Supplemental Material: Permian plume-strengthened Tarim lithosphere controls the Cenozoic deformation pattern of the Himalayan-Tibetan orogen

Geochemical compilation and details for magnetic data processing, plume-head restoration, and rheological profiles.


INTRODUCTION
The interior of the Tarim Basin, northwestern China, has remained undeformed during the development of the Cenozoic Himalayan-Tibetan orogen, expressed by flat-lying Cenozoic strata (e.g., Jia, 1997). Although its rigidity has been attributed to the presence of Precambrian basement (e.g., Neil and Houseman, 1997), this is inconsistent with the observation of an early Paleozoic thrust belt developed across the Tarim Basin (e.g., Yin and Nie, 1996;Jia, 1997;Carroll et al., 2001), which requires the acquisition of greater strength to resist Cenozoic deformation between the early Paleozoic and end of the Mesozoic. Alternatively, the rigidity of the Tarim lithosphere could have developed via the presence of a Precambrian oceanic plateau that was strengthened from below by a Permian plume (e.g., Deng et al., 2017). This model implies that Tarim was already strong prior to the Phanerozoic, inconsistent with aforementioned Paleozoic shortening across the basin. Because Early Permian (ca. 300-270 Ma) mafic magmatism, interpreted to have been derived from a plume (Yang et al., 1996;Chen et al., 2006;Xu et al., 2014), was the only regionally extensive event affecting the Tarim Basin after the early Paleozoic (e.g., Jia, 1997;Carroll et al., 2001;Guo et al., 2005), it is possible that plume-induced processes profoundly strengthened the earlier Tarim mantle lithosphere. Assessing this possibility requires a sound knowledge of the root zone of the plume, which is obscured by ∼10 km of overlying Mesozoic-Cenozoic strata (Jia, 1997;Guo et al., 2005).
Aeromagnetic anomalies have proven to be effective in locating and quantifying the geometry and extent of mafic intrusions associated with a mantle plume (e.g., Finn and Morgan, 2002). The success of this approach inspired us to reexamine the aeromagnetic data across Tarim. The extent of Permian intrusive rocks beneath the Tarim cover sequence has never been systematically determined before with aeromagnetic data sets (e.g., Xiong et al., 2016). Our reprocessed data reveal a 300-400-kmdiameter radial pattern of magnetic lineaments that emanate from a central region characterized by mixed positive and negative anomalies. Constrained by the existing geochemical and geochronological data from the plumerelated igneous rocks, we argue that Tarim's mantle lithosphere was thickened, depleted, and dehydrated by the Permian plume to transform Tarim into a rigid craton-like continent in situ through the development of a thick mantle keel and input of mafic intrusions into the crust (e.g., Lee et al., 2011). Tarim's stronger lithosphere deflected strain from Cenozoic India-Asia convergence around the Tarim Basin to shape the deformation pattern of the Himalayan-Tibetan orogen.

GEOLOGICAL SETTING
The Precambrian basement of the Tarim Basin consists of continental blocks that amalgamated along the Neoproterozoic central Tarim suture zone (Guo et al., 2005;Yang et al., 2018). The margins of the Tarim Basin were affected episodically by arc development and arc-continent collision throughout the Paleozoic during the development and closure of the Paleo-Asian oceans to the north (i.e., present southern Tian Shan; e.g., Yin and Nie, 1996;Xiao et al., 2005), and the Tethyan oceans to the south (i.e., present Tibetan Plateau; e.g., Yin and Harrison, 2000;Wu et al., 2016). The subduction polarity of the Carboniferous-Permian arc systems is enigmatic due to poor exposure and Cenozoic modification, but based on correlative subduction systems to the east (e.g., Xiao et al., 2003;Wu et al., 2016), there is a strong possibility that both subduction systems dipped beneath the Tarim continent. *E-mail: hlchen@zju.edu.cn The Tarim Basin's interior was affected by two regional events in the Phanerozoic. The first occurred in the early Paleozoic, expressed by reactivation of the Neoproterozoic suture zone as evidenced by thrusting and folding ( Fig. 1; Jia, 1997;Carroll et al., 2001;Guo et al., 2005), which was coeval with the closure of the Qilian and Kunlun oceans in northern Tibet . The second event occurred during the Early Permian (300-270 Ma), expressed as widespread igneous activity, defining a large igneous province (LIP), possibly induced by a mantle plume from below (Chen et al., 2006;Xu et al., 2014). The full extent of the LIP is obscured by a >∼10 km section of sedimentary cover, but its wide distribution is indicated by surface exposures, subsurface samples, and extrapolation of these findings by analysis and interpretation of available seismic reflection profiles ( Fig. 2A; e.g., Xu et al., 2014). The Permian plume-head location has been variably assigned to the Tian Shan and southwestern or northern Tarim (Chen et al., 2006;Li et al., 2014;Liu et al., 2016;Cheng et al., 2018). Subsequent Cenozoic India-Asia convergence resulted in concentrated shortening along the margins of Tarim Basin, including thin-skinned thrust belts in the north and west and strike-slip faulting in the southeast ( Fig. 1; Yin et al., 1998;Guo et al., 2005).

