Low palaeoelevation of the northern Lhasa terrane during late Eocene: Fossil foraminifera and stable isotope evidence from the Gerze Basin
Low palaeoelevation of the northern Lhasa terrane during late Eocene: Fossil foraminifera and stable isotope evidence from the Gerze Basin
Carmala N. Garzione
OPEN The Lhasa terrane is a key region for understanding the paleoelevation of the southern Tibetan Plateau after India-Asia collision. The Gerze Basin, located in the northern part of the Lhasa terrane, is a shortening-related basin. We discovered Lagena laevis (Bandy) fossils in upper Eocene strata of the Gerze Basin. This type of foraminifera is associated with lagoon and estuarine environments, indicating that the northern part of the Lhasa terrane was near sea level during the late Eocene. We speculate that these foraminifera were transported inland by storm surges to low elevation freshwater lakes during times of marine transgressions. This inference is consistent with the relatively positive ?18O values in carbonate from the same deposits that indicate low palaeoelevations close to sea level. Considering the palaeoelevation results from the nearby Oligocene basins at a similar latitude and the volcanic history of the Lhasa terrane, we infer that large-magnitude surface uplift of the northern Lhasa terrane occurred between late Eocene and late Oligocene time.
The Tibetan Plateau is widely considered to be the result of shortening related to convergence both preceding and
post-dating the initial collision between the Indian and Eurasia plates at 65?50 Ma1?3. However, with the advent
of quantitative palaeoelevation methods, the association between the timing of crustal thickening and the timing
of surface uplift of mountain belts has been challenged4,5. Consensus has not been achieved regarding the surface
uplift history of the Tibetan Plateau. Stable isotope paleoelevation estimates from the Nima and Lunpola basins
indicate paleoelevations similar to modern elevations by 26 Ma5,6 along the Bangong suture (Fig.?1). Wang et al.7
proposed that surface uplift of the southern and central Tibetan Plateau (Lhasa and southern Qiangtang
terranes) occurred by 40 Ma, based primarily on stratigraphy and thermochronology results. Arguing for an Eocene
?proto-Tibetan Plateau?, Kapp et al.8 suggested that the collision between the Lhasa terrane and the Qiangtang
terrane resulted in crustal shortening of the Lhasa terrane that may have raised southern Tibet to an elevation of
3?4 km during the Cretaceous, prior to India-Asia collision. This inference of a high proto-Tibetan plateau prior
to India-Asia collision is reinforced by paleoelevation reconstructions of the Linzhou Basin, which suggest that
the Gandese magmatic arc region in the southern Lhasa terrane attained an elevation of ~4500 m by the time of
India-Asia collision9. In contrast, Hetzel et al.10 performed thermal modelling of (U-Th)/He ages of apatite and
zircon, as well as apatite fission track data, from the northern Lhasa terrane, and inferred that peneplanation of
the northern Lhasa terrane occurred at least 3?4 km lower than its current elevation of ~5300 m. Moreover, the
formation of the peneplain at low elevation was completed by ca. 50 Ma, and the resistant bedrock surface has
undergone slow erosion since then. Zhang et al.3 also argued that the presence of the proto-Tibetan Plateau does
not match the sedimentary characteristics and overall topography during the suggested time period based on
a detailed summary of 121 geological maps (1:250000) of the Tibetan Plateau area, and they estimated that the
average palaeoelevation in central Tibet was ~500 m at ca. 34 Ma.
Based on conflicting evidence for high paleoelevations from sparse stable isotope studies and low
paleoevelations from fossil, stratigraphic, and thermochronology/thermal modelling studies, it remains unclear whether
the Lhasa terrane was at a high elevation at and following the collision of the Indian and Eurasian plates. To
determine the paleoelevation history of the Lhasa terrane, we collected both fossil and stable isotope data from a late
Eocene section in the Gerze Basin. We provide age information for these deposits, based on magnetostratigraphy
with additional age constraints from fossils and zircon U-Pb dates.
Lithostratigraphy. Eocene strata in the northern Lhasa terrane are defined as the Dingqinghu Formation.
