TY - JOUR
AU - Gamble, John, A
AB - Abstract The 2·08 Ma, ∼2500 km3 Huckleberry Ridge Tuff (HRT) eruption, Yellowstone, generated two fall deposits and three ignimbrite members (A, B, C), accompanying a ∼95 x 65 km caldera collapse. Field data imply that the pre-A fall deposits took weeks to be erupted, then breaks of weeks to months occurred between members A and B, and years to decades between B and C. We present compositional and isotopic data from single silicic clasts (pumice or fiamme) in the three ignimbrite members, plus new data from co-eruptive mafic components to reconstruct the nature and evacuation history of the HRT crustal magmatic complex. Geochemical data, building on field characteristics, are used to group nine silicic clast types into seven compositional suites (A1-A3; B1; C1-C3) within their respective members A, B and C. Isotopic data are then added to define four magmatic systems that were tapped simultaneously and/or sequentially during the eruption. Systems 1 and 2 fed the initial fall deposits and then vented throughout member A, accompanied by trace amounts of mafic magma. In member A, volumetrically dominant system 1 is represented by a rhyolite suite (A1: 73·0–77·7 wt % SiO2, 450–1680 ppm Ba) plus a distinct low-silica rhyolite suite (A2: 69·2–71·6 wt % SiO2, >2500 ppm Ba). System 2 yielded only a low-Ba, high-silica rhyolite suite (A3: 76·7–77·4 wt % SiO2, ≤250 ppm Ba). Glass compositions in pumices from systems 1 and 2 show clustering, indicative of the same multiple melt-dominant bodies identified in the initial fall deposits and earliest ignimbrite. Member B samples define suite B1 (70·7–77·4 wt % SiO2, 540–3040 ppm Ba) derived from magmatic system 1 (but not 2) that had undergone mixing and reorganisation during the A: B time break, accompanying mafic magma inputs. Mafic scoriae erupted in upper member B cover similar compositions to the member A clasts, but extend over a much broader compositional range. Member C clast compositions reflect major changes during the B: C time break, including rejuvenation of magmatic system 2 (last seen in member A) as suite C3 (75·3–77·2 wt % SiO2, 100–410 ppm Ba), plus the appearance of two new suites with strong crustal signatures. Suite C2 is another rhyolite (74·7–77·6 wt % SiO2, with Ba decreasing with silica from 2840 to 470 ppm) that defines magmatic system 3. Suite C2 also shows clustered glass compositions, suggesting that multiple melt-dominant bodies were a repetitive feature of the HRT magmatic complex. Suite C1, in contrast, is dacite to rhyolite (65·6–75·0 wt % SiO2, with Ba increasing with silica from 750 to 1710 ppm) that defines magmatic system 4. Compositions from magmatic systems 1 and 2 dominantly reflect fractional crystallization, but include partial melting of cumulates related to earlier intrusions of the same mafic magmas as those syn-eruptively vented. Country rock assimilation was limited to minor amounts of a more radiogenic (with respect to Sr) evolved contaminant. In contrast, systems 3 and 4 show similar strongly crustal isotopic compositions (despite their differences in elemental composition) consistent with assimilation of Archean rocks via partial melts derived from cumulates associated with contrasting mafic lineages. System 3 links to the same HRT mafic compositions co-erupted in members A and B. In contrast, system 4 links to olivine tholeiite compositions erupted in the Yellowstone area before, sparsely during, and following the HRT itself. All four magmatic systems were housed beneath the HRT caldera area. Systems 1 and 2 were hosted in Archean crust that had been modified by Cretaceous/Eocene magmatism, whereas systems 3 and 4 were hosted within crust that retained Archean isotopic characteristics. The extreme compositional diversity in the HRT highlights the spatial and temporal complexities that can be associated with large-volume silicic magmatism. INTRODUCTION Quaternary silicic (dacite to rhyolite in this context) magmatic systems have generated many caldera-forming explosive eruptions that would lead to global impacts and consequences if repeated today (Mason et al., 2004; Self, 2006). The largest of these (supereruptions) involve the eruption of >1015 kg (∼450 km3) of magma (Sparks et al., 2005; Wark & Miller, 2008). Of necessity, understanding their crustal magmatic systems has been incrementally developed, through studies of eruption deposits to reconstruct their geochemical nature and infer the characteristics of the parental magmatic system, and through numerical modelling of the generation, accumulation and eruptive triggering of these vast volumes of magma. Studies of past eruptions have included the analysis of a broad spectrum of erupted materials, including one or more of whole-rock samples (i.e. bulk-welded ignimbrite: e.g. Hildreth et al., 1984, 1991), single clasts (i.e. individual pumice/fiamme: e.g. Bacon & Druitt, 1988; Streck & Grunder, 1997; Hildreth & Wilson, 2007; Folkes et al., 2011), crystal cargoes (e.g. Bindeman et al., 2008; Ellis & Wolff, 2012; Rivera et al., 2014; Wolff & Ramos, 2014) and glass compositions (e.g. Nash et al., 2006; Roberge et al., 2013; Westgate et al., 2013). These works and others have highlighted significant diversity in silicic volcanism. A majority of silicic eruption deposits fall into two categories. First there are crystal-rich dacites, which are relatively homogeneous on the bulk-rock scale (e.g. Fish Canyon Tuff, Colorado; Bachmann et al., 2002) and have been labelled ‘monotonous intermediates’ (Hildreth, 1981; Christiansen, 2005; Bachmann & Bergantz, 2008; Bachmann & Huber, 2016). These are thought to represent rejuvenation/mobilisation of a crystal-rich magmatic mush body (e.g. Huber et al., 2011, 2012; Parmigiani et al., 2014). Second, there are crystal-poor rhyolites that are widely, although not exclusively (e.g. Streck, 2014), thought to represent rhyolitic melts extracted from a crystal-rich mush zone (Brophy, 1991; Bachmann & Bergantz, 2004; Hildreth, 2004) with subsequent minor crystallization. We focus on the latter here. Crystal-poor rhyolites generally show some intra-eruptive compositional variations (Hildreth, 1981; Christiansen, 2005; Bachmann & Bergantz, 2008; Bachmann & Huber, 2016). Works reviewed in Hildreth (1981) related whole-rock, single clast, mineral and/or glass compositional heterogeneities and zonations in eruption deposits to the inverted vertical zonation of the magmatic system, and overwhelmingly invoked a single magma body as the source. Subsequent work has used this framework to explain compositional variations in a similar way, whether those variations are gradational (e.g. Bishop Tuff, California: Hildreth & Wilson, 2007, Chamberlain et al., 2015; Carpenter Ridge Tuff, Colorado: Bachmann et al., 2014) or stepwise (e.g. Crater Lake, Oregon: Bacon & Druitt, 1988; Rattlesnake Tuff, Oregon: Streck & Grunder, 1997; Grizzly Peak Tuff, Colorado: Fridrich & Mahood, 1987). This compositional variability has, in turn, been related to variations in T, fO2 and H2O in the magmatic system that are manifested in one or more of: fractional crystallization-controlled evolution of the magma (e.g. Streck & Grunder, 1997; Hildreth & Wilson, 2007; Chamberlain et al., 2015), variations in crystal contents (e.g. Bacon & Druitt, 1988), or partial melting of a crystalline residuum (Deering et al., 2011; Wolff et al., 2015; Wolff, 2017). In other systems, however, unzoned crystal-poor deposits have also been recorded (e.g. Oruanui, Taupo Volcanic Zone: Wilson et al., 2006; Allan et al., 2017). Whether chemically variable or not, eruption deposits are typically inferred to represent the evacuation of a single melt-dominant body. However, some examples show compositional diversity inferred to represent the simultaneous or sequential evacuation of multiple melt-dominant bodies (e.g. Oruanui, Kidnappers/Rocky Hill, and Mamaku-Ohakuri eruptions, Taupo Volcanic Zone: Allan et al., 2012; Cooper et al., 2012, 2016; Bégué et al., 2014; and the Snake River Plain: Ellis & Wolff, 2012; Ellis et al., 2014). These bodies, in turn, may be interpreted to represent melt extractions from a unitary, heterogenous mush zone (Bégué et al., 2014; Ellis et al., 2014) or from multiple magmatic systems (Allan et al., 2012; Cooper et al., 2012, 2016). In contrast to these complexities, numerical models for such volcanism focus on the formation and evacuation of a single magma body (e.g. Jellinek & DePaolo, 2003; Gregg et al., 2012, 2015; Caricchi et al., 2014; de Silva & Gregg, 2014) and do not, therefore, take into account effects of the associated stress field (and hence eruption behaviour) of multiple melt-dominant bodies (Gudmundsson, 2012; Cashman & Giordano, 2014). The formation of parental magma bodies to super-eruptions is typically inferred to take of the order of 105–106 years (e.g. Jellinek & DePaolo, 2003). This arises from a presumption of a relationship between eruption size and the pre-eruptive repose period (Smith, 1979; Reid, 2008) although other studies demonstrate no such relationship exists (e.g. Wilson, 1993; Wilson et al., 2009). Although some examples show evidence for prolonged accumulation of the melt-dominant body (e.g. Crater Lake: Bacon, 1983; Bishop Tuff: Chamberlain et al., 2014a), others have demonstrated very rapid accumulation (e.g. Oruanui and Taupo: Wilson & Charlier, 2009; Allan et al., 2013, 2017; Pamukcu et al., 2015; Barker et al., 2016). In addition, there is also evidence for the rapid (sometimes decadal) recovery and regeneration of magmatic systems following large caldera-forming eruptions (e.g. Barker et al., 2014, 2016; Cooper et al., 2016, 2017; Hildreth et al., 2017). These studies highlight a dichotomy of timescales associated with silicic systems. Although melt-dominant bodies can accumulate rapidly, the generation of their chemical signatures, through some combination of fractional crystallization, assimilation, or magma mixing (Christiansen & McCurry, 2008; Bindeman & Simakin, 2014), is a longer term process, as indicated by the long-term broad chemical similarities of erupted materials (e.g. Mazama: Bacon, 1983; Taupo: Barker et al., 2015; Bruneau-Jarbidge: Cathey & Nash, 2004; Yellowstone: Girard & Stix, 2010, Loewen & Bindeman, 2015; Stelten et al., 2015; Troch et al., 2017). Broad chemical uniformity may be observed over long periods at such systems that masks the diversities seen on smaller scales in isotopic and/or crystal compositions. Such diversities may indicate that the melt-dominant bodies result from amalgamation of melt and/or crystals from a wide range of sources, including silicic fractionates from mafic precursors, partial melts of solidified intrusions and enriched partial melts from silicic country rocks (e.g. Bindeman et al., 2008; Charlier et al., 2008; Christiansen & McCurry, 2008; Wolff & Ramos, 2014; Stelten et al., 2017). A constant factor amongst these diversities, however, is the role of mafic magma, which provides thermal and chemical inputs into silicic systems, modulating their behaviour and potentially triggering eruption (e.g. Sparks et al., 1977; Hildreth, 1981; Loewen & Bindeman, 2015; Barker et al., 2015, 2016; Till et al., 2015; Stelten et al., 2017). To provide a case-study of the wide range of possible processes and products in large-scale silicic volcanism and associated magmatism we present a geochemical investigation of the Huckleberry Ridge Tuff (HRT) at Yellowstone. In particular, we document the major element, trace element and isotopic compositions of single pumices, fiamme and their associated groundmass glasses and crystal cargoes, from the three ignimbrite members of the tuff. These results are combined with limited new data from co-erupted mafic materials, plus published information (Myers et al., 2016; Swallow et al., 2018a, 2018b), to provide insights into the variability of the HRT erupted products, the nature of the multiple magmatic and crustal sources, and the way in which such huge systems are generated and evacuated during a supereruption. GEOLOGICAL BACKGROUND Yellowstone Plateau volcanic field Quaternary volcanic products in the Yellowstone area, including the HRT, are generally grouped together as the Yellowstone Plateau volcanic field (YPVF). The YPVF lies at the eastern end of the Snake River Plain, a 700 km long, ENE-trending volcano–tectonic depression comprising a series of eastward-younging silicic caldera centres and intercalated basaltic lava flows (Armstrong et al., 1975; Pierce & Morgan, 1992, 2009). The ∼16 Ma to present, time-progressive volcanism, elevated heat flow, fluid geochemistry and tectonic setting are consistent with movement of the North American plate over a stationary mantle plume, with complexities arising from asthenospheric flow around the foundered Farallon slab (Pierce & Morgan, 1992, 2009; Lowenstern & Hurwitz, 2008; James et al., 2011; Lowenstern et al., 2014; Zhou et al., 2018). Volcanism along the Snake River Plain is bimodal, with basaltic lavas (dominantly olivine tholeiites) and high-silica rhyolitic lavas and tuffs forming nearly all the eruption products (Christiansen, 2001; Christiansen & McCurry, 2008). The ∼2·1 Myr history of the YPVF is grouped into three volcanic cycles, with dominantly effusive, basaltic and silicic activity bracketing three climactic, caldera-forming silicic eruptions, of which the HRT was the first (Christiansen, 2001). Geochemical investigations of YPVF silicic eruptives have primarily focused on (a) whole-rock samples (i.e. whole pieces of welded ignimbrite), either for bulk chemical characteristics (Doe et al., 1982; Hildreth et al., 1984, 1991; Christiansen, 2001) or focused single-crystal studies (Bindeman & Valley, 2001; Bindeman et al., 2008; Ellis et al., 2012; Rivera et al., 2014, 2016; Singer et al., 2014; Wotzlaw et al., 2015); or (b) samples of lavas, principally from younger effusive episodes (Girard & Stix, 2009, 2010; Watts et al., 2012; Loewen & Bindeman, 2015; Stelten et al., 2015, 2017, 2018; Till et al., 2015; Befus & Gardner, 2016; Troch et al., 2017). Broad-scale investigations have indicated that the high-silica rhyolites of the YPVF are not simply highly fractionated melts from a mafic precursor, due to the absence of intermediate compositions and the offset between the Sr and Pb isotopic fields of YPVF rhyolites and tholeiitic basalts (Doe et al., 1982; Hildreth et al., 1991; Christiansen & McCurry, 2008). It was therefore inferred that the rhyolites were derived from partial melting of middle crust comprising Archaean basement and Cenozoic tholeiitic basalt intrusions (Doe et al., 1982; Hildreth et al., 1991), followed by extensive fractionation of feldspars and ferromagnesian minerals (Christiansen & McCurry, 2008). Heterogeneity observed in the crystal cargoes from single eruptive units attest to a diverse range of initial melt compositions that accumulate and hybridise to smooth out isotopic variations (Bindeman et al., 2008; Wotzlaw et al., 2015; Stelten et al., 2017). Larger-scale variations in isotopic systematics were attributed to shallow (upper crustal) processes, including the partial melting of the foundering magma chamber roof or assimilation of hydrothermally altered country rock (Doe et al., 1982; Hildreth et al., 1984, 1991; Bindeman & Valley, 2001; Bindeman et al., 2008; Troch et al., 2018). What is universally accepted, however, is the prominent role mafic magmas play in Yellowstone rhyolitic volcanism (e.g. Hildreth et al., 1991), whether through providing chemical and thermal inputs into a long-lived crystal mush zone (Stelten et al., 2015), generating melt-dominant bodies through partial melting of upper crustal rocks (Loewen & Bindeman, 2015; Troch et al., 2018), providing fractionated melts to the rhyolitic system (Stelten et al., 2017), or causing eruptive triggering (Till et al., 2015). Huckleberry Ridge Tuff The 2·08 Ma, ∼2, 500 km3 HRT (Fig. 1) represents the first and largest of the three caldera-forming eruptions in the YPVF (Christiansen, 2001). It comprises three ignimbrite members (A, B and C) and two fall deposits: initial pre-A fall deposits and others between members B and C. The only significant difference from the presentation by Christiansen (2001) is that most of the rheomorphic ignimbrite exposed to the southwest of the caldera is recognised by us as being member A, not member B. The overall published consensus is that the HRT was erupted geologically rapidly from discrete parts of a single, zoned magma body and was emplaced in thermal terms as a composite sheet (Reynolds, 1977; Christiansen, 1979, 2001; Hildreth et al., 1984, 1991). Fig. 1. Open in new tabDownload slide Simplified geological map of the Yellowstone Plateau volcanic field, adapted from Christiansen (2001) and Christiansen et al. (2007). Outcrop areas of the Huckleberry Ridge Tuff (HRT) are shown in red and key sampling localities for this study are numbered. Fig. 1. Open in new tabDownload slide Simplified geological map of the Yellowstone Plateau volcanic field, adapted from Christiansen (2001) and Christiansen et al. (2007). Outcrop areas of the Huckleberry Ridge Tuff (HRT) are shown in red and key sampling localities for this study are numbered. However, fieldwork by C. J. N. Wilson, to be presented elsewhere, has indicated that the eruption was episodic, with evidence for time breaks observed within and between eruptive deposits (Fig. 2; Supplementary DataFig. S1; supplementary data are available for downloading at http://www.petrology.oxfordjournals.org). Several lines of evidence point to this conclusion. First, the initial fall deposits display reworking structures and fine ash tops to individual horizons, indicative of episodic activity and hiatuses in the eruption. Coupled with information from volatile concentrations in quartz-hosted melt inclusions and re-entrants that suggest slow magma rise rates, it is inferred that fall deposition occurred over a period of some weeks (Myers et al., 2016). Second, there are welding contrasts between members A and B (particularly where deposits are thinnest), and the rheomorphic folding of member A (as recognised by us) is in contrast to a planar member B in exposures along the Teton River valley (Supplementary DataFig. S1). Estimates for the timing of rheomorphism are problematic, but field observations and palaeomagnetic data imply that the folding occurred before any devitrification of member A, and was almost entirely finished by the time member B was emplaced (Embree & Hoggan, 1999; Geissman et al., 2010). These constraints suggest a time gap of weeks to months occurred between the emplacement of members A and B. Third, widespread breaks in welding intensity occur across the contact between members B and C, but these are balanced against the fumarolic alteration of pre-C fall deposits (Supplementary DataFig. S1) and/or member C ignimbrite by hot gases from the underlying A + B packages. Using comparisons to the Valley of Ten Thousand Smokes ignimbrite from the 1912 eruption of Katmai, where fumaroles emanating from an ignimbrite of roughly comparable thickness and temperature diminished within years and were largely extinct within 20 years (see summary in Hildreth & Fierstein, 2012), we infer there to be time gap between B and C of years to decades in length. This revised chronostratigraphic framework provides context to the geochemical overview presented here. Fig. 2. Open in new tabDownload slide Schematic diagram of the stratigraphy of the HRT and estimates for intra-eruption timescales based on fieldwork by C. J. N. Wilson (Wilson, 2009 and Wilson et al., in prep.). Initial fall deposits were episodic, with short time breaks indicated by reworking and fine ash tops to individual horizons (Myers et al., 2016; Swallow et al., 2018a). A cessation in activity of weeks to months between members B and C is inferred from welding breaks, and rheomorphic deformation of member A but not B (see Supplementary DataFig. S1). A years to decades time break between members B and C is estimated from cooling breaks indicated by welding intensities, but also the presence of vapour-phase alteration of pre-C fall deposits or the base of member C ignimbrite where it rests directly on the top of ignimbrite member B (Supplementary DataFig. S1). Fig. 2. Open in new tabDownload slide Schematic diagram of the stratigraphy of the HRT and estimates for intra-eruption timescales based on fieldwork by C. J. N. Wilson (Wilson, 2009 and Wilson et al., in prep.). Initial fall deposits were episodic, with short time breaks indicated by reworking and fine ash tops to individual horizons (Myers et al., 2016; Swallow et al., 2018a). A cessation in activity of weeks to months between members B and C is inferred from welding breaks, and rheomorphic deformation of member A but not B (see Supplementary DataFig. S1). A years to decades time break between members B and C is estimated from cooling breaks indicated by welding intensities, but also the presence of vapour-phase alteration of pre-C fall deposits or the base of member C ignimbrite where it rests directly on the top of ignimbrite member B (Supplementary DataFig. S1). Although reported whole-rock major and trace elemental compositional variations are minor, significant isotopic heterogeneity is present (Doe et al., 1982; Hildreth et al., 1984, 1991; Christiansen, 2001). Whole-rock Sr, Pb and Nd isotopic data show that members A and B are similar, whereas member C is distinctly more crustally influenced, interpreted to reflect the rapid assimilation of a foundering upper crustal chamber roof following eruption of members A and B (Doe et al., 1982; Hildreth et al., 1984, 1991). However, any crustal contaminant is precluded from being hydrothermally altered, based on the uniformity of whole-rock O isotopic values in the HRT, in sharp contrast to subsequent variations (Hildreth et al., 1984, 1991). Isotopic heterogeneity is also observed within crystal populations from single HRT samples, with zircons showing clustering in Hf, but smaller variations in O isotopes, interpreted as reflecting a heterogeneous magmatic system comprising distinct magma reservoirs with separately dominant crustal and mantle signatures (Bindeman & Valley, 2001; Bindeman et al., 2008; Wotzlaw et al., 2015). However, these heterogeneities are difficult to relate to specific magmatic domains due to the lack of sample context, arising from the use of whole-rock samples (as distinct from the individual juvenile clasts used in this study). Diversities in crystal characteristics can thus not be linked to specific magma compositions, only to the broad stratigraphic units. Consequently, current inferences about the architecture of the HRT magmatic system, which is proposed by the above authors to have been heterogeneous and to have undergone rapid formation and evolution, are limited. Studies on the pre-A fall deposits and basal member A ignimbrite indicated that the upper levels of the HRT silicic magmatic system comprised multiple, laterally-adjacent, discrete, melt-dominant bodies that represented two distinct host magmatic lineages (Myers et al., 2016; Swallow et al., 2018a). The initial fall stages of the eruption saw the simultaneous and sequential tapping of three melt-dominant bodies before escalating into the evacuation of nine bodies (including the earlier three) at the onset of ignimbrite member A deposition and associated caldera collapse after the eruption of ∼50 km3 of magma (Swallow et al., 2018a). However, the nature of these bodies and their two host magmatic lineages was not clear. They could represent: complexity only within the upper levels of a magmatic system that merged downwards into homogeneity at deeper levels tapped during later stages of the eruption, two magmatic lineages that remained discrete throughout eruption (e.g. Kidnappers: Cooper et al., 2012, 2016), or the first stages of an increasing magmatic complexity from the fall deposits to later in the eruption (e.g. Mamaku/Ohakuri: Bégué et al., 2014). Here we present data from single pyroclasts from members A, B and C in order to test these three possibilities and provide an overall petrogenetic story for the HRT. For the most part, however, the HRT is densely welded, devitrified and vapour-phase altered. Juvenile materials, that were inferred to have been molten at the time of eruption, were preferentially collected from the limited number of sites where macroscopic glassy clasts were present (Fig. 1; Supplementary Data Electronic Appendix 1), but otherwise fiamme were collected from devitrified but otherwise unaltered tuff. Our data thus delimit the full compositional range and degree of compositional clustering in the silicic juvenile materials, but (unlike in the fall deposits: Myers et al., 2016; Swallow et al., 2018a) we are not able to interrogate any systematic compositional zonation that might have been present within the individual ignimbrite members. In the light of the complexities encountered in the field and subsequently revealed in the analytical data for the silicic materials we present a consistent descriptive framework (Fig. 3). We categorise the sampled juvenile clasts, using field-based observations of clast characteristics (e.g. colour, crystal contents and sizes, pumice textures) and eruptive unit, into nine types (Table 1). We then use our geochemical data to group these juvenile materials into seven compositional suites that share common features or distinctive trends and are distinct to specific ignimbrite members. The suites in turn are attributed, with the addition of isotopic data, to four distinctive magma systems which, together with the mafic components (Swallow et al., 2018b and this study) collectively make up the magmatic complex that yielded the total juvenile products of the overall HRT eruption. Fig. 3. Open in new tabDownload slide Schematic flow chart to show the logic structure underlying the data presentation in this paper. The sampled felsic materials are labelled on field criteria into the nine clast types (see Table 1 for descriptions), then grouped on geochemical and mineralogical grounds into the seven compositional suites and then, when coupled with isotopic data, allocated to four magmatic systems. The felsic and mafic components combined contribute to the overall magmatic complex that fed the HRT eruption. Fig. 3. Open in new tabDownload slide Schematic flow chart to show the logic structure underlying the data presentation in this paper. The sampled felsic materials are labelled on field criteria into the nine clast types (see Table 1 for descriptions), then grouped on geochemical and mineralogical grounds into the seven compositional suites and then, when coupled with isotopic data, allocated to four magmatic systems. The felsic and mafic components combined contribute to the overall magmatic complex that fed the HRT eruption. Table 1: Descriptions of silicic clast types in the HRT and their characteristic petrographic features Clast type Members where clasts sampled Physical characteristics Geochemical suite(s) J1 A and B (C as surficial pick-ups) Pumices: white, brown where oxidized (Fig. 4a, b) Moderately to highly vesicular (rounded to elongate vesicles) Fibrous, common schlieren and ‘adobe-type’ textures Coarse crystals (up to 8 mm), crystal-rich (11–20 wt %), common glomerocrysts Flattened, devitrified fiamme equivalents also sampled A1, A3, B1 J2 A and B Dense, poorly vesicular Dark grey–black through microlite growth (Fig. 4c) Coarse crystals (up to 8 mm), crystal rich (27–35 wt %) A1, B1 J3 A and B Black to dark brown (Fig. 4d) Moderately to poorly vesicular 10–20% euhedral feldspars up to 10 mm A2, B1 J4 B (upper) and C (pickups) Mingled, scoria-bearing rhyolitic clasts (Fig. 4e) Cm-scale lenses of scoria within pumices similar to J1-type Sparse incorporation of J3-type material B1 J5 C Black, devitrified material (Fig. 4f) Non-vesicular Angular rigid to flattened, fluidal fiamme or rounded clasts Equant (welded) to elongate (rheomorphic ignimbrite) Small crystals (<3 mm) and crystal poor (<15%); feldspar and rare quartz C1, C2 J6 C (Red Mountains) Visually identical to J5-type clasts Minor intrusions into hot, still-plastic member C ignimbrite C1 J7 C (Teton River: loc. 332, Fig. 1) Rounded pumices with equant vesicles (Fig. 4g) Small crystals (<3 mm), crystal poor-moderate (11–15 wt %) C3 J8 C (Bitch Creek: loc. 057, Fig. 1) Dark brown–grey, mid-grey or pale buff pumices (Fig. 4h), sometimes colour-mingled Elongate to platey in shape Moderately vesicular with sub-100 μm bubbles to extremely vesicular, coarsely cellular to fibrous with 1–10 mm bubbles Small crystals (<3 mm) and crystal poor-moderate (7–11 wt %) C2 J9 C (Red Mountains and Huckleberry Mountain) Pale-coloured yellow to grey devitrified fiamme Sometimes vuggy appearance Occasionally micro- to macro-spherulitic Similar crystal sizes and abundance to J8-type pumices C2 Clast type Members where clasts sampled Physical characteristics Geochemical suite(s) J1 A and B (C as surficial pick-ups) Pumices: white, brown where oxidized (Fig. 4a, b) Moderately to highly vesicular (rounded to elongate vesicles) Fibrous, common schlieren and ‘adobe-type’ textures Coarse crystals (up to 8 mm), crystal-rich (11–20 wt %), common glomerocrysts Flattened, devitrified fiamme equivalents also sampled A1, A3, B1 J2 A and B Dense, poorly vesicular Dark grey–black through microlite growth (Fig. 4c) Coarse crystals (up to 8 mm), crystal rich (27–35 wt %) A1, B1 J3 A and B Black to dark brown (Fig. 4d) Moderately to poorly vesicular 10–20% euhedral feldspars up to 10 mm A2, B1 J4 B (upper) and C (pickups) Mingled, scoria-bearing rhyolitic clasts (Fig. 4e) Cm-scale lenses of scoria within pumices similar to J1-type Sparse incorporation of J3-type material B1 J5 C Black, devitrified material (Fig. 4f) Non-vesicular Angular rigid to flattened, fluidal fiamme or rounded clasts Equant (welded) to elongate (rheomorphic ignimbrite) Small crystals (<3 mm) and crystal poor (<15%); feldspar and rare quartz C1, C2 J6 C (Red Mountains) Visually identical to J5-type clasts Minor intrusions into hot, still-plastic member C ignimbrite C1 J7 C (Teton River: loc. 332, Fig. 1) Rounded pumices with equant vesicles (Fig. 4g) Small crystals (<3 mm), crystal poor-moderate (11–15 wt %) C3 J8 C (Bitch Creek: loc. 057, Fig. 1) Dark brown–grey, mid-grey or pale buff pumices (Fig. 4h), sometimes colour-mingled Elongate to platey in shape Moderately vesicular with sub-100 μm bubbles to extremely vesicular, coarsely cellular to fibrous with 1–10 mm bubbles Small crystals (<3 mm) and crystal poor-moderate (7–11 wt %) C2 J9 C (Red Mountains and Huckleberry Mountain) Pale-coloured yellow to grey devitrified fiamme Sometimes vuggy appearance Occasionally micro- to macro-spherulitic Similar crystal sizes and abundance to J8-type pumices C2 Open in new tab Table 1: Descriptions of silicic clast types in the HRT and their characteristic petrographic features Clast type Members where clasts sampled Physical characteristics Geochemical suite(s) J1 A and B (C as surficial pick-ups) Pumices: white, brown where oxidized (Fig. 4a, b) Moderately to highly vesicular (rounded to elongate vesicles) Fibrous, common schlieren and ‘adobe-type’ textures Coarse crystals (up to 8 mm), crystal-rich (11–20 wt %), common glomerocrysts Flattened, devitrified fiamme equivalents also sampled A1, A3, B1 J2 A and B Dense, poorly vesicular Dark grey–black through microlite growth (Fig. 4c) Coarse crystals (up to 8 mm), crystal rich (27–35 wt %) A1, B1 J3 A and B Black to dark brown (Fig. 4d) Moderately to poorly vesicular 10–20% euhedral feldspars up to 10 mm A2, B1 J4 B (upper) and C (pickups) Mingled, scoria-bearing rhyolitic clasts (Fig. 4e) Cm-scale lenses of scoria within pumices similar to J1-type Sparse incorporation of J3-type material B1 J5 C Black, devitrified material (Fig. 4f) Non-vesicular Angular rigid to flattened, fluidal fiamme or rounded clasts Equant (welded) to elongate (rheomorphic ignimbrite) Small crystals (<3 mm) and crystal poor (<15%); feldspar and rare quartz C1, C2 J6 C (Red Mountains) Visually identical to J5-type clasts Minor intrusions into hot, still-plastic member C ignimbrite C1 J7 C (Teton River: loc. 332, Fig. 1) Rounded pumices with equant vesicles (Fig. 4g) Small crystals (<3 mm), crystal poor-moderate (11–15 wt %) C3 J8 C (Bitch Creek: loc. 057, Fig. 1) Dark brown–grey, mid-grey or pale buff pumices (Fig. 4h), sometimes colour-mingled Elongate to platey in shape Moderately vesicular with sub-100 μm bubbles to extremely vesicular, coarsely cellular to fibrous with 1–10 mm bubbles Small crystals (<3 mm) and crystal poor-moderate (7–11 wt %) C2 J9 C (Red Mountains and Huckleberry Mountain) Pale-coloured yellow to grey devitrified fiamme Sometimes vuggy appearance Occasionally micro- to macro-spherulitic