High-silica rhyolites represent a significant endmember component of continental crustal evolution, and their genesis provides insights into magmatic differentiation processes. Two processes are often cited to explain the origins of these types of magmas: fractional crystallization of a mafic or intermediate parental magma (Bacon and Druitt 1988;Hildreth et al. 1991) or partial melting of the shallow crust Fort valley at 865 ka, then the Bailey Ridge obsidian flow in the Mineral Mountains at 851 ka, and followed by additional mafic lavas in Cove Fort valley. Eruptions continued in the Mineral Mountains with the 786 ka Wildhorse Canyon obsidian flow, and the extrusion of the "middle domes" between 766 ka and 758 ka. After a ~ 170 kyr hiatus, volcanism resumed with the eruption of the "high domes", including Little Bearskin Mountain dome (588 ka) and Bearskin Mountain dome (539 ka). Subsequent eruptions shifted to higher elevations, producing the other high domes of North Dome (520 ka), Ranch Canyon Dome (501 ka), South Twin Flat (483 ka), and North Twin Flat (undated). These eruptions were grouped into two eruptive phases of rhyolitic compositions: obsidians and "middle domes" from 850 to 750 ka, and "high domes" from 590 to 480 ka (Rivera et al. 2024). Mafic eruptions in Cove Fort valley continued until at least 417 ka (Rivera et al. 2024). Combining the geochronology with whole rock major, trace, and rare earth element geochemistry, Rivera et al. (2024) note that compositions become increasingly evolved over time, with eruption sites migrating towards higher elevations. Previous workers suggested these trends resulted from progressive differentiation of a shared magma body, with a composition similar to that of the Bailey Ridge obsidian flow (Evans and Nash 1978).
This study seeks to understand the genesis and evolution of the Pleistocene high-silica rhyolites and their connection to each other, the contemporaneous basalts and andesites of Cove Fort valley, and the granitic batholith. Fractional crystallization and partial melting models of whole rock and zircon geochemical data are used to assess potential relationships, which help to elucidate the nature of the petrogenetic origins of these rhyolites. The models provide no evidence for derivation via extreme fractional crystallization of the basalts or for a long-lived crystal mush or scavenging of batholith-related crystals. Partial melting of various granitoid lithologies followed by rapid ascent to the surface may explain the petrogenesis of the Quaternary rhyolites of the Mineral Mountains (Fig. 1).
Hand samples from seven rhyolite domes were collected and processed using standard mechanical, density, and magnetic techniques to separate zircon grains from the host rock. Zircon were handpicked, mounted in epoxy, ground to expose an interior cross-section, and cathodoluminescence Contributions to Mineralogy and Petrology (2025) 180:56
(CL) imaged at Boise State University's Isotope Geology Laboratory. CL images were used to guide the placement of laser ablation spot analyses, described below. Whole rock major, trace, and rare earth element data, along with new 40 Ar/ 39 Ar eruption ages for these samples and Cove Fort mafic lavas were published in Rivera et al. (2024). Here, we supplement the previously published data with zircon geochemical information from the same rhyolitic units.
Hand samples of granitoids from five different lithologies were collected and processed for major and trace element geochemistry. Based on the map units of Kirby (2019), these include the 26 Ma hornblende granodiorite (Aleinikoff et al. 1987), 18 Ma biotite hornblende granite (Coleman 1991), 18 Ma biotite-hornblende monzonite (Coleman 1991), 18 Ma biotite hornblende diorite (Coleman 1991), and 17.5 Ma biotite granite (Coleman 1991). Samples were prepared as Li-borate glasses and analyzed by wavelength dispersive x-ray fluorescence (WDXRF) using a Thermo ARL Perform'X at the Hamilton Analytical Lab following methods of Conrey et al. (2023) and Johnson et al. (1999). Following, samples were analyzed by laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS) at Colorado School of Mines using the protocols of Conrey et al. (2023).
