UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Isotopic and chemical heterogeneity of the Hawaiian mantle plume : evaluating mantle geodynamics and… Harrison, Lauren Nicole 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_november_harrison_laurennicole.pdf [ 50.36MB ]
Metadata
JSON: 24-1.0357222.json
JSON-LD: 24-1.0357222-ld.json
RDF/XML (Pretty): 24-1.0357222-rdf.xml
RDF/JSON: 24-1.0357222-rdf.json
Turtle: 24-1.0357222-turtle.txt
N-Triples: 24-1.0357222-rdf-ntriples.txt
Original Record: 24-1.0357222-source.json
Full Text
24-1.0357222-fulltext.txt
Citation
24-1.0357222.ris

Full Text

ISOTOPIC AND CHEMICAL HETEROGENEITY OF THE HAWAIIAN MANTLE PLUME: EVALUATING MANTLE GEODYNAMICS AND CHARACTERIZATION OF THE LOA GEOCHEMICAL TREND byLauren Nicole HarrisonB.Sc., The University of Wyoming, 2011A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES(Geological Sciences)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)October 2017© Lauren Nicole Harrison, 2017iiAbstract Oceanic island basalts provide the exceptional opportunity to study deep mantle geochemical reservoirs, mantle geodynamics and, for long-lived systems, the time evolution of their mantle sources. The Hawaiian-Emperor chain represents the geologic record of the long-lived (>81 Ma) and deeply sourced Hawaiian mantle plume. The geochemical record of the entire chain is now complete with analysis of Pb-Hf-Nd-Sr isotopes and elemental compositions of the Northwest Hawaiian Ridge (NWHR), which consists of ~51 volcanoes spanning ~42 Ma between the bend in the chain and the Hawaiian Islands. This segment of the chain previously represented a significant data gap where Hawaiian plume geochemistry changed markedly: only Kea compositions have been observed on Emperor Seamounts (>50 Ma), whereas the Hawaiian Islands (<6 Ma) present both Kea and Loa compositions. Statistical analysis of the new isotopic compositions of NWHR shield-stage basalts confirms the ephemeral presence of the Loa composition. The Hawaiian plume sampled only Kea-type material from the deep Pacific mantle during formation of the Emperor Seamounts and most of the oldest NWHR. Plume movement up the gently sloping edge of the large low-shear velocity province (LLSVP) resulted in entrainment of greater amounts of LLSVP-enriched material over time, which explains why the Hawaiian mantle plume strengthens. Geochemical investigation of NWHR postshield and rejuvenated basalts shows that the rejuvenated component is homogeneous and different from both that of postshield basalts and the depleted component sampled by the oldest Emperor Seamounts. The oldest Emperor Seamounts record interaction between the Hawaiian plume and a mid-ocean ridge. The high Hf isotope ratios of rejuvenated basalts require a source imprinted by ancient partial melting. The robustness of Pb isotope analyses is assessed by investigation of a potential contaminant, ferromanganese crusts, precipitated hydrogenetically on NWHR lavas. Lithium isotopes of Hawaiian Island basalts distinguish between Loa trend heterogeneities; the Loa Abstractiiitrend represents singular components sampled for finite time periods. These conclusions demonstrate how integrated geochemical and geophysical studies of oceanic island basalts can further resolve mantle heterogeneity and challenge us to rethink models of how mantle plumes sample the lower mantle. ivLay Summary Earth is composed of layers: crust, mantle, and core. The mantle represents 84% of Earth’s volume and this reservoir is key to understanding how the planet operates. As scientists only have access to the outermost crust, investigation of the deep mantle requires a geological process that brings deep mantle specimens to where they can be sampled. Mantle plumes transport deep mantle material to the surface, where it melts to create volcanoes such as Hawai‘i. Analyzing the elemental and isotopic compositions of Hawaiian volcanoes, and how they vary between volcanoes of different ages, allow scientists to reconstruct the composition and structure of the deep mantle over time. In this study, the chemistry of Hawaiian volcanoes are analyzed and compared to volcanoes from the entire 5800 km-long chain. As their chemistry varies temporally, the deep mantle must contain different reservoirs that are sampled by the Hawaiian plume as it evolves through time.Lay SummaryvPreface The main thesis chapters (Chapters 2 through 6) were prepared in manuscript format suitable for submission to international geoscience journals. I am the lead author on all five of these manuscripts and all are co-authored by my supervisor, Dr. Dominique Weis, who provided guidance and edits on this entire dissertation. Dr. Michael O. Garcia, a member of my supervisory committee, is a co-author on Chapters 3, 4, and 5 and he provided both samples and expertise on Hawaiian volcanism.  Funding for these studies came from multiple sources. Isotope analyses, major and trace element concentrations of Northwest Hawaiian Ridge basalts were funded by a U.S. National Science Foundation (NSF) Grant EAR-1219955 to Dr. Michael O. Garcia of the University of Hawai‘i at Mānoa and a Canadian NSERC Discovery Grant to Dr. Dominique Weis of UBC (Chapters 3, 4 and 5). Dr. Dominique Weis’ NSERC Discovery Grant also funded the Li isotope study of Hawaiian Island lavas (Chapter 2). Chapter 6 and Appendix B isotope and trace element study of NWHR ferromanganese crusts were funded through a Geological Society of America (GSA) Graduate Student Research Grant awarded to L. Harrison and Dr. Weis’ Discovery Grant. A University of British Columbia Four Year Fellowship supported L. Harrison for part of this PhD. The Pacific Centre for Isotopic and Geochemical Research (PCIGR) was funded in part by NSERC MFA and NSERC MRSP over the duration of this research.   Samples for Li isotope analysis were a part of the existing sample collection of Dr. Dominique Weis. NWHR samples were provided by Drs. Michael O. Garcia and John Smith (Hawaii Undersea Research Laboratory), the Scripps Institution of Oceanography and University of Hawai‘i at Mānoa’s rock collections. Many NWHR samples were prepared, cut, and crushed by Jonathan Tree (University of Hawai‘i at Mānoa). All rock powdering, analytical preparation and instrumental analyses was carried out by me at UBC and using Prefacevithe PCIGR labs. I carried out all data reduction and interpretation, produced all figures and tables, and wrote all drafts of the submitted manuscripts and chapters included in this thesis. Contributions from coauthors are detailed below. My supervisor and co-authors also provided ideas, advice, insightful comments, careful reviewing and edits, and financial support for all of this work.  Chapter 2The Li Isotopic Signature of Hawaiian Basalts –a version of this chapter has been published in the American Geophysical Union “Hawaiian Volcanoes: From Source to Surface” Geophysical Monograph Series volume 208. Harrison, L.N, Weis, D., Hanano, D., Barnes, E., 2015. The lithium isotopic signature of Hawaiian basalts. In: Carey, R., Cayol, V., Poland, P., Weis, D., (Eds.), Hawaiian volcanoes: From source to surface. Geophysical Monograph Series Vol. 208, American Geophysical Union, 74-104.  Dr. Elspeth Barnes analyzed the first ~20 Hawaiian samples for Li isotope compositions that was included in this study, and Diane Hanano provided editorial and intellectual feedback. Dr. Shichun Huang (University of Nevada Las Vegas) and an anonymous reviewer reviewed the manuscript. Dr. Michael Poland edited the manuscript for inclusion in the AGU Geophysical Monograph 208. Chapter 3The Link Between Hawaiian Mantle Plume Composition, Magmatic Flux, and Deep Mantle Geodynamics–a version of this chapter has been published in the journal Earth and Planetary Science Letters. Harrison, L.N., Weis, D., Garcia, M.O., 2017.  The link between Hawaiian mantle plume composition, magmatic flux, and deep mantle geodynamics. Earth and Planetary Science viiLetters 463, 298-309. Dr. Michael O. Garcia provided NWHR samples, insightful comments and edits. Drs. Marcel Regelous (Friedrich-Alexander-Universität Erlangen-Nürnberg) and Matt Jackson (University of California Santa Barbara) reviewed the submitted manuscript for EPSL. Chapter 4The Spatial Distribution and Composition of the Loa Geochemically Enriched Component Along the Northwest Hawaiian Ridge–is intended for publication. Authors: Lauren N. Harrison and Dominique WeisDr. Michael O. Garcia provided NWHR samples, insightful comments and edits. Chapter 5 The Hawaiian Mantle Plume Over >81 Million Years: Tapping Multiple Depleted Sources–is intended for publication. Authors: Lauren N. Harrison, Dominique Weis, and Michael O. GarciaDr. Michael O. Garcia provided NWHR samples, insightful comments and edits. Chapter 6 A ~50 Million Year Record of Pacific Ferromanganese Crust Mineralogical, Trace Element, and Pb Isotope Compositions from the Northwest Hawaiian Ridge–is intended for publication.Authors: Lauren N. Harrison and Dominique Weis viiiTable of ContentsTable of ContentsAbstract                                                                                    iiLay Summary                                                                             ivPreface                                                                                       vTable of Contents                                                                         viiiList of Tables                                                                               xvList of Figures                                                                            xviList of Abbreviations                                                                      xxAcknowledgements                                                                      xxiiDedication                                                                                xxivChapter 1 Introduction to the Hawaiian-Emperor Chain, the Deep Mantle                  and Isotope Geochemistry                                                                  11.1. INTRODUCTION AND RATIONALE OF RESEARCH . . . . . . . . . .11.2. MANTLE PLUMES AND THEIR GEOLOGIC RECORDS  . . . . . . . .51.3. THE HAWAIIAN-EMPEROR CHAIN   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .81.3.1 Geology and Previous Work . . . . . . . . . . . . . . . . . . . . . .81.3.2 Volcanic Stages  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 121.3.3 Magmatic Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131.4. THE DEEP MANTLE  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 151.5. GEOCHEMICAL TOOLS IN THE STUDY OF OCEANIC ISLAND BASALTS   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 181.5.1 Radiogenic Isotopes . . . . . . . . . . . . . . . . . . . . . . . . . 191.5.2 Age Correction of Altered Basalts . . . . . . . . . . . . . . . . . . 211.5.3 Stable Isotopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221.6. FERROMANGANESE CRUSTS  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 23ix  1.7. OVERVIEW OF THE DISSERTATION . . . . . . . . . . . . . . . . . . 261.8. FIELD CONTRIBUTIONS TO RESEARCH  . . . . . . . . . . . . . . . 30Chapter 2 The Lithium Isotopic Signature of Hawaiian Basalts                       342.1. INTRODUCTION  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 342.2. SAMPLES AND METHODS   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 372.3. RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412.3.1 Alteration Control . . . . . . . . . . . . . . . . . . . . . . . . . . 432.3.2 Effect of Crystal Fractionation   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 452.3.3 Lithium Isotopic Signature   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 472.3.3.1 Shield Basalts . . . . . . . . . . . . . . . . . . . . . . . . . . 472.3.3.2 Postshield Basalts.  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 492.3.3.3 Rejuvenated and Preshield Basalts . . . . . . . . . . . . . . . 492.3.3.4 Lithium Isotopes of Hawai‘i Versus Global Oceanic Island  Basalts   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 502.4. DISCUSSION   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 502.4.1 Lithium Isotopic Signature in Hawaiian Basalts: A Source  Signature?  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 502.4.2 Hawaiian End-Member Components in Lithium Isotopes   .  .  .  .  . 522.4.2.1 The Kea Component  . . . . . . . . . . . . . . . . . . . . . . 522.4.2.2 Lō‘ihi: Heterogeneity in Lithium Isotopes . . . . . . . . . . . 552.4.2.3 Hualālai: Light Lithium End-Member  . . . . . . . . . . . . . 582.4.2.4 Lithium Isotopes of the Makapu‘u Section of Ko‘olau  Volcano: A Sedimentary Component . . . . . . . . . . . . . . 612.4.2.5 Lithium Isotopic Heterogeneity in Loa Trend Volcanoes   .  .  . 632.4.3 Why Are Postshield Basalts Lighter in Lithium Isotopic  Signature Than Shield Basalts?  . . . . . . . . . . . . . . . . . . . 64x2.4.4 Is MORB-related Lithosphere or Asthenosphere Assimilated  into the Hawaiian Plume? . . . . . . . . . . . . . . . . . . . . . . 662.4.5 Global Lithium Isotope Systematics . . . . . . . . . . . . . . . . . 662.5. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Chapter 3 The Link Between Hawaiian Mantle Plume Composition, Magmatic Flux, and Deep Mantle Geodynamics                                                    703.1. INTRODUCTION  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 703.2. GEOLOGICAL SETTING AND SAMPLING . . . . . . . . . . . . . . . 733.3. ANALYTICAL TECHNIQUES AND AGE CORRECTION  .  .  .  .  .  .  . 753.4. RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 823.5. DISCUSSION   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 873.5.1 Role of the Lithosphere in Magmatic Flux and Pb Isotopic  Composition of NWHR Basalts . . . . . . . . . . . . . . . . . . . 873.5.2 Role of the LLSVP in Magmatic Flux and Pb Isotopic  Composition of NWHR Basalts . . . . . . . . . . . . . . . . . . . 893.6. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95Chapter 4 The Spatial Distribution and Composition of the Loa Mantle         Component Along the Northwest Hawaiian Ridge                                      974.1. INTRODUCTION  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 974.2. GEOLOGICAL SETTING AND SAMPLE LOCATIONS   .  .  .  .  .  .  .  . 994.3. ANALYTICAL TECHNIQUES  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1004.4. ALTERATION OF SAMPLES AND AGE CORRECTION OF  ISOTOPIC RATIOS  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1054.5. RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1084.5.1 Major Elements  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1084.5.2 Trace Elements   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  111xi4.5.3 Nd-Sr-Hf Isotopic Compositions  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1154.6. DISCUSSION   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1214.6.1 The Spatial Distribution of the Loa Geochemical Trend Along  the NWHR: A Logistic Regression Analysis  .  .  .  .  .  .  .  .  .  .  .  1214.6.2 Elemental and Isotope Systematics of NWHR basalts  .  .  .  .  .  .  1274.6.3 A New Hf-Nd Hawaiian Array   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1314.6.4 The Lō‘ihi End-Member Component Along the NWHR  . . . . . 1324.6.5 A New Perspective on Loa Trend Heterogeneity From the  NWHR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1344.7. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136Chapter 5 The Hawaiian Mantle Plume Composition Over >81 Million Years: Evidence for Multiple Depleted Sources                                               1385.1. INTRODUCTION  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1385.2. GEOLOGICAL SETTING AND SAMPLE LOCATIONS   .  .  .  .  .  .  .  1435.3. TEMPORAL AND SPATIAL EXTENT OF REJUVENATED  BASALTS ALONG THE HAWAIIAN-EMPEROR CHAIN  .  .  .  .  .  .  1455.4. ANALYTICAL TECHNIQUES  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1475.5. RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1495.5.1 Assigning NWHR Basalts to a Volcanic Stage  .  .  .  .  .  .  .  .  .  .  1495.5.2 Major and Trace Elements of NWHR Postshield Basalts . . . . . 1545.5.3 Major and Trace Elements of NWHR Rej-type Basalts . . . . . . 1605.5.4 Pb-Hf-Nd-Sr Isotopic Compositions of NWHR Postshield  Basalts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1615.5.5 Pb-Hf-Nd-Sr Isotopic Compositions of NWHR Rej-type  Basalts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1675.6.  DISCUSSION  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  168xii5.6.1 NWHR Postshield Basalts: Preservation of the Shield-Stage  Loa and Kea Trend Distinctions . . . . . . . . . . . . . . . . . . 1685.6.2 NWHR Rej-type Basalts: A Homogeneous Rejuvenated  Component Over ~13 Myr   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1715.6.3 The Depleted Mantle Component of the Oldest Emperor  Seamounts: Mid-Ocean Ridge Interaction  . . . . . . . . . . . . 1725.6.4 Hawaiian Rejuvenated Magmatism: A Time-Tested Process  . . . 1775.6.5 Hawaiian Rejuvenated Volcanism Compared to Other                   Oceanic Islands  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1805.7. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182Chapter 6 A ~50 Million Year Record of Pacific Ferromanganese Crust Mineralogical, Trace Element, and Pb Isotope Compositions from the Northwest Hawaiian Ridge                                                              1846.1. INTRODUCTION  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1846.2. SAMPLE LOCATIONS  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1866.3. ANALYTICAL METHODS  . . . . . . . . . . . . . . . . . . . . . . . 1896.3.1 X-Ray Diffraction (XRD) Method   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1906.3.2 Pb Isotope Analysis Method . . . . . . . . . . . . . . . . . . . . 1906.3.3 Trace Element Analysis Method . . . . . . . . . . . . . . . . . . 1916.4. RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1916.4.1 Ferromanganese Crust XRD Results  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  1936.4.2 Pb Isotope and Trace Element Results   .  .  .  .  .  .  .  .  .  .  .  .  .  .  1986.5. DISCUSSION   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  2046.5.1 Heterogeneous and Nano-sized Mineralogy of NWHR  Ferromanganese Crusts  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  2046.5.2 Trace Element Signature of NWHR Ferromanganese Crusts: xiiiHydrogenetic Origin . . . . . . . . . . . . . . . . . . . . . . . . 