TY - JOUR
AU1 - Jeffery, A, J
AU2 - Gertisser,, R
AU3 - Self,, S
AU4 - Pimentel,, A
AU5 - O’Driscoll,, B
AU6 - Pacheco, J, M
AB - Abstract The recent (< 100 ka) volcanic stratigraphy of Terceira, Azores, includes at least seven peralkaline trachytic ignimbrite formations, attesting to a history of explosive eruptions. In this study, the petrogenesis and pre-eruptive storage conditions of the ignimbrite-forming magmas are investigated via whole-rock major and trace element geochemistry, melt inclusion and groundmass glass major element and volatile compositions, mineral chemistry, thermobarometric models, and petrogenetic modelling. Our primary aims are to develop a model for the magmatic plumbing system from which the ignimbrite-forming trachytes of Terceira were produced by evaluating various petrogenetic processes and constraining pre-eruptive magma storage conditions. We also place the ignimbrite-forming magmas into the context of the Terceira suite and discuss the potential implications of pre-eruptive magma conditions for eruptive behaviour. Results indicate that ignimbrite-forming, comenditic trachytes are generated predominantly by extended fractional crystallization of basaltic parental magmas at redox conditions around 1 log unit below the fayalite–magnetite–quartz buffer. This is achieved via a polybaric fractionation pathway, in which mantle-derived basalts stall and fractionate to hawaiitic compositions at lower crustal depths (∼15 km), before ascending to a shallow crustal magma storage zone (∼2–4 km) and fractionating towards comenditic trachytic compositions. The most evolved pantelleritic magmas of Terceira (not represented by the ignimbrites) are plausibly generated by continued fractionation from the comenditic trachytes. Syenite autoliths represent portions of peralkaline trachytic melt that crystallized in situ at the margins of a silicic reservoir. Trachytic enclaves hosted within syenitic autoliths provide direct evidence for a two-stage mingling process, in which ascending hawaiites are mixed with trachytic magmas in the shallow crustal magma storage zone. The resulting hybridized trachytes then ascend further and mix with the more evolved peralkaline trachytes in the uppermost eruptible cap of the system, passing first through a syenitic crystal mush. The reduced viscosities of the peralkaline silicic magmas of this study in comparison with their metaluminous counterparts facilitate rapid crystal–melt segregation via crystal settling, generating compositionally zoned magma bodies and, in some instances, relatively crystal-poor erupted magmas. Reduced viscosity may also inhibit highly explosive activity (e.g. formation of a sustained eruption column), and limit the majority of explosive eruptions to low pyroclastic fountaining or ‘boil-over’ eruption styles. The formation of intermediate composition magmas within the system is considered to be limited to episodic mixing between mafic and silicic magmas. INTRODUCTION Terceira, one of the nine islands of the Azores archipelago, exhibits a number of petrological features that are atypical of oceanic island silicic centres. In contrast to the frequently alkali basalt-dominated volcanism of oceanic islands, a significant proportion (86 vol. %; Self, 1976) of recently (< 20–23 ka) erupted products on Terceira have been silicic and peralkaline, extending to pantelleritic compositions. Furthermore, in addition to recent silicic lava domes and coulées, the volcanic stratigraphy of Terceira includes at least seven ignimbrite-bearing pyroclastic formations, some of which exhibit variably welded units, basal pumice falls, and dilute pyroclastic density current (surge) deposits (Gertisser et al., 2010), attesting to a spasmodic history of explosive eruptions of silicic magmas (Self, 1974, 1976; Gertisser et al., 2010). Although ignimbrite-forming eruptions have occurred on other Azorean islands, such as Faial (Pacheco, 2001; Pimentel et al., 2015), São Miguel (Duncan et al., 1999; Gaspar et al., 2015), and Graciosa (Gaspar, 1996), they represent a relatively minor portion of each island’s eruptive history. Such phenomena have also been reported at different locations in contrasting geodynamic settings worldwide (e.g. Pantelleria, Mahood & Hildreth, 1986; Gran Canaria, Araña et al., 1973; Ascension, Daly, 1925; Socorro, Bryan, 1966). Studies of peralkaline magmatic systems have highlighted their complexity, revealing the interplay of petrogenetic processes such as fractional crystallization, crustal assimilation, magma mixing, and remobilization and/or partial melting of cumulate material (e.g. Roux & Varet, 1975; Harris, 1983; Mahood, 1984; Davies & Macdonald, 1987; Macdonald, 1987; McBirney, 1993; Mungall & Martin, 1995; Black et al., 1997; Bohrson & Reid, 1997; Scaillet & Macdonald, 2001; Macdonald & Scaillet, 2006; Ren et al., 2006; White et al., 2006, 2009; Macdonald et al., 2008; Markl et al., 2010; Hong et al., 2013; Shao et al., 2015; Jeffery et al., 2016a, b). Considering the further complexity introduced by P–T–fO2 conditions and compositional variability, such systems are individually unique, to some extent (Macdonald, 2012). However, many peralkaline complexes are, in some respects, unified by the frequent occurrence of compositional zonation of their magma reservoirs (e.g. Civetta et al., 1984; Mahood 1984; Mahood & Hildreth, 1986; Macdonald et al., 1994; Troll & Schmincke, 2002; Peccerillo et al., 2003; Sumner & Wolff, 2003; Macdonald, 2012). Mungall & Martin (1995) proposed that the most recently extruded (< 20–23 ka) peralkaline silicic magmas on Terceira could be generated via extended fractional crystallization of an alkali basalt parental magma composition. In this study, we apply whole-rock major and trace element geochemistry, melt inclusion and groundmass glass major and volatile element analyses, mineral chemistry, thermobarometry, and petrogenetic modelling to the ignimbrites of Terceira, erupted between ∼86 and ∼20–23 ka (Gertisser et al., 2010), and a suite of associated syenite autoliths, aiming to (1) elucidate the petrogenesis of the peralkaline, ignimbrite-forming silicic magmas of Terceira, (2) constrain the pre-eruptive magma storage conditions, (3) place the ignimbrite-forming magmas of Terceira within the context of the identified magma series of Mungall & Martin (1995), as well as the overall magmatic trend of the island, and (4) evaluate the effects of the pre-eruptive magma system on eruption dynamics. GEOLOGICAL BACKGROUND The Azores archipelago comprises nine islands in the North Atlantic Ocean (Lat. 37°N to 40°N, Long. 25°W to 32°W), ∼1300 km west of the Portuguese mainland. The islands themselves are divided into three geographical groups (western, central and eastern), and represent the subaerial expression of the Azores Plateau, a triangular-shaped bathymetric and gravity anomaly reflecting a morphologically complex area (∼5·8 × 106 km2) of elevated oceanic crust that formed between 20 and 7 Ma (Kaula, 1970; Searle, 1980; Lourenço et al., 1998; Gente et al., 2003). The unique geodynamic setting of the Azores results from the triple junction of the North American, Eurasian, and African lithospheric plates. This area of the North Atlantic is marked by three major tectonic features: the Mid-Atlantic Ridge (MAR), the East Azores Fracture Zone (EAFZ) and the Terceira Rift (e.g. Krause & Watkins, 1970; Ridley et al., 1974; Vogt & Jung, 2004; Luis & Miranda, 2008; Madeira et al., 2015; Miranda et al., 2015 and references therein; Weiß et al., 2015). The MAR delimits the Eurasian and African plates to the east from the North American plate to the west. The EAFZ, located to the south of the archipelago, corresponds to an abandoned fault system that probably represents the ancient boundary between the Eurasian and African plates, which extends eastward as the Azores–Gibraltar Fracture Zone (Ridley et al., 1974; Luis et al., 1994; Silveira et al., 2006; Madeira et al., 2015). The Terceira Rift runs for ∼600 km in an oblique direction between the MAR in the NW and the EAFZ in the SE. The rift corresponds to the westernmost segment of Eurasian–African plate boundary and is characterized by a complex alignment of alternating basins and volcanic edifices, including seamounts and the islands of Graciosa, Terceira, and São Miguel. It is considered one of the world’s slowest spreading centres, with a spreading rate of 2–4 mm a−1 (Ridley et al., 1974; Searle, 1980; Madeira & Brum da Silveira, 2003; Vogt & Jung, 2004; Fernandes et al., 2006; Madeira et al., 2015; Marques et al., 2015). All of the Azorean islands are volcanic in origin and magmatism in the area is widely believed to result from the complex interaction between the MAR and a melting anomaly, often referred to as the Azores mantle plume, although the precise nature of the anomaly remains a matter of some debate (e.g. White et al., 1976; Schilling, 1991; Widom & Shirey, 1996; Courtillot et al., 2003; Beier et al., 2012; Métrich et al., 2014). Terceira Island belongs to the central group of the Azores and is the third largest in the archipelago, with an area of ∼400 km2 (Fig. 1). The island comprises four central volcanoes (Cinco Picos, Guilherme Moniz, Pico Alto and Santa Bárbara) that sit astride a 2 km wide basaltic fissure zone that bisects the island from NW to SE (Self, 1974, 1976). The oldest volcanic centre, Cinco Picos (also known as Serra do Cume-Ribeirinha; e.g. Pimentel et al., 2016) (> 401 ka; Hildenbrand et al., 2014), comprises a heavily eroded, 9 × 7 km caldera that dominates the SE sector of the island, with volcanism showing a compositional range that extends from basalt to peralkaline trachyte (Self, 1974; Self & Gunn, 1976). Guilherme Moniz volcano (> 270 ka in age; Calvert et al., 2006) is located slightly south of the centre of the island and, like Cinco Picos, comprises a 4 × 2 km caldera, with basaltic rocks in its floor and peralkaline trachytic rocks exposed in the caldera walls (e.g. Self, 1974; Self & Gunn, 1976). Fig. 1. Open in new tabDownload slide (a) Map showing the Central and Eastern Groups of the Azores archipelago in relation to key structural features. Inset: map highlighting the location of the Azores archipelago in the North Atlantic Ocean. (b) Map of Terceira Island showing the volcanic centres and major infrastructure. Contours (20 m intervals) generated using GeoMapApp©. The axis of the fissure zone that bisects the island from NW to SE is shown as a dashed line. Fig. 1. Open in new tabDownload slide (a) Map showing the Central and Eastern Groups of the Azores archipelago in relation to key structural features. Inset: map highlighting the location of the Azores archipelago in the North Atlantic Ocean. (b) Map of Terceira Island showing the volcanic centres and major infrastructure. Contours (20 m intervals) generated using GeoMapApp©. The axis of the fissure zone that bisects the island from NW to SE is shown as a dashed line. Pico Alto (for which the oldest available age is 141 ka; see Gertisser et al., 2010) lies on the northern flank of Guilherme Moniz, and has been considered by Calvert et al. (2006) to represent the younger portion of the same volcanic centre. Unlike the other volcanic centres, Pico Alto lacks a well-defined morphological structure, and comprises an assemblage of comenditic and pantelleritic lava domes and coulées partially filling and overflowing a caldera (Self, 1974; Pimentel, 2006; Gertisser et al., 2010). The eruptive history of Pico Alto shows evidence of explosive eruptions, recorded by major pyroclastic formations dominated by ignimbrites, erupted between ∼86 and ∼20–23 ka (Gertisser et al., 2010). The youngest volcano, Santa Bárbara (> 65 ka; Hildenbrand et al., 2014), has a distinctive conical shape, rising to 1021 m above sea level, truncated by two small nested calderas. Compositionally, this landform is made up of recent (< 20–23 ka) peralkaline silicic lava domes and coulées, and pumice falls, which overlie mafic rocks including hawaiites and mugearites (Self, 1974, 1976; Self & Gunn, 1976). The fissure zone (> 43 ka; Calvert et al., 2006) that bisects the island is defined by alignments of scoria cones, spatter cones, lava flows, and collapse pits (Self, 1976; Mungall & Martin, 1995; Zanon & Pimentel, 2015). It shows a general progression towards younger ages from SE to NW, and traverses the extinct volcanic centres Guilherme Moniz and Cinco Picos, covering the floor of both calderas with young basalts and hawaiitites. Three historical eruptions occurred along the fissure zone, in the centre of the island as well as offshore (Zbyszewski, 1966; Gaspar et al., 2003; Pimentel et al., 2016). The fissure zone is considered to represent the surface expression of the Terceira Rift (Self, 1974). IGNIMBRITE STRATIGRAPHY Ignimbrite-forming eruptions on Terceira appear to have been limited to periodic, short-lived eruptive episodes that each led to multiple depositional units. These are interspersed with longer periods of quiescence or eruption of various pyroclastic deposits and lava flows. The stratigraphy and chronology of ignimbrites on Terceira was established by Gertisser et al. (2010) (Fig. 2), who identified seven distinct pyroclastic formations containing ignimbrites or composed of ignimbrites based upon field characteristics, stratigraphical relationships, 14C and 40Ar/39Ar chronology, as well as major and trace element geochemistry. Each formation is bounded by unconformities and records an eruptive event or, more often, a sequence of eruptions closely spaced in time, and was named following the original scheme of Self (1974, 1976), where possible. The most likely source of the ignimbrites was identified by Gertisser et al. (2010) to be Pico Alto, and possibly Guilherme Moniz in the case of the older ignimbrite formations. The same researchers also identified two further pyroclastic density current deposits (the Quatro Ribeiras pyroclastic flow deposit and the Posto Santo spatter flow deposit), which are not considered further in this study owing to their comparatively isolated occurrence. Fig. 2. Open in new tabDownload slide Summarized ignimbrite stratigraphy and ignimbrite distribution maps for Terceira, after Gertisser et al. (2010). White diamonds indicate sampling locations for each ignimbrite formation. Full details on each location, including field photographs, have been given by Jeffery (2016). Unnamed portions of the stratigraphy comprise various basaltic to trachytic or rhyolitic lava flows and pyroclastic deposits. Fig. 2. Open in new tabDownload slide Summarized ignimbrite stratigraphy and ignimbrite distribution maps for Terceira, after Gertisser et al. (2010). White diamonds indicate sampling locations for each ignimbrite formation. Full details on each location, including field photographs, have been given by Jeffery (2016). Unnamed portions of the stratigraphy comprise various basaltic to trachytic or rhyolitic lava flows and pyroclastic deposits. The stratigraphy of the island has been studied in depth by Self (1974, 1976) and Gertisser et al. (2010), with a number of ages reported by the latter. The stratigraphy is divided into the Upper Terceira Group (UTG) and the Lower Terceira Group (LTG), each comprising basaltic to trachytic and rhyolitic lava flows, pumice and scoria falls, and ignimbrites (Fig. 2). The base of the UTG is marked by the youngest and most extensive pyroclastic formation, the Lajes–Angra Ignimbrite Formation (LAI). At least 116 separate eruptions of Santa Bárbara and Pico Alto, alongside fissure zone activity, are recorded by the pumice falls, scoria falls and lava flows, lava domes and coulées of the UTG, overlying the LAI. The LAI itself comprises two distinct members; the Angra Ignimbrite (exposed on the southern coast) and the Lajes Ignimbrite (exposed on both the northern and southern coasts), and is dated between 20 and 23 ka (Gertisser et al., 2010). The Lajes member (20 110 ± 470 to 23 150 ± 730 uncalibrated 14C years bp) is a relatively thin ignimbrite (3·5 m on average) with a welded lower part and a non-welded upper ignimbrite unit, whereas the Angra member (21 220 ± 120 to 22 310 ± 800 uncalibrated 14C years bp) is a thicker (up to 14 m) and remarkably monotonous, almost totally non-welded ignimbrite. Stratigraphically below the LAI, interstratified pumice falls, lava flows and at least six other ignimbrite formations and two other pyroclastic density current deposits combine to form the Lower Terceira Group (LTG) (Gertisser et al., 2010). These include the Linhares–Matela Ignimbrite Formation (LMI), the Vila Nova–Fanal Ignimbrite Formation (VFI), the Calderia–Castelinho Ignimbrite Formation (CCI), the Pedras Negras Ignimbrite Formation (PNI), the Grota do Vale Ignimbrite Formation (GVI) and the Ignimbrite-i Formation (Ign-i). The LMI is the uppermost ignimbrite formation in the LTG, comprising the Linhares and Matela members, which appear to be limited to the south of the island. The LMI is separated from the overlying LAI by a lava flow and c. 10 m of pyroclastic fall deposits, and is 14C dated at 34 690 ± 7500 to 37 320 ± 4960 uncalibrated years bp (Gertisser et al., 2010). The VFI is made up of multiple pyroclastic density current units and associated pumice fall units, and is divided into two members: the Vila Nova member, exposed on the northern coast, and the Fanal member, exposed on the southern coast. They are 40Ar/39Ar dated at 50 ± 10 and 58 ± 20 ka, respectively (Gertisser et al., 2010). The CCI also comprises two members: the Caldeira member in the north and the Castelinho member in the south, both of which are stratigraphically below the VFI. Of all the ignimbrites on Terceira, it is the only one to exhibit a well-developed fall unit and overlying dilute pyroclastic density current deposit, featuring cross-bedding of both fine- and coarse-grained pumice beds. Although small fine-grained syenitic clasts may be found in at least one of the other ignimbrite formations (LAI), the CCI is characterized by abundant coarse-grained, syenitic autoliths (termed cognate xenoliths by Gertisser et al., 2010), which can reach sizes of 25 cm in diameter. Gertisser et al. (2010) provided two 40Ar/39Ar ages of 71 ± 4  and 83 ± 18 ka for the CCI, showing that the CCI is considerably older than the overlying VFI. The PNI is found stratigraphically below the CCI and exposed exclusively on the northern coast, typically as dark-weathering, heavily eroded remnants with a welded basal layer. Ignimbrite-i is found only as a small outcrop on the north coast, where it is welded and positioned stratigraphically below the CCI. Gertisser et al. (2010) reported a single 40Ar/39Ar age of 86 ± 9 ka for Ignimbrite-i, although its relationship to the PNI, exposed further westward along the northern coast, remains unclear. The GVI (Gertisser et al., 2010) is the lowest ignimbrite on the south coast and, owing to the unusual occurrence of biotite, does not correlate with either of the lowest ignimbrite formations cropping out along the north coast (PNI and Ign-i). The exposure is only ∼1·5 m thick, but the top is eroded, so the original thickness is unknown. PETROGRAPHY Ignimbrites Juvenile clasts sampled from the ignimbrites of Terceira range from pumice to dense vitrophyres and exhibit similar mineral assemblages, characterized by the presence of alkali feldspar (anorthoclase), augite, Ti-magnetite, and apatite ± olivine and ilmenite (Self, 1974; Gertisser et al., 2010). The GVI is a notable exception to this, in that it also contains phenocrysts of biotite, a mineral that is not observed in any of the other ignimbrite formations (Gertisser et al., 2010). Additionally, the LMI is distinguished by the occurrence of plagioclase phenocrysts in addition to alkali feldspar. In all of the ignimbrites anorthoclase is the dominant phase, although it typically does not exceed ∼10 vol. % on a vesicle-free basis. Anorthoclase phenocrysts are generally unzoned, tabular, and up to ∼4 mm in length (Fig. 3a). However, crystal fragments and heavily resorbed and embayed examples are also common, the latter being especially abundant in the LMI (Fig. 3b). Augite is generally restricted to subhedral microphenocrysts up to 0·2 mm in length, although comparatively large, euhedral phenocrysts (up to ∼3 mm) are occasionally found (Fig. 3c). Augite frequently displays a spatial association with Fe–Ti oxides; the latter are often partially or entirely included within augite crystals (Fig. 3d). Olivine is generally found as equant phenocrysts that do not exceed 2 mm and are frequently resorbed and embayed (Fig. 3e). Biotite in the GVI occurs as small, euhedral phenocrysts that generally do not exceed ∼1 mm, and often contain inclusions of apatite (Fig. 3f). Ti-magnetite and ilmenite exhibit equant, subhedral forms that rarely exceed 1 mm in size. Some examples display optically visible zonation patterns in reflected light, with prominent, irregular cores, and rims of highly variable thickness and brighter reflectance (Fig. 3g). Rare examples of Ti-magnetite exhibit regular exsolution lamellae of ilmenite. Apatite is present in trace amounts throughout, and is generally restricted to small, acicular inclusions within other phases. Fig. 3. Open in new tabDownload slide Representative photomicrographs of mineral phases from ignimbrites of Terceira. (a) Large, euhedral alkali feldspar crystal surrounded by vesicular glass (LAI). (b) Highly resorbed and embayed alkali feldspar crystal surrounded by vesicular glass (LMI). (c) Large euhedral diopside crystal surrounded by vesicular glass (PNI). (d) Small augite crystal with associated Fe–Ti oxide crystals (LAI). (e) Strongly resorbed and embayed olivine crystal (LAI). (f) Euhedral biotite crystal with small acicular apatite inclusions (GVI). (g) Small zoned Fe–Ti oxide crystals (Ign-i). (h) Alkali feldspar-dominated syenite bleb surrounded by vesicular glass (LAI). An alkali feldspar phenocryst is visible in the top left of the image. Fig. 3. Open in new tabDownload slide Representative photomicrographs of mineral phases from ignimbrites of Terceira. (a) Large, euhedral alkali feldspar crystal surrounded by vesicular glass (LAI). (b) Highly resorbed and embayed alkali feldspar crystal surrounded by vesicular glass (LMI). (c) Large euhedral diopside crystal surrounded by vesicular glass (PNI). (d) Small augite crystal with associated Fe–Ti oxide crystals (LAI). (e) Strongly resorbed and embayed olivine crystal (LAI). (f) Euhedral biotite crystal with small acicular apatite inclusions (GVI). (g) Small zoned Fe–Ti oxide crystals (Ign-i). (h) Alkali feldspar-dominated syenite bleb surrounded by vesicular glass (LAI). An alkali feldspar phenocryst is visible in the top left of the image. Syenitic autoliths The quartz-syenitic autoliths of the CCI exhibit a variety of macroscopic textures, including inter- and intra-autolith grain-size variations (Fig. 4a) and schlieren structures rich in mafic minerals (Fig. 4b), and contain fine-grained, trachytic enclaves, with rounded, lobate forms and chilled margins (Fig. 4c). Individual autoliths can contain up to ∼10 vol. % unfilled intercumulus void space in the freshest samples. Schlieren may anastomose or bifurcate, and are characterized by an abundance of Na-clinopyroxene, Na-amphibole, and aenigmatite, which may be either intercumulus or megacrystic (up to ∼1 cm). Schlieren typically form at the contacts between two texturally or mineralogically distinct varieties of syenite (Fig. 4b). Fig. 4. Open in new tabDownload slide Representative hand specimen photographs of syenitic autoliths from CCI. (a) Coarse- and medium-grained varieties of syenite. (b) Schlieren structures, comprising abundant Na-clinopyroxene, Na-amphibole, and aenigmatite, within syenite autoliths. The surrounding syenite often exhibits contrasting grain sizes or modal mineralogies on either side of the structure. (c) Trachytic enclaves within a large block of syenite. Individual enclaves exhibit rounded, lobate boundaries, with distinctive chilled margins, and contain numerous miarolitic cavities. Fig. 4. Open in new tabDownload slide Representative hand specimen photographs of syenitic autoliths from CCI. (a) Coarse- and medium-grained varieties of syenite. (b) Schlieren structures, comprising abundant Na-clinopyroxene, Na-amphibole, and aenigmatite, within syenite autoliths. The surrounding syenite often exhibits contrasting grain sizes or modal mineralogies on either side of the structure. (c) Trachytic enclaves within a large block of syenite. Individual enclaves exhibit rounded, lobate