Miocene silicic volcanism in southwestern Idaho: geochronology, geochemistry, and evolution of the central Snake River Plain
- First Online:
- Cite this article as:
- Bonnichsen, B., Leeman, W.P., Honjo, N. et al. Bull Volcanol (2008) 70: 315. doi:10.1007/s00445-007-0141-6
- 524 Downloads
New 40Ar-39Ar geochronology, bulk rock geochemical data, and physical characteristics for representative stratigraphic sections of rhyolite ignimbrites and lavas from the west-central Snake River Plain (SRP) are combined to develop a coherent stratigraphic framework for Miocene silicic magmatism in this part of the Yellowstone ‘hotspot track’. The magmatic record differs from that in areas to the west and east with regard to its unusually large extrusive volume, broad lateral scale, and extended duration. We infer that the magmatic systems developed in response to large-scale and repeated injections of basaltic magma into the crust, resulting in significant reconstitution of large volumes of the crust, wide distribution of crustal melt zones, and complex feeder systems for individual eruptive events. Some eruptive episodes or ‘events’ appear to be contemporaneous with major normal faulting, and perhaps catastrophic crustal foundering, that may have triggered concurrent evacuations of separate silicic magma reservoirs. This behavior and cumulative time-composition relations are difficult to relate to simple caldera-style single-source feeder systems and imply complex temporal-spatial development of the silicic magma systems. Inferred volumes and timing of mafic magma inputs, as the driving energy source, require a significant component of lithospheric extension on NNW-trending Basin and Range style faults (i.e., roughly parallel to the SW–NE orientation of the eastern SRP). This is needed to accommodate basaltic inputs at crustal levels, and is likely to play a role in generation of those magmas. Anomalously high magma production in the SRP compared to that in adjacent areas (e.g., northern Basin and Range Province) may require additional sub-lithospheric processes.
KeywordsCaldera Crustal foundering Ignimbrite flare-up Lava flow Rhyolite Rifting Snake River Plain 40Ar-39Ar dating
This paper presents a synthesis of the regional geology and our current understanding of the volcanic stratigraphy, followed by a description of magmatic evolution in the CSRP. We then consider the petrogenetic and tectonic processes underlying silicic magmatism and their implications for development of the SRP-Yellowstone province as a whole. Our intent is to present what we know or can infer from work to date, and to focus attention on some of the remaining outstanding problems. A major take-home message is that the scale of magmatism associated with this province is enormous. We propose that silicic magmatism was driven primarily by infusion of great volumes of mantle-derived basalt into the crust, and conclude that much of the underlying crust was affected by partial melting and by deformation required to accommodate the volumes of magma involved. Origin(s) of the basaltic magmas and their link to sublithospheric processes are fundamental questions that will be addressed elsewhere.
The earliest manifestation of silicic magmatism that is considered to be related to the SRP province was eruption of peralkalic silicic volcanics associated with the ca. 16–14 Ma McDermitt caldera complex (Rytuba and McKee 1984; Conrad 1984) and vents in nearby ranges (Brueseke et al. 2007) in north-central Nevada. This volcanism is closely associated in time and space with early phases of Columbia Plateau flood basalt activity (cf. Camp and Ross 2004), leading to the view that it is hotspot-related. Rhyolitic activity directly associated with the western Snake River Plain (WSRP) first occurred in southwestern Idaho (Juniper Mountain/Owyhee-Humboldt eruptive center; JM) around 15–13 Ma (Ekren et al. 1984; Manley and McIntosh 2002).
Between roughly 12 and 7 Ma, NE–SW crustal extension in the WSRP produced a conspicuous NW-trending fault-bounded graben or rift that diverges from the main hotspot track (Fig. 1). This WSRP rift zone, the eastern SRP (ESRP), and the Owyhee Plateau (including the JM center) converge in the CSRP where the exceptionally prolific Bruneau-Jarbidge (BJ; Bonnichsen 1982a) and Twin Falls (TF; Wright et al. 2002) eruptive centers produced numerous regionally extensive ignimbrites and large-volume rhyolite lavas in late Miocene time. Overall, the cumulative volume of silicic magma produced in the CSRP may dwarf that produced at Yellowstone by several-fold (Perkins and Nash 2002). CSRP volcanism is the main focus of this paper, and is described in more detail in the next section.
A notable flare-up between 11.7–10.2 Ma produced numerous distinctive rhyolites from widely distributed vents in the BJ center, along the margins of the WSRP graben (e.g., Owyhee Front, OF; Bonnichsen et al. 2004), the Mount Bennett Hills to the north (MBH; Leeman 1982b, c; Honjo 1990), and as far east as the Pocatello area (e.g., Arbon Valley tuff, or AVT; cf. Kellogg et al. 1994). Eruptions of these units conspicuously diverge from the widely accepted age-space pattern for early SRP volcanism. The AVT is also anomalous with regard to its composition and mineralogy—particularly the presence of abundant biotite phenocrysts, which are notably absent in other early SRP rhyolites.
With exception of the AVT and related rhyolites, silicic volcanism in the ESRP was concentrated between 7 to 4 Ma and produced several regionally extensive ignimbrites and associated airfall tuffs that are attributed to the Heise volcanic center (Morgan et al. 1984; Morgan and McIntosh 2005). This sequence was onlapped to the east by ash flow tuffs produced from the well-known Yellowstone eruptive center beginning around 2 Ma (Christiansen 2001).
Most rhyolites of the SRP-Yellowstone region are metaluminous in composition, have high temperature anhydrous mineral assemblages (cf. Perkins et al. 1995), and have distinctive Sr and Nd isotopic compositions compared to those from north-central Nevada and eastern Oregon (Leeman et al. 1992a; Streck and Grunder 2007; Nash et al. 2006; Brueseke et al. 2007). This distinction coincides closely with the inferred locus of the western Idaho suture zone between cratonic North America to the east and accreted oceanic terranes to the west. Thus, the nature of the underlying lithosphere appears to have influenced compositions of the rhyolites—with a crustal signature more prevalent to the east.
Regarding the hypothesis that SRP magmatism is related to the Yellowstone hotspot, we also note that patterns of SRP volcanic activity differ in several important respects from typical oceanic hotspot magmatism. In any given area within the SRP, initial silicic volcanism persisted for several million years prior to onset of basaltic volcanism but, once initiated, the latter continued sporadically (up to Quaternary time locally). Delayed appearance of basaltic volcanism in the SRP may be explained by a crustal ‘density filter’ effect, but its persistence long after the North American plate migrated past the Yellowstone hotspot is seemingly inconsistent with basalt derivation from a narrowly focused source (e.g., a plume tail). Also, whereas silicic volcanism is a minor component of oceanic volcanism, the cumulative volume of rhyolitic magma (on the order of tens of thousands of km3; Perkins and Nash 2002) places the SRP among the most productive silicic igneous provinces known (Mason et al. 2004).
To understand the origin of the silicic magmatism, as well as the influence of the underlying basement, we have developed a comprehensive stratigraphic and petrologic framework for CSRP Miocene volcanism that spanned some 6–7 m.y. We describe the most important eruptive units, present new stratigraphic and age information, and document evolutionary trends of eruptive style, volume, geochemistry, and petrology.
Overview of central Snake River Plain geology
The main focus area for this study covers some 140 km SE to NW and 170 km SW to NE in south-central Idaho, exceeding an area of 20,000 km2. Representative stratigraphic sequences are described from the following areas (see Fig. 1): (1) Bruneau-Jarbidge (BJ) eruptive center [with partly equivalent units from the Grasmere escarpment (GE) and Jacks Creek (JC) areas to the west]; (2) the Rogerson-Twin Falls region (RTF)—which includes units from the Twin Falls (TF) eruptive center and the Rogerson graben-Browns Bench (BB) area near the Idaho-Nevada boundary (Andrews et al. 2007); and (3) the Mount Bennett Hills (MBH) along the northern margin of the SRP (Bonnichsen et al. 1989; Honjo 1990; Wood and Clemens 2002). Most silicic volcanic rocks in this broad region are inferred to have erupted from vents within the CSRP that are now largely buried beneath a veneer of younger basalts and fluvial-lacustrine sediments.
