G EODYNAMIC CONDITIONS FOR C ENOZOIC ACTIVATION OF TECTONIC STRUCTURES IN S OUTHEASTERN M ONGOLIA

: The knowledge of the neotectonic structures in Southeastern Mongolia, that is considerably distant from the active plate boundaries, is important for determining a source of tectonic deformation and regular features of ac‐ tivation in the intracontinental setting. Our research was focused on the East Gobi and South Gobi depressions located in Southeastern Mongolia, which developed since the Mesozoic and were activated to various degrees in the neotec‐ tonic stage. The study aimed to assess the paleostress state of the crust in Southeastern Mongolia, identify the stages, factors and mechanisms of the Cenozoic activation of the regional structures of different strike, and determine the sources of activation. The analysis of the available literature suggests a similar history of their development in the Late Jurassic – Early Cretaceous (rifting) and Late Cretaceous – Paleogene (tectonic quiescence). In the Cenozoic stage, the depressions experienced activation of completely different styles. In the East Gobi depression, left‐lateral strike‐ slip faults were activated in the Tertiary, and the post‐Late Cretaceous thrusting took place along the northeastern faults on the northern slope of the Totoshan uplift. In the Early Cenozoic, the N‐S and N‐W compression was dominant as evidenced by the deformed Late Cretaceous sediments and the reconstructed stress tensors typical of the compres‐ sion and transpression regimes. An overview of the published data suggests that the most probable cause of such de‐ formation was the impact of the Western Pacific zone of plate interaction. However, a potential influence of compres‐ sion at the early stages of the Indo‐Asian collision cannot be completely excluded. The East Gobi depression was low active in the second half of the Cenozoic. In contrast to the East Gobi depression, the South Gobi activation began in the Late Cenozoic (Late Miocene – Early Pliocene). Young uplifts and forbergs (Gobi Altai eastern termination) deve‐ loped actively and ‘cut’ the sediments of the basins originating from the Mesozoic. The W‐E and N‐W strike‐slip and thrust faults were active in the Pliocene–Quaternary. The stress field reconstructions show compression, transpres‐ sion and strike‐slip regimes with the NE‐trending axis of compression. Deformation in the East Goby Altay (as well as in Western and Southwestern Mongolia) is driven by the India‐Eurasia collision.


INTRODUCTION
Neotectonic structures in Southeastern Mongolia are an interesting object of studies aimed at establishing the regularities in activation of tectonic deformation in the intracontinental conditions. It is challenging to determine the source of deformation in this region located far away from the active borders of the lithospheric plates. Besides, the change in the strike of the main elements from W-E to northeastern was revealed in the ancient regional setting [Badarch et al., 2002]. Moreover, the modern terrain varies from mountains in the west (Gobi Altai) to weakly dissected flat areas in the east (Gobi plain). In the Late Cretaceous -Paleogene, the entire study area developed in condition of tectonic quiescence and terrain denuda-tion, and the sediments of that period are overlying in most of the area [Yanshin, 1975]. Our study was carried out in Southeastern Mongolia with the focus on the East Gobi and South Gobi depressions originating from the Mesozoic and activated to various degrees in the neotectonic stage (Fig. 1).
The traces of the Cenozoic tectonic activity significantly differ in the studied depressions. The South Gobi depression was replaced by the mountains represented now by the eastern termination of Gobi Altai, including the Gurvan-Saikhan and Hurhe-Ula uplifts (Fig. 2, a), while low-amplitude linear uplifts and NE-trending basins developed within the limits of the East Gobi depression (Fig. 2, b).
The above-mentioned regions differ in the levels of modern seismic activity (Fig. 3). In the Gobi Altai, many high-magnitude earthquakes occurred, and their epicenters formed clusters along the main W-WS and E-SE-trending faults. In the Gobi plain, dissipated seismicity is dominant, and the recorded earthquake magnitudes do not exceed 5.6. The reconstructions show focal mechanisms of strike-slip, reverse fault, and reverse fault with strike-slip component. The reconstructed stress tensors for the modern stress field show that the Gobi Altai is dominated by transpression with the NE-SW-trending axis of maximum compression, while the Gobi plain (i.e. the eastern and NE areas) is under compression with the ENE-WSW-trending axis of maximum compression.
The study area is thus divided into the highly active western and low active eastern parts in the Late Cenozoic and the modern period. Considering the microplate tectonics of Asia mapped in [Zonenshain, Savostin, 1979;Bird, 2002], the eastern area is a part of the Amur lithospheric microplate, and the western zone belongs to the orogen. The presence of a larger submeridional structural boundary along approximately 95-105°E, 1 -areas with platform cover sediments; 2 -strike-slip faults; 3 -thrusts.
which divides the entire continental Asia into the western and eastern parts, was proved by Semen I. Sherman in his publications on the modern geodynamics of Asia (e.g. [Sherman et al., 2015[Sherman et al., , 2017). He stated that the main causes for the existence of this boundary included strong meridional compression due to the plate collision and the high rates of modern deformation in the western part, and lacking/less active collision processes in the eastern part.
