The 36–18 Ma Indian Peak–Caliente ignimbrite field and calderas, southeastern Great Basin, USA: Multicyclic super-eruptions

The southern Great Basin ignimbrite province (Fig. 1) comprises on the order of 250 cooling units of silicic ash-flow tuff and 43 at least partly exposed calderas formed during the middle Cenozoic ignimbrite flareup in southwestern North America. A major focus of the explosive activity in the Great Basin was the Indian Peak–Caliente caldera complex from which over 50 eruptions (Table 1) broadcast ash flows to at least 150 km distant in southwestern Utah and southeastern Nevada to cover a present area of ∼60,000 km². The aggregate volume of the eruptions is estimated to be 32,600 km³; at least seven had volumes of more than 1000 km³ (Table 2) and, thus, qualify as super-eruptions (Miller and Wark, 2008; de Silva, 2008; volumes are corrected for an assumed uniform 50% east-west crustal extension post-dating volcanism; see below). The eruptions created six partly exposed, mostly overlapping source calderas to the north and three to the south (Fig. 2). Some calderas are nested, or multicyclic. Several additional source calderas have not been accurately located because of concealment beneath younger deposits; some were engulfed in younger calderas. The calderas and their surrounding related outflow ignimbrite sheets constitute the Indian Peak–Caliente ignimbrite field, the subject of this article.

As is true for volcanism generally throughout the Great Basin, caldera-forming eruptions in the Indian Peak–Caliente ignimbrite field migrated southward through time (Best and Christiansen, 1991). (In our usage, the Great Basin [Fig. 1] encompasses western Utah and nearly all of Nevada, rather than just the smaller hydrographic basin.) It is generally agreed that this southward sweep in volcanism throughout the Great Basin was a result of southward steepening in dip, or rollback, of a formerly “flat” subducting oceanic lithosphere that had extended far inland from the continental margin during the early Cenozoic. Volcanic rocks older than ca. 18 Ma in the Great Basin bear an arc chemical signature indicative of their subduction-related origin. Younger volcanic rocks lack this characterizing signature and were formed during a subsequent extensional tectonic regime supplanting the earlier subduction regime.

The calc-alkaline, subduction-related ignimbrites and minor contemporaneous lavas in the Indian Peak–Caliente ignimbrite field range in composition from andesite to rhyolite (Table 1), like those in other middle Cenozoic volcanic fields in southwestern North America. But two compositionally distinct classes of ignimbrite resulted from three super-eruptions of phenocryst-rich, relatively uniform dacite, or monotonous intermediate, magma followed by voluminous eruption of higher-temperature, drier, near-liquidus trachydacitic magma. The burst of three super-eruptions of 2000, 5900, and 4400 km³ occurred at 31.13, 30.06, and 29.20 Ma, respectively, while eruption of possibly as much as 3600 km³ of trachydacitic magma was from 27.90 to 27.25 Ma. Yet another distinct phenocryst-rich latite-andesite ignimbrite was erupted at 22.56 Ma with an apparent volume of 2200 km³. No other part of the southern Great Basin ignimbrite province nor other volcanic fields in southwestern North America involved in the middle Cenozoic ignimbrite flareup, to our knowledge, has such compositionally distinct ignimbrites in such volumes (Table 2).

Prior to the distinct tandem monotonous intermediate and trachydacitic eruptions, the long-lived ignimbrite activity in the Indian Peak–Caliente ignimbrite field began at ca. 36 Ma with eruption of small volumes of rhyolite from two small calderas to the north of the main Indian Peak caldera cluster. Precursory rhyolitic activity shifted southward until ca. 32 Ma with eruption of seven more cooling units whose total volume is ∼700 km³. Source calderas were apparently engulfed within younger large calderas. Several ash-flow cooling units of rhyolite and one of dacite totaling ∼2400 km³ were erupted ca. 30–29 Ma within and peripheral to the very large, 31–29 Ma monotonous intermediate calderas.

Following a lull in activity of 3 m.y. after emplacement of trachydacitic tuffs, major caldera-forming eruptions shifted to the south into the Caliente area (Fig. 2). Three eruptions at ca. 24–23 Ma and several from 18.88 to18.51 Ma were all of rhyolite and had an aggregate volume of 10,000 km³.

Contemporaneous lavas in the 36–18 Ma Indian Peak–Caliente ignimbrite field are of minor volume compared to that of ash-flow tuff. As first revealed in the compilation of Stewart and Carlson (1976) for the approximately similar 34–17 Ma time frame (Fig. 1), the volume of intermediate composition lavas, mainly andesitic, constitutes only a fraction of the volume of silicic ignimbrites. From another perspective, no major composite, or strato-, volcanoes existed prior to or during most of the ignimbrite flareup. In striking contrast, intermediate-composition lavas dominate over silicic ignimbrite in the two contemporaneous volcanic fields to the east on the margins of the Colorado Plateau. In the Marysvale field on the west margin, lavas are ten times more voluminous than ignimbrites (Cunningham et al., 2007). In the Southern Rocky Mountain field on the east margin, lavas are 1.7 times more voluminous (Lipman, 2007). In these two fields, stratovolcanoes dominated the landscape prior to the explosive eruptions of silicic magmas. Truly, the Indian Peak–Caliente ignimbrite field is just that, and resulted from a great flareup in explosive activity exceeding extrusion of andesitic lava flows.

This article consists of two main parts dealing with the basic stratigraphy, composition, chronology, and correlation of the ignimbrite units and their dimensions and source calderas in the Indian Peak–Caliente ignimbrite field; a brief description of contemporaneous lava flows follows.

The first main part of the article deals with 36.02–24.55 Ma ignimbrites and the Indian Peak caldera cluster displayed in Figure 2. This part of the article is a major update of the summary by Best et al. (1989a) and revises some stratigraphic relations, presents data on new ignimbrite units not recognized earlier, and refines the distribution and thickness of ignimbrites as well as the margins of calderas. We describe in detail marginal segments of two of the largest calderas and propose interpretive models for their complex origin during subsidence. Many local interpretations and conclusions drawn two decades ago in our published geologic maps have been changed in the light of newly developed regional concepts for the entire Indian Peak–Caliente caldera complex and its relation to other parts of the southern Great Basin ignimbrite province. The outline of this first part is essentially stratigraphic, beginning with the oldest ignimbrite units. However, because the ignimbrites are of three distinct compositions—rhyolite, dacite, and trachydacite—that are intermingled with respect to age (Table 1), we first describe all of the rhyolite tuff units from oldest to youngest, then, second, all of the dacites, and, third, the two trachydacite formations.

In the second main part of the article, the Caliente caldera cluster to the south and younger 24.03–18.51 Ma ignimbrites are described in stratigraphic order. All of these units are rhyolite, except for the latite-andesite Harmony Hills Tuff emplaced at 22.56 Ma. For this region, we relied heavily on the published research of other geologists, especially Peter Rowley, through their extensive 1:24,000-scale geologic mapping, mostly under the auspices of the U.S. Geological Survey (e.g., Rowley et al., 1995).

Brief comments on petrogenesis of the magmas are included where appropriate. Sample numbers, their stratigraphic unit assignment, and locations are in Supplemental File 1¹. Chemical and modal analyses are in Supplemental Files 2² and 3³, respectively; for analytical methods see Supplemental File 4⁴. Paleomagnetic data for the Indian Peak–Caliente field has been provided by Sherman Gromme and Mark Hudson (2012, personal commun.) and ⁴⁰Ar/³⁹Ar chronology by Alan Deino (2013, personal commun.). All⁴⁰Ar/³⁹Ar ages cited herein are referenced to 28.20 Ma for the Fish Canyon sanidine.

Although previously treated as separate entities (e.g., Best et al., 1989a; Rowley et al., 1995), the Indian Peak and Caliente fields and their associated source calderas are here combined into one unified whole. The 36–29 Ma Indian Peak calderas in the north are separated from the 23–18 Ma Caliente calderas in the south by a gap of ∼25 km (Fig. 2, Table 1). No large-scale geologic maps cover this gap, but the 1:250,000-scale map of Ekren et al. (1977; see also Rowley et al., 1995) shows it is underlain by variably altered Late Oligocene–Miocene lava flows and tuffs and younger, basin-fill sedimentary rocks deposited south of Panaca, Nevada. (Fig. 3 is an index map to geographic place names cited in this article. For clarity, these place names are omitted from following figures.) Source calderas for ca. 24 Ma ignimbrites are likely concealed in this gap, or possibly the southern part of these sources were engulfed in the younger Caliente calderas.

A further reason for combining the Indian Peak and Caliente fields is the substantial overlap of the ignimbrite outflow sheets from the two caldera clusters (refer to later figures in this article and to Best et al., 2013, their fig. 7).

The unity of the 36–18 Ma Indian Peak–Caliente caldera complex and ignimbrite field is consistent with its evolutionary parallelism to the Central Nevada caldera complex and ignimbrite field to the west, where calc-alkaline subduction-related ignimbrite eruptions took place from 36–18 Ma calderas (Fig. 2). Arc volcanism associated with this caldera complex was succeeded by non-arc, alkalic activity from 15.3 to 7.5 Ma to the south in the Southwestern Nevada volcanic field (Sawyer et al., 1994). A gap of ∼20 km separates the southernmost caldera in the former from the northernmost caldera in the latter. In the Caliente area, several relatively small-volume ignimbrites that were deposited ca. 16–12 Ma generally lack the definitive arc chemical signature possessed by the 36–18 Ma ignimbrites derived from the Indian Peak–Caliente caldera complex. The younger ignimbrites have, instead, affinities with the alkalic late Miocene and younger bimodal basalt-rhyolite assemblage, including the Kane Springs Wash ignimbrites and associated caldera source to the south (Rowley et al., 1995; Nealey et al., 1995).

Although there is an overlap of the larger ignimbrite outflow sheets derived from the Indian Peak–Caliente caldera complex and from the Central Nevada caldera complex to the west (Fig. 2; see also Best et al., 2013, their figures 6 and 7; Supplemental File 5⁵), these two caldera complexes are separated by an east-west gap of ∼60 km, which is more than twice the separation of the Indian Peak and Caliente calderas. These contrasting separations, together with the lack of any major, caldera-forming ignimbrite erupted in this 60 km gap, provide further justification for the unity of the Indian Peak–Caliente caldera complex and keeping it separate from the Central Nevada caldera complex.

The Seaman volcanic center (Fig. 3; du Bray, 1993) is the only focus of silicic volcanism between the Central Nevada and Indian Peak–Caliente caldera complexes. This small center, at ∼37°55′ N, 115°9′ W, is a mostly dacitic stratovolcano that was active at ca. 27 Ma, was 10 km in diameter, and had a restored volume of ∼20 km³.

The northern limit of the Indian Peak–Caliente field cannot be set on compositional grounds as for its southern limit because calc-alkaline ignimbrites possessing an arc chemical signature occur far to the north. Rather, the limit is geographic. In easternmost central Nevada, more than 60 km to the north beyond the Indian Peak calderas, there are only scattered remnants of small- to modest-volume silicic ash-flow tuffs (Hose et al., 1976; Gans et al., 1989; Grunder, 1995); none are known to occur in the Indian Peak–Caliente field. Among these tuffs is the phenocryst-rich rhyolite Charcoal Ovens Tuff that has conspicuous titanite phenocrysts (Hose et al., 1976) and a ⁴⁰Ar/³⁹Ar age on sanidine of 35.82 ± 0.11 Ma (sample ELY-1; A. Deino, 2013, personal commun.). An additional 30 km or so to the north of the Charcoal Ovens outcrops is the southern outflow margin of the ca. 35 Ma Kalamazoo Tuff, a phenocryst-poor, normally zoned rhyolite-dacite that has an estimated volume of more than 500 km³ (Gans et al., 1989). We do not consider these distant ignimbrites to be part of the Indian Peak–Caliente ignimbrite field.

In western Utah, 130 km to the north of the Indian Peak calderas, is the east-west–trending Tintic–Deep Creek magmatic belt of granitic intrusions and volcanic rocks (Stewart and Carlson, 1976; Hintze and Kowallis, 2009) that was determined to be active from 39 to 32 Ma based chiefly on ⁴⁰Ar/³⁹Ar age determinations. This belt is not deemed a part of the Indian Peak–Caliente field. Between this magmatic belt and the Indian Peak–Caliente caldera complex are three centers of magmatic activity, including 28–17 Ma granitic intrusions in the Mineral Range 15 km east of Milford and 34–21 Ma granitic intrusions and andesitic extrusions to the west in the southern San Francisco and central Wah Wah Mountains and Shauntie Hills. The third center is the wholly concealed Crystal Peak caldera in the northernmost Wah Wah Mountains (Steven, 1989) ∼50 km northeast of the Indian Peak–Caliente complex. The Tunnel Spring Tuff (Bushman, 1973), which originated from this caldera, has an age of 35.26 ± 0.03 Ma and an estimated volume of only 50 km³. Younger ignimbrites derived from sources to the south in the Indian Peak caldera complex overlie the Tunnel Spring Tuff. Despite having a volume less than an order of magnitude smaller than that of the 35.30 Ma Pancake Summit Tuff in the Central Nevada field, the Tunnel Spring Tuff is a similar phenocryst-rich, high-silica rhyolite that is among the earliest expressions of explosive silicic activity in the Indian Peak–Caliente field. Like the Crystal Peak caldera, the source caldera of the Pancake Summit Tuff also lies appreciably north, by ∼60 km, of the main caldera cluster in the Central Nevada field. Accordingly, we include the Tunnel Spring Tuff and its source caldera within the Indian Peak–Caliente ignimbrite field.

We separate the Indian Peak–Caliente ignimbrite field from the contemporaneous Marysvale volcanic field on the western margin of the Colorado Plateau (Fig. 1; Cunningham et al., 2007) because of an 80 km gap between the calderas of the two fields, even though ignimbrite outflow sheets from them overlap. The Marysvale field has relatively smaller calderas and ignimbrites constitute only one-tenth the volume of intermediate composition lavas and volcanic debris flows in coalescing composite volcanoes, one of which is located in the 80 km gap between the two fields.

Without a knowledge of the volume of ignimbrites, as well as their composition and age, as detailed in this article, it is impossible to fully understand the evolution of the explosive magma system from which they were derived. For this reason, we devote the following paragraphs to ignimbrite dimensions.

The area of a particular unit (Table 2) was calculated within the zero isopach using ArcGIS software (Environmental Systems Research Institute [ESRI]; www.esri.com/software/arcgis) and corrected for an assumed uniform 50% east-west crustal extension post-dating volcanism (Appendix). From published geologic maps, supplemented by field measurements, more than 800 determinations of the thickness (in meters) of individual ignimbrite units were made in the Indian Peak–Caliente field (Supplemental File 6⁶; see also Sweetkind and du Bray, 2008). Most thicknesses (in meters) of outflow sheets were measured at sites where older and younger deposits are exposed to constrain the entire cooling unit; erosion of any of the upper part of the sheet prior to deposition of the younger unit is assumed to be nil. A zero thickness is recorded where the particular unit is absent in the stratigraphic interval where it would be expected to lie. At sites where the top, or bottom, or both, of the unit are not exposed, the measured thickness is prefixed by a greater-than symbol (>). At sites where the unit thickness is somewhat variable the upper limit is prefixed by a less-than symbol (<). In some places, nearby differing measured thicknesses overlap at the map scale on which they are shown, in which case the upper and lower values are indicated (e.g., 180–235); if the range is less than a few tens of meters, a single average value is prefixed by a tilde (∼).

Thicknesses were plotted on a base map for each unit and contoured by eye as isopach lines at appropriate intervals. Because of the progressive smoothing of the landscape during the ignimbrite flareup, contours for most units were fairly regular, but maps of older units reveal local paleohills and paleovalleys.

For the ignimbrites in the Indian Peak–Caliente field, we base volume estimates on the four ideal models shown in Figure 4; results of the calculations are in Table 2.

In Model 1, we ignore any intracaldera volume and from the isopach lines calculate the volume of the pre-collapse tuff using a triangulated irregular network (TIN) in ArcGIS to fit the lines. Model 1 is most useful for ignimbrite units in which (1) the source caldera is partly or wholly concealed, or (2) so little of the caldera margin is known that it could not be drawn with any degree of certainty, or (3) the caldera margin is relatively certain but the thickness of the intracaldera tuff inside its associated source caldera is unknown, or (4) the thickness of the intracaldera tuff is represented by only one very minimal value, e.g., 200 m. This method was also used for some ignimbrites with identified caldera margins for comparison with the estimates by alternate models. For Model 1 ignimbrites, the estimated thickness at the caldera rim, typically 300–400 m, was extrapolated across the entire area of the caldera. In cases for which Model 1 was used, the calculated volume of the pre-collapse tuff sheet is shown in Table 2 and the preferred volume of the total ignimbrite unit is double this value. The doubling factor stems from the concept of Lipman (1984) that the volume of the outflow tuff beyond the source caldera is equivalent to that within the caldera. For the super-eruptions of the Lund and Wah Wah Springs ignimbrite units, this approach appears to work well and yields volume estimates comparable to other methods. However, doubling is known to yield incorrect volumes in some instances. For example, the outflow tuff of Deadman Spring and the intracaldera ignimbrite within its Kixmiller caldera source have volumes of 20 and apparently 180 km³, respectively. Even if these two values might be considerably in error, there is no doubt as to their nonequivalence. In the Central Nevada field, the volume of the outflow ignimbrite sheet of the Windous Butte Formation is ∼1400 km³ and that of the intracaldera tuff at least 3000 km³. In the Western Nevada field (Henry and John, 2013), voluminous ignimbrites within the adjacent 34.0 Ma Caetano and 24.9 Ma Fish Creek Mountains calderas have almost no outflow counterparts, though erosion of the outflow deposits cannot be discounted. Salisbury et al. (2011) noted that in the Altiplano-Puna volcanic complex in the central Andes where only limited relief exists within ignimbrite deposits, outflow:intracaldera ratios are as much as 1:5.

In Model 2, we assume that the caldera floor subsided as a simple piston and that the intracaldera tuff has a uniform thickness—the maximum found for the intracaldera deposit. Isopach lines within the caldera are nested to make a steep caldera wall and the volume calculated using a triangular irregular network (TIN) in ArcGIS fitted to isopachs drawn for outflow and intracaldera tuffs. Eruptions of several hundred to thousands of cubic kilometers from large calderas (>15 km diameter) typically yield intracaldera deposits that are kilometers thick (e.g., Acocella, 2007). Lipman (1997, p. 210) stated that the “best solutions for total subsidence depths at large ash-flow calderas are typically at least 3–4 km.” In the Western Nevada field (Henry and John, 2013), four exceptionally well-exposed calderas—Caetano, Elevenmile Canyon, Job Canyon, and Poco Canyon—all reveal more than 4 km of subsidence. In the Central Nevada field, two calderas—Cathedral Ridge and Williams Ridge—have exposed minimum intracaldera thicknesses on the order of 2000 m.

In Model 3, we use a similar approach to Model 2 but assume a perhaps more realistic, nonuniform thickness for the intracaldera ignimbrite, which is contoured to represent asymmetric subsidence using observed thicknesses.

Model 4 is the method used by Lipman (1997) to estimate the volume of the 5000 km³ Fish Canyon Tuff, among others. Actually, it is the volume of the magma erupted from the subterranean chamber and does not use the volume of the outflow, or pre-collapse, tuff in the calculation. Instead, the volume estimate is derived from the area of the structural caldera inside its ring fault multiplied by the subsidence depth, which is assumed to be uniform across the caldera. The subsidence depth is the thickness of the intracaldera tuff plus the unfilled collar height (Fig. 4), which is equivalent to what we refer to as caldera-filling, or post-collapse, tuff. Subtracted from the volume is a value for the estimated amount of wall-collapse breccia incorporated into the intracaldera deposit.

Several factors impact the accuracy of estimates of ignimbrite volumes, including:

1. Uncertainties in the perimeters of outflow deposits and calderas and thickness of ignimbrite within them.

2. Correction for east-west, mostly post-ignimbrite crustal extension. The 50% value used throughout this article could be in error, but an inventory of north-south versus east-west outflow-sheet dimensions indicate this value is reasonable (see Summary and Conclusions section below). The assumption of uniform extension throughout the whole Indian Peak–Caliente ignimbrite field cannot be justified in detail, but is adopted as a matter of expedience because of the lack of explicit quantitative information on individual strain domains within the field.

3. The doubling of the pre–caldera collapse, or outflow, ignimbrite volume to obtain the total for the unit in Model 1 where little or no data are available on the source caldera. The ratio of intracaldera to outflow volume can be expected to range widely. Examples cited above indicate ratios >1, so that the assumption of equivalent volumes (ratio of 1) for may err conservatively.

Hence, ignimbrite volumes cited in this article are, at best, working rough estimates, subject to modification as more data become available. As order-of-magnitude estimates—that is, thousands versus hundreds of cubic kilometers—we feel they are meaningful. An approximate and subjective uncertainty for each volume estimate is listed in Table 2.

A series of variation diagrams summarizes the composition of 36–18 Ma ash-flow tuffs in the Indian Peak–Caliente ignimbrite field (Figs. 5 and 6). See Table 1 for a list of letter symbols used in Figures 5 and 6 and following figures. Representative chemical analyses of the major ignimbrite units are listed in Table 3.

The total alkalies-silica classification diagram shows that most ignimbrites in the Indian Peak–Caliente field are rhyolite and dacite (Fig. 5). A few samples of the Wah Wah Springs (letter symbol W) and Cottonwood Wash (C) monotonous intermediates are andesite whereas all analyzed samples of the Harmony Hills (H) are andesite and latite, making Harmony Hills the only such mafic ignimbrite of regional extent in the middle Cenozoic southern Great Basin ignimbrite province, so far as we are aware. Most of the Isom-type tuffs (I) are trachydacite but some are low-silica, K₂O-rich rhyolites. The Isom-type designation stems from the occurrence of most of these ignimbrites in the Isom Formation.

Most ignimbrites are calc-alkalic and calcic, magnesian, and high-K but the Isom-type and some of the Condor Canyon Formation tuffs (B and T) are alkalic and alkalic-calcic, ferroan, and shoshonitic (Fig. 5). Nonetheless, these latter ignimbrites, like all of the 36–18 Ma ash-flow tuffs and lavas in the southern Great Basin ignimbrite province, possess an arc chemical signature manifesting their subduction kinship. Although a few samples fall in the anorogenic or within-plate field, they nonetheless show the negative Nb anomaly characteristic of arc rocks (Fig. 5).

On element variation diagrams, dacite (C, K, L, W) and more evolved rhyolite (S, A, E, D, O, U) ignimbrites plot in rather tight, continuous linear clusters and constitute what we call “main-trend” tuffs (Fig. 6). Trachydacitic Isom-type tuffs (I and P) plot well off the main trend because of relatively greater TiO₂, K₂O, and Zr (and Ba, Y, Nb, not shown) and lower CaO (and MgO). The low-silica, high-K₂O, rhyolitic Condor Canyon (B, T), Ryan Spring (M, N), and Lamerdorf (F) ignimbrites also lie to varying degrees off trend, especially with regard to Zr. These off-trend ignimbrites also have slightly lower ⁸⁷Sr/⁸⁶Sr ratios than their main-trend counterparts (Fig. 6D).

Of all diagrams, that for Zr-TiO₂ (Fg. 6C) most clearly distinguishes among the ignimbrite units, including the petrographically and otherwise chemically similar trachydacitic Isom-type cooling units (P, X, Y, Z).

Table 4 gives the Sr isotopic compositions of volcanic rocks from the Indian Peak–Caliente field. These data, together with those from Unruh et al. (1995) for volcanic rocks and shallow intrusions from the Caliente caldera complex, are plotted in Figure 6D. Initial Sr isotopic ratios range from 0.7064 for the 18 Ma andesite lava of Buckhorn Spring of the Caliente caldera complex to 0.7124 for the Sawtooth Peak Formation—the oldest rhyolite for which we have isotopic data.

The high ⁸⁷Sr/⁸⁶Sr ratios show that most of the magmas do not have sources solely in the mantle; significant proportions of old continental crust have been incorporated into most of the magmas. Basement rocks are poorly exposed in the Basin and Range province, but it is presumed to be underlain by the Proterozoic Mojave province (Whitmeyer and Karlstrom, 2007). Wright and Wooden (1991) estimated the Sr isotopic composition of the Proterozoic basement in the Mojave province to range from ∼0.710 to 0.735. Precambrian gneiss and schist from the Mineral Mountains, ∼100 km east of the center of the Indian Peak caldera, had very high ⁸⁷Sr/⁸⁶Sr ratios of 0.852–0.726 in the Oligocene (Coleman and Walker, 1992). Nelson et al. (2002) reported the isotopic composition of Proterozoic igneous and metamorphic rocks (n = 6) from the Santaquin complex (∼200 km northwest of Indian Peak). For these rocks, the Sr isotopic ratios today range from 0.708 to 0.720 with an outlier at 0.752. The average value is 0.720. Crustal xenoliths (n = 23) brought up in the Navajo minettes on the Colorado Plateau have ⁸⁷Sr/⁸⁶Sr ratios ranging from 0.704 to 0.767 with two outliers above 0.800. The average value, minus the outliers, is 0.725. Thus, the volcanic rocks from the Indian Peak–Caliente complex with initial ratios higher than 0.710 could include substantial crustal contaminants ranging from 50% to as much as 100% depending upon the composition of the initial mantle-derived magmas and of the contaminating crust.

However, there is little correlation between the Sr isotopic composition of the volcanic rocks and their silica concentrations. In fact, ratios for mafic Oligocene rocks (<60 wt% SiO₂) span nearly the entire isotopic range (Fig. 6D); Sr isotope ratios range from 0.7064 (the Buckhorn Spring andesite) or 0.7071 (basaltic andesite BRN-3) to 0.7121 (andesite LAM-9-71-2). Rather, initial Sr isotope ratios are highest for oldest rocks regardless of silica content. Thus, lavas and tuffs of the ca. 32 Escalante Desert Formation and the 33.5 Ma rhyolitic ignimbrite of the Sawtooth Peak Formation with ratios of 0.7099 to 0.7124 have incorporated large amounts of ancient continental crust. Igneous rocks associated with the 24–22.56 Ma Caliente caldera complex have distinctly lower isotopic compositions of 0.7066–0.7074 regardless of silica content.

All ignimbrites contain biotite phenocrysts, except for the Isom-type tuffs (Fig. 7; Table 1). Small concentrations of Fe-Ti oxides, essentially magnetite and minor ilmenite, are ubiquitous in the ignimbrites, as are trace amounts of apatite and zircon.

Only about half of the ignimbrites contain sufficient sanidine that can be dated by ⁴⁰Ar/³⁹Ar analyses. Notably, most phenocryst-poor, low-silica rhyolites, as well as trachydacitic Isom-type tuffs, which together compose the off-trend class of ignimbrites, lack sanidine; however, the off-trend sanidine-bearing Bauers is an exception.

Regional rhyolite ignimbrites older than 29 Ma include the phenocryst-rich, main-trend Deadman Spring (D) and Sawtooth Peak (S) which contain all three felsic phenocrysts (plagioclase, sanidine, and quartz) in addition to biotite. Other older rhyolites—Lamerdorf (F), Mackleprang (M), Greens Canyon (N), and Ripgut (U)—as well as the younger Swett (T) and Bauers (B) are phenocryst poor and contain mostly, or only, plagioclase and biotite as silicate phenocrysts; none of this group of off-trend rhyolite tuffs contains quartz phenocrysts. Other younger rhyolites—Leach Canyon (E), Racer Canyon (R), and Hiko (O)—are main trend, are phenocryst rich, and contain all three felsic phenocrysts plus biotite and hornblende.

Main-trend dacite ignimbrites, most of which are the voluminous and distinct monotonous intermediates, are phenocryst rich and contain plagioclase as the major phenocryst together with lesser biotite, hornblende, and minor pyroxene; quartz occurs in most samples whereas sanidine is absent in the Wah Wah Springs (W) and Cottonwood Wash (C) ignimbrites. The Wah Wah Springs, as a unit, has the highest hornblende-to-biotite ratio of any ignimbrite of which we are aware in the middle Cenozoic of the Great Basin. The andesite-latite Harmony Hills (H) is mineralogically like the dacite tuffs.

The off-trend trachydacitic Isom-type tuffs (Isom Formation and Petroglyph Cliff Ignimbrite) are another distinct class of as many as nine cooling units; they contain no sanidine, quartz, hornblende, or biotite and have plagioclase and subordinate clino- and orthopyroxenes and Fe-Ti oxides as their only phenocrysts.

Exposed calderas composing the Indian Peak caldera complex were created from ca. 30 to 29 Ma as the result of explosive eruptions from a long-lived crustal magma system (Table 1). This complex constitutes the northern of the pair of caldera clusters composing the greater Indian Peak–Caliente caldera complex. The roughly elliptical Indian Peak caldera complex lies across the Utah-Nevada state line and spans across five mountain ranges east-west over a present distance of 115 km (Figs. 2 and 8); compensating for 50% post-caldera extension, the east-west dimension is 77 km. The north-south, unextended dimension of the complex is ∼65 km.

Following precursory rhyolitic eruptions of modest volume (Sawtooth Peak, Marsden, Lamerdorf), the Indian Peak caldera probably engulfed their sources. This caldera is in turn largely eclipsed on the southwest by the equally large White Rock caldera. These two calderas are the sources of the super-eruptive 30.06 Ma Wah Wah Springs and the 29.20 Ma Lund monotonous intermediates, respectively. Based on the distribution and thickness of the outflow ignimbrite of the 31.13 Ma Cottonwood Wash monotonous intermediate, it too was the result of a super-eruption (Table 2) and its caldera was engulfed by one or both of the younger calderas; no evidence has been found for any intracaldera facies. Overlapped by the westernmost segment of the White Rock caldera, and extending somewhat beyond, is the older and smaller Kixmiller caldera related to eruption of the 30.00 Ma rhyolite tuff of Deadman Spring. Nested entirely within the older White Rock caldera is the small Mount Wilson caldera that was the source of the 29.0 Ma rhyolite tuff of the Ripgut Formation. Little is known of this small caldera because of incomplete mapping and concealment beneath younger deposits on the south, southwest, and southeast. Offset ∼5 km from the approximately located southwestern margin of these overlapping calderas in the complex is the very small (3 km diameter) and partly buried Blind Mountain caldera source of the 29.1 Ma Petroglyph Cliff Ignimbrite. The source caldera for the 30.01 Ma Mackleprang Tuff Member of the Ryan Spring Formation is entirely buried beneath younger deposits at the southern end of Pine Valley in the eastern segment of the Indian Peak–Caliente caldera complex. The source of the voluminous trachydacitic tuffs of the Isom Formation emplaced mainly from 27.90 to 27.25 Ma cannot be located with certainty. It is shown to lie to the southeast of the complex where it is entirely concealed beneath the broad alluvial valley of the Escalante Desert and flanking younger lava flows (Fig. 8).

The northern margin of the Indian Peak caldera complex and its constituent calderas is clearly defined, and the eastern and western less so. The southern margin is hidden by younger deposits but a distinct gradient in the Bouguer gravity field provides insight into its position (Fig. 8C).

Basin-and-range faulting and tilting of horst blocks subsequent to formation of the calderas have revealed the internal structure of the calderas (Fig. 8A). This is especially true of the northeastern sector of the Indian Peak caldera where the east-tilted Needle Range horst exposes the collar zone of wall-collapse breccias and caldera-filling tuff that lies between the topographic margin and the ring-fault structural margin; to the southeast, intercaldera tuff and breccia that is 4–5 km thick has been intruded by granodiorite porphyry. The northern half of the collar zone lies in a transecting low area in the central Needle Range between what is designated on post–A.D. 1974 maps as the Mountain Home Range to the north and the Indian Peak Range to the south (Fig. 3). On maps published before ca. 1974 these two ranges were designated as the Needle Range. Where we refer to both ranges the older appellation Needle Range will be retained for brevity. Decades of usage distinguishes the geographic place Needle Range from the stratigraphic unit Needles Range (see Historial section below).

Additional features of the Indian Peak caldera complex include: (1) resurgent uplift in the two larger calderas; (2) downsagging of the northern and eastern perimeter of the Indian Peak caldera in addition to deep subsidence of its central part along ring faults, and development of an extensive layer of collar-zone breccia; and (3) ring-fault extrusions in the White Rock caldera.

Because of the general pattern of southward-sweeping volcanism in the Great Basin, the earliest expressions of explosive silicic volcanism in the vicinity of the Indian Peak caldera complex occurred at two small centers to the north, at Crystal Peak and in the Fortification Range.

