Limited sulfur degassing and muted environmental impact of Ontong Java Plateau lavas

Large igneous provinces (LIPs) are the largest volcanic events on Earth, emplacing 10⁵–10⁷ km³ of lava over 1–2 m.y. The largest LIP in the geologic record is the Ontong Java Plateau (OJP) with a total crustal volume of ~3–6 × 10⁷ km³ (Fig. 1A; Coffin and Eldholm, 1994). Main emplacement occurred ca. 123–121 Ma, with a second, disputed minor phase ca. 90 Ma (Tejada et al., 1996; Chambers et al., 2004); however, a recent study suggests younger dates of 117–108 Ma for the main emplacement of the OJP (Davidson et al., 2023). LIPs emit vast amounts of volatiles with potentially severe environmental consequences, motivating proposed links to multiple mass extinctions (Rampino and Stothers, 1988; Bond and Wignall, 2014).

CO₂ and SO₂ are the volcanic volatiles that exert the strongest controls on climate, with CO₂ driving long-term warming and SO₂ driving short-term cooling (Bond and Wignall, 2014; Schmidt et al., 2016; Black et al., 2018). Early Cretaceous warming, ocean acidification, and Oceanic Anoxic Events 1a and 1b have been attributed to CO₂ emissions during OJP emplacement (Tejada et al., 2009; Erba et al., 2015; Davidson et al., 2023). It is curious that despite its apparently large CO₂ emission rates, which are suggested to play a role in driving mass extinctions (Clapham and Renne, 2019; Green et al., 2022), and the documented environmental consequences of CO₂ release, OJP emplacement did not coincide with a mass extinction (Erba et al., 2015). Unlike CO₂, submarine pressure (P) may be high enough to suppress S degassing, effectively decoupling S degassing from CO₂ emissions during submarine eruptions (Gaillard et al., 2011; Green et al., 2022). Therefore, limited S emissions from the OJP may be connected to the absence of a coeval mass extinction (Roberge et al., 2004; Reekie et al., 2019)—a link that has been proposed for S-poor continental LIPs (Callegaro et al., 2014).

Previous studies that measured [S] (where [ ] denotes concentration) in OJP glasses noted high [S] and suggested limited S degassing (Roberge et al., 2004; Reekie et al., 2019). However, these studies, and a study that included measurements of [S] in recrystallized melt inclusions (Jackson et al., 2015), did not attempt to quantify OJP emissions. Here, we present new measurements of [S] in pillow basalt glasses and the first [S] measurements of naturally glassy, olivine-hosted melt inclusions (MIs) from the OJP. We compared the new measurements of [S] in MIs and pillow basalt glasses to estimates of initial [S] to quantify S emissions during OJP emplacement. We also compared the differences in S emissions between the OJP and continental LIPs to test the hypothesis that limited S emissions at OJP could have played a role in the muted environmental impact of the largest LIP in the geologic record.

OJP lavas were sampled during Ocean Drilling Program (ODP) Leg 192, at Deep Sea Drilling Program Site 289, and at ODP Sites 803 and 807 (Fig. 1A; Mahoney et al., 2001). Thin sections of exceptionally fresh, glassy pillow basalt rims from Site 1187 (ODP Leg 192) included naturally glassy, bubble-free, olivine-hosted MIs (Fig. 1). Major element, S, and Cl concentrations in glassy rims and olivine-hosted MIs were measured via electron probe micro-analysis on the JEOL JXA-8200 Superprobe at Rutgers University (Tables S1 and S2 in the Supplemental Material¹).

Sulfur concentrations in OJP pillow basalt glasses and MIs (corrected for postentrapment crystallization; see Supplemental Material) from this study, along with previous measurements of [S] in OJP glasses, are shown in Figure 1. Sulfur concentrations in glasses covered a wide range of ~600–1400 ppm, while sulfur concentrations in MIs covered a smaller range of ~800–1000 ppm. Sulfur concentration appears to increase as [MgO] decreases, with a change in slope at ~8 wt% MgO. Several processes control [S] in silicate melts, including fractional crystallization, sulfide saturation, and degassing (Reekie et al., 2019). In the following, we discuss how these processes worked in concert to control [S] in the OJP melts.

To understand S degassing, we first considered fractional crystallization and sulfide saturation. Figure 1 shows [S] in the melt calculated via bulk partition coefficients for S (Callegaro et al., 2020) along the liquid line of descent and calculated [S] at sulfide saturation (SCSS; see Supplemental Material). At [MgO] >~8 wt%, [S] in OJP glasses slightly increases with decreasing [MgO] and plots along trends controlled by fractional crystallization, while also being significantly lower than SCSS (Fig. 1). Therefore, it is likely that high-MgO OJP melts were not sulfide saturated, and [S] was controlled by olivine fractionation, consistent with chalcophile systematics (Reekie et al., 2019).

