Volcanological evolution of the Rivi–Capo Volcanic Complex at Salina, Aeolian Islands: magma storage processes and ascent dynamics

Abstract

Lava flows and pyroclastic deposits from strombolian fallout related to the activity of the Rivi and Capo volcanoes, which are representative of early subaerial volcanoes on Salina (Aeolian Islands), have been investigated through a geological–petrological approach. Our geological field survey shows that Rivi and Capo volcanoes are part of a single N50°E aligned volcanic complex, here named Rivi–Capo Volcanic Complex (RCVC). Stratigraphically specific rock sampling has allowed reconstruction of the magma feeding processes through time. Whole rock major element compositions, together with core-to-rim profiles of plagioclase and clinopyroxene crystals, show a general evolution toward more basic compositions through the three formations constituting the Capo volcano and within the Rivi center. MELTS simulations and mass balance modeling suggest that the RCVC rocks are the result of fractional crystallization of plagioclase, clinopyroxene, and olivine (ca. 45 % of solid removed) from a primary magma. In addition to fractional crystallization, continuous recharge and mixing with more basic magma coming from deeper parts of the magmatic plumbing system contributed to the final volcanic rock compositions. Our textural and microanalytical data on plagioclase and clinopyroxene crystals allow the definition of a multilevel magmatic storage system with reservoirs at ~20 and ~3 km below sea level. When processes of magma differentiation, ascent, and storage are considered together with the stratigraphic position of each sample, a history of continuous modification of the RCVC plumbing system can be constructed. Volcanism may have been characterized by fissure-type eruptions during the early stages (Lower Capo, Lower Rivi, and Middle Capo Formations), gradually changing later to central-type volcanism (Upper Capo and Upper Rivi Formations).

Appendix

Geometry of the RCVC plumbing system has been here redrawn with respect to that of literature (cf. Nazzareni et al. 2001, 2011; Zanon and Nikogosian 2004; Lucchi et al. 2013c). The presence of a shallow magma reservoir where plagioclase can grow and develop oscillatory zoning and sieve textures at the rim is essential because these features cannot be acquired in storage zones placed at the Moho. In order to better define the articulation of the feeding system, three simulations of crystal fractionation have been performed by using the MELTS software (Ghiorso and Sack 1995; Asimow and Ghiorso 1998). In addition, recharge by more basic magma in the shallow magma reservoir and selective addition of mineral phases at various levels have been also modeled.

Composition of primary magmas

The first step of crystal fractionation simulated through MELTS is aimed at obtaining the thermodynamic parameters (P, T, fO₂, H₂O content) for the definition of geometry of the shallow plumbing system at RCVC in the further steps of simulation. We have simulated the geochemical evolution of a primary magma starting from a mantle-equilibrated bulk rock composition. This primary composition has been obtained from the less differentiated and porphyritic lava of the entire Rivi–Capo volcanic sequence (MCF7 sample; Table 1), which is a basalt with anhydrous contents of SiO₂ = 49.8 wt%, MgO = 5.63 wt%, and Mg# = 48. The mantle-equilibrated composition was obtained through mass balance calculations by adding to the sample MCF7 an ultramafic mineral assemblage composed of Fo₈₃ olivine (11 %), clinopyroxene with Mg# = 84 (8.5 %), An₉₀ plagioclase (3 %), orthopyroxene with Mg# = 77 (1 %), and Ti-magnetite (0.5 %) until the crystallizing olivine within the magma was equilibrated with that in the mantle (assumed at Fo₈₈). Mineral compositions for olivine, plagioclase, and clinopyroxene have been chosen from the most mafic of this work (samples LCF1, LCF 4, and MCF7; ESM 2 and ESM 3), whereas Ti-magnetite from Calanchi et al. (1993) and orthopyroxene from Di Maggio (1994). The resulting primary composition has SiO₂ = 48.8 wt%, MgO = 10.3 wt%, and Mg# = 63 (ESM 4).

Crystal fractionation through MELTS simulations

The first step of crystallization (CF1) has been performed starting from the MCF7 sample equilibrated to the mantle, which can be considered the first magma entering in the deep reservoir at the Moho level. This step models isobaric crystal fractionation at 550 MPa. The chosen pressure, corresponding to ~21 km (Fig. 13), well matches with data provided by De Ritis et al. (2013 and references therein), with the reliable assumption that depth of the mantle–crust boundary is not significantly changed during the last 240 ka. Used temperature for the first step is between 1,260 and 1,220 °C, a value obtained through the “Find liquidus” option of the MELTS software (Ghiorso and Sack 1995; Asimow and Ghiorso 1998). More than 40 runs of simulation by MELTS have been performed to have plausible H₂O content and the fO₂ in the starting composition of the primary magma. The used H₂O content is 1.9 wt%, which is higher than 0.9–1.4 wt% proposed by Nazzareni et al. (2011) for products of Rivi and Corvo volcanoes. Nazzareni et al. (2011) calculated water contents on the basis of H in clinopyroxene. However, clinopyroxene crystallization at depth of 21 km is prevented if water contents are below 1.9 wt%. Gertisser and Keller (2000) proposed values of oxygen fugacity at the NNO buffer for Salina magmas. Our MELTS simulations (>40 runs) suggest, however, that oxygen fugacity at the QFM buffer is more consistent with the petrographic observations. The resulting crystallization trends under these conditions show that magma residing at 21 km produces fractionation of ~5 % of solid phases constituted by 4 % of Fo₈₅₋₈₃ olivine and 1 % of orthopyroxene with Mg# = 76 (ESM 4). These mineral compositions match well with those of the ultramafic assemblage used for mantle equilibration (see previous chapter) and with the Fo contents measured at cores of olivine of the RCVC products (ESM 4). The final magma composition is basaltic, with SiO₂ = 49.2 wt% and MgO = 8.7 wt% (Fig. 11 and ESM 4).