This study is focused on the upper part of the Dingqinghu Formation, located at 31?54.655?N, 85?7.693?E
(4618 m), with a thickness of 330 m (Fig.?2a). The base of the section is marked by a thick layer of conglomerate,
and the top of the section is the core of a syncline. The lower part (layers 1?26) of the section consists of
mediumto thickly bedded brown conglomerate, pebbly sandstone, and cross-stratified sandstone. Occasionally
interbedded with clastic deposits are thick beds of grey bioclastic limestone that contain freshwater ostracods, gastropods
and charophyte fossils. These facies suggest that the depositional environment for the lower part of the section
was lacustrine to marginal lacustrine. The upper part (layers 27?86) mainly consists of thin-medium beds of grey
limestone intercalated with grey shale and two pebbly sandstone layers (<20 cm thick). These facies suggest that
the depositional environment of the upper part of the section was a stable shallow lacustrine environment. The
bedding thins upward in the section, suggesting a transition to deeper water environments.
Chronostratigraphy. We use three different methods to constrain the ages of the investigated section:
detrital zircon U-Pb ages, index fossils and magnetostratigraphy. We collected three detrital zircon samples from
layers 1, 12, and 25, which in total yielded 198 analysis points for the selected zircon grains. The three youngest
detrital zircon grains yielded a mean age of ca. 42 Ma (Fig.?2b), and the youngest zircon population provides the
maximum depositional age for our section.
Previous research has shown that the ostracoda genus Austrocypris is an important index fossil that
was short-lived in the late Eocene11,12. This genus is widely observed in the lower member of the late Eocene
Xiaganchaigou Formation in the Qaidam Basin11?13 and in the fourth member of the Shahejie Formation in
the Dongpu depression14 and Bohai Sea Basin in eastern China15. Their extinction may correspond with the
Eocene-Oligocene climate transition that occurred at ca. 34 Ma12,16. This ostracoda is observed in our study
section between layer 8 and layer 76 and can be used for regional stratigraphic correlation to bracket the timing of
events. Moreover, Gyrogona qianjiangica, which is another species of chara, is an index fossil of the late Eocene
Xiaganchaigou Formation13 that occurs from layer 9 to layer 83.
Based on the above late Eocene fossil age constraints, the age of this section is bracketed between 42 Ma and
34 Ma. A total of 197 oriented palaeomagnetic samples were collected in our section to acquire high-resolution
(average ~1.6 m) magnetostratigraphic data (see Supplementary Table S1 and Fig. S1). The results can be
correlated with the standard polarity time scale 2004 (GTS 2004)17, which covers an age range of 39.4?35.7 Ma.
Foraminifera fossils. We found a foraminifera fossil shell in layer 2 and five foraminifera fossil shells in
layer 23 (Fig.?2c). The lithology of layer 2 is purple-red medium-thick bedded laminated siltstone, whereas the
lithology of layer 23 is purple-red thinly bedded massive mudstone.
These fossils are well preserved with obvious identifying characteristics, indicating that they are the same
species. The fossils are globular, ovoid, and widest just below the middle section and presented circular cross
sections. Additionally, they have smooth rounded bases, short, round, tapering necks approximately one-ninth
the length of the body, and smooth, finely perforated walls. The plesiotype length is 0.23mm, and the diameter is
0.20 mm. The elemental composition was detected via an energy dispersion spectrometry analysis, and the results
indicate that they are mainly calcareous (Fig.?2c). Based on the neck shape, clear-walled burrow structure and
standard calcareous composition, these fossils are identified as Lagena laevis Bandy. This species belongs to the
Lagena genus, Nodosariacea family, and Rotaliina order and has been found in the Eocene Oujiang Formation
in the East China Sea shelf and late Eocene strata in Alabama, USA18. Lagena laevis Bandy is a benthic19 warm
water foraminifera species that occupies a narrow range of salinities and can adapt to a strong hydrodynamic
environment. This organism occurs in estuaries or coastal bay sedimentary environments and has been reported
in offshore gulfs around the world20?23.
Oxygen isotope paleoelevation reconstructions. 126 limestone samples were collected for oxygen
isotope analysis to reconstruct palaeoelevation. All of the carbonate results are reported in ?18Oc (VPDB-Vienna
PeeDee belemnite). Evaporation during deposition and post-depositional diagenesis are two major factors that
can influence oxygen isotope values, resulting in biased paleoelevation reconstructions, and thus we carry out a
detailed evaluation of these factors. Analysis of diagenetic phases provides insights into whether the ?18Oc values
of the primary micritic carbonates have been modified24. Thin section analyses show that the lacustrine
carbonates from layers 1?30 of our section contain significant (>50%) coarse-grained spar that is characterized by
extensive dissolution cavities filled with sparry calcite (Fig.?3a,b). However, the micrites from layers 31?86 of our
section generally preserve primary pore spaces (Fig.?3c) that are usually filled with fine-grained blocky calcite
textures associated with early diaganesis, when pore fluids reflect surface water compositions and temperatures.
