PUBLICATIONS. Tectonics

Transcription

1 PUBLICATIONS Tectonics RESEARCH ARTICLE Key Points: Two late Cretaceous paleopoles are reported from the western Lhasa Terrane The southern margin of Asia near west syntax was located at ~14 N A quasi-linear southern margin of Eurasia prior to the collision is defined Supporting Information: Table S1 and Figure S1 Correspondence to: Z. Yi, yizhiyu09@gmail.com Citation: Yi, Z., B. Huang, L. Yang, X. Tang, Y. Yan, Q. Qiao, J. Zhao, and L. Chen (2015), A quasi-linear structure of the southern margin of Eurasia prior to the India-Asia collision: First paleomagnetic constraints from Upper Cretaceous volcanic rocks near the western syntaxis of Tibet, Tectonics, 34, , doi: / 2014TC Received 2 MAR 2014 Accepted 26 JUN 2015 Accepted article online 29 JUN 2015 Published online 21 JUL 2015 A quasi-linear structure of the southern margin of Eurasia prior to the India-Asia collision: First paleomagnetic constraints from Upper Cretaceous volcanic rocks near the western syntaxis of Tibet Zhiyu Yi 1,2, Baochun Huang 2,3, Liekun Yang 2, Xiangde Tang 2, Yonggang Yan 2, Qingqing Qiao 2,4, Jie Zhao 2, and Liwei Chen 2 1 State Key Laboratory for Continental Tectonics and Dynamics, Institute of Geology, Chinese Academy of Geological Sciences, Beijing, China, 2 State Key Laboratory of Lithospheric Evolution, Institute of Geology and Geophysics of the Chinese Academy of Sciences, Beijing, China, 3 Key Laboratory of Orogenic Belt and Crust Evolution, Ministry of Education, School of Earth and Space Sciences, Peking University, Beijing, China, 4 Xinjiang Research Center for Mineral Resources, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi, China Abstract We report the first combined geochronologic and paleomagnetic study of volcanic rocks from the Shiquanhe and Yare Basins at the westernmost Lhasa Terrane, which aims to provide an accurate constraint on the shape and paleoposition of the southern margin of Asia prior to the India-Asia collision. Three new 40 Ar/ 39 Ar ages of 92.5 ± 2.9 Ma, 92.4 ± 0.9 Ma, and 79.6 ± 0.7 Ma determined by fresh matrix or feldspar from lava flows suggest a Late Cretaceous age for the investigated units. Characteristic remanent magnetizations have been successfully isolated from 38 sites which pass positive fold and/or reversal, conglomerate tests and are hence interpreted as primary in origin. The two paleopoles obtained from Yare and Shiquanhe yield consistent paleolatitudes of 13.6 N ± 9.6 N and 14.2 N ± 2.7 N, respectively (for a reference site of 31.5 N, 80 E), indicating that the southern margin of Asia near the western syntaxis was located far south during the Late Cretaceous time. A reconstruction of the Lhasa Terrane in the frame of Eurasia with paleomagnetic data obtained from its western and eastern parts indicates that the southern margin of Eurasia probably had a quasi-linear orientation prior to the collision formerly trending approximately 315 E. This is compatible with the shape of the Neo-Tethys slab observed from seismic tomographic studies. Our findings provide a solid basis for evaluating Cenozoic crustal shortening in the Asian interior and the size of Greater India near the western syntaxis American Geophysical Union. All Rights Reserved. 1. Introduction The large-scale lithospheric deformation resulting from the India-Asia collision has created the Himalayan Orogen and Tibetan Plateau [Tapponnier et al., 1982; Chang et al., 1986; Yin and Harrison, 2000] with profound geodynamic and environmental consequences [Raymo and Ruddiman, 1992; Molnar et al., 1993; Royden et al., 1997]. Restoration of the precollisional shape of continental India and Asia is closely related to a series of important scientific questions. These include the timing and position of the India-Asia collision and the amount and pattern of postcollisional crustal shortening [e.g., Dupont-Nivet et al., 2010a; van Hinsbergen et al., 2011; Yi et al., 2011]. The Lhasa Terrane is critically positioned at the southern margin of Asia, and its paleoposition is important for reconstructing the southern margin of Asia prior to the collision [Achache et al., 1984; Dupont-Nivet et al., 2010a; Liebke et al., 2010; Chen et al., 2010, 2012, 2014; Sun et al., 2010, 2012; Lippert et al., 2014]. The shapes of Asia and India prior to collision can potentially be restored from paleomagnetic study of rocks on either side of the suture zone in blocks comprising the Tethyan Himalaya and Lhasa Terranes [e.g., Dupont-Nivet et al., 2010a; Yi et al., 2011]. From the Asian side, paleolatitude of the Lhasa Terrane is constrained largely by paleomagnetic studies of the Paleogene Linzizong Group (including the Dianzhong, Nianbo, and Pana Formations) and underlying Upper Cretaceous red beds (the Shexing Formation) from the eastern sector. Based on an inadequate data set, pioneering paleomagnetic investigations during the last two decades of the last century proposed an approximate paleolatitude of 10 N [Pozzi et al., 1982; Westphal and Pozzi, 1983; Achache et al., 1984; Achache, 1991].However,onthe YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1431