Along the north margin of the CSRP, in the MBH, Miocene silicic volcanic rocks similar to those to the south unconformably overlie Eocene volcanic rocks (Challis Volcanics) and granitic rocks of the Idaho batholith. The western MBH rhyolites comprise two sequences of multiple tuffs and/or lavas erupted from southerly sources, also buried within the CSRP. Many of the individual units are equivalent at least in age to rhyolites of the BJ area. Structural relations between these packages suggest that early development of the WSRP bounding faults was in part syn-volcanic. Later, we show that rhyolites from the MBH and the BJ area define similar temporal trends in composition—suggesting that all are related to a common (or similarly) evolving magmatic system(s). Volcanism in the eastern MBH differs from that in the southern highlands in having a unique rejuvenation of silicic magmatism between 6–3 Ma, represented by the Magic Reservoir eruptive center (MREC) (Leeman 1982c; Honjo et al. 1986). The earliest MREC products include ferrolatites that clearly formed by mixing between coeval basalt and rhyolite magmas (Honjo and Leeman 1987). Because MREC rhyolites differ from older CSRP rhyolites with regard to mineralogy and composition and appear to be derived from an unrelated magmatic system (Leeman 2004), they are not considered further in this paper.
For brevity, considerable auxiliary supporting information is presented in electronic Appendices as follows: A1, locations of dated samples; A2, summary descriptions of stratigraphic units; A3, all available major/trace element data; A4, averaged analyses for each stratigraphic unit; A5, compiled temperature estimates for CSRP rhyolites; and A6, discussion of stratigraphic relations.
Geochronology of rhyolite units
Critical to understanding the systematics of CSRP magmatism, and that of the province as a whole, is establishment of a regional geochronological framework. Early reconnaissance studies (Armstrong et al. 1975, 1980) provided the basis for time-transgressive magmatism that is often cited in support of the hotspot hypothesis for origin of the SRP. More recent work substantiates the impression that silicic magmatism generally decreases in age northeasterly toward Yellowstone (e.g., Pierce and Morgan 1992; Manley and McIntosh 2002; Morgan and McIntosh 2005) but, as previously noted, also confirms that there are notable exceptions to a simple age progression. Except for Yellowstone (Doe et al. 1982; Hildreth et al. 1991; Christiansen 2001; Bindeman and Valley 2001a, b; Vazquez and Reid 2002), evolutionary details remain sparse for these older eruptive centers, although recent tephrochronology studies of SRP rhyolite tuffs and ashfall deposits (Perkins et al. 1995, 1998; Perkins and Nash 2002) provide improved chronology for the regional stratigraphic framework.
New 40Ar-39Ar dates for west and south-central SRP rhyolite units
K/Ca ± 2σ
Age ± 2σ (Ma)
27.0 ± 6.0
12.67 ± 0.08
26.9 ± 6.2
12.64 ± 0.06
13.2 ± 2.2
11.56 ± 0.07
15.5 ± 3.6
11.22 ± 0.07
14.6 ± 4.6
10.82 ± 0.06
Near House Ck.
14.3 ± 2.7
10.80 ± 0.06
Ross Pasture Ck.
12.2 ± 9.0
10.75 ± 0.07
Cedar Tree rhy.
17.2 ± 6.3
10.16 ± 0.09
Bruneau Jasper rhy.
15.8 ± 3.1
9.50 ± 0.06
Rogerson-Twin Falls Region
13.7 ± 3.1
10.91 ± 0.07
Browns Bench Escarp.
19.8 ± 16.5
10.22 ± 0.09
Browns Bench Escarp.
Rabbit Springs ignim.
19.1 ± 3.7
10.37 ± 0.13
Western SRP Region
Jump Ck., Shares Snout mbr.
21.6 ± 3.0
11.69 ± 0.06
Jump Ck., Buck Mountain mbr.
0.3 ± 0.1
11.56 ± 0.25
Reynolds Creek rhy.
21.3 ± 8.3
11.48 ± 0.09
Reynolds Ck. Can.
Wilson Creek ignim.
30.8 ± 10.6
11.42 ± 0.08
near Wilson Ck.
Wilson Creek ignim.(?)
40.2 ± 11.7
11.41 ± 0.05
Hill 2597, Owyhee Front
Wilson Creek ignim.
21.3 ± 2.2
11.34 ± 0.11
Upper Browns Ck. ignim.
30.6 ± 4.5
11.20 ± 0.02
SW of Oreana
Cerro Otono, Hill 2471
61.3 ± 14.5
11.14 ± 0.03
Hill 2471, Owyhee Front
Cerro Otono, Hlll 3036
56.3 ± 13.4
11.03 ± 0.07
Hill 3036, Owyhee Front
Older Rhyolite Units, Owyhee Mtns.
Swisher rhy., Poison Ck.
14.2 ± 8.0
14.21 ± 0.11
SW of Grand View
Silver City Rhyolite
14.6 ± 4.2
16.66 ± 0.08
Silver City Range
Silver City Rhyolite
13.2 ± 3.8
16.33 ± 0.12
Silver City Range
Where direct comparison can be made with previously available age information, the new ages generally agree within the stated analytical uncertainties (typically less than 0.1 Ma). For the earliest CPT ignimbrite (CPT III), two samples from the BJ area give ages (12.64, 12.67 Ma) that are indistinguishable from that reported (12.67 Ma; Perkins et al. 1995) for a correlative sample from Bruneau Canyon. The three determinations average 12.66 ± 0.02 Ma.
Perkins et al. (1995) infer ages of other CPT units from analyses of distal ash deposits from the Trapper Creek (TC, Fig. 1) section east of the Cassia Mtns.—more than 100 km east of our sample localities. For example, they report an age (10.94 Ma) for tephra they correlate with the tuff of Big Bluff (Cassia Mtns.) as well as, possibly, the penultimate CPT ignimbrite (CPT XIII). We report a similar age for one of the BB ignimbrites (BBU-7; 10.91 Ma). However, our three samples of CPT XIII from the BJ area give slightly lower ages (10.75, 10.80, 10.82; avg. = 10.79 ± 0.04 Ma), raising the possibility that multiple ignimbrites erupted over a short time span. On the other hand, these dates are indistinguishable within their respective analytical uncertainties. The new ages confirm the earlier chronology of Perkins et al. (1995, 1998). Other dates are discussed in a stratigraphic context in Appendix 6.
Along the Owyhee Front (OF; Fig. 1) and northern flank of the WSRP graben, several rhyolite units erupted, mainly as lavas, between 11.7–11.0 Ma (cf. Bonnichsen et al. 2004; Wood and Clemens 2002; Table 1). They clearly post-date activity in the nearby Juniper Mountain center, and are difficult to relate to a simple time-transgressive model for onset of SRP silicic magmatism. Although these rhyolites are compositionally distinct from the CPT ignimbrites, activity in these two contiguous areas clearly overlapped in time. Remarkably, WSRP eruptive activity falls within the time span of the three largest eruptive and collapse episodes in the BJ center (ca. 11.6, 11.2, and 10.9 Ma), suggesting the intriguing possibility that eruptions from widely separated and compositionally distinct reservoirs could have been triggered by related factors (e.g., episodes of faulting). Moreover, the unusual biotite-bearing Arbon Valley tuff (AVT) also erupted from vents in the ESRP (ca. 10.2 Ma; Kellogg et al. 1994) contemporaneous with waning stages of explosive volcanism in the BJ area—in this case, significantly earlier than postulated migration of its source region (Pocatello area) over the Yellowstone hotspot. Although different in terms of composition and eruptive volume, these examples demonstrate that silicic magmatism was widely distributed (over 400 km!) across southern Idaho between 10–12 Ma.