Similar crystal sizes and abundance to J8-type pumices C2 Clast type Members where clasts sampled Physical characteristics Geochemical suite(s) J1 A and B (C as surficial pick-ups) Pumices: white, brown where oxidized (Fig. 4a, b) Moderately to highly vesicular (rounded to elongate vesicles) Fibrous, common schlieren and ‘adobe-type’ textures Coarse crystals (up to 8 mm), crystal-rich (11–20 wt %), common glomerocrysts Flattened, devitrified fiamme equivalents also sampled A1, A3, B1 J2 A and B Dense, poorly vesicular Dark grey–black through microlite growth (Fig. 4c) Coarse crystals (up to 8 mm), crystal rich (27–35 wt %) A1, B1 J3 A and B Black to dark brown (Fig. 4d) Moderately to poorly vesicular 10–20% euhedral feldspars up to 10 mm A2, B1 J4 B (upper) and C (pickups) Mingled, scoria-bearing rhyolitic clasts (Fig. 4e) Cm-scale lenses of scoria within pumices similar to J1-type Sparse incorporation of J3-type material B1 J5 C Black, devitrified material (Fig. 4f) Non-vesicular Angular rigid to flattened, fluidal fiamme or rounded clasts Equant (welded) to elongate (rheomorphic ignimbrite) Small crystals (<3 mm) and crystal poor (<15%); feldspar and rare quartz C1, C2 J6 C (Red Mountains) Visually identical to J5-type clasts Minor intrusions into hot, still-plastic member C ignimbrite C1 J7 C (Teton River: loc. 332, Fig. 1) Rounded pumices with equant vesicles (Fig. 4g) Small crystals (<3 mm), crystal poor-moderate (11–15 wt %) C3 J8 C (Bitch Creek: loc. 057, Fig. 1) Dark brown–grey, mid-grey or pale buff pumices (Fig. 4h), sometimes colour-mingled Elongate to platey in shape Moderately vesicular with sub-100 μm bubbles to extremely vesicular, coarsely cellular to fibrous with 1–10 mm bubbles Small crystals (<3 mm) and crystal poor-moderate (7–11 wt %) C2 J9 C (Red Mountains and Huckleberry Mountain) Pale-coloured yellow to grey devitrified fiamme Sometimes vuggy appearance Occasionally micro- to macro-spherulitic Similar crystal sizes and abundance to J8-type pumices C2 Open in new tab JUVENILE COMPONENTS OF THE HRT Members A and B Juvenile silicic components (abbreviated to J) in members A and B are clasts that range from poorly to extremely vesicular (but are mostly pumice in non-welded tuff). The most common pumice (type J1: Table 1; Figs 4a, b) is white where freshest, tan where thermally oxidized, and exhibits a continuous spectrum in vesicle textures from a rounded, cellular foam through to elongated vesicles yielding a fibrous texture, with common crystal schlieren and ‘adobe-type’ textures, the latter matching those in some Bishop Tuff pumices (Hildreth & Wilson, 2007). Variations from type J1 pumice (and their equivalent devitrified fiamme) include dense, poorly vesicular juvenile material that, where glassy, is dark grey to black through the exsolution of iron oxide microlites (type J2: Fig. 4c). A third type is represented by black to dark brown, poorly vesicular clasts that are visually distinctive through containing 10–20% euhedral feldspar crystals up to 8–10 mm in length (type J3: Fig. 4d). In upper member B (and sparsely in A), mingled clasts occur (type J4: Fig. 4e), with centimetre-scale lumps or lenses of scoria plus (occasionally) type J3 material, distributed within type J1 pumice. Fig. 4. Open in new tabDownload slide Photographs showing representative of clast types sampled for this study. (a), (b) two examples of typical crystal-rich type J1 clasts (see Table 1) from member B at locality 65 (Fig. 1: YP028: 74·36 wt % SiO2; YP085: 75·23 wt % SiO2: Supplementary Data Electronic Appendix 3). Note the large crystal sizes and presence of glomerocrysts. (c) Type J2 clast YP251 (75·01 wt % SiO2) from member A, exhibiting a characteristic crystal-rich, dense texture with darkened groundmass glass. (d) Type J3 clast (not sampled) from member A showing coarse, intact euhedral feldspars within dark groundmass glass. (e) Type J4 mingled pumice from upper member B at locality 65 (Fig. 1). The scoria component was analysed as YP083SCORIA (55·56 wt % SiO2: data presented in Swallow et al., 2018b, and in Supplementary Data Electronic Appendix 3). (f) Outcrop photograph of characteristically non-vesicular, crystal-poor and devitrified type J5 clasts within member C ignimbrite from the Red Mountains. (g) Crystal-poor type J7 pumice (YP134: 76·61 wt % SiO2) from locality 332 (Fig. 1) in member C. Note the characteristic small crystal sizes, and the pink-oxidized rinds. (h) Examples of type J8 pumices from locality 057 (Fig. 1). Although uniformly crystal-poor to -moderate, and highly to extremely vesicular, note the wide variations in vesicle texture reflected in the colour variations. Fig. 4. Open in new tabDownload slide Photographs showing representative of clast types sampled for this study. (a), (b) two examples of typical crystal-rich type J1 clasts (see Table 1) from member B at locality 65 (Fig. 1: YP028: 74·36 wt % SiO2; YP085: 75·23 wt % SiO2: Supplementary Data Electronic Appendix 3). Note the large crystal sizes and presence of glomerocrysts. (c) Type J2 clast YP251 (75·01 wt % SiO2) from member A, exhibiting a characteristic crystal-rich, dense texture with darkened groundmass glass. (d) Type J3 clast (not sampled) from member A showing coarse, intact euhedral feldspars within dark groundmass glass. (e) Type J4 mingled pumice from upper member B at locality 65 (Fig. 1). The scoria component was analysed as YP083SCORIA (55·56 wt % SiO2: data presented in Swallow et al., 2018b, and in Supplementary Data Electronic Appendix 3). (f) Outcrop photograph of characteristically non-vesicular, crystal-poor and devitrified type J5 clasts within member C ignimbrite from the Red Mountains. (g) Crystal-poor type J7 pumice (YP134: 76·61 wt % SiO2) from locality 332 (Fig. 1) in member C. Note the characteristic small crystal sizes, and the pink-oxidized rinds. (h) Examples of type J8 pumices from locality 057 (Fig. 1). Although uniformly crystal-poor to -moderate, and highly to extremely vesicular, note the wide variations in vesicle texture reflected in the colour variations. The silicic clasts are moderately porphyritic, with 11–22 wt % crystals, but extending to 35 wt % in some samples, as determined through crystal–glass separation using heavy liquids (Supplementary Data Electronic Appendix 2). The crystal phases are generally coarse grained, with single crystals up to 7–8 mm and crystal aggregates or clots occasionally up to 2 cm across. The crystal assemblage is dominated by quartz, sanidine and plagioclase, either as single crystals or crystal aggregates. Ferromagnesian minerals are a minor component, and include pyroxene, olivine and Fe–Ti oxides. Member C Juvenile silicic components in member C are visually distinctive, both with respect to each other and to the clast types in members A and B (Table 1; Fig. 4), and are described as five types. Non-vesicular black juvenile material occurs as angular to rounded clasts which range in shape from equant in normally welded tuff to highly elongate, stretched clasts in rheomorphic ignimbrite (type J5: Table 1; Fig. 4f). These clasts have not been found in non-welded or glassy material. They contain few crystals: sparse feldspar and only rare quartz. Material visually identical to type J5 clasts (and labelled here as type J6 material) is found as minor intrusions (up to tens of metres across) invading densely welded member C ignimbrite in exposures along the northern scarp of the Red Mountains. The intrusions are inferred to be closely related in time to the emplacement of member C as their contacts show crenulated and embayed contacts with the host ignimbrite and have only poorly developed chilled margins. The presence of these intrusions implies a local source beneath this area, and that the corresponding magma suite was able to reach shallow levels in the aftermath of the eruption in a small-scale episode of resurgence. Glassy pumices were collected from non-welded to sintered ignimbrite at sites along the Teton River (particularly locality 332: Fig. 1; Supplementary Data Electronic Appendix 1). These clasts are rounded, contain equant vesicles and are crystal-moderate (type J7: 11–15 wt % crystals: Fig. 4g). The finer-scale vesicular texture in these pumices and notably smaller crystal sizes preclude them from being simply pick-ups from earlier deposits. Glassy pumices of contrasting appearance to type J7 were collected from the non-welded base of member C at locality 057 (Fig. 1; Supplementary Data Electronic Appendix 1). The clasts are light brown–grey or pale creamy brown in colour, elongate to platey in shape and extremely vesicular, fibrous and crystal-poor to moderate (type J8: 7–11 wt % crystals: Fig. 4h). Devitrified fiamme (type J9) from densely welded, often rheomorphic, ignimbrite in the Red Mountains and on Huckleberry Mountain. The clasts are pale yellow- to mid-grey, often vuggy, are sometimes micro- to macro-spherulitic and contain similar crystal proportions to pumices found elsewhere in non-welded member C (type J8). They contrast with type J5 clasts in being more flattened, and paler in colour due to coarser devitrification micro-textures. All but one of the juvenile clast types in member C have finer crystal sizes (<3 mm) and crystal-poorer natures (<15 wt % crystals) than nearly all clasts in members A and B. Quartz, commonly bi-pyramidal, and feldspar (both plagioclase and sanidine) dominate, with trace amounts of pyroxene, olivine and Fe–Ti oxides. The exception is pumices that are physically identical to J1-type or to J4-type pumices found in member B and are also chemically similar to those member B pumices (see below for further discussion). These pumices in C have been found only in areas along the Teton River southwest of the caldera where a non-welded top to member B was exposed at the time of member C emplacement. For these reasons we infer that these variant C clasts (labelled ‘B-in-C’) represent syn-eruptive pick-ups from the non-welded top of member B and are considered in with member B in terms of their compositions. Juvenile mafic materials As well as our silicic juvenile clast data, we also present new data from the juvenile mafic materials in members A and B, augmenting the data in Swallow et al. (2018b). Juvenile mafic material has not been found in member C deposits. Where found as individual clasts, these are, in general, poorly flattened (even in dense-welded tuff) yet show textures (e.g. chilled margins) indicating that they were quenched against the host tuff. The samples include poorly-vesicular material from parts of member A south of the caldera (including the earliest-erupted material), additional scoria from mingled type J4 pumices from member B south of the caldera, and six scoria clasts from glassy to devitrified (but not vapour-phase altered) member B north of the caldera. METHODS Sample selection All samples analysed are listed in Supplementary Data Electronic Appendix 1. For the silicic materials, glassy pumices were sampled whenever possible, but individual devitrified fiamme were also sampled to compare with the glassy material. In a few cases (mostly the basal vitrophyre in member A), groundmass glass was sampled from single fiamme where the whole clast could not be extracted. Mafic materials reported in Swallow et al. (2018b) represent single clasts. These are devitrified where found in devitrified material of members A and B, but are glassy when contained within glassy pumices sampled in member B. In cases where mingled clasts occur in upper member B, the individual components, distinguished by colour, crystal content and vesicle texture, were broken out separately. The additional mafic samples reported here are from devitrified and glassy material both south and north of the caldera and from members A and B. We stress that this study only uses single clast data, and no whole-rock (i.e. bulk tuff) material was analysed. Single clast analysis Samples were washed, trimmed of any adhering matrix and dried before being crushed in a Rocklabs Boyd crusher to ∼2 mm diameter. This material was then split. At least 100 g of crushed material, providing a representative sample, was milled using an agate TEMA ring mill to the homogenous fine powder which was then used for major and trace element, and isotopic ratio determinations. Single clast major (and Ba) element abundances were determined by X-Ray Fluorescence (XRF) on fused discs at the Open University, U.K. (OU), and the universities of Auckland (AU) and Waikato (WU), New Zealand, all based on the methods of Ramsey et al. (1995). Individual 2 s.d. % relative precisions, determined through repeated analysis of multiple standards in each laboratory, are <3% for almost all major element oxides, with most <1%, and offsets from preferred or recommended values are generally <5% for all elements. Samples duplicated between laboratories show offsets of <5% for all major and minor element oxides >0·1 wt % in abundance with Ba offsets ≤5% (Supplementary Data Electronic Appendix 3). Trace element analysis of rock powders was conducted using inductively-coupled plasma mass-spectrometry (ICP-MS) at Victoria University of Wellington, New Zealand (VUW). Due to the zircon-bearing nature of the samples, powders were digested using ammonium fluoride (Hu et al., 2013). Solutions were analysed on a ThermoFisher Element2 sector-field ICP-MS at VUW. Concentrations were determined using a calibration curve generated from gravimetric daughter dilutions of a 1 ppm mother solution comprised of certified and traceable standard solutions of all elements of interest. Procedural blanks were <0·1 ppb for all elements, with most close to, or at detection limits. Repeated analyses of rock standards GSP-2 and JR-1 yielded offsets from preferred values of <10% for all elements except Ho, Er (<15%), Y (<20%) and Zn (>20%), and 2 s.d. % similarly of <10% for all elements (Supplementary data Electronic Appendix 3). For the isotopic work, Sr, Nd, and Pb were separated from the same solution used for trace element determination. Sr and Nd separation methods broadly followed those of Pin et al. (2014), whereas for Pb separation, we employed the traditional bromide separation technique on an additional aliquot of the same solution (see Supplementary Data). An aliquot containing ≥200 ng Pb was dissolved in 1 M HBr and twice passed through AG1-X8 anion resin. For Nd and Sr, a separate aliquot was initially passed through AG50-WX8 cation resin, to reduce (in the case of the Sr) the abundance of Rb in the samples. This Sr fraction was further purified by passing through Sr-spec resin. Chromatographic separation of Nd from a REE-bearing column cut was undertaken using LN-spec resin. Isotopic ratio measurements were conducted on the same ThermoFisher Triton Thermal Ionisation Mass spectrometer (TIMS) instrument, initially at the OU and later at VUW. Dried Sr fractions were loaded onto a single Re filament assembly in concentrated HNO3 and a TaF2 emitter solution (Birck, 1986; Charlier et al., 2006). Measurements were corrected for fractionation to 86Sr/88Sr = 0·1194, and 85Rb was measured to correct for any 87Rb interference at mass 87 using 87Rb/85Rb = 0·385707 (Rosman & Taylor, 1998). The average of repeated analyses of NBS987 gave 87Sr/86Sr = 0·710255 ± 0·0000013 (2 s.d., n = 25). The 87Sr/86Sr ratios from different runs were normalised to a common value of 0.71025 for NBS987 (following Thirlwall, 1991). The procedural blank was 200 pg, and considered insignificant compared to the ∼1000 ng loaded onto each filament. The fractionation corrected 87Sr/86Sr ratios were age-corrected (AC) using 2·08 Ma (Rivera et al., 2014; Singer et al., 2014; Wotzlaw et al., 2015), to account for radiogenic ingrowth due to the large Rb/Sr in some samples. Nd fractions were loaded onto a double Re filament assembly using HCl and a H3PO4 loading solution. The 143Nd/144Nd was corrected for instrumental fractionation to 146Nd/144Nd = 0·7219, and corrected for Ce and Sm interferences using 140Ce/142Ce = 7·97279 and 144Sm/147Sm = 0·20667 (Rosman & Taylor, 1998). Repeated analyses of J&M and La Jolla standards yielded values of 0·511818 ± 0·000004 (2 s.d., n = 26) and 0·511845 ± 0·000002 (2 s.d., n = 6) respectively, both similar to the long-term laboratory average (J&M: 0·511821 ± 0·000002, 2 s.d.) and the value of Thirlwall (1991) for La Jolla (0·511856 ± 0·000007, 2 s.d.). Procedural blank was 300 pg. We express the Nd isotopic values in εNd notation (DePaolo & Wasserburg, 1976) which is calculated from measured 143Nd/144Nd ratios by: εNd= [(143Nd/144Nd)meas(143Nd/144Nd)CHUR−1]×104, where the 143Nd/144Nd value of CHUR (chondritic uniform reservoir) = 0·51263 (Bouvier et al., 2008). Pb fractions were analysed using the double-spike method (Todt et al., 1996). After column chemistry, the sample was divided and half loaded onto a single Re filament using a H3PO4-silica gel mix loading solution (Gerstenberger & Haase, 1997). Subsequently, the remaining half was mixed with the loading solution and a 207Pb/204Pb spike, then loaded onto a separate filament (Thirlwall, 2000). The unspiked natural and spiked samples were analysed during separate analytical sessions, followed by deconvolution using a modified Newton–Raphson inversion to determine the isotopic ratios. Values for standard NBS981 were 206Pb/204Pb = 16·945 ± 0·001, 207Pb/204Pb = 15·503 ± 0·001 and 208Pb/204Pb = 36·737 ± 0·004 (2 s.d., n = 32). Isotopic measurements from different analytical sessions were normalised to the NBS981 values of Todt et al. (1996). The procedural blank was 2 ng, considered insignificant when compared to the >200 ng analysed and thus no blank correction was applied. In situ analysis The remaining crushed material was sieved to 1 mm and any >1 mm crystals picked out. Material from glassy, non-mingled pumices was then immersed in lithium polytungstate heavy liquid (adjusted to ∼2·4 g/cm3) to separate glass and crystals. Weights of the starting materials and separated crystals were recorded in order to derive the crystal wt %. Samples were subsequently dry-sieved at 0·5 mm and wet-sieved at 250, 125 and 63 μm mesh. Crystals were picked from the 0·5–1 mm size fraction and set in 25 mm epoxy mounts with the crystals oriented parallel to their c-axes where possible. Glass was picked from the 250–500 μm sieve fraction and similarly mounted. The mounts were polished, coated with a ∼25 nm carbon film and analysed by wavelength dispersive electron probe microanalysis (EPMA) on a JXA 8230 Superprobe at VUW. All samples were initially imaged using backscattered electron imaging. Crystals were analysed using a 15 kV accelerating voltage, a 12 nA beam current, and a ∼1 μm spot size. All elements in feldspars and pyroxenes were analysed with 30 s count time on peak and a 15 s background count. Ti, Si, Mg, Fe and Al in olivines were similarly analysed with 30/15 s count times, whereas Ca, Mn and Ni were analysed over 60 s (peak) and 30 s (background) to improve precision and accuracy. Glass was analysed using a 15 kV accelerating voltage and a defocused 20 μm spot size. To reduce the effect of alkali migration a 2 nA beam current was used to analyse Si, Al, Na, K, Fe and Ca, with Si and Na analysed first, as recommended by Humphreys et al. (2006). Ti, Mg and Mn were then analysed using an 8 nA current. Peak and background count times were 15/8 (Na), 30/15 (Si, Al), 40/20 (K, Ca, Ti, Mg) and 60/20 s (Fe, Mn). Unknown samples were bracketed by internationally accredited samples to monitor accuracy and precision of the analytical sequence and to monitor instrumental drift. Standards data are given in Supplementary Data Electronic Appendices 4 and 5. In situ trace element analysis of glass fragments derived from crushed pumices, encompassing the same area as analysed by EPMA, was conducted at VUW using laser ablation inductively-coupled plasma mass spectrometry (LA-ICP-MS) on a RESOlution S155-SE 193 nm Excimer Laser coupled with an Agilent 7500CS ICP-MS. Two 70 μm diameter pre-ablation pulses were used to clean the sample surface at the site of ablation, followed by analysis using a 35 μm spot. Synthetic glass GSD-1G was used for calibration of 29Si concentrations, derived from EPMA analyses, and used as an internal standard. NIST 612, BCR-2G and BHVO-2G were also analysed repeatedly as secondary standards. Unknown samples were bracketed by calibration and secondary standards. Data were processed using LaserTRAM software (Loewen & Kent, 2012) to derive concentrations. Standard data are given in Supplementary Data Electronic Appendix 4. SINGLE CLAST GEOCHEMISTRY Major elements The silicic juvenile materials reported here show a wide diversity, from dacite to high-silica rhyolite (66–79 wt % SiO2, recalculated volatile free: Table 2; Supplementary Data Electronic Appendix 3). A number of distinct trends are observed for major (and trace) element compositions, even within single ignimbrite members, and we define seven suites (A1, A2, A3, B1, C1, C2, C3) based on their compositional characteristics and trends; these are letter-labelled by their host ignimbrite member. In general, the physical appearances and host stratigraphic units of the nine clast types are closely reflected in their compositional characteristics and groupings into the seven suites. Table 2: Representative major and trace element data for juvenile materials in the Huckleberry Ridge Tuff Sample YP114 YP459 YP251 YP191 YP255 YP078 YP061 YP266 YP139 YP271 YP504 YP611 Locality 263 397 397 99 397 65 65 64 332 332 502 509 Host unit (A) (A) (A) (A) (A) (B) (B) (B) (B in C) (B in C) (C) (C) Clast type J1 J1 J2 J3 J1 J1 J4 J4 J1 J1 J5 J5 Suite A1 A1 A1 A2 A3 B1 B1 B1 B1 B1 C1 C1 MS 1 1 1 1 2 1 1 1 1 1 4 4 SiO2 72·98 77·30 75·01 69·87 77·00 76·27 71·43 73·74 77·07 76·44 75·94 72·21 TiO2 0·26 0·12 0·17 0·47 0·11 0·14 0·33 0·27 0·13 0·11 0·31 0·56 Al2O3 13·56 11·69 12·72 14·05 11·85 12·21 13·75 13·14 11·99 11·62 11·88 12·05 Fe2O3 2·96 1·59 2·14 4·87 1·58 1·85 3·88 3·02 1·66 1·58 3·59 6·79 MnO 0·05 0·03 0·05 0·11 0·03 0·04 0·09 0·06 0·03 0·03 0·05 0·07 MgO 0·08 0·07 0·05 0·24 0·08 0·09 0·15 0·13 0·03 0·19 0·05 0·09 CaO 0·82 0·70 0·80 1·65 0·61 0·74 1·49 1·21 0·70 1·93 0·72 1·40 Na2O 4·05 2·86 3·58 3·68 2·89 3·39 3·68 3·06 3·26 3·11 3·52 2·99 K2O 5·22 5·64 5·45 4·94 5·83 5·27 5·14 5·31 5·11 4·96 3·92 3·36 P2O5 0·04 0·02 0·03 0·11 0·02 0·02 0·05 0·05 0·02 0·02 0·02 0·10 Total 99·24 100·24 99·80 99·91 100·36 99·32 99·20 99·89 100·22 100·57 99·47 100·53 LOI 0·27 2·25 1·85 2·76 2·27 2·40 2·72 2·96 2·31 3·52 0·76 2·67 K/Na 1·44 2·21 1·71 1·50 2·26 1·74 1·56 1·94 1·75 1·79 1·25 1·26 Li 24 26 37 26 30 18·8 20 19 24 26 41 93 Sc 3·2 1·4 2·1 5·8 1·5 1·8 4·5 3·5 1·8 1·6 4·7 20 V 1·8 0·3 1·2 7·5 0·2 0·6 3·3 2·8 0·4 0·5 1·2 2·6 Cr 1·0 1·3 0·4 2·1 2·3 1·1 1·1 1·3 3·8 1·5 2·3 3·7 Co 0·6 0·2 0·4 2·1 0·2 0·5 1·0 1·3 0·6 0·5 0·4 1·1 Ni 0·3 0·1 0·3 1·7 0·2 0·0 0·3 0·1 0·7 0·3 Cu 4·9 2·3 4·2 6·5 5·2 9·0 3·0 14·7 12·7 7·8 5·5 13·3 Zn 95 56 70 97 64 58 101 86 31 49 88 129 Ga 25 18 20 21 20 19 23 22 22 20 17·5 21 Rb 124 124 119 83 161 126 106 122 153 164 96 98 Sr 60 21 34 76 12 35 67 54 32 39 182 232 Y 53 50 52 51 62 51 54 54 56 59 36 49 Zr 347 158 222 604 155 186 488 357 176 149 481 615 Nb 46 45 46 40 53 44 46 46 47 50 28 26 Cs 1·22 2·5 2·2 1·2 3·1 2·1 1·50 1·82 2·6 2·8 2·5 2·8 Ba 2398 676 1358 3374 301 1024 3021 2167 928 542 1575 1465 La 78 70 90 75 81 87 79 89 90 68 53 63 Ce 172 136 176 141 157 163 154 172 173 130 104 82 Pr 17·0 14·4 18·8 16·9 16·7 17·6 17·3 19 18·5 14·4 11·4 14·4 Nd 63 50 67 63 58 62 64 68 67 52 43 56 Sm 12·2 10·4 12·3 12·5 12·1 11·8 12·9 13 12·6 11·0 8·8 11·4 Eu 3·6 1·0 1·8 3·8 0·69 1·5 3·8 2·7 1·5 0·94 3·6 3·8 Gd 13·1 10·3 12·2 11·9 12·0 12·1 12·4 12·5 13·6 10·9 8·7 12·1 Tb 2·1 1·7 1·9 1·85 2·0 1·8 1·9 1·9 2·1 1·8 1·4 1·9 Dy 11·2 9·5 9·9 9·9 11·3 9·7 10·3 10·3 11·1 10·5 7·5 10·1 Ho 2·20 1·8 1·9 1·92 2·2 1·9 2·0 2·0 2·2 2·1 1·44 2·1 Er 6·4 5·7 5·8 5·7 6·8 5·8 6·2 6·0 6·6 6·3 4·2 6·0 Tm 0·89 0·80 0·80 0·81 0·94 0·79 0·84 0·83 0·92 0·89 0·56 0·83 Yb 5·7 5·1 5·1 5·0 6·0 5·0 5·5 5·3 6·0 5·6 3·5 5·5 Lu 0·83 0·96 0·91 1·01 1·05 0·96 1·02 0·96 0·86 0·92 0·65 0·84 Hf 9·8 6·3 7·5 14·2 6·4 6·9 13·1 10·1 6·4 6·0 10·9 12·2 Ta 1·6 4·1 3·6 3·0 4·5 3·7 3·3 3·4 1·4 3·8 2·2 0·88 Pb 25 31 28 21 36 25 23 23 27 40 29 32 Th 19·4 20 21 13·7 24 22 16·1 19·2 25 22 13·0 13·0 U 2·2 4·7 4·0 2·8 5·5 4·3 3·4 3·8 5·0 5·1 3·9 4·3 Rb/Sr 2·1 6·0 3·5 1·10 12·9 3·6 1·58 2·3 4·9 4·2 0·53 0·42 La/Yb 13·6 13·8 17·7 14·9 13·5 17·2 14·5 16·7 14·9 12·1 15·1 11·5 Eu* 0·86 0·30 0·45 0·94 0·18 0·38 0·90 0·63 0·36 0·26 1·25 0·98 1/Sr 0·017 0·048 0·029 0·013 0·080 0·029 0·015 0·019 0·032 0·025 0·005 0·004 1/Nd 0·016 0·020 0·015 0·016 0·017 0·016 0·016 0·015 0·015 0·019 0·023 0·018 Sample YP114 YP459 YP251 YP191 YP255 YP078 YP061 YP266 YP139 YP271 YP504 YP611 Locality 263 397 397 99 397 65 65 64 332 332 502 509 Host unit (A) (A) (A) (A) (A) (B) (B) (B) (B in C) (B in C) (C) (C) Clast type J1 J1 J2 J3 J1 J1 J4 J4 J1 J1 J5 J5 Suite A1 A1 A1 A2 A3 B1 B1 B1 B1 B1 C1 C1 MS 1 1 1 1 2 1 1 1 1 1 4 4 SiO2 72·98 77·30 75·01 69·87 77·00 76·27 71·43 73·74 77·07 76·44 75·94 72·21 TiO2 0·26 0·12 0·17 0·47 0·11 0·14 0·33 0·27 0·13 0·11 0·31 0·56 Al2O3 13·56 11·69 12·72 14·05 11·85 12·21 13·75 13·14 11·99 11·62 11·88 12·05 Fe2O3 2·96 1·59 2·14 4·87 1·58 1·85 3·88 3·02 1·66 1·58 3·59 6·79 MnO 0·05 0·03 0·05 0·11 0·03 0·04 0·09 0·06 0·03 0·03 0·05 0·07 MgO 0·08 0·07 0·05 0·24 0·08 0·09 0·15 0·13 0·03 0·19 0·05 0·09 CaO 0·82 0·70 0·80 1·65 0·61 0·74 1·49 1·21 0·70 1·93 0·72 1·40 Na2O 4·05 2·86 3·58 3·68 2·89 3·39 3·68 3·06 3·26 3·11 3·52 2·99 K2O 5·22 5·64 5·45 4·94 5·83 5·27 5·14 5·31 5·11 4·96 3·92 3·36 P2O5 0·04 0·02 0·03 0·11 0·02 0·02 0·05 0·05 0·02 0·02 0·02 0·10 Total 99·24 100·24 99·80 99·91 100·36 99·32 99·20 99·89 100·22 100·57 99·47 100·53 LOI 0·27 2·25 1·85 2·76 2·27 2·40 2·72 2·96 2·31 3·52 0·76 2·67 K/Na 1·44 2·21 1·71 1·50 2·26 1·74 1·56 1·94 1·75 1·79 1·25 1·26 Li 24 26 37 26 30 18·8 20 19 24 26 41 93 Sc 3·2 1·4 2·1 5·8 1·5 1·8 4·5 3·5 1·8 1·6 4·7 20 V 1·8 0·3 1·2 7·5 0·2 0·6 3·3 2·8 0·4 0·5 1·2 2·6 Cr 1·0 1·3 0·4 2·1 2·3 1·1 1·1 1·3 3·8 1·5 2·3 3·7 Co 0·6 0·2 0·4 2·1 0·2 0·5 1·0 1·3 0·6 0·5 0·4 1·1 Ni 0·3 0·1 0·3 1·7 0·2 0·0 0·3 0·1 0·7 0·3 Cu 4·9 2·3 4·2 6·5 5·2 9·0 3·0 14·7 12·7 7·8 5·5 13·3 Zn 95 56 70 97 64 58 101 86 31 49 88 129 Ga 25 18 20 21 20 19 23 22 22 20 17·5 21 Rb 124 124 119 83 161 126 106 122 153 164 96 98 Sr 60 21 34 76 12 35 67 54 32 39 182 232 Y 53 50 52 51 62 51 54 54 56 59 36 49 Zr 347 158 222 604 155 186 488 357 176 149 481 615 Nb 46 45 46 40 53 44 46 46 47 50 28 26 Cs 1·22 2·5 2·2 1·2 3·1 2·1 1·50 1·82 2·6 2·8 2·5 2·8 Ba 2398 676 1358 3374 301 1024 3021 2167 928 542 1575 1465 La 78 70 90 75 81 87 79 89 90 68 53 63 Ce 172 136 176 141 157 163 154 172 173 130 104 82 Pr 17·0 14·4 18·8 16·9 16·7 17·6 17·3 19 18·5 14·4 11·4 14·4 Nd 63 50 67 63 58 62 64 68 67 52 43 56 Sm 12·2 10·4 12·3 12·5 12·1 11·8 12·9 13 12·6 11·0 8·8 11·4 Eu 3·6 1·0 1·8 3·8 0·69 1·5 3·8 2·7 1·5 0·94 3·6 3·8 Gd 13·1 10·3 12·2 11·9 12·0 12·1 12·4 12·5 13·6 10·9 8·7 12·1 Tb 2·1 1·7 1·9 1·85 2·0 1·8 1·9 1·9 2·1 1·8 1·4 1·9 Dy 11·2 9·5 9·9 9·9 11·3 9·7 10·3 10·3 11·1 10·5 7·5 10·1 Ho 2·20 1·8 1·9 1·92 2·2 1·9 2·0 2·0 2·2 2·1 1·44 2·1 Er 6·4 5·7 5·8 5·7 6·8 5·8 6·2 6·0 6·6 6·3 4·2 6·0 Tm 0·89 0·80 0·80 0·81 0·94 0·79 0·84 0·83 0·92 0·89 0·56 0·83 Yb 5·7 5·1 5·1 5·0 6·0 5·0 5·5 5·3 6·0 5·6 3·5 5·5 Lu 0·83 0·96 0·91 1·01 1·05 0·96 1·02 0·96 0·86 0·92 0·65 0·84 Hf 9·8 6·3 7·5 14·2 6·4 6·9 13·1 10·1 6·4 6·0 10·9 12·2 Ta 1·6 4·1 3·6 3·0 4·5 3·7 3·3 3·4 1·4 3·8 2·2 0·88 Pb 25 31 28 21 36 25 23 23 27 40 29 32 Th 19·4 20 21 13·7 24 22 16·1 19·2 25 22 13·0 13·0 U 2·2 4·7 4·0 2·8 5·5 4·3 3·4 3·8 5·0 5·1 3·9 4·3 Rb/Sr 2·1 6·0 3·5 1·10 12·9 3·6 1·58 2·3 4·9 4·2 0·53 0·42 La/Yb 13·6 13·8 17·7 14·9 13·5 17·2 14·5 16·7 14·9 12·1 15·1 11·5 Eu* 0·86 0·30 0·45 0·94 0·18 0·38 0·90 0·63 0·36 0·26 1·25 0·98 1/Sr 0·017 0·048 0·029 0·013 0·080 0·029 0·015 0·019 0·032 0·025 0·005 0·004 1/Nd 0·016 0·020 0·015 0·016 0·017 0·016 0·016 0·015 0·015 0·019 0·023 0·018 Sample YP648 YR225 YP282 YP105 YP366 YP502 YP133 YP575 YP186 YP227 YP212 Locality 498 351 57 57 465 502 332 361 109 46 363 Host Unit (C) Intr (C) (C) (C) (C) (C) (A) (A) (B) (B) Clast type J5 J6 J8 J8 J9 J9 J7 DM DM Scoria Scoria Suite C1 C1 C2 C2 C2 C2 C3 M1 M1 M1 M2 MS 4 4 3 3 3 3 2 1 1 1 1 SiO2 66·46 68·69 75·78 76·84 77·64 77·23 76·69 53·74 64·25 57·99 53·46 TiO2 1·09 0·86 0·25 0·12 0·13 0·16 0·10 3·13 1·42 1·27 1·19 Al2O3 11·54 11·55 12·77 12·35 12·00 11·72 12·04 14·92 15·02 14·46 14·53 Fe2O3 10·68 9·05 2·80 1·62 1·32 1·98 1·55 13·48 9·20 9·19 13·32 MnO 0·15 0·13 0·05 0·02 0·01 0·06 0·02 0·08 0·21 0·15 0·20 MgO 0·47 0·26 0·05 0·04 0·00 0·02 0·25 1·02 0·55 4·32 4·33 CaO 3·91 3·58 0·81 0·43 0·35 0·35 1·45 5·69 1·69 7·23 6·77 Na2O 2·91 3·02 2·72 2·85 3·55 3·43 3·03 4·00 3·04 3·65 3·87 K2O 2·18 2·60 4·76 5·72 4·97 4·80 4·85 2·13 4·12 1·59 2·19 P2O5 0·38 0·26 0·03 0·01 0·02 0·01 0·02 1·81 0·50 0·15 0·15 Total 100·22 99·78 100·09 99·15 100·28 100·51 100·41 99·39 99·70 99·61 99·71 LOI 1·75 0·62 2·98 3·32 0·26 0·75 4·26 1·64 2·32 0·32 0·93 K/Na 0·84 0·96 1·96 2·25 1·57 1·57 1·79 0·59 1·52 0·49 0·63 Li 35 17·0 40 40 54 36 27 13·5 39 25 56 Sc 39 38 3·2 3 2 2·8 1·2 35 15·1 23 24 V 6·1 5·6 1·0 0·2 0·1 0·5 0·2 108 25 193 163 Cr 7·7 1·8 2·1 1·6 2·6 2·7 1·6 5·7 2·7 59 100 Co 6·3 4·7 0·5 0·7 0·1 0·2 0·3 12·7 11·7 28 16·6 Ni 0·9 0·8 0·5 0·1 0·1 4·6 39 25 Cu 8·7 8·8 15·9 43 2·2 5·7 6·5 17·9 14·1 31 16·2 Zn 103 130 83 70 63 62 48 400 196 192 432 Ga 16·9 20 19·0 18·5 17·8 19·1 21 28 28 23 27 Rb 57 89·8 101 138 126 143 175 39 65 46 55 Sr 245 307 108 18·6 16·9 47 25 334 113 183 166 Y 27 38 39 50 14·5 28 68 184 139 40 67 Zr 310 491 379 210 197 280 158 1689 761 185 61 Nb 17·0 26 28 32 30 28 60 31 47 26 87 Cs 7·9 2·4 3·2 4·8 2·9 3·0 3·6 0·73 1·55 1·02 1·13 Ba 798 952 2559 449 470 1520 98 3571 2539 477 388 La 32 45 59 70 35 41 42 132 182 52 97 Ce 65 92 110 134 54 114 89 157 180 102 140 Pr 7·5 10·8 12·6 15·1 7·6 9·2 10·4 34 42 11·6 23 Nd 29 42 47 55 28 34 39 145 166 45 87 Sm 6·3 8·7 9·7 11·4 5·6 7·0 10·0 33 34 9·2 17·3 Eu 2·4 3·3 3·3 0·89 1·05 2·1 0·38 9·8 5·4 1·78 1·55 Gd 6·3 8·5 9·2 11·3 4·8 6·6 10·5 36 36 10·1 18·5 Tb 1·0 1·25 1·40 1·76 0·74 1·04 2·0 5·9 5·5 1·53 2·8 Dy 5·3 7·1 7·7 9·6 3·7 5·6 11·9 35 30 8·6 15·1 Ho 1·02 1·40 1·50 1·86 0·67 1·09 2·4 7·3 5·9 1·71 2·9 Er 3·0 3·9 4·4 5·7 2·0 3·3 7·4 22 17·1 5·0 8·3 Tm 0·40 0·56 0·61 0·77 0·27 0·45 1·06 3·1 2·2 0·67 1·10 Yb 2·6 3·5 3·9 4·8 1·8 3·0 6·7 21 14·0 4·5 6·9 Lu 0·50 0·54 0·71 0·89 0·35 0·50 1·19 3·2 2·1 0·63 0·97 Hf 7·0 10 9·5 7·2 6·8 7·8 7·0 31 28 6·6 5·1 Ta 1·48 1·44 2·4 2·9 2·9 2·1 5·3 2·3 6·4 2·0 2·5 Pb 16·8 27 35 42 31 29 42 27 31 31 17·2 Th 7·7 10·6 15 20·0 17·9 17·1 25 7·3 19·1 12·8 8·5 U 2·5 3·1 4·0 6·0 4·1 4·2 6·4 1·91 5·2 2·2 1·58 Rb/Sr 0·23 0·29 0·94 7·44 7·47 3·02 7·1 0·12 0·58 0·25 0·33 La/Yb 12·4 12·6 15·1 14·5 19·8 13·9 6·2 6·3 13·1 11·7 13·9 Eu* 1·17 1·18 1·05 0·24 0·62 0·93 0·11 0·87 0·47 0·56 0·26 1/Sr 0·004 0·003 0·009 0·054 0·059 0·021 0·041 0·0030 0·0089 0·0055 0·0060 1/Nd 0·034 0·024 0·021 0·018 0·036 0·030 0·026 0·0069 0·0060 0·0060 0·0115 Sample YP648 YR225 YP282 YP105 YP366 YP502 YP133 YP575 YP186 YP227 YP212 Locality 498 351 57 57 465 502 332 361 109 46 363 Host Unit (C) Intr (C) (C) (C) (C) (C) (A) (A) (B) (B) Clast type J5 J6 J8 J8 J9 J9 J7 DM DM Scoria Scoria Suite C1 C1 C2 C2 C2 C2 C3 M1 M1 M1 M2 MS 4 4 3 3 3 3 2 1 1 1 1 SiO2 66·46 68·69 75·78 76·84 77·64 77·23 76·69 53·74 64·25 57·99 53·46 TiO2 1·09 0·86 0·25 0·12 0·13 0·16 0·10 3·13 1·42 1·27 1·19 Al2O3 11·54 11·55 12·77 12·35 12·00 11·72 12·04 14·92 15·02 14·46 14·53 Fe2O3 10·68 9·05 2·80 1·62 1·32 1·98 1·55 13·48 9·20 9·19 13·32 MnO 0·15 0·13 0·05 0·02 0·01 0·06 0·02 0·08 0·21 0·15 0·20 MgO 0·47 0·26 0·05 0·04 0·00 0·02 0·25 1·02 0·55 4·32 4·33 CaO 3·91 3·58 0·81 0·43 0·35 0·35 1·45 5·69 1·69 7·23 6·77 Na2O 2·91 3·02 2·72 2·85 3·55 3·43 3·03 4·00 3·04 3·65 3·87 K2O 2·18 2·60 4·76 5·72 4·97 4·80 4·85 2·13 4·12 1·59 2·19 P2O5 0·38 0·26 0·03 0·01 0·02 0·01 0·02 1·81 0·50 0·15 0·15 Total 100·22 99·78 100·09 99·15 100·28 100·51 100·41 99·39 99·70 99·61 99·71 LOI 1·75 0·62 2·98 3·32 0·26 0·75 4·26 1·64 2·32 0·32 0·93 K/Na 0·84 0·96 1·96 2·25 1·57 1·57 1·79 0·59 1·52 0·49 0·63 Li 35 17·0 40 40 54 36 27 13·5 39 25 56 Sc 39 38 3·2 3 2 2·8 1·2 35 15·1 23 24 V 6·1 5·6 1·0 0·2 0·1 0·5 0·2 108 25 193 163 Cr 7·7 1·8 2·1 1·6 2·6 2·7 1·6 5·7 2·7 59 100 Co 6·3 4·7 0·5 0·7 0·1 0·2 0·3 12·7 11·7 28 16·6 Ni 0·9 0·8 0·5 0·1 0·1 4·6 39 25 Cu 8·7 8·8 15·9 43 2·2 5·7 6·5 17·9 14·1 31 16·2 Zn 103 130 83 70 63 62 48 400 196 192 432 Ga 16·9 20 19·0 18·5 17·8 19·1 21 28 28 23 27 Rb 57 89·8 101 138 126 143 175 39 65 46 55 Sr 245 307 108 18·6 16·9 47 25 334 113 183 166 Y 27 38 39 50 14·5 28 68 184 139 40 67 Zr 310 491 379 210 197 280 158 1689 761 185 61 Nb 17·0 26 28 32 30 28 60 31 47 26 87 Cs 7·9 2·4 3·2 4·8 2·9 3·0 3·6 0·73 1·55 1·02 1·13 Ba 798 952 2559 449 470 1520 98 3571 2539 477 388 La 32 45 59 70 35 41 42 132 182 52 97 Ce 65 92 110 134 54 114 89 157 180 102 140 Pr 7·5 