2 basis of a refined geochronologic and stratigraphic framework for these source rock units [Zhou et al., 2004; Dong et al., 2005; He et al., 2007; Lee et al., 2009], a new suite of paleomagnetic studies during the past 5 years have revised this conclusion [Dupont-Nivet et al., 2010a; Liebke et al., 2010; Tan et al., 2010; Sun et al., 2010, 2012; Huang et al., 2013; Chen et al., 2010, 2014] and derive significantly different paleolatitudes in the range of ~6 to 32 N. Studies of 20 sites from the lower part of the Linzizong Group, the Dianzhong Formation (~64 50 Ma), yield a paleolatitude of 6.6 N ± 6.3 N [Chen et al., 2010, 2014], while the paleolatitude resolved from the Nianbo Formation (~60 50 Ma) is ~10 N to 15 N [Liebke et al., 2010; Sun et al., 2010; Chen et al., 2014] which is comparable to estimates from the earlier studies. However, based on the studies of upper part of the Linzizong Group (the Pana Formation, ~50 44 Ma) and the Shexing Formation (~ Ma), some authors have argued that the Lhasa Terrane was located at ~20 to 24 N in Upper Cretaceous times [Dupont-Nivet et al., 2010a; van Hinsbergen et al., 2012; Huang et al., 2013; Lippertetal., 2014], and this opinion seems to be supported by paleomagnetic data from fore-arc sediments directly north of the Yarlung-Zambo suture [Meng et al., 2012]. The discrepancy in paleomagnetic data from the Lhasa Terrane is likely due to inclination shallowing typically found in sedimentary rocks [Huang et al., 2013], although it could also be due to inadequate sampling of the Earth s paleosecular variation (PSV), a feature more common in the study of volcanic rocks [Liebke et al., 2010; van Hinsbergen et al., 2012; Lippert et al., 2014]. A further difficulty with the paleomagnetic data set from the Lhasa Terrane is that most of the data come from the eastern sector and near the Linzhou Basin. In comparison, the length of the Indus-Yarlung Zambo suture is up to 2500 km, and most parts of the southern margin of Asia have been involved in large scale of deformation [Yin and Harrison, 2000; Johnson, 2002; van Hinsbergen et al., 2011]. Hence, more paleomagnetic data, with wider spatial and temporal distribution and stricter selection of the available results, are essential for reconstructing the southern margin of Asia prior to, and during, the India-Asia collision. This is key to determining the timing and position of the initial collision as well as postcollisional crustal shortening in Asia. In this study, we have performed a combined geochronologic and paleomagnetic study on the Upper Cretaceous volcanic rocks and intercalated sediments collected from the Yare and Shiquanhe Basins in the western part of the Lhasa Terrane. We thereby aim to provide accurate paleomagnetic constraints on the paleogeography of the southern leading edge of the Asian continent during the Late Cretaceous. We also provide a review of the available paleomagnetic results from the Cretaceous to early Cenozoic interval from both eastern and western segments of the Lhasa Terrane and discuss the precollisional shape and position of the southern margin of Asia and implications for the India-Asia collision. 2. Geological Setting and Paleomagnetic Sampling The Neo-Tethys suture, separating India-Africa from Eurasia, trends northwest to southeast from the Mediterranean to Southeast Asia with a length exceeding 15,000 km. The India-Asia suture, which constitutes the eastern part of the Neo-Tethys suture, protrudes northward significantly and forms two tectonic syntaxises near Pamir and Namche Barhwa, in the west and east, respectively. The structure of India-Asia collision zone is illustrated in Figure 1a, in which well-preserved trench-arc basin and foreland fold-thrust systems can be identified from north to south; these comprise the Gangdise Arc (granite belt), the Shigatse fore-arc basin (flysch belt), the Indus-Yarlung Zambo suture (ophiolite belt), and the Himalayas (foreland fold-thrust belt). In the southern part of the Lhasa Terrane, a volcanic succession, named the Linzizong Group, developed along the north of the Gangdise and has been considered to record subduction of the Neo-Tethys and collision of India and Asia [Ding et al., 2003; Mo et al., 2003, 2008] (Figure 1a). In accordance with spatial distribution and rock ages, the Linzizong Group has been classified as a lower Cenozoic volcanic division with intercalated sediments distributed along the southern margin of the Lhasa Terrane with ages ranging from 69 to 43 Ma [He et al., 2007; Lee et al., 2009; Chen et al., 2014]. It is subdivided into three members, namely, the Dianzhong, Nianbo, and Pana Formations [Team of Regional Geological Survey of the Bureau of Geology and Mineral Resources of Tibet Autonomous Region, 1991]. The latter division also includes volcanic rocks which crop out to the north of Cuoqin area [Yin and Harrison, 2000] (Figure 1a) and have distinctly different ages of approximately Ma [Coulon et al., 1986; Lee et al., 2009]. Although the designations of Dianzhong, Nianbo, and Pana are also used for stratigraphic divisions in this part of the Linzizong Group on the newly completed 1:250,000 regional geologic maps, they are unlikely to be related to the three members of the Linzizong Group in narrow YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1432

3 Figure 1. (a) Schematic map of the Lhasa Terrane and adjacent area (after Gansser [1964]) showing main tectonic units and distribution of the Linzizong Group in this area. Simplified regional geologic map of the (b) Yare and (d) Shiquanhe Basins showing the Mesozoic and Cenozoic stratigraphy and paleomagnetic sampling locations; lower hemisphere stereonet projection for the bedding attitudes of the sampled sites defines a fold axis plunging 4 toward 195 and 26.5 toward 235 for the strata from the (c) Yare and (e) Shiquanhe Basins, respectively. sense. In this paper the designation Upper Cretaceous volcanic rocks is used to represent this part of volcanic sequence mainly distributed within the northern and western parts of the Lhasa Terrane. To avoid confusion K 3 1,K 3 2, and K 3 3 formations are used for the further division of this unit and correspond, respectively, to the Dianzhong, Nianbo, and Pana Formations indicated on the 1:250,000 regional geologic maps, although we note that the precise age framework of these units is somewhat unclear. Our sampling area near the village of Yare is shown in Figures 1b and 2g where the sampled succession is also named the Chalicuo Group and mainly composed of volcanic rocks with intercalated sandstones (Figures 2a 2c and 2g). A total of 15 sites were collected from the K 3 2 formation with 13 sites collected from at least 10 independent lava flows and another 2 sites (xy190 and xy219) drilled from interbedded tuffaceous agglomerates (Figure 2g). The boundaries between different lava flows are easily identified by sandstone layers (Figures 2a 2c and 2g). One hand sample (xy191) was collected from the middle part of the section for 40 Ar/ 39 Ar dating (Figure 1d). Sampling near the city of Gar in the Shiquanhe Basin is shown in Figure 1d. The Upper Cretaceous volcanic rocks in this basin composed mainly of mafic lavas in the lower part and tuffs with intercalated agglomerates and lava flows in the upper part; these are readily assigned to the K 3 1 and K 3 2 formations, respectively (Figures 2d 2f). The K 3 1 formation is gently tilted and unconformably underlain by Lower Cretaceous limestones of the Jiega Formation and volcanic rocks of the Zenong Group (Figures 1d and 2d). The K 3 2 formation is unconformably overlain by the Oligocene to Miocene Rigongla Formation (Figure 1d). Ten (xa ) and 13 sites (xa ) were collected from the K 3 1 and K 3 2 formations, respectively; two hand samples xa161 and xa179 were collected for 40 Ar/ 39 Ar dating from the base of the K 3 1 and K 3 2 formations, respectively (Figure 1d). YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1433

