Cracked Mountain edifice was mapped at 1:7,000 scale, with detailed stratigraphic logs and lithofacies sampling done to characterize the entire eruptive sequence (e.g., Harris et al., 2022). Hand samples were catalogued in the field with mineral and textural descriptions. Each sample was given a unique identification code linked to their precise sample coordinates and elevation. Fresh, unaltered geologic samples were cut into 25 mm × 15 mm × 5 mm billets and sent out to Precision Petrographics Ltd., Langley British Columbia, Canada, for thin-section construction. Fifty grams of each volcanic sample was separated into clean, labeled containers, and sent to Acme Analytical Labs Ltd., Vancouver, BC, Canada, for powdering and whole-rock geochemical analyses.

Whole-rock major geochemical compositions were measured for 26 samples by Acme Analytical Labs Ltd., Vancouver, BC, Canada. X-ray fluorescence analyses were conducted with an ARL SMS-3500 Automated XRF using lithium borate fused discs. Trace element contents were measured by inductively coupled plasma-mass spectrometry on acid digested aliquots with an Agilent 7900 ICP-MS. Compositions of whole rocks and glass major element (wt%) compositions are reported in Table 1 and trace element contents (ppm) in Table 2; analytical uncertainties are based on replicate analyses performed by Acme Analytical Labs Ltd. Ferrous iron was measured directly for a subset of samples by volumetric analysis (i.e., titration) by ALS Canada Ltd., North Vancouver, BC, Canada. Supplementary Appendix SA contains six additional sample data and analytical uncertainties for whole-rock geochemistry conducted at the Peter Hooper Lab, Washington State University, USA. X-ray fluorescence analyses were conducted with a ThermoARL Advant’XP+ spectrometer using lithium tetraborate fused beads. Trace element contents were measured by inductively coupled plasma-mass spectrometry on acid digested aliquots with an Agilent 7700 ICP-MS.

Major element concentrations of phenocryst and groundmass phases were determined with a Cameca Sx50 scanning electron microprobe (EMP) at the University of British Columbia, Vancouver, Canada. Polished thin sections were carbon-coated before analyses. EMP conditions include an acceleration voltage of 15 keV, a beam current of 20 nA, and a beam diameter of 30 µm. A full compilation of EMP mineral data is presented in Supplementary Appendix SB.

We used the thermodynamic model Rhyolite-MELTS _v1.0 (Gualda et al., 2012; Gualda and Ghiorso, 2015) to constrain the pre-eruptive storage depths of crystallization for CM magmas. Rhyolite-MELTS is a macro-enabled excel program that has been recalibrated to better capture the evolution of silicic systems while still maintaining the effective calibration for mafic and alkalic systems found in the original MELTS program (Gualda and Ghiorso, 2015). Additionally, Rhyolite-MELTS can be readily used with Windows operating systems. Similar petrochemical studies on the mafic occurrences of Mt. Etna (e.g., Giacomoni et al., 2021) and the Lillooet Glacier basalts of the GVB (e.g., Wilson and Russell, 2017) have also successfully implemented rhyolite-MELTS in their thermodynamic investigation of crustal storage conditions. The version of Rhyolite-MELTS we selected was based on our presumed magma systematics, as we are not dealing with high-pressure mantle melting (e.g., pMELTS) nor saturated quartz phases (e.g., Rhyolite-MELTS_v.1.1).

Compositions were normalized anhydrously and values were stored in a separate file for use in subsequent simulations. We used measured values of FeO and values of Fe₂O₃ calculated from total FeO to constrain oxygen fugacity (fO₂) based on the models of Ghiorso and Sack (1995). These values of fO₂ (ΔQFM 0.6-1.3) were then used for each sample simulation. For samples where FeO was not measured directly (i.e., ferric titration) total FeO was input into MELTS and the compute redox function was used to calculate FeO and Fe₂O₃, respectively. Samples without known FeO were modeled assuming fO₂ was buffered at QFM.

Sixty-three isobaric crystallization simulations using Rhyolite-MELTS were performed on ten whole-rock compositions from Cracked Mountain. Each isobaric simulation ran from liquidus conditions (< 1,300°C) to a temperature where the system was ∼95% crystallized (∼950°C). The pressure range explored was from surface pressures (0.1 kbar) to 4 kbar (∼15 km depth) over increments of 0.5 kbar. Volatile content was also varied by changing the H₂O wt% (0–1.5) in steps of 0.25%. The crystallization assemblage was manually restricted to olivine, plagioclase, and clinopyroxene to match the petrographic properties of CM rocks. The resulting outputs were then used to create mineral stability phase diagrams of the system at 10% crystallization. The value of 10% crystallinity was chosen based on petrographic inspection of the CM holocrystalline glomeroporphyritic clots (see below).

