thick carbonate succession characterized by limestones and dolostones deposited in a carbonate platform environment (i.e. Latium-Abruzzi Carbonate Platform) between Late Triassic and Late Cretaceous time (Parotto & Praturlon, 1975;Damiani et al., 1991;Cosentino et al., 2010). Syn-orogenic deposits includẽ 100-200 m thick carbonate ramp limestones (Lithothamnion and Bryozoan limestones Formation; Sabbatino et al. 2021), capped by~100 m-thick hemipelagic marls (Orbulina marls Formation) and~1600 m thick flysch (e.g. Milli & Moscatelli, 2000). These three units form the infill of the late Miocene foredeep basin developed due to the flexure of the subducting Adriatic lithosphere (e.g. Royden et al. 1987;Cipollari et al. 1995).
The subduction and flexure of the foreland Adria plate generated bending and bulging accommodated by early orogenic extensional structures oriented both parallel and perpendicular to the compressive front (Doglioni, 1995;Bigi & Costa Pisani, 2003;Carminati et al. 2014b).

2.b. Stratigraphy and main structural features of the study area
The study area is located in the Carseolani Mountains, which represent a tectonic unit characterized by NW-SE-trending folds and NW-SE-striking thrust faults with tectonic transport towards the NE ( Fig. 1a; Mostardini & Merlini, 1986;Patacca et al. 2008). The study area is characterized by the Pietrasecca Anticline, an open anticline with a steeply SW-dipping axial plane ( Fig. 1c; Smeraglia et al. 2016). The Pietrasecca Anticline is located at the NW termination of the Caresolani Mountains ( Fig. 1), where the fold axis and the thrust sheet are characterized by an arcuate shape in map view. The fold axis trends NW-SE (~320°N trend) south of Pietrasecca village and rotates~20°counterclockwise (tõ 300°N trend) northwards, near Tufo village (Fig. 1a, c). The Pietrasecca Anticline is dissected by two ENE-WSW-striking and SSE-dipping normal faults (Fig. 1c). Of these, the best-exposed fault is the Pietrasecca Fault, which is~3 km long, dips~80°t owards SSE, and crops out~500 m E of Pietrasecca village (structural station 13 in Fig. 1c). Major and minor slip surfaces in this outcrop suggest that the Pietrasecca Fault accommodated mainly dip slip movements. The Pietrasecca Fault shows a maximum vertical displacement of~100 m (Smeraglia et al. 2016(Smeraglia et al. , 2021. Pre-and syn-orogenic rocks can be identified in the study area (  where the upper stratigraphic boundary is also exposed. The syn-orogenic Lithothamnion and Bryozoan (LB) limestones paraconformably rest on the Radiolitidae limestones and are~100 m thick (Fig. 1b). Three main lithostratigraphic units of LB limestones can be recognized: LB1, LB2 and LB3 from bottom to top ( Fig. 1b; Civitelli & Brandano, 2005). LB1 and LB3 have a thickness of~25 m and consist of havana-brown bioclastic calcarenite deposits with grainstone to packstone texture (Civitelli & Brandano, 2005). LB2 consists of~50 m thick white packstones, wackestones and floatstones with Bryozoans and large bivalves (Civitelli & Brandano, 2005). Lithothamnion and Bryozoan limestones are affected by two sets of pressure-solution cleavages. One set is oriented roughly parallel to bedding and developed during burial and diagenesis (Civitelli & Brandano, 2005). The other set is oriented oblique to bedding and developed during the incipient phases of shortening of the central Apennines (Tavani & Cifelli, 2010). Although pressure-solution cleavages mainly affect LB3, they are present (and locally pervasive) also in LB1 and LB2. Lithothamnion and Bryozoan limestones are topped by~60 m thick hemipelagic marls with planktonic foraminifers (Orbulina marls), and by~1600 m of siliciclastic sandstones and marls (flysch; Milli & Moscatelli, 2000).

Methods
We collected data from 13 structural stations located in different positions of the Pietrasecca Anticline (Fig. 1c). Despite the study area being densely vegetated (Fig. 2), locations of the structural stations were chosen to best cover all the structural positions of the Pietrasecca Anticline (i.e. backlimb, forelimb, periclinal closure, hinge) and all the different stratigraphic units (Fig. 1c). Due to vegetation cover and/or the high erodibility of rocks, the Orbulina marls and most of the flysch exposures in the study area did not allow the collection of structural data. We therefore collected data within the same flysch in six supplementary structural stations located near the study area (F1-6 in Fig. 1a, c).
