Understanding the role of the deep ocean in governing glacial/interglacial cycles has been a central theme of paleoceanography since its inception. Historically attempts at resolving the changes that occurred in the deep ocean have focused on tying together sedimentary time series from disparate locations. However, the sparse geographic distribution and variable depositional conditions of the sedimentary cores that make up such compilations have been significant obstacles for the creation of a detailed picture of changes in circulation and carbon storage over ice-age cycles.

Vertical tracer profiles represent an alternative approach that circumvents a number of these problems. These profiles (‘depth transects’) compile proxy measurements from sediment cores collected at multiple depths in a single geographic area into a single ‘steady-state’ picture. By densely sampling the water column in a single location, vertical profiles act as ‘paleo-CTD’ casts that provide detailed observations of changes in water mass geometry using sediment cores that have experienced nearly identical depositional conditions.

In this thesis I present proxy observations from a depth transect in the Southeastern Atlantic Ocean for 6 different time points over the last glacial/interglacial cycle. Together these observations provide information about the timing and magnitude of changes in deep ocean temperature, salinity, and carbon storage as glacial conditions evolved over the last 140,000 years.

The efficacy of the vertical depth transect approach depends entirely on how accurately the cores that comprise the depth transect can be correlated and dated. In Chapter 1, I present X-Ray Fluorescence (XRF) observations of bulk sedimentary elemental abundances for the 20 sediment cores that comprise the core collection. In this chapter I demonstrate that these XRF-derived elemental traces have characteristic variability that can be used to correlate cores across the entire range of depths in the collection, and that they can be used to develop a precise stratigraphic framework.

With this knowledge in hand, Chapter 2 focuses on documenting the evolution of the deep ocean temperature structure over the last ice age cycle using paired measurements of benthic foraminiferal Mg/Ca and δ18O. From these measurements I infer that both a change in structure and a significant cooling occurred in the Southern Atlantic between 2-3 km across the Marine Isotope Stage (MIS) 5E/5D transition. At MIS 5A/4 a second cooling took place at mid-depths, and was likely accompanied by an increase in the production of cold, salty southern-sourced water. The deep South Atlantic appears to have mostly cooled to LGM temperatures by MIS 4, though the gradient between northern- and southern-sourced water continued to sharpen between MIS 4 and the LGM.

The changes in stratification implied by the conservative tracers presented in Chapter 2 should have had consequences for the cycling of carbon and nutrients in the deep ocean. Chapter 3 presents concurrent measurements of 7 different tracers (δ13C, U/Ca, Mn/Ca, Nd/Ca, B/Ca, Sr/Ca, and Na/Ca) that provide a way to infer the relative significance of remineralization, air-sea gas exchange, and other nonconservative properties over the last glacial/interglacial cycle. Taken together, the vertical profile measurements in Chapter 3 are consistent with the phenomenon of nutrient deepening in the South Atlantic during cold episodes of the full glacial cycle, and suggest that the increased production of cold, salty southern-sourced water trapped nutrients and carbon at depth in the Southern Atlantic glacial conditions strengthened.

Chapter 4 outlines a complimentary perspective by using newly developed records of the δ13C of atmospheric CO2 in tandem with two time series of deep ocean δ13C from this core collection to document the amount of excess carbon stored in the deep South Atlantic ocean during cold episodes. The magnitude and timing of the δ13C divergence record match other independent attempts to calculate vertical δ13C gradients in the Subantarctic South Atlantic and can be tied directly to the ice core atmospheric CO2 records. Each of these comparisons suggests that enhanced carbon storage at depth in the ocean was indeed responsible for much of the atmospheric δ13C CO2 record, and, by extension, the atmospheric concentration of CO2. Additionally, the similarity of the record of δ13C divergence and a time-series record of circumpolar deep water temperature suggests that carbon trapping in the deep South Atlantic ocean was associated with colder, saltier water, consistent with the mechanisms proposed in Chapters 2 and 3.

This dissertation aims to understand the mechanisms and consequences of the low frequency “unforced” component of variability in the climate system using interannual and decadal variability as a platform to investigate phenomena thought to be sensitive to greenhouse gas emissions. To combat the lack of temporally extended instrumental observations, this dissertation relies on coral geochemistry and Earth System Models to identify the historical range of Pacific climate variability.

Chapter 2 describes the analysis of a roughly 200 year-long coral record from Clarion Island, in the Revillagigedos Archipelago, Mexico. This record expresses substantial decadal variability in the oxygen isotopes, strongly coherent with other paleoproxies in the central equatorial Pacific and in Southern California, indicative of a coordinated system of ocean-atmospheric interaction. To best resolve the key processes involved, “forward models” are employed to interpret the decadal variability in the coral record.

The third chapter complements the first by addressing a mechanism responsible for the modulation of decadal variability in the North Pacific; the Pacific Meridional Mode (PMM). The Community Earth System Model-Last Millennium Ensemble (CESM-LME) is used to estimate the historical range of PMM variability, highlighting the considerable power of natural variability within the system. I investigate the physical processes associated with adjustments in PMM variance, emphasizing that relatively small anomalies in the background state have a significant influence on PMM variability.

