Ash deposits link Oceanic Anoxic Event 2 to High Arctic volcanism

Oceanic Anoxic Event 2 (OAE 2) was a global carbon cycle perturbation (Arthur et al., 1988) that occurred near the Cenomanian–Turonian boundary at ca. 93.9 Ma (Jones et al., 2021). It is associated with rapid global warming (e.g., Robinson et al., 2019), ocean acidification (e.g., Jones et al., 2023), and widespread deoxygenation of the deep ocean (e.g., Clarkson et al., 2018), leading to ecosystem collapse and the extinction of marine biota (e.g., Parente et al., 2008). These changes are thought to have been driven by a large and rapid release of greenhouse gases from large igneous province (LIP) volcanism (e.g., Barclay et al., 2010), a hypothesis supported by evidence from seawater osmium isotopes (¹⁸⁷Os/¹⁸⁸Os; Turgeon and Creaser, 2008).

However, the specific LIP(s) responsible for triggering OAE 2 remain debated, partly due to the scarcity of reliable, high-resolution geochronology data for potential candidate LIPs, including the Caribbean LIP, the Madagascar Province, the Kerguelen Plateau (KP, southern Indian Ocean), the Greater Ontong Java Plateau (southwestern Pacific Ocean), and the High Arctic LIP (HALIP; Fig. S1 in the Supplemental Material¹). A recent study (Walker-Trivett et al., 2024) suggests that KP volcanism played a key role in triggering OAE 2, based on mercury enrichment in nearby sediments before and during the event. However, numerical models indicate that an emission of ∼24,000 Pg of volcanogenic carbon is required to explain the carbon cycle perturbation associated with OAE 2 (Papadomanolaki et al., 2022). Published estimates of CO₂ content in KP magma range from 0.0006 to 0.0025 Pg C km⁻³ (Jiang et al., 2022), suggesting that ∼9.5 × 10⁶ to 43.4 × 10⁶ km³ (or 38–176% of the entire KP–Broken Ridge complex; Coffin et al., 2002) would need to have erupted within a few hundreds of thousand years to match these degassing rates. Such rapid emplacement is not supported by radiometric data (Coffin et al., 2002; Jiang et al., 2021), highlighting a significant gap in our understanding of LIP activity and carbon emission dynamics during OAE 2.

We assessed the synchrony of HALIP volcanism and OAE 2 using ash deposits from a mid-Cretaceous sedimentary sequence in the Canadian High Arctic (Fig. 1). Building on field observations of bentonite beds (i.e., clay beds formed by weathering of volcanic ash; Schröder-Adams et al., 2014, 2019), we used high-resolution geochemical and grain-size profiles to pinpoint the exact timing, duration, intensity, and origin of ashfall. This record of subaerial volcanism was integrated with carbon isotope (δ¹³C) and ¹⁸⁷Os/¹⁸⁸Os data (Herrle et al., 2015; Schröder-Adams et al., 2019) to correlate HALIP volcanism with global carbon cycle dynamics and LIP activity trends.

Radiometric ages indicate that the HALIP was emplaced between ca. 130 and ca. 80 Ma, with three major magmatic pulses at 122 ± 2 Ma, 95 ± 4 Ma, and 81 ± 4 Ma (Dockman et al., 2018). In the Canadian High Arctic, HALIP magmatism manifested as flood basalt volcanism, silicic volcanism, and widespread intrusive activity forming the giant Queen Elizabeth dike swarm (Saumur et al., 2016). The second HALIP pulse at ca. 95 Ma broadly correlates with OAE 2 and includes the tholeiitic flood basalts of the Strand Fiord Formation (Ricketts et al., 1985) and their plumbing system (Kingsbury et al., 2018); alkaline flood basalts within the Hassel Formation (Osadetz and Moore, 1988) and associated sills and dikes of the Fulmar Suite (Bédard et al., 2021); as well as a volcanic-plutonic belt extending from Audhild Bay to Yelverton Bay on northern Ellesmere Island, which includes the predominantly silicic Audhild Bay Volcanic Complex and the bimodal Wootton Intrusive Complex (Fig. 1; Fig. S2; Embry and Osadetz, 1988). HALIP volcanism produced widespread ashfall, indicated by frequent bentonite beds within the marine sediments of the Kanguk Formation (Davis et al., 2017; Pointon et al., 2019), which was deposited on the Canadian shelf of the Arctic Ocean from Late Cenomanian times onward (Schröder-Adams et al., 2014; Herrle et al., 2015).

We studied a composite section of the basal Kanguk Formation at Glacier Fiord on southern Axel Heiberg Island (Fig. 1; 78°37.787′N, 89°54.136′W), which provides an expanded and complete record of OAE 2 with excellent age control. Sampling focused on mudstone intervals between bentonite beds, along with two bentonite beds sampled for U-Pb dating. The new U-Pb ages (Fig. S3) were combined with previously published ages (Davis et al., 2017) to construct a Bayesian age model (Fig. S4). Three legacy bentonite samples were analyzed for trace element composition to constrain volcanic sources (Fig. 2).

