Tectonic Subsidence and Its Response to Geological Evolution in the Xisha Area, South China Sea

: The evolution and mechanisms of tectonic subsidence in the Xisha area are poorly investigated, especially the spatiotemporal distribution features and reasons for the variations in tectonic subsidence. In this study, multi-channel seismic data and stratigraphic and lithologic features of wells are used to examine tectonic subsidence in the Xisha area from the Paleogene to Quaternary. The largest tectonic subsidence in the Xisha area is located in the Changchang Depression, with a maximum subsidence of 5.4 km, while the smallest tectonic subsidence is located on the Guangle Uplift and Xisha Uplift, which are close to 1.0 km and 1.5 km, respectively. Two rapid tectonic subsidence phases were mainly in the Oligocene, and from Middle to Late Miocene, with maximum subsidence rates of 0.45 m/ky and 0.32 m/ky, respectively. Five phases for the tectonic subsidence are proposed since the Paleogene based on our data. (1) The slow subsidence phase during the Eocene (53.5–32 Ma) was due to the transchronicity of the basement in the pro-rifted stage. (2) The rapid subsidence phase was common in the south and north margins of Qiongdongnan Basin, because of the faults triggered by the inherited stretched and thinned of crust in the Oligocene from 32 to 23.3 Ma. (3) The interim phase followed the rapid subsidence phase was in the Early Miocene (23.3–15.5 Ma) and marked the end of the rifted stage. (4) The accelerated rise phase started from the Middle Miocene (15.5 Ma) to the Late Miocene (5.5 Ma), and the reversal of the Red River Fault Zone may be tied to the acceleration of the tectonic subsidence. (5) The transitional phase started in the Pliocene (5.5 Ma) and lasts to the present. As the Red River Fault Zone changed from sinistral to dextral movement, the stress ﬁeld of the study area has changed. Our results are helpful to better understand the spatiotemporal coupling relationship between tectonic subsidence and regional geological evolution in the Xisha area


Introduction
The Xisha area is an extraordinary continental slope located on the northwestern margin of the South China Sea, as shown in Figure 1. Structurally, it is located at the elongated offshore end of the Red River Fault Zone in the Yinggehai Basin and close to the influence area of the Hainan mantle plume [1][2][3][4][5]. Multiple perspectives, including characteristics of faults, crust structure, and tectonic simulation, have been applied to study the tectonic evolution of the Xisha area [6][7][8]. The tectonic subsidence was also discussed in previous studies, which suggest at least three periods of rapid subsidence and one period of weak subsidence have been experienced in the Cenozoic [9,10]. The first and second rapid subsidence occurred during the rift period stage and the most recent With the increasing demands of petroleum exploration in the Xisha area, the acquisition of high-resolution seismic data and well-logging data provides a good opportunity for the analysis of tectonic subsidence history. There are two goals to study the history of tectonic subsidence in the Xisha area. One is to clarify the spatiotemporal distribution of tectonic subsidence of each tectonic unit and subdivide the subsidence phase. The second is to discuss the relationship between subsidence mechanisms, especially the hypothesis of reversal of Red River Shear Zone during the Middle Miocene, and regional geological factors, including tectonics, volcanism, and sedimentation.

Red River Fault Zone
The Red River Fault Zone consists of the onshore segment and offshore segment caused by the extrusion and rotation of the Indochina Block as a result of the India-Asia collision during the Oligocene. The onshore segment is exposed in four 10-20 km-wide With the increasing demands of petroleum exploration in the Xisha area, the acquisition of high-resolution seismic data and well-logging data provides a good opportunity for the analysis of tectonic subsidence history. There are two goals to study the history of tectonic subsidence in the Xisha area. One is to clarify the spatiotemporal distribution of tectonic subsidence of each tectonic unit and subdivide the subsidence phase. The second is to discuss the relationship between subsidence mechanisms, especially the hypothesis of reversal of Red River Shear Zone during the Middle Miocene, and regional geological factors, including tectonics, volcanism, and sedimentation.

