Magnetotelluric Evidence for Lithospheric Alteration Beneath the Wuyi‐Yunkai Orogen: Implications for Thermal Structure of South China

The Wuyi‐Yunkai Orogen experienced a polyphase tectonomagmatism and is a key region for deciphering the alteration and thermal structure of the South China Block lithosphere. Herein, an electrical resistivity model of the lithosphere is presented via the three‐dimensional inversion of broadband (0.003–3600 s) magnetotelluric (MT) data collected along a 380‐km‐long profile comprising 62 MT sites across the Wuyi‐Yunkai Orogen, and the robustness of this model is critically evaluated through a series of sensitivity tests. The resistivity model reveals that the upper crust of the Cathaysia Block and the Wuyi‐Yunkai Orogen is dominated by high‐resistivity sedimentary cover interposed with low‐resistivity features, mainly along fault zones. High‐resistivity bodies and strong conductors in the upper crust are interpreted as magmatic rocks and tectonic mélanges, respectively. Another feature of this resistivity model is the presence of zones featuring enhanced electrical conductivity (<30 Ωm) extending from the lower crust to the upper mantle beneath the Wuyi‐Yunkai Orogen. The conductors in the lower crust are attributed to saline fluids from either the dehydration of the subducting Paleo‐Pacific slab or the regional metamorphism‐induced dehydration of sandy argillaceous rocks. In contrast, the conductors in the upper mantle are attributed to 4%–7% partial melt, which corresponds to the analyses of mantle xenoliths in South China. These conductors in the upper crust and upper mantle supply volatiles and heat to shallow geothermal systems. This work ultimately shows that the lithospheric thinning of South China is controlled mainly by mantle upwelling caused by the retreat of the subducting Paleo‐Pacific slab.

The Wuyi-Yunkai Orogen ( Figure 1) is a vast NE-SW-trending block located in the transition zone between the Cathaysia Block and Yangtze Craton that is bounded by the Wuchuan-Sihui shear zone (WSSZ) in the east and the Bobai-Wuzhou fault zone in the west (Lin et al., 2008;Liu et al., 2018). The Wuyi-Yunkai Orogen is considered to be an ancient continental block that formed at least 1.0 Gyr and then underwent two periods of strong orogenesis in the early Paleozoic (∼467-413 Ma) and early Mesozoic (∼243-120 Ma) (Zhang et al., 2012), producing structures accompanied by large areas of gneissic granites and middle-to high-grade metamorphic rocks (Lin et al., 2008;Liu et al., 2018). The tectonic activities manifest mainly as strong tectonic deformation and middle-to high-grade metamorphism in the Wuyi-Yunkai Orogen, and the strongly deformed areas are oriented primarily NE-NNE (Lin et al., 2008;Wang et al., 2007). As evidenced above, the Wuyi-Yunkai Orogen has experienced multiple stages of tectonism, resulting in abundant ductile shearing signatures and fold-andthrust belts documenting the tectonic evolution and tectonomagmatic features of the South China Block. Thus, this block is a key area for studying the mechanisms responsible for the thermal structure and thinning of the lithosphere of South China.
Considering both the absence of island arc volcanic rocks and ophiolites to confirm the occurrence of early Paleozoic subduction and the extensive superimposition of tectonometamorphic structures, the South China Block have been influenced by multiple tectonic events, such as the opening and closing of the Paleo-Tethys Ocean (Qiu et al., 2020), the activity of the Emeishan mantle plume (Xu et al., 2004), and the initiation of subduction of the Paleo-Pacific Ocean (Li & Li, 2007). Therefore, several geodynamic models have been proposed to explain the dynamic mechanism of intracontinental orogeny in South China, such as the continental collision (Hsü et al., 1990), intracontinental subduction (Charvet et al., 2010), and the divergent double subduction model (Zhao, 2015), with the subduction of Paleo-Pacific Plate and lithosphere delamination being emphasized to different extents (Li & Li, 2007;Liu et al., 2018). Given that the lithosphere can retain the footprint of intracontinental orogeny, and the current thermal structure of the lithosphere can record the lithospheric alteration, geophysical Figure 1. (a) Regional topographic maps of South China and (b) simplified geological map (red box in (a)) illustrating the major tectonic features around the Wuyi-Yunkai Orogen . The blue and red circles showing the distribution of magnetotelluric (MT) sites. The red circles denote the representative MT sites whose data are plotted in Figure 2, and the purple triangles represent the heat flow sites in mW/m 2 (Jiang et al., 2019). F 1 : Bobai-Wuzhou fault; F 2 : Luchuan-Guangning fault; F 3 : Shiwo-Fengdongkou fault; F 4 : Lianjiang-Xinyi fault; F 5 : Yangjiang fault.
imaging of lithospheric structure has the potential to provide a new perspective on intracontinental orogeny in South China. However, the lack of detailed geophysical studies of lithospheric structure beneath Wuyi-Yunkai orogeny restricts the resolution of lithospheric alteration mechanism and thermal structure.
