Strike-slip faults are widely developed on the Earth Crust, particularly in mid-ocean ridges, pull-apart basins and convergent plate margins^[1⇓-3]. Andersonian mechanism has been widely used to interpret the origin of strike-slip fault in shallow crust. However, many factors have significant influences on the formation and development of fault, such as regional and local stress field, pre-existing structures and previous fractures, petrophysics and so on^{[1,4⇓⇓ -7]}. In addition, there are also non- Andersonian processes in the fault development, such as fault tip propagation and interaction, fault wear and dynamic rupture^[8-9]. Strike-slip faults are usually developed in such ways as reactivation of pre-existing structures, linkage of previous small structures, and localized slip of preexisting shear zones^[5]. The weak zones of preexisting and previous structures are favorable to the growth of fault nucleation and linkage, which constrains the development and distribution of later faults and affects the faulting mechanism^[5⇓-7]. The analogue modeling revealed upward growth of strike-slip faults and subsequent En echelon structure to form a complicated "helicoidally drag" pattern^[10], and then through-going by Y-shear fractures linkage^[11]. These suggest segment linkage and interaction are important mechanisms for the strike-slip fault growth. In addition, the heterogeneity and anisot-ropy in rock composition, texture and petrophysical properties also have strong effects on the fault formation and development^[5⇓-7], and have impacts on the mechanism of fault weakening^[12]. The complex mechanisms of strike-slip faults lead to the diversity of their structure style and evolution. As a result, fault zones generally have complex fault architectures, petrophysical properties and strong heterogeneity^{[9,12 -13]}, and subsequent challanges in earthquake prediction, hydrocarbon and minerals exploration, road and bridge engineering, etc. Further, the deformations of the early faults are often overprinted by late faulting activities in sedimentary basins, making it difficult to decipher the formation mechanisms of early faults with multi-stages of faulting activities. Moreover, it is difficult to dating multi-stage fault activities in the pre-Cenozoic fault zones with no effective dating methods and techniques^[14-15], which hinders the study on forming mechanisms of multi-stage strike-slip faults.

Much less oil and gas resources have been discovered along strike-slip fault zones than from the extensional and compressional fault zones in the petroliferous basins^{[16⇓⇓-19]}. In recent years, however, substantial oil/gas reserves have been found around a series of strike-slip fault zones in the central Tarim Basin, attracting much attention to the study of the impacts of the strike-slip faults on oil/gas accumulation^[20⇓-22]. The oil/gas reservoir evaluation and development show that the high production wells in the Ordovician weathering crust and reef-shoal carbonates of northern and central Tarim Basin are mainly distributed along strike-slip fault zones^{[23⇓⇓-26]}. Recent studies have described the geometry of the strike-slip faults from seismic data in different blocks, as well as some fault evolution history^{[27⇓⇓⇓-31]}. These data present a complicated features of the strike-slip faults with multiple fault activities in the Ordovician, Silurian, Devonian, Permian, Mesozoic and Paleogene^[32], and a linkage growth mechanism of the strike-slip faults in the Tarim Basin^[33]. However, the fault formation time and initiation mechanism are enigma in these small displacement (<2 km) but long length (>300 km) strike-slip fault zones, which constrains further study of the geometric and kinematic features of the strike-slip faults and the related oil/gas exploration and development.

Based on the geometric analysis of strike-slip fault, together with chronology and regional structural background analysis, this paper discusses the formation time and special growth mechanism of "small displacement but long length" strike-slip faults in the central Tarim intracratonic basin.

The Tarim Basin, with an area of 56×10⁴ km², is a superimposed basin formed by the Archean-Lower Neoproterozoic crystalline basement, the Paleozoic-Mesozoic Craton basin and Cenozoic forland basin^[34]. The Cryogenian-Quaternary sedimentary strata are well developed and composed of five tectonic layers, including the Pre- Nanhua basement, rift succession of the Nanhua-Sinian, intracratonic marine carbonate platform of the Cambrian- Ordovician and siliciclastics of the Silurian-Cretaceous, and terrestrial siliciclastics of the Cenozoic foreland basin^[35]. The Tarim Basin has undergone more than 10 stages of tectonic movements and developed multi-stage, multi-type and diverse faults^[34-35]. The strike-slip fault system controlled the Ordovician carbonate fracture-cave reservoir distribution and oil/gas enrichment^{[20⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓-33]}. So far, this area has proved oil/gas geological reserves of over 10×10⁸ t in oil equivalent, forming a strike-slip fault controlled super-deep (larger than 6000 m) petroleum system with an area of 9×10⁴ km² in the central Tarim Basin.