boundaries, with distinctive chilled margins, and contain numerous miarolitic cavities. The autoliths are characterized by more complex mineral assemblages than those of the various Terceira ignimbrite formations, comprising alkali feldspar (anorthoclase, sanidine, albite), Na-clinopyroxene, Na-amphibole, aenigmatite, Ti-magnetite, ilmenite, quartz, olivine, apatite, and biotite, in approximate decreasing order of abundance, with dalyite and eudialyte representing the most significant zirconosilicate accessory phases (Jeffery et al., 2016a). This mineral assemblage contains phases that are typical of both miaskitic and agpaitic rocks (see Marks et al., 2011), suggesting that the syenites should be considered transitional between the two. Alkali feldspar is the most abundant phase, constituting ∼75 vol. % of each autolith (including void space), and forming a cumulus framework, regardless of grain size. Individual crystals range from large, tabular crystals (up to ∼10 mm; Fig. 5a) to smaller laths (up to ∼2 mm; Fig. 5b), and from fresh and unaltered to heavily altered and perthitic. Alkali feldspar is also present as small, irregular crystals that, together with quartz, form granophyric patches. Fig. 5. Open in new tabDownload slide Representative photomicrographs of syenitic autoliths and syenite-hosted enclaves. Di, diopside; Aeg, aegirine; Agt, aegirine–augite; Na–Ca Amp, Na–Ca-amphibole; Na-Amp, Na-amphibole; Aen, aenigmatite; Afs, alkali feldspar; Mg-Ol, Mg-rich olivine. (a) Coarse-grained syenite comprising large interlocking alkali feldspar crystals. A small dalyite crystal and acicular aegirine crystals are visible, partially filling an interstitial void space in the lower right. (b) Medium-grained syenite comprising predominantly alkali feldspar laths with small intercumulus Na-amphiboles and aegirine. (c) Large irregularly zoned aegirine–augite crystal occupying an intercumulus void space, surrounded by alkali feldspar. A highly irregular core of Na-amphibole is visible within the aegirine–augite crystal. (d) An intercumulus amphibole crystal surrounded by alkali feldspar. The amphibole exhibits an optically distinguishable compositional transition from Na–Ca amphibole (centre), to Na amphibole (darker rims), and finally aegirine (upper and lower margins). (e) Complex intercumulus void, surrounded by alkali feldspar (lower right), filled with quartz (upper central image), aenigmatite and aegirine. The aegirine comprises a mush of acicular crystals, which appear to replace a large, irregular aenigmatite crystal. (f) Syenite-hosted enclave comprising large diopside and Na–Ca amphibole crystals set in a microcrystalline groundmass of alkali feldspar, aegirine–augite, Na–Ca amphibole, and Fe–Ti oxides. (g) Characteristically large alkali feldspar crystal from a syenite-hosted enclave. A pronounced, inclusion-rich rim is visible at its margins. (h) Mg-rich olivine double breakdown texture from a syenite-hosted enclave. The outer rim comprises a hydrous iddingsitic assemblage, whereas the interior represents an anhydrous breakdown assemblage that includes Fe–Ti oxides. Fig. 5. Open in new tabDownload slide Representative photomicrographs of syenitic autoliths and syenite-hosted enclaves. Di, diopside; Aeg, aegirine; Agt, aegirine–augite; Na–Ca Amp, Na–Ca-amphibole; Na-Amp, Na-amphibole; Aen, aenigmatite; Afs, alkali feldspar; Mg-Ol, Mg-rich olivine. (a) Coarse-grained syenite comprising large interlocking alkali feldspar crystals. A small dalyite crystal and acicular aegirine crystals are visible, partially filling an interstitial void space in the lower right. (b) Medium-grained syenite comprising predominantly alkali feldspar laths with small intercumulus Na-amphiboles and aegirine. (c) Large irregularly zoned aegirine–augite crystal occupying an intercumulus void space, surrounded by alkali feldspar. A highly irregular core of Na-amphibole is visible within the aegirine–augite crystal. (d) An intercumulus amphibole crystal surrounded by alkali feldspar. The amphibole exhibits an optically distinguishable compositional transition from Na–Ca amphibole (centre), to Na amphibole (darker rims), and finally aegirine (upper and lower margins). (e) Complex intercumulus void, surrounded by alkali feldspar (lower right), filled with quartz (upper central image), aenigmatite and aegirine. The aegirine comprises a mush of acicular crystals, which appear to replace a large, irregular aenigmatite crystal. (f) Syenite-hosted enclave comprising large diopside and Na–Ca amphibole crystals set in a microcrystalline groundmass of alkali feldspar, aegirine–augite, Na–Ca amphibole, and Fe–Ti oxides. (g) Characteristically large alkali feldspar crystal from a syenite-hosted enclave. A pronounced, inclusion-rich rim is visible at its margins. (h) Mg-rich olivine double breakdown texture from a syenite-hosted enclave. The outer rim comprises a hydrous iddingsitic assemblage, whereas the interior represents an anhydrous breakdown assemblage that includes Fe–Ti oxides. Na-clinopyroxene, Na-amphibole, and aenigmatite represent the dominant intercumulus phases, with a cumulative volume of up to ∼10 vol. %. All three phases are concentrated in schlieren, where their abundance may be as high as ∼50 vol. %, but are also present in subordinate quantities throughout the syenite. Within schlieren structures, Na-clinopyroxene is present as patches of acicular crystals that partially replace large aenigmatite crystals, of which only relict crystals with ragged edges remain. This relationship between Na-clinopyroxene and aenigmatite is not limited to schlieren structures and may be found throughout the syenite autoliths. Outside the previously described schlieren, Na-clinopyroxene exists as intercumulus crystals that can reach sizes of up to ∼3 mm, and often exhibits irregular or patchy zonation (Fig. 5c). Furthermore, Na-clinopyroxenes frequently show a spatial association with Na-amphiboles, appearing to have nucleated heterogeneously on, or to have replaced, pre-existing amphibole crystals (Fig. 5c and d). Similarly, amphiboles reach sizes of ∼4 mm and are frequently zoned, with a brownish amphibole generally making up the central portion of a given crystal, and blue amphibole forming the crystal margins (Fig. 5d). The margins between zones are almost exclusively irregular and gradational. Na-clinopyroxene is also present as small acicular crystals that form radiating bundles, typically projecting into unfilled cavities. Quartz is also limited to intercumulus pore spaces, where it occurs as aggregates of rounded crystals up to ∼1 mm in size. Together with phases such as acicular Na-clinopyroxene, dalyite, or eudialyte, quartz aggregates may either partially or entirely fill pores. Fe–Ti oxides are present as small (< 100 μm) equant crystals that are frequently included within other phases and do not account for more than 1 vol. % of the rock. More rarely, Fe–Ti oxides reach sizes of up to ∼400 μm and may represent an intercumulus phase rather than an inclusion. Olivine is uncommon in the syenites and, when present, exists as anhedral relict crystals that exhibit a complex reaction texture. Reaction rims are typically characterized by an inner, anhydrous zone including Fe–Ti oxides and an outer hydrous rim of iddingsite and comparatively rare biotite. Apatite is present in trace amounts, and is limited to small (< 100 μm) acicular inclusions within other phases. Biotite is uncommon and, where present, exists as small inclusions within alkali feldspars. Eudialyte is found as irregularly shaped crystals (generally < 1 mm) that partially or entirely fill intercumulus spaces, and is frequently associated spatially with Na-clinopyroxene. Examples of irregular, patchy or oscillatory zoning are common. Dalyite is typically present as small (< 0·5 mm) sub- to anhedral crystals, although it can reach sizes of 1–1·5 mm. It is almost exclusively anhedral and confined to the interstices, either filling or partially filling void spaces. It is often associated spatially with quartz, and in some cases can be found as inclusions within larger interstitial quartz crystals (Jeffery et al., 2016a). Syenite-hosted enclaves Dark enclaves found within individual syenite autoliths from the CCI are porphyritic, with large plagioclase, alkali feldspar, diopside, augite, and Mg-rich olivine phenocrysts up to ∼8 mm in length set in a fine-grained (< 0·2 mm) microcrystalline groundmass (Fig. 5f). Groundmass mineral assemblages include alkali feldspar (anorthoclase and albite), amphibole, diopside to aegirine–augite, Fe–Ti oxides, apatite, eudialyte, dalyite, aenigmatite and titanite, in approximate order of decreasing abundance. As in all of the rocks in this study, alkali feldspar is the dominant phase, occurring as large phenocrysts and as small (< 0·2 mm), anhedral groundmass crystals. Phenocrystic alkali feldspar is characterized by rounded cores, which are mantled by sieve-textured rims of variable width (∼50–750 μm) (Fig. 5g). In contrast to the interior (core–rim) boundary, which is frequently sharp, the exterior boundary between rim and groundmass is frequently diffuse and poorly defined. Phenocryst cores often exhibit patchy or, more rarely, oscillatory zoning patterns. Na–Ca amphibole is present both as a minor population of microphenocrysts, typically up to ∼500 μm in length, and as an abundant groundmass phase (< 150 μm). Augite and Na-clinopyroxene are found in abundance in the groundmass, but also exist as phenocrysts and microphenocrysts (∼250–1000 μm), which frequently have thin (< 50 μm) rims of iddingsite and exhibit resorption textures. Larger examples may also have concentric, oscillatory zoning patterns. Fe–Ti oxides are present as small groundmass crystals that do not exceed 150 μm. Olivine is found exclusively as ragged phenocrysts up to ∼3 mm in size, and surrounded by a distinctive double rim (Fig. 5h). Apatite is limited to small (< 100 μm), acicular crystals in the groundmass and included within other mineral phases. Eudialyte and dalyite are both present in trace amounts and are limited to miarolitic cavities, where they range from anhedral to euhedral morphologies, and reach sizes of up to ∼500 μm. Unlike in the host syenites, aenigmatite is rarely found in the enclaves and is restricted to miarolitic cavities, where it is predominantly occurs as small, irregular patches within clusters of acicular clinopyroxene. Plagioclase is uncommon, and can be found as crystals up to ∼8 mm in size. ANALYTICAL METHODS Whole-rock geochemistry Whole-rock major and trace element analyses were obtained at the Bureau Veritas Mineral Laboratories, Canada, using inductively coupled plasma atomic emission spectroscopy (ICP-AES) and inductively coupled plasma mass spectrometry (ICP-MS), respectively. Additional analyses were made by X-ray fluorescence spectrometry (XRF) using a Bruker AXS S4 Pioneer system at the University of East Anglia, UK. All samples were cleaned to remove altered surfaces and crushed in an agate mill prior to drying at 60°C. Loss on ignition (LOI) was reported as weight difference after ignition for 2 h at 1000°C. Samples analysed by ICP-AES and ICP-MS at Bureau Veritas Mineral Laboratories, Canada, were prepared with a LiBO2–Li2B4O7 flux and dilute nitric digestion. The instrument was calibrated using up to 12 international standards (AGV-1, BCR-2, BHVO-1, BHVO-2, BIR-1, RGM-1, WS-E, JB2, JB3, SO-18, DS9, OREAS45EA). The mean deviation from the accepted standard values was < 2% for major elements and < 3% for trace elements (Supplementary Data Electronic Appendix 3; supplementary data are available for downloading at http://www.petrology.oxfordjournals.org). For XRF analyses, fused glass discs for major element analysis were prepared using 0·7 g of rock powder mixed with 3·5 g of lithium metaborate. Trace element concentrations were determined using PVC-bound pressed powder pellets. For major elements, the instrument was calibrated using the following international standards: BCR-2, DTS-1, DTS-2, G2, GXR-1, GXR-2, GXR-3, BHVO-2, BCS-368, BCS-376, AC-E, BE-N, BX-N, GS-N, UB-N, LKSD-3, MRG-1, STSD-1, SARM-2. For trace elements, the Geoquant calibration of Bruker was applied. Data quality was evaluated using the following secondary standards: WS-E, OUG94, GSP-2, W2a, AC-E, BHVO-1, QLO-1, DNC-1, W-2, AGV-2, BCR-2, SDO-1, Mess-2, STSD-2. The mean deviation from the accepted standard values was < 5% for major elements, and typically < 10% for trace elements. Mineral and glass analyses Major element compositions of mineral phases and glass (both groundmass glass and melt inclusions) were analysed using a CAMECA SX 100 electron microprobe at The Open University, UK, a JEOL JXA 8900 RL electron microprobe at the University of Göttingen, Germany, and a CAMECA SX 100 electron microprobe at the University of Manchester. For mineral phases, peak counting times per element were 10–30 s using a 5–10 μm defocused beam, an acceleration voltage of 20 kV and a beam current of 20–27 nA. Major elements and volatiles (Cl, F, S) in groundmass glass and melt inclusions were analysed using peak counting times ranging from 90 to 120 s for volatiles and from 10 to 30 s for major elements, using a 10–20 μm defocused beam, an acceleration voltage of 15–20 kV and a beam current of 10–15 nA. To minimize Na loss, Na was always analysed first, with a peak count time of 10 s. Detection limits for Cl and F were 60 and 220 ppm, respectively. Detection limits for S were 300 ppm. The following natural minerals and synthetic materials (denoted as chemical formulae) were used as primary standards: olivine, albite, sanidine, TiO2, haematite, anorthite, wollastonite, Cr2O3, rhodonite, celsian, ZrSiO4 and HfSiO4. Mineral and volcanic glass standards (BCR-2G, VG-2, KN-18 and KE-12) were routinely analysed as secondary standards. Repeat analyses of secondary standards indicate accuracy of < 4%, and reproducibility of < 3% (mean standard deviation). Additionally, energy-dispersive spectrometry (EDS) spectra were produced using a Hitachi TM-3000 scanning electron microscope (SEM) equipped with a Bruker Quantax 70 EDS system at Keele University, UK. Fourier transform infrared spectroscopy The water content of alkali feldspar-hosted melt inclusions was determined by Fourier transform infrared (FTIR) spectroscopy using spectra collected with a Thermo Nicolet Nexus system coupled with a Continuμm IR microscope at The Open University, UK. Operation conditions included standard EverGlo mid-infrared source optics, a Ge-on-KBr beamsplitter, and a liquid nitrogen-cooled MCT-A* detector (11 700–750 cm–1). In all of the analyses, CO2 was below the detection limit (∼100 ppm; see Gertisser et al., 2012). The concentration of dissolved water was determined using the height of the total water (H2O + OH–) peak at 3550 cm–1 and the Beer–Lambert law: H2O (wt %)=100(MA/ρdε) where M is the molecular weight of H2O (18·02), A is the height of the absorption peak, ρ is the sample density (g l–1), d is the thickness of the sample (cm), and ε is the molar absorption coefficient (l mol–1 cm–1). The thickness of each sample (± 3 μm) was determined using a Mitutoyo Digimatic Indicator. The density of the trachytic glass at 298 K and 0·1 MPa was estimated to be 2510 g l–1, assuming a nonlinear temperature dependence of melt volume (e.g. Gottsmann & Dingwell, 2002). Because of the variability of the molar absorption coefficient as a function of [(Si/Al)/total cations] in glass (e.g. Mandeville et al., 2002), the approach given by Seaman et al. (2009) was used to calculate a molar absorption coefficient value of 73 for the 3550 cm–1 peak. RESULTS The entire dataset discussed in the following section is provided in two electronic appendices. Whole-rock, melt inclusion, and groundmass glass data are provided in full in Supplementary Data Electronic Appendix 1, and mineral chemical data are given in Supplementary Data Electronic Appendix 2. Details of applied data quality tests are given in Supplementary Data Electronic Appendix 3. Furthermore, a summary providing the major petrographical and geochemical features of the identified lithologies of this study is given in Table 1. To provide context for the geochemical data of this study, a number of additional published datasets are used for comparison, including: (1) whole-rock analyses derived from each of the volcanic centres of Terceira (Pico Alto, Santa Bárbara, Guilherme Moniz, and Cinco Picos) as well as the fissure zone (Self, 1974; Mungall, 1993; Madureira et al., 2011); (2) glass analyses from distal tephras of the youngest ignimbrite formations (LAI, LMI, VFI, and CCI; Tomlinson et al., 2015); (3) a suite of monzonitic and syenitic xenoliths from Santa Bárbara and Pico Alto, respectively (termed here S. Bárbara xenoliths and P. Alto xenoliths; Mungall, 1993); (4) a small number of whole-rock analyses of enclaves found within trachytic lava flows from Pico Alto (termed here P. Alto enclaves; Mungall, 1993); (5) glass analyses for interstitial glass found within syenitic xenoliths from Pico Alto lavas (referred to here as P. Alto xenolith glass; Mungall, 1993). Collectively, these data define the overall geochemical trend of Terceira and facilitate a discussion on the overall position of the ignimbrite-forming magmas within the context of their volcanic centre (Pico Alto or Guilherme Moniz), and also the island as a whole (see below). Table 1: Summary table of the key petrographical and geochemical features of the different lithologies identified in this study . Ignimbrites . Syenitic autoliths . Syenite-hosted enclaves . Texture Porphyritic Cumulate Porphyritic Whole-rock SiO2 62·2–65·9 wt % 64·4–65·6 wt % 61·1–62·2 wt % Peralkalinity index 0·98–1·43 1·08–1·14 0·97–1·01 Mineralogy Afs, Aug, Di, Ti-Mag, Ap ± Ol, Ilm, Bt, Pl Afs, Agt, Aeg, Na-Amp, Na–Ca-Amp, Aen, Ti-Mag, Ilm, Qtz, Ol, Ap, Bt, Dal, Eud Afs, Aug, Di, Agt, Ca-Amp, Na–Ca-Amp, Ti-Mag, Ol, Pl, Ap, Eud, Dal, Aen, Ttn Feldspar compositions Ph: Or1–39, Ab36–81, An0–62 Or17–40, Ab60–83, An0–4 Ph: Or4–63, Ab36–89, An0–12 Gm: Or2–35, Ab65–97, An0–6 Clinopyroxene compositions Ph: Wo39–46, En21–41, Fs16–36 Qd0–96, Aeg4–99, Jd0–7 Ph: Wo42–48, En31–46, Fs8–26 Gm: Qd43–97, Aeg3–57, Jd0–9 . Ignimbrites . Syenitic autoliths . Syenite-hosted enclaves . Texture Porphyritic Cumulate Porphyritic Whole-rock SiO2 62·2–65·9 wt % 64·4–65·6 wt % 61·1–62·2 wt % Peralkalinity index 0·98–1·43 1·08–1·14 0·97–1·01 Mineralogy Afs, Aug, Di, Ti-Mag, Ap ± Ol, Ilm, Bt, Pl Afs, Agt, Aeg, Na-Amp, Na–Ca-Amp, Aen, Ti-Mag, Ilm, Qtz, Ol, Ap, Bt, Dal, Eud Afs, Aug, Di, Agt, Ca-Amp, Na–Ca-Amp, Ti-Mag, Ol, Pl, Ap, Eud, Dal, Aen, Ttn Feldspar compositions Ph: Or1–39, Ab36–81, An0–62 Or17–40, Ab60–83, An0–4 Ph: Or4–63, Ab36–89, An0–12 Gm: Or2–35, Ab65–97, An0–6 Clinopyroxene compositions Ph: Wo39–46, En21–41, Fs16–36 Qd0–96, Aeg4–99, Jd0–7 Ph: Wo42–48, En31–46, Fs8–26 Gm: Qd43–97, Aeg3–57, Jd0–9 Afs, alkali feldspar; Aug, augite; Di, diopside; Ti-Mag, Ti-magnetite; Ap, apatite; Ol, olivine; Ilm, ilmenite; Bt, biotite; Pl, plagioclase; Agt, aegirine–augite; Aeg, aegirine; Na-Amp, Na-amphibole; Na–Ca-Amp, Na–Ca-amphibole; Aen, aenigmatite; Qtz, quartz; Dal, dalyite; Eud, eudialyte; Ca-Amp, Ca-amphibole; Ttn, titanite; Pheno, phenocrysts; Gm, groundmass; Or, orthoclase; Ab, albite; An, anorthite; Wo, wollastonite; En, enstatite; Fs, ferrosilite; Qd, quadrilateral components; Jd, jadeite; Ph, phenocrysts; Gm, groundmass. Table 1: Summary table of the key petrographical and geochemical features of the different lithologies identified in this study . Ignimbrites . Syenitic autoliths . Syenite-hosted enclaves . Texture Porphyritic Cumulate Porphyritic Whole-rock SiO2 62·2–65·9 wt % 64·4–65·6 wt % 61·1–62·2 wt % Peralkalinity index 0·98–1·43 1·08–1·14 0·97–1·01 Mineralogy Afs, Aug, Di, Ti-Mag, Ap ± Ol, Ilm, Bt, Pl Afs, Agt, Aeg, Na-Amp, Na–Ca-Amp, Aen, Ti-Mag, Ilm, Qtz, Ol, Ap, Bt, Dal, Eud Afs, Aug, Di, Agt, Ca-Amp, Na–Ca-Amp, Ti-Mag, Ol, Pl, Ap, Eud, Dal, Aen, Ttn Feldspar compositions Ph: Or1–39, Ab36–81, An0–62 Or17–40, Ab60–83, An0–4 Ph: Or4–63, Ab36–89, An0–12 Gm: Or2–35, Ab65–97, An0–6 Clinopyroxene compositions Ph: Wo39–46, En21–41, Fs16–36 Qd0–96, Aeg4–99, Jd0–7 Ph: Wo42–48, En31–46, Fs8–26 Gm: Qd43–97, Aeg3–57, Jd0–9 . Ignimbrites . Syenitic autoliths . Syenite-hosted enclaves . Texture Porphyritic Cumulate Porphyritic Whole-rock SiO2 62·2–65·9 wt % 64·4–65·6 wt % 61·1–62·2 wt % Peralkalinity index 0·98–1·43 1·08–1·14 0·97–1·01 Mineralogy Afs, Aug, Di, Ti-Mag, Ap ± Ol, Ilm, Bt, Pl Afs, Agt, Aeg, Na-Amp, Na–Ca-Amp, Aen, Ti-Mag, Ilm, Qtz, Ol, Ap, Bt, Dal, Eud Afs, Aug, Di, Agt, Ca-Amp, Na–Ca-Amp, Ti-Mag, Ol, Pl, Ap, Eud, Dal, Aen, Ttn Feldspar compositions Ph: Or1–39, Ab36–81, An0–62 Or17–40, Ab60–83, An0–4 Ph: Or4–63, Ab36–89, An0–12 Gm: Or2–35, Ab65–97, An0–6 Clinopyroxene compositions Ph: Wo39–46, En21–41, Fs16–36 Qd0–96, Aeg4–99, Jd0–7 Ph: Wo42–48, En31–46, Fs8–26 Gm: Qd43–97, Aeg3–57, Jd0–9 Afs, alkali feldspar; Aug, augite; Di, diopside; Ti-Mag, Ti-magnetite; Ap, apatite; Ol, olivine; Ilm, ilmenite; Bt, biotite; Pl, plagioclase; Agt, aegirine–augite; Aeg, aegirine; Na-Amp, Na-amphibole; Na–Ca-Amp, Na–Ca-amphibole; Aen, aenigmatite; Qtz, quartz; Dal, dalyite; Eud, eudialyte; Ca-Amp, Ca-amphibole; Ttn, titanite; Pheno, phenocrysts; Gm, groundmass; Or, orthoclase; Ab, albite; An, anorthite; Wo, wollastonite; En, enstatite; Fs, ferrosilite; Qd, quadrilateral components; Jd, jadeite; Ph, phenocrysts; Gm, groundmass. Whole-rock major element geochemistry Based on the total alkali–silica (TAS) scheme of Le Bas et al. (1986), the whole-rock juvenile samples of the ignimbrite formations are classified as trachyte (Fig. 6a) and exhibit little variation, with SiO2 contents clustering around 65 wt %, Al2O3 contents of ∼15 wt %, total alkali contents of ∼12 wt %, and uniformly low MgO (< 0·5 wt %) (see Gertisser et al., 2010). The samples are almost exclusively peralkaline (PI > 1), with calculated peralkalinity indices [PI = molar (Na2O + K2O)/Al2O3)] that range from 0·98 to 1·43. Syenite autoliths have similar whole-rock compositions to the ignimbrites, with calculated PI between 1·08 and 1·14 (Jeffery et al., 2016a). Enclaves within syenitic autoliths exhibit slightly different whole-rock compositions, with lower SiO2 and total alkali contents of 61–62 wt % and ∼11 wt %, respectively. Additionally, the enclaves lie on the boundary between metaluminous and peralkaline compositions, with PI of 0·97–1·01. On the basis of the Al2O3 versus FeOt classification scheme of Macdonald (1974), all peralkaline samples of this study are comenditic trachyte, with the exception of a basal pumice fall within the VFI, and a single anorthoclase-hosted melt inclusion from the CCI, which are classified as pantelleritic trachyte and pantellerite, respectively (Fig. 6c). Fig. 6. Open in new tabDownload slide Geochemical classification plots for Terceira. Data sourced from Self (1974), Mungall (1993), Gertisser et al. (2010), Madureira et al. (2011), Tomlinson et al. (2015), and Jeffery et al. (2016a). (a) Chemical compositions of various Terceira lithologies plotted on the total alkali–silica (TAS) diagram of Le Bas et al. (1986). Whole-rock, melt inclusion, and groundmass glass data are not distinguished. Errors (2σ) do not exceed symbol size. (b) Enlargement of (a), showing the compositions of the ignimbrite formations, Caldeira-Castelinho Ignimbrite Formation (CCI) syenite autoliths and CCI syenite autolith-hosted enclaves. Symbols are as in (a). Whole-rock data are shown using the same colours and symbols as in (a). The lighter variants of the same colours depict groundmass glass analyses. The darker variants indicate melt inclusion analyses [e.g. the green triangles mark whole-rock analyses of the VFI, as in (a), whereas the light green and dark green triangles reflect groundmass glass and melt inclusion analyses, respectively]. The 2σ error for melt inclusion and groundmass glass data is shown in the top right. Errors for whole-rock do not exceed symbol size. (c) Peralkaline compositions for Terceira plotted on the FeOt vs Al2O3 classification scheme for oversaturated peralkaline rocks (Macdonald, 1974). Symbols are as in (a). Errors (2σ) do not exceed symbol size. Fig. 6. Open in new tabDownload slide Geochemical classification plots for Terceira. Data sourced from Self (1974), Mungall (1993), Gertisser et al. (2010), Madureira et al. (2011), Tomlinson et al. (2015), and Jeffery et al. (2016a). (a) Chemical compositions of various Terceira lithologies plotted on the total alkali–silica (TAS) diagram of Le Bas et al. (1986). Whole-rock, melt