Volcanic origin of the mercury anomalies at the Cretaceous-Paleogene transition of Bidart, France

The cause of the Cretaceous-Paleogene (K-Pg) mass extinction is a long-standing debate and it is certainly the most disputed of the “Big Five” mass extinctions that punctuated the evolution of life on Earth. The compelling evidence that an asteroid impact was the primary cause of the K-Pg mass extinction (Alvarez et al., 1980; Schulte et al., 2010) has been opposed by the idea that the Deccan Traps (in India) drove the extinction instead (Courtillot et al., 1986; Keller et al., 2020). Recent high-precision radiometric ages indicate that the emplacement of the largest lava formations of the Deccan Traps began shortly after (Sprain et al., 2019) or before (Schoene et al., 2019) the Chicxulub (Mexico) impact. The sedimentary record of Deccan Traps eruptions is more challenging to discern and relies mostly on the correlation of the age of the Deccan lava flows with mercury (Hg) concentrations recorded in marine sediments, which are taken as a proxy for increased volcanic input (Sial et al., 2013, 2014, 2016; Font et al., 2016; Keller et al., 2018, 2020; Meyer et al., 2019). At Bidart (southwestern France), one of the best preserved K-Pg transitions in Europe, Hg anomalies are observed in the last meter of the Maastrichtian section and continue in the lower part of the early Danian (Font et al., 2016). The Hg anomalies are associated with an interval of low magnetic susceptibility (Fig. 1) located just below the K-Pg boundary, which also contains akaganeite, constituting another key piece of evidence for Deccan Traps volcanism in this sedimentary interval (see the review in Font and Bond [2021]). However, the high Hg contents have also been attributed to postdepositional Hg enrichment by downward remobilization of Hg deposited at the K-Pg boundary by the Chicxulub impact, or contained in anoxic sediments deposited immediately following the impact (Smit et al., 2016). Understanding the origin of these Hg anomalies is essential to resolving the timing of Deccan Trap eruptions relative to the K-Pg extinction and the Chicxulub impact. We examined the Hg isotope record at Bidart, and our results show that enhanced volcanism initiated prior to, and extended across, the K-Pg extinction boundary.

The Bidart section crops out along the Erreteguia beach 2 km north of the city of Bidart and can be accessed by highway R.N.10 (43°26′N, 1°35′W). It consists of hemipelagic to pelagic sediments deposited at upper-middle bathyal depths in the Basque-Cantabric Basin in southwestern France. The Maastrichtian interval is dominated by marls and calcareous marls, whereas Danian sediments are composed of pink and white biogenic limestone beds. The K-Pg boundary is identified in the field by a distinctive thin ∼1-cm-thick gray clay layer containing the globally recognized Ir anomaly (Bonté et al., 1984). Font et al. (2016) presented Hg concentration data from 37 samples (green points in Fig. 1).

We complemented our previous work by analyzing the bulk and isotopic Hg composition and total organic carbon (TOC), Ir, W, Cu, and Zn contents of 44 additional samples located in the upper Maastrichtian and Danian intervals of the Bidart section (Table S1 in Supplemental Material¹). Hg content was measured by atomic absorption spectrometry using a DMA-80 direct Hg analyzer (Milestone Srl, Italy) from MWS/MLS GmbH (Germany) at the GeoZentrum Nordbayern, Erlangen, Germany. Ir, W, Cu, and Zn concentrations were determined by quadrupole inductively coupled plasma–mass spectrometry (ICP-MS) at the GeoZentrum Nordbayern. Mercury isotope ratios were determined by multicollector (MC) ICP-MS (Nu-Plasma, Nu Instruments) at the Institute of Surface-Earth System Science of Tianjin University, China, following a method similar to Huang et al. (2015). Detailed analytical methods are available in the Supplemental Material.

Our new data show that Ir contents are generally below detection limit (0.40 ppb) in Maastrichtian samples and shift abruptly to values of 3.83 and 4.15 ppb at the K-Pg boundary (Fig. 1). These values are consistent with the concentration of 6 ppb previously found by Bonté et al. (1984). Ir concentrations gradually decrease in the next 50 cm above the K-Pg boundary, down to values lower than 0.60 ppb in the upper part of the early Danian carbonates. Other more-mobile elements like Zn, W, and Cu behave similarly to Ir and show no enrichment relative to background values below the K-Pg boundary.

For comparison, Figure 2 illustrates previously reported Hg content (Font et al., 2016) together with the new Hg measurements obtained here. Hg enrichments start in the upper Maastrichtian interval, ∼75 cm below the K-Pg boundary, and end in the lower part of the early Danian, ∼40 cm above the K-Pg boundary (Table S1). Two samples from the clay boundary yielded values of 43.7 and 44.3 ppb (21.8 ppb in the clay layer analyzed by Font et al. [2016]), while maximum values of 62.5 and 65.7 ppb were observed in samples located just above and below the boundary, respectively (Table S1). Because organic matter content is very low (TOC <0.15%), normalization of Hg by TOC is not needed.