Structural interpretation and mapping from seismic data show that strike-slip faults are well developed in the central Tarim Basin^[32], forming a large circum-Manxi strike-slip fault system (Fig. 1). Fault F_I1 divides the strike-slip fault system into two belts from west to east, with NW- striking faults in the west but more NE-striking faults in the east. From south to north, the strike-slip fault system is divided into three regions by the platform margin of the Upper Ordovician Lianglitage Formation in the Central Tarim Uplift and the Middle Ordovician Yijianfang Formation in the Northern Tarim Uplift (Fig. 1). The Central Tarim Uplift has largely NE-trending strike-slip faults, while the southern slope of the Northern Tarim Uplift has two sets of NE-trending and NW-trending strike-slip faults^[31]. Manxi area has fewer strike-slip faults, which connected with the southern and northern faults. More than 60 large strike-slip fault zones have been identified from seismic data^[32], with lengths of 30-80 km in general. Among them, the major fault F_I1 is up to 290 km in length. According to the offset displacements of platform margins, anticlines and paleo-channels by the strike-slip faults, the horizontal displacement along the strike-slip fault zone is generally less than 1.5 km, much less than the displacement of the same length faults in other areas^[36].

On seismic profiles, the strike-slip faults are mainly distributed in the Cambrian-Ordovician carbonates (Fig. 2). They show single vertical fault plane, half-flower, positive-flower and negative-flower patterns (Fig. 2), tending to change from single vertical fault to flower structure, and increase in displacement and deformation degree. There are many transpressional structures formed in the Cambrian-Ordovician carbonates, but mainly inherited transtensional structures upward to the overlying strata. In the Central Tarim Uplift, most strike-slip faults extend upward to the Silurian-Middle Devonian (Fig. 2a), and a few faults enter into the Carboniferous-Permian. In the Manxi area, some strike- slip faults can grow upward to the Permian^[30]. Whereas, some NE-striking faults can extend to the Paleogene in the North Tarim Uplift, (Fig. 2b). Multiple fault styles occur in the Cambrian-Ordovician and control the basic fault frame of the strike-slip fault system. A series of inherited faults, mostly of transtentional faults, expand upward the Silurian-Middle Devonian to form en echelon structures. The inherited fault zones in the Carboniferous-Permian are narrow but with vertical displacement of more than 200 m, mainly of transtensional structures in the northern Tarim areas. In the Mesozoic-Paleogene, the strike- slip faults are mainly distributed along the major NE- striking fault zones in the southern North Tarim Uplift, and show densely small en echelon faults. The strike-slip faults can combine into various structures on the plane and profile, such as linear structure, en echelon/oblique structure, flower structure, horse-tail structure, "X-type" conjugated structure, pull-apart structure and braided structure^{[21,27⇓⇓⇓⇓ -32]}.

Seismic coherent data shows that the short faults length less than 3 km are mostly isolated, disconnected segments (Fig. 3a), hundreds of meters apart from each other. When two fault segments overlap without interaction, they appear in soft linkage and both decrease in displacement to the overlap zone (Fig. 3b). In contrast, in the overlap zone where two fault segments connected and interacted with each other, they are linked directly or by secondary faults to form a hard linked zone (Fig. 3c). There are generally secondary faults, grabens or horsts with strong deformation, and rapid displacement increase in the overlap zones. By the seismic coherent data and seismic profile analysis, the isolated small strike-slip faults in oblique/en echelon or soft linkage patterns show distinct segmentation; while the faults in the hard-linked overlap zone are linked complicatedly with no obvious segment feature.