Red River Fault Zone
The Red River Fault Zone consists of the onshore segment and offshore segment caused by the extrusion and rotation of the Indochina Block as a result of the India-Asia collision during the Oligocene. The onshore segment is exposed in four 10-20 km-wide ranges, with numerous metamorphic massifs along the Red River Fault Zone [11]. The offshore strike slip fault is divided into four branches and extends continuously seaward across the Yinggehai Basin along eastern Vietnam, as shown in Figure 1 [4]. These faults converge near the southern end of the Yinggehai Basin, and whether they extend southeast to the Xisha area is unclear. Some investigations indicated that they are connected with the East Vietnam Boundary Fault on the west margin of the Guangle Uplift [4]. The offshore distribution of this strike slip fault approximately forms the shape of a spindle. The evolution of the Red River Fault Zone experienced three stages: sinistral movement from~36 to 16 Ma with a displacement of over 600 km, a reversal stage between 16 and 5.5 Ma when the slip velocity decreased and eventually ceased, and dextral movement with low rates after 5.5 Ma [11][12][13][14][15][16][17]. In addition, the first stage is synchronous with seafloor spreading of the South China Sea.

Seafloor Spreading of the South China Sea
In recent years, a consensus on the extensional process of the South China Sea has been reached after abundant geological surveys, which indicates that the first extension started in its northeastern area approximately 32 Ma, subsequently propagated southwestward, and terminated almost simultaneously circa 15 Ma in the East Subbasin and Southwest Subbasin [18,19]. Then, a southward ridge jump of approximately 20 km occurred around 23. 6 Ma in the East Subbasin and this time is coeval with the onset of seafloor spreading in the Southwest Subbasin. Simultaneously, the seafloor spreading in the Northwest Subbasin was deserted, which is the main interpretation of the dead rifted basin of Qiongdongnan Basin [6].

The Xisha Area
The Xisha area is a special continental slope in the northern margin of the South China Sea that developed with Qiongdongnan Basin, Zhongjiannan Basin, Zhongsha Trough, Guangle Uplift, and Xisha Uplift tectonic units, etc., as shown in Figure 2. A combination of well and seismic data studies indicate eight sequential boundaries from top to bottom in the area, as shown in Figure 3. The tectonic evolution in this area can be divided into three stages-the pre-rifted stage, the rifted stage, and the post-rifted stage-by the T 8 (32 Ma) and T 6 (23.3 Ma) unconformity boundaries in Figure 3, respectively [20,21]. Eight formations have been identified in the Xisha area in previous work. From bottom to top, they are the Lingtou Formation in the pre-rifted stage, the Yacheng Formation and Lingshui Formation in the rifted stage, and the Sanya Formation, Meishan Formation, Huangliu Formation, Yinggehai Formation, and Ledong Formation in the post-rifted stage [4].

Volcanism
The north margin of the South China Sea belongs to an atypical magma-poor rifted margin attributed to a certain number of volcanoes along the continent-ocean transitional zone [23]. Those volcanoes activated during the Cenozoic and pervasively developed on the northwest margin of the South China Sea [24,25]. The distribution of volcanoes in the study area is more extensive than other regions [5]. According to high-resolution multichannel seismic data and numerous drilling well data, magmatic activity can be divided into three periods: the active stage from 32 to 16 Ma, the quiet stage from 16 to 5.5 Ma, and the extremely active stage from 5.5 Ma to the present [26]. In the active phase, the magma erupted and intruded along the fracture surface caused by the extension and thinning of crust during the Paleogene. The magmatism in the extremely active period is speculated to be related to the activity of the Hainan plume. In addition, the activity of magma around the Xisha area is more intensive than that in the north margin of the Qiongdongnan Basin and Yinggehai Basin.

Seismic Data
Multi-channel seismic reflection data were collected by Guangzhou Marine Geological Survey from 2002 to 2011 across more than 20,000 km, with the dominant frequency 40-60 Hz. Seismic lines are located in the Xisha area and Qiongdongnan Basin, respectively, as shown in Figure 2. The seismic acquisition parameters include shot spacing of

Volcanism
The north margin of the South China Sea belongs to an atypical magma-poor rifted margin attributed to a certain number of volcanoes along the continent-ocean transitional zone [23]. Those volcanoes activated during the Cenozoic and pervasively developed on the northwest margin of the South China Sea [24,25]. The distribution of volcanoes in the study area is more extensive than other regions [5]. According to high-resolution multi-channel seismic data and numerous drilling well data, magmatic activity can be divided into three periods: the active stage from 32 to 16 Ma, the quiet stage from 16 to 5.5 Ma, and the extremely active stage from 5.5 Ma to the present [26]. In the active phase, the magma erupted and intruded along the fracture surface caused by the extension and thinning of crust during the Paleogene. The magmatism in the extremely active period is speculated to be related to the activity of the Hainan plume. In addition, the activity of magma around the Xisha area is more intensive than that in the north margin of the Qiongdongnan Basin and Yinggehai Basin.