At [MgO] <~8 wt%, clinopyroxene and plagioclase began to crystallize (Fig. 1), and a given decrease in [MgO] required more crystal fractionation, increasing the slope of [S] controlled by fractional crystallization. In contrast to high-MgO glasses, which followed the trend of [S] controlled by fractional crystallization with [S]_i = 775 ppm, no glasses with ≤8 wt% MgO reached [S] controlled by fractional crystallization with [S]_i = 775 ppm (Fig. 1). A single low-MgO sample, which we suspect assimilated seawater, exceeded [S] controlled by fractional crystallization with [S]_i = 775 ppm (see Supplemental Material). In glasses with <~8 wt% MgO, the highest [S] followed SCSS, suggesting sulfide saturation. Increases in [S] driven by fractionation of plagioclase and clinopyroxene likely drove sulfide saturation at [MgO] <~8 wt% (Reekie et al., 2019). In support of this interpretation, we only observed sulfides in samples with abundant plagioclase (Fig. 1B).

To quantify S degassing, we compared [S] of OJP glasses to S degassing curves modeled with Sulfur_X (Supplemental Material; Ding et al., 2023) to determine the samples that likely degassed S. For both high- and low-MgO samples, Sulfur_X predicted the onset of significant S degassing at P ~150 bars and tracked [S] in OJP glasses with low eruption P (assumed to equal saturation P calculated with [CO₂] and [H₂O] using the model of Dixon [1997] implemented in VESIcal [Iacovino et al., 2021]; Fig. 2). At P ≥ 150 bars, Sulfur_X predicted minimal S degassing, and the observed spread in [S] was likely controlled by fractional crystallization and sulfide saturation, as illustrated by increasing [S] with decreasing [MgO] (see Fig. 1). Thus, we considered samples with eruption P > 150 bars to be undegassed and samples with eruption P < 150 bars to be degassed with respect to S. Our new MI data also support minimal S degassing at Site 1187. If S degassing was important at Site 1187, we would expect pillow basalt glasses to have significantly lower [S] values compared to MIs from Site 1187, which they do not (Fig. 1).

To quantify S degassing, we calculated the difference between [S] in OJP glass and inferred initial [S]. We treated initial [S] as a function of [MgO], set by fractional crystallization at high MgO and SCSS at low MgO (Fig. 1). S degassing efficiency was defined as:

Assuming S degassing efficiencies = 0% for glasses with eruption P > 150 bars, the average S degassing efficiency of OJP glasses = 5% (Fig. 3), i.e., an order of magnitude less than the 75%–90% estimated for continental flood basalts (Black et al., 2012). If this assumption is ignored, average degassing efficiency increases to only 10% (Fig. S4). In support of our interpretation of minimal S degassing at P > 150 bars, increasing S degassing efficiency with decreasing eruption P was only observed in sample sets emplaced at P < 150 bars (Fig. S3).

It remains important to consider whether the large volume of the OJP could have resulted in S emissions similar to continental LIPs. If the OJP released a similar mass of S as the Siberian Traps, which has been implicated as a trigger for Earth’s largest mass extinction (Dal Corso et al., 2022), it would suggest that differences in S emissions alone cannot explain the differences in environmental impact between the two LIPs. Volume estimates of OJP extrusive products are ~1 × 10⁷ km³, while total crustal volume of the greater OJP is estimated at ~1 × 10⁸ km³ (i.e., including coeval Manihiki and Hikurangi Plateau and Nauru and East Mariana Basins; Fig. 1A; Coffin and Eldholm, 1994; Chandler et al., 2012). In the following, we explore the range of volume estimates for OJP; however, it is likely that only the extrusive products (i.e., 1 × 10⁷ km³) released appreciable amounts of S, due to the high lithostatic P of intrusive magmas (>>150 bars). With volume ≈1–10 × 10⁷ km³, initial [S] = 1000 ppm, and S degassing efficiency = 5%, the OJP would have emitted 1400–14,000 Gt S, as compared to the 6300–7800 Gt of magmatic S emissions estimated for the Siberian Traps (Black et al., 2012). However, we interpret the average S degassing efficiency (5%) to be a maximum value, with the true value being much lower. This is because high S degassing efficiencies of samples with eruption P < 150 bar are weighted equally in calculating an average S degassing efficiency of 5%, when they may have contributed only a small fraction to the total volume of the OJP. It is suggested that the volcaniclastic sequence at Site 1184 may have been on the order of 10¹–10² km³ (Thordarson, 2004), which would make its contribution to S degassing negligible. Additionally, over a depth interval of <100 m near the crest of the OJP edifice, S degassing efficiencies at Site 1183 decrease from ~20 to ~0% (Fig. S3). This may imply that OJP lavas emplaced at depths >100 m from the crest of the edifice had S degassing efficiencies of ~0%. Thus, we interpret 5% as a maximum S degassing efficiency for OJP and infer the true value is closer to 0%.

