Global mercury cycle during the end-Permian mass extinction and subsequent Early Triassic recovery

Earth and Planetary Science Letters 513 (2019) 144–155

Contents lists available at ScienceDirect

Earth and Planetary Science Letters www.elsevier.com/locate/epsl

Global mercury cycle during the end-Permian mass extinction and subsequent Early Triassic recovery Xiangdong Wang a , Peter A. Cawood b,c , He Zhao a , Laishi Zhao a,∗ , Stephen E. Grasby d , Zhong-Qiang Chen e,∗ , Lei Zhang a a

State Key Laboratory of Geological Processes and Resource Geology, China University of Geosciences (Wuhan), Wuhan 430074, China Department of Earth Atmosphere and Environment, Monash University, Australia c Department of Earth Sciences, University of St Andrew, KY16 9AL, UK d Geological Survey of Canada, Natural Resources Canada, Calgary, Alberta T2L2A7, Canada e State Key Laboratory of Biogeology and Environmental Geology, China University of Geosciences (Wuhan), Wuhan 430074, China b

a r t i c l e

i n f o

Article history: Received 30 September 2018 Received in revised form 17 February 2019 Accepted 18 February 2019 Available online xxxx Editor: T.A. Mather Keywords: end-Permian mass extinction Early Triassic recovery Siberian Traps Hg chemostratigraphy Hg isotope

a b s t r a c t The end-Permian mass extinction (EPME) at ∼252 Ma was the most severe extinction in the Phanerozoic. Marine ecosystems devastated by the EPME had a highly prolonged recovery, and did not substantially recover until after the Smithian–Spathian substage boundary (SSB) of the Lower Triassic (5 to 9 Ma after the EPME). While the Siberian Traps large igneous province (STLIP) has been invoked as the driver of the mass extinction, there remains controversy as to the cause of the protracted Early Triassic recovery; although renewed STLIP volcanism has been suggested. These previous studies though have drawn conclusions based on geochemical records of sediments deposited in northern latitude settings. To investigate the relationship between STLIP and extinction/recovery processes on a global base, we examined mercury chemostratigraphy, including mercury concentrations and isotopes, from high southern latitude and equatorial sections that span the Late Permian Changhsingian to Early Triassic Spathian substage successions; the Guryul Ravine section, Kashmir in northern India, and the Chaohu section in southern China. Organic and inorganic carbon-isotope data define the EPME horizon in the Chaohu section and the SSB in the Guryul Ravine section, respectively. Hg/TOC values are dramatically elevated approaching the EPME horizon and maintain high values until the lower Isarcicella Isarcica conodont zone, the base of which is believed to be the end of the mass extinction. In the stratigraphically overlying beds, Hg/TOC generally displays lower values with slight fluctuations through the two sections. These fluctuations are likely related to the increased terrestrial Hg influx associated with strong chemical weathering in the Early Triassic, as shown by a positive correlation between the contents of Hg and Al, and by less positive 199 Hg values in Early Triassic samples. Our data, presenting the first Southern Hemisphere Hg record from Guryul Ravine, in combination with previous results, indicates that anomalous high mercury deposition at the EPME occurred globally. The generally positive 199 Hg values at this time reflects atmospheric-derived Hg, consistent with a volcanic Hg source which we suggest indicates global impact of STLIP eruption. In contrast, there is no evidence for a global Hg/TOC anomaly during the protracted Early Triassic biotic recovery, suggesting that potentially renewed STLIP volcanism had only a northern hemisphere influence on the global Hg cycle. This more limited impact, may still have played a role in the delayed Early Triassic recovery. © 2019 Elsevier B.V. All rights reserved.

1. Introduction Increasing evidence shows that intense volcanic activity of the Siberian Traps Large Igneous Province (STLIP) was likely the pri-

*

Corresponding authors. E-mail addresses: [email protected] (L. Zhao), [email protected] (Z.-Q. Chen). https://doi.org/10.1016/j.epsl.2019.02.026 0012-821X/© 2019 Elsevier B.V. All rights reserved.

mary trigger of the end-Permian mass extinction (EPME) (Wignall, 2001; Svensen et al., 2009; Grasby et al., 2011; Burgess et al., 2017). New high precision age data for the STLIP (explosive, effusive, and intrusive eruption phases) demonstrate that twothirds of the estimated 3–4 × 106 km3 of magma erupted over a ∼300 ka interval (pyroclastic and lava eruptions), before and during the mass extinction interval (Burgess and Bowring, 2015; Burgess et al., 2017). In addition, a major pulse of sill intrusive activity began coincidently with the onset of the extinc-

X. Wang et al. / Earth and Planetary Science Letters 513 (2019) 144–155

tion, and continued for at least 500 ka after the cessation of the mass extinction (Burgess and Bowring, 2015; Burgess et al., 2017). Modeling of the STLIP, and associated emplacement of sills into carbonaceous-rich country rocks, shows a major release of CO2 to the atmosphere resulting in the negative excursions of carbon isotopes recorded at the EPME horizon (Svensen et al., 2009; Burgess et al., 2017). This hypothesis is supported by the observation of coal fly ash deposits just below to the EPME (Grasby et al., 2011), and could explain the globally recognized negative excursions of both δ 13 Ccarb and δ 13 Corg values at this level (Korte and Kozur, 2010). The recovery of marine ecosystems and invertebrate faunas following the mass extinction was protracted (Chen and Benton, 2012), and was not complete until after the Smithian–Spathian substage boundary (SSB) transition of the Lower Triassic. This interval was marked by another severe loss of marine biodiversity coupled with global extreme climate change (Orchard, 2007; Brayard et al., 2009; Sun et al., 2012). Though the ultimate cause of the protracted recovery and the SSB event is still under debate, environmental deterioration during the late Smithian, and amelioration in the early Spathian, have been interpreted to be the result of the strengthening and reduction in intrusive magmatic activity of the STLIP, respectively (Zhang et al., 2015; Grasby et al., 2013b). Volcanism and anthropogenic emissions (coal combustion) represent two major present-day sources of mercury (Hg) to the atmosphere (Selin, 2009), and major volcanic events can impact global and regional Hg cycles (Pyle and Mather, 2003). As Hg is one of the most toxic heavy metal pollutants, and has a long atmospheric residence time (∼1.5 a) and potential for global distribution, such a release could have devastating global impacts on the environment (Sanei et al., 2012; Blum et al., 2014). Volcanic Hg is largely released as gaseous Hg0 and removed from the atmosphere either by direct uptake and absorption, or by oxidation to Hg2+ species (Gehrke et al., 2009; Bergquist, 2017), then accumulates in ocean and terrestrial environments through rainfall or adsorption (Gehrke et al., 2009). In the aquatic realm, dissolved Hg2+ can be converted to be methyl-mercury that has a strong affinity for organic matter (OM) (Gehrke et al., 2009; Grasby et al., 2013a). Consequently, dissolved Hg2+ is principally adsorbed onto OM when deposited in sediments and a generally constant Hg/TOC ratio has been observed in modern ocean sediments (Gehrke et al., 2009), as well as in ancient marine shale records (Grasby et al., 2013a). Therefore, when a Hg/TOC spike in the rock record is associated with a peak in Hg concentration, rather than a drop in TOC content, this can be related to increased Hg loading to the environment (Grasby et al., 2013a). When seen in several sites across the globe a Hg/TOC spike can be assumed to relate to additional influx of Hg to the atmosphere on a global scale and possibly caused by a large igneous province event (Percival et al., 2017). The source of Hg spikes can be further resolved through application of stable isotope analyses. Hg stable isotopes can undergo both mass-dependent fractionation (MDF, reported as δ 202 Hg) and mass-independent fractionations (MIF, reported as 199 Hg) in the environment, and can be used as tracers to discriminate Hg sources and cycling (Blum et al., 2014). Compared to Hg-MDF, Hg-MIF is considered as a more conservative tracer for Hg source because of its more limited pathways (mostly photochemical) and lower possibility of alteration in postdepositional processes (Blum et al., 2014; Thibodeau et al., 2016; Thibodeau and Bergquist, 2017). Volcanic Hg has insignificant MIF (199 Hg ≈ 0h) (Zambardi et al., 2009). Once emitted to the atmosphere, volcanic gaseous Hg0 , can enter into both terrestrial systems and aquatic systems either by direct uptake and absorption, or through oxidation to Hg2+ species (Bergquist, 2017). Aquatic environments can receive Hg through atmospheric Hg deposition

