Volcanic related methylmercury poisoning as the possible driver of the end-Devonian Mass Extinction

The end-Devonian global Hangenberg event (359 Ma) is among the most devastating mass extinction events in Earth’s history, albeit not one of the “Big Five”. This extinction is linked to worldwide anoxia caused by global climatic changes. These changes could have been driven by astronomical forcing and volcanic cataclysm, but ultimate causes of the extinction still remain unclear. Here we report anomalously high mercury (Hg) concentration in marine deposits encompassing the Hangenberg event from Italy and Austria (Carnic Alps). The Hangenberg event recorded in the sections investigated can be here interpreted as caused by extensive volcanic activity of large igneous provinces (LIPs), arc volcanism and/or hydrothermal activity. Our results (very large Hg anomalies) imply volcanism as a most possible cause of the Hangenberg event, similar to other first order mass extinctions during the Phanerozoic. For the first time we show that apart from anoxia, proximate kill mechanism of aquatic life during the event could have been methylmercury formed by biomethylation of a volcanically derived, huge concentration of inorganic Hg supplied to the ocean. Methylmercury as a much more toxic Hg form, potentially could have had a devastating impact on end-Devonian biodiversity, causing the extinction of many pelagic species.

The D/C boundary in both sections is situated directly above the Hangenberg Black Shale 30,31,33 . The D/C boundary may be somewhat problematic and needs redefinition (caused by problems with discrimination of Siphonodella sulcata from its supposed ancestor Siphonodella praesulcata 35 ). The new criterion for definition of the base of the Carboniferous System proposed by the Working Group on the boundary is: identification of the base of the Pr. kockeli Zone, beginning of radiation and top of major regression (top of HSS) and end of mass extinction 32 . In the limestone overlying the HBS, the conodonts of the species Protognathodus kockeli 32 was found in both sections. Therefore, the position of the D/C boundary did not changed in comparison to previous studies 30,31,33 .
The D/C boundary successions in the Plan de Zermula A and Kronhofgraben were deposited in deeper paleoenvironment 31 . In the late Devonian, Carnic Alps represented the northern tips of Gondwana and belonged to the Gondwana-derived Bosnian-Noric Terrane accreted to the intra-Alpine Mediterranean terrane during the Carboniferous 36 . The investigated rocks outcropped in the Carnic Alps reflected strong thermal alteration (CAI) ranging from 4.5 to 5 in the Plan di Zermula A and Kronhofgraben, respectively 31,37 .

Results
The HBS intervals in the sections investigated display extremely high Hg values, with maxima of 20216 and 9758 ppb in Kronhofgraben ( Fig. 2; Table 1) and Plan di Zermula A ( Fig. 3; Table 2), respectively. The HBS from the Plan di Zermula A section contains Hg anomalies that are ~13-100 times higher than the ~100 ppb background, whereas in the Kronhofgraben section the anomalies are ~12-84 times higher than the background values.
Interestingly, significant concentrations of methylmercury were found in the whole Kronhofgraben section, where MeHg is in the range 13-348 pg/g d.w. (Fig. 2). Additionally, we found 55 pg/g d.w. of MeHg in Novchomok section in Uzbekistan and 72.72 pg/g d.w. of MeHg sampled from the uppermost Devonian part of the Woodford Shale from the Arbuckle Anticline in the Oklahoma, USA. Traces of MeHg were also found in the HBS interval at Kowala Quarry, Poland (20.66 pg/g d.w. of MeHg).
In comparison to MeHg levels found in modern sediments (reaching from1000 to 700000 pg/g d.w. in polluted basin 38 ), those detected in sedimentary rocks studied, are relatively low. However, the original amounts of MeHg in the investigated sediments would have been higher but impoverished during diagenesis (see below for details). The Hg enrichments are observed in organic-rich Hangenberg equivalent intervals such as Kronhofgraben (from 0.51 to 13.28% TOC) and Plan di Zermula A (from 0.7 to 12.53% TOC). The values of the Hg/TOC ratio in the HBS at Kronhofgraben range from 815 to 8096.5 (ppb/%), while the background samples show a range from 387.5 to 985 (ppb/%). In Plan di Zermula A, the values of Hg/TOC ratios in the HBS range from 779 to 3269 (ppb/%) and are higher than those from the background samples (ranging from 84.5 to 676.8 ppb/%Hg/TOC).