Zircon trace element geochemistry for 143 spot locations was determined by LA-ICP-MS at Boise State University with a ThermoElectron X-Series II ICP-MS system and 213 nm frequency-quintupled Nd-YAG NewWave laser. Operating conditions consisted of a 5 Hz at 3.5 J/cm 2 pulsed laser and 25 μm ablation spot diameter. Analytical protocols follow those of Rivera et al. (2016) and additional details can be found in the supplementary materials. An additional 35 LA-ICPMS spot analyses were obtained at the University of Missouri Research Reactor using a Teledyne Instruments Analyte Excite HelEx 193 nm laser ablation cell coupled to a PerkinElmer SCIEX NexION 300 quadrupole ICPMS. Operating conditions consisted of a 10 Hz at 7.5 J/cm 2 pulsed laser and 35 μm ablation spot diameter. Method details are available in Greer and MacDonald (2022), MacDonald et al. (2024), and Magnavita et al. (2024). Primary trace element standards were NIST SRM-610 and SRM-612.
Whole rock geochemistry of four new samples were compiled with major, trace, and rare earth element data from Coleman and Walker (1992;Fig. 2, supplementary materials). Granitoid lithology and mineralogical descriptions follow those of Kirby (2019). Examples of specimens and thin sections can be found in the supplementary materials. The oldest unit, the 26 Ma hornblende granodiorite, has SiO 2 between 65 and 70 wt%, TiO 2 ~ 0.4 wt%, Rb ~ 100 ppm, and Sr ~ 600 ppm. Younger biotite granite and biotite hornblende granite have more evolved compositions with SiO 2 > 70 wt%, total alkali (K 2 O + Na 2 O) ≥ 8 wt%, TiO 2 ≤ 0.4 wt%, and increasing Rb concentrations with decreasing Sr concentrations (Fig. 2). Biotite hornblende granites can be chemically distinguished from biotite granites by their lower Rb concentrations (150-200 ppm) and greater Sr concentrations (250-350 ppm). Additionally, the biotite granites show prominent negative Eu anomalies (Fig. 2), indicative of plagioclase fractionation. Other granitoid lithologies demonstrate minimal Eu anomalies. All units have a slight saddle shape for the middle REE, indicating the fractionation of hornblende and titanite (Coleman and Walker 1992).
Zircon crystal morphology can qualitatively reflect magmatic processes (Hanchar and Miller 1993;Zellmer 2021). Textures and zoning patterns help identify crystals or parts of crystals as autocrystic, antecrystic, or xenocrystic, revealing complex growth histories due to magmatic thermochemical changes (Miller et al. 2007). Here, we follow the conventions of Miller et al. (2007) and use the term antecryst to refer to zircon that crystallized within an earlier pulse of magma that was incorporated into a later magmatic pulse. This is in contrast to a xenocryst, which refers to zircon inherited from the country rock or other magmatic pulse significantly older than the erupted magma (Miller et al. 2007). Figure 3 shows CL images of zircon crystals from the middle and high domes. Middle dome zircons are typically elongated with oscillatory zoning around a central domain, sometimes exhibiting disequilibrium textures interpreted as antecrystic or xenocrystic origins (Miller et al. 2007). Most have a CLbright autocrystic rim with oscillatory and sector zoning, except the 766 ka Big Cedar Cove Dome zircons, which are more irregular and lack a CL-bright rim. High dome zircons are more equant with sector zoning and smaller central domains compared to the middle domes, and they lack a CL-bright rim. Zircon images with co-located LA-ICPMS spots are provided in the supplementary materials.