We utilize Pearce element ratios (PERs), first devised by Pearce (1968) to investigate the origin of chemical diversity in Cracked Mountain samples. The basis for PERs uses whole-rock oxide weight percentages, converted to stoichiometric ratios containing cations of crystallizing minerals normalized to a conserved element in a magmatic system (Pearce, 1968; Russell and Nicholls, 1988; Nicholls and Russell, 2016). The advantage of PERs is that you can produce geochemical plots that model mineral accumulation and loss (i.e., fractionation), and test petrologic hypotheses such as multiple magma sources, or magma contamination, without the distortions arising from closure (Pearce, 1968; Russell and Nicholls, 1988).

To effectively implement PERs, you must choose appropriate conserved elements for the denominator. In our case, we have used either titanium (Ti) or zirconium (Zr). Ti was chosen for plots involving major elements since it is effectively conserved during the early crystallization of most basaltic systems (Nicholls and Russell, 1991). We chose Zr as the conserved element for plots of trace elements because of its low analytical uncertainty (Nicholls and Russell, 2016).

Detailed mineral abundances were determined with standard polarizing light and scanning electron microscopy (Philips XL30) for all CM volcanic coherent lithofacies. The petrographic study of CM volcanic samples established two distinct groups 1) an olivine and plagioclase-phyric population (OP) (i.e., Figures 2A,C,E) and 2) olivine, plagioclase, and augite-phyric population (OPA) (i.e., Figures 2B,D,F). OP samples are texturally holo-to hypocrystalline, with glomeroporphyritic clots containing plagioclase (10–15%, 0.5–3 mm) and olivine (5–10%, 0.1–1 mm). OPA samples are texturally holo-to hypocrystalline, with glomeroporphyritic clots containing plagioclase (10–15%, 0.5–3 mm), olivine (10%, 0.1–1 mm), and augite (2–5%, 0.1–1 mm). The prevalence or lack of phenocrystic augite (< 3% volume) is the basis for our segregation. The OP phenocrystic suite includes pyroclasts separated from volcaniclastic deposits, pillow lavas, sheet lavas, and peperitic and non-peperitic intrusions. The OPA phenocryst suite is restricted to pillow lavas and late-stage dykes.

Olivine phenocrysts are subhedral and weakly zoned (see below). Plagioclase phenocrysts are tabular-shaped and petrographic study shows moderate zoning under cross-polarized light; a minor percentage (< 5%) are sieve textured. When present, augite crystals in glomerocrysts are subhedral (Figures 2B,D) and in rare cases, individual augite crystals have rounded edges suggesting partial resorption (Figure 2E).

Figure 3A illustrates the chemical diversity of CM samples after Le Bas et al., 1986 classification of volcanic rocks. All CM samples are subalkaline basalts (Irvine and Baragar, 1971) and display moderate spread in SiO₂ (∼47–51 wt%) and Na₂O-K₂O (∼3–4 wt%). CM OP samples are predominantly lower in SiO₂ wt% and Na₂O-K₂O wt% than OPA samples. All samples are hypersthene normative, although all samples lack Ca-poor pyroxenes (Figure 4C).

CM rare earth elements (REE) are normalized to a Chondrite (Sun and McDonough, 1989) and shown in Figure 3B. All CM samples (e.g., OP, OPA) show similar anomalies. Little variations in trace element concentrations are observed between samples, suggesting a similar mantle source for all CM eruptive deposits.

Furthermore, Figures 3C,D shows incompatible element ratio diagrams for minor elements (i.e., titanium and phosphorus) and trace elements [i.e., niobium (Nb) and yttrium (Y)] normalized to zirconium (Zr). In both plots, all CM samples (i.e., OP and OPA) overlap within 2σ of the mean of each respective data set (dashed and solid bars). These results correlate with the trends seen within the normalized REE diagram (Figure 3B) and indicate that all samples from CM likely originated from a single mantle melt.

