Mercury Anomalies Link to Extensive Volcanism Across the Late Devonian Frasnian–Famennian Boundary in South China

The Late Devonian Frasnian–Famennian (F–F) mass extinction has been long-time debated by non-volcanic causes, extra-terrestrial impacts, and large igneous province (LIP) eruptions. To better understand the ultimate cause of the F–F mass extinction, here we investigate the chemostratigraphy of mercury (Hg) and total organic carbon (TOC) on two marine F–F strata in the Dushan area, South China. In both sections, high Hg and Hg/TOC anomalies were observed near the F–F boundary. These anomalies are in line with those recently observed in Morocco, Germany, Poland, and north Russia, suggesting a global Hg flux. The Late Devonian LIP eruptions, which are believed to have emitted massive amounts of Hg, could be responsible for the global Hg and Hg/TOC anomalies around the F–F boundary. The observed Hg and Hg/TOC anomalies coincide with the extinction of Frasnian fauna in the Dushan area, implying a causal link between the Viluy, Kola, and Pripyat-Dnieper-Donets LIP eruptions and the F–F mass extinction.


INTRODUCTION
The Late Devonian Frasnian-Famennian (F-F) transition witnessed one of the "Big Five" mass extinctions which killed at least 40% of all species during the Phanerozoic (e.g., Stanley, 2016). This critical period was characterized by many global events including climate oscillations (Huang et al., 2018;Liu et al., 2020), eustatic sea-level changes (Sandberg et al., 2002;Bond and Wignall, 2008), atmospheric O 2 fluctuation (Dahl et al., 2010), seawater anoxic/hypoxic events (Kellwasser events, Joachimski and Buggisch, 1993), eutrophication events (Murphy et al., 2000), and instability in the sulfate reservoir (Cai et al., 2020). Three main Late Devonian extinction events occurred within an interval of 0.7 Ma from the latest Frasnian to the earliest Famennian stages. These three extinction events were named the Lower Kellwasser Event (372.5 Ma) from the latest Frasnian stage, the Upper Kellwasser Event near the F-F boundary (ca. 372 Ma), and the Homoctenid Event (371.8 Ma) during the earliest Famennian stage (McGhee and Racki, 2021). Impacts of these events were more severe in low-latitude areas and marine ecosystems, where a fatal impact was caused on shallow-water reefs and fish (McGhee and Racki, 2021). The Late Devonian Viluy LIP (Ricci et al., 2013) in Siberia and the Kola and Pripyat-Dnieper-Donets LIPs (Puchkov et al., 2016) in Baltica are temporally overlapped with the time of Devonian F-F boundary age (Percival et al., 2018a;Da Silva et al., 2020). Climate change, ocean anoxia, and other factors are proposed in concert as the killing mechanisms during these biotic crises (McGhee and Racki, 2021, and references therein), but the ultimate cause of the Late Devonian F-F mass extinction remains somewhat uncertain.
Sedimentary Hg anomalies have been proven to be used as a new proxy of LIP eruption (Bergquist, 2017;Grasby et al., 2019;Shen et al., 2020). In geological history, Hg was dominantly emitted to the environment via volcanism and cycled in the atmosphere-land-ocean system before ultimately sinking to the seafloor through organic matter burial . Excessive Hg emission often occurred during LIP eruption,  Blakey (2016), and assumed locations of coeval igneous activities and Hg enrichment sections are from Racki (2020). (B) Paleogeography of South China during the Late Devonian and location of the Dushan area (modified after Zhao, 1985). resulting in anomalously high Hg concentration to total organic carbon (Hg/TOC) ratios during critical events (e.g., mass extinction, oceanic anoxia) (Sanei et al., 2012;Grasby et al., 2019).
Compelling evidence of Hg enrichments in mass extinction horizons has manifested that the end-Ordovician (Gong et al., 2017;Jones et al., 2017), end-Permian (Sanei et al., 2012;Burgess and Bowring, 2015;Grasby et al., 2017Grasby et al., , 2020Shen et al., 2019), end-Triassic (Blackburn et al., 2013;Thibodeau et al., 2016;Percival et al., 2017), end-Cretaceous (Font et al., 2016;Schoene et al., 2019), and Paleocene-Eocene (Keller et al., 2018;Jones et al., 2019) extinctions had a causal link to LIP eruptions. High Hg anomalies were recently identified close to the Late Devonian F-F boundary from three sections located on the peripheral of Laurussia, implying that the Viluy LIP and other contemporaneous volcanism were the ultimate causes of the F-F mass extinction (Racki et al., 2018). However, to testify whether or not these LIP impacts are global, further examinations on different sections away from the volcanic center remain necessary.
The South China Craton preserves good sedimentary and PaleoBios records for the Upper Kellwasser Event across the F-F boundary (Carmichael et al., 2019, and references therein). This craton was located at the eastern Paleo-Tethys, which was far away from the center of the Late Devonian LIPs based on the global paleogeography reconstruction (Blakey, 2016, Figure 1A). If the Late Devonian volcanism did cause global-scale Hg perturbation and biota mass extinctions during the F-F interval, it may also have left sedimentary and biotic records in the South China Craton. Monotonous dolomite deposited in the Dushan area of South China during Late Devonian, making it an ideal location for recording the potential volcanic Hg signal without lithological interference (Bergquist, 2017;Percival et al., 2018b). Here, we report the high-resolution Hg chemostratigraphy in two continuous successions in the Dushan area to shed light on the correlation between the potential volcanic Hg signals and the biotic records during the F-F transition.

