Stress Field Evolution of the Southernmost Andean Cordillera From Paleostress Analysis (Argentine Tierra del Fuego)

The Argentine Tierra del Fuego comprises part of the roughly east‐west trending southern end of the Andean Cordillera intensely deformed since the Mesozoic. Mesostructures have been measured in Late Jurassic to Miocene rocks. Taking into account statistical criteria to provide a representative stress tensor from a fault population, this study defines 28 paleostress tensors pertaining to 22 sites. The orientation of σ1 shows two main modes trending E‐W to ESE‐WNW and NE‐SW. In addition, extensional sites reveal N‐S, NE‐SW, ESE‐WNW, and NW‐SE horizontal σ3 and vertical σ1. The stress fields obtained are congruous with a regional NE‐SW compressive stress direction active in the study zone since the Late Cretaceous. Shortening was coeval with a 30° counterclockwise rotation of the Patagonian orogenic curve and the indentation of the orogenic wedge against a basement high, the Río Chico Arch, up to the early Miocene. The indentation caused a modification in the orientation of the compressive stress trajectories, showing NE‐SW direction in Sorondo Range sector and NW‐SE in Mitre Peninsula area. Since the late Miocene, left‐lateral activity along the Magallanes‐Fagnano Fault System produced local deviations of the NE‐SW compressive stress toward an E‐W direction. The present‐day stress field is also characterized by NE‐SW subhorizontal P axis derived from earthquake focal mechanisms and geodetic studies.


Introduction
Obtaining paleostress tensors of a determined region allows the stress field acting during a certain geological period to be analyzed (Bergerat & Vandycke, 1994;Eyal & Reches, 1983, among others). Yet stress field is heterogeneous at both the spatial and the temporal scale. If we assume that the stress field remains constant over a particular time interval, it is therefore necessary to provide a sufficiently large data set to be able to detect orientation variations at the regional scale. The variations may be related to diverse tectonic states over time or to the activity of large faults (Armijo et al., 1986;Casas & Maestro, 1996;Maestro et al., 2014;Maestro & López-Martínez, 2011;Simón, 1984Simón, , 1986Tapponier & Molnar, 1976). The relationship between the activity of large faults and the deflection of stress trajectories has been studied through mathematical means and analog models (e.g., Casas et al., 1992, and references herein).
The Fuegian Andes, located in the southern end of the Andean Cordillera, also constitute the northwestern part of the Scotia Arc, which surround the Scotia and South Sandwich plates (Figure 1a). The geodynamic evolution of the Fuegian Andes is connected with the progression of the Scotia Arc, resulting in geological features that distinguish them from the rest of the Andean Cordillera. The external part of the orogenic chain is termed Fuegian thrust-and-fold belt (FTFB) with NW-SE, W-E, and WSW-ENE trends (Figure 1b). The FTFB involves Upper Jurassic-Lower Cretaceous sedimentary rocks deposited in the Rocas Verdes back-arc basin and in the Austral-Magallanes foreland basin during the Late Cretaceous-Neogene (e.g., Calderón et al., 2013). Sedimentary rocks were detached and imbricated along NE-to N-vergent thrusts (Torres Carbonell, Rodríguez Arias, & Atencio, 2017, and references therein; Figure 1c). A major left-lateral strike-slip fault system, the Magallanes-Fagnano fault deforms the FTFB (Klepeis, 1994;Tassone et al., 2005;Torres Carbonell et al., 2008;Figure 1b). This fault zone extends eastward along the North Scotia Ridge and represents the onshore expression of the left-lateral transcurrent boundary separating the Scotia and South American plates (Barker, 2001;Figure 1a). By analyzing brittle mesostructures, we were able to characterize a number of paleostress fields associated with these contrasting tectonic stages of the Fuegian Andes evolution.

MAESTRO ET AL. 7
Tectonics RESEARCH ARTICLE Figure 1. Situation of the study zone of the Argentine Tierra del Fuego in the Scotia Arc. (a) Tectonic and bathymetric map of northern Antarctic Peninsula and southern Scotia Arc regions (Galindo-Zaldívar et al., 2006). Bathymetric map has been carried out from satellite and ship track data (Smith & Sandwell, 1997). The study area is marked by a red box. Earthquake epicenters from National Earthquake Information Centre (USGS, 2013). (b) Simplified geological structural map of Argentine Tierra del Fuego (Olivero & Malumián, 2008;. Contours indicate foreland sediment thickness in kilometers (Biddle et al., 1986;Galeazzi, 1998). (c) Cross-section x-y of the Central Belt, the Magallanes-Fagnano Fault System, and the Fuegian Thrust and Fold Belt modified from Torres Carbonell, Cao, and Dimieri (2017).
The purpose of this investigation is (i) to characterize late Mesozoic and Cenozoic paleostress fields in the Fuegian Andes; (ii) to compare the obtained paleostress with the present-day stress pattern, deduced from focal mechanisms and GPS observations; and (iii) to determine the relationship between the evolution of the FTFB and the direction of the stress fields.

