Research Collection

Review Article

Low-δ18O silicic magmas on Earth: A review

Author(s): Troch, Juliana; Ellis, Ben S.; Harris, Chris; Bachmann, Olivier; Bindeman, Ilya N.

Publication Date: 2020-09

Permanent Link: https://doi.org/10.3929/ethz-b-000430958

Originally published in: Earth-Science Reviews 208, http://doi.org/10.1016/j.earscirev.2020.103299

Rights / License: Creative Commons Attribution 4.0 International

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library Earth-Science Reviews 208 (2020) 103299

Contents lists available at ScienceDirect

Earth-Science Reviews

journal homepage: www.elsevier.com/locate/earscirev

Low-δ18O silicic magmas on Earth: A review T ⁎ Juliana Trocha,b, , Ben S. Ellisb, Chris Harrisc, Olivier Bachmannb, Ilya N. Bindemand,e a Department of Earth, Environmental and Planetary Sciences, Brown University, 324 Brook Street, Providence, RI 02912, USA b Institute of Geochemistry and Petrology, ETH Zurich, Clausiusstrasse 25, 8092 Zurich, Switzerland c Department of Geological Sciences, University of Cape Town, 13 University Avenue, Rondebosch 7700, South Africa d Department of Earth Sciences, University of Oregon, 1272 University of Oregon, Eugene, OR 97403, USA e Fersman Mineralogical Museum, Leninskiy Prospekt, 115162 Moscow, Russia

ARTICLE INFO ABSTRACT

Keywords: Silicic magmas play an important role in the formation of continental crust and are responsible for some of the Oxygen isotopes most hazardous volcanic eruptions on the planet. Low-δ18O silicic magmas (δ18O < 5.5 ‰) have been a pet- Magmas rological conundrum as they require significant incorporation of rocks that were hydrothermally altered by Granite meteoric water at high water/rock ratios in the shallow, permeable, and relatively cold upper crust (< 400 °C), a Rhyolite region thought to be unfavorable for the production of large melt volumes. Their genesis is therefore crucial in Hydrothermal alteration understanding how silicic magma reservoirs interact with the upper crust, and how they can remain active and Assimilation produce extensive amounts of silicic magma over timescales of millions of years. In this paper, we compare low- 18 Crustal melting δ O silicic magmas from different tectonic settings, in order to identify general mechanisms for the production of low-δ18O silicic magmas on Earth. Low-δ18O magmas can be linked to either assimilation of pre-existing hydrothermally altered crust, or (more commonly) to assimilation of syn-magmatically altered rocks. Assimilation of syn-magmatically altered rocks may occur in a variety of volcanic settings, but is most likely in shallow, large-scale, long-lived caldera-forming systems that host extensive high-temperature hydrothermal systems and produce hot (> 800 °C) and dry silicic magmas. The relative scarcity of low-δ18O silicic magmas on Earth compared to normal- and high-δ18O magmas implies that coincidence of these factors is rare, and is most likely encountered in hotspot and rift settings characterized by bimodal basaltic-rhyolitic volcanism. Low-δ18O silicic magmas are usually generated by bulk assimilation of rocks that were hydrothermally altered at high temperatures (> 300 °C) by isotopically light meteoric water, prevalent at mid to high latitudes and altitudes and/or linked to global glaciation episodes in Earth’s history. We estimate that < 30-40 % assimilation can explain most of the oxygen isotope compositions of low-δ18O magmas, consistent with estimates from thermal models. At conditions optimal for oxygen isotope exchange towards lower δ18O values, alteration is not asso- ciated with hydration, and hydrothermally altered low-δ18O rocks do not melt more readily than average crust. Assimilation of co-genetic hydrothermally altered rocks rarely leaves identifiable traces in the major and trace element record of low-δ18O silicic magmas, and may often be obscured by assimilation of high-δ18O crustal rocks. These findings provide a framework for the assessment of low-δ18O silicic magmas on Earth, and the parameters that play a role in their genesis.

1. Introduction mantle interactions back to the Hadean (Valley et al., 2005). Silicic magmas with low δ18O values represent an unusual case of intracrustal Silicic magmas are crucial building blocks in the generation of recycling, as they record assimilation of rocks that were hydrothermally continental crust on Earth. They contribute to crustal growth either by altered by meteoric water at temperatures ≥300-400 °C at shallow forming plutons, or erupting in often devastating large-volume volcanic depths (< 10 km) in the Earth’s crust. The relatively low prevailing eruptions on the Earth’s surface (Bachmann and Huber, 2016; Bryan temperatures at such depths make the upper crust an unfavorable en- et al., 2010; Hildreth, 1981). Oxygen isotopes are important tracers for vironment for large-scale interaction between crust and mantle-derived crustal evolution and maturation (Simon and Lécuyer, 2005; Taylor, fractionated magmas, leading to the question of how silicic magmas can 1968), and have been used to track intracrustal recycling and crust- obtain exceedingly low δ18O values (reaching down to -6 ‰ or even

⁎ Corresponding author. E-mail address: [email protected] (J. Troch). https://doi.org/10.1016/j.earscirev.2020.103299 Received 4 May 2020; Received in revised form 17 July 2020; Accepted 21 July 2020 Available online 25 July 2020 0012-8252/ © 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/BY/4.0/). J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

+90°

High Arctic LIP Kanger- (130-79 Ma) lussuaq Iceland (50 Ma) (<10 ka) Aleutians North Atlantic Siberian Kamchatka LIP (62-58 Ma) Traps (<6 Ma) (<10 Ma) (252-250 Ma) Isle of Skye/Mull +60° Casto pluton (60-53 Ma) Dabie-Sulu Columbia River LIP (47-44 Ma) Yellowstone/Heise (<7 Ma) (870-600 Ma) Baekdusan (17-5 Ma) Central Snake River Plain (13-8 Ma) granites (AD 940) Mt. Mazama (167-110 Ma) New England Gyeongsan (<10 ka) (88-27 Ma) Avalon granites Tenerife Malani +30° Sierra Madre (600-550 Ma) (<12 Ma) (780-750 Ma) Occidental LIP (38-20 Ma) Central Atlantic Emeishan Magmatic Province Afar LIP Deccan Rajmahal Traps Timber Mountain complex (16-9 Ma) (204-191 Ma) (31-29 Ma) Traps LIP (265-259 Ma) Western Nevada volc. field (35-19 Ma) (68-60 Ma) (116-95 Ma) ±0° Seychelles (~750 Ma) Paraná-Etendeka LIP Kalkarindji LIP Whitsunday LIP (138-129 Ma) (512-509 Ma) (132-129 Ma) Imorona-Itsindro Koegel Fontein (850-750 Ma) Talbot -30° complex (144-133 Ma) Lebombo Calabozos Gawler Range LIP (1080-1040 Ma) (< 1 Ma) (183-180 Ma) (1600-1500 Ma)

Chon Aike (188-153 Ma) Karoo-Ferrar LIP (183-179 Ma) -60°

Meteoric δ18O values today -24 -18 -12 -8 -4 0

-90°

Fig. 1. World map with currently known occurrences of low-δ18O silicic magmas. Red dots indicate volcanic low-δ18O rocks, blue dots plutonic lithologies. Color- coding shows oxygen isotope compositions of modern meteoric precipitation based on Bowen and Revenaugh (2003). Note that these δ18O values for meteoric water are only applicable to settings with present-day magmatic activity (bold setting names; erupted in the last 1 Ma) and do not always account for regional topographic effects (e.g. meteoric water in Yellowstone estimated at -19 ‰). Black arrows point towards probable paleolatitude at the time of magmatic activity (Gregory et al., 2009; Pisarevsky et al., 2014; van Hinsbergen et al., 2015; Zhang et al., 2009); different models place Mesoproterozoic Australia at different paleolatitudes. Dark gray areas mark major large ignous provinces (LIP), which can be associated with low-δ18O magmas related to hotspot activity. lower). Following the initial discovery of plutonic and volcanic rocks commonly expressed in the relative delta notation δ18O, which com- that crystallized from magmas with large 18O-depletions in the Sey- pares its isotope ratio R with that of the standard reference material chelles (Taylor, 1968) and Yellowstone (Hildreth et al., 1984), an in- Vienna Standard Mean Ocean Water (V-SMOW): δ18 creasing number of localities with low- O silicic magmas has been 18 16 fi Rsample (O/O)sample identi ed around the world, and in many settings the interpretation of δ18O(‰)=−= [ 1]·1000 [ −1]·1000 18 18 16 the processes leading to low-δ O values is based on comparison with Rstandard (O/O)standard (1) the archetypical example of the Yellowstone volcanic field. For this Oxygen isotope fractionation describes the enrichment of one iso- study, we have compiled known low-δ18O settings (Fig. 1) and compare tope relative to another between two substances A and B in chemical the available data (electr. suppl. 1) in terms of their geotectonic en- equilibrium, expressed as the isotope fractionation factor vironment, field and textural relationships, major and trace elemental compositions, stable and radiogenic isotopes, and magmatic storage R (O/O)18 16 a ==A A parameters, in order to document the full range of low-δ18O magmas AB− 18 16 RB (O/O)B (2) and identify the parameters that promote interaction of magmas with hydrothermally altered crust. These information and data are by no Oxygen isotope fractionation factors are strongly temperature-de- means exhaustive, as new localities continue to be discovered and pendent and generally decrease towards higher temperatures by a 2 known settings investigated in more detail. The interested reader is factor of ~1/T (with T being the temperature in K). The difference in 18 referred to the literature referenced in the setting descriptions. Here, we δ O value between two materials is given by Δ, whereby review currently known key characteristics and interpretations of low- 18 18 18 ΔδδOAB−−=−≈ OA OBAB 1000 ln α (3) δ18O settings on Earth and provide a framework in which past and 18 18 future discoveries of low-δ O magmatism can be assessed. Mineral-melt equilibrium fractionation factors (Δ Omineral-melt) for mafic minerals are usually negative, so minerals such as , pyr- oxene and Fe-Ti oxides will have lower δ18O values than the melt, 2. Oxygen isotopes in magmatic systems leading to a progressive enrichment in 18O over 16O in magmas that undergo fractional crystallization of these phases. The vast majority of Oxygen has three stable isotopes 16O, 17O, and 18O, which typically unaltered mantle-derived have a narrow range of oxygen iso- occur in the rough proportions 99.76 % (16O), 0.04 % (17O) and 0.20 % tope compositions of 5.4 to 5.9 ‰ (Eiler, 2001 and references therein, (18O). The oxygen isotopic composition of water and silicates is Fig. 2). Considering typical fractionating mineral assemblages and

2 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

8.0 Calculated δ18O trends from LLD (Bucholz et al. 2017):

Bushveld Complex, 1 wt.% H2Oinitial, Van Tongeren et al. (2010) 7.5 18 high-δ O Ol-tholeiite, 3.6 wt.% H2Ointial, 0.7 GPa, Nandedkhar et al. (2014) normal-δ18O Dry Tholeiite, 1 GPa, Villiger et al. (2004) 7.0 Dariv Igneous Complex, hydrous high-K , Bucholz et al. (2014)

Ol-tholeiite, 0.4 wt.% H2Ointial, 0.4 GPa, 6.5 Whitaker et al. (2008) O (‰)

18 Kohistan Paleo-arc, Pakistan, Jagoutz (2010) δ Observed δ18O trends (Bindeman et al. 2004): 6.0 Ascension Island, Sheppard and Harris (1985) Hachijo-jima arc, Matsuhisa (1979) Anderson et al. (1971) low-δ18O 5.5 Galapagos spreading center, Muehlenbachs and Byerly (1982) MORB Tristan da Cunha, Harris et al. (2000) 5.0 Kamchatka, Bindeman et al. (2004) 45 55 65 75 85

SiO2 (wt%)

Fig. 2. Oxygen isotope trajectories for different magma differentiation pathways. Dashed lines indicate calculated δ18O trends from liquid lines of descent as presented by Bucholz et al. (2017). Here, an initial δ18O of +5.6 ‰ mantle-derived magma is assumed as a starting point for these trends. Liquid lines of descent are based on observations of natural systems (Bucholz et al., 2014; Jagoutz, 2010; Van Tongeren et al., 2010) or experimentally-derived (Nandedkar et al., 2014; Villiger et al., 2004; Whitaker et al., 2008). Solid lines represent observed δ18O trends in different settings (Anderson et al., 1971; Bindeman et al., 2004; Harris et al., 2000; Matsuhisa, 1979; Muehlenbachs and Byerly, 1982; Sheppard and Harris, 1985), as compiled in Bindeman et al. (2004). Gray area marks the approximate evolution of normal-δ18O magmas based on these models and observations; deviations would result in high- or low-δ18O magmas, respectively. fractionation factors, the oxygen isotopic composition of an evolving altered rocks, their mineral assemblage will re-equilibrate with the melt magma can be calculated along liquid lines of descent (e.g. Bucholz at magmatic temperatures. et al., 2017, compilation in Fig. 2). Magmas with δ18O values that In this paper, we focus exclusively on magmatic low-δ18O settings follow a trend of closed-system fractional crystallization are commonly where isotopic fractionation between different minerals or minerals and referred to as “normal-δ18O” magma, culminating in normal-δ18O glasses indicate that these are in high-temperature (> 600-700 °C) 18 magma with δ O values of +5.8 to +7.5 ‰ at SiO2 contents > 65 wt “magmatic” equilibrium and have not been modified by secondary al- %. These normal-δ18O trends depend on the exact differentiation teration. In most of the settings listed here, evidence for low-δ18O pathway and therefore may differ slightly between magmatic provinces magmatic values stems from analysis of multiple co-existing mineral (Fig. 2); the normal-δ18O range of +5.5 to +7.5 ‰ used here may phases and can usually be linked to supporting geochronological and therefore be slightly larger than other commonly cited ranges, which textural data. In a minority of the settings, zircons are the only record of may be based on a single magmatic setting. low-δ18O magmatism, either because only zircon analyses have been Open-system interaction with crustal rocks can lead to deviation reported so far, or because there is evidence that δ18O values of other from this normal-δ18O array, as crustal rocks have bulk δ18O values minerals may have shifted, for example due to later metamorphic anywhere between -10 and +32 ‰ (e.g. compilation in Bindeman, overprint or alteration. Zircon is extremely resistant to processes such 2008; Taylor, 1974). Magmas with lower δ18O values are designated as as high-temperature alteration, metamorphism and melting, and has “low-δ18O”, whereas those with higher values as “high-δ18O” magmas, exceptionally slow oxygen isotope diffusion rates (Peck et al., 2003). In respectively. Based on our current understanding of the different low-δ18O settings where oxygen isotopic data come solely from zircon, oxygen isotope reservoirs on Earth, high-δ18O magmas can be produced geochronological, textural and compositional information indicate that by assimilation of either sedimentary or carbonate rocks (δ18Oupto these zircons crystallized from low-δ18O magma. For better comparison +25 to 30 ‰) or of rocks altered at low temperatures (Taylor, 1968, of the different settings, we have recalculated magmatic δ18O values 1980; Valley et al., 2005). Low-δ18O magmas on the other hand are from different mineral δ18O values in the cited studies, which allows required to have inherited an isotopic signal from rocks that exchanged melt δ18O values to be calculated without having to estimate a mag- oxygen with surface water, since meteoric and sea water are the only matic temperature. Unless indicated otherwise, δ18O values stated reservoirs on Earth with 18O/16O ratios significantly lower than MORB throughout this text refer to the δ18O of the magma or melt as calcu- (Bindeman, 2008; Taylor, 1968). This exchange must occur during lated following the methods described in electr. suppl. 2. At magmatic high-temperature alteration, in order for hydrothermally altered rocks temperatures (~700-850 °C), δ18O values of quartz are commonly ~1 to approach the δ18O value of the altering fluid. Due to the large ‰ higher, δ18O values of feldspar are similar, and δ18O values of zircon fractionation of oxygen isotopes between rock and water at low tem- are ~2 ‰ lower than co-existing melt (electr. suppl. 2 and references 18 peratures (e.g. Δ O(granite-H2O) > 10 ‰ at 150 °C (Zhao and Zheng, therein). These fractionation factors are given here as examples to il- 2003)), low-temperature alteration would lead to higher δ18O in the lustrate the direction and extent of 18O-enrichment or depletion in altered rock, even when meteoric water with low δ18O values is in- minerals compared to their host melt, and may differ from those used to volved (e.g. Cerling et al., 1985; Donoghue et al., 2008). Efficient calculate melt δ18O values for the samples listed in electr. suppl. 1, lowering in the δ18O value of the altered rock therefore requires isotope many of which reflect magmas at higher pre-eruptive temperatures. exchange with isotopically light water at elevated temperatures, where 18 isotope fractionation is small (e.g. Δ O(granite-H2O) < 2 ‰ at 450 °C (Zhao and Zheng, 2003)). During melting or assimilation of these

