Rapid shifting of a deep magmatic source at Fagradalsfjall volcano, Iceland

Oceanic crust formation at mid-ocean ridges is commonly associated with centralized, mid- to shallow-crustal magma reservoirs and lateral magma transport, which can overprint deeper mantle signals. Rarely, eruptions are supplied directly from sub-Moho (>7 km) levels, but near-real-time observations of such deep-sourced events are limited, especially in submarine settings. Iceland, with thicker-than-typical oceanic crust but subaerial exposure of the MOR system, offers a unique opportunity for continuous, real-time sampling. Prior Icelandic rifting events (1975–84 Krafla; 2014–2015 Bárðarbunga) highlighted the importance of shallow storage and lateral transport, yet obscured deeper interactions. The Reykjanes Peninsula is a leaky transform fault with episodic rifting and volcanism; before 2021, the last eruptive episode occurred circa AD 700–1240. The 2021 Fagradalsfjall eruption followed intense seismic unrest and dyke injection, with eruption onset on 19 March 2021. The study aims to resolve processes operating near the crust–mantle boundary by integrating sequential geochemical and petrologic analyses of erupted products and gas emissions, testing how rapidly and by what mechanism deep mantle-derived melts reconfigure magma reservoirs and the erupted composition.

The paper situates the work within MOR magmatism studies showing centralized crustal reservoirs and lateral transport, and the concept of magmatic filtering of mantle compositions. It references Icelandic rifting episodes (Krafla 1975–84; Bárðarbunga 2014–2015) that revealed shallow storage but masked deeper mantle signals. It notes that MOR eruptions sourced from sub-Moho levels occur but are less commonly observed and rarely captured in real time due to deep-ocean inaccessibility. The Reykjanes Peninsula’s historical eruptions (circa AD 700–1240) provide context for mantle source variability. Prior work documents heterogeneous mantle sources beneath Iceland, progressive polybaric near-fractional melting, channelized melt transport, and that high-MgO lavas can best preserve mantle signatures. This study builds on these by capturing unprecedented temporal geochemical variability during a single eruption, enabling tests of models for transcrustal magmatism and melt supply dynamics.

- Study interval: Initial phase of the 2021 Fagradalsfjall eruption (21 March to 6 May 2021). - Sampling and materials: Sequentially erupted lava and tephra; quenched glasses; mineral phases and their melt inclusions (MIs); contemporaneous vent gas emissions. - Geochemical analyses: Whole-rock and glass major and trace elements; incompatible trace element ratios (ITERs) such as K2O/TiO2 and La/Yb; radiogenic isotope ratios (Sr, Nd, Pb). Emphasis on ITERs for mantle source/melting conditions because they are insensitive to fractional crystallization and modal mineral proportions. - Petrography and mineral chemistry: Characterization of primitive crystal cargo (Cr-rich spinel, Fo-rich olivine, An-rich plagioclase, green clinopyroxene with high Mg# and Cr2O3). Assessment of MI compositions, including high-MgO (>10 wt%, PEP-corrected) inclusions, and within-crystal compositional variability. - Post-entrapment process (PEP) corrections: Applied to MIs; K2O/TiO2 and La/Yb considered robust to PEP. - Thermobarometry: Olivine–plagioclase–augite–melt (OPAM) barometry using glass and MIs; clinopyroxene–liquid barometry from crystal cores and rims to estimate storage pressures. Kernel density estimates used to summarize pressure distributions. - Volatile constraints: MI CO2–H2O saturation pressures; surface vent gas CO2/SO2 ratios modeled assuming closed-system degassing to infer ascent depths and storage levels. - Eruption monitoring context: Effusion rates measured independently (1–8 m³ s−1 initially; increasing to 9–13 m³ s−1 after 27 April) and eruption style changes (onset of high lava fountaining >450 m) provide temporal framework for sampling and interpretation. - Comparative datasets: Historical Reykjanes Peninsula lavas and other Icelandic basaltic single-eruptive units used to benchmark the observed compositional range.