GEOLOGICAL BACKGROUND
The South China Craton was located at the eastern part of the Paleo-Tethys Ocean, near the equator during Late Devonian (Blakey, 2016, Figure 1A). It is one of the hotspots for the study of the Devonian system (Carmichael et al., 2019). After the Wuyi-Yunkai intraplate orogeny (Li et al., 2010;Zhang et al., 2016), the Early Devonian (Pragian Stage) transgression followed by the Late Devonian regression gives birth to numerous well-preserved Devonian sections ( Figure 1B) and provides ideal sedimentary and PaleoBios records (Ma et al., 2016;Qie et al., 2019) for global comparisons (Klapper et al., 1993). The eustatic fall and the mass extinction of the deeper-water brachiopods and conodonts and shallow-water coral and ostracods occurred correlative with the latest Frasnian Pa. linguiformis Zone to the Famennian Lower Pa. triangularis Zone (Ma et al., 2016). These patterns in South China are coincident with those of North America (Sandberg et al., 2002). The Late Devonian carbon isotope records of South China are also consistent with those from Gondwana, North America, and Europe (Zhang et al., 2019). A tuffaceous layer just above the F-F boundary from the basinal facies successions in South China yielded a SIMS zircon U-Pb age of 367.8 Ma (Zhang et al., 2020), which correlates with the precise F-F boundary age of ca. 372 Ma from Germany (Percival et al., 2018a;Da Silva et al., 2020). Low redox sensitive element concentrations (Mo, U, V) and low V/Cr ratios (<2) suggest a constant oxic ocean condition in South China during the F-F transition (Zhang et al., 2020). Ocean sulfate deficit (Cai et al., 2020), climate change (Huang et al., 2018), tectonic evolution, and volcanic/hydrothermal activities (Ma et al., 2016;Zhang et al., 2020) have been proposed as the killing mechanisms for F-F mass extinction in South China.
The Late Devonian sequence in the Dushan area represents typical semi-restricted and restricted platform facies which can be correlated with other sections in South China (Ma et al., 2016). Biostratigraphy, paleontology, and lithostratigraphy during the F-F stages have been previously studied (Wang and Chen, 1999;2006;Nie et al., 2016;Ma et al., 2016). This sequence displays a complete succession through the lastest Frasnian to the early Famennian. The latest Frasnian strata (the middle part of the Sifangpo Member of the Yaosuo Formation) in the Sifangpo (SFP) section are characterized by gray to light gray, medium to thick dolomite bearing abundant body fossils (stromatoporoids, coral, brachiopods). The F-F boundary can be roughly determined by a prominent stratigraphic transformational surface above the Frasnian fauna disappeared horizon which is also the horizon of sample SFP320 (Wang and Chen, 1999;2006). Body fossils disappeared near the F-F boundary. The earliest Famennian sequence (the middle part of the Sifangpo Member of the Yaosuo Formation) consists of light gray to grayish-white, thin-and medium-bedded dolomite and yields a few trace fossils (Wang et al., 2006). Locally hydrothermal and volcanic activities are absent in the Dushan area (GGS, 2017). Therefore, the well-preserved sedimentary and fossil records in the Dushan area ( Figure 2) show a unique advantage for studying the direct association between the potential volcanic activities and the Late Devonian mass extinctions.

METHODS
Both the fossil-rich SFP section (N 25°51′30.7″, E 107°32′11.3″) and the fossil-poor Shenheqiao (SHQ) section (N 25°55′4.8″, E 107°32′59″) in the Dushan area were selected in this study. Flesh rock samples (27 samples from the SFP section and 16 samples from the SHQ section) were systematically collected near the F-F boundary from both sections with an interval of 10-20 cm ( Figure 2). After removing the weathered surfaces, thin sections were prepared from 14 samples from the SFP section. The remaining samples were powdered to 200 mesh by the agate mortar for the following analytical testing.