In the basal Maastrichtian marls, Δ¹⁹⁹Hg shows positive values between 0.06‰ and 0.24‰ (2σ < 0.10‰). In the 50 cm above the K-Pg boundary, Δ¹⁹⁹Hg shows slightly more positive values of 0.16‰–0.46‰ (2σ < 0.05‰). The δ²⁰²Hg values are −2.17‰ to −0.12‰ in the Maastrichtian and lower part of the early Danian (biozones P0–P1a[1] in Fig. 2), whereas upper Danian carbonates show more negative δ²⁰²Hg values down to −4.87‰ (2σ < 0.31‰).

Studies of modern systems have shown that Hg, as well as other metals like Cu and Zn, is quite immobile and exhibits stable sediment profiles once deposited, except in some marine settings with extremely low sedimentation and mixing rates and organic matter–poor lakes (Gobeil et al., 1999; Feyte et al., 2012; Percival and Outridge, 2013). At Bidart, Hg could have been deposited by the impact or enriched/scavenged by anoxia on the seafloor and subsequently penetrated into uppermost 10–50 cm of the Maastrichtian interval (Smit et al., 2016). Ir has also been considered to be an element of relatively low mobility, but it can be remobilized by postdepositional processes and changes in sedimentary redox conditions (see the review in Evans and Chai [1997]; Martin-Peinado and Rodriguez-Tovar, 2010). In well-preserved K-Pg boundary records, Ir and other platinum group elements exhibit peak concentrations within the clay boundary layer, but anomalies are also described both above and below the boundary. High Ir concentrations above the clay boundary, as observed at Bidart, may result from erosion and redistribution of Ir-rich fallout deposited onto continental surfaces after the impact (e.g., Preisinger et al., 1986). Extension of the Ir anomaly below the clay boundary is better explained by physical reworking by bioturbation (Pospichal et al., 1990; Zhou et al., 1991), or remobilization during diagenetic alteration or the existence of a reducing environment at the time of deposition (e.g., Colodner et al., 1992). At Bidart, the absence of any Ir anomaly immediately below the K-Pg clay boundary demonstrates no downward Ir remobilization. W, Zn, and Cu show similar behavior to Ir, with low concentrations in the Hg-enriched sediments below the impact layer (Fig. 1). We conclude then that the Hg enrichment below the K-Pg boundary at Bidart did not result from the impactor, or remobilization of Hg from the boundary clay, and thus requires an alternative explanation.

Hg is derived from sources such as volcanoes, natural combustion of coal, biomass burning, and asteroid impacts. Thus, there could be a mixed source of Hg. Asteroid impactors can contain significant amounts of Hg, albeit with a large variability within the same meteorite and depending on the type of meteoritic material. Lauretta et al. (2001) reported bulk Hg abundance of ∼294 ppb and ∼30 ppb in the Murchison and Allende meteorite carbonaceous chondrites, respectively. Meier et al. (2016) measured Hg abundances between 12 ± 2 ppb (meteorite Dhofar 1432) and 13,600 ± 5600 ppb (meteorite Orgueil) for carbonaceous chondrites. The Hg concentrations of Meier et al.'s sample of Murchison yielded concentrations of ∼2500 ppb; i.e., one order of magnitude higher than that reported by Lauretta et al. (2001). Meier et al. (2016) showed that the large variability of Hg content found in different samples from the same meteorite is not due to terrestrial Hg contamination, but is rather controlled by the strongly heterogeneous distribution of the Hg-carrying phases at the centimeter scale. Despite the large differences in bulk Hg abundances in individual meteorites, it is possible that part of the Hg observed at, and/or immediately above, the K-Pg boundary came from the Chicxulub impactor. In many K-Pg sections, including Bottacione in Italy, Elles in Tunisia, and Stevns Klint in Denmark, the highest Hg contents and Hg/TOC ratios are observed at the clay boundary (Sial et al., 2013; Keller et al., 2020). At Bidart, Hg anomalies start 75 cm below the K-Pg boundary, with the highest values in the upper Maastrichtian interval, and anomalies continue into the lower part of the early Danian interval (Font et al., 2016; this study; Fig. 1). There is no distinct Hg enrichment at the K-Pg boundary clay.

