Exceptional eruptive CO2 emissions from intra-plate alkaline magmatism in the Canary volcanic archipelago

Volcanic gas emissions critically influence volatile cycles and climate, with CO2 exsolving early due to its low solubility in magmas. Quantifying magmatic volatile contents—especially CO2—is central to understanding eruption dynamics and the geological carbon budget. Prior work shows that CO2 often exsolves before melt inclusion entrapment, so inclusion-derived CO2 is a lower bound. While trace element ratios (e.g., CO2/Nb, CO2/Ba) can constrain mantle CO2, these are affected by source heterogeneity and melt mixing. Syn-eruptive gas measurements, when scaled by SO2 fluxes and mass eruption rates (MER), provide robust constraints on initial volatile contents. Recent studies suggest magmatic CO2 contents may be higher than previously assumed, with alkaline intra-plate ocean island magmas potentially especially CO2-rich. In the Canaries, El Hierro basanites were estimated to contain 2.5–5 wt% CO2, but no direct syn-eruptive CO2 measurements existed for the archipelago. The 2021 Tajogaite eruption at Cumbre Vieja, La Palma, offered an opportunity to quantify eruptive CO2 emissions and derive initial CO2 contents for alkaline mafic magmatism in this setting, addressing the question: how CO2-rich are Canary basanites and what are the implications for volcanic CO2 budgets?

Background research establishes CO2 as the least soluble major magmatic volatile, exsolving at mid- to lower-crustal pressures. Melt and fluid inclusion studies indicate CO2-rich fluid phases can exist at mantle and lower crustal depths, and in low-viscosity systems CO2 can segregate and accumulate as foams, potentially enhancing degassing relative to erupted volumes. Alternative constraints on mantle CO2 employ trace element ratios (CO2/Nb, CO2/Ba), though variability arises from mantle heterogeneity and mixing. Syn-eruptive gas measurements (OP-FTIR, Multi-GAS), combined with SO2 fluxes (ground and satellite) and MERs, have proven effective in retrieving initial volatile contents. Prior Canary work at El Hierro inferred high primary CO2 (2.5–5 wt%) from melt inclusions and trace-element systematics, and global studies highlight alkaline mafic magmas as significant CO2 sources. However, a direct syn-eruptive CO2 dataset for the Canary archipelago was lacking before this study.

The team conducted near-daily measurements of gas composition and fluxes at explosive (fountaining vents, FV) and effusive/spattering flank vents (SV/EV) during the 2021 Tajogaite eruption. Instrumentation and approaches included:

• Gas composition: In-situ Multi-GAS (ground-based mobile, fixed station, and drone-mounted) measured CO2, SO2, H2S+H2, temperature, pressure, and relative humidity. UAS flights (DJI Matrice M210) sampled plumes near vents (passive degassing PV, FV, SV, EV) between 27 Sep–4 Oct 2021. Sensors were calibrated pre- and post-campaign; data processed with Ratiocalc.
• Remote OP-FTIR spectroscopy: Two MIDAC OP-FTIR spectrometers, operating in passive and solar absorption modes, retrieved H2O, CO2, SO2, HCl (fit windows: 2080–2150 cm−1 for CO, CO2, H2O; 2450–2550 cm−1 for SO2; 2690–2830 cm−1 for HCl) over multiple dates (3 Oct–2 Dec). Best-quality FV spectra obtained on 2 Dec.
• SO2 flux and plume height: Satellite-based Sentinel-5P TROPOMI data analyzed with the PlumeTraj back-trajectory method to retrieve injection heights, time-resolved SO2 masses, and flux densities, correcting for height-dependent sensitivity. On 3 Oct the average SO2 flux was 860 ± 450 kg/s; the eruption-wide average was 500 ± 200 kg/s.
• Mass eruption rates (MER): Explosive MER derived from plume height scaling; on 3 Oct, plume height ~4500 ± 200 m a.s.l. implied 22,600 ± 3300 kg/s. Effusive MER from lava effusion rates and density; on 3 Oct, 30 ± 15 m3/s with density 2618 ± 179 kg/m3 gave 78,500 ± 40,000 kg/s. Time-averaged MERs were 63,000 kg/s (effusive) and ~20,000 kg/s (explosive, including tephra and airborne ash components).
• Petrology: Olivine-hosted melt and fluid inclusions and groundmass glasses were analyzed by EPMA in Paris and Bristol. Melt inclusions show S = 3290 ± 390 ppm (initial S), H2O = 1.30–2.21 wt%, CO2 = 0.22–0.50 wt%, with entrapment pressures ~290–350 MPa (~10–12 km depth). Groundmass glass S ~474 µg/g indicates ~85% S outgassed.
• Degassing modeling: A C-O-H-S equilibrium saturation model (isothermal 1273 K) simulated closed-system ascent (from 400 MPa) and open-system shallow degassing (5 MPa to 0.1 MPa) for initial CO2 = 1–5 wt% and redox ΔNNO ≈ +1 (with sensitivity at +0.5). Gas fraction lost varied from 0.1–0.9 for open-system runs. Modeled gas compositions were compared to measured FV, SV, EV plume compositions to infer degassing pathways and branch depth (~10 MPa).
• Mass-balance inversion: Using measured CO2/SO2 in FV and SV/EV plumes, TROPOMI SO2 fluxes, MER partitioning, initial S content, and assumptions about S exsolved at the branch (5–10%) and the fraction of exsolved gas routed to FV (80–95%, best estimate 84 ± 5%), the initial magmatic CO2 content and total eruptive CO2 flux were calculated for 3 Oct and for the eruption-integrated case.