Optical and FESEM-EDS Mineralogy Microscopy
Optical and field emission scanning electron microscope imaging of the 14 thin sections was performed at the Center for Lunar and Planetary Sciences, Institute of Geochemistry, Chinese Academy of Sciences, using an FEI Scios DualBeam field emission scanning electron microscope system, which was equipped with EDAX energy dispersive spectroscopy (FESEM-EDS). The backscatter electron mode analysis and energy dispersive spectroscopy point analysis were carried out at 20 kV.

Hg Concentration Analysis
The sample powders were analyzed for Hg concentrations using a Lumex Hg analyzer RA 915F with the Hg detection limit of 0.5 ng/g at the State Key Laboratory of Ore Deposit Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences. Measurements of the standard reference material (GSS-4, soil) yielded Hg recoveries of 90 and 110%, and coefficients of variation for triplicate analyses were <10%.

Total Organic Carbon Content Analysis
Total organic carbon (TOC, wt%) analyses were carried out using the same samples previously used for analyzing Hg concentrations, using the potassium dichromate method (Schumacher, 2002). The sample powders were first leached by 2.5 N HCl to remove inorganic carbon and then rinsed by 18.2 MΩ cm water, freeze-dried, and measured using an Elementar vario MICRO cube analyzer with analytical errors <2.5%.

X-Ray Diffractometry Analysis
The mineralogy composition was performed on 27 powder samples (SFP) using a Rigaku Ultima IV X-ray diffractometer, with a carbon monochromator and Cu-Kα radiation, at 40 kV and 40 mA at Guizhou Central Laboratory of Geology and Mineral Resources, Bureau of Geology and Mineral Exploration and Development of Guizhou Province. Mineral composition was determined on randomly oriented powdered samples. The diffraction patterns were processed using Jade 6.5 software, and the primitive modal contents were calculated by the Rietveld method.

Mineral Composition Results
The mineralogy composition (in wt%) is presented in Table 1. Samples from both SFP and SHQ sections are fineto mediumgrained dolostones. The rocks mainly consist of dolomite (88-97%) with minor illite, zeolite, calcite, quartz, variable fossil fragments, and organic matter. Dolomite exhibited poikilitic texture and is 0.1-0.2 mm in size. Clay minerals and authigenic quartz are mainly distributed between dolomite. Fossil fragments can be found in the thin sections of SFP0 to SFP290 and disappeared in the thin sections of SFP300 to SFP400.   (Figure 3).
Samples from the SHQ section have Hg concentrations ranging from 1.12 to 43.1 ppb with a median value of 9.68 ppb and relatively low TOC contents (0.06-1.34%). To provide meaningful interpretation, three extremely low-TOC (<0.1%) samples are eliminated from the discussion Shen et al., 2020). Hg/TOC ratios of these samples range from 2.58 to 251 with a background value of 52.2. Hg and TOC are insignificantly correlated (r −0.25), and Hg/TOC displays significant positive correlations with Hg (r +0.97) but weak correlations with TOC (r < −0.32) (Figure 3).

