The termination of Neoproterozoic “Snowball Earth” glaciations is marked globally by laterally extensive neritic cap carbonates directly overlying glacial diamictites. The formation of these unique deposits on deglaciation calls for anomalously high calcium carbonate saturation. A popular mechanism to account for the source of requisite ocean alkalinity is the shallow-ridge hypothesis, in which initial spreading ridges surrounding fragments of Rodinia, assumed to be dominated by volcanic margins, were formed at sea level. The shallow ridges are inferred to have promoted widespread deposition and alteration of glassy hyaloclastite—a source of alkalinity. We test this hypothesis by quantifying the prevalence of shallow ridges along Pangea's passive continental margins, and by assessing Neoproterozoic reconstructions of tectonic plates. We find that the most frequently occurring depth range for incipient mid-ocean ridges is 2.1 ± 0.4 km. Ridges with initial elevations of approximately sea level are rare and have anomalous crustal thicknesses >14 km that only occur proximal to large igneous provinces (LIPs). Hyaloclastite is uncommon on mid-ocean ridges as it is generally restricted to water depths of <200 m for tholeiitic basalts, instead forming mostly on intraplate seamounts. Additionally, ocean drilling recently found hyaloclastite to be insignificant along the outer Vøring Plateau (offshore Norway)—an exemplar of a volcanic margin. Reconstructions of Rodinia and associated LIPs demonstrate that volcanic margins potentially hosting minor hyaloclastites were scarce during the late Neoproterozoic. We conclude that the shallow-ridge hypothesis fails to explain the formation of cap carbonates and suggest that other mechanisms such as enhanced continental weathering may be largely responsible.

The Neoproterozoic era was punctuated by three significant glaciations. The Sturtian (ca. 717–661 Ma) and the Marinoan (ca. 646–635 Ma) global glaciations were the most severe in Earth's history, with sea ice extending all the way to the equator in a “Snowball Earth” scenario (Hoffman et al., 2017). The Gaskiers glaciation (ca. 580 Ma), in contrast, was a short-lived (≤340 k.y.), mid-latitude, regional event, comparable to Cenozoic glaciations (Pu et al., 2016). Common to all Neoproterozoic glaciations is the global occurrence of cap carbonates—laterally continuous layers of neritic limestone or dolostone immediately overlying glaciogenic deposits or associated erosional surfaces (Grotzinger and James, 2000; Shields, 2005; Hoffman, 2011). The cap carbonates preserve a unique δ13C negative excursion (e.g., Kennedy, 1996; Hoffman et al., 2017) and display unusual textural and compositional features that distinguish them from other carbonate rocks (Kennedy, 1996; Grotzinger and James, 2000; Shields, 2005). Cap carbonate deposition is associated with shallow water supersaturated in carbonate, a warming climate, and continental flooding caused by a eustatic sea-level rise following deglaciation (e.g., Kennedy, 1996; Hoffman et al., 2017). Formation of these puzzling deposits remains controversial, with multiple competing mechanisms proposed concerning the source of the alkalinity and its delivery to the sites of carbonate deposition (Shields, 2005). A novel mechanism put forward by Gernon et al. (2016) and considered plausible by many (e.g., Hoffman et al., 2017; Youbi et al., 2020; Hood et al., 2022) posits that the weathering of hyaloclastite (glassy debris formed by subaqueous magmatic eruptions) along shallow spreading ridges during the break-up of the supercontinent Rodinia supplied alkalinity for cap carbonate deposition via syn-glacial alteration of hyaloclastite to palagonite—a product of volcanic glass hydration (Gernon et al., 2016; Fig. 1). We test the validity of this hypothesis using a quantitative approach to assess the prevalence of anomalously shallow spreading ridges during the break-up of Pangea as a proxy for the break-up of Rodinia. We subsequently use Cryogenian reconstructions of continents, plate boundaries, and large igneous provinces (LIPs) following the break-up of Rodinia to show that the shallow-ridge hypothesis is fundamentally flawed.

For our analysis, we first create an updated set of preserved boundaries between stretched continental and ocean crust (COBs) (see the Supplemental Material1 for information on the methodology and global gravity data set used in this study). We choose COB data at 200 km intervals along the COBs and select points 50 km seaward of each COB to sample the age of the ocean crust (Seton et al., 2020; Fig. 2A) and its crustal thickness (Reguzzoni and Sampietro, 2015; Fig. 2B). The distance of 50 km from the interpreted boundary of stretched continental crust was chosen to ensure that we sample crustal thickness based on a grid cell overlying pure ocean crust, given a grid resolution of 0.5 degrees (Reguzzoni and Sampietro, 2015). We then use PyBacktrack (; Müller et al., 2018) to reconstruct the initial elevation of the seafloor (Fig. 2C) shortly after the onset of seafloor spreading at these points, removing subsequently deposited sediments taken from a global sediment thickness grid (Straume et al., 2019). These initial depths are computed with and without long-term sea-level variations and mantle convection–driven dynamic topography (Young et al., 2022). We consider the difference between modeled dynamic topography at the present day and at the initiation of seafloor spreading (Fig. 2D), using the plate rotation model employed by Young et al. (2022), which is based on the model of Merdith et al. (2021) (see the Supplemental Material).