2066.5.3 Pb Isotopes of NWHR Ferromanganese Crusts: Are They a  Source of Post-eruption Contamination?  .  .  .  .  .  .  .  .  .  .  .  .  .  2086.5.4 Pb Isotopes of NWHR Ferromanganese Crusts: An  Oceanographic Record   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  2096.6. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215Chapter 7 Conclusions                                                                  2177.1. SUMMARY OF THE DISSERTATION AND KEY RESEARCH  FINDINGS  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2177.1.1 Li Isotopic Composition of Hawaiian Island Basalts  . . . . . . . 2177.1.2 Spatial Location and Geochemical Characteristics of the Loa Geochemical Trend Along the Northwest Hawaiian Ridge . . . . 2207.1.3 The Composition of NWHR Postshield and Rejuvenated  Basalts: Implications for Hawaiian Depleted Components                    and Hawaiian Volcanic Stage Evolution . . . . . . . . . . . . . . 2227.1.4 NWHR Ferromanganese Crusts . . . . . . . . . . . . . . . . . . 2237.2. SUGGESTIONS FOR FURTHER RESEARCH . . . . . . . . . . . . . 2247.2.1 The Northwest Hawaiian Ridge: Still Much to Learn . . . . . . . 2247.2.2 The Young Emperor Seamounts: Looking for Depleted  Components by Completing the Hf Isotope Record . . . . . . . . 2267.2.3 The Loa Trend Keystone in the Pacific? High-Resolution  Geochemical Study of Nīhoa, Middle Bank, and Kaua‘i  . . . . . 2267.2.4 A Geodynamic Modeling Test of the NWHR Hawaiian  Mantle Plume Geochemical Model  . . . . . . . . . . . . . . . . 2287.3. CONCLUDING STATEMENT   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  229Bibliography                                                                              230xivAppendix A Analytical Precision and Accuracy of Radiogenic Isotope Analyses                                                                                   272Appendix B Pb Isotope and Trace Element Characterizaion of USGS Reference Materials NOD-P-1 and NOD-A-1                                          281Appendix C Compiled Northwest Hawaiian Ridge Basalt Sample Locations       287Appendix D List of Publications and Abstracts for Presentations at National and International Meetings Resulting from Dissertation Research                  289xvList of TablesTable 2 1: Literature compilation of USGS reference material BHVO-2 lithium isotope analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40Table 2 2: Lithium isotopic compositions of Hawaiian basalts   .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 42Table 2 3: Input model parameters for Figures 2.8, 2.9, 2.10   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 58Table 3 1: Sample locations and collection methods . . . . . . . . . . . . . . . . . . . . . 74Table 3 2: Pb isotopic composition and Pb, U, and Th concentrations of Northwest Hawaiian Ridge basalts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81Table 4 1: Trace element concentrations of Northwest Hawaiian Ridge basalts . . . . . . 101Table 4 2: Sr-Nd-Hf isotopic compositions of Northwest Hawaiian Ridge basalts  .  .  .  . 104Table 4 3: Pb-Hf-Nd-Sr logistic regression analysis of NWHR geochemical affinity . . . 125Table 5 1: NWHR postshield and rej-type basalt sample locations   .  .  .  .  .  .  .  .  .  .  .  . 144Table 5 2: Trace element compositions of NWHR postshield and rej-type basalts   .  .  .  . 149Table 5 3: Pb isotopic compositions of NWHR postshield and rej-type basalts . . . . . . 151Table 5 4: Sr-Nd-Hf isotopic compositons of NWHR postshield and rej-type basalts . . . 152Table 6 1: Northwest Hawaiian Ridge ferromanganese crust locations  . . . . . . . . . . 189Table 6 2: Trace element concentrations and Pb isotope compositions of Northwest Hawaiian Ridge ferromanganese crusts   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 197Table A 1: External reproducibility of Pb isotope analyses   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 275Table A 2: External reproducibility of Hf, Nd, and Sr isotope analyses . . . . . . . . . . 276Table A 3: Internal reproducibility (replicates) of Pb isotope analyses   .  .  .  .  .  .  .  .  .  . 277Table A 4: Internal reproducibility (replicates) of Hf and Nd isotope analyses  .  .  .  .  .  . 278Table A 5: Pb isotope compositions by MC-ICP-MS of Hawaiian basalt reference materials analyzed in this study . . . . . . . . . . . . . . . . . . . . . . . . . 279Table A 6: Nd-Sr-Hf isotope compositions by MC-ICP-MS of Hawaiian basalt       reference materials analyzed in this study  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 280Table B 1: Trace element concentrations of ferromanganese reference materials . . . . . 285Table B 2: Pb isotopic analyses by MC-ICP-MS of ferromanganese reference materials . 286Table C 1: Compiled Northwest Hawaiian Ridge basalt sample locations   .  .  .  .  .  .  .  . 288List of TablesxviList of FiguresFigure 1 1: Bathymetric map of the >100 seamounts and islands of the Hawaiian- Emperor chain  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  9Figure 1 2: The Loa and Kea geochemical and geographic trends on the Hawaiian    Islands  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 11Figure 1 3: Estimated magmatic flux of the Hawaiian plume for the Hawaiian-        Emperor chain  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 14Figure 1 4: Seismic and geodynamic models of mantle structure and convection.  .  .  .  .  . 16Figure 1 5: Isotope geochemistry reveals substantial heterogeneity in the mantle   .  .  .  .  . 19Figure 1 6: Schematic illustration of lithium isotope systematics . . . . . . . . . . . . . . 22Figure 1 7: Ferromanganese crusts on NWHR samples and their potential to     contaminate Pb isotope analyses of basatls .   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 24Figure 1 8: Photographs showing the highlights of the Schmidt Ocean Institute 2014     Leg 2 PMNM bathymetric mapping expedition . . . . . . . . . . . . . . . . . 33Figure 2 1: Map of Hawai‘i and sample locations . . . . . . . . . . . . . . . . . . . . . . 37Figure 2 2: ODP Site 843 leaching results . . . . . . . . . . . . . . . . . . . . . . . . . . 39Figure 2 3: Alteration indices of Hawaiian basalts   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 44Figure 2 4: Sc (ppm) versus weight percent MgO plot.  . . . . . . . . . . . . . . . . . . . 46Figure 2 5: Histograms of lithium isotopic data of this study combined with data from Chan and Frey (2003).  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48Figure 2 6: Li/Y versus δ7Li for compiled oceanic island basalts.   .  .  .  .  .  .  .  .  .  .  .  .  . 51Figure 2 7: Comparison of the Pb and Li isotopic characteristics for preshield, shield, postshield and rejuvenated volcanism on Hawai‘i.  . . . . . . . . . . . . . . . 54Figure 2 8:  Nd versus Li isotopes for (a) shield and preshield basalts and (b) postshield and rejuvenated basalts.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56Figure 2 9: Sr versus Li isotopes for (a) shield and preshield basalts and (b) postshield   List of Figuresxvii and rejuvenated basalts.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57Figure 2 10: Plots of shield-stage (a) La/Nb and (b) Sr/Nb versus δ7Li .  . . . . . . . . . . 60Figure 2 11: A compilation of oceanic island basalts with lithium isotopic data  . . . . . . 67Figure 3 1: Bathymetric map of the ~51 seamounts and islands of the Northwest  Hawaiian Ridge and sample locations. . . . . . . . . . . . . . . . . . . . . . . 72Figure 3 2: 207Pb/204Pb versus 206Pb/204Pb of Northwest Hawaiian Ridge shield-stage   basalts   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 80Figure 3 3: Pb-Pb isotopic diagrams for Northwest Hawaiian Ridge shield-stage       basalts   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 83Figure 3 4: Radiogenic Pb and lithospheric thickness versus age of the seafloor at time     of eruption  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 85Figure 3 5: Northwest Hawaiian Ridge radiogenic Pb variations with Hawaiian plume estimated volume flux  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 86Figure 3 6: Model evolution of the Hawaiian plume source at the CMB  . . . . . . . . . . 91Figure 4 1: Bathymetric map of the ~51 seamounts and islands of the Northwest     Hawaiian Ridge and sample locations . . . . . . . . . . . . . . . . . . . . . 100Figure 4 2: Alteration indices of NWHR shield basalts  . . . . . . . . . . . . . . . . . . 106Figure 4 3: Total alkali versus silica plot of NWHR shield basalts  .  .  .  .  .  .  .  .  .  .  .  . 109Figure 4 4: MgO variation plots of selected NWHR shield basalt major and minor elements   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 110Figure 4 5: Normalized major element abundances of NWHR and Hawaiian Island    shield basalts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111Figure 4 6: Primitive mantle-normalized extended trace element patterns of NWHR   shield basalts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112Figure 4 7: Chondrite-normalized rare earth element patterns of NWHR shield basalts  . 113Figure 4 8: Trace element ratio variation plots of NWHR shield basalts  .  .  .  .  .  .  .  .  . 114Figure 4 9: Isotopic ratios of Sr, Nd, and Hf versus distance from Kīlauea for NWHR xviiishield basalts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116Figure 4 10: Nd, Sr, and Hf isotopic variation diagrams for NWHR shield basalts . . . . 118Figure 4 11: Pb, Sr, and Hf isotopic variation diagrams for NWHR shield basalts  . . . . 120Figure 4 12: Receiver operating characteristic curve of the Hawaiian Island logistic regression testing dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126Figure 4 13: Pb isotopic composition of Hawaiian basalts younger than the Hawaiian-Emperor bend   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 133Figure 5 1: Bathymetric map of the Hawaiian-Emperor chain   .  .  .  .  .  .  .  .  .  .  .  .  .  . 142Figure 5 2: Total alkali versus silica plot of NWHR postshield and rej-type basalts  .  .  . 154Figure 5 3: Trace element plots for classifying volcanic stage and mantle source   .  .  .  . 155Figure 5 4: MgO variation plots of selected NWHR rej-type and postshield basalt      major and minor elements  . . . . . . . . . . . . . . . . . . . . . . . . . . . 156Figure 5 5: Primitive mantle-normalized extended trace elements and chondrite-normalized rare earth elements of NWHR postshield and rej-type            basalts   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 157Figure 5 6: Trace element ratio variation plots of NWHR rej-type and postshield      basalts   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 159Figure 5 7: Trace element discrimination diagrams for NWHR rej-type basalts   .  .  .  .  . 161Figure 5 8: Nd, Sr, and Hf isotope compositions of NWHR rej-type and postshield   basalts   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 162Figure 5 9: Pb and Sr isotopic compositions of NWHR rej-type and postshield          basalts   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 163Figure 5 10: Nd and Hf isotopic variation of NWHR rej-type and postshield basalts . . . 165Figure 5 11: Isotopes versus distance from Kīlauea for the Hawaiian-Emperor chain   .  . 166Figure 5 12: Isotopic ratios versus primitive mantle-normalized La/Yb for the     Hawaiian-Emperor chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169Figure 5 13: Epsilon Hf versus Lu/Hf for the Hawaiian-Emperor chain   .  .  .  .  .  .  .  .  . 175xixFigure 6 1: Bathymetric map of the ~51 seamounts and islands of the Northwest   Hawaiian Ridge and ferromanganese crust sample locations  . . . . . . . . . 187Figure 6 2: Ferromanganese crust samples from NWHR seamounts  .  .  .  .  .  .  .  .  .  .  . 188Figure 6 3: Results from preliminary XRD method experiments from bulk XRD     analysis of reference material USGS NOD-P-1  . . . . . . . . . . . . . . . . 192Figure 6 4: XRD results from NWHR ferromanganese crusts . . . . . . . . . . . . . . . 194Figure 6 5: Comparison figure of all NWHR bulk powder ferromanganese crust           X-ray diffractograms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195Figure 6 6: Pb isotope variations of NWHR ferromanganese crusts . . . . . . . . . . . . 199Figure 6 7: NWHR ferromanganese crust Pb isotope ratios versus age . . . . . . . . . . 200Figure 6 8: Trace element compositions of NWHR ferromanganese crusts versus     distance from Kīlauea   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 201Figure 6 9: North American shale composite-normalized rare earth element patterns        of NWHR ferromanganese crusts  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 202Figure 6 10: Genetic discrimination diagrams of NWHR ferromanganese crusts . . . . . 203Figure 6 11: The full width half-maximum (FWHM) of the 2.10 Å vernadite peak in NWHR ferromanganese crusts . . . . . . . . . . . . . . . . . . . . . . . . . 204Figure 6 12: Pacific ferromanganese crust locations and the provenance of Pb isotope fluxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212Figure 7 1: Evolution of the oceanic island basalt and mantle plume theory paradigm . . 218Figure 7 2: Contributions of this research to Hawaiian geochemistry . . . . . . . . . . . 219Figure 7 3: Detailed map of the “keystone” section of the NWHR between Nīhoa           and Kaua‘i  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 227Figure B 1: NOD-P-1 and NOD-A-1 multielement trace element averages . . . . . . . . 282Figure B 2: NOD-P-1 and NOD-A-1 duplicate Pb isotope analyses   .  .  .  .  .  .  .  .  .  .  . 283xxList of AbbreviationsAbbreviations and Symbols:Analytical Terms:δ7Li = Delta notation for stable isotopic ratio relative to a standard‰ = parts per milεNd = epsilon neodymiumεHf = epsilon hafniumSD = standard deviationSE = standard error2σ = 2*standard deviationN = normality of an acid solutionppm = parts per millionppb = parts per billionwt% = weight percentkm = kilometersm = metersg = gramss = secondmg = milligramng = nanogrampg = picogramMa = millions of years before presentMyr = elapsed time interval, measured in millions of yearsGa = billion yearsKa = thousand yearsICP-MS = inductively coupled plasma mass spectromemterMC-ICP-MS = multicollector ICP-MSHR-ICP-MS = high resolution ICP-MSTIMS = thermal ionization mass spectrometerRSD = relative standard deviationXRF = X-ray fluorescenceList of AbbreviationsxxiXRD = X-ray diffractionUSGS = United States Geological SurveyHawaiian and Geochemistry:NWHR = Northwest Hawaiian Ridge H-E = Hawaiian-EmperorPMNM = Papahānaumokuākea Marine National MonumentOIB = oceanic island basaltLIP = large igneous provinceODP = Ocean Drilling ProgramEPR = East Pacific RiseMORB = mid-ocean ridge basaltHSDP = Hawai‘i Scientific Drilling ProjectEM-I = enriched mantle oneEM-II = enriched mantle twoHIMU = high-238U/204PbPREMA = prevalent mantleSC = seamount chainPM = primitive mantleREE = rare earth elementsHREE = heavy rare earth elementsLREE = light rare earth elementsAll Hawaiian place names are spelled using the correct Hawaiian spelling as referenced in Sherrod et al., 2007. xxiiAcknowledgements This PhD has been a six-year expedition that has benefitted from the contributions of many. First and foremost, I would like to thank my thesis advisor, Dr. Dominique Weis, whose personal and scientific support, confidence in me, guidance, humor, edits, mentorship, expertise, professional advocacy, and friendship were integral to the success of this research (all the margaritas, wine, and chocolate didn’t hurt either). Timely guidance from Dr. James Scoates always hit the mark when faced with a challenge, and his intellectual input over the years helped greatly to develop my research and professional skills (I will never forget his crash course in “How to Teach”). Many of the highlight experiences of my PhD, especially the chance to participate in the SOI 2014 PMNM bathymetric mapping expedition, were op-portunities extended to me by Dr. Michael Garcia, for which I am very grateful (in addition to all of the information on Hawaiian volcanology and petrology).  I am very grateful to the research staff at the PCIGR, who works tirelessly to keep the clean preparatory and analytical labs running smoothly and taught me almost everything I know about high-precision clean analytical chemistry. Thank you to Kathy Gordon, Vivian Lai, Richard Friedman, Liyan Xing, Bruno Kieffer, Jane Barling, Yeena Feng, Taylor Ocker-man, Hai Lin, and Margharay Amini. I also thank Dr. Mati Raudsepp, Edith Czech, Elisabetta Pani, Jenny Lai, and Lan Kato for training and assistance with XRD analysis of ferromanga-nese crusts and basalts. The PCIGR support staff made the administrative side of my research run smoothly, and for that I thank Diane Hanano and Cecilia Li. Development as a Hawaiian researcher was greatly augmented from weekly discussions, and for that I thank the ladies of HOT SASS: Nicole Williamson, Catherine Armstrong, Diane Hanano, and Dominique Weis (with occasional cameos by Rhy McMillan). Finally I thank my fellow PCIGR students, for all the support, friendship, editing, and coffee runs over the years with a special thanks to the people who assisted with thesis editing: Nicole Williamson, Anaïs Fourny, Genna Patton, AcknowledgementsxxiiiRhy McMillan, Nichole Moerhuis, and Laura Bilenker. Thank you to my friends, without whom the last six years would not have been the lively, rich experience it was.  Thank you to my family. Paul and Sherri-lyn, my parents, who have given me so much emotionally, spiritually, mentally, physically. My success in life is a result of the les-sons you have taught, the tools you have given me, your continued love and support. Thank you. xxivDedicationThis work is dedicated to Stephanie Elizabeth Harrison, my dear sister. I could not have done any of this without you.  