3 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

3. Occurrences of low-δ18O silicic magmas pattern, with low-δ18O units being limited to the area near the Idaho batholith and to units younger than 14.5 Ma (cf. Ellis et al., 2013, their 3.1. Continental hotspot settings Fig. 13). Boroughs et al. (2005) suggested that low-δ18O CSRP form by melting of hydrothermally altered lithologies of the Eocene 3.1.1. Heise and Yellowstone volcanic fields, Idaho/Wyoming, USA Idaho batholith that extend into the hotspot track. Intrusion of these The Heise and Yellowstone volcanic fields are the two youngest granitoid bodies was accompanied by large-scale circulation of me- eruptive centers along the Yellowstone hotspot track, spanning an area teoric-hydrothermal fluids around the plutons, creating some of the from eastern Oregon, through southern Idaho and northern Nevada to largest low-δ18O alteration zones on Earth (Criss et al., 1984). Based on western Wyoming. Yellowstone National Park as the current location of thermal, temporal and volume constraints, Boroughs et al. (2012) ar- the Yellowstone hotspot is the site of active crustal deformation, intense gued that the larger volume of low-δ18O in the CSRP compared seismic activity and an extremely high heat flow of 30 times the con- to Yellowstone requires input from such pre-existing hydrothermally tinental average, resulting in abundant hydrothermal activity altered crust, which is then fully or partially melted. Other studies (Lowenstern and Hurwitz, 2008; Lowenstern et al., 2006; Smith and argue for a syn-magmatic origin of wide-scale hydrothermal alteration Braile, 1994). Silicic magmas in the Yellowstone and Heise volcanic that affected a variety of crustal rocks (Colón et al., 2018; Drew et al., fields show both normal- and low-δ18O values. Rhyolite magmas in 2013). After investigating hydrothermally altered rocks in the Idaho Yellowstone reveal a cyclic decrease in magmatic oxygen isotope batholith and Challis volcanic field and assessing their melting beha- compositions, with low-δ18O occurring after each of the three vior, Troch et al. (2020) suggest that low-δ18O signatures in CSRP caldera-forming eruptions in Yellowstone (Bindeman and Valley, 2000, rhyolites result from ~30 % assimilation of silicic rocks that were first 2001; Hildreth et al., 1984). Following the initial sharp decrease, δ18O affected by pre-existing Eocene hydrothermal alteration and later by values gradually recover towards normal-δ18O compositions as assim- syn-magmatic caldera-hosted hydrothermal alteration, leading to ex- ilation and melting target less altered rocks and/or assimilation rates tremely low δ18O values in the assimilated material. decrease, with all caldera-forming events erupting normal-δ18O rhyo- lite. Bindeman and Valley (2001) interpreted this cyclic evolution as 3.1.3. Talbot rhyolites, Musgrave Province, Australia “bulk cannibalization” and wholesale melting of previously deposited The large-volume (> 7800 km3 erupted volume) Mesoproterozoic and subsequently hydrothermally altered caldera infill, while a more (1080-1040 Ma) Talbot rhyolites erupted through the thinned litho- recent characterization of the altered caldera infill suggest < 30 % as- sphere of the Musgrave Province in central Australia may represent the similation of partially melted caldera infill (Troch et al., 2018). Both in geochemically most similar system to the Snake River Plain- 18 18 18 the Heise and Yellowstone volcanic fields, a general decrease in magma Yellowstone low-δ O province. Here, low-δ O zircons (δ Ozircon δ18O values can be observed throughout the lifetime of the magmatic +0.8 to +4.3 ‰) were found in high-temperature ferroan rhyolites province, with the lowest-δ18O material being erupted in the youngest that are associated with basaltic magmatism (Smithies et al., 2015). In caldera-forming eruption in Heise (Kilgore Tuff, δ18O +3.4 ‰, contrast to the Yellowstone hotspot track, the locus of Talbot activity Bindeman et al., 2007; Watts et al., 2011) prior to recovery towards remained stationary for > 30 Ma, producing a large number of super- normal-δ18O compositions (Ellis et al., 2017), and the lowest-δ18O imposed caldera centers (Smithies et al., 2015a), in which syn-genetic erupted after the youngest caldera-forming eruption in Yellowstone hydrothermal activity led to low-δ18O alteration of previously erupted (Bindeman et al., 2008; Bindeman and Valley, 2001; Hildreth et al., volcanic deposits. Talbot rhyolites also differ from Yellowstone in that 1984). These trends are interpreted to be the product of several cycles the earliest preserved units are among the most 18O-depleted and vol- of caldera collapse and ingestion of hydrothermally material, sug- canism trends to less 18O-depleted compositions through time (Smithies gesting that the lowest-δ18O magmas form towards the end-stage of et al., 2015). Different paleomagnetism models place Australia at dif- nested caldera complexes (Bindeman and Simakin, 2014; Bindeman ferent paleolatitudes (Pisarevsky et al., 2014), but generally agree that and Valley, 2001; Bindeman et al., 2007; Watts et al., 2011). Australia was located at mid latitudes > ± 30° (Fig. 1) at the time of magmatic activity. 3.1.2. Central Snake River Plain, Idaho-Nevada, USA eruptive centers in the Central Snake River Plain (CSRP) 3.2. Other settings related to mantle upwelling constitute older sections of the Yellowstone hotspot track, which we have separated here from the younger Heise and Yellowstone volcanic 3.2.1. Caldera centers in Iceland fields here due to the slightly different processes involved in the for- Magmatism in Iceland relates to both rifting along the mid-Atlantic mation of low-δ18O magmas. The Bruneau-Jarbidge and Twin Falls spreading center and the activity of a hotspot. As rifting and hotspot volcanic fields have produced an enormous volume (> 20,000 km3 activity coincide in many of settings listed in the following, we group erupted material) of silicic intensely welded ignimbrites and lavas be- them here as generally related to mantle-upwelling. In contrast to other tween 13-8 Ma (Ellis et al., 2013), although occasional deposits may be low-δ18O settings, oxygen isotope anomalies in Iceland are found both slightly younger. Outcrops of these units can be found on both the north in silicic and mafic magmatic rocks, e.g. rift zone-related tholeiites and south side of the Snake River Plain, whereas the eruptive centers (Hemond et al., 1988) or large-volume silicic units from the volcanoes themselves are covered by thick sequences of younger basalts Askja, Hekla and Torfajökull (Bindeman et al., 2012). It has long been (Bonnichsen and Godchaux, 2002). Extreme welding and rheo- debated whether these isotope anomalies reflect changes in mantle morphism of ignimbrites in the CSRP has, among other characteristics, composition or different amounts of crustal input (Hemond et al., 1988 been suggested to represent a distinctive and unusual high-temperature and references therein). However, since the deviation towards lower type of silicic volcanism, so-called “Snake River-type” volcanism δ18O values is more prominent in silicic magmas (Pope et al., 2013), we (Branney et al., 2008). favor the interpretation of rhyolite formation as the result of upper- Both ignimbrites and lavas in the CSRP have δ18O values between crustal processes. -1.2 and +4.8 ‰ (Bindeman and Simakin, 2014; Boroughs et al., 2005; Most low-δ18O silicic magma in Iceland is found at large volcanic Boroughs et al., 2012), distinctly lower than older eruptive centers centers along the main axial rift system (Krafla, Askja, Torfajökull, and associated with the Yellowstone plume in Oregon and Nevada (e.g. Hekla), which also produce the largest volumes of silicic magma in Colón et al., 2015; Seligman et al., 2014) or in the younger Yellowstone general. Compared to other volcanic hotspot and rift settings (Table 1), and Heise volcanic fields (volume-weighted average δ18O of +5.3 ‰ however, the total volume of low-δ18O silicic magma is relatively small. (Boroughs et al., 2012)). Oxygen isotopic composition of rhyolites Both Krafla(δ18O +1.0 to +3.3 ‰) and Torfajökull represent promi- along the Yellowstone hotspot track follow both a spatial and temporal nent silicic centers with of 10 and 12 km diameter,

4 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

Table 1 Settings with low-δ18O silicic magmas in this study, see text for references. References for calculations of volumes are provided in electronic supplement 2.

Low-δ18O locations Activity vol low-δ18O lowest Tectonic setting Caldera Volc/plutonic Alteration δ18O*

Central Snake River Plain, USA 13 - 8 Ma > 20,000 km3 1.5 ‰ continental hotspot multiple nested volcanic pre-ex. + syn- calderas magm. Yellowstone/Heise, USA since 7 Ma ~4,400 km3 1.1 ‰ continental hotspot ≥ 3 nested calderas/ volcanic syn-magmatic center Talbot rhyolites, Australia 1080 - 1040 Ma ~22,000 km3 3.7 ‰ continental hotspot yes volcanic syn-magmatic Iceland since 10 ka < 250 km3 1.7 ‰ mantle upwelling ≤ 3 nested calderas/ volcanic syn-magmatic center Isle of Skye/Mull, Scotland 60-53 Ma < 64 km3 1.9 ‰ mantle upwelling plutonic syn-magmatic Kangerlussuaq intrusion, Greenland 50 Ma < 2,000 km3 2.3 ‰ mantle upwelling plutonic syn-magmatic Lebombo, South Africa 183 - 180 Ma > 30,000 km3 4.7 ‰ mantle upwelling likely volcanic syn-magmatic Koegel Fontein, South Africa 144 - 133 Ma < 1 km3 -0.3 ‰ mantle upwelling subvolcanic pre-existing Seychelles, Indian Ocean 750 Ma < 40 km3 2.4 ‰ ? plutonic pre-existing Tenerife, Canary Islands < 12 Ma < 30-210 km3 2.0 ‰ mantle upwelling ≥ 3 calderas volcanic syn-magmatic Malani igneous suite, NW India 780 - 750 Ma ? 4.2 ‰ mantle upwelling likely volcanic ? Imorona-Itsindro suite, Madagascar 850 - 750 Ma ? 6.2 ‰+ mantle upwelling plutonic ? Metagranites Dabie-Sulu orogen, 870-600 Ma ? 2.4 ‰ mantle upwelling plutonic syn-magmatic China New England Avalon granites, USA 600 - 550 Ma ? 6.7 ‰+ mantle upwelling plutonic pre-existing Timber Mountain complex, USA 16 - 9 Ma ~900 km3 5.3 ‰ extension/upwell. ≥ 5 nested calderas volcanic syn-magmatic Kamchatka, Russia 6 Ma to recent ~500 km3 4.9 ‰ volcanic arc yes volcanic syn-magmatic Aleutians/Alaska, USA 10 Ma to recent ~100 km3 4.6 ‰ volcanic arc 1-2 calderas volcanic syn-magmatic Mount Mazama/Crater Lake, USA 6845 ± 50 yr BP < 50 km3 5.3 ‰ volcanic arc yes plu. fragments syn-magmatic Calabozos, < 1 Ma to < 1,000 km3 5.3 ‰ volcanic arc w/ ext. ≥ 2 nested calderas volcanic syn-magmatic recent Western Nevada volcanic field, USA 35 - 19 Ma 300-550 km3 7.1 ‰+ volcanic arc ≥ 7 nested calderas volcanic syn-magmatic Casto pluton, USA 44 - 47 Ma < 100 km3 2.8 ‰ volcanic arc w/ ext. yes plutonic syn-magmatic ♦ Gyeongsang granites, SE Korea 88 - 27 Ma ? 6.5 ‰ w/ ext. plutonic ? Mesozoic granites, eastern China 167 - 110 Ma ? 4.9 ‰ subduction w/ ext. plutonic pre-existing Baekdusan/Changbaishan, North AD 940 < 24 km3 6.4 ‰ volcanic arc yes volcanic syn-magmatic Korea*

* to avoid single very low values and outliers, we use the lowest decile of the δ18O range; + affected by generally high δ18O in the field, the lowest unit/sample is at ♦ 5.6 ‰ in Nevada, 5.5 ‰ in Madagascar, and 4.2 ‰ for the New England granites; techniqually not a low-δ18O province according to definition in this study. respectively, with abundant hydrothermal activity (Soosalu and ‰, Hattori and Muehlenbachs, 1982; Sveinbjornsdottir et al., 1986)or Einarsson, 2004). Askja (δ18O -1.2 to +3.2 ‰ (Condomines et al., sampled by xenoliths (δ18O -9.9 to -4.7 ‰, Muehlenbachs et al., 1974). 1983; Martin and Sigmarsson, 2010; Sigmarsson et al., 1991)) consists of three nested calderas (Sturkell et al., 2006). Off-rift volcano Hekla 3.2.2. Isle of Skye and Isle of Mull granites, Scotland (δ18O +4.9 to 5.3 ‰ (Schattel et al., 2014; Sigmarsson et al., 1992; Compositionally bimodal magmatism on the Isle of Skye and Isle of Sigmarsson et al., 1991)) is a dominated by basaltic an- Mull, Scotland (60-53 Ma), belongs to the British Igneous desite. Low-δ18O values (δ18O down to +1.7 ‰) have also been magma Province, which is part of the North Atlantic Igneous Province con- reported from zircons from extinct silicic volcanic centers in eastern sidered to be related to the onset of the Iceland hotspot and the opening and western Iceland (Carley et al., 2017; Carley et al., 2020). of the North Atlantic (Fig. 1). Following the initial discovery of large- Early publications suggested that many of the silicic magmas in scale hydrothermal systems (Forester and Taylor, 1976; Taylor and Iceland are dominantly derived from fractional crystallization of mafic Forester, 1971), low-δ18O silicic magmas have been identified from parental melts with subordinate assimilation of crustal material (e.g. zircons. Zircon δ18O values in different granite bodies on Skye (~64 Carmichael, 1964; Furman et al., 1992; MacDonald et al., 1990; km2) range from +0.6 to +5.3 ‰ (Gilliam and Valley, 1997; Monani Nicholson et al., 1991), whereas more recent works prefer them to and Valley, 2001), with minimal intra-grain variability. Zircons from originate from partial melting of hydrothermally altered basalts or the Isle of Mull granites have δ18O values between +3.4 to +5.9 ‰ other crustal rocks (Bindeman et al., 2012; Gunnarsson et al., 1998; (Monani and Valley, 2001). For some samples in these studies, oxygen Gurenko et al., 2015). Other studies suggested variable combinations of isotope ratios in quartz are in high-temperature equilibrium with those crustal melting and fractional crystallization in different temporal order in zircon, however, in most of them, δ18O values are lower than (e.g. Eiler et al., 2000; Elders et al., 2011; Hemond et al., 1988; Pope Qtz δ18O , indicating later high-temperature hydrothermal alteration, et al., 2013; Sigurdsson and Sparks, 1981). Martin and Sigmarsson Zrc which is also evident from extremely low whole-rock δ18O values of (2007) proposed that the variable degrees of inferred crustal input may down to -3.9 ‰ (Monani and Valley, 2001). Monani and Valley (2001) be directly linked to the regional thermal state, with areas of elevated attribute low-δ18O silicic magmas to assimilation or partial melting of heat transfer allowing for larger degrees of crustal melting. continental crust that was variously altered at high temperatures during Bimodal volcanism and intense hydrothermal activity are typical preceding mafic magmatism. features of central volcanic complexes in Iceland, such as the Krafla central volcano (Pope et al., 2013; Sigmarsson et al., 1991). At Krafla, δ18O values of rhyolite lavas are between +1.0 and +3.3 ‰ 3.2.3. Kangerlussuaq intrusion, Greenland (Nicholson et al., 1991; Pope et al., 2013), lower than those of basalt The Palaeogene Kangerlussuaq intrusion (ca. 50 Ma) is also related from the same volcano (δ18O +3.9 to +4.5 ‰). Rhyolite magmas are to magmatism in the North Atlantic Igneous Province. The intrusion thought to be derived by partial melting of hydrothermally altered was emplaced into Archean gneiss and Palaeogene flood basalts and basalt (Jónasson, 1994; Pope et al., 2013; Zierenberg et al., 2013), shows a gradual transition from alkaline nepheline-syenitic composi- which is similar to lithologies recovered in drillcores (δ18O -10.5 to -3.4 tions in the center towards quartz-syenitic compositions at the margin (Riishuus et al., 2008). This transition is accompanied by a continuous

5 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299 decrease in mineral δ18O values, suggesting that the outermost parts of subsequently hydrothermally altered by meteoric water in a caldera the intrusion (δ18O > 3.2 ‰) crystallized from a low-δ18O magma setting, and that rhyolite magma in this environment results from low- (Riishuus et al., 2015). Quartz-syenites at the intrusion margin contain degree melting of basaltic precursors. Evidence for caldera collapse abundant blocks of hydrothermally altered flood basalt (bulk rock δ18O structures has not yet been reported, although the large inferred vo- -3.2 to +4.6 ‰), which are thought to have dehydrated without sub- lumes (30,000 km3 erupted material; Table 1) suggest it is very likely stantial melting or assimilation during their incorporation in the that they accompanied eruptions. magma. To avoid unrealistically large amounts of entrainment, water fluxing from dehydration of stoped basaltic blocks requires that the 3.2.6. Koegel Fontein complex, South Africa basalts contained abundant low-δ18O meteoric water in pore space or The Cretaceous Koegel Fontein igneous complex formed at 144-133 that the alkaline magma already contained a significant component of Ma during the initial rifting prior to break-up of the African and South partially melted hydrothermally altered basalt (Riishuus et al., 2015). American continents (De Beer and Armstrong, 1998). It is the only Early oxygen isotope studies recognized wide-scale hydrothermal known low-δ18O Cretaceous complex in southern Africa, whereas con- alteration and lowering in δ18O in basalts at the margins of the temporary intrusive complexes in Namibia do not exhibit comparable Skaergaard intrusion (Taylor, 1968; Taylor and Forester, 1979), a oxygen isotopic compositions. The main igneous body consists of the 18 layered mafic intrusion in eastern Greenland. Similar to what has been 15-km diameter Rietpoort Granite (δ OQtz +8.3 ± 1.0 ‰,n=7 proposed for the Kangerlussuaq intrusion, remelting of stoped, low- (Harris et al., 2018)). The overlying gneissic roof pendant is intruded by δ18O hydrothermally altered blocks has been suggested to be re- several quartz-porphyry dikes that show large variability in oxygen 18 18 sponsible for the formation of ferrodioritic magmas with δ O values isotopic compositions (δ OQtz -2.3 to +5.6 ‰) and reach extremely between +3 to +4 ‰ (Bindeman et al., 2008; Wotzlaw et al., 2012). low δ18O values where they crosscut a shear zone within the gneiss However, with the Skaergaard intrusion being dominated by mafic (Curtis et al., 2013; Harris et al., 2018). Based on the petrographic compositions and volumetrically minor proportions of granophyre and features and isotopic homogeneity of quartz within a given sample, pegmatite, we do not consider it further in this review of low-δ18O si- these δ18O values are thought to reflect extremely low-δ18O melts licic magmas. (Harris et al., 2018) derived from selectively melting the most altered shear zone-related portions of the hydrothermally altered gneiss. 3.2.4. Las Cañadas, Tenerife, Canary Islands Sheared gneiss in the area have bulk-rock δ18O values as low as -2 ‰. Tenerife is currently the only ocean island setting described to Due to the correlation of oxygen with strontium isotopes, the alteration feature low-δ18O evolved magmas. Together with the Kangerlussuaq event is inferred to significantly predate melting, and possibly took intrusion, it is the only setting of silica-undersaturated character. The place during a global glaciation event at 550 Ma (Harris et al., 2018). origin of magmatic activity in the Canary Islands is complex but gen- erally related to mantle upwelling (Carracedo et al., 1998). Tenerife’s 3.2.7. Seychelles, Indian Ocean early composite mafic alkaline shield formed between 12 and 3 Ma, The Seychelles are largely composed of granite, in- with younger activity being characterized by basaltic to phonolitic terpreted to reflect the origin of the Seychelles as a continental frag- compositions erupted from the Las Canadas edifice in at least three ment from the Gondwana break-up (Davies and Francis, 1964; Du Toit, caldera cycles (Martí et al., 1994). Whole-rock powders of interbedded 1937). The 750-Ma Seychelles granites were the first documented oc- phonolitic pyroclasts and basanite lavas in the Diego Her- currence of low-δ18O silicic magma (Taylor, 1968, 1977). In the Praslin nandez Formation were found to have low δ18O values (Wolff et al., granite suite, which is exposed on several of the north-eastern islands of 2000). Powders of apparently fresh groundmass separates reach values the Seychelles, whole-rock oxygen isotopic compositions range from as low as 0 ‰, whereas feldspar separates range from 5.4-7.9 ‰, -1.2 to +7.5 ‰, with minerals in these samples indicating isotopic consistent with results in Wiesmaier et al. (2012). We note that so far, equilibrium at magmatic temperatures (Harris and Ashwal, 2002). This no analysis of co-existing mineral separates has excluded the possibility is much lower than the 5.25 ± 0.65 ‰ range of the Mahé granite suite of post-eruptive exchange of the groundmass, suggesting that Tenerife’s considered to be consistent with differentiation of mantle-derived mafic designation as a low-δ18O setting should perhaps be treated with cau- magma (Harris and Ashwal, 2002). The low δ18O values of the Praslin tion. The Diego Hernandez formation is estimated to have a cumulative granite suite has been interpreted as being derived from interaction of erupted volume of 30-210 km3 (Edgar et al., 2007), of which low-δ18O the silicic melts with older crustal rocks that were hydrothermally al- magma would only constitute a small fraction. (Wolff et al., 2000) at- tered well before granite emplacement (Harris and Ashwal, 2002). tribute low δ18O values to assimilation of hydrothermally altered crust, These authors suggest that due to the very low δ18O values, hydro- similar to the explanation of Harris et al. (2000) for significant oxygen thermal alteration must have taken place at higher latitudes where isotope variability in basanitic to phonolitic magmas on the ocean is- meteoric water has lower δ18O than prevalent at the tectonic position of land Tristan da Cunha, although these are not low-δ18O according to the Seychelles at the time of granite formation, possibly related to si- our definition. milar Snowball Earth events as other Neoproterozoic low-δ18O settings (Rumble et al., 2002). Ashwal et al. (2002) suggested that chron- 3.2.5. Lebombo rhyolites, South Africa ological, petrological and chemical evidence supports an origin in a The normal- to low-δ18O rhyolites of the Lebombo Monocline are continental or Andean-type arc setting for magmatism in the Seychelles, part of the Karoo volcanic province (183-180 Ma, Allsopp et al., 1984; however, other works have proposed an extensional hotspot or rift-re- Duncan et al., 1997; Riley et al., 2004), a large igneous province that is lated setting (Plummer, 1995; Weis and Deutsch, 1984), reflecting the associated with the Mesozoic break-up of Gondwana and activity of the late Proterozoic break-up of Rodinia. In our figures, we group the Marion Island hotspot (Richards et al., 1989). The erupted volume of Seychelles as associated with mantle upwelling related to rift or hotspot this rhyolite group is estimated to about 30,000 km3, and magmatic activity due to their similarity with synchronous settings. δ18O values have been determined to be on average +5.6 ‰, but reaching down to +4.4 ‰ (Harris and Erlank, 1992). The rhyolites are 3.2.8. Malani igneous suite, NW India high-temperature ignimbrites whose lava-like features (Cleverly et al., The Malani Igneous Suite (780-750 Ma) is a compositionally bi- 1984) are indicative of extensive welding and rheomorphism. As ob- modal (rhyolitic-dacitic and basaltic) large-volume silicic complex in served in other low-δ18O settings, the Lebombo rhyolites do not contain the Jodhpur area in northwestern India (Wang et al., 2017), covering an hydrous (Cleverly et al., 1984; Harris and Erlank, 1992). area of 55,000 km2. At present, the only evidence for low-δ18O mag- Miller and Harris (2006) suggest that the low-δ18O signatures are de- matism stems from low-δ18O highly radiogenic (δ18O -1.1 to +8.2 ‰, rived from melting of previously deposited rhyolite deposits that were εHf +13.0 to +3.6) zircons (Wang et al., 2017), which have been