Our preferred interpretation (that OJP degassed <<5% S) implies S emissions <<1400–14,000 Gt. For example, if 10% of the OJP’s volume was emplaced at <150 bars, which is likely an overestimate, and we weigh its contribution accordingly, average S degassing efficiency = 0.5%, which would suggest the OJP emitted 140–1400 Gt S. This illustrates that even when the total volume of the greater OJP is considered, and surely when only the extrusive products are considered, OJP released significantly less S relative to continental LIPs like the Siberian Traps. Therefore, suppressed S emissions may have played a role in the muted environmental impact of the OJP, considering S emissions have been shown to cause rapid and drastic transitions in climate and ocean circulation when combined with CO₂ emissions (Black et al., 2018).

Results suggest a spatiotemporal evolution in degassing behavior during OJP emplacement. Early OJP eruptions emplaced at water depths >~1500 m (150 bars) (Figs. 2 and 4) erupted at high enough P to suppress S degassing while allowing significant CO₂ degassing. Growth of the volcanic edifice caused later OJP lavas to erupt at shallower water depths and lower P, allowing increased S degassing efficiencies of ~20%–30% (Figs. 3 and 4). It is also possible that some late-stage eruptions occurred on the flanks of the plateau, resulting in high eruption pressures and low S degassing efficiencies. Crustal thickening via sill emplacement and dynamic uplift from a plume may have also decreased eruption P of later OJP lavas, though the existence of a plume at OJP has been questioned (Korenaga, 2005).

Shallow-water OJP eruptions were likely explosive, as evidenced by volcaniclastic sequences at Site 1184 (Thordarson, 2004). Eruption column heights associated with flood basalt phreatomagmatic activity are uncertain (Ross et al., 2008), and SO₂ scrubbing in H₂O-rich volcanic plumes produced during submarine eruptions may decrease SO₂ delivery to the atmosphere (Carn et al., 2022). If, however, volcaniclastic sequences at Site 1184 resulted from vigorous explosive activity, then these eruptions may represent a narrow window where OJP magmas could have injected S higher in the troposphere, where its climatic impact would have been greater (Schmidt et al., 2016). However, delivery of S to the upper troposphere may have been impeded due to OJP emplacement at paleolatitudes of ~25°–50°S (Chandler et al., 2012) because of higher tropopause altitudes in mid-to low latitudes.

The temporal evolution in degassing behavior proposed for the OJP may provide insights into volatile release at other oceanic plateaus, and even some continental LIPs. Continental LIPs have been proposed to degas significant CO₂ during crustal intrusion, decoupling CO₂ emissions from rates of surface volcanism (Hernandez Nava et al., 2021; Tian and Buck, 2022). Due to the high lithostatic P of crustal intrusions (>>150 bars), S degassing may be minimal, like the early OJP eruptions. This comes with the caveat that the intrusive phase of continental LIPs may devolatilize surrounding S-rich sediments (Yallup et al., 2013), though explosive diatremes (Polozov et al., 2016) may be required for this process to have significant climatic effects. Magmatic S degassing would only become important during extrusive phases of continental LIP volcanism due to emplacement at atmospheric P. Shifts between high CO₂ plus low S degassing during intrusive volcanism and high CO₂ plus high S degassing during extrusive phases may have important implications for the tempo of environmental impacts of continental LIP volcanism (Green et al., 2022).

We interpret [S] in OJP MIs and pillow basalt glasses to result from fractional crystallization at [MgO] >~8 wt% and sulfide saturation at [MgO] <~8 wt%. We suggest that only OJP lavas that erupted at water depths <~1500 m degassed S, and due to the volumetrically minor fraction of OJP lavas emplaced at these depths, total S emissions may have been much lower compared to continental LIPs. Suppressed S degassing of lavas from the OJP may have curtailed its environmental impact compared to smaller continental LIPs, suggesting that a combination of CO₂ and SO₂ emissions may deepen ecological stress. Indeed, freezing temperatures resulting from LIP-derived sulfate aerosols have been implicated in the extinction selectivity of land animals and the rise of dinosaurs during the end Triassic mass extinction (Olsen et al., 2022). There was likely a temporal evolution in S emissions at the OJP, with early, deep-water eruptions suppressing S degassing and later shallow-water/subaerial eruptions degassing S more efficiently. This P-sensitive evolution in S degassing may be more broadly applicable to other LIPs.

We thank Paul Burger for his assistance with electron probe micro-analysis. B.A. Black acknowledges National Science Foundation (NSF) grant EAR 2238441, and S. Ding acknowledges NSF grant EAR 2017814. We thank Stephen Self and two anonymous reviewers for their constructive comments that improved the manuscript. We thank Urs Schaltegger for editorial handling.