Geological Setting of Tierra del Fuego
The geological evolution of Tierra del Fuego can be divided for the purpose of this article into three major tectonic stages: (1) marine back-arc basin formation during Late Jurassic rifting and Early Cretaceous, (2) inversion of the back-arc basin during Late Cretaceous and propagation of contractional deformation to the foreland until the early Miocene, and (3) strike-slip regime linked to the left-lateral transform boundary between the major plates during and after the late Miocene (~10 Ma).

Back-Arc Basin Development
The first stage involved Late Jurassic continental stretching in the southwestern part of Gondwana that led to development of a rift and to accumulation of volcanic and volcanosedimentary units above Paleozoic rocks (Calderón et al., 2016, and references therein). Lithosphere-scale extension induced sea floor spreading in the Rocas Verdes back-arc basin (Dalziel et al., 1974, among others). Lower Cretaceous Flysch units derived from the arc filled the southern basin border, now preserved in Tierra del Fuego (Olivero & Martinioni, 2001; Figure 1b). The back-arc basin was at that time part of the continental connection between the Antarctic Peninsula and southern South America (e.g., Dalziel et al., 1975, among others). An important basement high located along the northeastern border of the basin, the Río Chico Arch, notably influenced the geometry of the future fold-and-thrust belt (e.g., Biddle et al., 1986;Figure 1b). Its subsurface topography was controlled by high-dipping normal faults of the Jurassic rifting (e.g., Robbiano et al., 1996).

Late Cretaceous-Miocene Basin Closure and Orogenic Growth
The second stage involving back-arc basin closure, tectonic emplacement of ophiolitic complexes, and underthrusting of the South American craton occurred at the beginning of the Late Cretaceous (e.g., Calderón et al., 2012, and reference herein). This process finally led to the collision of the volcanic arc against the craton (Klepeis et al., 2010;Nelson et al., 1980). The collision gave rise to stacking and duplexing in the upper crust connected to thin-skin shortening along the FTFB, deforming the sedimentary cover toward the foreland (e.g., Torres Carbonell & Dimieri, 2013; Figure 1c). Shales and fine-grained sedimentary successions of the back-arc basin act as detachment levels active during the establishment of the Austral-Magallanes foreland basin (Torres Carbonell, Rodríguez Arias, & Atencio, 2017, and references therein).
Paleomagnetic constraints derived from samples of Late Jurassic to Late Cretaceous rocks from the Fuegian Andes core suggest that the initial deformation of the orogenic wedge may have involved counterclockwise rotation of about 30° (Rapalini et al., 2015), whereas data from younger sediments suggest that rotation ceased after 60 to 50 Ma (Maffione et al., 2010;Poblete et al., 2014). Therefore, structural data and analog models also suggest that part of the orogenic curve, the Patagonian curve (or arc), may have formed at the onset of Late Cretaceous orogenesis (e.g., Ghiglione et al., 2016). The final curvature may reflect interaction of the FTFB advance with the Río Chico Arch, which exerted a major buttress along the Mitre Peninsula recess (Torres Carbonell et al., 2016;Figure 1b). At the apex of the recess, this final curvature involved some orthogonal flexure, as can be interpreted from N-S oriented folds that overprint middle Eocene structures (Torres Carbonell et al., 2016).
Ocean floor spreading in the Scotia plate and opening of the Drake Passage and the Western Scotia Sea after the middle Eocene highly condition the evolution of the orogen. The breakup entailed northward displacement of a major block formed by the magmatic arc and the incipient Fuegian Andes orogenic wedge. This block, coeval with the mentioned counterclockwise bending (e.g., Rapalini et al., 2015), pushed the backarc and the synorogenic sequences of the foreland basin against the Río Chico Arch, until the early Miocene .

Late Neogene Strike-Slip Tectonic Regime
During the third tectonic stage, the evolution of the Scotia Arc was further conditioned by the tectonic regime in the studied area. A left-lateral transcurrent boundary between the South American and Scotia

10.1029/2018TC005158
Tectonics plates developed along the North Scotia Ridge (Sue & Ghiglione, 2016, and references therein). This plate boundary is connected to the Magallanes-Fagnano fault system that cuts the FTFB and has generated a maximum left-lateral strike off-set of nearly 50 km and remains active (Bohoyo et al., 2007;Klepeis, 1994;Torres Carbonell et al., 2008;Figure 1a). This transcurrent plate boundary may have originated from a coincidence of (Barker, 2001) the following: (i) the cessation of eastward displacement in the North Scotia Ridge after collision of the East Georgia Rise with the South Georgia Microcontinent, (ii) the beginning of oceanic spreading in the East Scotia Ridge, and (iii) the interruption of seafloor spreading in the West Scotia Ridge.

Materials and Methods. The Palaeostress Database
We studied new 1,496 brittle mesostructures (1,236 faults, 113 joints, 67 tension gashes, and 64 clastic dikes) obtained from 87 sites in Early Jurassic-Late Cretaceous metamorphic and igneous rocks and Paleogene to Quaternary sedimentary rocks ( Figure 2).