The oldest known explosive activity took place 15 km north of the northwestern margin of the Indian Peak caldera complex, depositing locally thick ejecta that are exposed to the north of a small rhyolite lava dome (Loucks et al., 1989). These silicic volcaniclastic rocks that were deposited directly on Paleozoic sedimentary rocks are here designated the formation of The Gouge Eye, so named for a topographic scallop in the western escarpment of the Fortification Range (Fig. 9). The upper part of the formation is mostly a sequence of numerous, thin, weakly to moderately welded ignimbrites that contain upwards of 75% clasts of pumiceous, felsitic, and massive vitrophyric rhyolite that are as much as 25 cm in diameter. This part of the formation was likely the result of explosive fragmentation of the growing extrusive dome, a remnant of which is a mass of lava and autoclastic debris ∼250 m thick and 1.5 km in diameter located at the north end of The Gouge Eye. Underlying the effusive dome is a sequence of generally loosely welded lapilli ash-flow tuffs and minor bedded tuffs emplaced by surge, plinian, and epiclastic processes that were the result of explosive venting of the more volatile-rich upper part of the magma chamber.

The lava dome–vent complex lies on the southern margin of a source caldera, inside of which the volcaniclastic deposits aggregate to as much as 1590 m thick. Distal extra-caldera deposits of the formation that are 185–250 m thick were found in well cuttings in Lake Valley. The total volume of the formation calculated by Model 2 in Figure 4 is ∼100 km³.

Phenocrysts make up less than 15% of the ignimbrites and consist of variable proportions of plagioclase, quartz, and biotite; sanidine is found in some samples as well as apparent xenocrysts of pyroxene and hornblende.

Chemically, the lava dome is a rather ordinary rhyolite with ∼74 wt% silica and 8 wt% total alkalies. Two analyses of the tuffs show perturbed concentrations of alkalies but silica ranging from 69 to 72 wt% (Figs. 5, 6, and 10).

Armstrong (1970) cited a K-Ar age of 34.2 ± 0.8 Ma (corrected for new decay constants) on biotite from the basal part of the oldest major ash-flow unit. Our ⁴⁰Ar/³⁹Ar plateau age on plagioclase from a younger tuff is 36.02 ± 0.20 Ma.

Fifty kilometers northeast of the main Indian Peak caldera complex is the small Crystal Peak caldera which was the source of the Tunnel Spring Tuff (Bushman, 1973; Steven, 1989). Identification of the wholly concealed, hypothetical caldera is based on gravity data and on the distribution and thickness of the ignimbrite, which crops out in a 35-km-long paleovalley that includes Crystal Peak where the Tunnel Spring Tuff is 310 m thick (Fig. 9). Exposures at this edifice are famous for the exceptional cavernous weathering (McBride and Picard, 2000). The pre-collapse ignimbrite volume is estimated to be 25 km³ (Table 2). A sanidine ⁴⁰Ar/³⁹Ar age on the tuff is 35.26 ± 0.03 Ma. The weakly welded, phenocryst-rich, high-silica (76.8 wt%) rhyolite tuff contains conspicuous clasts to as much as 2 m in diameter of carbonate rock and abundant small lapilli of pumice. The tuff has about 30% phenocrysts, consisting mostly of doubly terminated quartz, lesser sanidine and plagioclase, and a trace of biotite.

The next eruptive activity took place in the northeast sector of the Indian Peak caldera complex, creating ignimbrite of the Sawtooth Peak Formation (Conrad, 1969; Best and Grant, 1987). A K-Ar age of 33.5 ± 1.2 Ma on biotite and a fission-track age of 33.6 ± 1.8 Ma on zircon have been determined on the same sample (Best and Grant, 1987).

Three samples of the Sawtooth Peak contain 44%–39% phenocrysts (dense rock basis), consisting of plagioclase (49%–41%), quartz (41%–24%), sanidine (12%–7%), biotite (10%–8%), hornblende (5%–2%), and opaque phases (4%–2%). Samples of the Sawtooth Peak plot on the rhyolite-dacite field boundary (Figs. 5, 6, and 10). Relative to other tuffs of comparable silica content, the Sawtooth Peak has somewhat greater CaO, Fe₂O₃, and Sr and has the highest Ba of any rhyolite ignimbrite in the Indian Peak–Caliente field. The high Ba concentrations probably reflect extensive differentiation in the magma before the onset of sanidine crystallization, after which lesser concentrations developed in the residual melt. The Sawtooth Peak tuff also has the highest initial ⁸⁷Sr/⁸⁶Sr ratio (0.7124) of any rock in the Indian Peak–Caliente field (Table 4), indicating this early magma assimilated significant amounts of continental crust.

The only comparable phenocryst-rich rhyolitic ash-flow tuff in the northern part of the Indian Peak–Caliente field is the 30.00 Ma, high-silica rhyolite tuff of Deadman Spring (Table 1). The five other regional rhyolite ignimbrite cooling units in the field that are less than 29 Ma are off trend, phenocryst poor, and lacking in quartz and usually sanidine. Other phenocryst-rich, main-trend rhyolite ignimbrites (Leach Canyon, Racer Canyon, and Hiko) were not emplaced to the south until after 24 Ma.

The Sawtooth Peak ignimbrite occurs chiefly in the Needle Range as a compound cooling unit (Fig. 9). It is ∼300 m thick in a paleovalley at Sawtooth Peak (Fig. 11) just beyond the topographic margin of the Indian Peak caldera where several meters of a near-basal black vitrophyre are exposed. The unit pinches out within ∼9 km to the north (Best et al., 1987b), and then recurs again in a 60-m-thick section ∼5 km farther north. About 12 km southwest of Sawtooth Peak, altered rocks as thick as 200 m were designated as Sawtooth Peak tuff by Best et al. (1987b) but, on reexamination of this terrain, the unit does not appear to be present among andesitic lavas and other rock types. Still farther south within the Indian Peak caldera (Fig. 8A) in a structurally high, apparently resurgent block exposing the caldera floor, the Sawtooth Peak tuff is as much as 200 m thick but locally pinches out on the underlying Paleozoic rocks (Best et al., 1987a). Due west across Hamlin Valley in the southern White Rock Mountains a 7-km-long north-to-south exposure of Cambrian rocks is overlain by Sawtooth Peak tuff as thick as 150 m, but near the southern end of the exposure, younger rocks lie directly on the Cambrian rocks (Keith et al., 1994).

A possible correlative of the Sawtooth Peak ignimbrite is the petrographically similar tuff of Tower Point in the southeastern Wah Wah Mountains (Hintze et al., 1994b). The small isolated exposures which are as thick as 200 m lie in the same stratigraphic position as the Sawtooth Peak tuff, i.e., between the Marsden Tuff and Eocene and Jurassic rocks.

The Sawtooth Peak was deposited on an irregular erosion surface carved into the terrain of older rocks and appears to lie mostly in three more or less east-west–trending paleovalleys and is thin to absent between them (Fig. 9). We roughly estimate its volume at 150 km³. No caldera is exposed, having been engulfed in younger calderas in the Indian Peak Range vicinity. Assuming an equivalent volume in the concealed caldera gives a total of 300 km³ for the unit (Table 2).

In early work on older volcanic rocks east of the Indian Peak caldera complex, Grant (1978) named and described the Escalante Desert Formation as consisting of phenocryst-poor, lithic-rich rhyolite ignimbrites. Later, Best and Grant (1987) formally named two constituent members of the formation—the Marsden Tuff Member and the younger Lamerdorf Tuff Member; the latter unit was originally named and described by Campbell (1978). Best and Grant (1987) also expanded the formation to include local informal rhyolite and andesite lava flow members that interfinger with the ignimbrites and an overlying, mostly epiclastic unit, the Beers Spring Member, which was first named and defined by Conrad (1969).

Subsequent analytical work has disclosed that the Marsden and Lamerdorf, although both phenocryst-poor and lithic-rich rhyolite ignimbrites, have contrasting modal and especially chemical compositions (Table 3; Figs. 5, 6, and 10) and, therefore, were probably derived from different magma sources and evolved along contrasting paths; the Marsden is a main-trend rhyolite whereas the Lamerdorf is an off-trend rhyolite.

Here, we elevate the stratigraphic status of the Marsden, Lamerdorf, and Beers Springs to formation rank and the Escalante Desert to group rank. The Marsden and Lamerdorf Tuffs are described in the following sections while the informal andesite and rhyolite lava-flow formations are described at the end of this article.

The Beers Spring Formation is a discontinuous sequence of well-sorted, poorly bedded, green to brown sandstone that is as thick as 400 m in the central Needle Range and southern Wah Wah Mountains (Best et al., 1987b, 1987d), which is essentially its only area of occurrence. Grains of feldspar and pyroxene are locally visible in hand sample. Locally, the unit also includes: (1) poorly cemented gravel of pebble- to boulder-size clasts of carbonate rock, locally also of quartzite and andesitic rock; (2) loosely welded, crystal-poor tuff, a single sample of which has 12% phenocrysts consisting of 55% plagioclase, 13% quartz, 18% sanidine, 10% biotite, 3% amphibole, and 1% Fe-Ti oxides; and (3) volcanic debris-flow deposits with clasts of greenish-brown porphyritic andesitic rock.

This pale gray to locally pale green, orange, and pink ignimbrite is loosely compacted and contains as much as 50% lapilli of rarely flattened pumice; consequently, the very phenocryst-poor tuff is typically massive. Unlike most other ignimbrites in the Indian Peak–Caliente field, the Marsden contains a varied and commonly abundant xenolithic assemblage of (1) rare, small lapilli of andesitic rock and (2) more common, larger, gray carbonate rock, green phyllite, and white, red, pink, and purple quartzite whose source is likely the thick section of underlying late Proterozoic metasedimentary rocks.

No precise isotopic age has been determined for the Marsden because of the small size (generally <1 mm) and paucity (only a few percent of the whole rock) of the phenocrysts and especially the widespread and pervasive alteration that has typically left quartz as usually the only preserved phenocryst. The quartz grains are not xenocrysts disaggregated from the abundant quartzite clasts because they are unstrained and possess embayments and glass inclusions characteristic of a magmatic heritage. In addition to quartz, other phenocrysts include plagioclase, possible sanidine, and less biotite; the abundance of phenocrysts appears to increase upward in the unit.

Freshest samples of the Marsden Tuff are highly evolved, high-silica rhyolite (Figs. 5, 6, and 10) and have among the lowest TiO₂, CaO, Fe₂O₃, Sr, and Zr of any tuff in the Indian Peak–Caliente field. Moreover, the Marsden has a high ⁸⁷Sr/⁸⁶Sr ratio like the Sawtooth Peak tuff (Table 4; Fig. 6D).

At its type section in the southeastern Wah Wah Mountains (Hintze et al., 1994b), the Marsden is 110 m thick (Fig. 12) and consists of three similar cooling units. Metasedimentary xenoliths locally constitute upwards of one-half the tuff and are as much as 20 cm across. The ignimbrite thickens to 300 m a few kilometers to the northwest, pinches out farther north, but reappears in a section more than 70 m thick ∼20 km beyond that, all in the Wah Wah Mountains.

The Marsden is also exposed in the central Needle Range in a continuous belt ∼8 km long trending northwest from the inner ring fault of the Indian Peak caldera (Fig. 13; Best et al., 1987b). Near the ring fault, north of Indian Peak (Fig. 14), intact exposures of the Marsden are nonexistent because of closely spaced fractures resulting in angular rubble. This terrain may contain unexposed faults and, in the absence of either stratigraphic markers or a conspicuous foliation, an accurate thickness determination cannot be made; the unit here is likely hundreds of meters, possibly 1000 m, thick. To the northwest of the ring fault, increasing amounts of poorly exposed, gray to green, fairly well-sorted sandstone made up of volcanic detritus overlies the Marsden ignimbrite. This sandstone closely resembles the sandstone in the Beers Spring Formation but here it both overlies and underlies the Lamerdorf Tuff. Still farther northwest, in a graben bounded on the south by the Ryan Spring fault and on the north by the outer ring fault of the Indian Peak caldera, cooling units of the Marsden are interlayered with debris-flow deposits, each layer being no more than a few meters thick. These deposits have a sandy matrix between angular to subangular clasts less than 1 m across of Paleozoic rock and minor altered andesitic rock. This tuff–debris flow sequence is 300–400 m thick, continues for ∼2 km southeast of the Ryan Spring fault, and is overlain by very poorly exposed deposits of apparently monolithologic breccias of the same Ordovician units that are exposed to the northwest. North of the graben, the Marsden sequence and overlying breccias of Ordovician rocks are absent and instead of this sequence lying on the Ordovician terrain, there are landslide breccias composed mostly of clasts of Cottonwood Wash Tuff related to the collapse of the Indian Peak caldera.

Best and Grant (1987, p. 16) postulated that this Marsden sequence and its overlying breccias were deposited at the northern margin of a postulated Pine Valley caldera that was the source of the Marsden ignimbrite. The northern structural margin of this caldera was alleged to be the northern-bounding fault of the graben described in the previous paragraph, beyond which no Marsden occurs. Judging from the displacement of Paleozoic strata, this fault appears to have had ∼600 m of normal displacement associated with collapse of the putative Marsden caldera. Farther to the southeast, the proportion of tuff to epiclastic deposits increases and the latter disappear. Within this area (4 km from the inner ring fault of the Indian Peak caldera), a very phenocryst-poor rhyolite lava flow lies within the Marsden ignimbrite. No more can be said of the postulated Pine Valley caldera, because of overprinting and engulfment by the younger Indian Peak caldera and lack of exposures in alluvial valleys to the west. Moreover, apparently near-source Marsden containing large quartzite clasts at the type section in the southeastern Wah Wah Mountains presents further ambiguity regarding a source caldera.

Overlooking possible thick sections of the Marsden engulfed in the younger Indian Peak caldera, its volume is estimated at 260 km³, including ignimbrite assumed to be present in the concealed caldera (Table 2); the true volume is likely more. An alternate view has the ignimbrite confined mostly to two, more or less east-west paleovalleys in about the same position as those that are possible for the Sawtooth Peak ignimbrite. If this is actually the situation, less tuff would have been engulfed in the younger Indian Peak caldera than estimated from the contours in Figure 12 and the total unit volume could be reduced to roughly half.

This lapilli ash-flow tuff is partially to densely welded, locally possesses a near-basal black to dark-brown vitrophyre, and, where devitrified, is distinctively mottled in shades of gray, brown, orange, red, and purple. Contributing to this mottling are lapilli of compacted light-colored pumice and of angular, dark-colored, altered andesitic rock that constitute as much as one-fourth of the tuff. Compared to the Marsden Tuff, phenocrysts in the Lamerdorf are larger (to 3 mm) and more abundant (9%–24% of rock). Plagioclase is by far the dominant phenocryst (73%–83%) followed by biotite (4%–14%), hornblende (3%–10%), Fe-Ti oxides (2%–7%), and clinopyroxene (1%–3%). Unlike the Marsden, quartz is absent and sanidine occurs only in trace amounts in some samples. Anhedral clinopyroxene present in a few samples may be xenocrystic, as may be some of the hornblende. Some stratigraphic sections indicate a subtle upward diminution in the proportion of mafic phenocrysts.

The Lamerdorf is a rather distinct low-silica rhyolite (Fig. 10) that has, relative to other tuffs of comparable silica content, high TiO₂, Ba, and, especially, Zr (Fig. 6C). Chemically, this off-trend rhyolite resembles some of the younger Isom-type tuffs. Nonetheless, its ⁸⁷Sr/⁸⁶Sr ratio is relatively high (0.712) compared to the tuffs of the Isom Formation (0.708; Fig. 6D); this suggests that its parental magma assimilated more old continental crust than the younger Isom magmas.

The Lamerdorf Tuff has a larger areal extent than the older Sawtooth Peak and Marsden ignimbrites (Figs. 9, 12, and 15). It consists of as many as three simple cooling units whose total thickness is as much as 100 m on the west side of the Mountain Home Range. Weighted mean plagioclase ⁴⁰Ar/³⁹Ar ages on the second and third cooling units are 32.09 ± 0.10 Ma and 31.90 ± 0.16 Ma, respectively. An unusually thick section of ∼900 m occurs as a paleovalley deposit in the central Wah Wah Mountains. This paleovalley is ∼6 km wide and, in addition to the Lamerdorf, includes an intercalated andesitic lava flow ∼100 m thick, less than 70 m of Marsden ignimbrite, and underlying local lenses of conglomerate as thick as 80 m of Paleozoic rock clasts. The Lamerdorf pinches out on the north side of the paleovalley but persists as a 70–170-m-thick sheet to the south and a 30–100-m-thick sheet ∼50 km to the northwest. The Lamerdorf Tuff thus appears to have accumulated on a somewhat flattened terrain beyond the central Wah Wah Mountains paleovalley. There may be a possible extension of this paleovalley as far as 80 km to the east-southeast in what is now the High Plateaus where a possible correlative phenocryst-poor, densely welded tuff that underlies the Wah Wah Springs Formation has a biotite K-Ar age of 31.9 ± 0.5 Ma (unit Tvl in Anderson et al., 1990).

The volume of the Lamerdorf Tuff is estimated at 180 km³, inclusive of that assumed to occur in a concealed source caldera (Table 2).

Following emplacement of the Lamerdorf Tuff, two super-eruptions produced the Cottonwood Wash and Wah Wah Springs monotonous intermediates. They were followed by eruption of modest volumes of the Ryan Spring Formation (Tables 1 and 2). It consists of two tuff members of phenocryst-poor, low-silica rhyolite—the older Greens Canyon and the younger Mackleprang—as well as local andesite and rhyolite lava flows (see end of this article) and sedimentary deposits (Best and Grant, 1987; Best et al., 1989a). These two members are well exposed in a stratigraphic section on the west side of the central Needle Range at Ryan Spring in the collar zone of the Indian Peak caldera. They have identical modes (Fig. 16) and similar chemical compositions (Figs. 5, 6, and 10) but a different appearance in hand sample.

No compositional zoning is evident in our limited sampling of either of the two units which are almost entirely confined within the older Indian Peak caldera where they constitute post-collapse caldera-filling tuff (Fig. 17). Only three occurrences of the Mackleprang and none of the Greens Canyon have been found beyond the Indian Peak caldera.

Weighted mean ⁴⁰Ar/³⁹Ar ages of plagioclases for the Greens Canyon of 30.13 ± 0.13 Ma and the overlying Mackleprang of 30.01 ± 0.09 Ma indicate their deposition very soon after the creation of the Indian Peak caldera and eruption of the dacitic Wah Wah Springs ignimbrite at 30.06 ± 0.05 Ma (Table 1; taking into account analytical uncertainties, these ages are consistent with the stratigraphic relations of the units). The initial ⁸⁷Sr/⁸⁶Sr ratios of the Mackelprang and Wah Wah Springs are indistinguishable (Table 4), providing another link between their magmas.

We considered the possibility that the dated plagioclases in the Greens Canyon, which has only small, sparse grains of this phase, might be xenocrysts derived by disaggregation of the common xenoliths of Wah Wah Springs dacite tuff. However, the Greens Canyon has no hornblende, an abundant phenocrystic constituent in the Wah Wah Springs. Microprobe analyses of Greens Canyon plagioclases reveals that some small cores are as calcic as An_70–80 whereas most of the grains are An_36–64, thus overlapping the An content of plagioclases in the Wah Wah Springs. However, they have distinctly lower concentrations of the K-feldspar component in accordance with crystallization from the cooler Greens Canyon rhyolitic magma.

The Greens Canyon superficially resembles the older Marsden Tuff except that the lithic lapilli and small blocks are commonly of red Wah Wah Springs tuff rather than the sedimentary assemblage that is typical of the Marsden. Most outcrops of the Greens Canyon have holes a centimeter or so in diameter surrounded by haloes that are lighter in color than the surrounding pale orange to pinkish-brown matrix; these holes apparently formed by weathering out of reacted xenoliths of unknown character. Small lapilli of andesitic rock are visible in thin sections. Also unlike the Marsden, the Greens Canyon lacks quartz phenocrysts but does similarly contain plagioclase, biotite, and Fe-Ti oxides that are generally less than 1.5 mm. Another contrast with the main-trend Marsden is the fact that the Greens Canyon is an off-trend rhyolite and is not nearly as chemically evolved (Figs. 5, 6, and 10) with, for example, higher concentrations of CaO, TiO₂, and Zr.

In the west-central Needle Range (Best et al., 1987b) near Ryan Spring, the Greens Canyon is a moderately welded and entirely devitrified simple cooling unit ∼500 m thick. To the west, in the northeastern White Rock Mountains (Best et al., 1989d), as many as five cooling units, some with prominent, dark-brown, near-basal vitrophyres, lie in what appears to be a partially fault-bounded basin between the topographic wall and resurgently uplifted block of the Indian Peak caldera; if not repeated by unrecognized faulting, this compound cooling unit sequence is ∼650 m thick. Basal parts of individual cooling units contain compacted glassy orange or black pumice lapilli; several meters of thinly bedded sandstone are locally intercalated between them whereas a greater thickness of sandstone overlies the member to the southwest. The Greens Canyon is only 20 m thick to the northeast. A 1300 m section of the Ryan Spring ignimbrite farther northeast south of Atlanta was mapped as entirely Greens Canyon by Willis et al. (1987); however, subsequent examination reveals at least the top portion of this thick section is Mackleprang. The variations in thickness of the Ryan Spring ignimbrites possibly reflect relief on fault blocks on the floor of the resurgent Indian Peak caldera.

The source of the Greens Canyon lies within the older Indian Peak caldera, but we are unaware of direct evidence defining its location. Because the eruption followed closely after the collapse of the caldera, further subsidence might have occurred, rather than the creation of an independent source caldera.

For the volume of the Greens Canyon Tuff Member, we multiply its area of exposure by the average thickness of 300 m, obtaining 600 km³ (Table 2).

The mottled Mackleprang is similar to the older Lamerdorf Tuff but can be distinguished by stratigraphic position and by the lack of hornblende phenocrysts. Chemically, the Mackleprang differs from the Lamerdorf in that it has much lower TiO₂, Zr, and Nb (Fig. 6). From the older Greens Canyon Tuff the Mackleprang is distinguished by slightly larger (<2 mm) and relatively euhedral plagioclase phenocrysts, commonly by more abundant flattened pumice lapilli that impart a more pronounced foliation, and larger, more abundant xenoliths of varicolored andesitic rock. A small amount of quartz phenocrysts occurs in the ignimbrite near Atlanta. In thicker stratigraphic sections the Mackleprang appears to be a compound cooling unit whereas thinner sections are a simple cooling unit.

A thick (500 m) section of the Mackleprang ignimbrite occurs south of the resurgent uplift of the Indian Peak caldera in the Indian Peak Range (Fig. 17), where it is capped by several meters of sandstone. Other post–caldera-filling sections to the north are thinner, suggesting the source of the Mackleprang lies in the southeastern sector of the caldera. Because the thickness of the younger Lund ignimbrite is as much as 1400 m in hills in southernmost Pine Valley, this area is believed to harbor the Mackleprang source caldera in which the Lund ponded. The caldera shown in Figure 17 (not to be confused with a postulated Pine Valley caldera source for the Marsden Tuff above) is delineated on the west by a rhyolite lava that appears to be of about Mackleprang age (Best et al., 1987a). The Mackleprang caldera is assumed to be filled with an equivalent volume of ignimbrite as the estimated outflow, giving a total volume for the unit of 480 km³ (Table 2).

The phenocryst-rich rhyolite tuff of Deadman Spring (Taylor, 1990) was deposited at 30.00 ± 0.10 Ma and consists of a thick intracaldera deposit in the Fairview Range and thinner outflow to the southwest (Fig. 18). Stratigraphic relations are somewhat uncertain but the Deadman Spring appears to underlie the Mackleprang Tuff Member of the Ryan Spring Formation but it certainly overlies Wah Wah Springs ignimbrite.

The Deadman Spring outflow is a moderately to loosely welded, simple cooling unit. One sample has 27% phenocrysts that include 47% plagioclase, 30% quartz, 17% sanidine, 5% biotite, and 1% opaque grains. The compound cooling unit that comprises the 2000-m-thick intracaldera deposit is densely welded throughout and is only very locally vitrophyric. In some places it has alternating layers a few centimeters thick made of finer (<1 mm) and coarser (2–3 mm) felsic phenocrysts. A compaction foliation is seldom evident because of very rare lapilli of pumice and only sparse biotite and an abundance of relatively equant felsic phenocrysts; the tuff has a lava-like appearance in outcrop. One sample has 41% phenocrysts that include 35% plagioclase, 32% quartz, 23% sanidine, 9% biotite, and 1% opaque grains. Chemically, the Deadman Spring is a relatively silica-rich rhyolite (73.8%–76.6%; Figs. 5, 6, and 10).

The source of the tuff of Deadman Spring is the Kixmiller caldera that apparently encompasses all of the Fairview Range (Figs. 8 and 18; Best et al., 1998). This caldera formed several kilometers west of the slightly older Indian Peak caldera but was later partially engulfed in the younger White Rock caldera when the Lund magma erupted at 29.20 Ma. The Fairview Range comprises horst-and-graben blocks of Deadman Spring and younger rocks as well as Paleozoic rocks in the south. The sharply defined northern margin of the Kixmiller caldera is a west-southwest–striking, steeply dipping ring fault through Kixmiller Summit between the Fairvew Range and the Grassy Mountain mass of Paleozoic rocks to the north (Fig. 19). A single exposed slab of Wah Wah Springs outflow tuff, ∼300 m long and 20 m thick, embedded in the Deadman Spring tuff and lying just inboard of the southwestern segment of the ring fault is the only manifestation of a wall-collapse landslide deposit. The intracaldera tuff appears to be at least 2000 m thick just south of the ring fault but decreases southward to less than 380 m at the southwest end of the Fairview Range. Just to the south, the tuff thins markedly to only a few tens of meters across what we have interpreted to be a caldera margin. It could be an ill-defined topographic wall or a hinge of a trapdoor-like depression. A granitic porphyry with an incremental-heating ⁴⁰Ar/³⁹Ar age of 30.04 ± 0.07 Ma on biotite is believed to be a small, near–ring-fault intrusion marking this southern caldera margin (Best et al., 1998). Although the age is identical to that of the tuff of Deadman Spring, it should be noted that some such biotite ages are anomalously old compared to sanidine ages in the same rock. The bulk chemical composition of the porphyry has similarities to that of both the Deadman Spring and the Lund, whose caldera source margin also lies close to the porphyry intrusion. Small-scale silver mineralization and widespread silicification of carbonate rocks is spatially associated with the granitic porphyry (Best et al., 1998).

As an alternate interpretation for the southern Kixmiller caldera margin, Ekren and Page (1995) suggested the abrupt change in thickness of the tuff marks the margin of an east-trending “volcanic trough” that was subsiding during deposition; they relate it to the east-striking Blue Ribbon lineament of Rowley et al. (1978) that extends into central Utah and connects with the Warm Springs lineament in central Nevada. The easterly striking faults that are associated with near-vertical dips and easterly striking compaction foliation in tuffs in the southern Fairview Range are an unusual aspect of this part of the Indian Peak caldera complex (see discussion in Best et al., 1998).

We estimate the volume of the Deadman Spring ignimbrite from Model 3 in Figure 4 to be ∼200 km³, ∼90% of which is intracaldera tuff that lies within the asymmetric, or trapdoor, caldera.

Ignimbrite of the Ripgut Formation overlies the Lund monotonous intermediate. The name is taken from Ripgut Springs on the east side of the White Rock Mountains where a surrounding fence of sharpened poles embedded in the ground protects the water source for range stock; any large animal attempting to compromise the integrity and purity of the springs would suffer the intended consequence! Most of the Ripgut Formation consists of pumice-rich rhyolite ignimbrite but the unit also includes one small dike of similar composition, intracaldera breccias made of clasts of Lund and Wah Wah Springs tuffs, and tuffaceous and conglomeratic sandstone as much as 130 m thick (Willis et al., 1987). Somewhat rounded clasts in the conglomerate are of Wah Wah Springs and Greens Canyon tuffs.

Duplicate sanidine and plagioclase ages have weighted means of 28.96 ± 0.05 Ma and 28.99 ± 0.10 Ma, respectively (Table 1).

Above a meter or so of uncompacted and loosely welded lapilli tuff, the Ripgut ignimbrite grades upward from a densely welded and compacted black to dark-brown massive glass to devitrified, densely welded brown lapilli tuff to a loosely welded, uncompacted light-brown lapilli tuff at the top. Pumice lapilli typically constitute about one-third of the tuff and in devitrified parts are varicolored—black, gray, brown, orange, and white—within a single outcrop. Xenoliths of Lund and Wah Wah Springs tuffs as much as 10 cm in diameter are locally evident, as are light-colored haloes that surrounded weathered out, reacted(?) xenoliths of unknown character. In some outcrops, the Ripgut superficially resembles the Greens Canyon Tuff Member of the Ryan Spring Formation but the Ripgut mostly contains fewer phenocrysts (Figs. 16 and 20). In compacted vitrophyre, only ∼2% of the tuff consists of small (∼1 mm) phenocrysts that are mostly plagioclase and lesser sanidine, biotite, and Fe-Ti oxide. Only a few of the plagioclase phenocrysts in a sample are euhedral; the remainder are anhedral and may be xenocrystic, derived by disaggregation of clasts of dacite tuff.

Nearly all of the Ripgut ignimbrite is found within the White Rock caldera (Figs. 8 and 21). The unit thins and locally pinches out over the resurgent uplift in the caldera and along its northern topographic margin. Thicker sections that consist of two or three simple cooling units are locally separated by several tens of meters of loosely consolidated sediment. The only exposure outside the White Rock caldera is a simple cooling unit less than 120 m thick ∼2.5 km southwest of Modena.

Ekren et al. (1977) noted the presence of a thick monolithologic breccia of dacitic “Needles Range tuff” north of Mount Wilson in the Wilson Creek Range that they believed to be an intracaldera deposit. This, together with other features considered to indicate a caldera, led them to refer to the area as the “Mount Wilson volcanic center.” Subsequently, we found an area of ∼3 km² of breccias of the Lund tuff lying atop intact Lund on the northwestern slopes of Mount Wilson; Lund clasts are as much as 1 m across and lie in a matrix of Ripgut tuff or comminuted Lund (Willis et al., 1987). Nearby, clasts of Wah Wah Springs and Paleozoic quartzite and carbonate rock occur within the Ripgut. These breccias confirm the presence of the Mount Wilson caldera source for the Ripgut ignimbrite. East of the large area of breccia is an elongate east-west mass of subvertically flow-layered rock of Ripgut character that we interpret to be a dike possibly emplaced along the caldera ring fault. Reconnaissance mapping reveals a thickness of ∼2000 m of Ripgut tuff underlying Mount Wilson and overlying thick intracaldera Lund in the lower foothills of the northwestern part of the range (Fig. 22).

The exposed volume of 800 km³ of the tuff calculated by Model 2 in Figure 4 is a minimum for the unit because the thick intracaldera deposit is truncated 4 km south of Mount Wilson by a major east-west–striking fault system that has downdropped lower Miocene rhyolite lavas and tuffs on the south, concealing the southern sector of the caldera (Figs. 8A and 21). Likewise to the east, but here overlying Miocene rhyolites obscure the caldera.

Most of the exposed Ripgut ignimbrite is a very phenocryst-poor (<2%), highly evolved rhyolite with silica ranging upwards to 77.7 wt%; it has the highest concentration of Nb (35–47 ppm) of any ignimbrite in the Indian Peak–Caliente field, as well as among the lowest Zr (mostly ∼100 ppm) (Figs. 5, 6, and 10; Table 3). However, the uppermost exposed tens of meters of the 2000-m-thick intracaldera deposit on the southeast slope of the southeast peak of Mount Wilson (Fig. 22) is chemically less evolved (72.5 wt% SiO₂); phenocrysts are larger (to 2 mm) and more abundant (12%) and include hornblende and a little clinopyroxene. Analyses of the hornblende grains indicate they are not xenocrysts disaggregated from the common xenoliths of Lund tuff. This less-evolved tuff hosts dark-brown fiamme several centimeters long of compacted pumice that contains 8% phenocrysts as large as 3 mm across of euhedral plagioclase plus minor biotite and lesser quartz, sanidine, hornblende, Fe-Ti oxides, and a trace of aggregated pyroxene (Fig. 20). Chemically, the fiamme are trachydacite (Fig. 10), contain 68.3 wt% SiO₂, 0.5 wt% TiO₂, and 335 ppm Zr, and resemble the Lamerdorf Tuff and Hole-in-the-Wall Tuff Member of the Isom Formation.

