Seafloor Cores Reveal How Montserrat’s Eruptions Are Selectively Preserved Underwater

Beneath the turquoise waters east of Montserrat, the seafloor keeps a diary of one of the world’s most closely watched volcanoes. But according to a new study, that diary is heavily edited. Researchers analysing sediment cores recovered from the submarine slopes of the Soufrière Hills Volcano have shown that the marine record of eruptions is neither complete nor straightforward: some sites preserve thick, complex successions of volcanic material, while others, just a short distance away, retain only a whisper-thin layer of ash. The findings, published in Environmental Earth Sciences, reveal how topography, flow transformation and post-depositional reworking combine to filter the volcanic signal before it ever becomes part of the geological archive.

The study, led by Dina Hanifah of MARUM, the Centre for Marine Environmental Sciences at the University of Bremen, together with Steffen Kutterolf of GEOMAR and Katrin Huhn of MARUM, focused on the eastern offshore sector of Montserrat, directly downslope of the Tar River Valley. This valley served as the main conduit for pyroclastic density currents during the Soufrière Hills eruption of 1995 to 2010, a prolonged crisis of andesitic lava-dome growth and collapse that delivered more than 90 percent of the erupted material into the sea. Because the onshore chronology of that eruption is exceptionally well documented by the Montserrat Volcano Observatory, the surrounding seabed offers a rare opportunity to test how faithfully marine sediments record known eruptive events.

Two gravity cores, each roughly 88 centimetres long, were recovered in 2019 during RV METEOR cruise M154-2 from water depths of around 850 to 950 metres. Their locations tell much of the story. Core GeoB23710-1 was taken from a topographically confined depression near the margin of a large mass-transport deposit, in the direct submarine extension of the dome talus fan. Core GeoB23711-1, recovered only a short distance away but roughly 90 metres higher on the adjacent slope, sits outside the principal transport pathway. Despite their proximity, the two cores could hardly be more different. GeoB23710-1 contains about 30 centimetres of volcaniclastic layers underlain by a muddy-sandy debrite, whereas GeoB23711-1 preserves just a one to two centimetre ash horizon near the top of an otherwise homogeneous mud sequence.

To interrogate these contrasting successions, the team applied a multi-proxy toolkit that goes far beyond visual description. Multi-sensor core logging provided continuous profiles of gamma-ray density, magnetic susceptibility and P-wave velocity at one-centimetre resolution. X-ray fluorescence scanning at ten-millimetre resolution yielded elemental ratios such as Zr/Sr, which distinguishes volcaniclastic input from background hemipelagic sediment, and Zr/Rb, which tracks grain-size changes. Most strikingly, the archive half of GeoB23710-1 was scanned with a medical-grade computed tomography scanner at 0.3-millimetre voxel resolution, allowing the researchers to reconstruct internal structures invisible to the naked eye: erosional basal contacts, imbricated clasts, cross-lamination, loaded beds and subtle density laminae.

The CT data resolved the volcaniclastic succession in GeoB23710-1 into three internally distinct packages, labelled V1, V2 and V3. V1, about five centimetres thick, begins with a sharp erosional base and coarsens dramatically, with bulk densities reaching 1.9 grams per cubic centimetre and CT intensities exceeding 1450 Hounsfield Units in its clast-rich base. V2, the thickest interval at roughly 33 centimetres, displays alternating high- and low-density laminae, normal grading, small-scale loading and cross-lamination, all signatures of repeated fluctuations between tractional transport and suspension fallout. V3, the uppermost eleven centimetres, fines upward into increasingly structureless sediment. Radiocarbon dating of planktonic foraminifera from hemipelagic layers bracketing V1 returned calibrated ages of AD 181 plus or minus 57 years above the unit and around 2131 calibrated BC below it, confirming that V1 predates the modern eruption and is chronologically compatible with a regional andesitic turbidite deposited roughly 1.5 to 2 thousand years ago.