Published age determinations (adjusted to currently accepted decay constants), albeit often of lesser precision, further constrain timing and correlations of the units described in this paper (cf. Armstrong et al. 1975, 1980; Hart and Aronson 1982; Honjo et al. 1986; Clemens and Wood 1993). For example, in the MBH, available dates (11.0 ± 0.5, 11.0 ± 0.6 Ma) for rhyolites of the Mount Bennett group are consistent with the reversed magnetic polarity of most units. With reference to the geomagnetic polarity time-scale of Cande and Kent (1995), these units likely erupted during chron 5r (11.1–12 Ma) and so are slightly older than the indicated ages (albeit within the analytical errors). Rocks of the Danskin group and most eastern MBH rhyolites have normal polarity consistent with assignment to chron 5n (9.8–11.1 Ma); equivalent units are overlain in the eastern MBH by the City of Rocks rhyolite (9.15 ± 0.13 Ma; chron 4An). Thus, the entire MBH rhyolite section likely was emplaced between ca. 11.5–9.0 Ma.
Compositions of representative Cougar Point Tuff, Bruneau-Jarbidge, and Mount Bennett Hills rhyolites
12.66 ± 0.02
11.81 ± 0.03
11.22 ± 0.07
10.82 ± 0.06
9.50 ± 0.06
9.15 ± 0.13
XRF trace elements (ppm)
INAA trace elements (ppm)
Composition and time (CAT) groups
Provisional temporal relations have been established for most well characterized rhyolite units from the BJ, RTF, and MBH subareas of the CSRP for which there is both detailed stratigraphic control and fairly extensive analytical data for most units. To examine temporal trends in composition, we rely on unit averages for the major rhyolite units (Appendix 4) based on nearly 460 individual major and over 280 trace element analyses (Appendix 3).
Composition and time (CAT) groups, CSRP rhyolite units
Age Range (Ma)
Grasmere Escarpment and Jacks Ck areas
Bruneau-Jarbidge and area to SW
Rogerson-Twin Falls and Cassia Mts
North margin of SRP (Mt. Bennett Hills, etc.)
Shoshone Falls lf
Magic Reservoir lf, ign
Sand Springs ign
Dorsey Creek lf
Greys Landing ign
Lake Hills 4 ign
Juniper-Clover Area lf
McMullan Ck 2–4 ign
Three Creek ign
Castleford Crossing ign
City of Rocks rhy
Balanced Rock lf
Horse Butte Area rhy
Horse Basin lf
Sheep Creek lf
Up Crows Nest lf
Lake Hills 1–3 ign
Poison Creek lf
Perjue Canyon lf
Bruneau Jasper lf
Dry Gulch ign
King Hill Creek III ign
Tigert Springs lf
Indian Batt lf
BBU 10–12 ign
Dempsey Meadows rhy
Triguero Homestead lf
Browns View ign
Gwin Springs ign
Long Draw lf
Wooden Shoe Bu ign
King Hill Creek II ign
Cold Springs VI rhy
Cedar Tree lf
House Creek ign
Thorn Creek rhy
BBU 9 ign
King Hill Creek I rhy
Rabbit Springs ign
Cold Springs rhy
Low Crows Nest ign
CPT XV ign
BBU 8 ign
Rattlesnake/High Spr rhy
Jackpot Mem 7 ign
Steer Basin ign
CPT XIII ign
BBU 7 ign
Frenchman Spr/Dive Ck rhy
Jackpot Mems 1–6 ign
Big Bluff ign
Marys Creek lf
CPT XII ign
Bennett Mtn ign
Black Rock Escp lf
Fir Grove ign
Grasmere Escarp. ign
CPT XI ign
BBU 5–6 ign
Windy Gap rhy
Deer Springs ign
Magpie Basin ign
Unnamed Rev Pol rhy
CPT IX ign
BBU 4 ign
Unnamed Nor Pol rhy
CPT VII ign
BBU 3 ign
Willow Creek rhy
Crab Creek ign
Halfway Gulch lf
O X Prong lf
Rattlesnake Ck rhy
Badlands-Poison Ck rhy
CPT V ign
BBU 2 ign
CPT III ign
BBU 1 ign
Rattlesnake Dr ign
Whiskey Draw ign
Coherency of the CAT groups is most robust in the first phase of CSRP volcanic activity, prior to onset of regional rifting (ca. 9.5 Ma) that is marked by the earliest eruptions of basaltic lavas. During this early period the character of individual CAT groups oscillated between more or less felsic compositions, whereas later, both felsic and relatively mafic rhyolites are represented in certain time intervals. In such cases, these are designated as ‘A’ or ‘B’ CAT groups, respectively (Table 3), and the respective rocks generally occur in geographically distinct regions—likely signifying development of more diverse reservoirs or melt lenses within the crust. Paleomagnetic polarity is usually consistent for all units in a particular CAT group, although exceptions occur (e.g., CAT group 6).
Table 3 shows units assigned to each CAT group within the main geographic regions discussed in Appendix 6. Most CAT groups include units that are widely distributed across the CSRP; in some cases (e.g., CAT group 5) units of similar composition are found in all subareas. This is unsurprising for large ignimbrite outflow sheets, but is notable when both ignimbrite and lava flow units are included (e.g., CAT group 6). In the latter case we surmise that eruptive style or depositional facies varied laterally within a restricted time interval, but that the magma system being tapped was relatively homogeneous over wide distances or, if different magma reservoirs were tapped, they somehow evolved in a consistent fashion.
We emphasize that our definition of CAT groups is strictly empirical and does not imply that all units within any particular group are exactly correlative or precisely cogenetic. It is likely that some ignimbrites are the distal equivalents of others in the same CAT group, or that some successions of ignimbrites or lavas in the same CAT group may have been sequentially erupted from the same (or similar) magma batch(es). For example, CPT XI and the Grasmere escarpment ignimbrite (CAT group 5) occur in adjacent areas and probably represent deposits of the same eruptive event. On the other hand, the superposed CPT XII ignimbrite and Black Rock Escarpment lava flow (CAT group 6) have similar geochemical attributes and probably represent sequential eruptions from a common (or related) magma batch(es).
Because of incomplete exposure or preservation, or due to burial by younger rocks, the full extent of the large ignimbrites has not been established. Nor has detailed unit-by-unit stratigraphic correlation across the region yet been achieved. In the following discussion, we use CAT groups as stratigraphic proxies to overcome these limitations and to estimate broad-scale CSRP rhyolite volumes and temporal variations of magma generation and eruption rates. Using this approach we have developed a model for the evolution of CSRP volcanism and accompanying tectonism (detailed below) that allows us to compare the pattern of CSRP volcanism with that elsewhere in the SRP-Yellowstone hotspot track. This model also serves to guide future stratigraphic investigations that will hopefully lead to refinements in our understanding of SRP volcanic evolution.
Evolution of rhyolitic volcanism in the central Snake River Plain
During stage A (13.0–11.7 Ma, CAT groups 1–3, Fig. 4a), volcanism was concentrated in the western part of the region, and culminated with large eruptions that formed the voluminous CPT VII ignimbrite unit and its probable correlatives to the east (BBU-3), west (Buckhorn ignimbrite) and north (Willow Creek rhyolite). The combined volume of these units suggests that significant crustal subsidence occurred, probably resulting in formation of now-buried calderas.
