Search

Silicic magmatic systems are fundamental to the production and evolution of continental crust, the recycling of volatile elements from mantle to atmosphere, and the production of economic deposits. Additionally, silicic systems are responsible for producing some the most catastrophic volcanic eruptions on Earth. The processes and timescales governing the production, evolution, and ultimately eruption of silicic magmatic systems remain poorly constrained but are critical to accurate volcanic hazard assessment. Active debate continues regarding many aspects of silicic magmatic systems, such as: 1) the architecture of the sub-volcanic system (i.e., physical distribution of diverse magmatic components throughout the crust that contribute to erupted magma bodies); 2) the longevity of storage of different magmatic components (i.e. melts, crystals, volatiles) prior to eruption; 3) the timescales over which the erupted magma body (or bodies) assemble prior to eruption; and 4) the predominant thermal (and by proxy physical) state in which silicic magma bodies are stored within the crust. In this dissertation I combine a variety of geochemical tools to explore the thermal the chemical evolution of silicic magmatic systems of different size and eruption magnitude to address this outstanding uncertainty. In detail this dissertation focuses on the two most recent eruptive episodes (1980-86 and 2004-05) from Mount St. Helens (MSH), USA, and six eruptions spanning the last 30 kyrs of activity at the Taupo Volcanic Center (TVC) in New Zealand. In Chapter 1 I use new 238U-230Th-226Ra disequilibria data, in-situ plagioclase trace-element data, and Sr-in-plagioclase diffusion timescales from Mount St. Helens 1980 cryptodome and 2004-2005 dacite domes to evaluate the thermal storage conditions and compositional diversity of recent MSH magmas. Plagioclase trace-element data and U-series characteristics reveal a compositionally heterogenous magmatic system beneath MSH and require multi-stage plagioclase growth histories. The data also show that 2004-05 dacites contain a different plagioclase population relative to the 1980 cryptodome dacite, comprised of either a compositionally distinct plagioclase component or composed of the same plagioclase components but in significantly different proportions. Coupling Sr diffusion timescales with U-series measurements indicates that significant fraction (at least ~40%) of modeled MSH plagioclase spend 25-50% of their storage time in similar conditions. Our data combined with data from other arc systems imply that the process of remobilizing magma in arc systems towards successive eruptions requires thermally rejuvenating largely crystalline material, as opposed to sequential tapping of a persistent liquid-dominated magma body, even if successive eruptions are spaced closely in time (decades). Also, rejuvenation events responsible for successive eruptions may sample spatially localized, but potentially overlapping, portions of the broader magma reservoir. In Chapter 2 I present new 238U-230Th ages and trace-element data for zircons separated from both whole-rock (WR) samples and from plagioclase separates from the 25.4 ka Oruanui eruption within the TVC. These data provide a more complete record of zircon crystallization than previous analyses, which were restricted to whole-rock interiors, and demonstrate that plagioclase-hosted zircons and WR hosted zircon rims (surfaces) both record processes obscured in WR zircon interior data. WR zircon surface 238U-230Th ages yield a single eruption age peak whereas the WR zircon interior and both plagioclase-hosted zircon surface and interior age spectra record more complex multi-modal age distributions. WR zircon surfaces are generally more restricted in composition than WR interiors and display two clear compositional groups, the preservation of which requires the two groups of zircons must have been incorporated into the erupted Oruanui rhyolite within ~2 kyrs of eruption. Systematic compositional variations with 238U-230Th crystallization age recorded in plagioclase-hosted zircon surfaces indicate a portion of the Oruanui magmatic system experienced a shift to higher temperatures and less evolved compositions occurred beginning at ~40-45 ka and suggests at portion of Oruanui plagioclase crystallized more than ~13 kyrs prior to eruption. These new data clearly demonstrate the additional insight available by detailed microanalytical investigations of zircons extracted from multiple hosts within a magmatic system. More broadly the work presented here suggests that large volumes of silicic magma accumulate in the crust via amalgamation of multiple smaller compositionally distinct portions of