In continental-margin subduction zones, basalt magmas spawned in the mantle interact with the crust to produce a broad spectrum of volcanic arc associations. A distinct style of very voluminous arc volcanism develops far inland on thick crust over periods of 10–20 m.y. and involves relatively infrequent caldera-forming explosive eruptions of dominantly calc-alkaline rhyolite, dacite, and trachydacite with repose times of 104–106 yr. Volumes of individual eruptions are large (102–103 km3), and nested super-eruptions of thousands of cubic kilometers are common. Calderas are as much as 60–75 km in diameter, and surrounding individual ignimbrite outflow sheets extend outward as much as 150 km, blanketing upwards of 105 km2. Little or no basalt is extruded, whereas andesitic differentiates coeval with silicic ignimbrites range from minor to dominant in relative volume. A common feature in these flareups is essentially nonextending, thick, inland crust overlying a subducting oceanic plate with transverse tears that rolled back to a steeper dip from a previously flat configuration. Lithospheric delamination is locally possible. Large volumes of basalt that provide heat and mass for silicic magma generation in the crust form by fluid fluxing of the growing mantle wedge overlying the steepening dehydrating slab and from asthenospheric decompression. Variations in the mantle input, together with variations in crustal thickness, temperature, and composition, modulate the expression of the flareups. As a consequence of the high flux of mantle-derived magma into the thick crust, geotherms become elevated, and the brittle-ductile transition can rise to depths as shallow as 7 km. At this transition, diapirically rising magmas from a melting, assimilation, storage, and homogenization (MASH) zone are blocked and spread laterally into discoid chambers that grow until a thermomechanical threshold is attained, triggering climactic eruption and caldera collapse.
This ignimbrite flareup style of continental arc volcanism is exemplified by the mid-Cenozoic southern Great Basin ignimbrite province; other examples include the contemporaneous Southern Rocky Mountain, Mogollon-Datil, vast Sierra Madre Occidental volcanic fields, and the late Cenozoic Altiplano-Puna volcanic complex in the Central Andes. Rhyolitic and trachydacitic ignimbrites typically have erupted, but where the crust was predominantly felsic, prewarmed, and orogenically thickened, well-developed MASH zones have spawned multiple super-eruptions of phenocryst-rich dacite, or monotonous intermediates, and smaller volumes of calc-alkaline rhyolite ignimbrite. In the Great Basin, eruptions of dry, hot trachydacite magma followed the monotonous intermediates. Partial melting in thinner crust with a major mafic component yielded more alkalic rhyolite and related trachydacite.
Large-scale, subduction-related, explosive silicic volcanism has developed in many places around the globe, such as in the late Cenozoic Central high Andes of western South America (e.g., Salisbury et al., 2011). The moniker “ignimbrite flareup” was coined by Coney (1978) for the great burst of mid-Cenozoic silicic volcanism in southwestern North America. This flareup—recognized earlier by Lipman et al. (1971) and Noble (1972)—is manifested in the vast Sierra Madre Occidental province in western Mexico and the voluminous Mogollon-Datil, Southern Rocky Mountain, Great Basin, and Challis volcanic fields northward in the United States (Fig. 1). It ranks as one of the premier expressions of explosive silicic volcanism in a continental arc in the global geologic record and was a major episode of crustal growth and hybridization (Johnson, 1991). Significant ignimbrites were deposited from 50 Ma until recently, but the most voluminous explosive silicic volcanism, the peak of the ignimbrite flareup, occurred at ca. 36–18 Ma, near the close of arc volcanism, which began along the southwestern margin of North America in the early Mesozoic. During the flareup, on the order of 500,000 km3 of silicic magma explosively erupted.
Continental arc volcanism is controlled by many geodynamic factors. One of the most important is the dip of the subducting slab of oceanic lithosphere (e.g., Lipman et al., 1971; Keith, 1978; Ward, 1995; DeCelles et al., 2009). Typically, slabs dip at moderate angles of 30° to 70°, and near-trench volcanic eruptions are small (<1 km3) and frequent (decades to centuries). Basaltic to intermediate-composition lavas and minor silicic ignimbrites create stratovolcanoes at central vents over <105 yr. This more familiar normal, or steady-state, activity (de Silva, 2008) of an arc may persist for tens of millions of years. In another type of subduction, the slab dips at a low angle, arc magmatism is suppressed, and the continental crust commonly thickens by deformation. The mantle lithosphere may be cooled, hydrated, and/or eroded by the flat oceanic slab (e.g., Humphreys, 2009; English et al., 2003). Later, the slab may steepen, or rollback, locally accompanied by segmentation, tears, and lithospheric delamination (Kay and Coira, 2009; DeCelles et al., 2009). As used here, “rollback” refers to slab steepening, as opposed to hinge migration. After approximately 10–20 m.y. of rollback, the slab reestablishes a steep dip, but during the transition, large volumes of silicic magma are produced and explosively erupted with recurrence times of 104 to 106 m.y. far inland from the trench over broad areas of thick crust. The largest eruptions have volumes1 of at least 1000 km3 of tephra (volcanic explosivity index [VEI] 8), or 450 km3 of magma, qualifying as super-eruptions (Rampino and Self, 1992; Sparks et al., 2005; Self, 2006; Miller and Wark, 2008; de Silva, 2008; Wilson, 2008).
In this article, we document slab-rollback ignimbrite flareups as a distinct and significant genre in continental arc volcanism, present a comprehensive survey of their character, and evaluate the relevant geodynamic factors in their origin. To achieve this goal, we first describe one part of the mid-Cenozoic southwestern North American flareup—the southern Great Basin ignimbrite province (Fig. 2). It is ideally suited as an example of this flareup style of arc volcanism: First, it harbors on the order of 75,000 km3 of silicic ignimbrite and more than 50 calderas as much as 60 km in diameter. Of the >230 explosive eruptions, at least 25 had volumes of 1000 to 5900 km3, thus qualifying as super-eruptions. Second, many hundreds of modal, chemical, Sr isotope, 40Ar/39Ar, and paleomagnetic analyses characterize the province, and dimensional data are also available for most ignimbrite sheets. Third, geologic maps at scales as large as 1:24,000 cover most of the province, and its structural architecture is well documented, including inferences on its character during the mid-Cenozoic flareup. A detailed and comprehensive anatomy of the southern Great Basin ignimbrite province is available for the first time in this themed issue of Geosphere (Best et al., 2013a, 2013b, 2013c; Henry and John, 2013).
Following this review of the southern Great Basin ignimbrite province, we then compare attributes of other volcanic fields that have experienced a flareup, focusing on the mid-Cenozoic Southern Rocky Mountain, Mogollon-Datil, and Sierra Madre volcanic fields (Fig. 1), and the late Cenozoic Altiplano-Puna volcanic complex of the Central Andes. Similarities and contrasts among these areas provide opportunities to resolve uncertainties in the genesis of voluminous ignimbrite flareups related to slab rollback. Fundamentally, with the tectonic setting of flareups as a backdrop, how were the vast volumes of mantle magmas produced that powered the generation of the silicic magmas in the crust? What aspects of the crust and processes in it allowed the flareups to develop?
GEOLOGIC SETTING OF THE SOUTHERN GREAT BASIN AND VICINITY
Tectonic and Magmatic History
The crust beneath the Great Basin area consists of two terranes of contrasting composition and thickness, thus offering an opportunity to evaluate the role of these factors in ignimbrite flareups. To the west, the basement consists of an assemblage of accreted Phanerozoic oceanic terranes including relict island arcs intruded by Mesozoic granitic plutons. In contrast, to the east (Figs. 2 and 3), the basement is Precambrian crystalline rocks, mainly Proterozoic igneous and metamorphic rocks with arc affinities (Whitmeyer and Karlstrom, 2007). Following the rifting of Rodinia in the Neoproterozoic, a thick prism of marine sediment accumulated on the rifted and thinned margin (Dickinson, 2006; Yonkee et al., 2014). This prism lies between the north-trending Wasatch hinge line in central Utah and the edge of the Precambrian basement in central Nevada (Fig. 4). Paleozoic strata are only ∼1.5 km thick east of the hinge line, whereas to the west, the mostly carbonate rocks are ∼12 km thick, and underlying predominantly siliciclastic late Proterozoic sedimentary rocks are 4–10 km thick (Stewart, 1980).
Farmer and DePaolo (1983) concluded that the lower crust in the eastern Great Basin consists of mostly felsic rocks that were thinned during the Neoproterozoic rifting, whereas the nonrifted crust under the Colorado Plateau of eastern Utah and western Colorado consists of stronger, more mafic rocks (Wannamaker et al., 2008; Lipman et al., 1978).
Initial Sr isotopic compositions of igneous rocks across the region reflect this inferred crustal composition (Fig. 5), with low values (<0.706) in western Nevada rising to as high as 0.718 eastward where the orogenically thickened wedge of sediment is the thickest, and then falling back to lower values across the Colorado Plateau and the San Juan sector of the Southern Rocky Mountain volcanic field in Colorado.
The late Paleozoic–Mesozoic history of the region is dominated by the development of the Cordilleran orogenic belt during plate convergence (e.g., Burchfiel et al., 1992; DeCelles, 2004; Dickinson, 2006; Hintze and Kowallis, 2009). In central Nevada, Devonian–Jurassic thrust systems transported deep-water marine rocks eastward as much as 200 km onto shallower-water rocks (Fig. 4). In the eastern Great Basin area, during the Cretaceous Sevier orogeny, as many as eight extensive slices of Precambrian basement and overlying Phanerozoic sedimentary strata were transported eastward. Total east-west shortening across the Cordilleran orogenic belt in the Great Basin area at the latitude of central Utah is estimated to have been at least 350 km (DeCelles, 2004), creating thickened crust beneath a high-elevation plateau—the Great Basin altiplano (Fig. 3; Best et al., 2009). We believe it resembled in many respects the late Cenozoic Central Andean plateau, where the crust is 70–80 km thick (e.g., Yuan et al., 2002; McGlashan et al., 2008). In the Great Basin, the crust was as much as ∼70 km thick in a roughly north-south belt in easternmost Nevada but thinned to perhaps 50–60 km to the west (Best et al., 2009; Coney and Harms, 1984; DeCelles, 2004; Chapman et al., 2015). Peak metamorphic conditions in the thicker eastern belt in the Late Cretaceous were ∼800 °C at 35 km depth (DeCelles, 2004). For comparison, the temperature of typical continental crust at this depth would be 400 °C to 500 °C (Blackwell, 1971). Metamorphic grades were lower to the east and west, where the crustal thickness was less.