Mudstone samples were analyzed using a multi-proxy approach to detect and quantify dispersed volcanic ash. Absolute element concentrations were calculated from X-ray fluorescence (XRF) core scanning counts. Si/Ti and Zr/Ti ratios serve as proxies for ash abundance given the high concentrations of Si and Zr in Kanguk bentonites (Figs. 2 and 3A; Pointon et al., 2019).

To assess the influence of grain-size sorting on ash-diagnostic parameters, we analyzed the grain-size distribution of 58 samples using a laser diffraction particle size analyzer. XRF-derived K/Rb ratios were used as a complementary grain-size proxy that correlates with the sortable silt content (8–63 µm), while being largely insensitive to sediment provenance (Fig. S5).

The existing δ¹³C_org stratigraphy (Herrle et al., 2015; Schröder-Adams et al., 2019) was enhanced by the analysis of 50 additional samples to improve temporal resolution. Details on analytical methods, age modeling, and composite construction are provided in the Supplemental Material.

The distribution of bentonite beds, geochemical data, and grain-size analysis suggests that sediments between ∼15 and ∼75 m above the base of the Kanguk Formation are enriched in volcanic ash (Fig. 2). Ash-diagnostic Si/Ti and Zr/Ti ratios exhibit a consistent increase ∼12 m below OAE 2, coinciding with the first bentonite beds and marking the onset of ash deposition. The geochemical signature of most samples within the ashfall interval is consistent with binary mixing of volcanic ash and hemipelagic background sediment (Fig. 3A), represented by the weathering-corrected composition of bentonites (see the Supplemental Material) and pre-ashfall sediment composition, respectively. Most samples of the ashfall interval closely align with the mixing line (Fig. 3A), indicating little alteration of the pristine volcanic ash signature during transport, deposition, and diagenesis. This unaltered state implies rapid deposition with limited influences from weathering and sorting, supporting the interpretation that ash was deposited via fallout.

Ash abundance decreases from ∼75 m above base, as reflected by the return of Si/Ti and K/Rb ratios to pre-ashfall levels, and the last bentonite beds occur at ∼95 m (Fig. 2). Post-ashfall samples exhibit elevated Zr/Ti ratios (Fig. 2B) and an excess of Zr relative to the mixing line (Fig. 3A), indicating selective enrichment of Zr relative to Si. This pattern is explained by hydrodynamic sorting processes, preferentially retaining coarser Zr-bearing particles while winnowing away finer Si-rich particles. Our interpretation is supported by the well-sorted, unimodal grain-size distribution in post-ashfall sediments and the positive correlation between Zr/Si ratios and K/Rb ratios, reflecting grain-size partitioning of Zr and Si (Figs. S6 and S7). We propose that hydrodynamic sorting occurred during the redeposition of ash-rich sediments from surrounding land masses.

The amount of dispersed ash was quantified using binary mixing models based on Zr/Ti and Si/Ti ratios (see the Supplemental Material). These models indicate dispersed ash contents of up to 12% and a total amount equivalent to ∼3 m thickness (Fig. 4), indicating that subaerial HALIP volcanism was more continuous and intense than the bentonite beds alone suggest (cumulative thickness of ∼1.5 m).

Two volcanic sources have been proposed for the bentonites of the Kanguk Formation: (1) alkaline intra-plate volcanism associated with the HALIP, and (2) active continental margin volcanism, potentially linked to the Okhotsk-Chukotka volcanic belt in Russia (Pointon et al., 2019). To constrain the volcanic source, we analyzed the immobile trace-element geochemistry of three bentonite samples. The geochemical fingerprint of these samples closely matches that of HALIP-derived bentonites, supporting an intra-plate origin (Fig. S8). Comparison with igneous rock data points to the Wootton Intrusive Complex and the Audhild Bay Volcanic Complex on northern Ellesmere Island as likely sources. These complexes have been dated at ca. 93 Ma to ca. 92 Ma (Wootton Intrusive Complex) and and 95.6 ± 1 Ma (Audhild Bay Volcanic Complex) (Fig. 3B), and are located ∼400 km from the study site (Fig. 1), consistent with our age estimate for ashfall and source distance estimates of <1000 km based on bentonite bed thickness (Davis et al., 2017).

Our Bayesian age model places the onset of OAE 2 at 94.02 Ma (93.50–94.77 Ma; 2σ uncertainty range), consistent with previous estimates (Jones et al., 2021). The onset of ash deposition is constrained by two independent stratigraphic approaches. The Bayesian age model suggests that ashfall began ∼160 k.y. (30–500 k.y.; 2σ uncertainty range) before OAE 2. This estimate is corroborated by osmium isotope stratigraphy, which shows a pronounced negative ¹⁸⁷Os/¹⁸⁸Os excursion coincident with the onset of ashfall (Fig. 4). This excursion is a globally recognized stratigraphic marker dated to ∼60 k.y. before OAE 2 (Fig. S9). Together, these lines of evidence suggest that HALIP volcanism commenced tens of thousands of years before OAE 2, suggesting a strong temporal link.