the broader magma reservoir, and that large silicic eruptions may result from a complex interaction of thermal priming (i.e., rejuvenation) and tectonic triggering/modulation. In Chapter 3 I present in-situ plagioclase trace-element data and Sr-in-plagioclase diffusion timescales from six eruptions spanning the last 30 kyrs of activity from the TVC. I use these data to better understand the compositional and thermal evolution of a large silicic magmatic system across a caldera-forming cycle. Results show that plagioclase compositions do not vary systematically across the caldera cycle, but within each eruption plagioclase interiors are far more compositionally diverse than corresponding rims. Additionally, plagioclase rims appear to have grown in equilibrium with the erupted melt, with the possible exception of the Omega dacite. These data require plagioclase interiors to have crystallized within a more compositionally heterogenous portion of the TVC system (possibly the deeper-seated crustal mush) and have been subsequently transported to the final erupted magma body, where crystal rims grew. Diffusion timescales recovered from plagioclase rims are also remarkably similar across all six eruptions, suggesting that the final crystal-poor erupted magma bodies were constructed within years to decades (possibly centuries) prior to eruption, regardless of eruption size. These data suggest that the accumulation of large volumes of mobile silicic magma within the upper crust may not require significantly longer accumulation times relative to small volume melt bodies. Instead, other explanations are needed – for example, it may be possible that larger volumes of mobile magma accumulate when the mush system is more interconnected (i.e., more developed melt pathways), which could allow for more efficient melt extraction, or access to a greater volume of mush. Collectively, these chapters provide novel insights into the processes and timescales governing the construction and evolution of silicic magmatic systems. The work on MSH demonstrates silicic reservoirs at many arc systems are comprised of heterogenous crystal rich mushes that appear to exist predominantly at low temperatures. The MSH work also demonstrates that successive eruptions can sample compositionally distinct portions of the magma reservoir even when spaces closely in time (within decades). The work presented herein on the TVC emphasized the importance of detailed investigations of crystalline cargo in detangling the complex processes operating within silicic magmatic systems. Additionally, this work demonstrates that the production of large volumes of mobile silicic melt within the upper crust can occur on timescales of years to centuries and thus does not require protracted accumulation timescales relative to smaller volume melt bodies. Instead, larger volumes of mobile magma may accumulate when the mush system is more interconnected (i.e., more developed melt pathways), which could allow for more efficient melt extraction, or access to a greater volume of mush.

Both spreading rates and local magma supply to mid-ocean ridges affect crustal construction styles and ridge morphology, alternately leading to either asymmetrical (detachment faulting) or symmetrical faulting styles. Uranium-series isotopic disequilibria in mid-ocean ridge basalts (MORB) may provide insight into how melt supply variations relate to ridges’ accretion styles, a processes that are not well understood. I use Reactive Porous Flow (RPF) equilibrium and disequilibrium modeling to simulate U-series disequilibria at mid-ocean ridge (MOR) generated by melt supply variations at both asymmetrical and symmetrical ridge segments. Guided by my modeling, I predict that enhanced melt contributions from enriched pyroxenitic mantle produce elevated (230Th/ 238U) in young basalts due to the higher garnet modes in pyroxenites throughout the melting regime. If symmetrical segments of the Kane-Atlantis Supersegment overlie more pyroxenite-rich mantle zones, as postulated here, I would expect measurably higher (230Th/ 238U) in basalts from those areas. Lack of such a systematic signature along symmetrical segments would alternatively suggest that crustal magma pooling patterns dominantly control melt supply variations along slow-spreading ridges without an underlying mantle driving force. The (226Ra/ 230Th) and (231Pa/ 235U) ratios of the RPF equilibrium and disequilibrium models generate results that do not fully explain the global MORB data using a single melting lithology. The model results suggest that mixing melts from multiple mantle sources may be necessary to produce the full global data set. Comparing these preliminary results to traditional dynamic melting models and radiogenic isotopes may provide additional insight into how the mantle melting process truly affects both U-series isotopes and ridge symmetry. Advisor: Lynne J. Elkins