6 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299 interpreted as reflecting high-temperature bulk cannibalization of al- ignimbrites have normal to high δ18O values (δ18O 7.1-9.0 ‰) tered juvenile mafic crust. Due to the absence of xenocrystic zircons (Bindeman and Valley, 2003). Lipman and Friedman (1975) showed with low δ18O values, these authors suggest that the heat source for that oxygen isotope compositions become systematically lighter with hydrothermal alteration also produced the low-δ18O rhyolites. Like the decreasing age in cogenetic sequences, with feldspar δ18O values Seychelles, the Malani igneous suite is thought to be associated with the reaching down to +5.0 ‰. While these early studies suggest that late Proterozoic break-up of the Rodinia supercontinent, and thus po- magmas interacted directly with meteoric water (cf. Friedman et al., tentially related to the same rifting episode that produced low-δ18O 1974), later studies advocate that variations in Sr, Nd and O isotopes granites in south China and Madagascar, which are described in the are due to 20-40 % assimilation of hydrothermally altered wallrock following. (Farmer et al., 1991). By analogy with their proposed model for Yel- lowstone, Bindeman et al. (2006) suggest that the observed isotopic 3.2.9. Imorona-Itsindro Suite in central Madagascar diversity in zircons and magmas is due to recycling of buried hydro- The Imorona-Itsindro magmatic suite consists of compositionally thermally altered roof material in a series of four overlapping caldera bimodal intrusive bodies of granitoid, syenite and gabbro exposed over centers. During this process, altered material is thought to be “bulk an area of several hundred square kilometers in central Madagascar. digested” due to progressive melting, leading to continued lowering of While most of the suite was overprinted by high-grade metamorphism, δ18O values of magma that was originally high in δ18O. the presence of low-δ18O magmas has been implied from the finding of δ18 ‰ ff zircons with O < 4.2 (Archibald et al., 2016) that di er from 3.2.12. New England Avalonia granites, USA δ18 more common high- O zircons in the same lithological domain. As is Low-δ18O zircons have also been recovered from granites in the the case for the Seychelles, there is some disagreement of whether New England Avalon zone, namely the 586 ± 7 Ma Hope Valley magmatism recorded in the Imorona-Itsindro suite in Madagascar is Gneiss in Connecticut (USA) and the 413 ± 2 Ma Quincy related to arc magmatism due to the subduction of oceanic crust be- Granite in Massachusetts (Fu et al., 2014). While δ18O values of co- neath Madagascar (Kroener et al., 2000), to continental rifting (Kabete existing minerals such as quartz were modified during recrystallization et al., 2006), the activity of a mantle plume (Yang et al., 2015; Zhou associated with younger metamorphism, these low-δ18O zircons et al., 2015) or a combination of these processes. We group this setting 18 (δ Ozircon > +2.3 ‰) are interpreted to be of igneous origin and to here as generally related to mantle upwelling. record partial melting of crustal lithologies of the Avalonia terrane (Fu et al., 2014), which experienced pervasive hydrothermal alteration by 3.2.10. Metagranites in the Dabie-Sulu orogen, east-central China meteoric water during Neoproterozoic rifting from Gondwana (600-550 δ18 The widespread low- O rocks in the Dabie-Sulu orogen in eastern Ma, Potter et al., 2008). China represent probably one of the most complex settings for low-δ18O magmatism. Eclogites, gneisses, quartzites and schists in this area are extremely low in δ18O, with mineral δ18O values as low as -7.7 ‰ for 3.3. Volcanic arc/subduction quartz and -10.1 ‰ for garnet (Rumble and Yui, 1998; Zheng et al., 1996; Zheng et al., 1999). These rocks are thought to represent an 3.3.1. Caldera centers in the Kamchatka volcanic arc, Russia extensive Neoproterozoic (758 ± 15 Ma) hydrothermal system related The multi-caldera centers Ksudach, Karymsky and Kurile to rifting and break-up of Rodinia that was subsequently metamor- Lake-Iliinsky, and Late caldera centers Maly Semyachik, δ18 δ18 phosed at coesite-eclogite facies conditions during subduction Akademy Nauk, and Uzon all feature low- O eruptive deposits ( O ‰ of the South China Block beneath the North China Block (Rumble et al., down to +4.5 ), most of which are voluminous silicic and rhyoda- 2002; Zheng et al., 2004). With peak metamorphic conditions estimated citic ignimbrites and caldera-related lavas and domes (Bindeman et al., δ18 at 3.0-4.5 GPa and 700-850 °C (Zhang et al., 1995), this overprint led to 2004). By contrast, only few low- O volcanic deposits are found at re-equilibration of mineral δ18O values at seemingly magmatic tem- stratovolcanoes, and basaltic magmas at stratovolcanoes tend to be δ18 ’ peratures. In-situ U/Pb dating and analysis of oxygen isotopic compo- normal- to high- O, with Klyuchevskoy volcano erupting the world s δ18 δ18 sitions allows distinction of metamorphic overgrowth rims (200-259 highest- O (Auer et al., 2009). Low- O magmas are Ma, δ18O -10.0 to -2.2 ‰) from original magmatic cores (600-800 Ma, thought to be derived from re-melting of older crust that was hydro- δ18O -0.9 to +10.1 ‰) in zircon from gneisses and eclogites (Chen thermally altered during a 2.6-million year-long Pleistocene glaciation et al., 2011; Fu et al., 2013). Compared to eclogites in the area, the period (Bindeman et al., 2004). Mass-balance models indicate that lower δ18O values of Neoproterozoic magmatic zircon cores in many oxygen and strontium isotope data require input of ~15-25% crustal δ18 ‰ ‰ metagranites suggests that these were low-δ18O granites prior to me- material with O from -10 to +20 . tamorphism (Chen et al., 2011). Although Zheng et al. (2007) suggest that low-δ18O magmatism did not begin until 760-750 Ma, Yang et al. 3.3.2. Caldera centers in the Aleutian volcanic arc, Alaska (2016) note that the first low-δ18O zircons appeared around 870 Ma in Although not as common as in Kamchatka, low-δ18O silicic magmas the south-eastern Yangtze Block, about 30 Ma prior to their first ap- have been reported for two Quaternary caldera centers in the Aleutian pearance in northern Yangtze, indicating that low-δ18O magmatism volcanic arc in Alaska, USA, the Fisher caldera volcano on Unimak followed synchronous rifting and associated hydrothermal activity. In Island and Okmok volcano on Umnak Island (Bindeman et al., 2001; our compilation, we only use δ18O values of magmatic zircon cores with Finney et al., 2008). The Aleutian arc extends from the Alaskan main- ages between 900 and 600 Ma, which are most likely to be undisturbed land to the Kamchatka volcanic arc and formed as a consequence of by metamorphism. northward subduction of the Pacific Plate underneath the North American Plate. It is built on continental crust to the east and oceanic 3.2.11. Timber Mountain caldera complex in Nevada, USA crust to the west, with Okmok being situated close to this transition on The Timber Mountain caldera complex is hosted within the south- oceanic crust and Fisher caldera on continental crust. At both volca- western Nevada volcanic field (16-9 Ma), which is one of the largest noes, basalts are normal in δ18O and low-δ18O values are restricted to centers of silicic magmatism in the western US and produced more than , and erupted in caldera-forming eruptions 4,000 km3 silicic magma (Byers et al., 1989; Christiansen et al., 1977). and associated pre-caldera units (δ18O > 4.6 ‰ at Fisher (Bindeman Regionally, magmatism is related to extension in the Great Basin fol- et al., 2001), δ18O 4.4 to 4.9 ‰ at Okmok (Finney et al., 2008)). lowing the subduction of the Farallon plate (Lipman et al., 1972). With Magmatic low-δ18O values are thought to be due to assimilation of a δ18O value of 5.4-6.0 ‰, the trachyandesitic to rhyolitic Ammonia earlier magmatic products that were extensively altered by glaciation- Tanks Tuff (900 km3) is a low-δ18O unit, whereas other zoned derived meteoric water (Bindeman et al., 2001; Finney et al., 2008).

7 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

3.3.3. Mount Mazama/Crater Lake in Oregon, USA that crystallized between 44-47 Ma (K/Ar-ages on biotite) in the end The volcano Mount Mazama in Oregon is part of the Cascade vol- stage of magmatic activity in the Challis volcanic field (Fisher et al., canic arc in the northwestern US and has erupted mainly andesitic to 1992). The high-silica Casto pluton intrudes into earlier erupted and dacitic compositions. Overall, the modern Cascade volcanic arc with its strongly altered volcanic units of the Van Horn and Thunder Mountain low overall magmatic output has not produced abundant low-δ18O cauldron complexes of the Challis volcanic field (Criss et al., 1984; Criss magmas (Borg et al., 1997), and Mt. Mazama is the only volcanic center and Taylor, 1983), sedimentary sequences, and the Idaho batholith. in the Cascades where low δ18O values have been reported. Oxygen Although there is no direct relationship with any erupted volcanics, the isotope compositions for unaltered volcanic rocks erupted at Mount Casto granites are interpreted to record a magmatic system that is si- Mazama are usually between +5.8 and +7.0 ‰ (Bacon et al., 1989), milar to the reservoir that fed the ignimbrite eruptions in the Challis whereas rhyodacite erupted in the caldera-forming eruption volcanic field (Hardyman, 1985; Jellinek, 1994; Larson and Geist, (6845 ± 50 yr BP, 50 km3 volume dense-rock equivalent, Bacon, 1983) 1995). Initial magmatic δ18O values are thought to range from +2.7 to contains glass and plagioclase with δ18O<5 ‰. Co-erupted with this +5.4 ‰ (Criss et al., 1984; Criss and Taylor, 1983; Larson and Geist, material are granitoid blocks with δ18O values as low as -3.4 ‰, which 1995). Isotope fractionation between minerals in the same samples have been interpreted as hydrothermally altered wallrock material that suggest that these values reflect magmatic compositions, although we was partially fused and re-equilibrated at magmatic temperatures note that some samples in these studies have been overprinted by al- 18 (≥900 °C, Bacon et al., 1989). U/Pb and U/Th dating of zircons from teration. We have therefore omitted all samples with Δ Oquartz-alkali these blocks reveal ages between 101 to 174 ka, suggesting that the feldspar >3‰ from our database. Larson and Geist (1995) suggest that material is remobilized from granodioritic plutons related to precursor the low-δ18O magmas of the Casto plutons formed by 32-55 % assim- andesitic-to-dacitic magmatism (Bacon and Lowenstern, 2005; Bacon ilation of intracaldera material (δ18O -8.8 to -1.6 ‰) that was hydro- et al., 2000). Oxygen isotope diffusion modelling suggests that hydro- thermally altered by meteoric water at high temperatures. thermal alteration occurred over a period of 1000-63000 years, and melting of the granitoid blocks occurred at least 10-200 years prior to eruption (Ankney et al., 2017). 3.3.7. Granitoid complexes in the Gyeongsang continental arc, SE Korea 18 18 Slightly O-depleted zircons (δ Ozircon > +4.2 ‰) have been re- 3.3.4. Calabozos caldera complex, Southern Andes ported from eight Cretaceous to (88-27 Ma) calc-alkaline to The Pleistocene-to-recent volcanic center lies within a transition alkaline granitoid complexes in the Gyeongsang continental arc in zone of thinning crust in the Southern Volcanic Zone and has produced southeastern Korea (Jo et al., 2016). Small core-to-rim decreases of three voluminous ash flow tuffs. Whole-rock and plagioclase oxygen δ18O in zircons (< 1 ‰) from granites with high δ26Mg in biotite are isotopic compositions of these andesitic to rhyodacitic tuffs are between thought to reflect increasing crustal contamination by weathered and 5.0 and 6.3 ‰, relatively low compared to regionally erupted basalts hydrothermally altered supracrustal rocks or hydrothermally altered (Grunder, 1987). Oxygen and Sr isotope data has been inferred to re- volcanic caldera roof (Jo et al., 2016). cord early-stage assimilation of high-δ18O crustal material and late- stage assimilation of 5-30 % hydrothermally altered older volcaniclastic wall and roof material at δ18O between 0 and -4 ‰. A similar inter- 3.3.8. Mesozoic granites in eastern China pretation of variable assimilation of heterogeneous low- and high-δ18O Oxygen isotopic compositions of zircons range from δ18O +3.1 to crustal material is made by Feeley and Sharp (1995) in order to explain +5.4 ‰ in Mesozoic A-type granites (Nianzishan, Shanhaiguan, oxygen isotope data from the volcano Ollagüe in the Andean Central Laoshan, Suzhou and Kuiqi granite) in eastern China (Wei et al., 2008; Volcanic Zone. Wei et al., 2002). In most samples, zircons are in high-temperature equilibrium with quartz, indicating that the low-δ18O values are a 3.3.5. Western Nevada volcanic field, USA primary magmatic signal (Wei et al., 2008). These authors propose that The Stillwater caldera complex (30-25 Ma) is part of the western low-δ18O magmas are derived from partial melting of pre-existing Nevada volcanic field (35-19 Ma) and is the only caldera complex in the crustal rocks that were altered by meteoric or seawater. volcanic field to have produced low-δ18O magmas (Watts et al., 2019). Zircons recording several stages of crustal recycling and processing Volcanic activity in this nested caldera settings has produced ~7 ig- have been described in Jurassic granites (Laoshan and Jingshan nimbrites, two of which are thought to be low-δ18O, the tuff of Job granite) from the Bengbu uplift between the Dabie and Sulu orogens in Canyon and Poco Canyon (δ18O +5.5 to +6.0 ‰, Watts et al., 2019). east-central China (Wang et al., 2013). Normal-δ18O Neoproterozoic We note that these fall into the normal-δ18O range used in this paper, zircon cores are overgrown by heterogeneous (δ18O between -9.4 and but are significantly lower than the majority of ignimbrites in the +8.5 ‰) Triassic zircon domains (Wang et al., 2013), resulting from western Nevada volcanic field, which are characterized by relatively the same high-grade metamorphism that resulted in wide-spread high δ18O values of +7 to +10 ‰ (Watts et al., 2016; Watts et al., oxygen isotopic equilibration in the Dabie-Sulu orogeny (section 3.3.5). 2019). This ignimbrite flare-up is thought to be related to steepening or These domains are overgrown by Jurassic mildly 18O-depleted com- roll-back of the subducting Farallon plate beneath the North American positions (δ18O 3.3 ± 0.5 ‰). Wang et al. (2013) interpret the low- plate, and thus is thought to predate significant extension responsible δ18O Jurassic magmas to result from lower- to middle-crustal anatexis for magmatism in the younger Timber Mountain caldera complex in the of Neoproterozoic altered crustal rocks that experienced high-degree south western Nevada volcanic field (section 3.2.11). The ignimbrites metamorphism in the Triassic. We use only Jurassic zircon δ18O values are variable in crystal content (from < 15 to > 35 vol%) and in com- in our compilation. 18 position, ranging from trachydacitic to rhyolitic (63-78 wt% SiO2). Zircons with δ O values between +1.8 and +5.2 ‰ have been Dissection and tilting during late Cenozoic Basin and Range extension reported from Cretaceous (~116 Ma) metaluminous to weakly per- provides rare exposure of the subvolcanic and plutonic roots of the aluminous granites, granodiorites and diorites in the Xiaocuo granitoid Stillwater volcanic complex, including a cross section through hydro- complex in south-eastern China (Li et al., 2015) and from Early Cre- thermally altered caldera infill (John and Pickthorn, 1996; Watts et al., taceous alkaline A-type granites in the Baerzhe area (Yang et al., 2017). 2019). These Cretaceous low-δ18O magmas are interpreted to be derived from altered juvenile crust in a subduction-related regime of lithospheric 3.3.6. Casto pluton in Idaho, USA extension during Early Jurassic to Early Cretaceous times, which would The Casto pluton is part of the extension-related Eocene Challis have allowed deep circulation of meteoric or seawater (Li et al., 2015; volcanic field in Idaho (USA) and consists of a group of epizonal plutons Yang et al., 2017).