Some samples show vein calcite (Fig.?3d). To determine the possible influence of veins and coarse-grained blocky
calcite that filled dissolution cavities, primary micritic carbonate and late stage diagenetic phases were sampled
using a microdrill to separate the micrite from the sparite for analysis. Sparry calcite within veins and
dissolution cavities show ?18Oc values of ?20.8? to ?5.3? (see Supplementary Table S2), whereas micrite ?18Oc
(VPDB) values are much more positive, ranging from ?8.1? to ?2.1?, with an average of ?5.1 ? 1.1? (1?)
(see Supplementary Table S3).
In addition to diagenesis, evaporation is another factor that can influence the ?18O values of lake water and
carbonate. In a hydrologically open system, oxygen isotopes are less likely to be affected by evaporation. ?13C
and ?18O values that present variation coefficients of r >0.7 are considered to indicate a closed lake system or
long-term residence times25. Lacustrine micrite from layers 31?86 show poor correlations between ?18Oc and
??13Cc values (r = 0.0173) (Fig.?3e), which implies a hydrologically open condition. Sparite from layers 1?30 shows
a distinct isotopic composition from micrite, indicating that micrite is a faithful recorder of the primary isotopic
composition of carbonates. The very negative ?18Oc end-member of about ?20? for sparite indicates that these
diagenetic carbonates either precipitated from meteoric water with a low ?18O value, consistent with modern
high altitude precipitation, or from waters of similar composition to what precipitated the micrites, but at burial
temperatures higher than surface temperature24.
Paleogeography inferences from foraminifera. Although foraminifera fossils mainly occur in marine
strata, they are also distributed in Late Quaternary strata located hundreds of kilometres away from current
shorelines associated with marine transgressions26,27. Inland basins that have little contact with the ocean may
also contain a small amount of foraminifera28,29 associated either spatially or temporally with marine sources. The
airborne colonization of these inland water bodies via migratory birds has been suggested by previous research,
although this type of colonization is occasional and does not result in selectivity of species. The foraminifera
fossils from our study section all belong to the same species and have a relatively high abundance. This species has a
nearly hollow spherical shape, and its selectivity is likely related to the large specific surface area of its structure,
which facilitates its transport during storm surges. Research has also shown that foraminifera from older marine
sources can adapt to and thrive in suitable inland saltwater environments30. However, in the investigated section,
the underlying and overlying lacustrine limestone strata contain rich chara and lacustrine ostracoda and
represent clear and shallow lacustrine environments. Additionally, this species of foraminifera has not been previously
reported in older marine strata across the Lhasa terrane and thus is unlikely to have survived from older marine
The Lagena laevis Bandy foraminifera could have been transported hundreds of kilometres inland from the
ocean by marine transgression or by large storms if a passageway was available31. Two possible passageways
include the fluvial volcanic depression through Gangdese volcanic arc, which is connected to the Himalayan Sea
in southern Tibet, and the channels that occur through the coastal-alluvial plain connected to Sumxi Bay of the
Pamir Sea in western Tibet (Fig.?1)32. Both the Himalayan Sea and Pamir Sea belong to the Neo-Tethys Sea.
(1) Passageway from the Himalayan Sea. The Palaeogene Himalayan seaway in Tibet was an east-west trending
remnant sea that covered southern Tibet and the Gangdese areas. Burrard33 and Douville et al.34 discovered
Palaeogene Himalayan marine strata in Gamba and Tingri (Fig.?1), and researchers have also discovered
many foraminifera, ostracoda and chara fossils in these areas and conducted further studies of the marine
strata35,36. Previous research has shown that the Zongpubei Formation in Gamba and Tingri contain the
youngest documented marine strata37,38. The Zongpubei Formation is dated to the early Priabonian stage
of the Eocene (ca. 35 Ma) and in both areas yields larger foraminifera, including Nummulites willcoxi and
Nummulites manila, and other smaller foraminifera. The strata at the top of the Zongpubei Formation in
these areas are cut by faults and are sparsely outcropping. However, the deposition in southern Tibet was
stable during the Eocene37,38, thus, the latest time of regression may have been after the early Priabonian stage
of the late Eocene. Furthermore, float rocks that contain rich foraminifera have been found in both Niumagou
and Zhongba, which are located to the north of Lhasa City. The foraminifera are identified as Keramosphaera
tergestina (Stache), which only lived during the Palaeocene-Eocene39. Although the fossil positions are not
clear, seawater certainly reached the area near Lhasa City during the early Tertiary. Deltaic deposits have been
documented in several Eocene stratigraphic sections along the southern margin of the Gangdese volcanic
arc32, indicating a fluvial drainage basin sourced in the Gandese volcanic arc. This palaeogeography could
have enabled the foraminifera to be transported through channels to be deposited in the Gerze Basin during
transgressions and/or extreme storms if the Gerze Basin and the fluvial system was located at a low elevation.