- Near-Moho sourcing: Multimethod barometry (OPAM; cpx–liquid) and volatiles indicate a dominant magma storage zone at 0.55–0.65 GPa (most probable), equivalent to ~20 km depth; overall range 0.36–0.80 GPa. Vent gas CO2/SO2 and MI volatiles imply ascent from 19 ± 4 km. - Rapid geochemical shift: Within the first three weeks, erupted lavas show K2O/TiO2 and La/Yb increases by about a factor of 2, indicating a shift from shallow, higher-degree, geochemically depleted melts to deeper, lower-degree, enriched melts. - Isotopic evolution: Concomitant shift to more radiogenic Sr and Pb and less radiogenic Nd; strong correlations among K2O/TiO2, La/Yb, and Pb isotopes (R² > 0.97) confirm mantle-derived variability rather than shallow processing. - Exceptional diversity: The erupted compositions during the initial weeks encompassed and exceeded the spectrum of mantle source indicators observed across ~540 years of historical Reykjanes Peninsula eruptions. - Melt inclusions: Primitive, high-MgO MIs record wide K2O/TiO2 and La/Yb ranges, including enriched melts (K2O/TiO2 up to 0.39) entrapped prior to eruption, evidencing pre-eruptive mixing of depleted and enriched melts in the storage zone. - Shallow overprint limited: Early tephra/groundmass glass and evolved cpx rims/cores indicate transient equilibration at 0.05–0.25 GPa (<8 km), but high-pressure signatures are preserved, indicating limited shallow overprinting of the carrier melts. - Reservoir size and dynamics: Temporal trends and mixing are consistent with near-Moho melt lens(es) containing ~10⁷–10⁸ m³ of basaltic magma that were progressively recharged by deeper enriched melts. - Eruption dynamics: Initial low–modest effusion (1–8 m³ s−1) from multiple vents transitioned after 27 April to higher discharge (9–13 m³ s−1) from a single vent with >450 m fountaining, coincident with evolved (more enriched) compositions.

The observations directly capture a rapid reconfiguration of magma supply from a near-Moho reservoir, with the erupted composition transitioning from shallow-sourced, depleted melts to deeper-sourced, enriched melts over weeks. This addresses the question of how quickly and where mantle-derived variability is filtered and mixed: the near-Moho storage zone acts as a dynamic aggregation and mixing lens. The strong geochemical–isotopic correlations indicate that temporal changes reflect mantle source/melting conditions rather than shallow differentiation. Primitive mineral and MI data show that enriched melts were present pre-eruption but became increasingly dominant as deeper melt input continued, potentially triggering ascent. Barometry and volatile constraints place storage at ~20 km depth, with minor shallow transit effects. The results challenge models that emphasize only low-porosity crystal mushes at shallow levels by documenting rapid, deep-level reservoir compositional shifts, consistent with channelized melt transport and polybaric melting of heterogeneous mantle. Compared with other MOR and ocean-island eruptions, the magnitude and rate of within-eruption variability at Fagradalsfjall are unique, offering a benchmark for transcrustal magma system timescales and processes.

This study provides near-real-time evidence that the 2021 Fagradalsfjall eruption was fed by a near-Moho reservoir whose composition shifted rapidly due to progressive input of deeper, enriched mantle melts. The work demonstrates exceptional within-eruption geochemical variability and tightly links mantle melting conditions, deep storage, and erupted compositions on timescales of weeks. It constrains storage depths (~20 km), reservoir volumes (~10⁷–10⁸ m³), and the role of dynamic mixing in filtering mantle signals. Future research should assess how widespread such deep magmatic reconfigurations are along mid-ocean ridges and oceanic islands with thinner crust, and further integrate geophysical imaging with time-resolved geochemistry to quantify melt channel dynamics, reservoir connectivity, and triggers of ascent and eruption.

- Temporal scope: Analyses focus on the initial ~50 days (21 March–6 May 2021), so later eruptive phases are not covered. - Incomplete shallow overprint characterization: While high-pressure signatures are preserved, some crystal rims and groundmass indicate shallow equilibration; quantifying all shallow processes is challenging. - Proxy-based inferences: Depth and storage estimates rely on thermobarometry, volatile models (closed-system degassing assumptions), and geochemical proxies, which carry uncertainties. - Generalizability: The subaerial Icelandic setting allows unique sampling; extrapolation to submarine MOR systems, where near-real-time sampling is limited, may be constrained.