• Strong, persistent compositional contrast between vents: FV gases were exceptionally CO2-rich (mean CO2/SO2 molar 33 ± 9.7; mass ~23 ± 6.7), while SV averaged 7.3 ± 4.0 molar; EV as low as 0.3–0.04 molar. Best FV OP-FTIR (2 Dec) measured 37 mol% H2O, 61 mol% CO2, 1.7 mol% SO2 (CO2/SO2 molar ~36).
• Melt inclusions indicate initial S content 3290 ± 390 ppm, with entrapment at ~10–12 km; maximum dissolved CO2 in inclusions (0.50 wt%) underestimates parental CO2 content.
• Degassing pathway: Closed-system ascent to a shallow branch (~10 MPa), where gas-melt decoupling channelled most CO2-rich gas to FV; EV/SV gases reflect open-system, low-pressure degassing of residual volatiles.
• Mass-balance results: For 3 Oct, with SO2 flux 860 ± 450 kg/s, explosive MER 22,600 ± 3,300 kg/s and effusive MER 78,500 ± 40,000 kg/s, assuming 7.5% S exsolved at 10 MPa and 84 ± 5% of exsolved gas routed to FV, the required initial CO2 content is 4.4 ± 1.3 wt%. Eruption-averaged calculations yield ~4.6 wt%.
• Overall initial magmatic CO2 content: 4.1–5.9 wt% (best estimate 4.5 wt%).
• Total eruptive CO2: 28 ± 14 Mt over the 85–86 day eruption; average 0.33 ± 0.16 Mt/day. This is among the largest documented eruptive CO2 outputs.
• SO2 metrics: Eruption-wide average SO2 flux 500 ± 200 kg/s; on 3 Oct, 860 ± 450 kg/s. The model reproduces observed SO2 fluxes, supporting the inferred CO2 content.
• Geological implications: At storage depths (~10–15 km), only 0.25–0.6 wt% CO2 is soluble; thus ~4 wt% CO2 was already exsolved, consistent with abundant CO2-rich fluid inclusions. Extrapolations suggest La Palma’s construction could have emitted ~1.6×10^15 moles (subaerial) to ~1.1×10^16 moles (including submarine growth) of CO2, roughly 20% of a typical LIP’s eruptive CO2, implying Canary archipelago formation may approach LIP-scale CO2 emissions.

The contrasting gas compositions reflect a shallow conduit branching where rapidly ascending magma degassed in near closed-system conditions until ~10 MPa. At this branch, exsolved CO2-rich gas decoupled and rose to FV, maintaining very high CO2/SO2, while the majority of magma was diverted to SV/EV and continued low-pressure, open-system degassing, enriching H2O and SO2 relative to CO2. Modeling shows that FV compositions require parental basanite with 1–5 wt% CO2 undergoing closed-system ascent to shallow levels; EV/SV compositions demand open-system degassing below ~5–10 MPa. High melt oxidation states increased S solubility so that 90–95% of S remained dissolved at the branch, making SO2 emissions largely controlled by MER, while CO2 was preferentially emitted at FV. Mass-balance, constrained by measured CO2/SO2, SO2 fluxes, MERs, and initial S, yields an initial CO2 content of ~4.5 wt% and total eruptive CO2 of 28 ± 14 Mt. The persistence of high FV CO2/SO2 through the eruption indicates limited CO2 heterogeneity or segregation within the reservoir during eruption. At depth, low CO2 solubility implies a substantial exsolved CO2-rich fluid phase existed before eruption, potentially enhancing explosivity and magma compressibility. The findings substantiate that intra-plate alkaline ocean island volcanism can be exceptionally CO2-rich, with significant implications for the long-term carbon cycle.

This study provides the first syn-eruptive quantification of CO2 emissions and magmatic CO2 content for the Canary archipelago, demonstrating exceptionally CO2-rich degassing from the 2021 Tajogaite eruption. A mass-balance inversion constrained by multi-platform gas measurements, satellite SO2 fluxes, MERs, and petrology yields a best-estimate initial CO2 content of 4.5 wt% and a total eruptive CO2 output of 28 ± 14 Mt. The degassing architecture involves closed-system ascent to a shallow branch point where CO2-rich gas preferentially vents explosively. Extrapolations suggest that CO2 released during construction of La Palma—and potentially the entire archipelago—approaches a significant fraction of LIP-scale eruptive CO2, highlighting the role of ocean island volcanism in the geological carbon cycle. Future research could expand syn-eruptive volatile measurements across other Canary islands and similar archipelagos, quantify intrusive and submarine degassing contributions, refine gas-melt partitioning at shallow conduit branches, and integrate high-frequency gas data with geophysical imaging to better resolve storage and ascent dynamics.

• Affiliation of some authors to multiple institutions indicates collaborative logistics; however, the study area is a single eruption, limiting generalizability without broader archipelago sampling.
• Initial CO2 content depends on assumptions about branch depth (~10 MPa), the fraction of S exsolved there (5–10%), and the proportion of exsolved gas routed to FV (80–95%). Uncertainties in these parameters propagate into CO2 estimates.
• Melt inclusion H2O–CO2 data are sparse (three inclusions), and inclusion CO2 provides lower bounds due to pre-entrapment exsolution and possible bubble loss; fluid inclusions indicate a pre-existing gas phase that could contribute additional CO2.
• Rapid ascent may kinetically limit gas exsolution, affecting the amount of S exsolved at depth; modeled values may represent upper limits.
• Satellite SO2 retrievals require plume height corrections (PlumeTraj) and carry wind and retrieval uncertainties; ground-based coverage was limited early on.
• Potential pre-eruptive segregation and reservoir-scale accumulation of CO2-rich gas cannot be completely excluded, although stable high FV CO2/SO2 suggests limited CO2 differentiation during the eruption.