inclusion, and groundmass glass data are not distinguished. Errors (2σ) do not exceed symbol size. (b) Enlargement of (a), showing the compositions of the ignimbrite formations, Caldeira-Castelinho Ignimbrite Formation (CCI) syenite autoliths and CCI syenite autolith-hosted enclaves. Symbols are as in (a). Whole-rock data are shown using the same colours and symbols as in (a). The lighter variants of the same colours depict groundmass glass analyses. The darker variants indicate melt inclusion analyses [e.g. the green triangles mark whole-rock analyses of the VFI, as in (a), whereas the light green and dark green triangles reflect groundmass glass and melt inclusion analyses, respectively]. The 2σ error for melt inclusion and groundmass glass data is shown in the top right. Errors for whole-rock do not exceed symbol size. (c) Peralkaline compositions for Terceira plotted on the FeOt vs Al2O3 classification scheme for oversaturated peralkaline rocks (Macdonald, 1974). Symbols are as in (a). Errors (2σ) do not exceed symbol size. In Fig. 7, the major element compositions of this study, alongside available literature data for Terceira, are plotted versus MgO. For clarity, and because of the association of the ignimbrite formations with Pico Alto and Guilherme Moniz, compositions from Santa Bárbara and Cinco Picos are not shown. Initially, SiO2 shows a uniform concentration of ∼47 wt %, until ∼4 wt % MgO, and then increases steadily to 72 wt % SiO2. TiO2 and FeOt both exhibit a downward kink at ∼6 wt % MgO. The alkalis (Na2O and K2O) both increase with decreasing MgO, exhibiting curved profiles. Interstitial glass from syenitic enclaves from Pico Alto lavas analysed by Mungall (1993) indicate late-stage (< 1 wt % MgO) enrichment in Na2O, reaching concentrations of up to ∼12·5 wt %, although this is likely to reflect evolution of intercumulus melt pockets in a manner analogous to post-entrapment crystallization in melt inclusions. By contrast, CaO exhibits a gently curved, concave-downwards trend. Al2O3 contents increase slowly until ∼1 wt % MgO, at which point concentrations fall from ∼18 to ∼5 wt %. MnO shows uniform concentrations of ∼0·2 wt % until, at ∼1 wt % MgO, concentrations increase to ∼0·8 wt %. PI increases gradually until ∼1 wt % MgO, when it sharply increases to values of up to 5, if intercumulus glasses are included. Fig. 7. Open in new tabDownload slide Major and trace element compositional data for Terceira plotted versus MgO. The grey dashed line represents the most successful Rhyolite-MELTS model (polybaric fractional crystallization with a transition from 500 to 150 MPa set to occur at 1100°C, fO2 = FMQ – 1, initial water content = 1·5 wt %). The transition from 500 to 150 MPa is marked with a vertical dashed black line at 2·76 wt % MgO on each plot. The crystallization intervals for each of the predicted mineral phases are marked on the lower-right plot. For clarity, data for Cinco Picos and Santa Bárbara are not shown. All data sourced from Self (1974), Mungall (1993), Gertisser et al. (2010), Madureira et al. (2011), Tomlinson et al. (2015) and Jeffery et al. (2016a). All major element oxides are reported in wt %, trace elements as ppm. Errors (2σ) do not exceed symbol size. Ol, olivine; Cpx, clinopyroxene; Fsp, feldspar; Ti-Mag, Ti-magnetite; Ilm, ilmenite; Ap, apatite. Fig. 7. Open in new tabDownload slide Major and trace element compositional data for Terceira plotted versus MgO. The grey dashed line represents the most successful Rhyolite-MELTS model (polybaric fractional crystallization with a transition from 500 to 150 MPa set to occur at 1100°C, fO2 = FMQ – 1, initial water content = 1·5 wt %). The transition from 500 to 150 MPa is marked with a vertical dashed black line at 2·76 wt % MgO on each plot. The crystallization intervals for each of the predicted mineral phases are marked on the lower-right plot. For clarity, data for Cinco Picos and Santa Bárbara are not shown. All data sourced from Self (1974), Mungall (1993), Gertisser et al. (2010), Madureira et al. (2011), Tomlinson et al. (2015) and Jeffery et al. (2016a). All major element oxides are reported in wt %, trace elements as ppm. Errors (2σ) do not exceed symbol size. Ol, olivine; Cpx, clinopyroxene; Fsp, feldspar; Ti-Mag, Ti-magnetite; Ilm, ilmenite; Ap, apatite. Fig. 7. Open in new tabDownload slide Continued. Fig. 7. Open in new tabDownload slide Continued. Whole-rock trace element geochemistry Selected trace elements are plotted against MgO in Fig. 7, together with previously published data (Self, 1974; Mungall, 1993; Madureira et al., 2011). Overall, trace elements such as Zr, Nb, Rb, and Y behave incompatibly in the mafic and intermediate compositions (> 1 wt % MgO), with a steepening trend at trachytic compositions (< 1 wt % MgO). By contrast, Sr concentrations increase from ∼500 to ∼700 ppm in the mafic and intermediate portion of the trend, whereas the silicic portion of the trend is generally restricted to values below ∼150 ppm. Unlike Sr, Ba shows no clear trend. Instead, considerable scatter is observed throughout the suite, with mafic and intermediate compositions ranging from ∼200 to 1000 ppm and silicic compositions ranging from < 20 to ∼1550 ppm. Chondrite-normalized rare earth element (REE) patterns are shown in Fig. 8a-c and indicate a relatively uniform enrichment of the light REE (LREE) relative to the heavy REE (HREE), with a total range of LaN/YbN ratios between 8·3 and 12·4. All of the samples exhibit variable negative Eu anomalies, with Eu/Eu* = 0·31–0·83 in the ignimbrites, 0·41–0·52 in the syenites, and 0·76–0·86 in the enclaves. A single syenite sample deviates markedly from the other samples, with a significant depletion of middle REE (MREE). Fig. 8. Open in new tabDownload slide Trace element variation diagrams for Terceira lithologies. Chondritic and primitive mantle values taken from Sun & McDonough (1989). (a) Chondrite-normalized REE patterns for the ignimbrite formations. Grey field indicates the range of literature values for the four youngest ignimbrite formations (LAI, LMI, VFI, CCI) from Tomlinson et al. (2015). (b) Chondrite-normalized REE patterns for the CCI syenite autoliths. Data from Jeffery et al. (2016a). Grey field indicates the range of literature values for syenitic xenoliths taken from Mungall (1993). (c) Chondrite-normalized REE patterns for the CCI syenite-hosted enclaves. Grey field indicates range of literature values for the four youngest ignimbrite formations (LAI, LMI, VFI, CCI) from Tomlinson et al. (2015). (d) Multi-element trace element patterns for the ignimbrite formations. (e) Multi-element trace element patterns for the CCI syenite autoliths. Data from Jeffery et al. (2016a). (f) Multi-element trace element patterns for the CCI syenite-hosted enclaves. Fig. 8. Open in new tabDownload slide Trace element variation diagrams for Terceira lithologies. Chondritic and primitive mantle values taken from Sun & McDonough (1989). (a) Chondrite-normalized REE patterns for the ignimbrite formations. Grey field indicates the range of literature values for the four youngest ignimbrite formations (LAI, LMI, VFI, CCI) from Tomlinson et al. (2015). (b) Chondrite-normalized REE patterns for the CCI syenite autoliths. Data from Jeffery et al. (2016a). Grey field indicates the range of literature values for syenitic xenoliths taken from Mungall (1993). (c) Chondrite-normalized REE patterns for the CCI syenite-hosted enclaves. Grey field indicates range of literature values for the four youngest ignimbrite formations (LAI, LMI, VFI, CCI) from Tomlinson et al. (2015). (d) Multi-element trace element patterns for the ignimbrite formations. (e) Multi-element trace element patterns for the CCI syenite autoliths. Data from Jeffery et al. (2016a). (f) Multi-element trace element patterns for the CCI syenite-hosted enclaves. Primitive mantle-normalized multi-element diagrams are given in Fig. 8d-f. The ignimbrites are characterized by pronounced depletions in Ba, Sr, Eu, P, and Ti. A notable exception to this observation is Ign-i, which exhibits a slight enrichment in Ba relative to the other ignimbrites. Syenitic autoliths and enclaves contained therein display a similar geochemical profile to those of the ignimbrites, with the same troughs for Ba, Sr, Eu, P, and Ti. However, in the syenites, these troughs are deeper than in the ignimbrites, and in the enclaves they are shallower. As observed for Ign-i, the enclaves do not exhibit the same trough for Ba and instead indicate a slight enrichment. Melt inclusions Melt inclusions in the ignimbrites have broadly similar major element compositions to groundmass glass and whole-rock analyses (Figs 6 and 7). All of the melt inclusions are classified as trachyte, with SiO2 contents around 65 wt %, Al2O3 contents between 12 and 17 wt %, total alkali contents of ∼12 wt %, and MgO below 0·5 wt %. The majority of samples are peralkaline (PI = 0·99–1·43). Chlorine concentrations show a total range of 1510–6960 ppm (average = 2810, n = 114), and F contents vary from 620 to 4750 ppm (average = 1644, n = 112). Sulphur concentrations are frequently below the detection limit (300 ppm). FTIR analyses of selected melt inclusions indicate water contents that range from 2·5 to 4·2 wt %, with an average of 3·5 wt %. By contrast, CO2 was not detected in any of the inclusions and was therefore considered to be below the detection limit of FTIR spectroscopy (∼100 ppm; see Gertisser et al., 2012). Groundmass glass Major element compositions of groundmass glass are similar to the whole-rock compositions, and are classified as trachyte (Fig. 6). SiO2 contents cluster around 65 wt %, with Al2O3 contents of ∼15 wt %, total alkali contents of ∼12 wt % and MgO contents that rarely exceed 0·5 wt %. All of the groundmass glass analyses are peralkaline, with PI between 1·05 and 1·26, and are classified as comenditic trachyte. Volatile contents show significant variation; for example, Cl concentrations show a total range of 1460–3370 ppm (average = 2250 ppm, n = 23). Fluorine concentrations are similarly varied, ranging from < 220 to 2610 ppm (average = 1259 ppm, n = 21). By contrast, S concentrations are exclusively below detection (i.e. < 300 ppm). Mineral chemistry Mineral chemical data are provided for the major mineral phases found in the three lithologies of this study: feldspar, clinopyroxene, olivine, Fe–Ti oxides, biotite and amphibole. Feldspar Alkali feldspars in the ignimbrite formations generally range from oligoclase to sanidine in composition, with a range of Or10–39, Ab60–81, An0–11 (Fig. 9a). However, the LMI also contains some plagioclase feldspars, classified as oligoclase, andesine and labradorite (Or1–16, Ab36–77, An10–62). Concentrations of BaO and SrO reach maxima of 1·25 and 0·19 wt %, respectively, with the highest concentrations generally being found in the least potassic feldspars. Feldspars within the syenite autoliths exhibit a similar range to the ignimbrites (Or17–40, Ab60–83, An0–4), although the inclusion of perthitic feldspars extends this range towards the albite and orthoclase end-members. BaO and SrO concentrations are less than observed in the ignimbrites (up to 0·27 and 0·09 wt %, respectively). Groundmass feldspars in the syenite-hosted enclaves show a linear trend between anorthoclase and albite, with a compositional range of Or2–35, Ab65–97, An0–6, and BaO and SrO contents of up to 0·38 and 0·05 wt %, respectively. By contrast, analyses of the large enclave feldspar crystals reveal a bimodal population, with the majority of analyses being classified as anorthoclase or sanidine (Or4–63, Ab36–89, An0–12), and a smaller number of analyses indicating the presence of labradorite and bytownite (Or0–1, Ab18–38, An62–82). The latter population contains SrO concentrations that are somewhat higher than those of the alkali feldspars (0·15 compared with 0·09 wt %). Fig. 9. Open in new tabDownload slide Mineral compositions of the ignimbrite formations, CCI syenite autoliths, and syenite-hosted enclaves. Analyses from the groundmass of the enclaves are labelled ‘Enclave gm’, whereas the enclave phenocrysts are marked ‘Enclave pheno’. (a) Alkali feldspar compositions plotted into the ternary An–Ab–Or system. (b) Clinopyroxene compositions plotted into the pyroxene quadrilateral and, where relevant, the ternary Qd–Jd–Aeg system (Morimoto et al., 1988) and the ternary Di–Hd–Aeg system. (c) Fe–Ti oxide compositions plotted into the TiO2–FeO–Fe2O3 ternary system. (d) Biotite compositions of the GVI (Deer et al., 1966). (e) Ca-amphibole compositions for the syenite-hosted enclaves, plotted using the scheme of Leake et al. (1997). (f) Ca–Na-amphibole compositions for the CCI syenites and syenite-hosted enclaves, plotted using the scheme of Leake et al. (1997). (g) Na-amphibole compositions for the CCI syenites, plotted using the scheme of Leake et al. (1997). Fig. 9. Open in new tabDownload slide Mineral compositions of the ignimbrite formations, CCI syenite autoliths, and syenite-hosted enclaves. Analyses from the groundmass of the enclaves are labelled ‘Enclave gm’, whereas the enclave phenocrysts are marked ‘Enclave pheno’. (a) Alkali feldspar compositions plotted into the ternary An–Ab–Or system. (b) Clinopyroxene compositions plotted into the pyroxene quadrilateral and, where relevant, the ternary Qd–Jd–Aeg system (Morimoto et al., 1988) and the ternary Di–Hd–Aeg system. (c) Fe–Ti oxide compositions plotted into the TiO2–FeO–Fe2O3 ternary system. (d) Biotite compositions of the GVI (Deer et al., 1966). (e) Ca-amphibole compositions for the syenite-hosted enclaves, plotted using the scheme of Leake et al. (1997). (f) Ca–Na-amphibole compositions for the CCI syenites and syenite-hosted enclaves, plotted using the scheme of Leake et al. (1997). (g) Na-amphibole compositions for the CCI syenites, plotted using the scheme of Leake et al. (1997). Clinopyroxene In the three-component (Wo–En–Fs) system of Morimoto et al. (1988) the clinopyroxene populations of the ignimbrites are dominantly classified as augite, with a compositional range of Wo39–46, En21–41, Fs16–36 (Fig. 9b). A small number of crystals from the PNI lie outside this range, with Mg-rich compositions. In contrast, the syenite clinopyroxene is dominated by aegirine–augite to aegirine, with a total compositional range of Qd0–96, Aeg4–99, Jd0–7. Groundmass clinopyroxene from the enclaves also reveals a trend from standard quadrilateral clinopyroxene to aegirine–augite, with a total compositional range of Qd43–97, Aeg3–57, Jd0–9. The enclave phenocrysts exhibit a bimodal distribution of quadrilateral compositions; one comparable with the ignimbrite clinopyroxene (Wo42–43, En31–35, Fs22–26), and another that is more Mg-rich (Wo44–48, En40–46, Fs8–15). In the ternary Di–Hed–Aeg system [calculated using the 10-component scheme of Marks et al. (2008)], all of the analyses define a single trend in which hedenbergite contents increase with little change in aegirine content until approximately Hed55, at which point aegirine increases rapidly towards near end-member compositions. Olivine Olivine phenocrysts within the ignimbrites exhibit a compositional range of Fa59–82. A single Mg-rich olivine was identified in Ign-i, with a contrasting composition of Fo75. EDS spectra semi-quantitatively indicate that large olivine crystals in the syenite-hosted enclaves are Mg-rich in composition. Fe–Ti oxides Ti-magnetite is the most common Fe–Ti oxide present in the ignimbrites and exhibits a compositional range of Mt5–53, Usp45–94, Sp1–6, with Al2O3 and MgO contents of up to 3·0 and 2·1 wt %, respectively (Fig. 9c). MnO contents are generally high, ranging from 1·3 to 2·4 wt %. Ti-magnetite in the syenite autoliths shows a smaller compositional range of Mt32–57, Usp35–68, Sp0, and significantly lower Al2O3 and MgO contents below 0·2 wt %. However, MnO concentrations are similarly high, reaching 2·45 wt %. Ilmenite is less common in the ignimbrites than Ti-magnetite, and exhibits a restricted compositional range of Ilm90–92, Hem3–5, Pyr5. Al2O3 contents are exclusively below 0·1 wt % and MgO concentrations do not exceed 1·8 wt %. The Mn component is comparable with that of Ti-magnetite, with MnO contents up to 2·3 wt %. Ilmenite in the syenite autoliths also show a restricted range (Ilm90–94, Hem1–4, Pyr5–7), low Al2O3 and MgO (< 0·1 and < 0·25 wt %, respectively), and high MnO contents (up to 3·6 wt %). Biotite Biotite phenocrysts found in the GVI are characterized by high TiO2 contents (5·7–6·0 wt %), variable SiO2 (36·6–39·3 wt %), Na2O contents of up to 1·2 wt %, and Fe/(Fe + Mg) ratios of ∼0·35 (Fig. 9d). The hydroxyl sites are characterized by variably high F contents [0·630–0·881 atoms per formula unit (a.p.f.u.)] and low Cl contents (< 0·015 a.p.f.u.). Amphibole Following the nomenclature of Leake et al. (1997), the amphiboles from the syenite autoliths belong to the sodic–calcic (Fig. 9e) and the sodic group (Fig. 9f), and are classified as katophorite to ferrorichterite, and ferroeckermannite and arfvedsonite, respectively. Fluorine concentrations range from 0·714 to 1·055 a.p.f.u., whereas Cl contents are exclusively < 0·015 a.p.f.u. Individual amphibole crystals are often zoned, with Na–Ca-amphibole in the core and Na-amphibole rims. Amphiboles in the groundmass of syenite-hosted enclaves range from calcic to sodic–calcic, and can be classified as ferroedenite, and katophorite and ferrorichterite, respectively (Fig. 9f and g). Occupation of the hydroxyl site is characterized by a greater range than for amphiboles from the syenites, with F ranging from 0·093 to 1·247 a.p.f.u., and Cl not exceeding 0·015 a.p.f.u. DISCUSSION In this section, the combined dataset presented above is used to provide insights into the pre-eruptive magmatic system that fed the ignimbrite-forming eruptions of Terceira. First, the pre-eruptive P–T–fO2 conditions of the ignimbrite-forming magmas are explored, followed by a detailed examination of the relative roles of various petrogenetic processes. Second, a conceptual model is presented, considering the variability of magma rheology and chemical zonation, aiming to account for the petrological features observed within each of the ignimbrites. Third, the ignimbrites are considered within the context of the magmatic suite of Terceira. Finally, the pre-eruptive viscosity of the erupted magmas is considered in terms of its potential control on eruptive behaviour. Pre-eruptive magma storage conditions Temperature Where both Ti-magnetite and ilmenite were present, pre-eruptive temperatures were calculated using the ILMAT program of Lepage (2003), applying the model of Andersen et al. (1993), and utilizing the calculation scheme of Stormer (1983) to determine values for Xhem, Xilm, Xmag and Xulv (Supplementary Data Electronic Appendix 4). The application of alternative calculation schemes was shown to lead to variation of no more than 5% in the calculated results. Equilibrium between mineral pairs was evaluated using the Mn–Mg partitioning test of Bacon & Hirschmann (1988). Pre-eruptive magmatic temperatures were also estimated using the alkali feldspar–melt thermometer of Putirka (2008) (Supplementary Data Electronic Appendix 4). To minimize the error introduced by mineral–liquid disequilibrium, the KdAb–Or equilibrium test proposed by Mollo et al. (2015) was applied, allowing a single suitable liquid composition to be selected for each case. Input pressure values were set at 0·1 GPa, and it was observed that a variation of 0·1 GPa led to a change in temperature of only 0·1°C, suggesting that the thermometer is not significantly influenced by pressure variations. The standard error of estimate (SEE) associated with the thermometer is ±23°C. Because of the relative scarcity of ilmenite, estimates derived from two-oxide models were calculated only for the LAI juvenile clasts and the CCI syenite autoliths. Conversely, the abundance of alkali feldspar facilitated the calculation of magmatic temperatures for all of the ignimbrites and the syenite autoliths. Two-oxide temperature estimates for the LAI range from 773 to 873°C (average = 830°C, n = 590, SD = 21), whereas estimates for the syenite autoliths lie between 616 and 769°C (average = 687°C, n = 20, SD = 40) (Fig. 10a). The alkali feldspar–melt temperatures for the LAI, VFI, CCI, PNI, and Ign-i lie between 857 and 912°C (average = 880°C, n = 304, SD = 7) (Fig. 10b). Notably, the results for the LMI and the GVI deviate from this, with contrasting temperature ranges of 927–938°C (average = 932°C, n = 16, SD = 3) and 819–824°C (average = 821°C, n = 42, SD = 2), respectively. Temperature estimates for the CCI syenites are hotter than those predicted via two-oxide models, with a range of 864–880°C (average = 873°C, n = 31, SD = 4). Alkali feldspar-based temperature estimates for the syenite-hosted enclaves range from 876 to 894°C (average = 884°C, n = 15, SD = 6). Fig. 10. Open in new tabDownload slide Summarized results of thermometry and hygrometry using the ILMAT program of Lepage (2003) for two-oxide thermometry, the alkali feldspar–melt thermometer of Putirka (2008) and the feldspar–melt hygrometer of Mollo et al. (2015). A full discussion of each method is given in the text. (a) T–fO2 estimates for the LAI and the CCI syenite autoliths derived from two-oxide models. FMQ buffer reaction curve calculated for 100 MPa. (b) T–H2Omelt estimates for the ignimbrite formations, the CCI syenite autoliths, and the syenite-hosted enclaves, derived from alkali feldspar–melt thermo-hygrometry. The standard error of the estimate for both temperature and water content is shown in the top right corner. Fig. 10. Open in new tabDownload slide Summarized results of thermometry and hygrometry using the ILMAT program of Lepage (2003) for two-oxide thermometry, the alkali feldspar–melt thermometer of Putirka (2008) and the feldspar–melt hygrometer of Mollo et al. (2015). A full discussion of each method is given in the text. (a) T–fO2 estimates for the LAI and the CCI syenite autoliths derived from two-oxide models. FMQ buffer reaction curve calculated for 100 MPa. (b) T–H2Omelt estimates for the ignimbrite formations, the CCI syenite autoliths, and the syenite-hosted enclaves, derived from alkali feldspar–melt thermo-hygrometry. The standard error of the estimate for both temperature and water content is shown in the top right corner. The temperatures determined via alkali feldspar-based models for the ignimbrite-forming magmas are notably greater than those derived from two-oxide models. Temperature estimates for the CCI syenites exhibit a similar disparity between models, with alkali feldspar thermometry producing a slightly higher range of temperatures. This may reflect the earlier crystallization of the feldspars relative to the oxides, particularly in the syenite nodules. Alternatively, this may result from the rapid re-equilibration timescales of coexisting Fe–Ti oxides (e.g. Gardner et al., 1995; Venezky & Rutherford, 1999; Pimentel et al., 2015), meaning that the lower temperatures recorded by Fe–Ti oxide phases reflect the final pre-eruptive magma system and/or syn-eruptive conditions within the plumbing system. Oxygen fugacity The pre-eruptive redox conditions of the magmatic system were determined via two-oxide models as described above (Supplementary Data Electronic Appendix 4). Estimates could be determined for only the LAI and the CCI syenites. Ti-magnetite and ilmenite pairs in the LAI indicate redox conditions close to 1 log unit below the fayalite-magnetite-quartz (FMQ) buffer reaction (Fig. 10a). The CCI syenites yield results that extend from 1 to 2 log units below FMQ. Pre-eruptive volatile content Pre-eruptive volatile contents were determined via FTIR and electron microprobe analysis of alkali feldspar-hosted melt inclusions and estimated via feldspar–melt hygrometry (Mollo et al., 2015). FTIR analyses of melt inclusions indicate H2O contents that range from 2·5 to 4·2 wt % (average = 3·54 wt %; n = 8), and CO2 contents below detection (< 100 ppm; see Gertisser et al., 2012). Similarly, pre-eruptive concentrations of S are frequently below the detection limit (i.e. < 300 ppm), with only a small number of analyses from the CCI reaching concentrations of up to ∼1260 ppm. Melt inclusions from the LAI and VFI indicate average pre-eruptive concentrations of Cl and F of 2550 and 1550 ppm, respectively. Melt inclusions from the CCI record higher and more variable volatile contents (average  ∼3550 ppm Cl and ∼1760 ppm F). Available melt inclusion analyses for the older ignimbrites suggest