Dedication1Chapter 1 Introduction to the Hawaiian-Emperor Chain, the Deep Mantle and Isotope GeochemistryChapter 1 Introduction to the Hawaiian-Emperor Chain, the Deep Mantle, and Isotope Geochemistry1 1  INTRODUCTION AND RATIONALE OF RESEARCH Isotopic analysis of oceanic island basalts (OIB) provides essential information on the composition and homogeneity of the Earth’s mantle today and throughout geologic time. The mantle comprises more than three-fourths of the volume of the Earth and its internal dynamics drive many global processes such as plate tectonics and the formation of continental crust. Furthermore, deciphering this reservoir is key to modeling the Earth’s compositional evolution and budgeting the element fluxes between different Earth reservoirs. Oceanic island magmatism is one of the best “windows” into the mantle because this volcanism typically samples the mantle without interfering signals from interaction with continental crust or subcontinental lithospheric mantle. In addition, OIB can originate from different depths within the mantle, ranging from the upper mantle at mid-ocean ridges and shallow plume upwellings to the core mantle boundary at ~2,800 kilometers depth via deep mantle plumes, thus sampling a wide representation of mantle composition. Over the years, studies have increasingly shown evidence that the mantle is much more compositionally heterogeneous than previously estimated (Zindler and Hart, 1986; Stracke et al., 2005; Boyet and Carlson, 2006; Cabral et al., 2013; Hofmann et al., 2014; White, 2015; Mundl et al., 2017). The evidence for such heterogeneity is in part due to advances in mass spectrometry, which have greatly improved the precision of geochemical measurements and which allow for a finer resolution of the chemical and isotopic variations in mantle-derived 2rocks. Analysis of non-traditional and stable isotope systems such as W, S, Li, Si, B, Os, Ca, Mo, Fe, Mg, Xe, and Ne has allowed for deciphering more processes responsible for mantle heterogeneity (Galer, 1999; Thirlwall, 2000; Woodhead, 2002; Albarède et al., 2004; Kamenov et al., 2004; Boyet and Carlson, 2006; Weis et al., 2006, 2007; Nobre-Silva et al., 2009, 2010; Jackson et al., 2010; Cabral et al., 2013; Harrison et al., 2015; Mundl et al., 2017).  Variations in global shear wave velocity for the entire depth of the mantle can be imaged with much improved resolution, leading to better constrained global seismic tomography models. These geophysical models show that two domains of seismically slow material thousands of kilometers wide exist beneath the Pacific Ocean and continental Africa (‘large low-shear velocity provinces’, LLSVP; Ishii and Tromp, 1999; Garnero, 2000; Mégnin and Romanowicz, 2000; Ritsema, 2004; Garnero and McNamara, 2008; French and Romanowicz, 2015; Garnero et al., 2016). The seismic wave variations in the lower mantle may be caused by differences in temperature or composition, and the study of OIB geochemistry can provide key information as to the potential cause and magnitude of these variations. For example, recent geochemical studies have proposed that the sources of enriched material sampled by Hawai‘i and other OIB originate from these LLSVP thermo-chemical piles (Huang et al., 2011a; Weis et al., 2011; Chauvel et al., 2012; Payne et al., 2012; Harpp et al., 2014a; Williams et al., 2014; Hoernle et al., 2015; Garnero et al., 2016).  The mantle plays a fundamental role in the formation and subsequent recycling of oceanic and continental crust, is a storage reservoir for compositionally distinct material (e.g., primordial and recycled components), plays a major role in the Earth’s heat budget via both the decay of radioactive elements and thermo-chemical convection, and contains the sources of mid-ocean ridge and intraplate volcanism (White, 1985; Zindler and Hart, 1986; Hofmann, 1997; 2014; Albarède and van der Hilst, 2005; Labrosse et al., 2007; White, 2015). As such, the ability to understand Earth’s compositional evolution is dependent on being able 3to constrain mantle heterogeneities in terms of composition, size, age, and distribution. Mantle plumes are narrow, roughly cylindrical pipes of upwelling anomalously hot mantle that partially melt to form hotspot volcanoes, providing a “window” into the composition of the mantle. The Hawaiian mantle plume is a well-sampled, long-lived, and tectonically simple hotspot (Jackson, 1972; Clague and Dalrymple, 1987; Sherrod et al., 2007; Tanaka et al., 2008; Garcia et al., 2015). As such, it provides an excellent opportunity to study mantle heterogeneity in time and space. Because strong mantle plumes such as the Hawaiian plume originate at the core mantle boundary, the study of Hawaiian basalts is an optimal geochemical window into the type of material stored in the deep mantle, whether that be homogenized and previously melted mantle, recycled crustal materials, or isolated primordial reservoirs (Kurz et al., 1982; Blichert-Toft et al., 1999; Courtillot et al., 2003; Labrosse et al., 2007; Wolfe et al., 2009, 2011; Jackson et al., 2010; Cabral et al., 2013; Nobre Silva et al., 2013a; Williams et al., 2014; French and Romanowicz, 2015). In addition, Hawaiian volcanoes can be delineated into two geochemical and geographical trends: the “Kea” trend with higher 206Pb/204Pb, 143Nd/144Nd, and 176Hf/177Hf isotope ratios and the “Loa” trend with higher 87Sr/86Sr isotope ratios and incompatible trace element concentrations (Dana, 1891; Jackson, 1972; Tatsumoto, 1978; Abouchami et al., 2005; Tanaka et al., 2008; Weis et al., 2011; Frey et al., 2016). Much debate persists over the type and origin of distinctive source material that would account for the presence of these two trends and of the mantle dynamics that keep these reservoirs separated over >5 million years (Hauri, 1996; Bianco et al., 2008; Tanaka et al., 2008; Farnetani and Hofmann, 2009; Ballmer et al., 2011; Weis et al., 2011; Farnetani et al., 2012; Harrison et al., 2017). The Hawaiian-Emperor chain records >81 million years of hotspot activity, providing the opportunity to study the long-term evolution of a mantle plume over a period of time during which significant changes have taken place tectonically (e.g., the bend in the Emperor-Hawaiian seamount chain; Sharp and Clague, 2006), geochemically (e.g., the emergence of the Loa geochemical trend 4somewhere along the Northwest Hawaiian Ridge; Garcia et al., 2015), and in its magmatic flux (e.g., a 650% increase over the last 40 million years; Van Ark and Lin, 2003; Vidal and Bonneville, 2003; Wessel, 2016). The research presented in this dissertation aims to characterize the composition and spatial distribution of mantle heterogeneities sampled by the Hawaiian mantle plume over its lifetime. This goal is first accomplished by filling in a high-precision data gap that existed for ~47% of the Hawaiian-Emperor chain along the Northwest Hawaiian Ridge (NWHR) using trace elements and isotopes (both radiogenic and stable). These geochemical results provide the basis to explore causes of magmatic flux variations that may be directly related to the lithology of the mantle source (i.e., different materials have different pressures and temperatures of solidi that will result in melts with different elemental and isotopic compositions; Pertermann and Hirschemann, 2003; Sobolev et al., 2005, 2007; Gurenko et al., 2010; Putirka et al., 2011; Lambart et al., 2012; Matzen et al., 2017). In addition, the first large-scale study of Li isotopic composition of Hawaiian Island basalts is undertaken to investigate the nature of the chemical heterogeneity present in Hawaiian basalts (Lithium is a stable isotope system fractionated during subduction and can thus be used to trace recycled materials; Tang et al., 2010). The research in this dissertation investigates several disputed topics about Hawaiian volcanism including:1. The relation between of Loa and Kea trend composition over both short and long periods of Hawaiian hotspot activity;2. The location and timing of the emergence of the Loa geochemical trend;3. A potential relationship between the appearance of the Loa trend with the increase in magmatic flux;4. The existence of Hawaiian volcanic stages and possible mantle source variations between shield, postshield, and rejuvenated basalts;5. The spatial distribution and origin of deep mantle heterogeneities sampled by 5the Hawaiian plume over its lifetime and the large-scale mantle dynamics that produce, preserve, and eventually samples them.  This PhD research provides the perspective on a long-lived volcanic chain, which makes it a useful comparison to other studies of oceanic island basalts (OIB), large igneous provinces (LIP), and mid-ocean ridge basalts (MORB). It also supplies important constraints on the composition, size and origin of deep mantle heterogeneities sampled by Hawaiian volcanoes by testing the temporal distribution of Hawaiian Island enriched and depleted compositional components. This research contributes high-precision data to a poorly understood region along the Hawaiian hotspot track and as such provides the means to examine a long record of Hawaiian mantle plume geochemistry, one of the longest records available globally. This data, interpreted in the framework of available data from the Hawaiian Islands, provides the context necessary to test hypotheses regarding Hawaiian plume dynamics over time and thus may potentially transform the current understanding of Hawaiian geochemistry. The following sections in this introductory chapter cover the background information germane to the topics covered in this dissertation. These topics include mantle plumes, the geology of the Hawaiian-Emperor chain and Hawaiian volcanoes; previous geochemical research on the Hawaiian-Emperor chain; a brief summary of current knowledge of the deep mantle, radiogenic isotopes, stable isotopes, and age correction of old altered basalts; and background on ferromanganese crusts, a potential source of contamination for submarine basalts. This introductory chapter then concludes with an overview of the contents of this dissertation.  1 2  MANTLE PLUMES AND THEIR GEOLOGIC RECORDS The Earth is a rocky planet that contains heat from primordial accretion and differentiation as well as from the decay of radioactive elements (Holmes, 1931; Jaupart et al., 2007 and references therein). As a result, the Earth convects to lose heat and achieve 6thermal equilibrium (Holmes, 1931; Pearson, 1958; Bercovici, 2011). This convection is approximated by Rayleigh-Bénard style fluid dynamics and is accomplished by slow, plastic flow of mantle rocks, where mantle heated by the core buoyantly rises and cold, dense material (mantle or crust) sinks (Pearson, 1958; Wilson, 1963; Morsei et al., 1996; Tackley, 1998; Bercovici, 2011 and references therein). This dynamic process drives plate tectonics, i.e., the movement of lithospheric plates on the surface of the Earth that explains most of the occurrences of surface volcanism, major earthquake locations, plate velocities, seafloor magnetism, Earth chemical segregation, and gravity anomalies (Wegener, 1915; Bercovici, 2011 and references therein). The occurrence of volcanic oceanic islands such as Hawai‘i, Kerguelen, and Tristan da Cunha that are not located near plate boundaries cannot readily be explained by plate tectonic theory (Wilson, 1965; Morgan, 1971, 1972; Courillot et al., 2003). This volcanism is typically attributed to mantle plumes—narrow, roughly cylindrical pipes of upwelling anomalously hot mantle that partially melts to form hotspot volcanoes (Wilson, 1963; Morgan, 1971, 1972; Coffin and Eldholm, 1994; Jellinek and Manga, 2004; Montelli et al., 2006; French and Romanowicz, 2015). The existence of these features and their source in the deep mantle, however, is debated (Anderson, 2001; Saunders, 2005; Foulger, 2007; Anderson and Natland, 2014).  Mantle plumes typically exhibit higher magmatic production at their initial arrival at the surface that diminishes with time, i.e., from melting the plume head to the plume tail (LIP; White and McKenzie, 1989; Campbell and Griffiths, 1990; White, 1993; Jellinek and Manga, 2004; Kumagi et al., 2008). This geodynamic behavior produces predictable geologic features related to the melting and uplift produced by the substantial thermal anomalies of mantle plumes (Campbell, 2007). Plume-related features that have been observed in the geologic record include a connection between linear hotspot tracks and large igneous provinces (LIP), uplift and crustal thickening previous to LIP emplacement, the distribution of picrites in LIP, observed melt volumes in LIP that are impossible under 7normal mantle decompression melting, and finite frequency seismic tomography models that have detected the presence of plumes from their active surface volcanoes all the way to the core mantle boundary (Campbell and Griffiths, 1990; White and McKenzie, 1995; Montelli et al., 2006; Campbell, 2007; French and Romanowicz, 2015).  Other explanations for non-plate tectonic related volcanism include shallow, broad upwellings in the upper mantle, small-scale mantle convection, propagating lithospheric cracks from extensional stress fields, bolide impacts, lithospheric delamination, and back arc processes (Wilson, 1963; Morgan, 1971, 1972; Campbell and Griffiths, 1990; Anderson, 2001; Jellinek and Manga, 2004; Saunders, 2005; Foulger, 2007; Anderson and Natland, 2014). Currently, the scientific consensus favors a plume-related origin to account for non-plate tectonic related OIB volcanism (Putirka et al., 2005; Herzberg et al., 2007; Campbell, 2007; French and Romanowicz, 2015). There are different categories of mantle plumes. Primary plumes are characterized by seismic velocity anomalies of at least -1.5% for 1,000-2,800 km mantle depths and require the presence of a linear hotspot track with a sequential age progression, the presence of a LIP at the origin of the hotspot track, a large magmatic flux, and high 3He/4He (suggesting a geochemical source different from MORB) (Courillot et al., 2003; French and Romanowicz, 2015). Hawai‘i, Afar, Canary, Cape Verde, Comores, Iceland, Macdonald, Marquesas, Pitcairn, Samoa, Tahiti, and possibly the Galápagos, Cameroon, Easter, Kerguelen, Tristan, Louisville, St. Helena, and Réunion are all deeply sourced, potentially primary, mantle plumes (Courtillot et al., 2003; French and Romanowicz, 2015). Plumes can also be shallow, weak, and not underlain by continuous pipes of low shear wave velocity mantle such as many short-lived island chains (e.g., Azores, Jan Mayen, Crozet, Caroline, etc.), or continental hotspots (e.g., Yellowstone). There are also erratic volcanoes that do not appear to be associated with any plume or form any age progressive sequence at all, such as the individual volcanoes scattered over the Pacific Ocean abyssal plain (Courtillot et al., 2003; 8Kumagai et al., 2008; French and Romanowicz, 2015). The locations of deep, strong mantle plumes are correlated with the edges of the LLSVP, whose locations remain relatively fixed throughout geologic time (Burke and Torsvik, 2008). The variety observed in global mantle plumes illustrates a dynamic Earth system by which not all volcanism is produced through the same processes or in the same geodynamic environments. Factors such as whether plumes are shallow or deeply sourced, long-lived or weak, erupting through continental crust or near a mid-ocean ridge, should be kept in mind when interpreting the different geophysical and geochemical properties of different mantle plume systems (Kumagai et al., 2008).1 3  THE HAWAIIAN-EMPEROR CHAIN The Hawaiian-Emperor chain is the surface expression of the deep-sourced Hawaiian mantle plume that records activity as far back as >81 Ma (Jackson, 1972; Clague and Dalrymple, 1987; Regelous et al., 2003; Duncan and Keller, 2004). Older volcanic products have either been subducted in the Kamchatka-Aleutian Arc or were accreted to Kamchatka (Avdeiko, 1980; Portnyagin et al., 2008). The Hawaiian plume has been imaged all the way to the core mantle boundary by whole mantle radially anisotropic shear-wave velocity models (Zhao et al., 2004; Montelli et al., 2006; French and Romanowicz, 2015). The dynamic cause of the prominent bend in the Hawaiian-Emperor chain (at ~50-47 Ma; Sharp and Clague, 2006) is currently hotly debated, and proposed mechanisms include absolute plate motion change, southward motion (equivalent to 2-9° of latitude) of the Hawaiian plume between ~80-50 Ma, or a combination of the two (Norton, 1995; Tarduno et al., 2003, 2007, 2009; Steinberger et al., 2004; Sharp and Clague, 2006; Wessel and Kronke, 2008; O’Conner et al., 2013; Wright et al., 2015; Torsvik et al., 2017). 1 3 1 Geology and Previous Work The Hawaiian-Emperor chain is divided into three sections based on the orientation of linear segments of seamounts and proximity to the active mantle plume: the Emperor Seamounts, the Northwest Hawaiian Ridge (NWHR), and the Hawaiian Islands (Figure 90 km 500 km160˚E160˚E165˚E165˚E170˚E170˚E175˚E175˚E180˚180˚175˚W175˚W170˚W170˚W165˚W165˚W160˚W160˚W155˚W155˚W20˚N 20˚N25˚N 25˚N30˚N 30˚N35˚N 35˚N40˚N 40˚N45˚N 45˚N50˚N 50˚N55˚N 55˚N−6000 −4000 −2000 0Bathymetry (m)NEmperorSeamountsPacific OceanNorthwest Hawaiian RidgeHawaiianIslandsFigure 1 1: Bathymetric map of the >100 seamounts and islands of the Hawaiian-Emperor chainThe Hawaiian-Emperor chain is split into three sections: the Emperor Seamounts (>81-48 Ma); the Northwest Hawaiian Ridge (~48-6 Ma); and the Hawaiian Islands (~6 Ma to present). Bathymetry is 2-minute Gridded Global Relief Data ETOPO2v2 satellite altimetry dataset (U.S. Department of Commerce, National Oceanic and Atmospheric Administration, National Geophysical Data Center, 2006, downloaded March 14, 2014) and new multibeam bathymetry (Smith et al., 2014). 101.1). The oldest section (the Emperor Seamounts, >81-50 Ma) is composed of seamounts and guyots that stretch from the northwestern Pacific near the Aleutian Islands to the bend in the Hawaiian-Emperor chain (Jackson et al., 1972; Regelous et al., 2003; Sharp and Clague, 2006). The next oldest section, the NWHR, is the ~2,800 kilometer long section of seamounts, atolls, and small islands from the bend in the Hawaiian-Emperor chain at ~48-50 Ma to the Hawaiian Islands (Sharp and Clague, 2006; Cousens and Clague, 2015; Garcia et al., 2015). The Hawaiian Islands, a chain of subaerial islands beginning ~6 Ma at Ni‘ihau to currently active volcanism at Kīlauea and Lō‘ihi volcanoes constitute the youngest section (Sherrod et al., 2007; Cousens and Clague, 2015). The Hawaiian Islands are characterized by the presence of two geographical and geochemical trends, named the “Loa” and “Kea” trends after the two largest Hawaiian volcanoes, Mauna Loa and Mauna Kea (Figure 1.2) (Jackson et al., 1972; Jackson and Shaw, 1975; Tatsumoto, 1978; Abouchami et al., 2005; Tanaka et al., 2008; Weis et al., 2011). Only Kea-trend basalts have been observed on the Emperor Seamounts (Keller et al., 2000; Regelous et al., 2003; Shafer et al., 2005; Huang et al., 2005a; Frey et al., 2005; Tanaka et al., 2008).  