As large-scale contractile deformation waned during the Sevier orogeny, eastward-propagating crustal shortening began at ca. 75–66 Ma in the Rocky Mountain foreland in eastern Utah and western Colorado (Fig. 1), creating basement-cored uplifts. The magnitude of this Laramide deformation was much less than to the west. For example, in the Laramide foreland, including the Southern Rocky Mountain volcanic field, typical shortening was 10%–15%, whereas in the Sevier fold-and-thrust belt, shortening was 50% (Erslev, 1993). Laramide contraction continued until the beginning of the magmatic flareup, about 35 Ma in central Utah (Dickinson et al., 1988; Willis, 1999).
Subduction-related Jurassic magmatism extended far inland, reaching central Utah (Christiansen et al., 2015). Then, during the latest Cretaceous–Paleocene, arc magmatism ceased over much of the western United States (e.g., Lipman, 1992), manifesting shallow, or “flat,” subduction of the oceanic Farallon plate (Coney and Reynolds, 1977; Severinghaus and Atwater, 1990; Humphreys, 2009; for a critique of the flat slab model, see Jones et al., 2011). Magmatism in the Great Basin area during the Late Cretaceous consisted of a few, strongly peraluminous two-mica granite plutons (initial Sr isotope ratios as high as 0.7246) in eastern Nevada just north of the Indian Peak caldera complex (Fig. 6). These peraluminous magmas originated by partial melting of metasedimentary rocks in the miogeoclinal wedge without input of mantle-derived basaltic magma (Best et al., 1974; Lee and Christiansen, 1983; Farmer and DePaolo, 1983).
Beginning in the Eocene, arc magmatism resumed north of the Great Basin at ca. 54–38 Ma in the Challis and Absaroka volcanic fields (Fig. 1). At about 45 Ma, magmatism began in the northern Great Basin and swept southward into southern Nevada, where it stalled in the mid-Miocene at about 15 Ma (Lipman et al., 1972; Christiansen and Lipman, 1972; Best and Christiansen, 1991, their fig. 2; Christiansen et al., 2007b). This time-transgressive arc magmatism was more or less parallel to the trench and corresponded with progressive steepening, or rollback, of the subducting Farallon slab that began first in the north and moved southward. Steepening of slab dip probably resulted from the combined effects of Siletzia accretion (Humphreys, 2009), the passage of a subducted oceanic plateau (Saleeby, 2003), and a slowing rate of convergence (Severinghaus and Atwater, 1990).
No basalt (International Union of Geological Sciences [IUGS] classification of Le Maitre, 1989) was extruded in the Great Basin until after ca. 20 Ma as the arc signature in the volcanic rocks disappeared, and intraplate, extension-related compositions took their place (Christiansen et al., 2007a, their fig. 11). The extension-related bimodal suite included both aluminous and peralkaline rhyolites (e.g., Farmer et al., 1991). During the ignimbrite flareup, extrusion of intermediate-composition, chiefly andesitic, lavas was an order of magnitude smaller in volume than silicic explosive eruptions (Fig. 2; Best et al., 2013b, 2013c).
In western Utah and eastern Nevada, southward-sweeping magmatism (ca. 45–18 Ma) is expressed by more or less separate, subparallel, roughly east-west belts of volcanic rocks and minor granitic intrusions (e.g., Stewart and Carlson, 1976). The greatest volume of mid-Cenozoic volcanic rocks occurs in a swath of mostly mountain-range exposures of silicic ignimbrite and lesser andesitic lava extending from the southwestern corner of Utah westward across the southern Great Basin and beyond into the Sierra Nevada (Fig. 2). This is the 36–18 Ma southern Great Basin ignimbrite province.
Absence of Significant Regional Tectonic Extension during the Ignimbrite Flareup
Controversy surrounds the time when the orogenically thickened crust in the Great Basin altiplano was subjected to significant extensional faulting and thinning to its current thickness of ∼30 km (Allmendinger et al., 1987). The timing of extensional thinning of the crust—before, during, or after the ignimbrite flareup—has a critical bearing on the role of crustal thickness in the ignimbrite flareup.
Prevolcanic extension has been advocated by, for example, DeCelles (2004, p. 149), who concluded that, in view of the excess gravitational potential energy residing in a thick orogenic plateau, “Within limits of available temporal resolution, hinterland extension and frontal thrusting were coeval in the Cordilleran thrust belt during Late Cretaceous time.” However, he added that, “Paradoxically, no surface-breaking large-scale normal faults have been documented that might have facilitated this crustal extension.” Long (2012) concurred that large-magnitude, regionally distributed, prevolcanic extensional faults do not exist in the Cordilleran hinterland.
Broad synchroneity of mid-Cenozoic volcanism and extension in the Great Basin was advocated by, for example, Gans et al. (1989), Constenius (1996), Humphreys (2009), and DeCelles et al. (2009); volcanism and coincident extension in some metamorphic core complexes were extrapolated over the entire Great Basin. However, in at least one of these metamorphic complexes, in the Snake Range in easternmost central Nevada, fission-track dating has revealed that large-magnitude extension in the early Miocene at 17 Ma had been “seriously underestimated” relative to the originally conceived “synvolcanic” late Eocene–early Oligocene extension (Miller et al., 1999, p. 902). Ward (1995) concluded that subduction-related ignimbrite flareups were related to changes to trench-normal extension. Bryan and Ferrari (2013) considered the Sierra Madre Orientale, the southern continuation of the U.S. flareup, to be an intraplate silicic large igneous province related to extension.
Because of the persisting viewpoint of synvolcanic extension during the ignimbrite flareup, we offer nine independent lines of evidence that, in our opinion, argue against significant regional extension during the ignimbrite flareup in the southern Great Basin province.
(1) There is a lack of substantial and widespread angular discordances and deposits of erosional debris within outflow ignimbrite sequences (Best and Christiansen, 1991; Best et al., 2013a, their fig. 7; Best et al., 2013b, their fig. 60; Best et al., 2013c, their figs. 28, 47, and 49).
(2) The ignimbrites cover a great areal extent. Had there been significant fault-induced synvolcanic topography, younger ash flows would not have been dispersed so widely from their sources (Best et al., 2013b, 2013c).
(3) Paleomagnetic directions are consistent throughout single widespread ignimbrites, coupled with significant differences between mean directions in stratigraphically adjacent ignimbrites (Grommé et al., 1972; see also, for example, Best et al., 2013c, their fig. 48). Had the ignimbrites been draped over a highly faulted terrane, the distinctions between paleomagnetic directions of different ignimbrites would have been considerably blurred.
(4) In the eastern sector of the province, ratios of east-west to north-south dimensions of outflow sheets imply that east-west crustal extension took place after deposition of ignimbrites (Best et al., 2013b, their table 8).
(5) Structural and stratigraphic analyses of basin sediments in nearby areas of the Great Basin have shown that large-magnitude extension began in the Miocene, after ca. 23 Ma (e.g., Colgan and Henry, 2009; Henry et al., 2011; Smith et al., 1991; McQuarrie and Wernicke, 2005, their table 1; Anderson et al., 2013).
(8) Stable isotope data (e.g., Mix et al., 2011; Folkes et al., 2013; Cassel et al., 2014) indicate that the high (∼3.5 km) elevations of the Great Basin altiplano persisted to the late Oligocene, ca. 24 Ma.
(9) The elemental and isotopic compositions of the mid-Cenozoic volcanic rocks indicate the Great Basin crust was likely as thick as about 70 km in its eastern part, somewhat thinner westward, and had not yet been thinned by extension (Best et al., 2009).
We conclude that major regional tectonic extension and collapse of the Great Basin altiplano mostly followed the 36–18 Ma ignimbrite flareup in the southern Great Basin province. This evidence does not preclude local faulting and extension related to magma intrusion, caldera collapse, or resurgence.
CHARACTER OF THE SOUTHERN GREAT BASIN IGNIMBRITE PROVINCE
This mid-Cenozoic province is an exceptional example of the ignimbrite flareup style of continental arc volcanism because of its >230 explosive eruptions of silicic magma, at least 25 of which had volumes of 1000 to 5900 km3, a provincewide volume on the order of 75,000 km3, and ∼50 calderas to as much as about 60 km in diameter (Table 1).
Rhyolite ignimbrites of all ages (36–18 Ma) occur throughout the province and are accompanied by temporally more restricted dacite and trachydacite ignimbrites, especially in the eastern sector (Fig. 7). Most ignimbrites contain variable proportions of plagioclase, sanidine, quartz, biotite, hornblende, Fe-Ti oxides, pyroxene, and trace amounts of zircon, apatite, and, rarely, titanite, that equilibrated in relatively wet magmas at shallow crustal depths of 7–9 km. Dacites show phenocryst-rich compositions, upward of 50% of the tuff on dense rock equivalent basis (Best et al., 2013b); nearly all formed by super-eruptions, and their relatively unzoned, uniform nature qualifies them as monotonous intermediates in the sense of Hildreth (1981). These dacite magmas are interpreted to have originated by mixing of andesitic and rhyolitic magmas in the deeper crust (Best et al., 2013b). Unusual trachydacites (Best et al., 2013b) have sparse (<15%) phenocrysts of plagioclase, two pyroxenes, and Fe-Ti oxides that were derived from drier, hotter magmas equilibrated at greater crustal depth. These Isom-type tuffs (Figs. 7–9) have >300 ppm Zr and high TiO2/CaO ratios.
Intermediate-composition lavas in the southern Great Basin province are mostly high-K andesite (Best et al., 2009). Notably, their volume is an order of magnitude less than that of silicic ignimbrite (Table 1; Stewart and Carlson, 1976; Best et al., 2013b, 2013c). As previously indicated, basalt is absent until after ca. 20 Ma.
Lavas and ignimbrites have an arc geochemical signature—wet, oxidized, with low Fe/Mg ratios, enrichments of fluid-soluble elements, and depletions of high field strength elements, producing high Ba/Nb ratios (Best et al., 2013b, 2013c; Henry and John, 2013). Sr and O isotopic ratios are high and consistent with the assimilation of large proportions of felsic crust in the eastern sector of the Great Basin province; Sr isotope ratios are much lower in the western sector, where the crust is younger and more mafic (Figs. 5 and 10).