145

and direct uptake of Hg by terrestrial sources (Thibodeau and Bergquist, 2017). Recent research shows that marine sediments receive their Hg from both direct atmospheric and terrestrial inputs (Grasby et al., 2017; Wang et al., 2018; Them et al., 2019). Marine sediments dominated by atmospheric Hg deposition tend to have positive 199 Hg values because of the photochemical reduction in the surface environments (Grasby et al., 2017; Wang et al., 2018; Them et al., 2019). Conversely, sediments that receive Hg through terrestrial runoff tend to have more negative 199 Hg than atmospherically deposited Hg, as terrestrial sources may acquire negative 199 Hg values when plants and soils sequestrate Hg0 (g) (Thibodeau and Bergquist, 2017). Thus, Hg-MIF can provide important information on sources and pathways of a measured Hg spike in sediments. Hg chemostratigraphy has been used to explore the relationship between large igneous provinces and contemporaneous mass extinctions (e.g. Sanei et al., 2012; Grasby et al., 2013a, 2017; Percival et al., 2015, 2017; Thibodeau et al., 2016; Gong et al., 2017; Wang et al., 2018; Them et al., 2019). Anomalous Hg deposition was first observed at the EPME crisis in the Sverdrup Basin, Canadian High Arctic that occupied a paleogeographic position near the STLIP (Fig. 1; Sanei et al., 2012; Grasby et al., 2013a). Further studies of boreal records show a very small increase in Hg/TOC in the Sverdrup Basin and an additional, but less significant, Hg anomaly in Spitsbergen at the SSB (Grasby et al., 2013a, 2016a). Mercury anomalies across the EPME are also observed in both shallow and deep-water records in South China (Grasby et al., 2017; Wang et al., 2018). These studies demonstrated that the Hg enrichments are associated with the STLIP and that deep-water marine facies can more directly record the atmospheric-derived Hg signature from volcanism than near shore shallow marine environments. While studies of Hg chemostratigraphy have been used to suggest global impact of STLIP volcanism through the EPME and SSB, these have been largely restricted to boreal records. The global impact of these eruptions requires examination of southern latitude records to determine if observed Hg spikes are truly global. To further explore the spatio-temporal impact of Hg on the end Permian to the SSB ecosystems, and its possible link to the STLIP, we selected sections from South China and northern India, and determined Hg/TOC values and Hg isotopic composition to trace oceanic Hg levels as well as pathways into the ocean through this time interval. The variety of rock types and sedimentary facies in the two sections enable an assessment of lithology and sedimentary environment on measured values. Previous studies have focused on the Hg record from the Boreal realm across the EPME. The two regions we report on here were located significantly further away from the STLIP, which was located in the northern hemisphere at that time. Occupying paleogeographic positions that were equatorial (South China) and high southern (India) paleo-latitudes (Fig. 1), these records provide important new insight into the extent of global Hg loading to the environment. 2. Geologic setting 2.1. Chaohu, South China The Chaohu area is located in the northeast corner of the South China Block. During the Late Permian to Early Triassic the region occupied a low latitude location within the Paleo-Tethys ocean, and represents deposition in a wide range of relatively deep-water settings at water depths of 200–400 m, ranging from an epicontinental basin in the Late Permian to an inner and outer ramp in the Early Triassic (Fig. 1; Chen et al., 2011). We integrated two sections from this area; the West Pingdingshan section that is a candidate stratotype for the Induan-Olenekian boundary of

146

X. Wang et al. / Earth and Planetary Science Letters 513 (2019) 144–155

Fig. 1. (A) Early Triassic paleogeography of the World (base map courtesy of Ron Blakey (http://www2.nau.edu/rcb7/240moll.jpg)), (B) South China (after Huang et al., 2017), (C) Location map for Kashmir area (white area). (D) Geological map of the study area (modified from Tewari et al., 2015), showing the Guryul Ravine section.

the Lower Triassic (Zhao et al., 2002, 2004, 2005), and the Majiashan section. The west Pingdingshan section contains lower Griesbachian to lower Spathian strata, which in ascending order are the Yingken Formation, consisting of interbedded calcareous mudstone and argillaceous limestone, the Helongshan Formation, composed of mudstone and argillaceous limestone, and the lower Nanlinghu Formation, which is dominated by limestone (Fig. 2). The Induan– Olenekian boundary (IOB) lies at the very top of Bed 24 (Zhao et al., 2007, 2008), in the Yingken Formation, based on the first appearance of the conodont element Neospathodus (Ns). Waageni (Zhao et al., 2007; Lyu et al., 2018). The SSB lies 60 cm from the bottom of Bed 52 (upper Helongshan Formation), assigned on the first occurrence of Ns. pingdingshanensis (Fig. 2; Zhao et al., 2008). The Majiashan section includes the uppermost Permian Dalong Formation, dominated by siliceous mudstone, and the lowermost Triassic Yingken Formation, dominated by calcareous mudstone. 2.2. Guryul Ravine section, Kashmir, northern India The Guryul Ravine section near Srinagar in Indian Kashmir (Fig. 1) was a candidate Global Stratotype of the Permian-Triassic boundary (PTB) (Brookfield et al., 2003). During the Late Permian to Early Triassic, the Guryul Ravine section was located at a high southern latitude on the northern margin of Gondwana, with strata accumulating in a carbonate ramp setting at water depths of ∼50–100 m (Fig. 1; Brookfield et al., 2003). Strata exposed at this section belong to the Zawan and Khunamuh formations and record a time interval spanning from the Late Permian Wuchiapingian stage to the Early Triassic Spathian sub-stage (Baud et al., 1996). The Zewan Formation is about 10 m thick and is subdivided into four units, A to D (Nakazawa et al., 1975). The Khunamuh Formation is >100 m thick and is divided into six units, E to J, on the basis of variations in carbonate content (Baud et al., 1996). Units D to I were sampled in the current project (Fig. 3). Unit D is dominated by sandy limestone and marl, and yield abundant gastropods and bivalves (Tewari et al., 2015). The transitional Unit E1 (2.5 m thick) is characterized by greenish shale with interbeds of argilla-

ceous limestone. Unit E2 (6.1 m thick) is composed of calcareous mudstone with thin-bedded limestone and marl. Unit E3 (9.9 m thick) and Unit F (20 m thick) are lithologically similar to Unit E2. Units G to I are dominated by medium-bedded to massive limestone. Conodont biostratigraphy across the PTB of the Guryul Ravine section indicates that the element Hindeodus (H.) parvus first occurs 40 cm above the base of unit E2 (Sweet, 1970; Nakazawa et al., 1975), corresponding to the first appearance of Lower Triassic ammonoids, including the genera Otoceras, Lytophiceras and Glyptophiceras (Bando, 1981). Recently, Brosse et al. (2017) restudied the Griesbachian conodont biostratigraphy and established Isarcicella (I.) isarcica Zone, Neoclarkina (Nc.) krystyni Zone, and Sweetospathodus (Sw.) kummeli Zone from the upper Unit E2 to Unit E3, and pointed out that the conodont specimens previously assigned to I. isarsica by Sweet (1970) should be re-assigned to I. staeschei. Here we integrate the previously reported conodont zonations and tentatively assign Unit D to the late Changhsingian Clarkina (C .) changxingensis Zone and Unit E1 to the latest Changhsingian Hindeodus (H .) praeparvus–C. meishanensis Zone. Calibrations of conodont zones from Units E2–E3 are as here: H. parvus Zone (Beds 52–53), I. staeschei Zone (Beds 54–56), I. isarcica Zone (Beds 57–63), Nc. krystyni Zone (Beds 64–68), and Sw. kummeli Zone (starting at Bed 70) (Fig. 3). The PTB, marked by the first occurrence of H. parvus, is placed at the base of Bed 52 (also the base of Unit E2). 3. Methods Samples were collected from outcrop with weathered surfaces first removed. In the laboratory any remaining weathered surfaces were removed and then samples were powdered in a freeze grinding apparatus. Splits from a homogenized powder were then taken for subsequent analyses. All geochemical analyses were carried out at the State Key Laboratory of Geological Processes and Mineral Resources (GPMR), China University of Geosciences (CUG), Wuhan,

X. Wang et al. / Earth and Planetary Science Letters 513 (2019) 144–155

147

Fig. 2. Lithology, conodont zones (Zhao et al., 2008), carbon isotopic stratigraphy (inorganic carbon isotope from Tong et al., 2007), TOC contents, Hg concentrations and Hg/TOC ratios, Hg isotopic compositions (199 Hg values), clay minerals input index (Al contents)and ocean redox conditions (Mo/Al ratios) at the Chaohu (CH) section. Fm. = Formation; Ns. = Neospathodus; Ng. = Neogondolella; H. = Hindeodus; SSB = Smithian-Spathian boundary; EPME = end Permian mass extinction horizon; Spa. = Spathian; Nlh = Nanlinghu; Perm. = Permian; Ch. = Changhsingian; Dal. = Dalong. The dotted line at 200 ppb/wt. % of Hg/TOC corresponds with the maximum value in the Early Triassic. Beds 0–11 are from Majiashan and above Beds are from Pingdingshan.