Seismic Data
Multi-channel seismic reflection data were collected by Guangzhou Marine Geological Survey from 2002 to 2011 across more than 20,000 km, with the dominant frequency 40-60 Hz. Seismic lines are located in the Xisha area and Qiongdongnan Basin, respectively, as shown in Figure 2. The seismic acquisition parameters include shot spacing of 25 m, track spacing of 12.5 m, and a 2 ms sampling rate, while the maximum record length is 9 s. Spacing of the 2D seismic profiles was 32 × 32 km in the Xisha area and 4 × 8 km around the Qiongdongnan Basin. According to the layer velocity analysis of the strata, the vertical resolution of the seismic data reached 30-50 m. Seismic interpretation was conducted by Geoframe software from Schlumberger. Ages of seismic horizons were determined by the stratigraphic chronology from Well YC35-1-2 and constrained by previous results [20,21,27,28]. In addition, the reconstruction of stratigraphic development history is based on the balanced cross-section restoration using the software 2D move.

Seismic Sequences
Eight regional stratigraphic horizons-T g (53. , and T 2 (1.9 Ma)-have been identified based on the seismic profiles and wells data in the Xisha area, as shown in Figures 4 and 5 [27,29]. In this study, we focus on the T 31 seismic horizon because of its importance in describing regional geological events occurring in the post-rift phase [28]. The age of T g has been suggested as 53.5 Ma by deep-water wells, including well YC35-1-2 and well LS33-1-1, in the south margin of Qiongdongnan Basin [18,28]. Appl. Sci. 2023, 13, x FOR PEER REVIEW 6 of 18 25 m, track spacing of 12.5 m, and a 2 ms sampling rate, while the maximum record length is 9 s. Spacing of the 2D seismic profiles was 32 × 32 km in the Xisha area and 4 × 8 km around the Qiongdongnan Basin. According to the layer velocity analysis of the strata, the vertical resolution of the seismic data reached 30-50 m. Seismic interpretation was conducted by Geoframe software from Schlumberger. Ages of seismic horizons were determined by the stratigraphic chronology from Well YC35-1-2 and constrained by previous results [20,21,27,28]. In addition, the reconstruction of stratigraphic development history is based on the balanced cross-section restoration using the software 2D move.

Seismic Sequences
Eight regional stratigraphic horizons-Tg (53.  [27,29]. In this study, we focus on the T31 seismic horizon because of its importance in describing regional geological events occurring in the post-rift phase [28]. The age of Tg has been suggested as 53.5 Ma by deep-water wells, including well YC35-1-2 and well LS33-1-1, in the south margin of Qiongdongnan Basin [18,28].

Restoration Principle of Tectonic Subsidence
Basement subsidence consists of loading subsidence and tectonic subsidence. In order to acquire the tectonic subsidence, it is necessary to remove the loading subsidence from the total subsidence. Presently, the back-stripping technique is widely used to calculate the amounts and rates of tectonic subsidence of the basement. Based on the backstripping technique of Sclater and Christie [30], we used Thermodel for Windows 2004, 25 m, track spacing of 12.5 m, and a 2 ms sampling rate, while the maximum record length is 9 s. Spacing of the 2D seismic profiles was 32 × 32 km in the Xisha area and 4 × 8 km around the Qiongdongnan Basin. According to the layer velocity analysis of the strata, the vertical resolution of the seismic data reached 30-50 m. Seismic interpretation was conducted by Geoframe software from Schlumberger. Ages of seismic horizons were determined by the stratigraphic chronology from Well YC35-1-2 and constrained by previous results [20,21,27,28]. In addition, the reconstruction of stratigraphic development history is based on the balanced cross-section restoration using the software 2D move.