In the Tarim Basin, many large strike-slip fault zones are through-going along fault strike in the Cambrian-Ordovician (Fig. 1). They compose of multiple segments overlapping with each other by different fault types to show different segmentation along fault strike. For example, fault F_I1 can be divided into five segments horizontally (Fig. 4). The southern linear segment I has smaller displacement, soft-linked oblique segments, and fewer secondary faults. Segment II has hard-linkage faults in the overlap zone, and transpressional and transtensional faults showing oblique/flower structure segments. The transtensional faults generally terminate upward in the Silurian-Devonian, but the transpressional faults mainly terminate in the Ordovician. Segment III in the center are pull-apart grabens with vertical throws of more than 50 m, and reach upward to the Permian even Mesozoic. The northern segment IV has mainly transpressional faults, and segment V show braided structure of intersecting transpressional and transtensional faults, and horsetail structure or linear faults in the fault tip by fault termination.

Comprehensive analysis shows that the small-scale strike-slip fault zones are mostly composed of a series of oblique/en echelon segments with small displacement without secondary branches and interaction in isolation and/or soft-linkage. In contrast, the large strike-slip fault zones are mostly through-going with overlap zones by hard-linkage and interaction, resulting in strong deformation zones. The fault segmentation is mainly shown in the changes of structural style and fault throw along the fault strike.

According to termination horizons of the strike-slip faults from seismic interpretation, it is inferred that the strike-slip fault system in central Tarim formed in the late Ordovician^{[27⇓⇓⇓⇓-32,37]}. It is generally assumed that the strike-slip faults in the North Tarim Uplift formed a little later than those in theCcentral Tarim Uplift^{[27⇓⇓⇓-31]}, given that the Middle-Late Ordovician strata in the northern Tarim are continuous and its paleo-uplift development is later than that in the Central Tarim Uplift, and the strike-slip fault system probably developed from south to north.

The Central Tarim Uplift occurred before the deposition of late Ordovician Lianglitage Formation, where the Middle Ordovician Yijianfang Formation and the Upper Ordovician Tumuxiuke Formation are almost missing, suggesting a distinct regional unconformity^[35]. Besides some thrust faults terminating at the top of the Yingshan Formation from new 3D seismic interpretation data, some strike-slip fault tops were truncated by the overlain Lianglitage Formation in the Central Tarim Uplift (Fig. 5a), with a large unconformity under the Lianglitage Formation. The strike-slip faults in the Cambrian- Yingshan Formations are steep and straight, with smaller vertical fault displacements and transpressional features. In contrast, faults in the overlying strata form negative flower structure, with obvious transtensional features and larger vertical displacements, quite different from the early vertical linear faults. Although the strike-slip faults merge downward in the major fault zone and have inherited evolution, the faults show distinct feature above and below the unconformity under the Upper Ordovician, suggesting a strike-slip fault activity occurred before the deposition of the Lianglitage Formation. In addition, as the strike-slip faults in central Tarim played a role of adjusting the deformation of thrust faults^[35] that formed during the formation period of the Central Tarim Uplift before the deposition of the Lianglitage Formation^[27,35], it is inferred that the strike-slip faults in the Central Tarim Uplift were also formed before the deposition of the Lianglitage Formation.

Although the Ordovician in the northern Tarim had a relatively continuous succession, the shoal facies grainstone of the Yijianfang Formation varied obviously from the marl of the Tumuxiuke Formation (Fig. 1b), indicating obvious hiatus before the Tumuxiuke Formation deposition^[35,38]. Tectonic uplift and large area of karst landform occurred before the deposition of the Lianglitage Formation^[39], which had strong control on the development of high-quality karst reservoirs on the top of the Yijianfang Formation. Statistics on core petrophysical properties (Fig. 6) show that the Middle-Upper Ordovician limestone samples have very low porosity, except for some slightly higher porosity samples from the platform margin of the Lianglitage Formation. In contrast, the permeability from the Yijianfang Formation and Yingshan Formation are one order higher (on average 5.78×10^-3 μm² and 4.48×10^-3 μm² respectively) than the Lianglitage Formation (on average 0.86×10^-3 μm²) and Tumuxiuke Formation (on average 0.56×10^-3 μm²). This is possibly related to fault activity before the deposition of the Upper Ordovician, which resulted in more fractures and abnormally high permeability in the Lower-Middle Ordovician. This means that strike-slip fault activity probably initiated in the North Tarim Uplift before the deposition of the Upper Ordovician too. The seismic profiles show (Figs. 2b and 5b) that some strike-slip faults in the North Tarim Uplift terminate at the top of the Yijiangfang Formation with chaotic reflections. In addition, there is karst landform at the top of the Yijianfang Formation, and the overlying carbonate has thickness change in laterally, suggesting a fault activity occurred and affected the paleogeomorphology and deposition at the end of the Middle Ordovician. Similar to the Central Tarim Uplift, there are mainly transpressional faults in the Cambrian-Ordovician carbonate formations, but transtensional faults in the overlain formations in the northern area. The seismic profile shows that the upper transtensional faults extend downward and offset the lower transpressional structures, forming a micro-graben at the core of the Yijianfang Formation anticline (Fig. 2b), which is distinct from the transpressional anticline structure in the lower strata. In addition, the strike-slip faults above the Ordovician rejuvenated from some localized strike-slip faults, characterized by en echelon structure, graben and linear structure, showing different distribution and combinations from the lower faults. The faults in the upper formations have shorter lengths, but vertical throw up to more than 200 m.