4 Figure 2. Lithology of the sampled rocks in (a c) Yare and (d f) Shiquanhe Basins. (g) Cross section of the volcanic-sediment succession exposed in Nile, west of the Chalicuo Lake, showing the stratified units and distribution of sampling sites. For the better averaging of PSV with the aim of avoiding selective sampling of the geomagnetic field, field strategy concentrated on distinguishing different lava flow. In most situations, the observed layers of intercalated sandstones between two lavas could serve as markers of different lava flows; in other cases separation was achieved by lithological comparison. In the case of discontinuous outcrops we sampled the rocks at intervals of at least ~10 m separation. Generally, 8 10 paleomagnetic samples were collected from each site, and all were collected using a portable gasoline-powered drill and oriented in situ by both magnetic and Sun compasses. 3. The 40 Ar/ 39 Ar Analytical Methods and Results Groundmass and feldspars were selected from samples xa161, xa179, and xy191 for 40 Ar/ 39 Ar dating. They were irradiated for 24 h together with Bern4M Muscovite standards in the 49 2 reactor, Beijing, China. The reference age for the Bern4M is ± 0.06 Ma [Baksi et al., 1996; McDougal and Harrison, 1999]. Argon isotopic analyses were performed at the 40 Ar/ 39 Ar Geochronology Laboratory, Institute of Geology and Geophysics, Chinese Academy of Sciences, on a MM5400 mass spectrometer. Argon was extracted by conventional furnace incremental heating or IR laser fusion method. All the data in Tables 1 and 2 were corrected for system blanks, mass discriminations, interfering Ca, K-derived argon isotopes, and the decay of 37 Ar since the time of the irradiation. The decay constant used here is λ = (5.543 ± 0.010) a 1,as recommended by Steiger and Jäger [1977]. Details of the analysis and data processing procedures are outlined in Wang et al. [2006], Yang et al. [2008], and Yang et al. [2014]. The results of the 40 Ar/ 39 Ar experiments are plotted as age spectrum and isotope correlation diagrams in Figure 3, and essential data are summarized in Tables 1 and 2. The feldspar from basaltic lava xa161 yields an essentially concordant age spectrum with the exception of several initial steps (Figure 3a). Seventeen consecutive steps account for 84.1% of the total 39 Ar released and define a plateau age of 92.5 ± 2.9 Ma (2σ, mean square weighted deviation (MSWD) = 2.0) (Figure 3a). YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1434

5 Table 1. The 40 Ar/ 39 Ar Incremental Heating Results for Samples From the Shiquanhe Lava Flows, Western Segment of the Lhasa Terrane Temperature ( C) 40 Ar/ 39 Ar 37 Ar/ 39 Ar 36 Ar/ 39 Ar 40 Ar*/ 39 Ark 40 Ar* (%) 39 Ar(k) (%) Age ± 2σ (Ma) Xa161 Feldspar, Weight = 5.0 mg, J = ± C ± C ± C ± C ± C ± C ± C ± C ± C ± C ± C ± C ± C ± C ± C ± C ± C ± C ± C ±17.90 Xa179 Groundmass, Weight = 10.0 mg, J = ± C ± C ± C ± C ± C ± C ± C ± C ± C ± C ± C ±1.25 An inverse isochron age of 91.3 ± 7.6 Ma (2σ, MSWD = 2.1), calculated from the same steps, is in agreement with the plateau age (Figure 3b). For the 40 Ar/ 36 Ar intercept of ± 5.8 (2σ) is no different with the air ratio (Figure 3b); the plateau age is hence interpretable. The plateau and inverse isochron ages are slightly younger than the total fusion age of 94.0 ± 2.4 Ma. Table 2. The 40 Ar/ 39 Ar Single Grain Total Fusion Results for Sample From the Yare Lava Flows, Western Segment of the Lhasa Terrane Laboratory ID 40 Ar/ 39 Ar 37 Ar/ 39 Ar 36 Ar/ 39 Ar 40 Ar*/ 39 Ark 40 Ar* (%) 39 Ar(k) (%) Age ± 2σ (Ma) Xy191 Feldspar, 16 grains, J = ± D-114D ± D-114F ± D-114G ± D-114H ± D-114A ± D-114B ± D-114C ± D-114D ± D-114E ± D-114F ± D-114G ± D-114H ± D-114I ± D-114J ± D-114K ± D-114L ±2.30 YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1435

6 Figure 3. Apparent age spectrum and isochron diagrams from 40 Ar/ 39 Ar analysis of samples (a and b) xa161, (c and d) xa179, and (e and f) xy191. All errors are shown at the 2σ level. The groundmass from basaltic lava xa179 also yields a concordant age spectrum with the exception of two initial steps showing slightly disturbed ages (Figure 3c). Nine consecutive steps accounting for 92.9% of the total 39 Ar released define a plateau age of 92.4 ± 0.9 Ma (2σ, MSWD = 1.2) (Figure 3c). An inverse isochron age of 93.0 ± 1.9 Ma (2σ, MSWD = 2.2) is in agreement with the plateau age (Figure 3d). The 40 Ar/ 36 Ar intercept of ± 2.6 (2σ) suggests no excess argon in this sample (Figure 3d). The plateau and inverse isochron ages are also in agreement with the total fusion age 92.7 ± 1.1 Ma. The feldspar from basaltic lava xy191 yields a concordant age spectrum (Figure 3e). All 16 analyses give a weighted mean age of 79.6 ± 0.7 Ma (2σ, MSWD = 1.1). The MSWD value is close to 1, which means that all analytical results record the time of eruption without contamination from mixed xenocrysts. An inverse isochron age of 79.2 ± 1.6 Ma (2σ, MSWD = 1.2) is in agreement with the weighted mean age. The 40 Ar/ 36 Ar intercept of ± 4.6 (2σ) is no different with the air ratio indicating that no excess argon is present in sample xy191 (Figure 3f). In this study, the sample has the initial 40 Ar/ 36 Ar ratio as air, and inverse isochron and plateau or weighted mean ages are essentially no different, but they have contrasting uncertainties. Here we prefer to use plateau and weighted mean age estimate eruption ages of the lava flows, i.e., 92.5 ± 2.9 Ma, 92.4 ± 0.9 Ma, and 79.6 ± 0.7 Ma for samples xa161, xa179, and xy191, respectively. YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1436

