Introduction

The Vulcano Island in the Aeolian Archipelago (Southern Tyrrhenian Sea, Italy) is a closed-system Quaternary volcano characterized by relatively long periods of volcanic rest periodically interrupted by low-to-high intensity degassing crises and eruptions. The peculiar volcano-tectonic setting of the island favored complex magma evolutionary processes characterized by high degree of fractional crystallization, mixing and crustal assimilation. These processes have led to the formation of multiple storage regions over time, extending from the Moho to very shallow crustal levels and characterized by variable differentiation degrees1,2,3,4,5,6,7,8. The evolution of the magmatic system has been a subject of intense research, mostly aimed at defining the mechanisms responsible for the primary magma modifications over time. In this context, the analysis and modeling of major and trace element distributions in volcanic minerals and rocks allowed quantitative constraints on the dynamics of the plumbing system5,9,10,11,12. Such information combined with diffusion chronometry and thermobarometric applications has proved to be effective for assessing the pre-eruptive magma conditions and defining the timescales required for their mobilization9,11,12. Despite petrological methods allow a direct view of the magmatic system, these methods alone are ineffective to forecast future eruptions. The integration of petrological data with real-time monitoring signals may enhance our capability of interpreting monitoring data, linking the potential changes recorded at the surface with magma dynamics developing at depth. The challenge in this direction occurs when dormant volcanoes undergo hundreds to thousands of years of quiescence. Indeed, the analyses of volcanic products can only be referred to past eruptions in such cases and cannot be directly connected with signals of volcanic unrest obtained by instrumental networks.

The last eruption at Vulcano Island occurred between 1888 and 1890 CE at the crater La Fossa (Fig. 1), while the volcanic surveillance on the island just began in the late 1970s. The volcano quiescence during the 20th and 21st centuries hampered the possibility of correlating petrological data with the geochemical and/or geophysical signals of volcanic unrest. Since the last eruption, Vulcano experienced, however, various phases of volcanic unrest, manifesting a visible escalation of the degassing activity in the area of the crater La Fossa, sometimes accompanied by diffuse soil degassing across the island. Intense degassing crises occurred particularly during 1920s and 1988–199313,14. Recently, since September 2021, the island entered into a new phase of sustained degassing output, which has been accompanied by shallow seismicity and ground inflation15,16,17. The crisis underwent a progressive declining phase since 2022 and was declared concluded at December 2023, although some geochemical parameters were not at the background level. Notably, the mechanisms proposed to explain these periodic unrests on the island are still unclear due to the complex interplay between magmatic and hydrothermal processes. Some authors stress the role of open-system degassing and increasing gas supply into the shallow plumbing system during the ascent of deep volatile-rich melts15,18. Other authors invoke a mechanism of overpressurization of the hydrothermal system due to accumulation of magmatic gases coming from stationary magmas crystallizing at depth7,19.

Fig. 1
figure 1

(a) Geological sketch map of the Vulcano island modified after Ref.6, showing the subaerial volcanic succession referred to the eight Eruptive Époques (EEs); (b) View from northwest of the Gran Cratere of La Fossa with some fumarolic fields.

Full size image

Typical geochemical anomalies detected during degassing crises include increasing gas temperature and flux from fumaroles and increasing of the magmatic component in the emitted gases (i.e., increase of CO2 and He contents in fumaroles, anomalous SO2 fluxes from the crater7,13,15,18,19). Volcanic gas manifestations carry information about the origin and quantity of magma involved in the degassing activity, as well as about the depth reached by magmas beneath the surface. Among volcanic gases, SO2 is released when magma is relatively near the surface, and it is therefore a reliable marker of the depth and quantity of magma reaching the shallowest portions of the plumbing system20,21. Overall, an increase in the gas output, the appearance of a new fumarole or changes in the chemistry of the gas mixture are signs ofheightened volcanic activity and can therefore indicate the volcano reawakening. The repetition of such phenomena with major escalation periods justifies the need for intensifying the monitoring activities to continuously check the fluxes of gas and their composition.

Selected fumaroles located along the northern rim and the inner flank of the La Fossa crater have been constantly monitored by the Istituto Nazionale di Geofisica e Vulcanologia (INGV) since 198422,23,24,25. The emitted gases have been collected and analyzed over time with weekly to monthly frequency as a function of the Volcanic Alert Level, with the aim of detecting anomalous deviations from the standard geochemical trends.

The SO2 flux time series at the La Fossa crater since 1978 is presented here in order to place constraints on the shallow dynamics of the magmatic system during the transition between stages of quiescence and unrest. These data have been connected to magma dynamics resulting from the inspection of zoned clinopyroxenes from pyroclastic products of the last 1888-90 eruption. As an early mineral phase growing across the entire crustal column, clinopyroxene zoning holds key information on the architecture of magma plumbing systems, and it is especially suitable for tracking the movement of primitive magma ascending from depth, as well as for detecting pre-eruptive magma interactions26,27,28,29,30. In this regard, we have used the Fe-Mg interdiffusion in clinopyroxene to gain insights into the timing of magma residence and mixing events prior to the 1888-90 eruption. When combined with the SO2 temporal trend, this information emphasizes the link between past and present dynamics during the volcanic unrest and offers clues into the processes that ultimately may lead to a new eruption at Vulcano.