The southern Great Basin province is conveniently divided into three contrasting sectors, each corresponding to discrete clusters of calderas surrounded by more or less separate ignimbrite outflow fields (Fig. 6; Table 1). In the eastern sector, the Indian Peak–Caliente field (Best et al., 2013b) surrounds the nested calderas of the Indian Peak–Caliente caldera complex astride the Utah-Nevada state line. In the central sector of the southern Great Basin province, the Central Nevada caldera complex of nested sources is surrounded by the Central Nevada field, the defining ignimbrite outflow sheets of which partly overlap those of the Indian Peak–Caliente field. All but one ash flow from Central Nevada caldera sources lie entirely to the east of a north-south topographic barrier, or drainage divide, near the western edge of the Precambrian basement (Figs. 2, 3, and 6; Best et al., 2009, 2013c). In the western sector of the Great Basin province, the Western Nevada ignimbrite field surrounds 23 known source calderas that are more widely scattered and apparently smaller than those to the east.
In the western sector, dacites are rare, and rhyolites dominate (Figs. 7, 11, and 12); subordinate trachydacites have mineral assemblages that equilibrated—like rhyolites—in relatively wet magmas. Notably, ignimbrites are relatively more alkalic overall (Figs. 8 and 9) because of greater concentrations of Na2O. In contrast, all trachydacites to the east are Isom type. These tuffs are especially voluminous in the eastern sector, where immediately older phenocryst-rich dacites, or monotonous intermediates, are the dominant ignimbrite. In the central sector, both Isom-type tuffs and monotonous intermediate ignimbrites are less voluminous than to the east.
Figure 13A reveals further details of the ignimbrites in the eastern sector. After a few million years of small, precursory rhyolite eruptions, three super-eruptions totaling 12,300 km3 of monotonous intermediate magmas occurred from 31.1 to 29.2 Ma (all 40Ar/39Ar ages are based on an age of 28.20 Ma for the Fish Canyon Tuff). The source calderas of these three eruptions overlap (Fig. 14), indicating a sustained and narrowly focused supply of a large volume of crystal-rich dacite magma to the shallow crust in just a few million years. Following this burst of activity, at least four Isom-type trachydacite ignimbrites totaling 3600 km3 were erupted from a concealed source just to the southeast at 27.9 to 27.3 Ma (Fig. 6). After a hiatus in explosive activity of ∼4 m.y., voluminous eruptions from nested calderas in the Caliente complex (Fig. 6) occurred episodically until ca. 18 Ma; rhyolite dominated, but a super-eruption of 2200 km3 at 22.6 Ma created an unusual phenocryst-rich, andesite-latite ignimbrite that may be a monotonous intermediate unit.
The central sector of the province (Fig. 13B) shares aspects with the adjacent sectors to the west and east. To the west, most eruptions were of rhyolite, but, in the central sector, three super-eruptions totaling 4500 km3 of monotonous intermediate dacite occurred in rapid succession at 27.6 Ma. Only about 600 km3 of slightly younger Isom-type trachydacite is recognized. In lieu of an earlier monotonous intermediate ignimbrite paralleling the brief burst of this activity in the eastern sector, 4800 km3 of zoned rhyolite-dacite erupted at 31.7 Ma.
As previously noted, during the mid-Cenozoic ignimbrite flareup, the crust in the western sector of the southern Great Basin province was thinner (∼50 km), younger, and more mafic (accreted Phanerozoic oceanic terranes) than the two eastern sectors, which were founded on thicker Precambrian metamorphic-granitoid basement covered by a thick wedge of sedimentary rock (Fig. 3). How these east-west contrasts in the crust influenced the contrasting nature of the ignimbrites in the province will be considered after a review of other Cenozoic volcanic fields.
COMPARISONS WITH OTHER CENOZOIC VOLCANIC FIELDS
Mid-Cenozoic fields in southwestern North America (Fig. 1; Table 2) developed on continental crust above subducting oceanic lithosphere (e.g., Severinghaus and Atwater, 1990; DeCelles, 2004; Dickinson, 2006; Humphreys, 2009). The mid-Cenozoic volcanic rocks in these fields possess arc chemical signatures consistent with a subduction heritage: Marysvale (Cunningham et al., 1997); Southern Rocky Mountain (Lipman et al., 1978; Askren et al., 1997; Bachmann et al., 2002; Parat et al., 2005); Mogollon-Datil (Bornhorst, 1980; Davis et al., 1993); Sierra Madre Occidental (Ferrari et al., 2007). The flareups, which totaled perhaps 400,000 km3 of ignimbrite in these volcanic fields, were nearly synchronous from ca. 36 to 18 Ma, with the greatest eruptive volumes ca. 32 to 23 Ma.
Marysvale Volcanic Field and Colorado Plateau Laccoliths: No Ignimbrite Flareup
This modest-sized field, which contains a cluster of small calderas, lies just east of the southern Great Basin ignimbrite province on the northwestern margin of the Colorado Plateau; it was active ca. 28–19 Ma (Table 2; Figs. 1 and 2; Steven et al., 1984; Cunningham et al., 2007; Rowley et al., 2002). The time-composition pattern from early phenocryst-rich dacite ignimbrite to trachydacite and then rhyolite mimics the trend in the adjacent eastern sector of the Great Basin province (Fig. 15). However, the Marysvale field differs from the Great Basin province. The number (7) and the total volume (∼1000 km3) of ignimbrite eruptions are an order of magnitude smaller. The dominant volume (11,000 km3) of the Marysvale field consists of coalesced andesitic-dacitic stratovolcanoes that were built primarily about 26–23 Ma. The initial Sr isotopic composition of the tuffs and lavas is significantly lower than for the adjacent Great Basin (Fig. 5; Cunningham et al., 1997).
These contrasts with the contemporaneous Great Basin ignimbrite flareup are likely related, at least in part, to thinner crust (currently ∼30 km) beneath the Marysvale field. This crust might have been thicker during volcanism, because late Cenozoic extensional tectonism has thinned the crust below the western part of the field (Wannamaker et al., 2008). Crust to the east, under the essentially unextended Colorado Plateau (<3%; according to Davis, 1999), is ∼45 km thick. The low Sr isotope ratios and some of the compositional differences are probably due to the more mafic lower crust beneath this region, as noted previously.
Farther east on the Colorado Plateau (Fig. 1), mid-Cenozoic magmatism generated local, highly alkaline dikes (e.g., Tingey et al., 1991), explosive breccia pipes (Usui et al., 2003), and shallow, composite laccolithic complexes mostly intruded at 29–23 Ma (Nelson et al., 1992). The laccoliths have a relatively modest cumulative volume (estimated ∼140 km3) and are composed almost entirely of a sodic diorite porphyry that is depleted in Nb and Ti, manifesting an arc signature (Nelson and Davidson, 1993, 1998). Low initial Sr isotope ratios <0.706 and common xenoliths of amphibolite indicate derivation from a mafic crustal source. This magmatism is strikingly different from that in contemporaneous volcanic fields to the west and east.
Southern Rocky Mountain Volcanic Field
In this major well-documented field on the northeastern margin of the Colorado Plateau (Fig. 1; Table 2; Lipman, 2007; Lipman and Bachmann, 2015), 28 regional ignimbrites (>100 km3) were emplaced from 21 exposed calderas at 37 to 23 Ma. At about 26 Ma, magmatism switched to a more bimodal assemblage related to opening of the Rio Grande rift immediately east of most of the source calderas (Lipman et al., 1978). Thus, the Southern Rocky Mountain field lacks the youngest 5 m.y. of explosive activity evident in the southern Great Basin province. The south-southwest sweep of volcanism in the field resembles that in the Great Basin: Ignimbrites erupted at 37–33 Ma in the north, at 25 Ma in the south, and as young as 23 Ma in the west; peak activity was from 32 to 27 Ma, without a clear space-time pattern in the central, or San Juan Mountains, sector. Significantly, caldera fills are dominated by intermediate-composition lavas, with only modest proportions of sediment, whereas southern Great Basin calderas are mostly filled by ignimbrite.
As in the southern Great Basin, regional faulting that could generate major topographic barriers was absent during the main ignimbrite flareup; ash flows from western sources were able to traverse the future Rio Grande rift and spread widely to the east. Most ignimbrites (37–27 Ma) erupted prior to the surface expression of extensional faulting, through crust characterized by a neutral stress field, as indicated by development of radial dike swarms around central volcanoes. The two youngest major ignimbrites, the peralkaline Amalia Tuff (25 Ma) and Sunshine Peak Tuff (23 Ma), bridge the transition to inception of regional extension along the Rio Grande rift zone. Another similarity to the Great Basin is the near absence of basalt during the flareup.
Ignimbrites in the Southern Rocky Mountain field have a comparable range in silica to those in the southern Great Basin province, but they tend to be more alkalic (Figs. 16A and 7A) because of relative enrichment in Na; in this respect, the field is like the Western Nevada field (Fig. 7B). Southern Rocky Mountain tuffs tend to have less hornblende and quartz than Great Basin tuffs, but clinopyroxene is more common. This implies that the magmas were drier, hotter, or crystallized at a lower pressure than their equivalents in the Great Basin. Sr and Nd isotopic compositions suggest less of a metasedimentary upper-crustal component, especially for the San Juan sector; no thick miogeoclinal section underlies this region (Fig. 5; Lipman et al., 1978; Riciputi et al., 1995). The δ18O values of quartz are also lower, about 7.5‰ to 8‰, in the Fish Canyon Tuff, and as high as 8.2‰ in a few rhyolites (Larson and Taylor, 1986), compared to 8.3‰ to 11.1‰ for quartz from dacites and rhyolites in the central and eastern sectors of the Great Basin province (Fig. 10; Hart, 1997; Larson and Taylor, 1986). Just as in the southern Great Basin, compositions of ignimbrites from the Southern Rocky Mountain field display distinct temporal and geographic diversity. In comparison to the peak flareup in the central San Juan region, early northern ignimbrites (Sawatch trend) tend to be more alkalic (higher Zr and incompatible elements, more sodic sanidine) and have higher Sr isotope and δ18O values (Fig. 10). Ignimbrites erupted from southeastern and western San Juan sources also differ from the central locus, mainly crystal-rich dacites in the southeast, crystal-poor rhyolites from western calderas, all lacking large compositional zonations, and also characterized by subtle regional isotopic differences (Lipman et al., 1978; Lake and Farmer, 2015).