except for Hg isotopes, which were analyzed at the Institute of Geochemistry, Chinese Academy of Sciences, Guiyang, China. 3.1. Major and trace elements analysis Whole-rock major element analysis was performed via dissolution by Atomic Fluorescence Spectrometer at CUG. Before analysis, pre-ignition was used to determine the loss on ignition. Uncertainty was generally less than 5%. Trace elements were analyzed at SKLGPMR of CUG with ICP–MS (Agilent 7500a). Blank and national standards AGV-2, BHVO-2, BCR-2, and RGM-2 were selected for calibrating element concentrations. Descriptions of the sample dissolution procedure, the analytical process, and the accuracy of analyses are outlined by Liu et al. (2008). 3.2. Carbon isotope analysis Analyses were conducted for organic carbon isotopes on the siliceous mudstones from the Majiashan section and for inorganic carbon isotope on the carbonates from the Guryul Ravine section. For organic carbon isotopes, samples were first acid washed to remove any inorganic component. Then ∼ 0.5–0.8 mg of powder was added to a clean tin cup and analyzed using a MAT 253 IR-MS in-

terfaced with a Flash EA 2000 auto-sampler. δ 13 Corg results are reported relative to the Vienna Pee Dee belemnite (V-PDB) standard with a precision better than ±0.1h based on repeated analyses of national standard GBW04407 (δ 13 C = −22.4h). For inorganic carbon-isotope analyses, 80–120 mg of limestone or 180–200 mg of calcareous mudstone powder was placed in a 10 ml Na glass vial and was reacted with 100% phosphoric acid at 72 ◦ C. δ 13 Ccarb was analyzed from the generated CO2 with a MAT 253 mass spectrometer. The analytical precision is better than ±0.10h according to the repeated analysis of the reference standard GBW-04416 (δ 13 C = 1.61h). 3.3. Total organic carbon measurement Total organic carbon (TOC) analyses were conducted by an Elementar various micro cube analyzer. About 10 g of sample powder was placed into the 50 mL tube, then injected with 50% HCl to dissolve any carbonate minerals. After multiple centrifugal and lyophilization steps, the residue was weighed and powdered for later TOC measurement. Data reliability was assessed through measurement of a standard sample DP-1 (65.44%). Both a standard sample and a repeat were analyzed after every 12 unknowns, yielding an analytical accuracy of 2.5% of the reported values.

148

X. Wang et al. / Earth and Planetary Science Letters 513 (2019) 144–155

Fig. 3. Lithology, condont zones (Sweet, 1970; Nakazawa et al., 1975; Brosse et al., 2017), carbon isotopic stratigraphy, TOC content, Hg concentration, Hg/TOC ratios, Hg isotopic compositions (199 Hg values), clay minerals input index (Al contents) and ocean redox conditions (Mo/Al ratios) at the Guryul Ravine section. Sub-s. = Sub-stage; Fm. = Formation; Sw. = Sweetospathodus; Nc. = Neoclarkina; I. = Isarcicella; H. = Hindeodus; C. = Clarkina; H. prae. = Hindeodus praeparvus; C. mei. = Clarkina meishanensis; SSB = Smithian-Spathian boundary; EPME = end Permian mass extinction horizon; Gries. = Griesbachian; Ch. = Changhsingian. The dotted line at 200 ppb/wt. % of Hg/TOC corresponds with the maximum value in the Early Triassic.

3.4. Mercury (Hg) analysis Mercury concentration was measured using a LECO AMA254 mercury analyzer. All the samples were freeze-dried to prevent the decomposition of Hg. About 100 mg for mudstone or shales and 150–200 mg for limestone were analyzed. Data reliability was ensured by analysis of international standard coal sample 502–685 (40 ppb) after every 12 unknowns, yielding reproducibility of sample concentrations being within 10%. 3.5. Hg isotopic composition For samples containing more than 10 ppb Hg, sample preparation and dissolution procedure follow the methods described previously (Grasby et al., 2017; Wang et al., 2018). Approximately 1 g of homogenized powder was digested with 12 mL of aqua regia (HCl:HNO3 = 3:1, v-v) in a 50 mL centrifuge tube and heated in water at 95 ◦ C for 3 h with repeated shaking. The supernatants were collected after centrifuging, followed by filtration. Before conducting Hg isotope analysis, the supernatants were diluted by adding Milli-Q water to Hg concentrations of ∼2 ng/mL or 1 ng/mL in solutions with acid concentration of <20%. The pyrolysis method was used to extract Hg from the samples that have low level Hg (less than 10 ppb). About 4 g powder was heated at 950 ◦ C for ∼4 h in a thermal decomposition furnace with oxygen injected for blowing the decomposed gaseous Hg0 into a absorption bottle, in which 6 mL 40% versa aqua regia (HCl:HNO3 = 1:3, v-v) was loaded in order to oxidize Hg0 to Hg2+

and reserve Hg2+ . To prevent photochemical reduction of mercury, the absorption bottle was encased in tinfoil. Standard reference materials (GSS-4, soil) were also prepared during both acid and pyrolysis dissolution procedures to ensure that no additional Hg isotopic fractionation occurred in the pretreatment processes. Hg isotopic ratios were determined using a Nu-Plasma multi-collector inductively coupled plasma mass spectrometer (MC-ICP-MS) with high sensitivity X skimmer cone. An international standard NIST SRM 997 Tl was used for simultaneous instrumental mass bias correction of Hg and 4 ng/mL SnCl2 solution was used to generate elemental Hg0 before being introduced into the plasma. International standard NIST SRM 3133 was measured after every 3 unknowns to monitor the stability of the instrument. We also analyzed NIST SRM 3177 after every 10 unknowns to examine the instrument accuracy. Hg concentrations of ∼2 ng/mL or 1 ng/mL of NIST SRM 3133 and NIST SRM 3177 solutions were prepared for matching measured sample solutions to reduce the matrix dependent mass bias. Hg isotopic composition is reported in δ 202 Hg notation in units of permil (h) referenced to the NIST SRM 3133 Hg standard:

δ 202 Hg (h)      = 202 Hg/198 Hgsample / 202 Hg/198 HgNIST SRM 3133 − 1 × 1000

(1)

X. Wang et al. / Earth and Planetary Science Letters 513 (2019) 144–155

MIF of Hg isotopes is expressed in  notation (xxx Hg) in units of permil (h), describing the difference between the measured δ xxx Hg and the theoretically predicted δ xxx Hg value:

  199 Hg ≈ δ 199 Hg − δ 202 Hg × 0.2520   200 Hg ≈ δ 200 Hg − δ 202 Hg × 0.5024   201 Hg ≈ δ 201 Hg − δ 202 Hg × 0.7520

149

(2)

negative values occurring at the PTB. This trend is typical of that seen globally, with negative shifts in isotope values on the order of ∼3h to ∼4h (Luo et al., 2014; Grasby et al., 2013b, 2015) at the PTB. The negative δ 13 Corg excursion calibrates the EPME horizon to ∼60 cm below the base of the Yinkeng Formation (Fig. 2).

(3)

4.4. Inorganic carbon isotope in the Guryul Ravine section

(4)

Aluminum (Al) is considered as a classical proxy of terrigenous fluxes in carbonates due to its incorporation into aluminosilicate minerals such as clays and its stability during transport and diagenesis (Shi et al., 2016). The Al content varies from 0.67 to 9.99 wt. % through the Chaohu section (Fig. 2; Table S1), yielding high values around the EPME (3.4–9.5 wt. %), at Beds 23–25 (1.8–9.9 wt. %) and Beds 29–36 (6.4–9.5 wt. %) and Beds 51–53 (2.3–9.1 wt. %) (Fig. 2). Other horizons have relatively low Al contents, mostly less than 5.0 wt. % (Fig. 2). Compared to Chaohu, the Guryul Ravine section generally has lower Al contents, mostly less than 4.0 wt. % except for four samples in Unit D and three samples in Units G–H, where relatively high values are recorded (4.04–9.58 wt. %) (Fig. 3; Table S2).