Our study aimed to assess the paleostress state of the crust in Southeastern Mongolia, identify the stages, factors and mechanisms of the Cenozoic activation of the regional structures of different strike, and determine the sources of activation. In the first stage of investigations, we collected and processed the data on tectonic fracturing, displacements and morphotectonic features of active faults in the South Gobi and East Gobi depressions. In the second stage, local stress tensors were reconstructed and analyzed. The regional component of the stress field was indentified for different stages of development and compared with the published data on the Mesozoic-Cenozoic deformation of the adjacent territory of China.

RESEARCH METHODS
The database on tectonic fracturing and tectonic deformation in the study area was consolidated using conventional geological, geomorphological, and geological-structural methods. Special attention was given to collecting the data on fault zones that were active in the Late Cenozoic and/or active at the modern stage. Initially, a map of active faults was compiled on the basis of satellite images and GTOPO-30, a global raster Digital Elevation Model (DEM). The map was then validated by field observations of the identified linear structures.
Dating of the fractures and deformations was challenging as most of the fracturing data was collected in the crystalline basement rocks, while outcrops of the Cenozoic rocks with easily traceable fractures were rarely observed. Fractures with calcite, zeolite and iron hydroxide (secondary mineralization) were considered indicative of subsurface conditions of their formation and, correspondingly, of the Late Mesozoic or Cenozoic age. Therefore, striations on the rocks with secondary mineralization can provide a basis for reconstructing the Cenozoic stress field. Besides, of great importance is dating of multidirectional striations on the same plane: the age relationships are used to distinguish the evolution stages of the stress field. In our study, striations on gouge in the rock crushing zones were considered as markers of the Cenozoic shear. Secondary mineralization, containing chlorite and epidote, was dated pre-Cenozoic and pre-Mesozoic due to its formation at depth. Accordingly, the reconstructed stress fields were dated pre-Cenozoic. The near-surface defor- Рис. 3. Карта механизмов очагов землетрясений (а) и современное напряженное состояние Юго-Восточной Монголии (b). mations recorded in the active fault zones, such as the shifts of the valleys of temporary streams, deformations of the Cenozoic sediments, and etc. are also support for determining the Cenozoic stress fields.

OF THE STUDY AREA
The paleogeodynamic reconstructions show that after the Mongol-Okhotsk (MO) Ocean closure due to the collision of the Northern Mongolian block and the Siberian plate at the turn of the Early -Middle Jurassic, a vast orogen was formed in the territory of the contemporary Transbaikalia, Mongolia and Northern China, and existed till the beginning of the Late Jurassic [Zorin, 1999, Wang et al., 2012, Meng, 2003. In the Late Jurassic and Early Cretaceous, a system of depressions occurred from Transbaikalia and Mongolia and Northeastern China (see Fig. 1): the Transbaikalian system of basins, East Gobi, South Gobi, Tamtsag, Hailar, Erlian, Yingen and Songliao depressions. Some of the basins in Transbaikalia, Northeastern Mongolia and Northeastern China were accompanied by the formation of the metamorphic nuclei complexes. They bordered by gently dipping normal faults and have long and narrow boundaries (e.g. [Wang et al., 2013;Zorin, 1999;Daoudene et al., 2009;Davis et al., 1996Davis et al., , 2002Ritts et al., 2010;Mazukabzov et al., 2006Mazukabzov et al., , 2011Sklyarov et al., 1997]). Some basins, specifically the East Gobi, Songliao, Yangen, Erlian and Hailar, are described as rifts with thick sediments, which are bordered by steeplydipping faults (e.g. [Wang et al., 2012;Meng, 2003;Graham et al., 2001;Johnson 2004;Feng et al., 2010]). According to [Wang et al., 2012], the Late Jurassic-Early Cretaceous extension was not took place simultaneously across the entire Northeastern Asia, but began in Mongolia and the China-Mongolia border areas and then spread into other regions in the southeast. To the west of this region, the territory belonging now to West China, West Mongolia and the Tien Shan was under compression due to collision processes at the southern boundary of Eurasia. From the Triassic to the Early Cretaceous, the collision between the Qiangtang, Lhasa blocks and Eurasia took place. In the Late Jurassic -Early Cretaceous, sedimentation occurred in ramp, foreland and pull-apart basins [Caroll et al., 2010;Vincent, Allen, 1999;Johnson et al., 2015;Cunningham, 2013]. At the same time, evidence of the Late Jurassic-Early Cretaceous rifting is found practically everywhere in the Gobi Altai. The intermountain depressions are filled with Jurassic-Cretaceous sediments, overlain by young Neogene-Quaternary sediments, and the Jurassic-Cretaceous deposits are uplifted and eroded in young ridges [Cunningham et al., 2009;Cunningham, 2010Cunningham, , 2013Horton et al., 2013]. The passively uplifted Cretaceous normal faults were discovered in the Ih Bogt ridge of the Gobi Altai [Cunningham, 2013]. Extension within the modern Gobi Altai is also evidenced by metamorphic core complexes, such as an MCC exhumed by the Cenozoic uplift in the northwestern part of the Altan Uul ridge [Cunningham et al., 2009;Cunningham, 2013], and the Yagan Onch Hairhan MCC at the southern border of Mongolia and China [Webb et al., 1999]. According to the recent publications, the Valley of Lakes, in particular its northeastern part covered by seismic surveys, was also involved in rifting in the Late Jurassic -Early Cretaceous, and the accumulated sediment thickness amounted to 4-5 km. It is suggested that this Mesozoic depression formed due to transtension. Starting from the Neogene, transpression, folding and reverse thrusting on normal faults took place in this region Johnson et al., 2015]. Extension and rifting in the Late Jurassic -Early Cretaceous and the inversion of the regime in the Late Cenozoic was discovered up to the southern Mongolian Altai, including the Dzereg, Dariv and Shargiin Gobi basins [Howard et al., 2003[Howard et al., , 2006Cunningham, 2013]. However, the NW orientation of these basins may be questioned, considering rifting at the background of general NW extension.