Arvicoline rodents (voles, lemmings, and muskrats) are important taxa for characterizing and defining the North American Land Mammal Ages (NALMAs) of the Pliocene and Pleistocene. They are the basis for finer-scale temporal and spatial resolution and have been used to generate the divisions of the Blancan and Irvingtonian NALMAs. The profuse fossil record of arvicolines has never been reconciled with modern phylogenetic techniques, because the fossil record is almost exclusively isolated teeth or tooth-bearing elements. Here I present the most taxonomically robust (n=134) and genetically well sampled (n=5 genes) molecular phylogeny of arvicoline rodents. I used Parsimony, Maximum Likelihood, and Bayesian Inference methodologies to explore the systematic relationships of the clade. Seven fossil calibrations were used to place a series of temporal constraints on the tree to generate the first time-calibrated phylogeny of arvicoline rodents and I found that the clade originated in the Miocene at 6.5 Ma. This hypothesis was subsequently used to model speciation rates, and the patterns of paleobiogeography of the group in the R program BioGeoBEARS. I hypothesize that arvicoline rodents originated in Asia and that dispersal was asymmetric from west (Asia) to east (North America) with at least nine dispersal events between the two continents. My data indicate an origination of Lemmiscus at 5 Ma and the origination of the Synaptomys clade at 3.8 Ma, which supports previously problematic reports of both taxa from Hagerman, ID. This leaves the Blancan V division (as previously defined by the immigration of Synaptomys at 2.5 Ma) undefined and this leaves the Blancan undifferentiated from 4. Ma to the start of the Irvingtonian at 1.9 Ma. This is a novel approach of reconciling biochronology and probabilistic paleobiogeography. I also obtained the first radiometric age of the fauna of Cumberland Bone Cave using a combined Uranium Series and Electron Spin Resonance (ESR) approach, the first date of a middle Pleistocene site in Northeastern North America. Results indicate an age of the fauna of 756 +87/-98 ka which provides an important temporal anchor point from which to begin to understand the faunal dynamics of the region during the Pleistocene.

Arvicoline rodents (voles, lemmings, and muskrats) are important taxa for characterizing and defining the North American Land Mammal Ages (NALMAs) of the Pliocene and Pleistocene. They are the basis for finer-scale temporal and spatial resolution and have been used to generate the divisions of the Blancan and Irvingtonian NALMAs. The profuse fossil record of arvicolines has never been reconciled with modern phylogenetic techniques, because the fossil record is almost exclusively isolated teeth or tooth-bearing elements. Here I present the most taxonomically robust (n=134) and genetically well sampled (n=5 genes) molecular phylogeny of arvicoline rodents. I used Parsimony, Maximum Likelihood, and Bayesian Inference methodologies to explore the systematic relationships of the clade. Seven fossil calibrations were used to place a series of temporal constraints on the tree to generate the first time-calibrated phylogeny of arvicoline rodents and I found that the clade originated in the Miocene at 6.5 Ma. This hypothesis was subsequently used to model speciation rates, and the patterns of paleobiogeography of the group in the R program BioGeoBEARS. I hypothesize that arvicoline rodents originated in Asia and that dispersal was asymmetric from west (Asia) to east (North America) with at least nine dispersal events between the two continents. My data indicate an origination of Lemmiscus at 5 Ma and the origination of the Synaptomys clade at 3.8 Ma, which supports previously problematic reports of both taxa from Hagerman, ID. This leaves the Blancan V division (as previously defined by the immigration of Synaptomys at 2.5 Ma) undefined and this leaves the Blancan undifferentiated from 4. Ma to the start of the Irvingtonian at 1.9 Ma. This is a novel approach of reconciling biochronology and probabilistic paleobiogeography. I also obtained the first radiometric age of the fauna of Cumberland Bone Cave using a combined Uranium Series and Electron Spin Resonance (ESR) approach, the first date of a middle Pleistocene site in Northeastern North America. Results indicate an age of the fauna of 756 +87/-98 ka which provides an important temporal anchor point from which to begin to understand the faunal dynamics of the region during the Pleistocene.