Our assessment of the shallow-ridge hypothesis begins with preserved ocean crust along passive continental margins, where break-up tectonic history is well constrained (Müller et al., 2019). The analysis (Fig. 3) allows us to consider the subsidence history of passive margins that underpin the validity of the shallow-ridge hypothesis (Gernon et al., 2016). Our aim is to test a critical assumption of the hypothesis that the ridge crest in the initial stage of seafloor spreading after break-up is located at sea level (i.e., basement depth is 0) and likely persists at depths of <2 km for 30–35 m.y. after the onset of mid-ocean ridge formation (supplementary information fig. 2 in Gernon et al., 2016).

We find that the present-day age of the initial ocean crust ranges from the Late Paleozoic in the eastern Mediterranean to the Neogene along young rifts, with the majority of break-up ages clustered between the early Jurassic (ca. 200 Ma) and the Late Eocene (ca. 35 Ma), reflecting the protracted break-up of Pangea (Fig. 2A). The initial oceanic basement paleo-elevation ranges from a depth of 4 km to shallow subaerial elevation (<0.5 km). The most frequently occurring depths are centered at 2.1 ± 0.4 km (Fig. 3A), which represents an offset toward shallower values from the initial basement depth of 2.6 ± 0.3 km computed by Richards et al. (2018). We primarily attribute this offset to shallowing of the initial paleo-elevation due to subsequent dynamic subsidence along most of Pangea's passive margins (Figs. 2D and 3C; see the Supplemental Material). The tail of shallower-than-expected values (Fig. 3A) includes outliers of initial elevations around sea level and at modest subaerial elevations that are proximal to large-scale mantle upwellings and LIPs (Fig. 2B).

The thickness of the initial ocean crust ranges from <1 km where the mantle is exhumed (e.g., the Somali margin; Mortimer et al., 2020) to ~28 km for a small number of sites in regions of plume-related mantle upwelling such as the Iceland plume in the North Atlantic and the Réunion plume, which caused excess volcanism on the western Indian margin (Figs. 2B). The highest proportion of sites occurs within the average range of global thicknesses of ocean crust (Fig. 3C) and agrees with the mean thickness of normal ocean crust (~7 km) distal from anomalous regions such as hot spots (White et al., 1992). The number of sites with crustal thickness >14 km represents the tail end of the distribution, with thicknesses >20 km representing only a very small proportion of all sites (Fig. 3C). These sites typically occur on volcanic passive margins in the vicinity of LIPs (Fig. 2), which are mostly related to mantle-plume activity producing excess volcanism (Coffin and Eldholm, 1994). Despite the abundance of LIPs during the protracted break-up of Pangaea (Fig. 2), very few sites were initially close to sea level, contrary to the prerequisite of the shallow-ridge hypothesis (Fig. 3A; Fig. S1), and these are all associated with extensive LIPs (Fig. 2C).

Hyaloclastite formation is considered to be more important above the critical depth of seawater for explosive volcanism, and typically occurs in the final growth stages of seamounts (Staudigel and Schmincke, 1984), rather than on mid-ocean ridges (Bonatti and Harrison, 1988). Tholeiitic magmatic explosivity at mid-ocean ridges is limited to <200 m, and to 1 km for seamount-forming alkalic magmas (Kokelaar, 1986). Rare hyaloclastite deposits have been reported from mid-ocean ridge settings, but these settings are unusual and include fissures on the Eastern Rift Zone of Iceland (Bergh and Sigvaldason, 1991) where the crust has been significantly thickened and uplifted by the Iceland plume, resulting in an anomalously shallow basement depth (Fig. S1), and on the ultraslow-spreading Gakkel Ridge in the Arctic Ocean (Sohn et al., 2008).

Although the seafloor is peppered with millions of seamounts, the vast majority are small (<100 m tall), and located on young lithosphere (Wessel et al., 2010). These seamounts are quickly buried by sediment (Wessel, 2007), and in just a few million years subside below the average mid-ocean ridge depth of ~2.6 km as the ocean lithosphere cools, unlikely to ever produce hyaloclastite. Most of the larger seamounts (>1 km tall) were formed in intraplate settings on old ocean crust by hotspot activity (Wessel, 2007), and are common sites for hyaloclastite formation (Batiza, 1982; Staudigel and Schmincke, 1984; Bonatti and Harrison, 1988). However, intraplate seamounts are estimated to comprise a tiny fraction (~0.5%) of all seamounts (Wessel et al., 2010), and only a few are active at any one time (Wessel, 2007).