Deformation structures in the study area, in addition to minor faults, mostly consist of joints and pressure-solution cleavages, with rare and usually mm-thick veins. In the following, we will use the term fracture to collectively indicate an assemblage that includes joints, veins and pressure-solution seams. In each station we collected data on bedding, faults (orientation and, if visible, slickenline orientations), oblique-to-bedding pressure-solution cleavage and joints. Structural data were collected through scanlines, scan-areas and virtual outcrop model analysis (Fig. 1c). The orientations of bedding, pressure-solution cleavage, fault surfaces and slickenlines were collected in all the structural stations. The joint data were collected using sub-horizontal scan-lines (e.g. Priest & Hudson, 1981) in most structural stations, allowing us to obtain the linear intensity of joints (see below for further details). The orientation of scan-lines with respect to that of the outcrop produces an orientation bias during sampling of joints (Terzaghi, 1965), with joints striking sub-parallel to the trend of scan-line being relatively under-sampled. To minimize the sampling bias, we avoided scan-lines with orientations parallel to the strike of joints. Orientation data were collected using FieldMove Clino App on iPhone 6S. During data-recording sessions, we carefully avoided interference of metal objects and electronic devices that could affect the magnetic field (Allmendinger et al. 2017). Furthermore, we periodically checked the accuracy of structural measurements taken using iPhone by comparing them with a handheld compass. For each scan-line, we manually classified the joints in different orientation sets using 1 % area contouring of poles to fractures. For each set, we determined its (1) mean orientation (dip and dip azimuth), (2) mean orientation after data rotation for bed dip removal, (3) mean spacing and (4) linear intensity. For the calculation of mean spacing (s) and linear intensity (I), we accounted for the orientation bias using the Terzaghi correction (Terzaghi, 1965): where n is the number of fractures collected along the scan-line, and L 0 is the length of the scan-line component orthogonal to the mean orientation of the fracture set: where α is the angle between scan-line and vector normal to the mean orientation of the joint set. To avoid an overestimation of the bias correction, we fixed the maximum correction angle α to 75°. In addition to scan-lines, we performed 14 scan-areas (Dershowitz & Herda, 1992) on bedding surfaces to better characterize (1) the abutting and/or cross-cutting relationships between joints and (2) the orientation of fractures on the top of Radiolitidae limestones (station 1; Fig. 1c). It was possible to perform scan-areas only for structural stations 1 (11 scan-areas) and 2 (3 scan-areas). The orientation of fractures in each scan-area was retrieved using the NetworkGT plugin for QGIS (Nyberg et al. 2018).
Regarding structural stations 9, 10, 11, 12 and F1-6, outcrops were too discontinuous to perform useful scan-lines. In these stations, only bedding, joint and pressure-solution cleavage orientations were collected, and no scan-line data were recorded.
For structural station 13, which represents the best exposure of the Pietrasecca Fault (see Fig. 1c), we combined field data with data derived from a virtual outcrop model (Fig. 3). In detail, we collected joint data along two scan-lines to constrain joint data retrieved from the virtual outcrop model (Fig. 3). The virtual outcrop model was built using Unmanned Aerial Vehicle imagery and the Structure from Motion Multi View Stereo photogrammetry algorithm (Remondino & El-Hakim, 2006;James & Robson, 2012). The model consists of a textured mesh of 17 × 10 6 triangular faces obtained from a point cloud containing 143 × 10 6 points, representing a surface of~14 000 m 2 . The virtual outcrop model was employed to collect a large amount of data, also from inaccessible portions of the outcrop. Structural data (fault, fractures and bedding orientations) were collected from the virtual outcrop model by manual digitization of traces (i.e. the intersection of fault, fractures and bedding surfaces with the outcrop topography) using the OpenPlot software ( Fig. 3; Tavani et al. 2011). Due to the parallelism between the Pietrasecca Anticline and the Miocene foredeep, fractures striking roughly parallel and perpendicular to the Pietrasecca Anticline do strike parallel and perpendicular to the Miocene foredeep too. Accordingly, the terms 'longitudinal' and 'transverse' will henceforth be used to refer only to the orientation of joints with respect to the local fold axis, without any genetic correlation to folding, which will instead be addressed in the discussion section.