Because the Ripgut ignimbrite was erupted soon after the Lund, and from within the source caldera of the Lund, most of the erupted, highly evolved Ripgut magma might have been a late differentiate of the residual unerupted Lund magma. Niobium concentrations, however, are high, which leads us to conclude this cannot be true. The lower part of the erupted Ripgut magma chamber was apparently invaded by and mixed with magma similar to that which formed the Isom and Lamerdorf ignimbrites, possibly triggering the eruption.

Three of the largest cooling units in the Indian Peak–Caliente ignimbrite field (Table 2) are of relatively uniform, phenocryst-rich dacite; they include the 31.13 Ma Cottonwood Wash, the 30.06 Ma Wah Wah Springs, and the 29.20 Ma Lund. With volumes of 2000, 5900, and 4400 km³, respectively, they easily fall in the super-eruptive category of Miller and Wark (2008) and de Silva (2008). As will be detailed below, their tightly overlapping source calderas indicate a recurrently eruptive, or multicyclic, magma system beneath the Indian Peak caldera complex.

Because of their relatively uniform composition and intermediate silica content, Hildreth (1981) designated such voluminous phenocryst-rich dacite ignimbrites “monotonous intermediates,” citing, as examples, among others, the Monotony Tuff in the Central Nevada field and the Needles Range tuffs, by which the three in the Indian Peak–Caliente field were formerly designated (see historical note below). As a distinct end-member in the broad spectrum of ash-flow deposits now known in the geologic record, monotonous intermediates generally lack the pronounced systematic modal and chemical zoning seen in many other and typically smaller ignimbrite deposits. Because the actual extent of compositional zoning in monotonous intermediates and compositional gradients in the pre-eruption magma chambers were not fully understood when Hildreth (1981) recognized this distinct class of ignimbrite, we will devote some attention to these aspects below.

All together, the three super-eruptive monotonous intermediates constitute a unique attribute of the Indian Peak–Caliente field not found, to our knowledge, in other volcanic fields in southwestern North America where the middle Cenozoic ignimbrite flareup is expressed. In the Great Basin, the only other monotonous intermediate is the Monotony Tuff. Beyond the Great Basin, the only other is the Fish Canyon Tuff in the Southern Rocky Mountain field (Lipman, 2007).

Before crustal extension, individual monotonous intermediates in the Indian Peak field cropped out over as much as 32,000 km² (Table 2) as simple outflow cooling units tens to hundreds of meters thick. Normal zoning in welding, compaction, devitrification, and vapor-phase crystallization (Ross and Smith, 1961) is typical; they are relatively low-grade non-rheomorphic ignimbrites in the grade continuum of Branney and Kokelaar (1992). Outflow sheets have a near-basal black vitrophyre as much as 10 m thick (Fig. 23). Xenoliths of other rock types are notably absent in these monotonous intermediates, except for the intracaldera tuff of the Wah Wah Springs Formation. Poorly bedded surge deposits a couple of meters thick are very locally exposed beneath the Cottonwood Wash and Wah Wah Springs ash-flow tuffs as much as 90 km north of the caldera complex (Fig. 24). Precursory plinian deposits seem to be lacking, but we acknowledge that typically poorly or nonwelded pyroclastic material beneath the resistant overlying densely welded ignimbrite and above the underlying welded tuff of an older unit is very seldom exposed.

Densely welded, pore-free monotonous intermediates contain as much as ∼50% phenocrysts that consist of mostly plagioclase with lesser biotite, hornblende, quartz, and magnetite, much smaller concentrations of pyroxene and ilmenite, and trace amounts of zircon, apatite, and sulfides. We have verified sanidine only in the Lund ignimbrite, although some workers have reported some in the Cottonwood Wash and Wah Wah Springs. The Lund also has trace amounts of titanite. Phenocrysts lie in a vitroclastic matrix of high-silica rhyolite glass (Table 5; Christiansen, 2005). Mineral thermobarometers indicate pre-eruption magmas had temperatures of ∼740–830 °C at pressures of 2.0–2.5 kbar, corresponding to a depth of 7–9 km. Oxygen fugacities were ∼2 log units above the QFM buffer and magmas were not water saturated (Maughan et al., 2002; Woolf, 2008). Monotonous intermediates are dominantly high-K dacite, magnesian, and calc-alkalic or alkalic (Fig. 6).

A turning point in the earliest pioneering work on ignimbrites in southwestern Utah was the study by Mackin (1960; see also Anderson et al., 1975, p. 13–16, for a comprehensive history going back to the late 1800s). He recognized a sequence of several cooling units in the Iron Springs mining district west of Cedar City and designated the lowermost two composed of phenocryst-rich dacite as members of the Needles Range Formation. This formation was subsequently redefined by Best et al. (1973) and later by Best and Grant (1987) who elevated its status and made the older Cottonwood Wash Tuff and the younger Wah Wah Springs Formation constituent formations in the Needles Range Group; they also included a third, still younger phenocryst-rich dacite tuff designated the Lund Formation in the group. Apparently the first to recognize these three distinct dacite cooling units in the eastern Great Basin was Conrad (1969), a student of Mackin, who mapped them in the Needle Range. He referred to the middle unit as the Wah Wah Springs but did not name the oldest and youngest units, which were named later by Best et al. (1973). Geologic maps and stratigraphic literature dealing with the volcanic geology of southwestern Utah over the past decades commonly refer to these phenocryst-rich dacite ignimbrites as “Needles Range tuffs,” following the original designation of Mackin (1960). We now refer to these three distinct ignimbrites as monotonous intermediates.

This oldest monotonous intermediate deposited at 31.13 ± 0.13 Ma is known only as an outflow sheet; no caldera and associated intracaldera deposit have been found. In addition to its stratigraphic position, this unit is distinguished from other monotonous intermediates with which it is coextensive over broad areas of the Indian Peak field (Figs. 2, 23, and 25) by the presence of uncommonly large books of biotite as much as 8 mm in diameter and similarly large, embayed quartz phenocrysts. Some quartz and plagioclase grains in tuffs are smaller broken phenocrysts, or phenoclasts, blown apart during eruption (Best and Christiansen, 1997), which, at least in part, may account for the generally larger size (<9 mm) of these phases in pumice compared to tuff (<5 mm). Plagioclases in tuff commonly have inclusions of melt as well as of all other phases present in the tuff; plagioclases with diverse mineral inclusions might be restite grains. Although about equally abundant as biotite (Figs. 7 and 26), euhedral hornblendes are rarely visible in hand sample because of their small size (<1 mm). Lesser amounts of magnetite, ilmenite, and clinopyroxene—some mantled by hornblende in an apparent reaction relation—together with trace amounts of zircon, apatite, sulfides, and very rare orthopyroxene are found in the Cottonwood Wash (Ross et al., 2002).

The terrain on which the Cottonwood Wash outflow ignimbrite was deposited had some local relief that was not entirely smoothed by earlier ignimbrites (Fig. 25). For example, in the southern Fairview Range in Nevada the Cottonwood Wash is as thick as 280 m but pinches out a few kilometers to the northwest over a pile of lava (Best et al., 1998). Southward, the Cottonwood Wash thins and pinches out over a northeasterly trending highland of Paleozoic rocks and then reappears southward. A similar, but more subdued, pinchout is evident in the Wah Wah Mountains in Utah at ∼38°23′ N. To the south, the unit is as much as 170 m thick but pinches out a few kilometers to the southwest over a paleohill of Mesozoic and Paleozoic rocks. Mindful of these irregularities in the depositional surface, we estimate the volume of the outflow Cottonwood Wash to be 1000 km³ (Table 2).

No underlying plinian deposits have been found but at four sites surge beds no more than 2 m thick are exposed (Fig. 24; see also Ross et al., 2002).

Distal fallout deposits add hundreds of cubic kilometers to the erupted Cottonwood Wash ejecta. As much as 9 m of thick-bedded pumiceous tuff in southwestern Nevada ∼200 km southwest of the caldera has a modal composition (including the presence of biotite books to as much as 5 mm in diameter) and age consistent with that of the Cottonwood Wash (Barnes et al., 1982). Fine-grained, co-ignimbrite deposits occur in northeastern Utah and as far as western Nebraska. The ash of Diamond Mountain Plateau along the south flank of the Uinta Mountains in northeastern Utah has a similar mineral assemblage (including large and abundant biotite), zircon morphology, and age of 30.90 ± 0.22 Ma as the Cottonwood Wash (Kowallis et al., 2005). Our age of the upper ash of the Whitney Member of the Brule Formation (White River Group; Tedford et al., 1996; LaGarry, 1998; Larson and Evanoff, 1998) in western Nebraska of 31.29 ± 0.17 Ma, as well as the compositions of biotites and hornblendes, match those of the Cottonwood Wash (Blaylock, 1998). Thicknesses of the Utah and Nebraska fallout deposits are as much as 0.5 m and 2 m, respectively. If a narrow elliptical area with axes of 1300 and 250 km is drawn to encompass the three fallout sites, the calculated area, minus that of the Cottonwood Wash outflow ignimbrite of 12,000 km² (Table 2), multiplied by an average tuff thickness of 1 m yields a very conservative volume of ∼240 km³. A similarly narrow elliptical area of fallout occurred for the 18 May 1980 eruption of Mount St. Helens (Washington State; Houghton et al., 2000). However, more equant and larger fallout distributions have been documented for the 760 ka Bishop Tuff erupted from the Long Valley, California, caldera (Izett et al., 1970) and the 640 ka eruption of the Lava Creek B Tuff from the Yellowstone, Wyoming, caldera (Izett and Wilcox, 1982). The latter ash deposit covers an area of 4,000,000 km².

We previously postulated (e.g., Best et al., 1989a) a concealed source caldera for the Cottonwood Wash in the area between the Mountain Home and Fortification Ranges and south of the Snake Range. However, low hills of unaltered limestone lie here, leaving insufficient space for a caldera that, judging from the outflow volume, would have a diameter of ∼25 km or more. This inconsistency, together with our updated distribution and thickness data for the outflow deposit, lead us to now believe that the caldera was engulfed in younger calderas. The probable area in which the caldera lies in the Wilson Creek Range and White Rock Mountains, shown in Figure 25, coincides with the lowest Bouguer gravity anomaly in the Indian Peak caldera complex (Fig. 8C). Assuming the volume of hidden tuff inside the buried caldera is at least equivalent to that of the pre-collapse tuff (Model 1 in Fig. 4), a total volume of 2000 km³ is estimated for the Cottonwood Wash (Table 2). But this might be a minimum value for the total eruptive volume because of the significant amount of distal fallout ash.

On many chemical variation diagrams (Figs. 5 and 6), element concentrations for the Cottonwood Wash Tuff overlap those for the Lund and especially the Wah Wah Springs ignimbrites. For the Cottonwood Wash, SiO₂ ranges from 61.0 to 68.3 wt%, TiO₂ from 0.5 to 0.8 wt%, and Fe₂O₃ from 4.2 to 6.8 wt% (Fig. 27). Of all the monotonous intermediates, the Cottonwood Wash has the highest ⁸⁷Sr/⁸⁶Sr ratio (0.711; Fig. 6D) and must have incorporated significant proportions of Proterozoic crust. Hart (1997) also showed that the Cottonwood Wash has the highest δ¹⁸O of the monotonous intermediates, with δ¹⁸O quartz of 9.8–10.4 ‰.

Two samples of ignimbrite plotting in the andesite field were collected a meter or so from the base of sections north of the source; samples higher in these sections contain more silica. Otherwise, no systematic lateral or vertical compositional zoning is evident in the ignimbrite deposit.

Cognate pumice inclusions to as much as 25 cm in longest dimension were collected from the Cottonwood Wash outflow sheet at sites north of the probable source of the tuff (Fig. 25). Chemical analyses were made of inclusions from the base of the outflow sheet at proximal sites ATL-1-70-1 and GOUGEWL-1-5 and at the distal BRN-2 and WARM-2 sites, and of inclusions from near the top of the sheet at proximal site GOUGEWL-3 and distal site KNOLL-1-38-1 (see Supplemental File 1 [see ^{footnote 1}] for locations). Pumices can be divided into two groups (Fig. 27): five less evolved with <66.3 wt% SiO₂ and >0.63 wt% TiO₂ that occur at the base of the outflow sheet at the ATL-1-70-1, WARM-2, and GOUGEWL-1-5 sites, and the remainder more evolved with >67.5 wt% SiO₂ and <0.58 wt% TiO₂. The more-evolved pumice clasts occur at all sites, both at the base and top of the stratigraphic sections. The more-evolved pumices contain rare sanidine, plagioclase phenocrysts with more sodic rims, and slightly more Fe-rich biotites and hornblendes than samples of tuff whereas the less-evolved pumices plot with less-evolved tuffs.

The spatial variations in composition of ignimbrite and cognate pumice clasts have no consistent explanation for withdrawal from a simply zoned, pre-eruption magma chamber. In whatever manner the evacuation of the chamber progressed, there appears to have been compositionally contrasting parts.

Several of the cognate pumices have more-evolved bulk compositions than the overall population of tuff samples (Fig. 27), which is consistent with fractionation or elutriation of high-silica rhyolite vitroclasts from the ash flow during eruption and emplacement. Nonetheless, some pumices are as little evolved as some tuffs, indicating that not all of the chemical variation in tuffs is the result of fractionation. Moreover, the more-evolved pumice clasts also have more-evolved mineral compositions than ignimbrites. Pumices have similar total phenocryst concentrations as tuffs (Fig. 26A; adjustment to a dense rock equivalent does not change this similarity); this also supports the lack of major fractionation of glass shards in the Cottonwood Wash ash flows.

This unit was named by Mackin (1960) for its occurrence near the large complex of Wah Wah Springs on the eastern flank of the Wah Wah Mountains (Figs. 28 and 29; see also Best et al., 1973, Figures 4 and 5). The unit includes an informal non-lithic outflow tuff member and, at its Indian Peak caldera source, an intracaldera member, which includes lithic dacite tuff, wall-collapse breccia, and intrusive granodiorite porphyry (Skidmore et al., 2012). Because we now realize that some of the non-lithic outflow ignimbrite occurs inside the topographic margin of the caldera as slabs in wall-collapse breccias, it is more accurate to refer to it as pre–caldera collapse tuff and the intracaldera tuff as caldera-collapse tuff.

The Wah Wah Springs ignimbrite shares compositional aspects with other monotonous intermediates (Figs. 5–7). Like the Cottonwood Wash, some samples are andesite but unlike that unit some ignimbrite samples are more silica rich and barely reach the rhyolite field (Fig. 30). Although other element concentrations are similar between these two tuffs, there are some significant differences. One intriguing distinction of the Wah Wah Springs (including ignimbrite, cognate inclusions, and intracaldera granodiorite porphyry) from other monotonous intermediates is the unusually high concentration of Cr (Fig. 30C); only the phenocryst-rich, andesite-latite Harmony Hills Tuff has still more Cr. The Wah Wah Springs can generally be distinguished in outcrop because of the presence of conspicuously large hornblende phenocrysts that are more abundant than biotite (Figs. 26B, 31, and 32), an aspect that is unique among all middle Cenozoic Great Basin ignimbrites of which we are aware. Whereas other monotonous intermediates contain proportionately more quartz phenocrysts, upwards of 25%, the Wah Wah Springs contains less than 10% and in some samples it is absent. Sanidine is notably absent throughout the ignimbrite unit but is found in one cognate pumice clast described below.

The weighted average age of the formation based on 16 published K-Ar determinations (corrected to new decay constants as needed) on biotite and hornblende from outflow tuff is 30.1 ± 0.3 Ma (Best and Grant, 1987). Eleven ⁴⁰Ar/³⁹Ar analyses of plagioclase in seven outflow and one intracaldera samples yield a weighted mean age of 30.06 ± 0.05 Ma (Table 1).

Correlation of the Wah Wah Springs throughout the vast outflow sheet based on stratigraphic, petrographic, and compositional attributes is confirmed by its reversed paleomagnetic direction at 30 distributed sites (S. Gromme and M. Hudson, 2006, personal commun.); the other two monotonous intermediates in the Indian Peak–Caliente field are normally magnetized.

The outflow (Fig. 29) was the most extensive at 32,000 km² and had the greatest volume of 3000 km³ (Table 2) of any ignimbrite in the Indian Peak–Caliente ignimbrite field, and as well in the Central Nevada field. It is found over a present north-south distance of 240 km and an east-west distance of 370 km from the High Plateaus of south-central Utah westward into south-central Nevada.

Sections are as thick as 500 m immediately north of the topographic margin of the caldera source in the central Needle Range and 460 m to the east in the Wah Wah Mountains. Both sections appear to lie in paleovalleys. In the central Needle Range (Best et al., 1987b), the outflow tuff is synformal with a northerly trending axis, whereas in the Wah Wah Mountains, the tuff completes the filling of a deep easterly trending paleovalley occupied by older tuff deposits (Abbott et al.,1983; see also Figs. 9 and 12). Beyond these paleovalleys, sections are as thick as 375 m.

On the western margin of the High Plateaus in south-central Utah, a thick pile of 34–33 Ma andesitic and dacitic lavas and tuffs in the Marysvale field (Hintze et al., 2003; Steven et al., 1979) prevented accumulation of the distal outflow in that area, but Wah Wah Springs ash flows traveled farther east around the north and south sides of the pile. On the north of the pile, three stratigraphic units of phenocryst-rich dacite tuff that are compositionally and petrographically similar to the Wah Wah Springs outflow tuff overlie a heterogeneous sequence of tuffs and lava and volcanic debris flows. These three units are the Three Creeks Tuff Member of the Bullion Canyon Volcanics, the volcanic rocks of Wales Canyon, and the tuff of Dog Valley. The latter unit is clearly older, at 33.6 Ma, than the Wah Wah Springs whereas the former two units that locally interfinger with one another are of roughly the same age as the Wah Wah Springs. Caskey and Shuey (1975) obtained reconnaissance paleomagnetic directions from four sites (their numbers 224, 225, 226, 227) in what they thought was an upper unit in their Clear Creek tuff, as the Three Creeks was then designated. The sites in this tuff are widely distributed in the east-central part of the Red Ridge 7.5-minute quadrangle where the tuff is less than 50 m thick, and are located either on or north of the hinge zone of the obscure trap-door caldera that was the source of the younger Three Creeks Tuff Member (Steven et al., 1979). Taken together, these paleomagnetic data are perfectly representative of those we have obtained from the Wah Wah Springs sites to the west. Hornblende phenocrysts in our sample REDR-CS227 have the same composition as those in the Wah Wah Springs and are unlike hornblendes in other ignimbrites in the central and eastern Great Basin, including other monotonous intermediates (Fig. 33). ⁴⁰Ar/³⁹Ar ages on plagioclase from our samples REDR-CS226 and REDR-CS227 are 29.82 ± 0.16 and 29.87 ± 0.15 Ma, respectively. Hence, we do not doubt the existence of the Wah Wah Springs north of Marysvale, as well as elsewhere on the western margin of the High Plateaus. Parenthetically, we note that apparently the most widespread of the Three Creeks cooling units occurs as far west as the Wah Wah Mountains (Fig. 28; Abbott et al., 1983; Best and Grant, 1987) where it lies between the Lund and older Isom tuffs (Table 1) as well as local andesitic lavas and, thus, has an age of ca. 28 Ma.

The distal outflow sheet of the Wah Wah Springs south of the Indian Peak caldera in westernmost Utah is as thick as 17 m and either lies conformably above a thick sequence of lake sediments and conglomerates (Hintze et al., 1994a) or is conformably interbedded within a sequence of lacustrine limestones (Blank, 1959). In distal sections southwest of the caldera source, outflow tuff is as thick as 120 m in the North Pahroc Range where beds of lacustrine limestone and sandstone interfinger between ignimbrite cooling units all the way up section from the Wah Wah Springs to the Bald Hills Tuff Member of the Isom Formation (Table 1), whereas conglomerate beds of rounded Paleozoic rock lie between cooling units below the Wah Wah Springs (Scott et al., 1994). This sequence of interbedded sedimentary rock and ignimbrite has an aggregate thickness of about 1 km, with about one-half of each, and was deposited in a basin that apparently existed for at least 5 m.y. The rounded clasts in the conglomerate appear to have been derived from a nearby long-lasting highland and are not a result of erosion off contemporaneously created fault blocks. Farther north in the North Pahroc Range, beds of ash-fall and reworked tuff ∼5 m thick lie between the Cottonwood Wash and the Wah Wah Springs tuffs and have a “Needles Range” modal composition (Swadley et al., 1994).

Proximal to the caldera source in the area of Condor Canyon (Gary J. Axen, 1989, personal commun.; also Best and Williams, 1997), the pre–caldera collapse ignimbrite pinches out over Cambrian carbonate rocks.

Some of the fine fallout (locally reworked) ash deposits in western Nebraska (Tedford et al., 1996; LaGarry, 1998; Larson and Evanoff, 1998) might have originated from the Wah Wah Springs super-eruption. A candidate correlative of the Wah Wah Springs is one of three ash beds that constitutes the widespread Nonpareil (NP) ash zone. In a 20-m-thick section of this ash zone studied by Tedford et al. (1996, p. 317), the lowest of three distinct ash beds has reverse polarity, like that of the Wah Wah Springs tuff, whereas the overlying two ash layers are normally polarized. They also indicate (p. 314) that at Roundtop an ⁴⁰Ar/³⁹Ar age of 30.05 ± 0.19 Ma was obtained on biotite by Swisher and Prothero (1990) on an ash correlated with the uppermost layer (NP₃). This age has not been adjusted to the Fish Canyon reference standard of 28.20 Ma. A sample (AJ-94) of the NP₃ ash from Swisher and Prothero’s Roundtop locality was kindly provided by James B. Swinehart. As expected for distal fallout ash, the sample has abundant, very fine glass particles (and plentiful interspersed secondary carbonate) in which sparse but larger crystals to as much as 0.4 mm are embedded; for comparison, phenocrysts in the ignimbrite in the Great Basin are about ten times larger. The proportions of different types of crystals are entirely consistent with the Wah Wah Springs; especially noteworthy is the greater amount of hornblende than biotite. Microprobe analyses of the hornblendes in the sample reveal that they are indistinguishable from those in the Wah Wah Springs ignimbrite for all of the nine elements analyzed (Fig. 33). Moreover, the NP₃ hornblendes are distinct from those in other hornblende-bearing dacite ignimbrites in the Indian Peak field as well as ignimbrites in the Central Nevada field.

Further study of the Nonpareil ashes is clearly warranted. Nonetheless, we feel confident, considering the vast volume and areal extent of the Wah Wah Springs ignimbrite, that hundreds of cubic kilometers of fallout, presumably finely sorted co-ignimbrite, ash was deposited from this super-eruption in the mid-continent, as for the Cottonwood Wash discussed above.

Unlike the lithic-free, pre–caldera collapse ignimbrite, the caldera-collapse Wah Wah Springs contains lapilli and rare blocks as much as a meter in diameter of varied volcanic rocks that generally constitute less than 10%, but locally make up as much as about half, of the tuff (Fig. 34). In most places in the Indian Peak Range, most of the lithic clasts are dark-colored, essentially aphyric volcanic rocks that resemble rhyolite lavas of the Escalante Desert Group; fewer clasts are of Cottonwood Wash tuff. In the northern White Rock Mountains, lapilli of sedimentary rock are common. These older rock fragments were apparently caught up in the pyroclastic mass as it vented along the margin of the collapsing caldera. None of this lithic ash-flow tuff is known to occur beyond the topographic margin of the caldera, implying ash flows venting along the ring fracture had insufficient energy to surmount this barrier. Pumice lapilli are obvious in many places and, although commonly densely compacted, the ignimbrite lacks an easily discerned foliation. An additional distinguishing aspect of the caldera-collapse ignimbrite is its widespread and variably intense, generally propylitic, alteration. The associated intrusive granodiorite porphyry is also so altered. Because variable degrees of alteration have affected nearly all samples of the ignimbrite, exact comparisons of modal and chemical composition with the pre-collapse tuff can be somewhat uncertain; however, chemical variation diagrams (Fig. 30) indicate it overlaps the less silica-rich, pre–caldera collapse ignimbrite whereas modal compositions of the two overlap completely (Fig. 31A).

The lithic-rich ignimbrite was sampled for paleomagnetic direction at two sites, 0.8 km east-northeast of Indian Peak by Shuey et al. (1976, their site WW 135) and 9 km southeast of Atlanta, Nevada by S. Gromme and M. Hudson (2006, personal commun., their site B98-8; our petrologic sample ATL-33 was collected at this site). The directions are identical but significantly steeper in inclination and more easterly in declination than directions in the pre–caldera collapse ignimbrite. Before tilt corrections, these two directions were closer to that of the mean Wah Wah Springs, which suggests that there was probably resurgence within the Indian Peak caldera before the included thick pile cooled below its Curie temperature.

To evaluate vertical zoning in the Wah Wah Springs pre–caldera collapse ignimbrite, five complete stratigraphic sections were sampled, including: (1) a 130-m-thick section (SILVRWL-1D) ∼45 km west of the caldera source at the south end of the Egan Range, (2) a 340-m-thick proximal section (HAM-10) on the west side of the Mountain Home Range north of the caldera, (3) a 106-m-thick section (FRSC-8) at the Wah Wah Springs type locality (Fig. 28) on the east flank of the Wah Wah Mountains northeast of the caldera, (4) a 270-m-thick section (LUND-10) at the southeast end of the Wah Wah Mountains, and (5) a distal 45-m-thick section (PANGNW-1) north of Panguitch, Utah, in the High Plateaus. A partial basal proximal section (HFW-8-153-3) was sampled on the east flank of the Mountain Home Range.

No systematic chemical zoning is evident in bulk rock samples from the FRSC-8 and LUND-10 sections whereas the other three disclose zoning in, for example, SiO₂ and TiO₂ (Fig. 35A). However, the zoning is inconsistent from one section to another; the HAM-10 section in the Mountain Home Range shows up-section decreases in SiO₂ and increases in TiO₂ whereas the SILVRWL-1D and PANGNW-1 sections west and east of the caldera are the reverse.

Bulk rock samples from four stratigraphic sections (SILVRWL-1D, HFW-8-153-3, FRSC-8, and PANGNW-1) were analyzed to evaluate vertical zoning in proportions among phenocrysts. No systematic modal variation in any phenocryst was found in the PANGNW-1 section. But, despite considerable scatter, the other three sections reveal poorly defined zoning with the proportion of plagioclase increasing up section (Fig. 35B). In the SILVRWL-1D and FRSC-8 sections, adjustment of total phenocryst concentrations to a dense rock equivalent revealed no systematic variation.

Analyzed samples of intracaldera, or caldera-collapse, ignimbrite are among the least evolved of the Wah Wah Springs data set (Fig. 30) but some outflow samples are equally so. The latter are mostly in the SILVRWL-1D, LUND-10, and PANGNW-1 sections.

The HFW-8-153-3 section that comprises the lower several meters of an ∼250-m-thick outflow sheet is atypical in several respects. Whereas ash-flow tuff beneath a near-basal vitrophyre is typically poorly welded and compacted and almost never exposed, in this section, less than one-third meter of tuff beneath the 3 m of black vitrophyre, and above the non-exposed base, is moderately welded, is glassy red-brown, and contains an unusually small proportion (14%, DRE [dense rock equivalent]) of unusually small (<1 mm) phenocrysts. This basal quartz- and clinopyroxene-free tuff (HFW-8-153-3A) is the most evolved (70.2 wt% SiO₂; 0.37 wt% TiO₂) of any Wah Wah Springs tuff and plots barely into the field of rhyolite on the total alkalies-silica diagram (Fig. 30A). The overlying black vitrophyre (HFW-8-153-3B; 24% phenocrysts) is only slightly less evolved at 70.1 wt% SiO₂ whereas the overlying red-brown devitrified tuff (HFW-8-153-3C; 30% phenocrysts) immediately above the vitrophyre is a more normal dacite with 68.4 wt% SiO₂.

In addition to the HFW-8-153-3 section, other samples of near-basal, densely welded, black vitrophyres have total phenocryst concentrations as low as 25%, but ranging more or less continuously to as high as 50% (adjusted to a dense rock equivalent; see Fig. 36). As well, the lowest parts of the Wah Wah Springs outflow sheet commonly have relatively smaller phenocrysts compared to overlying parts. Smaller, less abundant phenocrysts are conceivably the result of dynamic processes at the base of the pyroclastic flow as it moved across the depositional surface (e.g., Branney and Kokelaar, 2003, p. 29–30). On the other hand, the fewer, smaller phenocrysts in the lowest parts of the Wah Wah Springs outflow sheet might reflect a gradient in the pre-eruption magma chamber. The existence of compositional gradients in the chamber is supported by our observations on cognate inclusions.

Cognate pumiceous inclusions were collected at six sites (Table 6; for locations see Supplemental File 1 [see ^{footnote 1}]). Inclusions from all sites in the outflow and intracaldera ignimbrites are phenocryst-rich dacites, but one small (4 cm) pumice lapilli (sample BRN-1PC) collected from the non-welded base of the distal outflow is unique in being rhyolite (73.8 wt% SiO₂ and 0.31 wt% TiO₂) and very phenocryst poor (6%, DRE) (Figs. 30, 31, and 36). Because most of the phenocrysts appear to be resorbed, or are anhedral, the possibility arises that the rhyolite pumice might be a piece of fused Precambrian roof rock from above the magma chamber caught up in the early erupting magma. However, its initial ⁸⁷Sr/⁸⁶Sr composition of 0.7094 (Table 4) is identical to that of the Wah Wah Springs ignimbrite (0.7093–0.7095) and inconsistent with this origin; thus, it appears to represent a small parcel of more-evolved magma from near the top of the chamber. Proportions of phenocrysts in the rhyolite pumice are similar to less-evolved dacite tuff, except for the presence of substantial quartz and euhedral sanidine that has a relatively high Or content similar to that of sanidine in the Lund ignimbrite (see below). Other phenocryst compositions, such as that of hornblende (Fig. 33), are like those in less-evolved Wah Wah Springs tuff. Glass in the pumice is slightly more evolved than that in tuff (Table 5). All together, the compositions of phases in the rhyolite pumice indicate it is likely the low-temperature part of the main dacitic magma body and probably differentiated from that part of the Wah Wah Springs magma chamber.

Data from the dacite inclusions indicate that the main part of the pre-eruption chamber possessed compositional gradients. Inclusions hosted in intracaldera tuff have 65.7–66.1 wt% SiO₂ whereas those in pre–caldera collapse, or outflow, tuff contain slightly more at 67.1–67.6 wt%; these contrasts indicate the upper, early erupted part of the chamber had as much as 2 wt% more silica. Cognate inclusions hosted in intracaldera ignimbrite tend to have less plagioclase and more quartz than the outflow-hosted inclusions, whereas proportions of other phenocrysts overlap (Fig. 31B). Plots of quartz, plagioclase, and biotite in the entire population of ignimbrite and inclusion samples disclose no systematic patterns whereas plots of quartz versus hornblende and quartz versus clinopyroxene reveal subtle correlations (Fig. 37). Hornblende correlates negatively with respect to quartz, a pattern also seen in the Lund (Maughan et al., 2002, their figure 5). Clinopyroxene correlates positively with respect to quartz. Table 6 shows that two pumices from the GOUGE-3P site at the top of the outflow sheet have the lowest total quartz plus clinopyroxene (0.4%) whereas the highest totals (8%–14%) occur at the top of the intracaldera deposit at site GLE-6-98-1X. The intracaldera granodiorite porphyry exposed south of Indian Peak plots within the higher quartz-plus-clinopyroxene range. These data imply that the late-erupted and lower part of the Wah Wah Springs magma chamber had slightly less silica but a greater proportion of quartz and clinopyroxene, apparently at the expense of plagioclase and hornblende. Pumices at the MLLR-6-63 site at the top of the intracaldera host were apparently extracted from only the upper part of the chamber (0.2%–2.0% quartz plus clinopyroxene) whereas other pumices constitute a mixed population from different levels.

Experiments by Clemens and Wall (1981) on a granitic composition that is quite close to the most silica-rich Wah Wah Springs (except for half as much CaO and slightly more K₂O) reveal that the crystallization field of quartz broadens at higher pressure and lower water concentration in the system. Johnson and Rutherford (1989) showed the same for the dacite Fish Canyon magma.

Cognate inclusions from the top of the intracaldera deposit are mostly less vesicular (measured densities 1.59–2.33 g/cm³, average 2.0) than ones from the outflow sheet (0.88–1.81 g/cm³, average 1.4) (Table 6), indicating lesser concentration of volatiles in the corresponding deepest magma erupted.