Values of δ 13 Ccarb obtained from the Guryul Ravine section, Kashmir, vary from −3.9 to 7.1h and show two negative and two positive excursions (Fig. 3; Table S4); from which we define four carbon isotope intervals. In interval I, δ 13 Ccarb decreases to a minimum of 3.85h at N1, corresponding with the lower isarcica zone ∼6 m above the EPME and ∼1 m above the PTB. The negative excursion of interval I is comparable with the main Cisotopic shift in other PTB sections (Payne et al., 2004; Korte and Kozur, 2010), but also displays a regional component. First, the onset of the negative shift in the study section began in the C. changxingensis Zone ∼2.5 m below the EPME, earlier than C. meishanensis zone in the South China and Iranian PTB sections. Second, the values of δ 13 Ccarb in interval I are ∼2h lower than those in the South Chinese and Iranian PTB sections (Payne et al., 2004; Algeo et al., 2007). Those signals may indicate that regional events have enhanced the C-isotope negative shift. Algeo et al. (2007) ascribed the negative shift to a large sea-level rise commencing approximately 3 m below the EPME. In interval II, the δ 13 Ccarb starts a progressive positive shift, achieving a maximum at P1 that corresponds to the IOB (Fig. 3). The positive excursion of interval II correlates with that in the South China sections (Payne et al., 2004; Tong et al., 2007) as well as those at the Smithian Stratotype (Grasby et al., 2013b). Interval III is characterized by δ 13 Ccarb values decreasing to a minimum at N2 (Fig. 3), and interval IV is distinguished by δ 13 Ccarb values increasing to a maximum at P2. In other global localities, δ 13 C values usually display a positive carbon isotope excursion near the SSB (Payne et al., 2004; Tong et al., 2007; Galfetti et al., 2007; Grasby et al., 2013b; Zhang et al., 2015). Given this, we infer that the SSB in the Guryul Ravine section may be at, or near, P2.

4.2. Molybdenum/aluminum (Mo/Al)

4.5. TOC contents

Enrichments of molybdenum (Mo) in sediments only occur when a low level of oxygen and a high level of H2 S are present in seawater (Takahashi et al., 2014). Mo/Al is used to evaluate the input of terrestrial Mo and is widely used as a redox index (Grasby et al., 2013a, 2016a). Ratios of Mo/Al vary from 0.02 to 1.42 ppm/wt. % through the Chaohu section with two positive shifts occurring at the EPME and just prior to the SSB at Beds 50–52. In other horizons, Mo/Al is relatively low (mostly less than 0.1 ppm/wt. %), except for Beds 23–25 where a minor positive shift is observed (0.05 to 0.30 ppm/wt. %) (Fig. 2; Table S1). Mo/Al ratios in Guryul Ravine display relatively high values, but with a gradually declining trend in the uppermost Changhsingian Unit D. The highest Mo/Al values (up to 4.8 ppm/wt. %) occur between the EPME and the lower I. isarcica conodont zone, before declining and maintaining low values until the uppermost part of Unit F that yields a secondary positive shift (up to 0.75 ppm/wt. %) (Fig. 3; Table S2).

In Chaohu, TOC contents have relatively high values in the Dalong Formation and near the SSB, at Beds 51–53, with TOC up to 1.14% in the former and 1.56% in the latter (Fig. 2; Table S3). Moreover, slightly higher values occur in Beds 31–37 (Fig. 2) and correspond with preservation of abundant organic matter in the calcareous mudstone and shale present in this part of the section. In other horizons, TOC is negligible, generally <0.1%. Throughout the Guryul Ravine section TOC is low, usually less than 1% (Fig. 3; Table S4). Near the PTB (Unit D, Unit E1 and the bottom of E2), TOC concentrations are higher than other beds (Fig. 3), possibly correlating with the high proportion of shale in units D, E1 and E2.

Standard reference GSS-4 during acid and pyrolysis dissolution procedures produce results (n = 2) of δ 202 Hg = −1.53 ± 0.24h (2 sd); 199 Hg = −0.29 ± 0.02h (2 sd); 200 Hg = −0.02 ± 0.03h (2 sd); 201 Hg = −0.3 ± 0.02h (2 sd), and δ 202 Hg = −1.40 ± 0.08h (2 sd); 199 Hg = −0.28 ± 0.04h (2 sd); 200 Hg = 0.00 ± 0.04h (2 sd); 201 Hg = −0.27 ± 0.03h (2 sd), respectively, which all are consistent with the results reported by Estrade et al. (2009). Replicate analysis of the NIST 3177 Hg intra lab isotope reference standard (n = 5 analytical sessions) were as follows: δ 202 Hg = −0.50 ± 0.07h (2 sd); 199 Hg = −0.02 ± 0.02h (2 sd); 200 Hg = −0.01 ± 0.04h (2 sd); 201 Hg = −0.02 ± 0.02h (2 sd). 4. Results 4.1. Aluminum (Al) contents

4.3. Organic carbon isotope composition in the Majiashan, Chaohu section In total 36 samples were analyzed for δ 13 Corg across the Permian-Triassic boundary (PTB) in the Chaohu section, South China. Values of δ 13 Corg vary from −22.8 to −26.3h (Fig. 2; Table S3), with rapid and large shifts from the EPME to the most

4.6. Hg concentration and Hg/TOC ratios Though both the contents of Hg and TOC have large variance, the residuals between them are in range from −2 to 2 at both the Chaohu and Guryul Ravine sections (Figs. S1, S2), indicating our values are reliable and the Hg/TOC ratios can be used to assess if variation in Hg concentrations is driven by lithologically controlled changes in organic matter content. Hg concentrations vary from 0.6 to 199.1 ppb in the Chaohu section, with three sets of high values occurring near the PTB, the SSB, and in the middle Smithian (Fig. 2; Table S3). TOC values are also high in these horizons. Positive Hg/TOC excursions were observed near the EPME

150

X. Wang et al. / Earth and Planetary Science Letters 513 (2019) 144–155

horizon, with a maximum of ∼440 ppb/ wt. % at the EPME horizon. A second Hg/TOC (∼350 ppb/wt. %) peak occurs in the earliest Triassic. Hg/TOC falls to low levels in the lower Griesbachian, with only minor oscillations higher in the section (Fig. 2; Table S3). The Guryul Ravine section records a short-lived positive shift in Hg concentrations, and abrupt positive excursion in Hg/TOC ratios coincident with the PTB (Fig. 3). Hg concentrations vary from 1.2 ppb to 64.0 ppb through the section with positive peaks between the EPME and the base of I. isarcica zone. Hg concentrations generally fall to low levels above the I. isarcica zone except for there being some sporadic high Hg values that are associated with high TOC contents (Fig. 3; Table S4). Hg/TOC values vary in the range of 10 to 450 ppb/wt. % throughout the section. In Unit D, Hg/TOC values remained below 200 ppb/wt. %, whereas in the C. meishanensis zone and at the base of the I. isarcica zone, Hg/TOC values are distinguished by positive shifts to a maximum of ∼450 ppb/wt. %. Higher in the section, Hg/TOC return to pre-extinction levels and maintain low values (<200 ppb/wt. %; Fig. 3; Table S4). 4.7. Hg isotopic compositions The detailed analytical results of Hg isotope values for the Chaohu and Guryul Ravine sections are provided in Online Supplementary Tables S3 and S4. Here we focus on Hg-MIF signature (199 Hg) because of its greater reliability in tracing the pathway of the Hg source (Thibodeau et al., 2016). 199 Hg values vary from −0.09 ± 0.05h (2sd) to 0.33 ± 0.01h (2 sd) though the Chaohu section, yielding the most positive values (0.08 ± 0.02h to 0.33 ± 0.01h) at the EPME, and yielding low values (∼0h) in the Induan stage. Elevated values occur near the IOB and remain positive (0.05 ± 0.09h to 0.16 ± 0.04h) until near the SSB (Beds 51–54), where negative signatures were measured (−0.09 ± 0.05h to 0.08 ± 0.02h) (Fig. 2). Guryul Ravine displays a similar trend in 199 Hg values to Chaohu (Figs. 2, 3), with sustained positive values (0.09 ± 0.01h to 0.14 ± 0.01h) in the Late Changhsingian and at the EPME. Similarly, the MIF (199 Hg) show a decrease in the I. staeschei conodont zone and maintain lower values (∼0h) through the early Triassic except for three sporadic high positive values (Fig. 3). 5. Discussion 5.1. Hg cycle during the EPME and subsequent Early Triassic recovery Hg concentrations are typically strongly linked with TOC values. Hence, high Hg concentrations along with high Hg/TOC ratios recorded in sediments may point to increased Hg input being derived from an external source such as volcanism (Sanei et al., 2012; Grasby et al., 2013a; Percival et al., 2017). However, Hg concentration is strongly influenced by lithology due to varied TOC content in different lithological samples. Hg can also be associated with sulfides and clays, although considered less important. Hence, Hg/TOC ratios may be sensitive to local or regional factors, such as lithology or terrestrial clay input and oceanic redox conditions (e.g. Sanei et al., 2012; Percival et al., 2018). Here we assess possible controls on the measured Hg/TOC values in our records, including Hg sources and sinks, and examine the Hg cycle during the EPME as well as the subsequent Early Triassic recovery. 5.1.1. Global Hg deposition event across the EPME The highest Hg concentrations and largest Hg/TOC anomalies were observed around the EPME in both the Chaohu and Guryul Ravine sections. This is similar to Hg records across the EPME in the Sverdrup Basin, located close to the STLIP (Sanei et al., 2012; Grasby et al., 2013a, 2016a, 2017), as well as those observed in South China from shallow and deep depositional environments