Forsterite contents (Fo%) of olivine phenocryst and groundmass olivine are plotted vs. NiO in Figure 4A. OP phenocrysts have higher forsterite (Fo_79-83) and NiO contents (0.1–0.32%) than OPA phenocrysts (Fo_77-80 and 0.1–0.26% respectively). Olivine groundmass phases for OP samples range from Fo_73-79 while OPA has a tighter spread of Fo_77-78. Both OP and OPA groundmass olivine have lower NiO% than the phenocryst phases (0.1–0.2%).

Anorthite contents (An%) of plagioclase phenocryst and groundmass plagioclase are displayed in Figure 4B. Both OP and OPA phenocrysts overlap in anorthite content (OP: An_60-65) (OPA: An_56-65). Similarly, OP and OPA groundmass phases overlap in anorthite ranges (OP: An_49-63) (OPA: An_47-62).

Clinopyroxene phenocryst and groundmass phase compositions (after Morimoto, 1988) are Ca-rich (Figure 4C). The OPA phenocrysts are chromian augite (i.e., Cr³⁺ > 0.01 cation) while groundmass phases in both OP and OPA samples are aluminian augite (i.e., Al³⁺ > 0.1 cation) (Morimoto, 1988). We also tested whether the clinopyroxene phenocryst compositions were compatible with crystallization from their host basaltic magmas (i.e., versus a xenocrystic origin) using the equilibrium model of Putirka (2008). We adopted his range of equilibrium constant values (K_D = 0.28 ± 0.08) for Fe-Mg partitioning between clinopyroxene and the melt (i.e., samples 19-045; 15-168). Based on the model K_D values the augite compositions (see Supplementary Appendix Table SB1 for microprobe analyses) are consistent with crystallization from melt compositions roughly approximated by the bulk rock compositions (Figure 4D) reflecting the low modal abundance of phenocrysts.

Lastly, none of the three phenocryst types (e.g., olivine, plagioclase, and augite) show strong chemical zonation nor is there evidence for pronounced discontinuities between the core phenocryst compositions and their rims. In contrast, the compositions of the phenocryst rims partially overlap the range of core compositions (Figure 4; Supplementary Appendix Table SB1) suggesting stable, near-equilibrium crystallization conditions prior to eruption. The overall trends in mineral compositions between the two Cracked Mountain phenocryst suites are that OP phenocrysts are slightly elevated in Fo% and An% when compared to OPA phenocrysts. This difference in mineral chemistry and phenocryst assemblage (i.e., +/− augite) suggests disparate fractionation dynamics for the two respective mineral systems (OP and OPA). This topic is explored in detail throughout the discussion below.

Phenocryst assemblages in volcanic rocks can provide critical quantitative information on pre-eruptive temperatures, H₂O contents, and pressures (i.e., depth) of magmas. Different minerals (i.e., olivine, plagioclase, and pyroxene) are stable at different P-T conditions (Gualda and Ghiorso, 2015). Thermodynamic models, in conjunction with the knowledge of true phenocryst proportions and estimates of magma composition, can uniquely constrain these conditions in the magma chamber before the eruption.

Our modeling presumes equilibrium magma chamber conditions which are strongly supported by the following evidence. Firstly, glomeroporphyritic clots are observed to be 10% or less in modal abundance in all CM samples. Texturally, these clots display minerals locked together, indicating cognate growth of all phases present (i.e., olivine and plagioclase or olivine with plagioclase and augite; Figure 2). Secondly, the weak normal, chemical zoning recorded by the phenocrysts is consistent with a static closed magma system. Lastly, the phenocryst phase that distinguishes the two suites (i.e., augite) has compositions that are entirely consistent with equilibrium crystallization (Figure 4D) and a cognate origin. This is in contrast to many other basaltic systems (cf. Coote and Shane, 2018; Coote et al., 2018, 2019) where the crystal cargo is at least partially antecrystic or xenocrystic in origin. In summary, we assume the glomerocrysts originated during static, pre-eruptive magma crystallization (e.g., within the magma chamber) while smaller, isolated (micro-) phenocrysts are less constrained in origin and could result from cooling during ascent, or syn/post-eruptive cooling.

In the case of the OP samples (Figures 5A,B), the stable phases for only olivine and plagioclase crystallization range from <2 to 0 kbar (∼5.5–0 km depth), and 0–1.5% H₂0 wt% (see dashed fields). For OPA samples, (Figures 5C,D), the stability fields for olivine, plagioclase, and clinopyroxenes are restricted to pressures between 2.5 and 1 kbar (∼10–4 km depth), and 0–1% H₂O wt% (see dashed fields).