A total of 178 laser ablation spot analyses were obtained from samples of seven rhyolite domes. Zircon trace element geochemistry is presented in Fig. 4. Laser ablation spot concentrations between 2.9 and 9.8 ppm with an anomalous analysis of 27.2 ppm. (Figure 4A and B). Nb contents for the high domes are generally > 100 ppm, Hf contents > 10,000 ppm, and Th/Y ratios > 0.7 (Fig. 4C and D). Zircon crystals from the oldest of the high domes, Little Bearskin Mountain (588 ka), preserve an Eu/Eu* ratio intermediate between the middle domes and high domes (Fig. 4A and B). Deepening of the Eu/Eu* anomaly in zircon from the middle domes to the high domes zircons is apparent (Fig. 5), along with the enrichment in heavy REEs. Overall, the younger eruptions contain zircon that crystallized from a more fractionated magma, which is consistent with the whole-rock trace element trends (Rivera et al. 2024). locations on individual zircon grains and the full analytical data are available in the supplementary materials. In correlating the geochemistry to zircon morphology, there are no significant geochemical differences between interior domains and rim zircon. The 766 -758 ka middle domes have Eu/Eu* ranging from 0.05 to 0.3, U contents generally up to 1,500 ppm, and Ti contents between 3.3 and 11.9 ppm (Fig. 4A and B). Additionally, the middle domes have Nb contents < 100 ppm, Hf contents between 7,000 and 10,000 ppm, and Th/Y ratios between 0.1 and 1 (Fig. 4C and D). Of note, the oldest of the middle domes, the 766 ka Big Cedar Cove Dome, has elevated Hf relative to the two other middle domes (Fig. 4D).
Incompatible trace element concentrations and ratios for the 588 -483 ka high domes vary widely. Uranium contents range between 2,000 and 8,000 ppm, Eu/Eu* < 0.1, and Ti 1 3 Fractional crystallization of a shared magma source? Evans and Nash (1978) proposed a petrogenetic relationship between the Cove Fort valley mafic lavas and the Mineral Mountains rhyolites, which is evaluated here using the whole rock composition of the oldest and compositionally most primitive erupted basalt of the Cove Fort valley, the 865 ± 30 ka Black Rock tholeiitic lava (Rivera et al. 2024), in a Rayleigh fractional crystallization model. REE concentrations for the liquid evolution were calculated at 10%, 30%, and 50% crystallization of plagioclase and olivine in a 3:1 ratio based on modal proportions (Fig. 6). Partition coefficients (Kd) for fractionating phases were obtained using the Geochemical Earth Reference Model partition coefficient database (GERM-KdD) and were selected based on the closest matching rock type. In most cases, the Kd used in the modelling is an average of Kd high and Kd low values determined from phenocryst-matrix experiments. As crystallization proceeds, the REE contents become more enriched in each subsequent liquid. The REE patterns generated from these models do not match the observed rhyolite compositions, indicating that fractional crystallization does not explain the relationship between the coeval 851 ± 3 ka Bailey Ridge obsidian and the 865 ± 30 ka Black Rock lava flow. Additionally, fractional crystallization of any Cove Fort lava cannot produce the compositions of the middle or high domes. Basaltic andesites and trachyandesites (~ 750 ka) were excluded from modeling due to their more evolved compositions, and the ~ 300 ka andesite (Johnsen et al. 2010) indicates it was not a precursor to any of the Mineral Mountains rhyolites.
Evans and Nash (1978) proposed that the younger domes were derivatives of the older obsidian flows. To test this, a Rayleigh fractionation model was applied to the obsidian flows in an attempt to reproduce the observed compositions of the middle and high domes (Fig. 7A). The model removes a eutectic composition of 30% quartz, 35% potassium feldspar, and 35% plagioclase from the nearly aphyric (< 0.5% phenocrysts; Nash 1976) 851 ± 3 ka Bailey Ridge obsidian. While the calculated liquids for 10%, 30%, and 50% crystallization begin to reproduce the negative Eu anomalies of the middle and high domes, they became enriched in other REE, whereas observed concentrations were depleted with respect to the parental magma. In a bivariate plot of La/Sm versus Sm (Fig. 7B), modeled liquids clearly diverge from observed middle and high dome concentrations. Removal of accessory titanite, allanite, or zircon in addition to the eutectic phases from a magma with the composition of the obsidian could not produce residual liquids with compositions similar to those observed in the middle or high domes (Fig. 7B). Thus, fractional crystallization models
Earlier researchers in the Mineral Mountains identified chemical differences between the middle and high domes and suggested they shared a common source, proposing that later domes were fractionates of magmas with compositions similar to those of the earlier erupted obsidian flows (Evans and Nash 1978). The high domes contain a few percent (< 5%) of phenocrysts, largely quartz, two-feldspars, and minor biotite; the middle domes have the greatest percent of phenocrysts ranging from 5 to 8% (Nash 1976). Petrography and electron microprobe analyses of feldspar phenocrysts (supplementary material) do not suggest significant antecrystic or xenocrystic inheritance. Given the low phenocryst abundance coupled with lack of significant inheritance, we assume that the bulk rock compositions determined by Rivera at al. ( 2024) are representative of melt compositions and thus use whole rock compositions in the following petrogenetic models.