10.1029/2018TC005158
Tectonics Paleostress Determination (Galindo-Zaldívar & González-Lodeiro, 1988) methods. The use of different fault analysis methods guarantees the reliability of the obtained results. The paleostress results are mainly based on the fault population analysis; nevertheless, the orientations of joints, tension gashes, and clastic dikes are also useful as a complementary record of paleostress directions. There are numerous studies focused on the mechanical interpretation of joints as stress indicators (e.g., Engelder & Geiser, 1980;Hancock & Engelder, 1989). Tension and shear joints (e.g., Hancock, 1985) provide information of the maximum and minimum horizontal stress direction, but the estimation of the stress ratios cannot be established. Finally, the tension gashes and clastic dikes develop orthogonal to the minimum horizontal stress axis (Ramsay & Huber, 1983). Results are given with orientation of maximum (σ 1 ), intermediate (σ 2 ), and minimum (σ 3 ) stress axis. Furthermore, we indicate σ y (azimuth of the maximum horizontal stress) and the value of R (stress ratio in Bott's equation; Bott, 1959), Rb = (σ z -σ x )/(σ y -σ x ), where σ z is vertical and σ y > σ x (see Table S1 in the supporting information). By plotting the poles of the joints, tension gashes, and clastic dikes planes, a representative minimum stress axis (σ 3 ) is obtained (see Table S2 in the supporting information).
The measured fractures are distributed in 87 stations, where the populations measured in each of them range between 1 and 109 data. Theoretically, only four faults are necessary to obtain stress tensors; nevertheless, this amount of data is not enough to derive a statistically representative solution. The minimum size of the fault population to establish the credibility or significance of the stress tensor solution obtained after the analysis has been widely discussed (e.g., Lacombe, 2012;Simón, 2018). Angelier (1979) and Etchecopar (1984) suggested using at least15 fault data; Casas et al. (1990) propose that this number could be lower and consider acceptable populations between 10 and 12 faults; in other cases, for example, in the case of nearly uniaxial stress regimes, Arlegui and Simón (1998) increases the number of data necessary for the solution to be acceptable to 25-30. Orife and Lisle (2006) establish a method to discriminate between significant and nonsignificant stress tensors. They take into account both the absolute number of faults that define the stress solution and the average misfit angle, and it is based on comparing with artificial solutions inverted from random data sets. In addition to these proposals, other authors (Delvaux et al., 1997;Simón et al., 2000;Sperner & Zweigel, 2010) have attempted to establish a categorization of stress tensor solutions based on a series of parameters that classify them from excellent to unacceptable solutions. In general, the main criteria used in such evaluation are the degree of directional dispersion of the results, the proportion of explained faults respect to the full population, and the mechanical compatibility of fault orientations with the obtained stress tensor. We have thought that it is more appropriate to establish a clear differentiation between acceptable and unacceptable solutions than to classify the solutions according to their quality, for that reason we have used the separation between significant and nonsignificant solutions proposed by Orife and Lisle (2006) to eliminate the most debatable solutions. We have plotted our solutions on the diagram of significance proposed by these authors where the representation shows the average pitch misfit versus the number of faults that explain the solution (Figure 3). The 5% curve determines the boundary between significant (less than 5% of probability of being spurious solutions resulting from a random combination of fault-slip data) and nonsignificant stress tensors. It has also been taken into account, as proposed by the authors, that this quality indicator is not very useful when considering solutions defined from a population of less than 10 samples. Finally, we have only considered as significant solutions in this study 28 paleostress tensors pertaining to 22 sites (see Figure 3). The equal area and lower hemisphere projections with the orientation of faults and slickenside striations data of the sites with significant stress tensors is showed in Figure S1 in the supporting information.
The stress tensors, as well as the compression and tensional directions, are represented using two types of graphics. The spherical projections projection of σ 1 , σ 3 , and σ y axes of the stress tensors obtained from fault-slip data samples, joints, tension gashes, and clastic dikes are shown in Figures 4a-4c. In addition, a two-coordinate diagram shows the relationship between the azimuth of the maximum horizontal  Table S1 in the supporting information) on the significance diagram proposed by Orife and Lisle (2006).

10.1029/2018TC005158
Tectonics stress (σ y ) and the shape ratio of the stress ellipsoid (Rb; Figure 4d). The latter representation only shows the stress tensors assuming one of their axes to be vertical; otherwise, the axis that approaches the vertical is regarded as vertical for plotting the stress tensor. This representation resembles the graphics of the y-R method (Simón, 1986).
The present-day stress orientation has been established from the analysis of the earthquake focal mechanism data (see Table S3 in the supporting information) and even geodetic measurements. There are numerous studies that use the focal mechanisms to establish the recent stress field (e.g., Zoback, 1992;Müller et al., 1992;Heidbach et al., 2007Heidbach et al., , 2010Olaiz et al., 2009;Lund Snee & Zoback, 2016). Lacombe (2012) compares a couple of studies that relate paleostresses obtained from the fault population analysis by stress inversion methods with modern stresses (earthquake focal mechanisms and breakouts) and geodetic measurements, observing a high level of congruence between them. This author considers that their joint use of paleostress orientations from fault slip data and recent stress orientations can be an important contribution to establish the reliability of the reconstruction of paleostresses in areas where the stress orientations have remained stable with time. The focal mechanisms data were analyzed using the 'exact method' of FSMI software (Gephart, 1990) based on Gephart and Forsyth (1984). This method applies a search grid to determine a single stress tensor (three main axes and axial ratio, R = (σ 2 -σ 3 )/(σ 1 -σ 3 ); if σ 1 = σ z , Rb = 1/R; if σ 2 = σ z , Rb = R; and if σ 3 = σ z , Rb = R/(R -1)) that is responsible for the focal mechanisms in regions affected by a homogeneous geological structure. It analyzes the orientations of fault planes and slip directions indicated by a population of earthquake focal mechanisms using inverse techniques similar to those used for calculating the stress  Table S1 in the supporting information). σ y is the azimuth of the major horizontal axis; Rb = (σ z À σ x )/(σ y À σ x ), where σ z is vertical and σ y > σ x (Bott, 1959). The stress regimes corresponding to the different values of the stress ellipsoid shape ratio are indicated at the right of the graph.