10·8 12·6 15·1 7·6 9·2 10·4 34 42 11·6 23 Nd 29 42 47 55 28 34 39 145 166 45 87 Sm 6·3 8·7 9·7 11·4 5·6 7·0 10·0 33 34 9·2 17·3 Eu 2·4 3·3 3·3 0·89 1·05 2·1 0·38 9·8 5·4 1·78 1·55 Gd 6·3 8·5 9·2 11·3 4·8 6·6 10·5 36 36 10·1 18·5 Tb 1·0 1·25 1·40 1·76 0·74 1·04 2·0 5·9 5·5 1·53 2·8 Dy 5·3 7·1 7·7 9·6 3·7 5·6 11·9 35 30 8·6 15·1 Ho 1·02 1·40 1·50 1·86 0·67 1·09 2·4 7·3 5·9 1·71 2·9 Er 3·0 3·9 4·4 5·7 2·0 3·3 7·4 22 17·1 5·0 8·3 Tm 0·40 0·56 0·61 0·77 0·27 0·45 1·06 3·1 2·2 0·67 1·10 Yb 2·6 3·5 3·9 4·8 1·8 3·0 6·7 21 14·0 4·5 6·9 Lu 0·50 0·54 0·71 0·89 0·35 0·50 1·19 3·2 2·1 0·63 0·97 Hf 7·0 10 9·5 7·2 6·8 7·8 7·0 31 28 6·6 5·1 Ta 1·48 1·44 2·4 2·9 2·9 2·1 5·3 2·3 6·4 2·0 2·5 Pb 16·8 27 35 42 31 29 42 27 31 31 17·2 Th 7·7 10·6 15 20·0 17·9 17·1 25 7·3 19·1 12·8 8·5 U 2·5 3·1 4·0 6·0 4·1 4·2 6·4 1·91 5·2 2·2 1·58 Rb/Sr 0·23 0·29 0·94 7·44 7·47 3·02 7·1 0·12 0·58 0·25 0·33 La/Yb 12·4 12·6 15·1 14·5 19·8 13·9 6·2 6·3 13·1 11·7 13·9 Eu* 1·17 1·18 1·05 0·24 0·62 0·93 0·11 0·87 0·47 0·56 0·26 1/Sr 0·004 0·003 0·009 0·054 0·059 0·021 0·041 0·0030 0·0089 0·0055 0·0060 1/Nd 0·034 0·024 0·021 0·018 0·036 0·030 0·026 0·0069 0·0060 0·0060 0·0115 Major elements and Ba by XRF; trace elements by solution ICP-MS (see Supplementary Material for details). Major element values are recalculated to 100% but with the original totals and LOI figures given. Fe2O3 is total iron as determined by XRF. Full data are given in Supplementary Data Electronic Appendix 3. See Table 1 for definition of clast types, and text for characterization of compositional suites and their host magmatic systems. MS, magmatic system; DM, dense mafic clasts, Intr, intrusive body cutting plastic, still-hot member C ignimbrite. Open in new tab Table 2: Representative major and trace element data for juvenile materials in the Huckleberry Ridge Tuff Sample YP114 YP459 YP251 YP191 YP255 YP078 YP061 YP266 YP139 YP271 YP504 YP611 Locality 263 397 397 99 397 65 65 64 332 332 502 509 Host unit (A) (A) (A) (A) (A) (B) (B) (B) (B in C) (B in C) (C) (C) Clast type J1 J1 J2 J3 J1 J1 J4 J4 J1 J1 J5 J5 Suite A1 A1 A1 A2 A3 B1 B1 B1 B1 B1 C1 C1 MS 1 1 1 1 2 1 1 1 1 1 4 4 SiO2 72·98 77·30 75·01 69·87 77·00 76·27 71·43 73·74 77·07 76·44 75·94 72·21 TiO2 0·26 0·12 0·17 0·47 0·11 0·14 0·33 0·27 0·13 0·11 0·31 0·56 Al2O3 13·56 11·69 12·72 14·05 11·85 12·21 13·75 13·14 11·99 11·62 11·88 12·05 Fe2O3 2·96 1·59 2·14 4·87 1·58 1·85 3·88 3·02 1·66 1·58 3·59 6·79 MnO 0·05 0·03 0·05 0·11 0·03 0·04 0·09 0·06 0·03 0·03 0·05 0·07 MgO 0·08 0·07 0·05 0·24 0·08 0·09 0·15 0·13 0·03 0·19 0·05 0·09 CaO 0·82 0·70 0·80 1·65 0·61 0·74 1·49 1·21 0·70 1·93 0·72 1·40 Na2O 4·05 2·86 3·58 3·68 2·89 3·39 3·68 3·06 3·26 3·11 3·52 2·99 K2O 5·22 5·64 5·45 4·94 5·83 5·27 5·14 5·31 5·11 4·96 3·92 3·36 P2O5 0·04 0·02 0·03 0·11 0·02 0·02 0·05 0·05 0·02 0·02 0·02 0·10 Total 99·24 100·24 99·80 99·91 100·36 99·32 99·20 99·89 100·22 100·57 99·47 100·53 LOI 0·27 2·25 1·85 2·76 2·27 2·40 2·72 2·96 2·31 3·52 0·76 2·67 K/Na 1·44 2·21 1·71 1·50 2·26 1·74 1·56 1·94 1·75 1·79 1·25 1·26 Li 24 26 37 26 30 18·8 20 19 24 26 41 93 Sc 3·2 1·4 2·1 5·8 1·5 1·8 4·5 3·5 1·8 1·6 4·7 20 V 1·8 0·3 1·2 7·5 0·2 0·6 3·3 2·8 0·4 0·5 1·2 2·6 Cr 1·0 1·3 0·4 2·1 2·3 1·1 1·1 1·3 3·8 1·5 2·3 3·7 Co 0·6 0·2 0·4 2·1 0·2 0·5 1·0 1·3 0·6 0·5 0·4 1·1 Ni 0·3 0·1 0·3 1·7 0·2 0·0 0·3 0·1 0·7 0·3 Cu 4·9 2·3 4·2 6·5 5·2 9·0 3·0 14·7 12·7 7·8 5·5 13·3 Zn 95 56 70 97 64 58 101 86 31 49 88 129 Ga 25 18 20 21 20 19 23 22 22 20 17·5 21 Rb 124 124 119 83 161 126 106 122 153 164 96 98 Sr 60 21 34 76 12 35 67 54 32 39 182 232 Y 53 50 52 51 62 51 54 54 56 59 36 49 Zr 347 158 222 604 155 186 488 357 176 149 481 615 Nb 46 45 46 40 53 44 46 46 47 50 28 26 Cs 1·22 2·5 2·2 1·2 3·1 2·1 1·50 1·82 2·6 2·8 2·5 2·8 Ba 2398 676 1358 3374 301 1024 3021 2167 928 542 1575 1465 La 78 70 90 75 81 87 79 89 90 68 53 63 Ce 172 136 176 141 157 163 154 172 173 130 104 82 Pr 17·0 14·4 18·8 16·9 16·7 17·6 17·3 19 18·5 14·4 11·4 14·4 Nd 63 50 67 63 58 62 64 68 67 52 43 56 Sm 12·2 10·4 12·3 12·5 12·1 11·8 12·9 13 12·6 11·0 8·8 11·4 Eu 3·6 1·0 1·8 3·8 0·69 1·5 3·8 2·7 1·5 0·94 3·6 3·8 Gd 13·1 10·3 12·2 11·9 12·0 12·1 12·4 12·5 13·6 10·9 8·7 12·1 Tb 2·1 1·7 1·9 1·85 2·0 1·8 1·9 1·9 2·1 1·8 1·4 1·9 Dy 11·2 9·5 9·9 9·9 11·3 9·7 10·3 10·3 11·1 10·5 7·5 10·1 Ho 2·20 1·8 1·9 1·92 2·2 1·9 2·0 2·0 2·2 2·1 1·44 2·1 Er 6·4 5·7 5·8 5·7 6·8 5·8 6·2 6·0 6·6 6·3 4·2 6·0 Tm 0·89 0·80 0·80 0·81 0·94 0·79 0·84 0·83 0·92 0·89 0·56 0·83 Yb 5·7 5·1 5·1 5·0 6·0 5·0 5·5 5·3 6·0 5·6 3·5 5·5 Lu 0·83 0·96 0·91 1·01 1·05 0·96 1·02 0·96 0·86 0·92 0·65 0·84 Hf 9·8 6·3 7·5 14·2 6·4 6·9 13·1 10·1 6·4 6·0 10·9 12·2 Ta 1·6 4·1 3·6 3·0 4·5 3·7 3·3 3·4 1·4 3·8 2·2 0·88 Pb 25 31 28 21 36 25 23 23 27 40 29 32 Th 19·4 20 21 13·7 24 22 16·1 19·2 25 22 13·0 13·0 U 2·2 4·7 4·0 2·8 5·5 4·3 3·4 3·8 5·0 5·1 3·9 4·3 Rb/Sr 2·1 6·0 3·5 1·10 12·9 3·6 1·58 2·3 4·9 4·2 0·53 0·42 La/Yb 13·6 13·8 17·7 14·9 13·5 17·2 14·5 16·7 14·9 12·1 15·1 11·5 Eu* 0·86 0·30 0·45 0·94 0·18 0·38 0·90 0·63 0·36 0·26 1·25 0·98 1/Sr 0·017 0·048 0·029 0·013 0·080 0·029 0·015 0·019 0·032 0·025 0·005 0·004 1/Nd 0·016 0·020 0·015 0·016 0·017 0·016 0·016 0·015 0·015 0·019 0·023 0·018 Sample YP114 YP459 YP251 YP191 YP255 YP078 YP061 YP266 YP139 YP271 YP504 YP611 Locality 263 397 397 99 397 65 65 64 332 332 502 509 Host unit (A) (A) (A) (A) (A) (B) (B) (B) (B in C) (B in C) (C) (C) Clast type J1 J1 J2 J3 J1 J1 J4 J4 J1 J1 J5 J5 Suite A1 A1 A1 A2 A3 B1 B1 B1 B1 B1 C1 C1 MS 1 1 1 1 2 1 1 1 1 1 4 4 SiO2 72·98 77·30 75·01 69·87 77·00 76·27 71·43 73·74 77·07 76·44 75·94 72·21 TiO2 0·26 0·12 0·17 0·47 0·11 0·14 0·33 0·27 0·13 0·11 0·31 0·56 Al2O3 13·56 11·69 12·72 14·05 11·85 12·21 13·75 13·14 11·99 11·62 11·88 12·05 Fe2O3 2·96 1·59 2·14 4·87 1·58 1·85 3·88 3·02 1·66 1·58 3·59 6·79 MnO 0·05 0·03 0·05 0·11 0·03 0·04 0·09 0·06 0·03 0·03 0·05 0·07 MgO 0·08 0·07 0·05 0·24 0·08 0·09 0·15 0·13 0·03 0·19 0·05 0·09 CaO 0·82 0·70 0·80 1·65 0·61 0·74 1·49 1·21 0·70 1·93 0·72 1·40 Na2O 4·05 2·86 3·58 3·68 2·89 3·39 3·68 3·06 3·26 3·11 3·52 2·99 K2O 5·22 5·64 5·45 4·94 5·83 5·27 5·14 5·31 5·11 4·96 3·92 3·36 P2O5 0·04 0·02 0·03 0·11 0·02 0·02 0·05 0·05 0·02 0·02 0·02 0·10 Total 99·24 100·24 99·80 99·91 100·36 99·32 99·20 99·89 100·22 100·57 99·47 100·53 LOI 0·27 2·25 1·85 2·76 2·27 2·40 2·72 2·96 2·31 3·52 0·76 2·67 K/Na 1·44 2·21 1·71 1·50 2·26 1·74 1·56 1·94 1·75 1·79 1·25 1·26 Li 24 26 37 26 30 18·8 20 19 24 26 41 93 Sc 3·2 1·4 2·1 5·8 1·5 1·8 4·5 3·5 1·8 1·6 4·7 20 V 1·8 0·3 1·2 7·5 0·2 0·6 3·3 2·8 0·4 0·5 1·2 2·6 Cr 1·0 1·3 0·4 2·1 2·3 1·1 1·1 1·3 3·8 1·5 2·3 3·7 Co 0·6 0·2 0·4 2·1 0·2 0·5 1·0 1·3 0·6 0·5 0·4 1·1 Ni 0·3 0·1 0·3 1·7 0·2 0·0 0·3 0·1 0·7 0·3 Cu 4·9 2·3 4·2 6·5 5·2 9·0 3·0 14·7 12·7 7·8 5·5 13·3 Zn 95 56 70 97 64 58 101 86 31 49 88 129 Ga 25 18 20 21 20 19 23 22 22 20 17·5 21 Rb 124 124 119 83 161 126 106 122 153 164 96 98 Sr 60 21 34 76 12 35 67 54 32 39 182 232 Y 53 50 52 51 62 51 54 54 56 59 36 49 Zr 347 158 222 604 155 186 488 357 176 149 481 615 Nb 46 45 46 40 53 44 46 46 47 50 28 26 Cs 1·22 2·5 2·2 1·2 3·1 2·1 1·50 1·82 2·6 2·8 2·5 2·8 Ba 2398 676 1358 3374 301 1024 3021 2167 928 542 1575 1465 La 78 70 90 75 81 87 79 89 90 68 53 63 Ce 172 136 176 141 157 163 154 172 173 130 104 82 Pr 17·0 14·4 18·8 16·9 16·7 17·6 17·3 19 18·5 14·4 11·4 14·4 Nd 63 50 67 63 58 62 64 68 67 52 43 56 Sm 12·2 10·4 12·3 12·5 12·1 11·8 12·9 13 12·6 11·0 8·8 11·4 Eu 3·6 1·0 1·8 3·8 0·69 1·5 3·8 2·7 1·5 0·94 3·6 3·8 Gd 13·1 10·3 12·2 11·9 12·0 12·1 12·4 12·5 13·6 10·9 8·7 12·1 Tb 2·1 1·7 1·9 1·85 2·0 1·8 1·9 1·9 2·1 1·8 1·4 1·9 Dy 11·2 9·5 9·9 9·9 11·3 9·7 10·3 10·3 11·1 10·5 7·5 10·1 Ho 2·20 1·8 1·9 1·92 2·2 1·9 2·0 2·0 2·2 2·1 1·44 2·1 Er 6·4 5·7 5·8 5·7 6·8 5·8 6·2 6·0 6·6 6·3 4·2 6·0 Tm 0·89 0·80 0·80 0·81 0·94 0·79 0·84 0·83 0·92 0·89 0·56 0·83 Yb 5·7 5·1 5·1 5·0 6·0 5·0 5·5 5·3 6·0 5·6 3·5 5·5 Lu 0·83 0·96 0·91 1·01 1·05 0·96 1·02 0·96 0·86 0·92 0·65 0·84 Hf 9·8 6·3 7·5 14·2 6·4 6·9 13·1 10·1 6·4 6·0 10·9 12·2 Ta 1·6 4·1 3·6 3·0 4·5 3·7 3·3 3·4 1·4 3·8 2·2 0·88 Pb 25 31 28 21 36 25 23 23 27 40 29 32 Th 19·4 20 21 13·7 24 22 16·1 19·2 25 22 13·0 13·0 U 2·2 4·7 4·0 2·8 5·5 4·3 3·4 3·8 5·0 5·1 3·9 4·3 Rb/Sr 2·1 6·0 3·5 1·10 12·9 3·6 1·58 2·3 4·9 4·2 0·53 0·42 La/Yb 13·6 13·8 17·7 14·9 13·5 17·2 14·5 16·7 14·9 12·1 15·1 11·5 Eu* 0·86 0·30 0·45 0·94 0·18 0·38 0·90 0·63 0·36 0·26 1·25 0·98 1/Sr 0·017 0·048 0·029 0·013 0·080 0·029 0·015 0·019 0·032 0·025 0·005 0·004 1/Nd 0·016 0·020 0·015 0·016 0·017 0·016 0·016 0·015 0·015 0·019 0·023 0·018 Sample YP648 YR225 YP282 YP105 YP366 YP502 YP133 YP575 YP186 YP227 YP212 Locality 498 351 57 57 465 502 332 361 109 46 363 Host Unit (C) Intr (C) (C) (C) (C) (C) (A) (A) (B) (B) Clast type J5 J6 J8 J8 J9 J9 J7 DM DM Scoria Scoria Suite C1 C1 C2 C2 C2 C2 C3 M1 M1 M1 M2 MS 4 4 3 3 3 3 2 1 1 1 1 SiO2 66·46 68·69 75·78 76·84 77·64 77·23 76·69 53·74 64·25 57·99 53·46 TiO2 1·09 0·86 0·25 0·12 0·13 0·16 0·10 3·13 1·42 1·27 1·19 Al2O3 11·54 11·55 12·77 12·35 12·00 11·72 12·04 14·92 15·02 14·46 14·53 Fe2O3 10·68 9·05 2·80 1·62 1·32 1·98 1·55 13·48 9·20 9·19 13·32 MnO 0·15 0·13 0·05 0·02 0·01 0·06 0·02 0·08 0·21 0·15 0·20 MgO 0·47 0·26 0·05 0·04 0·00 0·02 0·25 1·02 0·55 4·32 4·33 CaO 3·91 3·58 0·81 0·43 0·35 0·35 1·45 5·69 1·69 7·23 6·77 Na2O 2·91 3·02 2·72 2·85 3·55 3·43 3·03 4·00 3·04 3·65 3·87 K2O 2·18 2·60 4·76 5·72 4·97 4·80 4·85 2·13 4·12 1·59 2·19 P2O5 0·38 0·26 0·03 0·01 0·02 0·01 0·02 1·81 0·50 0·15 0·15 Total 100·22 99·78 100·09 99·15 100·28 100·51 100·41 99·39 99·70 99·61 99·71 LOI 1·75 0·62 2·98 3·32 0·26 0·75 4·26 1·64 2·32 0·32 0·93 K/Na 0·84 0·96 1·96 2·25 1·57 1·57 1·79 0·59 1·52 0·49 0·63 Li 35 17·0 40 40 54 36 27 13·5 39 25 56 Sc 39 38 3·2 3 2 2·8 1·2 35 15·1 23 24 V 6·1 5·6 1·0 0·2 0·1 0·5 0·2 108 25 193 163 Cr 7·7 1·8 2·1 1·6 2·6 2·7 1·6 5·7 2·7 59 100 Co 6·3 4·7 0·5 0·7 0·1 0·2 0·3 12·7 11·7 28 16·6 Ni 0·9 0·8 0·5 0·1 0·1 4·6 39 25 Cu 8·7 8·8 15·9 43 2·2 5·7 6·5 17·9 14·1 31 16·2 Zn 103 130 83 70 63 62 48 400 196 192 432 Ga 16·9 20 19·0 18·5 17·8 19·1 21 28 28 23 27 Rb 57 89·8 101 138 126 143 175 39 65 46 55 Sr 245 307 108 18·6 16·9 47 25 334 113 183 166 Y 27 38 39 50 14·5 28 68 184 139 40 67 Zr 310 491 379 210 197 280 158 1689 761 185 61 Nb 17·0 26 28 32 30 28 60 31 47 26 87 Cs 7·9 2·4 3·2 4·8 2·9 3·0 3·6 0·73 1·55 1·02 1·13 Ba 798 952 2559 449 470 1520 98 3571 2539 477 388 La 32 45 59 70 35 41 42 132 182 52 97 Ce 65 92 110 134 54 114 89 157 180 102 140 Pr 7·5 10·8 12·6 15·1 7·6 9·2 10·4 34 42 11·6 23 Nd 29 42 47 55 28 34 39 145 166 45 87 Sm 6·3 8·7 9·7 11·4 5·6 7·0 10·0 33 34 9·2 17·3 Eu 2·4 3·3 3·3 0·89 1·05 2·1 0·38 9·8 5·4 1·78 1·55 Gd 6·3 8·5 9·2 11·3 4·8 6·6 10·5 36 36 10·1 18·5 Tb 1·0 1·25 1·40 1·76 0·74 1·04 2·0 5·9 5·5 1·53 2·8 Dy 5·3 7·1 7·7 9·6 3·7 5·6 11·9 35 30 8·6 15·1 Ho 1·02 1·40 1·50 1·86 0·67 1·09 2·4 7·3 5·9 1·71 2·9 Er 3·0 3·9 4·4 5·7 2·0 3·3 7·4 22 17·1 5·0 8·3 Tm 0·40 0·56 0·61 0·77 0·27 0·45 1·06 3·1 2·2 0·67 1·10 Yb 2·6 3·5 3·9 4·8 1·8 3·0 6·7 21 14·0 4·5 6·9 Lu 0·50 0·54 0·71 0·89 0·35 0·50 1·19 3·2 2·1 0·63 0·97 Hf 7·0 10 9·5 7·2 6·8 7·8 7·0 31 28 6·6 5·1 Ta 1·48 1·44 2·4 2·9 2·9 2·1 5·3 2·3 6·4 2·0 2·5 Pb 16·8 27 35 42 31 29 42 27 31 31 17·2 Th 7·7 10·6 15 20·0 17·9 17·1 25 7·3 19·1 12·8 8·5 U 2·5 3·1 4·0 6·0 4·1 4·2 6·4 1·91 5·2 2·2 1·58 Rb/Sr 0·23 0·29 0·94 7·44 7·47 3·02 7·1 0·12 0·58 0·25 0·33 La/Yb 12·4 12·6 15·1 14·5 19·8 13·9 6·2 6·3 13·1 11·7 13·9 Eu* 1·17 1·18 1·05 0·24 0·62 0·93 0·11 0·87 0·47 0·56 0·26 1/Sr 0·004 0·003 0·009 0·054 0·059 0·021 0·041 0·0030 0·0089 0·0055 0·0060 1/Nd 0·034 0·024 0·021 0·018 0·036 0·030 0·026 0·0069 0·0060 0·0060 0·0115 Sample YP648 YR225 YP282 YP105 YP366 YP502 YP133 YP575 YP186 YP227 YP212 Locality 498 351 57 57 465 502 332 361 109 46 363 Host Unit (C) Intr (C) (C) (C) (C) (C) (A) (A) (B) (B) Clast type J5 J6 J8 J8 J9 J9 J7 DM DM Scoria Scoria Suite C1 C1 C2 C2 C2 C2 C3 M1 M1 M1 M2 MS 4 4 3 3 3 3 2 1 1 1 1 SiO2 66·46 68·69 75·78 76·84 77·64 77·23 76·69 53·74 64·25 57·99 53·46 TiO2 1·09 0·86 0·25 0·12 0·13 0·16 0·10 3·13 1·42 1·27 1·19 Al2O3 11·54 11·55 12·77 12·35 12·00 11·72 12·04 14·92 15·02 14·46 14·53 Fe2O3 10·68 9·05 2·80 1·62 1·32 1·98 1·55 13·48 9·20 9·19 13·32 MnO 0·15 0·13 0·05 0·02 0·01 0·06 0·02 0·08 0·21 0·15 0·20 MgO 0·47 0·26 0·05 0·04 0·00 0·02 0·25 1·02 0·55 4·32 4·33 CaO 3·91 3·58 0·81 0·43 0·35 0·35 1·45 5·69 1·69 7·23 6·77 Na2O 2·91 3·02 2·72 2·85 3·55 3·43 3·03 4·00 3·04 3·65 3·87 K2O 2·18 2·60 4·76 5·72 4·97 4·80 4·85 2·13 4·12 1·59 2·19 P2O5 0·38 0·26 0·03 0·01 0·02 0·01 0·02 1·81 0·50 0·15 0·15 Total 100·22 99·78 100·09 99·15 100·28 100·51 100·41 99·39 99·70 99·61 99·71 LOI 1·75 0·62 2·98 3·32 0·26 0·75 4·26 1·64 2·32 0·32 0·93 K/Na 0·84 0·96 1·96 2·25 1·57 1·57 1·79 0·59 1·52 0·49 0·63 Li 35 17·0 40 40 54 36 27 13·5 39 25 56 Sc 39 38 3·2 3 2 2·8 1·2 35 15·1 23 24 V 6·1 5·6 1·0 0·2 0·1 0·5 0·2 108 25 193 163 Cr 7·7 1·8 2·1 1·6 2·6 2·7 1·6 5·7 2·7 59 100 Co 6·3 4·7 0·5 0·7 0·1 0·2 0·3 12·7 11·7 28 16·6 Ni 0·9 0·8 0·5 0·1 0·1 4·6 39 25 Cu 8·7 8·8 15·9 43 2·2 5·7 6·5 17·9 14·1 31 16·2 Zn 103 130 83 70 63 62 48 400 196 192 432 Ga 16·9 20 19·0 18·5 17·8 19·1 21 28 28 23 27 Rb 57 89·8 101 138 126 143 175 39 65 46 55 Sr 245 307 108 18·6 16·9 47 25 334 113 183 166 Y 27 38 39 50 14·5 28 68 184 139 40 67 Zr 310 491 379 210 197 280 158 1689 761 185 61 Nb 17·0 26 28 32 30 28 60 31 47 26 87 Cs 7·9 2·4 3·2 4·8 2·9 3·0 3·6 0·73 1·55 1·02 1·13 Ba 798 952 2559 449 470 1520 98 3571 2539 477 388 La 32 45 59 70 35 41 42 132 182 52 97 Ce 65 92 110 134 54 114 89 157 180 102 140 Pr 7·5 10·8 12·6 15·1 7·6 9·2 10·4 34 42 11·6 23 Nd 29 42 47 55 28 34 39 145 166 45 87 Sm 6·3 8·7 9·7 11·4 5·6 7·0 10·0 33 34 9·2 17·3 Eu 2·4 3·3 3·3 0·89 1·05 2·1 0·38 9·8 5·4 1·78 1·55 Gd 6·3 8·5 9·2 11·3 4·8 6·6 10·5 36 36 10·1 18·5 Tb 1·0 1·25 1·40 1·76 0·74 1·04 2·0 5·9 5·5 1·53 2·8 Dy 5·3 7·1 7·7 9·6 3·7 5·6 11·9 35 30 8·6 15·1 Ho 1·02 1·40 1·50 1·86 0·67 1·09 2·4 7·3 5·9 1·71 2·9 Er 3·0 3·9 4·4 5·7 2·0 3·3 7·4 22 17·1 5·0 8·3 Tm 0·40 0·56 0·61 0·77 0·27 0·45 1·06 3·1 2·2 0·67 1·10 Yb 2·6 3·5 3·9 4·8 1·8 3·0 6·7 21 14·0 4·5 6·9 Lu 0·50 0·54 0·71 0·89 0·35 0·50 1·19 3·2 2·1 0·63 0·97 Hf 7·0 10 9·5 7·2 6·8 7·8 7·0 31 28 6·6 5·1 Ta 1·48 1·44 2·4 2·9 2·9 2·1 5·3 2·3 6·4 2·0 2·5 Pb 16·8 27 35 42 31 29 42 27 31 31 17·2 Th 7·7 10·6 15 20·0 17·9 17·1 25 7·3 19·1 12·8 8·5 U 2·5 3·1 4·0 6·0 4·1 4·2 6·4 1·91 5·2 2·2 1·58 Rb/Sr 0·23 0·29 0·94 7·44 7·47 3·02 7·1 0·12 0·58 0·25 0·33 La/Yb 12·4 12·6 15·1 14·5 19·8 13·9 6·2 6·3 13·1 11·7 13·9 Eu* 1·17 1·18 1·05 0·24 0·62 0·93 0·11 0·87 0·47 0·56 0·26 1/Sr 0·004 0·003 0·009 0·054 0·059 0·021 0·041 0·0030 0·0089 0·0055 0·0060 1/Nd 0·034 0·024 0·021 0·018 0·036 0·030 0·026 0·0069 0·0060 0·0060 0·0115 Major elements and Ba by XRF; trace elements by solution ICP-MS (see Supplementary Material for details). Major element values are recalculated to 100% but with the original totals and LOI figures given. Fe2O3 is total iron as determined by XRF. Full data are given in Supplementary Data Electronic Appendix 3. See Table 1 for definition of clast types, and text for characterization of compositional suites and their host magmatic systems. MS, magmatic system; DM, dense mafic clasts, Intr, intrusive body cutting plastic, still-hot member C ignimbrite. Open in new tab Suite A1 incorporates most of the type J1 pumices and fiamme (see Table 1) found in member A. Suite A1 forms a coherent trend (73·0–77·7 wt % SiO2; Figs 5–7; Supplementary Data Figs S2–S4) that is elevated in alkalis (Na2O + K2O = 8·3–9·3 wt %) and Al2O3 (11·7–13·8 wt %) but depleted in Fe2O3 (1·2–3·0 wt %), P2O5 (0·01–0·04 wt %) and TiO2 (0·09–0·39 wt %) relative to other HRT juvenile materials. Suite A2 is represented by the visually distinctive type J3 clasts with their large intact feldspars, plus a lesser number of clasts which visually resemble the other J1 clasts in having fragmented and/or aggregated feldspar/quartz crystals. Collectively, these clasts are chemically distinctive with 69·2–71·6 wt % SiO2. Suite A2 is enriched in Al2O3 (13·8–14·9 wt %), Fe2O3 (3·8–5·0 wt %) and TiO2 (0·39–0·57 wt %), but slightly depleted in total alkalis (Na2O + K2O = 8·5–9·2 wt %) relative to the least-evolved end of suite A1 (Figs 5–7; Supplementary Data Figs S2–S4), with an intervening compositional gap. Suite A3 covers a narrow range of compositions that is continuous with the most-evolved end of suite A1 on the basis of major elements (e.g. 76·7–77·2 wt % SiO2, 11·9–12·4 wt % Al2O3: Figs 5–7; Supplementary Data Figs S2–S4), but is separated out on the basis of trace elements, particularly Ba (see below). Suite B1 is represented by a single compositional trend in member B. Suite B1 comprises clasts of types J1 (including the ‘B in C’ samples found as pick-ups in member C), J3 and the rhyolitic component in type J4. Suite B1 compositions cover most of the compositional range seen in member A (suites A1 and A2) but with no compositional gap evident (e.g. 70·6–77·4 wt % SiO2, 11·4–13·8 wt % Al2O3, 1·5–4·6 wt % Fe2O3, 7·1–9·1 wt % Na2O + K2O: Figs 5–7; Supplementary Data Figs S2–S4), setting aside two anomalous (slightly silicified) samples (Supplementary Data Electronic Appendix 3). Suite C1: dacite to rhyolite (66–75 wt % SiO2) that is represented by clast types J5 and J6. Suite C1 is notably enriched in CaO (1–4 wt %), Fe2O3 (3·6–11·5 wt %) and P2O5 (0·03–0·39 wt %), and depleted in Al2O3 (11·4–12·1 wt %) relative to compositions from members A and B (Figs 5–7; Supplementary Data Figs S2–S4). At its silica-rich end, this suite overlaps with part of suite C2. Suite C2: high-silica rhyolite (75–78 wt % SiO2) that is represented by clast types J8 and J9, with rare type J5 examples. This suite is moderate in its Fe2O3 (1·3–3·3 wt %) and TiO2 contents (0·12–0·29 wt %: Supplementary Data Figs S2 and S3). Suite C3: another high-silica rhyolite suite (75–77 wt % SiO2), represented by clast type J7. It is further depleted in Fe2O3 (1·6–1·8 wt %) and TiO2 (0·10–0·11 wt %) relative to suites C1 and C2 (Supplementary Data Figs S2 and S3) and is closely similar to suite A3 compositions from member A. Fig. 5. Open in new tabDownload slide Total alkalis–silica plot for samples analysed in this study for the entire HRT (and relevant regional mafic eruptives) and individual members. All major-element data plotted here represent values plotted on a volatile-free basis to total 100%. Symbol shapes relate to the host ignimbrite member from which the clast was sampled. Symbol colours denote the clast type (see Table 1 for details). Silicic suites (A1, B1 etc.) are those defined from compositional groupings (see text for details). Mafic suites 1–3 were defined in Swallow et al. (2018b) and labelled as M1–M3 here for clarity. For the mafic materials, new data presented in this study are shown in larger symbols; data from Swallow et al. (2018b) in smaller symbols. For the full data set presented in these and other bivariate plots, see Supplementary Data Electronic Appendix 3. Classification fields from Le Bas & Streckeisen (1991). Analytical 2 s.d. precisions are smaller than the symbol size. Fig. 5. Open in new tabDownload slide Total alkalis–silica plot for samples analysed in this study for the entire HRT (and relevant regional mafic eruptives) and individual members. All major-element data plotted here represent values plotted on a volatile-free basis to total 100%. Symbol shapes relate to the host ignimbrite member from which the clast was sampled. Symbol colours denote the clast type (see Table 1 for details). Silicic suites (A1, B1 etc.) are those defined from compositional groupings (see text for details). Mafic suites 1–3 were defined in Swallow et al. (2018b) and labelled as M1–M3 here for clarity. For the mafic materials, new data presented in this study are shown in larger symbols; data from Swallow et al. (2018b) in smaller symbols. For the full data set presented in these and other bivariate plots, see Supplementary Data Electronic Appendix 3. Classification fields from Le Bas & Streckeisen (1991). Analytical 2 s.d. precisions are smaller than the symbol size. Fig. 6. Open in new tabDownload slide Al2O3 vs SiO2 for samples analysed in this study for the whole HRT, individual members and relevant regional mafic eruptives (the latter from Swallow et al., 2018b). Analytical 2 s.d. precisions are smaller than the symbol size. Other information as in Fig. 5. Full data set in Supplementary Data Electronic Appendix 3. Fig. 6. Open in new tabDownload slide Al2O3 vs SiO2 for samples analysed in this study for the whole HRT, individual members and relevant regional mafic eruptives (the latter from Swallow et al., 2018b). Analytical 2 s.d. precisions are smaller than the symbol size. Other information as in Fig. 5. Full data set in Supplementary Data Electronic Appendix 3. Fig. 7. Open in new tabDownload slide P2O5 vs SiO2 for samples analysed in this study for the whole HRT, individual members and relevant regional mafic eruptives (the latter from Swallow et al., 2018b). Analytical 2 s.d. precisions are smaller than the symbol size. Other information as in Fig. 5. Full data set in Supplementary Data Electronic Appendix 3. Fig. 7. Open in new tabDownload slide P2O5 vs SiO2 for samples analysed in this study for the whole HRT, individual members and relevant regional mafic eruptives (the latter from Swallow et al., 2018b). Analytical 2 s.d. precisions are smaller than the symbol size. Other information as in Fig. 5. Full data set in Supplementary Data Electronic Appendix 3. Compositional suites A1, A2, B1, and C2 in their major element characteristics trend back towards HRT mafic compositions (Swallow et al., 2018b), whereas suite C1 shows a marked contrast, leading to bifurcations in many of the major element plots. All member C suites are depleted in total alkalis relative to the suites in members A and B (C1 and C3: Na2O + K2O = 7·4–8·6 wt %) with suite C1 particularly so (Na2O + K2O = 5·0–7·1 wt %: Fig. 5). Overall, there are no discernible differences in major element compositions between devitrified and glassy pumice and fiamme, and in particular there are no signs of significant alkali mobility. Although pumices typically have higher loss on ignition (LOI) measurements, K/Na ratios, which would be expected to increase with hydration (Hildreth & Wilson, 2007), broadly overlap (Supplementary DataFig. S5). Higher LOI values (3–5·6 wt %) in some clasts (including mafic clasts) seem to partly reflect minor contamination by a carbonate phase, identified in greater amounts (up to ∼20 wt %) in other clasts (data not presented here) from thin section inspection and testing with dilute acid. Trace elements The trace element characteristics of the HRT silicic samples show extraordinary diversity (Table 2; Supplementary Data Electronic Appendix 3) and link to the compositional suites identified primarily from field observations of pyroclast types and their major element compositions. In particular, Ba abundances measured by XRF in all samples highlight these groupings (Fig. 8). In member A, suite A2 is Ba-rich (2540–4170 ppm) and distinct from suite A1. A low-Ba cluster of analyses distinguishes suite A3 (Ba = 49–300 ppm) from the A1 suite (Ba = 410–2400 ppm). The trace element characteristics of member B samples, as with major elements, define a single suite (B1: Ba = 540–3040 ppm), but highest and lowest Ba abundances comparable to suites A2 and A3, respectively, are absent. In contrast, the three suites in member C based on major element compositions are also delineated in Ba abundances (C1 = 750–1710 ppm; C2 = 450–2840 ppm; C3 = 100–280 ppm). Notably, the last of these is closely comparable to the A3 suite in its low values. Fig. 8. Open in new tabDownload slide Ba (by XRF techniques) vs SiO2 for samples analysed in this study for the whole HRT, individual members and relevant regional mafic eruptives (the latter from Swallow et al., 2018b). Analytical 2 s.d. precisions are smaller than the symbol size. Other information as in Fig. 5. Full data set in Supplementary Data Electronic Appendix 3. Fig. 8. Open in new tabDownload slide Ba (by XRF techniques) vs SiO2 for samples analysed in this study for the whole HRT, individual members and relevant regional mafic eruptives (the latter from Swallow et al., 2018b). Analytical 2 s.d. precisions are smaller than the symbol size. Other information as in Fig. 5. Full data set in Supplementary Data Electronic Appendix 3. Full trace element analyses for selected samples that were chosen on the basis of their XRF Ba concentrations were conducted by solution ICP-MS (Supplementary Data Electronic Appendix 3). Suite A1 clasts show wide compositional variation in many trace elements (e.g. 110–162 ppm Rb, 16–60 ppm Sr, 158–347 ppm Zr, 40–56 ppm Nb: Figs 8–11, Supplementary Data Figs S6–S8). Weak compositional gaps distinguish suites A1 from A2 (69–110 ppm Rb, 76–100 ppm Sr, 509–704 ppm Zr, 40–46 ppm Nb) and A3 (161, 201 ppm Rb, 12 ppm Sr, 155, 158 ppm Zr, 53, 61 ppm Nb). Coherency in major-element compositions within the B1 suite is mirrored in the trace elements, with compositions straddling the A1–A2 divide but not extending to the most extreme compositions present in suites A2 or A3 (overall 59–173 ppm Rb, 15–67 ppm Sr, 149–488 ppm Zr; Fig 8, Supplementary Data Figs S6–S8). Fig. 9. Open in new tabDownload slide Sr vs SiO2 for samples analysed in this study for the whole HRT, individual members and relevant regional mafic eruptives (the latter from Swallow et al., 2018b). Analytical 2 s.d. precisions are smaller than the symbol size. Other information as in Fig. 5. Full data set in Supplementary Data Electronic Appendix 3. Fig. 9. Open in new tabDownload slide Sr vs SiO2 for samples analysed in this study for the whole HRT, individual members and relevant regional mafic eruptives (the latter from Swallow et al., 2018b). Analytical 2 s.d. precisions are smaller than the symbol size. Other information as in Fig. 5. Full data set in Supplementary Data Electronic Appendix 3. Fig. 10. Open in new tabDownload slide Nb vs SiO2 for samples analysed in this study for the whole HRT, individual members and relevant regional mafic eruptives (the latter from Swallow et al., 2018b). Analytical 2 s.d. precisions are smaller than the symbol size. Other information as in Fig. 5. Full data set in Supplementary Data Electronic Appendix 3. Fig. 10. Open in new tabDownload slide Nb vs SiO2 for samples analysed in this study for the whole HRT, individual members and relevant regional mafic eruptives (the latter from Swallow et al., 2018b). Analytical 2 s.d. precisions are smaller than the symbol size. Other information as in Fig. 5. Full data set in Supplementary Data Electronic Appendix 3. Fig. 11. Open in new tabDownload slide Sc vs SiO2 for samples analysed in this study for the whole HRT, individual members and relevant regional mafic eruptives (the latter from Swallow et al., 2018b). Analytical 2 s.d. precisions are smaller than the symbol size. Other information as in Fig. 5. Full data set in Supplementary Data Electronic Appendix 3. Fig. 11. Open in new tabDownload slide Sc vs SiO2 for samples analysed in this study for the whole HRT, individual members and relevant regional mafic eruptives (the latter from Swallow et al., 2018b). Analytical 2 s.d. precisions are smaller than the symbol size. Other information as in Fig. 5. Full data set in Supplementary Data Electronic Appendix 3. Member C suites are distinct in other trace elements both from each other and (with one exception) from the suites in members A and B. Suite C1 is characterised by overall higher Sc (4·7– 46 ppm) and Sr (182–321 ppm), lower Ba (750–1710 ppm) and comparable Zr (310–890 ppm) concentrations with respect to the maximum concentrations of members A and B suites (Figs 8, 9, and 11, Supplementary Data Fig. S7), with the latter two positively correlated with SiO2. Suite C2 exhibits large ranges in trace elements over small ranges in SiO2. This suite is characterised by Ba (470–2380 ppm), Sr (17–120 ppm) and Zr (200–360 ppm) values that are negatively correlated with SiO2 (Figs 8 and 9, Supplementary Data Fig. S7). Both suites C1 and C2 are distinctly lower in Nb (17–33 ppm: Fig. 10) relative to the suites from members A and B. Suite C3 is, however, comparable to suite A3 in most elements (e.g. 100–280 ppm Ba, 5–42 ppm Sr, 59–66 ppm Nb, 166–189 ppm Zr: Figs 8–10; Supplementary DataFig. S7). Chondrite-normalised (McDonough & Sun, 1995) rare-earth element (REE) diagrams for materials from members A, B and C are shown in Fig. 12. Juvenile clasts from suites A1-3 and B1 in members A and B have broadly similar patterns, with moderate light-REE enrichments (La/Yb = 12–21; Fig. 12). Distinctions between suites are observed in Eu anomalies (where Eu* = EuN / [SmN x GdN]0·5) in member A suites (A1 = 0·25–0·56; A2 = 0·94; A3 = 0·12–0·18); but Eu* has a narrower but continuous range in suite B1 (0·26–0·90). Suite C1 samples have flatter REE patterns than the suites in members A and B (La/Yb = 11–15) with small, both positive and negative, Eu anomalies (Eu* = 0·80–1·18). Suite C2 has similarly flat trends (La/Yb = 11–20) but more variable Eu anomalies (Eu* = 0·24–1·25). In contrast, suite C3 has flatter REE trends (La/Yb = 6–10) relative to other member C suites and its suite A3 comparator, but a strong negative Eu anomaly (Eu* = 0·11–0·16), comparable to that in suite A3. Fig. 12. Open in new tabDownload slide C1 chondrite-normalised rare earth-element (REE) plots for samples from members A, B and C of the HRT. Compositional fields (grey and pink shaded regions) of mafic suites M1–M3 are from Swallow et al. (2018b), plus new data from this study (see Supplementary Data Electronic Appendix 3). Normalising values from McDonough & Sun (1995). Fig. 12. Open in new tabDownload slide C1 chondrite-normalised rare earth-element (REE) plots for samples from members A, B and C of the HRT. Compositional fields (grey and pink shaded regions) of mafic suites M1–M3 are from Swallow et al. (2018b), plus new data from this study (see Supplementary Data Electronic Appendix 3). Normalising values from McDonough & Sun (1995). New data from juvenile mafic components We also present new data from mafic materials in the HRT beyond that given in Swallow et al. (2018b; Supplementary Data Electronic Appendix 3). The three materials represented by these data fall into somewhat different compositional areas. Dense to poorly vesicular mafic clasts from member A scatter over a wider field than previously reported, although with some remaining coherent trends (e.g. P2O5, Fe2O3, TiO2, CaO, Nb, Sc; Figs 7, 10 and 11, Supplementary Data Figs S2–S4). Scoria enclaves from type J4 pumices at locations south of the caldera plot with the other examples shown by Swallow et al. (2018b). However, the scoria from upper member B sampled north of the caldera plots along trends for some elements that contrast sharply with the other mafic clast data. In particular, P2O5, Fe2O3, TiO2, Ba, Sr, and Zr have lower abundances and form coherent trends for some clasts that generally align with compositions in the olivine tholeiite field (Figs 5, 7–9; Supplementary Data Figs S2–S4) defined from lavas erupted before and after the HRT (Swallow et al., 2018b). Juvenile materials with any kind of tholeiitic affinity had not previously been recognised to be co-erupted with the HRT. The more silicic of the HRT mafic compositions overlap in SiO2 contents with the melt inclusion compositions reported by Szymanowski et al. (2015), but have sharply contrasting characteristics with respects to other major and trace elements that preclude direct comparisons. Isotopic ratios Single-clast, age-corrected hereafter (to 2·08 Ma) values of 87Sr/86Sr in clasts from members A and B are clustered (0·70958–0·71256: Table 3; Supplementary Data Electronic Appendix 3) with suite A2 clasts falling at the less radiogenic end of the spectrum (0·70958–0·71009: Fig. 13). 87Sr/86Sr ratios show correlations with indicators of melt evolution, e.g. positive with Rb/Sr and negative with Ba, with the trend levelling off at higher Rb/Sr and lower Ba values (Fig. 14). Suite C3 clasts are less radiogenic than the suites in members A and B for a given elemental value (e.g. 1/Sr) and exhibit a flat trend with indicators of magma evolution (e.g. Rb/Sr). In sharp contrast, all other member C juvenile components are significantly more radiogenic, as first recognised by Doe et al. (1982). 