The main dacitic part of the pre-eruption Wah Wah Springs magma chamber apparently possessed gradients in silica concentration of ∼2 wt% and in proportions of phenocrysts. The more silica-rich, upper part of the chamber had little or no quartz and clinopyroxene and the less silica-rich, lower erupted part had as much as 14% quartz plus clinopyroxene combined and less hornblende and volatiles. The uppermost level of the main part of the chamber had smaller and lesser concentrations of total phenocrysts. Apparently, a relatively very small differentiated cap on the main dacitic chamber consisted of a phenocryst-poor rhyolite that contained 74 wt% silica and somewhat similar proportions of phenocrysts of basically the same compositions as the dacitic part, but with additional sanidine. The more silica-rich, phenocryst-poorer uppermost level of the magma chamber, including the rhyolite cap, comprised but a few percent of the whole chamber.

Fractionation of fine particles of high-silica rhyolite glass from the ash flows appears to have had limited influence on compositional variations in the Wah Wah Springs ignimbrites. Although chemical compositions of some ignimbrite samples arguably reflect this process (Figs. 30A, 30C), modal data on a smaller set of samples discount a role. Total phenocryst concentrations on a dense rock equivalent basis in the cognate inclusions that range from 33%–48% overlap the most phenocryst-rich tuffs (Fig. 36). Limited fractionation is not necessarily contradicted by the presence of fine, glass-rich fallout ash in Nebraska, which represents the distal, winnowed fraction of the co-ignimbrite portion of the eruption.

On the basis of regional geologic relations, Shuey et al. (1976) suggested that the source of the Needles Range tuffs, and the Wah Wah Springs in particular, might lie in the vicinity of the Indian Peak Range. Mapping by Grant (1979) and Best et al. (1987b) revealed a sequence of ignimbrite and intercalated breccias that is 3500 m thick faulted against an apparently thick mass of Marsden Tuff in the vicinity of Indian Peak (Figs. 14, 38, and 39); these relations define the ring fault and northeastern structural margin of the Indian Peak caldera. Southward from the ring fault, breccias are composed of clasts of the Cottonwood Wash and Lamerdorf ignimbrites as well as crystal-poor andesitic and rhyolite rocks that resemble those of the Escalante Desert. Clasts of Paleozoic rock and the Marsden Tuff appear to be absent here.

The thickness of the caldera-collapse ignimbrite inside the structural margin cannot be determined directly because of the lack of internal stratigraphic markers, which precludes determination of displacements on normal faults repeating parts of the pile. But because the bottom of the pile, which has been tilted by basin-and-range faulting and by resurgence, is not exposed and the top has been eroded off, we assume that the unknown amount of missing section counterbalances the amount of section possibly repeated by faulting. The apparent thicknesses measured in four places from geologic maps and shown in Figure 29 are 2100, 3500, 4000, and 5000 m.

In the following sections we trace the evidence for and the character of the Indian Peak caldera, mostly along its margin in a clockwise direction, beginning in the central Needle Range. Reference to Figures 3, 8, and 29 will aid the reader in this circumnavigation.

Because of fault-related tilting subsequent to volcanism, the internal stratigraphy and structure of the northeastern sector of the Indian Peak caldera are well exposed in the central Needle Range (Figs. 8 and 39; see also Best et al., 1987a, 1987b; Best and Grant, 1987, p. 10–13; Best et al., 1989b, p. 121, especially their figure R32). In this sector, the collar zone of the caldera lies between its topographic margin on the northwest and its structural margin, marked by the inner ring fault, on the southeast. This collar-zone terrane extends ∼11 km parallel to the range and is composed of the pre-caldera section of Paleozoic rocks and an overlying Marsden sequence, described above, overlain in turn by a caldera-collapse breccia layer as thick as ∼700 m, described in detail below. Post-collapse, caldera-filling tuffs of the Ryan Spring Formation thicken from zero at the topographic margin of the caldera to as much as 500 m north of Indian Peak. The overlying, caldera-filling Lund ignimbrite is uniformly ∼600 m thick throughout the eastern Indian Peak Range south of the topographic margin, but 55 m thick north of it; thus, Lund ash flows barely spilled over the topographic margin. Overlying the Lund north and south of the topographic margin, the Isom ignimbrite is uniformly tens of meters thick.

The topographic margin of the Indian Peak caldera trends easterly for several kilometers across the central Needle Range. Immediately west of Sawtooth Peak, the margin is marked by outcrops of brecciated Cottonwood Wash Tuff, shown as unit Twib in Best et al. (1987b; see also next section). A narrow exposed segment of topographic-wall rocks, chiefly brecciated Cottonwood Wash, is offset several kilometers to the south of the peak by a major, south-striking, basin-and-range(?) fault. The topographic margin continues for about 4 km to the east between alluvium and overlying caldera-filling Lund. Reexamination of this narrow 4-km-long segment, which is as little as ∼7 km from the southern terminus of the collar breccia layer at the inner ring fault, has revealed rocks similar to those in the breccia layer.

The northwestern two-thirds of the breccia layer in the collar zone (Twb in Fig. 39A) is composed mostly of a matrix-supported, monolithologic breccia made of Cottonwood Wash ignimbrite that has a pervasive cataclastic fabric down to sub-phenocryst scale (Fig. 40). Clasts in outcrop have local, and commonly multiple, slickenside surfaces but no through-going interclast shears are evident. Also included in the breccia layer are local zones of brecciated Lamerdorf Tuff and andesitic lava, the latter likely derived from the post–Cottonwood Wash, pre–Wah Wah Springs lava dome on the topographic margin of the caldera (Fig. 39). Near the top of the layer are slabs of lithic-free, pre–caldera collapse Wah Wah Springs; these slabs are shown in Figure 39 but some are too small to appear in the cross section. No internal deformation is evident in the larger slabs.

Cataclastic Cottonwood Wash still constitutes most of southeastern third of the breccia layer, 2–2.5 km north of the inner ring fault. However, near the top of the layer, beneath the overlying caldera-filling Ryan Spring ignimbrites, many outcrops reveal more or less planar seams that are as much as a few meters square and generally <1 cm thick in which a substantial proportion consists of broken grain fragments of submicron dimensions (Fig. 41). These seams of ultracataclasite (Snoke et al., 1998, p. 8) lie on the upper parts of outcrops of brecciated Cottonwood Wash. In places, seams lie on two near-orthogonal sides of the outcrop, or a thin, branching subsidiary seam penetrates several centimeters into the exposed block. Variably cataclastic Wah Wah Springs and Cottonwood Wash tuffs commonly lie on opposite sides of a seam with the former generally on top of the outcrop. Thin laminae within the ultracataclasite seams appear isotropic under cross-polarized light and some parts appear to be glass that contains scattered minute grain fragments. Because ultracataclasite and pseudotachylyte are commonly intimately associated and are considered to form a continuum as extreme brittle comminution leads into frictional melting under very high strain rates (e.g., Magloughlin and Spray, 1992; Spray, 1995; Snoke et al., 1998), it is possible that the glassy seams are the quenched product of frictional melting. However, because none of the devitrification textures typically seen in pseudotachylyte exist, we tentatively conclude that the glass represents segregated and fused vitroclasts from the host tuff.

Although caldera collapse was undoubtedly involved in the creation of the remarkably extensive breccia layer and its local seams of ultracataclasite near the ring fault, we are uncertain of many details regarding the mechanics of their origin. It appears unlikely that the entire layer represents a single massive landslide calved off the topographic-wall margin that now lies 11 km outboard from the structural-margin ring fault. As kilometers of caldera subsidence was taking place, why was there no apparent breakup of the Marsden or Paleozoic section in the collar zone, only of the overlying Cottonwood Wash and Lamerdorf tuffs? Why is there is so little of the pre–caldera collapse Wah Wah Springs ignimbrite in the breccia layer, given that a thickness as much as 500 m is exposed immediately north of the topographic margin (Fig. 29)? Why does no lithic Wah Wah Springs occur in the collar zone, as might be expected if the collar zone were downsagging to generate sufficient slope to cause extensional breakup and brecciation of the Cottonwood Wash and other rock units? Although the upper contact of the breccia layer with the overlying Ryan Spring Formation is presently ∼200 m higher near the inner ring fault than near the topographic margin, this slope must have reversed since the layer originated, presumably as a result of resurgent uplift of the caldera.

First of all, we speculate that a brecciating layer of the Cottonwood Wash tuff and thinner underlying Lamerdorf moved southward and down on a weak, enabling detachment surface—such as provided by the sandstone and debris-flow deposits that are locally exposed at the top of the Marsden sequence.

Further speculation regarding the setting in which the layer originated as well as the evolution of the whole collar zone follows from a wide range of independent observations on calderas, subsidence into other types of evacuating subsurface voids, and experimental studies using scaled analog models (e.g., Acocella et al., 2012; Acocella, 2007; Branney, 1995; Burchardt and Walter, 2010; Howard, 2010; Kennedy et al., 2004; Roche et al., 2000). All of these studies reveal that following downsag, caldera collapse is accompanied by development of an annular collar zone bounded by an inner, bell-shaped reverse ring fault and a complementary curved outer, normal ring fault; together these two faults define a downward-tapering, triangular annulus (Fig. 42) whose apex meets the margin of the erupting magma chamber. In plan view, these bounding faults form a set of segmented arcuate faults. For model experiments in which chambers have a horizontal diameter much larger than thickness, Roche et al. (2000) found that the initial subsidence involves downsag, which continues with displacement on the normal and reverse faults, whose dips at the surface are as little as 50° but steepen downward. Initially, a reverse fault develops at the margin of the chamber followed by a normal fault on the outside, or a normal fault followed on the outside by a reverse fault. In all cases, subsidence is asymmetric, with the maximum on the side of the first reverse fault. As subsidence increases, (1) successive reverse ring faults develop stepwise outward into the annulus as (2) extension also progresses outward in the annulus, (3) marker layers rotate down and into the depression, and (4) blocks break away and slide into the depression. Roche et al. (2000, p. 410) speculated that the topographic margins of many large calderas coincide with the outer limit of extension.

The evolution of the collar zone of the Indian Peak caldera is shown schematically in the idealized cross sections of Figure 43 that are essentially coincident with the plane of the section in Figure 39B. Our speculative model is designed to be consistent with the findings described in the previous paragraph and with our observations on the Indian Peak caldera. We do not know how the engulfed older source calderas of certainly the Cottonwood Wash and possibly of earlier rhyolite ignimbrites (Sawtooth Peak, Marsden, Lamerdorf) might have impacted the evolution of the collar zone.

Figure 43A represents the situation before collapse. The thickness of the Cottonwood Wash Tuff is shown to increase somewhat southward, consistent with a postulated source to the southwest where it was engulfed into the Indian Peak or overlapping White Rock caldera. A further and intuitively reasonable assumption is that surface tumescence resulting from growth of the Wah Wah Springs magma chamber in the shallow crust caused the pre–caldera collapse ignimbrite to pinch out southward; the flat surface in the section reflects initial down flexure after it erupted.

Figure 43B shows a small increment of downsag accompanied initial subsidence of the caldera on the inner reverse ring fault shown dashed in A. The inner reverse and outer normal ring faults converge to the margin of the erupting magma chamber, shown at a depth of more than 7 km, as indicated by the pressure of equilibration of the phase assemblage in the Wah Wah Springs ignimbrite (Wolff, 2008). As a result of the subsidence of the footwall caldera block, the unstable edge of the layer of Cottonwood Wash Tuff fails and brecciates; the potential void along the reverse ring fault is filled with boiling magma that erupts through the brecciating mass, entraining fragments into the ejecta. A major proportion of the lithic clasts in the caldera-collapse tuff north of Indian Peak are of aphyric lava of the Escalante Desert type (Fig. 34), indicating venting of boiling magma through a large mass of this rock at depth.

Figure 43C indicates that 2000 m of further subsidence of the caldera has occurred, schematically, by additional downsag but mostly by movement on the two reverse faults located to the north-northwest of the inner ring fault, which are dashed in B. Part of the layer of the Cottonwood Wash fails and brecciates in the downward-sloping hanging-wall block. Lamerdorf Tuff overlying the hypothetical detachment layer at the top of the Marsden also slides downward and brecciates, as does Escalante Desert rhyolite lava (see Fig. 39). All of these brecciating units slide into the accumulating, thickening pile of intracaldera ejecta—forming, in part, mappable lenses—and contribute to the lithic component of it. In the deeper levels of the breccia layer, sufficient load exists to create intense cataclasis at the base of sliding blocks. Boiling magma continues to ascend along the inner reverse ring fault; we speculate it might have propped up the Paleozoic- and Marsden-bounded margin of the hanging-wall block so it did not fail and fragment. Magma must, now, erupt through the thickening accumulating pile of intracaldera ejecta, thereby consuming energy and constraining the lithic ash flows within the topographic margin of the caldera. Incremental failure and brecciation of the Cottonwood Wash layer toward the topographic margin of the caldera is accomplished by activation of successive reverse faults stepping in that direction that results in concomitant extension of the collar-zone, hanging-wall block. Additional faults stepping farther to the north-northwest that are activated in the next cross section are dashed.

Figure 43D shows that a total of 4500 m of subsidence has occurred as the eruptive activity ends. The last ∼2000 m of movement takes place on faults in the northern half of the collar zone, mostly on a normal fault located ∼3 km south of the topographic margin and labeled the outer ring fault in Figure 39. Paleozoic sedimentary strata have dropped down to the south apparently by ∼2600 m, based on projection of a marker unit across the fault. The strata have also been rotated almost 50° from an easterly strike and north dip of ∼30° to a northwesterly strike and east dip of ∼60°. (Compensation has been made in these measurements for the east tilt of the range by post-volcanic basin-and-range faulting.) About 600 m of displacement on this fault occurs as the sequence of Marsden ignimbrites, debris flows, and breccias of Paleozoic rocks, described above, accumulates south of the fault, and 200 m of displacement occurs after the caldera-filling Ryan Spring and Lund ignimbrites are emplaced. This leaves possibly ∼700 m of throw during subsidence of the Indian Peak caldera and during which time an equivalent thickness of caldera breccia is deposited. Further brecciation of the Cottonwood Wash layer occurs back to the andesitic lava dome, the south part of which also collapses and brecciates. Slabs of the tapered wedge of pre–caldera collapse Wah Wah Springs ignimbrite slide downward on the brecciating Cottonwood Wash. More magma invades along the inner reverse ring fault and already initiated resurgent uplift of the collar zone block. Note that no lithic Wah Wah Springs ash flows advance beyond the midpoint of the collar zone.

Figure 43E (cf. Fig. 39B) indicates still more magma intrudes beneath the southern half of the collar zone, causing resurgent uplift by reversal of displacement along the three inner ring faults ornamented with half arrows. Paleomagnetic data discussed above are consistent with resurgence before the intracaldera tuff cooled below its Curie temperature. The magma beneath the resurgent dome solidifies as granodiorite porphyry. Note that hypothetical, lithic caldera-collapse Wah Wah Springs ignimbrite has been eroded off the resurgently uplifted southern half of the collar zone, in agreement with its observed absence there. As much as 1100 m of post-collapse, caldera-filling Ryan Spring and Lund ignimbrites accumulate in the caldera moat between the topographic margin and resurgent intracaldera pile.

Caveat. Our proposed evolution of the northeastern segment of the Indian Peak caldera collar zone with its unusually broad breccia layer satisfies the constraints of the observed geology as well as observations on naturally occurring, smaller-scale subsidence overlying voids and on experimental studies. However, we note that our proposed evolution for the Indian Peak caldera finds no support in the Caetano and Stillwater calderas in the Western Nevada field (Henry and John, 2013), from which more than 1000 km³ of ejecta were vented. Post-collapse tilting that has exposed 4000–5000-m-thick cross sections through the calderas reveals subvertical to steeply inward-dipping master ring faults and topographic margins less than 1 km outboard of the ring fault. The contrast between these calderas and our interpretation of the northeastern collar zone of the Indian Peak is striking.

Southward of Indian Peak for ∼9 km (Figs. 8 and 39B) are intrusive masses, likely connected at depth, of granodiorite porphyry (Grant, 1979; Best et al., 1987a). Although propylitically altered, this porphyry has the distinct Wah Wah Springs modal and chemical composition, including relatively high Cr. For another 9 km southward of the largest intrusion and exposed over ∼5 km², the sequence of lithic tuff and minor wall-collapse breccia is tilted to the south from 20° to 65° over the intrusion.

At Arrowhead Pass (Best et al., 1987a), an ignimbrite-breccia sequence ∼4000 m thick to the north is juxtaposed across an easterly striking fault against a section of Paleozoic sedimentary rocks. We interpret this pre–Ryan Spring fault to be a ring fault that marks the southern structural margin of the caldera. Overlying the Paleozoic section is a south-dipping sequence no more than 200 m thick of the Sawtooth Peak ignimbrite overlain by less than 400 m of lithic Wah Wah Springs followed by ignimbrites and lava flows of the caldera-filling Ryan Spring Formation. The relatively thin, lithic Wah Wah Springs continues southward for several kilometers before disappearing beneath a 1500-m-thick section of Lund and Isom ignimbrites on the margin of the younger White Rock caldera (see below). No wall-collapse breccias associated with the subsidence of the Indian Peak caldera are exposed in this southern collar zone of the caldera like those in the northern collar zone. The topographic margin of the Indian Peak caldera must lie farther south, buried beneath younger deposits. That the Wah Wah Springs ignimbrite south of the ring fault at Arrowhead Pass is not pre-collapse tuff but caldera-collapse tuff is indicated by the presence of lithic clasts and especially by the occurrence within it in the western foothills of the range of a one-half square kilometer exposure of altered and brecciated Paleozoic rocks of unknown stratigraphic identity (Best et al., 1987a). Drilling has shown that these Paleozoic rocks are underlain by altered volcanic rocks (Gerald Park, consulting geologist, Salt Lake City, Utah, 1992, personal commun.) and are, therefore, not an in situ caldera floor but a slide mass.

Relations near Arrowhead Pass (1) as detailed above, indicate that the Indian Peak caldera inside its southeastern margin did not subside nearly as much as in its northeastern sector, that is, caldera subsidence was asymmetric; and (2) are consistent with resurgent uplift of an intracaldera structural block and subsequent erosion prior to deposition. The post-caldera, 500-m-thick Ryan Spring tuff is draped across the northern end of the ring fault and the rocks to the north and south. However, this unit pinches out 7 km to the north and for several kilometers to the south it is <50 m thick and generally absent.

On the west side of the southern Wah Wah Mountains in hills at the south end of Pine Valley, Best et al. (1987d) found no wall-collapse breccias but mapped ∼1 km² of lithic tuff (unit Twi on their geologic map) like that which occurs within the Indian Peak caldera in the Indian Peak Range to the northwest; the tuff is at least 500 m thick and mostly is in fault contact with Lund and Mackleprang ignimbrites. Reexamination of variably altered exposures in low hills to the southeast, which were designated as outflow, or pre–caldera collapse, Wah Wah Springs (unit Two) on the geologic map, reveals that they actually contain a few percent of clasts to as much as 0.3 m across of aphyric red volcanic rock and should, therefore, also be considered as caldera-collapse Wah Wah Springs. The unit pinches out southward on a paleohill of Mesozoic and Paleozoic rocks (Fig. 8). We interpret this area of lithic Wah Wah Springs to lie within the southeastern collar zone of the Indian Peak caldera between the ring fault and an obscure topographic margin to the east and south. The relatively large areal extent of the southeastern collar zone sector of the caldera is related in part to post-volcanic extension.

South of the Indian Peak Range, the position of the caldera margin lacks geologic constraints because of burial beneath younger deposits, but gravity data (Fig. 8C) indicate no apparent extension into the Escalante Desert. West of the desert, 25–20 Ma lava flows further conceal the caldera margin.

In the southern White Rock Mountains along the Utah-Nevada state line, in the Rice Mountain quadrangle (Keith et al., 1994; Figs. 8, 29, and 44), no wall-collapse breccia is exposed, but less than 250 m of lithic Wah Wah Springs overlies older volcanic rocks deposited on Cambrian limestone. Drastic changes in thickness of the lithic tuff across a fault indicate displacement on the fault after its deposition followed by erosion before the younger Ryan Spring ignimbrite was emplaced. These relations are interpreted to have resulted from resurgent uplift in the Indian Peak caldera.

At the south end of the Wilson Creek Range, over a distance of ∼6 km northeast of Ursine, reconnaissance mapping discloses a stratigraphic section that consists of Cambrian limestone overlain by a few hundred meters of Escalante Desert andesite lavas and Cottonwood Wash and Wah Wah Springs ignimbrites; this section is typical of those outside the Indian Peak caldera margin. However, to the west of the major Meadow Valley Wash through Ursine, this pre-Lund volcanic section is entirely absent and the Cambrian strata are overlain by roughly 500 m, or more, of Lund tuff that has local black vitrophyre near its base. The underlying limestone is variably bleached and locally intensely silicified for several meters below the contact. We believe this area northwest of Ursine lies in the Indian Peak caldera collar zone and the missing pre-Lund volcanic units became detached from the sequence to the northeast and slid to the north into the caldera before deposition of the Lund; downsag in this part of the collar zone could have facilitated the slippage.

Five to ten kilometers to the northwest of Ursine on the western flank of the southern Wilson Creek Range, more than 260 m of non-lithic, pre-collapse Wah Wah Springs is exposed. The southwestern caldera margin likely passes between this area and the lithic Wah Wah Springs exposed in the Rice Mountain quadrangle (Fig. 44). The caldera margin must also lie between this area of exposure and one farther to the northwest, ∼17 km northwest of Ursine, where reconnaissance mapping reveals a 5 km² window of generally poorly exposed volcanic rocks amongst alluvium and surrounded on virtually all sides by early Miocene rhyolite lavas and tuffs. On the west side of the window, apparently in situ, mostly unaltered Paleozoic carbonate rocks are overlain to the east by Cottonwood Wash Tuff. Still farther east is a northerly trending, 100-m-long by tens of meters wide train of boulders of white Eureka Quartzite of Ordovician age and minor gray carbonate rocks lying in alluvium. The nearest exposures today of the quartzite lie 45 km to the north, at the north end of the range at Atlanta (Willis et al., 1987). It is unlikely the boulders are a part of the alluvial debris, but instead represent erosionally resistant fragments of wall breccia shed off the caldera margin from the west that are imbedded in weathered unexposed caldera-collapse ignimbrite; such wall breccia is well exposed 2 km southeast of Atlanta, as described below. Beyond this train of apparent wall-breccia boulders are exposures of non-lithic Wah Wah Springs tuff with randomly oriented compaction foliation that appear to represent chaotic blocks of a megabreccia. In one outcrop, a slab ∼0.5 m thick of non-lithic Wah Wah Springs is embedded in lithic Wah Wah Springs (Fig. 45). Extending for ∼2 km east of this intracaldera assemblage is a moderately north-dipping section of Lund ignimbrite that appears to be a few hundred meters thick and disappears to the east beneath flat-lying Isom ignimbrite and overlying rhyolitic tuff and lavas. Locally, at the unconformable contact between the Lund and the Isom, is another outcrop area of only a few hundred meters square of Wah Wah Springs megabreccia that here is a heterogeneous assortment of lithologically contrasting megablocks of igneous rocks of unknown stratigraphic identity in a lithic Wah Wah Springs matrix.

Considering the position of this window of older rocks in the central part of the eclipsing younger White Rock caldera (see below), we interpret the entire window as an exposure of the floor of the resurgent uplift of this caldera that reveals a segment of the margin of the older Indian Peak caldera.

West of the northern Wilson Creek Range and across Lake Valley in the northern Fairview Range (Figs. 8 and 29), a westernmost segment of the Indian Peak caldera is no longer believed to be present, as once thought (Best et al., 1998). This erroneous conclusion was based on a correlation of a local plagioclase-pyroxene tuff with the caldera-collapse Wah Wah Springs. However, the local tuff has a different composition and paleomagnetic direction than the Wah Wah Springs and our age on sample GRASSY-1-213-2 of 33.94 ± 0.18 Ma is much too old.

Geologic relations in the northern Wilson Creek Range (Willis et al., 1987) further constrain the position of the Indian Peak caldera margin. West of the Atlanta mine, the caldera margin is apparently eclipsed by the younger White Rock caldera (see below). Isolated exposures of altered ignimbrite—apparently both Wah Wah Springs and Lund—lie south of a kilometer-square area of rhyolite lava that is about the same age as the formation of The Gouge Eye to the northwest in the Fortification Range. This rhyolite lava was erroneously assigned to the younger 30 Ma Ryan Spring Formation (Willis et al., 1987) but fission-track dating of zircon yields significantly older ages of 34.8 ± 3.2 Ma (sample ATL-1-70-3) and 35.6 ± 3.3 Ma (HOR-1-70-4) (Kowallis and Best, 1990).

About 2 km southeast of the Atlanta mine, some exposures consist of breccia of silicified Ordovician and Silurian sedimentary rock clasts in a matrix of intensely argillized, quartz-free Wah Wah Springs, or with the altered tuff occuring as thin septa within the brecciated rock. These exposures represent the broken caldera wall that was invaded by or fell into the Wah Wah Springs ejecta. About 9 km south-southeast of the Atlanta mine, and farther into the caldera, a slab more than 1 km in diameter of broken Silurian dolomite (unit Twbd in Fig. 46A) overlies slightly altered caldera-collapse lithic Wah Wah Springs. Nearby, smaller remnant slabs of brecciated Ordovician Eureka Quartzite rest on tuff that is more than 2100 m thick.

For several kilometers to the southeast, the Indian Peak caldera margin is obscured by younger volcanic rocks, but critical relations reappear in the northern White Rock Mountains (Fig. 46B). Here, in the “moat” between the topographic margin and the resurgent uplifted interior of the caldera, upwards of 700 m of post-collapse, caldera-filling Ryan Spring and Lund ignimbrites covers the structural margin ring fault. The thickness of the resurgently tilted lithic Wah Wah Springs tuff appears to be on the order of 5000 m; although this extraordinary thickness might be the result of repetition by undetected faults, its base is not exposed, and hence these two factors may counterbalance. No wall-collapse breccias occur in this intracaldera section that is ∼10 km from the structural margin.

The northern margin of the Indian Peak caldera trends east-southeast across Hamlin Valley into the central Needle Range.

As drawn in Figures 8 and 29 from the evidence cited in the preceding paragraphs, the Indian Peak caldera spans four mountain ranges. The topographic margin extends ∼40 km north-south and the area within it was 2000 km² prior to extension. The position of the deeply subsided, structural caldera inside the inner reverse ring fault, whose area was one-half that within the topographic margin, is well constrained in the Indian Peak Range, but elsewhere is less certain. In the north, the ring fault may lie close to the exposed topographic wall, and is shown coincident in Figure 29 just north of the intracaldera section southeast of Atlanta that is more than 2100 m thick. Southwestward into the Wilson Creek Range, the position of the structural margin is drawn conservatively; the topographic margin is believed to lie just north of Ursine and then swings northwest up Lake Valley.

Because of the large dimensions of the Wah Wah Springs ignimbrite and its Indian Peak caldera source, and with the uncertainties in the margins of the latter, we employ four models to estimate the volume of the ash-flow deposit (Fig. 4 and Tables 2 and 7). Our intent is to find some agreement amongst the different model estimates because sufficient data exist to make valid comparisons.

In Model 1, the volume of the contoured pre–caldera collapse ignimbrite is 3000 km³, using a 400 m maximum thickness across the future caldera. The volume of the pre-collapse ignimbrite within the topographic margin of the future caldera is 800 km³ because the area is 2000 km² (Table 7), making the volume of the outflow beyond the margin 2200 km³. These figures assume no tumescence, or doming, of the land surface prior to eruption of the Wah Wah Springs ash flows resulting from growth of the underlying shallow crustal magma chamber, as was assumed in Figure 43A. We note that there is no basis in fact for this intuitively reasonable assumption of surface tumescence. In his review of Tertiary calderas in western North America, Lipman (1984, p. 8813) concluded that evidence for inflation “is elusive.” In the well exposed, essentially three-dimensional exposures of the Caetano caldera in the Western Nevada field (Henry and John, 2013) there is no direct evidence for pre-collapse doming. If, say, 100 m of tumescence did occur over the caldera area with negligible downsag, then the 3000 km³ volume would be reduced by less than 100 km³. An additional uncertainty in these calculations is the unknown thick volume of Wah Wah Springs ignimbrite filling the older engulfed Cottonwood Wash caldera. In any case, doubling the volume of the pre–caldera collapse ignimbrite, according to the concept of Lipman (1984), yields a total volume for the Wah Wah Springs ignimbrite of 6000 km³.

In Model 2, the volume of the contoured tuff beyond the structural caldera is added to the volume of the intracaldera tuff within it, assuming an average thickness of 3650 m and a simple piston geometry; this estimate yields a total of 6200 km³ for the entire unit.

The volume estimate for Model 3 is similar but assumes a more realistic asymmetric caldera in which the intracaldera tuff ranges from a thickness contour of 1000 m coincident with the southern structural margin to a maximum contour of 4000 m inside the caldera (Fig. 29); this estimate yields a total of 5700 km³ for the entire unit.

In Models 2 and 3, we have not compensated for the volume of wall-collapse breccia within the intracaldera tuff. But the fallout ash deposited beyond the outflow sheet to as far as western Nebraska which can easily amount to hundreds of cubic kilometers might counterbalance the subtracted wall-collapse breccia.

Model 4 in Figure 4 is an entirely different approach that avoids some of the uncertainties of the previous estimates; it was employed by Lipman (1997) to determine the 5000 km³ volume of magma erupted from the La Garita caldera represented in the Fish Canyon Tuff in the Southern Rocky Mountain volcanic field. This volume estimate is based on a simplified geometry in which the assumed symmetric subsidence of a uniform plate, or piston, multiplied by the structural caldera area yields the erupted magma volume. For the subsidence in the Indian Peak caldera, we use the maximum thicknesses of the intracaldera tuff, 5000 m, and for the “unfilled collar height” of Lipman (1997), the 1100 m thickness of the post-collapse, caldera-filling Ryan Spring and Lund tuffs deposited in the collar zone in the central Needle Range. Southwest of Atlanta, no Lund is exposed but the Ryan Spring is 1300 m thick (Willis et al., 1987). The Model 4 calculation yields 6100 km³ for the volume of the erupted Wah Wah Springs magma. Because the intracaldera pile includes wall-collapse breccias as well as ignimbrite, the actual thickness of the ignimbrite is some lesser value than 5000 m. Using the formulation of Lipman (1997, his appendix 1) for the collar-rock breccia volume (Cv) collapsed into an equivalent circular structural caldera from an equivalent circular topographic caldera, we calculate ∼200 km³, or ∼3% of the magma volume, reducing the erupted magma volume to 5900 km³.

All four models yield similar volumes, ranging from 5700 to 6200 km³ (Table 7). These two extreme values are for asymmetric and uniform caldera collapse models, 3 and 2, respectively. From the foregoing discussion of the Indian Peak caldera, there is little doubt that it collapsed asymmetrically, so the 6200 km³ volume is unrealistically high. Interestingly, the 6000 km³ volume derived from the doubling assumption is not significantly different from volumes found by the other models. The average of the four models is 5900 km³, which we take as our preferred total volume for the Wah Wah Springs ignimbrite (Table 2). This preferred volume of 5900 km³ equals the estimated volume of the erupted magma, which would argue that not accounting for the wall-collapse breccia in Models 2 and 3 is counterbalanced by the ignored fallout-ash volume.

A puzzling aspect of the Indian Peak caldera is the apparent dearth of eroded rock debris shed off the topographic wall and resurgent uplift into the depression. Despite the fact that the caldera-filling Ryan Spring and Lund tuffs are as much as 1100 m thick in the Indian Peak Range segment, the total thickness of observed sediment is generally no more than a few tens of meters. The sediment is typically crudely bedded, but well sorted, sandstone that occurs between cooling units of the Ryan Spring Formation and locally overlying this unit and below the Lund. Although the volume of tuff eroded off the resurgent uplift is unknown, this small amount of exposed sediment seems problematic. Breaching of the topographic margin of the caldera may have allowed eroded rock to be flushed out of the depression, but no significant sediment is known between the Wah Wah Springs and younger tuffs beyond the caldera. The location of the caldera on the Great Basin altiplano (Best et al., 2009, p. 617) where little precipitation apparently occurred might account for the sparse erosional sediment; such is the case on the modern central Andean plateau.

The 29.40 ± 0.06 Ma Silver King Tuff lies stratigraphically above the rhyolite tuff of Deadman Spring (Table 1) in the Fairview Range in the northwestern Indian Peak–Caliente field (Fig. 47; Best et al., 1998). The Silver King was previously designated as an informal stratigraphic unit (Maughan et al., 2002), but here we make it a formal unit.