During stage B (11.7–11.0 Ma, CAT groups 4–6, Fig. 4b) widespread rhyolite volcanism in the western and middle parts of the CSRP formed the voluminous CPT XI ignimbrite unit in the BJ region and its equivalents (the Grasmere escarpment ignimbrite and Windy Gap rhyolite) in the west and north. Stage B ended with the relatively mafic units of CAT group 6 and includes extrusion of the first documented rhyolite lavas in the BJ region (the Black Rock Escarpment and Marys Creek flows). The voluminous ignimbrite eruptions of CAT group 5 undoubtedly resulted in significant subsidence, and possibly caldera formation, in the western and middle part of the CSRP. The relatively mafic CAT group 6 rhyolites erupted at the end of stage B could represent continued evolution of the earlier magma system, following new recharge of the system by basaltic magmas. Stage B is coeval with eruption of genetically unrelated rhyolites along the Owyhee Front and southwestern margin of the WSRP graben (Bonnichsen et al. 2004).
During stage C (11.0–10.4 Ma, CAT groups 7, 8A, 8B, Fig. 4c) the zone of active volcanism spread eastward, producing enormous eruptions that formed the tuff of Big Bluff in the Cassia Mtns., and Jackpot ignimbrite members 1–6 (BBU-7 unit) in the Rogerson graben. The source region for these units likely is centered in the vicinity of Hollister and Rogerson, just west of the Cassia Mtns.—an area that is now buried under basalt flows. The CPT XIII ignimbrite may be slightly younger; although this unit was distributed farther west, its composition is similar to other magmas of CAT group 7, and we consider them to be related genetically. The compositionally similar Henley and Frenchman Springs rhyolites (WBM) also belong to CAT group 7. Because they were erupted respectively before and after faulting that separates the older Bennett Mountain and younger Danskin Mountain groups, their eruption brackets a time of caldera development and/or major subsidence in the northwestern part of the CSRP. Stage C closed with eruption of numerous ignimbrites from CAT groups 8A and 8B. All of these except CPT XV (CAT group 8B) are more mafic than the voluminous ignimbrites of CAT group 7, and they may represent a more mafic and hotter continuation of that magmatic phase. The timing of major ignimbrite eruptions during stage C suggests that the source calderas formed in a piecemeal fashion and ultimately resulted in major crustal subsidence.
During stage D (10.4–9.5 Ma, CAT groups 9, 10A, 10B, Fig. 4d) rhyolitic volcanism was broadly distributed across the CSRP. In the BJ region, volcanism switched from an explosive to an effusive mode, whereas a number of ignimbrites were erupted in the eastern part of the CSRP. Stage D coincides with a dramatic decrease in the overall eruption rate for silicic magmas. The earliest basalts intercalated within the rhyolite succession also appear at the end of stage D in the BJ (cf. Appendix 6); basalts of similar age also occur in the eastern MBH region (McHan basalt, Leeman 1982c), and as far east as the Malta Range (southeast of Twin Falls; Armstrong et al. 1975).
Stage E (9.5–7.5 Ma, CAT groups 11, 12A, 12B, Fig. 4e) is represented by rhyolitic volcanism in many areas of the CSRP—but mostly concentrated in the east. Both ignimbrites and rhyolite lava flows formed during this interval. Particularly notable are the very large and comparatively mafic rhyolite lavas erupted in the BJ region, even though the total volume of rhyolite erupted and the eruption rate were diminished considerably compared to earlier stages.
During the final phase of CSRP rhyolitic volcanism, stage F (7.5 to 5.5 Ma, CAT group 13; Fig. 4f), the focus of volcanic activity shifted entirely to the east, where comparatively small ignimbrite and lava flow units were erupted. Following stage F, rejuvenated silicic magmatism of the MREC is conspicuously anomalous with regard to the generally eastward-younging trend of the SRP-Yellowstone province silicic magmatism.
Huge rhyolite volume and catastrophic crustal foundering
The subsided interior of the CSRP region occupies about 14,000 km2. We believe it contains a fairly continuous rhyolite accumulation that is partly obscured by younger basalts and sediments. By analogy with Yellowstone, the subsided material likely includes thick deposits of intracaldera ignimbrites and lavas filling depressions that formed above partially emptied magma chambers. Nearly all of the rhyolite eruptions appear to have originated from sources in the subsided interior of the CSRP, whereas extracaldera ignimbrite outflow sheets and airfall tuffs dominate exposures in the flanking areas. The BRE, BB, WBM, and Cassia Mtns. sections expose ignimbrite successions on the order of at least 500 m in thickness. Thus, to estimate the minimum (i.e., ignoring intracaldera fill) rhyolite volume erupted from the 14,000 km2 subsided zone we have assumed an average thickness of 500 m. The cumulative volume of all CSRP units is estimated conservatively to be at least 7000 km3, most of which erupted in a little more than two million years (12.7–10.5 Ma) with recurrence intervals of roughly 200–300 ka. Inclusion of the volumes of dispersed airfall tuff deposits (that could constitute more than 30% of the mass of explosive eruptions; Mason et al. 2004), regionally correlative ignimbrites, and intracaldera fill, if known, would significantly increase estimates of the cumulative volume of material erupted from the BJ region, perhaps by 50 to 100%. From studies of tephras correlative with the CPT, Perkins and Nash (2002) estimate a cumulative eruptive volume on the order of 104 km3.
Finally, another important facet of CRSP magmatism concerns evidence for magmato-tectonic subsidence and crustal deformation. It is clear that most of the CSRP underwent significant subsidence in concert with, and possibly punctuated by, a series of episodic ignimbrite eruptions, as evidenced by major normal faulting cited earlier (e.g., at WBM, BB, and Rogerson graben areas). Of particular note are (1) the outbreak of compositionally distinct rhyolites in the WSRP graben essentially coeval with the period (11.7 to 11.0 Ma) of most intense activity of the CSRP magmatic system, and (2) eruption of the 10.2 Ma Arbon Valley tuff and related rhyolites in the ESRP (Kellogg et al. 1994) coinciding with major eruptions at the end of the CSRP ignimbrite flare-up (CAT group 9). Considering that CSRP activity postdates the initial manifestation of regional magmatism by several million years (and thus is unlikely to be associated with a plume head), the contemporary residence of multiple silicic magma bodies within the crust over horizontal distances of hundreds of kilometers is enigmatic in the context of the hotspot model. Assuming that crustal melting is driven by intrusion of basaltic magma, then regionally extensive basalt production is implied rather than localized, time-transgressive production. Also, the coincidence of regional rhyolite eruptions with major ignimbrite events in the CSRP suggests that all may be triggered by a common factor—perhaps, activation of a system of normal faults or triggered by major eruptive events. If so, this, too, seems inconsistent with a simple eastward progression of Basin and Range style extension at rates comparable to passage of North America over the Yellowstone hotspot, as proposed by Rodgers et al. (1990). Prevailing N- to NW-trending orientations of the principal syn-volcanic normal faults in this region suggest that the crust was extended primarily in a SW–NE direction—essentially parallel to the axis of the hotspot track. We propose that this deformation style reflects a combination of regional extension and crustal weakening due to injections of enormous volumes of mantle-derived basaltic magma and large-scale partial melting of crustal rocks. Widespread rhyolite outbreaks could have been triggered by catastrophic foundering of the weakened crust (cf. Elkins-Tanton and Hager 2000). Further support for this type of scenario is provided by petrologic considerations.
Compositional evolution of CSRP silicic volcanism
Here we review temporal variations in compositional and petrologic evolution of CSRP silicic magmatism in the context of our inferred regional stratigraphy. Discussion is focused mainly on the BJ and RTF areas, where stratigraphic relations and timing are best understood, and on the MBH where partly equivalent units have been studied in some detail. Ages are based on available radiometric dates combined with conservative interpolations for intervening units as constrained by stratigraphic correlations between sections. Although sequences of eruptive units from these areas are considered to be representative, we recognize that stratigraphic sections may be incomplete due to limited exposure or restricted areal distribution of some units. Also, particularly for long-traveled ignimbrites, local stratigraphic sections could include units derived from multiple sources having different petrogenetic histories. Despite these limitations, such sections provide useful insights into large-scale magma reservoir processes as well as constraints on petrologic scenarios for magma production.