87Sr/86Sr values range from 0·72465–0·73026 in clasts ranging from dacite to high-silica rhyolite (Figs 13, 14). These samples show a weak positive correlation between isotopic ratios and Rb/Sr and 1/Sr, but similar isotopic ratios across significantly different Ba abundances. Fig. 13. Open in new tabDownload slide 143Nd/144Nd vs 87Sr/86Sr (where 87Sr/86Sr is age-corrected to an eruption age of 2·08 Ma) for samples from the HRT and relevant regional mafic eruptives. Field of Archean crust from McCurry & Rodgers (2009). εNd calculated from DePaolo & Wasserburg (1976: see text for details). Individual 2 s.e. analytical uncertainties are smaller than symbol size unless shown. Full data set in Supplementary Data Electronic Appendix 3. Fig. 13. Open in new tabDownload slide 143Nd/144Nd vs 87Sr/86Sr (where 87Sr/86Sr is age-corrected to an eruption age of 2·08 Ma) for samples from the HRT and relevant regional mafic eruptives. Field of Archean crust from McCurry & Rodgers (2009). εNd calculated from DePaolo & Wasserburg (1976: see text for details). Individual 2 s.e. analytical uncertainties are smaller than symbol size unless shown. Full data set in Supplementary Data Electronic Appendix 3. Fig. 14. Open in new tabDownload slide 87Sr/86Sr vs indices of evolution for samples from the HRT and relevant regional eruptives. See Fig. 5 for symbol and suite details. Individual 2 s.e. analytical uncertainties are smaller than symbol size unless shown. Full data set in Supplementary Data Electronic Appendix 3. Fig. 14. Open in new tabDownload slide 87Sr/86Sr vs indices of evolution for samples from the HRT and relevant regional eruptives. See Fig. 5 for symbol and suite details. Individual 2 s.e. analytical uncertainties are smaller than symbol size unless shown. Full data set in Supplementary Data Electronic Appendix 3. Table 3: Representative single clast isotopic ratios for silicic juvenile components from members A, B and C in the HRT. Sample Host Clast Suite SiO2 Rb Sr 87Sr/ 2se 143Nd/ 2se 206Pb/ 2se 207Pb/ 2se 208Pb/ 2se unit type wt% ppm ppm 86Sr 144Nd 204Pb 204Pb 204Pb YP459 (A) J1 A1 77·30 124 21 0·71213 3·6E-06 0·51215 6·6E-06 17·01 8·0E-04 15·49 8·8E-04 38·06 2·6E-03 YP457 (A) J1 A1 74·10 110 38 0·71163 3·6E-06 0·51216 4·5E-06 17·01 8·5E-04 15·49 9·9E-04 38·05 3·0E-03 YP255 (A) J1 A3 77·00 161 12·5 0·71256 6·0E-06 0·51214 2·3E-06 17·02 9·6E-04 15·50 1·1E-03 38·06 3·4E-03 YP251 (A) J2 A1 75·01 119 34 0·71188 3·5E-06 0·51216 2·9E-06 16·99 8·2E-04 15·49 9·2E-04 38·05 2·7E-03 YP191 (A) J3 A2 69·87 83 76 0·71009 3·9E-06 0·51220 2·5E-06 16·95 1·0E-03 15·48 1·3E-03 38·02 3·9E-03 YP078 (B) J1 B1 76·27 126 35 0·71179 3·8E-06 0·51215 2·7E-06 16·99 2·2E-03 15·49 3·0E-03 38·03 9·6E-03 YP208 (B) J1 B1 72·23 98 54 0·71091 4·1E-06 0·51219 3·6E-06 16·98 1·1E-03 15·49 1·3E-03 38·03 3·9E-03 YP061 (B) J4 B1 71·43 106 67 0·71076 2·6E-06 16·90 5·4E-03 15·46 5·2E-03 37·95 1·4E-02 YP299 (C) J5 C1 69·39 46 228 0·72486 3·5E-06 0·51172 2·7E-06 18·16 1·3E-03 15·76 1·5E-03 37·88 4·7E-03 YP648 (C) J5 C1 66·46 57 245 0·72464 3·4E-06 0·51174 8·0E-06 18·16 8·8E-04 15·75 9·2E-04 37·92 2·6E-03 YP282 (C) J8 C2 75·78 108 39 0·72590 3·9E-06 0·51172 2·6E-06 17·91 1·1E-03 15·70 1·4E-03 37·90 4·2E-03 YP105 (C) J8 C2 76·84 138 18·6 0·72735 4·1E-06 0·51170 2·8E-06 17·89 1·5E-03 15·69 1·8E-03 37·89 5·7E-03 YP133 (C) J7 C3 76·69 175 25 0·71102 3·7E-06 0·51214 2·6E-06 16·99 1·4E-03 15·48 1·8E-03 38·03 5·8E-03 Sample Host Clast Suite SiO2 Rb Sr 87Sr/ 2se 143Nd/ 2se 206Pb/ 2se 207Pb/ 2se 208Pb/ 2se unit type wt% ppm ppm 86Sr 144Nd 204Pb 204Pb 204Pb YP459 (A) J1 A1 77·30 124 21 0·71213 3·6E-06 0·51215 6·6E-06 17·01 8·0E-04 15·49 8·8E-04 38·06 2·6E-03 YP457 (A) J1 A1 74·10 110 38 0·71163 3·6E-06 0·51216 4·5E-06 17·01 8·5E-04 15·49 9·9E-04 38·05 3·0E-03 YP255 (A) J1 A3 77·00 161 12·5 0·71256 6·0E-06 0·51214 2·3E-06 17·02 9·6E-04 15·50 1·1E-03 38·06 3·4E-03 YP251 (A) J2 A1 75·01 119 34 0·71188 3·5E-06 0·51216 2·9E-06 16·99 8·2E-04 15·49 9·2E-04 38·05 2·7E-03 YP191 (A) J3 A2 69·87 83 76 0·71009 3·9E-06 0·51220 2·5E-06 16·95 1·0E-03 15·48 1·3E-03 38·02 3·9E-03 YP078 (B) J1 B1 76·27 126 35 0·71179 3·8E-06 0·51215 2·7E-06 16·99 2·2E-03 15·49 3·0E-03 38·03 9·6E-03 YP208 (B) J1 B1 72·23 98 54 0·71091 4·1E-06 0·51219 3·6E-06 16·98 1·1E-03 15·49 1·3E-03 38·03 3·9E-03 YP061 (B) J4 B1 71·43 106 67 0·71076 2·6E-06 16·90 5·4E-03 15·46 5·2E-03 37·95 1·4E-02 YP299 (C) J5 C1 69·39 46 228 0·72486 3·5E-06 0·51172 2·7E-06 18·16 1·3E-03 15·76 1·5E-03 37·88 4·7E-03 YP648 (C) J5 C1 66·46 57 245 0·72464 3·4E-06 0·51174 8·0E-06 18·16 8·8E-04 15·75 9·2E-04 37·92 2·6E-03 YP282 (C) J8 C2 75·78 108 39 0·72590 3·9E-06 0·51172 2·6E-06 17·91 1·1E-03 15·70 1·4E-03 37·90 4·2E-03 YP105 (C) J8 C2 76·84 138 18·6 0·72735 4·1E-06 0·51170 2·8E-06 17·89 1·5E-03 15·69 1·8E-03 37·89 5·7E-03 YP133 (C) J7 C3 76·69 175 25 0·71102 3·7E-06 0·51214 2·6E-06 16·99 1·4E-03 15·48 1·8E-03 38·03 5·8E-03 For analytical details, see Supplementary Material, and for full data set see Supplementary Data Electronic Appendix 3. 87Sr/86Sr values are age-corrected to 2·08 Ma, using measured parent-daughter rations (EA 3). Open in new tab Table 3: Representative single clast isotopic ratios for silicic juvenile components from members A, B and C in the HRT. Sample Host Clast Suite SiO2 Rb Sr 87Sr/ 2se 143Nd/ 2se 206Pb/ 2se 207Pb/ 2se 208Pb/ 2se unit type wt% ppm ppm 86Sr 144Nd 204Pb 204Pb 204Pb YP459 (A) J1 A1 77·30 124 21 0·71213 3·6E-06 0·51215 6·6E-06 17·01 8·0E-04 15·49 8·8E-04 38·06 2·6E-03 YP457 (A) J1 A1 74·10 110 38 0·71163 3·6E-06 0·51216 4·5E-06 17·01 8·5E-04 15·49 9·9E-04 38·05 3·0E-03 YP255 (A) J1 A3 77·00 161 12·5 0·71256 6·0E-06 0·51214 2·3E-06 17·02 9·6E-04 15·50 1·1E-03 38·06 3·4E-03 YP251 (A) J2 A1 75·01 119 34 0·71188 3·5E-06 0·51216 2·9E-06 16·99 8·2E-04 15·49 9·2E-04 38·05 2·7E-03 YP191 (A) J3 A2 69·87 83 76 0·71009 3·9E-06 0·51220 2·5E-06 16·95 1·0E-03 15·48 1·3E-03 38·02 3·9E-03 YP078 (B) J1 B1 76·27 126 35 0·71179 3·8E-06 0·51215 2·7E-06 16·99 2·2E-03 15·49 3·0E-03 38·03 9·6E-03 YP208 (B) J1 B1 72·23 98 54 0·71091 4·1E-06 0·51219 3·6E-06 16·98 1·1E-03 15·49 1·3E-03 38·03 3·9E-03 YP061 (B) J4 B1 71·43 106 67 0·71076 2·6E-06 16·90 5·4E-03 15·46 5·2E-03 37·95 1·4E-02 YP299 (C) J5 C1 69·39 46 228 0·72486 3·5E-06 0·51172 2·7E-06 18·16 1·3E-03 15·76 1·5E-03 37·88 4·7E-03 YP648 (C) J5 C1 66·46 57 245 0·72464 3·4E-06 0·51174 8·0E-06 18·16 8·8E-04 15·75 9·2E-04 37·92 2·6E-03 YP282 (C) J8 C2 75·78 108 39 0·72590 3·9E-06 0·51172 2·6E-06 17·91 1·1E-03 15·70 1·4E-03 37·90 4·2E-03 YP105 (C) J8 C2 76·84 138 18·6 0·72735 4·1E-06 0·51170 2·8E-06 17·89 1·5E-03 15·69 1·8E-03 37·89 5·7E-03 YP133 (C) J7 C3 76·69 175 25 0·71102 3·7E-06 0·51214 2·6E-06 16·99 1·4E-03 15·48 1·8E-03 38·03 5·8E-03 Sample Host Clast Suite SiO2 Rb Sr 87Sr/ 2se 143Nd/ 2se 206Pb/ 2se 207Pb/ 2se 208Pb/ 2se unit type wt% ppm ppm 86Sr 144Nd 204Pb 204Pb 204Pb YP459 (A) J1 A1 77·30 124 21 0·71213 3·6E-06 0·51215 6·6E-06 17·01 8·0E-04 15·49 8·8E-04 38·06 2·6E-03 YP457 (A) J1 A1 74·10 110 38 0·71163 3·6E-06 0·51216 4·5E-06 17·01 8·5E-04 15·49 9·9E-04 38·05 3·0E-03 YP255 (A) J1 A3 77·00 161 12·5 0·71256 6·0E-06 0·51214 2·3E-06 17·02 9·6E-04 15·50 1·1E-03 38·06 3·4E-03 YP251 (A) J2 A1 75·01 119 34 0·71188 3·5E-06 0·51216 2·9E-06 16·99 8·2E-04 15·49 9·2E-04 38·05 2·7E-03 YP191 (A) J3 A2 69·87 83 76 0·71009 3·9E-06 0·51220 2·5E-06 16·95 1·0E-03 15·48 1·3E-03 38·02 3·9E-03 YP078 (B) J1 B1 76·27 126 35 0·71179 3·8E-06 0·51215 2·7E-06 16·99 2·2E-03 15·49 3·0E-03 38·03 9·6E-03 YP208 (B) J1 B1 72·23 98 54 0·71091 4·1E-06 0·51219 3·6E-06 16·98 1·1E-03 15·49 1·3E-03 38·03 3·9E-03 YP061 (B) J4 B1 71·43 106 67 0·71076 2·6E-06 16·90 5·4E-03 15·46 5·2E-03 37·95 1·4E-02 YP299 (C) J5 C1 69·39 46 228 0·72486 3·5E-06 0·51172 2·7E-06 18·16 1·3E-03 15·76 1·5E-03 37·88 4·7E-03 YP648 (C) J5 C1 66·46 57 245 0·72464 3·4E-06 0·51174 8·0E-06 18·16 8·8E-04 15·75 9·2E-04 37·92 2·6E-03 YP282 (C) J8 C2 75·78 108 39 0·72590 3·9E-06 0·51172 2·6E-06 17·91 1·1E-03 15·70 1·4E-03 37·90 4·2E-03 YP105 (C) J8 C2 76·84 138 18·6 0·72735 4·1E-06 0·51170 2·8E-06 17·89 1·5E-03 15·69 1·8E-03 37·89 5·7E-03 YP133 (C) J7 C3 76·69 175 25 0·71102 3·7E-06 0·51214 2·6E-06 16·99 1·4E-03 15·48 1·8E-03 38·03 5·8E-03 For analytical details, see Supplementary Material, and for full data set see Supplementary Data Electronic Appendix 3. 87Sr/86Sr values are age-corrected to 2·08 Ma, using measured parent-daughter rations (EA 3). Open in new tab Similar relationships are observed in the Nd isotopic compositions measured on the same dissolutions of the same samples. Samples from members A and B have 143Nd/144Nd ratios from 0·51214–0·51220 (εNd = −8·5 to −9·6), that are negatively correlated with 87Sr/86Sr (Fig. 13). Suite C3 sample YP133 has 143Nd/144Nd = 0·51214 ± 0·000001 (2 se; εNd = -9·5), identical to suite A3 sample YP255. Other juvenile samples from suites C1 and C2 in member C are less radiogenic: 143Nd/144Nd ratios range from 0·51174–0·51170 (εNd = −17·3 to −18·2) and are negatively correlated with 87Sr/86Sr (Fig. 13). There is no variation in Nd isotopic compositions with any elemental measures of magmatic evolution (e.g. Ba: Supplementary DataFig. S9). Pb isotopic compositions of members A and B samples are tightly clustered (Fig. 15), with positive correlations observed between 206Pb/204Pb (16·90–17·02), 207Pb/204Pb (15·46–15·50) and 208Pb/204Pb (37·95–38·06). Values from suite C3 pumices are tightly clustered and comparable those from members A and B (206Pb/204Pb = 16·99–17·00, 207Pb/204Pb = 15·35–15·49 and 208Pb/204Pb = 38·03–38·06). Other samples from suites C1 and C2 in member C show distinctly different Pb isotopic signatures, with more radiogenic 206Pb/204Pb (17·89–18·16) and 207Pb/204Pb (15·69–15·78), which are positively correlated and fall along a linear extension of the A and B trend. However, they have less radiogenic 208Pb/204Pb (37·87–37·92), which is negatively correlated with 206Pb/204Pb and 207Pb/204Pb (Fig. 15). Fig. 15. Open in new tabDownload slide Pb isotopic compositions for samples from the HRT and relevant regional eruptives. Insets show enlarged fields of data from all suites from members A and B plus suite C3. Individual 2 s.e. analytical uncertainties are smaller than symbol size unless shown. Full data set in Supplementary Data Electronic Appendix 3. Fig. 15. Open in new tabDownload slide Pb isotopic compositions for samples from the HRT and relevant regional eruptives. Insets show enlarged fields of data from all suites from members A and B plus suite C3. Individual 2 s.e. analytical uncertainties are smaller than symbol size unless shown. Full data set in Supplementary Data Electronic Appendix 3. Groundmass glass chemistry Groundmass glass compositions (by EPMA and LA-ICP-MS on separated shards from crushed pumices) in members A and B are dominantly high-silica rhyolite (75–78 wt % SiO2: Supplementary Data Electronic Appendix 4) with extensions to low-silica rhyolitic compositions observed in glass from types J3 (mostly 72–74 wt % SiO2) and J4 clasts (71–77 wt % SiO2). Glass from members A and B is alkali-rich (Na2O + K2O = 7·6–9·4 wt %), with glass from type J2 samples clustered to higher alkalis (8·5–9·3 wt %) and lower CaO (0·24–0·49 wt %) concentrations for a given value of SiO2 (Supplementary Data Figs S10–S12). Low-CaO and elevated alkali contents in type J2 sample groundmass glass, which is darkened by Fe–Ti oxide crystallization, indicates that some alkali mobility may have occurred, and these compositions are set aside. Types J3 and J4 samples extend to higher CaO values (up to 2·2 and 1·5 wt %, respectively). The negative correlations between Al2O3, alkalis and CaO vs SiO2 are consistent with a dominant role for feldspar crystallization on the glass compositions (Supplementary Data Figs S10–S12). Groundmass glasses from members A and B form an overall continuum in incompatible trace elements (e.g. 3·7–10·6 ppm U, 100–260 ppm Rb; Supplementary Data Electronic Appendix 4). In detail, however, glasses from suite A1 samples are clustered in composition for some compatible elements (110–730 ppm Ba, 80–140 ppm La, Ba/Rb = 0·49–4·2, Sr/Rb = 0·01–0·1 excluding one outlier; Figs 16, 17). Glasses from suite A2 samples are distinctly Ba-richer (1310–3360 ppm Ba, 90–110 ppm La, Ba/Rb = 9·2–24, Sr/Rb = 0·22–0·47) and from suite A3 samples Ba-poorer (31–62 ppm Ba, 49–81 ppm La, Ba/Rb = 1·3–0·24, Sr/Rb = 0·01–0·04, excluding two outliers). The clustered nature of the glass compositions parallels that observed in the initial fall deposits and basal ignimbrite (Myers et al., 2016; Swallow et al., 2018a), with close similarities in the cluster averages between the fall deposit/basal ignimbrite and groundmass glasses (Figs 16 and 17). Groundmass glass fragments derived by crushing from single clasts in member A are compositionally homogeneous within themselves, with only limited compositional variation within single pumices, particularly when compared to the overall range observed (Supplementary DataFig. S13). Glass compositions from type J1 clasts in member B focus at the less-evolved end (e.g. 120–1680 ppm Ba, 62–140 ppm La) of the compositional spectrum shown by suite A1 glass data. Rhyolitic glass from scoria-bearing (type J4) clasts covers the entire glass compositional range of J1 samples in member B (≥180 ppm Ba), but also defines an antithetic trend, extending from the highest-Ba J1 pumice glass to higher Ba and lower La compositions (3100 ppm Ba, 83 ppm La; Fig. 16), similar to the type J3 clast glass in member A. Fig. 16. Open in new tabDownload slide Ba vs La concentrations of groundmass glasses from pumice from members A, B and C of the HRT. Large crosses show the means of each compositional cluster in the fall deposit (fall glass cluster: FGC; obsidian [pyroclast] cluster: OGC) and basal ignimbrite (ignimbrite glass cluster: IGC; from Swallow et al., 2018a). Black crosses in member A plot represent 2 s.d. analytical uncertainties at the plotted concentrations (400, 50 and 50, 120 ppm for Ba and La, respectively) derived through error slope regression (from Swallow et al., 2018a). Full data set in Supplementary Data Electronic Appendix 4. Fig. 16. Open in new tabDownload slide Ba vs La concentrations of groundmass glasses from pumice from members A, B and C of the HRT. Large crosses show the means of each compositional cluster in the fall deposit (fall glass cluster: FGC; obsidian [pyroclast] cluster: OGC) and basal ignimbrite (ignimbrite glass cluster: IGC; from Swallow et al., 2018a). Black crosses in member A plot represent 2 s.d. analytical uncertainties at the plotted concentrations (400, 50 and 50, 120 ppm for Ba and La, respectively) derived through error slope regression (from Swallow et al., 2018a). Full data set in Supplementary Data Electronic Appendix 4. Fig. 17. Open in new tabDownload slide Ba/Rb vs Sr/Rb for groundmass glasses from pumice from members A, B and C of the HRT. Large crosses show the means of each compositional cluster in the fall deposit (fall glass cluster: FGC; obsidian [pyroclast] cluster: OGC) and basal ignimbrite (ignimbrite glass cluster: IGC: from Swallow et al., 2018a). Black crosses in member A plot represent 2 s.d. analytical uncertainties at the plotted concentration derived through error slope regression (as in Swallow et al., 2018a). Full data set in Supplementary Data Electronic Appendix 4. Fig. 17. Open in new tabDownload slide Ba/Rb vs Sr/Rb for groundmass glasses from pumice from members A, B and C of the HRT. Large crosses show the means of each compositional cluster in the fall deposit (fall glass cluster: FGC; obsidian [pyroclast] cluster: OGC) and basal ignimbrite (ignimbrite glass cluster: IGC: from Swallow et al., 2018a). Black crosses in member A plot represent 2 s.d. analytical uncertainties at the plotted concentration derived through error slope regression (as in Swallow et al., 2018a). Full data set in Supplementary Data Electronic Appendix 4. Groundmass glasses, where available from member C clasts (pumice types J7 and J8), are all high-silica rhyolite (76–78 wt % SiO2). No glass could be recovered from suite C1 clasts. There are no groundmass glasses found in any member C clasts that plot along the trend defined by suites A1 or B1 clasts. Glasses from type J7 clasts are highly evolved, with low Ba (18–120 ppm), La (45–64 ppm) and Ba/Rb and Sr/Rb ratios (0·07–0·54 and 0·01–0·04, respectively; Figs 16 and 17), i.e. similar values to glasses from suite A3 samples. Glasses from type J8 clasts are compositionally unique, however, with elevated, but clustered, Ba compositions (360–2390 ppm) for a given, but relatively uniform, La content (63–89 ppm) when compared to glasses from pumices in members A and B. The type J8 glasses are also poorer in U (5·1–8·5 ppm) for a given value of Rb (140–180 ppm) and plot below the Ba/Rb–Sr/Rb curve (Ba/Rb = 2·2–16·0 and Sr/Rb = 0·1–1·0) relative to the trend in members A and B (Fig. 17). They are also characteristically depleted in Nb (28–36 ppm) and Th (17–30 ppm) relative to glasses from clasts in members A and B (Supplementary Data Electronic Appendix 4). MINERAL CHEMISTRY Feldspar Potassium-rich feldspars in members A and B are dominantly sanidine, but with minor anorthoclase, and commonly contain a significant celsian (Cn: Ba-feldspar) component (An1–10Ab36–65Or29–59Cn0–11; Figs 18 and 19; Supplementary Data Electronic Appendix 5). Rim compositions have a more restricted range, with lower An and Cn values (An1–7Ab40–65Or29–58Cn0–6) with overgrowths (e.g. anti-rapakivi feldspars) being more Cn rich (1·7–7·5 mol %; Fig. 19). Sanidine crystals from suite A3 clasts lie at the potassic end of the range (An1–2Ab39–44Or53–59Cn0–1). Potassium-rich feldspars in member C form a similar trend to members A and B, but isolated crystals extend to higher Or values (An0–10Ab21–59Or29–78Cn0–7). Sanidine crystals from suite C3 clasts are typically more potassic and have lower Cn contents (An1–4Ab39–46Or49–59Cn0–2) than those from suite C2 clasts (An1–9Ab40–59Or29–56Cn0–7; Fig. 19). Fig. 18. Open in new tabDownload slide Ternary diagrams showing feldspar compositions for crystals from members A, B and C of the HRT. (a) All analyses from members A and B samples. (b) All analyses from member C samples. Details of clast types are in Table 1. Compositional fields are labelled with the classification of Deer et al. (1966). Fig. 18. Open in new tabDownload slide Ternary diagrams showing feldspar compositions for crystals from members A, B and C of the HRT. (a) All analyses from members A and B samples. (b) All analyses from member C samples. Details of clast types are in Table 1. Compositional fields are labelled with the classification of Deer et al. (1966). Fig. 19. Open in new tabDownload slide Mol % Cn vs Or contents of sanidine rim compositions (including overgrowths) in samples from HRT members A, B and C. Compositional fields of interior zones (cores and intermediate zones) are shown for comparison. Histogram bin widths are 0·1 mol % Cn and red lines are pdf curves derived from Isoplot (Ludwig, 2008) using a 2 s.d. error of ±0·3 mol % Cn. Cross in the top right indicates 2 s.d. analytical precision. Full data set in Supplementary Data Electronic Appendix 5. Fig. 19. Open in new tabDownload slide Mol % Cn vs Or contents of sanidine rim compositions (including overgrowths) in samples from HRT members A, B and C. Compositional fields of interior zones (cores and intermediate zones) are shown for comparison. Histogram bin widths are 0·1 mol % Cn and red lines are pdf curves derived from Isoplot (Ludwig, 2008) using a 2 s.d. error of ±0·3 mol % Cn. Cross in the top right indicates 2 s.d. analytical precision. Full data set in Supplementary Data Electronic Appendix 5. Plagioclase feldspars from members A and B are mostly andesine to oligoclase, but form a continuum to anorthoclase (An11–46Ab50–76Or3–28Cn0–4; Fig. 18). The more potassic compositions are dominated by interior domains and ‘plutonic-textured’ crystals (see below). Grains from suite A3 samples plot at the albite-richer end of the trend (An15–20Ab71–76Or8–11Cn0). Plagioclase compositions in member C are more clustered than in members A and B and are grouped by suite: (a) a trend extending to higher An contents (An33–52Ab46–61Or2–5Cn0) dominantly defined by grains in suite C1; (b) an oligoclase-centred cluster (An14–30Ab62–77Or4–15Cn0–2) with grains from suite C3 (An14–22Ab68–75Or8–12Cn0–1) at the albite-richer end of the cluster; and (c) an anorthoclase cluster (An10–15Ab60–66Or17–27Cn1–4) dominated by grains from suite C2. Texturally, 80% of the sanidines from suite A3 clasts are homogenous (Supplementary Data Electronic Appendix 5) whereas sanidines in clasts from suites A1 and B1 are more commonly zoned in backscattered electron images (57% and 70% of crystals, respectively; Fig. 20) but with no coherent sense of zoning. Suite A2 samples exhibit equal numbers of homogeneous and zoned (to lower Or) sanidines, but also contain crystals with high-Ba overgrowths, represented as either anti-rapakivi textures or sanidine rims (7% of feldspar crystals). Anti-rapakivi textures are also sparsely present in member B samples (2% of feldspar crystals; Fig. 20). Sanidines from member C show approximately equal proportions of zoned and homogenous crystals irrespective of the compositional suite. In contrast to the sanidine record, plagioclase is dominantly homogeneous in all compositional suites from the entire HRT (>80% of crystals in each suite). Fig. 20. Open in new tabDownload slide Open in new tabDownload slide Backscattered electron images of representative feldspar textures found in the HRT members and juvenile clast types (Table 1), annotated with compositional data from electron microprobe analyses (Supplementary Data Electronic Appendix 5). (a) Homogenous sanidine crystal from YP416 (member A, type J1), typical of suite A3 and C3 sanidines. (b) Plagioclase and sanidine crystals from YP023 (member B, type J1) showing the contrast between homogeneous plagioclase (found in all suites) and zoned sanidines characteristic of suites A1 and B1. (c) Sanidines from YP191 (member A, type J3) showing textural diversity in suite A2 samples. High Ba rims on sanidines are present, as are homogenous crystals. (d) Ba-rich rim characteristic of sanidines from suite C2 (YP282, type J8). (e), (f) Anti-rapakivi feldspars present in suites A2 (YP191, type J3) and B1 (YP056, type J4), respectively, with a plagioclase interior and Ba-rich sanidine overgrowth. (g), (h) Bright, high-Ba rims found in suite B1 type J4 scoria-bearing pumices (YP056 and YP061, respectively). (i), (j) Carbonate-cemented sanidine (YP307: type J1) and plagioclase (YP414: type J1) crystals in member A. (k), (l) ‘Plutonic-textured’ feldspars present in type J4 (YP071: member B) and J8 clasts (YP282: member C) marked by feldspar intergrowths and dissolution textures (e.g. corroded cores). Fig. 20. Open in new tabDownload slide Open in new tabDownload slide Backscattered electron images of representative feldspar textures found in the HRT members and juvenile clast types (Table 1), annotated with compositional data from electron microprobe analyses (Supplementary Data Electronic Appendix 5). (a) Homogenous sanidine crystal from YP416 (member A, type J1), typical of suite A3 and C3 sanidines. (b) Plagioclase and sanidine crystals from YP023 (member B, type J1) showing the contrast between homogeneous plagioclase (found in all suites) and zoned sanidines characteristic of suites A1 and B1. (c) Sanidines from YP191 (member A, type J3) showing textural diversity in suite A2 samples. High Ba rims on sanidines are present, as are homogenous crystals. (d) Ba-rich rim characteristic of sanidines from suite C2 (YP282, type J8). (e), (f) Anti-rapakivi feldspars present in suites A2 (YP191, type J3) and B1 (YP056, type J4), respectively, with a plagioclase interior and Ba-rich sanidine overgrowth. (g), (h) Bright, high-Ba rims found in suite B1 type J4 scoria-bearing pumices (YP056 and YP061, respectively). (i), (j) Carbonate-cemented sanidine (YP307: type J1) and plagioclase (YP414: type J1) crystals in member A. (k), (l) ‘Plutonic-textured’ feldspars present in type J4 (YP071: member B) and J8 clasts (YP282: member C) marked by feldspar intergrowths and dissolution textures (e.g. corroded cores). Feldspar crystals cemented or coated in a carbonate matrix are present in sparse abundance (3% of all feldspar crystals) in suite A1 samples. ‘Plutonic-textured’ feldspars, characterised by feldspar intergrowths and dissolution textures (Parsons, 1978; Fig. 20), are present in all suites. They typically comprise <10% of feldspar crystals, with the exception of scoria-bearing, type J4 pumices in member B (17%). Pyroxene Pyroxene compositions in members A and B show bimodality in their interior compositions (Fig. 21). Low-Ca pyroxenes (hereafter pigeonite for simplicity, despite extending to lower Ca than the pigeonite field of Poldervaart & Hess, 1951) are present as cores, inclusions or intermediate zones (En13–72Fs25–78Wo0–12) in ∼9% of crystals. Alternatively, interior domains of clinopyroxene composition (i.e. augite) occur, spanning a wide range (En4–40Fs22–55Wo34–43). In contrast, rims in members A and B are Fe-rich augite and are tightly clustered (En6–17Fs44–53Wo38–44) with no discernible difference between clast compositional types or eruptive suites. Fig. 21. Open in new tabDownload slide Pyroxene quadrilaterals for crystals from members A, B and C of the HRT. (a) All pyroxene analyses from members A and B. (b) Pyroxene rim compositions from member A and B samples. Compositional fields (grey shaded regions) copied across from panel (a) for comparison. (c) All pyroxene analyses from member C. (d) Pyroxene rim compositions from member C. Compositional fields (grey shaded regions) copied across from panel (c) for comparison. Compositional labels from Poldervaart & Hess (1951). Full data set in Supplementary Data Electronic Appendix 5. Fig. 21. Open in new tabDownload slide Pyroxene quadrilaterals for crystals from members A, B and C of the HRT. (a) All pyroxene analyses from members A and B. (b) Pyroxene rim compositions from member A and B samples. Compositional fields (grey shaded regions) copied across from panel (a) for comparison. (c) All pyroxene analyses from member C. (d) Pyroxene rim compositions from member C. Compositional fields (grey shaded regions) copied across from panel (c) for comparison. Compositional labels from Poldervaart & Hess (1951). Full data set in Supplementary Data Electronic Appendix 5. Member C pyroxenes present variable compositions, depending on the clast type and/or locality. Pyroxenes from suite C1 clasts show a continuum from Ca-poor to Ca-rich (En3–45Fs43–68Wo0–43) but with only crystals from sample YP648 plotting in the sub-calcic augite field (Fig. 21). In contrast, pyroxenes from samples in suites C2 and C3 show a more clustered compositional nature. Although there is a subset of suite C2 grains with pigeonite compositions (En11–37Fs62–79Wo1–11), these are only found in interior domains (similarly to pyroxenes in samples from members A and B) and in crystals apparently in disequilibrium (e.g. containing sieve textures). Augite compositions from suite C2 form a similar compositional trend to those of members A and B, but extend to more Fe-rich compositions (En3–28Fs32–56Wo38–43) with rim compositions tightly clustered at the Fe-rich end (En4–41Fs53–55Wo41–43). Suite C3 grains are homogenous, tightly clustered and plot within the A and B augite field (En7–11Fs48–52Wo39–44) with the exception of one pigeonite grain (out of 15 analysed). Pyroxene crystals in the HRT are dominantly (normally) zoned (>60% of crystals in all suites: Fig. 21; Supplementary Data Electronic Appendix 5). Olivine All olivines are fayalitic but compositional diversity is present within the different juvenile clast types. Olivine crystals are unzoned and homogenous in all samples studied. However, they commonly show oxidation alteration to iron oxides and hydroxides on rims and along fractures (Fig. 22), as noted also in other eruptive units from the YPVF (Troch et al., 2017). Olivines in members A and B are similar in composition and encompass the entire range in the HRT (Fa86–100; Supplementary Data Electronic Appendix 3), but some compositional clustering is present. There is bimodality in CaO contents: one cluster has CaO contents ≥0·32 wt %, dominated by crystals from J4-type (mingled) pumices, while the other has modal concentrations of 0·18–0·30 wt % CaO in crystals from J1-type pumices (Supplementary DataFig. S14). There is a similar bimodality in MgO: a lower MgO cluster (1·4–1·8 wt %) is defined by crystals from suite A1 samples, whereas olivines from other clast types define a cluster with a peak at 1·9–3·0 wt % MgO (Supplementary DataFig. S15). Member C fayalite compositions (Fa94–100) vary less than in members A and B but show a greater range in CaO (0–0·41 wt %; Supplementary DataFig. S14). Olivines from suites C1 and C2 are predominantly elevated in CaO (0·24–0·41 wt %) with a minor group with ≤0·04 wt % CaO. They also extend to lower MgO values (0·44–1·48 wt %, with a subset at <0·03 wt %). There are distinct differences, however, between olivines from suites C1 and C2 when compared to those from suite C3. Olivines from suite C3 are more similar to those from members A and B, and compositionally overlap with suite A3 olivines (1·43–2·34 wt % MgO; Supplementary DataFig. S15). Fig. 22. Open in new tabDownload slide Backscattered electron images of representative ferromagnesian minerals and textures found in the HRT and compositional data from electron microprobe analyses (Supplementary Data Electronic Appendix 5). (a) Bright pigeonite core within a augite crystal from YP140 ( ‘B in C’ type J1 clast, suite B1). (b) Dark Mg-rich augite core within lighter Fe-rich augitic rim from YP414 (type J1, suite A1). (c) Zoned augite crystal from YP363 (type J1, suite A1) Adhering crystals are Fe–Ti oxides. (d) Altered olivine with bright, homogeneous unaltered regions and darker oxidized domains, commonly around rims and along fractures (YP562: type J9, suite C2). (e), (f) Zoned chevkinite crystals in YP255 (type J1, suite A3). Fig. 22. Open in new tabDownload slide Backscattered electron images of representative ferromagnesian minerals and textures found in the HRT and compositional data from electron microprobe analyses (Supplementary Data Electronic Appendix 5). (a) Bright pigeonite core within a augite crystal from YP140 ( ‘B in C’ type J1 clast, suite B1). (b) Dark Mg-rich augite core within lighter Fe-rich augitic rim from YP414 (type J1, suite A1). (c) Zoned augite crystal from YP363 (type J1, suite A1) Adhering crystals are Fe–Ti oxides. (d) Altered olivine with bright, homogeneous unaltered regions and darker oxidized domains, commonly around rims and along fractures (YP562: type J9, suite C2). (e), (f) Zoned chevkinite crystals in YP255 (type J1, suite A3). Accessory phases As documented by Macdonald & Belkin (2002), chevkinite is present in the HRT (Fig. 22; Supplementary Data Electronic Appendix 3). Individual crystals (0·5–1 mm) are most common in suite A3 samples (9% of picked ferromagnesian minerals) but also occur in J1 clasts from suite B1 (<1%) and suite C3 (3%). Accessory apatite and zircon are also present. Fe–Ti oxide pairs are also present, but almost always one or both of these phases show exsolution textures (indicative of sub-solidus oxidation) and as such were not analysed. ESTIMATES OF INTENSIVE VARIABLES Temperature Because of the scarcity of non-exsolved Fe–Ti oxide pairs, zircon saturation calibrations (Supplementary Data Electronic Appendix 4) for groundmass glasses were used to estimate pre-eruptive melt temperatures for samples where zircon was inferred to be stable through petrography or Zr abundances decreasing with SiO2 (i.e. all except type J5, suite C1 samples). We used the Watson & Harrison (1983) calibration due to its consistency with other temperature estimates for the HRT (Myers et al., 2016; Swallow et al., 2018a). Types J1 and J2 clasts from members A and B yield a wide temperature range (790–890°C; Supplementary DataFig. S16 and Supplementary Data Electronic Appendix 4). Suite A3 pumices cluster at the cooler end of the range (790–840°C), while suite A2 pumices yield higher temperatures (890–960°C). Similarly, rhyolitic glasses from scoria-bearing J4 clasts in member B yield a wide temperature range (820–950°C). Type J7 clasts (suite C3) yield comparably lower estimates to their suite A3 counterparts (800–840°C). In contrast, the diverse glass compositions from type J8 clasts (suite C2) yield a wide range of temperature estimates (820–900°C) and show a bimodal distribution (Supplementary DataFig. S16). If single-clast compositions are used for devitrified member C clasts (type J9; also suite 2) from the Red Mountains (810°C) and Huckleberry Mountain (870–900°C), their model temperatures are comparable to those of type-J8 samples. Two-feldspar thermometry (Elkins & Grove, 1990) was also applied to plagioclase inclusions in sanidine and vice versa (Supplementary Data Electronic Appendix 5). All feldspar compositions were normalised to Cn-free (following Swallow et al., 2018a). Types J1 and J2 clasts from members A and B yield a large range in model temperatures (770–950°C) with medians of 840°C and 860°C, respectively, slightly hotter than the respective median zircon model temperatures for each group (830 and 840°C; Supplementary DataFig. S16). Type J1 clasts in suite A2 pumices yield hotter estimates (840–920°C) that overlap with the cooler end of the zircon saturation