As with other phenocryst-rich dacitic tuffs in the southern Great Basin ignimbrite province, the Silver King contains abundant plagioclase and lesser hornblende, biotite, quartz, sanidine, and magnetite (Fig. 48), and trace amounts of ilmenite, apatite, and zircon. A small amount (<2%) of titanite is characteristic of this ignimbrite. Hornblende is distinctly less abundant than biotite (ratio <0.47 in Fig. 26B). Chemically, the average Silver King is more evolved than most samples of monotonous intermediates (Figs. 5, 6, 27A, 30A, and 49). A more significant difference from the three super–monotonous intermediates in the Indian Peak field is the modest areal extent and order-of-magnitude lesser volume of the Silver King, estimated by Model 1 in Figure 4 to be ∼350 km³ (Table 1).

No direct evidence of a source caldera for the Silver King ignimbrite has been found. It is debatable whether the considerable thickness of the unit, as much as ∼1400 m, within the older Kixmiller caldera source of the tuff of Deadman Spring in the Fairview Range represents post-collapse caldera fill or is a syn-collapse deposit within a Silver King caldera nested in the older Kixmiller caldera. Widespread areas in the southwestern Fairview Range of brecciated and cataclastically deformed Silver King tuff, which were previously interpreted to be related to the collapse of the younger source caldera of the Lund tuff (Best et al., 1998), could have taken place during or shortly after emplacement of the Silver King at its source.

The loosely welded ash-flow tuff underlying the Lund in the northern White River Narrows is not the Silver King; its paleomagnetic direction (sample 9P058, S. Gromme and M. Hudson, 2006, personal commun.) is significantly different and the modal composition also differs. Instead, the phenocryst assemblage is similar to that of the lava flows erupted from the 29–27 Ma Seaman volcanic center (duBray, 1993) ∼13 km to the west. This pre-Lund ignimbrite may represent a small-volume explosive eruption during the development of that volcano.

Ignimbrite of this formation resulted from the youngest, at 29.20 ± 0.08 Ma, of the three super-eruptions of monotonous intermediates in the Indian Peak field. The Lund Formation also includes local intracaldera wall-collapse breccias as well as small lava domes and possible shallow intrusions along the caldera ring fracture.

The Lund ignimbrite (Maughan et al., 2002) is petrographically and compositionally similar to the other monotonous intermediates but can usually be distinguished in the field by the presence of trace amounts of titanite phenocrysts; however, as this phase is commonly obliterated by even slight alteration, other criteria must be employed to distinguish the unit. These criteria include:

1. Typically smaller mafic phenocrysts, generally no more than 2 mm, in contrast to larger mafic phenocrysts in the other two monotonous intermediates, especially the biotites in the Cottonwood Wash and the more conspicuous abundant hornblendes in the Wah Wah Springs (Figs. 26 B, 32, and 50).

2. Generally more quartz and presence of sanidine in most samples. Sanidine has not been verified in the Cottonwood Wash and Wah Wah Springs ignimbrites.

Although petrographically and modally similar to the slightly older Silver King Tuff, the hornblende/biotite ratio of the Lund (>0.6) exceeds that of the Silver King (<0.5) (Fig. 26B). The Sr isotopic composition of the Lund is also slightly lower than that of the Silver King (0.7092 versus 0.7099; Table 4). Paleomagnetic analysis shows the Silver King to be reversely magnetized instead of normal as for the Lund. These two cooling units are in direct contact in the Fairview Range and nearby locales, and in other places a thin andesitic lava separates them.

Samples of the Lund ignimbrite range across the dacite field and slightly into the rhyolite field; SiO₂ ranges from ∼64 to 71 wt% (Figs. 5, 6, and 49). Relative to the Cottonwood Wash and Wah Wah Springs ignimbrites, the Lund has more TiO₂, Sr, Ba, and Zr at a given silica content. The Lund ranges to lower silica contents than the Silver King and has distinctly higher concentrations of P₂O₅, Fe₂O₃, Sc, and V at similar TiO₂ (Maughan, 1996). Because some element variation diagrams show separate but sub-parallel trends for the Lund and Silver King tuffs, it is unlikely that the Lund magma evolved from the same system as the 0.2-m.y.-older Silver King.

The relatively large variation in the proportions and compositions of phenocrysts in the Lund (Fig. 50), as well as the order-of-magnitude variation in some of their ratios, reflect compositional heterogeneity in the pre-eruption magma chamber (Maughan et al., 2002, p. 139); the increase in sanidine and quartz relative to declining mafic phases is that expected in a crystallizing magma. The variation cannot be entirely the result of emplacement processes, e.g., fractionation of fine vitroclasts.

Despite these overall modal variations in the Lund tuff, no consistent spatial pattern of variation in the proportions of phenocrysts can be discerned laterally in the outflow sheet or in four vertically sampled stratigraphic sections. However, an upward decrease of ∼10% in the phenocryst/glass ratio in two of the sections may account for subtle and inconsistent reverse chemical zonation manifest in decreasing MgO and TiO₂ and increasing K₂O. The average outflow seems to have significantly fewer phenocrysts, on a dense rock basis, than the intracaldera tuff (34% versus 51%), which would be consistent with a pre-eruption chamber possessing deeper, more crystal-rich magma. The modal proportion of quartz in the outflow is lower (9% versus 17%) and the total mafics higher (32% versus 25%) than in the intracaldera tuff. Because the outflow and intracaldera tuffs are chemically indistinguishable there is, thus, a puzzling disconnect between the elemental and modal composition of the Lund ignimbrite (Maughan et al., 2002).

None of the Lund contains lithic clasts except for at one locale north of Ursine noted above. The general absence of lithic fragments in the caldera-collapse Lund is unlike in the older, lithic intracaldera Wah Wah Springs monotonous intermediate.

The thickest exposed sections of intracaldera Lund ignimbrite occur in the northwest sector of the White Rock caldera source, whereas the thickest outflow lies to the east of the caldera (Fig. 51). Dispersal of pre-collapse ash flows appears to have been constrained by their eruption from the eastern sector of the caldera source which was located, at least in part, within the incompletely filled depression of the older Indian Peak caldera. Consequently, 300–350 m of outflow ignimbrite is exposed in the southern Wah Wah Mountains whereas just to the west within the Indian Peak depression the Lund is 450 to as much as 1400 m thick, the latter thickness reflecting accumulation within the nested Mackleprang caldera.

Although the Lund is 230 m thick in Condor Canyon (G.J. Axen, 1989, personal commun.), immediately to the north in the southwestern corner of the Rose Valley quadrangle (Best and Williams, 1997) the ignimbrite pinches out over a highland of Cambrian rock that apparently extended to the west.

The present east-west extent of the Lund ignimbrite is more than twice that of the north-south extent, exceeding that which would have resulted solely from crustal extension (an assumed uniform 50%) after emplacement. The topographic wall on the older Indian Peak caldera apparently mostly blocked dispersal of Lund ash flows to the north. Similarly, the Indian Peak topographic wall that is postulated to lie in Lake Valley (Fig. 29) could have prevented dispersal of thick ash flows to the west. An alternate factor that possibly influenced the relatively large east-west distribution of the Lund is obscure extensional faults of the same orientation that had possibly created a concealed graben where the ignimbrite was deposited. Although this implied stress field—north-south extension—is seemingly at variance with prevailing wisdom for the middle Cenozoic Great Basin, it is consistent with widespread east-west dikes of about this age (Best, 1988). North-south extension is also evident in the late Oligocene–early Miocene, west-northwest–striking normal faults in the High Plateaus of south-central Utah (Anderson, 2001).

In contrast to the well-exposed internal structure of the Indian Peak caldera in the Needle Range, no similar deep exposures exist for the White Rock caldera, the source of the Lund ignimbrite. The following paragraphs describe successive segments of the caldera in a clockwise direction, beginning with the best-exposed northern part (Figs. 8 and 51).

In the low foothills of the northwestern Wilson Creek Range north of Mount Wilson (Fig. 22), lenses of bleached Paleozoic carbonate rock fragments interfinger with Lund tuff in an almost continuously exposed sequence 2500 m thick (Willis et al., 1987). As is commonly the case in thick sections of tuff that lack stratigraphic markers, there may be some repetition by unrecognized normal faults; however, the base of the section is cut off by the transecting younger Mount Wilson caldera. One breccia lens has pieces of andesitic lava that were probably derived from the andesite flows at the base of the ignimbrite sequence overlying the Pennsylvanian–Permian Ely Limestone in the divide between Grassy Mountain and Fairview Range to the west across Lake Valley (Fig. 17; Best et al., 1998). Patches of ash-flow tuff and minor breccia lenses surrounded by alluvial cover occur northward for ∼12 km to the approximately located northern margin of the intracaldera deposit. In this area west of the Atlanta mine, the margin of the White Rock caldera appears to merge with the margin of the older Indian Peak caldera (Fig. 8). Partially filling in the northern part of the White Rock caldera southwest of the Atlanta mine is ∼400 m of ignimbrite of the Ripgut Formation.

For ∼22 km southeast of the Atlanta mine, the caldera margin is concealed beneath a tableland of early Miocene rhyolite tuffs and lava flows (Willis et al., 1987).

Beyond and to the southeast of the tableland, in the northern White Rock Mountains (Fig. 46B; Best et al., 1989d), more than 900 m of Lund tuff is banked against thick lithic Wah Wah Springs ignimbrite, defining the topographic margin of the White Rock caldera. As much as 600 m of Ripgut tuff and more than 350 m of the overlying Isom tuff constitute post-caldera fill. In the southern White Rock Mountains, in the Rice Mountain 7.5-minute quadrangle (Fig. 44; Keith et al., 1994), faulted and tilted Lund is unconformably overlain by Ripgut tuff. To the south of this resurgent uplift of the White Rock caldera, the Ripgut thickens to greater than 650 m.

The caldera margin continues southeastward across Hamlin Valley into the southwestern Indian Peak Range. Here, a post-Lund andesite lava dome that is 400 m thick and 2 km in diameter is overlain by as much as 900 m of Isom ignimbrite that thins abruptly to less than 250 m to the northeast, hence, beyond the topographic margin of the White Rock caldera. Outside the White Rock caldera but within the northern collar zone of the Indian Peak caldera (Figs. 8 and 51), the Lund section between older and younger deposits is 570 m thick.

Farther east within the collar zone between the structural and topographic margins of the Indian Peak caldera, unusually thick sections of Lund tuff occur in the hills at the south end of Pine Valley. Three stratigraphic sections have thicknesses of 630, 1100, and 1400 m (Best et al., 1987d) but these are approximate as sections are incomplete without exposed bases and tops; sections might also have unrecognized faults. Not far to the north (Abbott et al., 1983) and east (Hintze et al., 1994b) the Lund is only 200–500 m thick (Fig. 51). We believe the unusually thick Lund marks a caldera source (Fig. 8) of the immediately older Mackleprang ignimbrite.

To the west, the roughly east-west–trending southern margin of the White Rock caldera is concealed beneath younger deposits and is constrained only by a gradient in Bouguer gravity that limits the southern margin of the greater Indian Peak caldera complex (Fig. 8C).

More than 15 km north of the approximately located southern caldera margin is an exposed area of the resurgent floor of the White Rock caldera. The southern part of this area, immediately north of Ursine, occupies the northwest corner of the map by Williams et al. (1997), whereas the remaining part is covered only by unpublished reconnaissance mapping. The western half of this area consists of a northerly trending synclinal outcropping of Cambrian carbonate rocks (see maps of Tschanz and Pampeyan [1970] and Ekren et al. [1977]). These rocks are overlain by a compound cooling unit of rather densely welded Lund ignimbrite more than 600 m thick. The Cambrian rocks are variably silicified along the depositional contact with the overlying Lund tuff, which is locally a vitrophyre. The Lund here is atypical in that it contains sparse lapilli of red aphanitic volcanic rock. Within 2 km east of Ursine, the Lund grades locally into zones where clasts of older rock, as much as a meter across (including Wah Wah Springs and Ryan Spring tuffs as well as andesitic rock), are embedded in a matrix of comminuted Lund tuff.

About 5 km north of Ursine in the northern part of the area, the synclinal terrain of Cambrian rocks and overlying thick intracaldera Lund are truncated by a high-angle east-west fault that is on trend with the marked eastward jog in Meadow Valley Wash. North of this fault, which appears to be caldera related, a mélange that overlies the Wah Wah Springs ignimbrite consists of pieces of somewhat altered Lund vitrophyre and cataclastically deformed granitic rock and rare altered Wah Wah Springs(?) as well as abundant crystal-rich rhyolite of unknown stratigraphic identity. The mélange, which is overlain by a thick sequence of younger andesitic and dacitic lava flows, is interpreted to be a landslide deposit that involved some stratigraphically unrecognized rock units on the floor of the White Rock caldera and has been exposed at the surface by considerable resurgence and erosion.

For ∼12 km farther northwest along the southern Wilson Creek Range, thick deposits of lower Miocene rhyolite tuff and lava conceal older units. But in the western foothills the Lund is mostly absent between patches of older outflow Wah Wah Springs and younger Isom ignimbrites, presumably because the intracaldera Lund was removed by erosion off the resurgent uplift of the caldera prior to deposition of the Isom.

About 17 km northwest of Ursine is the window area in the southwestern Wilson Creek Range (Fig. 8A) along the margin of the Indian Peak caldera described above. For ∼2 km east of the intracaldera Wah Wah Springs assemblage exposed in this window is a moderately north-dipping section of Lund tuff that is possibly as much as 700 m thick and disappears beneath flat-lying Isom tuff and overlying rhyolitic tuff and lavas. Locally at the top of the exposed section of Lund and at the unconformable contact between it and the overlying Isom is an exposure of a heterogeneous assortment of lithologically contrasting megablocks in a lithic Wah Wah Springs matrix. We interpret the entire window area to be a segment of the resurgently uplifted floor of the White Rock caldera that includes the margin of the older Indian Peak caldera.

We believe the Pioche Hills horst and flanking grabens represent a significant defining segment of the southwestern collar zone of the White Rock caldera (Fig. 8A).

Exposures in the relatively small, ∼10 × 4 km, anomalously northwest-trending horst consist of a sequence of Cambrian sedimentary rocks almost 2000 m thick (Fig. 52A; see also Park et al., 1958; Tschanz and Pampeyan, 1970). The Pioche Hills have been the site of significant zinc, lead, silver, and manganese mining (Gemmill, 1968); these ore deposits are possibly related, at least in part, to the Lund magma system and attendant caldera collapse.

Four periods of extensional faulting are recognized in the region encompassing the Pioche Hills (Axen, 1998; Axen et al., 1988; Burke and Axen, 1997; Rowley et al., 1994; G.J. Axen, 1989, personal commun.). The latest period involved normal faulting after the mid-Miocene that formed horsts and grabens, or the present-day ranges and valleys. Mid-Miocene (ca. 15 Ma) faulting is dominated by the Highland detachment (Fig. 8A) that is exposed west of the Pioche Hills and has more than 10 km of westward displacement. Late Oligocene, synvolcanic (ca. 29–27 Ma) faulting is implied by contrasts in dip of less than 30° of Oligocene and early Miocene ignimbrite sheets. Evidence for pre-volcanic extension is represented by closely spaced normal faults and east-west–trending tear faults that are found in the footwall of the Highland detachment and that sole into the Stampede detachment, according to Axen et al. (1988); this detachment is a zone of bedding-parallel faults between the middle Pioche Shale and lower Highland Peak Formation. The Stampede detachment cuts through the Pioche Hills and was recognized by Tschanz and Pampeyan (1970), who believed it to be an early Tertiary (“Laramide”) regional thrust sheet.

Significant aspects of the earlier geologic studies and large-scale mapping that bear on the elucidation of the White Rock caldera margin are summarized in the following paragraphs. Especially pertinent is the work of Tschanz et al. (2009) done in the mid-1900s but recently edited and made available to us by D.R. Shawe; this work contains detailed observations of the extensive underground mine workings.

Faulting in the generally gently dipping Cambrian rocks in the Pioche Hills is unusually intricate and complex, in marked contrast to the “simple structure” of the nearby Highland Range to the west and the Bristol Range to the northwest. Faults that cause bedding offset have normal extensional displacements, but it is common for a fault to have been offset by a cross fault, after which block movements continued on both. Relative ages of thrust faults, bedding faults, and normal faults are difficult to evaluate, but in most places the evidence is conflicting or at least confusing. In many places contemporaneity is indicated and one type of fault may bend and merge with a fault of another system. In one instance, as many as four cross-cutting generations of faults can be documented that have different orientations and senses of movement. Late deformation was so intense and widespread that it likely reactivated and/or reoriented earlier faults.

Everywhere associated with the complex faults are zones of breccia that occur in all lithologic units—limestone, dolomite, shale, and quartzite. The breccia zones can be as much as hundreds of meters thick and continuous along strike for kilometers with only minor interruptions and offsets. In many places, both on the surface and underground, breccia is all that remains of beds of quartzite hundreds of meters thick. Why deformation caused massive quartzite to shatter rather than to be taken up in the overlying Pioche Shale is puzzling. But, locally, the shale unit is highly sheared. The almost complete absence of folding and crumpling in the Pioche Hills, even in shales, and the very widespread brecciation of quartzite and carbonate rocks are thought to indicate that the faulting took place under comparatively light loads, or shallow burial.

The overall structure of the Pioche Hills shown in the east-west cross section in Park et al. (1958, their section CC′) and Tschanz and Pampeyan (1970, their plate 4) is a system of normal faults that have downdropped subhorizontal strata across the hills in a synformal manner. However, underground workings in Tschanz et al. (2009) reveal a more complex, even chaotic, structure than portrayed in this generalized published section. This general pattern of dominantly downward displacements inward toward the center of the horst is suggestive to us of that seen in some extensional grabens in the Basin and Range province (e.g., Effimoff and Pinezich, 1986; Liberty et al., 1994). It is unlike the outward and downward displacements marginal to horsts typically seen in the province.

In contrast to the relatively flat strata in the fault synform, bedding in the Prospect Mountain Quartzite in the eastern part of the Pioche Hills has a moderate east dip that we interpret to be the result of initial downsagging along the margin of the White Rock caldera to the northeast as magma began to be withdrawn from the underlying chamber (Fig. 52A). On the other hand, G.J. Axen (2010, personal commun.) believes this discordance in dip resulted from faulting related to the Stampede detachment.

The anomalous northwest-southeast trend of the relatively small Pioche Hills horst may reflect the local structural grain in this part of the Great Basin; the northwest-trending Kern Mountains east of Ely may be another example. G.J. Axen (2010, personal commun.) suggested the anomalous range trend is the result of vertical-axis rotational displacement that opened the valley to the south, a possibility that could be tested by paleomagnetic data. Yet another explanation, which we favor, is that the unusual trend of the Pioche Hills horst, compared to the typical north-south trend of larger ranges in the Great Basin, simply reflects a controlling influence by marginal faults along the White Rock caldera.

Although Tschanz et al. (2009) noted the existence of welded andesite and dacite ash-flow tuff in the Pioche Hills, only three surface exposures totaling no more than 0.01 km² are shown on the geologic map (Park et al., 1958). One in the northwest is Cottonwood Wash Tuff whereas the other two are of Wah Wah Springs ignimbrite. One of the exposures of the Wah Wah Springs is a 100-m-long “dike” as much as 8 m wide exposed near the crest of the range at the head of Slaughterhouse Canyon near coordinates 16,500 N and 10,000 E on Park et al. (1958). The “dike” rock is crushed and the broken rock along its borders contains angular fragments of limestone up to 20 cm in diameter. “Dikes” of welded tuff have also been found in the underground workings. Descriptions and a few available photomicrographs of the tuffs in the Pioche Hills (Tschanz et al., 2009) are consistent with the occurrence of any one or all of the three phenocryst-rich dacite ignimbrites—Wah Wah Springs, Cottonwood Wash, and Lund. Although they report that pieces of welded tuff can be found on the mine dumps, we were unable to find any with diligent searching. Tschanz et al. (2009) wondered whether the “dikes” might represent vents for eruption of the welded tuff sheets exposed, for example, in Condor Canyon to the southeast. This may be the case for “dikes” of the dacitic Lund that was derived from a magma chamber whose margin, in our interpretation, lies under the Pioche Hills. However, our examination of all that remains exposed of the Slaughterhouse Canyon “dike” of welded Wah Wah Springs lends little support for the feeder “dike” interpretation, for the following reasons:

1. The eutaxitic foliation is perpendicular in places to the contact with the Cambrian country rock; this fact, together with the crushed and broken character of the rock in the “dike,” seems inconsistent with it being a feeder vent (cf. Ekren and Byers, 1976), unless brecciation somehow occurred after the “dike” was emplaced.

2. A feeder dike of Wah Wah Springs tuff is unlikely 25 km southwest of the margin of its Indian Peak caldera source (Fig. 8).

3. No thermal alteration, such as bleaching, of the adjacent limestone wall rock is apparent, as would be expected in a hot 8-m-wide feeder “dike.”

For these three reasons, we believe the “dike” originated as a deposit of ignimbrite capping the Cambrian rocks that was drawn into a dilatant tensional fissure in the Cambrian rock during deformation, such as described by Lipman (1964) for a similar occurrence in southern Nevada. It is likely that at least some of the “dikes” of welded tuff in underground workings are also dilatant fissure fillings.

Vikre and Browne (1999) listed K-Ar ages on biotites from two underground vitrophyre dikes and the Slaughterhouse Canyon Wah Wah Springs dike; the three ages (29.2 ± 1.2, 28.3 ± 0.8, 28.8 ± 0.9 Ma) are within analytical error of one another and have a weighted average of 28.6 ± 0.5 Ma, which seems more consistent with the Lund (29.20 Ma) rather than the Wah Wah Springs (30.06 Ma). However, these K-Ar ages on biotites cannot be compared directly with the ⁴⁰Ar/³⁹Ar ages on feldspars reported in this article, which were determined in a different analytical facility.

Numerous exploratory churn-drill holes drilled in the late 1940s and early 1950s sought extensions of the ore bodies in the grabens flanking the Pioche Hills horst. At depths of less than a few hundred meters beneath alluvium the drills commonly encountered Cambrian limestone that overlay welded tuff, or alternations of the two, which Tschanz and Pampeyan (1970) referred to as “megabreccia” (see cross sections in Gemmill, 1968; Park et al., 1958; Tschanz et al., 2009). Our examination of the few available drill cuttings of tuff revealed them to be of Wah Wah Springs and Lund. One especially deep hole on the west flank of the hills encountered these megabreccias overlying a thick section of volcanic rocks that continued to a depth of more than 760 m. “Megabreccia” was not reported in the underground workings in the Pioche Hills horst.

“Megabreccias” in the flanking graben were interpreted by early workers to have resulted from thrust faulting or from gravity sliding, conceivably off a rising horst. On the other hand, we believe they are entirely consistent with intercalation of ignimbrites and wall-collapse breccias in the collar zone of the White Rock caldera.

Salient geologic attributes of the Pioche Hills, as contained in the early reports, include the following:

1. Unusual northwest-southeast trend of the Pioche Hills horst.

2. Complex and intricate faulting.

3. Widespread, pervasive brecciation, including of thick-bedded quartzites.

4. Generalized synformal structure in flat-lying strata in the core of the Pioche Hills horst; the structure in underground mine workings appears to be more chaotic.

5. A surface exposure of a dilatant fissure filling of welded Wah Wah Springs tuff embedded in broken Cambrian rocks, and by implication, additional such features in the underground workings. Underground occurrences of Lund tuff could be either fissure fillings or feeder dikes.

6. “Megabreccias” of alternating welded tuff and Cambrian limestone and of limestone overlying tuff in grabens flanking the central horst.

Some structural aspects are possibly related to regional extensional tectonism. For example, G.J. Axen (2010, personal commun.) noted that some steep faults with fairly large stratigraphic separations in the Pioche Hills die out as they approach the Pioche Shale; he believes this geometry is consistent with soling of the steep faults into the Stampede detachment. Overprinting by extensional basin-and-range faulting is an additional complexity.

However, considered all together, we believe the six salient geologic aspects of the Pioche Hills are entirely consistent with collapse along the southwestern collar-zone margin of the White Rock caldera. Our interpreted evolution of the caldera margin, shown schematically in Figure 52, was inspired by the work of Branney (1995, especially his figures 1, 2, and 5), Roche et al. (2000), and Burchardt and Walter (2010) and profited from suggestions by Gary J. Axen and Daniel R. Shawe. Our interpretation follows the same sort of caldera-related extensional collapse that we believe took place in the northeastern collar zone of the Indian Peak caldera (Fig. 43), but differs in some details. Roche et al. (2000) and Burchardt and Walter (2010) used natural examples in concert with analogue model experiments to demonstrate that caldera subsidence commonly begins as a downsag but displacement generally progresses along outward-stepping and outward-dipping reverse faults whose dips steepen with depth; then, as displacement continues, peripheral extensional fractures develop in the hanging-wall block and propagate downward to form inward-dipping normal faults that merge at depth with the earlier reverse faults and take up continued displacement (Figs. 42 and 43).

In Figure 52A, as magma begins to be withdrawn from the underlying Lund chamber, a downsag develops along the caldera margin. We interpret the east-dipping Prospect Mountain Quartzite to be evidence for this downsag, but, as noted above, this dip could be related to the pre-caldera Stampede detachment and related extensional faulting, which, for simplicity, is shown here only schematically and is omitted in succeeding parts of the figure. Caldera-collapse Lund ignimbrite begins to pond within the caldera depression.

In Figure 52B, further eruption of Lund ejecta results in further downsagging as well as initial reverse movement on an outward-dipping ring fault. The hanging-wall block, whose unmodified form is indicated by the blue dashed line, experiences extensional deformation strain, including fracturing and fissuring (long dashed lines), to compensate for the wedge-shaped potential void along the fault. A scallop of the steep topographic margin of the hanging-wall block collapses into landslides that cascade into the deepening depression as more Lund is deposited, creating intercalations of wall-collapse breccia of Cambrian limestone (and minor Wah Wah Springs ignimbrite) within the Lund.

In Figure 52C, considerable further subsidence of the caldera floor during continuing eruption produces large displacement along the reverse ring fault. The large potential void above the fault is compensated for by large-magnitude unconfined collapse, a major component of which is a synformal structure within the hanging-wall block that is bounded on the southwest by a major normal fault and on the northeast by the remnant of the east-dipping downsag. This sort of structure, called a “keystone graben” by Branney (1995, e.g., his figure 5), is drawn to correspond to the northeast-southwest cross section of the Pioche Hills shown in Park et al. (1958, their section CC′) and in Tschanz and Pampeyan (1970, their plate 4). The rapid unconfined collapse during the geologically instantaneous subsidence of the caldera imparts a significant amount of observed complexity to the extensional faulting as well as widespread brecciation that is overprinted on the basic synformal structure. Rapid, large-magnitude unconfined collapse differs from deformation in typical extensional tectonic regimes of more or less confined and slow incremental movement. Significantly, dilatant fissures would draw in both the older, cold Wah Wah Springs as well as the newly deposited Lund, creating “dikes.” Steep topographic walls would collapse to create lenses of breccia mingled within the thick intracaldera Lund.

In Figure 52D, later basin-and-range faulting drops grabens flanking the central Pioche Hills horst, where the preserved keystone graben of Cambrian strata is now exposed. “Megabreccias” of interbedded Lund and wall-collapse breccia in these flanking grabens were intersected in exploratory drill holes.

In their work on the ore deposits in the Pioche mining district that includes the Highland and Bristol Ranges to the west of the Pioche Hills, Vikre and Browne (1999) documented two periods of magmatic activity and mineralization, namely, Cretaceous at 100–90 Ma and late Oligocene at 29–27 Ma (biotite K-Ar). The latter period is roughly that of the Lund activity, with due allowance for different methods of dating, and carries the implication that localization of the Cu-rich Oligocene ores could have been related to the extensional structures along the caldera margin as well as to the underlying Lund magma system. Apart from the ages on “dikes” in the Pioche Hills indicated above, late Oligocene magmatism is manifest by the small Ida May dike of altered granitic porphyry in the northern Bristol Range, just to the west of the hypothesized western margin of the White Rock caldera; this dike has a K-Ar age on sericite of 26.7 ± 0.7 Ma. Farther north, in the southern Fairview Range (Best et al., 1998), a granitic porphyry with a ⁴⁰Ar/³⁹Ar age on biotite of 29.84 ± 0.07 Ma is spatially associated with silver mineralization. This porphyry was linked (above) with the tuff of Deadman Spring and formation of the Kixmiller caldera. On the other hand, owing to the commonly recognized older ⁴⁰Ar/³⁹Ar age of biotite compared to that of sanidine in the same sample (e.g., Salisbury et al., 2011), this porphyry may instead be co-genetic with the Lund.

Additional direct evidence for the White Rock caldera margin is in the Fairview Range 30–40 km north-northwest of Pioche where lenses of breccia of older rocks, chiefly tuff of Deadman Spring, are intercalated within a section of Lund that is more than 800 m thick on the east flank of the range (Best et al., 1998). Similar breccias, but not in contact with Lund, occur elsewhere in the range. Some fault blocks expose greater than 1600 m of Lund without intercalated breccias, as well as sections of Deadman Spring and Silver King ignimbrites more than 1000 m thick. All of these thick sections of tuff were likely deposited within their own source calderas or in older depressions nested within the northwestern segment of the Indian Peak caldera complex.

Six, mostly small (<1 km diameter) lava flows and a vent complex appear to mark the northern ring fault of the White Rock caldera (Fig. 51) and crop out less than 4 km inboard (mostly south) of the exposed topographic margin of the caldera. The lava flows are porphyritic, holocrystalline to glassy, and mostly flow layered. Outer contacts of the smaller ones are concealed beneath alluvium, leaving open the possibility that they could be shallow intrusions. The large (1 × 1.5 km) lava flow in the Wilson Creek Range ∼5 km south-southwest of Atlanta (Willis et al., 1987) is a rhyolite (Fig. 49) that is more evolved than the Lund ignimbrite samples but plots along the same element-variation trends. Three samples from ring-fault lavas in the White Rock Mountains (Best et al., 1989d) are dacitic like the Lund. The easternmost mass has a fission-track age on zircon of 27.6 ± 2.5 Ma (Kowallis and Best, 1990). The vent complex in the western White Rock Mountains has a bedded surge deposit at the base that grades upwards into massive tuff containing clasts as much as 1 m in diameter of Lund-like vitrophyre and capped by a mass of a similar tuff. Small blocks of granodiorite and feldspar-quartz-biotite-hornblende gneiss also occur in the massive tuff. The lava flow in the Fairview Range differs from the other ring-fault occurrences in being larger (∼2.5 km diameter) and having conspicuous phenocrysts of hornblende (lava flow of Chokecherry Spring in Best et al., 1998). Nonetheless, its stratigraphically constrained age and chemical composition are consistent with a co-genetic relation to the Lund magma system (Maughan, 1996).

On the basis of geologic relations in the several sites just described, we construct the outline of the White Rock caldera shown in Figures 8 and 51; the topographic caldera could be as much as 60 km north-south and have an equivalent diameter of ∼50 km after correction for extension (Table 2).

Although less is known of the internal character of the White Rock caldera relative to the Indian Peak caldera, some comparisons can be made. Both show clear evidence for resurgent uplift after collapse. Both disclose only little intracaldera sediment; only 130 m was found in the caldera-filling Ripgut Formation in one part of the White Rock caldera. Significant differences between the two calderas include:

1. Ring-fault magmatic activity is expressed in the White Rock caldera whereas in the Indian Peak caldera post-collapse magmatic activity is manifest by resurgent intrusion of granodiorite porphyry in the eastern core of the caldera; this large post-collapse porphyry may be a factor in the widespread alteration of the Wah Wah Springs intracaldera tuff, in contrast to the general absence of alteration in the intracaldera Lund.

2. Comparison of Figures 29 and 51 reveals a wider collar zone between the topographic and structural margins of the Indian Peak caldera than inferred for the White Rock caldera, implying a greater amount of subsidence resulting in a greater retreat of the topographic wall in the former. This contrast is evident in the estimates of subsidence for the Indian Peak caldera of 6100 m (see above) and for the White Rock caldera of 2900 m (see below).

3. A possible consequence of the greater subsidence of the Indian Peak caldera is the abundance of lithic clasts in the intracaldera or collapse ignimbrite in it relative to the intracaldera Lund tuff in the White Rock caldera that is mostly lacking in lithics.

To estimate the total volume of the Lund ignimbrite we use the same multi-model approach (Fig. 4 and Tables 2 and 7) as for the Wah Wah Springs.