Mercury has a complex and active biogeochemical cycle, including redox, biological, and phase transformations, which may induce mass-dependent fractionation (MDF) and/or mass-independent fractionation (MIF) of Hg isotopes, providing useful information about the sources and the mechanisms responsible for Hg enrichments (Bergquist and Blum, 2009; Blum et al., 2014; Grasby et al., 2019). Although the exact carbonaceous chondrite type of the Chicxulub impactor is not known, nor how representative the MDFs measured in individual meteorites really are for asteroid-sized bodies (Meier et al., 2016), Hg isotopes could help to distinguish the volcanic and meteoritic contributions to the Hg anomalies found at Bidart. For instance, Hg emitted by volcanoes generally have no MIF (noted as Δ¹⁹⁹Hg), whereas carbonaceous chondrites have a MIF signature (Δ¹⁹⁹Hg excess). Both have MDF (noted as δ²⁰²Hg) signatures, but whereas volcanic gas and particulates have MDF of −1.74‰ ± 0.36‰ for plume gaseous elemental Hg⁰_(g) and −0.11‰ ± 0.18‰ for plume particulate Hg^II_(p) (Zambardi et al., 2009), carbonaceous chondritic meteorites possess much more negative values of −7‰ to −1‰ (Meier et al., 2016; Moynier et al., 2020). Once emitted into the atmosphere, Hg⁰_(g) is readily oxidized into Hg²⁺ species or directly absorbed by vegetation and soils. Aqueous photoreduction of Hg²⁺ in cloud droplets and surface waters induces large MIF, leading to positive Δ¹⁹⁹Hg values in the residual aqueous Hg²⁺ pool (Bergquist and Blum, 2007). Therefore, atmospheric Hg²⁺ and sediments dominated by atmospheric Hg²⁺ deposition generally have δ²⁰²Hg values close to 0 and slightly negative (∼−2‰ to 0‰) or positive Δ¹⁹⁹Hg values. In contrast, in terrestrial reservoirs, additional MIF and MDF occur during the sequestration of Hg⁽⁰⁾_(g) by plant and soils, leading to negative Δ¹⁹⁹Hg values and much more negative δ²⁰²Hg values than atmospheric Hg⁽⁰⁾_(g). The sediments carrying the Hg anomalies at Bidart have positive Δ¹⁹⁹Hg (0.08‰ ± 0.06‰ to 0.46‰ ± 0.00‰) and slightly negative δ²⁰²Hg (−2.01‰ ± 0.00‰ to −0.12‰ ± 0.07‰) values. These values are consistent with a volcanic origin for the Hg anomalies (Fig. 2), where positive Δ¹⁹⁹Hg values reflect photoreduction of Hg²⁺ during atmospheric transport (Grasby et al., 2019). The δ²⁰²Hg values of the Bidart sediments are also consistent with volcanic gases and particulate values like those of Vulcano (Italy) (Zambardi et al., 2009). Based on these isotope data, we argue that the Hg anomalies initiating prior to, and extending across, the K-Pg boundary are related to greatly enhanced volcanism at that time, with the most reasonable source being the Deccan Traps, which are dated to have been erupting contemporaneously (Schoene et al., 2019).

Hg isotope compositions of meteorites are less well documented (Lauretta et al., 1999, 2001; Meier et al., 2016). Lauretta et al. (2001) studied the Murchison and Allende carbonaceous chondrites and suggested that the relative abundances of Hg isotopes in both meteorites are identical to terrestrial δ²⁰²Hg values within 0.20‰–0.50‰. Conversely, Meier et al. (2016) showed that most carbonaceous chondrites exhibit slight MIF of the odd-numbered isotopes ¹⁹⁹Hg and ²⁰¹Hg (∼–0.1‰ < Δ¹⁹⁹Hg < ∼0.4‰; ∼–0.18‰ < Δ²⁰¹Hg < ∼0.23‰), while noncarbonaceous chondrites and most achondrites show no MIF. Distinctively, all meteorites studied by Meier et al. (2016) showed isotopically light Hg (δ²⁰²Hg =–7‰ to −1‰) relative to Earth's average crustal values, suggesting that Earth has lost a significant fraction of its primordial Hg during planetary differentiation or during late accretion of chondritic materials to Earth (Moynier et al., 2020), resulting in different Hg isotope signatures between the surface and deep-Earth reservoirs. At Bidart, δ²⁰²Hg values in the Danian sediments above 50 cm contrast with the Hg isotopic composition of the Maastrichtian–lower Danian interval (Fig. 2), showing a large shift to negative values typical of carbonaceous chondrites, possibly representing remobilized impactor Hg that was redeposited on the seafloor. Thus, while our data may show an Hg isotope signature of the impactor, the very low Hg concentrations indicate it could not be the source of the Hg enrichments observed below the impact level. The high Δ¹⁹⁹Hg, Δ²⁰¹Hg, and δ²⁰²Hg values observed above the boundary may also represent a period of increased photoreduction during the time when there was strong Hg mobilization from biomass burning, volcanic eruptions, and impactor input. The slightly but significantly different isotopic compositions in Δ¹⁹⁹Hg and Δ²⁰¹Hg within the 50 cm above the boundary and ∼100 cm below the boundary virtually exclude the interpretation that the Hg found in the lower part was simply mixed down from the boundary level.

The results of our study reveal that Deccan Traps eruptions led to enhanced Hg deposition prior to and during the Chicxulub asteroid impact, and continued into the earliest Danian. This period of intense volcanism is consistent with recent high-precision U-Pb ages that place the largest eruptions of the Deccan Traps prior to or straddling the K-Pg boundary (Schoene et al., 2019). Deccan Traps volcanism clearly had a global effect, and whether the Chicxulub impact was the proverbial “straw that broke the camels back”, or was an inconsequential random event that occurred during an ongoing extinction, remains to be elucidated.

This work was funded by Natural Science Foundation of 225 China (grants 41625012, 41961144028, 41830647) (J. Chen) and the Fundação para a Ciência e Tecnologia / UIDB/50019/2020–IDL (E. Font). We thank Stephen Grasby, Matthias Meier, and Hamed Sanei for their help in improving the quality of the manuscript.