8 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

3.3.9. Baekdusan/Changbaishan volcano, North Korea normal- 18O range 0.5 n = 92 Central Snake Recently, low-δ18O magmas have been inferred from zircon 0 River Plain (USA) (δ18O +3.7 to +5.0 ‰) in trachyte of the AD 940 zircon 0.5 n = 139 Yellowstone- “ ” Millennium eruption of the volcano Baekdusan/Changbaishan in 0 Heise (USA) North Korea (Cheong et al., 2017). Although so far not well con- 1 n = 7 0.5 Talbot rhyolites strained, magmatic activity of the Changbaishan volcano is thought to cont. hotspot 0 (Australia) fi 0.4 n = 93 be related to subduction of the Paci c plate beneath the Eurasian plate 0.2 Iceland (Tang et al., 2014). Low-δ18O signatures are inferred to reflect canni- 0 0.4 n = 61 Isle of Skye/Mull balization of hydrothermally altered roof material (Cheong et al., 0.2 0 (Scotland) 2017). 0.4 n = 11 0.2 Kangerlussuaq 0 (Greenland) δ18 4. Characteristics of low- O silicic magmas 0.4 n = 61 0.2 Lebombo 0 (South Africa) 18 4.1. O-depletion and tectonic setting 0.4 n = 91 0.2 Koegel Fontein 0 (South Africa) δ18 0.5 n = 19 Our compilation shows that low- O silicic magmas can occur in Seychelles continental (and to a minor extent oceanic) hotspot settings, areas of 0 1 n = 34 active rifting and lithospheric extension and continental island arcs 0.5 Tenerife 0 (Canary Islands) (Fig. 1, Table 1), encompassing all major settings for silicic magma 1 n = 5 Malani province genesis. However, the different settings show differences in how effi- 0.5 0 (India) fi 18 δ18 ciently they can produce signi cant O-depletions. The lowest O 0.5 n = 24 δ18 Madagascar values and largest volumes of low- O magma are found in continental 0 hotspot settings and other areas of mantle upwelling (Table 1, Fig. 3). In 0.4 n = 41 0.2 Dabie-Sulu volcanic arcs, however, it is often only a small subset of units, samples, 0 (China) other mantle upwelling (hotspot/rift) or even zircons that classify as low-δ18O (e.g. in the Western Nevada 0.4 n = 87 New England 0.2 granites (USA) volcanic field or Calabozos volcano). Of the arc settings with evidence 0 0.5 for low-δ18O magmas listed here, a number are suggested to be asso- n = 65 Timber Mountain 0 complex (USA) ciated with crustal thinning and/or periods of extension, e.g. Calabozos 0.5 n = 136 Kamchatka volcano or Casto pluton. 0 (Russia) A comparison of the times of magmatic activity of low-δ18O settings 1 n = 5 Mt. Mazama ff δ18 0.5 from di erent tectonic environments (Fig. 4) shows that most low- O 0 (USA) settings associated with mantle upwelling related to rifting or hotspot 1 n = 23 0.5 Calabozos activity coincide with major dispersal periods of the supercontinents 0 (Andes) Rodinia and Pangaea (e.g. Seychelles, Malani suite, Madagascar, Dabie- 0.5 n = 57 Western Nevada Sulu, New England, Isle of Skye, Kangerlussuaq intrusion, Lebombo 0 volc. field (USA) 1 n = 11 rhyolites, and Koegel Fontein). This distribution correlates with the 0.5 Casto pluton (USA) frequency of rifting events in general (Condie, 2002). Neoproterozoic 0 1 n = 8 rifting associated with the dispersal of Rodinia was particularly efficient 0.5 Gyeongsan 0 granites (S Korea) 18 in producing multiple settings with large volumes of low-δ O magma 1 volcanic arc / subduction n = 87 Mesozoic 0.5 (Fig. 4; Seychelles, Malani suite, Madagascar, Dabie-Sulu, and New 0 granites (E China) England), probably because contrary to the dispersal of Pangaea, it 2 n = 2 1 Baekdusan/Chang- coincided with a period of global cooling, which included multiple 0 baishan (N Korea) Snowball or “Slushball” Earth events and resulted in meteoric water 1 n = 47 Aleutians/Alaska 0.5 (USA) with very low δ18O values (e.g. Rumble et al., 2002; Zheng et al., 2008). 0 δ18 -4-2024681012 Low- O rocks from intraplate settings related to hotspot or rifting 18O (‰) activity are likely overrepresented in the geological record as indicated magma by a complete lack of low-δ18O arc settings older than 90 Ma (Fig. 4). Fig. 3. Probability density distribution of oxygen isotope compositions of The past 100 Ma show a similar number of arc-related and intraplate magmas from different settings (in bins of 0.5 ‰). Gray bar marks normal-δ18O settings, suggesting a similar rate of occurrence of low- δ18O magmas in magmatic range. both, despite intraplate settings producing larger volumes of low-δ18O 18 18 magmas and more significant O-depletions. The paucity of a low-δ O nepheline-syenitic Kangerlussuaq intrusion). The Kangerlussuaq intru- arc-related magma record is consistent with the efficient reworking and sion and Tenerife phonolites represent the only low-δ18O magmas with limited preservation of subduction-related rocks compared to their in- peralkaline/alkaline affinity, generally containing more than 11 wt% traplate counterparts (Hawkesworth et al., 2009), although the me- Na2O+K2O. All others are metaluminous to slightly peraluminous 18 chanisms of this preservation bias remain a matter of on-going research (Fig. 5A+B). Low-δ O magmas tend to be Fe-rich (> 2 wt% FeOtotal) (Spencer et al., 2017 and references therein). with Fe numbers higher than 0.6 (Fig. 5C). Only magmas from Kam- chatka and the Aleutians generally contain more Mg than Fe (Fig. 5D). 4.2. Major and trace element compositions and mineral assemblages In terms of the modified alkali-lime index (MALI), most low-δ18O silicic magmas are alkali-calcic to calc-alkalic, whereas magmas from Kam- Most silicic rocks in these studies (> 70 %) are rhyolite/granite chatka and the Aleutians are more calcic (Fig. 5). All whole rock data with bulk silica contents exceeding 70 wt% SiO2, with > 30 % counting and data references can be found in electr. suppl. 1. 18 as high-silica rhyolites (> 75 wt% SiO2). Within each setting, the low- Silicic rocks from low-δ O settings commonly have high con- 18 δ O magmas are among the most silica-rich (Fig. 6), and where they centrations of light and heavy REE, with moderate to strong negative Eu are not of rhyolitic composition, they represent the most silica-rich anomalies (Fig. 6), similar to the “seagull-shaped” REE distributions 18 magma in the province (e.g. low-δ O rhyodacite in predominantly typical of hot and dry volcanic rocks (Bachmann and Bergantz, 2008). andesitic Mt. Mazama volcano or quartz-syenite in the predominantly

9 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

time (Ma) Fig. 4. Overview of episodes of low-δ18O magmatism 1000 800 600 400 200 0 compared to assembly and dispersal periods of su- percontinents (gray thick and dashed bars) and gla- Rodinia Pangaea cial events (bright blue), (A) in the time range 1050 Ma to present, and (B) close-up of past 100 Ma. Gondwana Colors for different settings mark tectonic environ- Talbot, Skye ments continental hotspot (red), mantle upwelling Australia Seychelles related to rifting and/or hotspot activity (yellow) and Kangerlussuaq subduction-related/arc volcanism (blue). Note the Malani Lebombo bias towards intra-continent settings with increasing Madagascar age. Neoproterozoic glacial era (*) includes multiple Koegel Fontein Snowball or slushball Earth events. Dabie-Sulu Gyeongsang New England Casto Andean- Mid-Proterozoic Neoproterozoic Saharan Karoo Stillwater A non-glacial era glacial era* ice age ice age

CSRP Yellowstone-Heise Skye/Mull Iceland Kangerlussuaq Tenerife Timber Mtn. Aleutians Gyeong- Kamchatka sang Baekdusan Casto W Nevada B Late Cenozoic ice age 10080 60 40 20 0 time (Ma)

Most low-δ18O silicic rocks have high concentrations in Nb, Ta, Y, and This observation is in stark contrast to the observation that oxygen and Yb, and classify as “within-plate granite” according to the trace element strontium isotopes often correlate positively for high-δ18O rocks scheme of Pearce et al. (1984) (Fig. 7). This indicates that low-δ18O (Taylor, 1980), as a result of assimilation of high-δ18O high-87Sr/86Sr silicic magmas (1) are dominated by rhyolitic/granitic compositions sediments and crustal metamorphic rocks. that are ferroan, alkali-calcic to calc-alkalic, and meta- to peraluminous The lack of an obvious correlation between oxygen and radiogenic (“A-type”), (2) represent extremely fractionated compositions, and (3) isotope compositions has been described independently for a variety of are more likely to be generated in large volumes in with-in plate hot- settings (e.g. Ellis et al., 2013; Harris and Ashwal, 2002; Monani and spot or rift settings. In major and trace elements, there is no discernible Valley, 2001), particularly those where low-δ18O values are attributed compositional difference between lower- and higher-δ18O magmas in to melting or assimilation of rocks that were hydrothermally altered in the same setting, except for the fact that lower-δ18O silicic magmas tend a caldera setting. Only minor co-variation has been observed in post- to show particularly pronounced Eu anomalies and high silica contents caldera low-δ18O rhyolite lavas from Yellowstone that show higher (Fig. 6). While these generally overlap with the range of higher-δ18O 87Sr/86Sr and 206Pb/204Pb ratios compared to normal-δ18Otuffs rocks, this observation points towards low-δ18O silicic magmas being (Hildreth et al., 1984), reflecting that in such a setting, hydrothermally dominated by extremely fractionated melts. altered rocks proximal to the magma reservoir are often co-genetic and Mineral assemblages in low-δ18O silicic magmas generally reflect not significantly older. This is different in the cases where low δ18O the evolved character of their host magma and are therefore dominated values are attributed to alteration episodes that precede magmatic ac- by the minerals quartz, plagioclase and sanidine/alkali feldspar. Among tivity by millions of years. Both the Koegel Fontein dykes and Sey- mafic phases, anhydrous minerals seem to dominate over amphibole chelles granites show a larger range of radiogenic isotope compositions and biotite, particularly in settings with voluminous low-δ18O magma than other settings, consistent with the suggested greater involvement production (e.g. Yellowstone hotspot track, Lebombo rhyolites). For of variably altered older crustal lithologies (Harris and Ashwal, 2002; example, Bindeman et al. (2004) noted that the majority of low-δ18O Harris et al., 2018). Similarly, some co-variation between δ18O values magmas in Kamchatka contain , whereas higher-δ18O magmas and 87Sr/86Sr ratios in silicic magmas in Kamchatka suggests that hy- usually contain amphibole. Except for the peralkaline Kangerlussuaq drothermal alteration also affected older basement rocks, e.g. by intrusion, the quartz porphyry dikes from Koegel Fontein, Tenerife leaching radiogenic Sr from them (Bindeman et al., 2004). phonolites, and relatively mafic volcanic deposits from Alaskan caldera centers, all settings are described to be zircon-saturated. 4.4. Magmatic temperatures

4.3. Radiogenic isotopes Although many of the low-δ18O silicic magmas documented here are relatively hot and dry, there seems to be no strict correlation between Initial strontium isotope and neodymium isotope compositions vary δ18O value and temperature within any individual setting (Fig. 9). Pre- little with oxygen isotope compositions in low-δ18O settings (Fig. 8). eruptive temperatures are particularly high in silicic magmas from

10 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

Fig. 5. Granite classification diagrams (Frost et al., 2001) for major element compositions of different low-δ18O settings. Panels next to each other show same data, 18 with data points in left panels (A, C, E) being color-coded for their δ Omagma value and data points in the right panels (B, D, F) indicating their setting. (A+B)

Alumina saturation expressed as molar Al2O3/(CaO+Na2O+K2O) versus molar Al2O3/(Na2O+K2O). (C+D) Fe number classification distinguishing ferroan A-type from magnesian arc granites. (E+F) Modified alkali-lime index (MALI) vs. SiO2 indicates most settings here are alkali-calcic to calc-alkalic. Note that this figure only includes settings where major element data and oxygen isotope data are available for the same samples (see electr. suppl. 1 for the full compilation).

11 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

Fig. 6. Oxygen isotope variation in magmas relative to (A) their silica content, and (B) their Eu anomaly. Gray bar indicates normal-δ18O range. Red arrows indicate chemical trend for assimilation with concurrent partial melting of high-δ18O crustal source material, blue arrows for low-δ18O hydrothermally altered crustal source material. hotspot and rift settings (Fig. 9), e.g. mineral thermometry suggests that of quartz-zircon pairs give temperatures of 702-945 °C, whereby lower pre-eruptive temperatures range between 750-900 °C in Yellowstone temperatures in this range overlap with temperatures estimates from (Troch et al., 2017; Vazquez et al., 2009; Watts et al., 2012) and 800- Al-in-hornblende thermobarometry (727-842 °C) on low-Al hornblende 900 °C in Heise (Watts et al., 2011). These estimates are consistent with interpreted as phenocrysts crystallized at shallow depths (Watts et al., experimental work suggesting ~845-875 °C and ~1.5-2.5 wt% H2O for 2019). Crystallization temperatures based on oxygen isotopic compo- Heise (Bolte et al., 2015), and ~790-870 °C and ~2.8-4.7 wt% H2O sitions in quartz-zircon pairs in the Gyeongsang granites (South Korea) indicated by phase equilibria and melt inclusions for Yellowstone are 730 ± 70 °C, consistent with Zr saturation temperatures of (Myers et al., 2016; Tollan et al., 2019; Troch et al., 2017). Geother- 727 ± 18 °C (Li et al., 2015). mometry on mineral phases in the Central Snake River Plain suggests high pre-eruptive temperatures (850-1050 °C), low water contents (< 3 4.5. Magmatic storage pressure estimates wt% H2O) and reducing conditions around QFM ± 0.5 (Almeev et al., 2012; Cathey and Nash, 2009; Christiansen and McCurry, 2008; Ellis Storage for silicic magmas in hotspot and rift settings is fairly et al., 2010; Honjo et al., 1992; Knott et al., 2016). Schattel et al. (2014) shallow. Geobarometry based on normative whole-rock compositions estimate temperatures of 886-1008 °C for rhyolites from Askja, Iceland, results in pressure estimates of 100-200 MPa for units from Heise based on plagioclase-melt and ilmenite-magnetite pairs, and 836-928 °C (Watts et al., 2011). Experimental work suggests storage pressure of for rhyolites from Öræfajökull based on olivine-melt and ilmenite- ~130-250 MPa for Heise (Bolte et al., 2015) and < 150 MPa by phase magnetite pairs. Similarly, temperatures of 830-1020 °C were de- equilibria simulations and volatiles in melt inclusions for Yellowstone δ18 termined for rhyolites from Hekla (Portnyagin et al., 2012). Low- O (Befus and Gardner, 2016; Tollan et al., 2019; Troch et al., 2017). rhyolites from Askja are slightly hotter and drier (< 2.7 wt% H2O) and Portnyagin et al. (2012) calculate entrapment pressures of 180-220 more oxidized (QFM +1.4 ± 0.1) than more normal-δ18O rhyolites MPa (6-7 km depth) for H2O-rich melt inclusions in fayalite from Hekla, from Hekla and Öræfajökull (< 6.2 wt% H2O, QFM -0.1 to -0.9) Iceland, consistent with the geophysically derived depth of a magma (Schattel et al., 2014). For the Lebombo rhyolites, mineral thermometry chamber between 5 and 14 km (Sturkell et al., 2006). Schattel et al. on pigeonite-augite and Fe-Ti oxide pairs suggests extremely high (2014) proposed that low-δ18O rhyolites from Askja crystallize at lower crystallization temperatures of 900-1100 °C (Betton, 1978; Harris and pressures (minimum 50 MPa or 1.8 km) than off-rift rhyolites from Erlank, 1992), in the same range as those of the CSRP. Estimated Hekla and Öræfajökull (minimum 120 MPa or 4 km) based on entrap- temperatures for the Kangerlussuaq intrusion range between 750-850 ment pressures of melt inclusions in plagioclase, fayalite, clino- and °C based on single-feldspar and nepheline geothermometry (Kempe and orthopyroxene. During borehole drilling at Krafla, rhyolite melt was fi Deer, 1970). Bindeman and Valley (2003) nd variable pre-eruptive encountered at 2.1 km depth (16-55 MPa, Zierenberg et al., 2013). temperatures in units of the Oasis Valley-Timber Mountain volcanic Riishuus et al. (2015) suggest an emplacement pressure of ~100 MPa complex, ranging from 700 to 900 °C based on Fe-Ti oxide thermometry for the Kangerlussuaq intrusion based on observations from phase and oxygen isotope fractionation between quartz-magnetite and sani- petrology. dine-magnetite pairs. Available barometry data for low-δ18O arc settings suggest slightly Pre-eruptive temperatures are slightly lower in most arc settings deeper pre-eruptive storage compared to hotspot and rift settings. “ ” (Fig. 9), as expected for common cold-wet arc rhyolites (Loewen and Pressure estimates for dacitic ignimbrites from the Gorely center in Bindeman, 2016). Dacitic ignimbrites from the Gorely centre in Kam- Kamchatka range between 500-600 MPa based on pyroxene thermo- chatka range between 891 and 926 °C based on pyroxene thermometry barometry (Seligman et al., 2014a), although MELTS simulations sug- (Seligman et al., 2014a). While these overlap with temperature esti- gest that magmas may have experienced storage at 100-150 MPa. Such mates from Yellowstone or Iceland rhyolites, the less evolved whole- shallow pre-eruptive storage is consistent with phase equilibria ex- fi rock compositions con rm their comparably cooler magmatic char- periments on compositions from Ksudach volcano (Bindeman et al., acter. For the Stillwater caldera complex, oxygen isotope compositions 2010a and references therein). Applying Al-in-hornblende barometers,

12 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

Fig. 7. Granite classification diagrams (Pearce et al., 1984) for trace element compositions of different low-δ18O settings. Panels next to each other show same data, 18 with data points in left panels (A+C) being color-coded for their δ Omagma value and data points in the right panels (B+D) indicating their setting. Note that low- δ18O silicic magmas can occur in any tectonic setting, but the largest 18O-depletions are found in with-in plate granites corresponding to continental hotspot and rift settings.

Watts et al. (2019) calculate pressures of 2.7-5.9 kbar for high-Al am- (roof) material (Bindeman and Valley, 2001; Leeman et al., 2008; Watts phiboles and 0.8-2.7 for low-Al amphiboles in the Tuff of Elevenmile et al., 2011). Canyon of the Stillwater volcanic complex. They interpret the high-Al Rates for diffusive oxygen exchange between a magma reservoir and amphiboles as xenocrysts and suggest a preferred crystallization pres- surrounding country rocks are generally insufficient to produce changes sure of 100-220 MPa or 4-8 km depth. in δ18O in magmatic bodies, although small-scale diffusive exchange of These barometry data focus on volcanic units; very little data is 18O/16O between magma and wall rock is indicated by systematic in- available for granites in this study. However, all of them are A-type creases in δ18O at the margins of some plutons (e.g. Shieh and Taylor, granites and thus likely to have originated at similarly shallow pres- 1969; Turi and Taylor, 1971). Direct addition of water into the magma sures (Frost and Frost, 2010; Patiño Douce, 1997). is usually considered unlikely due to the limited transportability and

solubility of H2O in ductile rocks and silicate melts (Taylor, 1974; Taylor, 1977, 1986). Due to the great volume change in H O upon the δ18 2 5. Generation processes of low- O silicic magmas liquid/vapor transition, water generally moves away from hot magma bodies, not toward it. This has for example been demonstrated by a 5.1. Crustal melting and assimilation study on the behavior of water and low-δ18O alteration near ignimbrite deposits at Crater Lake (Hudak and Bindeman, 2018). Mass-balance Three major processes have been proposed as possible mechanisms reasons also argue against direct addition of meteoric water into δ18 for creating low- O magmas: (1) direct addition and dissolution of magmas, as even with the most 18O-depleted and isotopically unshifted δ18 low- O meteoric water into the magma (Friedman et al., 1974; meteoric water (e.g. present meteoric water in Yellowstone at δ18Oof Hildreth et al., 1984), (2) assimilation of hydrothermally altered -19 (Hildreth et al., 1984)), a normal-δ18O magma of δ18O+6‰ country rocks into the magma (Bacon et al., 1989; Grunder, 1987; 18 would require input of ca. 20 wt% H2O to create the lowest-δ O Taylor, 1986) and (3) bulk or partial melting of hydrothermally altered

13 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

Fig 8. Strontium (initial 87Sr/86Sr) and neodymium (εNd) isotopic compositions in low-δ18O settings in different tectonic settings illustrate characteristic lack of correlation between oxygen and radiogenic isotopes. Gray bars indicate MORB and normal-δ18O ranges, respectively.

Sharp, 1995; Hildreth et al., 1984). Assimilation and melting as used in this study mainly differ in their proportions of mantle-derived to crustal-derived melt, with the degree of assimilation defined as the percentage of crustal melt in the total melt mass. Assimilation usually refers to tens of percent of crustal input (< 30-40 %), whereas crustal melting implies that melt almost exclusively consists of crustal mate- rials (> 80-90 %). The degree of assimilation depends on the ratio between the amount of energy required to heat crustal rocks to mag- matic temperatures (usually leading to partial melting) and the energy available for this process, e.g. due to crystallization of replenishing magma in the magma reservoir (Bohrson and Spera, 2001; Spera and Bohrson, 2001). In the following sections, we will discuss (1) general characteristics of low-δ18O settings, (2) the nature and melting behavior of hydrothermally altered rocks, (3) their potential to be assimilated by different magmas as a function of magma composition and other parameters, and (4) the mode of assimilation and the geochemical traces (or lack thereof) that assimilation leaves in low-δ18O magmas.