In some publications, the final closure of the MO Ocean is shifted to the Late Jurassic-Early Cretaceous, and extension is excluded from that period. In the Late Jurassic, the MO Ocean was not completely closed, as shown by the paleomagnetic studies of the Upper Jurassic basalts from the volcanic tectonic structures of Transbaikalia and the calculations of the paleomagnetic pole of rotation [Metelkin et al., 2007]. It should be noted that the authors make a reservation that the difference in the positions of the Middle-Late Jurassic paleomagnetic poles of Northern China and Mongolia and South Siberia may be due to not only of the open MO Ocean, but also the contraction and deformation of the continental crust during the collision [Metelkin et al., 2007]. Based on the AFT analysis, Van Der Beek et al. [1996] relate the phase of rapid cooling in the Early Cretaceous and, accordingly, uplifting in the Baikal region and Transbaikalia to the Late Jurassic-Early Cretaceous orogeny caused by the final closure of the MO Ocean. They associate the exhumation of MCC in Transbaikalia with the same phase. Based on the paleostress reconstructions from the data on tectonic fracturing of the Early Cretaceous rocks in Transbaikalia, the stage of N-S compression was identified [Delvaux et al., 1995]. This stage was dated to the Late Jurassic -Early Cretaceous and also associated with the closure of the MO Ocean and the formation of the orogen. However, later publications on Southeastern Mongolia and Northeastern China consider the inversion stage with the NW-trending compression at the turn of the Early and Late Cretaceous [Graham et al., 2001;Johnson, 2004].
The source of the Late Jurassic -Early Cretaceous extension remains a topic of debate. Some publications explain it by back-arc spreading due to the Paleo-Pacific plate subduction westwards underneath the Asian continent (e.g. [Traynor, Sladen, 1995;Watson et al., 1987]). Most researchers, however, relate the extension to the gravitational collapse of the Middle Jurassic orogen (e.g. [Zorin, 1999;Wang et al., 2012;Meng, 2003]). Q.-R. Meng [2003] proposed a model combining the gravitational collapse of the orogen and active rifting and stated that active rifting took place as a result of the subduction and separation of the Mongol-Okhotsk oceanic plate slab pushed underneath the Siberian continent. The latter in its turn, caused extension and weakening of the lithospheric bottom of the Mongolia-North China block, ascending asthenospheric flows, melting of the mantle and crust, extensive volcanism and magmatism, as well as active formation of the rift basins bordered by gently dipping listric faults. In the Late Cretaceous and Paleogene, the study area was in tectonic quiescence, with gradual post-rifting subsidence in the basins and planation of the existing terrain. The next stage of tectonic activation is related to the collision of the Indian and Eurasian plates and the development of the modern mountain systems and depressions [Cunningham, 2007[Cunningham, , 2010[Cunningham, , 2013.
Our study was focused on the Mesozoic East and South Gobi depressions located in Southeastern Mongolia (see Fig. 1 and 2). As mentioned above, in the Late Jurassic -Early Cretaceous, rifting took place in the East and South Gobi depressions; the basins were formed, and the sediments were accumulated [Graham et al., 2001;Johnson, 2004].The development of the basins was accompanied by extensive volcanic activity that generated the terrigenous sedimentary strata containing a large amount of volcanic material. In total, the thickness of the Late Jurassic -Early Cretaceous sediments in the East Gobi depression varies from 1.0 to 3.0 km [Yanshin, 1975]. The NE strike of the structureforming rift-related faults in this depression suggests the NW extension in the Late Jurassic -Early Cretaceous. The analysis of seismic profiles across the Dzunbai and Uneget basins (East Gobi depression) [Johnson, 2004] shows that the development of the basins was inverted when sedimentation ceased at the turn of the Late Jurassic -Early Cretaceous, and resulted in folding and reverse thrusting of the synrift sediments that were later overlain by non-deformed Upper Cretaceous and Cenozoic sediments. Some of the rift-related faults experienced reverse overthrusting. Folding deformation of the Late Jurassic -Early Cretaceous sediments is also observed in the South Gobi depression [Yanshin, 1975]. At the turn of the Early -Late Cretaceous, compression could have occupied a wider area, as evidenced by the folds in the Late Jurassic-Early Cretaceous sediments in other Cretaceous basins of Mongolia and China, e.g. the Songliao basin in Northeastern China [Graham et al., 2001] and the Baganur basin in the southern part of the Hentei dome [Dill et al., 2004].The NW-N-S compression is evidenced by the trends of the fold axes and activated reverse faults.