Together with previous study^{[27⇓⇓⇓⇓-32,37]}, it is inferred that the strike-slip faults were formed before the Late Ordovician, following multi-stage fault activities in the Late Ordovician-Devonian, Carboniferous-Permian, and Mesozoic-Paleogene, etc. (Figs. 1, 2, and 5). The strike-slip faults inherited successively from the Late Ordovician to Paleogene from the Middle Ordovician strike-slip faults (Fig. 2), but are different in pattern and distribution from the early strike-slip faults. It is worth noting that the faulting time cannot be determined only by the fault termination layer from a few seismic profiles, as the strike-slip faults were small in initial displacement and failed in subsurface, and were overprinted by late faults.

The fault activity periods can be roughly inferred from the regional tectonic evolution, and the fault crosscutting relationship and termination layers of faults on the seismic profile from the subsurface. However, it is difficult to determine faulting time accurately and even to know the initiation time of faults experiencing multi-stage fault reactivations (Figs. 2 and 5). Thermochronological method has been applied in the dating of fault activation^[14], but they are challenged by low precision in dating pre-Cenozoic faults and hard to apply to subsurface sedimentary strata. Recently, U-Pb dating of carbonate cements has made significant advance^[15,40], providing a high-resolution dating method for faults in carbonate formations.

We collected syn-faulting calcite samples from fractures in the Ordovician carbonate strike-slip fault zone in the Tarim Basin, and carried out in-situ LA-ICP-MS dating^[15] in the Radioisotope Laboratory of the University of Queensland, Australia, to obtain high-resolution U-Pb ages of the fault cements (Fig. 7). The calcite sample from the top weathering crust of the Yingshan Formation in well TZ2 was dated at 460±12 Ma (Fig. 7a; see reference [15]), another calcite sample from the fracture at the top of the Yijianfang Formation of well RP4 was dated at 462.6±6.8 Ma (Fig. 7b). These age data define the time of cement precipitation in the fault zones at the end of the Middle Ordovician. In addition, these data are consistent with the 460-450 Ma calcites from the Central Tarim Uplift^[15].

Since fault activity should be earlier than or contemporaneous with the cement precipitation, the fault initiation should be earlier than or at the end of the Middle Ordovician. Further, the sample from the Yijianfang Formation top can define the faulting time at or after the end of the Middle Ordovician. In addition, there are quite different lithology and sedimentary facies between the Yijianfang Formation and Tumuxiuke Formation before and after this age^[38]. On these context, it is inferred that the strike-slip fault initiation time was at around 460 Ma. Although well TZ2 is also located in the thrust fault zone, the dating data can define the age of the strike-slip faults because the strike-slip fault was transfer fault to accommodate the thrust fault deformation in the syn-tectonic period^[35]. The fault activity at around 460 Ma ago is consistent with the time of the regional uplift after the deposition of Yijianfang Formation^[35], as well as the large-scale subduction of the proto-Tethys Ocean^[41-42]. It is concluded through comprehensive analysis that the strike-slip fault activity in the Tarim Basin initiated at around 460 Ma ago of the end of the Middle Ordovician.

Previous studies revealed that the Proto-Tethys Ocean (Paleo-Kunlun Ocean) at the southern margin of the Tarim Plate subducted under the Middle Kunlun island arc at around 480-460 Ma ago^{[34-35,41 -42]}. The subducted plate breakoff occurred at 450-428 Ma ago, leading to the closure of the Proto-Tethys Ocean^[42], and subsequently formation of the foreland basin in the southwest Tarim^[34]. Hence this period is an important tectonic transition period of the Tarim plate. At this stage during the southern Tianshan Ocean expansion period^[43], the northern Tarim plate was stable with little impact on the structural deformation by the northern Tarim plate activation.