In the easternmost Basin and Range Province (BRP) in central Utah, the east-west trending Dry Canyon Graben formed perpendicular to the Gunnison Fault, which bounds the San Pitch Mountains horst block. A detailed investigation of the Rock Canyon fault zone, bordering the northern side of the Dry Canyon Graben, using field mapping, structural kinematic analysis, textural analysis of fault rocks, and radiogenic dating, has helped reveal the linkages between the graben and regional Basin and Range structure. Discovery of both dip-slip and oblique-slip lineations on faults in both the Rock Canyon fault zone and the south Dry Canyon fault zone indicate at least two phases of displacement occurred along boundary faults of the Dry Canyon Graben. Rock Canyon fault zone is likely connected to the Gunnison Fault, and Dry Canyon Graben may be part of a complex accommodation zone between three overlapping segments of the Gunnison Fault. A precise U-series date of 346.95 + 0.000098 Ka documents late Pleistocene fault activity. This suggests that the Dry Canyon Graben is a Basin and Range structure that formed concurrent with the Gunnison fault up until at least 347 Ka before present. Rock Canyon Fault zone is associated with multiple phases of fault rupture and sealing, indicated by polyphase cement-rich chaotic breccias, juxtaposed layers of slickenfibers, and composite zones of alternating breccias and veins along faults. The main faulting process is dilational faulting and collapse of subsurface fault voids as indicated by the dominance of cement-rich clast-supported chaotic breccias. Repeated rupture in the same zone is indicative of a fault weakening system. The crystal textures, travertine forms and layered carbonate deposits in the fault zone point to a near-surface environment for the deformation processes, where low lithostatic pressures permit open space to form and remain open. CO2-rich fluids ascended through fracture conduits along the Gunnison and Rock Canyon zone faults, likely derived from the Navajo Sandstone reservoir in the subsurface Sanpete-Sevier Valley anticline.Tufa deposits along the Gunnison Fault trace at Birch Canyon are reinterpreted as cascade deposits draping over and depositing in-situ over a pre-existing escarpment on a segment of the Gunnison Fault. If the Birch Canyon escarpment marks a fault scarp related to the Gunnison Fault, then the last displacement on that fault strand must be older than ~3200 years, the oldest dated tufa deposited in the linked upper and lower alluvial fans.

The development of soil is driven by environmental conditions, and soil properties are reflective of those conditions. Pedogenic carbonate is widespread in arid and semi-arid soils and its stable C and O isotope composition is often used as a paleoclimate indicator. However, questions remain about the precise meaning of those signals, and most importantly the season(s) that the isotopic composition most closely reflects. Chapter 1 examines the soil conditions that lead to carbonate formation, and describes research on a climosequence of four sites on Holocene soils in Fish Lake Valley, Nevada (western USA), where Mean Annual Precipitation ranges from ~80mm yr-1 to ~220mm yr-1. Continuous measurements of soil temperature, soil volumetric water content, air temperature, and air humidity were recorded in 0.5hr intervals. Precipitation and samples of soil atmosphere from 10, 25, 50 and 100cm were collected at various intervals for CO2 concentration, δ13C and δ18O values (which are a proxy for soil water δ18O). This climosequence indicates that the C and O stable isotope composition of soil CO2, H2O and carbonate change systematically with elevation and climate. At the lowest elevation site, soil carbonate δ13C and δ18O values reflect constant conditions nearly year round, especially in the 25-50cm soil depth interval. At the mid-elevation sites, spring through summer conditions appear to be recorded in δ13C and δ18O values of pedogenic carbonate. At the highest elevation site of 2602m, the C and O isotopes of carbonate never reflected the measured soil CO2 and H2O isotopes. These results indicate that there is inherent difficulty or complexity in assigning a definitive timing of carbonate formation based on short-term (one or two year) field measurements of soil CO2 and H2O, but also suggests that short-term monitoring, despite its weaknesses, helps to reveal