The MT method has proven to be effective for imaging the electrical resistivity distribution of the crust and mantle for the investigation of tectonic deformation and lithospheric thinning in Precambrian terranes (e.g., Bologna et al., 2019;Haugaard et al., 2021;Heinson et al., 2018;Hill et al., 2021;Kay et al., 2022;Padilha et al., 2019;Roots et al., 2022). Electrical resistivity is sensitive to temperature and composition but especially to the presence of fluids, which appear in the form of free water or partially molten rock (Chave & Jones, 2012). The aim of this study is to constrain the electrical structure of the lithosphere beneath the Wuyi-Yunkai Orogen to better define the features of tectonic deformation. In this study, new dense MT data were collected crossing the Wuyi-Yunkai Orogen in South China to image the detailed electrical structure of the lithosphere, which potentially bears fossil signatures to shed light on the mechanisms of lithospheric alteration in the South China Block and lithospheric thermal structure related to subsequent tectono-thermal events.

Geological Setting
The Wuyi-Yunkai Orogen, nestled in the interior of the South China Block, is located on the SW margin of the Cathaysia Block and is adjacent to both the Indosinian Block and the Yangtze Craton in the west (Lin et al., 2008;Wang et al., 2007). The orogen has experienced five evolutionary stages: the growth of its continental core during the pre-Jinningian, the propagation of tension fractures during the Jingningian (∼1350-950 Ma), folding and uplift during the Caledonian (∼490-390 Ma), collisional orogenesis during the Hercynian-Indosinian (∼380-210 Ma), and intracontinental extension during the Yanshanian-Himalayan (∼205-66 Ma) Wang et al., 2013). The main geological periods are characterized by magmatic (predominantly felsic) activity ( Figure 1). The axis of the Wuyi-Yunkai Orogen is predominated by Caledonian magmatic rocks, while Hercynian-Indosinian (∼380-210 Ma) magmatic rocks are mostly distributed on the two limbs of the Wuyi-Yunkai Orogen, and Yanshanian (∼205-135 Ma) magmatic rocks are distributed mainly along fault zones and the margins of Mesozoic grabens (Figure 1). The basement of the Wuyi-Yunkai Orogen comprises two layers. The deep basement is composed of Paleo-Mesoproterozoic middle-to high-grade metamorphic rocks derived from terrigenous clastic protoliths that underwent amphibolite-facies metamorphism (partially superimposed by granulite-facies metamorphism) and granites that were remelted to varying degrees during the Jingningian-Caledonian (∼1000-390 Ma) tectonothermal event. In contrast, the shallow basement is composed mainly of Neoproterozoic (∼1000-540 Ma) to early Paleozoic (∼467-413 Ma) metamorphic rocks derived from terrigenous clastic rocks dominated by sandy argillaceous rocks that have been metamorphosed into greenschist facies .
The Wuyi-Yunkai Orogen is characterized by large-scale metamorphic rocks and mylonites and overall presents as a modified dome structure (Lin et al., 2008;Wang et al., 2007). Metamorphic complexes with granitic rocks compose the main body of the orogen and are widely exposed; according to the lithological and metamorphic characteristics of their constituents, these complexes can be divided into two sets of rock formations. The upper set, called the Yunkai Group, is composed of metamorphic, sedimentary, and volcanic rocks (greenschist facies and epidote-amphibolite facies) and extends more than 100 km WNW-ESE along the northern Wuyi-Yunkai Orogen; the lower set, called the Gaozhou Complex, is composed of high-grade metamorphic rocks, namely, biotite granulite, biotite monzonite gneiss, plagioclase amphibolite and gneiss granite (including charnockite), which are widely exposed in the core of the Wuyi-Yunkai Orogen Zhang et al., 2012). The metamorphosed granitic rocks distributed in the core of the Wuyi-Yunkai Orogen have chemical compositions and structural features that are typical of granites, which are present mainly as granodiorites (Li et al., 2010); however, in a later period, these granites were subjected to near-horizontal shear, resulting in partial melting of the rock that generally resulted in the formation of gneissic structures (Wang et al., 2007). The attitude of the intrusions within the host rock is consistent with that of the foliation; that is, the intrusions exhibit gentle dips (generally less than 35°) that slope overall toward the NE and present as planar sheets (Lin et al., 2008).
The tectonic deformation of the Wuyi-Yunkai Orogen occurred in multiple stages from the Paleozoic to the Mesozoic (Lin et al., 2008;Wang et al., 2007). The first event was characterized mainly by top-to-the-NW ductile shear in conjunction with NE-SW-trending lineations and amphibolite facies metamorphism, which probably transpired during a postorogenic (∼467-413 Ma) migmatization event due to extension. The second event occurred in the early Mesozoic (∼243-120 Ma) and mainly developed the high-strain shear zone in the northern Wuyi-Yunkai Orogen characterized by top-to-the-NE ductile shear possibly linked with the subduction of Indochina beneath South China. The third (compressional) event resulted in thrusting and folding with associated NE-SW-trending foliations and lineations in sedimentary strata and gneissic banding in the basement linked with either the northwestward subduction of the Pacific plate beneath or the southeastward thrusting of the Songpan-Ganzi fold belt onto and over the Yangtze Craton. The final deformation event is represented by NE-SW-trending high-angle normal faults and sinistral strike-slip faults, which collectively signify formation under regional extension.