Seismic Sequences
Eight regional stratigraphic horizons-Tg (53.  [27,29]. In this study, we focus on the T31 seismic horizon because of its importance in describing regional geological events occurring in the post-rift phase [28]. The age of Tg has been suggested as 53.5 Ma by deep-water wells, including well YC35-1-2 and well LS33-1-1, in the south margin of Qiongdongnan Basin [18,28].

Restoration Principle of Tectonic Subsidence
Basement subsidence consists of loading subsidence and tectonic subsidence. In order to acquire the tectonic subsidence, it is necessary to remove the loading subsidence from the total subsidence. Presently, the back-stripping technique is widely used to calculate the amounts and rates of tectonic subsidence of the basement. Based on the backstripping technique of Sclater and Christie [30], we used Thermodel for Windows 2004,

Restoration Principle of Tectonic Subsidence
Basement subsidence consists of loading subsidence and tectonic subsidence. In order to acquire the tectonic subsidence, it is necessary to remove the loading subsidence from the total subsidence. Presently, the back-stripping technique is widely used to calculate the amounts and rates of tectonic subsidence of the basement. Based on the back-stripping technique of Sclater and Christie [30], we used Thermodel for Windows 2004, which has been widely used in the north margin of the South China Sea with certain success [9,10,31]. Three parameters-decompaction of stratum, paleo-water depths, and fluctuation in sea level-should be corrected in this system.
The lithology of the stratigraphic units is an important influential factor in porosity restoration for decompaction. The systematic analysis of 33 wells in the Xisha area and its adjacent areas, including the newest wells-YL19-1-1, CD26-1-1, XK-1, and 120-CX-1X [32,33]-was used to calculate the lithologic thickness percentages of different stratigraphic units and their generalized formation. Mudstone, shale, sandstone, limestone, and coral are the main ingredients of the Cenozoic strata in the study area. Moreover, their porosity-depth relationships have been analyzed in the literature [30,34,35]. This study adapted the modified compaction parameters from Sclater and Christie [30] and ODP Site 1148, as shown in Table 1. Although numerous volcanoes are distributed in this area, there is no distinct relationship between porosity and depth obtained. Furthermore, the synthesized pseudo-wells by extracted common depth points (CDPs) in the center of basins used to calculate the tectonic subsidence probably represented regional subsidence. The largest uncertainty originates from estimating paleo-water depths. In former research, sedimentary environment analysis from ODP site 1148 provided some useful information to estimate the paleo-water depths in the present abyssal area of the Qiongdongnan Basin since the Oligocene [21,29,36]. The increment in wells in the abyssal area surrounding the Xisha area, such as LS33-1-1, YC35-1-2, and CC26-1-1, is helpful to estimate paleo-water depths. The sedimentary environments were lacustrine and the water depths less than 100 m during the Eocene. From the Early to Late Oligocene, it gradually changed from the initial marine-continental transitional face into the neritic face, which indicated that the water depths rose from approximately 100 m to over 200 m. From the Early Miocene to the present, the water depths in each center of those basins speculated by successive changes in benthic combination was over 200 m in the earliest Miocene, and ascended to 500 m at the beginning of the Middle Miocene, reaching 1000-1500 m subsequently during the Later Miocene. Greatly benefiting from the retrogression of carbonate platform and the change in sedimentary environment [21,29,36], the estimation of the paleo-water depths on the slope was less than 100 m in the Early Miocene, ranged from 200-500 m during the Middle Miocene, and exceeded 500 m and rose to about 1000 m at the end of the Late Miocene.
Sea-level change is also an influential factor in tectonic subsidence calculation. In this study, the sea-level curve is extracted from Miller et al. [22]. It is clear that the restoration of tectonic subsidence should be tied on strata. However, the basement age of each tectonic unit is not always parallel with the age of its initial sedimentation. Therefore, the restoration of tectonic subsidence represents subsidence process since receiving deposits. In addition, the progradational slope prisms occurring on the north margin of the Qiongdongnan Basin since 10.5 Ma were significant in restoring tectonic subsidence [9], while progradation had little influence on our study area, proved by 3D data in Huaguang Depression [28].