Geological setting and eruption history of the Vulcano Island

Vulcano is one of the seven islands of the Aeolian Archipelago, a chain of volcanic islands located on the southern rim of the Tyrrhenian Sea, about 25 km from the northern coast of Sicily in Southern Italy (Fig. 1a). Together with the islands of Lipari and Salina, Vulcano forms a linear volcanic chain that is placed into a graben structure. The graben is tectonically controlled by the NNW-SSE oriented Tindari-Letojanni dextral transtensive fault system, extending from the south Tyrrhenian plain to the northeastern Sicily31,32,33,34. The Tindari-Letojanni system has major impact on the crustal dynamic processes along the Vulcano-Lipari-Salina alignment, controlling the interaction between deep and shallow structures and hydrothermal processes affecting the islands31,32,33.

Vulcano is a relatively young island of post-Tyrrhenian age (~ 127 ka6,35). The eruptive history of the island has been divided into eight Eruptive Epochs (EEs), each defining a major stage of the construction of Vulcano and the associated volcano-tectonic collapses6 (Fig. 1a). The most recent eruptive epoch, the EE 8 (8.9 ka to present), includes the eruption sequences building up the edifice of La Fossa that is active since 5.5 ka36, and the peninsula of Vulcanello, which entirely formed in historical times (1–2 ka) along the northern sector of the island6,35,37,38. The volcanic activity at La Fossa and Vulcanello closely alternated between the 9th and 17th centuries, according to the historical record6.

At La Fossa, the volcanic activity occurred in various eruptive phases separated by prolonged period of quiescence6,39. The cone grew up in the middle of La Fossa caldera through recurrent phreatomagmatic explosive phases, lasting for months or years, often accompanied by ballistic ejections of bombs and blocks that define the typical Vulcanian style40. The formation of multiple pyroclastic density currents, fallout deposits and some lava flows have been also documented in the last ~ 5.5 ka. The La Fossa stratigraphic succession has been divided into three main chrono-stratigraphic units6. They include from bottom to top: (a) Punte Nere and Grotta dei Palizzi 1 formations (~ 5.5–2.2 ka); (b) Grotta dei Palizzi 2, Caruggi and Forgia Formations (~ 2.2 ka − 770 CE); (c) Pietre Cotte (1739 CE) and the 1888-90 deposits (Gran Crater 2 Formation). With reference to this stratigraphic sequence, some discrepancies exist with the position of the Punte Nere unit proposed by Refs. 39 and 41, who postponed its age at 1170 ± 20 CE based on geological and petrographic observations and the archaeomagnetic dating of Ref.42.

The present-day cone (391 m a.s.l.) has a roughly circular ~ 500 m long crater rim showing at its bottom traces of the two vents that originated from the last 1888-90 eruption. This last eruption occurred at La Fossa after 149 years of rest since the 1739 CE eruption of Pietre Cotte, interrupted only by small and sporadic phreatic explosions over the period 1771-1886 CE (i.e., 1771, 1786, 1822-24, 1873, 1876-79, 1886), which are documented thanks to historic chronicles40. Ref.1 has provided a thorough description of the eruption sequence during 1888-90 and the associated deposits, ranging from latitic to rhyolitic compositions. These authors recognized three main eruptive phases on the basis of the tephro-stratigraphic and petrochemical characteristics of the different layers composing the sequence: (a) an initial phase characterized by the emission of rhyolitic phreatomagmatic breccias containing latitic magmatic enclaves; (b) an intermediate phase during which pyroclastic surges and fallout material, first trachytic and then latitic in composition, were emitted; (c) a final phase dominated by the emission of rhyolitic and trachytic tephra, during which decimeter-scale bread crust bombs and blocks were emplaced. Products preserve various evidence of magma mixing and mingling, which is highlighted by reaction rims in various mineral phases, reversely zoned clinopyroxenes, banded pyroclastic products and the presence of magmatic enclaves1,6,43.

Previous studies on the plumbing system configuration

A number of studies collecting petrological, geophysical, gas and mineral chemistry and fluid inclusions data have contributed to develop models for the plumbing system of Vulcano Island1,2,4,6,9,10,11,12,37,38,44,45. The proposed models depict a polybaric system made of several magmatic storage zones of various compositions (basalts to rhyolites) that distribute from ~ 21 km (the Moho depth) to 1–2 km depth beneath La Fossa. These multiple magma batches evolved in independent reservoirs, but they could interact before eruptions, as documented by the mixing/mingling textures found in several volcanic rocks. Previous studies concord in relating the complexity of the plumbing system with the volcano-tectonic setting of the island that favored the differential activation of the reservoirs over time2,6,46,47. Major changes in the magma differentiation mechanisms and, more in general, the transition towards more complex magma dynamics have been especially evidenced in the last 30 ka, with a dominance of fractional crystallization with respect to assimilation and recharge/mixing processes4,6.

The magmatic feeding system of Vulcano can be view as a complex, vertically-extended storage zone hosting basalts, shoshonites, latites, trachytes and rhyolites from depth to the surface. Throughout the whole section, some preferential storage levels have been identified at specific depths. In this regard, the deepest level of magma storage of basaltic composition is located at the mantle-crust transition zone between 18.5 and 21 km of depth2,11,48. These magmas can reach the surface directly (as documented for the activity of Vulcanello4,9) or they can evolve while rising and stalling at intermediate to shallow crustal levels (e.g., magmas that fed the recent < 1 ka La Fossa eruptions9,10,43). A shoshonitic storage region is hypothesized between ~ 17 and ~ 12 km, gradually changing to latitic compositions in the depth range ~ 12 and ~ 8 km2,4,6,11,49. The upper part of the plumbing system is the site of latitic to trachytic magma reservoirs, located at depth between ~ 7–8 and ~ 2 km, finally shifting to small rhyolitic magmatic reservoirs above 2 km b.s.l1,2,4,6,9,11. All the plumbing system section above ~ 10 km is characterized by evolutionary processes dominated by fractional crystallization, which results in crystal mush zones and highly differentiated melts9,11. Specifically, the formation of high-silicic melts is explained by fractionation and extraction of interstitial melts from the shallow latitic-trachytic crystal mushes11,49, which can be remobilized via multiple injections of hotter shoshonitic-latitic magmas. Petrological and gas geochemistry data have confirmed the occurrence of small magma batches with different compositions that might interact, to a variable extent, during the pre-eruptive phases9,15,50. Specifically, these data point to the current presence of a latitic magma reservoir at 4–5 km of depth, passively degassing below La Fossa cone, which receives fluid and melt inputs from the deeper and more mafic reservoirs14,15,50.