The heat flow of the South China Block is controlled predominantly by the tectonic setting: the Yangtze Craton in the NW is relatively stable with low heat flow, while the Cathaysia Block to the SE is close to the plate boundary and thus exhibits high heat flow (Jiang et al., 2019). The local thermal structure of South China has been shown to control the heat flow distribution, which gradually increases overall from west to east (Sun et al., 2013). Within the study area, the highest heat flow was recorded inside the Wuyi-Yunkai Orogen at 98.2 mW/m 2 , while the heat flow reduces to 85.7 mW/m 2 at the boundary, and the value is approximately 75 mW/m 2 just beyond its boundaries ( Figure 1). All of these values are greater than the average heat flow (60.2 mW/m 2 ) of mainland China (Jiang et al., 2019), in which the distribution of hot springs is closely related to magmatic activities, with most hot springs distributed on the edges of magma chambers or the boundaries between intrusive rocks and other strata (Luo et al., 2022). The hot springs in the Wuyi-Yunkai Orogen are exposed mainly in the Yangjiang and Xinyi areas. The maximum hot spring temperature in the Yangjiang geothermal field is 97°C, and the bottom-hole temperature of a 1,000-m well is 110.2°C (Lu et al., 2017), while the hot springs in the Xinyi geothermal field are naturally exposed along the riverbed, and the maximum temperature reaches 75°C (Lu & Liu, 2015).
The brittle-ductile transition zone beneath the Cathaysia Block is located at depths of 15-20 km according to the electrical structure of the lithosphere, and an asymmetric simple shear extension model was proposed for the South China rift system (Xu et al., 2019). Teleseismic body-wave attenuation and receiver function analyses suggested the presence of a large low-velocity anomaly beneath the Cathaysia Block in South China that is probably associated with mantle upwelling (Deng et al., 2021;He & Santosh, 2021). The average crustal thicknesses of the Cathaysia Block and the Yangtze Craton are significantly different; nevertheless, they present almost the same V p /V s values, which reflect the typical bimodal distribution of felsic and gabbroic rocks within the crust of South China (Ji et al., 2016). Furthermore, the Bouguer gravity anomaly distribution across South China displays low values in the north and high values in the south; in particular, the Xinyi region in the Wuyi-Yunkai Orogen exhibits low-value anomalies (Xi et al., 2018). Low gravity anomalies reflect thicker sedimentary cover, while high gravity anomalies indicate thinner sedimentary cover and uplifted crystalline basement (Deng et al., 2014).

Field Data
MT data are acquired via a passive source, namely, the ultra-low-and extremely low-frequency geomagnetic field components produced by solar wind and lightning, where the penetration depth of electromagnetic waves increases with the subsurface resistivity and the period of electromagnetic waves (Chave & Jones, 2012). Two electric field components (E x , E y ) and three magnetic field components (H x , H y , H z ) were measured for more than 20 hr at 62 broadband stations in the period range from 0.003 to 3600 s, which is suitable for obtaining the electrical structure of the crust and upper mantle. The 380-km-long MT profile was oriented NW-SE across the Wuyi-Yunkai Orogen with an MT station interval of 4-10 km depending on the topography (Figure 1). A Phoenix MTU-5 (Phoenix Geophysics Ltd., Canada) commercial acquisition system was employed during the survey. The magnetic field components were detected by induction coil magnetic sensors, and the telluric field components were detected by nonpolarized Pb-PbCl 2 electrodes with an electrical dipole length of 50 m. The electric and magnetic field sensors were oriented in alignment with the principal (N-S and E-W) axes of the geomagnetic field. To eliminate man-made electromagnetic noise, reference data were simultaneously collected at a remote MT station 500 km from the profile (Gamble et al., 1979).
The MT time series were processed by the SSMT2000 program (Phoenix Geophysics Ltd., Canada), and MT transfer functions were calculated using remote reference processing with statistically robust algorithms (Gamble et al., 1979;Jones et al., 1989). The power spectra were manually selected by using MTEditor (Phoenix Geophysics Ltd., Canada) software to obtain the high-quality MT impedances and vertical field transfer functions. The apparent resistivity and phase curves for four representative MT sites are displayed in Figure 2. Site a is located in the foreland basin within the Yangtze Craton, and the apparent resistivity and phase curves exhibit responses to an almost one-dimensional subsurface structure at periods less than 100 s and are relatively conductive at periods less than 10 s, which indicates the presence of a low-resistivity layer in the shallow crust. The site in the center of the MT profile (site b within the Lingshan-Tengxian fault zone) is characterized by elevated resistivities indicating the existence of a complex subsurface electrical structure. Site c within the intermountain basin features elevated apparent resistivities at all periods; the data show that the xy-and yx-mode apparent resistivity curves almost overlap at periods less than 1 s, which indicates the existence of a typical one-dimensional sedimentary structure beneath this site. Site d at the southeastern end of the MT profile shows a shift in the MT apparent resistivity curves caused by locally inhomogeneous structures; this shift was corrected by a statics correction during the subsequent inversion. Figure 2. Apparent resistivity and phase curves for the four representative magnetotelluric sites used in the inversion, which were indicated as red circles in Figure 1b. The red and blue circles represent XY-mode and YX-mode data, respectively, in the observation coordinate system. The X-axis points toward geomagnetic north.