(Grasby et al., 2017; Wang et al., 2018). When examining our new data along with these previous publications, the sections provide a more global perspective, from the high southern palaeolatitude, to close to the palaeoequator, to high northern palaeolatitude. We are able to demonstrate that the high Hg levels around the EPME, first reported in the Sverdrup Basin (Sanei et al., 2012), reflect a truly globally significant spike in Hg deposition. The Hg enrichments in South China and the Sverdrup Basin do not show any correlation with lithology, redox conditions, and variations in sedimentation rate during that time, and are interpreted to be associated with STLIP activity (Grasby et al., 2013a, 2017; Wang et al., 2018). In many sections though, the EPME is marked by significant changes in lithology, weathering input (Algeo and Twitchett, 2010), and anoxic conditions (Wignall and Twitchett, 1996). The interplay of these changes on the Hg record needs to be assessed before volcanic impact on the EPME can be elucidated. To further test the hypothesis that the global Hg spike across the EMPE is related to STLIP activity, we evaluate the possible causes of the Hg anomalies in the sections of this study. We put particular focus on the Guryul Ravine section, which is the first reported Southern Hemisphere Hg record across the EPME, to assess if it also reflects a volcanic source. High Hg levels are recorded in both shale and argillaceous limestone across the EPME, but other argillaceous limestone horizons have low Hg levels at the Guryul Ravine section (Fig. 3), indicating that lithology is not the cause of Hg anomalies. Changing inputs of aluminous phyllosilicate clay minerals as a driver for changes in Hg values can also be negated on the basis of Al data through the section. Low Al content was recorded around the EPME in Guryul Ravine (Fig. 3), implying that there is no connection between the Hg anomalies and the input of clay minerals. There is also an absence of correlation between Hg and Al for the EPME samples (Fig. 4E). High Mo/Al ratios are observed around the EPME at both the Chaohu and Guryul Ravine sections (Figs. 2, 3), suggesting the EPME was associated with an anoxic environment of deposition. Anoxia does not seem to be the driver of enhanced Hg deposition though, because high Mo/Al ratios with no associated Hg spikes are observed in other horizons, such as near the IOB and SSB in Chaohu, and the uppermost part of Unit F in Guryul Ravine (Figs. 2, 3). No correlation are observed in plots of Hg/TOC versus Mo/Al of the studied sections (Figs. 4G, H), which is similar to Late Permian to Early Triassic Hg records in the Sverdrup Basin and Spitsbergen (Fig. 4I; Grasby et al., 2013a, 2016a). These results imply that anoxia has no direct influence on Hg sequestration in sediments because the normally strong Hg-OM association inhibits highly insoluble Hg sulfide from precipitating in marine sediments. Such HgS drawdown is thought to only be significant when excess Hg in seawater exceeds the capacity of organic matter drawdown (Sanei et al., 2012). The Hg stable isotope data can help to further elucidate the origin of the EPME Hg anomalies, because MIF (199 Hg) is thought to only occur in photochemical reactions in surface environments (Blum et al., 2014). The Chaohu and Guryul Ravine sections both have the most positive 199 Hg values (∼0.15h) at the EPME horizon, indicating that sediments in these sections received Hg primarily through atmospheric Hg deposition instead of terrestrial runoff. Thus, the observed Hg peaks at the EPME are consistent with elevated volcanic activity and associated Hg flux to the atmosphere (Grasby et al., 2017; Wang et al., 2018). However, the 199 Hg values decreased to ∼0.0h at the second Hg peak horizon just above the EPME, in the earliest Triassic (Figs. 2, 3), indicating that a higher proportion of terrestrial Hg was imparted to the sediments at that time. This profile could be explained by enhanced weathering and continental denudation that was suggested to have occurred during the Early Triassic (Algeo and Twitchett, 2010). Enhanced weathering at this time is fur-

X. Wang et al. / Earth and Planetary Science Letters 513 (2019) 144–155

151

Fig. 4. (A, B, C) Plots of Hg concentration versus TOC content at the Chaohu, Guryul Ravine and Spitsbergen sections, respectively; (D, E, F) Plots of Hg concentration versus Al content at the Chaohu, Guryul Ravine and Spitsbergen sections, respectively; (G, H, I) Plots of Hg/TOC ratio versus Mo/Al ratio at the Chaohu, Guryul Ravine and Spitsbergen, respectively. Data of Spitsbergen are from Grasby et al. (2016a).

ther supported by elevated Al contents at the second Hg peak horizon in both Chaohu and Guryul Ravine (Figs. 2, 3). It is also possible that sediments received volcanic Hg through both atmospheric deposition as well as Hg from terrestrial runoff, with the Hg-MIF signatures of the terrestrial source (199 Hg ≈ 0.0h; Thibodeau et al., 2016) overwhelming that of the volcanic signature. Our results confirm the previous hypothesis of Sanei et al. (2012) that eruption of the Siberian Traps had an associated significant enhanced flux of Hg to the atmosphere, which Grasby et al. (2015) estimate to be as great as 10,000 Mg/a (∼14× increase over background). Our findings demonstrate that this Hg flux was truly global, with an interhemispheric distribution that left a Hg signature in the world’s oceans from high southern to high northern latitudes. These findings further support the hypothesis that not only does Hg make an effective stratigraphic marker of LIP events (rather than just relatively close to a LIP centre), but could have also exerted a globally significant toxic shock on marine ecosystems (Grasby et al., 2015) given its persistence in the environment.

5.1.2. Hg deposition in the Early Triassic Throughout the Early Triassic, Hg/TOC values are low (<200 ppb/wt. %) relative to the EPME anomalies in both the Chaohu and Guryul Ravine sections, implying that this represents normal or background Hg deposition. Hg/TOC values are also lower around the SSB relative to the spikes near the EPME. However, the Hg/TOC values throughout the Early Triassic in our sections display frequent minor fluctuations. We assess the cause of these fluctuations below. Plots of Hg versus TOC for the Early Triassic limestone samples from the two studied sections display positive correlations (Figs. 4A, B), indicating again that organic matter exerts an important role in sequestration of Hg. The Early Triassic also witnessed an increased flux of terrestrial material to the oceans owing to accelerated rates of continental weathering (Algeo and Twitchett, 2010). Plots of Hg versus Al for Early Triassic samples at both Chaohu and Guryul Ravine also yield positive correlations (Figs. 4D, E), suggesting that significant terrestrial Hg was transferred to the ocean along with clay minerals. This interpretation is reinforced by Hg isotopic data. As marine sediments, can receive

152

X. Wang et al. / Earth and Planetary Science Letters 513 (2019) 144–155

Fig. 5. 5 Global correlation of mercury deposition through the Late Permian to Early Triassic. Hg/TOC values of the Sverdrup Basin and Spitsbergen are from Grasby et al. (2013a, 2016a). Ages of the EPME, and the IOB and SSB are from Burgess et al. (2014), and Galfetti et al. (2007), respectively. Gries. = Griesbachian; Ch. = Changhsingian.