The Late Cretaceous sedimentation is associated with post-rift thermal sinking [Johnson, 2004]. Starting from the end of Cretaceous and in the Paleogene, the study area was in tectonic quiescence, the terrain was denudated, and the accumulation of sediments was moderate. The Upper Cretaceous and Cenozoic sediments are positioned sub-horizontally with an angular unconformity on the eroded surface of the Late Jurassic -Early Cretaceous sediments. Their thickness, varying from 150-200 m to 500-700 m, is significantly less than the thickness of the synrift sediments [Yanshin, 1975]. The Cenozoic sediments are less thick and less abundant than the Upper Cretaceous ones. In addition to the Quaternary and recent sediments, the Cenozoic section is mainly represented by the Paleocene, Eocene and Lower Oligocene sediments. A distinctive feature of the Upper Cretaceous sediments is their red colour. The sediments are of the polyfacial genesis and include proluvial, lacustrine, alluvial and lacustrine-alluvial deposits with thin layers of basalts. The Paleogene sediments are of the lacustrine, alluvial and lacustrinealluvial origin and have variegated and red colors.
The distribution of the sediments of the Upper Cretaceous-Paleogene cover reveals the Gobi basin, which occupies the entire southeastern and southern part of Mongolia and extends southward to the territory of China [Yanshin, 1975].The South Gobi depression includes the Bulgan, Argalintin, Baindalai, Sangindalai, Gunkhuduk, and Manlai basins and the Gurvansaikhan, Khurkha, Tsetsei, Dzurumtai, and Yamata uplifts (see Fig. 2, a).The NW and W-E trends are most typical of these structures. In the East Gobi depression (see Fig.  2, b), the largest structures are the Saishand, Unegt, Zuunbayan, Ulegei, and Galbyn-Gobi basins separated by the Mantakh, Saikhandulan, Tsagan Subarga and Totoshan uplifts. These structures trend in the NE and W-E directions.
Most uplifts in Southeastern Mongolia are syndepositional and represented by the sedimentation source areas, and the modern relief is plain-hilly. However, some of these uplifts are either newly formed in the Late Cenozoic (i.e. the recent stage) or have been activated and involved in the recent uplifting. The new/ac-tivated uplifts are characterized by the high-or lowmountain, strongly dissected relief and bordered by faults. In the study area, such structure is the eastern termination the Gobi Altai, including the Gurvan-Saikhan, Buryn Har and Hurhe Ula ridges. The Gobi Altai is described in [Cunningham, 2007[Cunningham, , 2010[Cunningham, , 2013 as the intracontinental transpression mountain building zone, and its eastern termination is viewed as a restraining bend of horsetail splay termination zone of the largest and longest Gobi-Tien Shan fault zone.
The Alpine-type relief Gurvan-Saikhan ridge separates the Bayandalai and Bulgan basins. According to [Yanshin, 1975], these basins developed from the single structure in the Late Cretaceous -Paleogene, as evidenced by the similar sedimentation sections of both basins and the lack of facial changes in the Upper Cretaceous -Paleogene sediments in the vicinity of the ridge. The newly formed structure is the large part of the Khurkh-Ula range (the Khurkhan uplift), as evidenced by the strongly dissected relief, antecedent valleys cutting through the ridge and the Lower and Upper Cretaceous deposits in its southeastern part, raised and dislocated on the sides of the ridge. Recent activity at the edges of the Tsetsei, Dzurumtai and Yamata uplifts is evidenced by clear fault scarps on their sides. In South Gobi depression, the Late Cenozoic activation is poorly manifested, and most of its uplifts are syndepositional. The traces of the Cenozoic activity are revealed only at the Totoshan [Yanshin, 1975] and Tsagan Subarga [Graham et al., 2001;Johnson, 2004;Webb, Johnson, 2006;Yue, Liou, 1999] uplifts. The Tsagan Subarga uplift is dissected in its central part, and the southern side of its western half and the northern side of its eastern half are bordered by the East Gobi fault zone (EGFZ) originating from the Mesozoic, which currently separates the Unegta and Zuunbayan basins.

THE EAST GOBI DEPRESSION
Our field observations were focused on the sides of the small Dalain-Khunda basin located east of EGFZ and the faults bordering the small ridges of the Totoshan uplift ( Fig. 4, a, and Fig. 5). In this area, the main structures are NE-trending. Practically all the studied local fault zones show reverse faulting and thrusting of the crystalline basement over the Upper Cretaceous red beds that fill the basin. One of the NE-trending scarps located north of Erdene Somon (Fig. 5, a, and Fig. 6, a) is the basin's northern border. The main fault planes along the scarp have a north-western and a northern dip at angles from 25° to 45°. The fragmented cataclastic crystalline rocks (crystalline schists and pegmatites) thrust onto the Upper Cretaceous red sediments (argillites) (Fig. 6, b). In the contact zones, the red sediments were thermally hardened as a brick. By reconstructing the stress tensors for the outcrops located in the contact zones, compression regime with the N-S-trending axis of the maximum compression is revealed, as well as strike-slip regime with the WNW-ESE-trending axis of maximum compression (Fig. 7, a, b, c; Table). For the rear side of the thrust, the subhorizontal Lower Cretaceous sediment layers (interbedded brown, yellowishbrown sandstones, conglomerates, and gravelites) and the crystalline basement rocks show the relationships of two types. The sediments overlay the crystalline basement rocks at low angles, and, at the same time, there is a steep tectonic contact (340°80°) that may represent a rift-forming fault on the ground surface. Based on the analysis of tectonic fracturing, the NWtrending extension regime is reconstructed for this zone (Fig. 7, d; see Table).