Dimensionality Analysis
The dimension of the MT impedance tensor can reflect the complexity of the subsurface resistivity structure, so before inversion and interpretation, a detailed dimensional analysis is indispensable for determining the dimension of the subsurface structure in order to select an appropriate inversion method (Martí, 2014). The phase tensor is free from influence of the galvanic distortion produced by heterogeneities in the near-surface structure, where the MT phase tensor is defined as the ratio between the real and imaginary parts of the impedance tensor, Φ = Im(Z)/Re(Z) (Bibby et al., 2005;Caldwell et al., 2004). The phase tensor skew angle (β) defines the asymmetry of the phase tensor, describing the complexity of the subsurface structure.
For a one-dimensional (1-D) or two-dimensional (2-D) subsurface structure, |β| is relatively small. Considering the observation error and artificial electromagnetic noise, if the absolute value of β is greater than 5°, then this indicates the existence of a three-dimensional (3-D) geoelectrical structure (Cai et al., 2017). Figure 3a shows the ellipses of the phase tensor with the values of |β| at different periods. The values of |β| at most MT sites are less than 5° at periods shorter than 1 s, indicating the existence of a 1-D or 2-D electrical resistivity structure in the shallow crust of the Wuyi-Yunkai Orogen. However, the values of |β| increase at periods greater than 1 s, implying the presence of a complex 3-D geoelectric structure in the middle-lower crust. The orientation of the principal axis of the phase tensor in the WSSZ is NW-SE, which is either parallel or perpendicular to the geological strike due to the 90° ambiguity in the polarization of electromagnetic fields (Booker, 2014;Peacock et al., 2021). Overall, the phase tensor results show significant 3-D structural features along the MT profile, and thus, the 3-D inversion and interpretation of the MT data from this profile are essential for realistically imaging the subsurface electrical structure.
Likewise, Figure 3b shows the ellipses of the phase tensor with the values of Φ 1 , which denotes the geometric mean of the maximum (Φ max ) and minimum (Φ min ) phases and represents the variation in resistivity with depth (Hill et al., 2009. In the Cathaysia Block and Wuyi-Yunkai Orogen, the Φ 1 values of short-period data increase with the period at most MT sites, which corresponds with the presence of shallow magmatic rocks with highly resistivity. Furthermore, low-resistivity bodies are present in the middle-lower crust beneath the Cathaysia Block and the Wuyi-Yunkai Orogen. The Φ 1 values at the MT sites in the Yangtze Craton decrease with depth, which corresponds with the sedimentary cover of the foreland basin.

3-D Inversion
The 3-D electrical resistivity structure was derived from the data set of 62 MT sites using the parallel inversion code ModEM based on the nonlinear conjugate gradient algorithm (Egbert & Kelbert, 2012;Kelbert et al., 2014). We removed obvious outliers and reduced the weights of suspicious data points by increasing the standard deviation, especially in the period of 1-20 s, which is characterized by strong artificial electromagnetic noise. All four components of the unrotated impedance tensor were used in the 3-D inversion with 40 periods in the period range from 0.003 to 3600 s ( Figure S1 in Supporting Information S1). However, magnetic induction vectors were not employed in the inversion to reduce the impact of artificial electromagnetic noise considering that tipper data are sensitive to external source bias (Parkinson, 1962;Samrock & Kuvshinov, 2013). The inversion was run with an error floor of 5% for the antidiagonal impedance tensor and 10% for the diagonal impedance tensor. The central region of the inversion model was 336 km (E-W) × 168 km (N-S) and was discretized into horizontal cells with a spacing of 3.5 km in the E-W direction and 4 km in the N-S direction. The grid outside the core area was padded with 10 cells (whose sizes were increased by a factor of 1.5 around the four sides of the model to eliminate boundary effects over an area of 1550 km (E-W) × 1550 km (N-S). In the vertical direction, the thickness of the grid was set to 20 m in the first layer, and the mesh size of subsequent layers was successively increased by a factor of 1.2 to a maximum depth of 800 km. Thus, the final model grid contained 122 (E-W) × 68 (N-S) × 60 (vertical) cells (including 10 padded grids in all directions and 7 air layers). A series of 3-D inversions with different initial models (50, 100, 200, and 500 Ωm) were performed, and those resistivity models have similar main features ( Figures S2-S5 in Supporting Information S1). The model derived from a 100 Ωm initial resistivity has the smallest final root mean square (RMS) value. Therefore, the initial model was assigned a homogeneous resistivity of 100 Ωm with a smoothing factor of 0.3 in all directions during the inversion. The model derived from the inversion with topography was very similar to the model without topography, but it has a higher RMS value ( Figure S6 in the Supporting Information S1). The regularization parameter (λ) was assigned a value of 3,000 as the initial value and divided by 10 when the RMS barely changed. The final resistivity model with a normalized RMS misfit of 2.82 was obtained after 191 iterations. The site-by-site RMS misfit distribution is shown in Figure 4, the results in which indicate that the final model ( Figures 5 and 6b) has a high degree of confidence and can match the observed data at most MT sites ( Figure S1 in Supporting Information S1).