In Middle Ordovician, the Tarim Basin changed from extensional to compressional background, and differentiated in strata, deposition and structure^[35,38]. The Northern, Central and Southwestern carbonate paleo-uplifts in nearly EW direction started to take shape at this time (Fig. 8). Southwestern and Central paleo-uplifts had strong tectonic activities and large areas of uplift and erosion, resulting in absence of the Yijianfang and Tumuxiuke formations^[35]. During the late Middle Ordovician, the deposition of the Yijianfang Formation changed from the east-west subdivision to the north-south zonation. During the depositional period of the Lianglitage Formation, the sedimentary differentiation took shape from south to north of "one basin sandwiched between two platforms” (Northern Tarim-Manxi-Central Tarim)^[35,38]. Subsequently, a large amount of siliciclastic rocks filled in the basin, and the carbonate platforms faded away gradually. In this way, we prefer to a progressively northward subduction rather than a southward subduction of the Proto-Tethys Ocean during the Middle-Late Ordovician (Fig. 8). This is consistent with the tectono-sedimentary evolution of the Tarim Basin^[35] and East Kunlun Orogen^[41] during the Early Paleozoic.

In sum, the dynamic origin of the intracratonic strike- slip fault system was related with the tectonic transition from regional extension to compression resulted by the closure of the Proto-Tethys Ocean during the Middle Ordovician in the Tarim Basin. If so, strike-slip faults may also be identified from the southern Tarim Basin where few strike-slip faults discovered yet by poor quality of seismic data and strong tectonic overprinting.

Although it is difficult to restore the paleostress field, and there is a large range of intersection angle between the strike-slip fault strike and the principal stress direction, the conjugate strike-slip faults formed in the early stage can indicate the principal stress direction^[1].

Symmetrical NNE- and NNW-trending conjugate strike-slip faults of the Ordovician carbonates in the North Tarim Uplift recorded the fault pattern during the Middle-Late Ordovician period^[31], which can be used to speculate the principal stress direction during the fault formation period. The NNW- and NNE-trending strike- slip faults strike at ∠330°-∠360° and ∠16°-∠30° respectively, forming a dihedral angle at about ∠2° (Fig. 8b). This means the principal stress was near S-N direction (inferred from the current position) during the formation of the strike-slip faults, as the dihedral angle of the conjugate strike-slip faults is generally consistent with the direction of the maximum compressive stress^[31]. The principal stress in S-N direction was nearly perpendicular to the E-W striking northern and southwestern paleo-uplifts in the Tarim (Fig. 8a), forming the intracratonic folding uplifts^[35]. In the Middle Ordovician, conjugate strike-slip faults was likely to occur by the remote S-N direction compression in the Harahatang area with weak tectonic uplift and uniform geological architecture. Under the regional stress field, there is preferential development of NNE-trending and NNW-trending strike-slip faults in the east and west of the northern Tarim respectively, which are separated by the central conjugate faults (Fig. 1). Meanwhile, some strike-slip faults from southern margin developed from north to south and terminated in horsetail structure, suggesting that the fault grew from north to south, possibly indicating backward compression from north to south in the North Tarim Uplift.

The NW-trending central Tarim paleouplift and its NW-trending thrust faults intersected obliquely with the S-N principal stress^[35], leading to the NE-trending strike-slip faults initiation with accommodation effect during the folding and thrusting of the paleo-uplift. The 11 main NE-trending faults strike from ∠30° to ∠39° in the northern slope of the Central Tarim Uplift, and intersect with the near S-N principal stress at low angles, which is consistent with Andersonian model. Thus, it can be inferred that the near S-N regional stress field controlled the formation of the intracratonic NE- and NW-trending strike-slip faults.

Considering weak zones of preexisting structures and petrophysical heterogeneity are favorable for fault initiation and development^[5], the basement of Tarim Basin with complex architecture and structures had significant effects on the structural pattern of Phanerozoic cover^[35].