Thermodynamic modeling places constraints on the pre-eruptive history of the two suites of volcanic rocks comprising Cracked Mountain. Here we use the major and trace element chemistry to constrain the differentiation processes that have generated moderate variations in SiO₂ content and the differences in mineral chemistry. Our approach is to use the whole rock chemistries to constrain the nature of magmatic differentiation through the use of Pearce element ratios (PER).

Assemblage test diagrams (Stanley and Russell, 1989) for crystal sorting of olivine alone are shown in Figures 6A,B. If olivine fractionation was responsible for the observed chemical variation seen in CM samples, then the data would fall along the model line with a slope of 1. A comparison of the data to the model line allows us to reject these hypotheses for both OP and OPA suites. The data plot above the model line, indicating that SiO₂ decreases more than the prescribed stoichiometric MgO + FeO in a single-phase olivine fractionation system.

A two-phase assemblage test for crystal sorting of olivine and plagioclase is shown in Figures 6C,D. In this case, nearly all OP samples fall within 2σ of the model slope line (Figure 6C) while the OPA samples fall slightly above the model slope line (Figure 6D). These models indicate that for the samples containing olivine and plagioclase only (i.e., OP), the observed chemical variations can be closely attributed to the crystal fractionation of the two-mineral system. However, the samples containing augite (i.e., OPA) reject the two-phase hypothesis as the data do not fall within the uncertainty of the model line and plot systematically above the model line. This suggests SiO₂ decreases greater than the prescribed stoichiometric CaO + Na₂O and MgO + FeO for the two-phase plagioclase and olivine fractionation system.

Finally, a combined three-phase assemblage test for crystal sorting of olivine, plagioclase, and augite, is shown in Figure 7. In this diagram, two model lines are drawn through the most primitive sample of each respective phenocryst assemblage, and the results show that all CM data fall within the upper (OP) and lower (OPA) limits of self-mixing (e.g., Russell and Stanley, 1990). These findings show that the CM samples preserve signatures of two discrete fractionation systems but reflect a degree of internal mixing. The differences in equilibrium pressures in the two mineral systems (e.g., Figure 5) suggest mixing of the two systems attended ascent and eruption (e.g., Perugini et al., 2015).

The results from the thermodynamic models of stable mineral phases allowed for a wide range of possible H₂O contents, particularly in stable olivine-plagioclase crystallization conditions (e.g., Figure 5). Through combined methods of PERs and MELTS simulations, we can further constrain the temperatures at which pre-eruptive crystallization would have occurred and bracket the maximum amount of water (H₂O wt%) in the system at the time of crystallization. Figure 8 shows chemical plots with axes designed for PERs as seen in Figures 6, 7. Major element oxide data were extracted from the outputs of MELTS simulations for the two most primitive samples in each petrographic suite (20-016 and 19-031) at varying water contents (0.1–1 H₂O wt%) and transposed onto the PERs as dashed vectors. Figures 8A–D show that the chemical trends for both OP and OPA samples mimic that of a system modeled with ≤0.5 H₂O wt%. Furthermore, the simulations for the two respective primitive samples containing 0.5 H₂O wt% are labeled for temperature (°C). These labels illustrate the range of magmatic temperatures for each phenocryst suite; all OP samples fit within the temperature interval ∼1,240–1,155°C whereas the OPA samples fit within the interval 1,250–1,150°C (Figure 8E). A summary of predicted mineral saturation temperatures for representative CM compositions is given in Table 3.

Given these findings, both OP and OPA samples crystallized in a nearly “dry” system with less than 0.5 H₂O wt%. Higher water (wt%) is shown to suppress the crystallization of plagioclase in favour of olivine and olivine with clinopyroxene at higher pressures (e.g., Figure 5). Only with low water contents (∼0.5%), are we able to simulate the plagioclase fractionation coinciding with olivine, or olivine with clinopyroxene in the OP and OPA samples respectively. These findings differ from the typical subduction-related basaltic water content. The global average (H₂O%) for volcanic arcs is 3.9% and 3.2% for the Cascade arc (Wallace et al., 2015). However, recent studies within the Garibaldi belt (GVB) have shown evidence of low-water basalts, with some melt inclusions containing as little as 0.1% H₂O (Venugopal et al., 2020), further supporting our findings of “dry” magmas at Cracked Mountain.