10.1029/2018TC005158
Tectonics tensor from field observations of the orientation of striations on exposed fault surfaces. On the other hand, it is necessary to take into account the magnitudes of the earthquakes to establish possible uncertainties in the estimation of the focal mechanism. These uncertainties could occur in focal mechanisms obtained from earthquakes of small magnitude (Mw < 5). Mooney (1989) considers for earthquakes with magnitude greater than 5 (earthquakes of moderate-high magnitude) that have been recorded since the mid-1960s, which a reliable focal mechanism can be obtained. In this study, the focal mechanisms considered are earthquakes produced from 1970 to 2015 and the Mw values are mostly greater than 5 (98%: 33% are Mw ≥ 5, 53% are Mw ≥ 6 and 12% are Mw ≥ 7, see Table S3 in the supporting information). Therefore, we consider that the data of the earthquakes focal mechanisms used in our study are of sufficient quality and confidence, making not necessary to weigh each focal solution according to its magnitude.

Mesozoic-Cenozoic Stress Fields From Brittle Mesostructure Analysis
Analysis of the measured fault orientations in Mesozoic and Cenozoic rocks shows a maximum striking NE-SW, and an ESE-WNW secondary direction, with modal dips between 65°and 80°. From the total population, 208 dextral strike-slip, 182 sinistral strike-slip, 275 reverse, and 571 normal faults were identified. The dextral strike-slip faults show two main orientations ENE-WSW and E-W, and a N-S secondary direction, with a modal dip of 85°. The sinistral strike-slip faults show three main directions NE-SW, ENE-WSW, and N-S, with a modal dip of 85°. The reverse faults show two maxima striking ESE-WNW and NNE-SSW, with a modal dip of 80°. The normal fault planes show a NE-SW main strike, with a modal dip angle of 65°and ENE-WSW to E-W secondary orientations.
The joint data set shows four relative maxima with N-S, NE-SW, ENE-WSW, and NW-SE orientations. The tension gashes show N-S and NE-SW orientation and thicknesses ranging from millimeters to centimeters. Most joints and tension gashes have vertical planes. The clastic dikes have a maximum trending NE-SW to ENE-WSW. Most clastic dikes show dips from 70°to subvertical and an average thickness of 35 cm. The significant stress tensors show a relative maximum of compression direction E-W to ESE-WNW, with Rb relationships varying between wrench (55% of the tensors with E-W to ESE-WNW direction) to triaxial compressional regimes (45% of the tensors with E-W to ESE-WNW direction), but being close to an axial compressional regime (27% of the tensors with E-W to ESE-WNW direction). There is another relative maximum compression with NE-SW orientation, the stress tensors having Rb relationships within a wrench regime (50% of the tensors with NE-SW direction) close to an axial compressional regime (50% of the tensors with NE-SW direction; Figures 4a and 4d). There is also a relative maximum of σ 1 direction NNW-SSE to NNE-SSW, with Rb values within the wrench regime (Figures 4a and 4d). The orientation of σ 3 shows two main modes trending N-S, NE-SW, ESE-WNW, and NW-SE, with Rb relationships within a tensional regime (82% of the tensors) tending to radial tensional regime (18% of the tensors; Figures 4b and d). σ y orientations show a ESE-WNW main trend and three secondary modes with NE-SW, NNE-SSW, and NW-SE directions (Figure 4c).

Spatial Distribution of the Compression Directions
The compression directions (Figure 5a) derived from fault population analysis can be grouped in sets, considering their location and their relationship with existing geological structures:

Spatial Distribution of the Tensional Directions
It is possible to establish some generalizations of the stress tensors with horizontal σ 3 and vertical σ 1 regarding their spatial distribution and their relationship with the tectonic macrostructures ( Figure 5b):   Table S2). Density equal area, lower hemisphere projections, and contour interval 1%, representing plane poles of joints (b), tension gashes (c), and clastic dikes (d).

10.1029/2018TC005158
Tectonics 6. One site shows the direction of the compressive stress tensor and the tensional stress tensors with similar orientation patterns (site 8).
The tensional brittle mesostructures (tension gashes, joints, and clastic dikes) are mainly detected in the southwestern part of the study zone (see Figure 6). From σ 3 orientations, a main NW-SE tensional direction is observed. There is also a relative maxima of σ 3 directions NE-SW. The principal axes of these tensors are arranged parallel (sites 1, 4, 7, 8, 9, and 10), perpendicular (site 3), and oblique (sites 2, 5, 6, and 8) to the trace of the main tectonic structures mapped (Figure 6a).