5.2. Characteristics of low-δ18O settings

Low-δ18O alteration and large degrees of assimilation are usually constrained to very different depth intervals in the Earth’s crust, making low-δ18O silicic magmas a conundrum. Circulation of hydrothermal Fig 9. Zircon saturation temperatures (Watson and Harrison, 1983) versus fluids is restricted to pathways such as faults that can only persist in the 18 oxygen isotope ratios in low-δ O settings. Continental hotspot and other brittle upper crust (e.g. Ingebritsen and Manning, 1999). Assimilation mantle upwelling settings show higher magmatic temperatures compared to and melting, however, is most efficient at high temperatures prevalent most volcanic arc settings, but there is no correlation between temperature and in the lower crust (Annen et al., 2006; Hildreth and Moorbath, 1988). δ18O within any particular setting. Blue dashed lines indicate liquidus tem- Several factors therefore need to come together for the formation of peratures for different water contents for rhyolite from Yellowstone at 2 kbar as low-δ18O silicic magmas: simulated by rhyolite-MELTS in Tollan et al. (2019). Gray bar marks normal- δ18O range. (1) large volumes of rock that were altered at high temperatures (> 300-400 °C) for extended periods in order to reach widespread ‰ magmas (< 0 ). This is several times higher than the maximum water lowering of δ18O, solubility in shallow rhyolitic melts (~4-6 wt%). The Kangerlussuaq (2) highly 18O-depleted meteoric water, i.e. the hydrothermal system intrusion is the only setting where multiple lines of evidence suggest must be located at sufficiently high latitude and/or altitude, or be that water was added directly into magma in the form of stoped un- active during global glaciation episodes (Figs. 1 and 4); sufficiently melted basaltic blocks containing pore water (Riishuus et al., 2015). high water/rock ratios are required for water to be only little ffi Two processes have been invoked as most e cient to transfer modified during water-rock oxygen isotope exchange, δ18 crustal O signatures into silicate melts: Assimilation and melting of (3) a shallow, high-enthalpy magmatic system that is sufficiently large crustal rocks that were previously hydrothermally altered by meteoric to fuel long-term magma storage, high degrees of assimilation and water (Bindeman and Valley, 2001; Boroughs et al., 2012; Feeley and

14 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

hydrothermal fluid circulation, and Although only a small fraction of the erupted volume of the Stillwater (4) vertical displacement mechanisms that bring hydrothermally al- complex is low in δ18O, this proximity suggests a high potential for tered rocks to magmatic depths and/or magma to sufficiently exchange between magma and hydrothermally altered extremely low- shallow depths at which hydrothermally altered rocks can be en- δ18O rocks in these syn-magmatic settings, because magmatism and countered. low-δ18O altered source material are already spatially associated. In settings with pre-existing alteration, in contrast, magmatism would by Two main relationships between hydrothermal alteration and silicic chance assimilate low-δ18O rocks, likely leading to much dilution of the magmatism can be distinguished for low-δ18O settings, (1) pre-existing δ18O signal as magma invariably will also assimilate higher-δ18O rocks. hydrothermal alteration that significantly pre-dates magmatic activity Multiple caldera collapses help to make hydrothermally altered rocks (often by tens or hundreds of Ma) and may be unrelated from the accessible for assimilation by both leading to deep dissection and in- current tectonic setting, or (2) syn-magmatic hydrothermal alteration creasing permeability for hydrothermal fluids in the caldera roof and by that is driven by magmatic activity in the same volcanic field or its transporting hydrothermally altered rocks to larger depths as they be- greater province in the same tectonic context. come progressively buried by caldera infill (Bindeman and Simakin, Settings with low-δ18O silicic magmas related to pre-existing hy- 2014; Simakin and Bindeman, 2012). drothermally altered crust include the granitic suite of the Seychelles, In such a system characterized by syn-magmatic hydrothermal al- Mesozoic granites from China, as well as some of the New England teration, it would be the uppermost portion of a magma reservoir clo- Avalonia granites and the Koegel Fontein igneous complex. The tem- sest to the overlying hydrothermally altered rocks that is most likely to poral decoupling of hydrothermal activity and magmatism places fewer be depleted in 18O. Low-δ18O magma accumulated in the shallow, constraints on the origin of low δ18O values in these settings, as tectonic crystal-poor cap of the mushy magma reservoir would be particularly activity may have transported hydrothermally altered low-δ18O rocks to prone to erupt and thus be part of the volcanic record (Fig. 11). This lower latitudes, or brought them to larger depths in the crust, where could explain the large number of volcanic settings with evidence for they are easier to melt and assimilate. Most of these settings seemingly low-δ18O magmas, although low-δ18O plutons should be more easily have only produced small volumes of low-δ18O silicic magma (Table 1), preserved over geological timescales. A comparison of different erosion although it is possible that much of the low-δ18O record in these old levels in such a system (Fig. 11 A-C) illustrates the limited probability centers has been eroded. for exposure of this low-δ18O shallow portion in a pluton, and in par- Most of the low-δ18O settings, however, result from interaction with ticular for its exposure together with the overlying hydrothermally al- syn-magmatically altered rocks, which tends to both produce larger tered source material (cf. “Low-18O silicic magmas: Why are they so volumes of low-δ18O magma and lead to lower δ18O values (Table 1). rare?” by Balsley and Gregory (1998)). The largest volumes of low-δ18O silicic magmas are generated in con- tinental hotspot or rift settings (Table 1) that are associated with mantle 5.3. The nature and melting behavior of hydrothermally altered rocks upwelling and therefore have extremely high geothermal gradients. In these settings, faulting associated with extension may provide pathways The hydrothermally altered source material for assimilation has for deep hydrothermal circulation. Magmas in these settings pre- been investigated and characterized in several different settings, e.g. dominantly consist of hot and relatively dry rhyolite that was generated along the Yellowstone hotspot track (Criss et al., 1984; Criss and Taylor, at shallow crustal levels (< 10 km), as indicated by thermobarometry 1983; Drew et al., 2013; Ellis et al., 2017; Troch et al., 2018), the Lake (section 4.3). At their time of activity all of these syn-magmatic low- City Caldera in the San Juan Mountains, Colorado (Garden et al., 2020; δ18O settings were located at fairly high latitudes (> 30 °N or °S) and/ Larson and Taylor, 1986), and the Stillwater volcanic complex, Nevada or altitude (Fig. 1), where δ18O values of meteoric water are sufficiently (John and Pickthorn, 1996; Watts et al., 2019). These rocks tend to be low to permit a decrease in δ18O value in the altered material at rea- propylitically altered and decrease in their 18O/16O ratios with depth, listic water-rock ratios, e.g. meteoric water in Yellowstone, Iceland and down to δ18O values of < -6 ‰. Metamorphically equilibrated zircons Kamchatka has δ18O values below -10 ‰. The central Snake River Plain from the Neoproterozoic Dabie-Sulu terrane in east-central China in- may be the only currently known hybrid setting (Table 1), where the dicate that in extreme cases, large-scale hydrothermal activity can lead superposition of pre-existing and syn-magmatic alteration has led to the to bulk rock δ18O values of ~-9 ‰ (Fu et al., 2013). Particularly when generation and eruption of extremely large volumes of low-δ18O formed during periods of extensive glaciation leading to extremely low- magma (Troch et al., 2020). δ18O meteoric water, e.g. during “snowball” or “slushball” Earth events, Settings with syn-magmatic hydrothermal alteration are commonly hydrothermally altered rocks can exhibit extremely low δ18O values. associated with one or multiple caldera collapses, suggesting that cal- Metasediments and metavolcanics with δ18O values of -16 to -28 ‰ in deras are particularly suitable environments for bringing together silicic the Paleoproterozoic Karelian rift are the most 18O-depleted silicate magmas and hydrothermally altered rocks, and generating low-δ18O rocks on Earth and likely represent the limit of possible hydrosphere- magmas. It is possible that some of plutonic suites listed here were lithosphere interaction (Bindeman et al., 2010b). This tremendous associated with caldera systems, however, the geological record rarely range in δ18O values requires that for every setting, the likely δ18O allows the identification of such calderas, as exposure of the pluton range of the assimilated material has to be assessed independently, in usually involves the removal of overlying rocks. Several studies have due consideration of paleolatitude, -altitude and extent of glaciation at shown that bulk δ18O values of pervasively propylitically altered cal- the time of activity as well as scale of water-rock interaction and al- dera roof systematically decrease with depth as a function of increasing teration temperatures. alteration temperatures at closer proximity to the underlying magmatic The melting behavior of a rock is a function of its major element heat source, e.g. in Yellowstone (Troch et al., 2018), the Lake City composition and water content (Johannes and Holtz, 1996 and refer- Caldera in the San Juan Mountains, Colorado (Larson and Taylor, 1986) ences therein). In terms of major and trace element compositions, hy- and the Stillwater volcanic complex, Nevada (John and Pickthorn, drothermally altered low-δ18O rhyolites and granites along the Yel- 1996; Watts et al., 2019). Cross-sections (≥2 km) through hydro- lowstone hotspot track are more variable but remain similar to their thermal fields in Iceland reveal similar decreases in δ18O(Hattori and unaltered counterparts (Troch et al., 2020; Troch et al., 2018). Some Muehlenbachs, 1982; Sveinbjornsdottir et al., 1986). The exposed hy- samples are affected by silicification and show an increase in normative drothermal roots of the Stillwater volcanic complex (Watts et al., 2019) corundum due to the removal of alkali elements during alteration illustrate that in such a setting, hydrothermally altered rocks with bulk (Troch et al., 2018), leading to a more peraluminous character (in- δ18O values < -6 ‰ can be found in the vicinity (< 1 km) of granites cluding crystallization of cordierite) when these rocks are melted (δ18O>+6 ‰) that reflect the depth of the main magma reservoir. (Troch et al., 2020).

15 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

Fig 10. Rhyolite-MELTS simulations of (A) enthalpy release and uptake during cooling and crystallization of rhyolite magma (solid lines) and basaltic magma (dashed lines) with different water contents at 2 kbar, and (B) melting of hydrothermally altered granite (solid lines) and basalt (dashed lines) at 2 kbar at different water contents. Small numbers indicate melt fractions in steps of 10 wt%.

For silicic rocks, water contents exert the strongest control on the granitoid rocks, whether they be hydrothermally altered or not, as long melting behavior (Troch et al., 2018). Although we will show in the as they have not experienced low-temperature alteration (~100-200 following paragraph that low-δ18O alteration usually does not lead to °C). While it is yet unclear how alteration may affect the fertility of less higher water contents, a simple thought experiment illustrates the silicic protoliths, there is currently no evidence that would support that possible relationships between water contents, melting behavior and low-δ18O hydrothermally altered rocks are more prone to melting and the oxygen isotope composition of the resulting melt. If oxygen isotope assimilation than average crustal rocks. exchange were associated with hydration, crustal rocks that were hy- drothermally altered at high temperatures would be more fertile (i.e. 5.4. Quantifying assimilation melt more easily) than average crustal rocks, e.g. due to alteration leading to the formation of hydrous alteration minerals, such as micas, Magmatic settings with the largest volumes of low-δ18O magma and clays and zeolites. This would lead to selective melting and assimilation the highest extent of 18O-depletion are associated with continental of hydrothermally altered source material, resulting in magma assim- hotspots and other sites of mantle upwelling, and tend to produce ilating low-δ18O melts that may be significantly lower in δ18O than rhyolite magma with high pre-eruptive temperatures and low water average crust. Such a model has for example been suggested for the contents (Fig. 9, section 4.4, Table 1). A similar observation led Balsley Koegel Fontein igneous complex (Curtis et al., 2013). In their early and Gregory (1998) to suggest that hot and dry silicic magmas can model for Yellowstone, Bindeman and Valley (2001) similarly proposed assimilate larger amounts of low-δ18O altered material (or other rocks). that hydrothermally altered caldera infill consists of near-cotectic, A correlation between magmatic temperatures and δ18O values would water-rich (> 3 wt%) rhyolite that melts completely at tempera- be consistent with a process whereby the energy/heat spent on assim- tures < 850 °C. However, detailed investigations of hydrothermally ilation results in slightly lower pre-eruptive temperatures for lower- altered silicic source material for assimilation in Yellowstone and along δ18O magmas that experienced more assimilation. However, there is no the Snake River Plain later showed that oxygen isotope exchange in evidence of systematic decreases in temperature with decreases in these rocks is not associated with hydration, with the vast majority of magmatic δ18O values within any of the settings investigated in this them having bulk water contents < 1 wt% (Troch et al., 2020; Troch study (Fig. 9). This indicates that within an individual setting, δ18O et al., 2018). This lack of correlation is in stark contrast to observations values are primarily controlled by the accessibility of low-δ18O source from high-δ18O rocks altered at lower temperatures (100-200 °C), materials for assimilation and less by changes in the degree of assim- where higher whole-rock δ18O values correlate with increasing water ilation. When comparing different settings, however, the drier and contents, with loss on ignition (LOI) values reaching up to 11 wt% (Berg hotter magmas associated with mantle upwelling may indeed lead to et al., 2018). At the high temperatures (> 300 °C) required for efficient higher degrees of assimilation. Calculating the enthalpy release for a lowering in δ18O, however, alteration of high-silica rocks does appar- typical silicic magma in this study at different water contents with ently not lead to significant formation of hydrous minerals such as rhyolite-MELTS (Gualda et al., 2012) indicates that due to their crys- clays, zeolites and micas (as noted already by Taylor and Forester, tallization over a larger temperature interval, drier magmas release 1979). more energy during crystallization and cooling (Fig. 10A) for a given Melting simulations on drillcore samples from Yellowstone, which temperature interval, and may therefore be indeed more prone to as- penetrate the hydrothermally altered caldera infill suggested to be the similating crustal rocks. This effect is even more pronounced if the in- source material for assimilation, result in average melt fractions truding magma is initially intermediate or basaltic (Fig. 10A). of < 20 wt% at 750 °C, ~35 wt% at 850 °C, and ~50 wt% at 950 °C, For a simple heat balance comparison, we calculate the sensible respectively, with water being the most limiting factor during melting heat and latent heat of fusion/crystallization for a mafic and rhyolitic (Troch et al., 2018). These samples have an average water content of magma as a function of different water contents (Fig. 10A+B), illus- 0.55 wt% H O (analyzed by Karl-Fischer titration), which correlates 2 trating that most energy is released during cooling and crystallization of well with average LOI values of 0.91 wt% or median LOI values of 0.73 mafic, dry magma, and least energy required by melting of hydrous wt% in granites from the GEOROC database since LOI values can be rhyolite. Assimilation of crustal rocks is thus most efficient where large affected by CO and oxidation. We therefore recommend using 0.5-1.0 2 amounts of dry basaltic magma encounter highly-evolved silicic rocks, wt% H O for simulating upper crustal melting and assimilation of 2 as is the case in many bimodal basaltic-rhyolitic settings. With their

16 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

Fig. 11. Schematic illustration of oxygen isotope variations in magmatic settings with syn-magmatic caldera-hosted alteration. Erupted volcanic products may be more likely to sample the low-δ18O portion of a magma reservoir as they represent the eruptible (i.e. mobile) and shallowest fraction of the reservoir, which is closest to overlying low-δ18O hydrothermally altered caldera roof. A comparison of different erosion levels through this system (A-C) indicates why low-δ18O granitic plutons are relatively rare, despite their presumably better preservation potential during erosion compared to their volcanic counterparts. Shallow erosion (A) only exposes a large-scale low-δ18O anomaly without an associated pluton or other record of low-δ18O magma. Deep erosion (C) may expose the normal-δ18O or only mildly 18O-depleted portion of a granitic pluton that has not experienced assimilation of hydrothermally altered rocks, without any record of any overlying, potentially hydrothermally altered units. Only a narrow window of exposure around an intermediate erosion level (B) would expose both low-δ18O plutonic rocks and associated overlying hydrothermally altered material. Note that the color-coding is not meant to imply that eruptible magma caps are commonly zoned in oxygen isotopic compositions, but different assimilation degrees of source materials with variable δ18O may explain some of the oxygen isotope heterogeneity in volcanic rocks. Also note that average continental crust tends to have higher δ18O values than most magmas differentiated from mantle compositions, which is omitted here for simplicity. thermal model, Karakas and Dufek (2015) showed that crustal melting rock may remain, but disintegrate (e.g. Huber et al., 2011) and minerals and assimilation are enhanced in areas of significant extension and re-equilibrate chemically and isotopically over time. Incomplete equi- during injection of high fluxes (> 0.01 m3/m2/yr) of dry basaltic libration of these minerals explains the often observed inter-grain iso- magma. For Yellowstone, a thermomechanical model taking into ac- topic variability of many low-δ18O rhyolites, e.g. the heterogeneity of count dike propagation and rheology of the crust suggests that the quartz grains correlates with the degree of lowering in δ18O in post- overall contribution of upper crustal melt in the total melt volume does caldera lava flows in Yellowstone (Troch et al., 2017). This inter-grain not exceed 50 wt%, and mostly remains below 25 wt% (Colón et al., heterogeneity is most apparent in zircon as one of the minerals most 2019). Whereas the amount of assimilation may be larger locally, e.g. in resistant during remelting (Bindeman and Valley, 2001; Bindeman and cm- to m-sized pockets during beginning anatexis, we propose ~30 wt% Valley, 2003). Quartz and zircon are also interesting in that they do not to be a good rule of thumb for the maximum amount of assimilation exchange during high-temperature alteration and retain their original possible during the generation of km3-sized volumes of low-δ18O magmatic oxygen isotopic compositions, which many zircons preserve rhyolite. This is in good agreement with independently constrained as normal-δ18O or variable cores overgrown by rims that are in equi- assimilation estimates in a variety of settings, e.g. Kamchatka (15-25 %, librium with the melt (Bindeman et al., 2012; Bindeman et al., 2008; Bindeman et al., 2004), Alaska (5-10 %, Bindeman et al., 2001), Iceland Bindeman and Valley, 2001; Watts et al., 2019). Other than inter-grain (10-20 %, Zakharov et al., 2019), Yellowstone and the CSRP (25-45 %, heterogeneity, very little evidence usually remains of the assimilated Colón et al., 2019; Troch et al., 2020; Troch et al., 2018), Calabozos altered wall rock, as Mt. Mazama and Kangerlussuaq are yet the only caldera complex (5-30 %, Grunder, 1987), Timber Mountain complex low-δ18O settings described to contain such blocks. (20-40 %, Farmer et al., 1991), the Stillwater caldera complex (20-30 As oxygen isotope exchange is not associated with hydration (see %, Watts et al., 2019), and Casto pluton (32-55 %, Larson and Geist, section 5.1), hydrothermally altered granite/rhyolite does not, or only 1995). rarely, contain any extremely low-δ18O hydrous mineral phases, whose decomposition and disequilibrium melting could result in partial melts 5.5. Mode of assimilation and geochemical traces being much lower in δ18O than their bulk source material. Partial melts of hydrothermally altered granite and rhyolite along the CSRP have Bulk assimilation is the most commonly invoked mechanism for experimentally been shown to inherit the bulk δ18O of their source assimilation (Beard et al., 2005), whereby blocks of country rock or material (Troch et al., 2020). A comparison of the resulting melt earlier crystallized material are incorporated into the magma, heated compositions in the examples used in Fig. 10 shows that such partial up and partially melted. Evidence for this process is given by the par- melts are generally water-undersaturated compared to the basaltic tially fused granitoid wall blocks of Mt. Mazama, USA (see images in fractionate that assimilates them, and assimilation could therefore Ankney et al., 2017; Bacon and Lowenstern, 2005), where melting further promote the anhydrous character of most low-δ18O silicic proceeds along feldspar-quartz contacts and other mesostasis compo- magmas. This is in contrast to suggestions by Balsley and Gregory nents such as microcrystalline groundmass. Less fusible parts of the (1998) that assimilation drives magmas to water oversaturation,