V2 and V3, which overlie the younger dated horizon, are interpreted as the most likely offshore expression of the 1995 to 2010 eruptive phase. Yet the researchers are careful not to over-interpret. The two intervals are separated only by a thin carbonate-rich layer, V2b, which likely records a brief lull in volcaniclastic supply, possibly during the relative quiescence around 2000 to early 2001, though the chronological resolution cannot confirm this. Critically, the composite architecture of V2 and V3, with its erosional contacts, renewed coarse basal influxes and tractional reworking, demonstrates that bed number cannot be equated with event number. A single dome collapse entering the sea can partition into multiple submarine flows, each depositing sediment at different times and places, while successive pulses amalgamate into what appears to be a single bed.

Microscopic point counts of smear slides added a compositional dimension to this story. V1 contains a comparatively high proportion of juvenile pyroclasts, glassy and vesicular particles freshly fragmented by explosive eruptions, alongside volcanic lithics, dome fragments, dense minerals and bioclasts. In contrast, V2 and V3 are dominated by dense minerals, at 35 to 55 percent, and lava-dome fragments, with juvenile pyroclasts generally falling below 15 percent. This shift suggests that the younger package incorporated substantial reworked volcaniclastic detritus and dome-derived material during repeated submarine transport, rather than receiving pristine eruptive fallout. The bioclasts and sediment clasts scattered through the volcaniclastic layers provide direct evidence that the currents eroded and entrained the seafloor as they travelled.

The comparison between the two cores carries the study’s most consequential message. Under a simple model, the off-axis site should preserve a thinned but ordered equivalent of the axis-proximal succession. Instead, GeoB23711-1 holds only a thin ash veneer that cannot be confidently correlated with any specific interval in GeoB23710-1. The two sites evidently retained different fractions of the passing currents: the confined depression accumulated coarse, concentrated flow components repeatedly, while the elevated slope received only dilute, ash-bearing portions, or was bypassed entirely. Analogous relief-controlled routing has been documented in the Valparaíso Basin off Chile, the Shikoku Basin and the Izu-Bonin region, but the Montserrat pair provides an unusually crisp core-scale demonstration that depositional position, not eruption magnitude alone, determines what the archive retains.

The implications extend beyond stratigraphy into hazard assessment. The sharp density and fabric contrasts between coarse volcaniclastic beds and fine-grained interlayers define a heterogeneous shallow mechanical stratigraphy that, in principle, could host weak layers prone to failure during future loading or earthquakes. The authors are explicit, however, that their data do not establish mechanical weakness or tsunami-generating landslide potential; CT intensity and P-wave velocity are not measurements of shear strength, and direct geotechnical testing would be required. What the study does establish is a practical framework: repeat bathymetric surveys to map confined pathways and depocentres, dated core transects spanning axes, margins and elevated slopes, and integrated multi-proxy analysis to distinguish fresh eruptive input from reworked material. For volcano monitoring and seabed infrastructure planning around dome-dominated island volcanoes, the lesson is clear: a thin or absent deposit at one site does not mean the currents were weak or absent, and a thick succession does not simply count eruptions. The seafloor archive off Montserrat is a dynamically filtered record, and reading it demands that the filter itself be understood.

Subject of Research: Marine volcaniclastic stratigraphy and preservation of Soufrière Hills eruption records offshore Montserrat

Article Title: Volcanic stratigraphy offshore East Montserrat: Selective preservation and subaqueous remobilisation in marine records

Article References: Hanifah, D., Kutterolf, S., & Huhn, K. (2026). Volcanic stratigraphy offshore East Montserrat: Selective preservation and subaqueous remobilisation in marine records. Environmental Earth Sciences, 85(16), Article 414. https://doi.org/10.1007/s12665-026-13151-5

Image Credits: AI Generated

DOI: 10.1007/s12665-026-13151-5

Keywords: Montserrat, Soufrière Hills Volcano, volcaniclastic sediments, marine stratigraphy, pyroclastic density currents, turbidity currents, computed tomography, XRF core scanning, mass-transport deposits, submarine slope stability, Lesser Antilles arc, radiocarbon dating