The Emperor Seamounts have been relatively well-studied with four Ocean Drilling Program expeditions (Legs 19, 55, 145 and 197; Jackson et al., 1980; Duncan and Keller, 2004), several dredging excursions (detailed in Clague and Dalrymple, 1987), high-precision modern geochemical and isotopic characterization (Keller et al, 2000; Regelous et al., 2003; Huang et al., 2005a; Frey et al., 2005; Shafer et al., 2005), and modern 40Ar/39Ar geochronology (Duncan and Keller, 2004; Sharp and Clague, 2006). Similarly, the Hawaiian Islands have been well-characterized for geochemistry and isotopic signatures with thousands of analyses of major and trace element compositions and at least 362 high-precision isotopic analyses of Pb-Hf-Nd-Sr isotopes on the same sample powder (e.g., Tanaka et al., 2008; Weis et al., 2011; Jackson et al., 2012). In contrast, high-precision isotopic data coverage of the Northwestern Hawaiian Ridge is surprisingly sparse with only three previous published110 100KilometersKaua‘iKo‘olauWai‘anaeE. Moloka‘iW. Moloka‘iW. MauiHana RidgeKohalaKīlaueaHualālaiLāna‘iKaho‘olawePenguin BankW. Ka‘enaNi‘ihauMauna KeaMauna LoaHawaiianIslandsLō‘ihiLoa Kea 37.637.737.837.938.038.138.238.317.7 17.9 18.1 18.3 18.5 18.7208Pb/204Pb206Pb/204Pb01234567890.7030 0.7035 0.7040 0.7045ɛNd87Sr/86SrNFigure 1 2: The Loa and Kea geochemical and geographic trends on the Hawaiian IslandsMap figure is a simplified topographic map of the Hawaiian Islands modified from the Main Hawaiian Islands Chart 750-001 Version 17 downloaded January 20, 2013 from the Hawai‘i Mapping Research Group at the University of Hawai‘i at Mānoa’s website (Main Hawaiian Islands Multibeam Bathymetry Synthesis: www.soest.hawaii.edu). The Loa and Kea geochemical and geographical trends are shown in blue and red, respectively. Geochemical plots are compiled Hawaiian Island shield-stage data normalized to the same standard values (data references are listed in the caption of Figure 4.10). Loa trend volcanoes are blue circles, and Kea trend volcanoes are red circles. The black line is the Loa-Kea divide line from Abouchami et al. (2005).12studies. Lanphere et al. (1980) examined strontium isotopes on basalts from 8 NWHR volcanoes and Basu and Faggart (1996) followed this pioneering work by analyzing Pb, Nd, and Sr isotopes from five NWHR islands. Compared to modern standards, these studies are limited by the lack of acid leaching and lower precision of the instrumental analysis (TIMS analyses versus MC-ICP-MS; Galer, 1999; Thirlwall, 2000; Woodhead, 2002; Albarède et al., 2004). Modern, high-precision Pb, Hf, and Nd isotopes have been analyzed on only two of the oldest NWHR volcanoes, Yuryaku and Daikakuji (Regelous et al., 2003). There is no Hf isotope data for the entire ridge, despite the importance of this isotopic system in resolving the presence and type of recycled material in the source of Hawaiian basalts (e.g., Blichert-Toft et al., 1999, 2003). This paucity of data is striking considering the 40 million years of time, ~2,800 kilometers, and roughly 52 volcanoes that constitute this section of the Hawaiian-Emperor chain, as well as the importance of the Hawaiian-Emperor chain as a defining case study of mantle plume dynamics (Wilson, 1963; Morgan, 1972; Garcia et al., 2015).1 3 2 Volcanic Stages Hawaiian volcanoes follow predictable stages of volcanic activity as the Pacific plate moves over the Hawaiian mantle plume (Clague and Dalrymple, 1987; Sherrod et al., 2007; Clague and Sherrod, 2014). The first stage of Hawaiian volcanism is the preshield-stage, dominated by highly heterogeneous alkalic magmas that have only been observed on Lō‘ihi Seamount and possibly on the submarine flank of Hualālai, the Hilo Ridge rift zone of Ko-hala, and the submarine south flank of Kīlauea (Moore et al., 1982; Holcomb et al., 2000; Hammer et al., 2006; Kimura et al., 2006). The shield-stage is the main shield-building stage where greater than 95% of the volcano is built of tholeiitic lavas (Clague and Dalrymple, 1987; Garcia et al., 2006; Sherrod et al., 2007; Clague and Sherrod, 2014). The post-shield-stage is characterized by transitional basalts, alkalic basalts, and evolved trachytes, phonolites, mugearites, benmoreites, and hawaiites; this stage represents ~1-4% of the 13volcano. The final stage of a Hawaiian volcano is the highly isotopically depleted rejuven-ated-stage, characterized by very small amounts of silica-undersaturated alkalic basalts that typically erupt after a period of quiescence (up to ~2 million years), but can also directly follow postshield activity without an age gap (Sherrod et al., 2007; Garcia et al., 2010; Cous-ens and Clague, 2015). These stages are controlled by the movement of the volcano over the Hawaiian mantle plume, starting and finishing with small, low-degree melts generated by the lower temperature of the leading and trailing edges of the plume and high-volume production over the center of the hot plume stem (DePaolo and Stolper, 1996; Ribe and Christensen, 1999; DePaolo et al., 2001). The mechanism leading to the formation of rejuvenated volcan-ism is debated between variations in source (e.g., lithosphere or depleted plume components; Chen et al., 1990, 1991; Lassiter et al., 2000; Yang et al., 2003; Frey et al., 2005; Fekiacova et al., 2007; Garcia et al., 2010; Hofmann and Farnetani, 2013) and dynamic origins (up-lift and flexure of the lithosphere away from the active plume; Ribe and Christensen, 1999; Bianco et al., 2005; Paul et al., 2005). 1 3 3 Magmatic Flux One result of the impingement by a hot mantle plume upon a rigid, cold lithospheric plate is geodynamic uplift and basal erosion of the lithosphere (Crough, 1983; Sleep, 1990; Phipps Morgan et al., 1995; Ribe and Christensen, 1999; DePaolo and Weis, 2007; Wessel, 2016). Around the Hawaiian Islands, the “Hawaiian Swell” is ~1,200 km wide, ~3,000 kilometers long, and uplifted ~100 km from surrounding oceanic crustal depth (Crough, 1983; Ribe and Christenson, 1999; Wessel, 2016). Long-period Rayleigh-wave low-velocity anomalies underneath Hawai‘i are consistent with excess temperatures 200-250°C above the mantle geotherm, and petrologic modeling (e.g., olivine thermometry and primary basalt composition) results are consistent with excess temperatures of ~100-300°C in the Hawaiian mantle relative to mid-ocean ridge mantle (Herzberg et al, 2007; Putirka et al, 2007; Laske et al., 2011). As such, there is considerable evidence for exceptionally hot mantle (i.e., a mantle14plume) supplying Hawaiian volcanoes.  Over the >81 million year lifetime of the Hawaiian mantle plume, the magmatic flux of the system has varied strongly, independent of uncertainties in the method of calculation (Figure 1.3) (Sleep, 1990; White, 1993; Van Ark and Lin, 2004; Vidal and Bonneville, 2004; Wessel, 2016). The Emperor Seamounts have uniformly low magmatic flux followed by order of magnitude increases along the NWHR and Hawaiian Islands. The strengthening of the Hawaiian plume with time is an unusual feature for a mantle plume; plumes typically exert the highest magmatic output upon arrival at the base of the lithosphere that decreases with time (i.e., from melting the plume head to the plume tail; White, 1993; Coffin and 40 90420 0681010 20 70605030 80Age (Ma)Estimated Magmatic Flux (m3 /sec)468MeijiDetroitSuikoOjinKokoKimmeiKaua‘iHawai‘iYuryakuDaikakujiMidwayPioneer      1214Gardner!"#$"#$ Magmatic Flux Van Ark & Lin, 2004  Vidal & Bonneville, 2004 Wessel, 2016   Northwest Hawaiian Ridge Emperor SeamountsFigure 1 3: Estimated magmatic flux of the Hawaiian plume for the Hawaiian-Emperor chainFigure modified from Van Ark and Lin, (2004), Vidal and Bonneville (2004), and Wessel (2016) and shows variations in magmatic flux in m3/s assuming flexural compensation. The black curve is the average flux variation with uncertainty shown in red (Wessel, 2016). The flux of the Emperor Seamounts is from Van Ark and Lin (2004; green line) and Vidal and Bonneville (2004; purple line) or not calculated (Wessel, 2016). The magmatic flux of the Hawaiian plume has increased ~650% during the formation of the NWHR.15Eldholm, 1994; Jellinek and Manga, 2004; Kumagai et al., 2008). The causes of variations in magmatic flux are either plume-related, such as changes in pressure, temperature, and lithology of the mantle source, or lithosphere-related, such as fluctuations in the stress field of the Pacific plate (Wessel and Kroenke, 2007; Garcia et al., 2015). Geochemical studies can help assess whether changes in melting conditions or variations in source material contribute to magmatic flux increases as changes in mantle source lithology may be detected from canonical mantle trace element ratios (e.g., Zr/Nb, Th/Ce, La/Nb) and radiogenic isotope compositions (Hofmann, 1988, 2014; Sleep, 1990; Pertermann and Hirschmann, 2003; Sobolev et al., 2007; Gurenko et al., 2013).1 4  THE DEEP MANTLE Study of the “Deep Earth” and the composition of mantle reservoirs has been transformed by advances in diverse scientific fields. In seismology, the resolution of global tomographic models has been greatly improved; in geodynamics, inclusion of the different viscosities, densities, and rheologies of compositional heterogeneities in the mantle has produced more accurate models; in geochemistry, expansion into non-traditional and stable isotopes and improvement in analytical precision have provided greater insight into the composition of the deep mantle. These three fields—geophysics, geodynamics, geochemistry—have revealed the mantle’s structure and dynamics to be much more complex than previously envisioned (Figure 1.4).  In the last 20-30 years, advances in both global seismometer coverage and geophysical technology has resulted in the first detailed geophysical images of the deep mantle all the way to the core mantle boundary (Ishii and Tromp, 1999; Garnero, 2000; Mégnin and Romanowicz, 2000; Ritsema, 2004; Garnero and McNamara, 2008; French and Romanowicz, 2015; Garnero et al., 2016). In addition to pronounced anisotropy and heterogeneity of shear wave velocities at smaller (sub-1,000 km) scales, the most important finding was the discovery of the existence of two large low-shear velocity provinces16Hawaii+2.0–2.0δVs/Vs (%)7Hawai‘i‘Primary’ plumes Clearly Resolved–2.0+2.0δVs /Vs  (%)Min –3.4Max +3.9SEMUCB-WM1 at 2,800-km depthabcTemperatureComposition Subducted oceanic crust More primitive material Background mantle0 10 1Velocity magnitude = 300 (non-dimensional)TemperatureCompositionVelocity magnitude = 300 (non dimensional)kground mantleMor  ri itive materialS cted oceanic crustA. The deep-sourced Hawaiian mantle plumeB. Generalized mantle convection, structure, and compositionFigure 1 4: Seismic and geodynamic models of mantle structure and convectionFigures modified from French and Romanowicz (2015) and Li et al. (2014). (A) Seismic tomography model of the deep mantle (2,800 km depth) and the Hawaiian mantle plume modified from French and Romanowicz (2015). SEMUCB-WM1 is a global seismic tomography model that shows the spatial extent of the LLSVPs underneath the Pacific Ocean and continental Africa and the correlation of strong, deeply sourced plumes with the location of LLSVPs. Also shown is a cross section of the Pacific mantle underneath Hawai‘i that confirms the presence of a deep mantle plume supplying Hawaiian volcanism (French and Romanowicz, 2015). (B) Mantle convection dynamic model results of Li et al. (2014) that shows generalized mantle convection patterns, the locations of subducted material and mantle plumes, and compositional mixing on the scale of the entire mantle and within the LLSVPs. The LLSVPS are constantly mixing primitive material and subducted oceanic crust, and this mélange of heterogeneous material is entrained in mantle plumes in varying amounts throughout time (Li et al., 2014).(LLSVP) below continental Africa and the Pacific Ocean (To et al., 2011 and references therein). These large-scale provinces also contain ultra-low velocity zones (ULVZ) of between 5-40 km thick where S- and P-wave velocity drops by a further 10%; these zones may contain partial melt or leaking core fluids (Ritsema et al., 2004; McNamara et al., 2010; Cottaar and Romanowicz, 2012; Li et al., 2017).  Improved resolution in seismic models has also shown greater complexity of 17subducting slabs and rising mantle plumes. Some slabs penetrate the 660 km discontinuity to the lower mantle, whereas others stagnate at intermediate depth in the mantle (Albarède and van der Hilst, 2002; Adam et al., 2017). Similarly, mantle plumes also show stagnation at ~1,000 and ~600-400 km depths that may be due to buoyantly unfavorable phase changes or perturbations from the interaction of materials with different densities (Ballmer et al., 2015; Adam et al., 2017). Dynamic modeling of the interaction of compositional materials in the mantle has shown that some materials may mix only intermittently and sluggishly, providing a mechanism to preserve both old recycled and primordial material in the deep mantle (Li et al., 2014). Geochemical observations can be used in tandem with seismic and geodynamic models to test geophysical problems such as a thermal versus a compositional cause of the ~5% decrease in seismic wave velocity of the LLSVP. Systems that provide a long-lived record in a single tectonic setting, such as the Hawaiian-Emperor chain, provide an ideal test location to investigate possible compositional variations in deep mantle reservoirs. Many geophysical and geodynamic studies have also observed that hotspots are preferentially located above mantle LLSVP at the time of initial eruption (Burke and Torsvik, 2004). Mantle convection models and second degree seismic observations suggest that the LLSVP can be stable for at least 200 million years and up to several billion years (Dziewonski et al., 2010; Mulyukova et al., 2015). Thus, the LLSVP provide a chemically or thermally distinct layer that can stabilize the roots of mantle upwellings for sustained periods of geologic time, including the entire ~81 Ma lifetime of the Hawaiian mantle plume (Regelous et al., 2003). In addition, deeply sourced mantle plumes located at the edges of LLSVP may help explain the presence of two geochemical trends on some oceanic islands by tracing their origins to the LLSVP for the enriched geochemical trend, and to ambient “normal” lower mantle for the more depleted geochemical trend (Huang et al., 2011a; Weis et al., 2011; Chauvel et al., 2012; Payne et al., 2012; Nobre-Silva et al., 2013a; Harpp et al., 2014a; Hoernle et al., 2015). The study of oceanic island chains provides rich information 18on how the chemistry of these seismologically resolved features in the deep may mantle vary spatially and temporally.1 5  GEOCHEMICAL TOOLS IN THE STUDY OF OCEANIC ISLAND BASALTS The compositional variation apparent in mantle-derived rocks requires that the mantle also be compositionally heterogeneous (Gast et al., 1964; White and Hofmann, 1982; Zindler and Hart, 1986; Hofmann, 1997, 2003, 2014; White, 2015). Early isotopic and geochemical studies in the 1970s and 1980s showed oceanic island basalts varied in composition, much more than could be accounted for by simple one-stage depletion of the mantle through creation of the continental crust at roughly 2.9-2.5 Ga (Figure 1.5) (Allègre et al., 1979; White and Hofmann, 1982; Zindler and Hart, 1986; Shirey et al., 2008). This has led to the current understanding that the composition of the Earth’s mantle evolves through continuous depletion of the upper mantle by partial melting and oceanic crust formation and recycling of enriched crustal material back into the mantle by subduction (Hofmann and White, 1982; Hofmann, 1997, 2014; Albarède and van der Hilst, 2002; Stracke et al., 2003). Fractionation of parent and daughter elements can occur during partial melting, fluid partitioning during subduction, differential erosion, and mineral fractionation. These processes lead to variations in the parent-daughter elemental ratio and daughter isotope ratios between different Earth reservoirs with time. Considerable debate continues over how much and what type of material is recycled into the mantle (e.g., crust, lithosphere, sediments), the efficiency by which this material is mixed into the mantle, and the geochemical “fingerprints” it will carry in subsequent mantle-derived melts. Despite these questions, geochemists have identified distinct mantle geochemical reservoirs. These are enriched mantle type I and II (EM-I and EM-II), high μ=238U/204Pb (HIMU), prevalent mantle (PREMA), and depleted mantle (represented by MORB) (Figure 1.5) (Zindler and Hart, 1986; Hofmann 1997, 2014; Stracke et al., 2005). These reservoirs are well-characterized geochemically, but there is still debate about when, how, and what are the source material(s) of these different mantle “flavors” 19 (Stracke et al., 2005).1 5 1 Radiogenic Isotopes Radiogenic isotopes (Pb, Hf, Nd, Sr) are powerful tools for evaluating the mantle source and the magmatic processes that affect the geochemical signature of volcanic rocks (e.g., Weis et al., 2011; Hofmann, 2014; White, 2015). One of the reasons they are such useful mantle source tracers is because radiogenic isotope ratios such as Pb, Hf, Nd, and Sr do not fractionate during melting or crystallization processes. This allows geochemists to effectively “see through” these processes to assess the nature of the mantle source. There is 206Pb/204Pb206Pb/204Pb206Pb/204Pb208Pb*/206Pb*87Sr/86Sr176 Hf/177Hf143 Nd/144 Nd206 Pb/204 Pb17 18 19 20 21 2217 18 19 20 21 2217 18 19 20 21 220.9 1.0 1.1 1.20.51340.51320.51300.51280.51260.51240.51220.28340.28320.28280.28260.28240.28300.28360.7080.7070.7060.7050.7040.7030.7022221201918HIMUHIMUHIMU17DMHawai‘iEMPREMAEMDMPREMAHIMUPREMAEMDMEMDMPREMAHawai‘iHawai‘iHawai‘i(a) (c)(d)(b)Figure 1 5: Isotope geochemistry reveals substantial heterogeneity in the mantleFigure modified from Stracke et al. (2012) and shows the (a) 87Sr/86Sr, (b) 206Pb/204Pb, (c) 176Hf/177Hf, and (d) 143Nd/144Nd composition of mantle end-members: EM for enriched mantle, DM for depleted mantle, PREMA for prevalent mantle, and HIMU for high μ=238U/204Pb (Zindler and Hart, 1986). Plotted data includes Atlantic, Indian, and Pacific MORB and oceanic island basalts (OIBs): Hawai‘i (grey, labeled), Galápagos, Society Islands, Samoa, Marquesas, St. Helena, Kerguelen Heard, Pitcairn, Walvis Ridge, Tristan de Cunha, Gough, Comores, Azores, Cape Verdes, Fernando de Noronha, Madeira, Canary Islands, Ascension, Iceland, and the Austral Cook Islands (Stracke et al., 2012).20also a time-integrated dimension to using radiogenic isotopes. The measured isotope ratios in OIB are directly controlled by the time-integrated parent to daughter ratio in the OIB mantle source (this relationship likewise holds for other mantle-derived rocks such as MORB). This information provides constraints for the age of mantle heterogeneities in addition to their composition. The use of these four isotopic systems is well established; advances in mass spec-trometry and lab methods have greatly improved precision and quality of isotope data in the last 15 years (Galer, 1999; Thirlwall, 2000; White et al., 2000; Woodhead, 2002; Kamenov et al., 2004; Weis et al., 2006, 2007; Nobre Silva et al., 2009; 2010). Each of these radiogenic isotope systems has its specific application in resolving and characterizing mantle heterogen-eity. Pb isotopes are the most statistically robust indicator of Loa and Kea trend differences, reflecting the time-integrated contrasts in U and Th concentrations of these two geochemical reservoirs (Tatsumoto, 1978; Abouchami et al., 2005). The Hf and Nd isotopic systematics in all major Earth reservoirs (e.g., OIB, sediments, continental basalts, granitoids) are cor-related, indicating efficient fluxes of Sm and Lu between the mantle and the crust with no large-scale fractionation of Sm from Lu (Vervoort et al., 1999). Hf isotopes, because Lu is incorporated in clays and Hf in zircon-rich sands during weathering, provide an excellent tool for tracing the type of recycled sedimentary material that may contribute to OIB mantle sources (Blichert-Toft et al., 1999; Vervoort et al., 1999; Carpentier et al., 2010). The nega-tive correlation between 143Nd/144Nd and 87Sr/86Sr isotopes in Earth reservoirs, termed the “mantle array,” represents large-scale mixing between mantle and crustal reservoirs (De-Paolo and Wasserburg, 1979; Hofmann, 1997, 2014). This relationship is a result of partial melting processes integrated over time. Rb is more incompatible during mantle melting than Sr, and Nd than Sm. Thus partial melting will leave the mantle with low Rb/Sr and high Sm/Nd, which radiogenically decays into low 87Sr/86Sr and high 143Nd/144Nd over time (with the complimentary opposite relationship in the extracted crust) (DePaolo and Wasserburg, 1979). 211 5 2 Age Correction of Altered Basalts It has long been recognized that alteration and contamination of samples can strongly affect the measured isotope ratios of samples (e.g., Abouchami et al., 2000; Nobre-Silva et al., 2009, 2010). To achieve accurate isotopic measurements on altered submarine basalts such as those of the NWHR, removal of the secondary phases by acid leaching is essential. This is especially the case for Pb and Sr, which are susceptible to post-eruption mobility and contamination from ferromanganese crusts for Pb (Abouchami et al., 2000; Nobre Silva et al., 2009) or seawater for Sr (Hart, 1971). Sequential acid leaching removes secondary alteration phases, which are more susceptible to acid dissolution and mobilization, and leaves behind the more resistant primary magmatic phases that record magmatic radiogenic isotope ratios (Hanano et al., 2009). Basalts are the only rock types leached in this study, and it has been documented that acid leaching of basaltic rocks increases the reproducibility of isotope ratio measurements and the accuracy of radiogenic isotope analyses (Dupré and Allègre, 1980; McDonough and Chauvel, 1991; Abouchami et al., 2000; Eisele et al., 2003; Weis et al., 2005, 2006; Nobre Silva et al., 2009, 2010; Fourny et al., 2016).  To correct isotopic measurements to the initial primary magmatic values, both accurate ages and radiogenic isotope parent element concentrations are required. However, obtaining magmatic elemental concentrations of parent-daughter elements, undisturbed by secondary alteration, is a challenge as the measured elemental concentrations in submarine basalts are commonly not representative of the primary erupted concentration. This results in inaccurate corrections of the measured isotope ratios for post-eruption in-situ decay. Careful use of average oceanic island basalt normalization ratios (e.g., Hofmann et al., 1986) can help correct this problem and this method is applied to measured trace element concentrations during this study (Nobre-Silva et al., 2013b). The reliability of existing NWHR isotopic data was strongly compromised by the uncertainties associated with inadequate treatment of these problems (Garcia et al., 2015). Addressing these challenges 22through acid leaching and applying canonical trace element normalization ratios has produced an accurate isotope dataset of the Hawaiian mantle plume along the NWHR.1 5 3 Stable Isotopes Stable isotopes fractionate during equilibrium or kinetic processes, which provides a way to trace geological processes. Lithium is one such stable isotope that undergoes equilibrium fractionation in low-temperature aqueous environments due to the preferential partitioning of 7Li into the fluid phase over 6Li. Lithium primarily fluxes into oceanic crust from seawater as it is hydrothermally altered and this signature is, in turn, transformed by prograde metamorphism and dewatering during subduction (Figure 1.6) (Zack et al., 2003; Elliott et al., 2004; Tomascak et al., 2004; Tang et al., 2010). Through this complex cycling, fresh OIB (e.g. Hawai‘i)+0.75 to +5.7‰seawater +32‰ marine sediments -4 to +24‰island arc lavas-6 to +12‰eclogite-35 to +8‰normal upper mantle+3 to +5‰altered oceanic crust -2 to +21‰dehydratedlow δ7Li slabhigh δ7Li fluidshydrate mantlerecycling oflow δ7Licomponentsinto plumeFresh MORB+1 to +6‰Figure 1 6: Schematic illustration of lithium isotope systematicsFigure is modified from Tang et al., 2010 and numbers in the plot indicate δ7Li values of different Earth reservoirs. Flux of Li is indicated with arrows. Low temperature alteration of oceanic crust by seawater enriched in heavy 7Li from river runoff increases the δ7Li of altered oceanic crust. This heavy signature is driven off during dewatering when the crust is subducted, resulting in heavy lithium isotope ratios in decollement fluids (δ7Li ≈+22‰) and arc lavas. Oceanic crust, now with low δ7Li, is recycled into the mantle, where incorporation of this material in mantle plumes may contribute a light lithium isotopic signature to oceanic island basalts.23Li isotopes have the potential to be used as a tracer for the presence of altered oceanic crust and sediments in the sources of oceanic island basalts such as Hawai‘i (Figure 1.6) (Zack et al., 2003; Elliott et al., 2004, 2006; Kobayashi et al., 2004; Ryan and Kyle, 2004; Tomascak et al., 2004; Nishio et al., 2005; Chan et al., 2009; Vlastélic et al., 2009; Tang et al., 2010; Krienitz et al., 2012).  The application of Li isotopes as tracers of recycling processes is potentially compromised by Li diffusion at small scale (mineral-matrix) during residence time in the mantle, a process implied by Li isotope ratios of OIB that do not vary significantly from those of MORB (Richter et al., 2003; Chan and Frey, 2003; Seitz et al., 2004; Lundstrom et al., 2005; Jeffcoate et al., 2007; Parkinson et al., 2007; Halama et al., 2008; Tomascak et al., 2008). Li isotope ratios heavier than those in MORB were also not observed in arc lavas, which suggested the fractionation of Li in subduction zones may not follow predicted behavior (Chan et al., 1999; Chan et al., 2001; Chan et al., 2002b). Evaluating the practicality of Li isotopes as a mantle tracer in well-characterized Hawaiian basalts with applications to other OIB studies is addressed in Chapter 2 of this dissertation. Compositional differences between Loa and Kea trend basalts on Hawai‘i has been proposed to result from different proportions of recycled material such as altered oceanic crust and sediments in their respective mantle sources (Hauri, 1996; Blichert-Toft et al., 1999; Gaffney et al., 2004, 2005; Huang et al., 2005b, 2009; Sobolev et al., 2005, 2007; Jackson et al., 2012; Pietruszka et al., 2013; Frey et al., 2016). If Li can be applied as a tracer of recycled altered oceanic crust and sediments, the application of this isotope system to Hawai‘i may provide a fresh perspective into the Loa-Kea origin debate.1 6  FERROMANGANESE CRUSTS Hydrogenetic ferromanganese oxides form crusts, nodules and stains on submarine substrates by slow precipitation from seawater (Figure 1.7) (Hein and Koschinsky, 2014). These encrustations are a well-documented feature of samples from the Hawaiian-Emperor24chain, and represent a significant potential contamination source in the analysis of radiogenic isotopes such as Pb and Nd of Hawaiian basalts as they have Pb concentrations 9-14 times the amount found in an average Kīlauea basalt (Figure 1.7) (Glasby and Andrews, 1977; McDonough et al., 1991; Staudigel et al., 1995; Abouchami et al., 2000; Thirlwall, 2000; Eisele et al., 2003; Regelous et al., 2003; Moore and Clague, 2004; Huang et al., 2005a; Nobre Silva et al., 2009). Precise and accurate measurements of radiogenic isotopes such as Pb are critical to the success of the isotopic study of NWHR basalts, so assessing sources  39.0208Pb/204Pb206Pb/204Pb38.738.438.137.818.1 18.3 18.5 18.7 18.91 cm2 cm1 cm(a)(b)(c)(d)Figure 1 7: Ferromanganese crusts on NWHR samples and their potential to contaminate Pb isotope analyses of basalts. (a), (b), and (c) show the variation in ferromanganese crust morphology found on NWHR samples. (a) A thick (~4 cm) ferromanganese crust on samples collected from Pioneer Seamount by submersible. (b) Heavily altered basalt sample from Gardner Seamount with an immersive ferromanganese crust that penetrates within the most altered portions of the basalt. (c) Thin, botryoidal ferromanganese crust deposited on a sample from the submarine flank of Ka‘ula Island. (d) A plot of 208Pb/204Pb versus 206Pb/204Pb modified from Nobre Silva et al. (2009) that shows the effect of ferromanganese crust contamination on measured Pb isotopic analyses (Pacific Fe-Mn nodules field is data from Abouchami and Galer, 1998). If contaminated by ferromanganese crusts, submarine basaltic samples will form a mixing line off of the mass fractionation lines (red lines) towards the field of Pacific Fe-Mn crusts.25of Pb contamination is critical (Hanano et al., 2009; Nobre Silva et al., 2009). Lead isotope ratio analyses by multicollector inductively coupled plasma mass spectrometry (MC-ICP-MS) are 6-10 times better than with older technology that does not allow such a precise correction for mass fractionation (Abouchami et al., 2000; White et al., 2000; Weis et al., 2005). This increase in analytical precision requires cleaner sample preparation and thorough mitigation of contamination to a degree that was unnecessary before. The assessment of this potential contaminant is the principal objective of geochemical investigation of NWHR ferromanganese crusts in this dissertation. However, because ferromanganese crusts are systematic records of ocean chemistry over time and are a reservoir of economically interesting elements, the analysis of NWHR ferromanganese crusts provides the potential for much richer interpretations in addition to the contamination characterization.  Although most global ferromanganese crusts are of hydrogenetic origin, some may have formed hydrothermally or diagenetically, in which case they will exhibit a different mineral composition and trace metal signature (Halbech et al., 1981; Hein et al., 2000; Bau et al., 2014). Hydrogenetic crusts tend to have the highest enrichments of economically interesting elements (e.g., Co, Ni, Cu, Mo, REE, Y; Hein et al., 2000). The trace element and mineralogical analysis of NWHR ferromanganese crusts will determine the mechanism of formation, and therefore provide constraints on the grade of this potential ore deposit (Hein et al., 2000).  The residence time of Pb in deep ocean water is short (80-100 years), so it is possible to have local variations in Pb isotopic signature recorded at different locations in the Pacific Ocean (von Blackenburg and O’Nions, 1996). Because ferromanganese crusts incorporate the local seawater Pb isotope composition as they grow, they can be used to resolve local inputs of Pb and regional ocean circulation patterns (Christenson et al., 1997; Ling et al., 1997; Frank et al., 1999; Frank, 2002; van de Flierdt et al., 2004a; Hein and Koschinsky, 2014; Koschinsky and Hein, 2017). This makes ferromanganese crusts valuable records of 26paleo-ocean-chemistry that can be used as a proxy for climate change, ocean circulation changes, tectonic opening and closing of oceanic passages, and mass balance of Pb fluxes in the Pacific Ocean over time (Frank, 2002). The NWHR ferromanganese crust Pb isotopic analyses will be interpreted within this context of existing paleoceanographic studies.1 7  OVERVIEW OF THE DISSERTATION This dissertation examines the geochemical characteristics (isotopic and elemental) of the Hawaiian mantle plume over a significant period of time (~47 Ma to the present). The main body of this research focuses on the Northwest Hawaiian Ridge, a section of the Hawaiian-Emperor chain that was previously largely unstudied geochemically. The results of this study have identified the spatial location of Loa trend occurrences along the NWHR, formulated a model to explain the geochemical trends observed in radiogenic isotopes, major and trace elements, and provide constraints on the changing mantle sources of Hawaiian volcanism as reflected in the evolution and longevity of different Hawaiian mantle components. To confidently assign the isotopic signatures and trends observed in NWHR samples as representative of erupted signatures, characterization of a major potential source of potential post-magmatic contamination, ferromanganese crusts, was also examined for Pb isotope ratios and trace element concentrations. Finally, the Loa geochemical trend was further probed using a stable isotope system, lithium, to explore possible recycled contributions to Hawaiian Island magmas. Chapters 2, 3, 4, 5, and 6 were written in manuscript form for submission to international journals and therefore contain some overlap in geological setting and analytical methods sections. Below is a brief synopsis of each thesis chapter and appendices presented in this dissertation.  Chapter 2 (“Li Isotopic Signature of Hawaiian Basalts”) consists of the largest Li isotopic study conducted on a single oceanic island to date and presents 89 new Li isotopic measurements on samples from the Hawaiian Islands. The results of this work were presented at the Goldschmidt Conference in Montreal, Canada (2012), the AGU Chapman 27Conference on Hawaiian Volcanoes held in Waikoloa, Hawai‘i (2012), and the AGU Fall Meeting in San Francisco, California (2013). This research was published as a chapter in the AGU Monograph on Hawaiian Volcanoes: From Source to Surface (Harrison et al., 2015). The major results of this study are that Li isotopes are a useful mantle source tracer, behave systematically on a global scale, and provide an additional perspective from other radiogenic isotopes used to trace the mantle source of oceanic island basalts. On Hawai‘i, Li isotopes differentiate between Loa- and Kea-type basalts, and between basalts from different volcanic stages (preshield, shield, postshield, and rejuvenated). Based on geochemical modeling of measured Li isotope ratios, possible input of carbonate into the mantle source of Ko‘olau Makapu‘u basalts and a possible incorporation of subduction eroded lower continental crust into the source of Hualālai basalts is proposed. Because of the sizeable fractionation of Li isotopes in low-temperature, aqueous environments, Li isotopes were not deemed an acceptable tool to study Northwest Hawaiian Ridge basalts as they are likely irredeemably contaminated by a seawater lithium isotopic signature.  Chapter 3 (“The Link between Hawaiian Mantle Plume Composition, Magmatic Flux, and Deep Mantle Geodynamics”) is the synthesizing work of this dissertation where the connection between observed geochemical trends in NWHR shield-stage basalts is interpreted in the context of deep mantle seismic observations and global mantle dynamics. The results of this work were presented at the Goldschmidt Conference in Sacramento, California (2014), the AGU Fall Meeting in San Francisco, California (2015), where it won an outstanding student paper award, the Goldschmidt Conference in Yokohama, Japan (2016), and at the GAC-MAC Conference in Kingston, Ontario (2017). A version of this chapter is published in Earth and Planetary Science Letters (Harrison et al., 2017). This study presents the first high-precision Pb isotope analyses from NWHR shield-stage basalts and shows a correlation of radiogenic Pb with magmatic flux. This correlation suggests a relationship between volcano chemistry and magmatic flux that is interpreted as reflecting 28increasing participation of the Loa source component with time along the NWHR. The absence of the Loa component during formation of the Emperor Seamounts, increasing participation during the NWHR, and volumious presence during the Hawaiian Islands is used to develop a model of Hawaiian plume movement through different lower mantle geochemical domains with time. Initially, only ambient Pacific lower mantle was entrained by the plume (Emperor Seamounts). Movement of the plume base southward up the northern edge of the Pacific LLSVP resulted in gradual entrainment of enriched LLSVP (e.g., Loa-trend) material with time (NWHR).  