Inversion Result
The final model obtained from the 3-D inversion is presented in Figure 5, which presents horizontal resistivity slices at depths of 5, 10, 20, 30, 40, and 60 km, and in Figure 6b, for which the vertical cross section was extracted along the MT profile. Overall, the electrical resistivity structure of the lithosphere in the study area shows significant heterogeneity in both the vertical and the horizontal distributions, which is in accordance with the remarkable 3-D features exposed by the dimensional analysis. The final model shows a series of highly conductive anomalies (1-30 Ωm) labeled C1-C4 and a highly resistive anomaly (greater than 1,000 Ωm) marked R. The upper crust beneath the Cathaysia Block and the Wuyi-Yunkai Orogen is dominated by high-resistivity cover (R), which is interposed with low-resistivity features, mainly within the fault zone. Similar crustal structures have also been imaged in other areas of the South China Block (Cheng et al., 2021;Li et al., 2022;Xu et al., 2019;Yin et al., 2021). The high-resistivity cover can be explained by the widespread distribution of magmatic rocks.
In contrast, the upper crust of the Yangtze Craton is characterized mainly by a low-resistivity layer (C1), which predominantly reflects the sedimentary cover and tectonic mélange (Liang & Li, 2005;Liu et al., 2018). In addition, a low-resistivity zone (C2), which is imaged as a nearly upright conductor, appears beneath the Bobai-Wuzhou and Luchuan-Guangning fault zones and extends from the near-surface region to a depth in excess of 50 km. Similar to conductor C2, the WSSZ appears as a subvertical conductor (C3) in the crust beneath the Wuyi-Yunkai Orogen, consistent with the shallow structure imaged by dense seismic array data (Gao et al., 2022;Li et al., 2021), which can be interpreted as a tectonic fracture zone or flower structure caused by wrench faulting. The conductor C4 in the upper mantle beneath the Cathaysia Block may be related to the lithospheric alteration. Furthermore, the mantle beneath the Cathaysia Block and the eastern Wuyi-Yunkai Orogen exhibits a low resistivity at depths of 40-60 km, whereas the mantle beneath the Yangtze Craton and the western  Wuyi-Yunkai Orogen is highly resistive at the same depths, which is similar to the velocity structure of the mantle (Figure 6d).

Sensitivity Tests
Some resistivity anomalies may be spurious or be poorly resolved when a sparse grid is used to reduce the roughness and residuals of the inversion model (Jones, 1999). Thus, to improve the stability and accuracy of the interpretations based on these MT data, a series of sensitivity tests were performed to evaluate the robustness of the structural features within the inversion model. The tests were carried out by modifying specific features in the original model (Figures S7-S11 in Supporting Information S1). 3-D forward modeling was performed on these test models to examine the differences between the calculated data and the observed data. The sensitivity tests were divided into two groups.
In the first group, conductor C1 was replaced with the average crustal resistivity (1,000 Ωm) block. The RMS misfit originating from the revised model was greater than the RMS misfit produced by the original model, and the misfits of the data at MT sites above conductor C1 were particularly poor ( Figure S8b in the Supporting Information S1). Similar test was also performed to verify the robustness of conductor C3. The revised model was generated by replacing C3 in the original model with a constant-resistivity (1,000 Ωm) block, and the RMS misfit produced by the revised model was greater than that of the original model ( Figure S8d in the Supporting Information S1). The resulting variations in the site-by-site RMS values indicate that the data at the MT sites above conductor C3 were poorly fitted. We similarly replaced mantle conductors (C2 and C4) with the crustal resistivity (1,000 Ωm) and the mantle resistivity (100 Ωm), respectively, and generated modified models, which produced a similar phenomenon to crustal conductor tests, with a larger RMS mismatch than the original model (Figures S9 and S10 in the Supporting Information S1). The data misfits at MT sites above mantle conductors (C2 and C4) were of poor quality after they were replaced with more resistive blocks. According to these tests, the highly conductive anomalies (C1-C4) are essential to the final model, indicating that these low-resistivity bodies are reliable features and are required by the MT data.
The second group of tests was conducted to verify the reliability of the final inverted model at different depths. To do this, we replaced the final model beneath depths of 20, 40, and 60 km with the crustal resistivity (1000 Ωm) and the mantle resistivity (100 Ωm), respectively, and calculated the response of the modified model (Figures S10 and S11 in the Supporting Information S1). When the fixed depth is set to 60 km, the RMS misfit of the modified model is almost the same as the RMS misfit of the original model, indicating that the final model is reliable to a depth of 60 km. Therefore, we focused on discussing and interpreting the final model in detail from the surface to a depth of 60 km in the following sections.

