Late Miocene constrictional strain in the northern Apennines: A case study from the Barabarca metaconglomerate (Elba Island, Italy)

Finite strain analyses were performed on a deformed metaconglomerate from the Calamita Unit in the Island of Elba. The Calamita Unit is a synkinematic contact aureole that was shaped by Late Miocene contractional deformation coeval with high‐temperature metamorphism. The metaconglomerate occurs as an L‐tectonite in the footwall of a major thrust, entirely surrounded by S‐tectonites developed in schistose rocks. Object lineations, defined by the preferred orientation of clasts, trend subparallel to the stretching lineations in the associated rocks. Quartz microstructures registered ductile deformation of clasts by grain boundary migration to bulging recrystallization, suggesting temperature decrease during deformation. Rf/ϕ analyses were carried out on three metaconglomerate samples using quartzite clasts as markers. The finite strain data show that the metaconglomerates are strongly deformed in the constrictional field with K values between ~3 and ~7. The constrictional deformation registered by the metaconglomerate with respect to the surrounding metapelites, which likely deformed under plane strain, can be interpreted as the result of flow partitioning in rheologically heterogeneous sequences during deformation. These results suggest the presence of significant strain gradients in the Calamita Unit, strictly associated with heterogeneously distributed ductile shear zones.

. The presence of localized L-tectonites has been linked to specific geological settings, such as (a) closures of isoclinal folds, (b) shear zones, (c) and around diapirs and plutons (Yang, Jiang, & Lu, 2019;Sullivan, 2013 and references therein).
Some authors have suggested that flow partitioning may produce bodies of L-tectonites in competent bodies surrounded by less-competent lithologies (e.g. Chen, 2014;Sullivan, 2008). This has been confirmed, based on multiscale micromechanical approach, by Yang, Jiang, and Lu (2019), which showed that L-tectonites produced by strain partitioning tend to develop only in lithologies with moderately strong relative viscosity with respect to their host rocks. According to these authors, stronger lithologies tend not to register enough strain, while softer lithologies always deform in the flattening field.
In the present study, we document an example of heterogeneous deformation, marked by the occurrence of L-tectonites in quartz metaconglomerate surrounded by schistose rocks that deformed as Stectonites. Finite strain analyses were carried out on the deformed metaconglomerate using quartz clasts as strain markers. The described deformation is associated with a major shear zone developed under high-to low-grade metamorphic conditions, in a cooling contact aureole.

| GEOLOGICAL OUTLINE
The northern Apennines are a fold-and-thrust belt that resulted from the Oligocene closure of the Ligurian Tethys Ocean and the subsequent westward subduction of the Adria microplate beneath the Corsica-Sardinia margin (Boccaletti, Elter, & Guazzone, 1971). The eastward retreat of the hinge of subduction caused the propagation of the northern Apennines towards the east, leading to the Miocene to present-day opening of the northern Tyrrhenian Sea as a back-arc basin in the hinterland sector of the belt (Brunet, Monié, Jolivet, & Cadet, 2000;Jolivet et al., 1998). Intrusive and extrusive bodies of dominant crustal signature, ranging in age from 14 Ma to 0.2-0.3 Ma and interpreted as related to crustal extension, mark the onset of anatectic magmatism in the area (Serri, Innocenti, & Manetti, 1993).
Back arc extension occurred discontinuously from Miocene to present-day times, interrupted by at least one episode of shortening in the late Miocene (see Bonini et al., 2014 and references therein), which is well-documented on the Island of Elba.