Chronology of the Stress Tensors
In general, it is very difficult to determine the age of the obtained stress tensors. In our case we do not have the cross-cutting relationships between superposed striae on one fault plane, so the only criterion that gives us information about the relative age of the tensors is the depositional age of the rocks affected by the measured mesostructures, which gives a maximum age. Moreover, the uniformity in the distribution of some tensor orientations in rocks of different age makes possible a first approximation to constrain the span for the related stress field.
The small number of significant tensors obtained in the study area means that any approach to establish the age of the tensors has a certain degree of speculation. In spite of this, taking into account the above premise, we can establish a series of time intervals in which the different directions of both compressional and extensional tensors have acted in a preeminent way. Then, taking into account the age of the rock units affected by the faults, we could interpret that the NNE-SSW and NNW-SSE compression are not recorded in rocks younger than the Lower Cretaceous and NE-SE and E-W to ESE-WNW σ 1 are recorded at least until the Eocene, which is the youngest age of the rock units in which faults have been measured (Figure 7a). On the other hand, the N-S minimum stresses are not recorded in rocks younger than Upper Cretaceous to Paleocene and NE-SW, ESE-WNW, and NW-SE σ 3 are present at least until the Eocene to Miocene (Figure 7b).

Discussion
We compared the interpreted chronology of the obtained stress tensors (see above), with the known geodynamic history of southernmost South America, as established in previous research. The following discussion, therefore, puts forward a model to explain how the stress fields evolved during the evolution of the southernmost Andean Cordillera, which honors both our data and previous tectonic models.

. Mesozoic to Cenozoic Compressive Stress Field
The closure of the Rocas Verdes back-arc basin started in the beginning of the Late Cretaceous due to the rapid westward motion of the South American Plate relative to a mantle reference frame, enhancing compression along the western boundary of South America (Dalziel, 1986). During the closure of the basin, obduction of the oceanic crust and underthrusting of the South American continental margin took place (Calderón et al., 2012;Klepeis et al., 2010;Nelson et al., 1980), ending with the collision of the South American continental margin with the magmatic arc that rimmed the Rocas Verdes basin on its Pacific sector (Cunningham, 1995;Klepeis et al., 2010;Maloney et al., 2011;Nelson et al., 1980). This produced large-scale folding and thrusting of the continental margin (Klepeis et al., 2010;. Subduction of the former Phoenix Plate below the Pacific-side arc continued coeval with closure and collision (Mukasa & Dalziel, 1996).
In our schematic model (Figure 8), we interpret that this orogenesis started under a compressional stress regime with NE-SW direction, caused by the tectonic underthrusting within the closing Rocas Verdes basin. Progression of deformation resulted in the development of the mesoscale fractures when the rocks in the Fuegian Andes were already subjected to an initial stage of ductile deformation and partial uplift, which favored the development of brittle structures (Figures 8a and 8b). In this context, the collision of the magmatic arc involved up to 30°counterclockwise rotation which, in the deformed Rocas Verdes basin rocks (Rapalini et al., 2015). Likewise, Poblete et al. (2016) show paleomagnetic results obtained from Cretaceous to Eocene rocks that indicate counterclockwise rotations varying from~30°to more than 90°in the magmatic arc (Fuegian backstop). In our study area, rotation of the orogenic wedge is younger than~70 and older than~50 Ma (Maffione et al., 2010;Rapalini et al., 2015).
The regional counterclockwise rotation accordingly developed a progressive arc structure related to the folding with vertical axis of the fold-and-thrust structures, the original NW trend acquiring a WNW to E-W orientation (Poblete et al., 2016;Torres Carbonell et al., 2014). This process rotated the successively forming mesofracture sets, which therefore indicate rotated paleostress tensors (e.g., NE-SW to N-S; Figures 7a and 7b). It also implies a replacement of the regional NE-SW compression by a local NNW-SSE to NNE-SSW compression due to the push of the rotating magmatic arc on the foreland sequences, eventually incorporated into the orogenic front. This effect can be locally seen from the orientation of σ 1 derived from the Stage 1 disjunctive foliations in the Late Cretaceous paleo-orogenic front (cf. Torres Carbonell et al., 2013; Figure 8a). The consequence of a rotating backstop on changing contraction directions along the belt has also been reproduced using analog models with a tectonic configuration almost identical to that of the Fuegian Andes (Torres Carbonell et al., 2016).
The FTFB propagation toward the foreland during the Late Cretaceous to early Neogene was hindered the buttressing effect of a notable basement promontory called the Río Chico Arch (Torres Carbonell et al., 2016;Torres Carbonell, Rodríguez Arias, & Atencio, 2017). This promontory was located in the NE border of the Rocas Verdes basin and was inherited from the Late Jurassic rifting stage (Biddle et al., 1986;Yrigoyen, 1989). There is a rough coincidence of the southern border of the Río Chico Arch with the FTFB front in eastern Tierra del Fuego (Mitre Peninsula). The effect of the interaction between the Río Chico Arch and the FTFB propagation was registered in the stress field variation observed in the studied area (Figure 8b): the western area of the study zone shows a NE-SW dominant orientation of the σ 1 , and in the northern sector of Mitre Peninsula, coinciding with the southern border of the Rio Chico Arch, the stress tensor obtained from the fault population analysis shows a main NW-SE orientation for the average maximum compressive direction (Figure 8b). Deformation in the FTFB continued until the Miocene through the northward propagation of thrust sheets that molded the Río Chico promontory (Ghiglione et al., 2010;Ponce et al., 2008;Torres Carbonell, Rodríguez Arias, & Atencio, 2017).