7 Figure 4. Representative results of hysteresis loops of the pilot samples from the (a d) Yare and (e i) Shiquanhe Basins with maximum field of 1.0 to 2.0 T. Acquisition of isothermal remanent magnetization (IRM), back-field demagnetization curves (in the second quadrant of each diagram) with maximum fields of (Figure 4e) 0.5 T, (Figures 4a, 4b, 4d, 4f, and 4g) 1.0 T, (Figures 4c and 4i) 1.5 T, and (Figure h) 2.0 T. 4. Paleomagnetic Analysis 4.1. Laboratory Methods Standard cm cylindrical specimens were cut from field core samples, and some fresh end material was selected for rock magnetic analysis. To characterize the composition and domain state of the magnetic carriers, thermomagnetic analysis (variation of susceptibility versus temperature) of six pilot samples was carried out in argon from room temperature to 700 C using an AGICO MFK1 Kappabridge equipped with a CS-3 high-temperature furnace in a field of 200 A/m at a frequency of 976 Hz. Hysteresis loops and remanence coercivity (Hcr) of 10 pilot samples were measured on a Princeton/MicroMag 3900 Vibrating Sample Magnetometer at room temperature with a field range of ±1 T to ±2 T. Anisotropy of magnetic susceptibility (AMS) of 127 specimens from 37 sites was measured using a KLY-4 Kappabridge. All specimens were subject to progressive thermal demagnetization using an ASC Model TD-48 oven with residual magnetic field minimized to less than 10 nt inside the cooling chamber. Demagnetization was performed at successive steps with intervals of 50 C or 100 C to 500 C, subsequently reduced to 30 C, YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1437

8 Figure 5. Temperature dependence of magnetic susceptibility (κ-t) for representative samples from the (b d) Yare and (a and e h) Shiquanhe Basins. The heating and cooling processes are represented by the solid and dotted lines, respectively. 20 C, or 10 C as the maximum unblocking temperatures of the remanence carriers were approached. All demagnetization treatments and remanence measurements were performed in a magnetically shielded room at the Paleomagnetism and Geochronology Laboratory of the Institute of Geology and Geophysics, Chinese Academy of Sciences. Demagnetization results were evaluated using stereographic projections and orthogonal diagrams [Zijderveld, 1967]. Most specimen directions were determined using principal component analysis [Kirschvink, 1980] to obtain line fits of characteristic remanence (ChRM) from at least four successive steps, while a few were derived from the best fit to remagnetization great circles [Halls, 1978]. Site mean directions were calculated using the well-determined ChRMs with maximum angular deviations of less than 10, and sample directions greater than two standard deviations away from the mean direction were excluded from the final mean. The site-mean ChRM directions and interval-mean results were determined using the standard Fisherian method[fisher, 1953], and spherical statistics was performed with paleomagnetic software offered by PMGSC (version 4.2) by R. Enkin Measurement Results Yare Basin In a valley northwest of the village of Yare, 158 samples from 15 sites were collected from K 3 2 formation, with 13 sites collected from lava flows and 2 sites from the matrix of intercalated agglomerates (Figures 1b, 2a 2c, and 2g). Complete saturation is reached in the basalt specimens before 0.3 tesla (T) with remanence coercivity (Hcr) less than 0.1 T, indicating that these samples are dominated by low-coercivity ferromagnetic phase (Figures 4a and 4d); thermomagnetic analysis displays a fast drop in susceptibility at ~580 C indicating that the main magnetic mineral is magnetite (Figures 5a and 5b). The ratios, Mrs/Ms = and Hcr/Hc = , calculated from the hysteresis loops (Figures 4a and 4d) further indicate a pseudosingle-domain (PSD) state for the magnetite grains [Day et al., 1977]. For the agglomerate specimens, the isothermal remanent magnetization (IRM) intensity increases quickly before 0.3 T but is not saturated up to 1.0 T to 2.0 T with Hcr up to ~0.4 T (Figure 4b); thermomagnetic analysis for pilot specimens shows significant drop of susceptibility at ~580 C and ~675 C (Figures 5c and 5d) indicating that the main magnetic carriers in the specimens probably comprise both magnetite and hematite. The AMS measurements reveal a large scatter of principle directions (Figures 6a and 6b) and a relatively weak anisotropy degree with corrected anisotropy of magnetic susceptibility (P j ) less than 1.11 (84% samples P j < 1.06) (Figure 6c). This observation indicates that the sampling zones have not suffered appreciable deformation, at least at a sample scale. In general, a lower-temperature component is isolated by ~ C (Figures 7a 7f). However, this component shows large dispersion both before and after tilt adjustment; it shows no clear link to the YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1438