Results

Sampling and general petrography

The volcanic products analyzed in this study refer to the intermediate/last emitted pyroclastic fall-out products of the La Fossa 1888-90 eruption, i.e. those emitted before February 18891. Samples were collected along the southern rim of the summit crater, which is almost entirely mantled by the Gran Cratere 2 pyroclastic products6. They belong to the shoshonitic series51 and can be classified as latites and poorly evolved trachytes [SiO2 concentration 59.7–62.3 wt%; Mg# 389]. Samples are vesiculated and have low phenocryst content (< 10 vol%) composed by clinopyroxene, plagioclase, and minor olivine and Ti-magnetite (Fig. 2). Clinopyroxene is by far the most abundant mineral phase, being ~ 50% of the total phenocryst volume. The groundmass is hypohyaline and contains sanidine laths. At the microscopic scale, it is evident the presence of reaction rims in olivine and plagioclase, in line with observations from past studies1.

Fig. 2
figure 2

Optical microscope images of the pyroclastic fall-out material of the 1888-90 eruption at La Fossa, showing the texture of tephra samples and the mineralogical assemblage composed of clinopyroxene, olivine, plagioclase and Ti-magnetite (photo under plane-polarized light).

Full size image

Textural and compositional observations on clinopyroxene crystals

Clinopyroxene crystals occur as euhedral to subhedral grains up to 4 mm in size, often forming glomerophyric structures with plagioclase and minor olivine. BSE imaging highlights the prevalence of concentric zoning with evident geochemical contrasts across clinopyroxenes (Fig. 3). The presence of dissolution textures is ubiquitous among crystals. The effect of dissolution is particularly evident at the boundary of dark-gray, Mg-rich cores that are frequently irregular and embayed (Fig. 3). In other crystals, the chemical gradients are almost negligible under electron microscope observations, but it is still possible to recognize the presence of dissolved portions that distinguish from the adjacent zones for the contrasting gray-scale color.

Fig. 3
figure 3

(a) Plot of clinopyroxene core and rim compositions on the Wo-En-Fs diagram; (b) Back-scattered electron (BSE) images showing some of the typical textural features of clinopyroxene phenocrysts from the 1888-90 eruption at Vulcano.

Full size image

The clinopyroxene core composition (Wo30−46 En50−60Fs10−20) falls within the fields of diopside and augite, comparing rather well with other clinopyroxene compositions found in products erupted at Vulcano over the period 54 ka − 1739 CE eruption12 (see Supplementary Material 1). Augitic cores are by far more abundant than diopsidic cores, while rims are dominantly augitic (Fig. 3; see also Supplementary Material 1). Mg#, calculated as [XMg/(XMg + XFe2+)], along core-to-rim traverses revealed a broad compositional range of Mg# 60–93. In accordance with our observations on BSE images, the compositional contrasts between adjacent zones (core, mantle and rim) can be as high as to 20% in Mg#. By considering the composition of the core only, crystals have been classified in three groups (Figs. 4, 5 and 6), namely: Group 1 that identifies Mg# 88–93 cores; Group 2 including core compositions at Mg# 84–85; Group 3 for the Mg# 75–80 cores. Group 1 (32%) and Group 3 (58%) define the dominant clinopyroxene core populations, while Group 2 is little represented (10%).

Fig. 4
figure 4

Representative core to-rim compositional zoning and associated back-scattered electron (BSE) of the 1888-90 clinopyroxene pertaining to the Group 1 Mg# 88–93 phenocryst population.

Full size image
Fig. 5
figure 5

Representative core to-rim compositional zoning and associated back-scattered electron (BSE) of the 1888-90 clinopyroxene pertaining to the Group 2 Mg# 84–85 phenocryst population.

Full size image
Fig. 6
figure 6

Representative core to-rim compositional zoning and associated back-scattered electron (BSE) of the 1888-90 clinopyroxene pertaining to the Group 3 Mg# 75–80 phenocryst population.

Full size image

In Group 1 (Mg# 88–93), clinopyroxene cores are mantled by highly differentiated growth zones, typically at Mg# 75–81. The compositional profile toward the rim can be smooth or it exhibits one or two reversely zoned bands with significant Mg# increases up Mg# 92.

Group 2 crystals (Mg# 84–85) display compositional profiles somewhat similar to Group 1. The core boundaries are normally zoned (Mg# decreases to Mg# 80–82), shifting to high-Mg reversely zoned bands toward the rim (Mg# 85–93). Contrariwise to Group 1, these crystals do not exhibit low-Mg overgrowths enclosing the cores.

Group 3 clinopyroxene cores (Mg# 75–80) are mantled by a Mg-rich reversely zoned band, which is compositionally comparable to those observed for Group 1 and 2. A second Mg-rich band can be observed at the rim of some Group 3 crystals.