The scarp bordering the southern side of the basin strikes NE to ENE (see Fig. 4, a, and Fig. 5, b, c, d). Practically everywhere along this scarp, the crystalline basement rocks thrust onto the Upper Cretaceous red sediments (sandstones, gravelites, and conglomerates) (see Fig. 6, c, d, e, and Fig. 8, a, b, c, d, e). The dip direction of the main fault planes vary from 120°-150° to 180° with the change of the scarp's strike from western fault segment to eastern one. Both reverse- (Fig. 8, d, e) and thrust-type displacements (Fig. 8, a, b) are observed along the fault zone, and the dip angles change from steep to very low-angle. Cataclasites, tectonic breccias (kakirites), zones of rock crushing (to tectonic flour), gouge, and thin weathering gouge layers are typical for the contract zones (Fig. 6, e, and Fig. 8, b, e). In some places, the sediment layers are turned up underneath the thrust plane, and the layers dip at angles varying from 1-2° at a distance from the contact to 30-45° directly at the contact (N and NW dip). No evident strike-slip displacements were observed in the fault zones. The stress field reconstructions show mainly compression and transpression regimes with the NW and N-S axes of compression (Fig. 7, f, g, Fig. 9, a, b, d, e, f; see Table). There are a few reconstructions of strikeslip and compression regimes with the NE axes of compression (Fig. 7, e, Fig. 9, c; see Table). Due to the fact that the Upper Cretaceous red sediments are displaced, but there is no deformation of the Cenozoic sediments, the upper age limit for the observed deformation cannot be specified, while the lower limit may be dated as the post-Late Cretaceous. The Cenozoic age of the basin is suggested by the fact that the propagation of the Upper Paleogene sediments was controlled by the faults bordering the basin, and the border scarp are quite straight and weakly eroded. The indicators of the Late Cenozoic activity are traceable east of Erdene Somon wherein the local erosional network is actively    Рис. 7. Реконструкции палеонапряжений в зонах разломов Далайн-Хундийской впадины. Стереограммы -сетка Вульфа, верхняя полусфера. Гистограммы показывают отклонения (p) наблюдаемого скольжения от теоретического для каждой плоскости разрыва. Кружком обозначена ось сжатия σ 1 , треугольником -промежуточная ось σ 2 , квадратом -ось растяжения σ 3 . Цифрами обозначены угол падения и азимут падения соответствующих осей главных нормальных напряжений. Номер стресс-тензора соответствует номеру точки наблюдения на рис. 5.
developing. On the 2.53.0 km site at the southeastern side of the Dalayn-Khunda basin, we observed a new network of small gullies in the center and dry washes (locally named 'sairs') cutting the perimeter of the lowamplitude uplift (Fig. 10). On the same site, along layer detachment was observed in the gently dipping Upper Cretaceous red sediments (Fig. 8, f).The stress field reconstruction shows transpression regime with the N-Strending axis of maximum compression (Fig. 9, h; see Table).

THE SOUTH GOBI DEPRESSION
We studied fault zones bordering the Gurvan-Saikhan and Khurkhe-Ula ridges, some small ranges and forebergs in the surrounding areas, as well as active fault zones traceable in the inner field of the Gunkhuduk basin (Fig. 4, b, c).
The NE and SW sides of the NW-trending Khurkhe uplift are bordered by faults (Fig.4, b, and Fig. 11). Besides the NW-striking faults, we observed the W-E faults, the largest of which goes across the central part of the Khurkhe-Ula ridge (Fig. 11, a). In this fault zone, the left-lateral displacements of temporary water stream valleys amount to 30-35 m (Fig. 12, a). The fault is steeply dipping, 80 to 90°, and its main plane dips to S and S-SW. The stress field reconstructions for the observation points along this fault zone show strike-slip conditions regime with the NE-trending axis of compression (Fig. 13, a, b, c; see Table).