17 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299 although it supports their idea of a general link between water contents geothermal gradients. Magmatism in these settings is characterized and 18O-depletion in magmas. by bimodal basaltic-rhyolitic volcanism and hot and dry magmas of In terms of major elements, assimilation does not leave marked “A-type” affinity, which allow for higher degrees of assimilation. signals as the addition of melt with minimum/eutectic melt composi- We estimate that < 30-40 % assimilation can explain most of the tion does not significantly alter the chemistry of already highly frac- oxygen isotope compositions in low-δ18O magmas, consistent with tionated magma. Even if the intruding magma is initially basaltic or an estimates from thermal models. intermediate differentiate, it is likely to have fractionated towards (4) Assimilation of syn-magmatically altered rocks is often associated evolved compositions, as assimilation is controlled by the available with nested caldera complexes, in which δ18O values in the altered latent heat of crystallization. Trace element and radiogenic isotope caldera infill decrease with depth, bringing very low-δ18O altered trends are similarly obscured by the general likeness between the in- rocks in close contact with magma inside the magma reservoir. At truding fractionated magma and the altered assimilated material, which conditions optimal for oxygen isotope exchange towards lower δ18O is usually co-magmatic in origin. values, alteration is not associated with hydration, and therefore Shallow assimilation of hydrothermally altered rocks is easily low-δ18O hydrothermally altered rocks do not melt more readily identifiable when silicic magmas have clearly lower δ18O values com- compared to average crust. Due to their mostly co-genetic nature, pared to basalts in the same province. This however, may not always be assimilation of hydrothermally altered rocks rarely leaves identifi- the case, and assimilation can be cryptic when δ18O values are only able traces in the major and trace element record of low-δ18O silicic just < 5.5 ‰ (the lower limit of our normal-δ18O range) or when magmas, which may have experienced additional overprinting as- magmas have experienced additional assimilation of higher-δ18O rocks, similation of high-δ18O crustal rocks. Together, low-δ18O silicic such as metasedimentary rocks at lower crustal levels. This has been magmatic settings offer unique insights into magmatic processes in shown for magmas in the western Nevada volcanic field (Watts et al., the shallow upper crust and into the generation mechanisms for 2016; Watts et al., 2019), the volcanoes Calabozos and Ollagüe in the silicic magmas on Earth. Andes (Feeley and Sharp, 1995; Grunder, 1987), and may be the case for Andean volcanoes in general (Folkes et al., 2013). It may also ex- Declaration of Competing Interest plain why no low-δ18O rhyolites have yet been identified in the Taupo volcanic zone, New Zealand. As a back arc extension-related volcanic None. field at intermediate latitudes (Fig. 1) with long-lived magmatism and extensive hydrothermal activity, Taupo would be a prime candidate at Acknowledgements least for small 18O-depletions, particularly as its most recent magma- δ18 tism is hot and dry (Deering et al., 2010). Assimilation of high- O We thank Peter Ulmer, Chris Huber, Kathy Cashman and François greywacke, however, largely overprints potential shallow assimilation Holtz for fruitful discussions on the topic, and Nico Kueter for com- of hydrothermally altered rocks, which could be inferred from a ne- menting on an earlier version of the manuscript. We are very grateful δ18 gative correlation between O and SiO2 for magmas with > 65 wt% for the detailed and thoughtful reviews by John Wolff and Kathryn SiO2 (Blattner and Reid, 1982; McCulloch et al., 1994). A similar two- Watts, which significantly improved this paper, and to Arturo Gomez- δ18 stage assimilation process has been inferred from late decreases in O Tuena for the editorial handling. This work was supported by an ETH in quartz and zircon rims from the 75-ka Toba supereruption (Budd research grant (ETH-05 13-2 covering JT), an ETH Career Seed grant δ18 et al., 2017) and could explain a general decrease in O values over (SEED-14 19-1 to JT), the Swiss National Science Foundation (grant time in granites from the Transbaikalia region (Wickham et al., 1996; 178928 covering JT and OB) and the Russian Science Foundation (RNF fi δ18 Wickham et al., 1995). Only when setting-speci c normal- O arrays grant 19-17-00241 to INB). are established (Fig. 2) can subtle 18O-depletions be identified in silicic magmas around the world. Appendix A. Supplementary data

6. Conclusions Supplementary data to this article can be found online at https:// doi.org/10.1016/j.earscirev.2020.103299. Based on a compilation of low-δ18O magmatic settings around the world, we conclude: References

δ18 (1) Low- O silicic magmas are generated by partial melting and as- Allsopp, H., Manton, W., Bristow, J., Erlank, A., 1984. Rb-Sr geochronology of Karoo similation of rocks that previously experienced hydrothermal al- felsic volcanics. Petrogenesis of the volcanic rocks of the Karoo Province. Special teration at high temperatures (> 300 °C) by isotopically light me- Publications of the Geological Society of South Africa 13, 273–280. Almeev, R.R., Bolte, T., Nash, B.P., Holtz, F., Erdmann, M., Cathey, H.E., 2012. High- teoric water, prevalent at high latitudes and altitudes. This process temperature, low-H2O silicic magmas of the Yellowstone hotspot: an experimental can occur in any magmatic-tectonic setting. study of rhyolite from the Bruneau–Jarbidge Eruptive Center, Central Snake River (2) Two configurations characterize the relationship between magma- Plain, USA. J. Petrol. 53, 1837–1866. ‘ ’ Anderson, A.T., Clayton, R.N., Mayeda, T.K., 1971. Oxygen isotope thermometry of mafic tism and alteration: (1) pre-existing alteration whereby alteration igneous rocks. J. Geol. 79, 715–729. significantly predates magmatism and may even be associated with Ankney, M.E., Bacon, C.R., Valley, J.W., Beard, B.L., Johnson, C.M., 2017. Oxygen and U- adifferent tectonic setting, and (2) ’syn-magmatic alteration’, Th isotopes and the timescales of hydrothermal exchange and melting in granitoid whereby hydrothermal fluid circulation is driven by magmatic ac- wall rocks at Mount Mazama, Crater Lake, Oregon. Geochim. Cosmochim. Acta 213, 137–154. tivity in the same volcanic field. Assimilation of rocks with pre- Annen, C., Blundy, J., Sparks, R., 2006. The genesis of intermediate and silicic magmas in existing alteration is mainly restricted by the limited probability for deep crustal hot zones. J. Petrol. 47, 505–539. rocks with rare, large-scale 18O-depletion to be encountered by Archibald, D.B., Collins, A.S., Foden, J.D., Payne, J.L., Holden, P., Razakamanana, T., De Waele, B., Thomas, R.J., Pitfield, P.E., 2016. Genesis of the Tonian Imorona–Itsindro magmatism, whereas the spatial and temporal correlation between magmatic Suite in central Madagascar: Insights from U–Pb, oxygen and hafnium alteration and magmatism favor assimilation of hydrothermally isotopes in zircon. Precambrian Res. 281, 312–337. ff altered rocks in magmatic settings characterized by syn-magmatic Ashwal, L., Demai e, D., Torsvik, T., 2002. Petrogenesis of Neoproterozoic granitoids and related rocks from the Seychelles: the case for an Andean-type arc origin. J. Petrol. alteration. Assimilation of rocks with syn-magmatic alteration ap- 43, 45–83. 18 pears to result in larger volumes of low-δ O magma. Auer, S., Bindeman, I., Wallace, P., Ponomareva, V., Portnyagin, M., 2009. The origin of 18 (3) The most significant 18O-depletions in silicic magmas are found in hydrous, high-δ O voluminous volcanism: diverse oxygen isotope values and high magmatic water contents within the volcanic record of Klyuchevskoy volcano, hotspot and rift settings associated with mantle-upwelling and high

18 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

Kamchatka, Russia. Contrib. Mineral. Petrol. 157, 209. California. Can. Mineral. 35, 425–452. Bachmann, O., Bergantz, G.W., 2008. Rhyolites and their Source Mushes across Tectonic Boroughs, S., Wolff, J., Bonnichsen, B., Godchaux, M., Larson, P., 2005. Large-volume, Settings. J. Petrol. 49, 2277–2285. low-δ18O rhyolites of the central Snake River Plain, Idaho, USA. Geology 33, Bachmann, O., Huber, C., 2016. Silicic magma reservoirs in the Earth’s crust. Am. 821–824. Mineral. 101, 2377–2404. Boroughs, S., Wolff, J., Ellis, B., Bonnichsen, B., Larson, P., 2012. Evaluation of models for Bacon, C.R., 1983. Eruptive history of Mount Mazama and Crater Lake caldera, Cascade the origin of Miocene low-δ18O rhyolites of the Yellowstone/Columbia River Large Range, USA. J. Volcanol. Geotherm. Res. 18, 57–115. Igneous Province. Earth Planet. Sci. Lett. 313, 45–55. Bacon, C.R., Lowenstern, J.B., 2005. Late Pleistocene granodiorite source for recycled Bowen, G.J., Revenaugh, J., 2003. Interpolating the isotopic composition of modern zircon and phenocrysts in rhyodacite lava at Crater Lake, Oregon. Earth Planet. Sci. meteoric precipitation. Water Resour. Res. 39. Lett. 233, 277–293. Branney, M.J., Bonnichsen, B., Andrews, G.D.M., Ellis, B., Barry, T.L., McCurry, M., 2008. Bacon, C.R., Adami, L.H., Lanphere, M.A., 1989. Direct evidence for the origin of low-18O ‘Snake River (SR)-type’ volcanism at the Yellowstone hotspot track: distinctive pro- silicic magmas: quenched samples of a magma chamber's partially-fused granitoid ducts from unusual, high-temperature silicic super-eruptions. Bull. Volcanol. 70, walls, Crater Lake, Oregon. Earth Planet. Sci. Lett. 96, 199–208. 293–314. Bacon, C.R., Persing, H.M., Wooden, J.L., Ireland, T.R., 2000. Late Pleistocene grano- Bryan, S.E., Peate, I.U., Peate, D.W., Self, S., Jerram, D.A., Mawby, M.R., Marsh, J.G., diorite beneath Crater Lake caldera, Oregon, dated by ion microprobe. Geology 28, Miller, J.A., 2010. The largest volcanic eruptions on Earth. Earth Sci. Rev. 102, 467–470. 207–229. Balsley, S.D., Gregory, R.T., 1998. Low 18O silicic magmas: why are they so rare? Earth Bucholz, C.E., Jagoutz, O., Schmidt, M.W., Sambuu, O., 2014. Fractional crystallization of Planet. Sci. Lett. 162, 123–136. high-K arc magmas: biotite-versus amphibole-dominated fractionation series in the Beard, J.S., Ragland, P.C., Crawford, M.L., 2005. Reactive bulk assimilation: a model for Dariv Igneous complex, Western Mongolia. Contrib. Mineral. Petrol. 168, 1072. crust-mantle mixing in silicic magmas. Geology 33, 681–684. Bucholz, C.E., Jagoutz, O., VanTongeren, J.A., Setera, J., Wang, Z., 2017. Oxygen isotope Befus, K.S., Gardner, J.E., 2016. Magma storage and evolution of the most recent effusive trajectories of crystallizing melts: Insights from modeling and the plutonic record. and explosive eruptions from Yellowstone Caldera. Contrib. Mineral. Petrol. 171, Geochim. Cosmochim. Acta 207, 154–184. 1–19. Budd, D.A., Troll, V.R., Deegan, F.M., Jolis, E.M., Smith, V.C., Whitehouse, M.J., Harris, Berg, S., Troll, R., Harris, C., Deegan, F., Riishuus, M., Burchardt, S., Krumbholz, M., C., Freda, C., Hilton, D.R., Halldórsson, S.A., 2017. Magma reservoir dynamics at 2018. Exceptionally high whole-rock δ18O values in intra-caldera rhyolites from Toba caldera, Indonesia, recorded by oxygen isotope zoning in quartz. Sci. Rep. 7, Northeast Iceland. Mineral. Mag. 82, 1147–1168. 1–11. Betton, P.J., 1978. Geochemistry of Karoo volcanic rocks of Swaziland. PhD thesis. Byers, F., Carr, W., Orkild, P.P., 1989. Volcanic centers of southwestern Nevada: University of Oxford. Evolution of understanding, 1960–1988. J. Geophys. Res. Solid Earth 94, 5908–5924. Bindeman, I., 2008. Oxygen isotopes in mantle and crustal magmas as revealed by single Carley, T.L., Miller, C., Sigmarsson, O., Coble, M., Fisher, C., Hanchar, J., Schmitt, A., crystal analysis. Rev. Mineral. Geochem. 69, 445–478. Economos, R., 2017. Detrital zircon resolve longevity and evolution of silicic mag- Bindeman, I., Simakin, A., 2014. Rhyolites—Hard to produce, but easy to recycle and matism in extinct volcanic centers: A case study from the East Fjords of Iceland. sequester: Integrating microgeochemical observations and numerical models. Geosphere 13, 1640–1663. Geosphere 10, 930–957. Carley, T.L., Miller, C.F., Fisher, C.M., Hanchar, J.M., Vervoort, J.D., Schmitt, A.K., Bindeman, I.N., Valley, J.W., 2000. Formation of low-δ18O rhyolites after caldera collapse Economos, R.C., Jordan, B.T., Padilla, A.J., Banik, T.J., 2020. Petrogenesis of Silicic at Yellowstone, Wyoming, USA. Geology 28, 719–722. Magmas in Iceland through Space and Time: The Isotopic Record Preserved in Zircon Bindeman, I.N., Valley, J.W., 2001. Low-δ18O Rhyolites from Yellowstone: Magmatic and Whole Rocks. J. Geol. 128, 1–28. evolution based on analyses of Zircons and individual phenocrysts. J. Petrol. 42, Carmichael, I.S.E., 1964. The petrology of Thingmuli, a Tertiary volcano in eastern 1491–1517. Iceland. J. Petrol. 5, 435–460. Bindeman, I.N., Valley, J.W., 2003. Rapid generation of both high- and low-δ18O, large- Carracedo, J.C., Day, S., Guillou, H., Badiola, E.R., Canas, J., Torrado, F.P., 1998. Hotspot volume silicic magmas at the Timber Mountain/Oasis Valley caldera complex, volcanism close to a passive continental margin: the Canary Islands. Geol. Mag. 135, Nevada. Geol. Soc. Am. Bull. 115, 581–595. 591–604. Bindeman, I., Brooks, C., McBirney, A., Taylor, H., 2008. The low-δ18O late-stage ferro- Cathey, H.E., Nash, B.P., 2009. Pyroxene thermometry of rhyolite lavas of the diorite magmas in the Skaergaard Intrusion: Result of liquid immiscibility, thermal Bruneau–Jarbidge eruptive center, Central Snake River Plain. J. Volcanol. Geotherm. metamorphism, or meteoric water incorporation into magma? J. Geol. 116, 571–586. Res. 188, 173–185. Bindeman, I.N., Fournelle, J.H., Valley, J.W., 2001. Low-δ18O tephra from a composi- Cerling, T.E., Brown, F.H., Bowman, J.R., 1985. Low-temperature alteration of volcanic tionally zoned magma body: Fisher Caldera, Unimak Island, Aleutians. J. Volcanol. glass: hydration, Na, K, 18O and Ar mobility. Chem. Geol. 52, 281–293. Geotherm. Res. 111, 35–53. Chen, Y.-X., Zheng, Y.-F., Chen, R.-X., Zhang, S.-B., Li, Q., Dai, M., Chen, L., 2011. Bindeman, I.N., Ponomareva, V.V., Bailey, J.C., Valley, J.W., 2004. Volcanic arc of Metamorphic growth and recrystallization of zircons in extremely 18O-depleted rocks Kamchatka: a province with high-δ18O magma sources and large-scale 18O/16O de- during eclogite-facies metamorphism: evidence from U–Pb ages, trace elements, and pletion of the upper crust. Geochim. Cosmochim. Acta 68, 841–865. O–Hf isotopes. Geochim. Cosmochim. Acta 75, 4877–4898. Bindeman, I.N., Schmitt, A.K., Valley, J.W., 2006. U–Pb zircon geochronology of silicic Cheong, A.C.-s., Sohn, Y.K., Jeong, Y.-J., Jo, H.J., Park, K.-H., Lee, Y.S., Li, X.-H., 2017. tuffs from the Timber Mountain/Oasis Valley caldera complex, Nevada: rapid gen- Latest Pleistocene crustal cannibalization at Baekdusan (Changbaishan) as traced by eration of large volume magmas by shallow-level remelting. Contrib. Mineral. Petrol. oxygen isotopes of zircon from the Millennium Eruption. Lithos 284, 132–137. 152, 649–665. Christiansen, E.H., McCurry, M., 2008. Contrasting origins of Cenozoic silicic volcanic Bindeman, I.N., Watts, K.E., Schmitt, A.K., Morgan, L.A., Shanks, P.W., 2007. Voluminous rocks from the western Cordillera of the United States. Bull. Volcanol. 70, 251–267. low δ18O magmas in the late Miocene Heise volcanic field, Idaho: Implications for the Christiansen, R.L., Lipman, P.W., Carr, W., Byers Jr., F., Orkild, P.P., Sargent, K., 1977. fate of Yellowstone hotspot calderas. Geology 35, 1019–1022. Timber Mountain–Oasis Valley caldera complex of southern Nevada. Geol. Soc. Am. Bindeman, I., Leonov, V., Izbekov, P., Ponomareva, V., Watts, K., Shipley, N., Perepelov, Bull. 88, 943–959. A., Bazanova, L., Jicha, B., Singer, B., 2010a. Large-volume silicic volcanism in Cleverly, R., Betton, P., Bristow, J., 1984. Geochemistry and petrogenesis of the Lebombo Kamchatka: Ar–Ar and U–Pb ages, isotopic, and geochemical characteristics of major rhyolites. Petrogenesis of the volcanic rocks of the Karoo Province. Special pre-Holocene caldera-forming eruptions. J. Volcanol. Geotherm. Res. 189, 57–80. Publications of the Geological Society of South Africa 13, 171–194. Bindeman, I., Schmitt, A., Evans, D., 2010b. Limits of hydrosphere-lithosphere interac- Colón, D.P., Bindeman, I.N., Ellis, B.S., Schmitt, A.K., Fisher, C.M., 2015. Hydrothermal tion: Origin of the lowest-known δ18O silicate rock on Earth in the Paleoproterozoic alteration and melting of the crust during the Columbia River Basalt–Snake River Karelian rift. Geology 38, 631–634. Plain transition and the origin of low-δ18O rhyolites of the central Snake River Plain. Bindeman, I.N., Fu, B., Kita, N.T., Valley, J.W., 2008. Origin and evolution of silicic Lithos 224, 310–323. magmatism at Yellowstone based on ion microprobe analysis of isotopically zoned Colón, D.P., Bindeman, I.N., Wotzlaw, J.-F., Christiansen, E.H., Stern, R.A., 2018. Origins zircons. J. Petrol. 49, 163–193. and evolution of rhyolitic magmas in the central Snake River Plain: insights from Bindeman, I., Gurenko, A., Carley, T., Miller, C., Martin, E., Sigmarsson, O., 2012. Silicic coupled high-precision geochronology, oxygen isotope, and hafnium isotope analyses magma petrogenesis in Iceland by remelting of hydrothermally altered crust based on of zircon. Contrib. Mineral. Petrol. 173, 11. oxygen isotope diversity and disequilibria between zircon and magma with im- Colón, D.P., Bindeman, I.N., Gerya, T.V., 2019. Understanding the isotopic and chemical plications for MORB. Terra Nova 24, 227–232. evolution of Yellowstone hot spot magmatism using magmatic-thermomechanical Blattner, P., Reid, F., 1982. The origin of lavas and ignimbrites of the Taupo Volcanic modeling. J. Volcanol. Geotherm. Res. 370, 13–30. Zone, New Zealand, in the light of oxygen isotope data. Geochim. Cosmochim. Acta Condie, K.C., 2002. The supercontinent cycle: are there two patterns of cyclicity? J. Afr. 46, 1417–1429. Earth Sci. 35, 179–183. Bohrson, W.A., Spera, F.J., 2001. Energy-constrained open-system magmatic processes II: Condomines, M., Grönvold, K., Hooker, P., Muehlenbachs, K., o'Nions, R., Oskarsson, N., application of energy-constrained assimilation–fractional crystallization (EC-AFC) Oxburgh, E., 1983. Helium, oxygen, strontium and neodymium isotopic relationships model to magmatic systems. J. Petrol. 42, 1019–1041. in Icelandic volcanics. Earth Planet. Sci. Lett. 66, 125–136. Bolte, T., Holtz, F., Almeev, R., Nash, B., 2015. The Blacktail Creek Tuff: an analytical and Criss, R., Taylor, H., 1983. An 18O/16O and D/H study of Tertiary hydrothermal systems experimental study of rhyolites from the Heise volcanic field, Yellowstone hotspot in the southern half of the Idaho batholith. Geol. Soc. Am. Bull. 94, 640–663. system. Contrib. Mineral. Petrol. 169, 1–24. Criss, R., Ekren, E., Hardyman, R., 1984. Casto Ring Zone: A 4,500-km2 fossil hydro- Bonnichsen, B., Godchaux, M., 2002. Late Miocene, Pliocene, and Pleistocene geology of thermal system in the Challis , central Idaho. Geology 12, 331–334. southwestern Idaho with emphasis on basalts in the Bruneau–Jarbidge, Twin Falls, Curtis, C.G., Harris, C., Trumbull, R.B., De Beer, C., Mudzanani, L., 2013. Oxygen isotope and western Snake River Plain regions. Tectonic and Magmatic Evolution of the diversity in the anorogenic Koegel Fontein complex of South Africa: a case for Snake River Plain Volcanic Province. Idaho Geol. Surv. Bull. 30, 233–312. basement control and selective melting for the production of low-δ18O magmas. J. Borg, L.E., Clynne, M.A., Bullen, T.D., 1997. The variable role of slab-derived fluids in the Petrol. 54, 1259–1283. generation of a suite of primitive calc-alkaline lavas from the southernmost Cascades, Davies, D., Francis, T., 1964. The crustal structure of the Seychelles Bank. In: Deep Sea