The second step of crystal fractionation (CF2) simulates polybaric fractionation of the magma from 21 km toward the shallower levels of the plumbing system (Fig. 13). Starting composition for this second step of calculation coincides with the final composition and water content (2 wt%) resulting from the first step. Starting pressure of crystallization is 550 MPa, whereas final pressure is 80 MPa that corresponds to ~3 km of depth (Fig. 13). Pressure has been constrained considering the bottom depth of 1–3.3 km for the sedimentary/volcanic cover estimated by De Ritis et al. (2013) in the Vulcano-Lipari-Salina sector. These values are also in accordance with the available seismic data provided by Ventura et al. (1999). Also in this case, the starting temperature (1,220 °C) has been found with the “Find liquidus” option of the MELTS software and corresponds to the last temperature resulting from the first step of simulation (ESM 4). The final temperature for the second step has been set at 1,110 °C, which is rather in accordance with temperature of calcalkaline basaltic magmas at these depths. fO₂ buffer has been kept at QFM. After 32 runs, choice of the above-mentioned parameters put into evidence the fractionation of 1–13 % of solid phases during magma ascent from 21 to 3 km (ESM 4). The fractionated assemblage is constituted by 1–6 % of Fo₈₃₋₇₈ olivine, 0–5 % of clinopyroxene with Mg# = 76 and Wo₄₆₋₅₀En₄₀₋₃₆Fs₁₃, and 0–2 % of An₉₁ plagioclase (ESM 4). Olivine and plagioclase compositions match with those measured in the RCVC products (ESM 2 and ESM 4), whereas clinopyroxene resulted from MELTS simulation is slightly more calcic than that measured. Such discrepancy could therefore explain the CaO liquid line of descent (Fig. 11), which has lower CaO contents than what expected. Final magma composition is basaltic, with SiO₂ = 49.9 wt% and MgO = 6.1 wt% (Fig. 11 and ESM 4). The first two steps of crystal fractionation have produced results strongly consistent with the petrographic and chemical characteristics of the MCF7 sample before its re-equilibration to the mantle (Fig. 11, Table 1, and ESM 4).

The third step of differentiation (CF3) simulates isobaric crystallization at shallow levels (~3 km; Fig. 13) of a magma that has already fractionated 20 % of solid phases (Fig. 11). The starting composition of the third step used for MELTS calculations coincides with the final composition and water contents (2.25 wt%) resulting from the second step (ESM 4). Pressure of crystallization has been set at 80 MPa, which corresponds to ~3 km and agrees with crustal discontinuity found below the central sector of Aeolian islands (Ventura et al. 1999; De Ritis et al. 2013). Starting temperature coincides with the final values of temperature resulting from the second step of simulation (1,110 °C). Final temperature of the third step has been fixed at 1,065 °C, in accordance with emission temperature of worldwide calcalkaline basalts and basaltic andesites. fO₂ buffer has been set at QFM+1, which is a plausible and acceptable redox condition for differentiation of this shallow magma reservoir. After 46 runs, choice of these parameters has led to the fractionation of 1–34 % of solid phases constituted by 0–16 % of clinopyroxene with Mg# 71–75 and Wo₄₉₋₅₃En₀₃₅₋₃₆Fs₁₁₋₁₄, 1–14 % of An₈₅₋₉₁ plagioclase, 0–4 % of Fo₇₆₋₇₉ olivine, and 0–4 % of Ti-magnetite (ESM 4). Olivine compositions match well the measured compositions within the RCVC products (ESM 4). The calculated clinopyroxene and plagioclase compositions are slightly more calcic than those measured, which results in a liquid line of descent more depleted for CaO than what is expected (Fig. 10). This third step of isobaric crystallization and fractionation partly explains most of the whole rock compositions of Upper Rivi, Upper Capo, and Middle Capo Formation.

Recharge and mixing processes

Crystal removal from the liquid simulated through the MELTS software has evidenced that products of Lower Capo Formation and some products of the Middle Capo Formation cannot be accounted for by sole fractionation. Discrepancies observed for major elements coupled with evolution toward more basic terms through time and the presence of crystals with sieve-textured rims (generally acknowledged to be generated by mixing) have suggested the occurrence of other differentiation processes. Magma recharge and mixing by more basic magma have been therefore simulated to verify the consistency of whole rock compositions.

Modeling has been performed through mass balance calculations involving three magma compositions: (1) residing magma; (2) recharging magma; and (3) magma resulting from mixing between these two end-members (ESM 4). Composition of the residing magma is not constant and corresponds to that resulting from the third step of MELTS simulations at various degrees of fractionation (ESM 4). The recharging end-member is fixed as the starting composition used for the third step of MELTS simulation, i.e., a basalt already fractionated in the deep magma reservoir at 21 km of depth and during its ascent toward the shallow reservoir at 3 km of depth (ESM 4). Magma recharge and mixing are superimposed to crystal fractionation occurring in the shallow magma reservoir. For this reason, a mixed magma variable in composition is produced for each degree of fractionation. Whole rock compositions can be justified by assuming the progressive replenishment by 45 vol.% of recharging magma at each step of the modeling (Fig. 11).