4.a. Bedding and joints
The study area is mainly characterized by sub-horizontal to gently dipping bedding (Figs 1c, 2, 4a). In detail, dip angles are generally <18°with the exceptions of structural stations located in the forelimb of the Pietrasecca Anticline, where bedding generally dips 25-28°to the NE (i.e. stations 4, 10 and 12; Fig. 4a).
In all the structural stations, joints are barren, not filled by cement, and non-strata-bound. They do not show evidence of reactivation on shear (except for one joint set at station 13, as discussed later). Joints are arranged in two main orientations, roughly striking NE-SW and NW-SE, transverse and longitudinal with respect to the trend of the fold-and-thrust belt, respectively (Fig. 4a). Joints are orthogonal to bedding and generally maintain the same main strike also after data rotation for bed dip removal (Fig. 4a, b). In most structural stations, both the transverse and the longitudinal joints are evident (structural stations 1-3, 5, 7-9, 11, 12; Fig. 4a). The transverse joints are NE-SW to E-W oriented and are characterized by linear intensities ranging between 2 m −1 and 24 m −1 (Fig. 4a). The longitudinal joints are NW-SE to N-S oriented and are characterized by  intensities I = 2-11 m −1 (Fig. 4a). In structural station 12, the transverse joints are not observed (Fig. 4a). In some structural stations, groups of joints with a common orientation are not well defined (i.e. structural stations 4, 6 and 10; Fig. 4a). In these stations, joints are always orthogonal to bedding regardless of their highly variable strike angle, as their poles are distributed along the cyclographic representing bedding (Fig. 4a). In the additional structural stations F1-6, joints are arranged in longitudinal and transverse sets also within flysch (Fig. 4).
4b. Pressure-solution cleavage and cross-cutting relationships with joints The oblique-to-bedding pressure-solution cleavage dips gently (25-30°) to the WSW-SW (except for station 6, where it dips to the S), and their poles form a single cluster also when bedding is restored to horizontal (Fig. 5a). Pressure-solution cleavage forms angles between 10°and 40°with respect to bedding, independently of the dip angle of the latter (Fig. 5b, c). At the hand sample scale, pressure-solution cleavage is composed of closed stylolites, which show no evidence of reactivation in opening or shear mode (Fig. 5d).
Relative age relationships between joints and pressure-solution cleavages can be inferred from cross-cutting and/or abutting relationships (Fig. 6). The longitudinal and transverse joints are characterized by mutually cross-cutting relationships (Fig. 6a). In our study area, the analysis of cross-cutting or abutting relationships between joints and pressure-solution cleavage can be performed only in a few outcrops, where pressure-solution cleavage is incipent, due to the lack of joints on outcrops pervaded by pressure-solution cleavage (i.e. where the spacing between the seams of pressure-solution cleavage is~1-2 cm). Where pressure-solution cleavages and joints occur together, pressure-solution cleavage oblique to bedding abuts on or crosses the joints (Fig. 6b, c). Where both bedding-parallel and oblique-to-bedding pressure-solution cleavages are visible, we observe joints that often arrest on bedding-parallel pressure-solution cleavage (Fig. 6d, e), whereas joints cross (Fig. 6d) or are deformed (Fig. 6e) by the oblique-to-bedding pressure-solution cleavage.

4.c. Structural data from the main outcrop of the Pietrasecca Fault
The Pietrasecca Fault (Figs 3, 7) is characterized by a~15 m thick fault core, which accommodates~50 m of displacement, delimited by two SSE-dipping (~80°) main fault surfaces ( Fig. 3 and stereoplot 1 in Fig. 7a). The fault core separates the hangingwall damage zone hosted in LB3 from the footwall damage zone hosted in LB2 (Fig. 7a, b). Slickenlines and cm-to m-scale wavelength grooves along the main fault surfaces indicate pure dip-slip extensional kinematics (stereoplot 1 in Fig. 7a and c). Normal and reverse minor faults affect both hangingwall and footwall damage zones (stereoplots 2 and 3 in Fig. 7a). The hangingwall damage zone is characterized by sub-horizontal bedding with pressure-solution cleavages, oriented parallel and oblique to bedding, dipping~30°t o SW-WSW (stereoplots 3 and 4 in Fig. 7a). Reverse faults in the hangingwall damage zone dip 10-50°to E-ENE, and slickenlines indicate hangingwall motion towards the W-SW (stereoplot 3 in Fig. 7a). Steeply dipping (60-90°) extensional faults in the hangingwall strike NW-SE, NNW-SSE and ENE-WSW, and display mostly dip-slip to dextral transtensional kinematics (stereoplot 3 in Fig. 7a). The scan-line performed in the hangingwall damage zone (SL12; Fig. 7a) highlights a pervasive sub-vertical joint set striking E-W (I = 23 m −1 ; stereoplot 4 in Fig. 7a). Moreover, two minor sub-vertical joint sets strike NE-SW and have intensity of 2-4 m −1 (stereoplot 4 in Fig. 7a). The manual interpretation of the virtual outcrop confirms the predominance of NE-SW to E-W-striking sub-vertical joints, with poles that are distributed along the cyclographic representing the bedding orientation (stereoplot 5 in Fig. 7a).
