The Sierra Nevada massif is part of the Betic Cordillera–Rif–Tell orogens that extend along the European–African plate boundary from the southern Iberian Peninsula to northern Africa. These orogens were formed by slab rollback and western migration of the Gibraltar Arc throughout the Neogene (Rosenbaum et al., 2002). Coincident with the translation of the arc, the upper plate experienced shortening, the growth of doubly vergent thrust belts, crustal thickening, and rock uplift (Duggen et al., 2003; Soto et al., 2008; Platt et al., 2013). In the Betics, contraction across the plate boundary was initially directed northward (Sanz De Galdeano, 1990; Lonergan, 1993; Platt et al., 2013). As the contraction continued into the foreland during the late Miocene, it slowed and progressively rotated to the northwest into its present orientation (Mazzoli and Helman, 1994; Rosenbaum et al., 2002). Active tectonics in the Betic Cordillera today is dominated by distributed NW–SE convergence of 4–6 mm yr-1 (Fernández-Ibanez et al., 2007; Koulali et al., 2011; Gutscher et al., 2012; Mancilla et al., 2013) and is accommodated in part on NW–SE-trending normal faults (Martínez- Martínez et al., 2006; Stich et al., 2006; Fernández-Ibáñez and Soto, 2008; Giaconia et al., 2014, 2015; Fig. 2).

The Sierra Nevada massif is a doubly plunging, actively uplifting (Azañón et al., 2015) elongate dome, characterized by medium to low-grade metamorphic rocks stacked in north-verging thrust sheets (Martínez-Martínez et al., 2002). Previous interpretations are that the Sierra Nevada dome was uplifted following top-to-west extension and isostatic rebound after thrust belt formation (Martínez-Martínez et al., 2006). Alternatively, as many culminations exist in orogenic hinterlands, the massif could have been uplifted during contractional or transpressive strain (e.g., Bernini, 1990; Mitra et al., 1997).

To resolve whether the uplift of the Sierra Nevada dome was the result of extensional exhumation or a compressional orogenic culmination, we collected rock fabric (AMS) data in Plio-Pleistocene deposits around the massif to explore the presence of penetrative tectonic fabrics that can contribute additional constraints to the kinematics of dome emplacement. We focused sampling on unburied alluvial fan deposits in Neogene basins that surround the core of the structure (Fig. 3).

We collected samples from six sites distributed around Sierra Nevada, from all structural positions, around the massif in unburied Plio-Pleistocene fan deposits that range from poorly cemented to unconsolidated (Sanz de Galdeano and Vera, 1992; Table A1; Fig. 3). The ages of the deposits sampled were determined from published geologic maps (IGME-1:50000 scale) and bridged the temporal gap between the late Miocene age metamorphic fabrics and the present-day deformation field recorded by GPS geodesy and recent seismicity. At each site, three oriented samples were collected as independent blocks. Before removal from the outcrop, most blocks were hardened with a diluted (∼ 50 %) aqueous solution of sodium silicate (Fig. 4). In the laboratory, two to three oriented cubes (8 cm3) were cut from each block using non-magnetic Teflon knives and enclosed in standard cubic paleomagnetic boxes. The anisotropy of magnetic susceptibility (AMS) was determined with an Agico KLY-3S Kappabridge at Lehigh University. To determine magnetic mineralogy, a heating stage under the presence of an argon atmosphere and a cold stage accessory to the Kappabridge were used.

Results from heating and cooling experiments show a complicated magnetic mineralogy composed of nearly 100 % ferromagnetic (magnetite or hematite) to nearly 100 % paramagnetic mineralogy (clays and iron-rich micas; Fig. 5). Since the kinematic interpretation of each of the specimen is the same regardless of magnetic mineralogy, the details of each specimen are not important for subsequent analysis. There is no correlation between the bulk magnetic susceptibility (km) and the anisotropy of the magnetic ellipsoid (Pj), so a comparison of the principal axis of susceptibility across the various structural positions around the Sierra Nevada massif specimens can provide useful kinematic information (Fig. 6a). Site 6 has a much higher magnetic susceptibility than the other sites because of possible secondary sulfide minerals at the site indicated by the heating and cooling behavior of the MS vs. T experiments.

Nearly all AMS ellipsoids are characterized by a low anisotropy degree (Pj) and oblate ellipsoid shape (T) (Jelinek, 1981; Fig. 6b). The AMS axes' determinations record nearly the same axis orientations. At all sites around the Sierra Nevada, the minimum principal axes, k3, are nearly orthogonal to bedding. The principal elongation axes means are preferentially oriented NNE–SSW to NE–SW (Fig. 3). The orientation of the site-mean magnetic susceptibility axes, k1, is horizontal or very shallowly plunging to the NE or SW (Fig. 3). In general, k1 and k2 are in or near the bedding plane of the specimens, and k2 and k3 do not form a girdle pattern in this principle plane.