The largest eruption in the field was the 28.2 Ma Fish Canyon Tuff with a magma volume of 5000 km3 (Fig. 17A). The 34.2 Ma Badger Creek Tuff is a compositional clone, but only sparse remnants are preserved, making accurate volume estimates uncertain (>500? km3). Most samples of the Fish Canyon Tuff (Fig. 16A) are dacite, but some with about 68% silica and upwards of 8% total alkalis are trachydacite. Mineralogically, the Fish Canyon Tuff is nearly identical to the 29.20 Ma Lund Tuff and similar to other Great Basin phenocryst-rich dacites, or monotonous intermediates (Figs. 13; Bachman et al., 2002). Apparently all shared similar crystallization conditions—about 2.5 kbar and 750–800 °C under water-undersaturated and oxidized conditions (Johnson and Rutherford, 1989).
Tuffs compositionally similar to the Isom-type trachydacites are not obvious in the Southern Rocky Mountain field (Fig. 16). In comparison to the phenocryst-poor (<15%) plagioclase-pyroxene Isom-type tuffs, phenocrysts are more abundant in the Southern Rocky Mountain trachydacitic tuffs (typically 20%–25%). The mineral assemblage (plagioclase, sanidine, biotite, and sparse hornblende) appears to record wetter and cooler magmatic environments than for the Isom-type tuffs in the Great Basin. In the more alkalic Latir volcanic locus, the youngest and southernmost part of the Southern Rocky Mountain field, Isom-type tuffs are also absent, although some precaldera trachydacite lavas are chemically similar to western Nevada Isom types. In addition, comendite lavas and the Amalia Tuff could be evolved Isom-type magma in terms of Ca-Ti-Zr, and alkali contents. However, these high-Zr rocks are peralkaline rhyolites, not trachydacites like those from the Great Basin.
With the exception of the 5000 km3 Fish Canyon Tuff, volumes of individual ignimbrite eruptions are 1000 km3 or less (Lipman and Bachmann, 2015); however, extreme synvolcanic erosion of the early ignimbrites makes their volume estimates especially uncertain. Unlike the southern Great Basin province, where eruptive activity was spread more or less uniformly throughout the flareup (Fig. 13; see also Henry and John, 2013), the Southern Rocky Mountain field has a well-defined peak in explosive eruptions of silicic magma from 30.3 to 27.1 Ma, during which 16 ignimbrites, each 100 to 1000 km3, were erupted from the San Juan locus, before and following the 5000 km3 Fish Canyon culmination at 28.2 Ma (Fig. 17A). This peak activity involved two thirds of the ignimbrite eruptions in less than one fifth the time of all the activity, and it constituted four fifths of the total ignimbrite volume in the field.
A prominent feature of the Southern Rocky Mountain field is the subordinate total volume of silicic ignimbrite (>17,000 km3) relative to intermediate-composition lavas (∼40,000 km3) that built coalescing stratovolcanoes during early activity and makes up most of the postcollapse caldera fills (e.g., Lipman, 2007). In this respect, the field is like the Marysvale field, on the opposite side of the Colorado Plateau, in having more lava than ignimbrite (Table 2).
The crust beneath the Southern Rocky Mountain volcanic field is currently 40 to 45 km thick; high regional elevations are supported by low crustal densities, rather than greater crustal thickness such as that under the High Plains to the east of the mountain front (Prodehl and Lipman, 1989; Gilbert, 2012; Hansen et al., 2013; Lipman and Bachman, 2015). Modest late Cenozoic extension along the Rio Grande rift apparently has not significantly modified crustal thickness beneath the volcanic field.
Mogollon-Datil Volcanic Field
In this field on the southern side of the Colorado Plateau (Fig. 1), volume-time relations for ignimbrites are similar to those in the Southern Rocky Mountain field, with a less well-defined peak in eruptions but at about the same time (Fig. 17B). However, individual eruptive and total volumes are smaller; five eruptions were >900 to 2200 km3, and the total volume is >9000 km3 (McIntosh et al., 1992). Eruptive activity migrated west to northwest through time. The ignimbrites are mostly alkalic rhyolite and lesser trachydacite; dacite tuffs are rare (Fig. 18A), and Isom-type tuffs have not been reported. Thus, the field resembles the western sector of the southern Great Basin (Fig. 7B). Like the Marysvale and Southern Rocky Mountain volcanic fields elsewhere on the margin of the Colorado Plateau, but unlike the southern Great Basin province, intermediate-composition lavas constitute a major proportion of the volcanism (Table 3).
Sierra Madre Occidental Province
Products of the ignimbrite flareup in this remote province in Mexico (Fig. 1) cover an area of about 400,000 km2 to depths of as much as 1 km. Only about 10% has been mapped geologically, chiefly along its margins, and less than 10% of the estimated 350 major calderas are known (Swanson et al., 2006). The subduction setting was similar to that in the adjacent United States; a Laramide episode of flat slab subduction was followed by Oligocene rollback to eventually form a narrow arc above a steeply dipping subducting slab by the Miocene (Severinghaus and Atwater, 1990; Ferrari et al., 2007; McDowell and McIntosh, 2012). Almost three fourths (∼300,000 km3) of the explosive activity in the field occurred between ca. 34 and 28 Ma; a second pulse of magmatism occurred between 24 and 18 Ma, when another 100,000 km3 of silicic magma erupted (McDowell and McIntosh, 2012; Bryan and Ferrari, 2013).
Analyzed ignimbrites are dominantly rhyolite (Fig. 18B) with a tendency for alkalic compositions, but the full range in composition may not yet be documented. Dacite is rare, and no large-volume monotonous intermediate ignimbrites or Isom-type trachydacites have been reported (Ferrari et al., 2007; Bryan and Ferrari, 2013). Intermediate-composition lavas are subordinate to explosive silicic deposits; true basalt is rare to absent (Ferrari et al., 2007). Initial 87Sr/86Sr ratios range between 0.7041 and 0.7070 for the Eocene and Oligocene volcanic rocks (Ferrari et al., 2007). Thus, compositions of the tuffs in the Sierra Madre Occidental are similar to those in the western sector of the southern Great Basin province (Figs. 5 and 7B) and the Mogollon-Datil field (Fig. 18A). Basaltic andesite and andesite lavas became more common at the end of the Oligocene pulse (Cameron et al., 1989). They were accompanied by east-west extension on the margins of the province (Murray et al., 2013) and the eruption of high-silica rhyolites, including many F-rich topaz rhyolites (Christiansen et al., 1986). This Miocene pulse is similar to the transition to extensional magmatism seen in the United States. Murray et al. (2013) suggested that the ignimbrite flareup was synextensional but also recognized an unextended core in the province. Regional plate reconstructions show active subduction beneath central Mexico until about 12 Ma (e.g., Oskin et al., 2001; Dickinson, 2002). The geochemistry of the ignimbrites published so far is consistent with this tectonic setting.
Early determinations of crustal thickness in the Sierra Madre Occidental indicated values of 40–42 km, with thinning on the extended east and west flanks (Rivera and Ponce, 1986; Gomberg et al., 1988; Couch et al., 1991). However, a determination of 55 km (Bonner and Herrin, 1999) was used by Ferrari et al. (2007) to represent the unextended and magmatically thickened core of the province.
Central Andean Plateau
The late Cenozoic Central Andes has been recognized as a younger analog for the magmatism and tectonics of western North America, going back at least as far as Hamilton (1969). More recent detailed analyses include Coira et al. (1982, 1993), Allmendinger et al. (1997), and Kay and Coira (2009). These works establish the Central Andes as a useful model for the mid-Cenozoic Great Basin area. Thus, we review salient attributes of this region and show similarities in tectonic regime, volcanism, and geochemistry (de Silva et al., 2006; de Silva and Gosnold, 2007; Salisbury et al., 2011; Folkes et al., 2013) between these two regions.
The Altiplano-Puna plateau (Fig. 19A) is the widest and highest major element of the Central Andes. It is built on a platform of deformed Precambrian and Phanerozoic rocks. Since the Mesozoic, an oceanic plate has subducted beneath South America. To the north of ∼13°S and south of ∼27°S, where volcanism is currently inactive, the plate subducts at <30° to nearly flat. The sharp transitions in dip of the subducting slab and the variable age-dependent subduction dip along the continental margin (Capitanio et al., 2011) indicate the possibility of tears rather than abrupt warps in the slab. Plateau formation was a direct response to crustal shortening and thickening of the Central Andes, to as much as 70–80 km under the central plateau starting in the middle Cenozoic but continuing until as recently as the late Miocene. Crustal thickening is thought to be primarily the result of compressional failure of the thermally weakened (probably hydrated) continental lithosphere during flat slab subduction analogous to the North American Laramide orogeny (Isacks, 1988; Allmendinger et al., 1997). During this time, volcanism swept eastward to define a broad volcanic footprint. At approximately 30 Ma, a major change in subduction geometry is indicated by westward migration and focusing of volcanism, signaling the rollback and steepening of the slab essentially concurrent with an ignimbrite flareup. The Central Andean Neogene ignimbrite province produced by the flareup covers the length and breadth of the Central Andes (de Silva and Francis, 1991; de Silva et al., 2006; Freymuth et al., 2015) over an area of ∼250,000 km2, comparable to that of the flareup in the United States (Fig. 19). Like the mid-Cenozoic North American flareup, several distinct fields and associated calderas developed (Fig. 19). What little is known about the entire province suggests that it was time transgressive, with the age of volcanism decreasing from north to south (de Silva, 1989; Freymuth et al., 2015). Available isotopic data indicate that the older ca. 20 Ma ignimbrites in the north are less “crustal” than the younger <10 Ma ignimbrites south of 22°S (Freymuth et al., 2015). Seismic and petrologic evidence in the Central Andes indicates the presence of delaminated slabs of lower continental crust, which are believed to include dense garnetiferous residua from melting, as well as of mantle lithosphere sinking into the underlying hotter mantle (Kay, 2014); some of the uplift might have been the result of delamination of dense crustal material.
Ignimbrite Flareup in the Altiplano-Puna Volcanic Complex
A distinct flareup within the Neogene Central Andean ignimbrite province produced the most voluminous ignimbrite plateau in the southern half of the province from 22°S to 24°S, the Altiplano-Puna volcanic complex (Fig. 19A; de Silva, 1989). Here, the crust is now 58–76 km thick (Yuan et al., 2002; McGlashan et al., 2008). Ignimbrite eruptions began at ca. 10 Ma and continued until ca. 1 Ma, with a distinct pattern of waxing, climax, and waning in three main pulses: >2400 km3 at 8.41–8.33 Ma, >3000 km3 at 5.65–5.45 Ma, and >5400 km3 at 4.09–2.89 Ma (Fig. 20; Salisbury et al., 2011). This ignimbrite flareup created at least six major calderas as much as 60 km in diameter and several smaller ignimbrite shields (de Silva and Gosnold, 2007; Salisbury et al., 2011). The volume-time pattern of explosive activity is like that of the central and eastern sectors of the southern Great Basin (Fig. 13).