Discussion and conclusions
Volcanic and hydrothermal activities are considered to be the main sources of elevated Hg in sedimentary rocks 16,17,[39][40][41] . Besides Hg delivery to the atmosphere by volcanic activity, other processes can produce Hg spikes in the sedimentary record, including widespread wildfires, terrestrial input, magmatic emplacement or thermogenic processes related to bolide impact 19,39,40,[42][43][44] . Additionally, some authors suggested that Hg enrichments can be sulfide-hosted in euxinic facies, and high Hg spikes not necessary would be connected with volcanic activity 45 . However, in such a case, the Hg enrichments would be well-correlated with total sulphur (TS), which is not observed in our sections. Although extensive wildfires on land were confirmed during the Hangenberg event, based on the co-occurrence of charcoal and high concentrations of polycyclic aromatic hydrocarbons in sedimentary rocks 2,46,47 , these, however, could have also been induced by volcanism, as evidenced by the co-occurrence of charcoals and ash layers 2,46 . No conclusive evidence for bolide impact at the D/C boundary has been detected thus far 1 . In fact, at the D/C boundary, volcanic activity has frequently been documented, mainly on the basis of the presence of ash layers below, above and within the HBS (e.g. in the Holy Cross Mountains, Iberian Pyrite Belt (SW  www.nature.com/scientificreports www.nature.com/scientificreports/ Iberia) and Rhenish Massif 2,9,15,48 ), mercury spikes (e.g. 14,23-25 ), as well as the presence of abnormal or strongly altered spores (tetrads), which could reflect the mutagenic effect of regional acidification caused by explosive volcanism 47,49,50 . The most plausible sources of very large amounts of Hg during the end-Devonian interval are the massive Magdalen silicic large igneous province (LIP) 12 and the Siberian (Yakutsk-Viluy) and/or the Kola-Dnieper LIPs 12,51 ; however, the interval also overlaps with formation of the Almaden Hg deposit (last mineralization pulse episodes 52 ), which constitutes one of the largest geochemical anomalies on Earth and coincided with the first phase of the Variscan orogeny 53 , as considered for the Hangenberg crisis 15    www.nature.com/scientificreports www.nature.com/scientificreports/ continental-silic type) 12 . Moreover, we cannot exclude other additional Hg sources, for instance connected with explosive eruptions which could overlap with LIPs activity, (see e.g. Magdalen Basin 54 ). Mercury has a strong affinity to organic matter and to a minor extent can also be associated with sulfides and clay minerals; therefore, Hg is normalized to total organic carbon (TOC) content 19 . Importantly, the Hg spikes in our sections are also evident when normalized to TOC content, which can be interpreted as an effect of increased input of Hg to the basins independently of the potential influence of reducing depositional conditions. The Hg vs. Al 2 O 3 correlation in the investigated successions is very weak (R 2 = 0.25 in Kronhofgraben and R 2 = 0.09 in Plan di Zermula A), indicating no correlation of Hg with the clay fraction. However, Hg exhibits a good correlation with Mo in the all sections (R 2 = 0.88 in Kronhofgraben and R 2 = 0.76 in Plan di Zermula A). This could indicate that some Hg was associated with sulfides as a result of its intensified precipitation in a sulfide-rich (euxinic) water column 2,42,45 . In the sections investigated, Hg vs. TS correlation is very weak (R 2 = 0.10 in Kronhofgraben and R 2 = 0.16 in Plan di Zermula A) which does not confirm sulfides as host of Hg. However, the Hg vs. TOC correlation in the D/C boundary at Novchomok section is very low 22 , which confirm that Hg enrichments are facies independent and thus are indicative of volcanic activity during this time. For the Kronhofgraben and Plan di Zermula A sections this correlation is good (R 2 = 0.86 and 0.75 respectively), suggesting possible different sources of this element. However, as already emphasized, there are a number of lines of evidence for volcanic and hydrothermal activities, as well as widespread wildfires, during this time 2,15 , allowing for a firm statement that increased Hg input to the basins was connected with diverse volcanic activities and related combustion of biomass on land. Moreover, the Hangenberg event took place during an interglacial period 1 ; therefore, some Hg could have originated from permafrost melting 55 , but even if this process had taken place, Hg would have previously accumulated in the permafrost as a result of volcanic or pyrogenic processes. To summarize, based on all the available data, we can state that the main sources of Hg were volcanism and related hydrothermal activities ( Fig. 4; see also 23,24 ). In fact, volcanic processes are main sources of Hg in atmosphere 39 .