The bedding in the footwall damage zone is gently dipping (15-30°) towards the NE (stereoplot 2 in Fig. 7a). In the footwall damage zone, reverse faults dip~40°towards ESE, with hangingwall motion towards WNW (stereoplot 2 in Fig. 7a). Extensional faults steeply dip (70-90°) towards SE-SSE and show dip-slip to right lateral transtensional kinematics (stereoplot 2 in Fig. 7a). The contour stereoplot of poles to joint data collected along the scan-line (SL11) shows one cluster representing sub-vertical joints striking NE-SW to ENE-WSW (stereoplot 6 in Fig. 7a). This cluster can be regarded as the sum of two main subsets with slightly different strike angle, one with intensity 21 m −1 and the other with intensity 8 m −1 (stereoplot 6 in Fig. 7a). Other poles to joints are distributed along the cyclographic representing the mean bedding orientation (stereoplot 7 in Fig. 7a). Shear fractures with the same orientation of joints show, in places, slickenlines compatible with dextral strike-slip kinematics.

4.d. Joints vs stratigraphy and vs main tectonic structures
In this section, we report on the relationship between joints and stratigraphy (Fig. 8) and between joints and the main tectonic structures (i.e. the Pietrasecca Anticline and fault; Figs 9, 10). All the joints recognized in each structural station (see Figs 4, 7) have been classified as longitudinal or transverse depending on their orientation with respect to the axis of the Pietrasecca Anticline.
The longitudinal and transverse joints are clearly recognizable in all the investigated stratigraphic units except for LB3, where joints are variably oriented, with slightly higher concentrations for the NE-SW and WNW-ESE orientations (Fig. 8a). The orientations of the transverse and longitudinal joints remain constant throughout the stratigraphic column and are characterized by median strike angles of 60-80°N and 150-170°N, respectively (Fig. 8a). An evident rotation of the mean orientation of joints occurs at structural station 1, where the stratigraphic boundary between Upper Cretaceous rocks (RL) and middle Miocene limestones (LB1 unit) is exposed. Here the transverse and longitudinal joints in RL are oriented~N45°E and N130°E respectively, showing a 20-30°counterclockwise rotation with respect to the orientation measured immediately above within the LB1 unit ( Fig. 8a; see also Fig. 1 for the location of stations 1 and 3). Similarly, a clockwise rotation of the orientation of joints by~20°occurs among scan-lines SL1, SL2 and SL3, which have been performed moving horizontally towards ESE within LB1 at station 1 (Fig. 4).
All stratigraphic units are characterized by linear joint intensity with a median value of 3-4 m −1 (Fig. 8b). No relevant differences in linear intensity can be observed between the longitudinal and transversal joints in all the studied stratigraphic units (Fig. 7b).
The relationship between joints and the Pietrasecca Anticline has been evaluated by analysing changes in joint orientations and/or linear intensities with respect to the position of the fold axis (Fig. 9). Longitudinal joints have been tested against distance from the fold axis (Fig. 9a, c), whereas transverse joints have been tested along the fold axis (Fig. 9b, d). The orientation of both the longitudinal and the transverse joints has been expressed as relative orientation with respect to the Pietrasecca Anticline. In detail, we call α the acute angle that a longitudinal joint set makes with the local direction of fold axis (Fig. 9). Analogously, we call β the acute angle that a transverse joint set makes with the local orthogonal direction to the fold axis (Fig. 9).