These model constraints on magmatic water content further restrict the optimal storage pressure conditions for the two rock suites (i.e., OP vs. OPA; Figure 5). The OP samples must have been stored at pressures ≤2 kbar, with a mean upper pressure of 1.6 kbar (N = 5) (Figure 5; Table 3). The OPA samples are restricted to pressures 1.5 ≥ 2.5 kbar, with the mean conditions falling within 1.9 ≥ 2.4 kbar (N = 5) (Figure 5; Table 3).

Based on our multifaceted petrochemical investigation, we propose the following model for the magma bodies that fed the eruption and growth of the Cracked Mountain volcano (e.g., Figure 9):

• A common mantle melt, as evidenced by identical REE signatures, ascended from the lower crust and either had very low initial H₂O content or became dehydrated (∼dry melt). The magma lacks a conventional arc-like chemical signature (i.e., trace element composition and low volatile content) but mirrors other mafic magmas in the northern GVB which show limited subducted slab input (Mullen and Weis, 2013, 2015; Mullen et al., 2017; Venugopal et al., 2020). The single melt source is further attested to by the indistinguishable conserved major and trace (REE) element signatures of all CM samples (e.g., Figures 3B–D), while the “dehydrated conditions” are supported through the modelled chemical trends observed in Figure 8.

The two magmas erupted concurrently in geologic time (monogenetic) producing a small basaltic edifice with two unique phenocryst assemblages (i.e., OP and OPA). The evidence of short-lived and coeval timing for the two magmas is supported by the field evidence which shows no break-in time in volcanic stratigraphy and is bolstered by the paleomagnetic signatures of all deposits overlapping within one magnetic moment (Harris and Russell, 2021; Harris et al., 2022).

Monogenetic volcanoes are found globally within all major tectonic environments. These short-lived eruptive centres have garnered increasing interest, in part, due to their threat as natural hazards (Smith and Németh, 2017). A variety of methods have been used in support of their study, including seismic tomography, geochemistry, petrography, and experimental and numerical modeling. These studies have elucidated greater complexity in their magmatic underpinning that belies their apparent single eruptive histories.

The best line of evidence for understanding the past CM volcanic plumbing comes from our petrographic observations in conjunction with geochemical measurements. Our findings of two unique petrographic suites (i.e., OP and OPA) guide our subsequent chemical and thermodynamic models that we use to constrain the pre-eruptive crystallization conditions. These results require that magmas were stored in two discrete pressure fields (i.e., crustal depths) to erupt materials containing two unique mineral assemblages, indicating tapping of a poly-chamber system during the eruption of the CM volcano. Our schematic interpretation of the shape and interconnectivity of the OP and OPA magmas (e.g., Figure 9) is supported by a diverse range of other studies of monogenetic eruptions.

The presence of polymagmatic crustal magma chambers beneath active monogenetic systems has been shown seismically for volcanoes in Kamchatka, Russia (Kugaenko and Volynets, 2019), Jeju, Korea (Song et al., 2018), and Grímsvötn, Iceland (Reverso et al., 2014). Additionally, geochemical variations in erupted materials in Jeju, Korea (Brenna et al., 2011), Mt. Gambier, Australia (Van Otterloo et al., 2014), and Bahariya, Egypt (Khalaf and Sano, 2020) have led to speculation that these small-volume monogenetic volcanoes were connected to polymagmatic systems. The Icelandic Eyjafjallajökull (2010) eruption is another example of a heavily studied volcano that displayed polymagmatic activity, with an initial flank eruption fed by a deep-seated mafic magma followed by a summit eruption fed from a shallowly stored silicic magma (Keiding and Sigmarsson, 2012). Even within the mafic magmas at Eyjafjallajökull (2010), binary chemical signatures are recovered from melt inclusions (e.g., Moune et al., 2012), suggesting a mixing of two end-member sources and further highlighting the complexities of magmatic plumbing systems.

Furthermore, real-time seismic monitoring of the volcanic activity beneath the Eyjafjallajökull (2010) eruption suggested that the observed subsurface deformation was due to “hydrostatic” interconnectivity of vertically separated sills (Sigmundsson et al., 2010). Field-based studies of eroded volcanic regions in the McMurdo Dry valleys of Antarctica have also revealed networks of interconnected high-aspect-ratio sills (Marsh, 2004), further supporting our geometric depiction of the magma chamber dynamics beneath Cracked Mountain volcano in Figure 9.