In all groups, reversely zoned bands may affect the mantle portion surrounding the core, the outermost rim or both. The composition of these bands usually falls in the range Mg# 87–93 and Mg# 81–83. Group 1 and 2 also include crystals with no significant Mg# fluctuations across the core to rim transect (ΔMg# <3), hereafter referred to as unzoned crystals.

Time series of the SO2 fluxes

The SO2 flux inventory between 1978 and September 2024 is reported in Fig. 7. On average, excluding the degassing process experienced over the unrest periods of 1988-93 and 2021-23, the quiescent background SO2 flux stands to a median value of ~ 22 t/d (16.4 t/d IQR), enveloped between minimum and maximum values of 5 and 70 t/d, respectively (distribution of the records with Pearson’s second skewness coefficient of 0.97). Over the last 46 years, the emission rate has been almost stable except for periods during which the signal slightly exceeded the background value of ~ 50 t/d in 1996, 1999, 2005, 2006 and 2009. Significant temporal variations of the SO2 flux crater outgassing are associated with the two unrest phases occurred in 1988-93 and since 2021. Both periods display a similar waxing-waning pattern, with values of several order of magnitude higher than the steady background degassing rates. A comparison between the two periods reveals a similar mean SO2 flux of ~ 80 t/d (40 St. dev) and maximum values up to ~ 240 t/d. Insights into the two periods of unrest have been inspected quantifying the volume of magma that has degassed the measured mass of SO2 through the studied period. The volume of magma degassed sustaining the crater emission is classically estimated by scaling the measured SO2 flux values to outgassed mass fraction of Sulphur, well-approximated by the initial dissolved Sulphur content of olivine-hosted melt inclusion (e.g., Ref52). The volume of degassed magma (Vmin m3) is derived as:

Fig. 7
figure 7

Temporal inventory of bulk SO2 flux released from La Fossa crater of Vulcano between January 1978 and September 2024; data from Ref.15 and references therein are represented with the solid blue-line, whereas the solid grey line is for original data (see also the Supplementary Material 3). Throughout the 46 years, the emission rate has been measured with three different UV spectroscopy methods and at various sampling time depending on the sampling method and activity of Vulcano. The yellow-dashed areas envelop the two periods of unrest experienced by Vulcano during 1988-93 and 2021-23. The green-solid bar is the mean quiescent baseline degassing.

Full size image
$$\:{V}_{m}=\frac{0.5\times\:{fSO}_{2}}{\left[S\right]\times\:\rho\:\:\:\times\:(1-x)}$$

where fSO2 is the emission rate of SO2, [S] is the mean degassed mass fraction of Sulphur from the parental magma ranging between minimum and maximum 440 and 1470 mg/kg, respectively15,44, x is the crystal fraction (20%) and ρ is the density of the magma (2700 kg/m3). The magmatic origin of Sulphur feeding the quiescent background outgassing has been assumed to be ~ 10%, which increases to ~ 80% and ~ 50% during the climax stage and the declining periods of unrest, respectively.

Regarding the 2021 unrest, the temporal changes of the SO2 emission rate and geophysical monitoring data indicate the period September - November 2021 as the climax phase of the unrest. Then, the unrest progressively declined since December 2021. From January 2023 to September 2024, the SO2 outgassing went back close to the range of the quiescent background of La Fossa degassing, though moderate and short episodic SO2 increases were recorded during the entire period. Based on magma budget calculations, we have estimated that between September 2021 and September 2024 ~ 4.1–13.6 × 106 m3 of magma were intruded and degassed at shallow depth with a rate of 0.19–0.52 m3/s, in which a fraction of ~ 14% was outgassed quiescently from January 2023 to September 2024 (Table 1; see also the section of Methods for errors associated to the estimates). On the whole, these estimates are lower with respect to those derived for the 1988-93 unrest, which range from ~ 17.9 to 60.0 × 106 m3, assuming the climax phase from May 1988 and December 1991 and declining between January 1992 and December 1993. It is worth to note, however, that the 1988-93 unrest has been quite longer with respect to the most recent unrest. In this regard, we also need to point out that the degassed magma volume derived for the 1988-93 unrest might be affected by uncertainty due to the lesser SO2 flux measurements compared to those acquired over the unrest started on 2021. Considering the mean degassing rate in the quiescent period between 1994 and August 2021, we have calculated a volume of degassed magma of ~ 3.4–11.4 × 106 m3, which combined with the budget of magma degassed during the unrests delivers a total volume of intrusive degassed magma over the last 46 years of ~ 26.0–86.0 × 106 m3 (Table 1).

Table 1 Estimates of degassed magma volume during the two unrest periods of 1988-93 and 2021-23, and quiescently between January 1994 - August 2021 and January 2023 - September 2024. The total volume of intrusive degassed magma over the period January 1978 - September 2024 is also reported. The mean flux during the distinguished periods with the standard deviation, the outgassed mass of SO2 and the fraction of magmatic origin of Sulphur is indicated. Uncertainty in the inverted volumes depends on the method employed to record the bulk SO2 flux in the studied period.
Full size table

Discussion

Linking clinopyroxene composition and crystallization conditions beneath La Fossa