that the GVI and PNI are comparable with the CCI, whereas Ign-i exhibits the greatest degree of halogen enrichment, with ∼6960 ppm Cl and 4750 ppm F. Excluding samples from the VFI, the total dataset for Cl and F forms a linear trend that correlates positively with calculated peralkalinity indices, suggesting that their concentrations are controlled primarily by fractionation, and indicating volatile-undersaturated conditions with respect to Cl and F (Fig. 11). Notably, available analyses for Ign-i indicate a positive correlation between F and peralkalinity index, whereas Cl appears to reach a plateau at ∼7000 ppm. This may indicate the exsolution of a Cl-rich aqueous fluid and would suggest that the magma from which Ign-i was derived was stored at pressures below ∼180 MPa (Métrich & Rutherford, 1992). Fig. 11. Open in new tabDownload slide Halogen compositions of melt inclusions from each of the ignimbrite formations of Terceira plotted against Peralkalinity Index. Errors (2σ) do not exceed the symbol size. Fig. 11. Open in new tabDownload slide Halogen compositions of melt inclusions from each of the ignimbrite formations of Terceira plotted against Peralkalinity Index. Errors (2σ) do not exceed the symbol size. For hygrometry, temperature estimates derived from alkali feldspar–melt thermometry were used as primary inputs, alongside feldspar and potential equilibrium liquid compositions. The SEE associated with these results is ±0·53 wt %. When outliers are included, alkali feldspar–melt models applied to the ignimbrites, the CCI syenites, and the syenite-hosted enclaves predict a slightly larger range of pre-eruptive water contents than determined via FTIR, with an overall average of 4·7 wt % (3·0–5·9 wt %; n = 396, SD = 0·6) (Fig. 10b). Within this range, the GVI and LMI deviate from this average value. The GVI yields a restricted but somewhat higher range of 5·2–5·9 wt % (average = 5·7 wt %, n = 42, SD = 0·1), whereas the LMI exhibits a lower than average range of water contents of 3·0–3·7 wt % (average = 3·4 wt %, n = 14, SD = 0·2). Pressure The depth of the magma storage system was estimated quantitatively using the H2O solubility model of Di Matteo et al. (2004). If the maximum water content, determined via FTIR (4·2 wt %) is applied, and water-saturated conditions are assumed, then the minimum pressure associated with the magmatic system that generated the ignimbrite-forming eruptions is ∼80 MPa. This value increases to ∼135 MPa if the maximum estimate of water content derived from alkali feldspar hygrometry is applied (6·0 wt %). Assuming a crustal density of 2800 kg m–3, this equates to depths between 2·2 and 3·7 km. The crustal depths at which the Terceira ignimbrite-forming magmas evolved can also be investigated using the clinopyroxene population. The rocks of this study all contain clinopyroxene, ranging compositionally from diopside, augite, and aegirine–augite in the enclaves to augite in the ignimbrites, and finally to aegirine–augite and aegirine in the syenites. Owing to the Na-rich nature of the syenite clinopyroxene population, they were considered unsuitable for thermobarometry. Furthermore, thorough testing for equilibrium between crystals and melts (following Putirka, 2008, and Mollo et al., 2013; see Jeffery, 2016) indicates a general lack of equilibrium between clinopyroxene and any of the silicic rocks of this study, precluding the application of clinopyroxene-based thermobarometrical models. It is notable that, when applied to the clinopyroxene population of the ignimbrite-forming trachytes, the equilibrium test of Mollo et al. (2013) indicated a positive correlation between the abundance of Al and Na (and therefore the aegirine end-member) and the ‘proximity’ to equilibrium, suggesting that the diopside and augite phenocrysts of the ignimbrites belong to a less evolved melt composition. The aegirine-rich composition of clinopyroxene within the syenite nodules is therefore more likely to reflect a composition in equilibrium with a peralkaline melt, but remains unsuitable for thermobarometry owing to its somewhat extreme composition. It is also noteworthy that, if the diopside and augite compositions of the ignimbrite-forming trachytes are tested against a variety of more mafic compositions (basalt to benmoreite), only a small number of pairs indicate equilibrium at a time, indicating that no single melt composition is suitable. Instead, the clinopyroxene populations appear to originate from a range of melts that are generally mafic to intermediate in composition. Some qualitative insights may, however, still be gained from clinopyroxene chemistry. The TiO2 and Al2O3 contents of clinopyroxene throughout the suite show substantial variations (∼0·2–8·2 wt % and ∼0·1–6·6 wt %, respectively), and, if considered alongside enstatite (En) content, allow the distinction of two chemical trends (Fig. 12). The first trend is marked by a rapid increase in both TiO2 and Al2O3 over a relatively small decrease in En content, and comprises predominantly diopside phenocrysts from the enclaves, with a lesser contribution from the ignimbrites and the syenites. By contrast, the second trend is marked by consistently low TiO2 and Al2O3 contents (< 0·5 wt %) at En contents between 20 and 40 mol %, and primarily consists of augite phenocrysts from the ignimbrites, with a small number of enclave phenocrysts. The groundmass clinopyroxene of the enclaves and the aegirine from the syenites appear to continue this trend to extremely low En contents (< 5 mol %), where Al2O3 remains low but TiO2 rapidly increases to ∼8 wt %. We suggest that the two observed trends indicate two separate stages in the crystallization history of the erupted magmas. The transition between the first and second stages occurred at ∼En40, where a change in the surrounding conditions prompted the crystallization of low-Ti, low-Al clinopyroxene. Although undoubtedly dependent upon additional factors such as melt composition, the link between the Al content of clinopyroxene and crystallization pressure (e.g. Thompson, 1974; Beier et al., 2006) suggests that this may reflect a change in depth [i.e. initial storage at greater depth (termed here the mafic stage), followed by ascent and storage at a shallower depth (termed here the felsic stage)]. Such multi-stage models have been applied at other Azorean volcanoes (Sete Cidades, São Miguel, Beier et al., 2006; Caldeira, Faial, Zanon & Frezzotti, 2013; Zanon et al., 2013; Furnas, São Miguel, Jeffery et al., 2016b), and other North Atlantic oceanic islands (e.g. La Palma, Klügel et al., 2000; Madeira, Schwarz et al., 2004). The continuation of this trend towards En values below 10 mol % and the observed enrichment of TiO2, as defined by the enclave groundmass and the syenites (Fig. 12), are likely to reflect late-stage processes associated with near complete solidification of trachytic melt under low-pressure conditions. Overall, these observations can be accounted for by a two-stage differentiation history, in which primitive, mantle-derived melts stall at a given depth in the crust and differentiate to broadly hawaiitic compositions, before ascending further, where continued differentiation leads to the generation of the erupted peralkaline silicic magmas. Fig. 12. Open in new tabDownload slide Variations in TiO2 (a) and Al2O3 (b) contents of clinopyroxene vs En mol % in relation to the depth of crystallization (see Beier et al., 2006; Jeffery et al., 2016b). Fig. 12. Open in new tabDownload slide Variations in TiO2 (a) and Al2O3 (b) contents of clinopyroxene vs En mol % in relation to the depth of crystallization (see Beier et al., 2006; Jeffery et al., 2016b). The distribution of the dataset across these two trends shows that the majority of clinopyroxenes from the ignimbrites, the syenites, and the groundmass of the enclaves adhere to the felsic trend. The majority of mafic-stage clinopyroxenes are found as phenocrysts in the trachyte enclaves, where felsic-stage phenocrysts are also present. It is therefore suggested that the enclaves provide direct evidence not only for the mingling between trachyte and syenitic mush under comparatively low-pressure conditions (see above), but also for the mingling of trachytes stored in shallow crustal reservoirs with ascending mafic magmas from below. In fact, these enclaves are likely to represent hybridized magmas with multiple populations of crystals, including heavily reacted Mg-rich olivine, diopside, augite, and plagioclase derived from the mafic stage, alkali feldspar derived from the felsic stage, and, more rarely, aegirine-rich clinopyroxenes that are representative of the final portion of the felsic stage, having been clearly included from the surrounding syenite. The origin of the ignimbrite-forming peralkaline trachytes The petrogenetic processes that generate peralkaline silicic magmas have been envisaged to include the following: (1) where no compositional gap (Daly Gap) exists between mafic and silicic compositions, extended fractional crystallization of a mantle-derived alkali basalt parent magma is typically considered (e.g. Barberi et al., 1975; Civetta et al., 1998; Peccerillo et al., 2007), possibly including some assimilation of the crust (e.g. Peccerillo et al., 2003); (2) where a Daly Gap is present, either partial melting of alkali gabbro cumulates (e.g. Bohrson & Reid, 1997) or extended fractional crystallization is generally suggested (e.g. White et al., 2009; Neave et al., 2012); in the former case, peralkaline silicic magmas are envisaged to be produced directly (e.g. Avanzinelli et al., 2004), and may evolve further via fractional crystallization (e.g. Trua et al., 1999). In the following section, a number of petrogenetic processes are explored, aimed at identifying key processes that generate ignimbrite-forming magmas on Terceira. Fractional crystallization The prominent role of fractional crystallization in the generation of evolved magmas in peralkaline silicic systems is well established (e.g. Barberi et al., 1975; Peccerillo et al., 2003; Macdonald et al., 2008; Macdonald, 2012), and was demonstrated quantitatively for the series of young (predominantly < 20–23 ka) lavas from Terceira by Mungall & Martin (1995). However, in terms of their eruptive character and temporal occurrence (∼20–23 to ∼86 ka; Gertisser et al., 2010), the ignimbrites of Terceira may represent a petrologically distinctive system, and should be considered separately. Thus, to validate the role of fractional crystallization, a number of petrogenetic models are applied here. To act as a starting point for each model, a variety of potential compositions are available (Self & Gunn, 1976). Mungall & Martin (1995) recognized three distinctive basaltic compositions, which they termed the on-rift, off-rift, and primitive basalts. Of these, the last corresponds to the silica-undersaturated series of Self & Gunn (1976), whereas the on- and off-rift basalts form the silica-oversaturated trend, and were linked to Pico Alto and Santa Bárbara, respectively (Mungall & Martin, 1995). Gertisser et al. (2010) attributed the Terceira ignimbrites primarily to Pico Alto volcano and in some cases possibly to Guilherme Moniz. For this reason, the on-rift basalts are considered to be the most suitable starting composition for mass-balance and trace element models. To evaluate simple fractional crystallization processes, two least-squares mass-balance models (after Bryan et al., 1969) were performed using the IgPet software package (Carr, 1995), aiming to recreate a basalt to pantellerite fractionation trend (e.g. Barberi et al., 1974; White et al., 1979). Model results were considered acceptable if ∑r2 < 1. Each model comprised five compositional steps: (1) alkali basalt to hawaiite; (2) hawaiite to mugearite; (3) mugearite to benmoreite; (4) benmoreite to least evolved trachyte; (5) least evolved trachyte to most evolved trachyte. Whole-rock and mineral chemical data for alkali basaltic, hawaiitic, mugearitic, benmoreitic, and pantelleritic rocks were taken from Mungall (1993), whereas data for comenditic trachytic compositions were taken from the ignimbrites of this study (Table 2). Two on-rift basalts that plot close the the basalt–hawaiite boundary in the TAS diagram [samples 89-13 and 89-19 of Mungall (1993)] were selected as suitable parent magma compositions on the basis of having the highest Mg# [58 and 53, respectively; Mg# = 100Mg/(Mg + Fe2+) molar] and lowest Zr content (168 and 188 ppm, respectively), alongside no clear evidence for crystal accumulation. However, these compositions are not indicative of primary melts, and so the determined degrees of fractionation (i.e. percentage of solid removed as crystals) should be considered to be minima. Table 2: Mineral compositions used for major element mass-balance modelling Composition: . Basalt . . . . . . Hawaiite . . . . . . Mineral: . Pl . Ol . Cpx . Ti-mag . Ilm . Ap . Pl . Ol . Cpx . Ti-mag . Ilm . Ap . SiO2 51·31 39·20 51·16 54·14 37·28 49·20 TiO2 0·04 1·09 21·50 48·80 0·06 2·40 21·50 48·80 Al2O3 28·73 0·02 2·32 1·48 0·04 26·50 0·03 4·35 1·48 0·04 Fe2O3 FeO 0·70 18·94 6·07 68·53 45·37 0·73 30·02 8·40 68·53 45·37 MnO 0·28 0·16 0·66 0·67 0·46 0·20 0·66 0·67 MgO 0·17 41·58 16·71 1·74 1·73 0·11 33·38 13·79 1·74 1·73 CaO 13·60 0·31 18·61 0·02 0·19 55·70 10·75 0·36 21·57 0·02 0·19 55·70 Na2O 3·63 0·05 0·31 5·00 0·03 0·45 K2O 0·19 0·34 P2O5 41·82 41·82 H2O 0·59 0·59 Composition: Mugearite Benmoreite Mineral: Pl Ol Cpx Ti-mag Ilm Ap Pl Ol Cpx Ti-mag Ilm Ap SiO2 55·57 35·83 47·20 58·33 36·82 50·72 TiO2 0·05 2·33 16·50 35·80 0·02 1·13 16·50 35·80 Al2O3 25·62 0·02 4·36 2·57 1·74 23·81 0·01 2·69 2·57 1·74 Fe2O3 FeO 0·64 35·66 9·53 73·42 51·29 0·71 32·66 8·41 73·42 51·29 MnO 0·93 0·26 0·81 0·56 0·78 0·31 0·81 0·56 MgO 0·08 28·41 13·27 1·77 3·00 0·07 30·56 13·96 1·77 3·00 CaO 9·73 0·23 19·58 0·09 0·00 55·70 6·96 0·22 20·88 0·09 0·00 55·70 Na2O 5·66 0·02 0·56 6·94 0·00 0·52 K2O 0·41 0·77 P2O5 41·82 41·82 H2O 0·59 0·59 Composition: Trachyte Pantellerite Mineral: Afs Ol Cpx Ti-mag Ilm Ap Afs Ol Cpx Ti-mag Ilm Ap SiO2 66·18 33·10 50·52 67·02 30·09 49·97 TiO2 0·06 0·04 0·39 22·56 51·10 0·04 0·51 22·56 50·40 Al2O3 18·75 0·03 1·86 0·60 0·07 18·77 0·01 0·50 0·60 0·02 Fe2O3 FeO 0·49 50·39 15·94 71·77 45·00 0·34 56·85 16·81 71·77 45·78 MnO 3·48 1·33 1·78 2·30 4·04 1·17 1·78 2·19 MgO 0·01 14·28 10·05 1·12 1·68 0·00 7·93 9·15 1·12 0·19 CaO 0·29 0·26 19·89 55·70 0·20 0·47 20·25 0·00 55·70 Na2O 7·82 0·87 7·68 0·00 0·64 K2O 5·47 0·00 6·24 P2O5 41·82 41·82 H2O 0·59 0·59 Composition: . Basalt . . . . . . Hawaiite . . . . . . Mineral: . Pl . Ol . Cpx . Ti-mag . Ilm . Ap . Pl . Ol . Cpx . Ti-mag . Ilm . Ap . SiO2 51·31 39·20 51·16 54·14 37·28 49·20 TiO2 0·04 1·09 21·50 48·80 0·06 2·40 21·50 48·80 Al2O3 28·73 0·02 2·32 1·48 0·04 26·50 0·03 4·35 1·48 0·04 Fe2O3 FeO 0·70 18·94 6·07 68·53 45·37 0·73 30·02 8·40 68·53 45·37 MnO 0·28 0·16 0·66 0·67 0·46 0·20 0·66 0·67 MgO 0·17 41·58 16·71 1·74 1·73 0·11 33·38 13·79 1·74 1·73 CaO 13·60 0·31 18·61 0·02 0·19 55·70 10·75 0·36 21·57 0·02 0·19 55·70 Na2O 3·63 0·05 0·31 5·00 0·03 0·45 K2O 0·19 0·34 P2O5 41·82 41·82 H2O 0·59 0·59 Composition: Mugearite Benmoreite Mineral: Pl Ol Cpx Ti-mag Ilm Ap Pl Ol Cpx Ti-mag Ilm Ap SiO2 55·57 35·83 47·20 58·33 36·82 50·72 TiO2 0·05 2·33 16·50 35·80 0·02 1·13 16·50 35·80 Al2O3 25·62 0·02 4·36 2·57 1·74 23·81 0·01 2·69 2·57 1·74 Fe2O3 FeO 0·64 35·66 9·53 73·42 51·29 0·71 32·66 8·41 73·42 51·29 MnO 0·93 0·26 0·81 0·56 0·78 0·31 0·81 0·56 MgO 0·08 28·41 13·27 1·77 3·00 0·07 30·56 13·96 1·77 3·00 CaO 9·73 0·23 19·58 0·09 0·00 55·70 6·96 0·22 20·88 0·09 0·00 55·70 Na2O 5·66 0·02 0·56 6·94 0·00 0·52 K2O 0·41 0·77 P2O5 41·82 41·82 H2O 0·59 0·59 Composition: Trachyte Pantellerite Mineral: Afs Ol Cpx Ti-mag Ilm Ap Afs Ol Cpx Ti-mag Ilm Ap SiO2 66·18 33·10 50·52 67·02 30·09 49·97 TiO2 0·06 0·04 0·39 22·56 51·10 0·04 0·51 22·56 50·40 Al2O3 18·75 0·03 1·86 0·60 0·07 18·77 0·01 0·50 0·60 0·02 Fe2O3 FeO 0·49 50·39 15·94 71·77 45·00 0·34 56·85 16·81 71·77 45·78 MnO 3·48 1·33 1·78 2·30 4·04 1·17 1·78 2·19 MgO 0·01 14·28 10·05 1·12 1·68 0·00 7·93 9·15 1·12 0·19 CaO 0·29 0·26 19·89 55·70 0·20 0·47 20·25 0·00 55·70 Na2O 7·82 0·87 7·68 0·00 0·64 K2O 5·47 0·00 6·24 P2O5 41·82 41·82 H2O 0·59 0·59 Pl, plagioclase; Ol, olivine; Cpx, clinopyroxene; Ti-mag, Ti-magnetite; Ilm, ilmenite; Ap, apatite; Afs, alkali feldspar. Table 2: Mineral compositions used for major element mass-balance modelling Composition: . Basalt . . . . . . Hawaiite . . . . . . Mineral: . Pl . Ol . Cpx . Ti-mag . Ilm . Ap . Pl . Ol . Cpx . Ti-mag . Ilm . Ap . SiO2 51·31 39·20 51·16 54·14 37·28 49·20 TiO2 0·04 1·09 21·50 48·80 0·06 2·40 21·50 48·80 Al2O3 28·73 0·02 2·32 1·48 0·04 26·50 0·03 4·35 1·48 0·04 Fe2O3 FeO 0·70 18·94 6·07 68·53 45·37 0·73 30·02 8·40 68·53 45·37 MnO 0·28 0·16 0·66 0·67 0·46 0·20 0·66 0·67 MgO 0·17 41·58 16·71 1·74 1·73 0·11 33·38 13·79 1·74 1·73 CaO 13·60 0·31 18·61 0·02 0·19 55·70 10·75 0·36 21·57 0·02 0·19 55·70 Na2O 3·63 0·05 0·31 5·00 0·03 0·45 K2O 0·19 0·34 P2O5 41·82 41·82 H2O 0·59 0·59 Composition: Mugearite Benmoreite Mineral: Pl Ol Cpx Ti-mag Ilm Ap Pl Ol Cpx Ti-mag Ilm Ap SiO2 55·57 35·83 47·20 58·33 36·82 50·72 TiO2 0·05 2·33 16·50 35·80 0·02 1·13 16·50 35·80 Al2O3 25·62 0·02 4·36 2·57 1·74 23·81 0·01 2·69 2·57 1·74 Fe2O3 FeO 0·64 35·66 9·53 73·42 51·29 0·71 32·66 8·41 73·42 51·29 MnO 0·93 0·26 0·81 0·56 0·78 0·31 0·81 0·56 MgO 0·08 28·41 13·27 1·77 3·00 0·07 30·56 13·96 1·77 3·00 CaO 9·73 0·23 19·58 0·09 0·00 55·70 6·96 0·22 20·88 0·09 0·00 55·70 Na2O 5·66 0·02 0·56 6·94 0·00 0·52 K2O 0·41 0·77 P2O5 41·82 41·82 H2O 0·59 0·59 Composition: Trachyte Pantellerite Mineral: Afs Ol Cpx Ti-mag Ilm Ap Afs Ol Cpx Ti-mag Ilm Ap SiO2 66·18 33·10 50·52 67·02 30·09 49·97 TiO2 0·06 0·04 0·39 22·56 51·10 0·04 0·51 22·56 50·40 Al2O3 18·75 0·03 1·86 0·60 0·07 18·77 0·01 0·50 0·60 0·02 Fe2O3 FeO 0·49 50·39 15·94 71·77 45·00 0·34 56·85 16·81 71·77 45·78 MnO 3·48 1·33 1·78 2·30 4·04 1·17 1·78 2·19 MgO 0·01 14·28 10·05 1·12 1·68 0·00 7·93 9·15 1·12 0·19 CaO 0·29 0·26 19·89 55·70 0·20 0·47 20·25 0·00 55·70 Na2O 7·82 0·87 7·68 0·00 0·64 K2O 5·47 0·00 6·24 P2O5 41·82 41·82 H2O 0·59 0·59 Composition: . Basalt . . . . . . Hawaiite . . . . . . Mineral: . Pl . Ol . Cpx . Ti-mag . Ilm . Ap . Pl . Ol . Cpx . Ti-mag . Ilm . Ap . SiO2 51·31 39·20 51·16 54·14 37·28 49·20 TiO2 0·04 1·09 21·50 48·80 0·06 2·40 21·50 48·80 Al2O3 28·73 0·02 2·32 1·48 0·04 26·50 0·03 4·35 1·48 0·04 Fe2O3 FeO 0·70 18·94 6·07 68·53 45·37 0·73 30·02 8·40 68·53 45·37 MnO 0·28 0·16 0·66 0·67 0·46 0·20 0·66 0·67 MgO 0·17 41·58 16·71 1·74 1·73 0·11 33·38 13·79 1·74 1·73 CaO 13·60 0·31 18·61 0·02 0·19 55·70 10·75 0·36 21·57 0·02 0·19 55·70 Na2O 3·63 0·05 0·31 5·00 0·03 0·45 K2O 0·19 0·34 P2O5 41·82 41·82 H2O 0·59 0·59 Composition: Mugearite Benmoreite Mineral: Pl Ol Cpx Ti-mag Ilm Ap Pl Ol Cpx Ti-mag Ilm Ap SiO2 55·57 35·83 47·20 58·33 36·82 50·72 TiO2 0·05 2·33 16·50 35·80 0·02 1·13 16·50 35·80 Al2O3 25·62 0·02 4·36 2·57 1·74 23·81 0·01 2·69 2·57 1·74 Fe2O3 FeO 0·64 35·66 9·53 73·42 51·29 0·71 32·66 8·41 73·42 51·29 MnO 0·93 0·26 0·81 0·56 0·78 0·31 0·81 0·56 MgO 0·08 28·41 13·27 1·77 3·00 0·07 30·56 13·96 1·77 3·00 CaO 9·73 0·23 19·58 0·09 0·00 55·70 6·96 0·22 20·88 0·09 0·00 55·70 Na2O 5·66 0·02 0·56 6·94 0·00 0·52 K2O 0·41 0·77 P2O5 41·82 41·82 H2O 0·59 0·59 Composition: Trachyte Pantellerite Mineral: Afs Ol Cpx Ti-mag Ilm Ap Afs Ol Cpx Ti-mag Ilm Ap SiO2 66·18 33·10 50·52 67·02 30·09 49·97 TiO2 0·06 0·04 0·39 22·56 51·10 0·04 0·51 22·56 50·40 Al2O3 18·75 0·03 1·86 0·60 0·07 18·77 0·01 0·50 0·60 0·02 Fe2O3 FeO 0·49 50·39 15·94 71·77 45·00 0·34 56·85 16·81 71·77 45·78 MnO 3·48 1·33 1·78 2·30 4·04 1·17 1·78 2·19 MgO 0·01 14·28 10·05 1·12 1·68 0·00 7·93 9·15 1·12 0·19 CaO 0·29 0·26 19·89 55·70 0·20 0·47 20·25 0·00 55·70 Na2O 7·82 0·87 7·68 0·00 0·64 K2O 5·47 0·00 6·24 P2O5 41·82 41·82 H2O 0·59 0·59 Pl, plagioclase; Ol, olivine; Cpx, clinopyroxene; Ti-mag, Ti-magnetite; Ilm, ilmenite; Ap, apatite; Afs, alkali feldspar. The two major element mass-balance models are in broad agreement that the least evolved trachyte compositions can be reliably reproduced (average ∑r2 = 0·178) by 84–85% fractionation of an assemblage of plagioclase (46–50%), clinopyroxene (26%), olivine (10–11%), Ti-magnetite (4–7%), ilmenite (4%), and apatite (2–3%). The most evolved trachyte compositions can be produced by a further 14–19% fractionation of the remaining residual liquid (87% total from parent) of an assemblage that is dominated by alkali feldspar (89–92%), olivine (4–5%), clinopyroxene (0–2%), Ti-magnetite (3%), and apatite (1%) (average ∑r2 = 0·329) (Jeffery, 2016). In contrast to the formulations of Mungall & Martin (1995), the inclusion of amphibole at any stage of the models leads invariably to failure. To investigate the role of fractional crystallization under variable P–T–fO2 conditions, ∼200 fractional crystallization models were run using the Rhyolite-MELTS software v. 1.2 (Ghiorso & Sack, 1995; Asimow & Ghiorso, 1998; Gualda et al., 2012). For each model, sample 89-19 (also used for the mass-balance models above) of Mungall (1993) was considered to be the most suitable starting composition (Jeffery, 2016). Fractional crystallization models were generated over three starting water contents (0·5, 1·5, and 2·5 wt %), four isobaric pressures (50, 150, 250, and 500 MPa, representative of the uppermost, upper, middle, and lower crust, respectively) and redox conditions ranging from FMQ + 2 to FMQ – 2 (Jeffery, 2016). Additional polybaric models were run in which the pressure was changed from 500 to 150 MPa at either 1100 or 1000°C, which corresponds to hawaiitic or benmoreitic compositions, respectively. These polybaric models were intended to simulate a general transition from deep to shallow crustal conditions, corresponding to depths of ∼15 km (lower crust beneath the Azores; see Beier et al, 2006) and 2–4 km, respectively. The following mineral phases, having been identified in the relevant lithologies of Terceira, were allowed to crystallize in the model: feldspar, clinopyroxene, olivine, biotite, quartz, apatite, Ti-magnetite, and ilmenite. Hornblende was also permitted to crystallize owing to its potential importance during the earlier portion of the liquid line of descent (e.g. Mungall & Martin, 1995). Each model was evaluated based upon its capability to reproduce the major element compositions of the ignimbrite formations and the Terceira liquid line of descent. Overall, the results of modelling indicate that the major element compositions of the Terceira ignimbrites can be best reproduced by a polybaric model in which the melt differentiates at 500 MPa until it has reached a hawaiitic composition (∼1100°C), at which point the pressure is reduced to 150 MPa (Jeffery, 2016). Relatively reducing conditions (FMQ – 1) and a hydrous parental basalt composition (1·5 wt % H2O) produced the best fit with the observed major element trends (Fig. 7). For this model, olivine is the liquidus phase, and exhibits two crystallization intervals (1180–1150 and 1060°C onwards), which may be an artefact of the MELTS software. This is followed by clinopyroxene (1160–1090°C), ilmenite (1140°C onwards), apatite (1140–1100 and 1040°C onwards), Ti-magnetite (1100°C onwards), and feldspar (1050°C onwards). The total fractionation at 850°C is 74%, of a mineral assemblage dominated by clinopyroxene (38%), feldspar (37%), olivine (11%), Ti-magnetite (8%), apatite (3%), and ilmenite (3%). Predicted H2Omelt values are compatible with the results of FTIR and alkali feldspar–melt hygrometry, ranging from 3·97 to 5·94 wt % for modelled melts with peralkalinity indices of 0·97 and 1·43, respectively. Furthermore, predicted melt temperature values are broadly compatible with the results of both two-oxide and alkali feldspar–melt thermometry, ranging from ∼870 to ∼810°C for the same range in peralkalinity indices. At temperatures below ∼850°C, the models were less successful and frequently failed to run to completion, suggesting that the final portions of the liquid line of descent, represented by the syenite nodules and their intercumulus assemblages, cannot be adequately modelled using Rhyolite-MELTS. It is noteworthy that feldspar compositions predicted by Rhyolite-MELTS for temperatures similar to those predicted by themometric methods (∼940–800°C) were typically more calcic than those observed in the rocks, and compositions within the anorthoclase range were achieved only at the lowermost temperatures predicted by thermometric methods. However, it should be noted that at temperatures below 800°C the model predicted two feldspars simultaneously. By