Chapter 4 (“The Spatial Distribution and Composition of the Loa Mantle Component Along the Northwest Hawaiian Ridge”) reports the Sr-Nd-Hf isotope ratios and trace element concentrations of NWHR shield-stage basalts studied in Chapter 3. This work was presented at the Goldschmidt Conference in Yokohama, Japan (2016) and the GAC-MAC Conference in Kingston, Ontario (2017). A major aim of this study was to constrain the spatial location of any Loa-trend basalts along the NWHR, and explore any potential changes in composition of the Loa component over long periods of time. In order to confirm through statistical analysis any Loa-type affinity of new NWHR samples, a logistic regression model of Loa and Kea trend geochemistry was built based on high-precision Hawaiian Island literature normalized isotope data. The results of the logistic model provide a classification of new NWHR samples as Loa or Kea trend and confirm Daikakuji, Mokumanamana, West Nīhoa, Nīhoa (NIH-D-1-2), and Middle Bank as belonging to the Loa trend. This finding is corroborated by the trace element characteristics of these basalts. Isotope ratios of NWHR shield-stage basalts show that there was no participation of the Lō‘ihi geochemical component during the formation of the NWHR, and thus it is a rarely sampled geochemical component. A regression of NWHR Hf and Nd isotope ratios also defines a different slope from the newly calculated Hawaiian array, also indicating source changes over time. All of these constraints on the Loa geochemical composition suggest that instead of being a homogeneous 29component itself, it is a highly heterogeneous mélange of several different materials of finite size and temporal expression in the Hawaiian plume.  Chapter 5 (“The Hawaiian Mantle Plume Over >81 Million Years: Evidence For Multiple Depleted Sources”) presents Pb-Hf-Nd-Sr isotope analyses and major and trace element compositions of NWHR rejuvenated and postshield samples and compares their compositions to those from the entire Hawaiian-Emperor chain. The samples identified as rejuvenated are the oldest “rejuvenated” basalts observed on the Hawaiian-Emperor chain (i.e., alkalic, silica-undersaturated, depleted isotope signature, enriched in incompatible trace elements) and are homogeneous over ~13 million years. The depleted component expressed in the oldest Emperor Seamount shield-stage basalts are more isotopically depleted in both isotopes and trace elements than all other depleted Hawaiian magmas younger than the bend in the Hawaiian-Emperor chain, including NWHR rejuvenated samples analyzed here and Hawaiian Island rejuvenated basalts. The interpretation is that there are two different sources of depleted compositions in Hawaiian magmas, one resulting from interaction with a mid-ocean ridge (the oldest Emperor Seamounts) and the other from an old, ultra-depleted mantle reservoir unrelated to the production of the current existing oceanic crust (NWHR rejuvenated and Hawaiian Island rejuvenated basalts). The old(er) age of the source of the rejuvenated basalts is inferred from the high Hf isotope ratios, which require an ancient partial melt event that separated Sm from residual Lu during ancient oceanic crust formation.  Chapter 6 (“A ~50 Million Year Record of Pacific Ferromanganese Crust Mineralogical, Trace Element, and Pb Isotope Compositions from the Northwest Hawaiian Ridge”) characterizes NWHR ferromanganese crusts using Pb isotopes, trace elements, and X-ray diffraction (XRD) mineralogical analysis. Pb isotope analyses of NWHR ferromanganese crusts, along with all other published Pb isotope analyses of Pacific ferromanganese crusts, confirm that the crusts are not a source of contamination in the Pb isotope analyses of NWHR igneous samples. Furthermore, the Pb isotopic signature of 30NWHR ferromanganese crusts reflect mixing of water masses imprinted with two sources of Pb influxes into the ocean: Asian dust from the west and volcanogenic Pb from circum-Pacific arc volcanoes and mid-ocean spreading ridges from the east. NWHR ferromanganese crusts are composed primarily of vernadite, an amorphous δ-MnO2 that is the dominant mineral associated with hydrogenetically formed Fe-Mn crusts. NWHR ferromanganese crusts have an average Fe/Mn ratio of 0.72, high REE concentrations, and moderate trace metal compositions suggest a purely hydrogenetic origin for these ferromanganese crusts.  A summary of significant research findings and future avenues of research is pre-sented in the final concluding Chapter 7 of this thesis. The appendix is a repository of sup-plementary datasets and figures to the dissertation including: A) analytical precision and accuracy of radiogenic isotopic analyses; B) Pb isotope and trace element characterization of USGS reference materials NOD-P-1 and NOD-A-1; C) compiled Northwest Hawaiian Ridge basalt sample locations D) list of publications and abstracts for presentations at national and international meetings resulting from dissertation research.genetic origin for these Fe-Mn crusts. 1 8  FIELD CONTRIBUTIONS TO RESEARCH Within the framework of this research I had the opportunity to participate in several remarkable professional and scientific development opportunities. I attended field courses exploring volcanology at Long Valley Caldera and Medicine Lake, California in 2014, as well as Yellowstone, Wyoming in 2016. I assisted in field sampling for Catherine Armstrong on O‘ahu (2014), Nicole Williamson on Kaua‘i (2015), and Matt Manor at the Giant Mascot Ni-Cu-PGE Deposit near Hope, British Columbia (2013). I was also a contributing student research scientist during a research expedition to map the bathymetry of the Papahānaumokuākea Marine National Monument on the research vessel Falkor in 2014. The Multidisciplinary Applied Geochemistry Network (MAGNET), an NSERC-funded Canada-wide research program aimed at training the next generation of geochemists, 31ran the volcanology field courses. The first trip took place June 14-21, 2014 in California and was led by Drs. James Scoates, Dominique Weis, and John Stix. The trip started at the Long Valley Caldera created by the 0.76 Ma supervolcanic eruption of the Bishop Tuff, and also visited columnar-jointed basalt, the Mammoth Mountain lava dome complex, banded obsidian at Panum Crater, and a series of spectacular silicic domes that make up the Mono and Inyo craters. Investigation of the volcanology of evolved arc systems continued at Medicine Lake, California, during the second half of the field trip. Highlights included zoned dacite-to-rhyolite flows, the High Hole cinder cone, and the Upper Sentinel cave lava tube. The second MAGNET field trip took place in August 2016 and covered the geology of the Stillwater Complex and Beartooth Mountains in Montana and the young Yellowstone volcanic system in Wyoming. Highlights include a traverse of Picket Pin Mountain across the Banded Series of the Stillwater Complex, a spectacular drive through the 4 billion years of Earth history in the mostly Precambrian Beartooth Mountains, and investigation of the formation of the travertine deposits at Mammoth Hot Springs, Yellowstone, Wyoming.  There were two Hawai‘i sampling expeditions: one in February 2014 to O‘ahu and Kaua‘i and another more extensive trip to Kaua‘i in June 2015. The goal of the February 2014 trip was to collect samples from the most enriched Hawaiian end-member volcano, Ko‘olau on O‘ahu, for the M.Sc. research of Catherine Armstrong at UBC and a reconnaissance survey of the geology and sampling potential on Kaua‘i for the Ph.D. research of Nicole Williamson at UBC. This trip benefited from the guidance of Drs. Michael Garcia, James Scoates, and Dominique Weis. The O‘ahu trip was led by Dr. Garcia, and started by sampling Makapu‘u point, a stack of the late shield-stage flows emplaced between 2.5-2.2 Ma on the eastern shore of O‘ahu. Sampling of the Ko‘olau caldera dike swarm along Highway H3 continued the next day and showed mostly aphyric narrow basaltic dikes with chilled and in places glassy margins, suggesting deep (~1 km) emplacement. The Kaua‘i portion of the trip included a sampling expedition up the Kalalau trail on the northern Nā 32Pali coastline to Hanakāpi‘ai beach through the shield-stage Nā Pali member. This section has numerous and mostly aphyric dikes. Waimea Canyon was surveyed next and represents a superb example of a highly dissected Hawaiian volcano. All three of the Kaua‘i shield-stage members (Nā Pali, Makaweli, Olekele) are exposed, along with crosscutting dikes and a major fault that separates the younger Olekele flows from the older Nā Pali flows. The primary goal of the 2015 Kaua‘i sampling expedition was to collect fresh shield-stage basalts to test Loa or Kea trend affinity of the Kaua‘i shield. Targets included the exposed mountains on the largely unsampled east side of the island (NouNou Mountains, Kālepa Ridge, Makaleha Mountains) where we were challenged to find fresh rock in the tropical environment (success was achieved by seeking out cliff walls created by waterfalls or landslide scarps). The experience these field excursions lent me in volcanological fieldwork was invaluable.  The most extraordinary opportunity I had during my PhD was to participate as a student research scientist on Leg 2 of Schmidt Ocean Institute 2014 Papahānaumokuākea Marine National Monument (PMNM) bathymetric mapping expedition. This 36-day cruise was the second of two related expeditions that mapped the bathymetry of the PMNM, the majority of which had not been mapped in detail (only satellite altimetry at a resolution of ~1 km) (Figure 1.8). In addition to detailed seafloor maps at a resolution of 5-50 meters, these expeditions also collected gravity and magnetic data (Figure 1.8). The expedition was led by Drs. Christopher Kelley and John Smith (both of the Hawai‘i Undersea Research Laboratory). As a student researcher, I contributed to bathymetric data collection and quality control, prepared composite maps of detailed features along the PMNM, participated in marine mammal observation, prepared scientific results communication via blog posts, and monitored shipboard scientific equipment, including evaluating gravity and magnetometer equipment, and conducted XBT and CTD (salinity, temperature, density) casts necessary for correction of bathymetric data. Preliminary results from this expedition were presented33in four presentations at the AGU Fall Meeting in San Francisco, California (2014), in two presentations at the Ocean Sciences Meeting in New Orleans, Louisiana (2016), and published in EOS (Kelley et al., 2015).   Fig. 20: Map showing the multibeam coverage in the monument (purple lines) after the cruises.  Four thousand meter contours (black) and UTM zones (grey) are also shown. ABC DEFGN100kmFigure 1 8: Photographs showing the highlights of the Schmidt Ocean Institute 2014 Leg 2 PMNM bathymetric mapping expedition(A) Map of newly collected multibeam bathymetric data from Legs 1 and 2 of the expedition modified from Kelley et al., (2014). The boundary of the Papahānaumokuākea Marine National Monument is shown in purple and the 4,000 meter depth contour is shown in black. (B) Photograph of the R/V Falkor taken from the Schmidt Ocean Institute website: https://schmidtocean.org/rv-falkor/ship-specifications/ The Falkor is a 82.90 meter ship that was originally built in 1981 as a Fishery Protection vessel and was refit from 2009-2012 as an oceanographic research vessel. (C) Photograph of Ka‘ula Island, a tuffaceous cone remnant located ~33 kilometers southwest of Ni‘ihau. We also sailed near and viewed Gardner Rock, Midway Island, and Laysan Island. (D) Student Researcher Brian Shiro (U. of Hawai‘i) preparing the magnetometer for launch and data collection. (E) Bathymetric and backscatter data could be made into three-dimensional maps using Fledermaus software. (F) Previous coverage of NWHR seamounts could be uploaded in systems software so ship tracks could easily be planned each day. (G) Launch of a CTD cast to collect salinity, temperature, and density measurements for correction of sidescan and backscatter bathymetric data.34Chapter 2 The Lithium Isotopic Signature of Hawaiian BasaltsChapter 2 The Lithium Isotopic Signature of Hawaiian Basalts2.1. INTRODUCTION Hawaiian volcanism delineates into two distinctive geographical volcanic trends, the Kea and the Loa trends, which are also different in their geochemical and isotopic characteristics (Jackson et al., 1972; Tatsumoto et al., 1978; Abouchami et al., 2005; Huang et al., 2011a; Weis et al., 2011). Traditionally, the geochemical differences between Loa and Kea trend shield basalts have been attributed to mixing of different components in the source of the Hawaiian mantle plume, i.e., an enriched end-member (Ko‘olau) characterized by the presence of recycled oceanic crust ± sediment, a more depleted end-member (Kea) with high εNd, εHf, and 206Pb/204Pb, and a more undegassed component defined by high 3He/4He (Lō‘ihi; Kurz et al., 1983; Stille et al., 1986; Frey et al., 1994; Hauri, 1996; Tanaka et al., 2008; Jackson et al., 2012). Furthermore, some postshield and rejuvenated samples require a fourth and more isotopically depleted source to account for observed isotopic variation (Huang et al., 2005b; Xu et al., 2005, 2007; Fekiacova et al., 2007; Hanano et al., 2010). Although many studies have focused on resolving, characterizing, and determining the proportions these different sources contribute to Hawaiian volcanism during the stages of a volcano’s life, many questions still remain that up to now radiogenic isotopic systems such as Pb, Hf, Nd, and Sr have not fully answered. In this study, we use the lithium stable isotopic system to obtain additional insights on the source characteristics of Hawaiian basalts. Geochemically, lithium is a moderately incompatible element and is thus concentrated in the continental crust. The two isotopes of lithium, 6Li and 7Li, are fractionated in low-35temperature aqueous environments as 7Li preferentially partitions into the fluid phase. Thus, different Earth reservoirs have unique signatures in lithium isotopes (e.g., Tang et al., 2010 and references therein). The flux of lithium between these reservoirs is largely controlled by interactions of solid geological materials with hydrous fluids and the temperatures of those fluids, i.e., at metamorphic temperatures, lithium is preferentially extracted from mineral phases, whereas at hydrothermal temperatures, lithium is added through incorporation in secondary minerals (Chan et al., 1992; Huh et al., 2004; Gao et al., 2012). Lithium primarily fluxes into oceanic crust from seawater as it is hydrothermally altered and this signature is in turn altered by prograde metamorphism and dewatering during subduction (Zack et al., 2003; Elliott et al., 2004; Tomascak et al., 2004; Tang et al., 2010). Through this complex cycling, lithium isotopes have the potential to be a unique tracer for the presence of altered oceanic crust and sediments in the sources of oceanic island basalts such as Hawai‘i (Zack et al., 2003; Elliott et al., 2004, 2006; Kobayashi et al., 2004; Ryan and Kyle, 2004; Tomascak et al., 2004; Nishio et al., 2005; Chan et al., 2009; Vlastélic et al., 2009; Tang et al., 2010; Krienitz et al., 2012).  Recently, however, the application of lithium isotopes as tracers of recycling processes has been questioned for several reasons. First, the discovery of lithium diffusion at small scale (mineral-matrix) that results in lithium isotopic heterogeneities has led to concerns that the original source lithium isotopic signature may be obliterated during transport and cooling of melts (Richter et al., 2003; Seitz et al., 2004; Lundstrom et al., 2005; Jeffcoate et al., 2007; Parkinson et al., 2007). Diffusion also presents the possibility of homogenization of lithium isotopes during residence time in the mantle, further complicating the use of lithium as a source tracer (Halama et al., 2008). Second, a lack of predictable behavior of lithium isotopes, including the absence of a significantly heavy lithium isotopic signature in arc basalts (Chan et al., 1999, 2001, 2002b), a lack of correlation of lithium isotopes with other isotopic systems (Chan and Frey, 2003), or the lack of significant 36variation in lithium isotopic signature from MORB mantle in other mantle-derived melts (Chan and Frey, 2003; Tomascak et al., 2008), has called into question the robustness and systematics of the processes that control the lithium isotopic system. Evaluating the practicality of lithium isotopes as a mantle tracer in well-characterized Hawaiian basalts is a major aim of this study. This study uses lithium isotopes to 1) corroborate the usefulness of this isotopic system in oceanic island basalt studies (Chan et al., 2009; Vlastélic et al., 2009; Krienitz et al., 2012); 2) further characterize Hawaiian plume source characteristics for preshield, shield, postshield and rejuvenated basalts; and 3) assess the prevalence and impact of recycled materials in creating observed mantle heterogeneities in Hawai‘i and especially in the difference between shield-stage Loa and Kea trend basalts. After controlling for different degrees of source melting, fractionation of lithium isotopes during crystal fractionation, and perturbation of lithium in samples by alteration, we have found relative differences in lithium isotopes between Kea and Loa trend volcanoes and between shield and postshield volcanics. We also show that lithium isotopes in Hawaiian basalts correlate with Pb, Hf, Nd, and Sr radiogenic isotopic systems, indicating that the lithium isotopic signature of erupted basalts can be used as a tracer of the mantle source in Hawaiian basalts. An analysis of lithium isotopes and various potential geochemical indices of sediment recycling (La/Nb, Sr/Nb, 87Sr/86Sr, 143Nd/144Nd) suggest the presence of a heterogeneous sedimentary contribution in the composition of Loa trend volcanoes. Trace element and isotopic data indicates that this sedimentary source may be ancient carbonate for the Makapu‘u section of Ko‘olau Volcano and lower continental crust for Hualālai (Willbold and Stracke, 2006; Huang et al., 2009, 2011b). Finally, a comparison of this Hawaiian data with published lithium isotopic data for other oceanic island basalts (OIB) shows systematic behavior of lithium in the framework