9 Figure 6. Stereoplots of three principal axes of anisotropy of magnetic susceptibility (AMS) episodes and plots of AMS degree (P j ) versus AMS shape (T) for volcanic and sedimentary rocks in (a c) Yare and (d f) Shiquanhe Basins. present field direction and is likely contaminated by magnetizations acquired during storage. Following removal of this component, 138 specimens out of 158 demagnetized yield well-defined ChRMs generally determined in the temperature interval of ~300 C to ~580 C (Figures 7a 7f and Table S1 in the supporting information). All the 15 sites yield well-grouped site-mean ChRM directions with an interval-mean direction of D = 12.8, I = 39.2 (α 95 = 5.6 ) before and D = 346.6, I = 25.6 (α 95 = 3.5 ) after tilt correction (Figures 8d and 8e and Table 3). The data grouping is significantly improved by tilt correction with k s /k g = 2.58, and analysis using the Watson and Enkin s [1993] test yields an optimal concentration at 107.9% unfolding with 95% uncertainty ranging from 94.0 to 121.5% unfolding (Figure 8f and Table 3). Application of McFadden s [1990] fold test indicates a positive result at 95% confidence level with the ξ 2 statistic equal to before and after tilt correction; the critical value is at this confidence level. The positive fold test is strongly indicative of a prefold origin for the ChRM (Table 3) Shiquanhe Basin Ten sites were collected from K 3 1 formation near the G219 National Road (Figure 1d) where the sampled rocks consist mainly of basalts and basaltic andesites (Figures 2d and 2e). Another 13 sites (xa ) were collected from the K 3 2 formation in the eastern part of the Shiquanhe Basin, where the lithology is mainly agglomerates, tuffs of brick-red color, and basalts (Figure 2f). To carry out a conglomerate test, one site (xa178) was collected from a deposit comprising volcanic bombs, lava cobbles, and coarse-grained tuffs at the bottom of the K 3 2 formation. For the basalt and basaltic andesite specimens from both localities, rapid saturation of IRM intensity below 0.3 T and the observed low Hcr (generally between ~3 and T) suggest low-coercivity ferromagnets as the main magnetic carriers (Figures 4e 4g and 4i). Rapid fall in magnetic susceptibility at ~580 C and in the range of ~300 to 400 C observed by thermomagnetic analysis for these specimens further indicate that the main magnetic minerals are magnetite and Ti-rich titanomagnetites. Hysteresis data indicate a pseudosingle-domain (PSD) state for the magnetite grains [Day et al., 1977] with Mrs/Ms = and Hcr/Hc = (Figures 4e 4g and 4i). YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1439

10 Figure 7. Orthogonal (Zijderveld) vector plots of representative specimens from the (a f) Yare and (g l) Shiquanhe Basins. Directions are plotted in situ; the solid and open circles represent the vectors and endpoints projected onto horizontal and vertical planes, respectively. YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1440

11 Figure 8. Equal-area projections of site-mean directions of the ChRMs before and after tilt corrections for the K 3 1 and K 3 2 formations from the Yare and Shiquanhe Basins. Incremental unfolding analysis [Watson and Enkin, 1993] for ChRMs isolated from Late Cretaceous K 3 2 formation indicative of a postfolding origin. The symbols are as for Figure 6. The magnetic carriers resident in the tuffs are complicated. Rapid increase of IRM intensity below 0.3 T (Figure 4h) and significant drop in susceptibility at ~580 C (Figure 5g) suggest the presence of magnetite, while the unsaturated IRM intensity up to 2.0 T and high Hcr of ~0.3 T (Figure 4h) further suggest the high-coercivity ferromagnets. The laboratory unblocking temperature of ~680 C indicates that the high-coercivity ferromagnet in these specimens is hematite (Figures 5g and 5h). AMS data from representative specimens from both K 3 1 and K 3 2 formations reveal large dispersion of principal directions (Figures 6d and 6e) and relatively low-anisotropy degree with P j < 1.06 (Figure 6f), suggesting insignificant latter deformation, at least at the hand-sample scale. After removal of a low-temperature viscous component, the ChRMs are generally defined in the temperature interval of ~300 C to ~580 C or close to 680 C (Figures 7g 7l and Table S1). For the K 3 1 formation, 7 out of 10 sites yield valid site-mean ChRM directions with α 95 less than 15, two sites were excluded from calculation of the interval-mean direction because they had distinctly outlying directions. The interval-mean of five remaining sites is D =343.0, I =29.7 (α 95 = 17.9 ) before and D =10.0, I = 34.9 (α 95 = 12.3 ) after tilt correction (Table 4). The data grouping is improved after tilt correction with k s /k g = 2.031, although the fold test on the basis of McFadden s [1990]methodprovestobeinconclusive.FortheK 3 2 formation, 7 out YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1441