The basement of southern Tarim Basin has a series of NE-trending high magnetic anomaly zones^[34,44]. They may formed at 1.9 Ga magmatism by the assemblage of the northern and southern Tarim during the Columbia supercontinent convergence period^[44]. The NE-trending zones constituted weak zones in the southern Tarim basement, consistent with the NE-trending strike-slip fault zones in the Central Tarim Uplift. Although there are relatively uniform metamorphic basement in the northern Tarim, the basement architecture and faults study by gravity, magnetic and electrical data revealed that the northern Tarim had NE- and NW-trending preexisting structures in the basement^[35]. These basement structures were favorable for the strike-slip fault nucleation and rejuvenation from the preexisting basement structures^[4⇓⇓-7]. During the strike-slip fault development from the basement to the cover strata, the S-N principal stress exerted mainly on the early preexisting structures of NE and NW strikes in the Tarim basement, constraining the in-situ weakening zones for the initiation and development of the strike-slip faults. Meanwhile, the weak basement zones may influence the petrophysical properties of the Cambrian-Ordovician carbonate rocks to facilitate the faults to grow upward. On these context, it is easier for the strike-slip faults initiation in small intersection angle with the principal stress direction by the oblique compression to the NWW-trending thrust fault zones and the NE-trending basement preexisting structures in central Tarim area in the Middle-Late Ordovician^[35]. Consequently, a series of NE-trending strike-slip fault zones formed in the Central Tarim Uplift to accommodate the deformation of the NW-trending compressional structures (Figs. 1 and 8a), which are different from the conjugate strike-slip fault zone in the northern Tarim area with uniform metamorphic basement or/and NE- and NW-trending basement faults.

It should be noted that the strike-slip fault system is roughly divided into three zones from south to north by the platform margins of the Lianglitage Formation on the Central Tarim Uplift and the Yijianfang Formation on the Northern Tarim Uplift (Fig. 1). The distribution of the rimed platform margin of the Lianglitage Formation and the unconformity of the Yingshan Formation is constrained along the northern boundary of the central paleo-uplift^[35], showing a tectonic and lithofacies boundary of the northern Central Uplift. The strike-slip faults in the Central Tarim uplift mostly terminate in horsetail structures at the platform margin (Fig. 1), and only a few large strike-slip faults extend to Manxi area. The conjugate strike-slip faults in the southern slope of the North Tarim Uplift are distributed on the wide and gentle platform of the Yijianfang Formation, and pinch-out southward in the lithofacies transition area along the platform margin, indicating lithofacies difference has a certain impact on the growth of the strike-slip fault development and distribution. Therefore, the pre-existing lithofacies may also affect the development and distribution of strike-slip faults.

In summary, the near S-N direction remote compression from the closure of the Proto-Tethys Ocean in the Middle Ordovician controlled the formation of the strike- slip fault system in the intracraton, and the pre-existing structure in basement and lithofacies impact the faulting location and fault distribution from south to north parts of the central Tarim Basin.

The mechanism of conjugate fault system is generally interpreted by Andersonian model, in which symmetrical faults are formed at intersection angles of 25°-30° with the principal stress direction^[8-9]. The basement of Harahatang area had uniform architecture, gentle topography, few pre-existing and precursor structures, and relatively even petrophysical properties^[31], which were favorable to the nucleation and development of conjugate fractures in Andersonian fracturing under remote compression. In this way, the conjugate fractures would gradually expand to form symmetrical conjugate shear fault zones (Fig. 9a). However, the dihedral angles of the conjugate strike-slip faults are in range from 26° to 51°, about 40° on average^[31], which are much lower than the ideal dihedral angle of 50°-60°. It is found that the large confining pressure differences of the surrounding rocks from more than 3000 m thick sedimentary formations can reduce the shear failure angle^[45]. In addition, variations in stress state and rock mechanics also affect the symmetry and dihedral angle of conjugate fractures under weak long-term remote compression^[46]. The pressure solution, mutual truncation of multi-phase fractures^[47], and upward movement to form flower-structure can adjust and maintain the volume balance in the intersection area of the conjugate faults. In addition, the faults could die out by gradually decreasing displacement or volume through downward movement^[48]. Therefore, non-Anderson mechanism might contribute to the growth and development of the conjugate strike-slip faults in the northern Tarim area.