Compositional data used in this paper are based mostly on XRF (some 355 major element and 175 selected trace element) analyses of multiple samples from the described BJ and RTF units; all but the lowermost RTF units and one BJ unit are represented by at least three major element analyses. Data for MBH rhyolite units are based on combined XRF (n = 39) and inductively coupled plasma optical emission spectroscopy (ICP-OES) analyses (n = 65). Precision and accuracy of these data are comparable based on replicate analyses of standards and multiple samples from common units (Appendices 3, 4). Analyzed samples are mostly whole-rock vitrophyres representing rapidly quenched magmatic material; such samples were obtained from basal parts of ignimbrites and from basal and upper parts of lavas. Where vitrophyres were not preserved, devitrified interior samples were analyzed; in general, compositions of the two types of samples are similar for any given unit. Averages and standard deviations for all well-studied stratigraphic units are given in Appendix 4, with major elements normalized to 100% on an anhydrous basis and all iron reported as FeO*; with exception of a few probably altered samples, all available data are included in these averages. These averages are used in compositional diagrams unless otherwise noted. Statistics presented in Appendix 4A demonstrate that overall compositional variability (RSD, or coefficient of variation) within individual lava flow units is generally comparable to analytical precision; data for ignimbrites tend to have RSDs about twice analytical precision on average for most reported elements. Compared to literature data for many ignimbrites (e.g., Hildreth 1979, 1981), CSRP units are relatively homogeneous (see below). However, welded basal airfall ash occasionally is distinctive from the overlying bulk ignimbrite (e.g., CPT III, CPT V, CPT XV); such analyses are excluded from the unit averages and treated separately. For a representative subset of BJ samples, the XRF data are complemented by more comprehensive trace element analyses using combined instrumental neutron activation and ICP-OES methods (Table 2). Sr and Nd isotopic analyses were obtained on these samples using thermal ionization mass spectrometry; these data are corrected for isotopic fractionation and reported relative to 87Sr/86Sr = 0.71025 for NBS-987 and 143Nd/144Nd = 0.51184 for Johnson–Matthey Nd and 0.51264 for BCR-1 standards. Published Nd isotopic data for CPT tuffs (Cathey and Nash 2004) were normalized to the same standard values to enable direct comparison with our results. Initial ratios are reported based on measured or inferred ages of each sample. Available δ18O data for feldspars or quartz in BJ area rhyolites show them to be variably depleted in 18O compared to most rhyolites worldwide (Boroughs et al. 2005); representative δ18O values from their work and one new datum for a MBH rhyolite are listed in Table 2.
Electron microprobe analyses of phenocryst minerals of representative CSRP rhyolites have been used to calculate two-feldspar, two-pyroxene, and Fe–Ti oxide mineral temperatures based on analyses of contiguous coexisting phases that are most likely to have attained equilibrium (Appendix 5; data references given therein). The two-pyroxene thermometers tend to produce the highest and most consistent temperature estimates (cf. Honjo et al. 1992); these were used to ‘re-tune’ the Zr-saturation thermometry of Watson and Harrison (1983), which in turn was applied to infer magmatic temperatures using average compositions for all sampled eruptive units. These values correlate well (not shown) with total FeO* in the rhyolites. Thus, FeO* is a useful and simple proxy for both relative magmatic temperature and degree of evolution.
Overview of compositional variations
Reliance on whole-rock samples to represent magmatic compositions for silicic ignimbrites can be problematic for several well-documented reasons (cf. Hildreth 1979, 1981; Streck and Grunder 1997, 2007) including: presence of initial compositional variations (e.g., tapping of compositionally zoned magma systems), heterogeneities related to eruption and deposition (e.g., crystal winnowing or concentration, incorporation of lithic clast impurities, etc.), or post-eruptive alteration (e.g., alkali exchange with meteoric fluids). Ideally, vitric pumice clasts are considered to be most representative of magmatic samples, but such clasts are rarely found in SRP ignimbrites–presumably due to high eruptive temperatures and extensive welding in most units. Alternatively, glass matrix from flows or shards from airfall deposits can be used to estimate melt composition. Such data suggest that several CSRP ignimbrites exhibit weak compositional zoning from base to top (though not in any consistent sense), and that glasses typically are slightly more evolved (e.g., higher SiO2, lower FeO*) than equivalent bulk-rock samples (Honjo 1990; Cathey and Nash 2004). In detail, some units carry multiple compositionally distinct populations (or ‘modes’) of glass shards–signifying (syn-eruptive?) extraction and incomplete mingling of magmas from multiple distinct or zoned reservoirs at depth (Cathey and Nash 2004; Andrews et al. 2007). The complexity of CSRP ignimbrites also is recorded by the presence of multiple ‘modes’ of phenocryst mineral populations, occurrence of xenoliths and/or xenocrysts derived from magmas as mafic as basalt, diverse glomerocrysts presumably derived from silicic magmas at differing stages of evolution, and rare lithic clasts of diverse crustal origins (Bonnichsen and Citron 1982; Honjo et al. 1992; Wright et al. 2002; Cathey and Nash 2004). Although such complexities must ultimately be factored into interpretation of the petrogenesis of CSRP rhyolites, this task largely awaits more detailed studies.
Temporal compositional trends
A second key point is that, between 12.7 and 9 Ma, the magma system as a whole tended to become more mafic (or less evolved) over time, with compositions gradually varying from relatively high- to low-SiO2 rhyolite. This evolutionary trend is clearly defined by increasing TiO2 with time. TiO2 is positively correlated with FeO*, MgO, CaO, Zr, Sr, and Ba, and inversely correlated with SiO2, K2O, and Rb. Superimposed small-scale fluctuations within the general progression suggest variations in magma chamber or magma supply dynamics. For example, following the large CPT VII and XI eruptions and again near 10.5 Ma, the system followed small excursions back toward higher-SiO2 rhyolite. Between 9 and 8 Ma (not precisely dated), we note a marked shift to relatively evolved rhyolite; rhyolites younger than 8 Ma (and mainly restricted to the RTF area) resume a trend toward less evolved compositions similar to that seen earlier. This part of the record could correspond to a waning phase in the CSRP system or, alternatively, the younger RTF units may sample a different magma system having a distinct evolutionary history.
Considering the heterogeneity in mineralogy, shard chemistry, etc. observed at the hand sample scale (Cathey and Nash 2004), it seems unlikely that these magmas tapped an evolving well-mixed single reservoir. An alternative scenario is that the magma system consisted of a plexus of melt lenses distributed over a wide region in the crust, each of which evolved in parallel over time—producing an array of similar though not identical compositions at any given time. Eruptions apparently triggered extraction and intimate mingling of magmas from multiple lenses to simultaneously produce broadly similar eruptive products characterized by small-scale heterogeneities. The physical processes by which this occurred are unclear and deserve further scrutiny.
The similarity of long-term compositional changes in eruptive products over a regional length-scale appears to record significant evolutionary patterns in the CSRP ‘magma system’ proper. Such trends are reasonably attributed to competing effects of recharge by less evolved magma vs. differentiation within the magma system. For example, coherent upsection increases in TiO2 and decreases in SiO2 (e.g. between 12.7–9.0 Ma) are inconsistent with progressive fractional crystallization (FC) of a simple evolving magma reservoir, but could reflect either recharge of less evolved magmas into the system or fractional melting of progressively more refractory source volumes over nearly 4 m.y. Such trends are commonly observed within zoned ignimbrites (e.g., Bishop Tuff) that, over the geologically instantaneous time scale of eruptions, presumably tap deeper, less evolved magma from compositionally stratified reservoirs (e.g., Hildreth 1979, 2004). However, this explanation seems unlikely for the long-term CSRP trends because (1) the time scales involved are much greater than expected residence times (perhaps on the order of 105 years) for most silicic magma bodies (cf. Vazquez and Reid 2002), and (2) the overall variations are significantly larger than those within individual CSRP eruptive units. Notable reversals in the long-term trend (e.g., following the large CPT VII or CPT XI eruptions, or recorded by rhyolite lavas erupted between 9–8 Ma) are consistent with crystallization of magma reservoirs over long repose intervals (ca. 250 ka) between ignimbrite pulses. Such reversals could record episodes of cooling owing to waning rates of basalt influx into the crust.