19 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

Research and Oceanographic Abstracts. Elsevier, pp. 921–927. southern Andes. Contrib. Mineral. Petrol. 95, 71–81. De Beer, C., Armstrong, R., 1998. Age and tectonic setting of Mesozoic anorogenic Gualda, G.A., Ghiorso, M.S., Lemons, R.V., Carley, T.L., 2012. Rhyolite-MELTS: a mod- complex west of Bitterfontein, Namaqualand, South Africa, IAVCEI International ified calibration of MELTS optimized for silica-rich, fluid-bearing magmatic systems. Volcanological Congress. Magmatic Diversity: Volcanoes and their roots. pp. 15. J. Petrol. 53, 875–890. Deering, C., Gravley, D., Vogel, T., Cole, J., Leonard, G., 2010. Origins of cold-wet-oxi- Gunnarsson, B., Marsh, B.D., Taylor, H.P., 1998. Generation of Icelandic rhyolites: silicic dizing to hot-dry-reducing rhyolite magma cycles and distribution in the Taupo lavas from the Torfajökull central volcano. J. Volcanol. Geotherm. Res. 83, 1–45. Volcanic Zone, New Zealand. Contrib. Mineral. Petrol. 160, 609–629. Gurenko, A.A., Bindeman, I.N., Sigurdsson, I.A., 2015. To the origin of Icelandic rhyolites: Donoghue, E., Troll, V.R., Harris, C., O'Halloran, A., Walter, T.R., Torrado, F.J.P., 2008. insights from partially melted leucocratic xenoliths. Contrib. Mineral. Petrol. 169, 49. Low-temperature hydrothermal alteration of intra-caldera tuffs, Miocene Tejeda Hardyman, R., 1985. The Twin Peaks caldera and associated ore deposits. U.S. Geol. Surv. caldera, Gran Canaria, Canary Islands. J. Volcanol. Geotherm. Res. 176, 551–564. Bull. 97–106. Drew, D.L., Bindeman, I.N., Watts, K.E., Schmitt, A.K., Fu, B., McCurry, M., 2013. Crustal- Harris, C., Ashwal, L.D., 2002. The origin of low δ18O granites and related rocks from the scale recycling in caldera complexes and rift zones along the Yellowstone hotspot Seychelles. Contrib. Mineral. Petrol. 143, 366–376. track: O and Hf isotopic evidence in diverse zircons from voluminous rhyolites of the Harris, C., Erlank, A.J., 1992. The production of large-volume, low-δ18O rhyolites during Picabo volcanic field, Idaho. Earth Planet. Sci. Lett. 381, 63–77. the rifting of Africa and Antarctica: the Lebombo Monocline, southern Africa. Du Toit, A.L., 1937. Our Wandering Continents: an Hypothesis of Continental Drifting. Geochim. Cosmochim. Acta 56, 3561–3570. Oliver and Boyd, Edinburgh. Harris, C., Smith, H.S., le Roex, A.P., 2000. Oxygen isotope composition of phenocrysts Duncan, R.A., Hooper, P., Rehacek, J., Marsh, J., Duncan, A., 1997. The timing and from Tristan da Cunha and Gough Island lavas: variation with fractional crystal- duration of the Karoo igneous event, southern Gondwana. J. Geophys. Res. Solid lization and evidence for assimilation. Contrib. Mineral. Petrol. 138, 164–175. Earth 102, 18127–18138. Harris, C., Mulder, K., Sarkar, S., Whitehead, B., Roopnarain, S., 2018. Petrogenesis of Edgar, C.J., Wolff, J.A., Olin, P., Nichols, H.J., Pittari, A., Cas, R.A.F., Reiners, P., Spell, T., low-δ18O quartz porphyry dykes, Koegel Fontein complex, South Africa. Contrib. Martí, J., 2007. The late Quaternary Diego Hernandez Formation, Tenerife: Mineral. Petrol. 173, 30. Volcanology of a complex cycle of voluminous explosive phonolitic eruptions. J. Hattori, K., Muehlenbachs, K., 1982. Oxygen isotope ratios of the Icelandic crust. J. Volcanol. Geotherm. Res. 160, 59–85. Geophys. Res. Solid Earth 87, 6559–6565. Eiler, J.M., 2001. Oxygen isotope variations of basaltic lavas and upper mantle rocks. Rev. Hawkesworth, C., Cawood, P., Kemp, T., Storey, C., Dhuime, B., 2009. A matter of pre- Mineral. Geochem. 43, 319–364. servation. Science 323, 49–50. Eiler, J.M., Grönvold, K., Kitchen, N., 2000. Oxygen isotope evidence for the origin of Hemond, C., Condomines, M., Fourcade, S., Allegre, C., Oskarsson, N., Javoy, M., 1988. chemical variations in lavas from Theistareykir volcano in Iceland’s northern volcanic Thorium, strontium and oxygen isotopic geochemistry in recent tholeiites from zone. Earth Planet. Sci. Lett. 184, 269–286. Iceland: crustal influence on mantle-derived magmas. Earth Planet. Sci. Lett. 87, Elders, W.A., Friðleifsson, G.Ó., Zierenberg, R.A., Pope, E.C., Mortensen, A.K., 273–285. Guðmundsson, Á., Lowenstern, J.B., Marks, N.E., Owens, L., Bird, D.K., 2011. Origin Hildreth, W., 1981. Gradients in silicic magma chambers: implications for lithospheric of a rhyolite that intruded a geothermal well while drilling at the Krafla volcano, magmatism. J. Geophys. Res. Solid Earth 86, 10153–10192 (1978–2012). Iceland. Geology 39, 231–234. Hildreth, W., Moorbath, S., 1988. Crustal contributions to arc magmatism in the Andes of Ellis, B.S., Barry, T., Branney, M.J., Wolff, J.A., Bindeman, I., Wilson, R., Bonnichsen, B., central . Contrib. Mineral. Petrol. 98, 455–489. 2010. Petrologic constraints on the development of a large-volume, high tempera- Hildreth, W., Christiansen, R.L., O'Neil, J.R., 1984. Catastrophic isotopic modification of ture, silicic magma system: The Twin Falls eruptive centre, central Snake River Plain. rhyolitic magma at times of caldera , Yellowstone Volcanic Field. Lithos 120, 475–489. J. Geophys. Res. 89, 8339. Ellis, B., Wolff, J., Boroughs, S., Mark, D., Starkel, W., Bonnichsen, B., 2013. Rhyolitic Honjo, N., Bonnichsen, B., Leeman, W.P., Stormer, J.C., 1992. Mineralogy and geother- volcanism of the central Snake River Plain: a review. Bull. Volcanol. 75, 1–19. mometry of high-temperature rhyolites from the central and western Snake River Ellis, B.S., Szymanowski, D., Wotzlaw, J.F., Schmitt, A.K., Bindeman, I.N., Troch, J., Plain. Bull. Volcanol. 54, 220–237. Harris, C., Bachmann, O., Guillong, M., 2017. Post-caldera volcanism at the heise Huber, C., Bachmann, O., Dufek, J., 2011. Thermo-mechanical reactivation of locked volcanic field: implications for petrogenetic models. J. Petrol. 58, 115–136. crystal mushes: Melting-induced internal fracturing and assimilation processes in Farmer, G.L., Broxton, D.E., Warren, R.G., Pickthorn, W., 1991. Nd, Sr, and O isotopic magmas. Earth Planet. Sci. Lett. 304, 443–454. variations in metaluminous ash-flow tuffs and related volcanic rocks at the Timber Hudak, M.R., Bindeman, I.N., 2018. Conditions of pinnacle formation and glass hydration Mountain/Oasis Valley Caldera, Complex, SW Nevada: Implications for the origin in cooling ignimbrite sheets from H and O isotope systematics at Crater Lake and the and evolution of large-volume silicic magma bodies. Contrib. Mineral. Petrol. 109, Valley of Ten Thousand Smokes. Earth Planet. Sci. Lett. 500, 56–66. 53–68. Ingebritsen, S., Manning, C.E., 1999. Geological implications of a permeability-depth Feeley, T., Sharp, Z., 1995. 18O/16O isotope geochemistry of silicic lava flows erupted curve for the continental crust. Geology 27, 1107–1110. from Volcán Ollagüe, Andean Central Volcanic Zone. Earth Planet. Sci. Lett. 133, Jagoutz, O.E., 2010. Construction of the granitoid crust of an island arc. Part II: a 239–254. quantitative petrogenetic model. Contrib. Mineral. Petrol. 160, 359–381. Finney, B., Turner, S., Hawkesworth, C., Larsen, J., Nye, C., George, R., Bindeman, I., Jellinek, A.M., 1994. The Twin Peaks caldera, Challis, Idaho: a unique window into the Eichelberger, J., 2008. Magmatic differentiation at an island-arc caldera: Okmok emplacement and evolution of a caldera-filling ignimbrite. University of Idaho. Volcano, Aleutian Islands, Alaska. J. Petrol. 49, 857–884. Jo, H.J., Chang-sik Cheong, A., Ryu, J.-S., Kim, N., Yi, K., Jung, H., Li, X.-H., 2016. In-situ Fisher, F.S., McIntyre, D.H., Johnson, K.M., 1992. Geologic map of the Challis 1° x 2° oxygen isotope records of crustal self-cannibalization selectively captured by zircon quadrangle, Idaho: U.S. Geological Survey Map I-1819, scale 1:250,000. crystals from high-δ26Mg granitoids. Geology 44, 339–342. Folkes, C.B., de Silva, S.L., Bindeman, I.N., Cas, R.A., 2013. Tectonic and climate history Johannes, W., Holtz, F., 1996. Petrogenesis and Experimental Petrology of Granitic Rocks. influence the geochemistry of large-volume silicic magmas: New δ18O data from the Springer, Berlin. Central Andes with comparison to N America and Kamchatka. J. Volcanol. Geotherm. John, D.A., Pickthorn, W.J., 1996. Alteration and stable isotope studies of a deep me- Res. 262, 90–103. teoric-hydrothermal system in the Job Canyon caldera and IXL pluton, southern Forester, R.W., Taylor Jr., H.P., 1976. 18O-depleted igneous rocks from the Tertiary Stillwater Range, Nevada. In: Coyner, A., Fahey, P. (Eds.), Geology and Ore Deposits complex of the Isle of Mull, Scotland. Earth Planet. Sci. Lett. 32, 11–17. of the American Cordillera: Reno/Sparks. Nevada, Geological Society of Nevada Friedman, I., Lipman, P.W., Obradovich, J.D., Gleason, J.D., Christiansen, R.L., 1974. Symposium Proceedings, pp. 733–756. Meteoric water in magmas. Science 184, 1069–1072. Jónasson, K., 1994. Rhyolite volcanism in the Krafla central volcano, north-east Iceland. Frost, C.D., Frost, B.R., 2010. On ferroan (A-type) granitoids: their compositional varia- Bull. Volcanol. 56, 516–528. bility and modes of origin. J. Petrol. 52, 39–53. Kabete, J., Groves, D., Mcnaughton, N., Dunphy, J., 2006. The geology, SHRIMP U–Pb Frost, B.R., Barnes, C.G., Collins, W.J., Arculus, R.J., Ellis, D.J., Frost, C.D., 2001. A geochronology and metallogenic significance of the Ankisatra-Besakay District, geochemical classification for granitic rocks. J. Petrol. 42, 2033–2048. Andriamena belt, northern Madagascar. J. Afr. Earth Sci. 45, 87–122. Fu, B., Kita, N.T., Wilde, S.A., Liu, X., Cliff, J., Greig, A., 2013. Origin of the Tongbai- Karakas, O., Dufek, J., 2015. Melt evolution and residence in extending crust: Thermal Dabie-Sulu Neoproterozoic low-δ18O igneous province, east-central China. Contrib. modeling of the crust and crustal magmas. Earth Planet. Sci. Lett. 425, 131–144. Mineral. Petrol. 165, 641–662. Kempe, D.R.C., Deer, W.A., 1970. Geological Investigations in East Greenland; Part 9: the Fu, B., Cliff, J., Zartman, R.E., 2014. Zircon oxygen isotopic constraints from plutonic Mineralogy of the Kangerdlugssuaq Alkaline Intrusion, East Greenland. CA Reitzel. rocks on the magmatic and crustal evolution of the northern Appalachians in Knott, T.R., Branney, M.J., Reichow, M.K., Finn, D.R., Coe, R.S., Storey, M., Barfod, D., southern New England, USA. Can. J. Earth Sci. 51, 485–499. McCurry, M., 2016. Mid-Miocene record of large-scale Snake River–type explosive Furman, T., Frey, F.A., Meyer, P.S., 1992. Petrogenesis of evolved basalts and rhyolites at volcanism and associated subsidence on the Yellowstone hotspot track: The Cassia Austurhorn, southeastern Iceland: the role of fractional crystallization. J. Petrol. 33, Formation of Idaho, USA. Bulletin 128, 1121–1146. 1405–1445. Kroener, A., Hegner, E., Collins, A.S., Windley, B.F., Brewer, T.S., Razakamanana, T., Garden, T.O., Chambefort, I., Gravley, D.M., Deering, C., Kennedy, B.M., 2020. Pidgeon, R.T., 2000. Age and magmatic history of the Antananarivo Block, central Reconstruction of the fossil hydrothermal system at Lake City caldera, Colorado, Madagascar, as derived from zircon geochronology and Nd isotopic systematics. Am. USA: Constraints for caldera-hosted geothermal systems. J. Volcanol. Geotherm. Res. J. Sci. 300, 251–288. 393, 1–23 106794. Larson, P.B., Geist, D.J., 1995. On the origin of low-18O magmas: Evidence from the Casto Gilliam, C.E., Valley, J.W., 1997. Low δ18O magma, Isle of Skye, Scotland: evidence from pluton, Idaho. Geology 23, 909–912. zircons. Geochim. Cosmochim. Acta 61, 4975–4981. Larson, P.B., Taylor, H.P., 1986. An oxygen isotope study of hydrothermal alteration in Gregory, L.C., Meert, J.G., Bingen, B., Pandit, M.K., Torsvik, T.H., 2009. Paleomagnetism the Lake City caldera, San Juan Mountains, Colorado. J. Volcanol. Geotherm. Res. 30, and geochronology of the Malani Igneous Suite, Northwest India: implications for the 47–82. configuration of Rodinia and the assembly of Gondwana. Precambrian Res. 170, Leeman, W.P., Annen, C., Dufek, J., 2008. Snake River Plain - Yellowstone silicic vol- 13–26. canism: implications for magma genesis and magma fluxes. Geol. Soc. Lond., Spec. Grunder, A.L., 1987. Low δ18O silicic volcanic rocks at the Calabozos Caldera Complex, Publ. 304, 235–259.