However, previous research suggested that portions of the magmatic arc were at high elevation at this time9,
which seems inconsistent with a low elevation fluvial volcanic depression through the Gangdese volcanic arc.
(2) Passageway from Sumxi Bay. Gulf lagoons developed in the Sumxi area at the easternmost part of the Pamir
Sea during the Eocene preserved as the Kashi Group40. The Kashi Group records three
transgression-regression cycles in the Tarim Basin41,42 that occurred northwest of the Sumxi area. Among these transgressions,
the last marine incursion occurred at 41?38 Ma (late Eocene), overlapping the time frame during which
foraminifera appeared in the Gerze Basin. Sedimentary facies in the Sumxi area indicate a neritic
environment, with several delta systems located along the eastern margin of Sumxi bay. An extensive area in the
southwestern part of Sumxi Bay contained an alluvial plain and lakes32,43 (Fig.?1). We infer that rivers
connected the Sumxi Bay to inland lakes, similar to the Yangtze River in eastern China and the Mississippi River
in the USA. These channels may have enabled eastward intrusions of seawater during large transgression
events, which could have transported the foraminifera Lagena laevis Bandy from an estuary environment
to the Gerze Basin on the northern margin of the Lhasa terrane. This process requires that this region was
located at low paleoelevation to allow for occasional intrusion if marine water containing foraminifera.
Evaluation of palaeoelevation. Based on the discovery of foraminifera fossils, the Gerze Basin appears to
have been close to sea level during the late Eocene (ca. 39 Ma). Oxygen isotopes from lacustrine micrites enable
evaluation of the paleoelevation of this region. Comparisons between the ?18O values of modern surface water
and palaeo-surface waters can be used to evaluate temporal changes in the surface elevation and climate44. Study
of modern surface water and air mass trajectories across the Tibetan Plateau shows that most of the moisture that
reaches the southern plateau (Lhasa terrane and the area to the south) is derived from a southern source region45.
The ?18O values of modern surface water in the southern plateau, including the Lhasa terrane, conform well with
a simple Rayleigh distillation model of vapor mass depletion in 18O, indicating that stable isotopes of meteoric
water provide reliable reconstructions of palaeoelevation44.
We assume that 25 ? 10 ?C reflects a reasonable temperature uncertainty for lacustrine carbonate
precipitation that could account for the overall warmer Eocene conditions and low latitude46, even at moderate to
high paleoelevations. We then used the mean ?18Oc (VPDB) of lacustrine micrites of ?5.1 ? 1.1? (1?) and the
temperature-dependent fraction equation between water and calcite47 to calculate the ?18Opsw (psw - paleo-surface
water) relative to SMOW - Standard Mean Ocean Water of ?2.8 ? 1.7?. The 1? uncertainty in this ?18Opsw
estimate is determine by a bootstrap Monte Carlo approach that propagates errors associated with uncertainty in the
temperature of carbonate precipitation, error in the temperature-fractionation equation47, and scatter in the ?18O
values of lacustrine micrite. Because of global cooling and the build-up of the continental ice sheets, the oxygen
isotope value of the sea water has changed by approximately +1? between the Eocene and the present day48.
For comparison to modern meteoric water, the ?18Opsw value is therefore corrected to ?1.8 ? 1.7?. We
compared this value of ?18Opsw to modern meteoric water isotope results (annual average weighted by precipitation
amount) of the GNIP (Global Network of Isotopes in Precipitation) sites in Asia49. Considering the mean ?18Opsw
of ?1.8 ? 1.7?, with values as low as ?3.5? at 1?, these values are observed in modern rainfall that occurred
at 50 ? 600 m asl (Fig.?4). The high ?18Opsw values in Eocene deposits of the northern Lhasa terrane are therefore
consistent with fossil foraminifera results, which both indicate low paleoelevations, close to sea level. This low
palaeoelevation inference agrees with sedimentary characteristics and the overall pattern of deposition and erosion
of the Tibetan Plateau during the Eocene3 (Fig.?5a).