Hg through both the atmosphere and runoff, 199 Hg values obtained in samples can be expressed as: ∗ 199 Hgsediments = f atmospheric 199 Hgatmospheric ∗ + f terrestrial 199 Hgterrestrial

f atmospheric + f terrestrial = 1 199

(5) (6)

199

where  Hgsediments is the  Hg values obtained in the samples. 199 Hgatmospheric and 199 Hgterrestrial are the Hg isotopic compositions of atmospheric-derived Hg and terrestrialderived Hg, whereas f atmospheric and f terrestrial are the fraction of atmospheric-derived Hg and terrestrial-derived Hg, respectively. As 199 Hgatmospheric and 199 Hgterrestrial values are not known, we are limited to a qualitative discussion. Compared to the latest Permian, the Early Triassic samples have less positive 199 Hg values. For instance, Chaohu has 199 Hg values of 0.08 ± 0.02h to 0.33 ± 0.01h in the latest Permian, but yields values of −0.09 ± 0.05h to 0.16 ± 0.04h in the Early Triassic (Fig. 2). The Guryul Ravine depositional environment was shallower than that at Chaohu, and 199 Hg display sustained positive values (0.09 ± 0.01h to 0.14 ± 0.01h) in the Late Changhsingian, and show generally lower values (∼0h) through the Early Triassic. These results indicate that the fraction of terrestrial-derived Hg ( f terrestrial ) increased from the latest Permian to Early Triassic time. The Early Triassic strata of the studied sections are dominated by argillaceous limestone or limestone. These strata have low Hg (mostly less than 10 ppb) and TOC (mostly less than 0.2%) contents, except for mudstone at Beds

32–37 and at SSB Beds 51–52 at Chaohu, where relatively high Hg (up to ∼100 ppb) and TOC (up to ∼2.0 wt. %) are recorded (Figs. 2, 3). For the sediments that have low Hg and TOC contents, receiving quantitatively changed terrestrial Hg can result in significantly varied Hg/TOC ratios. Therefore, the frequent minor fluctuations in Hg levels in the studied sections are more likely to result from terrestrial signals rather than reflecting changing atmospheric Hg levels at that time. The overall low values suggest a normal or background Hg deposition rate through the Early Triassic. This inference is reinforced by the almost constant background Hg/TOC values through most of the Early Triassic in the Sverdrup Basin, Canada (Fig. 5; Grasby et al., 2013a). Despite the overall trend of background Hg levels, studies from the Arctic region nearer the STLIP reveal that there is a small Hg peak around the SSB in Spitsbergen, and there is a slight shift to higher Hg/TOC ratios in the late Smithian in the Sverdrup Basin (Fig. 5; Grasby et al., 2013a, 2016a). Compared to the large Hg peak (up to 1000 ppb/wt. % in the Sverdrup Basin) at the EPME, the elevated Hg/TOC ratios around the SSB are minor with values of less than 200 ppb/wt. %, which is within the range that Hg/TOC values fluctuate in the Early Triassic samples in sections studied here (Fig. 5). There are two possible explanations for these patterns. One is that the Siberian Traps may have reactivated during this period, but the magnitude of activity was considerably reduced, and thus had only limited effects on the Hg cycle in the northern hemisphere as recorded in the Spitzbergen section. Another possibility is that the elevated Hg/TOC values were due to

X. Wang et al. / Earth and Planetary Science Letters 513 (2019) 144–155

153

Fig. 6. Integrated excursions of Hg/TOC ratios of the Chaohu (CH), Guryul Ravine (GR), Sverdrup Basin (SB; Grasby et al., 2013a) and Spitsbergen (SP; Grasby et al., 2016a) sections, δ 13 Ccarb values (Payne et al., 2004) and sea surface temperature (Sun et al., 2012) in South China from the Changhsingian to Spathian sub-stage, and biodiversities of conodonts and ammonoids during the Early Triassic (Stanley, 2009).

local effects other than volcanism, as this is not the only possible driver for the increased sedimentary Hg budget during environmental crises. For example, other inputs (e.g., terrestrial Hg) and outputs (e.g., decreased Hg sequestration on organic matter controlled by changes in primary productivity), can result in elevated Hg/TOC values in sediments (Grasby et al., 2017; Wang et al., 2018; Them et al., 2019; Percival et al., 2018). The Late Smithian was also an interval during which climate and environments underwent remarkable change, including a thermal maximum (Fig. 6; Sun et al., 2012), increased continental weathering (Zhang et al., 2015), ocean anoxia, and productivity crises in the Arctic region (Grasby et al., 2013b, 2016b). Additional data, such as Hg isotopic evidence and more detailed chronology of the STLIP, are needed to identify the cause of the elevated Hg/TOC ratios at the SSB in these Arctic sections.

5.2. STLIP, background Hg deposition, and protracted Early Triassic recovery Following the EPME, marine ecosystems experienced marked variations during the Early Triassic, as documented by both geochemical and fossil records. For example, marine sea surface temperatures were lethally hot and displayed a maximum in the latest Smithian followed by abrupt oscillations to a cooling event slightly prior to the SSB (Fig. 6; Sun et al., 2012). Biodiversity of ammonoids and conodonts showed rapid radiation in the early and middle Smithian, but underwent a severe loss in the latest Smithian, followed by a stepwise increase in the early Spathian (Fig. 6; Orchard, 2007; Stanley, 2009). Sizes of conodonts also display the Lilliput effect at the SSB, before return to normal size in the early Spathian (Chen et al., 2013). Significantly, anoxic conditions occur repetitively during the Early Tri-

154

X. Wang et al. / Earth and Planetary Science Letters 513 (2019) 144–155

assic (Grasby et al., 2013b; Huang et al., 2017). The Early Triassic interval also witnessed several negative carbon isotope excursions that coincided with climatic upheavals (e.g., late Smithian thermal maximum) and biodiversity drops (Fig. 6; Payne et al., 2004; Chen and Benton, 2012). The causes of these variations in the marine ecosystem in the Triassic are unknown, but have been linked to potentially renewed Siberian volcanism (Sun et al., 2012; Grasby et al., 2013b), as there is evidence that STLIP magmatism continued into the Middle Triassic (Ivanov et al., 2013). Recent high precision U–Pb age data high precision U–Pb age data establishes that STLIP magmatism continued at least until 251.4 Ma (Burgess and Bowring, 2015), which is later than the end of I. isarica conodont zone in age. However, positive Hg/TOC excursions only occurred during the C. meishanensis to lower I. isarica conodont zones (this study; Wang et al., 2018). This phenomenon raises two possibilities to explain the background Hg deposition in the Early Triassic: 1) STLIP related activity in the Early Triassic was relatively weak, and thus did not release enough Hg to influence the global Hg cycle. If this is the case, then there may be other primary drivers for the delayed biotic recovery. For instance, long-term astronomical forcing has been suggested to be involved in the repeated climatic and biotic upheavals that took place throughout the Early Triassic (Li et al., 2016). Another possibility is the development of nutrient limited oceans related to extreme heating and depression of the marine nutricline (Grasby et al., 2016b). 2) LIPs have differing styles of volcanism over their lifetime, and not all of them may necessarily release enough Hg to be preserved in the geologic record. For example, Percival et al. (2018) recently measured mercury records corresponding to two Cretaceous LIP events, one of which is submarine LIP, but both do not show geographically consistent signal of sedimentary Hg enrichment. Several factors could influence the potential impact of LIP eruptions on the geological Hg record, such as volcanic intensity, magmatic vent, explosivity, and the potential contribution of thermogenic Hg from reactions between the magma and country rocks (Percival et al., 2018). The STLIP eruption was divided into three stages, including initial pyroclastic eruptions followed by lava effusion, then medium-term sill intrusion, and then the terminal stage of both extrusive and intrusive magmatism, but dominated by extrusive lava (Burgess et al., 2017). The general elevation of Hg/TOC values during the C. meishanensis to lower I. isarica conodont zones is consistent with medium-term sill intrusion into the thick sediments in the Tunguska Basin, and the highest values at the EPME corresponds to the onset of widespread sill-complex intrusion (Wang et al., 2018). Thus, if the STLIP was still responsible for the protracted recovery in the Early Triassic, perhaps only pulses of sill intrusion into the thick sediment sequences were able to influence the global Hg cycle during the EPME. Other factors, such as lack of explosivity and Hg reactions between the magma and country rocks limited the release of Hg to the atmosphere during the Early Triassic. One possible explanation could be that eruption rates were lower, such that Hg with a relative short atmospheric residence time does not increase significantly as compared to the longer residence time of greenhouse gases such as volcanic CO2 . In this case, volcanic CO2 emissions may drive global warming but not leave a significant Hg spike in the rock record. 6. Conclusion Our results, based on high southern latitude and equatorial records, along with previous data from northern latitudes, provide the first global assessment of the impact of Siberian Traps large igneous province (STLIP) on the stratigraphic Hg record from late Permian to Early Triassic time. Our results show that there was a global Hg spike recorded in sediments at the end-Permian mass extinction horizon. Our new Hg stable isotope data is consis-