Our studies of the NE slope of the Khurkhe-Ula ridge, that borders the Sangindalai basin (see Fig. 11, a,  b), revealed mainly thrust-type and reverse displacements along the NW-striking faults (Fig. 12, b to g). The faults are marked by cataclastic and crushed rocks and weathering gouge (Fig. 12, c, g). The bordering fault planes dip in the SW direction, underneath the ridge, at the angles from 45° to 85°. In some places, we observed the crystalline basement thrust onto the Mesozoic (Cretaceous) sediments. A thrust of granite onto loose conglomerates was discovered near the Mogoin-Khuduk site (Fig. 12, b, c, d). The S-SW dipping thrust plane is accompanied by a linear weathering gouge zone (to 2 m thick), and its dip angle ranges from 30° to 18° (Fig. 12, c). In the same area, the Mesozoic conglomerates overthrust onto the Quaternary terrace, and the fault plane dips at 225°67°. Along the SW side of the Sangiyndalai basin, there are dry water stream valleys with weakly cemented sandstones, conglomerates and gravelites, which are disturbed by steeply dipping rock crushing zones and along layer disruptions, with obvious reverse displacements at some locations (Fig. 12,  h). The fault planes dip to SW (the dip azimuth of 215-230°) at the angles of 58-75°. As we approach the sides of the basin, we observe the twisting of sedimentary layers (the dip azimuth of 185-220° with the angles from 28 to 65°). For the fault zones at the NE side of the Khurkhe-Ula ridge, the stress field reconstructions show transpression, compression and strike-slip regimes with the dominating NE trend of maximum compression (Fig. 13, d to k; see Table 1). The SW side of the Khurkhe uplift is clearly bordered by the fault (Fig. 11, c). Based on the tectonic fracturing in the Paleozoic rock outcrops along the NW-striking scarp, the stress field reconstruction shows compression and strike-slip regimes with the NE-trending axis of compression (Fig. 14, a, b, e, f; see Table).
The Gunkhuduk basin is located south of the Khurkhe-Ula ridge. On the Bordzongiyn Gobi site located at the northern side of this basin, we observed a series of W-E and NW-trending scarps that deform the gently sloping plain (see Fig. 4, b, Fig. 11, c, d, and Fig. 15, a to  d). Practically everywhere near the scarps, there are the outlets of relatively large water springs. The north wall of the fault is uplifted. The scarp height varies from 0.5 to 1.3 m along the strike. The left-lateral shear amplitude of the dry water stream valleys ranges from to 1.5 to 9.0 m ( Fig. 15, b, c). Further eastward, this series of scarps is conjugated with the faults bordering the SW side of the Khurkhe-Ula ridge. The stress field reconstructions for several Paleozoic rock outcrops along the scarp show compression, transpression and strike-slip conditions regimes with the NE and N-S axes of compression (Fig. 14, c, d, g, h; see Table).
Our studies of active faults bordering the Gurvan-Saikhan uplift were carried out in its eastern termination near the town of Dalan-Dzadgad (see Fig. 4, c, and Fig. 16). At the southern side of the uplift, there is the W-E fault zone north of the Khurmen Somon, wherein we observed a well-defined scarp that is generally trending at 295° (Fig. 16, a). The fault zone along the scarp is marked by tectonites, mylonites, tectonic breccias (kakirites), and higher density of fractures. Its total thickness is not less than 500 m. The fault is rather steeply dipping (70-80°). Along the strike of the scarp, the local water stream valleys are left-laterally shifted by 100-150 m. The foreberg located further to the south and penetrating the northern side of the Baindalai basin, is clearly bordered at the north by the faults, which main planes dip to the south (Fig. 16, a, and  Fig. 17, a). The stress field reconstructions for these fault zones shows strike-slip, transpression and compression regimes with the NE-trending axis of compression (Fig. 18, a to e; see Table).

Результаты реконструкций палеонапряжений
The stress field reconstructions for the Paleozoic rock outcrops along the scarp show strike-slip, compression and transpression regimes with the NE-trending axis of compression (Fig. 18, f, g, h; Fig. 19, a, b, c; see Table). The W-E-trending Tsetsei uplift located east of the Gurvan-Saikhan uplift is syndepositional in relation to the Mesozoic basins in this region [Yanshin, 1975] (see Fig. 4, c). Its southern side was activated in the Cenozoic as evidenced by clear fault scarps bordering some ranges in the south, as well as the forebergs that 'cut' the northern side of the Argalint basin. Near the Khan-Ula mountain, the southern side of the basic rock block is overlain by the Cenozoic (50 Ma) basalts and displaced by the two W-E-striking left-lateral strike-slip faults (Fig. 16, c).The offset is about 60 m along the northern fault branch and 250 m along the southern fault branch. The northern block is raised relative to the southern one, which indicates a reverse component of displacement along the fault. Left-lateral horizontal displacement with reverse component where found along southern boundary of the small W-E-trending ridge-forberg located at a distance of 6.0 km south (Fig. 16, d).The scarp is 3.0 to 4.0 m high. The tempo- Рис. 9. Реконструкции палеонапряжений в зонах разломов Далайн-Хундийской впадины. Номер стресс-тензора соответствует номеру точки наблюдения на рис. 5 и рис. 10. rary water stream valleys are shifted, and the leftlateral strike-slip component amounts are to 6.0-7.0 m (Fig. 17, d). The stress field reconstructions for this fault zone show compression and strike-slip regimes with the NE-trending and N-S axes of compression (Fig.  19, d to h; see Table).

DISCUSSION
Comparing the data on the geological development of the East and South Gobi depressions suggests a similar history of the evolution of these two neighbouring areas in the Late Jurassic -Early Cretaceous (rifting) and Late Cretaceous -Paleogene (tectonic quiescence).
Later on, in the Cenozoic, the depressions were activeted in completely different modes.