Chapter 4 examines the roles of physical and biologically mediated processes on coral calcification from within an equatorial Pacific coral reef. Motivated by the biogeochemical implications of the previous sections, this work examines the manifestation of pH tracers in coral skeletal tissue across a network of diverse, monitored reef sites at Palmyra Atoll. Corals were analyzed for δ11B, trace metal, δ18O, and δ13C composition. Despite a narrow range in the pH of seawater, high variance in δ11B were observed between sites, and within single sites, suggesting that individual coral colonies have differential capacities to regulate pH variability and other environmental stressors.

The origin and influence of the “4.2 kyr event”– a climate excursion characterized by century-long drought hypothesized to have affected civilizations throughout the northern subtropics– is uncertain, despite its registration in multiple paleoclimatic archives. As the dominant source of global interannual climate variability, the El Niño-Southern Oscillation (ENSO) might be expected to have shaped the expression of global precipitation and temperature across the 4.2 kyr event. However, the role of ENSO has not been explored, owing to a general lack of temporally extended observations in the tropical Pacific. Here I analyze a number of tropical Pacific Porites coral fossils spanning this interval, between ~4600 and 4100 years B.P., building on the previously published record of ENSO activity as well as the prevailing mean state of the tropical Pacific climate. Monthly resolved oxygen isotopic records of three corals from Fanning Island and one from Christmas Island overlap with and extend existing records to produce a nearly-continuous sequence that spans several hundred years. Oxygen isotopic values of the overlapping segments of the fossil corals strengthen the reliability of the paleoclimate record and the radiometric dating. The new coral data generated in this study confirm the previous inference of low ENSO activity over an extended time interval encompassing the 4.2 kyr event. Additionally, the new coral data show that the interannual variability, though reduced in amplitude relative to the modern climate, was expressed as repeated episodes of prolonged and intense La Niña conditions. This characterization stands in contrast with a single multi-decadal excursion, as initially hypothesized for the 4.2 kyr event. Finally, the mean values of overlapping coral segments from Fanning and Christmas Islands are similar, suggesting that the modern temperature and salinity gradient was not present during the interval of the 4.2 kyr event. The simplest explanation for this lack of temperature/salinity gradient is that the southern boundary of the North Equatorial Counter Current was shifted north of Fanning Island.

The progression of a geographically specific hydroclimate system over the course of global climatic anomalies and shifts like the Last Glacial Maximum (LGM) and Heinrich Stadial (HS) events gives rise to the strength and impact of these paleoclimate events on the worlds oceans and atmospheric conditions. The interplay of the Intertropical Convergence Zone (ITCZ) location in response to positive feedback loops derived from high latitude hemispheric specific aberrations (HS events) causes shifts in the intensity and spatial variability of global precipitation belts. Here, we show the implications of HS1, HS4, and the LGM on the hydroclimate variability of South Central Africa using barium ratios and the δ18O of individual planktonic foraminifera captured in an ultra-high deposition sequence drilled off the Mozambique Margin at the output of the Zambezi River. In addition to sea surface measurements, a high resolution benthic δ18O record was constructed to locate and observe the expression of these events in the Southern Hemisphere. We find abrupt excursions to low δ18O values in line with HS events indicating these events are characterized by a warming of sub-Antarctic surface waters. In addition, we find that during HS events and the LGM there is evidence for southern shifts in the ITCZ documented by increased precipitation and freshwater discharge into the Mozambique Channel captured by elevated barium levels and negative shifts in δ18O of planktonic foraminifera.

Persistent drought presents one of the greatest risks of climate change faced by the western United States. Highlighting this risk is from paleo hydrological evidence that during the medieval period (~900-1400 AD) this region experienced two extensive droughts of greater duration than any experienced in recorded history. However, much remains to be learned about these “mega droughts,” including the severity of deficit in precipitation and snowpack in California’s Sierra Nevada Mountain range, which currently serve as a crucial source of fresh water to California agriculture, industry and domestic users. To possibly extract a record of the hydrological conditions in the Sierra Nevada during this era, cores of annually laminated sediment from Swamp Lake, in the northwest portion of Yosemite Park were collected. Stable hydrogen isotope (δD) ratios of plant leaf-wax lipid compounds preserved in the lake sediments and the thickness of the annual sedimentary laminations (varves) show promise as potential proxies for recording past hydroclimate variability in the Sierra Nevada Mountains. δD measurements at annual to interannual resolution were made for two time periods: the 20th century and the 13th-15th centuries while varve thickness was measured over the entire period 1150-2006 AD. We find significant negative correlations between 20th century δD fluctuations and instrumentally recorded variability in total precipitation, snow water equivalence (SWE) and Palmer Drought Severity Index (PDSI). Medieval δD values correspond to independent regional hydrologic reconstructions at the decadal to multi-decadal scale, which appear to delineate relatively dry medieval episodes. An issue that is still unclear is that the mean δD during this period is little different from that during the relatively moist contemporaneous period. This finding is at odds with evidence from submerged tree stumps for two prolonged droughts separated by about 200 years. Varve thickness measurements are on going, but results thus far reveal difficulties in using this proxy for reconstructing past Sierra Nevada hydroclimate variability, the most significant being the complexity of the seasonal deposition cycle and sub-millimeter thicknesses. Overall, geochemical and geomorphological information gleaned from Swamp Lake sediment cores provide an independent means of reconstructing hydroclimate in the Sierra Nevada and provide a unique perspective on the mega-drought intervals in the Sierra Nevada Mountains.