contrast, predicted clinopyroxene compositions (occurring only at temperatures between 1160 and 1090°C, and therefore in mafic liquids) fit well with the observed compositions, particularly with the more Mg-rich compositions observed within the PNI, LMI, and the syenite-hosted enclaves. Similarly, the predicted olivine compositions compare well with the observed compositions. Over the temperature range predicted by thermometric methods (∼940–800°C), Rhyolite-MELTS predicts olivine compositions with a range of Fa65–92, compared with the observed range of Fa59–82. In addition to supporting fractional crystallization as the dominant mechanism of differentiation within the Terceira suite, this model also provides further validation of the pre-eruptive P–T–fO2 conditions determined via thermobaromety above. For example, isobaric models run at 500 MPa invariably fail to achieve the SiO2 and total alkali contents that are observed in the ignimbrites, suggesting that a significant proportion of the liquid line of descent is representative of shallow crustal conditions. This is consistent with the findings of Mungall & Martin (1995), who suggested that the on-rift basalts evolved to more silicic compositions at relatively shallow depths. However, isobaric models run at 150 MPa fail to achieve the total alkali contents observed in the data. The inclusion of an initial, comparatively brief step at higher pressure conditions (for our purposes, 500 MPa was used to represent lower crustal conditions) negated these discrepancies. Utilizing an initial step at conditions below 500 MPa (e.g. 250 MPa) failed to have the same effect. Altering the redox conditions to > FMQ leads to the development of peralkalinity at higher MgO contents than observed on Terceira, suggesting that the calculated values described above (FMQ – 1 to FMQ – 2) are feasible. Similarly, the highest and lowest initial water contents lead to the development of peralkalinity at lower or higher MgO contents, respectively, most probably owing to the predicted control of water content upon the onset of plagioclase crystallization. To further investigate the role of fractional crystallization, and to evaluate the applicability of partial melting of various crustal lithologies as a petrogenetic process, closed-system Rayleigh fractional crystallization (RalFC) and batch melting models for selected trace elements were produced using the Rayleigh fractionation equation and the batch melting equation (Fig. 13). The final RalFC trends comprise three stages, each calculated using phase assemblages and proportions predicted by the most suitable Rhyolite-MELTS model (see above): (1) clinopyroxene + olivine + ilmenite + Ti-magnetite + apatite; (2) plagioclase + olivine + Ti-magnetite + apatite; (3) alkali feldspar + olivine + Ti-magnetite + apatite. Partition coefficient values were taken from the GERM KD database (www.earthref.org/KDD; Nielsen, 2006) and are given in Table 3. These trends are most evident in Fig. 13b and c, where the modelled trends for both Sr and Ba are characterized by an initial, steep positive slope (stage 1), followed by a steeply negative slope (stage 2). Finally, both trends become distinctly flat (stage 3). RalFC models for incompatible elements such as Zr and Nb provide a good fit to the trend observed for Terceira and indicate that the compositional range of the ignimbrites can be accounted for by between ∼65 and 90% fractionation of an alkali basalt parent. The RalFC models for compatible elements such as Sr and Ba (as well as for Cr and Ni; not shown) are less well defined owing to the substantial scatter observed in the Terceira suite, particularly within the intermediate compositions. However, modelled trends adhere closely to the generally low concentrations of Sr, Cr, and Ni observed in the suite. The model fit to the Ba data is poor owing to the substantial scatter in the data, with the majority of the data plotting above the RalFC trend. In this case, the observed scatter is interpreted to reflect the variable degrees of alkali feldspar accumulation and assimilation, as evidenced by the presence of resorbed feldspars throughout the suite. Overall, these models are in agreement with incompatible element models, suggesting that the compositions observed in the ignimbrites can be produced by between ∼65 and ∼90% fractional crystallization. Table 3: Partition coefficients selected for trace element modelling Rayleigh fractionation—Step 1 and batch melting of gabbro . . Pl . Cpx . Ol . Ti-Mag . Ilm . Ap . Afs . Nb 0·1(2) 0·01(3) 0·9(4) 2(5)* Cr 0·08(1) 5·3(1) 2·8(1) 4·2(1) 4·2(1) Y Zr 0·13(1) 0·27(1) 0·06(1) 0·4(1) 0·4(1) Sr 2·7(1) 0·16(1) 0·02(1) 0·68(1) 0·68(1) 1·2(6) Ba 0·56(1) 0·04(1) 0·03(1) 0·4(1) 0·4(1) Rb 0·13(1) 0·04(1) 0·04(1) 0·47(1) 0·47(1) Ni 0·04(1) 2·5(1) 34(1) 3·5(1) 3·5(1) Rayleigh fractionation—Step 2 and batch melting of syenodiorite Pl Cpx Ol Ti-Mag Ilm Ap Afs Nb 0·135(7) Cr Y Zr 0·04(8) 0·44(8) 0·94(8) 0·94(8) Sr 10·5(7) 0·01(9) 0·33(7) Ba 1·77(9) 0·07(10) Rb 0·03(8) 0·04(8) 0·02(8) 0·34(8) 0·34(8) Rayleigh fractionation—Step 3 and batch melting of syenite Pl Cpx Ol Ti-Mag Ilm Ap Afs Nb 0·009(12) 0·051(13)* Cr 6(11) 5(11) 8(11) 8(11) Y 0·138(12) 0·064(13)* Zr 0·16(8) 0·5(11) 0·07(8) 0·25(11) 0·25(11) 0·056(13)* Sr 0·053(12) 8(11) 1·76(13)* Ba 0·023(12) 0·45(11) 5·27(13)* Rb 0·07(8) 0·04(8) 0·08(8) 0·01(8) 0·01(8) 0·31(13)* Rayleigh fractionation—Step 1 and batch melting of gabbro . . Pl . Cpx . Ol . Ti-Mag . Ilm . Ap . Afs . Nb 0·1(2) 0·01(3) 0·9(4) 2(5)* Cr 0·08(1) 5·3(1) 2·8(1) 4·2(1) 4·2(1) Y Zr 0·13(1) 0·27(1) 0·06(1) 0·4(1) 0·4(1) Sr 2·7(1) 0·16(1) 0·02(1) 0·68(1) 0·68(1) 1·2(6) Ba 0·56(1) 0·04(1) 0·03(1) 0·4(1) 0·4(1) Rb 0·13(1) 0·04(1) 0·04(1) 0·47(1) 0·47(1) Ni 0·04(1) 2·5(1) 34(1) 3·5(1) 3·5(1) Rayleigh fractionation—Step 2 and batch melting of syenodiorite Pl Cpx Ol Ti-Mag Ilm Ap Afs Nb 0·135(7) Cr Y Zr 0·04(8) 0·44(8) 0·94(8) 0·94(8) Sr 10·5(7) 0·01(9) 0·33(7) Ba 1·77(9) 0·07(10) Rb 0·03(8) 0·04(8) 0·02(8) 0·34(8) 0·34(8) Rayleigh fractionation—Step 3 and batch melting of syenite Pl Cpx Ol Ti-Mag Ilm Ap Afs Nb 0·009(12) 0·051(13)* Cr 6(11) 5(11) 8(11) 8(11) Y 0·138(12) 0·064(13)* Zr 0·16(8) 0·5(11) 0·07(8) 0·25(11) 0·25(11) 0·056(13)* Sr 0·053(12) 8(11) 1·76(13)* Ba 0·023(12) 0·45(11) 5·27(13)* Rb 0·07(8) 0·04(8) 0·08(8) 0·01(8) 0·01(8) 0·31(13)* Pl, plagioclase; Cpx, clinopyroxene; Ol, olivine; Ti-Mag, Ti-magnetite; Ilm, ilmenite; Ap, apatite; Afs, alkali feldspar. References used: (1) Villemant et al. (1981); (2) Wood & Trigila (2001); (3) McKenzie & O’Nions (1991); (4) Nielsen (1992); (5) Zack & Brumm (1998); (6) Watson & Green (1981); (7) Ewart & Griffin (1994); (8) Lemarchand et al. (1987); (9) Villemant (1988); (10) Luhr et al. (1984); (11) Mahood & Stimac (1990); (12) Larsen (1979); (13) White et al. (2003). * Average partition coefficients. Table 3: Partition coefficients selected for trace element modelling Rayleigh fractionation—Step 1 and batch melting of gabbro . . Pl . Cpx . Ol . Ti-Mag . Ilm . Ap . Afs . Nb 0·1(2) 0·01(3) 0·9(4) 2(5)* Cr 0·08(1) 5·3(1) 2·8(1) 4·2(1) 4·2(1) Y Zr 0·13(1) 0·27(1) 0·06(1) 0·4(1) 0·4(1) Sr 2·7(1) 0·16(1) 0·02(1) 0·68(1) 0·68(1) 1·2(6) Ba 0·56(1) 0·04(1) 0·03(1) 0·4(1) 0·4(1) Rb 0·13(1) 0·04(1) 0·04(1) 0·47(1) 0·47(1) Ni 0·04(1) 2·5(1) 34(1) 3·5(1) 3·5(1) Rayleigh fractionation—Step 2 and batch melting of syenodiorite Pl Cpx Ol Ti-Mag Ilm Ap Afs Nb 0·135(7) Cr Y Zr 0·04(8) 0·44(8) 0·94(8) 0·94(8) Sr 10·5(7) 0·01(9) 0·33(7) Ba 1·77(9) 0·07(10) Rb 0·03(8) 0·04(8) 0·02(8) 0·34(8) 0·34(8) Rayleigh fractionation—Step 3 and batch melting of syenite Pl Cpx Ol Ti-Mag Ilm Ap Afs Nb 0·009(12) 0·051(13)* Cr 6(11) 5(11) 8(11) 8(11) Y 0·138(12) 0·064(13)* Zr 0·16(8) 0·5(11) 0·07(8) 0·25(11) 0·25(11) 0·056(13)* Sr 0·053(12) 8(11) 1·76(13)* Ba 0·023(12) 0·45(11) 5·27(13)* Rb 0·07(8) 0·04(8) 0·08(8) 0·01(8) 0·01(8) 0·31(13)* Rayleigh fractionation—Step 1 and batch melting of gabbro . . Pl . Cpx . Ol . Ti-Mag . Ilm . Ap . Afs . Nb 0·1(2) 0·01(3) 0·9(4) 2(5)* Cr 0·08(1) 5·3(1) 2·8(1) 4·2(1) 4·2(1) Y Zr 0·13(1) 0·27(1) 0·06(1) 0·4(1) 0·4(1) Sr 2·7(1) 0·16(1) 0·02(1) 0·68(1) 0·68(1) 1·2(6) Ba 0·56(1) 0·04(1) 0·03(1) 0·4(1) 0·4(1) Rb 0·13(1) 0·04(1) 0·04(1) 0·47(1) 0·47(1) Ni 0·04(1) 2·5(1) 34(1) 3·5(1) 3·5(1) Rayleigh fractionation—Step 2 and batch melting of syenodiorite Pl Cpx Ol Ti-Mag Ilm Ap Afs Nb 0·135(7) Cr Y Zr 0·04(8) 0·44(8) 0·94(8) 0·94(8) Sr 10·5(7) 0·01(9) 0·33(7) Ba 1·77(9) 0·07(10) Rb 0·03(8) 0·04(8) 0·02(8) 0·34(8) 0·34(8) Rayleigh fractionation—Step 3 and batch melting of syenite Pl Cpx Ol Ti-Mag Ilm Ap Afs Nb 0·009(12) 0·051(13)* Cr 6(11) 5(11) 8(11) 8(11) Y 0·138(12) 0·064(13)* Zr 0·16(8) 0·5(11) 0·07(8) 0·25(11) 0·25(11) 0·056(13)* Sr 0·053(12) 8(11) 1·76(13)* Ba 0·023(12) 0·45(11) 5·27(13)* Rb 0·07(8) 0·04(8) 0·08(8) 0·01(8) 0·01(8) 0·31(13)* Pl, plagioclase; Cpx, clinopyroxene; Ol, olivine; Ti-Mag, Ti-magnetite; Ilm, ilmenite; Ap, apatite; Afs, alkali feldspar. References used: (1) Villemant et al. (1981); (2) Wood & Trigila (2001); (3) McKenzie & O’Nions (1991); (4) Nielsen (1992); (5) Zack & Brumm (1998); (6) Watson & Green (1981); (7) Ewart & Griffin (1994); (8) Lemarchand et al. (1987); (9) Villemant (1988); (10) Luhr et al. (1984); (11) Mahood & Stimac (1990); (12) Larsen (1979); (13) White et al. (2003). * Average partition coefficients. Fig. 13. Open in new tabDownload slide Results of Rayleigh fractionation and batch melting trace element modelling. The calculated Rayleigh fractionation curve is labelled RalFC. Batch melting models were calculated for hypothetical gabbroic, syenodioritic, and syenitic lithologies. Each field comprises the total area occupied by four individual batch melting curves, each curve generated by varying the original mineral proportions of the parental material. For example, the gabbro melting field indicates the area occupied by four separate batch melting curves, each produced by altering the relative proportions of plagioclase, clinopyroxene, and olivine. Fig. 13. Open in new tabDownload slide Results of Rayleigh fractionation and batch melting trace element modelling. The calculated Rayleigh fractionation curve is labelled RalFC. Batch melting models were calculated for hypothetical gabbroic, syenodioritic, and syenitic lithologies. Each field comprises the total area occupied by four individual batch melting curves, each curve generated by varying the original mineral proportions of the parental material. For example, the gabbro melting field indicates the area occupied by four separate batch melting curves, each produced by altering the relative proportions of plagioclase, clinopyroxene, and olivine. By contrast, batch melting models calculated for hypothetical gabbroic, syenodioritic, and syenitic crustal lithologies with variable mineral proportions are almost exclusively incompatible with the trends observed in the Terceira suite (Fig. 13). For example, partial melting of either gabbroic or syenodioritic rock compositions yields trends that deviate significantly from the Terceira suite (e.g. Zr vs Nb; Fig. 13a), and generally fail to achieve the high concentrations of incompatible elements and extremely low concentrations of compatible elements (e.g. Sr, Ba; Fig. 13b and c). Partial melting of a syenitic crustal lithology provides a better fit, but requires degrees of melting in excess of 50% and notably cannot generate the least evolved trachyte compositions of the ignimbrites (∼400–800 ppm Zr). In summary, the results presented here suggest that the ignimbrites of Terceira can be accounted for by extended fractional crystallization of a basaltic parental magma. Although the role of partial melting of crustal lithologies such as alkali gabbro or syenodiorite cannot be ruled out entirely, model results indicate that any contribution from such processes is small and probably limited to assimilation of syenitic rocks. Various petrogenetic models are in broad agreement that the entire compositional range exhibited by the ignimbrites of Terceira can be accounted for by between ∼65 and ∼90% fractionation. Overall, the ignimbrite-forming peralkaline trachytes of Terceira appear to have formed in a two-stage fractionation process, with an initial higher pressure stage in the lower crust, and a later, more significant shallow crustal stage. In situ crystallization Syenite whole-rock major element compositions are similar to those of the trachytes (Fig. 6a), and the depletion of compatible trace elements such as Sr, Ba, and P (attributed here to fractional crystallization; Fig. 8d and e) is also present, and may be even more extreme. If the syenites represent cumulate material derived from the fractional crystallization of trachytic magmas, then they would exhibit compositions that are uniformly less evolved than the ignimbrite-forming trachytes. Additionally, the syenites exhibit negative Eu anomalies that are comparable with those of the trachytes (Fig. 8a and b). The ignimbrite-forming trachytes exhibit a negative correlation between Eu/Eu* and differentiation indices such as Zr and Nb, suggesting that any accumulation of feldspar would buffer, or even counteract, the continued development of a negative anomaly. Based upon the petrographic and geochemical characteristics, it is therefore suggested that the syenitic autoliths do not represent fragments of the cumulate from which the trachytes were derived, but instead provide direct evidence for the role of in situ crystallization of trachytic magma, most probably in the thermal boundary layer at the edge of a magma reservoir (see Tait et al., 1989; Turbeville, 1993; Widom et al., 1993). Magma mingling and remobilization of crystal mush The rocks of Terceira provide abundant evidence for the role of open-system processes such as magma mingling and magma interaction with partially or totally solidified crystal mushes. Mungall (1993) reported trachytic lavas containing mafic enclaves, as well as disaggregated and partially melted syenitic autoliths in basaltic lavas, providing evidence for physical interaction of mafic and silicic magmas. He cited reverse zonation of phenocryst phases to infer the mixing of intermediate magmas shortly before eruption. The syenite-hosted enclaves of this study provide direct evidence for the mingling of variably evolved silicic magmas and/or crystal mushes. If the syenite autoliths are considered to represent randomly sampled portions of a crystal mush derived from in situ crystallization in a thermal boundary zone, then it is suggested that the syenite-hosted enclaves must indicate the injection of the least evolved trachytes of this study into another trachytic reservoir, passing through the marginal crystal mush. Although it cannot be ruled out that the syenites may instead represent significantly older, crustal lithologies, the prevalence of fresh, unaltered textures does not support this interpretation. Furthermore, the intruding trachyte contains a mixed phase assemblage, in which a basaltic assemblage of diopside, Mg-rich olivine, and bytownite is found alongside a more silicic assemblage of oligoclase, anorthoclase and augite, implying that the basaltic assemblage is antecrystic, derived from mixing of mafic magma with a trachytic magma (see Ferla & Meli, 2006). The described difficulty in establishing an equilibrium liquid composition for the clinopyroxenes of this study is likely to be a consequence of this mixing process. In fact, the trachytic assemblage may also be, to some extent, antecrystic. The sieve-textures observed at the rims of the largest anorthoclase crystals may have resulted from the mingling between ascending hawaiites and trachytes stored in the shallow crust, causing reheating of the latter, implying that these crystals represent the true phenocryst assemblage of the trachyte. Alternatively, these textures may indicate disequilibrium between the least evolved trachyte and large anorthoclase crystals originating from the surrounding syenitic mush, introduced during trachyte–syenite interaction and subsequent disaggregation of the latter. Some evidence exists for the remobilization and disaggregation of syenitic mush in the form of glomerocrystic fragments comprising large, cumulus (and often perthitic) alkali feldspar and intercumulus aegirine–augite or Na-amphibole, two phases that are not observed in any lithology other than the syenites. The presence of such glomerocrysts as well as individual feldspar crystals that do not exhibit a perthitic texture suggests that both processes occur. Further evidence for mingling may be seen in the calculated RalFC models, where a number of compositions deviate from the modelled trend (Fig. 13b). In particular, the syenite-hosted enclaves, intermediates (mugearites and benmoreites) and a number of the ignimbrite-forming trachytes exhibit Sr concentrations that form a mixing trend in which the trachytes of the LMI, PNI and Ign-i are mixed with hawaiitic compositions (Fig. 13b). This is consistent with the observed petrographic features for mingling in these lithologies (abundant resorbed crystals), mineral chemistry (rare Mg-rich olivine and plagioclase) and also, to some extent, thermometric and hygrometric evidence (higher temperatures and lower water contents predicted for the LMI; Fig. 10b). This mixing trend is also observed for other compatible elements such as Ba, Cr, and Ni (Cr and Ni not shown), but is less distinct owing to the data scatter. In particular, Ba shows a potential (but highly scattered) mixing trend between hawaiites and trachytes. Furthermore, Ba within the syenite-hosted enclaves reaches concentrations in excess of both the maximum concentration predicted by closed-system RalFC (∼800 ppm) and the proposed mixing trend, up to values as high as ∼1250 ppm. The enrichment of Ba in whole-rock analyses is a feature that is typical of peralkaline systems and is frequently attributed to the accumulation and resorption of alkali feldspars (e.g. Macdonald et al., 2008; Macdonald, 2012). The presence of feldspars with resorption textures throughout the ignimbrites provides evidence for mingling in the ignimbrite-forming trachytes of Terceira. The magmatic plumbing system of the ignimbrite-forming eruptions Implications of viscosity for differentiation Viscosity is of first-order importance for both the evolution and eruption of magmas, and is controlled by magma temperature, composition, abundance of volatile components, solid fraction, and bubble content (e.g. Shaw, 1972; Lejeune & Richet, 1995; Dingwell et al., 1996; Dingwell et al., 1998; Manga et al., 1998; Llewellin et al., 2002b; Costa, 2005; Giordano et al., 2006). To estimate the pre-eruptive viscosity of the ignimbrite-forming trachytes, the model of Giordano et al. (2008) was used, as this model has been calibrated for a wide range of magma compositions and has been shown by recent experimental work to be able to reproduce magma viscosity to within < 0·2 log units (Vona et al., 2011). The minimum and maximum temperature and H2Omelt estimates and the average composition of each ignimbrite were used as input parameters. The total range of crystal-free viscosity estimates varies from 103·1 to 104·0 Pa s. These values were converted into magma viscosities using the method of Dingwell et al. (1993): ηmagma=ηmelt{1+0·75[(f/fm)/(1−f/fm)]}2 where ηmagma and ηmelt are the viscosities of magma and melt, respectively, f is the volume fraction of crystals, and fm is the concentration of crystals required to achieve an infinite viscosity. The last value was set to 0·6, following Andújar & Scaillet (2012), and the volume fraction of crystals was set to values between zero and 0·2, in accordance with petrographic observations (see Self, 1971; Gertisser et al., 2010). Results exhibit a total range of viscosities extending from 103·1 (aphyric, high water) to 104·3 Pa s (20% crystallinity, low water), and indicate that the pre-eruptive viscosities of the ignimbrite-forming trachytes were relatively low, extending to values more than one order of magnitude lower than is typical for metaluminous silicic melts (∼104·5 Pa s, Scaillet et al., 1998). This observation has significant implications not only for eruptive behaviour (see Andújar & Scaillet, 2012) but also for the dominant mechanism of differentiation within the trachytes. Owing to the difficulties of two-phase flow (i.e. crystals settling or floating through a silicate melt) in relatively cool, high-viscosity magmas, the generation of crystal-poor silicic magmas is frequently attributed to the extraction of interstitial melt from a crystal mush via processes such as compaction, hindered settling, micro-settling, and gas-driven filter pressing (Sisson & Bacon, 1999; Bachmann & Bergantz, 2004; Hildreth, 2004; Pistone et al., 2015). However, the application of such models to peralkaline magmatic systems is hindered by the reduced viscosities associated with peralkalinity, which may allow efficient crystal–melt segregation via crystal settling (Macdonald, 2012), as well as by the apparent absence of erupted crystal-rich magmas, typically termed ‘monotonous intermediates’ in metaluminous systems (Hildreth, 1981). As such, the efficiency of two-phase flow in the pre-eruptive magma system of the Terceira ignimbrites was evaluated via the calculation of Stokes’ settling velocities for alkali feldspar crystals, supplemented by the hindered settling equation, which allows the estimation of settling rates in polydispersed suspensions (Bachmann & Bergantz, 2004): Uhs=UStokes×f(c) where Uhs is the hindered settling velocity, UStokes is Stokes settling velocity, and f(c) is a correction factor calculated as f(c)=(1−c)2(1+c1/3)5c/[3(1−c)] where c is equal to the crystal fraction. Crystal sizes were set to 2 mm, in accordance with petrographic observations. Melt densities were set to 2250 kg m–3, based upon the typical densities predicted by Rhyolite-MELTS. Results indicate that the alkali feldspars in the highest viscosity trachytes are capable of settling at rates of between 1·99 (unhindered) and 0·39 m a–1 (hindered settling, 40% crystallinity), whereas those of the lowest viscosity trachytes reach rates of between 19·00 (unhindered) and 3·78 m a–1 (hindered settling, 40% crystallinity). These estimates suggest that, unlike typical metaluminous rhyolites (where calculated settling rates are unlikely to exceed ∼0·67 m a–1, assuming unhindered settling and melt viscosity of 104·5 Pa s), the peralkaline ignimbrite-forming trachytes of Terceira may still be able to segregate melt from crystals efficiently via crystal settling despite their silicic compositions, as suggested at Furnas, São Miguel (Jeffery et al., 2016b), although a contribution from in situ crystallization at the margins of a magma reservoir and associated migration of residual liquids is also feasible. Rapid crystal–melt segregation would not only account for the often crystal-poor nature of the erupted magma, but could also promote the formation of density stratification. The scatter observed for Ba concentrations throughout the suite, and at other peralkaline volcanic centres (e.g. Macdonald et al., 2008; Macdonald, 2012), probably indicates the ease with which peralkaline trachytes can, through rapid crystal settling, lose Ba to underlying melt, or gain Ba by receiving alkali feldspar from overlying melt. Zoned magma bodies Despite their major element homogeneity, the ignimbrites of Terceira exhibit substantial internal trace element variations (e.g. ∼900 to ∼1350 ppm Zr in the CCI; Gertisser et al., 2010), which, based upon the often crystal-poor nature of the juvenile clasts (typically less than 10% crystallinity on a vesicle-free basis), are unlikely to reflect a mineralogical influence. A minor basal pumice fall of the VFI exhibits an even more extreme compositional variation when compared with the overlying ignimbrite units (∼2250 ppm Zr compared with ∼700 ppm, respectively), despite the similarly low crystal content of alkali feldspar. On the basis of the trace element RalFC models presented above, this reflects up to ∼15% fractionation. These phenomena are typically considered to originate from the eruption of a zoned magma body (e.g. Hildreth, 1981; Williams et al., 2014), assuming that the zoning patterns have not been altered by syn-eruptive conduit processes. Alternatively, the eruption of multiple discrete melt pockets within a crystal mush has also been invoked (e.g. Shane et al., 2008; Cooper et al., 2012; Ellis & Wolff, 2012; Zanon et al., 2013; Ellis et al., 2014; Pimentel et al., 2015). However, the data of this study indicate that the observed chemical zoning within