12 Table 3. Site-Mean ChRM Directions of the Middle Chalicuo Group (~80 Ma) From the Yare Basin, South Tibet, China a Coordinates Site ID Lithology Long Lat Strike/Dip n/n 0 Dg (deg) Ig (deg) Ds (deg) Is (deg) k α 95 (deg) Plat ( E) Plong ( N) K 3 2 Formation, Yare Basin (~80 Ma, Section A) xy186 Basalts /19 9/ xy185 Basalts /26 11/ xy184 Basalts /53 10/ xy187 Basalts /30 8/ xy188 Basalts /30 9/ xy189 Basalts /54 7/ xy190 Basalts /54 12/ Agglomerates xy191 Basalts /48 7/ xy216 Basalts /54 9/ xy217 Basalts /54 10/ xy218 Basalts /38 10/ xy219 Agglomerates /38 5/ xy220 Basalts /38 11/ xy221 Basalts /40 9/ xy222 Basalts /40 9/ Unit-mean / K = 198.6, A 95 = 2.7 a Fold tests: (1) Watson and Enkin [1993]: the optimal concentration is achieved at ± 13.8% unfolding level, indicative of a positive fold test; (2) McElhinny [1964]: N = 15, k s /k g = > F [28, 28] = of the grouping of the ChRM, at 95% confidence level, indicative of a positive fold test; (3) McFadden [1990]: N = 15, in situ ξ 2 = 9.192, tilt-corrected ξ 2 = 0.694, statistical threshold ξ = at 95% confidence level, indicative of a positive fold test. Abbreviations: Site ID, Site identification; Long and Lat, longitude and latitude of sampling site in geographic coordinates; Strike/Dip, strike azimuth and dip of bed; n/n 0, number of samples or sites used to calculation/yielded well-defined ChRM or demagnetized; Dg and Ig (Ds and Is), declination and inclination of in situ (after tilt-corrected) direction. κ and α 95, precision parameter, and 95% confidence limit of Fisher statistics; Plat and Plong, latitude and longitude of virtual geomagnetic pole (VGP) in stratigraphic coordinates; K and A 95, precision parameter and 95% confidence limit of Fisher statistics for locality-mean VGP. of 13 sites yield valid site-mean ChRM directions with α 95 less than 15 (Table 4). After removal of the 2 outlying sites (xa171 and 179), the remaining 5 sites yield an interval-mean direction of D = 30.0, I =41.6 (α 95 = 15.5 ) before and D = 30.5, I = 18.1 (α 95 =7.7 )aftertiltcorrection(figures8aand8b and Table 4). The data grouping is improved significantlyaftertiltcorrection(k s /k g = 3.96) (Figures 8a and 8b). Furthermore, both normal and reversed polarities were identified from the five sites of the K 3 2 formation and permit a reversal test on this data set (Figure 8b). The angular difference between the tilt-adjusted directions of each polarity is 4.2 which is less than the critical angle of 17.4 and indicative of a positive reversal test of class C [McFadden and McElhinny, 1990]. Site xa178 collected from volcanic bombs and lava cobbles at the bottom of the section permits a conglomerate test. The vector sum of 7 ChRM directions from xa178 yields a vector length of 1.33, which is lower than the critical value of 4.12 for N = 7 at 95% confidence level and indicative of a positive conglomerate test [Watson, 1956] (Figure 8c). The results of all these tests thus support a primary origin for the remanence The Reliability and Age of the Remanence The isotopic ages of approximately 92 Ma determined by 40 Ar/ 39 Ar dating of lavas from the base of the K 3 1 and K 3 2 formations in the Shiquanhe Basin (Figures 3a 3d and Table 1) are not significantly different and suggest that the K 3 1 formation probably erupted over a short time. However, the presence of reversed polarity in the K 3 2 formation (Table 3) suggests that the time period of deposition continued to the end of the Cretaceous Normal Superchron (CNS; approximately 120 to 84 Ma). To better average PSV, we include the 10 sites from the Shiquanhe Basin; the result then yields a mean direction of D =4.8, I = 38.0 (α 95 = 16.4 ) before and D = 21.1, I =26.8 (α 95 = 10.0 ) after tilt correction (Figures 8a and 8b and Table 4). This result produces a positive fold test and yields a paleomagnetic pole at 64.1 N, E (A 95 =9.6 ;Table4)withanassignedageofapproximately90 to 80 Ma. The virtual geomagnetic pole (VGP) scatter (S) for this paleomagnetic pole is 16.4, which is somewhat larger than the model-predicted value (S = ~11 ) for the CNS (~ Ma) at a YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1442

13 Table 4. Site-Mean ChRM Directions of the Upper Cretaceous Volcanic Rocks From the Shiquanhe Basin, South Tibet, China a Coordinates Site ID Lithology Long Lat Strike/Dip n/n 0 Dg (deg) Ig (deg) D b (deg) Ds (deg) Is (deg) k α 95 (deg) Plat ( E) Plong ( N) K 3 1 Formation, Shiquanhe Basin (~90 Ma, Section B) xa161 Basalts /29 9/ xa162 Basalts /29 7/ xa163 Basalts /29 7/ xa164 Basalts /45 6/ xa165 b Basalts /41 10/ xa166 Basalts /41 7/ xa167 Basalts / xa168 b Basalts /22 11/ xa169 Basalts / xa170 Basalts / Sub-mean / K = 30.9, A 95 = 14.0 K 3 2 Formation, Shiquanhe Basin (From ~93 Ma to Less Than 84 Ma, Section C) xa171 c Basalts /18 5/ xa172 Basalts /35 7/ xa173 Basalts / xa174 Tuffs / xa175 Tuffs / xa176 Tuffs /36 8/ xa177 Tuffs / xa179 b Basalts /19 6/ xa180 Basalts / xa181 Tuffs /16 11/ xa182 Tuffs /16 7/ xa183 Tuffs /16 9/ Sub-mean / K = 159.1, A 95 = 6.1 Group-mean / K = 26.3, A 95 = 9.6 a Field tests for site-mean directions of the K3 2 formation: fold test: (1) McElhinny [1964]: N = 10, k s /k g = > F [18, 18] = of the grouping of the ChRM, at 95% confidence level, indicative of a positive fold test; (2) reversal test [McFadden and McElhinny, 1990] for K 3 2 formation: the angular difference of 4.2 between the normal and reversed polarities is less than the critical angle of 17.4 indicative of a positive reversal test of C class; (3) McFadden [1990]: N = 10, in situ ξ 2 = 7.559, tilt-corrected ξ 2 = 1.819, statistical threshold ξ = at 95% confidence level, indicative of a positive fold test. Abbreviation: D b, tilt-corrected declination; Ds, tilt-corrected declination with inclined fold axis plunging 26.5 toward 235. Other abbreviations are as for Table 3. b Sites discarded from calculation of the locality-mean ChRMs for significant outlying directions, see text for more information. c Sites discarded from calculation of the locality-mean ChRMs for α95 > 15. corresponding latitude of ~15 N [Biggin et al., 2008] but is comparable with the one resolved from model of the last 5 Ma (S =~14 ) [Johnson et al., 2008]. Only one isotope age determination was obtained from the middle part of the K 3 2 formation in the Yare Basin, and this yields a 40 Ar/ 39 Ar plateau age of 79.6 ± 0.7 Ma (Figure 3c and Table 2). A paleomagnetic pole calculated from site-mean ChRMs of this unit lies at 68.4 N, E with A 95 of 2.7 (Table 3) and an interpreted age of approximately 80 Ma. Weak later deformation and the positive fold test support a primary origin for the ChRM directions. The lava-sandstone succession in the Yare Basin consists at least 10 eruption-sedimentary cycles (Figure 2g), suggesting a relative long duration for the sampled unit which is beneficial for averaging PSV. Furthermore, the calculated mean inclination (26.8 ) is not significantly different from the coeval paleoinclination obtained from the Shiquanhe Basin (25.6 ) where PSV appears to have been effectively sampled, and this supports the interpretation that the mean ChRM direction from the Yare unit is an unbiased estimate of mean geomagnetic field at approximately 80 Ma. Nevertheless, we note that the low VGP scatter of 7.3 could suggest inadequate sampling of the PSV. In order to better average PSV, a group-mean inclination of 26.7 ± 3.7 is calculated from the ChRM directions from the Yare and Shiquanhe Basins using the Web calculator for inclination data only [Arason and Levi, 2010] ( this yields a paleolatitude of 14.1 N ± 2.2 N at the reference site of 31.5 N, 80 E. YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1443