Crustal Electrical Structure
The final resistivity model correlates well with the known geological features of the near-surface region (0-2 km depth), indicating that the model accurately represents the shallow subsurface resistivity structure. In the shallow crust (0-20 km depth) of the Wuyi-Yunkai Orogen and Cathaysia Block along the MT profile, the resistivity model is generally resistive with the exception of several thin conductors (10-100 Ωm) near the surface. These conductors are located between stations B09-B11, B17-B24, B30-B32, and B39-B44 and are correlated with fault zones mapped at the surface. In contrast, highly resistive areas near the surface may correspond to outcrops of middle-to high-grade metamorphic rocks, granites and Precambrian basement rocks, as expected for dry, low-porosity crystalline rocks at low temperatures (Li, 2000;Shu et al., 2014). The upper crust of the Yangtze Craton is dominated by a conductive layer, which correlates with the sedimentary foreland basin covering the tectonic mélange (Liang & Li, 2005).
There are Late Permian-Middle Triassic sediments with a maximum thickness of 5 km in the low-velocity foreland basin, that formed on continental crust between the Wuyi-Yunkai Orogen and the adjacent Yangtze Block (Liang & Li, 2005). The surficial conductor (C1) has been attributed to either an accumulation of saline fluid in thick and coarse molasse deposits or to an increase in fluid salinity (or both) at the bottom of the sedimentary that underwent uplift in the Indosinian Orogen (Hu et al., 2015). The Bouguer gravity anomalies from the 2012 World Gravity Model (WGM2012) database (Figure 6a, Balmino et al., 2012) gradually increase from NW to SE, indicating that the upper crustal density of the Cathaysia Block is greater than that of the Yangtze Craton, which is consistent with our resistivity model. Moreover, the velocities of the upper crust in the Cathaysia Block are also greater than those in the Yangtze Craton (Shan et al., 2017).
Another low-resistivity feature (C3) observed below the WSSZ (sites B17-B24) extends from the surface into the upper mantle at high angles; this conductor likely reflects the large-scale and deep extension of the fault zone, which is consistent with the saw-tooth shape of Bouguer gravity anomalies. According to the temperature model, the temperature of the Moho is approximately 600°C (Figure 6c, Sun et al., 2013), which is lower than the melting temperature of water-bearing granite (Puziewicz & Johannes, 1990). Thus, the conductor C3 has been interpreted as interconnected saline fluids, possibly derived from either the partial melting of the uppermost mantle following the dehydration of the subducting Paleo-Pacific slab (Cheng et al., 2021;P. Zhou, Xia, et al., 2020) or the metamorphism-induced dehydration of sandy argillaceous rocks in the lower crust (Connolly & Thompson, 1989;Schorn, 2018;Thiel et al., 2016). Preexisting faults act as fluid pathways, and both the reaction of hydrothermal fluids with rock and the extension of the crust caused by upwelling of the mantle trigger dynamic, permeable networks. Similar low-resistivity features related to fluids have been revealed in other large-scale strike-slip shear zones (Becken et al., 2011;Karaş et al., 2020;Meqbel et al., 2016;Unsworth & Bedrosian, 2004).
This high resistivity is attributed mainly to dry rock (Jones, 1999;Selway, 2014). Combined with the volcanic rocks generated by partial melting of an old mantle wedge source metasomatized by slab-derived fluids , and the relatively low crust Vp/Vs ratios influenced by the distribution of large-scale Mesozoic granites , we speculate that the crust of the Cathaysia Block is composed mainly of felsic rocks and has undergone vertical thinning, which has been accompanied by the eclogitization and delamination of the lower crust.

Conductors in the Upper Mantle
The most conspicuous feature in the final resistivity model is the presence of large-scale conductors (resistivities <30 Ωm) extending through the lower crust and into the upper mantle beneath the Wuyi-Yunkai Orogen and Cathaysia Block. One conductor (C2) was detected beneath the Wuyi-Yunkai Orogen extending almost vertically into the upper mantle. Another strong conductor (C4) slopes southeastward from the lower crust of the Wuyi-Yunkai Orogen toward the upper mantle beneath the Cathaysia Block. Possible explanations for these highly-conductive bodies in the upper mantle include saline fluids, hydrogen diffusion in olivine, graphite films, solid-phase conduction by solid phases such as sulfide minerals, and partial melt caused by high temperatures Jones, 1999;Selway, 2014). Among these mechanisms, it would be difficult to achieve such high electrical conductivity of the upper mantle under the diffusion of hydrogen in olivine (Novella et al., 2017). Moreover, saline fluids can be discounted because such fluids would have difficulty forming an interconnected phase at the pressures and temperatures of the upper mantle (1.5 GPa, 800-1200°C) (Yardley & Valley, 1997), and this is known because the wetting angle of fluid in olivine is uniformly greater than 60° regardless of the salinity of the fluid (Watson & Brenan, 1987). In addition, sulfides do not easily exist as a stable phase in the upper mantle because the oxygen fugacity of the mantle is relatively high; instead, most sulfides form highly volatile H 2 S (Frost & McCammon, 2008). Likewise, graphite films on grain boundaries do not exist stably at high temperatures, higher interfacial energy and larger dihedral angle between graphite and olivine (Yoshino & Noritake, 2011;Zhang & Yoshino, 2017), so graphite films cannot explain the upper-mantle conductor based on the temperature range of 800-1200°C at depths of 40-80 km (Sun et al., 2013) and a heat flow range of 75-100 mW/m 2 (Jiang et al., 2019). Therefore, the low-resistivity anomalies of 10-30 Ωm in the depth range of 40-60 km are attributed mainly to partial melting in the upper mantle beneath the Wuyi-Yunkai Orogen and Cathaysia Block.