Estimating the size of past volcanic events is challenging due to factors such as erosion, limited exposure, incomplete mapping, and tectonic fragmentation. For the silicic component of HALIP volcanism recorded in sedimentary ash, we estimate a total volume of 2250 km³ (dense rock equivalent of ∼900–1600 km³), based on a conservative approach that considers a distance of ∼400 km to the eruptive centers and a constant ash thickness of 4.5 m over the depositional area of ∼500,000 km². We note that our approach does not account for the typical thickening of ash deposits toward the vent, which would increase the volume. Factoring in equivalent volumes of intracaldera and outflow deposits (Mason et al., 2004) raises the estimate to 2700–4800 km³, which remains modest compared to major LIP events (e.g., Bond and Wignall, 2014).

However, we emphasize that our approach is inherently limited by focusing on the products of explosive silicic volcanism, and thus it is prone to underestimate the total magmatic volume. In fact, silicic volcanism usually constitutes less than 5% of the total eruptive volume in mafic-dominated LIPs (Bryan et al., 2002) and ∼75–85% of the HALIP volume is estimated to be intrusive (Saumur et al., 2016). The exact eruptive history and magmatic volumes of mafic HALIP magmatism are poorly constrained. However, we note that several radiometric ages, including recent high-resolution dates for the feeder dikes of the voluminous Strand Fiord flood basalts (Kingsbury et al., 2018), align with our age estimate for ashfall (Fig. 3B), suggesting that multiple, and potentially interconnected, volcanic centers of the HALIP may have been active at the time of OAE 2.

LIP volcanism is widely considered a major driver of OAE 2 (e.g., Turgeon and Creaser, 2008), with numerical models suggesting the release of ∼24,000 Pg (18,000–46,000 Pg) of volcanogenic carbon before and during the event (Clarkson et al., 2018; Papadomanolaki et al., 2022). There is strong evidence that KP volcanism contributed to this carbon release (Walker-Trivett et al., 2024). However, radiometric ages that align with OAE 2 are largely restricted to the southern and central KP (Jiang et al., 2021), which together have an estimated volume of 13 × 10⁶ km³ (Coffin et al., 2002). Given a CO₂ content of 0.0006–0.0025 Pg C km⁻³ in KP magmas (Jiang et al., 2022), between ∼7800 and 32,500 Pg C could have been released during the emplacement of these parts of the plateau. This would require a nearly instantaneous eruption and complete degassing to trigger OAE 2, which is inconsistent with radiometric data indicating that the KP formed gradually over more than 32 m.y. (ca. 122–90 Ma; Jiang et al., 2021). Therefore, we consider it unlikely that the KP was the sole trigger of OAE 2.

Our results indicate that HALIP volcanism commenced tens of thousands of years before OAE 2, making it a plausible contributor of additional carbon. Consistent with this interpretation, we observe a marked decrease in δ¹³C_org values in response to the onset of HALIP volcanism (Fig. 4), suggesting large-scale release of ¹³C-depleted carbon. Although there is some uncertainty as to whether this signal is global or local—given that some OAE 2 records display similar excursions (e.g., Kuroda et al., 2007; Takashima et al., 2024; Walker-Trivett et al., 2024) while others do not (e.g., Jones et al., 2021)—we invoke two mechanisms to explain the δ¹³C_org decrease: (1) emission of volcanogenic carbon with a δ¹³C value of approximately −5‰, and/or (2) the release of thermogenic gases from contact metamorphism within the HALIP's plumbing system (Schröder-Adams et al., 2019). The latter mechanism represents a substantial carbon source within many continental LIPs (Svensen et al., 2023). Evidence for contact metamorphism is widespread in organic-rich sediments intersected by HALIP intrusions (Goodarzi et al., 2019), and thermal modeling suggests a high potential for carbon release of up to 6 × 10⁷ g km⁻² (Bédard et al., 2023).

We conclude that HALIP activity may have released significant amounts of carbon, contributing to the rapid global warming and carbon cycle perturbation associated with OAE 2. The concurrent activity of the HALIP and the KP may account for the exceptionally high rates of carbon release during OAE 2, making it one of the most severe environmental crises of the Mesozoic.

This research was funded by the German Research Foundation (grant HE3521/5-2 to J. Herrle). Research licenses have been granted by the Nunavut (Canada) Research Institute, the Nunavut Water Board, and the Nunavut Department of Culture and Heritage. We thank the community of Grise Fiord for their agreement to conduct research on northern lands. C. Schröder-Adams and J. Herrle thank the Canadian Polar Continental Shelf Program for logistical support during fieldwork. Alex Quesnel and Keenan Lindell are thanked for field support. Corey Wall (Pacific Centre for Isotopic and Geochemical Research, Vancouver, Canada) provided U-Pb geochronology. ICP-MS analysis was performed by Geoscience Laboratories (Sudbury, Ontario, Canada). We would like to thank Reishi Takashima and the two anonymous reviewers for their insightful comments on an earlier version of this manuscript.