High-silica rhyolitic melts can be produced through fractional crystallization of a basaltic melt (Geist et al. 1995) or anatectic melting of crustal material from basaltic underplating (Huppert and Sparks 1988;Hildreth et al. 2023). This study tests models of fractional crystallization versus partial melting to address the following questions: (1) Are the oldest Cove Fort valley mafic lavas parental to the Mineral Mountains obsidian flows or rhyolite domes? (2) Are the oldest obsidian flows parental to the younger middle and high domes, or do the middle and high domes share a common magma source? (3) Does partial melting of granitic country rock produce any of the Mineral Mountains rhyolites? Coupled assimilation and fractional crystallization (AFC) was considered but excluded because the mixing of fractionating basaltic compositions with potential assimilants (granitoids, Precambrian metamorphic rocks, or Paleozoic sedimentary rocks) would not generate the SiO 2 concentrations > 75 wt% observed in the Mineral Mountain rhyolites.
Following earlier workers, Rivera et al. (2024) noted that less evolved felsic Mineral Mountain magmas erupted first near the valley floor, with progressively more differentiated lavas erupting at higher elevations over time. Zircon compositional trends mirror whole rock geochemistry trends, with the middle domes showing higher Eu/Eu* values (less evolved magma) compared to lower Eu/Eu* values in the high domes (Fig. 4A and B). This trend of decreasing Eu/ Eu* with increasing U and other incompatible elements in zircon may suggest progressive differentiation of a shared magma body, which we test through petrogenetic modeling.
1. The Quaternary rhyolites of the Mineral Mountains were not derived from extreme fractional crystallization of the coeval basalts in the adjacent Cove Fort valley (Fig. 6). 2. The 850 -785 ka obsidian flows do not represent parental magmas for either the middle or high domes (Fig. 7A and B). 3. The 766 -758 ka middle domes do not represent parental magmas for the high domes (Fig. 8A). 4. The 588 -483 ka high domes may share a common parental magma, similar in composition to the 588 ka Little Bearskin Mountain, although not all observed zircon trace element compositions can be reproduced through fractional crystallization modeling (Figs. 7C and D and 8B, C and D). 5. At least four distinct magma bodies were involved over a period of ~ 400 ka: at least one for the basalts, one for the obsidian flows, one for the middle domes, and one for the high domes.
If the rhyolites are not related to the basalts, nor each other, could they be derived from partial melting of the granitoids through which they erupted? Nash (1976) proposed that the rhyolites were a "classic example" of crustal melts resulting from basaltic magma injection in an extensional tectonic regime, with no evidence of a long-lived shallow magma body (Nash 1986). The model involves crustal thinning induced by extension leading to asthenospheric upwelling, generating basaltic magmas in the upper mantle. These magmas ascend into the lower crust, inducing partial melting of the country rock and undergo subsequent fractional crystallization prior to eruption. Additionally, Evans and Nash (1978) proposed a petrogenetic relationship between the Cove Fort valley mafic lavas and the Mineral Mountains rhyolites, which was attributed to the high regional heat flow (Sass et al. 1976;Chapman et al., 1978) and periodic injection of mafic magmas. Coleman (1992) also suggested periodic partial melting due to extension for the generation of the granitic batholith. The oldest cooling ages (18 -16 Ma) are from the structurally highest parts of the batholith on the eastern and northern margins where it intrudes Paleozoic wall rocks, while the youngest ages (11 -9 Ma) are from the structurally deepest parts on the west-central margin where it intrudes Precambrian basement rocks (Coleman 2009).