The orientation of longitudinal and transverse joint sets is generally close to that of the fold axis and to its orthogonal direction, respectively (i.e. −15°< α, β < 15°; Fig. 9b-d). The orientation of joints significantly deviates from the trend of the Pietrasecca Anticline axis at stations 4-7, 9 and 12 (Fig. 9b, c; see also   (Fig. 9c). Similarly, transverse  The role of the Pietrasecca Fault on joints has been further investigated by testing the linear intensity of transverse and longitudinal joints against distance from and along the Pietrasecca Fault (Fig. 10). The intensity of longitudinal joints does not vary along the Pietrasecca Fault (Fig. 10a), whilst, as previously seen, the intensity of the transverse joints abruptly increases in the vicinity of the Pietrasecca Fault (Fig. 10b).

5.a. Pre-vs syn-folding fracturing
The fracture pattern of the Pietrasecca Anticline includes two joint sets and two sets of pressure-solution cleavages. Despite the occurrence of local variability, joints are mainly beddingperpendicular and oriented roughly parallel (longitudinal) and perpendicular (transverse) to the axis of the Pietrasecca Anticline (Figs 4, 7-10), which is in turn oriented parallel to the Apennines fold-and-thrust belt and associated foredeep. The pressure-solution cleavages are organized in a bedding-parallel and an oblique-to-bedding set. The oblique-to-bedding pressure-solution cleavage strikes roughly parallel to the fold axis and forms a constant angle of 30-40°with respect to bedding, regardless of the bedding dip (Fig. 5).
Previous studies conducted on several anticlines of the Apennines, specifically in the northern Apennines, highlighted a deformation pattern mainly composed of joints (and/or veins) and tectonic stylolites (e.g. Tavarnelli, 1997;Tavani et al. 2012;Beaudoin et al. 2016Beaudoin et al. , 2020Labeur et al. 2021). These studies showed that most of the joints and veins are oriented orthogonal to bedding and strike parallel and orthogonal to the fold axis. Joints and veins oriented orthogonal to bedding and striking oblique to the fold axis have also been reported (e.g. Labeur et al. 2021). Tectonic stylolites are characterized by two sets of orientations striking sub-parallel to the fold axis: one is sub-vertical and occurs in the near-vertical fold limbs, and the other, by far the more common, is oriented orthogonal to the bedding. These structures record all the compressive deformation stages from pre-folding foreland flexuring to the late stage of fold tightening and have been interpreted as follows: (1) joints and veins oriented orthogonal to bedding and striking oblique to the fold axis have been interpreted as structures related to foreland flexuring and bulging; (2) joints oriented orthogonal to the fold axis and tectonic stylolites oriented nearly orthogonal to bedding and striking parallel to the fold axis are considered as the result of a pre-to syn-folding layer-parallel shortening; (3) joints and veins striking parallel to the fold axis are associated with a syn-folding outer-arc extension; and (4) the  Fig. 1 for a more detailed version) with positions of structural stations projected along the fold axis and rosediagrams of strike angle of joints after bed dip removal. Petals of the rosediagrams are 10°wide and are colour-coded depending on the geological formation in the corresponding structural station (see also Fig. 1). Black arrows indicate the local orientation of the axis of the Pietrasecca Anticline. The actual location of the structural station is reported transparently in the background. (c) Relative orientation of longitudinal joints with respect to the trend of the fold axis (above) and linear intensity of longitudinal joints vs map distance from the fold axis (below). Relative orientation of joints is represented by their angular deviation from the local orientation of the axis of the Pietrasecca Anticline (α angle) as shown in the legend. The linear intensity of all the joints (i.e. number of all joints divided by the length of scan-line) is also represented. (d) Relative orientation of transverse joints with respect to the direction orthogonal to the trend of the fold axis (above) and linear intensity of transverse joints vs map distance from point 0 (see (b)) measured along the fold axis (below). Relative orientation of joints is represented by their angular deviation from the direction orthogonal to the local orientation of the axis of the Pietrasecca Anticline (β angle) as shown in the legend. Blue and red colours in (c) and (d) are graduated depending on the relative linear intensity of the joint set for a single scan-line.

1908
M Mercuri et al.
sub-vertical tectonic stylolites striking parallel to the fold axis are associated with late-stage fold tightening.