The organic form of mercury (CH 3 Hg + ) is a strong neurotoxin that is bioconcentrated in aquatic food chains and is able to cross the blood-brain barrier; thus, this form of Hg is much more toxic to living organisms than inorganic Hg 26,28 . In modern environments, methylmercury (MeHg) is generated predominantly by anaerobic microorganisms, such as sulfate-reducing bacteria (e.g., Geobacter sulfurreducens) 56 . Despite widespread Hg pollution, annual emissions of Hg have recently been higher from natural sources than anthropogenic ones, constituting as much as 70% of all Hg emissions 57 . However, the concentrations of Hg detected in all the end-Devonian sections are surprisingly high, similar to the present-day mercury concentrations found in highly polluted basins, e.g., some parts of the Baltic Sea 38 . The Hg concentrations of up to 20000 ppb in Kronhofgraben and 1000-10000 ppb in the Plan di Zermula A, and Hg spikes determined in Germany 22 , south Vietnam 23 , Czech Republic and south China 24 sections suggest, that global Hg concentrations were highly elevated during the Hangenberg event. This finding implies that, during favorable sedimentary conditions, very high concentrations of MeHg can be produced on the global scale. In the investigated samples we measured relatively minor amounts of MeHg (Table 1) in comparison with the MeHg levels in modern sediments. In polluted basins, concentrations of MeHg vary from 1000 to 700000 pg/g d.w. of MeHg 38 and are much higher relative to total MeHg concentration from our sections. However, the original amounts of MeHg in the investigated sediments would have been higher, assuming large enrichment of total Hg in anomalous samples. It is very probable that MeHg could have been demethylated during diagenesis as a result of the common diagenetic process of demethylation, which is influenced by temperature 58 . Because of the strong thermal alteration of the investigated rocks (conodont alteration index = 4.5-5 in Carnic Alps 31,37 ) the occurrence of demethylation seems to be very likely.
Therefore, regardless of the Hg source, its high level in the end-Devonian water column, subsequent trapping in sediment and biomethylation to the more toxic MeHg form by anaerobic bacteria, would have had an additional devastating impact on aquatic life during the Hangenberg event. This can be produced under conditions of extended anoxia/euxinia during this time 2,59 and the occurrence of rich sulfate-reducing bacteria communities 60 which can change Hg to its methyl form. Additionally, blooms of green algal phototrophs (prasinophytes) 8 during black shale events would have contributed, mostly indirectly, to MeHg production 61 . However, indisputable evidence for bacterial Hg methylation is the occurrence of notable concentrations of MeHg in the sediments investigated and the similarities in the distributions of Hg and MeHg in the Kronhofgraben section (Fig. 2).