Results
Tectonic subsidence curves and rates in the Xisha area have been calculated based on the multi-channel seismic and wells data, shown in  The subsidence of Group 1 in the structural highs is much smaller than that of Group 3 in the center of the basins. The minimum tectonic subsidence of Group 1 is close to 0.8 km and 1.5 km on the Guangle Uplift and Xisha Uplift, respectively. The tectonic subsidence of Group 2 is between 3.2 km and 3.8 km on the Zhongsha Trough and Zhongjian Depression, respectively. The maximum tectonic subsidence of Group 3 in the Changchang Depression is up to 5.4 km, which is approximate to the Qiongdongnan Basin, as shown in Figure 6.
Five phases of tectonic subsidence history in the Xisha area can be identified according to the thickness and rates of tectonic subsidence, as shown in Figures 6-8.  Figure 7.
Third phase-the interim phase in the Early Miocene (23.3-15.5 Ma). Relatively slow subsidence followed the first rapid subsidence that occurred in the Early Miocene. The rate of subsidence was about 0.05 m/ky or even lower in most of the tectonic units in Figure 7. During the interim stage (23.3-15.5 Ma), the tectonic subsidence was about 0.6 km in basins and 0.4 km on the uplifts, and the average rate of tectonic subsidence, much smaller than that in the rapid subsidence phase, was about 0.05 m/ky, even in basins, as shown in Figure 6.  The subsidence of Group 1 in the structural highs is much smaller than that of Group 3 in the center of the basins. The minimum tectonic subsidence of Group 1 is close to 0.8 km and 1.5 km on the Guangle Uplift and Xisha Uplift, respectively. The tectonic subsidence of Group 2 is between 3.2 km and 3.8 km on the Zhongsha Trough and Zhongjian Depression, respectively. The maximum tectonic subsidence of Group 3 in the Changchang Depression is up to 5.4 km, which is approximate to the Qiongdongnan Basin, as shown in Figure 6.
Five phases of tectonic subsidence history in the Xisha area can be identified according to the thickness and rates of tectonic subsidence, as shown in Figures 6-8.

Discussion
Generally, the tectonic subsidence of passive continental margin rift basins is due to the stretching and thinning of the lithosphere during the rifted stage and increased density of mantle material due to heat loss in the post-rifted stage [31,37,38]. There are a few controversies about the rapid subsidence occurring in the rifted stage. In many regions, usually there is a discrepancy between the theoretic value and actual computed value during the post-rifted stage [9,31]. To explain this anomaly, a number of mechanisms have been proposed [9,31,39,40]. Among these mechanisms for the northern South China Sea area, dynamic topography is a relevant mechanism that can interpret the discrepancy in tectonic subsidence rate between uplifts and basins during the post-rifted stage. However, to some extent, the dynamic topography is not very effective in explaining the cause of First phase-the slow subsidence phase in the Eocene (53.5-32 Ma). The tectonic subsidence was 1.0-1.5 km in the slow subsidence stage, with subsidence rate of 0.05-0.07 m/ky in Figures 6 and 7.
Second phase-the rapid subsidence phase in the Oligocene (32-23.3 Ma). The first rapid subsidence emerged in the Oligocene and the subsidence rate was up to 0.45 m/ky in the Changchang Depression. The subsidence rates in other tectonic units also exceeded 0.2 m/ky (Figure 7). The tectonic subsidence was widespread up to 2 km in the basins around the Xisha area in the rapid subsidence phase and could be further divided into two portions by the boundary of T 7 (29.3 Ma), as shown in Figures 6 and 7. The rate of tectonic subsidence of 32-29. 3 Ma was higher than that of 29.3-23. 3 Ma, as shown in Figure 7. The tectonic subsidence in the Zhongsha Trough was also restored with 0.15 m/ky in 29.3-23. 3 Ma, as shown in Figure 7.
Third phase-the interim phase in the Early Miocene (23.3-15.5 Ma). Relatively slow subsidence followed the first rapid subsidence that occurred in the Early Miocene. The rate of subsidence was about 0.05 m/ky or even lower in most of the tectonic units in Figure 7. During the interim stage (23.3-15.5 Ma), the tectonic subsidence was about 0.6 km in basins and 0.4 km on the uplifts, and the average rate of tectonic subsidence, much smaller than that in the rapid subsidence phase, was about 0.05 m/ky, even in basins, as shown in Figure 6. Fifth phase-the transitional phase since the Pliocene (5.5-1.9 Ma) to the present. In the transitional phase, the total tectonic subsidence was about 0.4 km in the basins and even less than 0.2 km on the uplifts. Although the rate of subsidence was low in the Pliocene, the rate of tectonic subsidence increased again in the Quaternary, as shown in Figure 7. The average rate of tectonic subsidence was as low as 0.05 m/ky in the Pliocene and grew to 0.1 m/ky in the Pleistocene, as shown in Figure 7.