Alternatively, Evans and Nash (1978) proposed that the chemically distinct rhyolites could be distinct melt pockets produced by repeated basalt injections into the crust that independently fractionated. This would reflect the chemical variability of the source rock rather than differentiation processes. Partial melting of mafic material to produce felsic using the oldest tholeiitic lava or high-silica rhyolite could not produce the observed compositions of the middle or high domes. Similarly, using a middle dome as the parental magma and the same fractionating phases yielded liquid compositions diverging from the high domes. Zircon geochemistry further indicates no genetic relationship between the middle and high domes, as the Eu anomalies of middle dome zircon remained unchanged with Nb content, indicating no significant fractional crystallization (Fig. 8A).
With fractional crystallization of basaltic and earlier erupted rhyolite compositions both ruled out as representing parental magmas, Rayleigh fractional crystallization modeling was conducted to determine if the 588 to 483 ka high domes, the youngest group of Mineral Mountains rhyolites, could be derived from a shared magma source (Fig. 7C and D; supplementary materials). Using Little Bearskin Mountain as the parental magma, which is the oldest of the high domes and located nearest to the valley floor, crystallization and subsequent removal of 0.05% allanite, along with eutectic proportions of quartz, potassium feldspar, and plagioclase can produce residual liquids approximating the compositions of the other high domes at 40-60% crystallinity (Fig. 7D). This suggests that the high domes could share a common source distinct from the magma bodies of the older middle domes and obsidian flows.
The same fractional crystallization model was applied to the zircon geochemistry, again using Little Bearskin Mountain as the parental magma (crystal proportions: 30% quartz, 35% plagioclase, 35% alkali feldspar, 0.01 to 0.05% allanite; Fig. 8). The model reproduced some, but not all, observed trace element patterns. It generally reproduced the trend of Eu/Eu* compositions and most Hf compositions (excluding Ranch Canyon Dome) but struggled with U and Y compositions. The poor reproducibility of observed zircon compositions using a fractional crystallization model that generates observed whole rock compositions indicates that fractional crystallization alone is unlikely to account for the compositional variability observed in the high domes.
Although magma recharge and mixing may account for the compositional variations within a magma batch, the morphology of the high domes zircon crystal do not show disequilibrium textures that could be attributed to mixing processes (Supplementary Materials).
In summary, the fractional crystallization models yield the following conclusions: 1 3 Page 7 of 17 56 Contributions to Mineralogy and Petrology (2025) 180:56 identical to the obsidians. Similarly, melting up to 30% of a biotite-hornblende granite collected near the southwestern edge of the range yields liquids resembling the compositions of the middle domes. The zircon and whole rock geochemical similarities among the three middle domes suggest they may be different pulses of a larger single magma body, though elevated Hf in Big Cedar Cove Dome zircon relative to Wildhorse Domes North and South indicates minor differentiation differences or compositional variations in the partially melted crust. High dome compositions can be melts has been suggested at ocean island volcanoes (e.g., Bohrson and Reid 1997; Sverrisdottir 2007; Carley et al. 2011) and may be responsible for creating bimodal basaltrhyolite volcanic fields (e.g., Charlier et al. 2013; Meade et al. 2014).