The broad compositional heterogeneity of clinopyroxene in the 1888-90 pyroclastic rocks delineates magmatic crystallization at variable evolutionary degree, and reconciles well with the idea of a complex system of reservoirs hosting primitive to highly differentiated melts beneath La Fossa. Recurrent disequilibrium textures and complex zoning profiles provide first clues for open system processes, such as magmatic recharge and mixing during magma evolution. However, understanding how these compositionally distinct magmas may have interacted with each other requires, preliminarily, to find a chemical correspondence between a zoned crystal and its coexisting melt. Accordingly, we have evaluated the clinopyroxene-melt equilibrium relations based on the Fe-Mg exchange equilibrium at Kd= 0.28 ± 0.0853. The 1888-90 tephra are relatively crystal-poor (crystals < 10 vol%) and thus the whole-rock composition has been assumed to be close to the host melt composition. Clinopyroxene cores exhibit high Mg# compared to the bulk rock Mg#, always falling outside the equilibrium range. This is indicative of a dynamic magma feeding system, whereby erupted magmas incorporated recycled crystals from more mafic magma batches. Equilibrium has been only found for Mg# between 64 and 75, which is generally attained in domains at the outermost crystal rim.

Recent studies have provided inferences on the crystallization of clinopyroxene in different magma storage zones beneath Vulcano Island on the basis of thermodynamic constraints and thermobarometric plus mass balance calculations11,12. As specified above, the composition of clinopyroxene cores, mantles and rims observed in tephra of the 1888-90 eruption, from Group 1 (Mg# 88–93) to Group 3 (Mg# 75–80), is consistent with data provided by Ref.11,12 for clinopyroxenes found in products erupted over the period 54 ka − 1739 CE eruption (see also Supplementary Material 1), reflecting crystallization in differently evolved reservoirs. For instance, the composition of the Group 1 diopsidic cores (Mg# 88–93) overlaps the range of crystal Mg# in equilibrium crystallization with mantle-derived basalts residing close to the Moho11,12. Being the most mafic end-member of the differentiation series exposed at Vulcano, these primitive magmas were found to host a clinopyroxene assemblage crystallizing at P ∼620–740 MPa and T ∼1190–1250 °C. The subsequent accumulation and differentiation of these basaltic melts in the upper crust would have led to the formation of more differentiated magmatic reservoirs. Existing thermodynamic constraints11,12indicate crystallization of Mg# 83–86 clinopyroxene (i.e., comparable to Group 2 crystals of this work) in shoshonitic and latitic melt at P ∼100–450 MPa and T ∼1100–1180 °C. The crystallization of clinopyroxene with Mg# 80–82 has been associated to equilibrium with trachytic melts at P ∼120 MPa and T ∼1080–1100 °C11, whereas Mg# 76–79 clinopyroxenes are supposed to be in equilibrium with rhyolitic melts at P ∼50 MPa and T < 1100 °C.

Based on our observations, the Group 3 clinopyroxene bears a wide range of Mg# at their cores (Mg# 75–80) that reflects equilibrium with hybridized melts spanning from trachytes to rhyolites. Rhyolitic products erupted during the 1888-90 eruption have been interpreted as the result of low-pressure fractional crystallization, possibly starting from a latitic magma, plus the assimilation of crustal material1,2. The trachytic compositions could have instead resulted from the hybridization of the latitic end-member with the accidental presence of residing rhyolitic magmas at very shallow levels (1–2 km b.s.l.). The crustal interval at depth of ∼2–5 km is considered the level of neutral buoyancy of ascending magmas, where hybridized trachytic/latitic melts preferentially develop due to accumulation and protracted differentiation of mafic melts in the upper crustal regions1,2,4,15. A buoyancy-controlled magma ascent behavior is in agreement with density data of the crustal layers obtained from Vp data and magma densities from fluid inclusions2,4,31, supporting the presence of a shallow, 2–5 km-depth zone of magma accumulation where the Group 3 clinopyroxene has realistically formed.

Fe-Mg zoning reveals magmatic processes and their timescales

As stated above, latitic and trachytic to rhyolitic magmas of Vulcano develop from basalt differentiation via fractional crystallization and assimilation1,2,4,6,9,11,12,37,44. As basalts migrate and evolve, the Fe-Mg composition of the crystallizing clinopyroxene changes in accordance with the degree of magma differentiation, keeping the trace in the zoning profile of the reservoirs they passed through30. Temperature and compositional fluctuations in a magma reservoir due to replenishment by more mafic magma can form Mg-rich mantle/rim overgrowths compared to the adjacent zones in the resident crystals. Instead, crystals carried by the ascending magma would develop the overgrowth of a new Mg-poor composition as a consequence of the intrusion and residence into more evolved magma bodies. We have observed such patterns in phenocrysts characterized by relatively high-Mg cores (Mg# >84). These cores have rounded/irregular boundaries and are mantled by highly differentiated growth zone at Mg# 75–81. The dissolution surface between the core and the overgrowth indicates that the overgrowth is not caused by progressive fractional crystallization. In accordance with experimental and petrological observations, we have interpreted this feature as the result of dissolution/destabilization caused by the rapid transport of crystals through the upper crust, owing to the strong dependence of clinopyroxene stability on pressure54,55,56,57,58. This finding supports the idea of progressive crystal transport from the deep-seated basaltic region to the shallow storage regions. The absence of overgrowths at Mg# 83–86 reinforces the previous assumption, indicating that the basaltic magma either accelerated upwards or that it just stalled briefly in the shoshonitic/latitic reservoir prior to the latest stage of storage at the trachytic crustal level.