In Model 1, the volume of the contoured, pre–caldera collapse Lund exclusive of the source caldera, but including the thick accumulation within the older Indian Peak and nested Mackleprang depressions, is 2300 km³. Doubling this value to account for the hypothetical intracaldera deposit yields 4600 km³.

For the White Rock caldera used in Models 2–4, we assume the concealed inner ring fault marking the margin of the structural caldera lies a uniform 5 km inside the topographic margin, which is the entire perimeter of the caldera shown in Figures 8 and 51. The 5-km-inboard assumption is constrained by the positions of ring-fault lavas. The area of the so-defined structural caldera is 1500 km². In Model 2, we use a uniform thickness of 2500 m of caldera-collapse Lund and add this volume (3750 km³) to that of the pre–caldera collapse ignimbrite, yielding 5400 km³. In Model 3, the contribution of contoured caldera-collapse tuff (maximum thickness 2000 m) in an asymmetric caldera has been added to the pre-collapse tuff, giving 4600 km³. It should be noted that the assumed asymmetry of the caldera is based on the thickest exposed caldera-collapse tuff lying in the northern sector of the caldera whereas the thickest pre–caldera collapse tuff lies to the east of the caldera.

In Model 4, to evaluate the amount of subsidence of the caldera floor, we use the thicknesses of the caldera-collapse Lund plus the overlying caldera-filling Ripgut and Isom, for which there are four constraints. In the southern Indian Peak Range, the southeastern extremity of the White Rock caldera at its topographic margin is filled by a 400-m-thick, post-Lund lava dome and ∼650 m of overlying Isom tuff, making the total caldera fill thickness 1050 m. In the Fairview Range, the Lund is more than 800 m thick. (A greater than 1600-m-thick section in another fault block has no intercalated breccias and could be filling an older depression related to eruption of the Silver King or Deadman tuffs.) In the central White Rock Mountains, the Lund is greater than 900 m thick and the overlying caldera-filling Ripgut and Isom tuffs are 600 m and more than 350 m, respectively, giving a subsidence greater than 1850 m. The fourth constraint is in the northern Wilson Creek Range, north of Mount Wilson, where 2500 m of Lund is exposed and, to the north, ∼400 m of Ripgut tuff, giving ∼2900 m of subsidence. Following the approach used by Lipman (1997), and assuming 2900 m of subsidence across the 1500 km² area of the structural caldera, we determine the erupted volume of the Lund magma to be 4350 km³. Because the intracaldera pile includes wall-collapse breccias as well as ignimbrite, the actual thickness of the latter is some lesser value than 2900 m. Using the formulation of Lipman (1997, his appendix 1) for the collar-rock breccia volume (Cv) collapsed into an equivalent circular structural caldera from an equivalent circular topographic caldera, we calculate a maximum of ∼250 km³, or 6% of the magma volume, reducing the erupted magma volume to 4100 km³.

Models 2–3 ignore the volume of fallout ash surely deposited beyond the Lund ignimbrite as well as the volume of wall-collapse breccias within the intracaldera tuff; these two volumes might counterbalance.

In summary, the estimated volumes using Models 1, 3, and 4 give values of 4600, 4600, and 4100 km³, respectively, whereas Model 2 gives significantly more at 5400 km³. Our preferred total volume of the Lund ignimbrite is 4400 km³.

Whereas rhyolite ignimbrites occur throughout the 36–18 Ma period of activity in the southern Great Basin ignimbrite province, super-eruptions of monotonous intermediates are restricted to a brief time period of 31.13–29.20 Ma in the Indian Peak caldera complex and to 27.57 Ma in the Central Nevada complex. Immediately following these monotonous intermediate eruptions, numerous cooling units of unique trachydacitic ignimbrite were erupted from both caldera complexes. These typically densely welded, mostly relatively thin cooling units were designated as the Isom compositional type by Best et al. (1989b), but because their characteristics go beyond composition, and for brevity, they are referred to here as Isom-type tuffs. They contain less than 15% phenocrysts, mostly plagioclase, together with lesser clino- and orthopyroxene and magnetite, which are commonly aggregated (Fig. 53). Ilmenite and apatite are rare and most samples lack zircon. Very sparse grains of sanidine, quartz, amphibole, and biotite occur inconsistently and appear to be, at least in part, xenocrystic, possibly derived from the older monotonous intermediates such as the Lund, which occurs only very locally as xenoliths. Lapilli of andesitic rock are common.

In the Indian Peak–Caliente field, Isom-type tuffs comprise a single cooling unit of the 29.1 Ma Petroglyph Cliff Ignimbrite and nine recognized cooling units constituting three tuff members of the 27.90–24.55 Ma Isom Formation (Table 1; another Isom-type cooling unit with an age of ca. 23 Ma and apparently of local origin in the Northern Pahroc Range in the southwestern sector of the field [Scott et al., 1992] is not discussed further here). During the 4.6 m.y. interval of Isom-type eruptions only one other ignimbrite was deposited—the rhyolitic Ripgut–—shortly after the Petroglyph Cliff. The sources of the Petroglyph Cliff Ignimbrite and the Isom Formation lie just beyond the northwest and the southeast margins of the Indian Peak–Caliente caldera complex, respectively (Fig. 8). The 4.6 m.y. Isom-type time interval corresponds to the time for production of dacite-rhyolite magmas in the crust to shift southward from the Indian Peak focus to the Caliente (Fig. 2).

On many chemical variation diagrams, the Petroglyph Cliff Ignimbrite and the three members of the Isom Formation are distinct from one another (Figs. 5, 6C, and 54).

Like other Isom-type tuffs, the Petroglyph Cliff Ignimbrite (Cook, 1965; originally designated as the White Rock Spring Ignimbrite by Martin, 1957) is generally partially to densely welded and sparsely porphyritic, but is readily distinguished from them by unusually abundant clasts of both cognate and foreign heritage; the unit is basically a tuff breccia. In many outcrops there is a continuum in fragment size from the finest ash to the largest clast, ∼25 cm, so there is no clearly distinguishable matrix, implying a low-energy eruptive process such as dome collapse creating a block-and-ash flow. Striking contrasts in the color of the fragments—shades of orange, red, purple, brown, gray, and black—lend a distinct mottled appearance to outcrops. Most of the clasts appear to be cognate and some clasts are themselves pieces of eutaxitic tuff that contain discernible clasts. Angular lapilli of purple-gray porphyritic andesitic rock are common. Pale orange to yellow haloes surround holes that are filled with fine unidentified material. The abundance of clasts and the potential for their disaggregation prior to final emplacement of the rock preclude an accurate determination of phenocryst proportions. The difficulty of excluding foreign material during sample preparation likely compromises the true chemical composition.

The most distinctive type of cognate clast, not seen in other Isom-type tuffs, is rounded, texturally distinct “blobs” (Fig. 55). Phenocrysts, constituting less than 10% of this clast type, consist of tabular plagioclase (87%), pyroxene (10%), and Fe-Ti oxides (3%) in a vesicle-free, black glassy matrix. These blobs are especially conspicuous in the intracaldera unit in the Blind Mountain caldera source and at its type locality at the north end of White River Narrows, but are also seen in the Fairview Range. Their shapes are consistent with incorporation into the ash flow in molten form. Chemically, the blobs are somewhat variable in composition and range from andesite to trachydacite (Figs. 54 and 56). Similar blobs characterize the dacitic San Jose ignimbrite in northern Peru (Longo, 2006) and have been reported in the Grizzly Peak Tuff in southwestern Colorado (Fridrich and Mahood, 1987).

The bulk chemical composition of the Petroglyph Cliff Ignimbrite is not as extreme for several elements as in other Isom-type tuffs (Figs. 5, 6, and 54); for example, TiO₂, Ba, and Zr are lower, as are total alkalies so that most samples are dacite and less are trachydacite, andesite, and latite.

Although not dated isotopically, the 29.1 Ma age of the Petroglyph Cliff is constrained by its stratigraphic position between the 29.20 Ma Lund and the 29.0 Ma Ripgut (Table 1).

Preliminary correlation of the Petroglyph Cliff Ignimbrite in the field is based on its unique physical appearance but is confirmed by chemical composition, stratigraphic position, and paleomagnetic direction. In the Ely Springs Range and at its White Rock Spring type section ∼6 km north of the White River Narrows (Fig. 57; see also Supplemental File 5 [see ^{footnote 5}]; du Bray and Hurtubise, 1994) it is a simple cooling unit. At the Petroglyph Cliff locale at the north end of the narrows, the ignimbrite lies below ground level and the petroglyph carvings are in the overlying 27.57 Ma Monotony Tuff derived from the Central Nevada caldera complex (Fig. 2). Just north of the petroglyphs and between the southward-dipping Monotony and Lund tuffs is a 3-m-thick bed of well-sorted, fine, dark-gray semi-consolidated ash of the Isom type. How this apparently correlative ash-fall deposit relates to the thicker (15 m) Petroglyph Cliff Ignimbrite in the same stratigraphic position 4 km to the north is not clear because of discontinuous exposures.

The Petroglyph Cliff Ignimbrite has been found throughout the North Pahroc Range where it is overlain by another Isom-type cooling unit called the upper member of the Petroglyph Cliff Ignimbrite by Scott et al. (1992, 1994, 1995a, 1995b). However, because this younger unit has a distinctly different paleomagnetic direction, is not everywhere petrographically like the Petroglyph Cliff, and is separated from it by bedded ash or by a more widespread andesitic lava flow, we do not include it as a part of the Petroglyph Cliff Ignimbrite.

The Petroglyph Cliff has a markedly lobate distribution north and southwest of the source caldera (Fig. 57); it is not found to the east across Lake Valley in the Wilson Creek Range nor to the west in the Bristol Well quadrangle (Page and Ekren, 1995). Isom-type tuffs in the latter area probably correlate with the Bald Hills Tuff Member of the Isom Formation, according to our chemical data. In the Coyote Spring quadrangle to the west, Ekren and Page (1995) mapped a compound cooling unit of what they designated Petroglyph Cliff Ignimbrite. However, the paleomagnetic direction of this unit (site 4L073, S. Gromme and M. Hudson, 2006, personal commun.) differs from the Petroglyph Cliff, as does the lithology.

The igneous center at Blind Mountain on the southwestern flank of the Bristol Range (Figs. 57 and 58) is believed to be the source of the Petroglyph Cliff Ignimbrite. Page and Ekren (1995) recognized that the dioritic intrusions in the center have modal compositions like Isom-type tuffs and considered it to be a local source for some of them. A K-Ar age on hornblende (plus minor biotite and chlorite) from an associated granite is 28.4 ± 0.9 Ma (Armstrong, 1970; corrected for new decay constants), which is analytically the same as the Petroglyph Cliff. However, Johnson (1972) obtained a (corrected) age on hornblende of 35.8 ± 3.3 Ma but this age is significantly older than ages on nearby intrusions and lava flows (Best et al., 1989b, their figure 3) and we consider it spurious.

Detailed mapping by G.J. Axen (2010, personal commun.) has disclosed that the igneous center surrounded by Cambrian sedimentary rocks is a semi-elliptical segment of a caldera exposed for ∼2 km in longest dimension along the range front (Fig. 58). Inside this Blind Mountain caldera, a mass of variably metamorphosed Devonian limestone is intruded by granite, quartz diorite, and porphyritic plagioclase-pyroxene andesite; one mass of andesite forms a ring dike adjacent to the caldera margin. Lenses of steeply dipping, foliated Petroglyph Cliff Ignimbrite, which contain an abundance of the distinctive black blobs (Fig. 55), trend parallel to the long axis of the exposed caldera and likely represent feeder vents; they are no more than 170 m thick.

Using the isopach map of Figure 57 and Model 1 of Figure 4, the volume of the Petroglyph Cliff Ignimbrite is estimated to be 40 km³.

Mackin (1960) realized that some dark-colored, phenocryst-poor rocks that occur as widespread and generally thin lava-like deposits in the eastern Great Basin and High Plateaus of south-central Utah (Fig. 59) are ash-flow tuffs, not lava flows as believed by earlier workers. He named these densely welded cooling units the Isom Formation. The formation is the basis of the designation “Isom-type tuffs” indicated above. Anderson et al. (1975) recognized three tuff members of the formation in the High Plateaus; in ascending stratigraphic order they are the Blue Meadows, the Bald Hills, and the Hole-in-the-Wall. The Blue Meadows (Fryman, 1987) is found only in the Markagunt Plateau in central Utah and is not considered further here. The other members are more widespread regional ignimbrites, especially the Bald Hills. In our mapping we have not distinguished between the Bald Hills and the Hole-in-the-Wall because of the similar modal composition and characteristics of their constituent cooling units in the field, except where other rock units intervene (e.g., in the Rose Valley area; Best and Williams, 1997) or where paleomagnetic data are available to distinguish the reversely magnetized Hole-in-the-Wall from the normally magnetized Bald Hills. In Condor Canyon, just southwest of the probable source area of the Isom Formation, an additional sequence of normally magnetized cooling units of the Isom type has been found lying between the 27.90–27.25 Ma Bald Hills and the 24.5 Ma Hole-in-the-Wall Tuff Members (Table 1). This proximal sequence (Fig. 60), referred to as the tuff member of Hamlight Canyon by Scott et al. (1995a), but here formalized as the Hamlight Tuff Member, is also petrographically indistinguishable from the other two members of the Isom Formation; absent stratigraphic control, only paleomagnetic or chemical analyses distinguish the Hamlight. The four Hamlight cooling units are presumed to have also originated from the same source as the rest of the formation (Fig. 59; see also below).

The generally cliff- or ridge-forming cooling units of the Isom Formation are relatively thin, typically less than 10 m to as thick as 20 m, densely welded, and usually occur in dark shades of purple, red, and brown. A black, near-basal vitrophyre a few meters thick commonly underlies the devitrified facies.

Most outcrops contain lighter-colored lenses as much as a meter in diameter but usually less than 2 cm thick whose parallelism imparts a foliation to the tuff. Some of these lenses may be compacted pumice that are devitrified and replaced by vapor-phase minerals. But others, called “lenticules” by Mackin (1960) and Anderson and Rowley (2002), are believed to be devitrified gas-rich portions of the ignimbrite that separated and buoyed intervening less-gaseous layers during emplacement of the relatively thin ignimbrites. According to these workers, these gassy zones facilitated laminar flow, allowing the mass to travel great distances. Chapin and Lowell (1979) came to the same conclusion in their study of lenticles in the Wall Mountain Tuff of south-central Colorado.

A summary of the stratigraphy and occurrence of the Isom units, based primarily on their paleomagnetic directions as known in the early 1990s, is provided by Scott et al. (1995a). Here, we add further pertinent details. For reasons that will become apparent, we discuss the members of the Isom Formation from youngest to oldest.

This reversely magnetized ignimbrite was emplaced at 24.55 ± 0.12 Ma as a single simple cooling unit. Thickest sections (50–150 m) of the member occur west and southeast of the Escalante Desert. Because of its similarity to other Isom cooling units in the field, it has likely been included with the Bald Hills Tuff Member in many places and has been verified by stratigraphic position and paleomagnetic and/or chemical analyses at only eleven sites (Fig. 59A). Hence, its true areal extent and volume estimated by Model 1 (Fig. 4) are unknown but were at least 5900 km² and 600 km³, respectively.

In Condor Canyon, this 130-m-thick middle member of the Isom Formation comprises four cooling units that are superposed directly atop one another; they underlie the Hole-in-the-Wall Tuff Member and are underlain by local andesitic lava flows (Fig. 60; Supplemental File 5 [see ^{footnote 5}]). ⁴⁰Ar/³⁹Ar ages on the cooling units range from 24.91 to 24.75 Ma (Table 1), indicating emplacement over a relatively brief time period of 160,000 years. Only a slightly longer time period separates the youngest Hamlight from the Hole-in-the-Wall Tuff Member.

The only other known occurrence of the Hamlight Tuff Member is in the North Pahroc Range (Supplemental File 5 [see ^{footnote 5}]; Scott et al., 1992) where two Isom-type cooling units with a total thickness of 30 m lie in the same stratigraphic interval and have the same paleomagnetic direction as the Hamlight cooling units in Condor Canyon.

This is the oldest and by far the most extensive member of the Isom Formation (Fig. 59). It consists of at least four cooling units, and perhaps more in some places near the source; they are everywhere superposed directly atop one another without intervening deposits. Distal occurrences are of a single cooling unit. In addition to age and stratigraphic position, the Bald Hills is chemically (e.g., Fig. 6C) and paleomagnetically distinct from younger members.

⁴⁰Ar/³⁹Ar analyses of plagioclase from the three exposed Bald Hills cooling units in Condor Canyon yield a weighted mean age of 27.90 ± 0.09 Ma for the lowest unit and 27.25 ± 0.09 Ma for the uppermost unit; analyses of the middle unit yielded stratigraphically inconsistent ages. Samples from two other locales have intermediate and indistinguishable ages between the lowest and uppermost units in Condor Canyon, indicating they might be the same cooling unit: A single distal cooling unit at the south end of the Fortification Range has an age of 27.59 ± 0.15 Ma. The lowest of three cooling units in a proximal section 148 m thick east of the source area has an age of 27.60 ± 0.14 Ma. Based on the dated units in the two sequences of three units, it appears, therefore, that the Bald Hills consists of at least four cooling units, the youngest and oldest dated ones being exposed in Condor Canyon. These four cooling units were emplaced over a time interval of 0.65 m.y., longer than the duration of emplacement of the four Hamlight cooling units. There was a lull in eruptions in the Indian Peak–Caliente field of 2.34 m.y. between the Bald Hills activity and the much less voluminous Hamlight and Hole-in-the-Wall eruptions that together lasted 0.36 m.y. (Table 1).

Paleomagnetic data and stratigraphic relations of dated units indicate that the westernmost outcrop of the extensive Bald Hills units is at Hancock Summit in eastern Nevada (Fig. 59B) where it is exposed as a 6-m-thick simple cooling unit of black vitrophyre lying between the Monotony Tuff (27.57 Ma) and the Lower Tuff Member of the Shingle Pass Formation (26.98 Ma), both of which had sources in the Central Nevada caldera complex (Fig. 2). Additional exposures of a thin simple cooling unit of Isom-type tuff have been found for ∼100 km northward at nearly the same longitude. The most easterly exposures of the Bald Hills are in the High Plateaus in south-central Utah (Kurlich and Anderson, 1997; Anderson et al., 1990), making the present east-to-west extent ∼250 km, or 170 km after correction for post-deposition extension. The most northerly certain occurrence of the Bald Hills is a single cooling unit 5 m thick at the south end of the Fortification Range (Loucks et al., 1989); its age of 27.59 Ma was indicated above. Southernmost (<37.5° N latitude) exposures of the Isom Formation on either side of the Utah-Nevada state line (e.g., Hintze et al., 1994a) are not distinguished on a member basis so the true, present north-to-south extent of the Bald Hills Tuff Member is uncertain but could be ∼170 km. Its extension-corrected area of exposure is 21,000 km².

Anderson et al. (1975, p. 18; see also Best et al., 1989a) suggested that the source of the Isom Formation lay to the southeast of the Indian Peak caldera complex in the Escalante Desert where thickest sections occur in surrounding hills (Fig. 59B). The aggregate thickness of the Bald Hills and Hole-in-the-Wall Tuff Members is 365 m to the southeast of the desert and of all three members of the formation is ∼480 m to the west in Condor Canyon; both sites lie outside of the Indian Peak caldera complex delineated in Figure 8 and do not, therefore, represent unusual thicknesses ponded within older caldera depressions. However, thickness of 350–800 m occur as caldera fill within the older White Rock caldera to the northwest of the desert whereas regional thicknesses of 100 m or so occur just beyond the caldera margin.

Cuttings from three drill holes in the Escalante Desert offer only inconclusive support for the presence of a buried source caldera. A well just south of the Beryl railroad siding reveals only several meters of Isom beneath hundreds of meters of the Bauers ignimbrite (Table 1) and above several hundred meters of a sequence of andesitic and carbonate rock before bottoming at a depth of 1862 m. A well 8 km east of Table Butte passed through the Leach Canyon ignimbrite from ∼1384 to 1686 m depth and then through a few tens of meters of siltstone before bottoming in the Isom at a depth of 1750 m. A 5644-m-deep well 3 km northwest of Table Butte discloses no Isom tuff between 200 m of the Leach Canyon and hundreds of meters of underlying Lund, Wah Wah Springs, and Lamerdorf tuffs. Thousands of meters of Isom tuff was anticipated in these wells, if indeed the source caldera is concealed beneath the desert, but was not found. However, there is the possibility that the sequence of andesitic and carbonate rock in the Beryl well and the Lund, Wah Wah Springs, and Lamerdorf tuffs found in the well northwest of Table Butte actually represent wall-collapse breccias in a concealed Isom caldera.

We cannot rule out a source to the west of the Escalante Desert in an area where the only exposed rocks are post-Isom lavas several hundreds of meters thick (Fig. 8A; Best, 1987) and where no evidence for a fault-bounded caldera exists. On the northeastern margin of this area, in the southern Indian Peak Range (Best et al., 1987a), the map relations described above for the southeastern margin of the White Rock caldera could equally be considered as evidence for the northeastern margin of a source caldera for the Isom Formation; especially noteworthy in this regard is the drastic thinning of the Isom tuff toward the northeast.

On the basis of the equilibration pressure of the phenocryst assemblage, Isom-type magmas appear to have erupted from a depth of ∼30 km (see below), which is at least three times deeper than that from which the rhyolite and dacite caldera-forming magmas erupted in the Great Basin. With the magma erupting from a chamber beneath a 30-km-thick roof and the chamber itself of perhaps a similar horizontal diameter, it might be thought that a fault-bounded depression would not have developed; instead, the surface might only have sagged downward as the magmas were withdrawn. However, in the model experiments of Roche et al. (2000) fault-bounded calderas develop for roof aspect ratios (thickness / width) to as much as 4.5. Surface subsidence diminishes relative to the magma drawdown as the ratio increases; but for an aspect ratio of 1, which might be approximately the case for the Isom caldera, the subsidence and drawdown are about the same. Gravity data (Fig. 8C) provide no constraint on the location of an Isom caldera.

The combined volume of all of the outflow cooling units of the Isom Formation is estimated at 2100 km³. Doubling this outflow volume according to Model 1 in Figure 4 gives a total volume for the formation of 4200 km³ (Table 1). At least four cooling units of the Bald Hills Tuff Member together had a volume of ∼3600 km³. Whether one or more of these four cooling units resulted from super-eruption of at least 1000 km³ cannot be answered without further study.

The modal composition of each of the three members appears to be uniform and similar to the other two members (Fig. 61). The only difference might be a tendency for the Hole-in-the-Wall to have a little less pyroxene than the other two older members. The lowest Hamlight cooling unit contains a few sanidines that do not appear to be xenocrystic, based on their stratigraphically consistent ⁴⁰Ar/³⁹Ar age (Table 1). Densely welded, near-basal vitrophyres of three Bald Hills cooling units sampled in the White Rock Mountains show an upward increase in the concentration of total phenocrysts (5.0% to 8.4% to 14.3%) but the proportions of constituent phenocrysts reveal no consistent variation.

In contrast to the apparent uniformity in modal composition of the three members of the formation, their bulk chemical compositions are distinctly different in some variation diagrams (Figs. 6C and 54). Especially noteworthy are the distinctly lower concentrations of TiO₂, Fe₂O₃, and Sr and higher SiO₂ in samples of the more evolved Hole-in-the-Wall, all of which are low-silica rhyolite, in contrast to samples of the other two older members that are less evolved and nearly all trachydacite. The Hole-in-the-Wall as well as the Hamlight have less Zr than most of the Bald Hills samples. The Hamlight samples mostly overlap the Bald Hills but lie at their high end for TiO₂, SiO₂, Fe₂O₃, CaO, and Sr. There is an increase in Al₂O₃/CaO ratios from the Bald Hills and Hamlight to the more evolved Hole-in-the-Wall, which is consistent with fractional crystallization of a less evolved Bald Hills–type magma to the Hole-in-the-Wall, but the 2.7 m.y. age difference precludes such a simple interpretation; a closed magma body would not persist that long. Moreover, the Hamlight magma that is only slightly older than the Hole-in-the-Wall does not have the proper composition to be parental to it because, for example, Zr concentrations are too low and TiO₂ is too high in the Hamlight (Fig. 6).

Relative to rocks of comparable silica concentration (mostly 66–69 wt%), alkali-calcic trachydacitic Isom-type tuffs have higher Al₂O₃, TiO₂, K₂O, P₂O₅, Ba, Rb, Ce, Zn, Zr, Y, and Th and generally higher Nb, Y, and U. On the other hand, they have distinctly lower concentrations of many compatible constituents such as CaO, MgO, Sr, Ni, Cr, and V.

Isom magmas were less crystallized, drier, and hotter compared to most rhyolite and dacite magmas in the Great Basin as indicated by the sparse phenocrysts that include pyroxenes rather than the more common hydrous assemblage of biotite and hornblende. Phase equilibria calculated using MELTS (Ghiorso and Sack, 1995) also indicate a greater depth of phenocryst equilibration. An average Isom-type composition at ∼8 kb (∼30 km depth), 950 °C, oxygen fugacity fixed at QFM + 1, and 2 wt% H₂O yields the phase assemblage seen in the Isom-type ignimbrites, i.e., clinopyroxene, orthopyroxene, plagioclase, and magnetite. Lower-pressure assemblages lack clinopyroxene and may have quartz or sanidine. Fractionation of this high-pressure phase assemblage develops high concentrations of many incompatible elements, including K₂O, without strong enrichment in SiO₂.

The intimate spatial and temporal association of Isom-type tuffs in some near-source occurrences with high-K pyroxene andesite lava flows, some of which also have high Zr concentrations, indicate a kindred relationship. We envisage fractionation of pyroxenes, plagioclase, and minor Fe-Ti oxides and apatite from andesitic parent magmas at depths of near 30 km in the crust to yield the near-liquidus Isom-type magmas. Slightly different parental magmas, environments of differentiation, and possible extent of contamination with crustal wall rock led to different erupted compositions.

Immediately following deposition of the Lund, which is the youngest monotonous intermediate in the Indian Peak field, the small-volume Petroglyph Cliff Ignimbrite was erupted from its Blind Mountain source caldera just west of the northwestern margin of the Indian Peak caldera complex. A little more than one million years later, voluminous eruptions of the Isom Formation cooling units began from a concealed source on the opposite, or southeastern, side of the caldera complex. These space-time relations and ⁸⁷Sr/⁸⁶Sr ratios suggest the following scenario: With continued influx of mafic mantle-derived magmas into the crust, less additional silicic magma could be created by partial melting because felsic components in the now less-fertile magma source volume had been largely extracted during generation of the preceding three super-eruptive monotonous intermediate magmas (Cottonwood Wash, Wah Wah Springs, Lund). Instead, accumulating masses of mantle-derived magmas evolved into andesitic derivatives that upon further fractional crystallization created the near-liquidus trachydacitic Isom-type magmas, into which little or no sialic crustal components could be mixed or assimilated. As a result, initial ⁸⁷Sr/⁸⁶Sr ratios are distinctly lower in both the Petroglyph Cliff (0.7084) and the Isom (Bald Hills Member, 0.7077) than any of the preceding intermediate to felsic magmas erupted in the field (Fig. 6D). The Isom-type magmas could not ascend buoyantly nor by diking through the residual magmas under the Indian Peak caldera complex but were able to erupt just beyond on the northwest and southeast margins of the complex.

Following voluminous eruptions of monotonous intermediate and subordinate rhyolite magmas from the multicyclic calderas of the Indian Peak complex, generation of rhyolite magmas shifted southward tens of kilometers while the Isom-type magmas were being erupted from 27.90 to 24.55 Ma. The new focus of eruptive activity after ca. 24 Ma lay in the multicyclic Caliente caldera complex where overlapping, or at least nearby, calderas developed (Fig. 2). Over the next 5.5 m.y., five significant eruptions of rhyolite magma took place, creating, from oldest to youngest, the Leach Canyon, Swett, Bauers, Racer Canyon, and Hiko ignimbrites (Table 1). An additional eruption of the unique andesite-latite Harmony Hills ignimbrite occurred at 22.56 Ma. Four of these six eruptions were of super magnitude (>1000 km³), including the Harmony Hills. The dominance of rhyolite over dacitic ignimbrites in the Caliente complex represents an inversion of what had taken place before 28 Ma in the Indian Peak caldera complex when and where monotonous intermediates dominated eruptive activity.

Another distinctive compositional difference between the volcanic rocks of the two caldera complexes is found in their ⁸⁷Sr/⁸⁶Sr ratios; for a given silica content, the initial Sr isotope ratios of rocks erupted from the Caliente complex are lower than those from the Indian Peak complex. For example, rhyolites from the Caliente complex have Sr isotope ratios that are lower than 0.7085 whereas all rhyolites from Indian Peak are higher than this, ranging to as high as 0.712 (Table 4, Fig. 6D). This difference implies less crustal contamination of magmas in the Caliente complex or a contrast in the source and composition of the mantle magma component.

It might be expected that some degree of inheritance or commonality among the chemical compositions of the rhyolite ignimbrites would be evident because of their eruption from a focused magma system in the crust. However, representative analyses (Fig. 62B) reveal that the five rhyolite ignimbrites evolved along distinctly separate but essentially parallel trends, lacking any simple age progression.

Williams (1967, p. 118) concluded that the Leach Canyon, Swett, and Bauers ignimbrites, and possibly the Hiko Tuff, originated in “a large caldera-like igneous complex centering about the town of Caliente.” He based this conclusion chiefly on the fact that the thickest sections of these tuffs surround this area. Reconnaissance work by Noble et al. (1968) and Noble and McKee (1972) further targeted the Caliente topographic depression as an ignimbrite source. Ekren et al. (1977) produced the first geologic map of the Caliente caldera complex and, on the basis of geologic, gravity, and aeromagnetic data, expanded its extent to the east and west and shifted its center southward. Larger-scale mapping by Peter Rowley and associates, summarized in Rowley et al. (1995), extended the margin of the complex still farther south as well as east into Utah (Fig. 2). They recognized the Clover Creek caldera source of the Bauers and the Delamar caldera source of the Hiko in the complex. Rowley et al. (2008) later identified the Telegraph Draw caldera source for the Racer Canyon in the complex. No geologic evidence has been found for caldera sources of the other ignimbrites, but their locations can be roughly approximated on the basis of the distribution and thickness of the outflow deposits.

According to Rowley et al. (1995, 2001), the 80 × 35 km Caliente caldera complex evolved from 23 to 13 Ma during regional tectonic extension in the area. East-striking syn-volcanic fracture systems, or transverse zones, bound the margins of the elongate caldera complex.

Local ash-flow tuff and local tuff breccia as thick as 500 m of the Rencher Formation are exposed in the Bull Valley and Pine Valley Mountains in the southwestern corner of Utah (Cook, 1957; Blank, 1959, 1993; Blank et al., 1992; Rowley et al., 2006; Biek et al., 2009). This unit has an age of 22.2 Ma, placing it between the Harmony Hills and the Racer Canyon Tuffs (Table 1; the Rencher is not listed in this table). The Rencher is chemically an andesite-dacite (60–64 wt% silica) and lacks quartz and sanidine, but is otherwise similar to the phenocryst-rich Harmony Hills. The Rencher is considered to be a small-volume, local eruptive facies of the Bull Valley intrusion, one of several shallow monzonitic intrusions along the “Iron Axis” trending west-southwest of Cedar City that have associated iron ore deposits (Biek et al., 2009). The Rencher is not considered further here.

Ignimbrite of the 24.03 ± 0.01 Ma Leach Canyon Formation is rather ordinary rhyolite that consists of slightly more phenocrysts of plagioclase than subequal amounts of sanidine and quartz; lesser biotite, hornblende, and Fe-Ti oxides; and trace amounts of pyroxene, titanite, zircon, and apatite (Figs. 62 and 63). Conspicuous lapilli of pumice are typical.

The Leach Canyon was subdivided by Williams (1967; see also Mackin, 1960; Cook, 1965; Rowley et al., 1995) into two units, an older Narrows Tuff Member and an overlying Table Butte Tuff Member. No definite cooling break between the two members was indicated by Williams (1967), an aspect observed by us in Condor Canyon (Fig. 60). In quadrangles northwest of Cedar City, Rowley (e.g., 1976) distinguished the upper Table Butte as a less resistant tuff that contains as much as 10%–15% dark red and purple aphanitic volcanic fragments underlain by the more resistant, moderately welded Narrows that contains less than 2% of the same type of lithic fragments. Although not optimal magnetic recorders, the two members have similar paleomagnetic directions. The two members appear to be a product of a more or less continuous eruptive event during which, possibly as a result of vent collapse, the later-erupted ejecta entrained more wall-rock fragments. Although the proportions of phenocrysts in the two members are quite similar, the older Narrows tuff tends to have a lower proportion of plagioclase and biotite and more sanidine and quartz than the Table Butte (Fig. 63); hence, considered together, the two members exhibit normal zoning.