The strike-slip kinematics of displacements along the fault zone is evidenced by the contrasting sediment formations across the fault strike [Johnson, 2004] and the stress field reconstructions showing strike-slip regime in the Tsagan Subarga uplift (Tsagan-Suburga, Tavan-Khar, Ulgei-Khid ridges) [Webb, Johnson, 2006]. The axes of compression and extension trend in the NNW and ENE directions, respectively (see the stress tensors marked in grey in Fig. 20). In [Webb, Johnson, 2006], the strike-slip amplitude is estimated at 100-150 km by displacement of Low Cretaceous markers and dated Cenozoic, similar to [Darby et al., 2005]. According to [Darby et al., 2005;Yue, Liou, 1999], it is assumed a kinematic relation between EGFZ and the Altyn Tagh fault through the system of left-lateral strike-slip faults in the Alxa region located north and northeast of Tibet. According to [Yue, Liou, 1999], the EGFZ, the Altyn Tagh and Alxa faults had developed as a single fault system since the Oligocene. The system ceased its activity 13-16 Ma ago as the left-lateral strike-slip displacements along the Altyn Tagh fault were transfered to thrusting, crustal shortening and uplifting in the Qilian Shan region. The Early Cenozoic displacement along EGFZ is estimated by [Yue, Liou, 1999] as about 400 km. According to [Darby et al., 2005], strike-slip faulting developed in the Alxa region in two stages. In the post Cretaceous -pre-Middle Miocene stage, large amplitude left-lateral strike-slip displacements took place along the Altyn Tagh fault and the system of the Alxa left-lateral strike-slip faults (i.e. the NE-trending continuation of Altyn Tagh fault). Further to NE, left-lateral strike-slip displacements where either transferred to EGFZ or transformed into normal faulting in the Periordos rift system. The Upper Cretaceous markers show the offset amplitudes of 70-150 km [Darby et al., 2005]. In the post Miocenerecent stage, there are the left-lateral strike-slip faults of limited amplitudes (less than 3.0 km) in the Alxa region and lower rates of strike-slip displacements along the Altyn Tagh fault as the left-lateral strike-slip in the Altyn Tagh fault was accommodated by thrusting in the Qilian Shan region. According to [Webb, Johnson, 2006], the redistribution of deformation in the NE periphery of the Indo-Asian collision zone correlates with changes in the boundary conditions at another interplate boundary, specifically the replacement of transtension with transpression in the NE segment of the Western Pacific interplate boundary in the Late Miocene [Worrall et al., 1996]. This may also be a cause of EGFZ inactivity since the Late Miocene. Рис. 14. Реконструкции палеонапряжений в зонах активных разломов юго-западного склона поднятия Хурхэ-Ула и северо-западного борта Гунхудукской впадины (урочище Бордзонгийн-Гоби). Номер стресс-тензора соответствует номеру точки наблюдения на рис. 11, (c), (d).  The post-Late Cretaceous thrusts observed in our study of the NE-trending scarps in the northern side of the Totoshan uplift (east of the EGFZ), the stress field reconstructions of compression and transpression regimes ( Fig. 20) indicate that the N-S and NW compression was also active at the beginning of the Cenozoic. According to [Tomurtogoo, 1999], the faults bordering the SW part of the Dalain-Khunda basin control the spread of the Cenozoic (Paleogene) sedimentation. The most probable age of the deformation is thus Paleogene.
Рис. 20. Cхема стресс-тензоров поля напряжений Южно-Гобийской (a) и Восточно-Гобийской (b) депрессий и классификация стресс-тензоров по [Delvaux et al., 1997] (c). located further to NE, which commenced in the Miocene under NW extension and continued until the Pleistocene. The magma flows were controlled by the NEand, more rarely, W-E-striking faults, as evidenced by the chains of volcanic structures and basalt dikes. The activity in this region coincided in time with active extension in the Baikal rift and the formation of the Inchuan-Hetao rift zone in the Periordos rift system, which many authors relate to the Indo-Asian collision. The current tectonic activity in the Dariganga plateau is weak. The NW compression source is still unknown, but possible causes for its occurrence are suggested in the literature. Firstly, local thrusting in the southeastern wing of EGFZ may be related to large-amplitude leftlateral strike-slip displacements along EGFZ in the Oligocene -Miocene. The Indo-Asian collision may be the source of the left-lateral strike-slip movements in this region. In this case, an explanation can be provided for the absence of similar deformation in the South Gobi depression, but the kinematic relation between the NEtrending thrusts and EGFZ still remains unclear. Secondly, there is evidence of the NW compression in North China in the post-Cretaceous period, which is driven by the Pacific subduction zone. According to [Feng et al., 2010], the movements of the Pacific plate were rearranged about 65 Ma ago and caused an inversion in the development of the Songliao basin, uplifting, folding and thrusting. According to [Zhang et al., 2018], the NNE trusting and folding occurred in a more southern region (Hanghua basin) and in an earlier period (Late Cretaceous). The long-distance action of plate interactions and the transfer of deformations into the lithospheric plate interior depend on the type of interaction. The most effective in this sense is the collision mechanism, which is clearly seen in the example of the Indo-Eurasian interaction. According to the new reconstructions [Yang, 2013], starting from 100 million years ago in the East Asia, there was a frontal collision and oblique interaction between the North China block and the Okhotsk block of Izanagi plate. With this interaction, the author associates intense intra-continental compressional deformations within East China, which occurred right up to the end of the Cretaceous. Thirdly, the back-arc spreading in the marginal seas of the Pacific subduction zone commenced 65 Ma ago, and later on, about 15 Ma ago (Middle Miocene), spreading was replaced by compression [Yin, 2010]. This is evidenced by the cessation of the back-arc spreading in the Japan and South China seas, the development of thrust and fold belts in the sediments of the East China and South China seas, and the subduction of the South China basin underneath the Philippine plate. Another evidence is the replacement of transtension with transpression in the Okhotsk Sea and the Sakhalin-Hokkaido fault zone in the Late Miocene [Worrall et al., 1996]. Thus, inter-continental deformation in East Asia might have been influenced by the change from extension to compression-transpression in the West Pacific interplate boundary in the Middle Miocene. In particular, this change could caused the redistribution of deformation in the NE periphery of the Indo-Asian collision zone and the cessation of displacements along the Alxa-East Gobi left-lateral strike-slip fault branch 13-16 Ma ago. The NE-striking thrusts may have occurred due to compression. In this case, however, an evidence of similar deformation needs to be discovered not only in the East Gobi depression. Besides, such compression in the Middle Miocene would have acted against the activity of the NE-trending structures in the Dariganga plateau.