range for the same suite. Type J1 clasts in suite A3 pumices are offset to cooler temperatures (750–820°C; median 790°C), ∼20°C cooler than the zircon median (810°C). Wide temperature ranges, comparable to those derived from zircon saturation estimates, are returned from member B samples; type J1 clasts yield a comparable range to member A (770–940°C), whereas scoria-bearing type J4 clasts extend to 760–990°C. In member C, type J7 clasts (suite C3) yield a wider range (770–860°C) than their suite A3 comparators, but a similar median (790°C). Type J8 samples (suite C2) show the largest temperature range of all compositional groups (720–920°C) encompassing the range in compositionally equivalent devitrified clasts from Huckleberry Mountain (780–900°C). Sparse suitable feldspars from type J5 (suite C1) clasts in member C yield a wide range of temperatures, extending to implausibly cooler temperatures (650–810°C). Utilising the presence of pigeonite compositions as cores or inclusions in augite crystals, two-pyroxene thermometry was also undertaken (Supplementary Data Electronic Appendix 5). Out of 58 potential pairings, however, only four passed the Putirka (2008) Fe–Mg exchange equilibrium test of KD = 1·09 ± 0·14. Using the calibration of Brey & Köhler (1990), temperature estimates were obtained in crystals from clasts of types J1 (870°C), J4 (1010°C), J8 (1100°C) and J5 (820°C). Temperature estimates using the calibration of Putirka (2008) for low Mg# clinopyroxenes yielded implausible results (temperatures up to 1810°C). Oxidation state and H2O content The presence of fayalite, magnetite and quartz in all compositional suites constrains the oxygen fugacity for the HRT magmas to be close to the QFM buffer, consistent with other studies of Yellowstone–Snake River Plain (YSRP) rhyolites (Christiansen & McCurry, 2008; Almeev et al., 2012; Befus & Gardner, 2016). The plagioclase-melt hygrometer of Waters & Lange (2015) was used to estimate pre-eruptive melt H2O contents (Supplementary Data Electronic Appendix 5), using zircon saturation thermometry estimates due to the wide range of two-feldspar model temperatures. Other inputs are measured rim compositions, mean major element groundmass glass compositions (or single clast compositions for devitrified member C samples), and a pressure of 200 MPa (Myers et al., 2016, their Supplementary Data Electronic Appendix 1). We use mean temperature estimates to calculate mean H2O contents. Samples from suite A1 yield 2·7–3·6 wt % H2O, a similar range to that from type J1 clasts in member B (suite B1: 2·1–3·6 wt % H2O). In comparison, maximum contents in suite A3 pumices are higher (3·8–4·0 wt % H2O), whereas contents in samples from suite A2 and type J4 clasts in suite B1 are lower (1·2–1·3 and 1·7–3·.0 wt % H2O, respectively). Suite C3 pumices yield similar H2O contents (3·7–4·0 wt %) to their suite A3 counterparts, whereas suite C2 samples yield H2O contents of 2·0–3·5 wt %, with one outlier at 4·1 wt %. For suite C1 samples the hygrometer yields unlikely maximum estimates of 8·5 wt % H2O, but this reflects the anomalously low model temperatures (which could only be done by two-feldspar thermometry) in these samples. Comparison of the hygrometer between melt inclusion data from the early HRT fall deposits (Myers et al., 2016; 3.3–4.9 wt % H2O) and the associated plagioclase and groundmass glasses (4·0–4·5 wt % H2O; E. J. Swallow, unpublished data cited in Myers et al., 2016) suggests that the hygrometer is accurate. However, consideration of likely uncertainties in model temperatures would suggest that variations of ± 1·2 wt % H2O could apply to individual determinations. The ranges listed above should thus be taken as indicative. Although pyroxenes are limited in constraining temperatures, the presence of pigeonite only as cores or inclusions is significant. In contrast to the HRT, the ‘hot and dry’ (Branney et al., 2008) Snake River Plain ignimbrites commonly have co-existing population(s) of phenocrystic pigeonite and augite within any one eruption (Cathey & Nash, 2004; Ellis & Wolff, 2012). Experimental work on these materials (Almeev et al., 2012; Bolte et al., 2015) provides constraints, as follows. At water contents of <0·5 wt %, pigeonite is a near-liquidus phase (∼1000 °C) but augite replaces pigeonite as the earliest-forming pyroxene at higher water contents (Almeev et al., 2012). Augite was not observed in experiments at H2O > 1·3 wt % (Almeev et al., 2012). Ca-rich augite compositions (i.e. >40 mol % Wo), as seen predominantly in rim compositions in the HRT, are only observed in experiments with moderate-high melt H2O contents (>1 wt % at 925 °C to >3 wt % at 825 °C; Almeev et al., 2012). Augite is generally stable at lower temperatures than pigeonite and the range in conditions where both can co-exist is narrow (Bolte et al., 2015). These constraints suggest that the HRT magmatic systems were initially H2O-poor, and that pigeonite, possibly with co-existing augite, was stable. Increasing H2O contents and decreasing temperatures during evolution of the various magma systems preferentially favoured augite stability and a continuing increase in H2O contents led to an increase in the augite Wo content, as displayed in augite rim compositions (Fig. 21). Pressure Pre-eruptive storage pressures were estimated from groundmass glass major element compositions using the Determining Rhyolite Pressures (DERP) geobarometer of Wilke et al. (2017; Supplementary Data Electronic Appendix 4), which requires glasses to have equilibrated with quartz and at least one feldspar. Although criticised by Gualda et al. (2019) the DERP model remains a valid approach for the HRT glasses as the glasses are mildly corundum normative and are closer in composition to those used to define the DERP model than those used to define the alternative rhyolite-MELTS model (Wilke et al., 2019). Types J1 and J2 clasts from member A (suite A1) yield a similar distribution and range of model pressures from ∼100–450 MPa (Supplementary DataFig. S17). The lower figure is consistent with the lowest pressures (80 MPa) derived from quartz-hosted melt inclusions in the initial fall deposits (Myers et al., 2016). Type J3 clasts (suite A2) yield a tail of greater model pressures (mostly >800 MPa; Supplementary DataFig. S17), although these are outside the range of model calibration (Wilke et al., 2017). Although some samples return impossible negative pressures, type J1 clasts in member B (suite B1) yield a broader distribution of pressures than their counterparts in member A, but have similar mean and median values (320 and 330 MPa, respectively, in member A; 310 and 320 MPa in member B). Scoria-bearing type J4 clasts in member B show a bimodal distribution, with one mode (∼330 MPa) similar to J1 clasts, and the other at ∼800 MPa (Supplementary DataFig. S17). In member C, type J8 (suite C2) samples extend to lower model pressures (mean 200 MPa, median 230 MPa) when compared to J7 samples (mean 260 MPa, median 270 MPa), while the latter estimates are closely similar to their member A, suite A3 comparators (mean 270 MPa, median 270 MPa). However, pressure estimates can only be as precise as the input data. We tested the DERP model using an average major element glass composition from all analyses in this study and absolute 2 s.d. EPMA uncertainties derived from multiple analyses of glass standards (e.g. SiO2 ± 1·16 wt %; K2O ± 0·48 wt %). Approximate differences in model pressures of ±50 MPa from SiO2 and ±100 MPa from K2O are calculated using these typical analytical uncertainties. In comparison, variations in input H2O contents of ±2 wt %, around a starting point of the mean glass composition and 4 wt % H2O, yield pressure differences of approximately ±20 MPa. As a result, we consider that pressure differences of less than ∼100 MPa are effectively unresolvable with the analytical uncertainties attained in this study (which are typical of EPMA data). This is important to note as pressures estimated using EPMA data on (hydrated) natural glasses can be a recurring focal point of debate regarding the nature of large magmatic systems (e.g. Gualda & Ghiorso, 2013; Chamberlain et al., 2015; Wilke et al., 2019). DISCUSSION The HRT provides an unparalleled example of compositional diversity within the deposits of a large silicic, caldera-forming eruption that in itself was complex and prolonged (Fig. 2, Supplementary DataFig. S1). The initial fall deposits represent episodic activity, inferred to have occupied some weeks, with time for reworking and numerous short hiatuses in eruption, matched by evidence from melt inclusion volatile abundances for slow magma ascent (Myers et al., 2016; Swallow et al., 2018a). Ignimbrite members A and B were separated by weeks to months, and members B and C by years to decades. This chronostratigraphic interpretation provides a framework within which the observed extraordinary compositional diversity can be placed. Magmatic systems in the HRT To provide a central focus for discussion, the elemental and isotopic characteristics of the seven silicic suites (A1, A2, A3; B1; C1, C2, C3) are used here to define four magmatic systems that at various stages simultaneously and successively fed the HRT eruption (Fig. 23). The defining chemical and isotopic characteristics of each system are summarised in Table 4 and reviewed below. Note that our geochemical and isotopic data indicate that the HRT magmatic systems were discrete entities (cf. Cooper et al., 2012, 2016) and not subtly different components of a heterogeneous single system (cf. Bégué et al., 2014). Fig. 23. Open in new tabDownload slide Summary schematic diagram showing the broad eruptive timescale and stratigraphy of the HRT along with the four inferred silicic magmatic systems (with their associated magmatic suites and mafic root zones) tapped during the eruption. See text for discussion. Colours based on symbol colours that distinguish the systems/suites in earlier figures. Fig. 23. Open in new tabDownload slide Summary schematic diagram showing the broad eruptive timescale and stratigraphy of the HRT along with the four inferred silicic magmatic systems (with their associated magmatic suites and mafic root zones) tapped during the eruption. See text for discussion. Colours based on symbol colours that distinguish the systems/suites in earlier figures. Table 4: Distinguishing geochemical and petrological characteristics of the four HRT magmatic systems Magmatic system Members, suites and clast types Major element characteristics Trace element characteristics Isotopic characteristics Glass characteristics Crystal characteristics 1 Members A and B· Suites A1 (J1-, J2- type clasts) and A2 (J3-type) in member A, that are linked in member B by suite B1 (J1-, J3- and J4- type clasts) High in alkalis, and Al2O3 Depleted in FeO, CaO Large range in Ba and Zr Low Sr and Sc abundances None to negative Eu anomaly Moderately radiogenic isotope systematics Increasing 87Sr/86Sr with increasing 1/Sr, decreasing Ba Constrained Pb isotopic ratios identical to HRT mafics Clustered compositions Similar to FGC/OGC3 and IGC2-8 in the fall deposit and basal ignimbrite Suite A2 unique to member A; gap with A1 infilled by J4-type pumices in member B >0·4 mol% Cn in sanidines Orthopyroxene cores/inclusions, clinopyroxene rims Olivines in J4-type clasts have higher CaO 2 Members A and C· Suites A3 (J1-type pumices) and C3 (J7-type clasts) Similar to magmatic system 1 Elevated Nb, Rb Low Ba, Sr, LREE Moderate to high negative Eu anomaly Similar to system 1 for Pb and Nd isotopes and Sr in member A Regeneration in member C contains less radiogenic 87Sr/86Sr Low Ba and Sr (similar to FGC/OGC1 + 2 and IGC1) Low (<0·35 mol %) Cn sanidines Lower MgO olivines relative to system 1 3 Member C· Suite C2 comprised of J8-type pumices together with J9- and rare J5-type clasts Offset to higher SiO2 from the system 1 trends Low in alkalis Moderate in FeO and TiO2 Trend back to higher Ba, Sr, Zr Low Nb, LREE No Eu anomaly to moderately –ve Eu anomaly Highly radiogenic isotopic systematics Negative 208Pb/204Pb correlation with 206Pb/204Pb Positive correlation between 87Sr/86Sr and 1/Sr Clustered compositions Offset to higher Ba, Sr and lower La, Zr compared to system 1 Negatively correlated Ba and La Fe-rich olivines and clinopyroxenes High Cn sanidine 4 Member C· Suite C1 comprised of J5- and J6-type juvenile clasts Elevated in FeO, CaO, P2O5 and TiO2 Low in Al2O3 and alkalis Elevated in Sr and Sc Low in Ba, Zr, Nb and LREE No Eu anomaly Highly radiogenic isotopic systematics Negative 208Pb/204Pb correlation with 206Pb/204Pb Positive correlation between 87Sr/86Sr and 1/Sr N/A- only ever found as devitrified clasts Continuum in compositions between orthopyroxene and clinopyroxene Calcic plagioclase Magmatic system Members, suites and clast types Major element characteristics Trace element characteristics Isotopic characteristics Glass characteristics Crystal characteristics 1 Members A and B· Suites A1 (J1-, J2- type clasts) and A2 (J3-type) in member A, that are linked in member B by suite B1 (J1-, J3- and J4- type clasts) High in alkalis, and Al2O3 Depleted in FeO, CaO Large range in Ba and Zr Low Sr and Sc abundances None to negative Eu anomaly Moderately radiogenic isotope systematics Increasing 87Sr/86Sr with increasing 1/Sr, decreasing Ba Constrained Pb isotopic ratios identical to HRT mafics Clustered compositions Similar to FGC/OGC3 and IGC2-8 in the fall deposit and basal ignimbrite Suite A2 unique to member A; gap with A1 infilled by J4-type pumices in member B >0·4 mol% Cn in sanidines Orthopyroxene cores/inclusions, clinopyroxene rims Olivines in J4-type clasts have higher CaO 2 Members A and C· Suites A3 (J1-type pumices) and C3 (J7-type clasts) Similar to magmatic system 1 Elevated Nb, Rb Low Ba, Sr, LREE Moderate to high negative Eu anomaly Similar to system 1 for Pb and Nd isotopes and Sr in member A Regeneration in member C contains less radiogenic 87Sr/86Sr Low Ba and Sr (similar to FGC/OGC1 + 2 and IGC1) Low (<0·35 mol %) Cn sanidines Lower MgO olivines relative to system 1 3 Member C· Suite C2 comprised of J8-type pumices together with J9- and rare J5-type clasts Offset to higher SiO2 from the system 1 trends Low in alkalis Moderate in FeO and TiO2 Trend back to higher Ba, Sr, Zr Low Nb, LREE No Eu anomaly to moderately –ve Eu anomaly Highly radiogenic isotopic systematics Negative 208Pb/204Pb correlation with 206Pb/204Pb Positive correlation between 87Sr/86Sr and 1/Sr Clustered compositions Offset to higher Ba, Sr and lower La, Zr compared to system 1 Negatively correlated Ba and La Fe-rich olivines and clinopyroxenes High Cn sanidine 4 Member C· Suite C1 comprised of J5- and J6-type juvenile clasts Elevated in FeO, CaO, P2O5 and TiO2 Low in Al2O3 and alkalis Elevated in Sr and Sc Low in Ba, Zr, Nb and LREE No Eu anomaly Highly radiogenic isotopic systematics Negative 208Pb/204Pb correlation with 206Pb/204Pb Positive correlation between 87Sr/86Sr and 1/Sr N/A- only ever found as devitrified clasts Continuum in compositions between orthopyroxene and clinopyroxene Calcic plagioclase Open in new tab Table 4: Distinguishing geochemical and petrological characteristics of the four HRT magmatic systems Magmatic system Members, suites and clast types Major element characteristics Trace element characteristics Isotopic characteristics Glass characteristics Crystal characteristics 1 Members A and B· Suites A1 (J1-, J2- type clasts) and A2 (J3-type) in member A, that are linked in member B by suite B1 (J1-, J3- and J4- type clasts) High in alkalis, and Al2O3 Depleted in FeO, CaO Large range in Ba and Zr Low Sr and Sc abundances None to negative Eu anomaly Moderately radiogenic isotope systematics Increasing 87Sr/86Sr with increasing 1/Sr, decreasing Ba Constrained Pb isotopic ratios identical to HRT mafics Clustered compositions Similar to FGC/OGC3 and IGC2-8 in the fall deposit and basal ignimbrite Suite A2 unique to member A; gap with A1 infilled by J4-type pumices in member B >0·4 mol% Cn in sanidines Orthopyroxene cores/inclusions, clinopyroxene rims Olivines in J4-type clasts have higher CaO 2 Members A and C· Suites A3 (J1-type pumices) and C3 (J7-type clasts) Similar to magmatic system 1 Elevated Nb, Rb Low Ba, Sr, LREE Moderate to high negative Eu anomaly Similar to system 1 for Pb and Nd isotopes and Sr in member A Regeneration in member C contains less radiogenic 87Sr/86Sr Low Ba and Sr (similar to FGC/OGC1 + 2 and IGC1) Low (<0·35 mol %) Cn sanidines Lower MgO olivines relative to system 1 3 Member C· Suite C2 comprised of J8-type pumices together with J9- and rare J5-type clasts Offset to higher SiO2 from the system 1 trends Low in alkalis Moderate in FeO and TiO2 Trend back to higher Ba, Sr, Zr Low Nb, LREE No Eu anomaly to moderately –ve Eu anomaly Highly radiogenic isotopic systematics Negative 208Pb/204Pb correlation with 206Pb/204Pb Positive correlation between 87Sr/86Sr and 1/Sr Clustered compositions Offset to higher Ba, Sr and lower La, Zr compared to system 1 Negatively correlated Ba and La Fe-rich olivines and clinopyroxenes High Cn sanidine 4 Member C· Suite C1 comprised of J5- and J6-type juvenile clasts Elevated in FeO, CaO, P2O5 and TiO2 Low in Al2O3 and alkalis Elevated in Sr and Sc Low in Ba, Zr, Nb and LREE No Eu anomaly Highly radiogenic isotopic systematics Negative 208Pb/204Pb correlation with 206Pb/204Pb Positive correlation between 87Sr/86Sr and 1/Sr N/A- only ever found as devitrified clasts Continuum in compositions between orthopyroxene and clinopyroxene Calcic plagioclase Magmatic system Members, suites and clast types Major element characteristics Trace element characteristics Isotopic characteristics Glass characteristics Crystal characteristics 1 Members A and B· Suites A1 (J1-, J2- type clasts) and A2 (J3-type) in member A, that are linked in member B by suite B1 (J1-, J3- and J4- type clasts) High in alkalis, and Al2O3 Depleted in FeO, CaO Large range in Ba and Zr Low Sr and Sc abundances None to negative Eu anomaly Moderately radiogenic isotope systematics Increasing 87Sr/86Sr with increasing 1/Sr, decreasing Ba Constrained Pb isotopic ratios identical to HRT mafics Clustered compositions Similar to FGC/OGC3 and IGC2-8 in the fall deposit and basal ignimbrite Suite A2 unique to member A; gap with A1 infilled by J4-type pumices in member B >0·4 mol% Cn in sanidines Orthopyroxene cores/inclusions, clinopyroxene rims Olivines in J4-type clasts have higher CaO 2 Members A and C· Suites A3 (J1-type pumices) and C3 (J7-type clasts) Similar to magmatic system 1 Elevated Nb, Rb Low Ba, Sr, LREE Moderate to high negative Eu anomaly Similar to system 1 for Pb and Nd isotopes and Sr in member A Regeneration in member C contains less radiogenic 87Sr/86Sr Low Ba and Sr (similar to FGC/OGC1 + 2 and IGC1) Low (<0·35 mol %) Cn sanidines Lower MgO olivines relative to system 1 3 Member C· Suite C2 comprised of J8-type pumices together with J9- and rare J5-type clasts Offset to higher SiO2 from the system 1 trends Low in alkalis Moderate in FeO and TiO2 Trend back to higher Ba, Sr, Zr Low Nb, LREE No Eu anomaly to moderately –ve Eu anomaly Highly radiogenic isotopic systematics Negative 208Pb/204Pb correlation with 206Pb/204Pb Positive correlation between 87Sr/86Sr and 1/Sr Clustered compositions Offset to higher Ba, Sr and lower La, Zr compared to system 1 Negatively correlated Ba and La Fe-rich olivines and clinopyroxenes High Cn sanidine 4 Member C· Suite C1 comprised of J5- and J6-type juvenile clasts Elevated in FeO, CaO, P2O5 and TiO2 Low in Al2O3 and alkalis Elevated in Sr and Sc Low in Ba, Zr, Nb and LREE No Eu anomaly Highly radiogenic isotopic systematics Negative 208Pb/204Pb correlation with 206Pb/204Pb Positive correlation between 87Sr/86Sr and 1/Sr N/A- only ever found as devitrified clasts Continuum in compositions between orthopyroxene and clinopyroxene Calcic plagioclase Open in new tab Magmatic system 1 Magmatic system 1 is inferred to be volumetrically dominant in terms of eruption products, based on the relative abundance of clasts observed and analysed (including the glass compositions in Swallow et al., 2018a), and the overall proportion (∼90 vol. %: Christiansen, 2001) of members A and B in the HRT. Magmatic system 1 hosted the melt-dominant bodies that were evacuated for most of the initial fall deposits (glass clusters FGC/OGC3 and IGC2 to IGC8 of Swallow et al., 2018a) and ignimbrite members A and B (suites A1, A2 and B1: Fig. 23). Products of this system show wide ranges in geochemical composition (e.g. 410–4170 ppm Ba) but a moderate isotopic range (e.g. 87Sr/86Sr = 0·71009–0·71256: Figs 8, 14). Single clast compositions are enriched in alkalis, Al2O3, Zr and Ba, and depleted in FeO, TiO2, CaO, Sr and Sc relative to the three other systems (Figs 5–12; Supplementary Data Figs S2–S4, S7). Clustered glass compositions are interpreted to reflect multiple melt-dominant bodies which capped the system (Figs 16, 17), some of which were tapped from the eruption onset (Myers et al., 2016) and others which first vented at the onset of member A emplacement (Swallow et al., 2018a). In its mineralogical characteristics, sanidines from this system contain >0·4 mol % Cn and pyroxenes are typically Fe-rich clinopyroxene, sometimes with pigeonite cores, and are tightly clustered in composition (Figs 19 and 21). No material has been found with the compositional characteristics of suite B1 but with the small crystal sizes present in all juvenile ejecta found in member C, implying that magmatic system 1 was not vented during member C activity, and that it was distinct from magmatic system 2. Magmatic system 2 Magmatic system 2, minor in volume when compared to system 1, is represented by the low-Ba glass in the initial fall deposits (FGC/OGC1 + 2, IGC1 of Swallow et al., 2018a), and clasts of suites A3 and C3 in ignimbrite members A and C, respectively (Fig. 23). Although suite A3 compositions arguably fall on a continuum with those of suite A1 (Figs 5–7), the contrasting evacuation history relative to magmatic system 1 (i.e. its absence in member B and reappearance in member C: Fig. 23), the distinctive sanidine (<0·35 mol % Cn: Fig. 19) and olivine compositions (Supplementary Data Figs S14 and S15), and the lack of mixing/mingling with any other compositional suite make it distinct. Although suite A3 samples have comparable isotopic compositions to the materials from magmatic system 1, suite C3 samples are less radiogenic with respect to Sr (Fig. 14) for a given elemental abundance (although no different in Nd or Pb isotopes: Figs 15, Supplementary DataFig. S9) and the pumices have contrasting textures and smaller crystal sizes. System 2 was vented from the very start of the eruption in the earliest fall deposits (Myers et al., 2016), then stopped, and re-started in early member A ignimbrite activity (Swallow et al., 2018a). It was not vented during eruption of member B, then was reactivated during the B: C time break to produce new melt-dominant magma that was vented in minor amounts during eruption of member C ignimbrite. Magmatic system 3 Magmatic system 3 is represented by suite C2 clasts, which are the dominant lineage in member C. It is inferred to be the second largest system represented within the HRT on the basis of the relative abundance of suite C2 clasts within the dominant observed volume of member C ignimbrite. Suite C2 covers a comparable compositional range to the suites in magmatic system 1 (e.g. Ba 450–2840 ppm: Fig. 8), but is offset to higher SiO2 abundances. Products of this system exhibit a unique compositional trend with large variations in trace elements occurring over a narrow SiO2 range (75–78 wt %; e.g. Figs 8, 9, 12). They are depleted in Nb (27–32 ppm) relative to member A and B samples, but have moderate Fe2O3 (1·3–3·3 wt %) and TiO2 (0·12–0·29 wt %) concentrations (Fig. 10, Supplementary Data Figs S2 and S3). Products of this magmatic system have strongly crustal isotopic signatures (e.g. 87Sr/86Sr = 0·72590–0·73026: Figs. 13, 14; cf. Doe et al., 1982) and in this are closely linked to system 4. Groundmass glass compositions from this system are clustered, again suggesting the presence of multiple melt-dominant bodies (Figs 16, 17; cf. Myers et al., 2016; Swallow et al., 2018a), although glass compositional trends are offset from those in members A and B. Characteristic crystal compositions include Fe-rich olivines and Cn-rich sanidines (Fig. 19, Supplementary Data Figs S14 and S15). Magmatic system 4 Magmatic system 4 is represented by suite C1 samples and is the minor of the two newly-erupted systems contributing to ignimbrite member C, based on the observed minor volumetric abundances of clasts of type 5 in member C ignimbrite. Despite their moderate SiO2 contents, suite C1 samples return remarkably crustal isotopic compositions (e.g. 87Sr/86Sr = 0·72464–0·72486: Figs 13–15). Suite C1 samples are distinctly elevated in CaO, Fe2O3, Sc and Sr, and lower in Al2O3, Ba and Nb relative to the other magmatic systems (Figs 6-11, Supplementary Data Figs S2 and S4), comparable in some respects to those compositions of magmatic system 3, but distinct in others (e.g. Sr, Zr: Fig. 9, Supplementary Data Fig. S7). Like other member C material, suite C1 samples have small crystal sizes (<3 mm) and contain pyroxenes along a continuum between the pigeonite and augite fields (Fig. 21). To add to these complexities, system 1 alone shows evidence from compositional variations and the direct eruption of homogeneous and mingled clasts in members A and B, for extensive interaction with mafic magmas. Although most of these compositions belong to the ‘HRT mafics’ grouping documented by Swallow et al. (2018b), some material with tholeiitic affinities also interacted with and was erupted from magmatic system 1 to be incorporated in upper member B material distributed north of the caldera. Magmatic system 2 shows no indicators of syn-eruptive interaction with mafic compositions, but has compositional and isotopic links to magmatic system 1 and thus an indirect linkage to the HRT mafic magmas. Magmatic system 3 also has some compositional links to the distinctive HRT mafics, but also shows no signs of direct interaction, and is isotopically distinct from both systems 1 and 2. In contrast, system 4 shows compositional trends with respect to SiO2 that are aligned in several respects to the fields defined by YPVF olivine tholeiite lavas (suite M2: Swallow et al., 2018b; e.g. Figs 5, 8, and 10, Supplementary Data Fig. S7), although, again, no evidence for syn-eruptive interaction has been found. Petrogenesis of HRT silicic melts Here we consider the processes that influenced the origins of the compositions erupted from the four HRT magmatic systems. Detailed quantitative petrogenetic modelling is not possible at this stage for two reasons. First, the overall range of mafic compositions erupted into member A is very broad, and no clear end-member ‘primitive’ composition can be identified. The spread of data increases markedly with the scoria erupted into member B, and it is apparent in particular from some elements (Fe2O3, P2O5, TiO2, CaO, Ba, Sr, Zr: Figs 7–9, Supplementary Data Figs S2–S4, S7) that renewed influxes of mafic magma into member B included other magmas with tholeiitic affinities. A distinctive mafic end-member is thus hard to define. Second, we lack the data on trace element or isotopic compositions in the mineral phases that would serve to accurately define the respective end-members or fractionation pathways. We thus discuss a range of possible processes below to arrive at generalised conclusions about the pathways to generation of the HRT silicic compositions. An overall summary flow chart of the processes discussed is given in Fig. 24 and the time-sequence of erupted materials summarised in Fig. 25. Fig. 24. Open in new tabDownload slide Flow chart showing the distinguishing features of each of the four HRT magmatic systems and their petrogenetic pathways. The overall characteristics of the silicic systems are controlled in turn by their parental mafic root zone, the degrees of interaction with crustal protoliths and preceding intrusions, and spatial relationship to the focus of mafic influx. See text for discussion. See also Swallow et al. (2018b) for background to the source regions for the mafic magma suites. Colours based on symbol colours that distinguish the systems/suites in earlier figures. Fig. 24. Open in new tabDownload slide Flow chart showing the distinguishing features of each of the four HRT magmatic systems and their petrogenetic pathways. The overall characteristics of the silicic systems are controlled in turn by their parental mafic root zone, the degrees of interaction with crustal protoliths and preceding intrusions, and spatial relationship to the focus of mafic influx. See text for discussion. See also Swallow et al. (2018b) for background to the source regions for the mafic magma suites. Colours based on symbol colours that distinguish the systems/suites in earlier figures. Fig. 25. Open in new tabDownload slide Semi-quantitative diagram showing the behaviour of the HRT magmatic complex throughout the eruption sequence. (a) Multiple laterally-adjacent melt-dominant bodies tapped at the onset of eruption (initial fall deposits) and into widespread deposition of member A ignimbrite were recognised on the basis of glass and feldspar compositions to represent two magmatic lineages (Swallow et al., 2018a). (b) Member A ignimbrite compositions reflect the continued tapping of these multiple melt-dominant bodies, with the lineages recognised here to represent magmatic systems 1 and 2. A deeper, separate suite A2 body was also tapped during member A deposition. Influx of mafic material predominantly in the form of dense clasts (Swallow et al., 2018b). (c) During a time break of weeks to months, extensive mixing occurred between a renewed influx of mafic material of at least two lineages (erupted at the surface as scoria in upper member B ignimbrite) and magmatic system 1 material. Member B then evacuated magmatic system 1 but there was a cessation in evacuation of system 2. (d) During a time break of years to decades, magmatic system 2 was rejuvenated, forming a new melt-dominant body (suite C3). Coevally, melt-dominant bodies from two previously unerupted magmatic systems (3 and 4) formed. Member C evacuated melt-dominant bodies from systems 2, 3 and 4, but no material representing system 1 has been found. Colours and suite names as in Swallow et al. (2018a) and Fig. 5. See text and Table 4 for classification of the magmatic systems. Fig. 25. Open in new tabDownload slide Semi-quantitative diagram showing the behaviour of the HRT magmatic complex throughout the eruption sequence. (a) Multiple laterally-adjacent melt-dominant bodies tapped at the onset of eruption (initial fall deposits) and into widespread deposition of member A ignimbrite were recognised on the basis of glass and feldspar compositions to represent two magmatic lineages (Swallow et al., 2018a). (b) Member A ignimbrite compositions reflect the continued tapping of these multiple melt-dominant bodies, with the lineages recognised here to represent magmatic systems 1 and 2. A deeper, separate suite A2 body was also tapped during member A deposition. Influx of mafic material predominantly in the form of dense clasts (Swallow et al., 2018b). (c) During a time break of weeks to months, extensive mixing occurred between a renewed influx of mafic material of at least two lineages (erupted at the surface as scoria in upper member B ignimbrite) and magmatic system 1 material. Member B then evacuated magmatic system 1 but there was a cessation in evacuation of system 2. (d) During a time break of years to decades, magmatic system 2 was rejuvenated, forming a new melt-dominant body (suite C3). Coevally, melt-dominant bodies from two previously unerupted magmatic systems (3 and 4) formed. Member C evacuated melt-dominant bodies from systems 2, 3 and 4, but no material representing system 1 has been found. Colours and suite names as in Swallow et al. (2018a) and Fig. 5. See text and Table 4 for classification of the magmatic systems. Magmatic systems 1 and 2 On the basis of their similar major element and isotopic features, we petrogenetically link magmatic systems 1 and 2 here. The HRT (and other YPVF rhyolites) were previously proposed to be generated by partial melting of tholeiitic intrusions (Doe et al., 1982; Hildreth et al., 1991; Christiansen, 2001; Christiansen & McCurry, 2008). In contrast, our data from magmatic systems 1 and 2 strongly imply a petrogenetic link to the distinctive HRT mafic material (suite M1, rich in alkalis, Ba and Zr: Swallow et al., 2018b), erupted with member A. In particular, the Pb isotopic compositions of samples from magmatic systems 1 