Discussion
Generally, the tectonic subsidence of passive continental margin rift basins is due to the stretching and thinning of the lithosphere during the rifted stage and increased density of mantle material due to heat loss in the post-rifted stage [31,37,38]. There are a few controversies about the rapid subsidence occurring in the rifted stage. In many regions, usually there is a discrepancy between the theoretic value and actual computed value during the post-rifted stage [9,31]. To explain this anomaly, a number of mechanisms have been proposed [9,31,39,40]. Among these mechanisms for the northern South China Sea area, dynamic topography is a relevant mechanism that can interpret the discrepancy in tectonic subsidence rate between uplifts and basins during the post-rifted stage. However, to some extent, the dynamic topography is not very effective in explaining the cause of tectonic subsidence in the Xisha area during the active tectonic period. Therefore, regional geological events, such as the reversal of the Red River Fault Zone and extensive magmatism, should be considered to explain the subsidence process during the post-rifted stage, as shown in Figure 9. The Eocene tectonic subsidence of the Xisha area is in the slow subsidence phase, and the total subsidence is usually less than 2 km relative to the northern Qiongdongnan Basin, which is consistent with the restoration results acquired from the wells in the Xisha area [10,27]. Inevitably, the tectonic subsidence on the Xisha Uplift and Guangle Uplift cannot be restored accurately due to thinner sediment and more uncertain paleo-water depths, as shown in Figures 5 and 8. Based on the intensity of faults on the uplifts, the inferred tectonic subsidence was smaller than that in the basins. Stretched and thinned lithospheric mantle and crust were universally considered the primary cause of the subsidence of passive continental margin rift basins, which has been confirmed by the imitation of crust stretching in the north margin of the South China Sea [41,42]. The driving mechanism of crust stretching remains under debate between two main views: one is the dominant role of extrusion of the Indochina Block, and the other is the sinistral motion of the Red River Fault Zone [14,43,44] or the subduction of a proto-South China Sea at the North Borneo Trench [45,46]. During the Eocene, the Xisha Block slowly moved towards the southeast with a slightly anticlockwise rotation caused by the primary rift in the western part of Qiongdongnan Basin. This movement was also accompanied by the NE-to ENE-trending normal faults, which were more in favor of the extrusion of the Indochina Block [47]. The rate of tectonic subsidence should be high in the initial stretching and thinning of lithospheric mantle and crust in theory. However, the actual rate of tectonic subsidence is only 0.05 m/ky in the Yinggehai Basin and north of the Qiongdongnan Basin [9], which may contribute to the diachronism of the basement in the Xisha area. Although the tectonic subsidence was a particular phenomenon in this period, it represented the process of the initial rifted stage. The Eocene tectonic subsidence of the Xisha area is in the slow subsidence phase, and the total subsidence is usually less than 2 km relative to the northern Qiongdongnan Basin, which is consistent with the restoration results acquired from the wells in the Xisha area [10,27]. Inevitably, the tectonic subsidence on the Xisha Uplift and Guangle Uplift cannot be restored accurately due to thinner sediment and more uncertain paleo-water depths, as shown in Figures 5 and 8. Based on the intensity of faults on the uplifts, the inferred tectonic subsidence was smaller than that in the basins. Stretched and thinned lithospheric mantle and crust were universally considered the primary cause of the subsidence of passive continental margin rift basins, which has been confirmed by the imitation of crust The rate of tectonic subsidence in the Early Oligocene was higher than that in the Late Oligocene, as shown in Figures 6 and 7. The tectonic subsidence in the Zhongsha Trough was 0.15 m/ky in this phase for the sediment depositing all over the study area. This rapid subsidence was universal in the north margin of the South China Sea [10,31,34,48]. The inherited stretched and thinned lithosphere and crust were considered to be responsible for the rapid subsidence and the crust-stretching factors of the Qiongdongnan Basin. Spanning from β = 1.2 of uplifts in the basin margin to β = 3.14 in the central basin [42] interpreted the change in tectonic subsidence from uplift to central basin.
The activities of faults that continued to the end of the Oligocene indicated that the process of stretching and thinning probably terminated at approximately 23.3 Ma (T 6 ), as shown in Figures 4 and 5. This couples well with the rapid subsidence occurring during the Oligocene, and the homeochronous high sedimentary rate also matched the rapid subsidence [10,27]. According to the analysis of seismic data, the process of the Xisha Block splitting from the South China Block initially occurred in the western Qiongdongnan Basin and gradually extended to the east. The homologous extension of the rift zone ranged from 15 km to 35 km, which indicated that the anticlockwise rotation of the Xisha Block was intensified. In the rift zone, there were three echelon fault zones: one of them was being along the No. 1 Fault and the other two were distributed in its eastern portion [47,49]. Seismic profiles revealing the half graben and titled block of structural styles also concluded that the nearly NE-trending faults around the Xisha area should be under the control of extrusion stress, as shown in Figures 2 and 5, although there have been other arguments on the structural differences between the western and eastern portions of Qiongdongnan Basin, which probably were related to the sinistral movement of the offshore Red River Fault and the subduction of the proto-South China Sea in the North Borneo Trench separately [6,8]. The characteristics of faults around the Xisha area and the anticlockwise splitting process of the Xisha Block suggest that the extrusion of the Indochina Block and the associated leftlateral slip of the Red River Fault Zone led to the stretching and thinning of the lithosphere and crust in the Xisha area, resulting in the rapid Oligocene subsidence in the Xisha area. This standpoint was also supported by the geophysical imitation of crust deformation of the Qiongdongnan Basin [2,50,51].