Significant Mg# increases, compared to the adjacent zones at the clinopyroxene mantles and rims, confirm the occurrence of mafic injections before the eruption. Such events typically leave a clearly identifiable record in clinopyroxene provided by abrupt Mg# increase. In the 1888-90 tephra, except for a few crystals with no evident zoning, clinopyroxene exhibits one or two high Mg# bands at Mg# 88–93 and Mg# 81–82 that alternate with low Mg# growth zones (Mg# <79). Two high-Mg# bands are well recorded by the clinopyroxene of Group 3 (Mg# 75–80 cores). These crystals also preserve mantle and/or rim overgrowths of comparable composition (Mg# <79) as evidence of subsequent stages of crystallization in the same highly evolved reservoir (Figs. 4, 5 and 6). Thus, we infer at least two main events of magmatic recharge affecting the shallow plumbing system at the latitic/trachytic crustal level, then mobilizing the latest residual melt prior to the final ascent.

Some evidence points to a third episode of mafic intrusion into the same section of the plumbing system. As mentioned before, some of the most mafic Group 1 (Mg# 88–93) and Group 2 (Mg# 84–86) crystals keep the record of this early event, which is testified by the important Mg# decrease at the core boundary and the subsequent overgrowth of a low-Mg zone as a result of the mafic magma input followed by a period of residence in the shallow plumbing system (Figs. 4, 5 and 6).

Although the sequence of zoning patterns attests the continuous replenishment of the shallow plumbing system, distinguishing the different phases of magmatic intrusion is not straightforward, especially because most of the 1888-90 crystals display only one evident recharge band with abrupt Mg# increase. The use of chemical diffusion to quantify the crystal residence time has provided an efficient solution to this problem, allowing to temporally correlate the identified events of magmatic intrusion. Specifically, diffusion chronometry has been applied here with the aim of determining time of crystal residence following each episode of magmatic injection documented along the clinopyroxene core-to-rim profile. This means that more than one diffusion time has been calculated for complex zoning profiles attesting multiple recharging events (Fig. 8; Supplementary Material 2). Our time determinations define three clusters of timescales in the range of 85–140 years, 16–36 years and 2–7 years, respectively (Fig. 8; Supplementary Material 2). The longest timescales of 85–140 years have been calculated for early formed Mg# 85–93 crystals (Group 1–2) initially residing in the deeper parts of the crust and then remobilized and transported upwards by mafic melts. These timescales refer to the total residence time after the crystal is introduced into the shallow storage region and relate, therefore, to the early recharge episode identified via chemical zoning. Such early event also appears as a Mg# 88–93 recharge band in a few crystals of the Group 3, which clearly resided at low pressure when the magma arrived. The arrival of this primitive magma may have been the cause for partial destabilization of the plumbing system that led only to the minor phreatic eruptions observed since 177140 (i.e., 117 years before the beginning of the 1888-90 cycle). However, this mafic recharge was almost entirely stored into the plumbing system. Indeed, chemical and temporal evidence indicates that the magma stalled in the upper crust after the intrusion, allowing later stages of crystal growth and diffusive re-equilibration of the newly formed chemical gradient (Fig. 8).

Fig. 8
figure 8

Compositional zoning patterns, diffusion profiles and calculated timescales of clinopyroxene crystals from the 1888-90 eruption at Vulcano Island. The left diagram illustrates the diffusion modeling for some representative zoned crystals. The analytical dataset (white diamonds) was modelled non-isothermally using diffusion curves (colored lines) that re-equilibrate at different temperatures to quantify the total time elapsed between each event of mafic intrusion and the eruption onset. T1, T2 and T3 indicate the best-fit diffusion curves fitting the first, second and third chemical gradient, respectively, from core to rim. The right diagram shows all the calculated timescales and the recharging event they refer to. Error bars represent the 2σ uncertainties associated to the time estimates. Error bars are within the symbols for the shortest timescales (green squares).

Full size image

We can assume that once reached the upper crust, crystals reside for decades into the trachytic/latitic reservoirs prior to come into contact again with a new mafic recharging magma. During the time interval between the first event of mafic replenishment and the eruption onset, the arrival of two new magma pulses has been documented. The temporal sequence of these two replenishment events is defined by distinct diffusion timescales of 16–36 years and 2–7 years that were obtained from the diffusive relaxation of the recharge bands at the crystal mantle and rim (Fig. 8; Supplementary Material 2). These two recharging events could have been responsible for plumbing system instability leading to the sporadic phreatic explosions observed before 1888 (i.e., 1873, 1876-79 and 1886). The latest phase of mafic recharge (2–7 years before the eruption onset) is typically recorded at the clinopyroxene outermost rim, and is also consistent with the short residence timescales (5–10 years) of plagioclase crystals in these levels of the plumbing system9. This final replenishment event must have produced sufficient pressurization of the upper plumbing system to remobilize the residing melt and favor its extraction from the dense crystal mushes up to the surface.