Ignimbrites in the Altiplano-Puna volcanic complex include dacite and rhyolite with minor trachydacite, andesite, and latite (Fig. 21). The compositional pattern of these ignimbrites resembles those of the eastern sector of the southern Great Basin ignimbrite province more closely than the dominantly rhyolitic and more alkalic fields previously discussed (Fig. 7A). Summaries in de Silva et al. (2006), de Silva and Gosnold (2007), Lipman and McIntosh (2008, their table 4), and Kay et al. (2010) underscore the similarities between the dacite ignimbrites in the Altiplano-Puna volcanic complex and eastern sectors of the Great Basin and the Southern Rocky Mountain volcanic field. These are phenocryst-rich, calc-alkaline, high-K dacitic to rhyodacitic ignimbrites of the “monotonous intermediate” type of Hildreth (1981). The 40% to 50% phenocrysts are a low-pressure assemblage of quartz, plagioclase, biotite, and Fe-Ti oxides with accessory apatite and titanite. Amphibole is common in some northern Puna ignimbrites near the arc, minor amounts of clinopyroxene and orthopyroxene can occur in ignimbrites with <67% SiO2, and sanidine is found in some dacite and rhyolite units. Pre-eruptive temperatures are generally near 700 °C to 850 °C, with pressures before eruption corresponding to depths of 5 to 10 km, while magmatic oxygen fugacities (∼2 log units above Quartz-Fayalite-Magnetite [QFM]) are similar to many silicic tuffs of the Great Basin. Detailed information on specific monotonous intermediate ignimbrites in the Central Andes is available from Lindsay et al. (2001) for the 3.9 Ma, 2500 km3 Atana ignimbrite, and Schmitt et al. (2001) for the 1 Ma, 100 km3 Purico ignimbrite and 3.69 Ma, >800 km3 Tara ignimbrite. Rhyolites are less common, the largest being the ∼300 km3 Toconao ignimbrite, which is interpreted to be from the upper zone of the Atana-Toconao magma reservoir (Lindsay et al., 2001).
A suite of peraluminous ignimbrites in the eastern Altiplano-Puna volcanic complex (Figs. 10 and 19) has the most “crustal” Sr and Nd isotopic compositions, attesting to the influence of a metapelitic basement in the east (Caffe et al., 2012). These are also monotonous intermediates and testify to hypersolidus temperatures in the orogenically thickened crust, in a similar manner as the Late Cretaceous peraluminous granites just north of the caldera complex in the eastern sector of the Great Basin province.
No Isom-type trachydacitic tuffs have been recognized in the Altiplano-Puna volcanic complex.
Andesitic components are found as rare individual pumices in some ignimbrites or more commonly as bands and inclusions in pumices in most ignimbrites (de Silva, 1991; Lindsay et al., 2001; Schmitt et al., 2001; Burns et al., 2015). While these are volumetrically insignificant, they provide evidence for the role of intermediate magmas in the evolution of the dacitic magmas, representing the thermal and material input from the deep roots into the upper-crustal system (Burns et al., 2015). The most mafic lavas in the Altiplano-Puna volcanic complex are small andesitic lava flows with 54%–58% silica in many of the calderas. These lavas have compositions similar to the andesitic pumices and mafic enclaves (e.g., Burns et al., 2015), and they reveal the ubiquity of relatively mafic compositions at depth. Much of the underlying zone of partial melt is considered to be of similar composition to these mafic lavas. Ages are rare and of low precision for scattered, rare, large andesitic-dacitic composite volcanoes; some are clearly older than the flareup based on superposition, while a few are intercalated with the ignimbrites, or have ignimbrites draped around them. In the several locations where pre-ignimbrite basement is seen beneath the ignimbrites, no evidence of mafic lavas preceding the flareup has been found. Current volume estimates emphasize the dominance of silicic ignimbrite over contemporaneous andesitic lavas; a ratio of 10:1 is considered a minimum, because erosion rates may artificially bias the estimate toward the andesites.
Oxygen isotope ratios in the volcanic rocks of the Altiplano-Puna volcanic complex and Cerro Galan are high (δ18OQtz from +7.8‰ to +10.2‰), and initial 87Sr/86Sr ratios (0.709–0.718) are the highest of the flareup provinces (Fig. 10; Kay et al., 2010, 2011; Folkes et al., 2013). They are interpreted to be caused by extensive assimilation of high Rb/Sr, high δ18O metasedimentary rock during magma generation. Folkes et al. (2013) posited that the absence of low δ18O magmas may result from the plateau’s high elevation and related arid climate. Low δ18O magmas generally originate by assimilation of hydrothermally altered crust in wetter environments. However, the Southern Rocky Mountain tuffs lack low δ18O values (+6.5‰ to +9.9‰), despite a wet environment—implied by the presence of caldera-filling lake sediments containing conifer pollen and leaf debris (e.g., Axelrod, 1987). Among large-volume ignimbrites of North America (e.g., Southern Rocky Mountain, Snake River Plain, Yellowstone), the closest analogs to the Central Andes are the ignimbrites of the mid-Cenozoic of the eastern Great Basin, where δ18OQtz ranges from 8‰ to 11‰ (Fig. 10; Larson and Taylor, 1986; Hart, 1997). These findings constitute indirect support for the existence of a similarly high, dry, and unextended Great Basin altiplano during the mid-Cenozoic flareup (Fig. 3).
Viramonte et al. (1984) and Riller et al (2001) noted that many of the eruptive sources of the ignimbrites in the Altiplano-Puna volcanic complex lie along plateau-wide northwest-southeast–trending zones of right-lateral transtension. Riller et al. (2000) argued that the ignimbrite flareup coincides with a local change from compressional tectonics to plateau collapse. Indeed, the outlines of the calderas have some northwest-southeast control; however, a cause and effect relationship between tectonism and eruption remains to be demonstrated. We note that Kay and Coira (2009) made no mention of contemporary crustal extension during the latest Cenozoic ignimbrite volcanism; instead, contemporaneous deformation was contractional. Overall, none of the Central Andean, southern Great Basin, or Southern Rocky Mountain ignimbrite flareups was accompanied by obvious crustal extension.
Geophysical Underpinnings of the Altiplano-Puna Region
What is believed to be the largest active midcrustal zone of partial melt in Earth’s continental crust underlies the Altiplano-Puna volcanic complex—the so-called Altiplano-Puna magma body (e.g., Zandt et al., 2003; Prezzi et al., 2009; del Potro et al., 2013; Ward et al., 2014; Comeau et al., 2015). A negative 335 mGal Bouguer gravity anomaly (with “depth” of about 50 mGal) coincides with a low-seismic-velocity zone (Fig. 22A) that defines a 200-km-wide body between 9 and 30 km below the surface, which is interpreted to be 500,000 km3 of partially molten rock (Ward et al., 2014). Magnetotelluric studies (Fig. 22B) show a deep highly conductive zone (top at 25–30 km below the surface); fingers of conductive material (magma or magma plus saline fluid) extend above this nearly to the surface (Comeau et al., 2015). Gravity models also show what are interpreted to be diapirs of low-density material rising from the Altiplano-Puna magma body (del Potro et al., 2013). These geophysical studies suggest that the melt fraction below the Altiplano-Puna volcanic complex ranges from ∼4% to 25%. The Altiplano-Puna magma body is interpreted to have produced the magmas that erupted during the flareup (e.g., de Silva et al., 2006; Muir et al., 2014; Burns et al., 2015) and could be an extensive zone where melting, assimilation, storage, and homogenization (MASH) modify mantle-derived magma (e.g., Hildreth and Moorbath, 1988) below upper-crustal magma reservoirs (Fig. 22). Annen (2009) called this a hot zone.
Summary Comments on Comparisons of Ignimbrite Flareups
In the continental arc ignimbrite flareups described herein (Tables 1 and 2), numerous caldera-forming eruptions of explosive silicic magma occurred over 10–18 m.y., with repose times ranging from a few hundred thousand to as much as a few million years. Super-eruptions were common. Aggregate volumes of ignimbrite in an individual field are at least several thousands of cubic kilometers, blanketing upwards of 105 km2 surrounding source calderas tens of kilometers in diameter. Contemporaneous basalt is rare to absent, although significant andesitic precursors or evidence for mafic input through the silicic magmatic history is present. Flareups occurred far inland on thickened, nonextending crust, beneath which a previously flat subducting lithospheric slab rolled back to steeper dip. This distinct style of continental arc volcanism contrasts sharply with the near-trench, steady-state volcanism overlying steeply dipping slabs. Of all the ignimbrite flareups examined, the one most similar to the eastern sector of the southern Great Basin province is the Altiplano-Puna volcanic complex. Similarities include: calc-alkaline rhyolite and dacite ignimbrites (Figs. 7A and 21A), multiple super-eruptive monotonous intermediate ignimbrites (Figs. 13A and 20), similar Sr and O isotope ratios (Fig. 10), only minor coeval andesitic lavas (Figs. 2 and 19), and spatially associated production of peraluminous silicic magmas originating from crustal anatexis. Notably, both flareups occurred on the thickest (∼70 km) orogenic crust. Nonetheless, the Andean volcanic complex differs in its shorter flareup time span, smaller total volume of ignimbrite, and absence of Isom-type trachydacite.
The Southern Rocky Mountain and Mogollon-Datil volcanic fields on the margin of the Colorado Plateau (Fig. 1; Tables 1 and 2) resemble the Altiplano-Puna volcanic complex with respect to flareup duration, ignimbrite volume, and absence of Isom-type tuffs. However, these two fields differ from the Altiplano-Puna complex and the eastern sector of the southern Great Basin province in having smaller volumes of ignimbrite than intermediate-composition lava (Tables 1 and 2; Fig. 20) and the Mogollon-Datil volcanic field has no monotonous intermediate ignimbrite. Also, the Colorado Plateau crust had experienced less orogenic thickening than that in the Central Andes and the southern Great Basin, where we postulate a longer-lived mid- to upper-crustal MASH zone (Fig. 23) efficiently blocked ascending mafic magmas. The Colorado Plateau crust was not only thinner, but also more mafic; and partial melting resulted in more alkalic ignimbrites with lower initial 87Sr/86Sr ratios compared to the calc-alkaline ones of the eastern Great Basin and Central Andes (Figs. 5, 7A, 10, 16A, and 18). Exceptions to this generalization are the Grizzly Peak Tuff, which was underlain by crust that was isotopically distinct from that below the San Juan locus, and New Mexico’s Amalia Tuff, which was extensively contaminated by upper-crustal materials (Fig. 10; Johnson and Fridrich, 1990; Johnson et al., 1990).