Tectonic Subsidence of Miocene (23.5-5.5 Ma)
The tectonic evolution of the Xisha area shifted from rifted stage to post-rifted stage in the Early Miocene [5,21]. Theoretically, the tectonic subsidence caused by the heat loss of the deep mantle should be relatively slow. However, the rate was different from the recovery of tectonic subsidence in the study area, and this anomalous subsidence pervasively existed on the north margin of the Qiongdongnan Basin [9,10,48]. In former research, the interpretation usually focused on the rapid subsidence occurring in the Late Miocene, while little attention was given to its change during the Miocene. Based on an analysis of the relationship between the regional geological events and the change in tectonic subsidence rate, the process of subsidence was divided into two phases: the interim phase  and the accelerated rise phase .  At the end of the Oligocene (23.3 Ma), all faults ceased and formed the most important unconformity of the Xisha area-the breakup unconformity (T 6 ), as shown in Figure 9. The extensional strain of crust was inferred to be close to 0 km since 23.3 Ma based on the analysis of the balanced cross section in Figure 5. Both of the above showed the termination of stretched and thinned lithosphere and crust at this boundary, which should be the main cause of this slow tectonic subsidence. This standpoint has been extensively applied to interpret the low tectonic subsidence simultaneously occurring in the Pearl River Mouth Basin, northern Qiongdongnan Basin and Beibuwan Basin [9,10,27,34]. Although the Red River Fault Zone went forward in the Early Miocene [4], the ridge of seafloor spreading in the northeastern South China Sea was deserted. Subsequently, the southward ridge jump occurred near the end of the Oligocene (23. 6 Ma) in the East Subbasin, which should be the ending mechanism of the rifted stage. This ridge jump was also considered the origin of the post-rifted stage in the Pearl River Mouth Basin, despite lagging behind its breakup unconformity at T 7 (29.3 Ma) [31].