Deep mafic recharge drives degassing crisis and volcanic unrest

Aside from the clinopyroxene zoning, the presence of olivine crystals with orthopyroxene reaction rims and plagioclase with K-feldspar overgrowths provide additional evidence of magmatic recharge within the shallow magmatic system, in line with previous petrological observations1,44. The refilling of the shallow magmatic reservoirs by mafic magmas has occurred periodically at La Fossa during its 5.5 ka of volcanic activity1,5,6,9,11,12,43,50. The question is whether or not the ascent of fresh, volatile-rich magma can be responsible for the escalating degassing crisis documented on the island since the beginning of the 20th century. In fact, a model of increasing gas exsolution from a stationary magma body has been also hypothesized as an alternative to magma ascent and its shallow emplacement to explain the recent degassing unrests7,19. Whatever the scenario, the gas source fueling anomalous degassing has a clear deep magmatic signature. Remarkable increases in the SO2 flux from fumaroles were detected at La Fossa during the major periods of unrest in 1988-93 and after September 2021, with similar increasing patterns and maximum values up to ~ 230 t/d. In both circumstances, the composition of the Sulphur-rich emissions suggested a magmatic origin15,55,59. According to several authors, the source of magmatic gases feeding the fumaroles are latitic and trachytic magmas stored at 3–5.5 km of depth, which in turn are fed by deeper fluids released by more mafic magmas2,4,7,9,44,50. Degassing of a more differentiated and shallower (< 3 km) magmatic source is, instead, considered insufficient to justify the sustained emission of SO2 for prolonged periods of time during the crisis, an inference also drawn from the low Sulphur content measured in melt inclusions62. The degassing of deep-seated mafic melts has been, therefore, preferentially invoked to explain the SO2 output budget during the 1988-93 unrest60. On the same line of reasoning, based on volcanic gas numerical modelling and mass balance calculations, Ref.15 argue for the ascent-driven volatile exsolution of a mafic magma, ascending to final emplacement depths of 4–5 km, as the driver of degassing unrest observed since fall 2021. These calculations, accounting for the total SO2 mass output, are consistent with the emplacement of at least ~ 1.8 × 106 m3 of magma between mid-September and December 2021, which is in line with our lower estimation ( ~1.6 × 106 m3) for the same period.

These considerations reconcile well with our petrological dataset for the 1888-90 eruption, which emphasizes the migration of deep-seated, basaltic/shoshonitic magmas and subsequent emplacement in the shallow system characterized by trachytic-latitic reservoirs. During residence, these magmas received recurrent inputs of mafic, volatile-rich magma, likely releasing much of accumulated gases after recharge. Our petrological data, although clearly supporting the involvement of three distinct events of mafic replenishments in the 1888-90 eruption, do not allow us to assert whether analogous events also triggered degassing unrests during the 19th century. However, the similarity between past and present dynamics is doubtless. A common point throughout the recent (19th century to present day) history of La Fossa is that the ascent and low-pressure emplacement of mafic, volatile-rich magma batches do not necessarily culminate into an eruption. For instance, the 1888-90 eruption occurred only after three main recharging events that involved the same shallow section of the plumbing system. Similarly, during the last century, much evidence has been provided for magma/gas accumulation and pressurization of the shallow plumbing system, which culminated into intense gas release, without however triggering a new eruption. The estimated volumes of intruded magmas during the recent unrests are significant, being on average 17.9–60.0 × 106 m3 in 1988-93 and 4.1-13.6 × 106 m3 in 2021-24, but clearly not still sufficient to ultimately mobilize the resident magma. According to our results, magmas can therefore reside for decades in shallow reservoirs, undergoing subsequent recharging episodes, the last of which occurs just few years before the eruption onset. It is worth to note that intruded magma volumes since 1890 are significantly higher than the ~ 2.4 × 106 m3of products emitted during the 1739 CE Pietre Cotte eruption43 (Vetere et al., 2015) and comparable to the 107 m3of products erupted during the 1888-90 cycle39. These considerations lead to the idea that the shallow system might already be highly pressurized, and possibly prepared for a new eruption once some threshold pressure is reached.

Concluding remarks

The analysis and interpretation of the SO2 flux from the fumarole field of La Fossa during recent unrests allowed gathering information about the origin and quantity of magma involved in the degassing activity and depths reached by intruded magmas. There is a good correspondence between the modeled magma emplacement depth (4–5 km beneath surface) and the latitic-trachytic crustal interval that we have identified as the main zone of magma emplacement receiving recurrent mafic inputs before the most recent eruptive activity of La Fossa.

A time gap of 149 years separates the 1888-90 CE eruption from the 1739 CE eruption of Pietre Cotte, an important event in the volcano-stratigraphic record of the island, which was followed only by sporadic phreatic eruptions over the period 1771-1886 CE. In this time interval, three distinct recharge phases by mafic, volatile-rich magmas have been detected by clinopyroxenes, with a frequency of decades from each other. The first mafic injection occurred 85–140 years before the 1888-90 eruption, after ~ 30 years since the Pietre Cotte eruption. In such a view, this first mafic replenishment may have destabilized the plumbing system triggering the small phreatic eruptions since 1771, without significant upward mobilization and involvement of magma into the eruption, given the phreatic nature of the events. The second mafic recharge dates to 16–35 years before the volcano re-awakening of 1888, but only after a third replenishment event (2–7 years before 1888) the residing magma finally erupted. Although neither gas chemistry data nor historical reports documenting degassing unrests are available for the 19th century, we cannot exclude the possibility that each recorded replenishment episode produced sufficient accumulation of magmatic gases in the shallow system to produce an escalation in surface degassing, as observed in recent times.

Degassing of mafic magma, either during ascent or through the release of previously accumulated gas upon storage, is argued as a driver of unrest at La Fossa. Nevertheless, none of the unrest periods of the last century has evolved in a volcanic eruption. This trend mimics somehow the dynamics preceding the 1888-90 eruption. In terms of hazard, results of this study offer some points for reflection, indicating short-term residence of just 2–7 years at low pressure, once the shallow system has been sufficiently pressurized by multiple magmatic inputs.