The Nima Basin and Lunpola Basin are also located to the south of the Bangong-Nujiang orogenic belt at
approximately the same latitude as the Gerze Basin. According to previous palaeoelevation study results, in the
late Oligocene (~26 Ma), the Nima Basin had reached approximately 4,000 m5 and the Lunpola basin had reached
3,190 ? 100 m50,51. Combining these results with the results of our study suggests that the northern Lhasa terrane
experienced surface uplift of at least 3,000 m over approximately 10 million years from the late Eocene to the late
Evidence from low temperature thermochronometers and the history of magmatism in the Lhasa terrane are
consistent with a low paleoelevation of the northern Lhasa terrane in the late Eocene. Thermal modelling based
on the (U-Th)/He ages of apatite and zircon, as well as apatite fission track data from the northern Lhasa terrane
indicate that a peneplain formed at least 3?4 km beneath its current elevation of ~5300 m10. The formation of
this low elevation peneplain was completed by ca. 50 Ma, and the resistant bedrock surface has undergone very
slow erosion since then. Zircon age records52?54 indicate an approximately 10 million-year period of magmatic
quiescence in the Lhasa terrane between ca. 40?30 Ma with a significant magmatic event occurring after ca. 30Ma
(Fig.?5b(1)). Widespread magmatism initiated with adakitic rocks by ~26 Ma, followed by potassic-ultrapotassic
rocks by ~24 Ma55. The chemistry of these magmas has been attributed to break-off the Indian continental
lithosphere that had subducted beneath the Lhasa terrane and/or convective removal of the Tibetan lithospheric
mantle55,56. Ninety-six apatite fission track ages collected from the Lhasa terrane indicate three cooling age peaks
at 22?15, 10?7, and 5?0 Ma, whereas the Palaeogene ages are more scattered57 (Fig.?5b(2)). Magmatic zircon ages
and apatite fission track ages combined suggest minimal exhumation of the Lhasa terrane before the Oligocene.
Additionally, widespread unconformities and related basal-conglomerates are dated at ca. 34 Ma, and alluvial fan
and fluvial conglomerates resulting from rapid erosion are documented in the peripheral areas of the plateau,
such as the Shimagou, Huoshaogou and Tala Formations along the north of the plateau3.
Fossil foraminifera and oxygen isotope studies of Eocene Gerze basin lacustrine deposits indicate that the
northern Lhasa terrane resided at paleoelevations close to sea level between ~39 to 36 Ma. In combination with
results from the Nima basin and Lunpola basin that indicate high paleolevations of ~4000 m by 26 Ma, we infer
>3 km surface uplift of the northern Lhasa terrane between 36 Ma and 26 Ma. The temporal association between
this large magnitude surface uplift and widespread adakitic and potassic-ultrapostassic magmatism across the
Lhasa terrane after ~30 Ma suggests that this surface uplift event was associated with break-off Indian continental
lithosphere and/or convective removal of the lithospheric mantle beneath the Lhasa terrane.
Detrital zircon age. A total of 198 analysis points of selected zircon grains were analysed using a Neptune
MC-ICP-MS (multicollector inductively coupled plasma mass spectrometer) (Thermo Fisher, Lafayette,
Colorado, USA) with a UP193-FX ArF-excimer laser-ablation system (ESI, Sunnyvale, California, USA) at the
Isotope Laboratory of Tianjin Center, China Geological Survey (CGS). Data were collected with a 35 ?m spot and
the ablation depth was 20?40 ?m. Zircon 91500 was the external standard used for age calculation, NIST SRM
610 silica glass was the external standard used for calculation of concentrations, and Si was the internal standard.