Hyaloclastites can also form in volcanic rifts in early rift environments if rifting occurs below sea level, as documented along the Vøring (offshore Norway) and conjugate Greenland margins (Planke et al., 2000). The transition from rifting to seafloor spreading is commonly marked by an outer volcanic high interpreted from seismic reflection data as a sequence of hyaloclastite flows (Planke et al., 2000). However, recent ocean drilling of the outer high on the Vøring Plateau mainly recovered massive and pillow basalts with only minor hyaloclastite (Planke et al., 2023), implying that the original interpretation based solely on seismic images had significantly overestimated the abundance of hyaloclastite. There is no published evidence that any hyaloclastite deposition continues once seafloor spreading has commenced along volcanic margins, in contrast to the assumption made by Gernon et al. (2016) that hyaloclastite formation continues for up to 35 m.y. after seafloor spreading commences.

Although the break-up history of Rodinia is controversial, the plate motion model we used implies a protracted continental break-up lasting ~150 m.y. (Merdith et al., 2021; Fig. 4), comparable to that of Pangea (Müller et al., 2019). Neoproterozoic ocean crust was also similar to modern crust, attaining the average Phanerozoic thickness of 7 km since ca. 0.9 Ga (Moores, 2002). However, unlike the numerous large LIPs associated with the break-up of Pangea, such as the Karoo (southern Africa), High Arctic, and Paraná-Etendeka (South America) LIPs (Fig. 1A) that originally covered between 3.12 Mkm2 and 3.6 Mkm2 (Park et al., 2021), the late Neoproteroic LIPs were far less numerous and extensive, and emplaced in continental interiors. We estimate that Pangea's passive margin length with depths in the range of hyaloclastite formation (0–200 m; Kokelaar, 1986) is ~2000 km—a mere ~1.3% of the total passive margin length of ~156,000 km, and marginally higher (3%) when dynamic topography and sea level are considered. The Franklin LIP (Canadian Arctic; Fig. 4A), emplaced 718 Ma (Dufour et al., 2023), was quite large (2.64 Mkm2), but the Central Iapetus Magmatic Province (CIMP; Fig. 4C) was emplaced in multiple pulses between ca. 615 Ma and 560 Ma over an area ranging from only 0.07–0.36 Mkm2 (Ernst et al., 2021). The absence of continental margins intersecting LIPs during Rodinia's break-up suggests that plume-related volcanic margins and associated dynamic uplift were rare. Consequently, spreading ridges did not form at or near sea level and the potential for any significant hyaloclastite formation via the shallow-ridge mechanism was greatly reduced. This is inadvertently substantiated by most sequences cited by Gernon et al. (2016, see their fig. 2 and their supplementary table 1) in support of the shallow-ridge hypothesis, in which small volumes of hyaloclastites were generated in convergent tectonic settings rather than on passive margins. Additionally, hypersaline and near-freezing bottom water consistent with near-global sea-ice cover (Hoffman et al., 2017) would have considerably slowed down weathering (Coogan and Dosso, 2015) and the alteration of hyaloclastite to palagonite by which alkalinity is supplied to the ocean in the shallow-ridge hypothesis (Fig. 1). Most spreading ridges form at depths of ~2.6 km in the absence of dynamic topography (Fig. 3A) and subside to ~4 km within 35 m.y. (Richards et al., 2018), which casts doubt on extensive hyaloclastite formation for 35 m.y. after the onset of mid-ocean ridge initiation proposed by Gernon et al. (2016).

We conclude that the shallow-ridge hypothesis fails to explain the formation of extensive cap carbonates marking the termination of Neoproterozoic glaciations globally. While the genesis of these unique carbonates remains controversial, it is more likely that other mechanisms such as enhanced continental weathering were largely responsible for supplying alkalinity to the Neoproterozoic ocean (Hoffman et al., 2017). It is unlikely that mid-ocean ridge evolution played any significant role in driving cap carbonate formation, even though plate tectonics may have played a role in modulating Cryogenian climate via changes in solid Earth degassing after Rodinia's break-up (Dutkiewicz et al., 2024). Future coupled plate-mantle models for the past billion years will be helpful for elucidating the effect of plate boundary evolution, mantle plumes, and dynamic topography on Rodinia's and Gondwana's passive margins, evolving ocean basins, and ocean chemistry.

We thank three anonymous reviewers for their detailed and constructive comments. This research was supported by the Australian Research Council (ARC) Future Fellowship FT190100829 to A. Dutkiewicz and by the National Collaborative Research Infrastructure Strategy (NCRIS) via AuScope.

1Supplemental Material. Methods, Figures S1–S4, Table S1, and boundaries between continental and oceanic crust. Please visit to access the supplemental material; contact with any questions.
Gold Open Access: This paper is published under the terms of the CC-BY license.