of traditionally defined mantle end-members (HIMU, EM-I, EM-II, PREMA; White, 1985, 2010; Zindler and Hart, 1986; Hofmann, 2003; Krienitz et al., 2012) where Hawai‘i has  37traditionally been classified as an EM-I-type oceanic island.2.2. SAMPLES AND METHODS The aim of this study is a wide-scale investigation of the lithium isotopic signatures present on Hawai‘i to both characterize Hawai‘i in the context of other oceanic islands and to test for the presence of recycled material in the source of Loa trend volcanoes. Samples 0 100KilometersODP Site 843Ka‘ula W. Ka‘enaLoa TrendKaua‘iKea TrendKo‘olauKohalaHualālaiMaunaKeaMaunaLoaKīlaueaLō‘ihiMauna Loa HualālaiMauna Loa* Kaua‘iKo‘olau* Ka‘ulaHualālai W. Ka‘enaW. Ka‘ena Lō‘ihiKaua‘iMauna Kea*KohalaKīlaueaMauna KeaKohalaLoa Trend Kea Trend19˚N20˚N21˚N22˚N-161˚W -160˚W -159˚W -158˚W -157˚W -156˚W -155˚W-161˚W -160˚W -159˚W -158˚W -157˚W -156˚W -155˚W19˚N20˚N21˚N22˚NFigure 2.1: Map of Hawai‘i and sample locationsThe basemap for this figure is the Main Hawaiian Islands Chart 750-001 Version 17 (released January 2011) downloaded January 20, 2013 from the Hawai‘i Mapping Research Group of the School of Ocean and Earth Science and Technology at the University of Hawai‘i at Mānoa’s website (Main Hawaiian Islands Multibeam Bathymetry Synthesis: www.soest.hawaii.edu). Approximate locations of samples are indicated by symbols that are also used in the following figures (Garcia et al., 1993, 1995, 1998, 2010, 2016; Chan and Frey, 2003; Greene et al., 2010; Hanano et al., 2010; Weis et al., 2011; Jicha et al., 2012; Feigenson, unpublished [Hualālai and Kohala shield samples]; Garcia, unpublished [Hualālai and Kohala shield samples]). Approximate locations of the Loa and Kea geochemical trends also shown as a blue and red line, respectively. Symbols for this and all following figures are divided into circles for shield basalts, squares for preshield basalts, and diamonds for postshield and rejuvenated basalts. Data noted in legend with an asterisk is from Chan and Frey (2003).38spanning the greatest compositional, geographical, and temporal range as well as different volcanic stages were analyzed to constrain the range of lithium isotopic signatures in Hawaiian basalts (Figure 2.1). Lithium isotopic ratios were measured on 89 samples of Hawaiian basalts, including preshield, shield, postshield, and rejuvenated volcanic stages and a temporal range from less than a century old to ~4.3 Ma (Kaua‘i; Garcia et al., 2010). Published descriptions and localities of many of these samples are provided in the references listed in Table 2.2. In addition, ten samples of ~110 Ma Pacific upper altered oceanic crust from ODP Site 843 (~320 km west of the island of Hawai‘i; see Figure 2.1; King et al., 1993; Fekiacova et al., 2007) were analyzed for lithium isotopes. Also included in the figures and discussion is the Mauna Loa, Mauna Kea, and Ko‘olau lithium isotopic data of Chan and Frey (2003). For this chapter, the term ‘Ko‘olau’ refers to the Makapu‘u section of Ko‘olau volcano. This dataset compliments existing Pb-Hf-Nd-Sr isotopic, major and trace element data from several sources (see Table 2.2 for a full compilation). In the cases where high precision trace element data did not exist (Hualālai, Kohala, and Mauna Loa shield samples, Lō‘ihi preshield samples, and Ka‘ula and Kaua‘i shield, postshield and rejuvenated samples), a full characterization was carried out at the Pacific Centre for Isotopic and Geochemical Research (PCIGR) at the University of British Columbia. Whole rock powders were analyzed for lithium isotopic compositions and concentrations at PCIGR. For lithium isotope analysis, whole rock powders were leached in 0.05 M HCl for one hour at room temperature, a procedure that has been shown to have no or limited effect on primary magmatic lithium isotopic ratios, but that is necessary to remove possible contamination from a heavy, post-eruption lithium isotopic signature contained in secondary clays (Huh et al., 2004; Vlastélic et al., 2009). Our results show that this heavy alteration signature may be effectively removed by leaching (Figure 2.2). However, for nine out of 10 altered oceanic crust samples from ODP Site 843 leaching did not change 39             the lithium isotopic signature beyond analytical error of unleached samples. Leaching also did not change the lithium isotopic signature of samples outside of the 2 standard deviation (2SD) of average East Pacific Rise MORB (Figure 2.2).  After leaching, an aggressive two-step combination HF and HClO4 digestion was carried out following Method C of Yokoyama et al. (1999) to minimize the presence of insoluble fluorides. The two ion-exchange column procedure of Jeffcoate et al. (2004) was used to separate and purify lithium. Total procedural blanks contained 0-0.2 nanograms lithium, negligible in comparison to the amount present in samples (~45 ng). Lithium isotopes were analyzed on a Nu Plasma MC-ICP-MS using the conventional blank-standard-sample-standard-blank bracketing technique and employing the NIST lithium carbonate standard L-SVEC. Each sample was analyzed a minimum of three times to statistically assess precision. During a typical three-day analytical session, the average 7Li/6Li value for the L-SVEC standard was 13.95 ± 0.06 (2 standard error, n=242). Lithium isotopic ratios are reported here as per mil variation relative to L-SVEC based on equation 2.1 below (Coplen et al., 1995; Carignan et al., 2004):                   δ7Li  7Li 6Li 6Li 7Li 1  1000sampleL-SVEC= _ X (2.1) 0.01.53.04.56.0   δ7Li!"#$%&'(((()(*++,+-Unleached LeachedAverage EPR MORBn=25Figure 2.2: ODP Site 843 leaching resultsAverage East Pacific Rise MORB is calculated from Elliott et al., 2006 and Tomascak et al., 2008. The grey field bounded by dashed lines indicates 2SD of average EPR MORB. Leached and unleached sample pairs are linked. For eight out of ten samples, lithium isotope ratios decrease with gentle leaching.40 Repeated analyses of USGS reference material BHVO-2, a Hawaiian basalt, resulted in an average δ7Li=3.2 ± 2.5‰ (2SD; n=55), slightly lower than the published range (4.0-5.5‰; see Table 2.1 and Brant et al., 2012). This may be a result of removal through leaching of the heavy lithium isotopic signature imparted to BHVO-2 by deposition of marine aerosols, a procedure only one other study carried out (Vlastélic et al., 2009). Analysis of unleached BHVO-2 resulted in δ7Li=4.3 ± 0.4‰ (2SD, n=6), confirming this hypothesis. In-house reference materials Kil-93 (Hawaiian basalt from the 1993 Kīlauea eruption, chilled) and the synthetic Li carbonate Puratronic® (Alfa Aesar 99.999% Li2CO3 LOT# 23267) were reproducible over the three year course of this study, with values of δ7Li=3.3 ± 1.7‰ (2SD, n=77) and 80.9 ± 1.5‰ (2SD, n=232), respectively. Reproducibility on individual days of analysis was much better, i.e., δ7Li=3.5 ± 0.6‰ (2SD, n=4) for Kil-93 and 80.8 ± 0.5‰ Table 2.1: Literature compilation of USGS reference material BHVO-2 lithium isotope analysesVlastélic et al., 2009 (n=4)Reference 7Li 2 SD Reference 7Li 2 SDnot leached 4.7 0.2 4.5 0.34.8  - 4.3 0.4 4.4 0.14.1 0.2 4.7 0.44.8 0.4 4.5 0.24.0 0.1 4.5 0.04.6 0.24.2 0.1 4.6 0.14.1 0.2not leached 4.8 0.24.8  - not leached 4.6 0.64.7 0.14.7 0.3not leached 4.6 0.2not leached 4.4 0.4not leached 4.5  - not leached 4.7 2.05.5 0.44.6 0.64.4  - 4.6 1.24.1  - 4.2  - 4.9  - 0.5 N HCl 3.2 2.54.7  - not leached 4.3 0.4Jeffcoate et al., 2004 (n=1)Krienitz et al., 2012 (n=2)leached in 0.5 N HClElliott et al., 2006 (n=3)not leachedZach et al., 2003 (n=1)Brant et al., 2012 (n=3)This Study (n=55)not leachednot leachedMagna et al., 2004 (n=9)not leachedMarschall et al., 2007 (n=1)Ko ler et al., 2009 (n=1)Huang et al., 2010 (n=1)Magna et al., 2006 (n=1)Kasemann et al., 2005 (n=1)Jochum et al., 2006 (n=5)not leached41(2SD, n=10) for Puratronic®. All samples analyzed have two standard deviations equal to or less than one per mil.   Trace element concentrations of samples without existing data were measured at the PCIGR using a Thermo Finnigan Element2 high-resolution ICP-MS and following the methods reported in Pretorius et al. (2006) and Carpentier et al. (2013). Powdered samples were digested in a mixture of 10:1 HNO3:HF with no prior leaching. After digestion, samples were fluxed in HCl to eliminate insoluble fluorides, diluted 5000 times, and analyzed on the Element2 using a 1 ppb In spike as an internal standard and the USGS basaltic standard BCR-2 for external calibration. BHVO-2 was analyzed with all samples, and resulting values were within two standard deviations of published values for all elements (Weis et al., 2006; Chauvel et al., 2011; Carpentier et al., 2013 and references therein). Full procedural duplicates and replicate measurements showed excellent agreement, and most elements were analyzed with one relative standard deviation (RSD) of less than 5%.  Finally, to place these results in the context of the worldwide geochemical signature of the oceanic mantle, we have compiled existing lithium isotopic data for other oceanic island basalts along with supporting major, trace element, and Pb-Nd-Sr data for these samples (references listed in caption of Figure 2.11). All literature radiogenic isotope data are normalized to the same standard values to ensure comparability (Weis et al., 2011). Nd isotopes are normalized to a 143Nd/144Nd La Jolla value of 0.511858; Sr isotopes to a 87Sr/86Sr SRM 987 value of 0.710248; Pb isotopes to a SRM 981 values of 16.9405 for 206Pb/204Pb, 15.4963 for 207Pb/204Pb, and 36.7219 for 208Pb/204Pb; and Hf to a 176Hf/177Hf JMC 475 value of 0.282160 (Weis et al., 2006, 2007).2.3. RESULTS Lithium isotopic compositions and concentrations are reported in Table 2.2. References to published major and trace element data and radiogenic isotopic data for these samples are also listed in Table 2.2. Shield samples for all volcanoes in this study are42 tholeiitic basalts. The postshield and rejuvenated samples are mostly alkalic basalts that range in composition from foidite to trachyte. All of the Lō‘ihi preshield samples are tholeiitic basalts. Lithium concentrations of all samples range from 3-14 ppm and increase with decreasing weight percent MgO, reflecting the moderately incompatible behavior of lithium. The ratio of lithium to elements of similar compatibilities (Y, Dy, Yb) does not show any systematic variation versus an incompatible element (Nb). This suggests that the Table 2.2: Lithium isotopic compositions of Hawaiian basaltsSampleVolcanic  StageVolcanic SeriesLoa or Kea TrendLi (ppm) 7Li 2 SampleVolcanic StageVolcanicSeriesLoa or Kea TrendLi (ppm) 7Li 2L 'ihi Mauna KeaATV-1-4b PreS Loa 4.4 5.5 0.4 02AMK-1a PS 5 Kea 10.0 5.2 0.2186-3b PreS Loa 3.0 4.2 0.0 02AMK-2a PS 6 Kea 3.2 0.31801-9b PreS Loa 4.3 1.7 0.3 02AMK-3a PS 6 Kea 4.0 3.6 0.31804-17b PreS Loa 5.8 0.9 0.4 02AMK-4a PS 5 Kea 10.0 5.0 0.3P287-1b PreS Loa 6.1 2.6 1.1 02AMK-5a PS 6 Kea 4.0 4.1 0.6158-10b PreS Loa 5.3 3.5 0.2 02AMK-6a PS 5 Kea 10.9 4.4 0.2158-1db PreS Loa 5.8 2.6 0.7 02AMK-7a PS 6 Kea 7.4 3.5 1.0P286-14b PreS Loa 4.1 2.8 0.5 02AMK-8a PS 5 Kea 8.1 4.7 0.81804-21b PreS Loa 4.4 2.0 0.8 02AMK-10a PS 5 Kea 8.3 2.7 0.5K lauea 02AMK-11a PS 5 Kea 10.5 3.7 0.6KIL-93 S Kea 4.4 3.7 1.3 02AMK-12a PS 6 Kea 10.5 3.4 0.9Mauna Loa 02AMK-13a PS 6 Kea 3.6 3.0 0.4ML-63 dd S 1 Loa 4.8 2.7 0.3 Hual laiML-190d S 1 Loa 5.3 2.6 0.3 02AHU-1a PS 7 Loa 6.3 2.0 0.4ML-95d S 1 Loa 4.7 3.0 0.3 02AHU-2a PS 7 Loa 4.8 2.5 0.3ML-41d S 1 Loa 4.5 2.9 0.3 02AHU-3a PS 7 Loa 9.0 2.9 0.9ML-91d S 1 Loa 4.2 4.0 0.4 02AHU-4a PS 7 Loa 7.4 2.6 0.1J2-019-08d S 2 Loa 3.8 3.1 0.1 02AHU-4da PS 7 Loa 3.0 0.4J2-019-08dd S 2 Loa 3.4 0.5 02AHU-5a PS 7 Loa 7.3 2.2 0.4J2-019-19d S 2 Loa 4.7 3.6 0.2 02AHU-6a PS 7 Loa 8.7 1.3 0.2J2-019-01d S 2 Loa 5.4 4.5 0.7 02AHU-7a PS 7 Loa 10.6 3.1 0.1J2-020-14d S 2 Loa 5.1 2.6 0.7 02AHU-8a PS 7 Loa 32.9 1.9 0.8184-7d S 3 Loa 4.4 1.8 0.3 02AHU-9a PS 7 Loa 5.6 0.8 0.3L91-55d S 4 Loa 4.7 2.8 0.7 02AHU-10a PS 7 Loa 5.6 1.9 1.0L00-503d S 4 Loa 4.0 3.7 0.5 02AHU-11a PS 7 Loa 6.0 3.5 0.5L00-003d S 4 Loa 4.6 2.8 0.6 02AHU-13a PS 7 Loa 6.5 1.5 0.3L00-506d S 4 Loa 4.2 3.5 0.9 KK-9-2b S Loa 4.7 3.4 0.8L87-235d S 4 Loa 4.5 3.0 0.3 KK-9-2 db S Loa 3.9 0.4L00-014d S 4 Loa 4.4 3.3 0.4 KK-11-2b S Loa 5.9 3.1 0.5PreS=Preshield; S=Shield; LS=Late Shield; PS=Postshield; Rej=Rejuvenated; AOC=Altered Oceanic Crust; d=duplicateVolcanic series are as follows:1Historic, 2Mile High Section,  3Pisces Dive, 4Prehistoric, 5Laup hoehoe, 6H m kua, 7Hual lai, 8H w , 9Polul , 10Jason2 Dive 296, 11Jason2 Dive 305, 12Jason2 Dive 306, 13Waimea Canyon, 14Kōloa Volcanics, 15Jason2 Dive 300, 16Leg 136 0843B-, 17Leg 136 0843 A-Supporting Data for the above samples (Major and trace elements; radiogenic isotopes) comes from: aHanano et al., 2010; bD. Weis unpub;  M.O. Garcia unpub.; cD. Weis unpub; Harrison et al., in prep, M. Feigenson unpublished; dWeis et al., 2011; eGreene et al., 2010; fGarcia et al., 2010; gGarcia et al., 2016; hKing et al., 1993; Fekiacova et al., 200743partition coefficient of lithium does not change significantly between the different samples of Hawaiian basalts analyzed in this study (Chan and Frey, 2003).2.3.1 Alteration Control It is important to have stringent controls on the degree of alteration of analyzed samples due to of the propensity of lithium to fractionate in low-temperature aqueous environments (Chan et al., 1992; Huh et al., 2004; Marschall et al., 2007). In addition, interaction of basaltic rocks with seawater (δ7Li≈31‰; Jeffcoate et al., 2004 and references  Table 2.2 (continued): Lithium isotopic compositions of Hawaiian basaltsSampleVolcanic  StageVolcanic SeriesLoa or Kea TrendLi (ppm) 7Li 2 SampleVolcanic StageVolcanic   SeriesLoa or Kea TrendLi (ppm) 7Li 2KK-11-5b S Loa 3.2 3.3 0.2 Kaua'iKK-13-1b S Loa 3.5 3.5 1.0 KV03-5f PS 13 Loa 5.1 4.2 0.3KK-14-5b S Loa 3.9 2.4 0.6 KV03-5dupf PS 13 Loa 3.6 0.2KK-78-12-11-1b S Loa 3.9 3.7 0.3 KV05-10f PS 13 Loa 4.4 3.3 0.1HU-01b PS 7 Loa 9.0 2.7 0.9 KV03-15f Rej 14 Loa 6.8 4.1 0.1HU-02b PS 7 Loa 7.7 3.0 0.1 KV03-17f Rej 14 Loa 6.6 3.3 0.3Kohala GA-650f PS 13 Loa 5.1 2.5 0.002AKA-2a PS 8 Kea 8.9 2.6 0.6 GA-565f PS 13 Loa 4.1 2.2 0.502AKA-1a PS 8 Kea 9.1 3.0 0.2 KV04-19f S-LS 13 Loa 5.3 3.1 0.302AKA-3a PS 8 Kea 9.1 3.5 0.9 KV04-21f S-LS 13 Loa 6.1 2.5 1.202AKA-4a PS 8 Kea 14.0 4.0 0.4 Ka'ulaKH-14c PS 8 Kea 5.3 2.1 0.5 300-10g Rej 15  - 7.2 4.2 0.4KH-11c PS 8 Kea 6.5 3.0 0.3 300-20g Rej 15  - 4.1 3.1 0.4KH-11dc PS 8 Kea 3.2 0.7 300-30g Rej 15  - 4.4 4.1 0.702AKA-5a LS 9 Kea 8.8 1.8 0.2 300-41g Rej 15  - 3.8 3.3 0.502AKA-6a LS 9 Kea 6.1 1.9 0.1 300-41dg Rej 15  - 3.7 1.002AKA-7a LS 9 Kea 7.2 2.0 0.3 ODP Site 843 - Altered Oceanic CrustMG-2Bc LS 9 Kea 7.2 3.1 0.3 001R 01W 28-30h AOC 16  - 2.1 0.6KO3-6c LS-PS Kea 5.5 3.3 0.4 004R 02W 29-31h AOC 16  - 3.5 0.4MG-5Ac LS 9 Kea 5.1 4.4 0.8 001R 01W 37-39h AOC 16  - 3.3 1.3MG-4Ac LS 9 Kea 4.1 3.8 1.0 001R 01W 56-58h AOC 16  - 1.7 0.9KO1-27b LS Kea 4.9 3.1 0.6 004R 01W 83-85h AOC 16  - 4.0 1.0Ko'olau 001R 06W 84-86h AOC 16  - 3.3 0.2USGS Ref. Mat. S Loa 3.8 4.3 0.9 001R 03W 94-96h AOC 16  - 1.6 0.6West Ka'ena 002R 01W 105-106h AOC 16  - 3.1 0.7296-01e PS-Rej 10 Loa 6.1 4.3 0.3 001R 02W 122-124h AOC 16  - 3.8 0.5J2-305-14e LS 11 Loa 8.2 4.4 0.6 001R 02W 122-124dh AOC 16  - 4.3 0.8J2-305-14de LS 11 Loa 5.0 0.7 003R 02W 8-12h AOC 17  - 3.5 1.3306-15e LS 12 Loa 4.6 5.2 1.1306-27e LS 12 Loa 6.3 3.2 0.3PreS=Preshield; S=Shield; LS=Late Shield; PS=Postshield; Rej=Rejuvenated; AOC=Altered Oceanic Crust; d=duplicateVolcanic series are as follows:1Historic, 2Mile High Section,  3Pisces Dive, 4Prehistoric, 5Laup hoehoe, 6H m kua, 7Hual lai, 8H w , 9Polul , 10Jason2 Dive 296, 11Jason2 Dive 305, 12Jason2 Dive 306, 13Waimea Canyon, 14Kōloa Volcanics, 15Jason2 Dive 300, 16Leg 136 0843B-, 17Leg 136 0843 A-Supporting Data for the above samples (Major and trace elements; radiogenic isotopes) comes from: aHanano et al., 2010; bD. Weis unpub;  M.O. Garcia unpub.; cD. Weis unpub; Harrison et al., in prep, M. Feigenson unpublished; dWeis et al., 2011; eGreene et al., 2010; fGarcia et al., 2010; gGarcia et al., 2016; hKing et al., 1993; Fekiacova et al., 200744therein) will enrich the rock in lithium and increase the lithium isotopic signature to heavier values (Chan et al., 1992, 2002a). The alteration index of Wright (1971), shown in Figure 2.3a, was used to eliminate samples that are significantly altered (Frey et al., 1990). Most samples plot within the range of magmatic K/Rb (380-630; Chan and Frey, 2003) and have 0.000.501.001.500.0 10.0 20.0 30.0K2O (Wt%)Rb (ppm) Loa Postshield & Rej Kaua‘i PSK/Rb=114302004006008000 1 2 3K/Rb K2O/P2O50.00.30.60.90.0 10.0 20.0 30.0 40.0K2O (wt%)Rb (ppm)Mauna Loa HualālaiMauna Loa* Kaua‘iKo‘olau* Ka‘ulaHualālai West Ka‘enaWest Ka‘ena Lō‘ihiKaua‘iMauna Kea*KohalaKīlaueaMauna KeaKohalaLoa Trend Kea Trenda.b.c. Kea Postshield PreshieldFigure 2.3: Alteration indices of Hawaiian basalts (a) K/Rb versus K2O/P2O5. Samples with anomalous δ7Li signatures and K/Rb>750 and K2O/P2O5<0.9 (not shown) were removed from figures and discussion. (b) Weight percent K2O versus ppm Rb for shield and preshield basalts. Coherent trends in these diagrams indicate that variation is due to melting or crystallization processes rather than alteration, which would not have a linear signature in this plot. Most shield samples fall along one trend (red line), with only Hualālai shield samples and Lō‘ihi preshield samples defining another trend (blue line). (c) Weight percent K2O versus ppm Rb for preshield, postshield, and rejuvenated basalts. Here the data defines three separate trends: red line for Kea trend postshield basalts; dark blue line for Loa trend postshield basalts and rejuvenated basalts; light blue line for preshield Lō‘ihi lavas. Data noted in legend with an asterisk is from Chan and Frey (2003). Symbols for this figure are as in Figure 2.1. 