and 2 and the HRT mafic samples overlap (Fig. 15), suggesting that they share a common root zone. Based on this feature, we explore the petrogenesis of magmatic system 1, focusing first on the least-evolved compositions from suite A1, plus suite A2, before discussing intra-suite variations and relationships between systems 1 and 2. Clasts from magmatic systems 1 and 2 have more radiogenic 87Sr/86Sr ratios than any HRT mafics (Fig. 13), although the differences are minor in comparison to those with suites C1 and C2, and 143Nd/144Nd ratios are similar in the rhyolitic and mafic samples (Fig. 13). These data thus preclude a significant petrogenetic role for the highly radiogenic (with respect to Sr) Archean country rocks (e.g. Doe et al., 1982; Wooden & Mueller, 1988; Hildreth et al., 1991; Frost et al., 2006). In detail, however, Sr isotopic compositions form an asymptotic increasing trend when plotted against indices of melt evolution (Fig. 14). This increase, and the offset of the silicic components in members A and B to higher 87Sr/86Sr values at similar elemental concentrations relative to the mafic suites (e.g. Ba: Fig. 14), implies some minor incorporation of an evolved, radiogenic (with respect to Sr) component during melt evolution. This component cannot be any of the strongly crustally influenced melts of suites C1 and C2 because corresponding mixing trends in either elemental abundances (e.g. Nb) or other isotopic (e.g. Pb) compositions (Figs 10, 15) are lacking. Overall, the moderate Sr isotopic diversity within magmatic systems 1 and 2 and slight positive correlation with 1/Sr (particularly relative to magmatic system 3), suggest that the final evolution of magmas in systems 1 and 2 was mostly controlled by fractional crystallization, but with minor assimilation. These inferences are also supported by the relative uniformity of Nd isotopic ratios with degrees of fractionation (Supplementary DataFig. S9; Wolff et al., 2015). Fractional crystallization model. Potential analogues to the HRT silicic compositions are the rhyolites of Cedar Butte (eastern Snake River Plain), interpreted to reflect extensive (∼80%, 4·3 kbar) fractional crystallization of a basaltic trachyandesite comparable to the HRT mafic compositions (McCurry et al., 2008; Whitaker et al., 2008). In parallel to the HRT rhyolites, McCurry et al. (2008) also reported minor increases in radiogenic Sr isotopes, correlated with melt evolution. We therefore consider fractional crystallization from an HRT mafic parent (or melting of a compositionally related protolith) for generation of the least evolved compositions within magmatic systems 1 and 2. In turn, as magmatic system 2 is of smaller volume, we focus on system 1, specifically suites A1 and A2 (as suite B1 is a combination of these). Overall, a simple fractional crystallization model is precluded, however, by the behaviour of highly incompatible elements such as Rb, Nb and U, which show increasing concentrations in member A mafics on either side of the compositional gap, but a uniformity across the gap (e.g. Fig. 10, Supplementary Data Fig. S6). Therefore, when plotted together, these incompatible elements show a broad continuum between HRT mafics, suite A2 and suite A1 material (Supplementary DataFig. S18). These relationships are inconsistent with a simple fractional crystallization model and would require coincidental changes in behaviour of these diverse elements at intermediate compositions. We therefore conclude that extreme fractional crystallization alone of a HRT mafic parent is implausible as the sole mechanism for generation of the silicic suites. Partial melting model. Alternatively, the HRT rhyolites could be generated through partial melting of mid-crustal intrusions of HRT mafic lineage type. Two simple, end-member, one-stage partial melting scenarios are considered. First, partial melting of a source similar in composition to the most-evolved of the member A mafics suite. This scenario is unlikely due to the comparable concentrations of strongly incompatible elements (e.g. Nb, Rb) between the most-evolved HRT mafic composition and suite A2 material (Supplementary Data Figs S6 and S18). Generation of melts with such similar incompatible element concentrations to the protolith would require either large degrees of melting (close to 100%) and/or an increase in compatibility of the elements in residual phases. The former is precluded by large increases (SiO2) and decreases (e.g. Ba, Zr) in other elements between the two compositional suites. Although the bulk compatibility of Rb would be elevated with residual alkali feldspar (suggested by the decrease in Ba), Nb concentration changes would require the presence of a Nb-rich phase, for which evidence is lacking (Swallow et al., 2018b). Second, partial melting of a cumulate derived from magmas similar to the member A mafics suite. From the trends in the mafics suite M1 (Swallow et al., 2018b) any hypothetical cumulate source should contain residual apatite (decrease in P2O5, LREE), plagioclase (broad decrease in Sr) and pyroxene (decrease in Sc with increasing SiO2). Partial melting of such a cumulate would therefore generate melts enriched in these elements relative to the melts from which the cumulate formed. However, the HRT silicic compositions are further depleted in P2O5, Sr and Sc than any compositions within the mafics suite. Fractionation-partial melting model. We infer that a multi-stage fractionation-partial melting model (similar to that of Reubi & Blundy, 2008) is required for the silicic melts of systems 1 and 2, involving compositions not sampled by us at the surface. Assuming an igneous protolith (Protolith 1 in Fig. 26) that is in part a cumulate, constraints on this model are as follows: Generation of a melt composition, continuing the fractionation trend from the HRT member A mafic materials, that is more evolved than any erupted mafic material sampled. This melt would have Nb, Rb and other incompatible element concentrations higher than those of the HRT mafics, suite A2 material and the least-evolved suite A1 material. To match the incompatible element concentrations between this melt and the rhyolites, there has to be further fractional crystallization of an assemblage from this melt that is incompatible element depleted, with some abstraction of an enriched melt (not known to be erupted, and/or not sampled here). Subsequent melting of the crystallized protolith would then generate silicic melts with moderately enriched concentrations of incompatible elements, similar to those in suite A2 and least-evolved suite A1 samples. The protolith would need to be sanidine- and zircon-bearing to explain the decreases in Ba and Zr contents across the SiO2 compositional gap. Continued decreases of P2O5, Sc and Sr throughout the entire sequence imply continuing roles for apatite, pyroxene and plagioclase as fractionating phases and/or as residual phases in any partially melted protolith. A role for pyroxene in the cumulate residuum is consistent with the depression of MREE in the silicic samples (Fig. 12), as seen elsewhere (Olin & Wolff, 2010). However, although P2O5 abundances decrease throughout the sequence, LREE concentrations become decoupled and plateau across the compositional gap before increasing between suites A2 and A1 (Supplementary Data Fig. S19). This decoupling could be related to the very low P2O5 contents in suites A1 and A2 (Fig. 7). Fig. 26. Open in new tabDownload slide Semi-quantitative diagram summarising the HRT magmatic complex from the upper mantle to the surface. Bracketed numbers indicate key processes, as follows. (i) HRT mafic melts are generated from regions of the mantle enriched by high P–T fluids derived from the subducted Farallon slab. Early HRT mafic melts ascend through contemporaneously or previously (in late Cretaceous or Eocene times) hybridised crust (see Swallow et al., 2018b). (ii) Melts stall and crystallize in the middle crust to form protolith 1. (iii) Partial melts of protolith 1 ascend and become parental melts that differentiate to form the melt-dominant bodies of magmatic systems 1 and 2. (iv) Crystalline residuum associated with these fractionates forms protolith 2, which is remelted to form suite A2. (v) Suite M1 HRT mafic melts ascend through the crust and intersect and mix with magmatic system 1. Magmatic system 2 is located away from the focus of mafic influx. (vi) Spatially separated from magmatic systems 1 and 2, as well as within a region of un-modified Archean crust (Swallow et al., 2018b), melts from enriched (suite M1) and non-enriched (suite M2) source zones ascend from the uppermost mantle, stall in the middle crust, hybridise with surrounding Archean crust, crystallize and solidify. (vii) Partial melts of these hybridised cumulates ascend and feed magmatic systems 3 (suite C2) and 4 (suite C1). Crustal/mantle boundary depths from Huang et al. (2015). Fig. 26. Open in new tabDownload slide Semi-quantitative diagram summarising the HRT magmatic complex from the upper mantle to the surface. Bracketed numbers indicate key processes, as follows. (i) HRT mafic melts are generated from regions of the mantle enriched by high P–T fluids derived from the subducted Farallon slab. Early HRT mafic melts ascend through contemporaneously or previously (in late Cretaceous or Eocene times) hybridised crust (see Swallow et al., 2018b). (ii) Melts stall and crystallize in the middle crust to form protolith 1. (iii) Partial melts of protolith 1 ascend and become parental melts that differentiate to form the melt-dominant bodies of magmatic systems 1 and 2. (iv) Crystalline residuum associated with these fractionates forms protolith 2, which is remelted to form suite A2. (v) Suite M1 HRT mafic melts ascend through the crust and intersect and mix with magmatic system 1. Magmatic system 2 is located away from the focus of mafic influx. (vi) Spatially separated from magmatic systems 1 and 2, as well as within a region of un-modified Archean crust (Swallow et al., 2018b), melts from enriched (suite M1) and non-enriched (suite M2) source zones ascend from the uppermost mantle, stall in the middle crust, hybridise with surrounding Archean crust, crystallize and solidify. (vii) Partial melts of these hybridised cumulates ascend and feed magmatic systems 3 (suite C2) and 4 (suite C1). Crustal/mantle boundary depths from Huang et al. (2015). We therefore propose a two-stage evolutionary model for the least-evolved suite A1 compositions (Figs 24, 26). Mafic magmatism associated with the Yellowstone plume injected HRT mafic compositions as intrusions into the lower to mid crust. Melt losses from these cooling and crystallising intrusions (to form other, unsampled, intrusive or extrusive rocks) depleted the magmas in Nb, Rb and U and the remaining magma solidified to broadly trachydacitic (∼65 wt % SiO2), sanidine-, plagioclase-, pyroxene-, apatite- and zircon-bearing intrusions. Subsequent partial melting of these intrusions then generated a Ba- and Zr-poorer low-silica rhyolite melt. Relationships between the A2 and A1 suites. Relationships between suite A2 and the least-evolved suite A1 compositions are constrained by: (a) the compositional gap and offset in single clast and glass compositions between the two suites (Figs 8, 16, 17); and (b) the similar, but slightly less crustally influenced isotopic signatures and absence of an Eu anomaly in suite A2 samples (Figs 12–14). Three possible origins for the suite A2 magmas then arise: Thorough mixing between mafic material and the least-evolved suite A1 magmas (e.g. Streck & Grunder, 1999). This origin is unlikely because suite A2 samples show no physical evidence of mingling and their compositions are offset from a tie-line between the least-evolved high-silica rhyolite and the HRT mafics (e.g. P2O5, Ba: Figs 7, 8). As the low-silica rhyolite feedstock material for the main high-silica rhyolite body in the HRT (e.g. Allan et al., 2017), consistent with the coherency of suites A1 and A2 compositions, particularly for compatible elements (e.g. Ba, Zr: Fig. 8, Supplementary Data Fig. S7). However, crystal textures are variable within the suite A2 material. Sanidine crystals occasionally contain Ba-rich overgrowths and the compositions of rims are typically less-evolved than the cores (Figs 18, 19), inconsistent with growth as phenocrysts in an evolving melt-dominant magma body. Furthermore, the zircon model temperature estimates are offset to higher temperatures than two-feldspar model temperatures in suite A2 samples (Supplementary DataFig. S16), in contrast to the alignment between model estimates in other juvenile components that are inferred to be controlled by fractional crystallization. As such, any interpretation of the suite A2 magma type as a direct precursor to the suite A1 rhyolites is not favoured. As partial melts of the cumulate, or mush, related to the evolution of the A1 suite. The HRT suite A2 compositions extend up to, but not across, the trachyte-rhyolite boundary (Fig. 5). Evolution of the A1 suite was driven by a feldspar-rich crystallization assemblage (predominantly sanidine), as indicated by the continued increase in SiO2 throughout the sequence and the earlier onset of sanidine crystallization relative to quartz at all H2O contents in experimental studies on similar materials (Almeev et al., 2012). Any resulting mush will thus be sanidine-rich (i.e. syenitic), and less-evolved than both the initial magma composition and any residual melt. Re-melting of this crystal-rich residuum would therefore yield trachytic magmas (Wolff, 2017) but would require an influx of heat and/or volatiles, generally attributed to mafic magma (Deering et al., 2011; Bachmann et al., 2014; Wolff et al., 2015; Wolff, 2017), which is demonstrably present in both members A and B (Christiansen, 2001; Swallow et al., 2018b). Such a mafic influx would be consistent with the less-evolved, Ba-richer rims seen on suite A2 sanidines, and higher zircon saturation model temperatures. Therefore, we propose that magma with a composition intermediate between the suite A2 and least-evolved suite A1 materials, i.e. in the compositional gap, was formed through partial melting of an intermediate protolith (protolith 1: Fig. 26) and began to evolve through the crystallization of a feldspar-dominant assemblage (with accessory ferromagnesian minerals, apatite and zircon). The resultant melts, parental to suite A1, became enriched in LREE, Nb, and Rb, and depleted in Ba, Zr and Sr. The onset of chevkinite crystallization began to lower LREE concentrations in the subsequent melts (e.g. McCurry et al., 2008). Variable degrees of crystallization led to a heterogeneous feldspar-rich mush (protolith 2) from which were extracted multiple melt-dominant bodies that experienced further evolution. An increase in temperature due to an influx of mafic magma into the lower parts of the mush zone then led to partial melting of protolith 2 to yield the LREE-depleted suite A2 magma. This petrogenetic model differs from that proposed for Cedar Butte rhyolites (McCurry et al., 2008) due to the contrasting LREE-enriched low-silica rhyolite suite A1 samples relative to LREE-depleted suite A2 material in the HRT. Variations within suites A1 and A3. Glass compositions in the initial fall deposits and basal member A (Swallow et al., 2018a) cover the same compositional ranges as glasses in all analysed suites’ A1 and A3 pumices, implying that the first-erupted ignimbrite tapped all the range in suite A1 and A3 melts. The overall trend can be modelled by ∼60% crystallization from the least- to most-evolved compositions. Isotopic data, showing increasing 87Sr/86Sr with degrees of evolution (Fig. 14), indicate some assimilation. However, although single clast elemental and isotopic data show a simple and coherent trend, glass data are clustered and show two evolutionary trends, one between clusters and another, contrastingly inclined series of intra-cluster trends (Figs 16, 17). For these relationships to be controlled by assimilation alone, two assimilants affecting all the distinct melt-dominant bodies would thus need to be present. This is considered implausible: highly enriched assimilants would skew the isotopic data (Doe et al., 1982), and while less enriched silicic rocks occur in the region (e.g. Meen & Eggler, 1989), these are not plausible contaminants because the amounts required to cause the isotopic variations are not matched by major element variations. Compositional variations in suite A1 could also reflect mixing between a highly-evolved magma (i.e. suite A3) and a less-evolved end-member that realistically would be suite A2 or mafic magmas. Single clast and glass LREE (e.g. La) compositional trends preclude both scenarios, however, with the suite A1 compositions trending away from both possible end-members (Fig. 16, Supplementary Data Fig. S8). Overall, the variations within the samples from suites A1 and A3 are inferred to be predominantly driven by fractional crystallization. This inference is supported by the curved trend between Rb or Rb/Sr and SiO2 in type J1 clasts in member A, in contrast to the scattered but linear trend shown by member B type J4 mingled pumices (Supplementary Data Figs S6 and S20). Elemental and isotopic diversities within suites A1 and A3 reflect a mildly heterogeneous mush source, with variations in concurrent, dominant sanidine-controlled fractional crystallization and minor assimilation, generating the inter-cluster trends. Extraction and ascent of magmas from this source initiated a more plagioclase-dominant crystallization assemblage, plausibly through changing pressure, generating shallower-sourced trends within each suite (Swallow et al., 2018a). It could be considered that the compositions seen in suite A3 samples (magmatic system 2) represent further crystallization of the same parental melts as for system 1. However, the distinctive nature of crystalline materials from system 2 (e.g. sanidine core and rim compositions: Figs 18, 19), the absence of suite A3 compositions from member B, and their independent rejuvenation and eruption as suite C3 in member C (see discussion below) indicate that magmatic system 2 was physically separated from system 1 at the time of the eruption. System 2 was also separated from the focus of mafic inputs, due to the absence of any evidence for mixing or mingling between suite A3 pumices and mafic compositions. A physical separation of magmatic system 2 would also explain the offset to cooler temperatures (by ∼30°C: Supplementary Data Electronic Appendix 4) and the compositional gap between suite A3 and the evolved end of the A1 suite in some trace elements (Fig. 8, Supplementary Data Fig. S16). Magmatic systems 3 and 4 Whilst a highly enriched (crustal) component is implicated in the generation of magmatic systems 3 and 4, what is striking is the isotopic uniformity, but chemical dichotomy, between these systems (Figs 6, 9, 13, 14). Variations in Sr isotopic ratios, coupled with relatively uniform Nd and Pb isotopic ratios, have previously been related to lower Sr concentrations in YPVF rhyolitic melts and therefore greater susceptibility to isotopic modification by assimilation (Hildreth et al., 1991; Wolff et al., 2015). However, the compositional suites representing the two, member C systems have contrasting Sr concentrations (C1 = 230–320 ppm; C2 = 17–180 ppm) and therefore require separate consideration. Both systems, however, have similar but more radiogenic Sr and Pb and less radiogenic Nd compositions (Figs 13, 15) and positive trends between 87Sr/86Sr and Rb/Sr and 1/Sr (Fig. 14), collectively indicative of crustal contamination. Although 206Pb/204Pb and 207Pb/204Pb ratios are more radiogenic than those from magmatic systems 1 and 2, 208Pb/204Pb ratios are less radiogenic and negatively correlated with other Pb isotope ratios (Fig. 15). This contrast points to a crustal component with low initial Th/U being involved in the generation of both member C systems. This component, however, is inferred to be Archean, due to the 206Pb/204Pb vs 207Pb/204Pb slope defined by HRT silicic samples yielding a pseudo-isochron of 3·06 Ga (Supplementary DataFig. S21), similar to the 2·6–3·0 Ga pseudo-isochrons from the mafic suites analysed in Swallow et al. (2018b). Consequently, the assimilant requires Pb isotopic ratios of 206Pb/204Pb ≥17·9, 207Pb/204Pb ≥15·7 and 208Pb/204Pb ≤37·9. Published geochemical data for crustal materials from the Wyoming craton (e.g. Wooden & Mueller, 1988; Hildreth et al., 1991; Frost et al., 2006) show wide ranges and suggest that a plausible Archean assimilant with crust-like isotopic characteristics, low initial Th/U and with variable elemental chemistry could be found, but no suitable end-member composition has yet been reported. Magmatic system 3. Compositions from this system span a wide range (e.g. 410–2280 ppm Ba: Fig. 8) and form the distinctive suite C2 trend. Correlations between increasing 87Sr/86Sr values and degrees of evolution in the suite (e.g. increasing Rb/Sr, decreasing Ba: Fig. 14) indicate that assimilation was also important in generating the intra-suite variation. However, the curved Rb and Rb/Sr trends in Supplementary DataFigs S6 and S20 suggests that fractional crystallization was the dominant process for controlling intra-suite trend variation (as in magmatic systems 1 and 2). Overall decreasing Sr, Ba, Zr and Sc abundances within the whole suite attest to feldspar, zircon and pyroxene fractionation within the system (Figs 8, 9 and 11, Supplementary Data Fig. S7), consistent with the increasingly negative Eu anomalies with evolution in the type J8 pumice samples (Fig. 12). Clustered glass and single clast compositions from type J8 pumices that almost cover the entire range in suite C2 (410-2160 ppm Ba) indicate that system 3 (like systems 1 and 2) included multiple independent melt-dominant bodies (Figs 16, 17). The kinked trend in the glass Sr/Rb vs Ba/Rb ratios (Fig. 17) is inferred to reflect variations in the crystallising assemblage, with initial plagioclase-dominant followed by sanidine-dominant fractionation, as reflected qualitatively through percentages of picked crystals (Supplementary Data Electronic Appendix 2). Similarly, the intra-cluster trends also indicate continued compositional evolution following formation of the individual melt-dominant bodies. The parental melt for magmatic system 3 is notably elevated in Ba, Zr and Sr, and depleted in Nb and Rb relative to that for magmatic systems 1 and 2. From the isotopic systematics, we consider partial melting/assimilation of country rocks as a main petrogenetic mechanism. The low Nb and Rb concentrations of suite C2 materials require the influence of at least one residuum component, comprising a feldspar and zircon rich assemblage. The high Ba and Zr contents of suite C2 samples further implies that any mafic component is related to the HRT mafics rather than olivine tholeiite compositions. The interaction could be direct (i.e. mixing of HRT mafic melts and partial melts of a silicic cumulate) or indirect (i.e. mixing of partial melts from cumulates of both HRT mafics and a highly silicic assimilant). The Pb isotopic compositions of suite C2 samples also align with the fields of HRT mafics and members A and B. However, any melt compositions associated with the HRT mafic materials, either as magmas or partial melts of derivative cumulates, would realistically have ≤70 wt % SiO2, i.e. much less evolved than the ∼75 wt % SiO2 of the least-evolved suite C2 compositions. Simple binary mixing to generate the least-evolved high SiO2 trend magma would therefore require a highly silicic end-member that would also need to be enriched in Sr (∼250–350 ppm) and depleted in Na2O+K2O (∼6–7 wt %), Ba (∼1500–2500 ppm), Zr (∼250–500 ppm) and Nb (∼15–25 ppm) relative to the more primitive end-member. Given that documented shallow crustal rocks in the region extend to measured 87Sr/86Sr = 0·89 (Frost et al., 2006), a ∼0·2 increase in 87Sr/86Sr isotopic values between the HRT mafics and magmatic system 3 is feasible with this model. Although it is challenging to constrain mixing models with limited constraints on the end-members, this is a logical scenario for the generation of the least-evolved compositions of magmatic system 3. The presence of intergrown feldspars and disequilibrium textures (e.g. sieve textures) in feldspars in suite C2 samples also provides evidence for partial melting of a feldspathic cumulate (Fig. 20). Magmatic system 4. In sharp contrast to the other HRT silicic compositions, Suite C1 products of magmatic system 4, for many trace elements (e.g. Ba, Zr, La: Fig. 8, Supplementary Data Figs S7 and S8), are aligned towards the field of YPVF olivine tholeiites (Swallow et al., 2018b) and the newly analysed member B scoria samples from north of the caldera. However, suite C1 diverges from all mafic end-members in other elements, including Al2O3, alkalis (both lower than in mafic end members), Fe2O3, Sc and Sr (all higher), suggesting a broad relationship to olivine tholeiites, but not simple mixing between a tholeiitic melt and a high-silica component (Figs 5–11, Supplementary Data Figs S2 and S7). The high Fe2O3 and Sc and low Al2O3 contents are indicative of a dominant role for clinopyroxene in genesis of this suite. Cryptic clinopyroxene fractionation at >25 km depths has been inferred to occur in YPVF olivine tholeiites (Thompson, 1975; Hildreth et al., 1991). Consequently, there would be large quantities of pyroxene-rich cumulates in the mid-lower crust beneath Yellowstone related to the olivine tholeiite mafic root zone and represented in xenoliths of this assemblage found in lavas erupted just west of the HRT caldera (Leeman et al., 1985; Hildreth et al., 1991). The moderate Sr, low Al2O3 and lack of large positive Eu anomalies in the suite C1 compositions preclude large degrees of plagioclase melting, consistent with an apparent preferential melting of pyroxene over plagioclase in these cumulate assemblages (Leeman et al., 1985) and the pyroxene-dominant cumulates predicted at higher pressures (Thompson, 1975; Hildreth et al., 1991). The enriched isotopic signatures and elevated Fe, Ti and P in some tholeiitic Yellowstone basalts (e.g. Gerrit and Snake River tholeiites) has also been attributed to interaction with apatite- and pyroxene-rich cumulates (Hildreth et al., 1991). We, therefore, propose that melting of such cumulates by further influx of olivine tholeiitic magmas, coupled with assimilation of surrounding enriched Archean crustal materials, generated the suite C1 compositions defining magmatic system 4. The HRT magmatic complex The four magmatic systems defined above are collectively referred to what we here, for convenience, term the HRT magmatic complex, that fed the overall eruption sequence. Here we consider the relationships within and between these different systems and their melt-dominant bodies in terms of their intensive characteristics and spatial relationships (Figs 25, 26). Temperature and pressure/depth characteristics For magmatic system 1, temperature estimates are broadly similar to those estimated from the basal member A ignimbrite (Swallow et al., 2018a; Supplementary DataFig. S16). As such, we cannot resolve a systematic temperature gradient within the suite A1 samples. Suite A2 yields zircon saturation temperatures hotter than those in suite A1 samples, consistent with this being a thermally discrete magma body. The scoria-bearing pumices in member B contain a mode (∼900°C) intermediate between those of suites A1 and A2, consistent with suite B1 representing a mingling of those two suites. In contrast, magmatic system 2 (both suites A3 and C3) was 20–30°C cooler (Supplementary DataFig. S16), consistent with its physical separation from magmatic system 1 and implying that rejuvenation processes in the B-C time gap generated magmas with similar temperatures as well as compositional characteristics. Magmatic system 3 samples yield widely variable temperatures, reflecting wide variations in single clast and glass compositions (Fig. 16), and these are interpreted to reflect the presence of multiple melt-dominant bodies in this system. The temperature of magmatic system 4 is unknown, although its generally less-evolved compositions imply that it would have been hotter than system 3. The wide temperature ranges across magmatic systems 1–3 are similar to, but offset to higher values than the ∼100°C range suggested by Rivera et al. (2014) from Ti-in-zircon thermometry (using an assumed, constant value for αTiO2). Our combined estimates imply that there were significant thermal variations (>170°C) within the melt-dominant parts of magmatic systems 1 and 3 prior to evacuation, that magmatic system 2 was slightly cooler and less variable, and system 4 hotter. In contrast to temperatures, pressure estimates are more consistent across and within compositional suites (Supplementary DataFig. S17), ranging from ∼100–450 MPa within all suites (except A2), corresponding to ∼4 to 19 km depths. The former is consistent with the lowest pressures estimated from quartz-hosted melt inclusions in the initial fall deposits (Myers et al., 2016) and the latter would roughly match the base of the modern geophysically-imaged inferred silicic magma reservoir (Smith et al., 2009; Farrell et al., 2014; Huang et al., 2015). Taken together, available geobarometry estimates for all rhyolite suites, except A2, yield similar pressures, despite their contrasting compositions, implying that the separate magmatic systems and their melt-dominant bodies were laterally, rather than vertically, separated (Figs 25, 26). This inference is supported by the absence of a simple composition–depth relationship from melt inclusions (Myers et al., 2016), as would be expected in a zoned system (Wallace et al., 1999; Cashman & Giordano, 2014; Chamberlain et al., 2015), and the simultaneous eruption of multiple, compositionally diverse melt-dominant bodies (Myers et al., 2016; Swallow et al., 2018a). Suite A2 yields inferences of significantly hotter temperatures and higher pressures, consistent with its absence from the early member A ignimbrite. The ∼95 x 65 km mapped HRT caldera (Christiansen, 2001: Fig. 1) offers ample space for an areally extensive magmatic complex to have been present in the shallow crust prior to eruption. In addition, the great and sharply defined isotopic contrasts observed between magma systems 1 plus 2 vs 3 plus 4 (Figs 13–15, Supplementary Data Fig. S9) are implausible in a vertically stacked configuration of magmatic bodies (cf. Cashman et al., 2017) where continuity of isotopic variations would be expected through either pre-eruptive magmatic lineages or syn-eruptive mixing processes. Compositional characteristics and their syn-eruptive evolution Analysed clasts of types 1 and 2 from member A form a well-defined, coherent high-silica rhyolite trend represented by suites A1 and A3. Groundmass glass compositions from these clasts indicate, however, that despite relative uniformity in major elements there is significant trace element diversity (cf. Figs 16, 17, Supplementary Data Fig. S10–S12). The range in trace element compositions and the clustered nature of the data mirror the pattern seen in the fall deposits and basal member A ignimbrite (Myers et al., 2016; Swallow et al., 2018a). Glass compositional ranges in these early products indicate that the entire range in melt compositions seen in suites A1 and A3 was tapped at the onset of widespread member A deposition. The clustering of the compositional data, inferred to reflect multiple discrete melt-dominant domains, then continued through members A and B. The relative homogeneity of the glass within single pumices, however, implies that these melt-dominant domains remained separate during continued eruption of member A (Fig. 16). This feature then implies that the contrasting domains identified from glasses in the basal member A ignimbrite did not sit atop a unitary, zoned, melt-dominant body that was progressively tapped, but actually reflected almost the entire extent of the subsurface magmatic systems 1 and 2. Pumices spanning the compositional spectrum show comparable crystal abundances (Table 1), suggesting that all the melt-dominant bodies had broadly similar physical properties (i.e. density, viscosity, crystal content) during the eruption (cf. Tramontano et al., 2017). For example, magmatic system 2, dominant at the onset of fall deposition and reappearing in the basal ignimbrite, is also represented by some pumices from member A (suite A3: e.g. YP255 [Table 2], YP416: Supplementary Data Electronic Appendix 3), as shown by the similarity in their groundmass glass compositions to clusters FGC/OGC1 + 2 and their clustered sanidine rim compositions with a