14 Table 5. Paleomagnetic Estimate of the Crustal Shortening in Asia Along ~80 E Since Approximately 80 Ma Age Reference Pole Reference Paleolatitude at 31.5 N, 80.0 E Observed Paleolatitude at 31.5 N, 80.0 E (This Study) Latitudinal Shortening (km) 85 Ma 80.0 N, E; A 95 = 5.5 [Schettino and Scotese, 2005] for Eurasia 80 Ma 73.5 N, E; A 95 = 3.9 [Torsvik et al., 2012] for Eurasia 80 Ma 79.5 N, E; A 95 = 3.4 [Cogné et al., 2013] for Asia Ma 54.7 N, E; A 95 = 6.3 [Huang et al., 2005] for Tuoyun 29.9 N ± N ± ± N ± N ± ± N ± N ± ± N ± N ± ± Discussion 5.1. Late Cretaceous Paleolatitude of the Western Segment of the Lhasa Terrane The paleolatitude of 14.1 N ± 2.2 N defined from the Yare and Shiquanhe Basins by this study strongly indicates that the western segment of the Lhasa Terrane was located farther south between about 90 and 80 Ma in contrast to its present latitude of ~32 N. In addition, an earlier study carried out by Chen et al. [1993] also reported paleomagnetic data from a 1000 km transverse along ~80 E longitude across the Tibetan Plateau and ChRMs isolated from Cretaceous limestones and sandstones from Shiquanhe, Longmucuo, and Aksaichin passed reversal and fold tests and hence were interpreted as primary. A group mean of 14 sites yielded a paleolatitude at ~7 N (at reference site of 31.5 N, 80 E) which is marginally lower latitude than our estimate. However, since sedimentary rocks typically show inclination shallowing during deposition and compaction [e.g., Tan and Kodama, 2002; Tauxe, 2005], this paleolatitude may be somewhat underestimated. The recalculated paleolatitude of ~12 N, assuming a typical flattening factor of 0.6, is in good agreement with our estimate and supports the view that the western segment of the Lhasa Terrane was located further south during Late Cretaceous times. Our estimate of a paleolatitude at ~14 N for the southern margin of Asia near Shiquanhe is not consistent with paleomagnetic study from the middle part of the Yarlung-Zambo suture zone where on the basis of a study of 62 specimens collected from fore-arc sediments of the Cuojiangding Group directly north of the suture zone (30 N, 84 E), Meng et al. [2012] proposed that the Lhasa Terrane was located at ~24 N during the late Paleocene (~57 54 Ma). However, using the ChRM directions obtained from this study to calculate a paleomagnetic pole (343.7 N, 80.1 E with K = 6.7 and A 95 = 7.6 ), we note that the VGP scatter is large (31 ) and much larger than a model-predicted value (S = ~15 at ~22 N [Johnson et al., 2008]); the A 95 of 7.6 also falls outside of the N-dependent A 95 envelope recommended by Deenen et al. [2011] which defines an acceptable interval of 3.2 to 5.2 for N = 62. The large VGP scatter and A 95 are generally indicative of measurement errors and/or overprinting beyond the effects of PSV, and this may lead to considerable uncertainty in estimation of paleolatitude. We hence prefer to discard the result of Meng et al. [2012] from further tectonic interpretation Shortening Along ~80 E Longitude North of the Indus-Yarlung Zambo Suture Zone The postcollisional crustal shortening within Asia can be estimated by comparison of the observed and reference paleolatitudes of the southern margin of Asia [e.g., Chen et al., 2010, 2014; Dupont-Nivet et al., 2010a; Liebke et al., 2010]. The calculated results are listed in Table 5 and show a range of estimates ranging from ~1100 to 2200 km, depending on the different apparent polar wander (APW) paths used. The recommended reference pole at approximately 80 Ma from Torsvik et al. [2012], which has been corrected for inclination shallowing for the sedimentary rocks with a typical flattening factor of 0.6, yields a maximum estimate of ~2200 km, while the value decreases to only ~1100 km when the APW path compiled more recently by Cogné et al. [2013] for Asia is considered. Furthermore, using the paleomagnetic pole obtained from the lower and upper basalt units from the Tuoyun Basin near the northern Pamir (40 N, 75 E) [Huang et al., 2005], a crustal shortening of 1380 ± 590 km is defined in a north-south direction between the southern margin of Asia and the southern Tianshan. However, we observe that the various APW paths for Asia are complicated in detail [e.g., Dupont-Nivet et al., 2010b] and YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1444

