Geochemical and isotopic insights into the assembly, evolution and disruption of a magmatic plumbing system before and after a cataclysmic caldera-collapse eruption at Ischia volcano (Italy)

Abstract

New geochemical and isotopic data on volcanic rocks spanning the period ~75–50 ka BP on Ischia volcano, Italy, shed light on the evolution of the magmatic system before and after the catastrophic, caldera-forming Monte Epomeo Green Tuff (MEGT) eruption. Volcanic activity during this period was influenced by a large, composite and differentiating magmatic system, replenished several times with isotopically distinct magmas of deep provenance. Chemical and isotopic variations highlight that the pre-MEGT eruptions were fed by trachytic/phonolitic magmas from an isotopically zoned reservoir that were poorly enriched in radiogenic Sr and became progressively less radiogenic with time. Just prior to the MEGT eruption, the magmatic system was recharged by an isotopically distinct magma, relatively more enriched in radiogenic Sr with respect to the previously erupted magmas. This second magma initially fed several SubPlinian explosive eruptions and later supplied the climactic, phonolitic-to-trachytic MEGT eruption(s). Isotopic data, together with erupted volume estimations obtained for MEGT eruption(s), indicate that >5–10 km³ of this relatively enriched magma had accumulated in the Ischia plumbing system. Geochemical modelling indicates that it accumulated at shallow depths (4–6 km), over a period of ca. 20 ka. After the MEGT eruption, volcanic activity was fed by a new batch of less differentiated (trachyte-latite) magma that was slightly less enriched in radiogenic Sr. The geochemical and Sr–Nd-isotopic variations through time reflect the upward flux of isotopically distinct magma batches, variably contaminated by Hercynian crust at 8–12 km depth. The deep-sourced latitic to trachytic magmas stalled at shallow depths (4–6 km depth), differentiated to phonolite through crystal fractionation and assimilation of a feldspar-rich mush, or ascended directly to the surface and erupted.

Appendix

Fluid-mineral equilibria

Following Fabbrizio et al. (2009), activity values for sanidine were taken by graphically reading the activity-composition plot in Waldbaum and Thompson (1969), those for magnetite by using the activity-composition expression of Woodland and Wood (1994) along the join magnetite-ulvospinel, and those for annite by the ionic model of Czamanske and Wones (1973). Calculations at 930 °C return fH₂O values of 668 bars and 498 bars for the least (NNO + 0.75) and most oxidized (NNO + 1.44) trachyte (or trachy/phonolites), respectively. Furthermore, we computed the fHF and fHCl by using the expressions for chemical exchanges between fluoroannite/phlogopite–annite/fluorophlogopite and chloroannite/phlogopite–annite/chlorophlogopite pairs from Munoz (1984), based on the compositional deconvolution of Gunow et al. (1980) for the siderophyllite and annite components in biotites. We obtained fHF = 0.4 bars and fHCl = 335 bars at NNO + 0.75, and fHF < 0.1 bars and fHCl = 18 bars at NNO + 1.44.

MELTS calculations

For the parental latite ascending from depth larger than 6 km, we lack a direct temperature estimate as well as information on volatile contents. By analogy with younger than 3 ka Ischia latites (Moretti et al. 2013), temperatures around 1,200 °C and maximum water contents of 3–3.5 wt% can be suggested for the Chiummano latite. Of note is that the application of the CaO-in-glass thermometer of Cioni et al. (1998) returns a temperature of 1,060 °C.

MELTS calculations were thus initialized with a starting composition represented by the Chiummano latite (Supplementary table, Table 3) having 2.5 wt% of water, fO₂ conditions of QFM + 1 (i.e. NNO + ~0.6), the latter more reduced than biotites in poorly evolved trachytes, and initial P and T at 200 MPa and 1,200 °C, respectively. The crystallizing solid phases are those consistent with petrographic description (see also Table 3). A ∇P/∇T gradient of 2.6 bar/°C was applied in such a way that 930 °C were reached at 130 MPa (5.5 km). In this P–T conditions, the poorly evolved trachyte magma forms. After crystallization of clinopyroxene, Ti-magnetite, plagioclase, minor olivine and biotite, the residual liquid (72 %) attains a composition close to MEGT 0308 trachyte sample (step 1 in Table 3). Interestingly, biotite starts crystallizing between 940 and 930 °C for the selected water content. An initial water content larger than 2.5 wt% would prevent crystallization of biotite.

The following step requires assimilation of felsic rocks to approach the composition of the MEGT 0312 sample; otherwise, fractional crystallization would only drive the system towards more silica-undersaturated compositions. The assimilant was composed by two feldspars (Na-sanidine and K-sanidine), whereas the assimilating magma was represented by the trachyte, derived by fractional crystallization from the parental latite at 930 °C. Table 3 in text reports the MELTS outcomes for the assimilation process, in which feldspars, initially at 600 °C, are isobarically and isothermally assimilated by the trachyte at 930 °C. Incremental assimilation of P₂O₅- and TiO₂-free feldspars gives results that are in line with the Schiappone trachy-phonolitic MEGT 0312 composition (step 2 in Table 3; Fig. 8), allowing a slight decrease of P₂O₅ and TiO₂ content (this latter favoured by concomitant precipitation of Ti-magnetite).

Alternatively, feldspar assimilation can involve directly the latite (1,200 °C), ascending between 130 and 100 MPa and infiltrating feldspar cumulitic residuals. If this is the case, during feldspar digestion, water is lost and dissolved H₂O falls down to 1 wt% in the final melt. In the absence of a detailed melt inclusion study, we will retain here the first scenario, involving feldspar assimilation from trachyte at 930 °C. It is also worth noting that during assimilation, fractionated solids, which amount are nearly equal in mass to that of assimilated feldspars, are dominated by feldspar Or_60–67. This implies that, after each assimilation event, feldspar cumulates are potentially left in place for future remobilization.

After feldspar assimilation, fractional crystallization drives again the residual liquid towards undersaturation, with SiO₂ reaching 55–56 wt% and Al₂O₃ dropping inevitably, accompanied by a significant decrease of FeO_tot and TiO₂. Only a depressurization taking place right after assimilation allows to keep high silica and alumina contents, nearly constant FeO_tot and to increase TiO₂, consistent with observations (step 3 in Table 3).

After assimilation and during this final stage of isothermal decompression, biotite, two feldspars, spinel and apatite crystallize. This association is observed only by keeping QFM + 2. Lower oxygen fugacities promote olivine, whereas higher values do not allow Na₂O to overcome K₂O. Nepheline crystallizes at the very end (T = 910 °C and P = 100 bar; see MELTS step 4 in Table 3), when the residual liquid amounts to 3.5 wt%. However, liquid composition at that stage would be even more evolved than that in step 3 (Table 3).

Finally, we remark that MELTS cannot be used to investigate accessory phases such as sphene, apatite and Na-amphiboles, at least for the studied compositions, for which the role of fluids other than water (S-species, halogens) may have dramatic effects on the mixing properties of such solid solutions.