low-Cn sanidine rim peak, characteristic of the fall deposit (Figs 16, 19; Swallow et al., 2018a). The only compositional suite not represented at the start of member A ignimbrite is suite A2. Although the A2 samples plot on a continuation of the A1 trend with respect to single clast compositions and glass elemental ratios, their groundmass glass concentration data lie off this trend (Figs 16, 17). On the sampling scale used for this study, this compositional separation is consistent with the presence of a discrete, hotter body containing suite A2 magmas within the deeper parts of magmatic system 1. Significant changes occurred within the HRT magmatic complex between the eruption of the different ignimbrite members. In member B, although suite A1 is largely replicated as suite B1, there are significant differences. First, magmatic system 2 was not tapped, as indicated by the absence of low-Ba pumice and glass, and lack of a significant mode in low-Cn sanidine rims (Figs 16, 19). The absence of evidence for mixing of HRT mafic material with magmatic system 2 also implies that this system was separate from magmatic system 1. Second, the compositional gap between suites A1 and A2 is infilled by the silicic component of scoria-bearing (type J4) pumices in member B from localities southwest of the caldera (e.g. Fig. 8). This infilling is also observed in the widely varying groundmass glass compositions, with type J4 pumice glass linking the least-evolved glass compositions of suites A1 (and non-mingled pumices in suite B1) to the field of suite A2 clasts (Fig. 16). The antithetic nature of this trend relative to suite A1 indicates that the filling in of the compositional gap is due to mixing between the less-evolved suite A1 melt and a Ba-enriched melt similar to that in the suite A2 magma. The mixing is not directly with HRT mafic scoria, due to the trend of mixed pumices to low LREE contents (Supplementary DataFig. S8). The bimodality in olivine compositions from J4-type pumices at sites southwest of the caldera also suggests the occurrence of mixing between a melt that contributed to the non-mingled J1-type clasts and a less-evolved component (Supplementary Data Fig. S14). All of these mixing processes are inferred to have occurred in magmatic system 1 in the time gap between eruption of members A and B. The member B scoria (Swallow et al., 2018b) indicates a renewed influx and increased degree of mafic mixing in member B relative to member A. Because the rhyolitic components of individual scoria-bearing pumices cover in composition the entire range of suite B1, mixing of the scoria component was widespread within magmatic system 1 and not limited to less-evolved compositions. However, and unexpectedly, scoria clasts that were not included in the samples documented by Swallow et al. (2018b) from member B ignimbrite north of the caldera have compositions that trend towards tholeiitic in several elements (particularly P2O5, Ba and Zr: Figs 7 and 8, Supplementary Data Fig. S7, and Supplementary Data Electronic Appendix 3). There were thus compositionally diverse mafic influxes into magma system 1 prior to and during eruption of member B. Further work is required to identify whether there is a contrasting ‘tholeiitic’ imprint on the rhyolitic clast compositions from member B north of the caldera, but material suitable for single clast analysis has not yet been found. The compositional record from member C is in marked contrast to that from members A and B. Following the years to decades time break between members B and C (Fig. 2, Supplementary Data Fig. S1), there was venting of magmatic systems 3 and 4, renewed venting of magmatic system 2, yet an absence of any magmatic system 1 material. With system 2, although glass LREE contents are similar between the relevant suites in members A and C, pumices show contrasting abundances of LREE in single clast groundmass glass. Suite A3 pumices have LREE depletions (i.e. glass/clast <1) whereas suite C3 pumices have enriched LREE contents (i.e. glass/clast >1: Supplementary DataFig. S22). The difference is inferred to reflect chevkinite abundances, being 9% of the picked ferromagnesian minerals in A3 compared to 3% in C3 (Supplementary Data Electronic Appendix 2). As temperature, pressure and H2O content estimates are comparable between suites A3 and C3, variations in chevkinite abundance may, therefore, be related to time constraints. Glass clusters from the fall deposits (related to magmatic system 2) show bimodal La concentrations (Swallow et al., 2018a), suggesting that the degree of LREE depletion is not uniform within this system and is more pronounced in the most evolved melts. This observation implies that chevkinite crystallization is primarily a late-stage feature, and that the rapid extraction of the melt for suite C3 from magmatic system 2 prior to eruption of member C precluded significant chevkinite fractionation. Furthermore, the less radiogenic (with respect to Sr) nature of suite C3 versus A3 pumices (Figs 13, 14) suggests also that rapid melt extraction hindered the extent of crustal contamination and therefore limited the Sr isotopic values. In suite C3 compared to suite A3 pumices, although sanidine compositions extend to hotter, more Cn-rich values (Fig. 19), these are predominantly in core compositions and crystals therefore record an overall down-temperature history. Furthermore, up-temperature signals are not observed in other crystal phases (e.g. Fig. 22), model temperature estimates are comparable between suites A3 and C3 (Supplementary DataFig. S16), and system 2 is inferred to lie away from any foci of mafic influx. The rejuvenation of magmatic system 2 during eruption of member C is thus inferred to reflect factors other than mafic influxes. Magmatic systems 3 and 4, with their unique trace element and crustal isotopic characteristics, are remarkably distinct from each other and magmatic system 1. We therefore link the highly enriched crustal isotopic signatures in member C material reported by Doe et al. (1982) and Hildreth et al. (1991) from their associated SiO2 contents to our suite C2 from magmatic system 3. These signatures were attributed (Hildreth et al., 1991; Christiansen, 2001) to syn-eruptive melting of the foundering magma chamber roof following caldera collapse associated with eruption of members A and B. In contrast, our data show that suite C2 is unrelated to rhyolitic melts from members A and B and derives from an independent magmatic system. Although magmatic system 3 shows a large compositional range (e.g. Fig. 8), compositions are clustered, particularly of groundmass glass (Figs 16, 17). This feature suggests that magmatic system 3 (like systems 1 and 2) contained multiple melt-dominant bodies that were, based on unimodal pressure estimates (Supplementary DataFig. S17), predominantly laterally separated. Despite the similar isotopic compositions (Figs 13-15) of magmatic systems 3 and 4, there is no evidence of significant mixing between them, suggesting a physical separation of the two. However, isotopic data equally imply that they have a shared petrogenetic pathway that is distinct from systems 1 and 2. The lower-silica nature of devitrified samples from system 4, if pressure estimates from groundmass glass from low-silica suite A2 are used for comparison (Supplementary DataFig. S17), implies that melt-dominant material in system 4 was stored at greater depths than that for system 3. Despite their inferred existence during eruption of members A and B, magmatic systems 3 and 4 did not vent at this time. However, the notably smaller crystal sizes (≥85% of crystals <1 mm) in clasts from these systems relative to those from magmatic system 1 (up to 4–8 mm: Table 1), suggests that all the member C melt-dominant bodies were generated during the B–C time gap, but from systems that had longer histories. The bi-pyramidal quartz in clasts from system 3 suites can be interpreted to reflect a pulse of accelerated crystallization shortly prior to eruption, as prolonged growth tends to result in crystals that have contrasting shapes (Laporte & Provost, 1994; Hosaka et al., 1995; Manley, 1996). Euhedral, bi-pyramidal quartz has been experimentally reproduced by moderate undercooling of hydrous silicate melts (Swanson & Fenn, 1986), suggesting that rapid extraction of magma from the root zone to shallow levels led to the onset of quartz crystallization. Published quartz growth rates support this inference, with 1 mm quartz crystals forming in ∼3 ka (using rim growth rates of 10-13 m/s: Gualda et al., 2012) to ∼60 days (using growth rates of 2 x 10-10 m/s: Swanson & Fenn, 1986), spanning the years to decades estimate for the B–C time break. The HRT magmatic complex was thus an extraordinary entity, covering four compositionally and isotopically distinct magmatic systems, each of which had a unique combination of factors controlling its erupted compositions. Magmatic systems 1 and 2 were hosted in crust that was less enriched, inferred to reflect large-scale modification during Eocene (and earlier) Absaroka-type magmatism. Magmatic systems 3 and 4 were hosted in highly enriched Archean crust, and a major crustal boundary must therefore lie across the overall HRT caldera area. The HRT mafic magmas with their notably high concentrations of LILE and HFSE (e.g. Ba and Zr; Fig. 8, Supplementary Data Fig. S7) acted, through their earlier crustal intrusions and cumulates, to control assimilation/fractionation trends in magmatic systems 1, 2 and 3, but only directly interacted syn-eruptively with magmatic system 1. In contrast, magmatic system 4 shows an influence from sources that are closer to the regional Snake River Plain olivine tholeiites in composition. Unexpectedly also, scoria with tholeiitic affinities has also been identified as a component in member B from its northern sector, and it appears that the two contrasting mafic lineages both interacted syn-eruptively into high levels of the re-organised magmatic system 1. Thermal and chemical drivers of the HRT magmatic complex The ultimate thermal and chemical drivers of the HRT silicic magmatic complex, as in the YPVF to the present day, are mafic magmas feeding and forming the root zone to the overall area (Hildreth, 1981; Hildreth et al., 1991; Christiansen, 2001; Huang et al., 2015). Most co-erupted mafic enclaves in members A and B (suite 1 HRT mafics of Swallow et al., 2018b; labelled suite M1 here) are related to Craters of the Moon-type eruptives seen elsewhere in the Snake River Plain (Leeman et al., 1976; Christiansen & McCurry, 2008; Putirka et al., 2009). The HRT suite M1 compositions owe their distinctive features to an origin in lithospheric mantle enriched by high P–T fluids emanating from the subducted Farallon slab (Swallow et al., 2018b). In contrast to earlier views (Christiansen, 2001; Christiansen & McCurry, 2008), we infer that the great majority of erupted silicic material in the HRT (from magmatic systems 1-3) is related to these HRT mafics, rather than olivine tholeiitic compositions (Suite 2 of Swallow et al., 2018b; suite M2 here: Fig. 26). The latter are petrogenetically linked to and are inferred to have been the driver to magmatic system 4, and form a minor component, opportunistically syn-eruptively vented, invading part of magmatic system 1. Both mafic root zones were simultaneously active prior to and during the HRT eruption, reflected in the pre-HRT olivine tholeiitic Junction Butte Basalt (Christiansen, 2001), the presence of suite M1 compositions in members A and B, compositions of both mafic types in member B, and co-eruption of magmatic systems 3 and 4 that show separate affinities to both mafic suites. This contemporaneity is paralleled by work reporting Craters of the Moon-type lavas sandwiched in olivine tholeiite lavas along the Snake River Plain as early as ∼3·7 Ma (Potter et al., 2018). The presence of mafic material in members A and B, and elevated temperatures derived from type J4 scoria-bearing pumices (Supplementary DataFig. S16), support a major thermal role for mafic magma in the HRT system. The distinct stratigraphic occurrence and vesicular nature of scoria in member B, relative to dense mafics in member A, suggest a renewed mafic influx associated with eruption of member B. Although a direct mafic triggering of the member B eruption is not established here, the mixing between melt-dominant bodies (suites A1 and A2 to yield suite B1) in the time break between members A and B may reflect a renewed influx of mafic magma into the silicic part of magmatic system 1. Furthermore, an earlier influx of mafic magma likely led to remelting of sanidine-bearing cumulates and the generation of suite A2 compositions. Although there is no physical evidence of mafic magma influx in member C, the presence of less evolved sanidine compositions in rims relative to cores in samples from magmatic system 3 is consistent with a late-stage thermal input during formation of the melt-dominant bodies. Rhyolite petrogenesis at Yellowstone Recent debate around the dominant mechanism of rhyolite petrogenesis in the YPVF, involves end-member views suggesting crustal partial melting (Bindeman et al., 2008; Loewen & Bindeman, 2015) and fractional crystallization from a long-lived crystal mush zone (Girard & Stix, 2009, 2010; Stelten et al., 2015, 2017). Our data indicate that systems conforming to both end-member views were present within the HRT magmatic complex. The diversity of compositions within the HRT, particularly isotopically, is not only unique to the YPVF (Hildreth et al., 1984, 1991; Christiansen, 2001) but also in regard to the Snake River Plain as a whole. Large-scale isotopic studies of YSRP volcanic products indicate that although there is an increased crustal signature linked to the volcanic focus traversing the western edge of the North American craton, glass isotopic signatures in silicic tephra erupted through cratonic crust are moderate (εNd ≥ -11: Nash et al., 2006) in contrast to the HRT magmatic systems 3 and 4 (εNd = −17·3 to −18·2: Fig. 13). Other rhyolitic pyroclastic products from the Snake River Plain with similar crustal isotopic signatures to the HRT systems, e.g. the Arbon Valley Tuff (Nash et al., 2006; Drew et al., 2013, 2016), have been linked to the commencement of activity at a new centre (Colón et al., 2018). This situation is interpreted as representing either crustally-dominated magmatism prior to the full impingement of a mantle plume, or that the ‘crustal’ magmatic system was isolated from subsequent more mantle-like isotopic systems (Drew et al., 2013). The chemical and isotopic diversity in the HRT show that the latter situation was the case. As the isotopic signatures of the sources to Yellowstone magmas are thought to result from hybridization of Archean crust, mantle-derived melts and mafic intrusives (Doe et al., 1982; Hildreth et al., 1991; Swallow et al., 2018b), more enriched isotopic signatures therefore logically indicate a greater crustal component. With continued mafic influx, this crustal component will be diluted and result in a trend towards less extreme isotopic compositions, as seen in the YPVF (Hildreth et al., 1984, 1991) and farther west in the Snake River Plain (Drew et al., 2013, 2016). However, the bimodality in isotopic compositions within the HRT indicates that in this case at least, dilution of the ‘crustal’ signature was a dominantly spatial rather than a temporal feature. Eocene magmatism of the Absaroka Volcanic Province overlaps with the Yellowstone area (Chadwick, 1970; Christiansen & Yeats, 1992; Feeley, 2003) and would have significantly hybridised and ‘pre-conditioned’ the lithosphere beneath for the onset of Quaternary hotspot-related volcanism (Swallow et al., 2018b). The isotopic contrasts between magmatic systems 1 + 2 vs 3 + 4 are thus interpreted to reflect the contrasts between these systems being lodged in crust that was affected by Eocene magmatism, versus crust that was not, respectively. Others have also suggested a complex nature to rhyolite petrogenesis, not only in the HRT but in the YPVF as a whole (e.g. Bindeman et al., 2008; Watts et al., 2012; Wotzlaw et al., 2015). These studies suggest that the melt-dominant bodies that have fed volcanism in the YPVF are heterogeneous and either represent discrete long-lived melt-dominant bodies, or the late-stage and rapid accumulation of multiple melt-dominant bodies. We demonstrate here, with a single clast-focused approach, that not only were multiple melt-dominant bodies present to feed the HRT eruption, but also multiple magmatic systems. Wotzlaw et al. (2015) report a bimodality of zircon Hf isotopic compositions in member A whole-rock samples, with ‘more crustal’ and ‘more mantle-like’ populations. The ‘more crustal’ population is absent in member B but reappears in member C, in contrast to the ‘mantle-like population’ that shows the reverse. The appearance of a distinct zircon Hf isotopic population in members A and C, but not B, is parallel to our results on the low-Ba magmatic system 2 which, however, is not strongly crustal in its single clast isotopic signature. Overview of the Yellowstone magmatic systems YPVF silicic volcanism has been categorised as a ‘hot-dry’ system (Loewen & Bindeman, 2016). Although this label is archetypically linked to the pumice-poor, intensely welded ignimbrites of the Snake River Plain, with magmas inferred to be >900°C and H2O-poor (Branney et al., 2008; Ellis et al., 2013), explosive eruption deposits in the YPVF are somewhat different and contain more abundant pumice (Christiansen, 2001). Although the HRT silicic magmas cover a large overall temperature range (e.g. zircon saturation temperatures of 790–960°C), they were cooler and H2O-richer (up to 5 wt %: Myers et al., 2016; this study) than the earlier Snake River Plain magmas. The HRT rhyolitic magmas are in turn hotter than recent YPVF lava-dominated volcanism, which yields temperature estimates as low as 720°C, depending on the proxy and calibration used (e.g. Watts et al., 2012; Loewen & Bindeman, 2015; Stelten et al., 2015; Befus & Gardner, 2016). Two-feldspar thermometry estimates of >900°C and the presence of pigeonite compositions as cores or inclusions also provide indications of a hotter and H2O-poorer beginning to the HRT magmatic complex before its evolution to a cooler and H2O-richer final state. These transitional characteristics, therefore, mean that the HRT straddles the division between ‘hot-dry’ and ‘cold-wet’ types of silicic magmas set by Loewen & Bindeman (2016) at ∼825°C. One could therefore propose that the HRT magmas be classified as ‘tepid-moist’ on the basis of such an arbitrary division. Heterogeneity is present in YPVF eruptive products on multiple scales, from crystal populations within a single eruption (e.g. Bindeman et al., 2008; Ellis & Wolff, 2012; Watts et al., 2012; Wotzlaw et al., 2015) to variable compositions across an eruptive episode (e.g. Girard & Stix, 2009, 2010; Loewen & Bindeman, 2015; Stelten et al., 2015; Troch et al., 2017), both related to heterogeneity within a single magmatic system. The perception of a single magmatic system has fitted in conveniently with the modern, unitary low-velocity zone (i.e. ‘magma body’) geophysically imaged beneath the YPVF (Smith et al., 2009; Farrell et al., 2014; Huang et al., 2015). Although broad-scale imaging, with a resolution of ∼10 km obscuring finer details (e.g. Lowenstern et al., 2006), can identify the broad dimensions of large-scale crustal anomalies, analysis of past eruption deposits sheds light on the nature of possible heterogeneities within the ‘magma body.’ Post-dating the HRT, eruptives of analytically indistinguishable ages, but contrasting magma sources, are known, including the domes post-dating the ∼1·3 Ma Mesa Falls Tuff (Troch et al., 2017; Stelten et al., 2018; Rivera et al., 2018), and the two magma types in the 0·63 Ma Lava Creek Tuff (Christiansen, 2001). Even so, the presence of four independent magmatic systems contemporaneously active within the YPVF and having these venting in one (albeit punctuated) eruption is remarkable. Furthermore, the HRT magmatic systems included multiple melt-dominant bodies that did not simply merge downwards into a unitary magma chamber, but retained their distinctive natures deep into the magmatic system. The presence of focused zones of eruptible melt-rich magma within a larger magmatic system is important in consideration of future eruptions and hazard modelling (Christiansen et al., 2007). This is because although the current percentage of melt in the YPVF upper crustal magmatic system is thought to be too low to be eruptible (Smith et al., 2009; Farrell et al., 2014; Huang et al., 2015), this melt could be concentrated in regions too small to be detected by present imaging techniques. The evacuation of multiple melt-dominant bodies in pre-Quaternary YSRP eruptions appears to be common (Perkins et al., 1995; Cathey & Nash, 2004; Bonnichsen et al., 2008; Ellis & Wolff, 2012). Although there is a lack of single-clast data (Branney et al., 2008), the presence of multiple modes in crystal and glass compositions is parallel to what is seen in the HRT. But why multiple melt-dominant bodies should form and remain molten in the crust prior to eruption is puzzling. Small melt-dominant bodies should solidify quicker than large bodies and therefore be transient features or be kept melt-dominant only through continuing inputs of heat and/or volatiles. Although volcanism in the Snake River Plain is high temperature (>900°C: Branney et al., 2008), and HRT temperature estimates extend to similar values, estimated final storage temperature for the HRT magmas are moderate (∼780–900°C). However, the modern thermal and gas fluxes in the YPVF reflect a large magma flux from the mantle and lower crust, related to the Yellowstone plume (e.g. Lowenstern et al., 2006, 2014; Huang et al., 2015). If this abundant thermal flux pre-dated the surface volcanism then it could have thermally pre-conditioned the crust to allow small melt-dominant bodies to survive in the upper crust. Another consideration is the tectonic regime along the YRSP related to Basin and Range extension (e.g. Pierce & Morgan, 1992, 2009; Smith et al., 2009), as multiple co-existing melt-dominant rhyolite bodies have been reported from rifting areas (e.g. Taupo Volcanic Zone: Shane et al., 2007, 2008; Cooper et al., 2012; Bégué et al., 2014). Extensional tectonics would likely enhance the geothermal gradient in such settings and aid the ascent of magma and rapid formation of melt-dominant bodies (e.g. Allan et al., 2013, 2017). It is inferred that a combination of elevated heat flow and extensional tectonics in the early stages of development of the YPVF helped develop the unusually heterogeneous magmatic complex that is represented in the HRT. Implications for global silicic magmatic systems Although focused on a single caldera-forming eruption, this study raises several general issues around silicic eruptions and their source magmatic systems. The HRT, previously considered to represent one eruption (Reynolds, 1977; Hildreth et al., 1991; Christiansen, 2001), raises questions surrounding the distinction between one, multi-phase event and multiple distinct eruptions. Were three ignimbrite caldera-forming eruptions to occur today, each separated by periods of months to decades, and exhibiting compositional differences between each eruption, they would likely be considered distinct, at least on human timescales. Evidence for time breaks within eruptive sequences is therefore crucial in providing a framework for geochemical studies and providing a more realistic representation of the eruption cycle. Although this is increasingly challenging further back in the geological record, eruptions separated by similar time gaps have been identified as distinct events (e.g. Barker et al., 2016; Cooper et al., 2017). The HRT is another example, albeit extreme in size and complexity, of a silicic eruption tapping multiple melt-dominant bodies (cf. Sampson & Cameron, 1987; Shane et al., 2007, 2008; Cooper et al., 2012, 2016; Westgate et al., 2013; Bégué et al., 2014). In contrast, numerical models associated with large silicic magmatic systems do not reflect this complexity and are based on single unitary bodies (e.g. Jellinek & DePaolo, 2003; Gregg et al., 2012, 2015; Caricchi et al., 2014). The complex configurations of silicic magmatic systems indicated by field-based studies would have important, and variable, thermomechanical effects on the magmatic system (Gudmundsson, 2012), and it is arguably important to narrow the gap between field-based evidence and numerical model assumptions. There was rapid (syn-eruptive) accumulation of melt-dominant bodies in the HRT, with changes from members A to B within months and B to C within decades. Such rapid behaviour has been inferred elsewhere (e.g. Oruanui: Charlier et al., 2008, Allan et al., 2013, 2017; Rocky Hill/Kidnappers: Cooper et al., 2016, 2017; Santorini: Druitt et al., 2012). However, this rapidity is in contrast to the assumption that large magma bodies take 105–106 years to grow and that their growth to a given size controls the repose period prior to eruption (e.g. Smith, 1979; Jellinek & DePaolo, 2003; Reid, 2008). This study, as with others (e.g. Druitt et al., 2012; Till et al., 2015; Barker et al., 2016; Allan et al., 2017) indicates that large-scale silicic magmatic systems can rejuvenate or create eruptible melt-dominant bodies on short timescales, such as within the years-decades time break between members B and C for magmatic systems 2 to 4. Not only is the presence of multiple independent melt-dominant bodies in the HRT notable, but so also is the presence of multiple independent magmatic systems. The simultaneous or systematic construction and evacuation of magma bodies related to spatially and/or chemically distinct magmatic lineages has been documented elsewhere (e.g. Long Valley: Sampson & Cameron, 1987; Taupo Volcanic Zone: Brown et al., 1998; Shane et al., 2007, 2008; Allan et al., 2012, 2017; Bégué et al., 2014; Campi Flegrei: Pistolesi et al., 2016), but not on such a large scale. On a broader spatial and temporal scale, periods of intense silicic volcanism comprising multiple caldera-forming eruptions over geologically short periods of time are labelled ignimbrite flare-ups (e.g. Bonnichsen et al., 2008; Best et al., 2013; Lipman & Bachmann, 2015; Gravley et al., 2016). In the Taupo Volcanic Zone, eight caldera-forming eruptions, evacuating 2–3000 km3 dense-rock equivalent of material, occurred within an area of 90 x 40 km over a ∼100 kyr period (Houghton et al., 1995; Gravley et al., 2016). In comparison, the HRT erupted ∼2500 km3 dense-rock equivalent of material and formed a caldera of ∼95 x 65 km (Christiansen, 2001) that encloses the entire collapse area of the TVZ examples. The HRT caldera thus covers an area as large or larger, than other terrestrial volcanic regions that contained multiple magmatic systems that, in the example quoted, were evacuated mostly as singular eruptions separated by periods long enough for erosion and soil formation. CONCLUSIONS The 2·08 Ma, ∼2500 km3 HRT comprises three ignimbrite members (A, B and C) and two sets of fall deposits, an initial one beneath member A and another beneath member C (Christiansen, 2001). The eruption was episodic and prolonged, with spasmodic initial fall activity over several weeks, and with weeks- to months-long time break between members A and B, and a years- to decades-long interlude between members B and C (Fig. 2, Supplementary Data Fig. S1). The eruption products include silicic juvenile material exhibiting a wide diversity of physical characteristics, with variations in the size and abundance of crystals, nature/degree of vesicularity, growth of microlites and mixing with mafic scoria. Crystal-rich (11–22 wt %), vesicular pumices and their devitrified fiamme equivalents (clast type J1) dominate in the sample suite from members A and B, where a coarse-grained (up to 8 mm) crystal assemblage is ubiquitous. Scoria-bearing, mingled pumices and fiamme (type J4) are present in member B. Also present in these members are two types (types J2, J3) of less vesicular clasts with darkened groundmass glass, with type J3 containing characteristic coarse euhedral feldspars. Member C juvenile clast types (types J5 to J9) are uniformly distinct for their consistently small crystal sizes (<3 mm) and crystal-poorer natures (<15 wt %). In contrast, however, member C clast types range in physical characteristics from dense black material (J5) to highly vesicular pumices (J8) and their devitrified fiamme equivalents (J9). The physical diversity of juvenile clasts in the HRT is mirrored in a remarkable degree of geochemical diversity within and between the ignimbrite members. This diversity reveals that the magmatic complex that gave rise to the HRT was extraordinary. Taken together, single clast major and trace element, and isotopic compositions, coupled with matrix glass and mineral compositions from samples constrained within a chronostratigraphically controlled framework show the following: Four magmatic systems, with distinct chemical and radiogenic isotopic signatures (Table 4), were involved during the HRT eruption. From the eruption onset, the initial fall deposits (Myers et al., 2016; Swallow et al., 2018a) and earliest member A ignimbrite evacuated two systems, the major magmatic system 1 and the minor low-Ba system 2. The former includes the dominant A1 suite of compositions, separated by a compositional gap from the low-silica A2 suite, which is notably elevated in Ba and Zr. Suite A1 shows wide variations (e.g. 450–1680 ppm Ba) but is separated from suite A3 from magmatic system 2 based on trace elements (e.g. 140–250 ppm Ba) and characterised by low-Cn (<0·35 mol %) sanidine rim compositions. Member B represents the evacuation of only magmatic system 1, with a cessation in evacuation of system 2. Mixing throughout system 1 occurred prior to the eruption of member B, indicated by the closing of the compositional gap in member A (suites A1 and A2 condensed into suite B1), the presence of mixed, scoria-bearing pumices, and greater mineral and glass compositional variations within single clasts relative to member A. The mafic end-member for this mixing event mostly belonged to the distinctive HRT mafics suite of Swallow et al. (2018b) (part of suite M1 here), consistent with its long-term role in the petrogenetic development of magmatic system 1. However, analysed scoria from north of the caldera has tholeiitic affinities, and was an apparently opportunistic infiltrator into the silicic magma body during eruption of member B (and subsequently: Pritchard et al., 2013). The coherency of crystal compositions between members A and B indicate that the mafic mixing was rapid enough such that there was no development of contrasting overgrowths on crystals (cf. Wark et al., 2007; Chamberlain et al., 2014b). Member C contains material from two new, compositionally distinct magmatic systems, but with similar, crustally-influenced isotopic compositions: the larger magmatic system 3 and the smaller system 4. System 3 generated high-silica rhyolite magmas that exhibit large trace element compositional variations (e.g. 410–2280 ppm Ba). System 4 generated a range of dacitic to rhyolitic compositions (65·6–75·9 wt % SiO2), with many compositional trends pointing towards broadly tholeiitic sources, rather than the ‘HRT mafics’ source. Member C also involved the reappearance of newly-generated magma from system 2 (suite C3) but, conversely, material from magmatic system 1 is absent. Although previous studies have highlighted heterogeneity of the HRT in its isotopic characteristics (e.g. Doe et al., 1982; Hildreth et al., 1991) and crystal cargo (e.g. Wotzlaw et al., 2015), the details presented here highlight the advantages of analysing single pyroclasts and their components. Elemental and isotopic compositions of products from the different magma systems suggest that a variety of petrogenetic mechanisms operated prior to the HRT eruption. Lead isotopic constraints, coupled with the high-Ba and Zr nature of magmatic system 1, indicate that this and system 2 are petrogenetically related to the co-erupted HRT mafic clasts (Swallow et al., 2018b). Moderate offsets to more crustally influenced isotopic compositions in magmatic systems 1 and 2 relative to HRT mafic compositions suggests a control on rhyolite compositions dominantly by fractional crystallization with only minor assimilation. Compositional variations and gaps within systems 1 and 2 can be explained through a multi-stage partial melting scenario, with the erupted high-silica rhyolites (e.g. suite A1) extracted from a sanidine-rich mush