the ignimbrites of Terceira is generally gradational rather than abrupt (one exception is discussed below). Although it cannot be ruled out that the observed zonation may be contributed to by processes such as remelting of cumulate material (in this case envisaged to be represented by the syenite nodules) (e.g. Wolff, 2015) or the presence and subsequent eruption of discrete magma pockets (e.g. Ellis et al., 2014), the strong evidence discussed above suggests that the pre-eruptive magma system that generated the ignimbrites of Terceira typically comprised a single, gradationally zoned melt body, with the most evolved and most volatile-rich trachytes overlying progressively less evolved trachytes. The preservation of this zonation until eruption indicates that convection, which might be expected to be enhanced by low melt viscosities, was not sufficient to allow the reservoir to homogenize, possibly owing to the formation of multiple, individually convecting layers (Huppert et al., 1986). One exception to this observation is the LMI, where available data indicate a more restricted range of incompatible trace elements (∼760 to ∼910 ppm Zr). The LMI exhibits abundant evidence for having been mixed with a hotter, less differentiated magma prior to eruption (e.g. presence of resorbed antecrysts including plagioclase, high pre-eruptive temperatures, mixing trend with hawaiitic compositions). The rather limited chemical zonation of the LMI may therefore reflect a large-scale homogenization event, in which an influx of hotter magma initiated convection within the reservoir, and triggered its eruption shortly afterwards. Although the model applied in this study is based around a single, compositionally zoned magma body, the CCI offers some further complexity that may provide some indication of a more complex pre-eruptive magmatic system. Juvenile pumice clasts from the basal pumice fall and dilute pyroclastic density current (surge) deposits of the CCI exhibit incompatible element enrichment compared with the overlying ignimbrite (e.g. ∼1000 versus ∼650 ppm Zr). However, a number of analyses from the overlying ignimbrite record incompatible enrichments that are even greater than those of the basal deposits (> 1000 ppm), suggesting that more complex models for the geometry of the underlying magma system may yet have some role in fully accounting for the observed chemical variability. A model for the magmatic system Here we present a conceptual model for the magma plumbing system from which the ignimbrite-forming episodes of Terceira were fed (Fig. 14). Considering the results of thermodynamic modelling and water solubility, we infer the presence of a magma storage zone at shallow crustal depths (∼2–4 km, assuming a crustal density of 2800 kg m–3). This is consistent with the low concentrations of S and CO2 found in melt inclusions. At least the upper portion of this zone is considered to be exclusively trachytic in composition, based upon the entirely trachytic compositions of alkali feldspar-hosted melt inclusions. Fig. 14. Open in new tabDownload slide Conceptual model for the magma plumbing system of Pico Alto volcano, comprising a prominent magma storage zone in the shallow crust. Byt, bytownite; Di, diopside; Mg-Ol, Mg-rich olivine; Anorth, anorthoclase; Olig, oligoclase; Aug, augite. (a) Ascending hawaiites (phase assemblage = Byt + Di + Mg-Ol) are mixed with trachytes (phase assemblage = Anorth), forming hybridized intermediate to silicic magmas (phase assemblage = Olig + Aug + Anorth + Byt + Di + Mg-Ol). (b) Injection of hybridized trachyte into peralkaline trachyte in the uppermost portion of the shallow crustal storage zone, passing through a marginal syenitic crystal mush and forming enclaves therein. Replenishment initiates further mixing and introduces antecrysts to the eruptible portion of the reservoir. (c) Efficient crystal settling in the uppermost eruptible cap of peralkaline trachyte contributes to the formation of crystal-poor magma and chemical zoning. Fig. 14. Open in new tabDownload slide Conceptual model for the magma plumbing system of Pico Alto volcano, comprising a prominent magma storage zone in the shallow crust. Byt, bytownite; Di, diopside; Mg-Ol, Mg-rich olivine; Anorth, anorthoclase; Olig, oligoclase; Aug, augite. (a) Ascending hawaiites (phase assemblage = Byt + Di + Mg-Ol) are mixed with trachytes (phase assemblage = Anorth), forming hybridized intermediate to silicic magmas (phase assemblage = Olig + Aug + Anorth + Byt + Di + Mg-Ol). (b) Injection of hybridized trachyte into peralkaline trachyte in the uppermost portion of the shallow crustal storage zone, passing through a marginal syenitic crystal mush and forming enclaves therein. Replenishment initiates further mixing and introduces antecrysts to the eruptible portion of the reservoir. (c) Efficient crystal settling in the uppermost eruptible cap of peralkaline trachyte contributes to the formation of crystal-poor magma and chemical zoning. The most applicable Rhyolite-MELTS model included a polybaric regime in which basaltic magmas undergo an initial fractionation period, moving compositionally towards hawaiitic compositions at comparatively high pressures (∼500 MPa), equivalent to the lower crust beneath the Azores (∼15 km; see Beier et al., 2006). We therefore suggest that beneath Terceira ascending mantle-derived basalts are likely to stall in the lower crust and differentiate to hawaiitic compositions via fractional crystallization of a clinopyroxene-dominated assemblage, before ascending further and stalling in the lower portion of the shallow crustal zone. The mixed crystal populations found in syenite-hosted trachytic enclaves facilitate the inference that the hawaiites are introduced into the dominantly trachytic magmas of the shallow crustal system in a mixing zone in the lower regions of the shallow system. Here, they are envisaged to mingle and mix with trachytes, generating hybridized trachytes represented by the syenite-hosted enclaves, with multiple phase assemblages: (1) anorthoclase; (2) diopside + Mg-rich olivine + bytownite; (3) oligoclase + augite (see Bacon & Metz, 1984; Bacon, 1986; Ferla & Meli, 2006). Considering the overwhelming dominance of anorthoclase in the ignimbrite-forming trachytes, the first assemblage is considered to be representative of the trachytes and is therefore the true phenocryst assemblage of the upper crustal system. The second is the phenocryst assemblage of the ascending hawaiites introduced during mixing. This is consistent with the high Al2O3 and TiO2 contents of the diopside phenocrysts, which indicate crystallization within a different magmatic environment (Fig. 12). Owing to the adherence of augite phenocrysts to the felsic trend (described above; Fig. 12), the third assemblage cannot be associated with the ascending hawaiites, and therefore must be attributed to the shallow magma storage zone. Oligoclase is found only in the LMI and is heavily resorbed. In addition, the augite population of the ignimbrite juvenile clasts is not in chemical equilibrium with the host trachyte. Furthermore, Rhyolite-MELTS and mass-balance models predict little to no clinopyroxene crystallization in trachytic compositions. The more sodic clinopyroxene found in the syenitic autoliths and the groundmass of the syenite-hosted enclaves clearly indicates its formation at temperatures lower than those of the trachytes (< 800°C). The oligoclase and augite observed in the enclave magmas are therefore considered to represent an assemblage derived from the benmoreitic or least evolved trachyte hybrid magmas, which result from the mixing of relatively small proportions of hawaiite with trachyte. The hybridized trachytes may then ascend further, out of a lower mixing zone and into the upper regions of the shallow magma storage zone, where the most differentiated magmas are inferred to be present. The mineral assemblages and calculated viscosities of these trachytic magmas suggest that differentiation is likely to be controlled by the crystallization of anorthoclase, driving the melt towards the most evolved trachyte compositions. The compositional zonation of these magma bodies is likely to form as the lower density, more evolved liquids migrate upwards to form a lens of low-viscosity, hydrous, crystal-poor highly evolved trachyte. Additional processes that may play a role in the construction of a zoned magma body include the ascent of bubbles through volatile-saturated magma, which may lead to upward migration of alkalis, halogens, and other fluid-mobile elements (Hildreth & Wilson, 2007), and the generation of low-density, hydrous melts via sidewall crystallization (e.g. Huppert et al., 1986). The preservation of this zonation until eruption may be linked to the development of a system of multiple, individually convecting layers (Huppert & Sparks, 1984), which would probably be enhanced by the low magma viscosities described above, or alternatively a strong density contrast associated with the range of water contents (see Hildreth & Wilson, 2007), as indicated by the water contents measured in alkali feldspar-hosted melt inclusions and calculated via alkali feldspar hygrometry (∼2–6 wt %). Because of the dominance of trachytic compositions at this uppermost level, individual magma reservoirs are likely to be surrounded by a marginal syenitic crystal mush developing via in situ crystallization in the thermal boundary zone. This mush may then be sampled during eruption, providing the syenite autoliths observed in the CCI and, to some extent, in the LAI. The presence of hybridized trachytic enclaves within these syenitic autoliths is considered to record the injection of hybridized trachyte into an upper trachytic reservoir, through the marginal (basal) mush zone. The abundance of miarolitic cavities in both enclave and host syenite provide further evidence for the shallow depth of this magma storage zone. Subsequent mixing of these hybrids with the overlying trachytes may contribute to the development of zonation of the magma body and probably accounts for the occurrence of resorbed oligoclase, augite, and, more rarely, calcic plagioclase, diopside, and Mg-rich olivine in the LMI and the PNI. Furthermore, the replenishment of the uppermost reservoir with hybridized trachyte may act as a trigger for eruption. The calculated rates of crystal settling imply that the antecrystic population introduced during mixing would settle rapidly, implying relatively rapid eruption following replenishment. Finally, a note should be made concerning the GVI, which, unusually for peralkaline silicic systems, contains abundant biotite phenocrysts. Based upon the study by Scaillet & Macdonald (2001), this indicates pre-eruptive temperatures below 700°C, at conditions close to water-saturation. This is consistent with the results of thermometric and hygrometric modelling applied in this study, which predict cooler and wetter pre-eruptive conditions for the GVI than for any of the other ignimbrite formations of Terceira (Fig. 10). Furthermore, the thermodynamic models of Jeffery (2016) predict that, for the most suitable pre-eruptive conditions found in this study, biotite crystallization occurs at redox conditions close to FMQ and FMQ + 1. It is therefore suggested that the GVI represents a magma body that was cooler, closer to water-saturation, and somewhat more oxidized than those that fed the other ignimbrite-forming eruptions. The ignimbrites in the context of the magmatic suite of Terceira When considered in the context of the Terceira suite, the ignimbrites, syenites, and enclaves lie close to the end (< 2 wt % MgO) of a single liquid line of descent that characterizes the volcanic products of the island (Figs 6 and 7). Mungall & Martin (1995) further subdivided this trend, noting the presence of a more oxidizing Santa Bárbara trend at low MgO contents (< 1 wt %). The mafic to intermediate region of this trend comprises the various lava flows and scoria cones associated with the fissure zone (Self, 1974, 1976; Mungall & Martin, 1995; Madureira et al., 2011), whereas the silicic region is made up of the syenite enclaves, the ignimbrites (including whole-rock, melt inclusions and groundmass glass ) and the syenites (Mungall, 1993; Gertisser et al., 2010; Tomlinson et al., 2015; this study), as well as the lava domes and coulées of Pico Alto, and residual glass in syenite xenoliths (labelled P. Alto xenolith glass; Mungall & Martin, 1995), listed in order of increasing differentiation and peralkalinity. The addition of a sixth step to the major element mass-balance models discussed above, in which the most evolved trachyte composition is utilized as a parental magma composition and a pantellerite is used as a daughter (see Table 2), indicates that the pantellerites, which are typical of Pico Alto, can be generated via 72–79% fractionation (93–97% total from basaltic parent) of an assemblage comprising predominantly alkali feldspar (92%), clinopyroxene (3%), olivine (2%), Ti-magnetite (2%), and apatite (< 1%) (average ∑r2 = 0·496). The ignimbrite-forming magmas would therefore have evolved to pantelleritic compositions, had they been stored long enough to allow continued fractionation. In the overall trend, Al2O3 contents show little variation until ∼1 wt % MgO, indicating only a minor role for plagioclase feldspar (Fig. 7). This point corresponds broadly to benmoreitic compositions, and is in agreement with the results of Rhyolite-MELTS modelling. At MgO contents of < 1 wt %, Al2O3 rapidly declines from ∼16 to ∼5 wt %, indicating significant fractionation of initially plagioclase and then alkali feldspar. A similar, kinked trend is observed in FeOt and TiO2, where concentrations show little change until ∼4 wt % MgO, when they decline from ∼12 to ∼4 wt %, and ∼3·5 to < 1 wt %, respectively. As is common in peralkaline systems, a late-stage (< 1 wt % MgO) FeOt enrichment is present, leading to concentrations of up to ∼15 wt % in the most evolved trachytic liquids. The PI of the system shows a gradual increase, becoming peralkaline close to the benmoreite–trachyte boundary, followed by a rapid increase during the final stages. A compositional gap is observed in the suite at ∼1·5–3·0 wt % MgO, which probably corresponds to the previously described Daly Gap for Terceira (Self & Gunn, 1976). Although Mungall & Martin (1995) used monzonitic autoliths to bridge this gap, no such autoliths are reported from Pico Alto. However, the benmoreitic lavas of Mungall (1993) and the syenite-hosted enclaves of this study begin to narrow this gap. At Pico Alto, this gap may reflect a genuine scarcity of intermediate compositions owing to rapid melt evolution at intermediate compositions. Based upon the results of this study, we also suggest that the Daly Gap is likely to reflect prevalent magma hybridization in the shallow magma storage zone (e.g. Storey et al., 1989; Araña et al., 1994; Peccerillo et al., 2003; Sumner & Wolff, 2003; Avanzinelli et al., 2004; Ferla & Meli, 2006; Lowenstern et al., 2006; Romengo et al., 2012). In effect, the intermediate melt compositions are rarely seen because they are likely to form only during the mixing of ascending hawaiite and comparatively large volumes of trachyte in an established silicic magma storage zone. As such, they are rapidly lost, and are recognized only through the preservation of their relict phenocryst assemblages. Overall, it should be noted that, despite its relatively low volume [typical eruptive volumes of ∼1–2 km3 dense rock equivalent (DRE); Self, 1976; Gertisser et al., 2010] and shallow crustal depth (∼2–4 km), the ignimbrite-forming magma system of Terceira (i.e. the underlying magma system of Pico Alto and potentially Guilherme Moniz) appears to have remained relatively stable for up to ∼60 kyr. At least seven ignimbrite-forming eruptions and many smaller scale eruptions have occurred within this time period, suggesting that the rates of magma supply to the shallow crustal system have remained relatively constant, allowing the system to remain thermally active in cold, upper lithosphere. Control of magma viscosity on eruptive behaviour Explosive activity derived from Pico Alto (and potentially Guilherme Moniz) is dominated by low pyroclastic fountaining eruptions with only briefly sustained eruption columns. This eruptive behaviour is likely to be linked to the previously described low pre-eruptive viscosities of the ignimbrite-forming magmas. To investigate the rheological changes induced by ascent-driven degassing, isothermal magma viscosities were calculated for each of the ignimbrite-forming trachytes for water contents ranging from 0 to 6 wt %, assuming a crystal fraction of 0·2 in each instance. Magma compositions used were taken from each of the ignimbrite formations, as well as a typical Terceira pantelleritic composition taken from Mungall & Martin (1995). Results indicate that even total degassing of the ignimbrite-forming trachytes, which leads to an increase in viscosity of about four orders of magnitude (∼107·5 Pa s), is insufficient to achieve the threshold typically required for brittle fragmentation (108–109 Pa s; Papale, 1999; Giordano et al., 2009) (Fig. 15a). Fig. 15. Open in new tabDownload slide Results of rheological modelling aiming to simulate the effects of degassing and microlite crystallization upon the viscosity of the peralkaline trachytic magmas of this study during ascent. (a) Isothermal magma viscosities calculated for each of the ignimbrite-forming peralkaline trachytes using the model of Giordano et al. (2008) and the melt to magma viscosity conversion of Dingwell et al. (1993). Magma water contents were varied from 0 to 6 wt % and crystal fraction was set to 0·2 (see text for full details). Each of the curves was calculated using a composition from one of the peralkaline ignimbrite formations of this study. The dashed line indicates the behaviour of a typical Terceira pantellerite. (b) Bubble-free, crystal-bearing magma viscosities for the peralkaline trachytes of this study, calculated using the model of Vona et al. (2011). Models were run for the minimum and maximum pre-eruptive melt viscosity estimates determined in this study (103·1 and 104·3 Pa s, respectively), using a variable crystal fraction (0–0·4) and a mean crystal aspect ratio of 7 (see text for details). Fig. 15. Open in new tabDownload slide Results of rheological modelling aiming to simulate the effects of degassing and microlite crystallization upon the viscosity of the peralkaline trachytic magmas of this study during ascent. (a) Isothermal magma viscosities calculated for each of the ignimbrite-forming peralkaline trachytes using the model of Giordano et al. (2008) and the melt to magma viscosity conversion of Dingwell et al. (1993). Magma water contents were varied from 0 to 6 wt % and crystal fraction was set to 0·2 (see text for full details). Each of the curves was calculated using a composition from one of the peralkaline ignimbrite formations of this study. The dashed line indicates the behaviour of a typical Terceira pantellerite. (b) Bubble-free, crystal-bearing magma viscosities for the peralkaline trachytes of this study, calculated using the model of Vona et al. (2011). Models were run for the minimum and maximum pre-eruptive melt viscosity estimates determined in this study (103·1 and 104·3 Pa s, respectively), using a variable crystal fraction (0–0·4) and a mean crystal aspect ratio of 7 (see text for details). The viscosity of the ascending trachytes may be further influenced by degassing-induced microlite crystallization (e.g. Sparks & Pinkerton, 1978). The rheological effects of such crystallization are related to crystal abundance, size, shape, packing density, and the shear strain of the magma (e.g. Llewellin et al., 2002a, 2002b; Caricchi et al., 2008; Costa et al., 2009; Di Genova et al., 2013). To explore the effects of syn-eruptive microlite crystallization upon the ignimbrite-forming trachytes, the viscosities of bubble-free, crystal-bearing magmas were calculated using the model of Vona et al. (2011), following Di Genova et al. (2013). Models were run for the minimum and maximum pre-eruptive melt viscosity estimates determined in this study (103·1 and 104·3 Pa s, respectively). In each case, the strain rate was set to γ = 1 s–1, the crystal fraction was varied from 0 to 0·4, and the mean crystal aspect ratio was set to 7 (Hammer & Rutherford, 2002). Results indicate that during crystallization, magma viscosity increases by 2·6 log units before approaching infinite values at a crystal fraction of ∼0·36 (Fig. 15b). The rate of viscosity increase is non-linear and the greatest viscosity increase occurs above a crystal fraction of ∼0·3 (see Di Genova et al., 2013). Based upon the total crystal fraction of microlites in the samples of this study (< 0·1), this process is therefore unlikely to contribute more than a total increase in viscosity of ∼0·2 log units. As such, even when considered alongside the viscosity increase induced by degassing, the ascending trachytes are unlikely to reach the viscosities required for brittle fragmentation. Other processes not considered directly here must therefore be invoked, such as closed-system degassing and the rheological effects of bubble formation. This approach demonstrates how the peralkalinity-induced transposition of pre-eruptive magma viscosities to values lower than those typical of metaluminous magmas can have profound effects, not only upon pre-eruptive processes such as magmatic differentiation, but also upon eruptive behaviour. In this instance, pre-eruptive viscosity may have played a major role in the generation of the pyroclastic density current-producing eruptions of Terceira by inhibiting highly explosive convective volcanic activity, instead promoting low pyroclastic fountaining. Implications for peralkaline magmatic systems The magmatic plumbing system from which the rocks of this study were derived presents a variety of features that are characteristic of peralkaline silicic complexes (see Macdonald, 2012). For example, the erupted material is exclusively peralkaline and silicic in composition, suggesting that the uppermost part of the magmatic system has been stable for a sufficient period of time to allow the development of a shadow zone. In effect, ascending mafic magmas, which provide both the thermal energy and matter required to maintain the system in relatively cool, upper crustal conditions, are prevented from reaching the surface, instead being incorporated into the silicic magmas via mixing. In the case of Pico Alto, the only erupted evidence for the presence of basaltic magmas within the system is a relict phenocryst assemblage. It therefore seems likely that this mixing process represents a primary contributor to the generation of the Daly Gap on Terceira. Historically, the presence of a Daly Gap has been utilized as fundamental evidence for the primary role of partial melting as a petrogenetic process (e.g. Chayes, 1963, 1977). However, various studies have highlighted alternative processes that could lead to the generation of bimodal magmatism, such as density-based redirection of mafic magmas to the periphery of a volcanic centre (e.g. Peccerillo et al., 2003), rapid differentiation of intermediate compositions (e.g. White et al., 2009), and high density and/or crystal content of intermediate magmas preventing their eruption (e.g. Weaver, 1977). The evidence provided in this study suggests that magma mixing processes, invoked at many peralkaline volcanic centres [e.g. Gran Canaria, Canary Islands (Troll & Schmincke, 2002; Sumner & Wolff, 2003), Pantelleria, Italy (Ferla & Meli, 2006; Gioncada & Landi, 2010; Landi & Rotolo, 2015)] represent an equally valid means of generating a Daly Gap (see Romengo et al., 2012). Overall, this suggests that peralkaline magmatic systems that are controlled primarily by fractional crystallization are likely to undergo the following key evolutionary stages: (1) initiation of volcanic activity with mantle-derived mafic magmas ascending from the mantle and/or lower crustal storage zones; (2) development of an upper crustal storage zone in which mafic magmas stall and differentiate towards peralkaline silicic compositions; (3) growth and stabilization of the upper crustal magma storage zone, such that a relatively large volume of silicic magma is stored and maintained via periodic replenishment with comparatively small volumes of mafic magma. Within this scheme, there is a key transition between stages 2 and 3 in which the