Since the conjugate faults restrict mutual horizontal movement, sequential slip rather than simultaneous movement may be an important mechanism for the development of conjugate faults^[31,46]. Following the formation of "X-type" conjugate fractures in the Harahatang area, the horizontal movement at the fault intersection was limited (Fig. 9b), but sequential slip could cause mutual offset of the conjugate faults^[31](Fig. 9b-9d), leading to continuous fault deformation. The sequential slip generally occurred during a relatively short time of a contemporaneous fault activity, giving rise to rhomboid micro-faults in the intersection to accommodate structural deformation^[49] (Fig. 9c, 9d). However, there is no distinct rhombic deformation in the intersection of the conjugate strike-slip faults in Halahatang area. This may be attributed to small mutual offset displacements after shortly sequential slip, and subsequent preferential development of the NW-trending faults. The more offset from NW-trending faults against the NE-trending fault (Fig. 9d), the more displacement in NW-trending faults to show asymmetric displacement distribution^[33]. Although the NE-trending faults had more reactivations in late stage, there were stronger fault deformation and maturity of the NW-trending strike-slip faults in the Cambrian-Ordovician^[33]. With the late reactivation of the NE-trending faults, they offset the NW-trending faults in some places^[31], but at small displacements of less than 200 m horizontally. Therefore, a rare intact intraplate conjugate strike- slip fault system has been preserved with more than 70 km in length in the Tarim Basin.

Non-Andersonian mechanism plays an important role in the formation and development of faults^[8-9]. The results of detailed seismic interpretation show that some strike-slip faults in the central Tarim have horsetail structure at the tips (Fig. 1). These faults generally spread out with gradually decreasing displacement and irregular fault surface. The formation of this kind of fault is mostly based on the nonlinear and post-yield fracture mechanics^[8-9]. Within a short time before fault tip expansion, an outward off-fault damage zone come up at the fault tip, forming fault tip propagation model (Fig. 9e). As the fault expands outward, the off-fault damage zone at the tip may form secondary faults. These fractures have a wide range of intersection angles with the main fault in Halahatang area, which is different from Riedel shear fracture and wing crack. After the fault tip propagates forward or links with adjacent fault (Fig. 9d, 9e), it is often inhibited the fracture growth with outward off-fault damage at the fault tip. Multiple horsetail structures are northward propagation in the Central Tarim Uplift, but southward development in the North Tarim Uplift^[44], suggesting different fault growth directions.

As the fault tips expand and approach each other, the faults would be linked along the interaction area^[9]. The major strike-slip fault zones are generally composed of 3-8 segments, which form strike-slip fault zones more than 50 km long through linkage growth^[33]. The linkage of the strike-slip fault segments multiple the fault in length (Fig. 3), but increase little in displacement. As a result, the displacement-length relationship of the faults does not consistent to the power-law. In this model, most secondary fault strikes are close to the direction of the maximum principal stress, but have deformation and displacement generally concentrated in the fault interaction zones^[33]. Different from the typicallinkage growth model, most deformation and strain concentrate in the overlap zones with hard-linkage (Fig. 9f), accommodate the fault zone deformation and escape volume imbalance caused by horizontal slip from cross-cutting. In addition, some strike-slip faults show multiple horsetail structures and multiple secondary faults, which may be resulted by the growth of multi-stage fault tip linkage. This type linkage growth at fault tip also consistent with the fault tip interaction model^[9]. The deformation mainly occurs at the fault tip linkage zone to form strong deformed overlap zone through the fault tip interaction. Statistical analysis shows that the strike-slip fault zones generallyhave small horizontal displacement less than 400 m. The displacements and deformations are mainly concentrated in the overlap zones and increase progressively in vertical throw, resulting in deformation localization in overlap zones by fault interactions to accommodate the deformation of the fault zones. This is different from the typical fault weakening mechanism^[12]. On this basis, the strike-slip faults can increase in length and maintain small horizontal displacements to form "small displacement but long fault zone”, which is inconsistent with the displacement-length power-law of faults in other areas^[36]. It should be noted that non-Andersonian faulting is also affected by preexisting Andersonian faulting, and they can occur simultaneously, resulting in complex distribution of faults.