Partial melting of the upper mantle beneath South China is expected for several reasons. Geochemical and Re-Os isotope analyses of mantle xenoliths reveal that the juvenile subcontinental lithosphere beneath South China is derived from low degrees of partial melting and asthenospheric upwelling (Liu et al., 2012). Furthermore, geochemical data of volcanic rocks and ophiolitic fragments indicate that they were generated by the partial melting of an old mantle wedge metasomatized by fluids from an ancient subducted slab Liu et al., 2018). According to the 3-D temperature model, regional heat flow data and crustal heat production (Jiang et al., 2019;Sun et al., 2013), the temperature of the upper mantle exceeds the melting temperature of water-bearing olivine (Katz et al., 2003). The upper-mantle seismic velocity structure beneath the Cathaysia Block determined by teleseismic Rayleigh wave tomography is 4.3-4.5 km/s depending on the model, which is approximately 6% lower than that of the surrounding upper mantle (Shan et al., 2017); partial melting of the mantle with melting fractions as low as 0.2% appears to be a viable explanation for this anomaly (Chantel et al., 2016). Low-resistivity anomalies in the upper mantle have also been found elsewhere beneath the Cathaysia Block Xu et al., 2019;Yin et al., 2021) and are attributed mainly to partial melting triggered by the metasomatism and dehydration of a subducted oceanic slab.
Nevertheless, it is difficult to accurately estimate the fraction of partial melt with the bulk resistivity because this percentage is closely related to temperature and the model used to for mixed-phase (partial melt and solid) rocks. Under the pressures and temperatures of the upper mantle, the molten phase may significantly contribute to the bulk conductivity, so the electrical conductivity of mantle melts must be accurately calculated. Subduction-related dacite derived from upper-mantle material is widely distributed throughout the Cathaysia Block (Li et al., 2015;Zhao et al., 2021); accordingly, the conductivity of the molten phase of the strong conductor in the upper mantle was estimated using the laboratory conductivity model of water-rich dacitic melts affected by subduction processes (Laumonier et al., 2015). The assumption of 8% water content in the dacitic melt beneath the Cathaysia Block is reasonable based on the water contents of alkali basalts and mantle xenoliths in South China and other juvenile mantle regions (Hao et al., 2014;Sisson & Layne, 1993). Within the temperature range of 800-1000°C at depths of 40-60 km, the range of estimated melt conductivities is 3-6 S/m, as shown in Figure 7a. To constrain the fraction of partial melt in the upper mantle, we employed a modified version of Archie's model to evaluate the bulk conductivity of the solid-melt system while assuming that the pockets of melt are moderately interconnected (Glover et al., 2000). The resistivity of solid rocks in the upper mantle was computed to be approximately 1000 Ωm based on the final electrical model, while the bulk resistivity of the conductor was estimated to be 10 Ωm based on the average value of the modeled feature. The estimated partial melt fraction ranges from 4% to 7% at depths of 40-60 km, as shown in Figure 7b, which is consistent with the degrees of partial melting originating from mantle xenoliths beneath the South China Block (Liu et al., 2012). Upper-mantle conductors have also been reported beneath western North America as a result of low-angle flat-slab subduction; these conductors are attributed to partial melt (McGary et al., 2014;Park et al., 1996;Soyer & Unsworth, 2006). Partial melting is thus a reasonable interpretation for the strong conductors in the upper mantle, while the low crustal conductivities are likely due to hydrothermal fluids released by the interaction between magma and the lower crust.

Insights Into Lithospheric Thinning
The lithosphere beneath the South China Block has been significantly thinned, and the amount of vertical thinning has reached more than 100 km since the Mesozoic (Shan et al., 2014). Mantle xenoliths from this region are composed predominantly of both fertile and moderately depleted peridotites, which may indicate that the Archean-Proterozoic lithospheric mantle was replaced by hotter, younger mantle material related to asthenospheric upwelling during the Mesozoic, which led to extensive lithospheric extension and abundant magmatic activity (Liu et al., 2012;Xu et al., 2002). The final resistivity model indicates that partial melting is present in the upper mantle beneath the South China Block. However, the relationship between the mechanism of lithospheric thinning and the partial melting of the upper mantle remains unclear. Flat-slab subduction was proposed based on fold-and-thrust belts and postorogenic Basin and Range-style magmatic activity (Li & Li, 2007). Lithospheric thinning and partial melting of the upper mantle can both be explained by slab-derived fluid metasomatism and the subsequent upwelling of asthenospheric mantle after slab break-off. We propose a model in which the South China Block has experienced intracontinental extensional deformation and lithospheric thinning (Figure 8). The crust and lithospheric mantle beneath the Wuyi-Yunkai Orogen were thickened during the intracontinental Orogen during the early Paleozoic. Then, fluid released by the subducting Paleo-Pacific slab was introduced into the thickened overlying lithosphere, and the increasing fluid content reduced the viscosity of the lithosphere, weakening the root. Consequently, the lithosphere became gravitationally unstable, which led to delamination, promoting asthenospheric upwelling and thermal convection (Calvert & Doublier, 2018;Calvert et al., 2021;Hill et al., 2021). Furthermore, the South China Block was subjected to large-scale extension caused by the eastward retreat of the subducting Paleo-Pacific slab, and the crust and lithospheric mantle were stretched. The thermal erosion generated by the ascending mantle accelerated the rate at which the lithosphere was thinned and produced numerous faults in addition to triggering partial melting of the upper mantle during the Late Cretaceous. At present, atmospheric precipitation infiltrates into the subsurface and is circulated at depth along fault planes, forming a series of hot springs Luo et al., 2022).