20 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

Li, Y., Ma, C.-Q., Xing, G.-F., Zhou, H.-W., Zhang, H., Brouwer, F.M., 2015. Origin of a Rumble, D., Yui, T.-F., 1998. The Qinglongshan oxygen and hydrogen isotope anomaly Cretaceous low-18O granitoid complex in the active continental margin of SE China. near Donghai in Jiangsu Province, China. Geochim. Cosmochim. Acta 62, Lithos 216, 136–147. 3307–3321. Lipman, P.W., Friedman, I., 1975. Interaction of meteoric water with magma: An oxygen- Rumble, D., Giorgis, D., Ireland, T., Zhang, Z., Xu, H., Yui, T., Yang, J., Xu, Z., Liou, J., isotope study of ash-flow sheets from southern Nevada. Geol. Soc. Am. Bull. 86, 2002. Low δ18O zircons, U-Pb dating, and the age of the Qinglongshan oxygen and 695–702. hydrogen isotope anomaly near Donghai in Jiangsu Province, China. Geochim. Lipman, P., Prostka, H., Christiansen, R., 1972. Cenozoic volcanism and plate-tectonic Cosmochim. Acta 66, 2299–2306. evolution of the Western United States. I. Early and middle Cenozoic. Phil. Trans. R. Schattel, N., Portnyagin, M., Golowin, R., Hoernle, K., Bindeman, I., 2014. Contrasting Soc. Lond. A 271, 217–248. conditions of rift and off-rift silicic magma origin on Iceland. Geophys. Res. Lett. 41, Loewen, M.W., Bindeman, I.N., 2016. Oxygen isotope thermometry reveals high mag- 5813–5820. matic temperatures and short residence times in Yellowstone and other hot-dry Seligman, A., Bindeman, I., Jicha, B., Ellis, B., Ponomareva, V., Leonov, V., 2014a. Multi- rhyolites compared to cold-wet systems. Am. Mineral. 101, 1222–1227. cyclic and isotopically diverse silicic magma generation in an arc volcano: gorely Lowenstern, J.B., Hurwitz, S., 2008. Monitoring a supervolcano in repose: Heat and vo- eruptive center, Kamchatka, Russia. J. Petrol. egu034. latile flux at the Yellowstone Caldera. Elements 4, 35–40. Seligman, A.N., Bindeman, I.N., McClaughry, J., Stern, R.A., Fisher, C., 2014. The earliest Lowenstern, J.B., Smith, R.B., Hill, D.P., 2006. Monitoring super-volcanoes: geophysical low and high δ18O caldera-forming eruptions of the Yellowstone plume: implications and geochemical signals at Yellowstone and other large caldera systems. Philos Trans for the 30–40 Ma Oregon calderas and speculations on plume-triggered delamina- A Math Phys Eng Sci 364, 2055–2072. tions. Front. Earth Sci. 2, 34. MacDonald, R., McGarvie, D., Pinkerton, H., Smith, R., Palacz, A., 1990. Petrogenetic Sheppard, S.M., Harris, C., 1985. Hydrogen and oxygen isotope geochemistry of evolution of the Torfajökull Volcanic Complex, Iceland I. Relationship between the Ascension Island lavas and granites: variation with crystal fractionation and inter- magma types. J. Petrol. 31, 429–459. action with sea water. Contrib. Mineral. Petrol. 91, 74–81. Martí, J., Mitjavila, J., Araña, V., 1994. Stratigraphy, structure and geochronology of the Shieh, Y., Taylor, H., 1969. Oxygen and hydrogen isotope studies of contact meta- Las Cañadas caldera (Tenerife, Canary Islands). Geol. Mag. 131, 715–727. morphism in the Santa Rosa Range, Nevada and other areas. Contrib. Mineral. Petrol. Martin, E., Sigmarsson, O., 2007. Crustal thermal state and origin of silicic magma in 20, 306–356. Iceland: the case of Torfajökull, Ljósufjöll and Snæfellsjökull volcanoes. Contrib. Sigmarsson, O., Hémond, C., Condomines, M., Fourcade, S., Oskarsson, N., 1991. Origin Mineral. Petrol. 153, 593–605. of silicic magma in Iceland revealed by Th isotopes. Geology 19, 621–624. Martin, E., Sigmarsson, O., 2010. Thirteen million years of silicic magma production in Sigmarsson, O., Condomines, M., Fourcade, S., 1992. A detailed Th, Sr and O isotope Iceland: Links between petrogenesis and tectonic settings. Lithos 116, 129–144. study of Hekla: differentiation processes in an Icelandic volcano. Contrib. Mineral. Matsuhisa, Y., 1979. Oxygen isotopic compositions of volcanic rocks from the East Japan Petrol. 112, 20–34. island arcs and their bearing on petrogenesis. J. Volcanol. Geotherm. Res. 5, Sigurdsson, H., Sparks, R., 1981. Petrology of rhyolitic and mixed magma ejecta from the 271–296. 1875 eruption of Askja, Iceland. J. Petrol. 22, 41–84. McCulloch, M., Kyser, T., Woodhead, J., Kinsley, L., 1994. Pb−Sr−Nd−O isotopic Simakin, A., Bindeman, I., 2012. Remelting in caldera and rift environments and the constraints on the origin of rhyolites from the Taupo Volcanic Zone of New Zealand: genesis of hot,“recycled” rhyolites. Earth Planet. Sci. Lett. 337, 224–235. evidence for assimilation followed by fractionation from basalt. Contrib. Mineral. Simon, L., Lécuyer, C., 2005. Continental recycling: the oxygen isotope point of view. Petrol. 115, 303–312. Geochem. Geophys. Geosyst. 6, 1–10. Miller, J.A., Harris, C., 2006. Petrogenesis of the Swaziland and northern Natal rhyolites Smith, R.B., Braile, L.W., 1994. The Yellowstone hotspot. J. Volcanol. Geotherm. Res. 61, of the Lebombo rifted volcanic margin, south east Africa. J. Petrol. 48, 185–218. 121–187. Monani, S., Valley, J.W., 2001. Oxygen isotope ratios of zircon: magma genesis of low Smithies, R., Howard, H., Kirkland, C., Korhonen, F., Medlin, C., Maier, W.D., de δ18O granites from the British Tertiary Igneous Province, western Scotland. Earth Gromard, R.Q., Wingate, M., 2015a. Piggy-back supervolcanoes—long-lived, volu- Planet. Sci. Lett. 184, 377–392. minous, juvenile rhyolite volcanism in Mesoproterozoic central Australia. J. Petrol. Muehlenbachs, K., Byerly, G., 1982. 18O-enrichment of silicic magmas caused by crystal 56, 735–763. fractionation at the Galapagos spreading center. Contrib. Mineral. Petrol. 79, 76–79. Smithies, R., Kirkland, C., Cliff, J., Howard, H., de Gromard, R.Q., 2015. Syn-volcanic Muehlenbachs, K., Anderson, A.T., Sigvaldason, G.E., 1974. Low-O18 basalts from Iceland. cannibalisation of juvenile felsic crust: Superimposed giant 18O-depleted rhyolite Geochim. Cosmochim. Acta 38, 577–588. systems in the hot and thinned crust of Mesoproterozoic central Australia. Earth Myers, M.L., Wallace, P.J., Wilson, C.J.N., Morter, B.K., Swallow, E.J., 2016. Prolonged Planet. Sci. Lett. 424, 15–25. ascent and episodic venting of discrete magma batches at the onset of the Soosalu, H., Einarsson, P., 2004. Seismic constraints on magma chambers at Hekla and Huckleberry Ridge supereruption, Yellowstone. Earth Planet. Sci. Lett. 451, 285–297. Torfajökull volcanoes, Iceland. Bull. Volcanol. 66, 276–286. Nandedkar, R.H., Ulmer, P., Müntener, O., 2014. Fractional crystallization of primitive, Spencer, C., Roberts, N., Santosh, M., 2017. Growth, destruction, and preservation of hydrous arc magmas: an experimental study at 0.7 GPa. Contrib. Mineral. Petrol. 167, Earth's continental crust. Earth Sci. Rev. 172, 87–106. 1015. Spera, F.J., Bohrson, W.A., 2001. Energy-constrained open-system magmatic processes I: Nicholson, H., Condomines, M., Fitton, J.G., Fallick, A.E., Grönvold, K., Rogers, G., 1991. General model and energy-constrained assimilation and fractional crystallization Geochemical and isotopic evidence for crustal assimilation beneath Krafla, Iceland. J. (EC-AFC) formulation. J. Petrol. 42, 999–1018. Petrol. 32, 1005–1020. Sturkell, E., Einarsson, P., Sigmundsson, F., Geirsson, H., Olafsson, H., Pedersen, R., de Patiño Douce, A.E., 1997. Generation of metaluminous A-type granites by low-pressure Zeeuw-van Dalfsen, E., Linde, A.T., Sacks, S.I., Stefánsson, R., 2006. Volcano geodesy melting of calc-alkaline granitoids. Geology 25, 743–746. and magma dynamics in Iceland. J. Volcanol. Geotherm. Res. 150, 14–34. Pearce, J.A., Harris, N.B., Tindle, A.G., 1984. Trace element discrimination diagrams for Sveinbjornsdottir, A., Coleman, M., Yardley, B., 1986. Origin and history of hydrothermal the tectonic interpretation of granitic rocks. J. Petrol. 25, 956–983. fluids of the Reykjanes and Krafla geothermal fields, Iceland. Contrib. Mineral. Petrol. Peck, W.H., Valley, J.W., Graham, C.M., 2003. Slow oxygen diffusion rates in igneous 94, 99–109. zircons from metamorphic rocks. Am. Mineral. 88, 1003–1014. Tang, Y., Obayashi, M., Niu, F., Grand, S.P., Chen, Y.J., Kawakatsu, H., Tanaka, S., Ning, Pisarevsky, S.A., Wingate, M.T., Li, Z.-X., Wang, X.-C., Tohver, E., Kirkland, C.L., 2014. J., Ni, J.F., 2014. Changbaishan volcanism in northeast China linked to subduction- Age and paleomagnetism of the 1210 Ma Gnowangerup–Fraser dyke swarm, Western induced mantle upwelling. Nat. Geosci. 7, 470. Australia, and implications for late Mesoproterozoic paleogeography. Precambrian Taylor, H.P., 1968. The oxygen isotope geochemistry of igneous rocks. Contrib. Mineral. Res. 246, 1–15. Petrol. 19, 1–71. Plummer, P.S., 1995. Ages and geological significance of the igneous rocks from Taylor, H.P., 1974. The application of oxygen and hydrogen isotope studies to problems Seychelles. J. Afr. Earth Sci. 20, 91–101. of hydrothermal alteration and ore deposition. Econ. Geol. 69, 843–883. 18 Pope, E.C., Bird, D.K., Arnórsson, S., 2013. Evolution of low- O Icelandic crust. Earth Taylor, H.P., 1977. Water/rock interactions and the origin of H2O in granitic batholiths. J. Planet. Sci. Lett. 374, 47–59. Geol. Soc. 133, 509–558. Portnyagin, M., Hoernle, K., Storm, S., Mironov, N., van den Bogaard, C., Botcharnikov, Taylor, H.P., 1980. The effects of assimilation of country rocks by magmas on 18O/16O 87 86 R., 2012. H2O-rich melt inclusions in fayalitic olivine from Hekla volcano: and Sr/ Sr systematics in igneous rocks. Earth Planet. Sci. Lett. 47, 243–254. Implications for phase relationships in silicic systems and driving forces of explosive Taylor, H.P., 1986. Igneous rocks: II. Isotopic case studies of Circumpacific magmatism. volcanism on Iceland. Earth Planet. Sci. Lett. 357, 337–346. Rev. Mineral. Geochem. 16, 273–317. Potter, J., Longstaffe, F.J., Barr, S.M., Thompson, M.D., White, C.E., 2008. Altering Taylor, H.P., Forester, R.W., 1971. Low-18O igneous rocks from the intrusive complexes of Avalonia: oxygen isotopes and terrane distinction in the Appalachian peri- Skye, Mull, and Ardnamurchan, Western Scotland. J. Petrol. 12, 465–497. Gondwanan realm. Can. J. Earth Sci. 45, 815–825. Taylor, H.P., Forester, R.W., 1979. An oxygen and hydrogen isotope study of the Richards, M.A., Duncan, R.A., Courtillot, V.E., 1989. Flood basalts and hot-spot tracks: Skaergaard intrusion and its country rocks: a description of a 55 my old fossil hy- plume heads and tails. Science 246, 103–107. drothermal system. J. Petrol. 20, 355–419. Riishuus, M.S., Peate, D.W., Tegner, C., Wilson, J.R., Brooks, C.K., 2008. Petrogenesis of Tollan, P., Ellis, B., Troch, J., Neukampf, J., 2019. Assessing magmatic volatile equilibria cogenetic silica-oversaturated and-undersaturated syenites by periodic recharge in a through FTIR spectroscopy of unexposed melt inclusions and their host quartz: a new crustally contaminated magma chamber: the Kangerlussuaq intrusion, East technique and application to the Mesa Falls Tuff, Yellowstone. Contrib. Mineral. Greenland. J. Petrol. 49, 493–522. Petrol. 174, 24. Riishuus, M.S., Harris, C., Peate, D.W., Tegner, C., Wilson, J.R., Brooks, C.K., 2015. Troch, J., Ellis, B.S., Mark, D.F., Bindeman, I.N., Kent, A.J.R., Guillong, M., Bachmann, O., Formation of low-δ18O magmas of the Kangerlussuaq Intrusion by addition of water 2017. Rhyolite generation prior to a Yellowstone supereruption: insights from the derived from dehydration of foundered basaltic roof rocks. Contrib. Mineral. Petrol. Island Park-Mount Jackson rhyolite series. J. Petrol. 0, 1–24. 169, 1–16. Troch, J., Ellis, B.S., Harris, C., Ulmer, P., Bachmann, O., 2018. The effect of prior hy- Riley, T., Millar, I.L., Watkeys, M., Curtis, M.L., Leat, P., Klausen, M., Fanning, C., 2004. drothermal alteration on the melting behaviour during rhyolite formation in U–Pb zircon (SHRIMP) ages for the Lebombo rhyolites, South Africa: refining the Yellowstone, and its importance in the generation of low-δ18O magmas. Earth Planet. duration of Karoo volcanism. J. Geol. Soc. 161, 547–550. Sci. Lett. 481, 338–349.

21 J. Troch, et al. Earth-Science Reviews 208 (2020) 103299

Troch, J., Ellis, B., Harris, C., Ulmer, P., Bouvier, A.-S., Bachmann, O., 2020. Experimental Wickham, S.M., Alberts, A.D., Zanvilevich, A.N., Litvinovsky, B.A., Bindeman, I.N., melting of hydrothermally altered rocks: Constraints for the generation of low-δ18O Schauble, E.A., 1996. A stable isotope study of anorogenic magmatism in East Central rhyolites in the central Snake River Plain, USA. J. Petrol. 60 (10), 1881–1902. Asia. J. Petrol. 37, 1063–1095. Turi, B., Taylor, H.P., 1971. An oxygen and hydrogen isotope study of a granodiorite Wiesmaier, S., Troll, V.R., Carracedo, J.C., Ellam, R.M., Bindeman, I., Wolff, J.A., 2012. pluton from the Southern California batholith. Geochim. Cosmochim. Acta 35, Bimodality of lavas in the Teide–Pico Viejo succession in Tenerife—the role of crustal 383–406. melting in the origin of recent phonolites. J. Petrol. 53, 2465–2495. Valley, J., Lackey, J., Cavosie, A., Clechenko, C., Spicuzza, M., Basei, M., Bindeman, I., Wolff, J., Grandy, J., Larson, P., 2000. Interaction of mantle-derived magma with island Ferreira, V., Sial, A., King, E., 2005. 4.4 billion years of crustal maturation: oxygen crust? Trace element and oxygen isotope data from the Diego Hernandez Formation, isotope ratios of magmatic zircon. Contrib. Mineral. Petrol. 150, 561–580. Las Cañadas, Tenerife. J. Volcanol. Geotherm. Res. 103, 343–366. van Hinsbergen, D.J., de Groot, L.V., van Schaik, S.J., Spakman, W., Bijl, P.K., Sluijs, A., Wotzlaw, J.-F., Bindeman, I.N., Schaltegger, U., Brooks, C.K., Naslund, H.R., 2012. High- Langereis, C.G., Brinkhuis, H., 2015. A paleolatitude calculator for paleoclimate resolution insights into episodes of crystallization, hydrothermal alteration and re- studies. PLoS One 10, e0126946. melting in the Skaergaard intrusive complex. Earth Planet. Sci. Lett. 355, 199–212. Van Tongeren, J.A., Mathez, E.A., Kelemen, P.B., 2010. A felsic end to Bushveld differ- Yang, X.-A., Chen, Y.-C., Liu, S.-B., Hou, K.-J., Chen, Z.-Y., Liu, J.-J., 2015. U–Pb zircon entiation. J. Petrol. 51, 1891–1912. geochronology and geochemistry of Neoproterozoic granitoids of the Maevatanana Vazquez, J.A., Kyriazis, S.F., Reid, M.R., Sehler, R.C., Ramos, F.C., 2009. Thermochemical area, Madagascar: implications for Neoproterozoic crustal extension of the evolution of young rhyolites at Yellowstone: Evidence for a cooling but periodically Imorona–Itsindro Suite and subsequent lithospheric subduction. Int. Geol. Rev. 57, replenished postcaldera magma reservoir. J. Volcanol. Geotherm. Res. 188, 186–196. 1633–1649. Villiger, S., Ulmer, P., Müntener, O., Thompson, A.B., 2004. The liquid line of descent of Yang, Y.-N., Wang, X.-C., Li, Q.-L., Li, X.-H., 2016. Integrated in situ U–Pb age and Hf–O anhydrous, mantle-derived, tholeiitic liquids by fractional and equilibrium crystal- analyses of zircon from Suixian Group in northern Yangtze: New insights into the lization—an experimental study at 1.0 GPa. J. Petrol. 45, 2369–2388. Neoproterozoic low-δ18O magmas in the South China Block. Precambrian Res. 273, Wang, S.-J., Li, S.-G., Liu, S.-A., 2013. The origin and evolution of low-δ18O magma re- 151–164. corded by multi-growth zircons in granite. Earth Planet. Sci. Lett. 373, 233–241. Yang, W.B., Niu, H.C., Hollings, P., Zurevinski, S.E., Li, N.B., 2017. The role of recycled Wang, W., Cawood, P.A., Zhou, M.F., Pandit, M.K., Xia, X.P., Zhao, J.H., 2017. Low-δ18O oceanic crust in the generation of alkaline A-type granites. J. Geophys. Res. Solid Rhyolites From the Malani Igneous Suite: A Positive Test for South China and NW Earth 122, 9775–9783. India Linkage in Rodinia. Geophys. Res. Lett. 44. Zakharov, D., Bindeman, I., Tanaka, R., Friðleifsson, G., Reed, M., Hampton, R., 2019. Watson, E.B., Harrison, T.M., 1983. Zircon saturation revisited: temperature and com- Triple oxygen isotope systematics as a tracer of fluids in the crust: A study from position effects in a variety of crustal magma types. Earth Planet. Sci. Lett. 64, modern geothermal systems of Iceland. Chem. Geol. 530, 119312. 295–304. Zhang, R.Y., Hirajima, T., Banno, S., Cong, B., Liou, J., 1995. Petrology of ultrahigh- Watts, K.E., Bindeman, I.N., Schmitt, A.K., 2011. Large-volume rhyolite genesis in caldera pressure rocks from the southern Su-Lu region, eastern China. J. Metamorph. Geol. complexes of the Snake River Plain: insights from the Kilgore Tuff of the Heise 13, 659–675. Volcanic Field, Idaho, with comparison to Yellowstone and Bruneau–Jarbidge Zhang, Q., Chu, X., Feng, L., 2009. Discussion on the Neoproterozoic glaciations in the rhyolites. J. Petrol. 52, 857–890. South China Block and their related paleolatitudes. Chin. Sci. Bull. 54, 1797. Watts, K.E., Bindeman, I.N., Schmitt, A.K., 2012. Crystal scale anatomy of a dying su- Zhao, Z.-F., Zheng, Y.-F., 2003. Calculation of oxygen isotope fractionation in magmatic pervolcano: an isotope and geochronology study of individual phenocrysts from vo- rocks. Chem. Geol. 193, 59–80. luminous rhyolites of the Yellowstone caldera. Contrib. Mineral. Petrol. 164, 45–67. Zheng, Y.-F., Fu, B., Gong, B., Li, S., 1996. Extreme 18O depletion in eclogite from the Su- Watts, K.E., John, D.A., Colgan, J.P., Henry, C.D., Bindeman, I.N., Schmitt, A.K., 2016. Lu terrane in East China. Eur. J. Mineral. 317–324. Probing the volcanic–plutonic connection and the genesis of crystal-rich rhyolite in a Zheng, Y.-F., Fu, B., Xiao, Y., Li, Y., Gong, B., 1999. Hydrogen and oxygen isotope evi- deeply dissected supervolcano in the Nevada Great Basin: source of the Late Eocene dence for fluid–rock interactions in the stages of pre-and post-UHP metamorphism in Caetano Tuff. J. Petrol. 57, 1599–1644. the Dabie Mountains. Lithos 46, 677–693. Watts, K.E., John, D.A., Colgan, J.P., Henry, C.D., Bindeman, I.N., Valley, J.W., 2019. Zheng, Y.-F., Wu, Y.-B., Chen, F.-K., Gong, B., Li, L., Zhao, Z.-F., 2004. Zircon U-Pb and Oxygen isotopic investigation of silicic magmatism in the Stillwater caldera complex, oxygen isotope evidence for a large-scale 18O depletion event in igneous rocks during Nevada: Generation of large-volume, low-δ18O rhyolitic tuffs and assessment of their the Neoproterozoic. Geochim. Cosmochim. Acta 68, 4145–4165. regional context in the Great Basin of the western United States. Bulletin 131, Zheng, Y.-F., Zhang, S.-B., Zhao, Z.-F., Wu, Y.-B., Li, X., Li, Z., Wu, F.-Y., 2007. 1133–1156. Contrasting zircon Hf and O isotopes in the two episodes of Neoproterozoic granitoids Wei, C.-S., Zheng, Y.-F., Zhao, Z.-F., Valley, J.W., 2002. Oxygen and neodymium isotope in South China: implications for growth and reworking of continental crust. Lithos evidence for recycling of juvenile crust in northeast China. Geology 30, 375–378. 96, 127–150. Wei, C.-S., Zhao, Z.-F., Spicuzza, M.J., 2008. Zircon oxygen isotopic constraint on the Zheng, Y., Gong, B., Zhao, Z., Wu, Y., Chen, F., 2008. Zircon U-Pb age and O isotope sources of late Mesozoic A-type granites in eastern China. Chem. Geol. 250, 1–15. evidence for Neoproterozoic low-18O magmatism during supercontinental rifting in Weis, D., Deutsch, S., 1984. Nd and Pb isotope evidence from the Seychelles granites and South China: Implications for the snowball earth event. Am. J. Sci. 308, 484–516. their xenoliths: mantle origin with slight upper-crust interaction for alkaline anoro- Zhou, J.-L., Shao, S., Luo, Z.-H., Shao, J.-B., Wu, D.-T., Rasoamalala, V., 2015. genic complexes. Chem. Geol. 46, 13–35. Geochronology and geochemistry of Cryogenian gabbros from the Ambatondrazaka Whitaker, M.L., Nekvasil, H., Lindsley, D.H., McCurry, M., 2008. Can crystallization of area, east-central Madagascar: Implications for Madagascar-India correlation and olivine tholeiite give rise to potassic rhyolites?—an experimental investigation. Bull. Rodinia paleogeography. Precambrian Res. 256, 256–270. Volcanol. 70, 417–434. Zierenberg, R., Schiffman, P., Barfod, G., Lesher, C., Marks, N., Lowenstern, J., Wickham, S.M., Litvinovsky, B.A., Zanvilevich, A.N., Bindeman, I.N., 1995. Geochemical Mortensen, A., Pope, E.C., Bird, D., Reed, M., 2013. Composition and origin of evolution of Phanerozoic magmatism in Transbaikalia, East Asia: A key constraint on rhyolite melt intersected by drilling in the Krafla geothermal field, Iceland. Contrib. the origin of K-rich silicic magmas and the process of cratonization. J. Geophys. Res. Mineral. Petrol. 165, 327–347. Solid Earth 100, 15641–15654.

22