PERMIAN PLUME HEAD
Data used in this study are mainly from the aeromagnetic data set of China (Xiong et al., 2016) with a 1 km × 1 km grid spacing and 1 km datum, supplemented by the Earth Magnetic Anomaly Grid Version 2 (EMAG2; Maus et al., 2009; see also Fig. S1A in the Supplemental Material 1 ). The combined data set was first processed by reduction to pole ( Fig. S2; e.g., Arkani-Hamed, 1988) and next by analytical signal transformation ( Fig. S3; see details in the Supplemental Material; Roest et al., 1992). A series of high-amplitude, long-wavelength anomalies are superimposed by four variably oriented, short-wavelength, positive linear anomalies across the northwestern Tarim Basin (e.g., the Bachu Uplift; i.e., L1-L4 in Fig. S2). Highamplitude, long-wavelength magnetic anomalies have previously been correlated to signals derived from the Tarim Precambrian crystalline basement (e.g., Xiong et al., 2016;Yang et al., 2018). The final reprocessed result ( Fig. 2A) reveals a radial pattern of linear anomalies: L1 lineaments trend southeast, L2 lineaments trend east and southeast, L3 lineaments trend east, and L4 lineaments trend northeast ( Fig. 2A; Fig. S3), converging near Bachu City (∼39.8°N, ∼78.6°E).
The radial linear anomalies overlap the Permian volcanic field ( Fig. 2A). They converge beneath the observed ca. 300 Ma kimberlite complex, which are the oldest Permian intrusive rocks in Tarim Basin that are proposed to be associated with, and derived from, the plume head ( Fig. 2A; e.g., Chen et al., 2006;Xu et al., 2014). This spatial correlation led us to interpret the linear magnetic anomalies to represent a giant dike swarm with arms radiating ∼150-200 km outward from the center interpreted as the Permian plume head. Additionally, isolated magnetic highs dispersed across Tarim Basin may have been related to plume-induced mafic intrusions ( Fig. 2A).
To assess whether the linear anomalies are related to Permian intrusions, we compiled 25,643 detailed rock-magnetic susceptibility measurements from within and around the Tarim Basin (Table S1). The measured susceptibilities concentrated in the Permian igneous suites range from ∼27 × 10 −5 to ∼11,700 × 10 −5 SI, with high mean values (>5300 × 10 −5 SI; Fig. S1B). The high values of Permian intrusions show that the radial pattern of the positive linear anomalies can indeed be explained by Permian mafic intrusions, corroborated by surface exposures and drill core (Fig. S4).
Although radially trending dikes should be distributed symmetrically around a plume center (Ernst and Buchan, 2003), our observed radial magnetic lineaments appear to occur only on the east side of the interpreted plume head. We note that the interpreted plume head is close to the southern Cenozoic Tian Shan thrust belt (Figs. 2B and 2C), which could have obscured the projected radial dikes northwest of the inferred plume head. To test this possibility, we schematically restored 20%-30% crustal shortening across the southern Tian Shan thrust belt and 10% shortening across the Tarim Basin ( Fig.  S5; e.g., Yin et al., 1998). This simple exercise yielded a magnetic anomaly pattern consistent with a radial dike swarm around the interpreted central plume head (Figs. 2C and 2D).
The impact and extent of the interpreted Permian plume have been constrained by the Carboniferous-Permian unconformity in the Tarim Basin (Figs. 3A and 3B;Chen et al., 2006;Li et al., 2014;Xu et al., 2014). The extent of this unconformity overlaps our interpreted region affected by the plume. Because the aeromagnetic 1 Supplemental Material. Geochemical compilation and details for magnetic data processing, plume-head restoration, and rheological profiles. Please visit https://doi .org/10.1130/GEOL.S.12869684 to access the supplemental material, and contact editing@ geosociety.org with any questions.

Figure 1. Topographic and tectonic map of the Tarim Basin (northwestern China). Early Paleozoic thrusts are after Jia (1997) and Yin and Nie (1996).
image of the plume is the integrated effect of intrusions in the Tarim crust, whereas the unconformity occurred over the life span of plume activity, this spatial coincidence of the two patterns suggests a relatively simple plume structure expressed by a center with a radial dike swarm that was relatively stationary, at least within the resolution of the two data sets. Within this resolution, a minor shift of the plume head is permissible ( Fig. 2; Li et al., 2014), possibly facilitated by the plume exploring a preexisting weakness in the Tarim lithosphere . The 300-400 km plume-head diameter observed in the middle-upper crust (above the Curie surface; Fig. 2) is consistent with a plume that flattened with a >500 km diameter at the base of the lithosphere (Ernst and Buchan, 2003).
Carboniferous-Triassic inward-dipping subduction along Tarim's northern and southern margins (e.g., Xiao et al., 2005;Wu et al., 2016) was contemporaneous with plume activity (Fig. 3). The temporal and spatial overlap of subduction and the plume suggests a complex causal relationship, possibly involving flow through or around slab windows/tears, as has been hypothesized for Yellowstone and the backarc of the Patagonian Andes (e.g., Gorring et al., 1997;Kincaid et al., 2013). The exact connection between the plume and these subduction systems remains to be explored.