The AMS principal axes show a consistency between sites (Fig. 3), so we combine the susceptibility axes' orientation data in Fig. 7. These combined data suggest that during deposition the phyllosilicate grains were oriented with their basal planes parallel or slightly imbricated to the depositional surface. Compaction during dewatering and lithification amplified the initial oblate depositional fabric and was coincident with the formation of the tectonic fabric. Regardless of the magnetic mineralogy of the specimens, a well-clustered minimum susceptibility axis (k3) is present, which we interpret as a compaction fabric in these sedimentary deposits. The possibility of a primary depositional current fabric (imbrication) is unlikely because of an independent paleocurrent study on clast imbrication at Site 3 and Site 4, which shows an eastward rather than westward transport direction during deposition (Carrigan et al., 2018).

Irrespective of the structural position around the Sierra Nevada massif, all sites show a preferred orientation of k1. The mean principal axis of maximum susceptibility is preferentially oriented at 30–210∘ (Figs. 7 and 8). We interpret this as a tectonic fabric due to the tight clustering of k1 and k2, the relationship between k1 and strike of dipping bedding at sites SN1, SN4, and SN6, and the lack of influence from depositional processes. In specimens dominated by phyllosilicate grains, it is difficult to create a strong lineation by aligning grain crystallographic axes; however, an intersection lineation between slightly rotated clay grains orthogonal to a shortening direction has been observed (Henry, 1997; Cifelli et al., 2004; Parés et al., 2007; Martín-Hermández and Ferré, 2007; Borradaile and Jackson, 2010). The orientation of k1 is consistent with the present-day GPS velocity field, being oriented almost perfectly orthogonal to the direction of convergence of the Betic Cordillera to stable Africa (Nubia; Fig. 2; Gutscher et al., 2012), in good agreement with the mineral lineations recorded in the massif's core (Martínez-Martínez et al., 2002), and the Neogene brittle extensional structures and recent seismicity (Mancilla et al., 2013) in the orogen (Fig. 2). Because of the low strains and the orthogonal relationship between contractional and extensional principal directions, it is not possible to distinguish the uplift processes of the Sierra Nevada massif with our results. The AMS ellipsoid orientations, mineralogic stretching lineation from the core of the Sierra Nevada massif, the nearby GPS velocity field, and recent fault slip all have orientations consistent with the same strain field (Fig. 8). The principal elongation direction is interpreted to have persisted across different structural levels from Miocene time to the present (> 10 Myr). The principal elongation direction, k1, was collected from only young sediments so this fabric must also be young. The AMS fabric points out that the various phases of deformation affecting the Betic Cordillera were nearly coaxial since the Miocene.

The Northern Apennines are an accretionary fold and thrust belt (Bally et al., 1986) where crustal deformation, rock uplift, and topographic growth result from the ongoing subduction of Adria beneath Europe (Picotti and Pazzaglia, 2008; Carminati and Doglioni, 2012). The Apennine orogenic wedge initiated ∼ 30 Ma along the southern flank of the Alps (Le Pichon et al., 1971) and has grown at variable rates through the Neogene dependent on the transfer of mass imbricated from the subducting plate (Picotti and Pazzaglia, 2008). Rapid rollback of Adria with respect to Europe results in retreat and stretching of the upper plate, forming a wide zone of back arc crustal extension. The Apennine wedge started to become emergent ∼ 4 Ma (Picotti and Pazzaglia, 2008), uplifting and exposing paired compressional and extensional deformation fronts near the trench and in the forearc, respectively, with the structural transition near the topographic culmination of the range (D'Agostino et al., 2001; Carminati and Doglioni, 2012). Balanced cross sections for the Apennines (Bally et al., 1986; Hill and Hayward, 1988) indicate ∼ 130 to 150 km of subduction over the 30 Myr history of the wedge, which indicates relatively slow long-term rates at ∼ 4 to 5 km Myr-1 (4–5 mm yr-1), similar to the GPS geodetic rates (Devoti et al., 2008; Caporali et al., 2011; Bennett et al., 2012).