The major element temporal patterns are mimicked by trace element variations (Fig. 8). Ba, Sr, Zr, and Rb contents define coherent fluctuations with time for the three areas. Notably, MBH samples appear to be slightly enriched in Ba (as reflected in both ICP and XRF analyses), yet concentrations of other elements are comparable to those in BJ/RTF rhyolites of similar age; we tentatively suggest that this could reflect small lateral variations in either source compositions or reservoir processes. With decreasing age, BJ area rhyolites become progressively depleted in Rb (280 to 150 ppm), and enriched in FeO* (1.1 to 4.7%), Ba (<250 to 1250 ppm), Sr (<20 to 140 ppm), Zr (300 to 610 ppm); 87Sr/86Sr and 143Nd/144Nd ratios also increase (see below). On a short time-scale all these parameters fluctuate noticeably—particularly during the early history of the system—and it is not always possible to correlate the variations in detail across the entire region. However, significant trend reversals are common to all sections within the precision of our stratigraphic control. Between ca. 12.7–9 Ma, Ba/Y (<10 to 24), Ba/Nb (<10 to 30), Zr/Nb (<8 to 13.5), Ba/Rb (<1 to 7.5), and K/Rb (<200 to 290) ratios exhibit general upsection increases, whereas K/Ba (ca. 250 to 30), Nb/Y (1.0 to 0.6), K/Sr (ca 3000 to 250)*, Rb/Sr (ca. 18 to 1)*, Pb/Nb (0.69 to 0.47)*, Rb/Y (4.5 to 2.2)*, and Th/Nb (0.75 to 0.45)* generally decrease, with significant trend reversals or oscillations at particular time horizons (Fig. 9 shows representative examples of these trends). In general, all of these patterns are consistent with the existence of a dynamic magma system that became progressively less evolved with time. Trends for element ratios marked above by (an asterisk) show small reversals consistent with a transient increase in the proportion of evolved magma between 9 and 7 Ma; such excursions are consistent with evolution of stored magma during lulls in the recharge sequence (although other explanations may be viable). Overall, these compositional trends suggest that the magma system (sensu latu) was open to short-term communication between differently evolving silicic magma reservoirs (more vs less evolved), and/or recharge by mantle-derived basaltic magma (Hildreth et al. 1991).
Possible origins of voluminous early rhyolites
Implied sources of rhyolite magmas
First-order constraints on degree of melting
To model the physical process of CSRP silicic melt production, an estimate of degree of melting is required. Assuming ‘average crust’ (Taylor and McLennan 1985) to be a valid source, the extent of melting can be constrained from the maximum rhyolite/crust enrichment factors. These are highest for U and Th (∼9–10 for most CPT samples and 7–8 for most BJ lavas; Fig. 12). If bulk mineral/melt partition coefficients approach zero for these elements (cf. Bea 1996), simple batch melting can produce the observed enrichments with 10–15% melting. The fact that enrichment factors for the younger lavas are lower than those for the earlier ignimbrites suggests that degree of melting could have increased with time. Also, these are likely minimal estimates because (1) enrichment factors could be enhanced by FC processes, and (2) distribution coefficients could be enhanced by the presence of accessory minerals (Bea 1996) or, for elements like Sr, Eu, and Ba, affinities for feldspar minerals. For example, enrichment factors (ca. 4–5) for the other incompatible elements can be explained by the same model if bulk distribution coefficients for those elements are slightly higher (∼0.10–0.15). For any given model, assuming a less fertile source (e.g., ‘average lower crust’) will require lower melt fraction, and vice versa. Further consideration of magmatic differentiation effects and more complex melting models will be presented elsewhere.
Temperature and composition relations
The absence of hydrous phenocrysts in most SRP rhyolites and available mineral thermometry (Appendix 5) indicate that SRP rhyolites are unusually hot and water-undersaturated magmas. Minimum temperatures exceeding 950°C are indicated by pyroxene thermometry in many CSRP samples (Honjo et al. 1992), and Zr-saturation thermometry suggests that most CSRP rhyolites had magmatic temperatures above 850–900°C. The general enrichment of FeO* with time fundamentally signifies that progressively more mafic and hotter rhyolites were produced by fractional melting of the crust. This scenario would be favored by repeated injection of basaltic magmas into zones of crustal melting, in which case successive melts would be more depleted in incompatible elements and enriched in compatible ones—consistent with the overall temporal variations defined by CSRP rhyolites. The observed small oscillations on these trends signify that, in reality, melt production is more complex and may involve other factors, including migration of melt generation zones within the crust, mingling of magmas produced over large crustal domains, heterogeneities in the crust, etc.
A physical model for SRP silicic magmatism
Scale of melt production and tectonic implications
Scenarios for estimating magma supply in the Y-SRP province
Silicic magma production
YP-extrusive rhyolite (scenario 1)
CSRP-minimum (scenario 2)
CSRP-likely (I:E=2) (scenario 3)
Duration of eruptive center (Ma)
Estimated volume of rhyolite magma (km3)
Average rhyolite supply rate (km3/year)
Source volume assuming 15% partial melting (km3)
Thickness (km) of melt zone (radius = 50 km)
Basaltic power source
Volume of basalt (km3) needed to heat crust & melt rhyolite source (assuming minimal basalt/rhyolite volume ratio = 2)a
Average (minimal) basalt supply rate (km3/year)b
Thickness (km) intruded (radius = 50 km)
Rate of layer thickening (mm/year)
If total rhyolite production in the CSRP is taken as 104 km3 (over 4 m.y.), the corresponding volume of basalt required is at least twice that amount. Taking the Yellowstone Plateau as an analog for horizontal scale, the underlying magmatic system could have a diameter up to about 100 km, which may be reasonable for the BJ center. This scale is also consistent with lateral extent of an inferred mafic sill complex beneath the ESRP (Rodgers et al. 2002). For basalt intrusion into a cylindrical domain of such diameter, and given the parameters in scenario , an equivalent thickness of about 2.6 km of basalt must be injected into the crust. The time-integrated magma influx rate is at least 0.01 km3/year—i.e., similar to the integrated rate (0.017 km3/year) estimated for Hawaiian magmatism (Crisp 1984; Robinson and Eakins 2006).
These estimates will increase to the extent that a significant fraction of the rhyolitic magma is intruded or that basaltic intrusion is confined to a smaller areal footprint. For example, assuming a conservative intrusive:extrusive ratio of 2:1 (cf. Annen and Sparks 2002; our scenario ), the total rhyolite volume would be about 3*104 km3, and the corresponding lens of basalt would be about 7.6 km thick. This calculation is likely to be an underestimate considering that the inferred mafic lens beneath the ESRP is several kilometers thicker (cf. Peng and Humphreys 1998; Rodgers et al. 2002). Annen and Sparks (2002) present a more detailed analysis in which it is demonstrated that, after an incubation time of up to 106 years, frequent small additions of basalt will maximize efficiency of secondary melt generation in the crust, and that crystalline differentiates of the basalt are likely to remelt along with country rocks to produce silicic magmas having a hybrid ‘crust-mantle’ composition. The petrologic implications of such models for the SRP will be explored elsewhere.