Thermal Structure Implications
The hot springs in the coastal areas of the Cathaysia Block are exposed mainly along faults, and the thermal reservoir temperature of the hot springs was calculated to be 120-150°C using chemical geothermometers (Mao et al., 2021). The intersection of a major fault with a secondary fault forms a pathway for the migration and storage of geothermal fluids, and the groundwater from atmospheric precipitation forms geothermal fluids that are circulated deep underground to a depth of approximately 4 km (Mao et al., 2021). The heat flow (mean of 83.4 mW/m 2 ) in the Cathaysia Block is distinctly higher than that elsewhere in mainland China (Jiang et al., 2019), which is consistent with the features of granitic rocks containing large amounts of radiogenic heat-producing elements and with the number of hot springs in the Cathaysia Block (Z.-M. Zhou, Ma, et al., 2020). Hot mantle material upwells along the WSSZ, bringing heat from depth directly to the middle and lower crust; the resulting superposition of hot mantle heat conduction and radiogenic heat production in the crust constitutes the main heat source of the geothermal field. The average rate of radioactive heat production of the late Mesozoic granitoids in the coastal areas of the Cathaysia Block is approximately 4 μW/m 3 . According to the final electrical resistivity structure, the thickness of the granitic rock mass beneath the geothermal system can reach 10 km, and the heat flow generated by radioactive elements in the upper crust is estimated to be approximately 40 mW/m 2 (Jaupart et al., 2016), while the average heat flow is 83.4 mW/m 2 in the geothermal system. The heat flow from the mantle is 43.4 mW/m 2 , accounting for more than half of the total heat flux, which is consistent with the heat flow quantified from the geochemical data of hydrothermal volatiles within the geothermal system (Tian et al., 2021).

Conclusion
In this paper, we present an electrical resistivity model for the crust and upper mantle beneath the Wuyi-Yunkai Orogen generated from the 3-D inversion of a single MT profile. Our model indicates the presence of significant conductors in the crust and upper mantle associated with interconnected saline fluids and partial melting, respectively, which implies that the lithosphere has been significantly thinned as a result of thermal erosion due to mantle upwelling. Sensitivity tests confirm that these conductors, which extend from the surface to a depth of 60 km beneath the Wuyi-Yunkai Orogen, are reliable model features for fitting the observed MT data. The model presented in this study is combined with the available geochemical data, revealing that the conductors in the lower crust are unlikely to be caused by crustal melt due to the rapid ascension of magma through the crust and limited crustal contamination. Instead, these lower-crustal conductors are interpreted as interconnected bodies of saline fluids derived from either the dehydration of the subducting Paleo-Pacific slab, which caused the mantle to partially melt, or the dehydration of sandy argillaceous rocks during regional metamorphism in the lower crust. Based on the heat flow values and mantle xenoliths, the enhanced conductivity anomalies in the upper mantle are attributed to partial melting. The partial melt fraction is estimated to range within 4%-7% according to a modified version of Archie's model, which is consistent with the results from mantle xenoliths. The partial melting of the upper mantle suggests that the destruction and thinning of the lithosphere of the South China Block was caused by thermal erosion of the lithospheric mantle, and the high mantle heat flow values provide both conduction-based heat and volatiles to shallow geothermal systems.

Data Availability Statement
The work was carried out at the National Supercomputer Center in Tianjin, and the calculations were performed on TianHe-1(A). For the purpose of review, the MT data used in this study are hosted and made freely available in the Figshare repository (https://doi.org/10.6084/m9.figshare.19128116.v2). ing the 3-D MT electromagnetic inversion module ModEM. We would like to thank Dr. Yujun Sun and Dr. Haijiang Zhang for providing the temperature model and shear-wave velocity model beneath the South China Block, respectively. We are also grateful to Editors Prof. Joshua Feinberg and two reviewers (Dr. Darcy Cordell and another anonymous reviewer) for their constructive comments and insightful suggestions that have greatly improve the quality of the manuscripts. Special thanks to Junfeng Guo and Libo Wang for their support in collecting the field data. The figures in this manuscript were generated by Generic Mapping Tools.