Through comprehensive analysis, the conjugate strike-slip faults in the North Tarim Uplift are formed by sequential slip and cross-cutting at the intersections of faults on the basis of the Andersonian mechanism. In constrast, the strike-slip faults are mainly resulted from non-Andersonian growth mechanisms, particularly of fault linkage growth, and propagation and interaction at fault tip. Further, the faults have distinct fault localization of displacements and deformations in overlap zones. As a result, long but small displacement fault zone progressively formed through linkage.

The fault interpretation suggests that the strike-slip faults in the Central Tarim Uplift also have distinct segmentation^[29,35], and length expansion and multiplication caused by linkage growth. But the propagation and growth of the strike-slip faults in this area are inconsistent with the conjugated faults in the North Tarim Uplift. The strike-slip faults in this area have larger displacement, higher through-going degree, and gradual transition from the linkage zones to bilateral fault walls. Thus, there are less localized displacement and strain in the overlap zones as those in the Harahatang area. The strike-slip faults in this area are larger in scale than in the northern area, showing larger displacement and deformation in the overlap zones and oblique linking zones.

Fault element measurement along the strike-slip fault zones in the Central Tarim Uplift shows the highest displacement and strongest deformation of the strike-slip faults are concentrated at the area near the thrust faults in the northern slope and the NW-trending transtensional graben. The fault displacement decreases quickly to northward segment of the graben. It is found through displacement analysis that the western wall of a strike-slip fault moved faster toward south, resulting in sinistral slip during the southward thrusting (Fig. 10a). The western block of the fault moved southward passively, giving rise to fractures at the fault tip and a narrow and deep graben with the increase of displacement (Fig. 10b). The displacement and deformation drop sharply to the northern segments of the grabens. Fault F_Ⅰ3 has a typical horsetail structure when propagating to the northern margin of the Central Tarim Uplift, forming fault tip of another segment of the strike-slip fault zone (Fig. 1).

Through comprehensive analysis, NE-trending strike-slip faults occurred to accommodating the thrust deformation under the effects of oblique compression and basement preexisting structures during the southward oblique movement of the thrust fault zone in the Central Tarim Uplift. And then these strike-slip faults grew progressively through tip propagation and linkage. With the increased displacements of inconsistent southward thrusting, the larger displacement of the western wall to the south led to wing crack at the fault tip, forming NW-trending graben progressively (Fig. 10b). Different from the stress and strain concentrated in the overlap zones in Halahatang area, the wing cracks had more strain and stress released due to the tip expansion, resulting in growth of the fault tip grabens in the Central Tarim Uplift. The grabens in the Ordovician carbonates are more than 400 m in depth. With the fault going-through and further propagation, sinistral strike-slip faults developed northward of the fault grabens (Fig. 10c). Similar to the conventional intraplate strike-slip faults, the northern fault segments formed horsetail structure on the uplift margin by fault tip propagation.

There are multiple kinds and stages strike-slip faults in the central Tarim Basin. Although the fault timing and mechanisms are big challenges in the deep subsurface, we can propose the following main conclusive points.

(1) The circum-Manxi strike-slip fault system in the central Tarim Basin is characterized by multi-stages inherited evolution. The formation time of the strike-slip faults is defined at the end of the Middle Ordovician (about 460 Ma ago) based on the U-Pb dating of syn-faulting calcite cements and seismic analysis.

(2) The formation of the strike-slip fault is controlled by the near S-N compressional stress caused by the closure of the Proto-Tethys Ocean in the Middle Ordovician, and the fault distributions are significantly influenced by the pre-existing basement structures and lithofacies.

(3) On the basis of the early Andersonian model of fracturing, the strike-slip faults grew fast by non-Andersonian mechanisms, mainly of linkage growth, and fault tips propagation and interaction, resulting in long fault zone with small displacement increase through fault segments linkage.

(4) Sequential slip and preferential growth of the NW-trending faults accommodated the deformation in the fault interaction zones, and the strongly localizaction in the overlap zones accommodated the main displacement and deformation of the conjugate strike-slip faults in the North Tarim Uplift. In contrast, the narrow-deep grabens at the fault tips, and strike-slip segments cross the thrust zones accumulated more deformation and strain of the strike-slip faults in the Central Tarim Uplift.

The authors thank the Tarim Oil Company for data support and project research. We are grateful to Wan Xiaoguo, Zheng Duoming and Liu Xin for their help in U-Pb age dating.