DISCUSSION AND CONCLUSION
Subaerial volcanism serves as the major natural source of Hg in the environment, with a normal annual flux of 75-700 tons . Volcanism emits massive amounts of Hg to the atmosphere, in the form of gaseous Hg(0). This form of Hg has a lifetime of ∼1 year in the atmosphere, allowing this element to be globally and uniformly distributed before deposition into marine and terrestrial ecosystems . Once deposited, Hg continues to cycle in the atmosphere-land-ocean system, but it ultimately sinks to the seafloor. Organic matter (OM) drawdown plays a critical role in Hg burial onto the seafloor due to its strong affinity to Hg (Sanei et al., 2012;Grasby et al., 2019). Under normal conditions, volcanic Hg emissions can be balanced by OM drawdown; however, excessive Hg emission often occurred during LIP eruption, resulting in high Hg concentration to TOC (Hg/TOC) ratios during critical events such as mass extinction events and/or oceanic anoxia events (Sanei et al., 2012;Blackburn et al., 2013;Burgess and Bowring, 2015;Font et al., 2016;Thibodeau et al., 2016;Gong et al., 2017;Jones et al., 2017;Percival et al., 2017;Keller et al., 2018;Jones et al., 2019;Schoene et al., 2019;Grasby et al., 2020;Shen et al., 2020). To be noticed, clay minerals (Kongchum et al., 2011;Uddin, 2017), sulfides (Bower et al., 2008;Duan et al., 2016;Shen et al., 2020), and Fe-Mn oxides (Quémerais et al., 1998) are also important hosts of Hg in marine sediments. Small Hg fluxes can be a localized source from the submarine volcanic and hydrothermal activity limited to vicinity of the eruption or emission center (Scaife et al., 2017;Moreno et al., 2018;Jones et al., 2019), wildfire (Shen et al., 2011), or erosion . Therefore, special attention must be paid to determine whether Hg anomalies are attributed to increased volcanic Hg loading, or other Hg enrichment mechanisms (Sanei et al., 2012;Grasby et al., 2019;Shen et al., 2020). In this study, given sulfides were not detected by X-ray diffractometry and the correlation between Hg concentrations and main rockforming minerals (dolomite, illite) is insignificant (not shown), manifesting these minerals are not the main hosts of Hg. Hg displays weak correlation with TOC ( Figure 3A), and Hg/TOC peaks are more correlative with the Hg peaks rather than TOC peaks ( Figures 3B,C) in both sections, which indicates that Hg is unlikely related to localized organic matter detention . The Hg/TOC values from both sections exhibit rhythmic and comparable variation through time, strongly supporting that the Hg enrichments in sediments are not associated with organic matter, redox conditions (e.g., Shen et al., 2020), or secondary weathering in random layers. The lithological and mineralogical similarity across the Hg and Hg/TOC peak horizons in both sections demonstrates that the terrestrial inputs, sedimentation rate, and diagenesis process are not the main causes for the observed Hg anomalies. In both sections, the Hg/TOC peaks correlate well with the Hg peaks near the F-F boundary (Figure 4), suggesting excessive Hg inputs to the ocean were responsible for the high Hg/TOC anomalies during the F-F transition. Two spurious Hg/TOC ratio peaks at 0.8 m above and 3.1 m below the F-F boundary were due to low TOC content rather than elevated supply of Hg ( Figure 4). Racki et al. (2018) suggested three criterions for detection of the real volcanic Hg signal in the sedimentary sequence: (1) Hg enriched above 3 × the median Hg content, (2) Hg/TOC enriched above 3 × the median Hg/TOC content, and (3) Hg abnormal identified worldwide. In this study, the SFP section shows the highest Hg enrichment of 3.1 × the background while the highest Hg/TOC enrichment of 5.7 × the background at the F-F boundary. Compatible Hg and Hg/ TOC enrichments can be observed in the SHQ section ( Figure 4). Localized volcanic and hydrothermal activities are limited in the Dushan area (GGS, 2017) which preclude the possibility of a local volcanic or submarine Hg contribution. Exotic volcanic matter was identified throughout the F-F boundary succession in the basinal facies in South China (Zhang et al., 2020), which indicated extensive volcanic activities occurred beyond South China during the F-F transition. Moreover, the Hg anomalies just under the F-F boundary have also been reported previously in Morocco, Germany, Poland, and north Russia (Racki et al., 2018;Racki, 2020). Although the low Hg contents in South China than other sections worldwide may be due to the effectiveness of different lithology in archiving the Hg record (Bergquist, 2017;Percival et al., 2018b), the Hg and Hg/TOC enrichments observed in our study are in line with those observed in other sections worldwide ( Figure 5). An ash bed named Center Hill Ash coincidently occurred in the US just prior to the F-F boundary based on associated conodonts (Over, 2002). Thus, the observed global Hg anomalies are tentatively correlated with the volcanic activities which produce the Center Hill Ash Bed (Racki, 2020). Considering the correlation of the complex global Hg patterns ( Figure 5) and the age of Devonian LIPs (Ricci et al., 2013;Puchkov et al., 2016;Ernst et al., 2020), we suggest the Hg perturbation in sediments is induced by the eruptions of the Late Devonian Viluy, Kola, and Pripyat-Dnieper-Donets LIPs. Volcanic activity can release massive CO 2 and geothermic gas into the atmosphere which could cause synergistic effects of global warming, ocean anoxia, acidification, and eventually marine mass extinction (Clapham and Renne, 2019;McGhee and Racki, 2021;Racki, 2021). The studied F-F volcanic Hg signals are stratigraphically tied to the extinction of Frasnian fauna and decline of organic matter (Figure 4), supporting a   Racki et al., 2018;Racki, 2020) showing Hg enrichments with highlights on their probable correlation with bentonite beds by Winter (2015). Abbreviation: CH, Center Hill; Sc, Scorpius; Sx, Sextans, PPP, Pictor-Phoenix-Pegasus.
Frontiers in Earth Science | www.frontiersin.org July 2021 | Volume 9 | Article 691827 causal link between global massive volcanism and biodiversity crisis, which further strengthens the hypothesis for LIP volcanism as the ultimate driver of F-F mass extinction.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, and further inquiries can be directed to the corresponding authors.