It is also instructive to consider effects of magmatism in terms of crustal modification within a zone of partial melting and melt migration. Assuming a melt fraction of 15% (minimal) and the cylindrical geometry used above, melting scenarios considered (Table 4) imply partial melt zone thicknesses ranging from roughly 4 to 25 km, the latter being a conservative figure for the CSRP. Even if melt fraction approaches 25%, scenario  predicts a vertical dimension for the zone of melting (15 km) approaching half the crustal thickness. Solidification of injected basalt and extraction of lower density rhyolitic magma will thus contribute to an overall densification of a significant volume of crust (Leeman 1982a) and may contribute to observed positive magnetic intensity and Bouguer gravity anomalies coincident with parts of the SRP (Mabey 1982). Below we elaborate on the physical significance of these calculations.
In passing, we note that similar volume considerations also effectively preclude FC of associated basaltic magmas as the origin of most SRP rhyolite. Such models require about 95% (or greater) FC to produce suitable daughter liquids (e.g., production of rhyolite with 150 ppm Rb from basalt with 5 ppm Rb requires a closed system F-value of 0.033, or ∼97% solidification). Using the same geometric assumptions as above, the corresponding volume ratio of basalt:rhyolite (approaching 20–30) implies a thickness (22–33 km) of basaltic cumulates exceeding half that of the present-day crust! This volume of basaltic magma far exceeds that required in the anatectic model.
Even if silicic magmas are produced solely by crustal anatexis, basaltic intrusion likely has resulted in significant reconstitution of the original crust. Notably, given that the Moho (base of crust) beneath the SRP is nearly flat at ∼40 km (Hill and Pakiser 1967; Sparlin et al. 1982; Peng and Humphreys 1998), intrusion of the requisite basalt volume creates a serious ‘room problem’. Second, seismic refraction studies (cf. Pakiser 1989; Smith and Braile 1994) indicate that the lower crustal layer (Vp = 6.7 km/s) is thicker beneath the WSRP (∼30 km) than under either the ESRP or Yellowstone (both ∼20 km); effectively, this translates to the presence of a higher velocity (i.e., denser) lower crust toward the west. A plausible explanation for this observation is that the density of the WSRP lower and middle crust has been selectively increased owing to greater inputs of basaltic magma (Leeman 1982a, 1989). The room problem may be alleviated by Basin and Range style extension (i.e., parallel to the ESRP axis or, equivalently, to the vector of North American plate motion), effectively elongating the SRP (cf. Leeman 1982a, 1989; Parsons et al. 1998). There is little evidence supporting significant NW-SE extension across the SRP (Peng and Humphreys 1998; Rodgers et al. 2002). Horizontal flow of warm and weak lower crust is also possible, but in itself will not accommodate the room problem.
In Fig. 14 we assume a constant strain rate (2% per m.y.) equivalent to that documented for the ESRP. Since 15 Ma, this corresponds to 30% (or about 150 km) cumulative extension, with significantly greater strain concentrated in the west-central SRP. To maintain a constant apparent crustal thickness, the model predicts that addition of at least 7.7 km of new crust must be added beneath the area (west-central SRP) that began to extend at ∼12 Ma; this agrees well with the above estimate (scenario ) of basalt addition required to produce CSRP silicic magmas. Also, according to this model, material with lower crustal seismic properties would appear to make up about 60% of the present-day crust, whereas the seismic estimate is closer to 75% (Hill and Pakiser 1967). This discrepancy also suggests that we have underestimated the actual basalt input into the crust. All of these perspectives can be brought into closer agreement if the input is about 1.5 times that calculated for scenario . In this case the room problem is more severe and requires greater extension—either at a higher constant rate (∼3%/Ma) or, if extension is non-uniform, selectively concentrated across the western and central SRP. Uncertainties in magma volumes, intrusive geometries, and other factors, allow considerable latitude in the model results, so we do not claim that they are unique—nevertheless, internal consistency between the various constraints considered point to the need for astounding modification of the crust. Clearly, the diachronous nature of lithospheric deformation and its relation to magmatic evolution of the SRP warrant further detailed investigation—particularly with regard to the feedback effects of magmatism on crustal strength and strain history.
In summary, this model can (1) accommodate significant input of basaltic magma leading to densification of the crust as well as extensive partial melting to produce SRP rhyolitic magmas, and (2) account for protracted inputs (and eventually throughputs) of basaltic magma over the duration of SRP magmatism – including an incubation period to generate the earliest rhyolites and extending to recent time over much of the province. Production of quartz-bearing silicic magmas having light oxygen isotopic compositions and other observed compositional features is consistent with melting at mid- to upper crust levels. The inferred heat source is contemporaneous injection of mantle-derived basaltic magmas that, upon cooling, form a mafic sill complex—i.e., imaged by receiver function studies at depths between about 10–20 km beneath the ESRP (Peng and Humphreys 1998), and implied to exist beneath much of the SRP on the basis of observed Bouguer gravity and magnetic intensity highs (Mabey 1982) and elevated heat flow (Blackwell and Steele 1992).
The dominant CSRP silicic magma system persisted for more than 4 Ma producing a volume of rhyolitic magma on the order of 104 km3. Broad distribution of multiple ignimbrite units and compositionally similar rhyolite lavas implies the existence of large reservoir systems that at times could deliver more than 1,000 km3 of magma. Whereas these magmas were relatively homogeneous at specific event horizons, they exhibit large and coherent compositional variation over the life of the magmatic system. In detail they typically comprise complex, sometimes multi-modal mineral assemblages that seemingly require contributions from multiple discrete magma lenses in the crust. The magma systems perhaps resembled a melt-filled ‘crustal sponge’ rather than large ‘convecting vats’. Parallel generation and storage in a crustal plexus (cf. Dufek and Bergantz 2005) could possibly overcome the problem of accumulating sufficient magma to feed the largest ignimbrite events (Jellinek and DePaolo 2003) but leaves open the question of how such magmas are efficiently extracted. In any case the extracted magmas appear to have mixed efficiently enough to produce relatively homogeneous eruptions with only small compositional zoning. High magmatic temperatures would lower melt viscosity and enhance mixing and homogenization. The models of Annen and Sparks (2002) imply that persistently high mineral equilibration temperatures (∼900–1000°C) for CSRP rhyolites require repeated injections (trickle feed?) of basaltic magma over protracted time intervals. Moreover a precursory incubation period (on the order of 105–106 years) may be required prior to onset of the earliest rhyolitic eruptive activity—in which case the CSRP crust could have received significant basaltic inputs as early as ∼13.5 Ma.
The eruption of distinct rhyolites over long horizontal distances (e.g., from western to eastern SRP at 11.7–10.2 Ma) implies the near-contemporaneous existence of discrete crustal melt and accumulation zones over distances of some 400 km. This implies widely distributed injection of basaltic magmas into the crust beneath much of the SRP well before 11 Ma. Also, intermittent continuation of basaltic magmatism to Pleistocene or younger time over much of the SRP is difficult to reconcile with a Hawaiian-style (‘point source’ or ‘plume tail’) hotspot model. We propose that both features are more consistent with tectonic control by lithospheric extension – as demonstrated for the Great Basin to the south (Harry and Leeman 1995).
Both the general eastward younging age progression of SRP silicic magmatism (and inferred mafic plutonism), and the persistence of this activity once initiated, could be related to Basin and Range style extension that, essentially, has lengthened the apparent hot-spot track. The apparent age progression is consistent with diachronous eastward migration of the onset of extension, coupled with sustained stretching once deformation began. A simple extensional model involving at least 2% stretching per m.y. (ca. 10 mm/y) can accommodate the volume of mafic magma needed to sustain silicic magmatism, while balancing crustal attenuation (i.e., maintaining near-constant crustal thickness), and simultaneously increasing lower crustal density and seismic P wave velocity to match inferred crustal structure.