The northeastern Apennines, including the Umbria–Marche target region of this research, exposes Mesozoic–early Cenozoic carbonates and middle–late Cenozoic mixed carbonate–siliciclastic rocks folded and imbricated into northeast-vergent thrust sheets (Porreca et al., 2018, and references therein; Fig. 9). In Marche, these thrust sheets are located with carbonate ridges and have inferred blind thrusts in their cores (Artoni, 2013). Further west in Umbria, the thrust sheets are dissected by both east- and west-dipping high-angle normal faults (Barchi et al., 1998; Fig. 9). Ongoing thrust earthquakes beneath the Po Plain and Adriatic Sea (Pondrelli et al., 2006; Boccaletti et al., 2011) and normal-fault-sense earthquakes beneath the high Apennines (Lavecchia et al., 1994; Doglioni et al., 1999; Ghisetti and Vezzani, 2002; Chiaraluce et al., 2017) speak to concurrent shortening and extension in the wedge.

The paired deformation fronts in the Northern Apennines, Italy, are convolved with an enigmatic, but active, east-dipping (towards Adria), 14–15 km deep detachment called the Alto Tiberina fault that projects to the surface west of the Apennine crest (Barchi et al., 1998; Pialli et al., 1998; Boncio et al., 2004; Chiaraluce et al., 1999; Eva et al., 2014; Lavecchia et al., 2016; Fig. 9). This detachment is one of only a handful of low-angle normal faults globally that are demonstrably seismogenic (Hreinsdóttir and Bennett, 2009; Valoroso et al., 2017), apparently in contradiction to frictional fault reactivation theory that predicts that slip on low-angle normal faults as extremely unlikely (reviewed in Collettini, 2011). Most of the destructive seismicity in the high Apennines tends to nucleate on west-dipping high-angle normal faults that are antithetic to and sole into this east-dipping detachment (Galadini and Galli, 2000; Boncio et al., 2004; Roberts and Michetti, 2004). The most destructive seismicity, including the 2016–2017 earthquake sequence, is tightly focused along the highest crest of the Apennines where it is co-located with young, underfilled, extensional basins, high-angle normal faults that rupture the surface (Fig. 9) and geomorphic evidence for an east-marching drainage divide. It is not known if the infrequent but large historic earthquakes east of the divide are indicative of new blind normal faults that have nucleated on the detachment, represent active shortening, or alternatively are responding to a different stress field.

Imbricated foredeep and wedge-top basins contain a time-transgressive range of poorly consolidated deposits that span the compressional and extensional regimes. Conceivably, shortening fabrics could be recorded in lithofacies at the base of one of these basins when it was formed and filled in the shortening part of the wedge, only to be superseded by stretching fabrics in overlying lithofacies as the basin was translated westward and into the extending part of the wedge. Adriatic slope transverse rivers (Alvarez, 1999) traverse both the extending and shortening parts of the wedge and contain Pleistocene alluvial deposits representing an AMS geodetic snapshot of the current crustal strains. Published AMS data from the thrust belt show a strike-parallel (NE–SW and horizontal) extension that is perpendicular to compression and shortening directions (Sagnotti, et al., 1998; Caricchi et al., 2016). To confirm these data towards the southeast and to better locate the kinematic transition region between the contracting and extending regions of the overlying Eurasian plate, we sampled AMS data in Oligocene and younger units, including Quaternary deposits in a NE–SW-oriented corridor across the thrust belt (Fig. 9).

Sampling in the Apennines was designed to identify the location of the modern extensional front. Field collection and specimen preparation occurred as in Example I from Spain, with unconsolidated samples being hardened with sodium silicate before or just after orienting and removal from the outcrop (Fig. 4). We collected samples from 17 sites from sedimentary rocks and poorly consolidated sediments from Late Eocene to late Pleistocene age, with a focus on late Miocene–Pliocene argillaceous marine deposits (Table A1). The Italian specimens were prepared and rock magnetic data were acquired in the Archeomagnetism Laboratory at Centro Nacional de Investigación de la Evolución Humana (CENIEH) (Spain). The AMS of the collected specimens was measured on a MFK1-FA Kappabridge (Agico Instruments), a fully automated inductive bridge, at a frequency of 976 Hz and a field of 200 A m-1. Analysis software (Saphyr6, by Agico) creates a complete susceptibility tensor. Rock magnetic measurements included isothermal remanent magnetization (IRM) acquisition experiments up to 1 Tesla and hysteresis curves to determine the relative contribution of ferromagnetism and paramagnetism to the total susceptibility tensor. These experiments were carried out with a vibrating sample magnetometer (VSM; Micromag 3900).