Based on the general pattern of southward migration of the inception of volcanism through time in the region, the source for the Leach Canyon likely lies south of the Indian Peak caldera complex. However, no evidence for a fault-bounded caldera created upon super-eruption of the exposed 1800 km³ of outflow ejecta has been found. Compilation of thicknesses of the Leach Canyon by Williams (1967, his figure 9) clearly targeted a source around Caliente and our updated compilation does the same (Fig. 64). Younger lava flows and sedimentary deposits in the Panaca Basin (Rowley and Shroba, 1991) conceal any direct evidence of a source in this area. Outcrops of the upper Table Butte member near the northeast margin of the basin near Ursine reveal the “largest known lithic fragments” (Williams, 1967, p. 116); the Narrows member is most densely welded in Utah where a near-basal vitrophyre occurs in some places. South of Caliente, younger calderas in the Caliente caldera complex have likely buried at least the southern part of the source caldera of the Leach Canyon ignimbrite.

The western distribution of the tuff beyond its presumed source area was influenced by piles of andesitic lava flows. In the Deadman Spring NE quadrangle (Swadley et al., 1994) at 37°59′ N and 114°52′30″ W, the Leach Canyon is absent and only several meters of the older Isom and Petroglyph Cliff are draped on top of a stack of lava flows as thick as 600 m. The Leach Canyon is also absent to the south in the Pahroc Spring SE quadrangle (Swadley and Rowley, 1994) at 37°32′ N and 114°47′ W where a cluster of andesitic-dacitic stratovolcanoes lie between the Lower Tuff Member of the 26.98 Ma Shingle Pass Formation and the Swett Tuff Member of the Condor Canyon Formation (Table 1); thus, they essentially bracket the time of eruption of the Leach Canyon ignimbrite. It is possible that these stratovolcanoes of about the age of the Leach Canyon lie on the west margin of the source caldera.

Taking into account an assumed equivalent volume of ignimbrite in the concealed source caldera as occurs as outflow (Model 1 in Fig. 4), the total volume of the Leach Canyon is 3600 km³.

The Condor Canyon Formation (Table 1; Mackin, 1960; Cook, 1965; Williams, 1967; Rowley et al., 1995) consists of two similar, low-silica, off-trend rhyolite ignimbrites (Fig. 62), the older Swett Tuff Member and the order-of-magnitude-larger Bauers Tuff Member. They have relatively high concentrations of alkalies and Zr but low CaO and Sc (Figs. 5, 6, and 62). In some element variation diagrams, Condor Canyon samples are relatively tightly clustered and lie apart from other main-trend ignimbrites of the Indian Peak–Caliente field. Each member contains 8%–23% phenocrysts, mostly plagioclase and lesser biotite, but no quartz (Fig. 65); the Bauers also has abundant sanidine and trace amounts of clinopyroxene. Both members contain moderate to abundant lapilli and blocks of pumice. Chemical and modal compositions are reminiscent of low-silica rhyolite tuffs (Lamerdorf and Ryan Spring) erupted from the Indian Peak caldera complex except for the general absence of lithic clasts in the Condor Canyon tuffs.

Condor Canyon cooling units are commonly relatively thin and moderately to densely welded, and thus commonly crop out as ledges. Thicker, proximal outflow sections are simple cooling units and display conspicuous secondary zonation (Williams, 1967). A black, near-basal vitrophyre a few meters thick is locally present and grades into an overlying zone of densely welded, devitrified tuff that is shades of brown, red, and purple. The gradation is manifest by an upward increase in the concentration of red-brown spherulites as much as several centimeters in diameter. This zone of gradational devitrification is overlain, in places with a rather sharp contact, by a strongly foliated zone (Fig. 66) that contains lighter-colored material in disc shapes; these discs are ∼5 cm thick and locally as much as 2 m in diameter and in the Bauers compose upwards of one-half of the cooling unit. Some of these discs could be flattened collapsed pumice clasts that have devitrified and been replaced to varying degrees by vapor-phase alkali feldspar, cristobalite, and tridymite, but others could be lenticules as described above for the Isom Formation. The strongly foliated zone grades upward within several centimeters into a zone of massive tuff lacking the discs whose matrix is rich in vapor-phase crystals. In thinner, more distal stratigraphic sections, these zones are not as well developed and the foliated zone is absent.

Though compositionally similar, ⁴⁰Ar/³⁹Ar ages indicate the two members are more than a million years apart (Table 1). The Bauers has a weighted mean age on sanidine of 23.04 ± 0.11 Ma. Rowley et al. (1994) reported a sanidine age by L.W. Snee of 23.13 ± 0.1 Ma on the intracaldera Bauers intrusion (see below). Duplicate analyses of plagioclase from one sample of the Swett from Condor Canyon yielded a weighted mean age of 24.15 ± 0.10 Ma. The Swett is very close in age to the underlying Leach Canyon at 24.03 ± 0.01 Ma.

The distribution of the Swett is quite irregular in the northeast (Fig. 67A). This unusual outcrop pattern might reflect the general thinness of the unit and some erosion after its emplacement. At least some of this pattern could be the result of topographic control at the time of deposition by northeast-southwest–trending paleovalleys carved into the older Leach Canyon tuff. The southeastern paleovalley could have developed on thicker parts of the Leach Canyon tuff (Fig. 64) that experienced relatively greater compaction. Control by faulting cannot be ruled out.

The volume of the outflow Swett is estimated at 200 km³ and the total volume of the unit including that assumed to be hidden in the concealed caldera is 400 km³.

There is no direct evidence for a source caldera of the Swett; it was either largely engulfed in the Bauers source caldera (see below) or is buried to the north of Caliente in the Panaca Basin that is filled with late Miocene–Pliocene sedimentary deposits. A possible intracaldera sequence more than 200 m thick of volcanic and minor sedimentary units that underlies the Bauers Tuff Member is exposed in the southwest corner of the Panaca Summit quadrangle (Williams et al., 1997) at 37°47′ N and 114°13′45″ W. This sequence, whose base is not exposed, consists of ∼100 m of intermediate-composition lava flows, beds of intervening sedimentary rock, and lithic- and pumice-rich ignimbrite cooling units, one of which has the paleomagnetic direction of the Swett (site 2P130, S. Gromme and M. Hudson, 2006, personal commun.). This sequence has not been seen anywhere else between the Leach Canyon and the Bauers where the Swett normally lies.

The Bauers ignimbrite contains conspicuous phenocrysts of sanidine in addition to slightly greater amounts of plagioclase and much less biotite, Fe-Ti oxides, and trace amounts of clinopyroxene (Fig. 65). The sparse feldspar phenocrysts in the Bauers are conspicuously euhedral (Best and Christiansen, 1997, their figure 1) relative to other crystal-poor ignimbrites in the Great Basin. Aggregates of plagioclase and smaller mafic phenocrysts are common.

Rowley et al. (1995) indicated that the Bauers outflow deposit (Fig. 67B) is reversely zoned, which we confirmed. TiO₂ and Fe₂O₃ decline from the bottom to the top of the unit (Fig. 62B). In three stratigraphic sections there is an upward decrease in the plagioclase-to-sanidine ratio, from an average of 2.0 to 1.4, and the amount of mafic phenocrysts diminishes somewhat upwards as well.

A small segment of the Clover Creek source caldera of the Bauers lies near Caliente in the northern sector of the Caliente caldera complex (Figs. 2 and 67B; Rowley et al., 1994). The intracaldera tuff is a compound cooling unit more than 400 m thick that contains as much as 20% pumice clasts as long as 20 cm and as much as 30% angular, mostly cognate volcanic rock fragments as much as 12 cm across. The intracaldera tuff is intruded by a flow-foliated, hypabyssal intrusion of similar composition and overlain by a sequence more than 150 m thick of tuffs and volcanic debris flows. An east-west–trending caldera margin is speculated to lie concealed beneath younger deposits to the northeast and west of the intracaldera rocks (Rowley et al., 1992).

Of the ignimbrites derived from the Caliente caldera complex, the Bauers has the largest outflow distribution at 23,000 km² (Table 1 and Fig. 67B). But throughout most of this area it is no more than 50 m thick. In only three sites is the outflow as much as 200 m thick north and east of the caldera source. There is a hint of east-trending paleovalleys in the distal eastern part of the outflow sheet; paleotopography in the Condor Canyon area also influenced its thickness.

Total volume of the Bauers ignimbrite, including assumed equivalent amounts as outflow and within the caldera, is 3200 km³, second only to the Leach Canyon in volume among the ignimbrites deposited after 24 Ma.

The Harmony Hills Tuff (Mackin, 1960; Cook, 1965; Williams, 1967; Rowley et al., 1995) generally occurs as a simple cooling unit throughout most of its 13,000 km² extent.

In the middle Cenozoic southern Great Basin ignimbrite province, the Harmony Hills is the most mafic and poorest in silica of major regional ignimbrites. Six available analyses are relatively tightly clustered near the junction of the dacite, latite, and andesite fields in Figure 49, but lie mostly in the latter two. Analyses overlap with the least-evolved samples of the Cottonwood Wash Tuff (Fig. 27A). The Harmony Hills contains relatively high concentrations of MgO, CaO, Fe₂O₃, Sr, and Cr (Figs. 5, 6 and 30C). Thirty modes indicate the ignimbrite is exceptionally phenocryst rich, with as much as 58% phenocrysts, including abundant plagioclase, lesser biotite, hornblende, quartz, clinopyroxene, and Fe-Ti oxides, and a trace of sanidine (Figs. 68 and 69).

We have no data on cognate inclusions nor on possible systematic zoning in the outflow sheet. Whether it should be classed as an unusually mafic monotonous intermediate remains to be evaluated.

Duplicate ⁴⁰Ar/³⁹Ar ages on plagioclase from a sample in Condor Canyon yield a weighted mean age of 22.56 ± 0.11 Ma.

No direct evidence has been found for a source caldera of the Harmony Hills Tuff. Whether it lies in Utah or Nevada has been controversial for decades (e.g., Blank, 1959; Ekren et al., 1977; Scott et al., 1995a).

In the southwestern corner of Utah, in the eastern Bull Valley Mountains, proximity to the source is suggested by the presence of two cooling units separated by 0.5 m of finely bedded ash and sandstone. Northwest of the Bull Valley Mountains, Rowley et al. (2007) documented a thickness of at least 275 m of the Harmony Hills (Fig. 70) and an unusual abundance (as much as 20% of the rock) of collapsed pumice clasts as long as 0.3 m. To the southwest, Anderson and Hintze (1993) also reported a near-source character for the Harmony Hills Tuff, including an unusual thickness of at least 330 m, the presence of two cooling units, zones rich in large cognate pumice clasts (to as much as 0.3 m), and as much as 5% inclusions of andesite lava rock in the lower and upper parts of the unit.

In Nevada, in the southern Clover Mountains, Ekren et al. (1977) found more than 100 m of Harmony Hills Tuff that contains basketball-size pumice clasts, which diminish rapidly in size and abundance southward. In the southern part of the Delamar caldera source of the Hiko Tuff (see below), a deposit of Harmony Hills breccia that appears to have been derived from a thick unit at the time the caldera collapsed was mapped by Rowley et al. (1995). To the north of the Clover Mountains, Rowley et al. (1994) documented ∼170 m of Harmony Hills in the older Clover Creek caldera (Fig. 67B); this thickness may be the result of accumulation as caldera fill. Nearby exposures are less thick.

These apparent near-source occurrences of the Harmony Hills delineate an area within the Caliente caldera complex where the source caldera probably lies (Figs. 2 and 70). This integrated area reconciles the two conflicting proposed sources for the Harmony Hills near the Bull Valley Mountains in Utah and the southeastern part of the Caliente caldera in Nevada. The dimension of a caldera enclosed within the source area is not unreasonable, given that the volume of the outflow tuff is 1100 km³. Doubling this volume gives 2200 km³ for the approximate total volume of the unit.

The Racer Canyon and Hiko Tuffs (Blank, 1959; Rowley et al., 1979; Dolgoff, 1963) are products of eruptions from calderas on the east and west sides of the Caliente caldera complex resulting in dispersal of the ash flows chiefly, but not entirely, into Utah and into Nevada, respectively (Figs. 71 and 72).

In most places, the Hiko is a single simple cooling unit whereas the mostly older Racer Canyon locally consists of multiple cooling units.

At sites within a few tens of kilometers on either side of the Utah-Nevada state line the two units overlap. The unusually thick section of cooling units of rhyolite ignimbrites near Panaca Summit, originally mapped as all Racer Canyon Tuff (Williams et al., 1997), has now been divided into 390 m of the Hiko Tuff above a prominent layer of vitrophyre (paleomagnetic sample 0P284 in Gromme et al., 1997) and 485 m of the underlying Racer Canyon Tuff.

The section more than 150 m thick southwest of Modena, Utah (Best, 1987), may be all Racer Canyon whereas just to the south, Siders (1991) described a lower cooling unit less than 122 m thick of the Racer Canyon that is overlain by beds of fine sediment and then an upper thinner (<30 m) weakly welded tuff that has less quartz and sanidine than the average Racer Canyon; we consider this thinner tuff to be Hiko. Much farther south, Anderson and Hintze (1993) reported a section 230 m thick of Hiko but Rowley et al. (2007) instead correlated this section with the Racer Canyon. Still farther south, Hintze et al. (1994a) and Hintze and Axen (1995) documented 155 m and 50 m, respectively, of Hiko Tuff but admit the possibility that some or all is Racer Canyon. This latter correlation seems more reasonable because the source of the Racer Canyon lies just to the north and is closer than the source of the Hiko (see caldera sources below). To the east, sections are 90–335 m thick but lie geographically between sections 50 m thick or less. Siders et al. (1990) and Rowley et al. (1995) pointed out that the Racer Canyon was deposited on uneven terrain created by syn-eruptive faulting as well as by earlier shallow laccolith emplacement that created local topographic domes.

Although considerable work has been done on these two similar rhyolite ignimbrites (e.g., Rowley et al., 1995; Gromme et al., 1997), full understanding of their stratigraphic relations, distribution, and compositional variations is incomplete. Further paleomagnetic and chronologic analyses will be needed to distinguish them throughout their extent in the eastern Great Basin (Figs. 71 and 72).

The Racer Canyon Tuff is loosely to moderately welded, is tan to light gray, and consists in most places of multiple cooling units (for a thorough description of the Racer Canyon see Siders et al., 1990). Pumice and lithic clasts constitute 10%–20% of the tuff.

The generally more densely welded, gray to brown Hiko Tuff in most places outside its caldera source appears to be a single cooling unit except in the South Pahroc Range (Moring, 1987) where it consists of two cooling units totaling ∼300 m thick; the upper unit appears to be the most widespread and voluminous. The Hiko commonly weathers into bulbous granite-like outcrops (Fig. 73) controlled by joint sets. Locally, a vitrophyre occurs near the base of the unit as well as flattened pumices and sparse xenoliths of volcanic and sedimentary rock.

Available modal data do not unequivocally distinguish between the Racer Canyon and Hiko because of overlap in proportions of phenocrysts (Rowley et al., 1995, their figure 15; Gromme et al., 1997, their figure 3). But in an attempt to possibly clarify differences we have plotted Hiko modes from samples well into Nevada and Racer Canyon modes from samples well into Utah, both from Rowley et al. (1995); these selected samples avoid uncertain stratigraphic identities near the state line as just described. These modes (Fig. 74) show differences in the relative proportions of felsic phenocrysts. The younger Hiko has more plagioclase than quartz whereas in the Racer Canyon the proportions of these two phenocrysts are similar. Proportions of quartz and sanidine are similar in the Hiko whereas in most samples of the Racer Canyon quartz exceeds sanidine.

An additional challenge in the field is distinguishing between the Racer and the Leach Canyon ignimbrites, especially the lower, Narrows member of the latter because of nearly identical appearance and modal composition (Rowley et al., 1995, their figures 6 and 15); their modes include trace amounts of titanite, which is generally not present in most silicic Great Basin tuffs and, therefore, serves as a useful tool in distinguishing most stratigraphic units. However, the Leach Canyon is significantly older, lying below the Harmony Hills Tuff and Condor Canyon Formation, so that the Leach Canyon can usually be recognized as such on stratigraphic grounds, except in terranes of complex faulting or isolated outcrop.

Chemically, the Racer Canyon and Hiko are also very similar but they form distinct trends on the Fe₂O₃ versus TiO₂ diagram that we have used to distinguish many of the tuffs from the region (Fig. 62B). For a given TiO₂ content, the Racer Canyon Tuff has ∼0.2 wt% more Fe₂O₃ than the Hiko. Moreover, the Racer Canyon Tuff has the highest Fe/Ti ratios of the silicic units erupted from the Indian Peak or the Caliente caldera complexes. The Racer Canyon and Hiko Tuffs are rather different from the older tuffs of the Quichipa Group—the Leach, Swett, and Bauers. Most significantly, the younger tuffs are less alkaline (Fig. 62A). Most samples of the three older units also have significantly higher Al₂O₃/CaO ratios than the Racer Canyon and Hiko Tuffs. The thick section (>450 m) of the Racer Canyon south of the Enterprise Reservoir displays strong normal zoning (Fig. 62); the base of the unit is a high-silica rhyolite (77.6 wt%) whereas the top is a high-silica dacite (68.9 wt%). The Hiko is less strongly zoned, from ∼74 to 69 wt% silica.

⁴⁰Ar/³⁹Ar ages of sanidine and paleomagnetic data provide insight into the stratigraphic relations of the Racer Canyon and Hiko Tuffs.

In the South Pahroc Range, a slight difference in the paleomagnetic directions of the two Hiko cooling units was interpreted by Hudson et al. (1998) to represent an elapsed time of as little as one or two centuries between their emplacement.

A sample of the Hiko in Condor Canyon in eastern Nevada is 18.47 ± 0.04 Ma whereas a sample in the North Pahroc Range is 18.56 ± 0.04 Ma. These samples have a reverse paleomagnetic polarity (Gromme et al., 1997). Rowley et al. (1995) reported identical preliminary plateau ages by L.W. Snee on sanidine of 18.5 ± 0.1 Ma for Hiko ignimbrite and a presumed co-genetic rhyolite dome on the rim of the source caldera (see below).

In the section of twelve cooling units of the Racer Canyon south of the Enterprise Reservoir (northeast corner of Goldstrike quadrangle; Rowley et al., 2007), an upper unit is reversely magnetized and has an age of 18.57 ± 0.03 whereas the lowest exposed cooling unit is normally magnetized and has an age of 18.85 ± 0.03 Ma (Gromme et al., 1997). Rowley et al. (1995) reported a similar age of 19.0 ± 0.1 Ma by L.W. Snee from the same section. Normally magnetized Racer Canyon cropping out on the highway between Panaca and Modena near the state line has an age of 18.88 ± 0.06 Ma.

Hence, available data indicate the multiple cooling units of the Racer Canyon are mostly older than the Hiko, but the youngest is about the same age as the Hiko.

The paleomagnetic direction of a sample from the lowest part of the Racer Canyon at Panaca Summit is reversely magnetized whereas the immediately overlying part of the unit is normally magnetized. Thus, the sampled sequence of Racer Canyon cooling units reveal two paleomagnetic reversals, i.e., R → N → R.

The poorly exposed source of the Racer Canyon Tuff is designated the Telegraph Draw caldera by Rowley et al. (2008). Near 114° W, the northern sector of the caldera (Fig. 71) is expressed by intercalated wall-collapse breccias in ignimbrite that has pumice clasts to as much as 0.6 m in diameter. In the southeastern sector, about twelve cooling units of Racer Canyon Tuff totaling at least 450 m thick are exposed (Rowley et al., 2007); this thick section south of the Enterprise Reservoir and its abrupt southward thinning mark the southeastern margin of the caldera. A largely coincident younger caldera apparently combines with the Telegraph Draw caldera to produce a large negative gravity anomaly in the eastern part of the Caliente caldera complex (Rowley et al., 2008).

Ekren et al. (1977) discovered that the source caldera of the Hiko Tuff constitutes the western lobe of the Caliente caldera complex (Figs. 2 and 72). Rowley and Siders (1988) named this source the Delamar caldera. As documented by Swadley and Rowley (1994), the Delamar caldera is expressed by more than 400 m of relatively pumice-rich tuff, which in its upper part occurs in multiple cooling units that interfinger with lenses of caldera-collapse breccias, chiefly of andesitic flow rocks. A lava dome similar in composition to the Hiko Tuff lies on the topographic margin of the caldera. An eruptive vent 2 km in diameter for the tuff lies on the southwestern margin of the caldera; the vent is filled with mostly Hiko Tuff that possesses a subvertical foliation defined by flattened pumice clasts (Rowley et al., 1995, p. 60). The margin of the caldera consists of complex faults that are believed to have been active during its development.

The outflow Hiko occurs as a thick lobe extending for at least 50 km southwest of the Delamar caldera and a smaller lobe extending to the northeast (Fig. 72).

The Hiko Tuff qualifies as a super-eruption of 1700 km³ if its largely concealed intracaldera volume is assumed equivalent to the outflow volume (Table 1). The total volume of the Racer Canyon Tuff so determined is 1100 km³.

Multiple Racer Canyon eruptions that resulted in as many as twelve cooling units zoned from high-silica rhyolite to high-silica dacite contrasts with eruption of no more than two cooling units of the zoned, low-silica (74–69 wt%) rhyolite Hiko (Fig. 62). The mostly older Racer Canyon and the Hiko also exhibit inter-unit normal compositional zonation. Because the extension-corrected distance between their sources at opposite ends of the Caliente caldera complex was no more than 8–10 km (Figs. 71 and 72), it is possible that their two zoned magma chambers were connected; venting began at the east end, withdrawing evolved high-silica rhyolite magma from the top of the chamber, followed by deeper withdrawal of less-evolved magma at both ends.

The Racer Canyon and Hiko were the terminal large-scale eruptions of the long-lived 36 to 18 Ma Indian Peak–Caliente magma system. Resulting calc-alkaline arc rocks possess pronounced Nb-Ta-Ti anomalies on normalized diagrams and plot in the Pearce et al. (1984) continental arc field (Fig. 75). Thereafter, from ca. 16 to 12 Ma, mostly more-alkaline, within-plate magmas lacking strong arc signatures were explosively erupted from the Caliente magma system and the Kane Springs Wash system to the southwest (Nealey et al., 1995). The 15–14.2 Ma local tuff of Kershaw Canyon in the Caliente caldera complex has a transitional composition, but it is not considered here because of its younger age.

To the west, in the Central Nevada ignimbrite field, the high-silica, phenocryst-rich rhyolite Fraction Tuff is similar in age (18.57 Ma) and composition to the Racer Canyon and Hiko Tuffs, including the presence of titanite. It is also the terminal calc-alkaline, arc-type eruption in the field and contrasts with the younger more alkaline activity to the south in the Southwest Nevada volcanic field (Sawyer et al., 1994).

Rowley et al. (1995) reported that the small, weakly welded, high-silica rhyolite tuff of Teepee Rocks that overlies the Hiko Tuff has an age of 18.1 ± 0.7 Ma and may have been erupted from a source to the east of the Delamar caldera. It resembles the Hiko in many respects, but is more evolved chemically and modally, although the proportions of phenocrysts are highly variable (Fig. 74C). It may represent a very late differentiate of the residual Hiko magma. Chemically, it belongs with the group of 36–18 Ma arc-type ignimbrites (Fig. 75; Nealey et al., 1995).

Rhyolitic to andesitic lava flows were extruded from numerous vents from 34 to 18 Ma in the Indian Peak–Caliente field (Fig. 76) during the time of explosive silicic activity. Minor volumes of granitic magma were intruded during this activity. It is especially noteworthy that no true basalt magma (International Union of Geological Sciences classification of Le Maitre, 1989) was extruded in the Great Basin during the ignimbrite flareup until after ca. 20 Ma and even younger in southwest Utah.

The lavas are described in the following sections for three time periods: (1) 34–31 Ma, prior to eruption of the very large-volume monotonous intermediates in the Indian Peak caldera complex; (2) 31–26 Ma, contemporaneous with eruption of the monotonous intermediates and trachydacitic Isom-type tuffs; and (3) 25–18 Ma after eruptive activity in the Indian Peak caldera complex and during ignimbrite eruptions in the Caliente caldera complex.

Oligocene lavas of the first two time periods have been described by Best et al. (1989a), Askren (1992), and Askren et al. (1997) whereas Best et al. (1987c) have described the Miocene lavas in the third period. For major- and trace-element analyses of mafic lavas for all three time periods see Best et al. (2009).

Extrusion of andesite and lesser low-silica dacite lavas at 34–31 Ma (E in Fig. 77) around all but the south flank of the Indian Peak caldera complex (Fig. 76) began when the first regional rhyolite ignimbrites were emplaced from sources within the complex, i.e., the Sawtooth Peak and Marsden tuffs. The greatest volume of lavas comprises remnants of small andesite-dacite stratovolcanoes immediately to the west of the Indian Peak caldera complex near Silver King Mountain (Ekren et al., 1977) and 30–60 km to the northeast of the complex. Lavas near Silver King Mountain that underlie the 31.69 Ma Windous Butte Formation have not been mapped nor studied in detail; distal flows may extend southward into the North Pahroc Range (Scott et al., 1995b). Northeast of the caldera complex, 34 Ma lavas, debris flows, and breccias are 360–600 m thick (Best et al., 1989c; Hintze et al., 1984) and contain phenocrysts of plagioclase, hornblende, Fe-Ti oxides, and biotite and less-common quartz and clinopyroxene; lavas in the east are associated with Pb-Ag-Cu mineralization. Small intrusions of diorite exposed in the low pass across the Wah Wah Mountains northwest of Wah Wah Springs (Hintze, 1974) might represent intrusive counterparts of lavas to the east. Southwest of Milford, 32 Ma lavas are ∼400 m thick and contain phenocrysts of plagioclase, two pyroxenes, and Fe-Ti oxides (Best et al., 1989c).

Much smaller erosional remnants of the 32–31 Ma andesite lava flow formation of the Escalante Desert Group occur in three places in and around the Indian Peak caldera complex:

1. Near Atlanta, Nevada, immediately north of the northern margin of the complex flow remnants are less than 200 m thick (Willis et al., 1987); one flow has a stratigraphically constrained age between 31.1 and 31.7 Ma.

2. In the southern Wah Wah Mountains, duplicate ⁴⁰Ar/³⁹Ar analyses of plagioclase from sample LAM-1C collected at the top of the 100–200-m-thick flow between cooling units of the Lamerdorf Tuff (Table 1; Fig. 15) yielded a stratigraphically consistent weighted mean age of 31.98 ± 0.16 Ma.

3. In the southern White Rock Mountains, as much as 350 m of andesite lavas lie between the 33.5 Ma Sawtooth Peak and 31.13 Ma Cottonwood Wash ignimbrites in the resurgently uplifted floor of the caldera complex (Fig. 44; Keith et al., 1994).

Local volcanic debris flows tens of meters thick that contain clasts of andesitic rock occur in the Beers Spring Formation of the Escalante Desert Group and at the top of the Marsden sequence, as described above, in the central Needle Range; these debris flows testify of a nearby unexposed extrusive mass.

The rhyolite lava flow formation of the Escalante Desert Group is generally nonvesicular and is exposed over an area of only ∼10 km² mostly on the north and east sides of the Indian Peak caldera complex. Oldest flows are more or less contemporaneous with the Marsden Tuff. On the north side of the complex, west of Atlanta (Fig. 76), altered flow remnants were originally mapped as the 30 Ma Ryan Spring Formation (Willis et al., 1987) but subsequent fission-track dating of zircons revealed ages of 36–32 Ma (Kowallis and Best, 1990). On the east side of the caldera complex, in the southern Wah Wah Mountains, finely flow-layered, locally glassy flows are as thick as 250 m and occupy a similar stratigraphic interval as the andesite lava flow formation of the Escalante Desert Group (Best, 1987; Best et al., 1987d). The variegated colors as well as the size and abundance of phenocryst are similar to those of the Lamerdorf Tuff, as are the types of phenocrysts, viz., plagioclase, biotite, and locally amphibole. Farther north in the Wah Wah Mountains, in the deep paleovalley filled with thick Lamerdorf Tuff (Fig. 15), an unusual low-silica rhyolite lava at the base of the volcanic pile above local conglomerate has altered phenocrysts of platy plagioclase as long as 1 cm and lesser smaller unidentifiable remnants of mafic phases (Abbott et al., 1983, their unit Toa designated as “older andesite”; alkali concentrations are perturbed in our sample LAM-9-18-2). A petrographically and chemically very similar flow lies directly on Paleozoic rocks just beyond the northwest margin of the caldera complex in the western Fairview Range (Best et al., 1998, their unit Tb, our sample FAIRV-3-160-1).

A perlitic rhyolite lava flow hundreds of meters thick was encountered between dacitic “Needles Range” tuffs and Cambrian carbonate rocks in two adjacent deep wells in Pine Valley at ∼38°32′45″ N and 113°43′00″ W (see Hintze and Davis, 2003, p. 257, for data on one of these wells).

Four small plutons of granitic rock northwest of Milford, Utah, that may be connected at depth were intruded at ca. 31 Ma (Best et al., 1989c). These plutons are associated with porphyry Cu mineralization.

Andesitic to dacitic lava and debris flows extruded during the monotonous intermediate and trachydacitic ignimbrite activity constitute small exposures within the Indian Peak caldera complex and to the southwest (Fig. 76). The most voluminous occurrence of andesitic rocks is in the latter area, where they are remnants of a volcanic pile that is as thick as 1200 m lying between the 30.06 Ma Wah Wah Springs and 29.1 Ma Petroglyph Cliff ash-flow tuffs (Swadley et al., 1994).

The largest silicic body is a dominantly dacitic stratovolcano ∼10 km in diameter with a volume of less than 20 km³ that developed at ca. 28 Ma (du Bray, 1993) in the Seaman Range midway between the Central Nevada and the Indian Peak–Caliente caldera complexes. Granitic to dioritic intrusions in the 29.1 Ma Blind Mountain caldera (Figs. 8 and 58) are too small to be shown in Figure 76. In the Fairview Range, a dacite lava was apparently extruded on the margin of the White Rock caldera (Best et al., 1998, their unit Tch). A small rhyolite lava in the southern Indian Peak Range (Best et al., 1987a, their unit Trr) lies between the Wah Wah Springs and Ryan Spring ignimbrites. West of Milford are small 31–30 Ma plutons of granitic rock.

Figure 76 displays a blossoming of andesitic and latitic lava extrusions from ca. 25–18 Ma as explosive silicic activity waned in the southern part of the Indian Peak caldera complex and progressed farther southward into the Caliente caldera complex and surrounding areas. On the northwest margin of the Escalante Desert, four types of chemically rather similar intermediate-composition lava flows have been recognized based on phenocryst assemblage; chemical distinctions are not clearly evident (Fig. 77; Best, 1987):

1. The most restricted in areal extent is the dacite of Spanish George Spring (S in Fig. 77), a lava dome complex ∼5 km in diameter and no more than 0.5 km thick, in which ∼20% of the rock is composed of blocky plagioclase grains as much as 3 cm across; smaller lesser phenocrysts include biotite, hornblende, quartz, Fe-Ti oxides, and, in some samples, a trace of titanite, clinopyroxene, and sanidine. Stratigraphic relations indicate an age of ca. 25 Ma.

2. Lava flows and flow breccias of hornblende-bearing andesite and latite (H) that are as thick as several hundred meters occur north and south of Modena (Siders, 1985). Sparse phenocrysts include conspicuous hornblende together with minor plagioclase, pyroxene, and biotite. Critical stratigraphic relations are lacking to fix the age of these flows but one K-Ar hornblende age is 24 Ma. East of Lund, the hornblende-bearing Mount Dutton Formation is mostly a vent-facies, volcanic debris flow as thick as 600 m and probably emplaced ca. 26–22 Ma (Rowley, 1978). The mostly younger Horse Valley Formation, also a stratovolcanic complex, appears to be more silicic, but neither unit has been chemically analyzed.

3. A large area of latite and trachydacite lava flows and flow breccias (L) as thick as 1500 m emplaced ca. 22 Ma are exposed north and west of Modena. Phenocrysts make up ∼10% of the rock and include prominent biotite, plagioclase, minor clinopyroxene, and local hornblende (Best, 1987; Williams et al., 1997).