Modal equilibrium partial melting models were generated for each group of rhyolites (Fig. 9; supplementary materials). Melting up to 30% of the hornblende granodiorite sample collected near the 786 ka Wildhorse Canyon sample site produces liquids with REE compositions nearly versus Eu/Eu*. C. Nb (ppm) versus Th/Y. D. Hf (ppm) versus Th/Y. E. Eu/Eu* versus 40Ar/39Ar eruption age (Rivera et al. 2024). Sample abbreviations are the same as in Fig. 1 1 3
Melt segregation from a pluton-scale crystal mush is often referenced as the process to extract liquids from parental rocks to generate high-silica rhyolites and granites (e.g., modeled by melting up to 30% of a biotite granite sampled in the western valley. However, some high dome compositions (RCD and ST) are predicted between 50 and 63% melting of the biotite granite (Fig. 9F). This seems unlikely in the upper crust and, as such, may instead represent a mixture of partial melting and fractional crystallization processes that were not modeled here or, alternatively, smaller degrees of partial melting of a slightly different or mixed granitoid lithologies. The latter is favored given that the vents for RCD and ST are within the biotite-hornblende granite near the contact with the biotite granite (Kirby 2019). In all scenarios, middle to heavy REE (Gd, Tb, Ho, Er) depletion in the modeled liquid suggests retention in the restite, likely due to the presence of titanite in the source granite. Thus, melting the dominant granitoid lithology exposed nearest the rhyolite lavas could produce the observed compositions, indicating that partial melting of different granitoids is likely responsible for the compositional variability in the rhyolites.
In summary, we propose the following scenario for the generation of the Mineral Mountains rhyolites (Fig. 10).
1. 865 ka: Injection of mafic material causes the eruption of the Black Rock tholeiitic lava. The associated heat partially melts the hornblende granodiorite, resulting Major phenocryst phases and accessory minerals record thermal, geochemical, and geochronologic evidence of extraction from a mush (Ellis et al. 2014;Stelten et al. 2015;Rivera et al. 2016;Allan et al. 2017). For instance, in the 1.3 Ma Mesa Falls Tuff (Yellowstone volcanic field), zircon with "dark cores" were modeled to represent > 90% crystallinity of the host magma based on colder crystallization temperatures, higher incompatible trace element concentrations (e.g., >20,000 ppm U), and U/Pb crystallization ages up to 50 kyr prior to eruption (Rivera et al. 2016). The authors interpreted the anetcrystic zircon as minor pulses of pre-eruptive magmatism, providing the heat needed to sustain a long-lived crystal mush with interstitial melt. The absence of highly differentiated cores in zircon from the Mineral Mountains' middle and high domes (Fig. 3) suggest that they were not extracted from a long-lived mush.
In contrast, zircon morphology, geochemistry, thermometry, and geochronology of the Alder Creek Rhyolite (ACR), Geysers Geothermal-Clear Lake Volcanic Field, California, record two unrelated zircon populations (Rivera et al. 2013). The first was related to the nascent ACR magma; however, the second population was morphologically and Hildreth 1981Hildreth , 2021;;Hildreth and Wilson 2007;Cashman et al. 2017;Schaen et al. 2017Schaen et al. , 2021;;Andersen et al. 2019;Lu et al. 2021;Pamukçu et al. 2022;Boulanger and France 2023). This model involves the thermal reactivation of a crystal mush from rheological lock up, followed by the extraction of interstitial liquids, which rise and accumulate into larger magma bodies that either erupt as high-silica rhyolites or crystallize as pods of evolved granite in the subsurface (Bachmann and Bergantz 2008;Huber et al. 2011;Escribano et al. 2022). Eruptable liquids extracted from the mush may occur as crystal-poor rhyolites, leaving behind a cumulate crystal pile (Bachmann and Bergantz 2004;Wolff and Ramos 2014;Ellis et al. 2023). The mush model has been used to explain various eruptions across tectonic settings (Bachmann and Bergantz 2008), including voluminous high-silica rhyolites (e.g., Wolff and Ramos 2014;Wotzlaw et al. 2014Wotzlaw et al. , 2015;;Shamloo and Till 2019;Wilson et al. 2021;Pamukçu et al. 2022), small volume rhyolites (Hildreth et al. 2014(Hildreth et al. , 2023;;Stelten et al. 2015;Burgess et al. 2021), and plutonic systems (Schaen et al. 2018;Tavazzani et al. 2020;Lu et al. 2021). olivine in a ratio of 3:1 to produce liquids at F = 0.9, F = 0.7, and F = 0.5 (i.e., 10%, 30%, and 50% liquids, respectively). Chondrite normalization values are from Sun and McDonough (1989) 1 3 56 pyramidal sectors (enriched in REE, Hf, Y, and U), and CLbright prismatic sectors. The combined morphology and lack of Miocene-aged central domains suggest rapid magma extraction without scavenging pre-existing zircon from the cooled pluton. On the contrary, it is possible that the melts hosted pluton-extracted zircon that were dissolved as the melts sat above the zircon saturation temperature. The