Samples from the Apennines have variable magnetic mineralogy and include a wider range of lithologies and ages than the Betics sampling. Samples from sites AP2 and AP7 (Bisciaro Fm.) are dominated by diamagnetic calcite and negative mean susceptibility, which precludes any meaningful analysis of the AMS axes' orientations. At the 1T field, the magnetization was not fully saturated, indicating the presence of hematite in addition to lower coercivity magnetite as the dominant ferromagnetic components (Heller, 1978). Still, the bulk magnetic susceptibility is dominated by paramagnetism, as revealed by the hysteresis curves (Fig. 10). The contribution of paramagnetic grains suggests that the measured magnetic fabric can be used as a proxy for phyllosilicate grains' preferred orientation; therefore, the AMS principal axes are indicators of the orientation of the strain axes' orientation (e.g., Soto et al., 2009).

Representative examples of AMS fabrics are shown in Fig. 11. The mean susceptibility shows no positive correlation with the shape parameter or anisotropy degree (T, Pj; Fig. 12). Similar to the data from Spain, the AMS ellipsoids from the Italian specimens indicate low Pj values, revealing a low degree of grain shape preferred orientation and low strains. The AMS axes' distribution is particularly clear in specimens of the argillaceous and semi-consolidated Pliocene Argille Azzurre Fm. At all sites, k1 axes' orientations are shown as a function of rock formation, as well as the sites at which k3 is perpendicular to bedding (Fig. 13). All interpretable specimens from the Apennine range samples, including the Pleistocene fluvial deposits, generate a site-mean AMS fabric consistent with contraction and shortening in the wedge.

The k1 axis orientation is orthogonal to the rock transport and crustal shortening directions, as recorded in GPS geodesy data and seismology (Fig. 9). Irrespective of sample age, we interpret AMS ellipsoids that have the magnetic lineation in a NW–SE orientation as recording contraction, as this is the main trend of the fault traces and strike of bedding and topography (Fig. 13). A few sites do not provide interpretable kinematic results because the axes directions are scattered and suggest inconsistent strain directions. The calcareous marls of the Bisciaro Fm. (AP2, AP7) have a poorly formed AMS fabric. In these specimens, the mean susceptibility is negative and dominated by diamagnetism, most likely calcite. The absence of a compactional fabric in carbonate-dominated specimens (AP2, AP7) likely indicates that these sediments lithified by cementation soon after deposition.

In general, the distribution of the principal axes of the AMS ellipsoid does not significantly vary with stratigraphic age or structural position. For example, the oldest specimens collected from Eocene–middle Miocene marls and Pliocene siliciclastic rocks (AP6, AP14, AP17) uniformly show AMS fabrics consistent with contractional deformation of the orogenic wedge (Fig. 13), similar to the results of Sagnotti et al. (1998). Most importantly, sites collected from thrust structures that are currently in an extending regime (AP11, AP12, AP13) imply that either the AMS fabric was locked after the original deformation due to the high strain required to rotate grain pairs, or that subsequent extension has not affected the previous AMS fabric (e.g., Larrasoña et al., 2004). The same is true for middle and late Miocene siliciclastic deposits next to the Marche ridge (AP3, AP9), where the current orientations of crustal stresses from fault and earthquake data are ambiguous. Pliocene and Pleistocene samples from near the toe of the orogenic wedge show an orientation consistent with ongoing shortening (AP4, AP5, AP8). Wegmann and Pazzaglia (2009) also report ongoing shortening in this region as evidenced by fluvial terrace folding above the Filottrano thrust, which we cross at the location of AP4.

The kinematic transition zone in central Italy aligns with the topography, the seismicity (Pondrelli et al., 2006), and the GPS geodesy (Bennett et al., 2012; Fig. 9). Our AMS data do not improve on the location of the transition zone because of the lack of samples from Plio-Pleistocene deposits directly northeast of the drainage divide (Fig. 13). Unfortunately, the one Pleistocene river terrace deposit northeast of the divide (AP10) has indeterminate axes. As such, our AMS results are not able to support the idea that there is an apparent rotation of the principal compressive stress between the Adriatic coast and the Marche ridge associated with wedge-scale pore-pressure variations (Peacock et al., 2017). Furthermore, the AMS is unable to determine the stress field responsible for the large historic earthquakes in the region between the drainage divide and the Marche ridge. If earthquakes in the region are related to blind normal faults with tips breaking upsection from the Alto Tiberina detachment (Fig. 9), a possible rationale is that, according to extensional critical wedge theory (Davis et al., 1983), a wedge with a taper greater than some critical value is unable to slide over its basal detachment until sufficient wedge thinning on connecting faults reduces the surface slope and wedge taper below the critical value (Xiao et al., 1991). Suitable deposits do outcrop in this critical region, so additional field work and AMS analyses may shed light on this problem.