Late Neogene Strike-Slip Stress Field
Regional neogene strike-slip faulting in the Fuegian Andes is related to creation of the sinistral transform fault that bounds the Scotia Plate to the north. In Tierra del Fuego this regime is manifested by the Magallanes-Fagnano Fault System, with a strike separation of~50 km imposed on the FTFB (Torres Carbonell et al., 2008). This transcurrent fault shows in some segments, a normal component with a vertical displacement below 3 km (Klepeis, 1994). Taking the geodetic slip rates established for this fault by Smalley et al. (2003) and Mendoza et al. (2011), which are about 6.6 ± 1.3 to 4.4 ± 0.6 mm/a, the strike-slip faulting would have started in the Late Miocene, between~7 and~11 Ma (Torres Carbonell et al., 2008).
This age is in accordance with the initiation of the sinistral regime along the South Scotia Ridge (Bohoyo et al., 2007). Bohoyo et al. (2007) propose that the oceanic spreading of the southern Scotia basins, between late Oligocene and middle Miocene, must have formed in a dextral strike-slip tectonic regime of the South

10.1029/2018TC005158
Tectonics Scotia Ridge in relation to the eastward movement of the continental blocks. The East Scotia Ridge was characterized by asymmetric spreading from the middle Miocene to~10 Ma, when the central part of the South Sandwich Arc was affected by a ridge jump and the western edge of the East Scotia Ridge was affected by slow spreading until~6.5 Ma (Larter et al., 2003). On the other hand, the Shackleton Fracture Zone in the western border of the Scotia Arc acted as a purely oceanic transform fault until spreading ceased in the West Scotia Ridge at 6.5 Ma (Maldonado et al., 2000) and the Antarctic-Phoenix Ridge at 3.3 Ma. Synchronous with a decrease in the West Scotia Ridge oceanic spreading, the Shackleton Fracture Zone began to uplift 8 Ma ago. This amount of tectonic processes occurring since 8 Ma ago along the Scotia-Antarctic plates edge has been related to the stratigraphy of mass-transport deposits in the region by Pérez et al. (2016).
Surface ruptures along the Magallanes-Fagnano Fault System form an E-W~10-km wide zone (Lodolo et al., 2003;Pedrera et al., 2014;Sue & Ghiglione, 2016;Torres Carbonell et al., 2008). Strike-slip and oblique faults in some scattered outcrops reveal kinematics consistent with regional sinistral strike-slip tectonics, everywhere postdating the compressive structures of the FTFB (Klepeis, 1994;Torres Carbonell et al., 2008. The sinistral movement of the fault system is congruent with a regional NE-SW late Miocene compressive stress field. Deviations of its trajectory occur in the Magallanes-Fagnano Fault System, as seen in some sites situated near to the fault zone in the northern and southern blocks; the tensors obtained in this work can be interpreted as the deflected stress trajectories due to displacement of the fault system ( Figure 8c). In general, mathematical models show how trajectories tend to be parallel to the fracture plane at the tips toward which the block moves and tend to be perpendicular to the opposite tips. The study area is located in the central segment of a large fracture zone that extends westward to the subduction zone of the Pacific Plate under the South American Plate and to the east extends far along of the North Scotia Ridge (see Figure 1). Therefore, we can underestimate the anchoring effect at the fault tips and attend to how the stress field deflection occurs in the central zones of the faults. Regarding this, there are numerous studies based on mathematical models and data obtained from the fault populations analysis and breakout wells (e.g., Aleksandroski et al., 1992;Homberg et al., 1997Homberg et al., , 2004Reynolds et al., 2005;Yale et al., 1994). The 2-D modeling of these studies shows that the orientation of the local stress perturbation closely parallels the fault orientation along the midsection of the fault in situations where there is a small angle between the maximum horizontal compressive stress orientation and the fault. Tingay et al. (2006) moreover considers the deflection processes of the stress field near geological structures to result from structures acting as a mechanical discontinuity. They consider that an open or very weak fracture in the subsurface will act as a free surface and be unable to sustain shear stresses. As a result, the stress field must be locally reoriented in the vicinity of the fracture so that one principal stress acts perpendicular to the fracture. Similarly, stresses will be locally deflected or refracted near the boundary between mechanical contrasts. In general, it is predicted that the maximum horizontal stress orientation will be deflected subparallel to mechanically weak structures. In the southern and northern blocks of the Magallanes-Fagnano Fault System, stress trajectories change from NE-SW to E-W direction near the fault zone. This agrees not only with a sinistral movement of the central part of the Magallanes-Fagnano Fault System but also with the NE-SW regional main compressive direction.