The main peculiarity of our study area is hence the orientation of the pressure-solution cleavage, which is neither sub-vertical nor orthogonal to bedding, contrary to what is reported for other anticlines in the northern Apennines (e.g. Beaudoin et al. 2020;Labeur et al. 2021). Such observation suggests that pressure-solution cleavage is not related to layer-parallel shortening or to late-folding tightening. We propose that the pressure-solution cleavage reported in this study formed in a pre-folding stage. Coherently with this inference, Tavani & Cifelli (2010) carried out a regional study on this pressure-solution cleavage oblique to bedding and showed that it regionally affects syn-orogenic Lithothamnion and Bryozoan limestones with rather constant strike, angle to bedding, and frequency, suggesting a pre-folding origin ( fig. 10 in Tavani & Cifelli, 2010).
The relative timing between pressure-solution cleavage and joints, with the former postdating the latter, can be deduced from field observations. The mutual cross-cutting relationships between longitudinal and transverse joints suggest they are coeval on a geological timescale (Fig. 6a). Most outcrops are characterized by pressure-solution cleavage oblique to bedding, where this cleavage crosses or abuts on the joints (Fig. 6b, c). In places, relicts of joints dissolved by pressure-solution cleavages also occur (Fig. 6e), providing further evidence that joints already existed during the formation of the pressure-solution cleavage. In this framework of abundant pressure-solution cleavages, the absence of mineralization within joints is enigmatic. In our interpretation, the already existing longitudinal and transversal joints should have represented potential reprecipitation loci for the carbonate dissolved during the development of oblique-to-bedding pressure-solution cleavage. Nevertheless, the vast majority of joints are barren, with a sporadic occurrence of mm-thick veins. Providing a plausible explanation for the absence of calcite within transversal and longitudinal joints is beyond the scope of this work. We can speculate that both joint sets were closed at the time of pressure-solution cleavage formation, each being non-orthogonal to the orientation of the minimum principal stress orientation (σ 3 ).
Although the orientation of longitudinal and transverse joints might suggest that joints formed in response to layer bending and stretching during fold growth (i.e. Ramsay, 1967;Dietrich, 1989;Lemiszki et al. 1994;Richard, 1994;Bobillo-Ares et al. 2000;Fischer & Wilkerson, 2000;Twiss & Moores, 2007;Watkins et al. 2015), we support a pre-folding origin. This preference is based on the pre-folding nature of the pressure-solution cleavage (postdating jointing) and on the lack of clear relationships between the intensity of joints and their structural position along the Pietrasecca Anticline (Fig. 9c, d). In fact, folding-related longitudinal joints should be characterized by: (1) orientation of joints parallel to the fold axis and (2) intensity related to the curvature of the fold (e.g. Ramsay, 1967). This is clearly not the case for the longitudinal joints of the Pietrasecca Anticline, where the joints strikingly occur in areas where no curvature exists, such as the anticline backlimb (Fig. 1c). Analogously, as the Pietrasecca Anticline does not show any evident along-axis curvature (see the section in Fig. 10b), transverse joints cannot be associated with folding. In summary, although we cannot rule out a syn-folding reactivation of longitudinal and transverse joints, as documented in other anticlines worldwide (e.g. Bergbauer & Pollard, 2004), our data suggest that both sets of joints exposed in the Pietrasecca Anticline originated before folding.
Finally, we remark the abrupt increase of the intensity of transverse joints in the vicinity of the Pietrasecca Fault (Fig. 10b), the high dip angle (i.e.~80°) of the main fault surface (the optimal dip angle of normal faults is~60°; Anderson, 1905), and the geometrical affinity between the transverse joints and the subsidiary faults within the damage zone of the Pietrasecca Fault (Fig. 7a). These observations clearly indicate a close relationship between the jointing event and the development of the Pietrasecca Fault and thus point towards a main pre-folding origin for the fault. Assuming that slip occurred under an Andersonian stress field, the fact that the fault rotaxes (also known as slip-normals, i.e. the direction perpendicular to striation and lying along the fault plane; Salvini & Vittori, 1982) of subsidiary faults in the footwall of the Pietrasecca Fault are contained in the bedding plane (stereoplot 2 in Fig. 7a) suggests that the Pietrasecca Fault has been passively rotated during folding.

5.b. Tectonic evolution of the study area
The interpretation of field data indicates that joints, pressure-solution cleavage and faulting predate the folding event that generated the Pietrasecca Anticline. Fracturing hence occurred in the uppermost Miocene, obviously after the deposition of flysch (late Miocene) and before the folding event. Moreover, since the Lithothamnion and Bryozoan limestones seal (and thus postdate) the forebulge unconformity of the central Apennines (Sabbatino et al. 2021), the tectonic setting of fracturing can be placed in the foredeep region in between the fold-and-thrust belt and the forebulge (Fig. 11).