dissolution of zircon is significantly influenced by temperature, degree of Zr undersaturation, size of the zircon, and volume of the melt cell surrounding the zircon (Bindeman and Melnik 2016). Larger bodies (> 200 m) can completely dissolve zircons, while smaller ones result in partial dissolution and rim overgrowth (Bindeman and Melnik 2016). Thermal erosion of zircon begins at temperatures typical of crustal melting (~ 750˚C to 900 °C), and complete dissolution of a xenocryst could occur on timescales of 10 2 to 10 4 years (Bindeman and Melnik 2016). However, zircon dissolution is often incomplete, leading to resorbed, embayed, or antecrystic cores in newly grown zircon overgrowths. geochemically distinct, and produced consistently older U-Pb dates. These zircon were attributed to the liberation and extraction of xenolithic zircon from partially melting earlier cooled intrusive granitoids, rather than forming within a long-lived crystal mush (Rivera et al. 2013). The Mineral Mountains zircon crystal cargo preserves a third process by which to generate zircon crystals within highsilica rhyolites. Here, the exposed granitic batholith is a cooled body rather than a near-solidus mush. Zircon U-Pb dates suggest batholith emplacement at 18 Ma, initial exhumation at 11 Ma, and final cooling at 8 Ma (Coleman et al. 2009). Mafic injections beneath the batholith likely caused partial melting of solidified granite with various lithologies, rather than reheating of a mush, generating ephemeral pools of magma in which new zircon crystallized. Zircon textures such as sector zoning may form as a result of rapid crystal growth, in which incompatible elements are pushed to the edges or corners of the crystal face (e.g., Chamberlain 2019; Watson and Liang 1995;Watson 1996). Sector zoning is present in all high domes investigated here, with CL-dark Fig. 7 Fractional crystallization models remove eutectic proportions of 30% quartz, 35% plagioclase, and 35% alkali feldspar. Liquid compositions are shown for F = 0.9, F = 0.7, and F = 0.5 (i.e., 10%, 30%, and 50% liquids, respectively). A and B Bailey Ridge obsidian as parental melt. Chondrite values from Sun and McDonough (1989). to occur along the periphery of the system, where magmas exist as individual pods prior to amalgamating into larger magma chambers (Bacon et al. 1980;Hildreth 1981). In the western United States, samples younger than 1 Ma are primarily found in areas experiencing localized thinning Coeval basalt-rhyolite eruptions in other Quaternary North American volcanic fields include the Yellowstone Volcanic Field (Wyoming), Long Valley caldera system and Coso Volcanic Field (California), and the San Francisco peaks (Arizona). In these settings, eruptions are more likely of Kirby 2019). In panels A, C, E, the mode of the source rock in the model is provided, and this mode is used in the models presented in panels B, D, F, respectively. D = partition coefficient; F = percent of melt Fig. 8 Zircon trace element fractional crystallization models of the middle and high domes. A Model of a middle dome parental composition to produce high dome compositions. B-D Model using the 588 ka Little Bearskin Mountain high dome as the parental magma to relate to other high dome compositions. Poor reproducibility of zircon compositions suggests fractional crystallization is not likely the process generating chemical heterogeneity. Percentages reflect amount of fractional crystallization 1 3
In the Mineral Mountains, the rhyolites do not originate from extreme fractional crystallization of coeval basalts from the neighboring Cove Fort valley. Fractional crystallization models indicate that neither the older obsidian flows nor the middle domes served as parental magmas for the younger high domes. The rhyolites appear to be products of partial melting of granitoid lithologies. Melting up to 30% of local granitoid samples can produce liquids that match the REE compositions of the obsidians, middle domes, and high domes. This suggests that the compositional variability observed in the rhyolites is due to partial melting of different granitoids. The preserved zircon record indicates rapid magma generation and differentiation (10 3 -10 4 years), with the presence of sector zoning and the lack of highly differentiated cores, supporting the absence of a highly differentiated mush. The proposed model can be summarized as:
1. Injection of mafic material leads to eruption of mafic lavas in the Cove Fort valley. Associated heat causes partial melting of granitoids, producing isolated, ephemeral pools of high-silica rhyolite melt. 2. Newly formed high-silica melts seed and crystallize zircon, with minimal scavenging of batholith-related zircon. New zircon form rapidly as evidenced by sector zoning and minimal within-grain compositional heterogeneity. These melts prohibit basalt eruptions within the range, confining those eruptions to the low-elevation valley. 3. Crystal-poor high-silica liquids ascend through the batholith to erupt on the surface as a series of domes.