Tensional Stress Field: Local Vs. Regional Interpretation
In order to interpret the significance of the obtained extensional stress tensors, it is necessary first to constrain their relative age with respect to the geodynamic evolution of Tierra del Fuego during the Mesozoic and the Cenozoic. In that sense, we highlight that many of the faults that explain the tensional tensors appear to postdate the contractional structures of the Fuegian Andes; these cases the maximum compressive direction is almost vertical, suggesting that no major titling of the tensional tensor due to folding occurred. In fact, it has been possible to determine, from the existence of some fault planes with superposed striae, that the tensors with σ 1 in the horizontal are prior to the tensors with σ 3 in the horizontal and σ 1 in the vertical. It is difficult to confirm the relationship between tensors and major structures in the field, given that the folds in the thrust-fold belt are of large wavelengths (a few kilometers), thus exceeding the scale to which one could constrain timing of the small scale fractures studied in our work. However, we can argue that there are no major extensional tectonic features of an age pertaining to the late Cretaceous-early Miocene evolution of the thrust-fold belt (e.g., Torres Carbonell, Rodríguez Arias, & Atencio, 2017). Therefore, it is more probable that the tensional tensors formed after the contractional structures, or even simultaneously with them, as explained in (2) below.

10.1029/2018TC005158
Tectonics In this context, we consider that there are two possible explanations for the tensional stress tensors: 1. That these tensional tensors pertain to the episode of late Cenozoic regional uplift in Tierra del Fuego and southern Patagonia, which is independently constrained by few apatite fission track ages as young as 9 Ma in Cordillera Darwin (Nelson, 1982), positive dynamic topography for the last 10 Ma in the region (Dávila et al., 2018), and an important Holocene uplift due to isostatic rebound and/or recent tectonics after the last glacial maximum (Gordillo et al., 1992;Ivins & James, 2004;Mendoza et al., 2011). In the case of recent tectonics, the uplift may be related to local deformation related to the Magallanes-Fagnano fault system. The late Cenozoic uplift could have caused some extent of collapse of the mountain belt through extensional deformation, explaining the tensional and radial tensional regimes existing in some places (sites 9 and 13). 2. Overall, the extensional tensor trajectories from many sites are very alike the pattern of σ 1 trajectories. In these cases both parallel and orthogonal relationships are seen between folds-thrusts and σ 3 directions. The similarity between the orientation of the compressional and extensional directions can be explained considering the interaction between the orientation of the regional stress field and the kinematics of thrusts and associated folds. Extension perpendicular to fault-bend folds is known to be accommodated by external hinge faults ( Figure 9) and has been described in the context of sequential compression and extension parallel to the transport direction in the associated thrust (Cooper, 1992;Turner & Hancock, 1989;Wojtal & Pershing, 1991). Conversely, slip on arcuate thrusts is known to cause extension parallel to the thrust-strike due to flexure of the convex hanging wall (Figure 9; Geiser, 1988;Marshak et al., 1992;Martínez-Peña et al., 1995).
There are several studies that show many examples of extension processes within folds (e.g., Cosgrove & Ameen, 2000;Lemiszki et al., 1994;Srivastava & Engelder, 1990). Normal faults formed in association with bend folding may be the result of local stresses generated due to extension in the external arc. The relationship between extensional structures-normal faults, joints and tension gashes-and fold geometry is usually quite simple. The local stress perturbations generated as a result of bending could produce an interchange of the principal stress axes that in turn causes these extensional structures to originate both parallel and perpendicular to the fold axis. Similarly, thrust faults form in the hinge regions below the inner arc of the fold, in the area of localized compression. The development of normal and/or thrust faults in the hinge region is controlled by the effective thickness of the layer which, in turn, may cause the migration of the neutral surface either up or down.
Thus, a second possibility is that the tensional stresses recorded are related to local stress perturbations associated to the development of folds in the thrust-fold belt (Figure 9). This is consistent with the lack of firstorder extensional tectonic features in the thrust fold belt and would limit the tensional tensors not to regional stress fields but to local conditions. Furthermore, it is not unusual to find in the same site or in very close sites in rocks of the same age, superposed tensors where the compressive stress direction (σ 1 ) of one phase is arranged subparallel to the extensional stress direction σ 3 of the other (see Figure 5). For example, between sites 1 and 10 and in stations 8 and 9, it is possible to observe the marked parallelism that exists between the compressive (σ 1 ) and extensional (σ 3 ) stress orientations, with N-S, NE-SW, and E-W directions, with an average angular variation of 4°between them ( Figure 5). This would corroborate that an important number of the extensional stress directions determined from the fault population analysis could correspond to local deformations linked to the development of folds and thrusts.