The longitudinal joints developed in the outermost sector of the foredeep in response to the Adria foreland subduction-related flexure (stage 1 in Fig. 11). Extensional structures that formed in a foreland-to-foredeep tectonic setting are widely documented worldwide (e.g. Tavani et al. 2015) and in the Apennines (Calamita & Deiana, 1980;Scisciani et al. 2001;Mazzoli et al. 2005;Tavani et al. 2012;La Bruna et al. 2018). Although our data do not unequivocally shed light on the origin of transverse strain, the mutual cross-cutting relationship between longitudinal and transverse joints (Fig. 6a) indicates that they are coeval on a geological timescale. Two possible explanations exist for the development of transverse joints at the same time as longitudinal joints. The first possibility is that the transverse joints are also related to foreland flexuring. In this scenario they can be cross-joints of longitudinal joints developing from temporal/local permutations between σ 2 and σ 3 causing transverse fracturing of blocks bounded by longitudinal joints (e.g. Gross, 1993;Bai et al. 2002;Billi, 2005;Ferrill et al. 2021), or can be related to the damage zone of release faults (Destro, 1995). A second possible explanation for crossjoints is that they accommodated along-foredeep stretching due to cross-sectional (e.g. Tavani et al. 2020) and/or planar (e.g. Doglioni, 1995) arching of the foredeep. This latter mechanism is favoured in our view, as it is supported by the occurrence of a major transverse element, i.e. the Pietrasecca Fault, and by the lack of any remarkable longitudinal fault.
After the development of longitudinal and transverse beddingperpendicular joints and formation of the Pietrasecca Fault, the advancement of the fold-and-thrust belt caused the development

1910
M Mercuri et al. of a regional brittle-ductile basal décollement level (Tavani & Cifelli, 2010). This décollement includes pressure-solution cleavages regionally oriented at 45°with bedding and consistent with NE shearing accommodated within the entire LB limestones formation (stage 2 in Fig. 11). This regional décollement decoupled the deformation of the overlying clastic foredeep infill from the underlying pre-orogenic multilayer. The peculiarity of this regional décollement level, developing in 'stiff' carbonates at shallow structural levels (i.e. <5 km depth) instead of the overlying apparently weaker marly layer (Orbulina Marls), has already been pointed out (Tavani & Cifelli, 2010) and discussion of the reason for this puzzling observation is beyond the scope of this work. With ongoing collision, the basal décollement of the accretionary wedge migrated to deeper structural levels. A blind thrust developed in the Mesozoic multilayer and the Pietrasecca Anticline grew (stage 3 in Fig. 11). Joints, pressure-solution cleavages and the Pietrasecca Fault were passively rotated with bedding during fold growth. In our opinion, the absence of folding-related deformation structures, contrary to what is observed in other anticlines of the Apennines (e.g. Beaudoin et al. 2016Beaudoin et al. , 2020Labeur et al. 2021), is attributable to joints and pressure-solution cleavage already present during the onset of folding. Pre-existing discontinuities likely inhibited the formation of new fractures and were likely reactivated in opening or shear-mode during folding.
Finally, as witnessed by purely dip-slip grooves and slickenlines on the main fault surfaces (Fig. 7c), after the folding event the Pietrasecca Fault was reactivated as a dip-slip fault (stage 4 in Fig. 11).

Conclusions
Structural field data of joints collected along the Pietrasecca Anticline, in the central Apennines (Italy), highlight pre-orogenic fracturing events (pressure-solution cleavage and joints) which occurred in a foredeep setting, unrelated to folding. Although the arrangement of joints in two sets of orientations, longitudinal and transverse to the fold axis, would suggest a folding-related origin, the following pieces of field evidence suggest a pre-folding origin of joints: (1) cross-cutting relationships show that joints predate pressure-solution cleavages oblique to bedding, where this cleavage, in turn, certainly predates the folding event; (2) the intensity of both longitudinal and transverse joints does not relate to the structural position on the studied anticline (e.g. hinge, backlimb, forelimb). We thus conclude that both longitudinal and transverse joints must have formed in response to a foreland flexure-related strain. The procedure adopted in this study represents a robust approach to determine the timing of fracturing in outcropping analogues of anticline reservoirs, with important implications for the modelling of fluid migration.
Supplementary material. To view supplementary material for this article, please visit https://doi.org/10.1017/S0016756821001291