ORIGIN OF SLAB-ROLLBACK IGNIMBRITE FLAREUPS
Voluminous ignimbrite flareups related to rollback of a formerly flat subducting plate are a distinct style of continental arc volcanism that are powered by an elevated influx of mantle basalt magma into thickened crust. However, little of this basalt magma is extruded at the surface. Before considering how these basalt magmas were created in the mantle, we examine how the thickness and composition of the crust modulated the influence of this flux of mantle-derived magma during ignimbrite flareups. The southern Great Basin province provides a focal point for our discussion.
All of the flareups developed on crust at least 40 km thick that was not subject to significant contemporaneous tectonic extension. Concurrent extrusions of andesitic lavas were subordinate to silicic ignimbrites where the crust was thicker in the southern Great Basin and Altiplano-Puna, whereas apparently thinner crust, as in the Southern Rocky Mountain and Mogollon-Datil volcanic fields, allowed more andesitic magmas to rise to the surface. As discussed in detail in the following, thicker crust played two significant roles in flareups:
(1) Adjustment of the thermal gradient after orogenic thickening led to elevated temperatures in the lower to middle crust.
(2) Ample fertile source rock was available to create voluminous silicic magmas, and the compositional profile of the crust governed the details of ignimbrite composition.
Elevated Temperatures in Thick Crust
In the eastern Great Basin, where the mid-Cenozoic crust was an estimated 70 km thick (Fig. 3), Late Cretaceous two-mica granites with initial Sr isotope ratios as high as 0.7246 (Best et al., 1974; Lee and Christiansen, 1983; Farmer and DePaolo, 1983) are exposed 50–150 km north of the Indian Peak caldera complex (Fig. 2). These strongly peraluminous granites originated without a mantle contribution by partial melting of metasedimentary rocks where orogenic thickening was greatest. After thickening and thermal re-equilibration, temperatures in the lower to middle crust can reach near-solidus values (e.g., England and Thompson, 1984; Patiño-Douce et al., 1990; Vanderhaeghe et al., 2003). As previously indicated, peak metamorphic conditions as great as 800 °C at about 35 km depth prevailed in eastern Nevada in the Late Cretaceous (DeCelles, 2004). From thermal models, England and Thompson (1984) concluded that temperatures in the middle and lower crust increased for 40 to 60 m.y. after the crust doubled in thickness and then slowly decreased. Thus, following the Late Cretaceous contraction, the Great Basin crust would have been anomalously warm in the Eocene and Oligocene. Then, as huge volumes of basaltic magma intruded the crust during the mid-Cenozoic slab rollback, the added heat and feedbacks among crustal temperature, thermal diffusivity, and melt production (e.g., Whittington et al., 2009; Nabelek et al., 2012) resulted in copious production of silicic magmas of the Great Basin ignimbrite flareup. Orogenically thickened and prewarmed crust invaded by mantle magmas resulted in elevation of the geothermal gradient as well as the brittle-ductile transition, creating a largely ductile crust (de Silva and Gregg, 2014). In the Southern Rocky Mountain volcanic field, where little orogenic thickening had occurred, precursory large-scale andesitic magmatism resulted in a similar modification of the thermomechanical regime of the crust. This modified regime had at least three significant consequences for flareups:
(1) Instead of ascent through extensional fissures, as in some arc volcanism (e.g., Hughes and Mahood, 2008; Cashman and Sparks, 2013), silicic magma transport through the ductile crust during ignimbrite flareups likely involved rising diapirs. This transport process probably prevailed in the southern Great Basin, the Altiplano-Puna, and most of the Southern Rocky Mountain volcanic field. Diapiric magma transport was the natural result of a neutral stress field in a ductile crust (e.g., Vigneresse and Clemens, 2000; Weinberg and Podladchikov, 1994). Maughan et al. (2002) argued that a quasi-isotropic horizontal stress field prevailed during the flareup in the Great Basin, as it did in the Southern Rocky Mountain volcanic field, with the exception of the late Questa-Latir, New Mexico, locus (Lipman, 1982, 1988), and also during much of the post-thickening history of the Central Andes, particularly in the deeper, more ductile parts of the crust. The diapirs were probably spawned in crustal MASH zones (Figs. 22 and 23). In the southern Great Basin and Altiplano-Puna, where the crust was thickest, the MASH zone appears to have been in the mid- to upper crust, based on geophysical data in the latter, whereas in the Southern Rocky Mountain field, it was probably in the lower crust, based on the isotopic compositions of the rocks (Lipman et al., 1978; Lake and Farmer, 2015).
In the Central Andes, seismic, gravity, geodetic, and magnetotelluric data gathered by Fialko and Pearse (2012), del Potro et al. (2013), Ward et al. (2014), and Comeau et al. (2015) indicate a 10- to 15-km-diameter diapir is actively rising from the top of the mushy Altiplano-Puna magma body 20 km below the surface (Fig. 22). Buoyant ascent probably results from Rayleigh-Taylor instability in the partially molten zone and is fed by inward migration and collapse; this radial mass transfer has created a broad zone of surface subsidence about 120 km in diameter.
(2) Barring other factors such as crystallization or crustal structures that might stall rising magma at even deeper levels, the ultimate height to which diapirs of silicic magma can ascend within the crust is not governed by the level of neutral buoyancy because their densities (2.3–2.5 g/cm3) are consistently less than that of most crustal wall rocks (2.5–2.9 g/cm3); even solid granite is less dense than most enclosing wall rocks. Instead, ascending diapirs of silicic magma are probably trapped at the brittle-ductile transition, where the stress required to fracture the brittle lid is two orders of magnitude greater than that provided by the driving force for diapirism resulting from the contrast in density between magma and host rock (Vigneresse, 1995). However, the transition level is not static, but it is further elevated as a result of upward transfer of heat from the top of the entrapped and spreading magma. For crystal-poor rhyolite ignimbrites of the Southern Rocky Mountain volcanic field, the pressure of crystallization has been estimated at only 1 to 2 kbar (4 to 8 km) on the basis of experimental phase equilibria, particularly, absence of quartz in quartz-normative magma (Lipman et al., 1978; Bachmann et al., 2014). The depth of the brittle-ductile transition beneath the eastern sector of the Great Basin province, as indicated by the equilibration pressure of phenocrysts in the three pre-eruption monotonous intermediate magmas (Best et al., 2013b), was as shallow as 7–9 km. Mineral barometers in the dacites and rhyolites in the central sector where the crust may have been slightly thinner (Fig. 3) indicate greater depths of ∼14 km (Phillips, 1989; Radke, 1992; Blaylock, 1998).
Whereas the Southern Rocky Mountain volcanic field experienced significant precursory andesitic volcanism before the ignimbrite flareup, which probably played some role in elevating the brittle-ductile transition, the Great Basin had no such significant precursory volcanism. Instead, its greater crustal thickening, sufficient to yield late Mesozoic S-type granite magmas, influenced the elevation of the transition.
(3) The characteristic accumulation during flareups of colossal volumes of silicic magma (several hundreds to thousands of cubic kilometers) in the shallow crust before eruption at intervals of 104 to 106 yr was made possible by a ductile crust. Recent theoretical considerations highlight the importance of the rheology of the wall rock in the growth of a magma reservoir (Jellinek and DePaolo, 2003; Gregg et al., 2012, 2013, 2015). Largely a function of the thermal energy associated with the reservoir and its history, small chambers (<100 km3) are likely to be surrounded by cool (<400 °C), brittle wall rocks, while larger chambers that yield super-eruptions are built under prograde conditions that result in ductile wall rocks (Fig. 24). Absent tectonic or other external influence, smaller eruptions are typically triggered by dike propagation when the elastic limit (∼2× tensile strength) of the wall rocks is exceeded by an overpressure (e.g., through recharge or fluid saturation) in the chamber. For larger chambers of eruptible magma surrounded by warmer ductile wall rocks, theoretical models indicate such diking is inhibited. Eruption is suppressed, and a reservoir grows until the roof breaks (Fig. 24). This mechanical threshold is a function of the aspect ratio of the roof; an uplifting and extending roof above the growing reservoir has a finite mechanical strength that is eventually exceeded, and faults may propagate from the surface to the chamber, triggering eruption and caldera collapse. In one scenario, when pressure on the magma is reduced, it becomes fluid-saturated and expands explosively through these fracture vents. The long repose times of large (>1,000 km3) caldera-forming eruptions during ignimbrite flareups (e.g., Figs. 13, 17, and 20) are consistent with some studies of zircon age spectra (Brown and Fletcher, 1999; Schmitt et al., 2003; Costa, 2008; Bachmann et al., 2007; Wotzlaw et al., 2013) and thermal models (Spera, 1980; Annen et al., 2009; Gelman et al., 2013; de Silva and Gregg, 2014) that indicate extended melt-rich lifetimes of several hundred thousand years.
Wholesale expulsion of magma along outward-dipping (reverse) ring faults along caldera margins provides increasingly open channels as subsidence progresses (Fig. 24; Best et al., 2013b, see their figs. 43 and 52). Resulting catastrophically high eruption rates yield sustained and continuously collapsing columns of ejecta. Precursory plinian eruptions (e.g., Druitt and Sparks, 1984) are effectively bypassed, as appears typical of all the flareup fields discussed (e.g., de Silva et al., 2006; Lipman et al., 2015).
Diapiric feeding of a shallow magma chamber is likely pulsed, with long periods of time between pulses, because it takes time to regenerate the gravitational instability of buoyant magma. This concept might apply to the exceptional burst at ∼1 m.y. intervals of three monotonous intermediate magmas erupted at the same focus from 31.1 to 29.2 Ma in the Indian Peak caldera complex in the eastern sector of the southern Great Basin province (Figs. 13A and 14). The three super-eruptions, totaling 12,300 km3, were fed from similar depths of about 7–9 km; each batch of magma evolved somewhat independently, as indicated by their distinctive chemical variation trends (Best et al., 2013b). We envisage a scenario in which a hybridized mixture of andesite and rhyolite magma rose diapirically from a midcrustal MASH zone and stalled at the brittle-ductile transition in the upper crust (Fig. 23). The mixed crystal-rich dacite magma accumulated until the roof of the chamber failed, resulting in eruption of the Cottonwood Wash Tuff and caldera formation at 31.1 Ma. Subsequent pulses of similar crystal-rich dacitic magma ascended through the weakened conduit already established in the crust. These new drafts of magma were also trapped in much the same place as the older chamber until the growing Wah Wah Springs magma reservoir grew to the point of roof failure and erupted at 30.06 Ma. Finally, the third super-eruptive monotonous intermediate magma—the Lund—accumulated and vented at 29.2 Ma in much the same way and in the same place as the two preceding ones, forming a nested caldera complex. Similar processes have been suggested for the Altiplano-Puna volcanic complex (de Silva et al., 2006; de Silva and Gosnold, 2007; de Silva and Gregg, 2014), and they probably apply to the accumulation of large magma bodies in other ignimbrite flareups.