the nature of the long term record isotopically recorded in soil carbonate. Chapter 2 presents the first continuous millennial-scale paleoclimate reconstruction based on O, C and U isotopes in dense, laminated pedogenic carbonate. The record is from the Wind River Basin of Wyoming, and provides isotopic proxies for climate and vegetation conditions in the North American mid-continent for the last 120ka. Accurate and precise 230Th/U ages were obtained from laser ablation ICP-MS spots of 93um diameter size. The resulting carbonate growth calculations indicate that ~100μm sampling regions represent 10ka to 0.1ka at growth rates of 10 to 1000μm ka-1. The U-series transects were coupled to δ13C and δ18O values obtained by ion probe on 10um diameter spots. Modern soil carbonate in the region has δ18O and δ13C values that are similar to the youngest (early Holocene) samples acquired in the laminations, strengthening the interpretation that the carbonate reflects soil isotopic conditions at the time of formation. The carbonate lamination record from 120 to 80ka is somewhat sparse due to slowly forming carbonate, but records conditions that change in phase with summer solar insolation, and indicate that relatively warm and dry conditions prevailed during the penultimate deglaciation and last interglacial period. The paleoclimate signal in the soil carbonate δ18O values during this time period are similar to that recorded in the well-studied Devils Hole phreatic carbonate δ18O record from the Mojave Desert in southwestern North America. From 75 to 55ka, carbonate δ18O values sharply increase by ~2‰, a change also recorded by speleothems from central North America, but not seen in the Devil’s Hole record. Atmospheric circulation during this time appears to have changed from a state dominated by westerly winds to one influenced by meridional flow from the south along the western limb of a strong and west-based North Atlantic Subtropical High, a condition that increased the proportion of summer to winter precipitation by 20-33%. δ13C and 234U/238Ui values from 75 to 55ka also indicate a shift towards wetter conditions and are supported by regional fossil pollen records suggesting a change to prairie-like ecosystems with wetter summers during this time. Subsequently, the carbonate record suggests that climate in mid-latitude central North America became progressively more arid during the onset of the last glaciation (both δ13C and 234U/238Ui values become more positive), culminating in peak aridity during the Last Glacial Maximum. Carbonate formation rates during these colder conditions increased and the temporal resolution of the Wind River record increased as a result, allowing it to capture sub-millenial abrupt climate changes associated with Dansgaard-Oeschger cycles 6 through 3, Heinrich Stadials 2 through 0 (Younger Dryas), as well as the dramatic Bolling-Allerod warming at 14.5ka. The Wind River pedogenic carbonate record shows the potential to develop similar high temporal resolution paleoclimate records form soil carbonate deposits that are relatively common worldwide. The Cenozoic paleoclimate of the Atacama Desert is not well known. Chapter 3 presents the results of a study of 14 Oligocene to early Miocene paleosols exposed in the El Tesoro Mine, near Calama, Chile. The paleosols developed on an aggrading alluvial fan system, and lie above mineralized gravels that host a copper ore body. Soil forming conditions that oscillated between chemical weathering and clay production (humid: analogous to modern Alfisols) to environments favoring the accumulation of pedogenic carbonate (arid to semi-arid: analogous to modern Aridisols) are indicated. In contrast, the region is presently hyperarid, and soils accumulate sulfate, chlorides, and nitrates. While total chemical analyses clearly reveal the accumulation of Ca by the carbonate rich paleosols, none of the soils revealed significant losses of elements by leaching. The δ18O values of the carbonates range from -13.81‰ to -3.16‰ (VPDB). The O isotope data, when combined with published data from the region, revealed a significant divergence in the O isotope composition of precipitation in the eastern and western margins of the Andean plateau since the Oligocene, suggesting that simple interpretations of declines in δ18O values of carbonate with increasing elevation may not be appropriate. These paleosols clearly indicate