Thus, the East Gobi depression is characterized by active deformation in the first half or in the middle of the Cenozoic due to compression. Three potential sources with slightly different directions of compression are the Indo-Asian collision (N-S compression in the Oligocene -Early Miocene), the collision of the North China block of Eurasia and the Okhotsk Sea block of Izanagi plate (the beginning of the Paleocene, NW compression) and the Pacific subduction zone (Middle Miocene, NW compression). This region was low active in the Late Cenozoic, although we revealed some indicators of activity, such as the networks of erosional gullies on the low-amplitude uplifts. This is explainable by the fact that the tectonic deformation caused by the interplate interactions due to the Indo-Asian collision was mainly concentrated in the Gobi Altai and the Periordoss rift system, i.e. resulted in the development of the large structures. Besides, the seismic records from this region show low seismic activity [Dugarmaa, Shlupp, 2000]. The regional earthquake focal mechanisms show strike-slip and thrust displacements in the earthquake foci with the NE and ENE-trending axes of maximum compression (see Fig. 3) [San'kov et al., 2011;Radziminovich et al., 2016]. Thus, the modern state of crustal stresses and the reconstructed paleostress state are characterized by completely different directions of the axis of maximum compression (see Fig. 3). The main faults in the East Gobi depression have the NE strike parallel to the direction of the axis of maximum horizontal compression in the modern stress field, which does not presuppose their high activity under compression.
The South Gobi depression was also in tectonic quiescence in the Late Cretaceous -Paleogene, but its activity (in contrast to the East Gobi depression) commenced in the Late Cenozoic (Late Miocene -Early Pliocene). From the Late Cenozoic, young uplifts and forebergs, that 'cut' the sediments of the Mesozoic basins, develop in this seismically active region (e.g. [Cunningham, 2007[Cunningham, , 2010[Cunningham, , 2013Yanshin, 1975;Bayasgalan et al., 1999]). In the eastern termination of the Gobi Altai, we revealed clear indicators of the Pliocene-Quaternary activity of the W-E and NW-striking strike-slip and thrust faults. The stress field reconstructions show compression, transpression and strike-slip regimes with the dominant NE-trending axis of compression (see Fig. 20). The regional earthquake focal mechanisms in the eastern termination of the Gobi Altai show strike-slip and reverse displacements with the NEtrending axis of maximum compression (see Fig. 3) [San'kov et al., 2011;Radziminovich et al., 2016]. The modern state of crustal stress correlates with the reconstructed Late Cenozoic paleostress state in both the type and direction of the axis of maximum compression (see Fig. 3). The source of deformation in the Eastern Gobi Altai, as well as in Western and Southwestern Mongolia, is the Indo-Eurasia collision.

CONCLUSION
Having consolidated and compared the field data with the earthquake focal mechanisms in Southeastern Mongolia, we identify two major deformation stages for the study area in the Cenozoic. In the East Gobi depression, the Early Cretaceous NE-trending faults were ac-tivated in the Early Cenozoic (most probably, Paleogene) due to the N-S and NW-trending compression, and the thrusts and left-lateral strike-slips were formed. The most probable source of compression was the West Pacific zone of plate interaction. The impact of the Indo-Eurasia collision, however, cannot be excluded. With time, the activity of the NE-striking faults was considerably reduced.
In the South Gobi depression, the structures in the eastern termination of the Gobi Altai were formed in the Pliocene -Quaternary due to the NE-trending compression caused by to the Indo-Eurasia collision. The W-E and WNW-striking left-lateral strike-slip faults and thrusts occurred in this region. These faults remain active at the present stage of the regional geological development.

ACKNOWLEDGEMENTS
Authors thank to Laura Webb and Damien Delvaux who carefully reviewed the manuscript and made numerous helpful suggestions for its improvement. The reported study was supported by RFBR (research project № 17-05-00826_a).