In regions where the crust was not as thick, and the pre-flareup geothermal gradient was lower, such as in the Southern Rocky Mountain field, a greater volume of mantle-derived magma may have been required to overcome this energy deficit and produce voluminous bodies of silicic magma. A higher flux might also explain the much greater volume of erupted andesite in the Southern Rocky Mountain province (estimated to exceed 40,000 km3; Table 2) than in the coeval Great Basin province, where intermediate composition magmas were much less voluminous (6000 km3; Table 1). Farmer et al. (2008) estimated the volume of basaltic magma that powered the San Juan segment of the Southern Rocky Mountain volcanic field to be 384,000 to 719,000 km3. In contrast, Christiansen and Best (2014) estimated that 300,000 to 400,000 km3 of basalt powered the Indian Peak–Caliente volcanic field.
Composition of the Crust
Hughes and Mahood (2008) argued that crustal composition, independent of its thickness, influences the style and composition of caldera-related magmatism in volcanic arcs. Arcs on oceanic crust have fewer rhyolite calderas per unit length than arcs built on Mesozoic or older continental crust. However, this is the opposite of the southern Great Basin ignimbrite province: Caldera-related, relatively alkalic rhyolite ignimbrites dominate the western sector, which is underlain by young, thinner, mafic crust, whereas dacites dominate the eastern sector, underlain by old, thicker, felsic crust (Figs. 3, 5, and 10).
Experiments show how magma generation in these two contrasting sectors of the Great Basin is related to different crustal sources and contaminants. Partial melting of arc basaltic rocks, like those that underlie the western sector, produces trachydacites and high-alkali rhyolites at low degrees of melting (e.g., Sisson et al., 2005; Whitaker et al., 2007). In contrast, melting of more felsic crustal rocks with quartz or with a large metasedimentary component creates calc-alkaline dacites and rhyolites (e.g., Patiño-Douce and Harris, 1998; Holtz and Johannes, 1991). These latter melts typically have lower Na/K ratios as well. These contrasting experimental melts are compared with magmatic trends in the southern Great Basin in Figure 25. Many of the more alkalic rhyolite and trachydacite ignimbrites in the western sector (Fig. 7B) are similar to the experimental melts of gabbro (i.e., arc basalt; 0.10 to 0.17 melt fractions in Sisson et al., 2005). The western rhyolites, therefore, could have been direct melts of this type of mafic crust; initial Sr isotope ratios are low and permissive of this interpretation (Fig. 5; Best et al., 2013c). Alternatively, they could have been produced when differentiating mafic magmas hybridized with such crustal melts. In any case, thinner, cooler mafic crust, like that below western Nevada, yielded small batches of alkali-rich rhyolite and lesser trachydacite where it was intruded and heated by mantle-derived basalt. On the other hand, thicker, hotter, felsic crust like that in the eastern sector yielded large batches of calc-alkaline rhyolite magma (melt fractions of 0.20 to 0.30). This rhyolite magma then interacted with the flux of mantle-derived magma to create the andesite-dacite-rhyolite spectrum, spawned in the midcrust. Sr isotope ratios show that the erupted ignimbrites are unlikely to have been pure crustal melts—a significant mantle component was required (Fig. 10).
Another way that crustal composition modulates ignimbrite compositions is revealed by the seemingly unique Isom-type trachydacite magmas in the southern Great Basin that are especially evident in the two eastern sectors. These hot, dry trachydacite magmas appear only to have formed when and where the crust was less fertile and incapable of yielding rhyolitic melts to mix with evolving mafic magmas. Infertility was caused by prior extraction of low-melting-point felsic material during the large-scale episode of monotonous intermediate magma production, or by hybridization of the crust by massive intrusion of mafic magma. Differentiation of mafic magmas without extensive crustal contamination is the hallmark of Isom-type magmas (Best et al., 2013b), but we cannot rule out the influence of other tectono-magmatic factors in their origin.
Mantle Contributions to the Southern Great Basin Ignimbrite Flareup
Input of mantle-derived basalt magmas into thick crust is fundamental to generation of ignimbrite flareups, but where and how were these empowering but unseen magmas generated?
Lithospheric Delamination or Dripping
One way to produce basaltic magma is by decompression of asthenospheric mantle as it rises and takes the place of dense delaminated mantle lithosphere and crust. Seismic data are consistent with such a mechanism for the ignimbrite flareup in the Central Andean plateau (e.g., Kay and Kay, 1993; Kay et al., 2014). These authors speculated that garnet was stabilized at the expense of plagioclase deep in the orogenically thickened crust below the plateau, causing subsequent delamination of the dense eclogitic lower crust. This triggered more magmatism by decompression melting of asthenosphere that flowed in to take its place; the extent of melting might have been limited by thick lithosphere, unless it had already sloughed away. DeCelles et al. (2009) have made lower-crustal delamination and decompression melting an integral element in their model of cyclicity in Cordilleran orogenic systems.
In view of the evidence for crust as thick as about 70 km beneath the southern Great Basin ignimbrite province during the flareup (Best et al., 2009), dense eclogite should have developed in the lower crust. If this was the case, then some of the lower crust of the Great Basin may have been removed by delamination, or dripping, in the closing stages of or after the ignimbrite flareup (Best et al., 2009). If the crust during the ignimbrite flareup was 70 km thick, after 50% postvolcanic extension (Best et al., 2013a), it should now be 45 km thick, compared to a current thickness of about 30–35 km (e.g., Allmendinger et al., 1987; Gilbert, 2012). Thus, the missing 10–15 km of crust might have been removed after the flareup. Another possibility to account for the missing 10–15 km of crust is that an eclogitic lower-crustal residue became seismically indistinguishable from the underlying mantle; in other words, the seismic Moho is imaged at a shallower depth than the petrologic Moho.
Steepening of the Subducting Plate
Rollback of a formerly flat slab is an essential feature of flareup models for southwestern North America (e.g., Coney, 1978; Lipman, 1980; Best and Christiansen, 1991) and for the Central Andes (e.g., de Silva and Gosnold, 2007; Kay and Coira, 2009). However, the exact nature of the flat slab subduction and especially the thermal regime of the slab and its dehydration, or lack thereof, are subjects of current debate (e.g., English et al., 2003; Usui et al., 2003; Humphreys, 2009; Jones et al., 2011, 2015); this lends a degree of uncertainty to inferences regarding the dynamic interaction of the crust and mantle during the flareup. For the southern Great Basin, Christiansen and Best (2014) concluded that it might have been difficult to derive enough mantle magma to drive silicic magma production from the lithospheric mantle alone (contrast this with the conclusions of Farmer et al.  and Lake and Farmer  for the Southern Rocky Mountain field). Thus, a major contribution from a sublithospheric source for the mafic magma may be required. Furthermore, English et al. (2003), Usui et al. (2003), and Liu et al. (2016) argued that the slab was only partly dewatered during the flat subduction stage (for more general models for slab dewatering, see also Rupke et al., 2004). We suggest that additional dehydration accompanied heating during slab rollback and immersion in the hotter asthenosphere, providing a major generating mechanism for basaltic partial melts in the overlying mantle wedge. We do not view this as a one-time event; ongoing subduction would continue to fuel dehydration and flux melting for several millions of years after rollback started.
Tears in the Subducting Plate
The manner of slab rollback was also significant for the flareup in western North America. Lipman (1980) identified the Jemez trend (a northeast-trending volcanic lineament) in New Mexico and interpreted it to be controlled by northeast -trending structures in the continental lithosphere and by an offset in the downgoing slab of oceanic lithosphere. McQuarrie and Oskin (2010) proposed that a major discontinuity in the pattern of migrating volcanism in southwestern North America (Fig. 26A) was the result of a tear in the subducting slab as it rolled back beneath the continental plate. This tear is roughly coincident with the Pioneer fracture zone and the hypothetical contact between the Hess oceanic plateau to the south and the Shatsky plateau to the north (Liu et al., 2010). The buoyant Shatsky Plateau apparently inhibited rollback and the development of a mantle wedge beneath the Colorado Plateau, thus accounting for the relatively small amount of unusual magmatism in its interior.
A nonuniform rollback of the Farallon slab is envisaged to have begun beneath the Pacific Northwest in the Eocene and migrated southward into Nevada-Utah-Colorado in the Oligocene (Lipman, 1980, 1982; Severinghaus and Atwater, 1990; Humphreys, 1995). Seismic tomography has confirmed the existence of remnants of the torn, segmented Farallon slab beneath the northern United States (e.g., Sigloch et al., 2008; Humphreys, 2009). Colgan et al. (2011) postulated that an Eocene–Miocene tear in the subducting Farallon plate resulted in the offset of the western Cascade volcanic arc in coastal Washington and Oregon from the Ancestral Cascade arc in northeastern California and Nevada.