tent with this Hg spike having an atmospheric-derived signature of volcanic Hg. These results demonstrate that the STLIP had global impact, with interhemispheric transport and mixing of Hg injected into the atmosphere and eventual deposition and sequestration into marine sediments. Such a massive increase in Hg deposition to global oceans may have had significant ecological effects, potentially playing a contributing factor to the mass extinction. During the Early Triassic Hg values are low, indicative of overall background Hg deposition, and suggesting that the STLIP did not perturb the global Hg cycle during the Early Triassic recovery. Previous studies of high precision U–Pb age dating of the STLIP rocks and northern hemisphere Hg records suggest that STLIP magmatism continued at least until 251.4 Ma and potentially renewed volcanism occurred during latest Smithian time. Our results suggest that renewed volcanism during the Early Triassic was either not as severe as at the EMPE, and/or its intensity and frequency did not impact global background Hg deposition with only northern hemisphere records close to the eruptive centre recording a Hg anomaly around the Smithian–Spathian substage boundary. As such this renewed event may have delayed overall Early Triassic recovery but did not have as severe ecological effect as the impact of volcanism appears to have been more restricted to the northern hemisphere. Acknowledgements We are grateful to journal editor Prof. Tamsin Mather and two anonymous reviewers for comments that improved this paper, and to Xinbin Feng and Guangyi Sun for their help of Hg isotopic analysis. This study was supported by NSFC grants (No. 41673011, 41473006, 41272025 to LSZ, 41803011 to LZ and 41821001 to XSC), by Natural Science Foundation of Hubei Province (No. 2018CFB263 to LZ), by the Fundamental Research Funds for the Central Universities, China University of Geosciences-Wuhan (No. CUGCJ1815 and CUGQYZX1728 to LSZ, CUG170683 to LZ). Peter A. Cawood acknowledges support from Australian Research Council grant FL160100168. Xiangdong Wang acknowledges support from the Fundamental Research Funds for National Universities, China University of Geosciences (Wuhan). Appendix A. Supplementary material Supplementary material related to this article can be found online at https://doi.org/10.1016/j.epsl.2019.02.026. References Algeo, T.J., Hannigan, R., Rowe, H., Brookfield, M., Baud, A., Krystyn, L., Ellwood, B.B., 2007. Sequencing events across the Permian–Triassic boundary, Guryul Ravine (Kashmir, India). Palaeogeogr. Palaeoclimatol. Palaeoecol. 252, 328–346. Algeo, T.J., Twitchett, R.J., 2010. Anomalous Early Triassic sediment fluxes due to elevated weathering rates and their biological consequences. Geology 38, 1023–1026. Bando, Y., 1981. Lower Triassic ammonoids from Guryul ravine and the Spur three kilometres north of Barus. The Upper Permian and Lower Triassic Faunas of Kashmir. Palaeontol. Indica New Ser. 46, 135–178. Baud, A., Atudorei, V., Sharp, Z., 1996. Late Permian and Early Triassic evolution of the Northern Indian margin: carbon isotope and sequence stratigraphy. Geodyn. Acta (Paris) 9, 57–77. Bergquist, B.A., 2017. Mercury, volcanism, and mass extinctions. Proc. Natl. Acad. Sci. USA 114, 8675–8677. Blum, J.D., Sherman, L.S., Johnson, M.W., 2014. Mercury isotopes in earth and environmental sciences. Annu. Rev. Earth Planet. Sci. Lett. 42, 249–269. Brayard, A., Escarguel, G., Bucher, H., Monnet, C., Brühwiler, T., Goudemand, N., Galfetti, T., Guex, J., 2009. Good genes and good luck: ammonoid diversity and the end-Permian mass extinction. Science 325, 1118–1121. Brookfield, M.E., Twitchett, R.J., Goodings, C., 2003. Palaeoenvironments of the Permian–Triassic transition sections in Kashmir, India. Palaeogeogr. Palaeoclimatol. Palaeoecol. 198, 353–371.

X. Wang et al. / Earth and Planetary Science Letters 513 (2019) 144–155

Brosse, M., Baud, A., Bhat, G.M., Bucher, H., Leu, M., Vennemann, T., Goudemand, N., 2017. Conodont-based Griesbachian biochronology of the Guryul Ravine section (basal Triassic, Kashmir, India). Geobios 50, 359–387. Burgess, S.D., Bowring, S.A., Shen, S.Z., 2014. High-precision timeline for Earth’s most severe extinction. Proc. Natl. Acad. Sci. USA 111, 3316–3321. Burgess, S.D., Bowring, S.A., 2015. High-precision geochronology confirms voluminous magmatism before, during, and after Earth’s most severe extinction. Sci. Adv. 1, 1500470e. Burgess, S.D., Muirhead, J.D., Bowring, S.A., 2017. Initial pulse of Siberian Traps sills as the trigger of the end-Permian mass extinction. Nat. Commun. 8, 164. Chen, Z.Q., Tong, J.N., Fraiser, M.L., 2011. Trace fossil evidence for restoration of marine ecosystems following the end-Permian. Palaeogeogr. Palaeoclimatol. Palaeoecol. 299, 449–474. Chen, Z.Q., Benton, M.J., 2012. The timing and pattern of biotic recovery following the end-Permian mass extinction. Nat. Geosci. 5, 375–383. Chen, Y.L., Twitchett, R.J., Jiang, H.S., Richoz, S., Lai, X.L., Yan, C.B., Sun, Y.D., Liu, X.D., Wang, L., 2013. Size variation of conodonts during the Smithian-Spathian (Early Triassic) global warming event. Geology 41, 823–826. Estrade, N., Carignan, J., Sonke, J.E., Donard, O.F.X., 2009. Measuring Hg isotopes in Bio-Geo-Environmental reference materials. Geostand. Geoanal. Res. 34, 79–93. Galfetti, T., Bucher, H., Ovtcharova, M., Schaltegger, U., Brayard, A., Brühwiler, T., Goudemand, N., Weissert, H., Hochuli, P.A., Cordey, F., Guodun, K., 2007. Timing of the Early Triassic carbon cycle perturbations inferred from new U–Pb ages and ammonoid biochronozones. Earth Planet. Sci. Lett. 258, 593–604. Gehrke, G.E., Blum, J.D., Meyers, P.A., 2009. The geochemical behavior and isotopic composition of Hg in a mid-Pleistocene western Mediterranean sapropel. Geochim. Cosmochim. Acta 73, 1651–1665. Gong, Q., Wang, X.D., Zhao, L.S., Grasby, S.E., Chen, Z.Q., Zhang, L., Li, Y., Cao, L., Li, Z.H., 2017. Mercury spikes suggest volcanic driver of the Ordovician-Silurian mass extinction. Sci. Rep. 7, 5304. Grasby, S.E., Sanei, H., Beauchamp, H., 2011. Catastrophic dispersion of coal fly ash into oceans during the latest Permian extinction. Nat. Geosci. 4, 104–107. Grasby, S.E., Sanei, H., Beauchamp, H., Chen, Z.H., 2013a. Mercury deposition through the Permo–Triassic biotic crisis. Chem. Geol. 351, 209–216. Grasby, S.E., Beauchamp, B., Embry, A., Sanei, H., 2013b. Recurrent Early Triassic ocean anoxia. Geology 41, 175–178. Grasby, S.E., Beauchamp, B., Bond, D.P.G., Wignall, P., Talavera, C., Galloway, J.M., Piepjohn, K., Reinhardt, L., Blomeier, D., 2015. Progressive environmental deterioration in northwestern Pangea leading to the latest Permian extinction. Geol. Soc. Am. Bull. 127, 1331–1347. Grasby, S.E., Beauchamp, B., Bond, D.P.G., Wignall, P., Sanei, H., 2016a. Mercury anomalies associated with three extinction events (Capitanian Crisis, Latest Permian Extinction and the Smithian/Spathian Extinction) in NW Pangea. Geol. Mag. 153, 285–297. Grasby, S.E., Beauchamp, B., Knies, J., 2016b. Early Triassic productivity crises delayed recovery from world’s worst mass extinction. Geology 44, 779–782. Grasby, S.E., Shen, W.J., Yin, R.S., Gleason, J.D., Blum, J.D., Lepak, R.F., Hurley, J.P., Beauchamp, B., 2017. Isotopic signatures of mercury contamination in latest Permian oceans. Geology 45, 55–58. Huang, Y.G., Chen, Z.Q., Wignall, P.B., Zhao, L.S., 2017. Latest Permian to middle Triassic redox condition variations in ramp settings, South China: pyrite framboid evidence. Geol. Soc. Am. Bull. 129, 229–243. Ivanov, A.V., He, H.Y., Yan, L.K., Ryabov, V.V., Shevko, A.Y., Palesskii, S.V., Nikolaeva, I.V., 2013. Siberian Traps large igneous province: evidence for two flood basalt pulses around the Permo-Triassic boundary and in the Middle Triassic, and contemporaneous granitic magmatism. Earth-Sci. Rev. 122, 58–76. Korte, C., Kozur, H.W., 2010. Carbon-isotope stratigraphy across the Permian–Triassic boundary: a review. J. Asian Earth Sci. 39, 215–235. Li, M.S., Huang, C.J., Hinnov, L., Ogg, J., Chen, Z.Q., Zhang, Y., 2016. Obliquity-forced climate during the Early Triassic hothouse in China. Geology 44, 623–626. Liu, Y.S., Hu, Z.C., Gao, S., Gunther, D., Xu, J., Gao, C.G., Chen, H.H., 2008. In situ analysis of major and trace elements of anhydrous minerals by LA-ICP-MS without applying an internal standard. Chem. Geol. 257, 34–43. Luo, G.M., Algeo, T.J., Huang, J.H., Zhou, W.F., Wang, Y.B., Yang, H., Richoz, S., Xie, S.C., 2014. Vertical δ 13 Corg gradients record changes in planktonic microbial community composition during the end-Permian mass extinction. Palaeogeogr. Palaeoclimatol. Palaeoecol. 396, 119–131. Lyu, Z.Y., Orchard, M.J., Chen, Z.Q., Zhao, L.S., Zhang, L., Zhang, X.M., 2018. A taxonomic re-assessment of the Novispathodus waageni group and its role in defining the base of the Olenekian (Lower Triassic). J. Earth Sci. 29, 824–836. Nakazawa, K., Kapoor, H.M., Ishii, K., Bando, Y., Okimura, Y., Tokuoka, T., 1975. The Upper Permian and the Lower Triassic in Kashmir, India. Mem. Fac. Sci., Kyoto Univ., Ser. Geol. Mineral. 41, 1–106. Orchard, M.J., 2007. Conodont diversity and evolution through the latest Permian and Early Triassic upheavals. Palaeogeogr. Palaeoclimatol. Palaeoecol. 252, 93–117. Payne, J.L., Lehrmann, D.J., Wei, J.Y., Orchard, M.J., Schrag, D.P., Knoll, A.H., 2004. Large perturbations of the carbon cycle during recovery from the end-Permian extinction. Science 305, 506–509.