Similar petrochemical methods to those used here have been conducted at other monogenetic volcanoes in Undirhlíðar (Iceland), Kaikohe-Bay of Islands (New Zealand), and Ma’anshan (China), and have provided insight into magma chamber dynamics (Pollock et al., 2014; Coote et al., 2018, 2019; Coote and Shane, 2018; Liu et al., 2021). Forensic studies of monogenetic basaltic volcanoes relying solely on chemical signatures may lack the capacity to recover the specifics concerning crustal magma storage conditions. As Annen et al. (2006) suggest, the chemical signature of magma is dictated by deep-crustal processes, while texture and mineralogy for magma are defined by the transport and ascent processes governing the magmas’ transit through the shallow crust.

As the study of monogenetic systems advances, so do the insights into their eruptive complexities. Monogenetic volcanoes commonly feature two or more eruptive stages (Kereszturi and Németh, 2012) reflecting combinations of internal (e.g., volatiles) and external (e.g., groundwater or fluids) factors that influence eruptive styles (i.e., explosive or effusive) and the resulting volcanic morphologies (e.g., maar/scoria cone) (Kereszturi and Németh, 2012). As a monogenetic edifice, Cracked Mountain most closely classifies as a “Surtseyan” tuff cone associated with spatter and lava flows (Kereszturi and Németh, 2012). Its glaciovolcanic origin would also classify as a transitional, tephra-dominated, tuya (Russell et al., 2014; Harris et al., 2022). The transitional classification is given because Cracked Mountain’s onset was explosive, followed by intrusive and effusive activity. Additionally, the bulk of the edifice is built of tephra (rather than lava-dominant or pillow-dominant), with the term “tuya” used to describe a positive relief volcano whose morphology and lithofacies directly result from ice confinement and/or meltwater interactions (Russell et al., 2014).

The presence of an impounding ice sheet at the time of the Cracked Mountain eruption represents an external factor that played a major role in its style of eruption, the lithofacies diversity, and the volcano’s unique morphology (Harris et al., 2022). However, the presence of ice covering SW British Colombia at the time of the CM eruption may have impacted magma-crustal dynamics as well. Recent studies have proposed causal linkages between volcanism and glacial cycles wherein the loading and unloading of the crust by glaciers causes lithostatic compression and decompression (Jellinek et al., 2004; Rawson et al., 2016; Wilson and Russell, 2020; Guillot and Ponce, 2021; Russell et al., 2021). In particular, the “glacial pumping” models of Wilson and Russell (2020) show that periods of glacial loading facilitate magma charging and entrapment within the upper crust, and during periods of deglaciation those stored magmas are released. Similarly, an investigation of Chile’s Mocho-Choshuenco volcanic system found that during peak glaciation eruption rates are low and magma storage times are high and imply extended periods of magma fluxing into the crust (Rawson et al., 2016). Conversely, the period of deglaciation was attended by efficient evacuation of crustally-stored magmas and a rise in eruption rate.

Glaciovolcanic edifices provide a powerful means to study paleo ice sheets (e.g., Kelman et al., 2002; Smellie, 2007; Wilson et al., 2020). Cracked Mountain erupted during the mid-Pleistocene at the waning of a global ice period, marine isotope stage (MIS) 12, and the onset of rapid global warming during, period MIS 11 (Lisiecki and Raymo, 2005; Harris et al., 2022) (Figure 10A). We propose that the storage of CM magmas in the upper crustal magma chambers may have resulted from the loading of the lithosphere by the paleo-cordilleran ice sheet. During this time, the glacially loaded lithosphere underwent magma ponding within the crust but compressional forces inhibited eruptive activity (e.g., Figure 10B). After a period of magma-crustal accumulation (in MIS 12), there was a period of relatively rapid deglaciation (in MIS 11) which led to decompression of the crust supporting the eruption of the crustally stored magmas beneath Cracked Mountain (Figure 10C).

The causal linkages between glacial events and volcanic eruptions are relevant to the long-term forecasting of future volcanic eruptions across the planet. With ice caps shrinking globally across regions with well-documented volcanism (e.g., Antarctica, British Columbia, Iceland, and South America) (Peltier et al., 2015), there will be an increased opportunity and need to further understand the role in which glaciers may impact magmatic crustal dynamics.