For ages younger than 1,000 Myr ago, the discordance is defined as (207Pb/235U?206Pb/238U)/ 207Pb/235U*100 and
206Pb/238U ages were adopted for them. All the ages of the three zircon grains in Fig.?2b present a low discordance
Magnetostratigraphy. A total of 197 oriented palaeomagnetic samples were collected in the field to acquire
high-resolution magnetostratigraphy data. Because most of the palaeomagnetic samples originated from
limestone, which contains small amounts of magnetic minerals, we used different methods to conduct the magnetism
tests. Samples with NRM (natural remanent magnetization) ?10?5 T were subjected to stepwise (averaging
17 steps) thermal demagnetization using a Thermal Demagnetizer (TD-48) (ASC Scientific Ltd, Carlsbad,
California, USA). The heating routine is as follows: 150?C/ 200 ?C/ 250 ?C/ 300 ?C/ 350 ?C/ 400 ?C/ 450 ?C/ 500 ?C/
525 ?C/ 550 ?C/ 585 ?C/ 610 ?C/ 625 ?C/ 640 ?C/ 660 ?C/ 680 ?C. Samples with NRM <10?5 T were subjected to
stepwise (averaging 17 steps) alternating demagnetization using a D2000 & ARM Alternating Demagnetizer (2G
Enterprises, Mountain View, California, USA). The demagnetization routine is as follows: heating up to 120?C,
then follow with alternating demagnetization steps as 30 mT/ 35 mT/ 40 mT/ 45 mT/ 50 mT/ 70 mT/ 100 mT/
150 mT/ 200 mT/ 250 mT/ 300 mT/ 400 mT/ 500 mT/ 600 mT/ 700 mT/ 800 mT. Magnetic remanence was
measured using a 2G superconducting magnetometer and JR6A two-speed rotation magnetometer (2G Enterprises,
Mountain View, California, USA) housed in field-free space. The tests were performed at the Centre of Analysis
and Testing, Institute of Earth Environment, Chinese Academy of Sciences. The characteristic remanence
directions were determined using principal component analysis, and directions were analyzed using Fisher statistics.
Microfossils. We collected 183 bulk samples for a microfossil analysis from the study section. A 15%
solution of H2O2 was used for the sample preparation of microfossils from mudstone and siltstone. The
heating-acid-digestion method with pure acetic acid (>99.5%) was used for the sample preparation of
microfossils from limestone. More than 1300 ostracoda fossils and 6 foraminifera fossils were handpicked and sorted. We
selected well-preserved specimens for photography using a scanning electronic microscope. An energy
dispersion spectrometry analysis of the foraminifera fossils was performed at the State Key Laboratory of Geological
Processes and Mineral Resources of the China University of Geoscience.
Oxygen isotope. We collected 126 limestones samples from the bottom to the top of the cross section, and
they were generally evenly distributed. The ?18O data are reported with respect to the V-PDB standard. Analyses
of the stable isotope were performed at the SIREAL (Stable Isotope Ratios in the environment, Analytical
Laboratory) of the University of Rochester using a ?Delta Plus XP? mass spectrometer (ThermoFinnigan, San
Jose, California, USA). ?18Opsw can be calculated from the ?18Oc, which was fractionated from paleo-waters. The
fractionation factor (?) is defined as: ?(calcite-water) = (1000 + ?18Ocalcite)/ (1000 + ?18Owater). The equation47 to
calculate (?) is: 1000ln?(calcite-water) = 18.03(103T-1)?32.42 (T in Kelvin). The equation58 to calculate ?18O(SMOW) from ?
18O(VPDB) is: ?18O(SMOW) = 1.03091 ?18O(VPDB) + 30.91.
We assume 25 ? 10 ?C as a reasonable lacustrine carbonate precipitation temperature to account for overall
warmer conditions, the low latitude position of southern Tibet in Eocene59.
Supplementary information accompanies this paper at http://www.nature.com/srep
Competing financial interests: The authors declare no competing financial interests. How to cite this article: Wei, Y. et al. Low palaeoelevation of the northern Lhasa terrane during late Eocene: Fossil foraminifera and stable isotope evidence from the Gerze Basin. Sci. Rep. 6, 27508; doi: 10.1038/srep27508 (2016).
This work is licensed under a Creative Commons Attribution 4.0 International License. The images
or other third party material in this article are included in the article?s Creative Commons license,
unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license,
users will need to obtain permission from the license holder to reproduce the material. To view a copy of this
license, visit http://creativecommons.org/licenses/by/4.0/
We thank Fang Han, Zenglian Xu, Gaolei Jiang and Keke Ai for their sampling assistance in the field . Bing Yang is acknowledged for providing help with the figure preparation and formatting. This study was funded by the Geological Survey of China (#1212011121261), National Natural Science Foundation of China (# 41302279) and the U.S . NSF (# 1211527 ).
Author Contributions Y.W. , K.Z. and C.G. wrote the main manuscript text , Y.X. and B.S. prepared Figure 1 and J.J. prepared Figure 2A. All of the authors reviewed the manuscript .