LITHOSPHERE STRENGTHENING BY PLUME ACTIVITY
Plume-lithosphere interactions may either destroy strong cratons (e.g., Hu et al., 2018) or strengthen the mantle lithosphere by developing a craton-like mantle keel (Lee et al., 2001(Lee et al., , 2011. The low heat flow and resistance to Cenozoic deformation in the Tarim Basin may be explained by plume-induced lithospheric strengthening. Here, we argue that the Permian plume transformed Tarim into a stable craton-like continent in situ (Lee et al., 2011) by simultaneously building a strong mantle keel beneath Tarim and adding voluminous strong mafic rocks to the lower crust.
As outlined above, the Tarim lithosphere was not strong prior to Permian plume impingement. Permian volcanic rocks were initially derived from volatile-rich metasomatized mantle lithosphere (Zhang et al., 2013;Cheng et al., 2018), which may reflect Tarim's pre-Mesozoic subduction history (e.g., Guo et al., 2005). However, the subtle arc-subduction signature in the plume-derived melts diminished in both time, after initial ca. 300 Ma impingement, and space, moving toward the plume center (see Fig. S6  for geochemical compilation). For example, Nb depletion tracked inversely by Nb/Yb ratios, commonly associated with subduction-zone influence, diminishes moving toward the plume center (Nb/Yb = 5-10 to ∼20-40; Fig. S6). We interpret this trend to reflect more direct plumederived mantle melts above the plume-head center, as opposed to mixed-source melts. At a given regional location, Nb/Yb increases through time from ca. 300 (Nb/Yb ∼7) to ca. 270 Ma (Nb/Yb ∼20-40; Fig. S6), which may reflect progressive dehydration/modification of the mantle lithosphere that removed the subduction-influenced signature to yield more classic ocean-island basalt (OIB) geochemistry. Hot plume activity (zircon saturation temperatures >800 °C; Xu et al., 2014) affected the Tarim Basin for ∼30 m.y. (Fig. 3), and the high heat flux would have led to high-degree partial melting, devolatilization, and thickening of the mantle lithosphere ( Fig. 3C; Fig. S7). Specifically, chemical-melt depletion (e.g., Fe relative to Mg) lowered mantle density and facilitated growth of the mantle lithosphere into thickened buoyant mantle (Lee et al., 2001). Water is highly incompatible during melting, and plume magmatism dehydrated the mantle lithosphere, generating hydrated flood basalts (5 wt% H 2 O; Selway et al., 2014;Liu et al., 2017). This process would have greatly strengthened the mantle lithosphere, given the inverse dependence of mantle viscosity on water content (Hirth and Kohlstedt, 1996).
Dehydration and chemical depletion together intrinsically strengthened the mantle keel. Injected and underplated mafic intrusions in the lower crust further increased crustal strength (Liu and Furlong, 1994). The flattened plume head may have been >500 km in diameter (Ernst and Buchan, 2003), and isolated signal highs across the basin ( Fig. 2A) suggest a greater area was affected by the plume. With minor plume migration, as discussed above, most of the Tarim Basin may have experienced the thermal pulse associated with this plume.
These integrated contributions built Tarim into a rigid craton-like continent in the Permian (Fig. 3C). Prior to the Cenozoic, much of Tibet and the Pamirs involved arc subduction and collisional systems between various terranes (Yin and Harrison, 2000;Wu et al., 2016) that led to broad-scale hydrologic weakening of most of the region. This explains why deformation was distributed thousands of kilometers inboard of the collisional front shortly after collision (Chen and Gerya, 2016). The stronger Tarim craton-like lithosphere deflected strain from Cenozoic India-Asia convergence around the Tarim Basin, including Pamir overthrusting to the northwest and Altyn Tagh left-slip displacement to the northeast.
The Sichuan Basin also behaved rigidly within the Pamir-Tibet orogen, and the region experienced hydrated LIP volcanism (2-4 wt% H 2 O; Liu et al., 2017) associated with a mantle plume at ca. 260 Ma that may have dehydrated, melt-depleted, and strengthened Sichuan's mantle. Voluminous mafic intrusions and underplating would have simultaneously strengthened the lower/middle crust. Accordingly, rheological conditioning by Permian mantle plumes may have exerted a first-order control on strain partitioning between deformed Tibetan lithosphere and the stronger Tarim and Sichuan Basins, which acted as conjugate rigid buttresses, thus shaping the geometry of the Tibetan Plateau (Fig. 4).