Observation of modern marine environments has confirmed that MeHg is highly toxic to animals at higher trophic levels (such as fish, birds and mammals) 26,28 . In this light it seems to be evident that severe extinction of marine and nonmarine fish and tetrapods 3 , as well as pelagic conodont animals 1 , during the Hangenberg event may also have resulted from MeHg poisoning that could have affected different aquatic habitats. Although the effect of MeHg on benthic invertebrates is regarded as minimal 62 , these organisms were significantly affected by concomitant, globally widespread anoxia. Such anoxia asphyxiation-methylmercury poisoning may have also been kill mechanisms in other mass extinctions, but this should be tested by searching for traces of methylmercury in other sedimentary rocks.

Methods
A total of 30 samples were analyzed for Hg abundances, using atomic absorption spectrometry (AAS) analyzer Milestone DMA-80 Direct Mercury (detection limit = 0.2 ppb). An Eltra CS-500 IR-analyzer was used for total organic carbon (TOC) determination [for more detail methodology see 22 ]. In addition, Hg as well as Mo and Al contents were analyzed using ICP-MS at the Bureau Veritas Acme Labs Canada Ltd. The results of Hg concentrations measured by AAS and ICP-MS are comparable 22 . Mercury enrichments are typically normalized to TOC 42 . For some samples, Hg/TOC ratios were deemed unreliable, due to very low TOC content (<0.2 TOC%; these samples are marked in red on Figs. 2-3). In these cases even low Hg concentration may generate very high, artificial Hg/TOC spikes 63 . Additionally, we examined 19 samples in terms of methylmercury (MeHg) occurrence (Kronhofgraben, n = 15; Novchomok, Uzbekistan n = 1; Arbuckle Anticline USA n = 1; Kowala, Poland n = 1 and one blank sample form Morocco La10 -Lahmida, which consist of only trace amounts of total Hg < 7 ppb). MeHg was determined using Automated Methylmercury System MERX-M (Brooks Rand, USA) following EPA Method 1630. Samples were extracted according to the modified method of De Wild et al. 64 . In brief, five grams of grinded rock was weighted in a 30 ml Teflon centrifuge tube. For each tube 5 ml of extraction buffer (180 g KBr, 50 ml concentrated H 2 SO 4 , 0.2 g NH 2 OH • HCl dilute to 1 liter), 1 ml of 1 M CuSO 4 and 10 ml of CH 2 Cl 2 were added. Tubes were left at room temperature in darkness for 1 h, and afterwards they were mixed on a vortex mixer for 1 h in darkness. Then, the samples were centrifuged at 3000 rpm for 30 min. From each sample 5 ml of CH 2 Cl 2 phase was added into a back-extraction vial containing 25 ml of reaction grade water. The back-extraction vials were placed in a heating block (45 °C) and attached to N 2 lines (flow at 100 mL per minute). After the extraction, 5 ml of the aqueous sample was added to 20 ml of reaction grade water in a glass reaction vial. The pH was adjusted to between 4 and 5 with the addition of 300 µl of 2 M acetate buffer. The last step was an addition of 50 µl of 1% NaBEt 3 . Samples were analyzed using Automated Methylmercury System MERX-M (Brooks Rand, USA) following EPA Method 1630.
The Method Detection Limit (MDL) value was determined based on a series of measurement results of blank samples spiked with extracts obtained from the reference material ERMCC580 (Estuarine sediment MeHg = 75 ng/g d.w. ± 4 ng/g d.w.). Nine determinations were performed for extracts diluted to the expected MDL value -1 pg. The MDL was calculated as 3x Standard Deviation (SD). Method Quantification Limit (MQL) was calculated as 3x MDL. The MDL value was estimated to be 0.7 pg what corresponds to MeHg concentration in sediments samples of an even 3.5 pg/g d.w. (calculated from the sample weight that was taken for extraction and including sample moisture -2 g d.w.). The MQL value was calculated as MQL = 3xMDL and equalling 2.2 pg what corresponds to a MeHg concentration of 11.0 pg/g d.w. (assuming the mass of the sample -2 g of d.w.).
The mercury-poor sample (Hg = 6.78 ppb) from the Lahmida section used here as reference blank contain MeHg below detection limit (2.73 pg/g d.w.).