Recent and Present-Day Stress Field
The present-day stress field in Tierra del Fuego can be obtained with earthquake focal mechanisms (Table S3 in the supporting information). An approximately regional NE-SW compression in the region is suggested by the World Stress Map (Heidbach et al., 2007) from the U.S. Geological Survey National Earthquake Information Centre (United States Geological Survey (USGS), 2013) and the Harvard Seismology Centroid Moment Tensor Tectonics Catalogue (Dziewonski et al., 1981;Ekström et al., 2012). This is consistent with the sinistral strike-slip between the Scotia and the South American plates (Galindo-Zaldívar et al., 1996;Giner-Robles et al., 2003;Pelayo & Wiens, 1989).
In order to identify the present-day stress, we analyzed earthquake focal mechanism data from Pelayo and Wiens (1989), the U.S. Geological Survey National Earthquake Information Centre (USGS, 2013), the Harvard Seismology Centroid Moment Tensor Catalogue (Dziewonski et al., 1981;Ekström et al., 2012), and Bollini and Sabbione (2014). The data include 56 events with magnitudes between 4.2 and 7 ( Figure 10). The analysis of the earthquake focal mechanisms using the exact method of FSMI software (Gephart, 1990) gives for Tierra del Fuego and the northern Scotia Arc (inset in Figure 9a) a NE-SW subhorizontal P axis (trend/ plunge: 056/10) and orthogonal T axis (trend/plunge: 325/05). The axial ratio (R) is about 0.35 (Rb = 0.35), supporting a triaxial stress compatible with the strike-slip and reverse-oblique fault types associated, in turn, with the individual focal mechanism and with the transpressional left-lateral regime that occurs along the North Scotia Ridge.
Geodetic studies determining recent displacement rates on Tierra del Fuego show an NE-SW direction, congruent with the compressional direction established from the earthquake focal mechanisms. The displacement direction is N036E south of the Fagnano Lake and N011E in the north (Mendoza et al., 2011(Mendoza et al., , 2015. The horizontal velocities average 1.3 cm/a (Mendoza et al., 2011(Mendoza et al., , 2015. The strongest deformation is concentrated along the Magallanes-Fagnano Fault System, approximately 30-km wide (Mendoza et al., 2011). The shortening orientation is NNE-SSW to NE-SW, and the extensional orientation is NW-SE (Mendoza et al., 2011), compatible with the left lateral movement of the Magallanes-Fagnano Fault System and the earthquake focal mechanisms obtained. Finally, the slip-rate reported from GPS data along the Fault System varies closely between 6.6 ± 1.3 (Smalley et al., 2003) and 4.4 ± 0.6 mm/a.

Conclusions
From the analysis of fracture data measured at 22 sites in Tierra del Fuego a total of 28 paleostress tensors have been obtained. In addition, 56 earthquake focal mechanisms were analyzed to determine the present-day orientations. The direction of σ 1 displays E-W to ESE-WNW and NE-SW main modes with most of the stress tensors having Rb relationships within a wrench regime close to an axial compressional regime. The orientation of σ 3 shows four modes oriented N-S, NE-SW, ESE-WNW, and NW-SE, whose Rb relationships indicate a tendency to radial tensional regime. Present-day stress orientation in the study zone shows NE-SW P axis and orthogonal NW-SE T axis with triaxial ellipsoids.
The stress orientations determined were interpreted in the tectonic context of Tierra del Fuego: 1. The closure of the Rocas Verdes basin during the Late Cretaceous involved first-order thrusting and folding of the continental margin under a contractional stress regime with NE-SW direction. Local deviations of the stress tensors were related to the orogenic curve of the backstop of the Andean chain that rotated Figure 10. Present-day stresses from earthquake focal mechanism solutions in the study region derived from different sources (see Table S3 in the supporting information). Mechanism solutions are plotted as lower hemisphere equal-area projection, white quadrants indicating compressional P wave. Seismicity of magnitude greater than 5 Mw from National Earthquake Information Centre database is represented by red circles. Inset with the regional focal mechanism solution derived from the integration of all focal mechanisms. The P axis and T axis using the exact method of FSMI software (Gephart, 1990) are indicated. Bathymetry map is derived from GEBCO2014 database.
the initially formed fractures, and caused a change in the direction of the local stresses at the orogenic front, from NE-SW to almost NNW-SSE to NNE-SSW. 2. From the latest Cretaceous to the early Neogene, the effect of the interaction between the Río Chico Arch and the advance of the Fuegian fold-and-thrust belt is registered by the different stress orientations. The western area of the study zone shows a predominantly NE-SW orientation of σ 1 , while in the eastern sector of the Mitre Peninsula recess the stress shows a main NW-SE average direction of compressive stress. That is, the stress trajectories converge toward the Río Chico Arch basement promontory. 3. After the late Miocene, the left-lateral movement of the Magallanes-Fagnano Fault System is congruent with a regional NE-SW compressive stress field. This stress field shows some deviations of its trajectory along the Magallanes-Fagnano Fault System, as seen in some sites near the fault zone in the northern and southern blocks, where the stress trajectories deviate from a NE-SW direction to E-W. 4. Many tensional tensors appear to postdate the contractional structures of the Fuegian Andes. In these cases they could be related with an episode of late Cenozoic regional uplift in Tierra del Fuego and southern Patagonia suggested by previous research. The late Cenozoic uplift could have caused some extent of collapse of the mountain belt through extensional deformation, explaining the tensional and radial tensional regimes determined in some sites. In other cases, the tensional tensors show subparallelism between superposed σ 1 and σ 3 trajectories, which in turn are parallel to folds and perpendicular to thrusts. We explain the tensional stresses in this context as the effect of local extension due to the mechanics of thrusts and associated folds formed by bending. 5. Finally, the present-day stress field in the Tierra del Fuego and the northern Scotia Arc is characterized by a NE-SW subhorizontal P axis.