15 merit discussion beyond the scope of this paper. A more precise estimate for the magnitude of shortening therefore remains open for discussion. The balanced cross-section approach can also be used to provide independent estimates on surface shortening in Asia. The updated geologic data suggest that only ~750 km total crustal shortening is observed along ~82 E [van Hinsbergen et al., 2011, and references therein], which is substantially smaller than the minimum estimate of ~1100 km derived from the paleomagnetic data. A possible reason for the discrepancy is the large uncertainties associated with the balanced cross-section method when being used to estimate crustal shortening within upper crust, especially when this has been subjected to substantial long-distance underthrusting. The Himalaya is a good test case. Only ~670 to 775 km of surface shortening has been observed by workers in the Himalayas [DeCelles et al., 2002; Yin et al., 2010], whereas the northern extension of the Indian lithosphere is generally estimated to be underthrust by more than 1500 km [Patzelt et al., 1996; Yi et al., 2011]. For the Asian side, a recent seismic global tomography study defined a lower mantle positive anomaly at depths between 1100 and 900 km beneath the India-Asia suture zone. This anomaly has been interpreted as a remnant of Asian lithosphere and suggests that a total subduction of Asian continent may exceed 600 km [Replumaz et al., 2013]. This implies that the pattern of Asia continent response to the India-Asia collision is somewhat similar to that of the Indian Plate; both continents are involved in considerable continental subduction. Such long-distance subduction can result in development of major thrust faults such as the Main Central Thrust Fault in the Himalayas and lead to underestimation of the total shortening in Asia Outline of the Southern Margin of Asia Prior to the India-Asia Collision Comparing with the western segment of the Lhasa Terrane, far more data are available from the Upper Cretaceous and Paleogene rocks in the eastern part as summarized by Yi et al. [2011]. A combination of data for accepted sites from different studies allows us to calculate three paleopoles for the Linzizong Group near the Linzhou Basin located at 66.0 N, E (A 95 = 8.5 ); 73.6 N, E (A 95 = 4.6 ); and 76.5 N, E (A 95 = 4.7 ) for the Dianzhong (64 60 Ma), Nianbo (60 50 Ma), and Pana (50 44 Ma) Formations, respectively. Accordingly, the Lhasa Terrane is predicted to have been located at paleolatitudes of 6.1 N ± 8.5 N, 12.9 N ± 4.6 N, and 19.3 N ± 4.7 N, respectively [Yi et al., 2011]. Because most studies so far have precluded an initial collision between India and Asia later than ~50 Ma [Ding et al., 2005; Najman et al., 2010; Cai et al., 2011], we propose that the most plausible paleolatitude location for the precollisional southern margin of Asia is ~6 to 13 N. This result is confirmed by recent paleomagnetic study of three formations of the Linzizong Group from the Linzhou Basin which also indicates that the Lhasa Terrane was located at ~7 to 11 N during this interval [Chen et al., 2014]. The paleomagnetic data set of the Linzizong Group is also revised by Lippert et al. [2014] who propose a former location of ~22 N for the Lhasa Terrane. However, this compilation did not include data from the Dianzhong and Nianbo Formations and only calculated a paleopole based on the ChRM directions from the Pana Formation, the uppermost unit of the Linzizong Group with a younger age of approximately 50 to 44 Ma [Zhou et al., 2004]. For this unit, the large dispersion of ChRM directions indicates a complex paleomagnetic record for which a number of causes, including inclination shallowing [Huang et al., 2013], inadequate and/or repeating sampling of the geomagnetic field [Liebke et al., 2010; Chen et al., 2014] could have resulted in a biased estimate of the paleolatitude. Moreover, in view of the young age of this unit, we propose that these data only provide an upper limit for the precollisional paleolatitude of the southern margin of Asia. We recognize that a paleomagnetic study of the red beds and intercalated lava flows from the Shexing Formation, which clearly has an age earlier than the collision ( Ma), yields a paleolatitude of ~24 N for the Lhasa Terrane, significantly higher than our estimate [Tan et al., 2010]. This result is widely cited as the paleolatitude where India and Asia initially collided [Dupont-Nivet et al., 2010a; Meng et al., 2012; van Hinsbergen et al., 2012]. Nevertheless, there are several key uncertainties with this conclusion. First, for the ChRM directions resolved from Tan et al. s [2010] study, although the inclinations corrected from the red beds appear to agree well with the one from lavas, the observed mean declinations are significantly different, as suggested by an angular difference of ~30 (Zhiming Sun, personal communication, 2013); moreover, the strata in section B near Pengbo farm in the Tan et al. s [2010] study was considered to be overturned, while our field observation in 2010 found clear evidence that this is not the case (Figure S1 in the supporting information). In view of the ambiguities raised in this study we consider that the data YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1445

16 Tectonics from the Tan et al. s [2010] study cannot be reliably used for tectonic interpretation. Recently, a new study of the upper Shexing Formation where a red bed-lava succession was investigated and found to have recorded comparable paleomagnetic inclinations, further paleomagnetic analysis, has yielded a paleopole at 71.0 N, E with A95 = 5.3 and indicating that the Lhasa Block was located at ~10 N between ~74 and 64 Ma (Zhiming Sun, personal communication, 2014). By plotting the paleolatitudes of the Lhasa Terrane from west to east between the longitudes of ~80 E and 90 E, the southern edge of Asia is reconstructed into a relatively straight structure with an orientation of ~310 E (Figure 9b). This finding is supported by the paleomagnetic declinations resolved from the Yare and Shiquanhe Basins of this study. In the study area south of the Shiquanhe and Yare Basins (Figure 9a), the northern branch of the Indus-Yarlung Zambo suture zone, characterized by a suit of ophiolites, is Figure 9. Reconstructions of the southern margin of Eurasia prior to the India-Asia collision at the western segment of the (a) Lhasa Terrane; (b) between the longitude interval of ~80 E to 90 E; and (c) in a Eurasian frame, Africa-Arabian, and Indian plates showing a quasi-linear structure in contrast to its present-day shape. The precollisional shape of the southern margin of Asia (solid lines and arrows in green in Figure 9a) was restored by aligning the documented declinations (solid lines and arrows in green) to the geographic north; precollisional positions of the southern margin of Asia in Figure 9b are constrained by paleomagnetic data from Chen et al. [2010, 2014], Tang et al. [2013], Yang et al. [2015], and this study; Eurasia, Africa, Arabia, and India in Figure 9c are positioned in accordance with the APW path of Torsvik et al. [2012]; the asterisk in red denotes the Euler pole of modeled southern margin of Eurasia (58 N, 170 E) on the basis of plate circuit and available paleomagnetic data, and the fitted radius of the small circle is ~8700 km, suggesting that the precollisional southern margin of Eurasia was approximately distributed along a great circle; the Neo-Tethyan subducted slab is constructed according to Van der Voo et al. [1999]. Abbreviations: IN, India; GI, Greater India ; AB, Arabia; AF, Africa. The Greater India of ~1500 km is defined according to Yi et al. [2011]. YI ET AL. RESTORING THE SOUTHERN MARGIN OF EURASIA 1446