| Deformed metaconglomerate from the Barabarca quartzite
The Barabarca Quartzite Fm. is made up of white to pink quartzite and violet cordierite-bearing schist with lenses of quartz metaconglomerate ( Figure 4a), referred to as anagenite (Anageniti grossolane of Rau & Tongiorgi, 1974). Barberi et al. (1967) correlated the Barabarca Quartzite Fm. with the Triassic Verrucano of mainland Tuscany, marking the onset of the alpine sedimentary cycle (Rau & Tongiorgi, 1974). The quartz metaconglomerate (i.e. Anageniti grossolane) deposited in a fluvial system, fed mostly by a cratonic source, and its clasts, showing sub-angular to well-rounded shape, indicate prolonged transport (see Cassinis, Perotti, & Santi, 2018 and references therein). On Elba, outcrops of metaconglomerates belonging to the Barabarca Quartzite Fm. are exposed along the coast in the Barabarca and Stecchi localities (Figures 2a,4a), in the footwall of the CSZ (Figures 1, 2a). Inland, the metaconglomerates are poorly exposed, due to the presence of a thick cover of Pleistocene aeolian sandstones (e.g. Figure 4a).
The metaconglomerate lenses contain polycrystalline quartz-and, locally, tourmaline-clasts (tourmalinoite Auctt.), ranging in size between some millimeters to several centimeters (Figure 4b,c), embedded in a quartz-rich matrix, with minor white mica, tourmaline, biotite, cordierite, ilmenite, and hematite. The foliation in the metaconglomerate is marked by sub-millimetric phyllosilicate-rich layers and pressure solution surfaces wrapping the deformed clasts ( Figure 4b). Based on the elongate shape of the clasts and the intensity of the foliation, the metaconglomerate can be classified as a LS-or, less frequently, a L-tectonite, according to the tectonite classification proposed by Ramsay (1967). The L-fabric is defined by the strong preferred orientation of elongated clasts  (Figure 4e). However, such aggregates are sometimes also randomly oriented, suggesting a heterogeneous partitioning of strain. The metapelites can be classified, hence, as S-tectonites to SL-tectonites, according to Ramsay (1967).
In the investigated area, three samples of quartz-metaconglomerate were collected for finite-strain analysis in all the available coastal outcrops (sampling locations are shown in Figure 2a

| STRAIN ANALYSIS
Finite strain analysis was conducted on three samples of quartz metaconglomerate (samples IESP3SP161, IESP3SP162 and IESP3SP164; shown in Figure 2a), representative of all the exposed outcrops (see Figure 2a and Figure 4a), using the Rf/ϕ method and the deformed clasts as strain markers. The Rf/ϕ method relies on several techniques developed to estimate finite strain starting from a dataset of deformed markers, whose shape can be approximated to that of an ellipse (Dunnet, 1969;Lisle, 1977Lisle, , 1979Lisle, , 1985Ramsay, 1967;Shimamoto & Ikeda, 1976), hence suitable for deformed conglomerates.
Samples were prepared cutting two orthogonal slabs parallel to the XZ (perpendicular to foliation, parallel to lineation) and YZ (perpendicular to foliation and lineation) principal planes of the finite strain ellipsoid. The polished slabs were analyzed using the best-fit ellipse tool of the EllipseFit software (Vollmer, 2015) that makes it possible to estimate the aspect ratio (Rf) and the angle with the foliation (ϕ) of each deformed clast. The original orientation (θ) and aspect ratio (Ri) that each deformed object had in the undeformed state are linked to the Rf and ϕ parameters and depend on the axial ratio of the finite strain ellipsoid (Rs). Dunnet (1969) defined a method to plot hyperbolic Ri-and θ-curves for a population of deformed objects in the Rf/ϕ space, which is, however, based on the assumption of initial random distribution.
Following Lisle (1985), the Rf values and the number of objects in a population (n) can be used to derive the harmonic mean H and to perform the symmetry (I SYM ) and χ 2 test. H allows to obtain an approximate estimation of Rs, I SYM estimates the statistical quality of a sample.
If I SYM is high (above the critical values listed in Lisle, 1985), then the data set is symmetrical and the assumption of initial random distribution is correct (see in detail Lisle, 1977Lisle, , 1985. Finally, the χ 2 test allows to obtain a precise estimation of Rs, by calculating the family of θ-curves providing the best fit for the data set. The calculation of H, Rs, and χ 2 for a given set of Rf/ϕ values was done using the Excel spreadsheet compiled by Chew (2003). The dataset containing the polished slab scans and the finite strain data is available for download at Mendeley Data (https://doi.org/10.17632/ymbcxhpgyr.1).