155

Percival, L.M.E., Witt, M.L.I., Mather, T.A., Hermoso, M., Jenkyns, H.C., Hesselbo, S.P., Al-Suwaidi, A.H., Storm, M.S., Xu, M., Ruhl, M., 2015. Globally enhanced mercury deposition during the end-Pliensbachian extinction and Toarcian OAE: a link to the Karoo–Ferrar Large Igneous Province. Earth Planet. Sci. Lett. 428, 267–280. Percival, L.M.E., Ruhl, M., Hesselbo, S.P., Jenkyns, H.C., Mather, T.A., Whiteside, J.H., 2017. Mercury evidence for pulsed volcanism during the end-Triassic mass extinction. Proc. Natl. Acad. Sci. USA 114, 7929–7934. Percival, L.M.E., Jenkyns, H.G., Mather, T.A., Dickson, A.J., Batenburg, S.G., Ruhl, M., Robinson, S.A., Woelders, L., 2018. Am. J. Sci. 318, 799–860. Pyle, D.M., Mather, T.A., 2003. The importance of volcanic emissions for the global atmospheric mercury cycle. Atmos. Environ. 37, 5115–5124. Sanei, H., Grasby, S.E., Beauchamp, B., 2012. Latest Permian mercury anomalies. Geology 40, 63–66. Selin, N.E., 2009. Global biogeochemical cycling of Mercury: a review. Annu. Rev. Environ. Resour. 34, 43–63. Shi, L., Shen, J., Ito, T., Chen, Z.Q., 2016. Proliferation of shallow-water radiolarians coinciding with enhanced oceanic productivity in reducing conditions during the Middle Permian South China: evidence from the Gufeng Formation of western Hubei Province. Palaeogeogr. Palaeoclimatol. Palaeoecol. 444, 1–14. Stanley, S.M., 2009. Evidence from ammomoids and conodonts from multiple Early Triassic mass extinctions. Proc. Natl. Acad. Sci. USA 106, 15264–15267. Sun, Y.D., Joachimski, M.M., Wignall, P.B., Yan, C.B., Chen, Y.L., Jiang, H.S., Wang, L.N., Lai, X.L., 2012. Lethally hot temperatures during the Early Triassic Greenhouse. Science 338, 366–370. Svensen, H., Planke, S., Polozov, A.G., Schmidbauer, N., Corfu, F., Podladchikov, Y.Y., Jamtveit, B., 2009. Siberian gas venting and the end-Permian environmental crisis. Earth Planet. Sci. Lett. 277, 490–500. Sweet, W.C., 1970. Permian and Triassic conodonts from a section at Guryul Ravine, Vihi district, Kashmir. Univ. Kans. Paleontol. Contrib., Pap. 49. 10 pp. Takahashi, S., Yamasaki, S.I., Ogawa, Y., 2014. Bioessential element-depleted ocean following the euxinic maximum of the end-Permian mass extinction. Earth Planet. Sci. Lett. 393, 94–104. Tewari, R., Pandita, S.K., McLoughlin, S., Agnihotri, D., Pillai, S.S., Singh, V., Kumar, K., Bhat, G.D., 2015. The Permian–Triassic palynological transition in the Guryul Ravine section, Kashmir, India: implications for Tethyan-Gondwanan correlations. Earth-Sci. Rev. 149, 53–66. Them II, T.R., Jagoe, C.H., Caruthers, A.H., Gill, B.C., Grasby, S.E., Gröcke, D.R., Yin, R., Owens, J.D., 2019. Terrestrial sources as the primary delivery mechanism of mercury to the oceans across the Toarcian Oceanic Anoxic Event (Early Jurassic). Earth Planet. Sci. Lett. 507, 62–72. Thibodeau, A.M., Ritterbush, K., Yager, J.A., West, A.J., Ibarra, Y., Bottjer, D.J., Berelson, W.M., Bergquist, B.A., Corsetti, F.A., 2016. Mercury anomalies and the timing of biotic recovery following the end-Triassic mass extinction. Nat. Geosci. 7, 11147. Thibodeau, A.M., Bergquist, B.A., 2017. Do mercury isotopes record the signature of massive volcanism in marine sedimentary records. Geology 45, 96. Tong, J.N., Zuo, J.X., Chen, Z.Q., 2007. Early Triassic carbon isotope excursions from South China: proxies for devastation and restoration of marine ecosystems following the end-Permian mass extinction. Geol. J. 42, 371–389. Wang, X.D., Cawood, P.A., Zhao, H., Zhao, L.S., Grasby, S.E., Chen, Z.Q., Wignall, P.B., Lv, Z.Y., Han, C., 2018. Mercury anomalies across the end Permian mass extinction in South China from shallow and deep water depositional environments. Earth Planet. Sci. Lett. 496, 159–167. Wignall, P.B., Twitchett, R.J., 1996. Oceanic anoxia and the end Permian mass extinction. Science 272, 1155–1158. Wignall, P.B., 2001. Large provinces and mass extinctions. Earth-Sci. Rev. 53, 1–33. Zambardi, T., Sonke, J.E., Toutain, J.P., Sortino, F., Shinohara, H., 2009. Mercury emissions and stable isotopic compositions at Vulcano Island (Italy). Earth Planet. Sci. Lett. 277, 236–243. Zhang, L., Zhao, L.S., Chen, Z.Q., Algeo, T.J., Yang, L., Cao, L., 2015. Amelioration of marine environments at the Smithian–Spathian boundary, Early Triassic. Biogeosciences 12, 1597–1613. Zhao, L.S., Tong, J.N., Zuo, J.X., Ming, H.L., 2002. Discussion on Induan-Olenekia boundary in Chaohu, Auhui Province, China. J. China Univ. Geosci. 3, 141–150. Zhao, L.S., Orchard, M.J., Tong, J.N., 2004. Lower Triassic conodont biostratigraphy and speciation of Neospathodus waageni around the Induan-Olenekian boundary of Chaohu, Anhui Province, China. Albertiana 29, 41–43. Zhao, L.S., Tong, J.N., Orchard, M.J., 2005. Study on the Lower Triassic Conodont Sequence and the Induan-Olenekian Boundary in Chaohu, Anhui Province. China University of Geosciences Press. 140 pp. Zhao, L.S., Orchard, M.J., Tong, J.N., Sun, Z.M., Zuo, J.X., Zhang, S.X., Yun, A.L., 2007. Lower Triassic conodont sequence in Chaohu, Anhui Province, China and its global correlation. Palaeogeogr. Palaeoclimatol. Palaeoecol. 252, 24–38. Zhao, L.S., Tong, J.N., Sun, Z.M., Orchard, M.J., 2008. A detailed Lower Triassic conodont biostratigraphy and its implications for the GSSP candidate of the Induan– Olenekian boundary in Chaohu, Anhui Province. Prog. Nat. Sci. 18, 79–90.