4. The most widespread but not the largest volume of the intermediate-composition lava flows are those assigned to the informal mafic flow member of the Blawn Formation (B) on geologic maps of the Indian Peak caldera complex and the southern Wah Wah Mountains and Shauntie Hills (e.g., Best et al., 1987c). Flows range from ca. 25 to 21 Ma (Best et al., 1989c). Some sequences are hundreds of meters thick. Typical phenocrysts include plagioclase, clinopyroxene, Fe-Ti oxides, and combinations of lesser orthopyroxene, hornblende, and either olivine or biotite. Chemically, the mafic flow member ranges widely and includes trachydacite, andesite, basaltic andesite, and, most common, latite (Barr, 1993; Best et al., 2009).

The large area in Utah south of 37°30′ N on the southern margin of the Caliente caldera complex (Fig. 76) comprises intermediate-composition stratovolcano deposits of lava and volcanic debris flows as thick as 1200 m emplaced mostly between the Leach Canyon and Racer Canyon tuffs (Table 1; see also Rowley et al., 2007). Westward into Nevada, this outcrop area is drawn from the generalized 1:250,000-scale Lincoln County map of Ekren et al. (1977). No chemical analyses are available for this large area of lavas.

To the southeast of the Escalante Desert, the northeast-trending, so-called “Iron Axis” (Shubat and Siders, 1988; Blank et al., 1992; Rowley et al., 2008; Biek et al., 2009) is composed of locally Fe-mineralized granitic laccoliths emplaced at 22–21 Ma. South of these intrusions is the very large, 20 Ma laccolith constituting the Pine Valley Mountains (not shown in Fig. 76).

North of the Escalante Desert within the Indian Peak caldera complex and to the east of it is a broad area in which 23–20 Ma rhyolite to dacite lava flows were extruded from many local vents (Best et al., 1987c); most of these are too small to be shown in Figure 76. High-silica, locally topaz-bearing, rhyolite extrusions of the Blawn Formation (B in Fig. 77; Willis, 1985; Christiansen et al., 1986, 2007b) were commonly preceded by explosive venting of pyroclasts of the same composition that formed local aprons of tuff; none of these eruptions were of sufficient volume to create a caldera.

The only expression of effusive activity after 20 Ma and contemporaneous with the eruption of the Hiko–Racer Canyon Tuffs are small 18 and 19 Ma rhyolite lavas south of Modena and west of the Delamar Mountains.

The compilation of Cenozoic volcanic rocks in the Great Basin by Stewart and Carlson (1976) clearly showed the subordinate volume of andesitic rocks relative to contemporaneous silicic ignimbrites for the 34–17 Ma time period (Fig. 1). We have made minor adjustments from geologic maps published in the decades since this compilation, especially by Crafford (2007), and have cast lava occurrences into the three narrower time frames as outlined above to generate Figure 76. This new compilation allows quantification the volumes of andesitic flow rocks for comparison with the volume of contemporaneous ignimbrites. Because of sparse chemical data and a range of apparent compositions in many occurrences as described in publications, it is impossible to cleanly distinguish andesitic and latitic rocks having <63 wt% silica from dacitic rocks containing more silica.

To further constrain the intended volumetric comparison, we have circumscribed the lava flow activity within an arbitrary distance of 40–50 km from the outer perimeter of the Indian Peak and Caliente caldera complexes. This distance corresponds to the diameter of the larger calderas and limits the activity to within all but the most distal reaches of the larger ignimbrite outflow sheets. To the north of the circumscribed area, the next nearest andesitic lavas are ∼125 km distant and ca. 36 Ma in age (Gans et al., 1989). To the south of the Caliente caldera complex, andesitic lavas more than 18 Ma are only present within less than 20 km of the complex. In the other two quadrants, the perimeter is more arbitrary. To the west, the circumscribed area overlaps by ∼25 km a similarly circumscribed area for the Central Nevada caldera complex (Fig. 2); the overlap encompasses the only locus of silicic volcanism—the 20 km³ Seaman volcanic center (du Bray, 1993)—between the two caldera complexes. To the east, the circumscribed area merges into the Marysvale volcanic field (Fig. 1).

Generalized, extension-corrected areas of lavas for the three time frames are, from oldest to youngest: 500, 300, and 2400 km². Multiplied by the approximate average thickness for each outcrop area, the volumes are 120, 35, and 1200 km³. Before comparing these volumes with those for silicic ignimbrites, some sort of correction must be made for the lavas buried beneath alluvium in the broad valleys. This correction would likely double the lava volumes. Another correction might be made for lavas buried within calderas. The tilted Needle Range and White Rock Mountains and resurgent cores of the Indian Peak and White Rock calderas expose partial intracaldera volcanic sections above Paleozoic rocks. Had major stratovolcanoes existed within the calderas, distal volcanic debris flows would be exposed in the extracaldera sections; however, only minor such flows occur in the central Needle Range, as indicated above. Nonetheless, major piles (stratovolcanoes?) of lava are likely buried in the Caliente caldera complex. Doubling the andesitic rock volumes for the first two time periods and tripling it for the third gives a total of ∼4000 km³ for all three time periods, which is ∼12% of the total volume of contemporary silicic ignimbrite (32,600 km³; Table 2). Remarkably, for the 31–26 Ma time frame when 18,000 km³ of monotonous intermediate and Isom-type tuffs was erupted, less than 100 km³ of andesitic lavas have been inventoried.

Despite the several uncertainties in our estimate of the volume of andesitic lavas, there can be no question of its small fraction relative to the colossal mass of ignimbrite. By way of comparison with contemporaneous volcanic fields, the Marysvale to the east on the western margin of the Colorado Plateau (Fig. 1) has an order of magnitude more intermediate-composition lava than ignimbrite (Cunningham et al., 2007) and the Southern Rocky Mountain field on the eastern margin of the plateau has 1.7 times more lava than ignimbrite (Lipman, 2007).

Although a huge input of mantle-derived basalt magma into the crust was necessary to provide energy and mass to drive crustal magma systems generating the silicic ignimbrite magmas, only a relatively small volume of derivative andesitic magma penetrated all the way through the crust to the surface. And this penetrating volume was apparently smallest during the time period from 31 to 26 Ma. Hence, while the ignimbrite-forming magma systems were operative in the unusually thick crust, the masses of low-density silicic magma, or possibly a widespread layer of crustal partial melt, effectively blocked the ascent of denser andesitic magmas. The circumscribed area over which volumes are compared might correspond to the greater source region, or “sphere of influence,” in the crust where magmas were being generated, and from this region magmas collected and ascended to beneath and erupted from the caldera complex.

Why the extruded volume of andesitic lava blossomed after 25 Ma is uncertain. Because ∼12,000 km³ of contemporaneous, mostly rhyolite ignimbrite was erupted from the Caliente calderas, an underlying melt region must still have existed to provide buoyancy blocking for andesitic magmas. Although the demise of arc magmatism wasn’t until ca. 18 Ma as plate subduction ceased, early effects of crustal extension might have been operative to facilitate, via fracturing and diking, the ascent of andesitic magmas.

The voluminous 25–18 Ma lavas, as well as shallow intrusions, include more-extreme compositions, as manifest in silica and total alkali contents (Fig. 77), than in older lavas in the Great Basin as a whole (Best et al., 1989b). A comprehensive interpretation of these time-volume-composition changes is beyond the scope of this article, but they obviously correspond to the changing tectonic regime of this part of the Great Basin near the beginning of the Miocene as volcanism related to subduction was supplanted by volcanism related to tectonic extension (Christiansen et al., 2007a). Fundamental shifts in the nature of the mantle magma input accompanied these changes at the surface. An additional factor might be the closer proximity to the northeast-trending margin of the Great Basin where the thickened and pre-warmed crust thinned into the Colorado Plateau (Figs. 1 and 76).

The 36–18 Ma Indian Peak–Caliente caldera complex and its surrounding ignimbrite field was a major focus of subduction-related, explosive silicic volcanism in the southern Great Basin ignimbrite province during the middle Cenozoic ignimbrite flareup. The 32,600 km³ of silicic ash-flow tuff was produced by continuously evolving, southward-migrating, crustal magma systems in unusually thick and pre-warmed crust into which invading mantle-derived basaltic magmas furnished heat as well as mass. These powering magmas were generated during southward “rollback” of the subducting oceanic lithosphere beneath the continental margin (see Best and Christiansen, 1991, their figure 2).

Twenty-two mapped regional ignimbrite units (generally >100 km³ each) that are exposed in more than one mountain range have been correlated over a present area of ∼60,000 km² on the basis of geologic mapping, position in stratigraphic sequence, composition, paleomagnetic direction, and ⁴⁰Ar/³⁹Ar age. Runout distances of ash flows were as great as 150 km.

Conformable sequences of outflow cooling units are commonly hundreds of meters thick and lack intervening erosional debris deposits, thus testifying to tectonic quiescence and absence of large-magnitude crustal extension during most of the ignimbrite flareup.

At least seven super-eruptions having volumes of more than 1000 km³ each took place from 31.13 to 18.51 Ma. Super-eruptions of rhyolite magma occurred after 24 Ma whereas smaller rhyolite eruptions took place earlier, to 36 Ma (Fig. 78). Three super-eruptions were of the distinctive monotonous intermediate magmas composed of relatively uniform phenocryst-rich dacite that erupted at 31.13 Ma (Cottonwood Wash), 30.06 Ma (Wah Wah Springs), and 29.20 Ma (Lund) in volumes of ∼2000, 5900, and 4400 km³, respectively. After this tremendous burst of activity from a multicyclic locus, almost 4000 km³ of trachydacitic magma (Bald Hills) was erupted from 27.90 to 27.25 Ma to the southeast of the locus. An eruption farther south of 2200 km³ of phenocryst-rich andesite-latite magma (Harmony Hills) occurred at 22.56 Ma. The volumes and compositions of these monotonous intermediate, trachydacitic, and andesite-latite eruptions are without parallel elsewhere in the southern Great Basin ignimbrite province and in other well-documented volcanic fields in southwestern North America where the middle Cenozoic ignimbrite flareup is expressed.

The apparent lack of significant Plinian deposits associated with the monotonous intermediates and the lack of evidence for significant fractionation of fine glass particles in the Wah Wah Springs indicate ash flows were erupted at high rates, “boiling over” from the vent, rather than by collapse of high-standing eruptive columns.

Eruptions created nine at least partly exposed calderas as much as 60 km in diameter in the unextended north-south dimension that are filled with as much as 5–6 km of syn-collapse intracaldera tuff and post-collapse caldera-filling tuff. Additional source calderas are buried beneath younger deposits in nested, multicyclic caldera clusters. Collar zones between inner reverse and outer normal ring faults in two of the largest calderas reveal complex evolution during caldera collapse that involved extensional faulting and brecciation of wall rock; these two calderas also display evidence for resurgence and post-collapse intrusive activity.

Dispersal of ash flows from their sources to create ignimbrite outflow sheets is influenced by eruption dynamics and landscape topography; on the Great Basin altiplano (Best et al., 2009) where the Indian Peak–Caliente field is located, only minimal relief existed when ash flows, especially the younger ones, were broadcast across the landscape. Present outcrop dimensions of outflow sheets were also controlled by subsequent erosion and by east-west crustal extension subsequent to their emplacement in the Indian Peak–Caliente field. Perusal of the maps in this article that display the distribution of the ignimbrite deposits clearly reflect the influence of east-west crustal extension on the present shapes of the larger outflow sheets for which adequate information is available. For all sheets except the Cottonwood Wash, their present east-west dimension exceeds their north-south. Caldera-fill deposits whose distribution was governed by an older depression, such as the Lund, and small deposits, such as the Petroglyph Cliff Ignimbrite, which are exposed only in a restricted area, do not necessarily show a similar influence or amount of extension on their distribution.

With these caveats in mind, it is of interest to use the outflow sheets as strain markers by comparing the present-day ratios of east-west to north-south dimensions of the larger ignimbrite sheets with their dimensions after compensating for three different amounts of uniform crustal extension (Table 8). Although only ten sheets were examined, the dimensional ratios indicate that, after compensation for an arbitrary uniform 35% extension, sheets are still, on average, somewhat elongate east-west. Correction for 65% extension is too much as most sheets are more elongate north-south. For 50%, or slightly less, extension, the sheets are, on average, equidimensional, supporting our use of the 50% value throughout this article for correction of areas and volumes of tuffs (Table 1). Even if the proper value for the amount of extension has been applied as a correction, an individual sheet may still be non-equidimensional because of the factors controlling sheet shape listed above; but for several sheets these factors cancel out in the average.

If crustal extension had accompanied the ignimbrite flareup in the Indian Peak–Caliente field, as advocated by some geologists, then present-day dimensional ratios would be larger for the older sheets and smaller for the younger. The absence of such temporal variation in the ratios in Table 8 is an argument for the lack of extension during most of the ignimbrite flareup.

From 36.02 Ma to 27.25 Ma, eruptions in the Indian Peak–Caliente field (Table 1; Fig. 78) were essentially continuous, with breaks of no more than about 1 m.y. A lull in activity of 2.3 m.y. followed and another lull of 3.6 m.y. after 22.56 Ma before the culminating eruptions from 18.9 to 18.5 Ma of many cooling units of phenocryst-rich rhyolite magma at the end of the ignimbrite flareup.

The earliest explosive eruptions, in the northern part of the field, totaled no more than ∼300 km³ of rhyolite magma. The 36.02 Ma formation of The Gouge Eye comprises a small lava dome and surrounding related pyroclastic deposits that apparently accumulated mostly within a small asymmetric caldera located immediately north of the Indian Peak caldera complex. The formation is exposed only in the Fortification Range but well cuttings have been found in adjacent Lake Valley. Fifty kilometers northeast of the caldera complex the 35.26 Ma Tunnel Spring Tuff was deposited in a paleovalley around its small concealed Crystal Peak caldera source. Sources of the larger 33.5 Ma Sawtooth Peak Formation, undated Marsden Tuff, and 32.09–31.90 Ma Lamerdorf Tuff are concealed beneath younger deposits and were likely engulfed, at least in part, in younger calderas in the northeastern part of the caldera complex. The phenocryst-rich Tunnel Spring and phenocryst-poor Marsden Tuffs are high-silica rhyolites whereas the other three ignimbrites are low-silica rhyolites.

Following this precursory rhyolitic activity, three super-eruptions of phenocryst-rich dacite occurred at 31.13, 30.06, and 29.20 Ma, creating monotonous intermediate ignimbrites of the Cottonwood Wash Tuff, Wah Wah Springs Formation, and Lund Formation whose volumes were ∼2000, 5900, and 4400 km³, respectively. Hundreds of cubic kilometers of Cottonwood Wash and Wah Wah Springs fallout ash occur as far as Nebraska. The Cottonwood Wash and Lund ignimbrites appear to be unzoned; limited variations in their composition are likely the result of slight contrasts in composition of parcels of magma within the pre-eruption chamber and, for the Cottonwood Wash, not of substantial fractionation of vitroclasts in their ash flows. On the other hand, the Wah Wah Springs magma chamber possessed significant gradients in chemical composition—a few weight percent more silica and more volatiles at the top; in the main dacitic part of the chamber there were fewer and smaller phenocrysts near the top and the proportion of quartz plus clinopyroxene increased from about nil to 14% in the deepest erupted level as a result increased pressure. Comparison of compositions of ignimbrite and cognate clasts indicate minimal fractionation of vitroclasts in the Wah Wah Springs ash flows.

The caldera source of the Cottonwood Wash Tuff was apparently engulfed in the younger Indian Peak caldera or overlapping White Rock caldera, sources of ash-flow tuffs of the Wah Wah Springs and Lund Formations, respectively.

Tilting of the Needle Range by post-volcanic extensional faulting has provided an exceptional exposure of the internal structure and stratigraphy of the northeastern sector of the Indian Peak caldera that includes a 3500-m-thick section of lithic caldera-collapse ignimbrite and intercalated breccia deposits inboard of the inner reverse ring fault. In the 11-km-wide collar zone between this ring fault and the topographic margin, a layer of cataclastic wall-collapse breccia as thick as 600 m underlies 1100 m of post-collapse, caldera-filling tuff. The breccia layer was apparently formed by incremental collapse accompanying downward displacement on progressively outward-stepping reverse faults and on a north-bounding normal fault in the extending annular collar zone. The breccia layer is made all the more unusual by the presence of seams of ultracataclasite that testify of extreme shearing in the collapsing milieu of pre-collapse Cottonwood Wash and Wah Wah Springs tuffs sliding into the deepening depression. Apparently immediate resurgent uplift of this northeastern caldera sector accompanied a major intrusion of a granodiorite porphyry of the same composition as the Wah Wah Springs ignimbrite.

The lower part of the caldera-filling tuff in the Indian Peak caldera comprises the phenocryst-poor, low-silica rhyolite tuffs of the Ryan Spring Formation whose volume is estimated at ∼1000 km³. Multiple cooling units of the 30.13 ± 0.13 Ma Greens Canyon Tuff Member from an undisclosed source accumulated mostly in the northern part of the collar zone. An equivalent volume of the 30.01 Ma Mackleprang Tuff Member had a source caldera in the southeastern collar zone that was buried beneath an unusually thick accumulation—as much as 1400 m—of the younger Lund ignimbrite.

Lund ash flows apparently mostly vented from within the southeastern sector of the older Indian Peak caldera at 29.20 Ma; they filled the remainder of this depression, and spilled over in thick accumulations to the east but progressively less to the south, west, and north. As much as 2500 m of intracaldera Lund ignimbrite and lesser wall-collapse breccias plus 1000 m of post-collapse, caldera-filling tuffs define the northern sector of the White Rock caldera. The Pioche Hills horst exposes a complexly faulted and intensely brecciated sequence of Cambrian rocks that is interpreted to have resulted from local collapse-related extension in the southwestern segment of the White Rock caldera collar zone between hypothesized inner reverse and outer normal ring faults. Like the older Indian Peak caldera, the White Rock caldera experienced resurgent uplift, but unlike in the Indian Peak no major intracaldera intrusion is exposed; instead, late, post-collapse magmatic activity is manifest by several small extrusions and possible shallow intrusions of dacitic lava along the hypothetical concealed ring fault.

After the Ryan Spring, explosive eruptions over the next million years occurred in the northwestern sector of the Indian Peak caldera complex. The 30.00 Ma tuff of Deadman Spring is the only phenocryst-rich, high-silica rhyolite in the complex. The source of this 180 km³ ignimbrite was the modest-size asymmetric Kixmiller caldera that formed several kilometers west of the slightly older Indian Peak caldera and was later partially engulfed in the younger White Rock caldera. The source of the 350 km³ phenocryst-rich dacitic Silver King Tuff that was emplaced at 29.40 Ma has not been verified but its localization in the northwestern part of the field and great thickness—as much as 1400 m—within the Kixmiller caldera indicate a nearby source, or one possibly nested within this earlier structure. The oldest Isom-type tuff is the small—40 km³—unusually clast-rich Petroglyph Cliff Ignimbrite that was erupted at ca. 29.1 Ma from the small Blind Mountain caldera located ∼5 km beyond the southwestern margin of the White Rock caldera. The 800 km³ ignimbrite volume of the 29.0 Ma Ripgut Formation constitutes caldera fill in the White Rock depression. Nested within this older caldera is the Mount Wilson caldera source of the Ripgut, but only its northern part is exposed. The 2-km-thick intracaldera tuff is normally zoned from high-silica, very phenocryst-poor rhyolite to a less evolved rhyolite that contains fiamme of trachydacite compositionally like the Lamerdorf ignimbrite.

Relatively dry, alkaline, high-temperature magmas forming the 4200 km³ ignimbrites of the trachydacitic Isom Formation were then erupted from a concealed source just beyond the southeast margin of the Indian Peak caldera complex. Most of this volume is made up of at least four cooling units of the Bald Hills Tuff Member that were emplaced from 27.90 to 27.25 Ma over a broad area of ∼21,000 km². After an eruptive lull of 2.3 m.y. as many as four cooling units of the Hamlight Tuff Member were deposited from 24.91 to 24.75 Ma, apparently only to the west of the concealed source. Following the Hamlight eruptions, at 24.55 Ma, a cooling unit of the Hole-in-the-Wall Tuff Member was emplaced east and west of the source; this is a chemically distinct low-silica rhyolite.

Thereafter, explosive activity shifted farther to the south where it was dominated by eruption of rhyolitic magmas rather than the dacitic magmas that were dominant in the north. Although the sources of the next two ignimbrites have not been located, they must lie between the Indian Peak caldera complex to the north and the Caliente caldera complex to the south, and possibly were engulfed, at least in part, within the latter complex. Ash-flow tuffs of the 24.03 Ma Leach Canyon Formation compose a normally zoned, somewhat phenocryst-rich rhyolite deposit with a volume of 3600 km³. Ignimbrites of the Condor Canyon Formation deposited next consist of two generally simple cooling units, the 400 km³Swett Tuff Member and the larger 3200 km³Bauers Tuff Member deposited at 23.04 Ma. Both are low-silica, phenocryst-poor, off-trend rhyolites that generally plot apart from other Great Basin ignimbrites on many variation diagrams and share some of the characteristics of the Isom-type tuffs, e.g., high K₂O, TiO₂, Ba, and Zr and low CaO. But they contain biotite as the major mafic phenocryst rather than pyroxenes. The source of the Bauers was the Clover Creek caldera that is poorly exposed in the northern sector of the Caliente caldera complex. This source is marked by hundreds of meters of clast-rich intracaldera tuff comprising numerous cooling units and by a hypabyssal intrusion of similar composition.

One-half million years after deposition of the Bauers, a super-eruption of 2200 km³ created the Harmony Hills Tuff at 22.56 Ma. The volume of this ignimbrite and its andesite-latite composition are apparently unique in all of the Great Basin; it may be of the same genre as the older monotonous intermediates. The Harmony Hills is not only unusually mafic but is also unusually phenocryst rich—phenocrysts make up as much as 58% of the tuff. The Harmony Hills apparently had a source in the central part of the Caliente caldera complex where it is concealed by younger deposits.

The last major eruptions of subduction-related magmas from the Caliente caldera complex followed a period of inactivity of ∼3.6 m.y., creating two phenocryst-rich ignimbrites. The 18.88 to18.57 Ma Racer Canyon Tuff comprises numerous cooling units totaling 1100 km³ that are normally zoned from high-silica rhyolite to high-silica dacite exposed chiefly in Utah near the Telegraph Draw caldera source in the eastern sector of the caldera complex. The 18.51 Ma, 1700 km³Hiko Tuff is generally a single, normally zoned, rhyolite cooling unit exposed principally in Nevada near the Delamar caldera source in the western sector of the Caliente caldera complex.

A very small volume of the nonwelded, high-silica rhyolite tuff of Tepee Rocks that is found as caldera-filling in the Delamar caldera is the last manifestation of arc magmatism at 18.1 Ma.

The relatively small volume, ∼300 km³, of andesitic lava flows extruded before 25 Ma probably reflects the existence of a widespread zone of silicic magma in the deeper crust that effectively blocked the ascent of the more dense magmas. After 25 Ma, the onset of crustal extension apparently allowed these mafic magmas to ascend through fractures so that thousands of cubic kilometers of lavas were extruded. This transition in tectonic regime was ultimately complete after 18 Ma when extensional geochemical signatures are evident in the volcanic rocks of the Great Basin.

In the southern Great Basin ignimbrite province, 36–18 Ma ignimbrites possess typical arc chemical signatures, indicating their subduction-related heritage. Magmas had unusually high concentrations of K, indicating their origin in unusually thick continental crust, likely to as much as 70 km thick. The absence of extruded basalt and the limited volume of andesitic lavas during the ignimbrite flareup are also a consequence of the unusually thick crust. The general southward migration of eruptive sources through time reflects the southward rollback of the subducting plate beneath the continent. An unusually high influx of mantle magma into the unusually thick crust was necessary to create the colossal volume of erupted silicic magma.

The main spectrum of magmas ranges from high-silica rhyolite (78 wt% SiO₂) to high-silica andesite (61 wt% SiO₂) and includes huge monotonous intermediates. These magmas equilibrated under water-rich conditions at temperatures of <830 °C and depths of ∼7–9 km with an assemblage of plagioclase, quartz, biotite, and Fe-Ti oxides with or without hornblende, sanidine, pyroxene, zircon, apatite, and titanite. Hotter (∼900 °C) and drier trachydacitic magmas of the Isom type contained unusually high concentrations of TiO₂, K₂O, P₂O₅, Ba, Rb, Ce, Zn, Zr, Y, and Th, and generally higher Nb, Y, and U, but distinctively lower concentrations of many compatible elements including MgO, CaO, Sr, Ni, Cr, and V for the same range of silica as in the dacites and rhyolites. These magmas equilibrated at depths of ∼30 km with an assemblage of plagioclase, two pyroxenes, and Fe-Ti oxides. Relatively small volumes of silica-poor, off-trend rhyolitic magmas apparently originated by combination of the main-spectrum magmas with those of the Isom type.

Relatively high initial ⁸⁷Sr/⁸⁶Sr ratios in the lavas and especially the ignimbrites indicate magmas were not derived solely from the mantle, but reflect varying proportions of old continental crust combined with the mantle-derived basaltic magmas. Older lavas and ignimbrites, especially the trachydacitic Isom types, have higher ratios than younger, testifying to the gradual diminution of fertile felsic crust added into the magmas.

In short, we conclude that the southern Great Basin ignimbrite province with its colossal volume of erupted silicic magmas owes it origin to distinctive tectono-magmatic conditions and processes related to thick pre-heated felsic crust and rollback of a subducting slab of oceanic lithosphere. Consequently, this province developed along a convergent margin during subduction, in contrast to other large silicic igneous provinces that developed during continental breakup.

Our understanding of the geology of the Indian Peak caldera complex profited from the collaborative mapping of Jeffrey Abbott, Kerry Grant, Jeff Keith, Mark Loucks, Hal Morris, Margo Toth, Van Williams, Julie Willis, and, especially, Lehi Hintze and some 600 geology students at Brigham Young University in conjunction with their summer field courses from 1967 to 1997. We gratefully acknowledge their very significant contributions. The field work of three of these students—Richard Holmes, Kim Sullivan, and Jack Rogers—was instrumental in the discovery of three calderas in the Indian Peak complex. Others worked out details of the petrology—Larissa Maughan, Keryn Tobler Ross, Kurtus Woolf, Garret Hart, and Glenn Blaylock. Many geologic maps were published under the auspices of and with the financial support of the U.S. Geological Survey’s Richfield 2° National Uranium Resources Evaluation Project under the direction of Thomas A. Steven. Tom’s patient encouragement, mentoring regarding ash-flow tuffs and calderas, as well as his ability to expedite government publications is gratefully acknowledged. H. Richard Blank, Jr., first provided a Bouguer gravity map of the Indian Peak caldera complex that served to constrain its margin. In the 1970s, Kerry Grant, Ralph Shuey, and Rick Caskey assisted in establishing the volcanic stratigraphy in southwestern Utah. Bart J. Kowallis, Ted McKee, and Harald H. Mehnert provided crucial isotopic ages. Daniel R. Shawe made available important unpublished work on the Pioche Hills. Gary J. Axen furnished copies of his unpublished geologic maps. Michael Laine, Thomas Dempster, and Thomas Chidsey at the Utah Geological Survey Core Research Center and David Davis at the Nevada Bureau of Mines and Geology made well cuttings available for our examination. Jim Toy and Virgil Frizzell in Caselton allowed us to examine and sample well cuttings from grabens flanking the Pioche Hills. Michael Dorais performed electron microprobe analyses and Kathleen Robertson helped draw some of the illustrations.

Because we have conducted almost no field work in the Caliente caldera complex and little with the associated ignimbrites, we relied heavily on the published and unpublished research of other geologists, including John Anderson, Richard Blank, Earl Cook, Gary Dixon, Bart Ekren, Robert Scott, Paul Williams, and, especially, Peter Rowley through their extensive 1:24,000-scale geologic mapping, mostly under the auspices of the U.S. Geological Survey. Except for Scott and Ekren, these geologists were mentored as students by J. Hoover Mackin, who laid the groundwork for the volcanic history of the southeastern Great Basin and adjacent High Plateaus of Utah through his seminal investigation of the Iron Springs mining district west of Cedar City (Mackin, 1960).

We are indebted to Robert Biek, Matthew Brueseke, Anita Grunder, and Peter Rowley for their constructive comments on an earlier version of this article.

Financial support for the Great Basin project was provided by the National Science Foundation through grants EAR-8604195, 8618323, 8904245, 9104612, 9706906, and 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.

In an east-west transect from ∼113°30′ to 117°20′ W longitude (westernmost Utah to central Nevada) between 39° and 40° N latitude, Smith et al. (1991) determined an overall extension of 55% that resulted from mostly early Miocene (23 Ma) and younger faulting. (Our measurement of their present-day cross-sectional length compared to the palinspastically restored section gives 44% overall extension.) However, the amount of extension in this transect varied greatly in different domains, from ∼110% in eastern Nevada to ∼40% in central Nevada and nil in between and in westernmost Utah. Although this transect lies just north of the Indian Peak–Caliente ignimbrite field (Fig. 2), it is reasonable to extrapolate the same overall amount of extension southward. This extrapolation follows from the paleomagnetic observation that the Sierra Nevada block moved uniformly westward relative to the Great Basin about a pole of rotation near the geographic north pole (Hillhouse and Gromme, 2011).

McQuarrie and Wernicke (2005, their table 1) have tabulated the amount of extension, mostly after ca. 18 Ma, within individual strain domains between 40°20′ and 38°40′ N in an east-to-west transect from 114°7′ to 117°23′ W. Strains are significantly greater on two adjacent structural domains in their transect, namely, the Sevier Desert detachment in western Utah and the Snake Range décollement in easternmost Nevada. Because interpretations of these structures are controversial, relevant conflicting information is provided as a basis for corrections to the areal extents of ash-flow sheets in the Indian Peak–Caliente field that lies principally to the south of these two domains.

The 40 km of extension claimed by many geologists (e.g., McQuarrie and Wernicke, 2005) for the Sevier Desert detachment in west central Utah has been challenged by, among others, Hintze and Davis (2003) and Wills et al. (2005). They question the very existence of the detachment—interpreted from seismic reflection data—and argue that the available information may permit substantially less total extension, to as low as 6 km across the Sevier Desert. McBride et al. (2010) re-analyzed the seismic reflection data and concluded it is consistent with either substantial or minimal extension.

For more than two decades, B.P. Wernicke and associates (e.g., Lewis et al., 1999, p. 50) have speculated that the Snake Range décollement is a major low-angle extensional shear zone that underlies several ranges in the eastern Great Basin and cuts deep into the upper crust. Bartley and Wernicke (1984) estimated 60 km of displacement on the décollement between the Egan Range and the Confusion Range. McQuarrie and Wernicke (2005, their table 1) indicated 79 km of extension over the same interval. In contrast to the Wernicke model, Gans and Miller (1983), Miller et al. (1983), and Gans et al. (1985) concluded that the décollement is a broadly domical brittle-ductile transition zone overlying a metamorphic core complex and that the décollement does not extend more than 60 km in any direction. Restoration of tilted fault slices in the brittle hanging wall to their pre-deformation configuration indicates as much as 500% extension whereas the ductile footwall has been stretched at least 330%. This deformation, along with slip on major faults to the north, constitutes a unified extensional system with at least 15 km of slip. Gans and Miller (1983, p. 136) believed that similar low-angle faults and large stratal rotations are characteristic of the Egan and Schell Creek Ranges where 95 km (238%) of extension occurred in a 40-km-wide corridor. Fifty kilometers of extension took place between the Schell Creek and Confusion Ranges.

Accepting maximum extensions of 40 and 79 km on the Sevier Desert detachment and Snake Range décollement, respectively, as cited in McQuarrie and Wernicke (2005, their table 1), we calculate there is 145 km (64%) of extension over a present-day distance of 372 km from 111°47′ to 116°7′ W, which is the longitudinal span of the Wah Wah Springs tuff—the largest sheet in the Indian Peak field. For the Caliente ignimbrite field the extension is 109 km (75%) over ∼254 km from 112°30′ to 115°30′ W. But if we assume more conservative displacements of, say, 10 and 20 km, respectively, for the two structural domains, the extensions for the two fields are reduced to 83 km (29%) and 62 km (32%), respectively. As a compromise and for convenience, as well as to be consistent with dimensions of ignimbrite deposits cited in earlier publications (e.g., Best et al., 1989a), we use a value of 50% uniform extension for the Indian Peak and Caliente fields. This value is consistent with the data for ignimbrite outflow sheets tabulated in Table 8.