This model aligns with observations from other Quaternary volcanic fields, where rapid magma generation and eruption are common. The isolated, ephemeral melt pockets formed by partial melting of granitoids due to mafic injections result in compositional diversity among the rhyolites. The absence of a long-lived crystal mush and the presence of rapid zircon crystallization suggest a dynamic magmatic system with transient magma bodies. Consequently, the Mineral Mountains' volcanic activity can be explained by a series of mafic injections causing localized partial melting and rapid ascent of high-silica melts, rather than by prolonged fractional crystallization or extraction from a crystal mush.
and are seldom observed in regions where the lithosphereasthenosphere boundary is deeper than 90 km (Golos and Fischer 2022).
During periods of high magmatic flux, batches of mafic melt are emplaced in the root zone below the Mineral Mountains, which triggers a partial melting event of the granitoid. The partial melts are thus ephemeral, isolated pockets of melt, which act as a barrier against further ascent of the basalt (Hildreth 1981;Huppert and Sparks 1988;Escribano et al. 2022;Ellis et al. 2023). The basalts are then confined to the valley, whereas the felsic magmas may continue to rise through the crust via a system of dikes, producing a cluster of domes at the surface (Hildreth 1981;Petford et al. 1993). This interpretation coincides with the discrete, isolated eruptions that occurred in the latest period of volcanism, which span ~ 10 km across the ridgeline (Fig. 1B). Our model removes the need for a shallow, long-lived (> 10 5 ) crystal mush (Costa 2008;Schoene et al. 2012). The paucity of intermediate compositions can also be explained by this model; mafic magmas ascend rapidly to the surface in the valley, with minimal interaction with the crust. Naturally, melting of high-silica granites will only yield highsilica melts.
Duration of magmatic activity suggests that heat must have been supplied to the magma body fairly continuously in order to prevent solidification. Nash (1986) notes that there is "ample evidence of supply" in the neighboring Cove Fort valley; Nash (1986) also notes observed quenched basalt xenoliths within the rhyolites, suggesting that mafic magmas were injected directly into felsic magmas. High heat flow has been recorded regionally (Sass et al. 1976;Chapman et al., 1978) and seismic velocities suggest the presence of melt below the lithosphere-asthenosphere boundary near the Mineral Mountains (Byrnes et al. 2023). Further, a 270 km 2 dilute mantle He anomaly was identified in the cold groundwater and geothermal waters of the nearby Roosevelt Hot Springs (Simmons and Kirby 2024). Interpretation of magnetotelluric data (Wannamaker and Maris 2024) indicates the presence of a strong lower crustal conductor in the 15-35 km depth range (Simmons and Kirby 2024). The combination of He isotopic analysis with the geophysical data led Simmons and Kirby (2024) to suggest that the predominant source of mantle-derived helium in the Milford Valley, the western basin adjacent to the Mineral Mountains, is linked to the underplating and intrusion of mafic magma. The presence of a thick and impermeable crystalline granitic basement rock at mid-crustal levels is noted as a barrier that influences the ascent of mantle He, creating separate flow paths for helium and heat (Simmons and Kirby 2024).