What processes drive SRP magmatism? In our view, this question fundamentally concerns origin of the associated basaltic magmas. Considering the basaltic inputs inferred to sustain voluminous SRP silicic magmatism, the integrated melt production rate for the SRP clearly exceeds that associated with Basin and Range extension to the south or north. While this observation provides a compelling incentive for invoking a mantle hotspot model, the causes for excess melt productivity associated with the SRP are far from clear. For example, origin of the basalts can be viewed in terms of two competing generic processes, both of which are potentially important:  decompression melting of lithospheric mantle due to tectonic stretching (Harry and Leeman 1995), and  decompression melting of hot ascending asthenospheric mantle (McKenzie and Bickle, 1988). Of these, the first could possibly account for latitudinal differences in melt production – for example, if the lithospheric mantle is compositionally more fertile beneath the SRP. It could also better account for the timing and spatial pattern of mafic magmatism. The second process could result in excess melt productivity by promoting a local positive thermal anomaly in underlying mantle. As noted earlier, the style of SRP magmatism does not conform well to patterns expected for this process, but it remains to be seen what influence the continental lithosphere imposes on expressions of deeper magma production and ascent. These issues will be addressed in greater detail elsewhere (cf. Leeman 2005).
Although the silicic magmatism associated with the SRP broadly tends to decrease in age from SW to NE along the trace of the Yellowstone hotspot track, in detail there are important exceptions at several scales. In the CSRP, rhyolitic volcanism occurred sporadically throughout the entire area, and apparently was related to piecemeal foundering of the crust in a series of major, perhaps catastrophic, caldera-forming events that did not follow a simple diachronous pattern in time and space.
CSRP rhyolitic volcanism consisted of an early (12.7 to 9.5 Ma), very voluminous, caldera-forming stage and a later (9.5 to 5.5 Ma), less voluminous, rifting stage. Ignimbrite eruptions dominated the caldera-forming stage, whereas rhyolite lava flow eruptions dominated the rifting stage; in fact, both eruptive styles occurred during each stage. Basaltic eruptions were coeval with the second stage of this activity.
A regional stratigraphic framework is proposed based on physical stratigraphy, geochronology, and compositional variations. These data allow us to define composition and time (CAT) groups, in which ignimbrites and/or lavas erupted within a single time interval, over distances as great as about 102 km, have similar bulk compositions and mineralogy. Implicitly, rhyolites of a single CAT group provide a time-slice sample of the evolving regional-scale reservoir system. In particular, the 12.7 to 9.0 Ma CAT groups define a coherent general temporal variation from more to less evolved compositions and toward higher temperatures over time, albeit with small short-term reversals along the way.
These coherent patterns over great distances suggest that the underlying magma systems are of regional extent, and are consistent with either increasingly refractory crustal sources, increasing inputs of basaltic magma, or both, with time. These trends, and the volume scale of CSRP silicic magmatism appear to be unique among eruptive centers of the SRP-Yellowstone province.
The overall volume of the rhyolite erupted over about 7 million years was immense. It exceeds 7,000 km3, and possibly is well within the 10,000–20,000 km3 range considering that volumes of intracaldera fill and distal airfall deposits have not been included. An unknown, but perhaps larger volume of magma may have been retained as intrusions within the crust.
The rhyolitic melts were produced dominantly by partial melting of pre-existing crustal materials, as indicated by the persistent presence of restitic glomerocrystic aggregates, low-δ18O signature and other geochemical characteristics of CSRP rhyolites. Incompatible trace element data are consistent with at least 10–15% melting, assuming an ‘average crust’ source. Sr and Nd isotopic data resemble compositions of Idaho batholith granitoids, and seemingly preclude significant involvement of either Archean crust or dominant contributions from the coeval basaltic magmas or melts of their intrusive counterparts.
Intrusion of huge volumes of basaltic magma into the crust provided the energy to drive CSRP silicic magmatism. Significant crustal deformation was required to accommodate the volume of basalt needed for energy balance. We present a model that conceptually integrates basalt intrusion and compositional reconstitution of the crust with Basin and Range style extension. Lithosphere-scale extension could also contribute to production of the basaltic magmas.
Eruptions between 11.7–10.2 Ma of voluminous mineralogically and compositionally distinct rhyolites across much of the SRP, during the height of the CSRP ignimbrite flare-up, indicate that (1) significant basaltic injections into the crust were widespread prior to that time, and (2) magmatism did not follow a simple hotspot-like time-transgressive pattern as previously believed. Rather, basaltic magmatism likely was significant across the province from mid-Miocene to Quaternary time. Also, the respective 11.7–10.2 Ma rhyolites demonstrate that large silicic magma systems were at least partly coeval over a distance of some 400 km. We suggest that magmato-tectonic events, such as proposed for the CSRP, may have triggered activity in adjacent areas if silicic magma accumulations were primed to erupt.
We envision a magmatic system that at any given time involves a wide swath of crust that is occupied by a swarm of coeval but distinct melt lenses, and that these reservoirs mimic each other not only in composition but also in the pace and degree of their evolution. Large eruptions presumably extract magma from multiple discrete lenses rather than from a single zoned magma chamber. Syneruptive mixing may partly homogenize these melt ‘parcels’, but is inefficient to the extent that mineralogical and even glass shard heterogeneities are preserved on a small (cm) length scale. This model implies that magma bulk composition may partly depend on the length-scale of melt extraction (hence the total volume of an eruption), on the rate at which melt parcels can be extracted and commingled, on the heterogeneity and melting history of the crustal source rocks, and on thermal gradients across the melt zone that influence the degree of melting for any given lens.
Modeling of such a dynamic melting regime is a fertile research goal for the future. We also leave as an open question the details of how crustal deformation and magmatism are linked. To constrain this relationship, and to test the concepts presented here, will require additional information regarding (1) the distribution and magnitude of tectonic strain across the entire SRP and its flanks, and (2) the density distribution and seismic structure of the underlying crust. Also, investigations are needed to quantify how magmatism at the scale we envisage will modify the thermal structure and strength of the crust; such information may provide insights into processes of magma accumulation and eruption. Finally, although we propose that tectonic processes have a significant influence on SRP magmatism, we are reluctant to rule out entirely some linkage between sublithospheric processes and generation of the associated basaltic magmatism (cf. Humphreys et al. 2000; Yuan and Dueker 2005; Waite et al. 2006).
Our efforts leading to this paper benefited from the assistance of many people over some 30 years. The Idaho Geological Survey, Rice University, New Mexico Bureau of Mines and Mineral Resources, and Mount Holyoke College supported aspects of this research, and the U.S. Geological Survey and the Idaho Department of Water Resources contributed financial and logistical support. Leeman acknowledges grants EAR-8018580 and EAR-8320358 from the National Science Foundation and continued support by the Division of Earth Sciences since joining NSF. We thank M. Rhodes and P. Dawson (University of Massachusetts), D. Cornelius (Washington State University), and G. Fitton (University of Edinburgh) for help with XRF analyses. E. Pestana and M. Dehn (Rice University) assisted with ICP analyses and N. Rogers (Open University) provided neutron activation analyses. Help in the field and many discussions were provided by the ‘Friends of Rhyolite’: J. Bernt, D. Kauffman, M. Jenks, J. Wolff, S. Boroughs, P. Larson, G. Gillerman, M. Branney, T. Barry, G. Andrews, B. Ellis, J. Sumner, M. McCurry, S. Hughes, M. Perkins, D. Clemens, B. Hirt, C. Manley, G. Citron, T. Gregg, A. Semple, C. White, S. Wood, C. Henry, M. Ferns, A. Grunder, M. Cummings, M. Norman, J. Aranda-Gomez, G. Aguirre-Diaz, G. Labarthe, J. McPhie, N. Riggs, R. Smith, R. Christiansen, L. Morgan, and W. Hackett. Finally, we are indebted to Henny Cathey, Martin Streck, Derek Schutt, Barbara Nash, and Lina Patino whose reviews and comments helped us improve this paper. Of course, we take responsibility for any errors in interpretation and hope that, as research continues on SRP rhyolites, our ideas will be modified for the better.