In the southwestern United States (Fig. 26B), steepening of the slab, as indicated by the pattern of migrating magmatism, was relatively uniform from the eastern to western margins of the Colorado Plateau, that is, from the Southern Rocky Mountain volcanic field to the Marysvale field (Fig. 1). However, to the west in the Great Basin, the migration isochrons are less uniform, prompting us to speculate on the existence of additional tears in the slab. An apparent southward jog in the migration isochrons implies a pair of tears existed in the subducting plate before ca. 24 Ma. Between the tears, the slab rolled back to a much greater extent by 36 Ma, less so at 38 Ma and 30 Ma, relative to the rollback to the east and west of the paired tears (Fig. 26C). The region of the subducting slab between the two postulated northerly trending tears coincides with the central and eastern sectors of the Great Basin where a colossal volume (58,000 km3) of mid-Cenozoic ignimbrite was erupted onto the thickest crust of the Great Basin altiplano east of the Precambrian continental margin and west of the stable Colorado Plateau (Table 1; Figs. 3 and 13). Between the two tears, the 36 to 24 Ma isochrons are closely spaced, implying a steeper slab. This might have resulted in a more focused release of fluids from the slab fragment and caused greater mantle melting and silicic magma production in the crust, thus accounting for the great eruptive volume in these sectors. As the slab dropped away from the overlying continental plate between the tears, hot asthenospheric mantle invaded the growing wedge region from around the edges of the slab (Fig. 26C). This decompressing asthenosphere could have experienced some partial melting, but the degree of melting might have been limited by the thickness of the overlying lithosphere, unless decompression was concurrent with lithospheric delamination or dripping. As the hot asthenosphere contacted the base of the partially hydrated lithosphere, additional melting could have taken place. In concert with fluids released from the heated underlying slab, these secondary melting processes are likely to have resulted in an unusually high flux of basaltic magma into the crust beneath the central and eastern sectors. The chemical and isotopic compositions of some ignimbrites, such as those of the Isom type (Best et al., 2013c, their p. 57), suggest they have components from flux melting and from decompression melting of the asthenosphere. Some of the basalt magma produced near the east tear in southwestern Utah otherwise destined for the Marysvale volcanic field could have moved west of the tear to feed the eastern sector (Fig. 26C), thus accounting for the striking contrast in the volume and nature of volcanism in these two fields.
The nested calderas in the central and eastern sectors of the Great Basin (Figs. 2 and 14) testify to pulsed ascent of diapirs along the same established path from the well-developed MASH zone fed by an unusually high flux of basaltic magma into the crust between the two tears.
The western of the paired tears (Fig. 26B) is close to the eastern boundary of the western sector of the Great Basin province (Figs. 2 and 6), where a more gradual slab rollback took place. In this sector, caldera-forming eruptions were distributed northward over a larger area, and apparently fewer were nested, compared to the east. Subsequently, the gently dipping segment of the plate seems to have taken on a “normal” subduction configuration with a northerly strike approximately along the present eastern Sierra Nevada front, creating the Ancestral Cascade arc (Christiansen and Lipman, 1972; Cousens et al., 2008; Colgan et al., 2011).
Volcanic activity related to tears and segmentation of subducting slabs appears to be common in other arcs. In the case of the late Cenozoic Andean ignimbrite flareup, a possible tear lies south of the Cerro Galan and Altiplano-Puna volcanic complexes (Fig. 19A). Elsewhere, tears have been discussed for the Sunda arc (Page et al., 1979, p. 575; Pesicek et al., 2008) and the Eolian arc (Faccenna et al., 2011). In the Central American arc, slab tears have been proposed by Dougherty and Clayton (2014), while Ferrari et al. (2007) argued for slab rupture beneath the Sierra Madre Occidental, and a tear associated with rollback has been identified beneath the Trans-Mexican volcanic belt (Ferrari et al., 2012).
IGNIMBRITE FLAREUPS AND SUBDUCTION
According to Bryan and Ferrari (2013) and Cather et al. (2009), the mid-Cenozoic ignimbrite flareup of southwestern North America was part of an intraplate silicic large igneous province. However, the criteria laid out by Bryan et al. (2008) and Bryan and Ferrari (2013) for such a province are not consistent with the southwestern North America flareup, which was clearly related to subduction beneath the margin of the North America plate. The ignimbrite flareup followed a long period of prior contractional deformation, whereas regional tectonic extension mostly followed the flareup. No basalt, much less flood basalt, accompanied the flareup, and the compositions of the volcanic rocks have more geochemical similarities to those in arcs than in intraplate settings. For example, flareup magmas in the mid-Cenozoic southern Great Basin province contrast strongly with those in silicic large igneous provinces like the Yellowstone hotspot (Fig. 10; Christiansen, 2001, 2005; Christiansen and McCurry, 2008) or Africa-Arabia (Ukstins-Peate et al., 2005). Magmas in large silicic igneous provinces are typically rhyolitic, with low fO2 and fH2O, elevated concentrations of high field strength elements (with low solubilities) like Nb and Zr, and low ratios of large ion lithophile (soluble) elements to high field strength elements (e.g., La/Nb; Ba/Nb). Additionally, they have high eruption temperatures for a given silica content and, importantly, form bimodal suites with chemically distinctive basalts. The volcanic rocks of the southern Great Basin and other Cenozoic ignimbrite flareups summarized here range from basaltic andesite to rhyolite and formed from magmas that were mostly cool, wet, oxidized, and typically, but not universally, magnesian rather than ferroan, and calc-alkalic rather than alkalic. The dacites and rhyolites are depleted in high field strength elements and have high Ba/Nb (and other large ion lithophile element to high field strength element ratios). These attributes reflect their subduction heritage. The distinctive Isom Formation ignimbrites could mark a switch to intraplate magmatism, but they retain enrichments of large ion lithophile elements, high Ba/Nb ratios, and higher fO2 like their predecessors and lack a bimodal association with erupted basalt. Moreover, subduction-related rhyolites in the southern Great Basin both preceded and followed after eruption of the Isom ignimbrites.
It is clear that Cenozoic voluminous ignimbrite flareups of the western American Cordillera were fundamentally subduction-related features and not formed by intraplate processes, although one can certainly transition to the other.
Continental margin arc volcanism above subducting oceanic plates is expressed by a broad spectrum of magma systems and corresponding surface manifestations (Fig. 27). The more familiar, near-trench, steady-state activity is represented by many arc systems around the globe. These volcanic arcs generally form on crust thinner than 40 km and have relatively frequent and small eruptions building stratovolcanoes of mafic to intermediate lavas and minor pyroclastic rocks. In contrast, slab-rollback ignimbrite flareups constitute a distinct, recognizable style of arc volcanism. Their attributes include: several tens of explosive, caldera-forming eruptions of dacitic, rhyolitic, and trachydacitic magma over 10–20 m.y periods; eruptions that are relatively infrequent (recurrence intervals of 104 to 106 yr) but individually of large magnitude (102 to 103 km3); and common super-eruptions (>103 km3). Individual outflow sheets extend to 150 km from the caldera source and cover as much as 30,000 km2 to thicknesses of several hundred meters, even where no prior topographic low existed. An entire composite field surrounding a multicyclic caldera complex can blanket 100,000 km2. Little or no basalt lava is extruded, but derivative andesite magmas play significant roles in production of silicic ignimbrite magma, even where their extruded volume is as little as one tenth the volume of contemporaneous silicic ignimbrite.
The mid-Cenozoic southern Great Basin ignimbrite province serves as a prime example of this ignimbrite flareup style of continental arc volcanism (Table 1). Other contemporaneous volcanic fields in southwestern North America that experienced voluminous ignimbrite flareups and share similar attributes (Table 2) include: the Southern Rocky Mountain and the Mogollon-Datil fields peripheral to the Colorado Plateau (Fig. 1) and the contiguous, vast Sierra Madre Occidental province in Mexico. The late Cenozoic flareup in the Central Andean Altiplano-Puna volcanic complex is an especially significant example that provides geophysical insights because of its recent activity.
An inventory of these Cenozoic flareup fields discloses that a dynamic interaction of key factors governs their development:
(1) After a period of “flat” subduction, slab rollback to a steeper dip beneath thick continental crust results in volcanism far inland from the trench.
(2) Copious basalt magma is generated by transient dehydration of the foundering slab and flux melting of the overlying mantle. Additional processes might include lithospheric delamination, decompression melting of ascending asthenosphere, and melting of the base of the overlying hydrated lithosphere as hot asthenosphere is contacted (Fig. 23).
(3) Tears in the slab produce variations in the rate of rollback and the amount of mantle magma that invades the crust (Fig. 26).
(4) Voluminous basalt magma intrudes nonextending crust, elevating the geothermal gradient, rendering much of the crust ductile, generating large MASH zones, and creating huge volumes of hybridized silicic magma.
(5) Hybridized silicic magmas ascend from the MASH zones as diapirs, rather than through dikes as in steady-state arc volcanism. Diapirs ascend to the elevated brittle-ductile transition in the shallow crust, where the magma spreads laterally. In this ductile storage regime, magma accumulates over 104 to 106 yr, forming large, broadly discoidal melt-rich zones prior to erupting, rather than erupting in small increments (Fig. 22).
(6) The variable thickness, temperature, and composition of the crust modulate the character of flareups. Where the crust is felsic, thickest, and most prewarmed by orogenic thickening, voluminous dacite (monotonous intermediate) is the dominant erupted product, accompanied by lesser calc-alkaline rhyolite. Voluminous, unusually hot and dry Isom-type trachydacites develop where crustal assimilation into evolving andesitic magmas is minimized by prior extraction of the low-melting-point materials. In places where the crust was thinner and more mafic, alkali-rich rhyolite and related trachydacite dominate the volcanic record.
Study of the southern Great Basin ignimbrite province by Best and Christiansen has benefited from the assistance of numerous individuals over the past decades; they are acknowledged in Best et al. (2013a, 2013b, 2013c). We especially recognize the collaboration of our colleagues Sherman Grommé and Alan Deino for their crucially important paleomagnetic and chronologic data and Chris Henry and Dave John for their work on the Western Nevada field that provided an important comparison for the volcanism in the central and eastern sectors of the province. Financial support was provided by the National Science Foundation through grants EAR-8604195, EAR-8618323, EAR-8904245, EAR-9104612, EAR-9706906, and EAR-0923495 to M.G. Best and E.H. Christiansen. The U.S. Geological Survey and Nevada Bureau of Mines and Geology supported quadrangle mapping. The continuing financial and material assistance of Brigham Young University is gratefully acknowledged. Also, de Silva acknowledges the many students and colleagues (too numerous to mention here but acknowledged in specific publications) who have been crucial to the work in the Central Andes. The late Peter Francis is singled out with deep gratitude for introduction to the area and early mentoring. Financial support from the Royal Society of the UK, National Aeronautics and Space Administration (grant NAGW-1167), and the National Science Foundation (EAR-0710545, EAR-0838536, and EAR-0908324) is gratefully acknowledged. Geologic mapping and other long-term research in the Southern Rocky Mountain volcanic field by Lipman have been supported by several parts of the U.S. Geological Survey, including the Volcanic Hazards, National Geologic Mapping, and Mineral Resources Programs. Especially critical in recent years have been geochronologic studies in collaboration with William McIntosh and Matthew Zimmerer at the New Mexico Bureau of Geology and Mineral Resources. We are grateful for the comments of Susanne McDowell and one anonymous reviewer and Lang Farmer for his editorial assistance.