which was in turn re-melted (e.g. Wolff, 2017) to generate suite A2 compositions. In contrast, magmatic systems 3 and 4 in member C contain significantly more enriched isotopic compositions, consistent with greater degrees of assimilation of Archean crustal rocks (Figs 13, 26). Whilst the high-Ba nature of magmatic system 3 indicates a continuing role for HRT mafic compositions, in contrast the low-Ba, high-Sc nature of system 4 links it to a pyroxene-bearing olivine tholeiitic root zone. Despite the contrasting elemental compositions between these two member C suites, their isotopic compositions are comparable and can be similarly explained through crustal contamination of partial melts of mafic cumulates, derived from contrasting mafic lineages (system 3-HRT mafics; system 4-olivine tholeiites). These newly-erupted member C systems, coupled with the rejuvenation of the fractional crystallization-controlled magmatic system 2, illustrate the wide diversity in petrogenetic mechanisms represented in the HRT. This variability suggests that rhyolite petrogenesis in the YPVF is complex and diverse, and may be more complex than considered in other end-member models (cf. Loewen & Bindeman, 2015; Stelten et al., 2015). Groundmass glass compositions from magmatic systems 1-3 (magmatic system 4 is only represented by devitrified material) show clustering, indicative of the presence of multiple, distinct melt-dominant bodies. As this clustering is consistent in the fall deposits (Myers et al., 2016; Swallow et al., 2018a) and the ignimbrite, this configuration in the relevant magmatic system was not confined to the upper (earlier-erupted) levels but persisted throughout evacuation of the systems. Broadly comparable estimates of storage pressures/depths from volatile contents in melt inclusions (Myers et al., 2016) and glass major element compositions (Supplementary DataFig. S17) imply that these bodies were predominantly laterally rather than vertically separated (Figs 25, 26). This arrangement appears to be a common feature of large silicic caldera-forming eruptions (e.g. Cooper et al., 2012, 2016; Ellis & Wolff, 2012; Bégué et al., 2014). However, this complex, multiple-body nature of magmatic systems is not incorporated into numerical models of magma bodies, leading to a gap that requires filling between investigations based on detailed field study and model frameworks, which require simplicity. Syn-eruptive magmatic processes within the HRT were rapid. Within the inferred weeks to months-long time break between members A and B, mixing related to renewed influx of HRT mafic material occurred, closing the compositional gap present in member A between suites A1 and A2. Furthermore, the absence of material derived from magmatic systems 3 and 4 in members A and B suggests that melt-dominant bodies in these systems were either not present, or were not in a state to be tapped at this stage of the eruption. This observation, at one extreme, constrains rapid formation of these melt-dominant bodies to in the years-decades time break between members B and C, further linked by the uniformly small (<3 mm) crystal size across all member C juvenile material. This time break also saw rejuvenation of magmatic system 2. Whilst further mineral-specific studies are required to better constrain timescales of melt-dominant body formation in the HRT, our inferences on the rapidity with which the HRT systems behaved are consistent with findings from other large-scale silicic systems (e.g. Charlier et al., 2008; Allan et al., 2013, 2017; Cooper et al., 2016, 2017). Temperature estimates from erupted material in the HRT show a broad range of values, indicating temporal and spatial heterogeneity within the HRT magmatic complex. These estimates (Supplementary DataFig. S16) yield temperatures of ∼780–900°C that lie between the higher temperatures associated with most Snake River Plain silicic volcanism (>900°C: e.g. Branney et al., 2008) and lower temperatures inferred from the youngest episodes of volcanism in the YPVF (<800°C: e.g. Loewen & Bindeman, 2015; Stelten et al., 2015; Befus & Gardner, 2016). Whether this represents a cooling overall magmatic system (e.g. Watts et al., 2012) or distinct differences between the various volcanic centres is not clearly established. However, incorporation of the YPVF into the ‘hot-dry’ rhyolite category associated with the Snake River Plain appears overly simplistic (e.g. Loewen & Bindeman, 2016), and Yellowstone silicic products in general and the HRT eruption products in particular defy simple pigeonholing. The mapped HRT caldera and total eruptive volume (Christiansen, 2001) are comparable in size to regions in other silicic volcanic provinces that experienced clusters of caldera-forming eruptions (‘flare-ups’) from multiple volcanic centres over periods of ∼105 years (e.g. Taupo Volcanic Zone: Houghton et al., 1995; Gravley et al., 2016). The HRT could be considered some ways as a ‘flare-up’, but on a compressed timescale, with the effectively collective (rather than separated) eruption of multiple independent magmatic systems. Subsequent volcanism in the YPVF shows evidence for complex mineral assemblages and stratigraphic relationships (e.g. Christiansen, 2001; Troch et al., 2017; Stelten et al., 2018; Wilson et al., 2018). Our results suggest that other eruptive units in the YPVF deserve closer scrutiny to determine if the situation for the HRT is the norm or exception for large eruptions at Yellowstone. Acknowledgments C. J. N. Wilson thanks R. L. Christiansen for an introduction to the Huckleberry Ridge Tuff, and the Yellowstone (YELL-05248) and Grand Teton (GRTE-00604) research offices for permission to work in the respective national parks, and the help and support of their staff over many years. Discussions on Yellowstone matters with Simon Barker, Jake Lowenstern, Madison Myers and Paul Wallace are gratefully acknowledged. We also acknowledge the late John Watson, Mike Rowe and Annette Rodgers for XRF analytical services, and thank Adam Kent, Dan Morgan, Ian Schipper and Dan Sinclair for their help with data collection and processing. Detailed reviews by Ben Ellis, Kurt Knesel, Madison Myers and Mike Rowe are much appreciated. FUNDING E. J. Swallow acknowledges with thanks the receipt of a Commonwealth Scholarship, administered by Universities New Zealand–Te Pōkai Tara. We also acknowledge past support from the Marsden Fund (grant VUW0813) and award of a James Cook Fellowship to C. J. N. Wilson, both administered by the Royal Society of New Zealand. REFERENCES Allan A. S. R. , Barker S. J. , Millet M.-A. , Morgan D. J. , Rooyakkers S. M. , Schipper C. I. , Wilson C. J. N. ( 2017 ). A cascade of magmatic events during the assembly and eruption of a super-sized magma body . Contributions to Mineralogy and Petrology 172 , 49 . Google Scholar Crossref Search ADS WorldCat Allan A. S. R. , Morgan D. J. , Wilson C. J. N. , Millet M.-A. ( 2013 ). From mush to eruption in centuries: assembly of the super-sized Oruanui magma body . Contributions to Mineralogy and Petrology 166 , 143 – 164 . Google Scholar Crossref Search ADS WorldCat Allan A. S. R. , Wilson C. J. N. , Millet M.-A. , Wysoczanski R. J. ( 2012 ). The invisible hand: tectonic triggering of a rhyolitic supereruption . Geology 40 , 563 – 566 . Google Scholar Crossref Search ADS WorldCat Almeev R. R. , Bolte T. , Nash B. P. , Holtz F. , Erdmann M. , Cathey H. E. ( 2012 ). High-temperature low H2O silicic magmas of the Yellowstone hotspot: an experimental study of rhyolite from the Bruneau-Jarbidge eruptive center, central Snake River Plain, USA . Journal of Petrology 53 , 1837 – 1866 . Google Scholar Crossref Search ADS WorldCat Armstrong R. L. , Leeman W. P. , Malde H. E. ( 1975 ). K-Ar dating Quaternary and Neogene volcanic rocks of the Snake River Plain, Idaho . American Journal of Science 275 , 225 – 251 . Google Scholar Crossref Search ADS WorldCat Bachmann O. , Bergantz G. W. ( 2004 ). On the origin of crystal-poor rhyolites: extracted from batholithic crystal mushes . Journal of Petrology 45 , 1565 – 1582 . Google Scholar Crossref Search ADS WorldCat Bachmann O. , Bergantz G. W. ( 2008 ). Deciphering magma chamber dynamics from styles of compositional zoning in large silicic ash flow sheets . Reviews in Mineralogy and Geochemistry 69 , 651 – 674 . Google Scholar Crossref Search ADS WorldCat Bachmann O. , Huber C. ( 2016 ). Silicic magma reservoirs in the Earth's crust . American Mineralogist 101 , 2377 – 2404 . Google Scholar Crossref Search ADS WorldCat Bachmann O. , Deering C. D. , Lipman P. W. , Plummer C. ( 2014 ). Building zoned ignimbrites by recycling silicic cumulates: insight from the 1, 000 km3 Carpenter Ridge Tuff, CO . Contributions to Mineralogy and Petrology 167 , 1025 . Google Scholar Crossref Search ADS WorldCat Bachmann O. , Dungan M. A. , Lipman P. W. ( 2002 ). The Fish Canyon magma body, San Juan volcanic field, Colorado: rejuvenation and eruption of an upper-crustal batholith . Journal of Petrology 43 , 1469 – 1503 . Google Scholar Crossref Search ADS WorldCat Bacon C. R. ( 1983 ). Eruptive history of Mount Mazama and Crater Lake caldera, Cascade Range, USA . Journal of Volcanology and Geothermal Research 18 , 57 – 115 . Google Scholar Crossref Search ADS WorldCat Bacon C. R. , Druitt T. H. ( 1988 ). Compositional evolution of the zoned calcalkaline magma chamber of Mount Mazama, Crater Lake, Oregon . Contributions to Mineralogy and Petrology 98 , 224 – 256 . Google Scholar Crossref Search ADS WorldCat Barker S. J. , Wilson C. J. N. , Allan A. S. R. , Schipper C. I. ( 2015 ). Fine-scale temporal recovery, reconstruction and evolution of a post-supereruption magmatic system . Contributions to Mineralogy and Petrology 170 , 5 . Google Scholar Crossref Search ADS WorldCat Barker S. J. , Wilson C. J. N. , Morgan D. J. , Rowland J. V. ( 2016 ). Rapid priming, accumulation and recharge of magma driving recent eruptions at a hyperactive caldera volcano . Geology 44 , 323 – 326 . Google Scholar Crossref Search ADS WorldCat Barker S. J. , Wilson C. J. N. , Smith E. G. C. , Charlier B. L. A. , Wooden J. L. , Hiess J. , Ireland T. R. ( 2014 ). Post-supereruption magmatic reconstruction of Taupo volcano (New Zealand), as reflected in zircon ages and trace elements . Journal of Petrology 55 , 1511 – 1533 . Google Scholar Crossref Search ADS WorldCat Befus K. S. , Gardner J. E. ( 2016 ). Magma storage and evolution of the most recent effusive and explosive eruptions from Yellowstone Caldera . Contributions to Mineralogy and Petrology 171 , 30 . Google Scholar Crossref Search ADS WorldCat Bégué F. , Deering C. D. , Gravley D. M. , Kennedy B. M. , Chambefort I. , Gualda G. A. R. , Bachmann O. ( 2014 ). Extraction, storage and eruption of multiple isolated magma batches in the paired Mamaku and Ohakuri eruption, Taupo Volcanic Zone, New Zealand . Journal of Petrology 55 , 1653 – 1684 . Google Scholar Crossref Search ADS WorldCat Best M. G. , Gromme S. , Deino A. L. , Christiansen E. H. , Hart G. L. , Tingey D. G. ( 2013 ). The 36-18 Ma Central Nevada ignimbrite field and calderas, Great Basin, USA: multicyclic super-eruptions . Geosphere 9 , 1562 – 1636 . Google Scholar Crossref Search ADS WorldCat Bindeman I. N. , Fu B. , Kita N. T. , Valley J. W. ( 2008 ). Origin and evolution of silicic magmatism at Yellowstone based on ion microprobe analysis of isotopically zoned zircons . Journal of Petrology 49 , 163 – 193 . Google Scholar Crossref Search ADS WorldCat Bindeman I. N. , Simakin A. G. ( 2014 ). Rhyolites-hard to produce, but easy to recycle and sequester: integrating microgeochemical observations and numerical models . Geosphere 10 , 930 – 957 . Google Scholar Crossref Search ADS WorldCat Bindeman I. N. , Valley J. W. ( 2001 ). Low- δ18O rhyolites from Yellowstone: magmatic evolution based on analyses of zircons and individual phenocrysts . Journal of Petrology 42 , 1491 – 1517 . Google Scholar Crossref Search ADS WorldCat Birck J. L. ( 1986 ). Precision K-Rb-Sr isotopic analysis: application to Rb-Sr chronology . Chemical Geology 56 , 73 – 83 . Google Scholar Crossref Search ADS WorldCat Bolte T. , Holtz F. , Almeev R. , Nash B. ( 2015 ). The Blacktail Creek Tuff: an analytical and experimental study of rhyolites from the Heise volcanic field, Yellowstone hotspot system . Contributions to Mineralogy and Petrology 169 , 15 . Google Scholar Crossref Search ADS WorldCat Bonnichsen B. , Leeman W. P. , Honjo N. , McIntosh W. C. , Godchaux M. M. ( 2008 ). Miocene silicic volcanism in southwestern Idaho: geochronology, geochemistry, and evolution of the central Snake River Plain . Bulletin of Volcanology 70 , 315 – 342 . Google Scholar Crossref Search ADS WorldCat Bouvier A. , Vervoort J. D. , Patchett P. J. ( 2008 ). The Lu-Hf and Sm-Nd isotopic composition of CHUR: constraints from unequilibrated chondrites and implications for the bulk composition of terrestrial planets . Earth and Planetary Science Letters 273 , 48 – 57 . Google Scholar Crossref Search ADS WorldCat Branney M. J. , Bonnichsen B. , Andrews G. D. M. , Ellis B. , Barry T. L. , McCurry M. ( 2008 ). ‘ Snake River (SR)-type’ volcanism at the Yellowstone hotspot track: distinctive products from unusual high-temperature silicic super-eruptions . Bulletin of Volcanology 70 , 293 – 314 . Google Scholar Crossref Search ADS WorldCat Brey G. P. , Köhler T. ( 1990 ). Geothermobarometry in four-phase lherzolites II. New thermobarometers, and practical assessment of existing thermobarometers . Journal of Petrology 31 , 1353 – 1378 . Google Scholar Crossref Search ADS WorldCat Brophy J. G. ( 1991 ). Composition gaps, critical crystallinity, and fractional crystallization in orogenic (calc-alkaline) magmatic systems . Contributions to Mineralogy and Petrology 109 , 173 – 182 . Google Scholar Crossref Search ADS WorldCat Brown S. J. A. , Wilson C. J. N. , Cole J. W. , Wooden J. ( 1998 ). The Whakamaru group ignimbrites, Taupo Volcanic Zone, New Zealand: evidence for reverse tapping of a zoned silicic magmatic system . Journal of Volcanology and Geothermal Research 84 , 1 – 37 . Google Scholar Crossref Search ADS WorldCat Caricchi L. , Annen C. , Blundy J. , Simpson G. , Pinel V. ( 2014 ). Frequency and magnitude of volcanic eruptions controlled by magma injection and buoyancy . Nature Geoscience 7 , 126 – 130 . Google Scholar Crossref Search ADS WorldCat Cashman K. V. , Giordano G. ( 2014 ). Calderas and magma reservoirs . Journal of Volcanology and Geothermal Research 288 , 28 – 45 . Google Scholar Crossref Search ADS WorldCat Cashman K. V. , Sparks R. S. J. , Blundy J. D. ( 2017 ). Vertically extensive and unstable magmatic systems: a unified view of igneous processes . Science 355 , eaag3055 . Google Scholar Crossref Search ADS PubMed WorldCat Cathey H. E. , Nash B. P. ( 2004 ). The Cougar Point Tuff: implications for thermomechanical zonation and longevity of high-temperature, large-volume silicic magmas of the Miocene Yellowstone hotspot . Journal of Petrology 45 , 27 – 58 . Google Scholar Crossref Search ADS WorldCat Chadwick R. A. ( 1970 ). Belts of eruptive centers in the Absaroka-Gallatin Volcanic Province, Wyoming-Montana . Geological Society of America Bulletin 81 , 267 – 274 . Google Scholar Crossref Search ADS WorldCat Chamberlain K. J. , Morgan D. J. , Wilson C. J. N. ( 2014b ). Timescales of mixing and mobilisation in the Bishop Tuff magma body: perspectives from diffusion chronometry . Contributions to Mineralogy and Petrology 168 , 1034 . Google Scholar Crossref Search ADS WorldCat Chamberlain K. J. , Wilson C. J. N. , Wallace P. J. , Millet M.-A. ( 2015 ). Micro-analytical perspectives on the Bishop Tuff and its magma chamber . Journal of Petrology 56 , 605 – 640 . Google Scholar Crossref Search ADS WorldCat Chamberlain K. J. , Wilson C. J. N. , Wooden J. L. , Charlier B. L. A. , Ireland T. R. ( 2014a ). New perspectives on the Bishop Tuff from zircon textures ages and trace elements . Journal of Petrology 55 , 395 – 426 . Google Scholar Crossref Search ADS WorldCat Charlier B. L. A. , Ginibre C. , Morgan D. , Nowell G. M. , Pearson D. G. , Davidson J. P. , Ottley C. J. ( 2006 ). Methods for the microsampling and high-precision analysis of strontium and rubidium isotopes at single crystal scale for petrological and geochronological applications . Chemical Geology 232 , 114 – 133 . Google Scholar Crossref Search ADS WorldCat Charlier B. L. A. , Wilson C. J. N. , Davidson J. P. ( 2008 ). Rapid open-system assembly of a large silicic magma body: time resolved evidence from cored plagioclase crystals in the Oruanui eruption deposits, New Zealand . Contributions to Mineralogy and Petrology 156 , 799 – 813 . Google Scholar Crossref Search ADS WorldCat Christiansen E. H. ( 2005 ). Contrasting processes in silicic magma chambers: evidence from very large volume ignimbrites . Geological Magazine 142 , 669 – 681 . Google Scholar Crossref Search ADS WorldCat Christiansen E. H. , McCurry M. ( 2008 ). Contrasting origins of Cenozoic silicic volcanic rocks from the western Cordillera of the United States . Bulletin of Volcanology 70 , 251 – 267 . Google Scholar Crossref Search ADS WorldCat Christiansen R. L. ( 1979 ). Cooling Units and Composite Sheets in Relation to Caldera Structure. In: Chapin C. E. , Elston W. E. (eds) Ash-Flow Tuffs. Geological Society of America Special Papers 180 , 29 – 42 . Google Preview WorldCat COPAC Christiansen R. L. ( 2001 ). The Quaternary and Pliocene Yellowstone Plateau Volcanic Field of Wyoming, Idaho and Montana. U. S. Geological Survey Professional Paper 729-G. Christiansen R. L. , Lowenstern J. B. , Smith R. B. , Heasler H. , Morgan L. A. , Nathenson M. , Mastin L. G. , Muffler L. J. P. , Robinson J. E. ( 2007 ). Preliminary assessment of volcanic and hydrothermal hazards in Yellowstone National Park and vicinity . U. S. Geological Survey Open File Report 2007-1071 , 1 – 94 . WorldCat Christiansen R. L. , Yeats R. S. ( 1992 ). Post-laramide geology of the US cordilleran region. In: Burchfiel B. C. , Lipman P. W. , Zoback M. L. (eds) The Cordilleran Orogen: Conterminous US. Boulder, Colorado, Geological Society of America, The Geology of North America G-3 , 261 – 406 . Google Preview WorldCat COPAC Colón D. P. , Bindeman I. N. , Wotzlaw J.-F. , Christiansen E. H. , Stern R. A. ( 2018 ). Origins and evolution of rhyolitic magmas in the central Snake River Plain: insights from coupled high-precision geochronology, oxygen isotope, and hafnium isotope analyses of zircon . Contributions to Mineralogy and Petrology 173 , 11. Google Scholar Crossref Search ADS WorldCat Cooper G. F. , Morgan D. J. , Wilson C. J. N. ( 2017 ). Rapid assembly and rejuvenation of a large silicic magmatic system: insights from mineral diffusive profiles in the Kidnappers and Rocky Hill deposits, New Zealand . Earth and Planetary Science Letters 473 , 1 – 13 . Google Scholar Crossref Search ADS WorldCat Cooper G. F. , Wilson C. J. N. , Millet M.-A. , Baker J. A. ( 2016 ). Generation and rejuvenation of a supervolcanic magmatic system: a case study from Mangakino volcanic centre, New Zealand . Journal of Petrology 57 , 1135 – 1170 . Google Scholar Crossref Search ADS WorldCat Cooper G. F. , Wilson C. J. N. , Millet M.-A. , Baker J. A. , Smith E. G. C. ( 2012 ). Systematic tapping of independent magma chambers during the 1 Ma Kidnappers supereruption . Earth and Planetary Science Letters 313-314 , 23 – 33 . Google Scholar Crossref Search ADS WorldCat Deer W. A. , Howie R. A. , Zussman J. ( 1966 ). An Introduction to the Rock-Forming Minerals. Harlow, U.K.: Longman. Deering C. D. , Bachmann O. , Vogel T. A. ( 2011 ). The Ammonia Tanks Tuff: erupting a melt-rich cap and its remobilized crystal cumulate . Earth and Planetary Science Letters 310 , 518 – 525 . Google Scholar Crossref Search ADS WorldCat DePaolo D. J. , Wasserburg G. J. ( 1976 ). Nd isotopic variations and petrogenetic models . Geophysical Research Letters 3 , 249 – 252 . Google Scholar Crossref Search ADS WorldCat de Silva S. , Gregg P. M. ( 2014 ). Thermomechanical feedbacks in magmatic systems: implications for growth, longevity, and evolution of large caldera-forming magma reservoirs and their supereruptions . Journal of Volcanology and Geothermal Research 282 , 77 – 91 . Google Scholar Crossref Search ADS WorldCat Doe B. R. , Leeman W. P. , Christiansen R. L. , Hedge C. E. ( 1982 ). Lead and strontium isotopes and related trace elements as genetic tracers in the upper Cenozoic rhyolite-basalt association of the Yellowstone Plateau volcanic field . Journal of Geophysical Research: Solid Earth 87 , 4785 – 4806 . Google Scholar Crossref Search ADS WorldCat Drew D. L. , Bindeman I. N. , Loewen M. W. , Wallace P. J. ( 2016 ). Initiation of large-volume silicic centres in the Yellowstone hotspot track: insights from H2O- and F-rich quartz-hosted rhyolitic melt inclusions in the Arbon Valley Tuff of the Snake River Plain . Contributions to Mineralogy and Petrology 171 , 10 . Google Scholar Crossref Search ADS WorldCat Drew D. L. , Bindeman I. N. , Watts K. E. , Schmitt A. K. , Fu B. , McCurry M. ( 2013 ). Crustal-scale recycling in caldera complexes and rift zones along the Yellowstone hotspot track: O and Hf isotopic evidence in diverse zircons from voluminous rhyolites of the Picabo volcanic field, Idaho . Earth and Planetary Science Letters 381 , 63 – 77 . Google Scholar Crossref Search ADS WorldCat Druitt T. H. , Costa F. , Deloule E. , Dungan M. , Scaillet B. ( 2012 ). Decadal to monthly timescales of magma transfer and reservoir growth at a caldera volcano . Nature 482 , 77 – 80 . Google Scholar Crossref Search ADS PubMed WorldCat Elkins L. T. , Grove T. L. ( 1990 ). Ternary feldspar experiments and thermodynamic models . American Mineralogist 75 , 544 – 559 . WorldCat Ellis B. S. , Bachmann O. , Wolff J. A. ( 2014 ). Cumulate fragments in silicic ignimbrites: The case of the Snake River Plain . Geology 42 , 431 – 434 . Google Scholar Crossref Search ADS WorldCat Ellis B. S. , Mark D. F. , Pritchard C. J. , Wolff J. A. ( 2012 ). Temporal dissection of the Huckleberry Ridge Tuff using the 40Ar/39Ar dating technique . Quaternary Geochronology 9 , 34 – 41 . Google Scholar Crossref Search ADS WorldCat Ellis B. S. , Wolff J. A. ( 2012 ). Complex storage of rhyolite in the central Snake River Plain . Journal of Volcanology and Geothermal Research 211-212 , 1 – 11 . Google Scholar Crossref Search ADS WorldCat Ellis B. S. , Wolff J. A. , Boroughs S. , Mark D. F. , Starkel W. A. , Bonnichsen B. ( 2013 ). Rhyolite volcanism of the central Snake River Plain: a review . Bulletin of Volcanology 75 , 745 . Google Scholar Crossref Search ADS WorldCat Embree G. F. , Hoggan R. D. ( 1999 ). Secondary Deformation within the Huckleberry Ridge Tuff and Adjacent Pliocene Units near the Teton Dam: Road log to the regional geology of the eastern margin of the Snake River Plain, Idaho. In: Hughes S. S. , Thackray G. D. (eds) Guidebook to the Geology of Eastern Idaho . Pocatello , Idaho Museum of Natural History , pp. 181 – 203 . Google Preview WorldCat COPAC Farrell J. , Smith R. B. , Husen S. , Diehl T. ( 2014 ). Tomography from 26 years of seismicity revealing the spatial extent of the Yellowstone crustal magma reservoir extends well beyond the Yellowstone caldera . Geophysical Research Letters 41 , 3068 – 3073 . Google Scholar Crossref Search ADS WorldCat Feeley T. C. ( 2003 ). Origin and tectonic implications of across-strike geochemical variations in the Eocene Absaroka volcanic province, United States . Journal of Geology 111 , 329 – 346 . Google Scholar Crossref Search ADS WorldCat Folkes C. B. , de Silva S. L. , Wright H. M. , Cas R. A. F. ( 2011 ). Geochemical homogeneity of a long-lived large silicic system; evidence from the Cerro Galán caldera, NW Argentina . Bulletin of Volcanology 73 , 1455 – 1486 . Google Scholar Crossref Search ADS WorldCat Fridrich C. J. , Mahood G. A. ( 1987 ). Compositional layers in the zoned magma chamber of the Grizzly Peak Tuff . Geology 15 , 299 – 303 . Google Scholar Crossref Search ADS WorldCat Frost C. D. , Frost B. R. , Kirkwood R. , Chamberlain K. R. ( 2006 ). The tonalite-trondhjemite-granodiorite (TTG) to granodiorite-granite (GG) transition in the late Archean plutonic rocks of the central Wyoming province . Canadian Journal of Earth Sciences 43 , 1419 – 1444 . Google Scholar Crossref Search ADS WorldCat Geissman J. W. , Holm D. , Harlan S. S. , Embree G. F. ( 2010 ). Rapid, high-temperature formation of large-scale rheomorphic structures in the 2.06 Ma Huckleberry Ridge Tuff, Idaho, USA . Geology 38 , 263 – 266 . Google Scholar Crossref Search ADS WorldCat Gerstenberger H. , Haase G. ( 1997 ). A highly effective emitter substance for mass spectrometric Pb isotope ratio determinations . Chemical Geology 136 , 309 – 312 . Google Scholar Crossref Search ADS WorldCat Girard G. , Stix J. ( 2009 ). Magma recharge and crystal mush rejuvenation associated with early post-collapse Upper Basin Member rhyolites, Yellowstone caldera, Wyoming . Journal of Petrology 50 , 2095 – 2125 . Google Scholar Crossref Search ADS WorldCat Girard G. , Stix J. ( 2010 ). Rapid extraction of discrete magma batches from a large differentiating magma chamber: the Central Plateau Member rhyolites, Yellowstone caldera, Wyoming . Contributions to Mineralogy and Petrology 160 , 441 – 465 . Google Scholar Crossref Search ADS WorldCat Gravley D. M. , Deering C. D. , Leonard G. S. , Rowland J. V. ( 2016 ). Ignimbrite flare-ups and their drivers: a New Zealand perspective . Earth-Science Reviews 162 , 65 – 82 . Google Scholar Crossref Search ADS WorldCat Gregg P. M. , de Silva S. L. , Grosfils E. B. , Parmigiani J. P. ( 2012 ). Catastrophic caldera-forming eruptions: thermomechanics and implications for eruption triggering and maximum caldera dimensions on Earth . Journal of Volcanology and Geothermal Research 241-242 , 1 – 12 . Google Scholar Crossref Search ADS WorldCat Gregg P. M. , Grosfils E. B. , de Silva S. L. ( 2015 ). Catastrophic caldera-forming eruptions II: the subordinate role of magma buoyancy as an eruption trigger . Journal of Volcanology and Geothermal Research 305 , 100 – 113 . Google Scholar Crossref Search ADS WorldCat Gualda G. A. R. , Bégué F. , Pamukcu A. S. , Ghiorso M. S. ( 2019 ). Rhyolite-MELTS vs DERP—Newer does not make it better: a comment on ‘The effect of anorthite content and water on quartz-feldspar cotectic compositions in the rhyolitic system and implications for geobarometry’ by Wilke et al. (2017; Journal of Petrology, 58, 789–818) . Journal of Petrology 60 , 855 – 864 . Google Scholar Crossref Search ADS WorldCat Gualda G. A. R. , Ghiorso M. S. ( 2013 ). The Bishop Tuff giant magma body: an alternative to the Standard Model . Contributions to Mineralogy and Petrology 166 , 755 – 775 . Google Scholar Crossref Search ADS WorldCat Gualda G. A. R. , Pamukcu A. S. , Ghiorso M. S. , Anderson A. T. Jr , Sutton S. R. , Rivers M. L. ( 2012 ). Timescales of quartz crystallization and the longevity of the Bishop giant magma body . PLoS ONE 7 , e37492 . Google Scholar Crossref Search ADS PubMed WorldCat Gudmundsson A. ( 2012 ). Magma chambers: formation, local stresses, excess pressures and compartments . Journal of Volcanology and Geothermal Research 237–238 , 19 – 41 . Google Scholar Crossref Search ADS WorldCat Hildreth W. ( 1981 ). Gradients in silicic magma chambers: implications for lithospheric magmatism . Journal of Geophysical Research 86 , 10153 – 10192 . Google Scholar Crossref Search ADS WorldCat Hildreth W. ( 2004 ). Volcanological perspectives on Long Valley Mammoth Mountain and Mono Craters: several contiguous but discrete systems . Journal of Volcanology and Geothermal Research 136 , 169 – 198 . Google Scholar Crossref Search ADS WorldCat Hildreth W. , Christiansen R. L. , O’Neill J. R. ( 1984 ). Catastrophic isotopic modification of rhyolitic magma at times of caldera subsidence, Yellowstone Plateau volcanic field . Journal of Geophysical Research 89 , 8339 – 8369 . Google Scholar Crossref Search ADS WorldCat Hildreth W. , Fierstein J. ( 2012 ). The Novarupta-Katmai eruption of 1912—largest eruption of the twentieth century: centennial perspectives . U. S. Geological Survey Professional Paper 1791 , 1 – 259 . WorldCat Hildreth W. , Fierstein J. , Calvert A. ( 2017 ). Early post-caldera rhyolite and structural resurgence at Long Valley caldera, California . Journal of Volcanology and Geothermal Research 335 , 1 – 34 . Google Scholar Crossref Search ADS WorldCat Hildreth W. , Halliday A. N. , Christiansen R. L. ( 1991 ). Isotopic and chemical evidence concerning the genesis and contamination of basaltic and rhyolitic magma beneath the Yellowstone Plateau volcanic field . Journal of Petrology 32 , 63 – 138 . Google Scholar Crossref Search ADS WorldCat Hildreth W. , Wilson C. J. N. ( 2007 ). Compositional zoning of the Bishop Tuff . Journal of Petrology 48 , 951 – 999 . Google Scholar Crossref Search ADS WorldCat Hosaka M. , Miyata T. , Sunagawa I. ( 1995 ). Growth and morphology of quartz crystals synthesized above the transition temperature . Journal of Crystal Growth 152 , 300 – 306 . Google Scholar Crossref Search ADS WorldCat Houghton B. F. , Wilson C. J. N. , McWilliams M. O. , Lanphere M. A. , Weaver S. D. , Briggs R. M. , Pringle M. S. ( 1995 ). Chronology and dynamics of a large silicic magmatic system: central Taupo Volcanic Zone, New Zealand . Geology 23 , 13 – 16 . Google Scholar Crossref Search ADS WorldCat Hu Z. , Zhang W. , Liu Y. , Chen H. , Gaschnig R. M. , Zong K. , Li M. , Gao S. , Hu S. ( 2013 ). Rapid bulk rock decomposition by ammonium fluoride (NH4F) in open vessels at an elevated digestion temperature . Chemical Geology 355 , 144 – 152 . Google Scholar Crossref Search ADS WorldCat Huang H.-H. , Lin F.-C. , Schmandt B. , Farrell J. , Smith R. B. , Tsai V. C. ( 2015 ). The Yellowstone magmatic system from the mantle plume to the upper crust . Science 348 , 773 – 776 . Google Scholar Crossref Search ADS PubMed WorldCat Huber C. , Bachmann O. , Dufek J. ( 2011 ). Thermo-mechanical reactivation of locked crystal mushes: melting induced internal fracturing and assimilation processes in magmas . Earth and Planetary Science Letters 304 , 443 – 454 . Google Scholar Crossref Search ADS WorldCat Huber C. , Bachmann O. , Dufek J. ( 2012 ). Crystal-poor versus crystal-rich ignimbrites: a competition between stirring and reactivation . Geology 40 , 115 – 118 . Google Scholar Crossref Search ADS WorldCat Humphreys M. C. S. , Kearns S. L. , Blundy J. D. ( 2006 ). SIMS investigation of electron-beam damage to hydrous rhyolitic glasses: implications for melt inclusion analysis . American Mineralogist 91 , 667 – 679 . Google Scholar Crossref Search ADS WorldCat James D. E. , Fouch M. J. , Carlson R. W. , Roth J. B. ( 2011 ). Slab fragmentation, edge flow and the origin of the Yellowstone hotspot track . Earth and Planetary Science Letters 311 , 124 – 135 . Google Scholar Crossref Search ADS WorldCat Jellinek A. M. , DePaolo D. J. ( 2003 ). A model for the origin of large silicic magma chambers: precursors of caldera-forming eruptions . Bulletin of Volcanology 65 , 363 – 381 . Google Scholar Crossref Search ADS WorldCat Laporte D. , Provost A. ( 1994 ). The equilibrium crystal shape of silicates: implications for the grain-scale distribution of partial melts . EOS, Transactions of the American Geophysical Union 75 , 364 . WorldCat Le Bas M. J. , Streckeisen A. L. ( 1991 ). The IUGS systematics of igneous rocks . Journal of the Geological Society, London 148 , 825 – 833 . Google Scholar Crossref Search ADS WorldCat Leeman W. P. , Menzies M. A. , Matty D. J. , Embree G. F. ( 1985 ). Strontium, neodymium and lead isotopic compositions of deep crustal xenoliths from the Snake River Plain: evidence for Archean basement . Earth and Planetary Science Letters 75 , 354 – 368 . Google Scholar Crossref Search ADS WorldCat Leeman W. P. , Vitaliano C. J. , Prinz M. ( 1976 ). Evolved lavas from the Snake River Plain: Craters of the Moon National Monument, Idaho . Contributions to Mineralogy and Petrology 56 , 35 – 60 . Google Scholar Crossref Search ADS WorldCat Lipman P. W. , Bachmann O. ( 2015 ). Ignimbrites to batholiths: integrating perspectives from geological, geophysical, and geochronological data . Geosphere 11 , 705 – 743 . Google Scholar Crossref Search ADS WorldCat Loewen M. W. , Bindeman I. N. ( 2015 ). Oxygen isotope and trace element evidence for three-stage petrogenesis of the youngest episode (260-79 ka) of Yellowstone rhyolitic volcanism .
TI - The Huckleberry Ridge Tuff, Yellowstone: evacuation of multiple magmatic systems in a complex episodic eruption
JF - Journal of Petrology
DO - 10.1093/petrology/egz034
DA - 2019-07-01
UR - https://www.deepdyve.com/lp/oxford-university-press/the-huckleberry-ridge-tuff-yellowstone-evacuation-of-multiple-magmatic-dEdwRyjpIB
SP - 1371
VL - 60
IS - 7
DP - DeepDyve
ER -