| Results
The estimated values of H, I SYM , χ 2 , and Rs are summarized in Table 1, whereas Figure 6 shows the resulting Rf/ϕ diagrams with Ri-and θ-curves and the vector mean (red lines). The investigated samples are characterized by very high values of I SYM (always above 0.89), indicating that the assumption of initial random distribution is correct. This is consistent with the sedimentological data reported from the anagenites in the northern Apennines, which contain clasts with very weak preferred orientation (e.g. Azzaro et al., 1976;Martini, Rau, & Tongiorgi, 1986;Rau & Tongiorgi, 1974). The samples display high Rf ratio associated with ϕ < 30 on XZ sections and low Rf ratio with a complete ϕ spread on YZ sections ( Figure 6). Estimated Rs value ranges between 2.50 and 2.92 on XZ sections and between 1.17 and 1.38 on YZ sections (Table 1). The K value of Flinn (1965) was derived from the Rs value calculated on XZ and YZ sections, based on the χ 2 test, following the relation: where Rs XY = Rs XZ /Rs YZ . The calculated K values are 3.29 (IESP3SP161), 6.68 (IESP3SP162), and 2.94 (IESP3SP164), which plot in the constrictional field of Flinn's diagram (Figure 7). In Figure 7  Fm. (Papeschi et al., 2017). Although geometrically linked to the Late Miocene contractional event that affected the Calamita Unit (see Musumeci et al., 2015;Viola et al., 2018), the reconstructed total finite-strain ellipsoid marked by the metaconglomerates might have registered a certain amount of pre-Late Miocene strain, related to regional metamorphism during continental collision in the northern Apennines. It is impossible to quantify the role of the pre-Late Miocene deformation in the Calamita Unit, where thermal metamorphism and deformation nearly erased pre-existing structures (see Papeschi et al., 2017Papeschi et al., , 2018. However, the anagenites from the Verrucano of mainland Tuscany, where regional metamorphic temperatures were mostly in the range of 350-500 C (e.g. Franceschelli, Leoni, Memmi, & Puxeddu, 1986;Lo Pò & Braga, 2014;Molli, Giorgetti, & Meccheri, 2000), is characterized by weakly deformed clasts retaining in large part their original shape and sedimentary features (e.g. Cassinis et al., 2018;Martini et al., 1986;Rau & Tongiorgi, 1974). It is, hence, likely that the metaconglomerate from the Barabarca Quartzite Fm. was weakly deformed before the contact metamorphic event, and that, hence, its fabric is mostly the result of the Late Miocene shearing related to the CSZ.
Thermal metamorphism might have induced a strong softening of quartz in the cordierite-bearing Barabarca Quartzite Fm., which allowed deformation to be accommodated within quartz-metaconglomerate. Indeed, deformation temperatures in the range of 400-600 C were suggested for this part of the aureole (Musumeci & Vaselli, 2012, based on the assemblages described by Pattison & Tracy, 1991). The structures preserved in quartz are indicative of dynamic recrystallization with no evidence of annealing or static growth (Figure 5a-d). Although quartz microstructures in the cores of the clasts may be inherited from the protolith, extensive recrystallization is evident along the rims (Figure 5a,b), in strongly stretched quartz, and in the matrix (Figure 5c,d), domains marking the stretching lineation of these rocks. Such structures are indicative of grain boundary migration recrystallization (amoeboid grain boundaries) overprinted by subgrain rotation and bulging (i.e. equant, serrated grains), based on a comparison with the structures observed by Stipp, Stünitz, Heilbronner, and Schmid (2002). High-temperature microstructures overprinted by low-temperature microstructures in quartz suggest progressive temperature decrease (see also Papeschi et al., 2017), assuming that no significant change in strain rate and presence of water during deformation.

| Origin of L-tectonites surrounded by Stectonites in the Barabarca quartzite
The general trend of stretching lineations (Figure 2b, c) and the analysis of deformation mechanisms at the microscale (see above) link the investigated L-and LS-tectonites (Figure 4b, c) to the activity of the CSZ. Any effect related to pluton ballooning (e.g. Ramsay, 1989;Sylvester, Ortel, Nelson, & Christie, 1978)  Holst & Fossen, 1987;Liu, Lin, & Song, 2017). This situation is in contrast with the observed parallelism between object and stretching lineations (e.g. Figure 2c). Moreover, the presence of folds would produce a gradient between hinge zones, where L-tectonites would form, and limbs that would experience deformation under conditions of plane strain, which are not present in the study area.
The localization of L-and LS-tectonites within homogeneous and competent lithologies surrounded by heterogeneous and incompetent lithologies that developed S-and SL-tectonites is similar to the example described by Sullivan (2008). According to this author, constriction in the Raft River shear zone (Utah) was absorbed by more competent metaconglomerates, while simple shear concen- Strikingly different tectonites in associated competent and competent lithologies can hence be explained as a result of strain partitioning (Figure 8b), as it is commonly documented also for folding, boudinage, and foliation development (e.g. Holst & Fossen, 1987;Smith, 1975;Treagus, 1988). Stretching lineations in incompetent metapelites tend to form, at the outcrop-to micro-scale (see also