that wetter conditions prevailed in what is now the Atacama Desert during the Oligocene to early Miocene. Chapter 4 presents findings from a study examining a special case of interactions between soil water oxygen isotopes and cations adsorbed to clay minerals in the soil matrix. In isotope-enabled hydrology, soil and vadose zone sediments have been generally considered to be isotopically inert with respect to the water they host. This is inconsistent with knowledge that clay particles possessing an electronegative surface charge and resulting cation exchange capacity (CEC) interact with a wide range of solutes which, in the absence of clays, have been shown to exhibit δ18O isotope effects that vary in relation to the ionic strength of the solutions. To investigate the isotope effects caused by high CEC clays in mineral-water systems, we created a series of monominerallic-water mixtures at gravimetric water contents ranging from 5 to 32%, consisting of pure deionized water of known isotopic composition with homoionic (Mg, Ca, Na, K) montmorillonite. Similar mixtures were also created with quartz to determine the isotope effect of non-, or very minimally-, charged mineral surfaces. The δ18O value of the water in these monominerallic soil analogs was then measured by isotope ratio mass spectrometry (IRMS) after direct headspace CO2 equilibration. Mg- and Ca-exchanged homoionic montmorillonite depleted measured δ18O values up to 1.55‰ relative to pure water at 5% water content, declining to 0.49‰ depletion at 30% water content. K-montmorillonite enriched measured δ18O values up to 0.86‰ at 5% water content, declining to 0.11‰ enrichment at 30% water. Na-montmorillonite produces no measureable isotope effect. The isotope effects observed in these experiments may be present in natural, high-clay soils and sediments. These findings have relevance to the interpretation of results of direct CO2-water equilibration approaches to the measurement of the δ18O value of soil water. The adsorbed cation isotope effect may bear consideration in studies of pedogenic carbonate, plant-soil water use and soil-atmosphere interaction. Finally, the observed isotope effects may prove useful as molecular scale probes of the nature of mineral-water interactions.

This dissertation consists of three projects that involve isotope and trace element investigations of magmatic processes and timescales in the Azores, Portugal. Two projects focus on the evolution of historical eruptions from Fogo and Furnas volcanoes on the island of São Miguel, addressing two fundamental questions: 1) What are the processes that generated the evolved (explosive) magma and chemical zonation in the magma chambers?; and 2) What are the timescales over which these processes occur? The Fogo 1563AD deposit (0.14 km3 d.r.e.) shows monotonic chemical zonation that can be explained by closed-system fractional crystallization. In contrast, the Furnas 1630AD deposit (0.65 km3 d.r.e.) shows more complex chemical zonation with variable Sr isotope signatures that require assimilation combined with fractional crystallization and a complex magma chamber geometry. The magma residence time estimated from U-series disequilibria for the Fogo 1563AD deposit is the shortest ever determined for silicic magmas, and reveals that alkaline silicic magmas can evolve extremely rapidly, with pre-eruptive magma residence timescale on the order of decades. The larger Furnas deposit has a substantially longer fractionation timescale (~500 years). The results of these studies are consistent with results from other silicic deposits, which display a global negative correlation between fractionation rate and eruptive volume.The third project constitutes the first platinum group element (PGE) abundance and 186Os/188Os isotope study of Azores basalts, aimed to address the origin of the Azores mantle plume and whether or not mantle plumes carry an outer core signature. The results of this study reveal that radiogenic 186Os/188Os ratios in some Azores basalts are consistent with a minor outer core contribution to the Azores mantle plume, which would suggest a deep (core-mantle boundary) origin of this plume. However, the lack of elevated PGE abundances in the Azores basalts suggest that alternative origins for the radiogenic 186Os/188Os also need to be considered, including a heterogeneous mantle with pyroxenite veins or metasomatic sulfides.