Methods

In situ micro-analytical investigations

The 1888-90 tephra of this study have been prepared for polished thin sections in order to be inspected for textural observations and microanalysis by SEM-EDS/WDS. High-contrast back-scattered electron images (BSE; 1024 × 864 pixels) and major element compositions on clinopyroxene crystals have been obtained at the Dipartimento di Scienze Biologiche, Geologiche e Ambientali of the University of Catania (Italy) by means of a Tescan Vega-LMU scanning electron microscope equipped with an EDAX Neptune XM4-60 microanalyzer operating by energy dispersive system (EDS) equipped by an ultra-thin Be window. The EDS microanalyzer is coupled with an EDAX WDS LEXS (wavelength dispersive low energy X-ray spectrometer) calibrated for light elements. Operating conditions were set at 20 kV accelerating voltage and ~ 8 nA beam current for obtaining high-contrast BSE images and 20 kV accelerating voltage and 2 nA beam current for the analysis of major element concentrations in clinopyroxene. Repeated analyses on an internationally certified standard of diopside (SPI 02753-AB Serial KF crystal #21) during the analytical runs ensure precision for all the collected elements on the order of 3–5%; accuracy is on the order of 5%. The analysis of major elements on clinopyroxene was performed along core-to-rim transects, assuming a distance between analytical spots on the order of 3 to 8 μm depending on the crystal size. All the compositional analyses on clinopyroxene crystals can be found in the Supplementary Material 1 associated to the online version of this article.

Diffusion modeling application on clinopyroxene

Fe-Mg compositional zoning in clinopyroxene has been investigated to quantify the crystal residence time via elemental diffusion chronometry. Twenty out of thirty-two clinopyroxene crystals examined in this study exhibit pronounced Fe-Mg gradients and were therefore selected as suitable for diffusion timescale estimations. Zoned crystals were modeled numerically with the method of finite differences61and using a non-isothermal diffusion incremental step approach62 due to the presence of complex profiles (i.e., alternation of normal and reverse zoning). Because of evident geochemical contrasts across clinopyroxenes, a sharp step-like initial profile, assuming a maximum concentration range at each boundary layer, was adopted as an extreme initial concentration to model the sequence of chemical gradients within crystals57,61 (see Supplementary Material 2).

The modeling temperatures for compositionally distinct boundary layers account for the crystallization temperatures for clinopyroxene crystals in equilibrium with different melts11 (i.e., 1220–1190 °C for basalts, 1180–1100 °C for shoshonites and latites, and 1100–1080 °C trachytes to rhyolites). Specifically, we have considered a minimum temperature of 1190 °C when modeling the diffusive relaxation of clinopyroxene chemical gradients that re-equilibrate at the composition of basaltic melts (Mg# ≥88), a temperature of 1100 °C for chemical re-equilibration at the concentration of shoshonitic-latitic melts (Mg# 83–86), and a temperature of 1080 °C for re-equilibration in more evolved trachytic to rhyolitic melts (Mg# ≤82). Calculated timescales are therefore referred to as ‘maximum’, as the modeling is carried out by using temperatures that are the lowest estimates for magmas at Vulcano.

The choice of temperature has major effect on the diffusivity of Fe-Mg in clinopyroxene, which is assumed to be independent from concentration (at least in the range Di93-Di65) and oxygen fugacity, in accordance with experimental data available63,64. Here, the diffusivity is calculated using a pre-exponential factor (D0) of 2.77 (± 4.27) × 10−7 m2s−1 and activation energy (Q) of 320.7 (± 16) kJ mol−1as experimentally determined for natural diopside crystals64. In terms of error calculations for our timescale determinations, we assumed an uncertainty of ± 15 °C errors on temperature data11,12. The error on the diffusion coefficient64 together with the error on the temperature, which affects the diffusion coefficient D, were then used to calculate the relative error propagation on the residence time, following standard procedures for error propagation65. Residence time estimates and calculated uncertainties are reported in Supplementary Material 2.

Measurements of the SO2 flux at Vulcano

Remote measurements of bulk Sulphur dioxide (SO2) emission rates from the fumarole field of the La Fossa Crater have been carried out since 1978 using ground-based ultraviolet spectroscopy techniques. Three different spectrometers and methods have been employed throughout the 46-year period: discrete sampling by COrrelation SPECtrometer (COSPEC66), using CCD-based ultraviolet devices67 and UV Camera between 2006 and 2008, and since 2009 by the DOAS FLAME scanning spectrometer network (FLux Automatic Measurement68). The method to perform SO2 measurements is common for the three techniques and consists in collecting UV spectra by crossing underneath the volcanic plume to get the average burden of SO2 cross-section. Integrated SO2column amounts are then multiplied by wind-plume transport speed to determine flux69,70. In the case of COSPEC and CCD spectrometers, plume traverses were conducted by traversing the volcanic plume installing the spectrometers in a car-ship at a mean distance of 4 km and walking along the crater rim. Instead, the FLAME station allows automatic measurements from a permanent site scanning the sky to the line between the station and the volcano summit17. Over the years, SO2 observations have been done with different frequency due to monitoring activities and timestamp of the three methods, spanning from tens to hundred sampling per day. Error in the SO2 flux is mostly due to the uncertainty in wind-plume speed estimation, and it ranges from ± 13% and 23% for COSPEC, ± 22% and 36% for FLAME, and between ± 26% and 43% considering a total uncertainty deriving from both COSPEC and UV traverses, and DOAS Scanning spectrometers errors71,72. In the present work, a complete time series of SO2 emission rates covering the time interval from January 1978 to September 2024 has been provided, combining the observations from the three aforementioned techniques and literature data from Ref.15 and references therein (Fig. 7; Supplementary Material 3).

Data Availability

All the petrological data produced in this work are available in the Supplementary Material 1 and 2. The original data related to the bulk SO2 flux covering the period January 1978 - September 2024 are available in the Supplementary Material 3.