length of the fractional crystallization-controlled liquid line of descent is split into two separate segments. In stage 2 the liquid line of descent extends from basalt to peralkaline trachyte, whereas in stage 3 there are two separate liquid lines of descent: basalt to hawaiite, and metaluminous trachyte to peralkaline trachyte. In stage 2, intermediate magmas exist as the fractionation products of more mafic magmas, and will go on to fractionate further to produce trachytic compositions. At this point, their apparent absence is more likely to result from their density, crystal content, or their relatively brief existence (see above). In stage 3, intermediate magmas are instead formed only through the mixing of ascending hawaiites with trachyte, suggesting that they exist only briefly, until such time as they have been mixed in with the trachytes, thereby acting as a buffer to continued fractional crystallization within the trachytes. CONCLUSIONS The ignimbrite-forming comenditic trachytes of Terceira can be generated by extended fractional crystallization of hydrous (1·5 wt %), mantle-derived basaltic parental magmas at redox conditions around 1 log unit below the fayalite–magnetite–quartz buffer. Pre-eruptive water contents measured in melt inclusions and estimated via alkali feldspar hygrometry extend from 2·5 to 6·0 wt % and, based upon solubility models, indicate the presence of a prominent magma storage reservoir at shallow crustal depths (∼2–4 km) in which silicic magmas are stored. Syenitic autoliths of similar whole-rock composition to the trachytes provide evidence for the in situ crystallization of trachytic magmas in a thermal boundary layer in the upper crustal reservoir. The abundance of miarolitic cavities in these rocks also indicates shallow crustal conditions. The results of thermodynamic modelling, as well as the minor presence of Al2O3- and TiO2-rich clinopyroxenes, provide evidence for an initial high-pressure fractionation step in the lower crust, in which basalts differentiate via fractional crystallization to hawaiitic compositions. Trachytic enclaves within syenite autoliths contain mixed crystal populations, indicating a two-stage mixing process in which ascending hawaiites are mixed with trachytic magmas in the base of the shallow crustal storage zone. This generates a hybridized trachyte, which then ascends further and is mixed with more evolved trachytes, passing first through a syenitic crystal mush at the margin of a magma reservoir. Calculated magma viscosities for the ignimbrite-forming trachytes extend to values more than one order of magnitude lower than is typical for metaluminous silicic magmas. Estimated crystal settling rates suggest that fractional crystallization is likely to remain a viable process in the trachytic magmas stored in the shallow crust, contributing to the substantial trace element compositional zonation observed in the ignimbrite formations. Major element mass-balance modelling indicates that the most evolved, pantelleritic compositions of Terceira can be generated by continued fractionation of alkali feldspar from the ignimbrite-forming comenditic trachytes. The low pre-eruptive viscosities of the ignimbrite-forming magmas increase the overall difficulty of brittle fragmentation, which may reduce the likelihood of highly explosive (e.g. sustained eruption columns) eruptive behaviour and limit the majority of explosive activity to low pyroclastic fountaining. The sporadic mixing of comparatively low volumes of mafic magma into an established upper crustal silicic reservoir is envisaged to contribute to the generation of a Daly Gap on Terceira, with intermediate magmas existing only briefly before being mixed into trachytic magmas, effectively buffering fractional crystallization in the silicic reservoir. ACKNOWLEDGEMENTS We gratefully acknowledge A. Tindle and A. Kronz for analytical support and access to electron microprobe facilities at the Open University, UK, and the University of Göttingen, Germany, respectively. We are grateful to B. Leze and K. Preece for their support in the production of additional XRF analyses at the University of East Anglia, UK. For diligent assistance in the field, K. Jeffery is also acknowledged. We are also grateful to P. Greatbatch and D. Wilde for the production of thin sections and invaluable assistance in the preparation of melt inclusions for analysis. The paper was greatly improved by the detailed and constructive comments of D. Neave, J. C. White, and G. Daly. We are also grateful to W. Bohrson for insightful comments and for editorial handling. FUNDING This work was supported by Keele University, which provided use of facilities and financial support. B. O’Driscoll acknowledges support from a Natural Environment Research Council (NERC) New Investigator Grant NE/J00457X/1 and a NERC Standard Grant NE/L004011/1. A. Pimentel was financially supported by CIVISA/IVAR. SUPPLEMENTARY DATA Supplementary data for this paper are available at Journal of Petrology online. REFERENCES Andersen D. J. , Lindsley D. H., Davidson P. M. ( 1993 ). QUILF: a Pascal program to assess equilibria among Fe–Mg–Mn–Ti oxides, pyroxenes, olivine and quartz . Computers and Geosciences 19 , 1333 – 1350 . Google Scholar Crossref Search ADS WorldCat Andújar J. , Scaillet B. ( 2012 ). Relationships between pre-eruptive conditions and eruptive styles of phonolite–trachyte magmas . Lithos 152 , 122 – 131 . Google Scholar Crossref Search ADS WorldCat Araña V. , Badiola E. R., Hernán F. ( 1973 ). Peralkaline acid tendencies in Gran Canaria (Canary Islands) . Contributions to Mineralogy and Petrology 40 , 53 – 62 . Google Scholar Crossref Search ADS WorldCat Araña V. , Marti J., Aparicio A., García-Cacho L., García-García R. ( 1994 ). Magma mixing in alkaline magmas: an example from Tenerife, Canary Islands . Lithos 32 , 1 – 19 . Google Scholar Crossref Search ADS WorldCat Asimow P. D. , Ghiorso M. S. ( 1998 ). Algorithmic modifications extending MELTS to calculate subsolidus phase relations . American Mineralogist 83 , 1127 – 1132 . Google Scholar Crossref Search ADS WorldCat Avanzinelli R. , Bindi L., Menchetti S., Conticelli S. ( 2004 ). Crystallisation and genesis of peralkaline magmas from Pantelleria Volcano, Italy: an integrated petrological and crystal-chemical study . Lithos 73 , 41 – 69 . Google Scholar Crossref Search ADS WorldCat Bachmann O. , Bergantz G. W. ( 2004 ). On the origin of crystal-poor rhyolites: extracted from batholithic crystal mushes . Journal of Petrology 45 , 1565 – 1582 . Google Scholar Crossref Search ADS WorldCat Bacon C. R. ( 1986 ). Magmatic inclusions in silicic and intermediate volcanic rocks . Journal of Geophysical Research 91 , 6091 – 6112 . Google Scholar Crossref Search ADS WorldCat Bacon C. R. , Hirschmann M. M. ( 1988 ). Mg/Mn partitioning as a test for equilibrium between coexisting Fe–Ti oxides . American Mineralogist 73 , 57 – 61 . OpenURL Placeholder Text WorldCat Bacon C. R. , Metz J. ( 1984 ). Magmatic inclusions in rhyolites, contaminated basalts, and compositional zonation beneath the Coso volcanic field, California . Contributions to Mineralogy and Petrology 85 , 346 – 365 . Google Scholar Crossref Search ADS WorldCat Barberi F. , Santacroce R., Varet J. ( 1974 ). Silicic peralkaline volcanic rocks of the Afar Depression (Ethiopia) . Bulletin of Volcanology 38 , 755 – 790 . Google Scholar Crossref Search ADS WorldCat Barberi F. , Ferrara G., Santacroce R., Treuil M., Varet J. ( 1975 ). A transitional basalt-pantellerite sequence of fractional crystallisation, the Boina Centre (Afar Rift, Ethiopia) . Journal of Petrology 16 , 22 – 56 . Google Scholar Crossref Search ADS WorldCat Beier C. , Haase K. M., Hansteen T. H. ( 2006 ). Magma evolution of the Sete Cidades volcano, São Miguel, Azores . Journal of Petrology 47 , 1375 – 1411 . Google Scholar Crossref Search ADS WorldCat Beier C. , Haase K. M., Turner S. P. ( 2012 ). Conditions of melting beneath the Azores . Lithos 144–145 , 1 – 11 . Google Scholar Crossref Search ADS WorldCat Black S. , Macdonald R., Kelly R. ( 1997 ). Crustal origin for peralkaline rhyolites from Kenya: evidence from U-series disequilibria and Th-isotopes . Journal of Petrology 38 , 277 – 297 . Google Scholar Crossref Search ADS WorldCat Bohrson W. A. , Reid M. R. ( 1997 ). Genesis of silicic peralkaline volcanic rocks in an ocean island setting by crustal melting and open-system processes: Socorro island, Mexico . Journal of Petrology 38 , 1137 – 1166 . Google Scholar Crossref Search ADS WorldCat Bryan W. B. ( 1966 ). History and mechanism of eruption of soda-rhyolite and alkali basalt, Socorro island, Mexico . Bulletin of Volcanology 67 , 42 – 56 . OpenURL Placeholder Text WorldCat Bryan W. B. , Finger L. W., Chayes F. ( 1969 ). Estimating proportions in petrographic mixing equations by least-squares approximation . Science 163 , 926 – 927 . Google Scholar Crossref Search ADS WorldCat Calvert A. T. , Moore R. B., McGeehin J. P., Rodrigues da Silva A. M. ( 2006 ). Volcanic history and 40Ar/39Ar and 14C geochronology of Terceira Island, Azores, Portugal . Journal of Volcanology and Geothermal Research 156 , 103 – 115 . Google Scholar Crossref Search ADS WorldCat Caricchi L. , Giordano D., Burlini L., Ulmer P., Romano C. ( 2008 ). Rheological properties of magma from the 1538 eruption of Monte Nuovo (Phlegrean Fields, Italy): an experimental study . Chemical Geology 256 , 158 – 171 . Google Scholar Crossref Search ADS WorldCat Carr M. ( 1995 ). Program IgPet . Terra Softa . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Chayes F. ( 1963 ). Relative abundance of intermediate members of the oceanic basalt–trachyte association . Journal of Geophysical Research 68 , 1519 – 1534 . Google Scholar Crossref Search ADS WorldCat Chayes F. ( 1977 ). The oceanic basalt–trachyte relation in general and in the Canary Islands . American Mineralogist 62 , 666 – 671 . OpenURL Placeholder Text WorldCat Civetta L. , Cornette Y., Crisci G., Gillot P. Y., Orsi G., Requejo C. S. ( 1984 ). Geology, geochronology and chemical evolution of the island of Pantelleria . Geological Magazine 121 , 541 – 668 . Google Scholar Crossref Search ADS WorldCat Civetta L. , Antonio M., Orsi G., Tilton G. R. ( 1998 ). The geochemistry of volcanic rocks from Pantelleria island, Sicily Channel: petrogenesis and characteristics of the mantle source region . Journal of Petrology 39 , 1453 – 1491 . Google Scholar Crossref Search ADS WorldCat Cooper G. F. , Wilson C. J., Millet M., Baker J. A., Smith E. G. ( 2012 ). Systematic tapping of independent magma chambers during the 1 Ma Kidnappers super eruption . Earth and Planetary Science Letters 313–314 , 23 – 33 . Google Scholar Crossref Search ADS WorldCat Costa A. ( 2005 ). Viscosity of high crystal content melts: dependence on solid fraction . Geophysical Research Letters 32 , L22308 . OpenURL Placeholder Text WorldCat Costa A. , Carricchi L., Bagdassarov N. S. ( 2009 ). A model for the rheology of particle-bearing suspensions and partially molten rocks . Geochemistry, Geophysics, Geosystems 10 , 1 – 13 . Google Scholar Crossref Search ADS WorldCat Courtillot V. , Davaille A., Besse J., Stock J. ( 2003 ). Three distinct types of hotspots in the Earth’s mantle . Earth and Planetary Science Letters 205 , 295 – 308 . Google Scholar Crossref Search ADS WorldCat Daly R. A. ( 1925 ). The geology of Ascension island . Proceedings of the American Academy of Arts and Sciences 60 , 1 – 80 . Google Scholar Crossref Search ADS WorldCat Davies G. R. , Macdonald R. ( 1987 ). Crustal influences in the petrogenesis of the Naivasha basalt–comendite complex: combined trace element and Sr–Nd–Pb isotope constraints . Journal of Petrology 28 , 1009 – 1031 . Google Scholar Crossref Search ADS WorldCat Deer W. A. , Howie R. A., Zussman J. ( 1966 ). An Introduction to the Rock-Forming Minerals . Longman . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Di Genova D. , Romano C., Hess K.-U., Vona A., Poe B. T., Giordano D., Dingwell D. B., Behrens H. ( 2013 ). The rheology of peralkaline rhyolites from Pantelleria Island . Journal of Volcanology and Geothermal Research 249 , 201 – 216 . Google Scholar Crossref Search ADS WorldCat Di Matteo V. , Carroll M. R., Behrens H., Vetere F., Brooker R. A. ( 2004 ). Water solubility in trachytic melts . Chemical Geology 213 , 187 – 196 . Google Scholar Crossref Search ADS WorldCat Dingwell D. B. , Bagdassarov N. S., Bussod J., Webb S. L. ( 1993 ). Magma rheology. In: Luth R. W. (ed.) Short Handbook on Experiments at High Pressure and Applications to the Earth’s Mantle. Mineralogical Association of Canada, Short Course Series 21 , 131 – 196 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Dingwell D. B. , Romano C., Hess K. U. ( 1996 ). The effect of water on the viscosity of a haplogranitic melt under P–T–X conditions relevant to silicic volcanism . Contributions to Mineralogy and Petrology 124 , 19 – 28 . Google Scholar Crossref Search ADS WorldCat Dingwell D. B. , Hess K.-U., Romano C. ( 1998 ). Extremely fluid behaviour of hydrous peralkaline rhyolites . Earth and Planetary Science Letters 158 , 31 – 38 . Google Scholar Crossref Search ADS WorldCat Duncan A. M. , Queiroz G., Guest J. E., Cole P. D., Wallenstein N., Pacheco J. M. ( 1999 ). The Povoação Ignimbrite, Furnas Volcano, São Miguel, Azores . Journal of Volcanology and Geothermal Research 92 , 55 – 65 . Google Scholar Crossref Search ADS WorldCat Ellis B. S. , Wolff J. A. ( 2012 ). Complex storage of rhyolite in the central Snake River Plain . Journal of Volcanology and Geothermal Research 211–212 , 1 – 11 . Google Scholar Crossref Search ADS WorldCat Ellis B. S. , Bachmann O., Wolff J. A. ( 2014 ). Cumulate fragments in silicic ignimbrites: the case of the Snake River Plain . Geology 42 , 431 – 434 . Google Scholar Crossref Search ADS WorldCat Ewart A. , Griffin W. L. ( 1994 ). Application of proton-microprobe data to trace element partitioning in volcanic rocks . Chemical Geology 117 , 251 – 284 . Google Scholar Crossref Search ADS WorldCat Ferla P. , Meli C. ( 2006 ). Evidence of magma mixing in the ‘Daly Gap’ of alkaline suites: a case study from the enclaves of Pantelleria (Italy) . Journal of Petrology 47 , 1467 – 1507 . Google Scholar Crossref Search ADS WorldCat Fernandes R. M. S. , Bastos L., Miranda J. M., Lourenço N., Ambrosius B. A. C., Noomen R., Simons W. ( 2006 ). Defining the plate boundaries in the Azores region . Journal of Volcanology and Geothermal Research 156 , 1 – 9 . Google Scholar Crossref Search ADS WorldCat Gardner J. E. , Rutherford M., Carey S., Sigurdsson H. ( 1995 ). Experimental constraints on pre-eruptive water contents and changing magma storage prior to explosive eruptions of Mount St Helens volcano . Bulletin of Volcanology 57 , 1 – 17 . Google Scholar Crossref Search ADS WorldCat Gaspar J. L. ( 1996 ). Ilha Graciosa (Açores). História vulcanológica e avaliação do hazard. PhD thesis, Universidade dos Açores, Ponta Delgada. Gaspar J. L. , Queiroz G., Pacheco J. M., Ferreira T., Wallenstein N., Almeida M. H., Coutinho R. ( 2003 ). Serreta submarine ridge eruption (Azores). In: White J. D. L., Smellie J. L., Clague D. A. (eds) Explosive Subaqueous Volcanism. American Geophysical Union, Geophysical Monograph 140 , 205 – 212 . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Gaspar J. L. , Guest J. E., Duncan A. M., Barriga F. J. A. S., Chester D. K. (eds) ( 2015 ). Volcanic Geology of São Miguel Island (Azores Archipelago) . Geological Society, London, Memoirs 44 , 309 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gente P. , Dyment J., Maia M., Goslin J. ( 2003 ). Interaction between the Mid-Atlantic Ridge and the Azores hot spot during the last 85 Myr: emplacement and rifting of the hotspot derived plateaus . Geochemistry, Geophysics, Geosystems 4 , 8514 . Google Scholar Crossref Search ADS WorldCat Gertiser R. , Self S., Gaspar J., Kelley S. P., Pimental A., Eikenberg J., Barry T., Pacheco J., Queiroz G., Vespa M. ( 2010 ). Ignimbrite stratigraphy and chronology on Terceira Island, Azores. In: Gropelli, G. & Viereck-Goette, L. (eds) Stratigraphy and Geology of Volcanic Areas: Geological Society of America Special Paper 464, 133 – 154 . Gertisser R. , Self S., Thomas L. E., Handley H. K., Van Calsteren P., Wolff J. A. ( 2012 ). Processes and timescales of magma genesis and differentiation leading to the Great Tambora eruption in 1815 . Journal of Petrology 53 , 271 – 297 . Google Scholar Crossref Search ADS WorldCat Ghiorso M. S. , Sack R. O. ( 1995 ). Chemical mass transfer in magmatic processes IV. A revised and internally consistent thermodynamic model for the interpolation and extrapolation of liquid–solid equilibria in magmatic systems at elevated temperatures and pressures . Contributions to Mineralogy and Petrology 119 , 197 – 212 . Google Scholar Crossref Search ADS WorldCat Gioncada A. , Landi P. ( 2010 ). The pre-eruptive volatile contents of recent basaltic and pantelleritic magmas at Pantelleria (Italy) . Journal of Volcanology and Geothermal Research 189 , 191 – 201 . Google Scholar Crossref Search ADS WorldCat Giordano D. , Mangiacapra A., Potuzak M., Russell J. K., Romano C., Dingwell D. B., Di Muro A. ( 2006 ). An expanded non-Arrhenian model for silicate melt viscosity: a treatment for metaluminous, peraluminous and peralkaline liquids . Chemical Geology 229 , 42 – 56 . Google Scholar Crossref Search ADS WorldCat Giordano D. , Russell J. K., Dingwell D. B. ( 2008 ). Viscosity of magmatic liquids: a model . Earth and Planetary Science Letters 271 , 123 – 134 . Google Scholar Crossref Search ADS WorldCat Giordano D. , Ardia P., Romano C., Dingwell D. B., Di Muro A., Schmidt M. W., Mangiacapra A., Hess K. U. ( 2009 ). The rheological evolution of alkaline Vesuvius magmas and comparison with alkaline series from the Phlegrean Fields, Etna, Stromboli and Teide . Geochimica et Cosmochimica Acta 73 , 6613 – 6630 . Google Scholar Crossref Search ADS WorldCat Gottsmann J. , Dingwell D. B. ( 2002 ). Thermal expansivities of supercooled haplobasaltic liquids . Geochimica et Cosmochimica Acta 66 , 2231 – 2238 . Google Scholar Crossref Search ADS WorldCat Gualda G. A. , Ghiorso M. S., Lemons R. V., Carley T. L. ( 2012 ). Rhyolite-MELTS: a modified calibration of MELTS optimized for silica-rich, fluid-bearing magmatic systems . Journal of Petrology 53 , 875 – 890 . Google Scholar Crossref Search ADS WorldCat Hammer J. E. , Rutherford M. J. ( 2002 ). An experimental study of the kinetics of decompression-induced crystallisation in silicic melt . Journal of Geophysical Research: Solid Earth 107 , ECV 8-1 – ECV 8-24 . Google Scholar Crossref Search ADS WorldCat Harris C. ( 1983 ). The petrology of lavas and associated plutonic inclusions of Ascension island . Journal of Petrology 24 , 424 – 470 . Google Scholar Crossref Search ADS WorldCat Hildenbrand A. , Weis D., Madureira P., Marques F. O. ( 2014 ). Recent plate reorganization at the Azores Triple Junction: evidence from combined geochemical and geochronological data on Faial, S. Jorge and Terceira volcanic islands . Lithos 210–211 , 27 – 39 . Google Scholar Crossref Search ADS WorldCat Hildreth W. ( 1981 ). Gradients in silicic magma chambers: implications for lithospheric magmatism . Journal of Geophysical Research: Solid Earth 86 , 10153 – 10192 . Google Scholar Crossref Search ADS WorldCat Hildreth W. ( 2004 ). Volcanological perspectives on Long Valley, Mammoth Mountain, and Mono Craters: several contiguous but discrete systems . Journal of Volcanology and Geothermal Research 136 , 169 – 198 . Google Scholar Crossref Search ADS WorldCat Hildreth W. , Wilson C. J. N. ( 2007 ). Compositional zoning of the Bishop Tuff . Journal of Petrology 48 , 951 – 999 . Google Scholar Crossref Search ADS WorldCat Hong W. , Xu X., Zou H. ( 2013 ). Petrogenesis of coexisting high-silica aluminous and peralkaline rhyolite from Yunshan (Yongtai), southeastern China . Journal of Asian Earth Science 74 , 316 – 329 . Google Scholar Crossref Search ADS WorldCat Huppert H. E. , Sparks R. S. J. ( 1984 ). Double diffusive convection due to crystallisation in magmas . Annual Review of Earth and Planetary Sciences 12 , 11 – 37 . Google Scholar Crossref Search ADS WorldCat Huppert H. E. , Stephen R., Sparks J., Wilson J. R., Hallworth M. A. ( 1986 ). Cooling and crystallisation at an inclined plane . Earth and Planetary Science Letters 79 , 319 – 328 . Google Scholar Crossref Search ADS WorldCat Jeffery A. J. ( 2016 ). Petrogenesis and contrasting eruption styles of peralkaline silicic magmas from Terceira and São Miguel, Azores. PhD thesis, Keele University, http://eprints.keele.ac.uk/2477/ Jeffery A. J. , Gertisser R., Jackson R. A., O’Driscoll B., Kronz A. ( 2016a ). On the compositional variability of dalyite, K2ZrSi6O15: a new occurrence from Terceira, Azores . Mineralogical Magazine 80 , 547 – 565 . Google Scholar Crossref Search ADS WorldCat Jeffery A. J. , Gertisser R., O'Driscoll B., Pacheco J. M., Whitley S., Pimentel A., Self S. ( 2016b ). Temporal evolution of a post-caldera, mildly peralkaline magmatic system: Furnas volcano, São Miguel, Azores . Contributions to Mineralogy and Petrology 171 , 42 . Google Scholar Crossref Search ADS WorldCat Kaula W. M. ( 1970 ). Earth’s gravity field: relation to global tectonics . Science 169 , 982 – 985 . Google Scholar Crossref Search ADS WorldCat Klügel A. , Hoernle K. A., Schmincke H.