Fourth
Phase-Accelerated Rise Phase from Middle to Late Miocene  The relatively rapid tectonic subsidence in the Late Miocene was consistent with the value restored in the Qiongdongnan Basin, which was considered to be associated with the dynamic topography [9]. In this period, the evolution of the Red River Fault Zone entered the reversal stage and terminated at the end of the Late Miocene [4]. Based on the information from the stress-field analysis, as shown in Figure 10A, there was an extrusive component to the Xisha Block generated by the sinistral movement of the Red River Fault Zone, which compelled the Xisha Block to split anticlockwise of the South China continent in the Paleogene [47]. Although this splitting process stopped in 23.3 Ma, the extrusive component remained and the extrusive intensity was interrelated with the displacement rate of sinistral movement. During the reversal stage of the Red River Fault Zone, the gradual weakening of extrusive intensity with the fall in displacement rate led to stress release of the Xisha area, which further resulted in tectonic subsidence. The accelerating leakage of stress might have caused the rise in tectonic subsidence rate from the early Middle Miocene (15.5 Ma) to the end of the Late Miocene (5.5 Ma).   Two sedimentary recordings in the Xisha area coming from the carbonate platform and the mass transport deposits (MTDs) responded well to the tectonic subsidence, as shown in Figure 3 [21,28,36]. Since the Middle Miocene, the spatiotemporal distribution of carbonate platform was identified in six periods and could be further divided into three stages ( Table 2). The scale of carbonate platform reduced slowly in the Middle Miocene and shrank quickly in the Late Miocene, which reflected an accelerating process of the tectonic subsidence in accordance with the restoration in this stage. The development of the Huaguang MTDs in the upper Huangliu Formation (8.2-5.5 Ma) provided evidence of coeval rapid tectonic subsidence [28]. The instability of sediment caused by rapid subsidence was considered to be the dominant trigger mechanism for MTDs, which were related to the reversal of the Red River Fault Zone from sinistral to dextral movement. It gradually weakened from west to east in the Xisha area, inferred from the effects of the heating event [52]. Additionally, the tectonic subsidence in the north slope of the Qiongdongnan Basin also gradually weakened from west to east in the Late Miocene, which corresponds to the effect that it gradually declined with the increased distance to the Red River Fault Zone [10]. The slow sedimentary rate also responded to the slow tectonic subsidence, as shown in Figure 8. We should be cautious in interpreting the tectonic subsidence as a result of the activity of regional geological events, such as the volcanism, the dextral movement of the Red River Fault Zone and the transgression of the continental shelf from west to east [4,5,9]. Theoretically, when the nature of the Red River Fault Zone changed from sinistral to dextral movement, the extrusive component could have disappeared and shifted to an extensive component according to stress-field analysis, as shown in Figure 10B, which would lead to regional stress to absolute release and further cause rapid tectonic subsidence. Indeed, this rapid tectonic subsidence took place and was accompanied by diapirs and high rate of sedimentation in the Yinggehai Basin, the Beibuwan Basin, and the northern Qiongdongnan Basin [8,17,38,48,53,54].
However, this rapid subsidence did not prospectively occur in the Xisha area, but slow subsidence did, perhaps due to volcanism in this period. The volcanism was fierce in the Xisha area, different from the weak volcanism to the west of study area, constrained by seismic stratigraphy and well data since 5.5 Ma [24,55,56]. Unlike the magmatism in the rifted stage, which overflowed along the major faults, the vertical intrusive magma appeared to uplift the overburden or extrude peripheral strata and lead them to rise after the late Miocene [5,57]. This vertical intrusion, associated with the Hainan Mantle Plume, probably hampered the regional tectonic subsidence. The statistics of drill cores also revealed that most of the volcanoes were primally active in the Pliocene and gradually weakened from 1.9 Ma [58], which should be responsible for the slow rise in subsidence rate in the Pleistocene.

Conclusions
Based on the information obtained from the wells and pseudo-wells extracted from seismic profiles, the tectonic subsidence of rifted basins surrounding the Xisha area was systemically recovered, and the following conclusions are drawn.
(1) The subsidence in the structural highs is much smaller than that in basins. The largest tectonic subsidence in the Xisha area is located in the Changchang Depression, with maximum subsidence of 5.4 km and maximum subsidence rate of 0.45 m/ky. (2) Subsidence history in the Xisha area, including two phases of relatively rapid subsidence, could be divided into five phases since the Eocene to the present: the slow subsidence phase , the rapid subsidence phase (32-23.