-U., White J. D. L. ( 2000 ). The chemically zoned 1949 eruption on La Palma (Canary Islands): Petrologic evolution and magma supply dynamics of a rift zone eruption . Journal of Geophysical Research: Solid Earth 105 , 5997 – 6016 . Google Scholar Crossref Search ADS WorldCat Krause D. C. , Watkins N. D. ( 1970 ). North Atlantic crustal genesis in the vicinity of the Azores . Geophysical Journal of the Royal Astronomical Society 19 , 261 – 283 . Google Scholar Crossref Search ADS WorldCat Landi P. , Rotolo S. G. ( 2015 ). Cooling and crystallization recorded in trachytic enclaves hosted in pantelleritic magmas (Pantelleria, Italy): Implications for pantellerite petrogenesis . Journal of Volcanology and Geothermal Research 301 , 169 – 179 . Google Scholar Crossref Search ADS WorldCat Larsen L. M. ( 1979 ). Distribution of REE and other trace-elements between phenocrysts and peralkaline undersaturated magmas, exemplified by rocks from the Gardar Igneous Province, South Greenland . Lithos 12 , 303 – 315 . Google Scholar Crossref Search ADS WorldCat Leake B. E. , Woolley A. R., Arps C. E. S., Birch W. D., Gilbert M. C., Grice J. D., Hawthorne F. C., Kato A., Kisch H. J., Krivovichev V. G., Linthout K., Laird J., Mandarino J. A., Maresch W. V., Nickel E. H., Rock N. M. S., Schumacher J. C., Smith D. C., Stephenson N. C. N., Ungaretti L., Whittaker E. J. W., Youzhi G. ( 1997 ). Nomenclature of amphiboles: report of the subcommittee on amphiboles of the international mineralogical association, commission on new minerals and mineral names . The Canadian Mineralogist 35 , 219 – 246 . OpenURL Placeholder Text WorldCat Le Bas M. J. , Le Maitre R. W., Streckeisen A., Zanettin B. ( 1986 ). A chemical classification of volcanic rocks based on the Total Alkali-Silica diagram . Journal of Petrology 27 , 745 – 750 . Google Scholar Crossref Search ADS WorldCat Lejeune A. , Richet P. ( 1995 ). Rheology of crystal-bearing silicate melts: an experimental study at high viscosity . Journal of Geophysical Research: Solid Earth 100 , 4215 – 4229 . Google Scholar Crossref Search ADS WorldCat Lemarchand F. , Villemant B., Calas G. ( 1987 ). Trace element distribution coefficients in alkaline series . Geochimica et Cosmochimica Acta 51 , 1071 – 1081 . Google Scholar Crossref Search ADS WorldCat Lepage L. D. ( 2003 ). ILMAT: an Excel worksheet for ilmenite–magnetite geothermometry and geobarometry . Computers and Geosciences 29 , 673 – 678 . Google Scholar Crossref Search ADS WorldCat Llewellin E. W. , Mader H. M., Wilson S. D. R. ( 2002a ). The constitutive equation and flow dynamics of bubbly magmas . Geophysical Research Letters 29 , 24 . Google Scholar Crossref Search ADS WorldCat Llewellin E. W. , Mader H. M., Wilson S. D. R. ( 2002b ). The rheology of a bubbly liquid . Proceedings of the Royal Society of London, Series A 458 , 987 – 1016 . Google Scholar Crossref Search ADS WorldCat Lourenço N. , Miranda J. M., Luís J. F., Ribeiro A., Victor L. A. M., Madeira J., Needham H. D. ( 1998 ). Morpho-tectonic analysis of the Azores Volcanic Plateau from a new bathymetric compilation of the area . Marine Geophysical Researches 20 , 141 – 156 . Google Scholar Crossref Search ADS WorldCat Lowenstern J. B. , Charlier B. L. A., Clynne M. A., Wooden J. L. ( 2006 ). Extreme U–Th disequilibrium in rift-related basalts, rhyolites and granophyric granite and the timescales of rhyolite generation, intrusion and crystallization at Alid volcanic center, Eritrea . Journal of Petrology 47 , 2105 – 2122 . Google Scholar Crossref Search ADS WorldCat Luhr J. F. , Carmichael I. S. E., Varekamp J. C. ( 1984 ). The 1982 eruptions of El Chichón volcano, Chiapas, Mexico: mineralogy and petrology of the anhydrite-bearing pumices . Journal of Volcanology and Geothermal Research 23 , 69 – 108 . Google Scholar Crossref Search ADS WorldCat Luis J. F. , Miranda J. M. ( 2008 ). Reevaluation of magnetic chrons in the North Atlantic between 35°N and 47°N: implications for the formation of the Azores Triple Junction and associated plateau . Journal of Geophysical Research 113 , B10105 . Google Scholar Crossref Search ADS WorldCat Luis J. F. , Miranda J. M., Galdeano A., Patriat P., Rossignol J. C., Victor L. A. M. ( 1994 ). The Azores triple junction evolution since 10 Ma from an aeromagnetic survey of the Mid-Atlantic Ridge . Earth and Planetary Science Letters 125 , 439 – 459 . Google Scholar Crossref Search ADS WorldCat Macdonald R. ( 1974 ). Nomenclature and petrochemistry of the peralkaline oversaturated extrusive rocks . Bulletin of Volcanology 38 , 498 – 505 . Google Scholar Crossref Search ADS WorldCat Macdonald R. ( 1987 ). Quaternary peralkaline silicic rocks and caldera volcanoes of Kenya. In: Fitton J. G., Upton B. G. J. (eds) Alkaline Igneous Rocks . Geological Society, London, Special Publications 30 , 313 – 333 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Macdonald R. ( 2012 ). Evolution of peralkaline silicic complexes: lessons from the extrusive rocks . Lithos 152 , 11 – 22 . Google Scholar Crossref Search ADS WorldCat Macdonald R. , Scaillet B. ( 2006 ). The central Kenya peralkaline province: insights into the evolution of peralkaline salic magmas . Lithos 91 , 59 – 73 . Google Scholar Crossref Search ADS WorldCat Macdonald R. , Navarro J. M., Upton B. G. J., Davies G. R. ( 1994 ). Strong compositional zonation in peralkaline magma: Menengai, Kenya Rift Valley . Journal of Volcanology and Geothermal Research 60 , 301 – 325 . Google Scholar Crossref Search ADS WorldCat Macdonald R. , Belkin H. E., Fitton J. G., Rogers N. W., Nejbert K., Tindle A. G., Marshall A. S. ( 2008 ). The roles of fractional crystallisation, magma mixing, crystal mush remobilisation and volatile–melt interactions in the genesis of a young basalt–peralkaline rhyolite suite, the Greater Olkaria Volcanic Complex, Kenya Rift Valley . Journal of Petrology 49 , 1515 – 1547 . Google Scholar Crossref Search ADS WorldCat Madeira J. , Brum da Silveira A. ( 2003 ). Active tectonics and first paleoseismological results in Faial, Pico and S. Jorge islands (Azores, Portugal) . Annals of Geophysics 46 , 733 – 761 . OpenURL Placeholder Text WorldCat Madeira J. , Brum da Silveira A., Hipólito A., Carmo R. ( 2015 ). Active tectonics in the Central and Eastern Azores islands along the Eurasia–Nubia boundary: a review. In: Gaspar J. L., Guest J. E., Duncan A. M., Barriga F. J. A. S., Chester D. K. (eds) Volcanic Geology of São Miguel Island (Azores Archipelago). Geological Society, London, Memoirs 44 , 15 – 32 Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Madureira P. , Mata J., Mattielli N., Queiroz G., Silva P. ( 2011 ). Mantle source heterogeneity, magma generation and magmatic evolution at Terceira Island (Azores archipelago): constraints from elemental and isotopic (Sr, Nd, Hf, and Pb) data . Lithos 126 , 402 – 418 . Google Scholar Crossref Search ADS WorldCat Mahood G. A. ( 1984 ). Pyroclastic rocks and calderas associated with strongly peralkaline magmatism . Journal of Geophysical Research 89 , 8540 – 8552 . Google Scholar Crossref Search ADS WorldCat Mahood G. A. , Hildreth W. ( 1986 ). Geology of the peralkaline volcano at Pantelleria, Strait of Sicily . Bulletin of Volcanology 48 , 143 – 172 . Google Scholar Crossref Search ADS WorldCat Mahood G. A. , Stimac J. A. ( 1990 ). Trace-element partitioning in pantellerites and trachytes . Geochimica et Cosmochimica Acta 54 , 2257 – 2276 . Google Scholar Crossref Search ADS WorldCat Mandeville C. W. , Webster J. D., Rutherford M. J., Taylor B. E., Timbal A., Faure K. ( 2002 ). Determination of molar absorptivities for infrared absorption bands of H2O in andesitic glasses . American Mineralogist 87 , 813 – 821 . Google Scholar Crossref Search ADS WorldCat Manga M. , Castro J., Cashman K. V., Loewenberg M. ( 1998 ). Rheology of bubble-bearing magmas . Journal of Volcanology and Geothermal Research 87 , 15 – 28 . Google Scholar Crossref Search ADS WorldCat Markl G. , Marks M. A. W., Frost B. R. ( 2010 ). On the controls of oxygen fugacity in the generation and crystallisation of peralkaline melts . Journal of Petrology 51 , 1831 – 1847 . Google Scholar Crossref Search ADS WorldCat Marks M. A. W. , Schilling J., Coulson I. M., Wenzel T., Markl G. ( 2008 ). The alkaline–peralkaline Tamazeght complex, High Atlas Mountains, Morocco: mineral chemistry and petrological constraints for derivation from a compositionally heterogeneous mantle source . Journal of Petrology 49 , 1097 – 1131 . Google Scholar Crossref Search ADS WorldCat Marks M. A. W. , Hettmann K., Schilling J., Frost B. R., Markl G. ( 2011 ). The mineralogical diversity of alkaline igneous rocks: critical factors for the transition from miaskitic to agpaitic phase assemblages . Journal of Petrology 52 , 439 – 455 . Google Scholar Crossref Search ADS WorldCat Marques F. O. , Catalão J., Hildenbrand A., Madureira P. ( 2015 ). Ground motion and tectonics in the Terceira Island: tectonomagmatic interactions in an oceanic rift (Terceira Rift, Azores Triple Junction) . Tectonophysics 651-652 , 19 – 34 . Google Scholar Crossref Search ADS WorldCat McBirney A. R. ( 1993 ). Differentiated rocks of the Galapagos Hotspot. In: Prichard H. M., Alabaster T., Harris N. B. W., Neary C. R. (eds) Magmatic Processes and Plate Tectonics . Geological Society, London, Special Publications 76 , 61 – 69 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC McKenzie D. , O’Nions R. K. ( 1991 ). Partial melt distributions from inversion of rare earth element concentrations . Journal of Petrology 32 , 1021 – 1091 . Google Scholar Crossref Search ADS WorldCat Métrich N. , Rutherford M. J. ( 1992 ). Experimental study of chlorine behaviour in hydrous silicic melts . Geochimica et Cosmochimica Acta 56 , 607 – 616 . Google Scholar Crossref Search ADS WorldCat Métrich N. , Zanon V., Créon L., Hildenbrand A., Moreira M., Marques F. O. ( 2014 ). Is the ‘Azores Hotspot’ a wetspot? Insights from the geochemistry of fluid and melt inclusions in olivine of Pico basalts . Journal of Petrology 55 , 377 – 393 . Google Scholar Crossref Search ADS WorldCat Miranda J. M. , Luis J. F., Fernandes R. M. S., Lourenço N. ( 2015 ). The structure of the Azores triple junction: implications for S. Miguel Island. in Volcanic Geology of S. Miguel Island (Azores Archipelago), Geological Society, London, Memoirs 44 , 5 – 13 . Google Scholar Crossref Search ADS WorldCat Mollo S. , Masotta M., Forni F., Bachmann O., De Astis G., Moore G., Scarlato P. ( 2015 ). A K-feldspar-liquid hygrometer specific to alkaline differentiated magmas . Chemical Geology 392 , 1 – 8 . Google Scholar Crossref Search ADS WorldCat Mollo S. , Putirka K., Misiti V., Soligo M., Scarlato P. ( 2013 ). A new test for equilibrium based on clinopyroxene–melt pairs: Clues on the solidification temperatures of Etnean alkaline melts at post-eruptive conditions . Chemical Geology 352 , 92 – 100 . Google Scholar Crossref Search ADS WorldCat Morimoto N. , Fabries J., Ferguson A. K., Ginzburg I. V., Ross M., Seifert F. A., Zussman J., Aoki K., Gottardi G. ( 1988 ). Nomenclature of pyroxenes . Mineralogical Magazine 52 , 535 – 550 . Google Scholar Crossref Search ADS WorldCat Mungall J. E. ( 1993 ). Compositional effects of magma mixing and diffusive mass transport on a basalt–pantellerite suite, Terceira, Azores. PhD thesis, McGill University, Montreal, ON. Mungall J. E. , Martin R. F. ( 1995 ). Petrogenesis of basalt–comendite and basalt–pantellerite suites, Terceira, Azores, and some implications for the origin of ocean-island rhyolites . Contributions to Mineralogy and Petrology 119 , 43 – 55 . Google Scholar Crossref Search ADS WorldCat Neave D. A. , Fabbro G., Herd R. A., Petrone C. M., Edmonds E. ( 2012 ). Melting, differentiation and degassing at the Pantelleria volcano, Italy . Journal of Petrology 53 , 637 – 663 . Google Scholar Crossref Search ADS WorldCat Nielsen R. ( 1992 ). BIGD.FOR: A FORTRAN program to calculate trace-element partition coefficients for natural mafic and intermediate composition magmas . Computers and Geosciences 18 , 773 – 788 . Google Scholar Crossref Search ADS WorldCat Nielsen R. ( 2006 ). Geochemical Earth Reference Model (GERM) partition coefficient (Kd) database. www.earthref.org/KDD/ Pacheco J. M. ( 2001 ). Processos associados ao desenvolvimento de erupções vulcânicas hidromagmáticas explosivas na ilha do Faial e sua interpretação numa perspectiva de avaliação do hazard e minimização do risco. PhD thesis, Universidade dos Açores, Ponta Delgada. Papale P. ( 1999 ). Strain-induced magma fragmentation in explosive eruptions . Nature 397 , 425 – 428 . Google Scholar Crossref Search ADS WorldCat Peccerillo A. , Barberio M. R., Yirgu G., Ayalew D., Barbieri M., Wu T. W. ( 2003 ). Relationships between mafic and peralkaline silicic magmatism in continental rift settings: a petrological, geochemical, and isotopic study of the Gedemsa volcano, central Ethiopian Rift . Journal of Petrology 44 , 2003 – 2032 . Google Scholar Crossref Search ADS WorldCat Peccerillo A. , Donati C., Santo A. P., Orlando A., Yirgu G., Ayalew D. ( 2007 ). Petrogenesis of silicic peralkaline rocks in the Ethiopian rift: geochemical evidence and volcanological implications . Journal of African Earth Sciences 48 , 161 – 173 . Google Scholar Crossref Search ADS WorldCat Pimentel A. ( 2006 ). Domos e coulées da ilha Terceira (Açores): contribuição para o estudo dos mecanismos de instalação. MSc thesis, Universidade dos Açores, Ponta Delgada. Pimentel A. , Pacheco J., Self S. ( 2015 ). The ∼1000-years BP explosive eruption of Caldeira Volcano (Faial, Azores): the first stage of incremental caldera formation . Bulletin of Volcanology 77 , 42 . Google Scholar Crossref Search ADS WorldCat Pimentel A. , Zanon V., de Groot L. V., Hipólito A., Di Chiara A., Self S. ( 2016 ). Stress-induced comenditic trachyte effusion triggered by trachybasalt intrusion: multidisciplinary study of the AD 1761 eruption at Terceira Island (Azores) . Bulletin of Volcanology 78 , 22 . Google Scholar Crossref Search ADS WorldCat Pistone M. , Arzilli F., Dobson K. J., Cordonnier B., Reusser E., Ulmer P., Marone F., Whittington A. G., Mancini L., Fife J. L., Blundy J. D. ( 2015 ). Gas-driven filter pressing in magmas: insights into in-situ melt segregation from crystal mushes . Geology 43 , 699 – 702 . Google Scholar Crossref Search ADS WorldCat Putirka K. D. ( 2008 ). Thermometers and barometers for volcanic systems. In: Putirka D. K., Tepley F. J., III (eds) Minerals, Inclusions and Volcanic Processes. Mineralogical Society of America and Geochemical Society, Reviews in Mineralogy and Geochemistry 69 , 61 – 120 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Ren M. , Omenda P., Anthony E., White J., Macdonald R., Bailey D. ( 2006 ). Application of the QUILF thermobarometer to the peralkaline trachytes and pantellerites of the Eburru volcanic complex, East African Rift, Kenya . Lithos 91 , 109 – 124 . Google Scholar Crossref Search ADS WorldCat Ridley W. I. , Watkins N. D., MacFarlane D. J. ( 1974 ). The oceanic islands: Azores. In: Nairn A. E. M., Stehli F., G. (eds) The Ocean Basins and Margins, Volume 2: The North Atlantic . Plenum , pp. 445 – 484 . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Romengo N. , Landi P., Rotolo S. G. ( 2012 ). Evidence of basaltic magma intrusions in a trachytic magma chamber at Pantelleria (Italy) . Periodico di Mineralogia 81 , 1 – 16 . OpenURL Placeholder Text WorldCat Roux J. , Varet J. ( 1975 ). Alkali feldspar liquid equilibrium relationships in peralkaline oversaturated systems and volcanic rocks . Contributions to Mineralogy and Petrology 49 , 67 – 81 . Google Scholar Crossref Search ADS WorldCat Scaillet B. , Macdonald R. ( 2001 ). Phase relations of peralkaline silicic magmas and petrogenetic implications . Journal of Petrology 42 , 825 – 845 . Google Scholar Crossref Search ADS WorldCat Scaillet B. , Holtz F., Pichavant M. ( 1998 ). Phase equilibrium constraints on the viscosity of silicic magmas 1. Volcanic–plutonic comparison . Journal of Geophysical Research: Solid Earth 103 , 27257 – 27266 . Google Scholar Crossref Search ADS WorldCat Schilling J.-G. ( 1991 ). Fluxes and excess temperatures of mantle plumes inferred from their interaction with migrating mid-ocean ridges . Nature 352 , 397 – 403 . Google Scholar Crossref Search ADS WorldCat Schwarz S. , Klügel A., Wohlgemuth-Ueberwasser C. ( 2004 ). Melt extraction pathways and stagnation depths beneath the Madeira and Desertas rift zones (NE Atlantic) inferred from barometric studies . Contributions to Mineralogy and Petrology 147 , 228 – 240 . Google Scholar Crossref Search ADS WorldCat Seaman S. J. , Dyar M. D., Marinkovic N. ( 2009 ). The effects of heterogeneity in magma water concentration on the development of flow banding and spherulites in rhyolitic lava . Journal of Volcanology and Geothermal Research 183 , 157 – 169 . Google Scholar Crossref Search ADS WorldCat Searle R. ( 1980 ). Tectonic pattern of the Azores spreading centre and triple junction . Earth and Planetary Science Letters 51 , 415 – 434 . Google Scholar Crossref Search ADS WorldCat Self S. ( 1971 ). The Lajes Ignimbrite, Ilha Terceira, Açores . Comunicações dos Serviços Geólogicos de Portugal 55 , 165 – 184 . OpenURL Placeholder Text WorldCat Self S. ( 1974 ). Recent volcanism on Terceira, Azores. PhD thesis, Imperial College London. Self S. ( 1976 ). The recent volcanology of Terceira, Azores . Journal of the Geological Society, London 132 , 645 – 666 . Google Scholar Crossref Search ADS WorldCat Self S. , Gunn B. M. ( 1976 ). Petrology, volume and age relations of alkaline and saturated peralkaline volcanics from Terceira, Azores . Contributions to Mineralogy and Petrology 54 , 293 – 313 . Google Scholar Crossref Search ADS WorldCat Shane P. , Nairn I. A., Smith V. C., Darragh M., Beggs K., Cole J. W. ( 2008 ). Silicic recharge of multiple rhyolite magmas by basaltic intrusion during the 22·6 ka Okareka eruption episode, New Zealand . Lithos 103 , 527 – 549 . Google Scholar Crossref Search ADS WorldCat Shao F. , Niu Y., Regelous M., Zhu D.-C. ( 2015 ). Petrogenesis of peralkaline rhyolites in an intra-plate setting: Glass House Mountains, southeast Queensland, Australia . Lithos 216–217 , 196 – 210 . Google Scholar Crossref Search ADS WorldCat Shaw H. R. ( 1972 ). Viscosities of magmatic silicate liquids—empirical method of prediction . American Journal of Science 272 , 870 . Google Scholar Crossref Search ADS WorldCat Silveira G. , Stutzmann E., Davaille A., Montagner J.-P., Mendes-Victor L., Sebai A. ( 2006 ). Azores hotspot signature in the upper mantle . Journal of Volcanology and Geothermal Research 156 , 23 – 34 . Google Scholar Crossref Search ADS WorldCat Sisson T. W. , Bacon C. R. ( 1999 ). Gas-driven filter pressing in magmas . Geology 27 , 613 – 616 . Google Scholar Crossref Search ADS WorldCat Sparks R. S. J. , Pinkerton H. ( 1978 ). Effects of degassing on rheology of basaltic lava . Nature 276 , 385 – 386 . Google Scholar Crossref Search ADS WorldCat Storey M. , Wolff J. A., Norry M. J., Marriner G. F. ( 1989 ). Origin of hybrid lavas from Agua de Pau volcano, São Miguel, Azores. In: Saunders A. D., Norry M. J. (eds) Magmatism in the Ocean Basins . Geological Society, London, Special Publications 42 , 161 – 180 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Stormer J. C. Jr ( 1983 ). The effects of recalculation on estimates and oxygen fugacity from analyses of multi-component iron–titanium oxides . American Mineralogist 68 , 586 – 594 . OpenURL Placeholder Text WorldCat Sumner J. M. , Wolff J. ( 2003 ). Petrogenesis of mixed-magma, high-grade, peralkaline ignimbrite ‘TL’ (Gran Canaria): diverse styles of mixing in a replenished, zoned magma chamber . Journal of Volcanology and Geothermal Research 126 , 109 – 126 . Google Scholar Crossref Search ADS WorldCat Sun S. , McDonough W. F. ( 1989 ). Chemical and isotopic systematics of ocean basins: implications for mantle composition and processes. In: Saunders, A. D. & Norry, M. J. (eds) Magmatism in the Ocean Basins, Geological Society of London Special Publication 42, 313 – 346 . Tait S. R. , Wörner G., Van Den Bogaard P., Schmincke H.-U. ( 1989 ). Cumulate nodules as evidence for convective fractionation in a phonolite magma chamber . Journal of Volcanology and Geothermal Research 37 , 21 – 37 . Google Scholar Crossref Search ADS WorldCat Thompson R. N. ( 1974 ). Some high-pressure clinopyroxenes . Mineralogical Magazine 39 , 768 – 787 . Google Scholar Crossref Search ADS WorldCat Tomlinson E. L. , Smith V. C., Albert P. G., Aydar E., Civetta L., Cioni R., Çubukçu E., Gertisser R., Isaia R., Menzies M. A., Orsi G., Rosi M., Zanchetta G. ( 2015 ). The major and trace element glass compositions of the productive Mediterranean volcanic sources: tools for correlating distal tephra layers in and around Europe . Quaternary Science Reviews 118 , 48 – 66 . Google Scholar Crossref Search ADS WorldCat Troll V. R. , Schmincke H.-U. ( 2002 ). Magma mixing and crustal recycling recorded in ternary feldspar from compositionally zoned peralkaline ignimbrite ‘A’, Gran Canaria, Canary Islands . Journal of Petrology 43 , 243 – 270 . Google Scholar Crossref Search ADS WorldCat Trua T. , Deniel C., Mazzuoli R. ( 1999 ). Crustal control in the genesis of Plio-Quaternary bimodal magmatism of the Main Ethiopian Rift (MER): geochemical and isotopic (Sr, Nd, Pb) evidence . Chemical Geology 155 , 201 – 231 . Google Scholar Crossref Search ADS WorldCat Turbeville B. N. ( 1993 ). Sidewall differentiation in an alkalic magma chamber: evidence from syenite xenoliths in tuffs of the Latera caldera, Italy . Geological Magazine 130 , 453 – 470 . Google Scholar Crossref Search ADS WorldCat Venezky D. Y. , Rutherford M. J. ( 1999 ). Petrology and Fe–Ti oxide reequilibration of the 1991 Mount Unzen mixed magma . Journal of Volcanology and Geothermal Research 89 , 213 – 230 . Google Scholar Crossref Search ADS WorldCat Villemant B. ( 1988 ). Trace element evolution in the Phlegrean Fields (Central Italy): fractional crystallization and selective enrichment . Contributions to Mineralogy and Petrology 98 , 169 – 183 . Google Scholar Crossref Search ADS WorldCat Villemant B. , Jaffrezic H., Joron J.-L., Treuil M. ( 1981 ). Distribution coefficients of major and trace elements; fractional crystallization in the alkali basalt series of Chaîne des Puys (Massif Central, France) . Geochimica et Cosmochimica Acta 45 , 1997 – 2016 . Google Scholar Crossref Search ADS WorldCat Vogt P. R. , Jung W. Y. ( 2004 ). The Terceira Rift as hyper-slow, hotspot-dominated oblique spreading axis: a comparison with other slow-spreading plate boundaries . Earth and Planetary Science Letters 218 , 77 – 90 . Google Scholar Crossref Search ADS WorldCat Vona A. , Romano C., Dingwell D. B., Giordano D. ( 2011 ). The rheology of crystal-bearing basaltic magmas from Stromboli and Etna . Geochimica et Cosmochimica Acta 75 , 3214 – 3236 . Google Scholar Crossref Search ADS WorldCat Watson E. B. , Green T. H. ( 1981 ). Apatite/liquid partition coefficients for the rare earth elements and strontium . Earth and Planetary Science Letters 56 , 405 – 421 . Google Scholar Crossref Search ADS WorldCat Weaver J. D. ( 1977 ). The Quaternary caldera volcano Emuruangogolak, Kenya Rift, and the petrology of a bimodal ferrobasalt–pantelleritic trachyte association . Bulletin Volcanologique 40 , 209 – 230 . Google Scholar Crossref Search ADS WorldCat Weiß B. J. , Hübscher C., Lüdmann T. ( 2015 ). The tectonic evolution of the southeastern Terceira Rift/São Miguel region (Azores) . Tectonophysics 654 , 75 – 95 . Google Scholar Crossref Search ADS WorldCat White J. C. , Holt G. S., Parker D. F., Ren M. ( 2003 ). Trace-element partitioning between alkali feldspar and peralkalic quartz trachyte to rhyolite magma. Part I: systematics of trace element partitioning . American Mineralogist 88 , 316 – 329 . Google Scholar Crossref Search ADS WorldCat White J. C. , Benker S. C., Ren M., Urbanczyk K. M., Corrick D. W. ( 2006 ). Petrogenesis and tectonic setting of the peralkaline Pine Canyon caldera, Trans-Pecos Texas, USA . Lithos 91 , 74 – 94 . Google Scholar Crossref Search ADS WorldCat White J. C. , Parker D. F., Ren M. ( 2009 ). The origin of trachyte and pantellerite from Pantelleria, Italy: Insights from major element, trace element, and thermodynamic modelling . Journal of Volcanology and Geothermal Research 179 , 33 – 55 . Google Scholar Crossref Search ADS WorldCat White W. M. , Schilling J.-G., Hart S. R. ( 1976 ). Evidence for the Azores mantle plume from strontium isotope geochemistry of the central North Atlantic . Nature 263 , 659 – 663 . Google Scholar Crossref Search ADS WorldCat White W. M. , Tapia M. D. M., Schilling J.-G. ( 1979 ). The petrology and geochemistry of the Azores islands . Contributions to Mineralogy and Petrology 69 , 201 – 213 . Google Scholar Crossref Search ADS WorldCat Widom E. , Shirey S. B. ( 1996 ). Os isotope systematics in the Azores: implications for mantle plume sources . Earth and Planetary Science Letters 142 , 451 – 465 . Google Scholar Crossref Search ADS WorldCat Widom E. , Gill J. B., Schmincke H.-U. ( 1993 ). Syenite nodules as a long-term record of magmatic activity in Agua de Pau Volcano, São Miguel, Azores . Journal of Petrology 34 , 929 – 953 . Google Scholar Crossref Search ADS WorldCat Williams R. , Branney M. J., Barry T. L. ( 2014 ). Temporal and spatial evolution of a waxing then waning catastrophic density current revealed by chemical mapping . Geology 42 , 107 – 110 . Google Scholar Crossref Search ADS WorldCat Wolff J. A. ( 2015 ). Remelting of cumulates as a process for producing chemical zoning in silicic tuffs: a comparison of cool, wet and hot, dry rhyolitic magma systems . Lithos 236-237 , 275 – 286 . Google Scholar Crossref Search ADS WorldCat Wood B. J. , Trigila R. ( 2001 ). Experimental determination of aluminous clinopyroxene–melt partition coefficients for potassic liquids, with application to the evolution of the Roman province potassic magmas . Chemical Geology 172 , 213 – 223 . Google Scholar Crossref Search ADS WorldCat Zack T. , Brumm R. ( 1998 ). Ilmenite/liquid partition coefficients of 26 trace elements determined through ilmenite/clinopyroxene partitioning in garnet pyroxene. In: Gurney, J. J., Gurney, J. L., Pascoe, M. D. & Richardson, S. H. (eds) Proceedings of the 7th International Kimberlite Conference, Red Roof Design, Cape Town, pp. 986–988. Zanon V. , Frezzotti M. L. ( 2013 ). Magma storage and ascent conditions beneath Pico and Faial islands (Azores islands): a study on fluid inclusions . Geochemistry, Geophysics, Geosystems 14 , 3494 – 3514 . Google Scholar Crossref Search ADS WorldCat Zanon V. , Pimentel A. ( 2015 ). Spatio-temporal constraints on magma storage and ascent conditions in a transtensional tectonic setting: the case of the Terceira Island (Azores) . American Mineralogist 100 , 795 – 805 . Google Scholar Crossref Search ADS WorldCat Zanon V. , Keuppers U., Pacheco J. M., Cruz I. ( 2013 ). Volcanism from fissure zones and the Caldeira central volcano of Faial Island, Azores archipelago: geochemical processes in multiple feeding systems . Geological Magazine 150 , 536 – 555 . Google Scholar Crossref Search ADS WorldCat Zbyszewski G. ( 1966 ). As observações de F. Fouqué sobre o vulcanismo dos Açores . Boletim do Núcleo Cultural da Horta 4 , 17 – 95 . OpenURL Placeholder Text WorldCat © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
TI - Petrogenesis of the Peralkaline Ignimbrites of Terceira, Azores
JF - Journal of Petrology
DO - 10.1093/petrology/egy012
DA - 2017-12-01
UR - https://www.deepdyve.com/lp/oxford-university-press/petrogenesis-of-the-peralkaline-ignimbrites-of-terceira-azores-Xw1yv0b8S5
SP - 2365
VL - 58
IS - 12
DP - DeepDyve
ER -