45K2O/P2O5 greater than one. Variations outside of this window may still be included if the cause for the variation is from differentiation and not alteration. On a plot of K2O versus Rb (Figure 2.3b), most samples form coherent arrays. Because the extremely high K/Rb values of Kaua‘i postshield sample GA-650 are coupled with a normal lithium isotopic signature and a tight linear correlation on Figure 2.3b, interaction with a lithium-poor fluid, such as fresh groundwater, is likely and probably did not impact the lithium isotopic signature (Chan and Frey, 2003).Samples with abnormal lithium isotopic signatures and K/Rb>750 were removed from all following plots and discussion, along with Hawai‘i Scientific Drilling Project (HSDP) samples with K2O/P2O5<0.9 (Chan and Frey, 2003).2.3.2 Effect of Crystal Fractionation Major processes with the potential to disturb mantle source lithium isotopic signatures are solid-state diffusion of lithium between mantle minerals, diffusion of lithium between minerals and melt, and isotopic fractionation during crystal fractionation (Chan et al., 2009). Understanding how these processes impact measured lithium isotopic values is key to the application of lithium isotopes as a mantle tracer. Analysis of Kīlauea Iki whole rocks found that lithium isotopes are not fractionated by crystallization of olivine, plagioclase, clinopyroxene, and Fe-Ti minerals (Tomascak et al., 1999). Subsequent studies have, however, found a significant isotopic disequilibrium between olivine and clinopyroxene phenocrysts and the groundmass of basaltic rocks (Seitz et al., 2004; Beck et al., 2006; Magna et al., 2006; Jeffcoate et al., 2007; Parkinson et al., 2007). 6Li diffuses ~3.35% faster than 7Li into clinopyroxene, with the result that clinopyroxene commonly records the lowest δ7Li signatures in a rock sample (Richter et al., 2003; Coogan et al., 2005). In terms of whole rock δ7Li measurements, isotopic disequilibrium presents a measurable problem only if clinopyroxene is fractionated (Chan et al., 2009; Vlastélic et al., 2009).  In our sample set, none of the shield samples show correlations of lithium isotopes with MgO, indicating that crystal fractionation at low pressure does not fractionate lithium 46isotopes for these samples. Some of the postshield basalts from Kohala, Mauna Kea, and Hualālai show evidence of clinopyroxene fractionation, as exhibited by the decrease of Sc below 5 weight percent MgO for these samples (Figur 2.4a) (Wolfe et al., 1997; Hanano et al., 2010). Clinopyroxene fractionation measurably impacts lithium isotopic signatures as observed in plots of Ba/Y and Nb/Y versus δ7Li (Figure 2.4b, c). These trace element ratios are used as a proxy for the degree of partial melting. High Ba/Y or Nb/Y indicate high degrees of partial melting as a result of decreases in residual mantle clinopyroxene abundances (Huang et al., 2011b). However, the positive slope in Figure 2.4b and c suggests that residual clinopyroxene (with preferentially lower δ7Li compositions than the rock matrix and other phenocrysts; Jeffcoate et al., 2007; Parkinson et al., 2007) left in the source mantle    0   10   20   30   40   50  0   5   10   15   20   25   30Sc (ppm)MgO (wt %)!"#"$%0123  1   2    3   4   5   6Nb/Y01530  1   2    3   4   5   6Ba/Y δ7LiMauna Loa HualālaiMauna Loa* Kaua‘iKo‘olau* Ka‘ulaHualālai West Ka‘enaWest Ka‘ena Lō‘ihiKaua‘iMauna Kea*KohalaKīlaueaMauna KeaKohalaLoa TrendKea TrendMauna KeaMauna KeaKohalaKohalaKaua‘iKaua‘iClinopyroxene Fractionationa. b.c.Olivine FractionationFigure 2.4: Sc (ppm) versus weight percent MgO plot (a) The inverse correlation for the majority of data on this plot arises from the concentration of incompatible Sc during olivine fractionation, and the inflection point, which occurs in samples below 5 weight percent MgO, reveals clinopyroxene fractionation (Vlastélic et al., 2009). Samples below 10 ppm Sc were removed from the discussion to control for the effects of clinopyroxene fractionation (as seen in Nb/Y and Ba/Y versus δ7Li in (b) and (c) of this figure) on lithium isotopic signature. (b) Plot of Nb/Y versus δ7Li and (c) Plot of Ba/Y versus δ7Li for postshield basalts from Kaua‘i, Kohala, and Mauna Kea. Ba/Y and Nb/Y usually increase with decreasing amounts of partial melt (Huang et al., 2011a). Here the positive correlation results from clinopyroxene fractionation affecting the lithium isotopic signatures (see discussion in text). When samples with less than 10 ppm Sc are removed, the positive correlation in these plots disappears. Data noted in legend with an asterisk is from Chan and Frey (2003). Symbols for this figure are as in Figure 2.1.47by undergoing less partial melting would decrease the lithium isotopic signature of the melt, rather than increase it as expected. The explanation for this unexpectedly heavy lithium isotopic signature is not due to source residual clinopyroxene, but instead to clinopyroxene fractionation after separation of melts from the mantle source and before eruption. As Sc is compatible in clinopyroxene, but not in olivine, it can be used as a proxy for samples that have fractionated clinopyroxene (see Figure 2.4a). When samples with Sc concentrations below 10 ppm are removed from this dataset, the correlation between Ba/Y, Nb/Y, CaO, and MgO versus δ7Li for postshield samples disappears; therefore, the low-Sc samples will not be considered in the discussion on source processes below. Finally, a correlation of Kaua‘i postshield basalts δ7Li with Nb/Y, Ba/Y, and MgO is a well-documented feature of these rocks that reflects a heterogeneous mantle source rather than perturbation of lithium isotopes by non-source processes (Garcia et al., 2010). Coincidentally, the relationship between clinopyroxene fractionation and lithium isotopic signature suggests that isotopic disequilibrium between phenocrysts and groundmass in basaltic rocks is a result of equilibrium or kinetic fractionation during magmatic differentiation rather than solid-state diffusion of lithium.  2.3.3 Lithium Isotopic Signature A description of the lithium isotope ratios of the different volcanic stages are described in the following subsections.  2.3.3.1 Shield Basalts Shield-stage basalts range from δ7Li=1.8 to 5.7‰. The Loa and Kea geochemical trends are characterized by systematic differences in lithium isotopes, with Kea trend shield basalts appreciably heavier (25 of 31 samples with δ7Li above 3.5‰) than Loa trend shield volcanics (only 19 of 39 above 3.5‰) (Figure 2.5). When known outlier volcano Ko‘olau (see Ko‘olau section in discussion below) is removed from consideration, the difference between Loa and Kea shields becomes more apparent (i.e., only 13 of 34 Loa trend shield48samples above 3.5‰) (Figure 2.5). This result is contrary to the findings of Chan and Frey (2003), who did not see a significant difference between Loa and Kea trend volcanoes when studying Mauna Loa, Mauna Kea, and Ko‘olau shield basalts (n=54). This discrepancy may 20 1 3 4 5 61086420108642020 1 3 4 5 620 1 3 4 5 61086420Kea Shield   N=31Kea PostShield       N=9Rejuvenated      N=720 1 3 4 5 61086420Loa PostShield        N=1820 1 3 4 5 61086420Preshield (Lō‘ihi)         N=920 1 3 4 5 61086420Loa Shield   N=36 Ko‘olau      δ7Li    δ7LiKohalaLow SiO₂Mauna KeaAv MORBFigure 2.5: Histograms of lithium isotopic data of this study combined with data from Chan and Frey (2003). The y-axis on each plot indicates the number of samples within 0.5‰ intervals of δ7Li variation. Kea trend shield basalts are displaced to higher average δ7Li than Loa trend shield basalts. Loa trend shield basalts have more heterogeneous lithium isotopic signatures than Kea trend shield basalts. For both Kea and Loa trend volcanics, postshield basalts are lower in average δ7Li than shield basalts. The dashed line in each histogram indicates the average δ7Li ≈ 3.7 ± 1‰ (1SD) of MORB (calculated from Moriguti and Nakamura, 1998; Elliott et al., 2006; Nishio et al., 2007; Tomascak et al., 2008; and Chan et al., 2009). Basalts from the Makapu‘u section of Ko‘olau volcano are distinguished from the rest of Loa trend basalts because the heavier lithium isotopic signature of this volcano is unique from the rest of the Loa trend volcanoes (see discussion). Also note that Kohala volcano, not Mauna Kea, exhibits the lowest lithium isotopic signature of Kea trend shield basalts. Low SiO2 Mauna Kea basalts are distinguished in dark red and are based on the classification of Rhodes and Vollinger (2004).49arise from the fact that the present study includes more volcanoes and focuses on volcanoes shown to have wide-ranging compositions (Weis et al., 2011). Loa trend volcanoes are also more heterogeneous in lithium isotopes than Kea trend volcanoes (Figure 2.5). Mauna Kea shield samples (from Chan and Frey, 2003) are separated into low- and high-SiO2 groups based on the characterization of Rhodes and Vollinger (2004). Generally, higher SiO2 Mauna Kea shield samples also exhibit heavier lithium isotopic signatures (Figure 2.5) (Chan and Frey, 2003).2.3.3.2 Postshield Basalts Postshield samples have systematically lighter lithium isotopic signatures than shield basalts (average of all postshields: δ7Li=2.7‰ (n=24); average all shields: δ7Li=3.9‰ (n=74). On Hawai‘i Island, lithium isotopic signature lightens with the transition from shield to postshield on Mauna Kea and Hualālai volcanoes (Figure 2.7c, d). This agrees with the Pb-Pb linear trends observed by Hanano et al. (2010) for the shield to postshield transitions of these volcanoes with Mauna Kea and Hualālai becoming less radiogenic with time. Finally, it is important to note that postshield samples from Hualālai extend the Hawaiian lithium isotopic signature to the lightest measured whole-rock values for all oceanic islands, and appear to be sampling an end-member in lithium isotopic space (Figures 2.7b, 2.8b, 2.9b). Hualālai postshield volcanics range from δ7Li=0.8 - 3.5‰ (n=11) with almost three quarters of samples below δ7Li=2.6‰. Hualālai postshield basalts also have among the lowest 206Pb/204Pb observed in Hawai‘i (Hanano et al., 2010). It is important to note that the source of this light lithium signature is not due to clinopyroxene fractionation, as samples exhibiting signs of this process have been excluded in the dataset (Figure 2.4). 2.3.3.3 Rejuvenated and Preshield Basalts Of all the volcanic stages, rejuvenated and preshield samples represent the most and least homogeneous stages, respectively, in lithium isotopic signatures. Rejuvenated samples have an average δ7Li=3.78 ± 0.99 (2SD, n=7), and appears to be the most homogeneous 50volcanic stage sampled in this study. The remarkable similarity of rejuvenated samples from Ka‘ula, Kaua‘i, and West Ka‘ena in lithium isotopes is similar to the homogeneity of Hawaiian rejuvenated samples in general as observed in trace elements and radiogenic isotopes (Yang et al., 2003; Frey et al., 2005; Fekiacova et al., 2007; Garcia et al., 2010). Preshield samples from Lō‘ihi range in δ7Li=0.9 - 5.5‰ (n=9), by far the largest variation for a volcano observed in the 88 samples analyzed.2.3.3.4 Lithium Isotopes of Hawai‘i Versus Global Oceanic Island Basalts Globally, oceanic island basalts range from δ7Li=2.5‰ in the Foundation Seamount Chain to 8.0‰ on Rurutu Volcano in the Cook-Austral Islands (Vlastélic et al., 2009; Krienitz et al., 2012). Hawaiian data (this study), particularly from Hualālai and Lō‘ihi volcanoes, extend the global range of values down to δ7Li=0.8‰. On a plot of Li/Y versus δ7Li (Figure 2.6), it is interesting to note that most OIB plot above the average MORB value of 0.20 ± 0.04 (Sun and McDonough, 1989; Bouman et al., 2004; Krienitz et al., 2012; Gale et al., 2013) and some below. Hualālai postshield volcanics are characterized by high Li/Y, much higher than the 0.20 of MORB, and systematically higher than any other oceanic island basalt. Lō‘ihi samples are among the lowest Li/Y values, whereas HIMU oceanic islands have consistently higher values.2.4. DISCUSSION The lithium isotopic signature of Hawaiian basalts varies between volcanic stages and different Hawaiian volcanoes and, as a group, compared to other oceanic islands. A discussion of these observations follows.2.4.1 Lithium Isotopic Signature in Hawaiian Basalts: A Source Signature? The variation observed in lithium isotopes in this study is small and generally cluster around average MORB (δ7Li≈3.7‰; Moriguti and Nakamura, 1998; Elliott et al., 2006; Nishio et al., 2007; Tomascak et al., 2008; Chan et al., 2009). However, lithium isotopic signatures correlate systematically with other trace elements and radiogenic isotopes,  51implying that lithium isotopes do behave systematically enough to be used as a geochemical tracer. Hawaiian volcanoes that are geochemically distinct in Pb-Hf-Nd-Sr isotopes are also distinct in lithium isotopes (Figures 2.7, 2.8, 2.9; Hf isotopes not shown). Kea trend volcanoes are systematically higher in δ7Li, 206Pb/204Pb, 207Pb/204Pb, 208Pb/204Pb, 143Nd/144Nd, and lower in 87Sr/86Sr than Loa trend volcanoes. Because this trend holds for both shield and postshield basalts, it is more likely that variations in lithium isotopes arise from mantle heterogeneities rather than from non-linear diffusion, melting, or crystallization processes. Although Ko‘olau is distinctly a Hawaiian end-member in radiogenic isotopic systems (Pb, Hf, Nd, Sr), in lithium isotopes it apprears that Hualālai and Mauna Kea form the extremes of an isotopic mixing array that can be resolved in lithium isotopes as well as in Pb, Hf, Nd, and 00.20.40.60.80 1 2 3 4 5 6 7 8 9Li/YMORB Li/Y= 0.2HualālaiLō‘ihi   δ7LiMauna LoaMauna Loa*Ko‘olau*HualālaiWest Ka‘enaKaua‘iMauna Kea*KohalaKīlaueaMauna KeaKohalaHualālaiKaua‘iKa‘ulaWest Ka‘enaLō‘ihiCook-Austral SCAzoresEaster SCFoundation SCJuan FernandezMarquesasPitcairnRéunionSocietySt HelenaLiterature DataHawai‘iFigure 2.6: Li/Y versus δ7Li for compiled oceanic island basalts. The heavy dark line denotes the MORB Li/Y value of 0.2 ± 0.04, which is calculated from N-MORB values of Sun and McDonough (1989) and Gale et al. (2013; Bouman et al., 2004; Krienitz et al., 2012). The grey field indicates two standard deviations (2SD) of MORB Li/Y. SC is shorthand for seamount chain. Data sources listed in caption of Figure 2.11. Symbols for this figure are as in Figure 2.1.52Sr isotopes. The covariation between lithium and radiogenic isotopes suggests that lithium can preserve old, low-temperature fractionation signatures throughout long residence times in the mantle (up to two billion years according to the calculations of Vlastélic et al., 2009).2.4.2 Hawaiian End-Member Components in Lithium Isotopes Hawaiian geochemical variation is typically explained by mixing of different end-member components in the mantle source of volcanoes (Tanaka et al., 2008). This section addresses apparent end-members as identified by lithium isotopes.  2.4.2.1 The Kea Component The depleted shield-stage source component in Hawaiian basalts is the Kea geochemical component. Although the Kea component is most dominant in Kea trend volcanoes, it contributes to all Hawaiian volcanism and has been shown to be a long-lived component in the Hawaiian mantle plume (>80 Ma; Blichert-Toft et al., 2003; Regelous et al., 2003; Frey et al., 2005; Portnyagin et al., 2008; Tanaka et al., 2008; Jackson et al., 2012). Although there is still debate on the lithology of the source of Hawaiian shield volcanoes, a strictly peridotitic source for Kea trend volcanoes will first be considered. This peridotitic source can originate in the upper mantle as either entrained asthenosphere or entrained lithosphere or from the deep mantle as either plume matrix or recycled oceanic lithosphere (Chen and Frey, 1985; Eiler et al., 1996; Hauri, 1996; Lassiter et al., 1996; Lassiter and Hauri, 1998; Weis et al., 2011; Jackson et al., 2012; Nobre-Silva et al., 2013a). The lithium budget of peridotite is controlled by olivine, a mineral that is remarkably homogeneous with respect to lithium isotopes (Seitz et al., 2004; Magna et al., 2006; Jeffcoate et al., 2007). The range of observed lithium isotopic compositions of olivine in fertile peridotitic mantle xenoliths with minor to no metasomatism or melt extracted is δ7Li=3.0 – 5.2‰ with an average bulk composition of δ7Li≈3.5‰ (Jeffcoate et al., 2007). This composition is both near average MORB and a common composition for Kea trend samples (Figure 2.5). The range observed by Tomascak et al. (2008) for global MORB extends to δ7Li=5.6‰, very 53near the heaviest Kea trend volcanic lithium isotopic signature of δ7Li=5.7‰ (Mauna Kea, Chan and Frey, 2003). Thus, although the average of all Kea trend volcanoes (δ7Li≈4.3‰) is heavier than average MORB (δ7Li≈3.7‰), a depleted peridotite similar to MORB may explain all observed Kea-trend variation, as can fertile, previously unmelted peridotite. There is the possibility that the slightly heavier average of Kea trend volcanism relative to average MORB is due to incorporation of very small amounts of a HIMU-type mantle, which is characterized by heavier lithium isotopes in the range of δ7Li=2.8 – 8.0‰  (Chan et al., 2009; Vlastélic et al., 2009; Krienitz et al., 2012). However, because the Kea component lithium isotopic signature is so near average MORB and the composition of prevalent mantle (PREMA; see discussion below and Figure 2.11), it is more likely that it records the average lithium isotopic signature of peridotitic mantle. As lithium isotopes cannot distinguish between depleted, recycled or fertile peridotite, they are unsuitable for distinguishing between a deep or shallow source for the Kea component on Hawai‘i.  Several authors have put forth a different source model of Hawaiian shield volcanoes, including those from the Kea trend, that includes a mixture of both peridotite and a pyroxenite formed from eclogitic melts reacting with peridotite (Sobolev et al., 2005, 2007; Herzberg, 2006, 2011). This model was proposed to explain high Ni concentrations and silica contents of some Hawaiian parental magmas. According to the model, Hawaiian basalts with higher weight percent SiO2, such as Ko‘olau and Mauna Loa, reflect the addition of greater proportions of this hybrid pyroxenite source than Lō‘ihi, Mauna Kea, and Kīlauea (Sobolev et al.,2005). As discussed in the sections above, peridotite and pyroxenite theoretically should produce different lithium isotopic signatures in the resulting melt due to lower δ7Li in clinopyroxene (Richter et al., 2003; Jeffcoate et al., 2007; Parkinson et al., 2007). Thus, lithium isotopes may provide some insight in this debate. First, consider Mauna Kea, proposed to include ~45% of this hybrid pyroxenitic source (Sobolev et al., 2005). If an increase in the amount of SiO2 in Mauna Kea basalts is due to a greater addition of a54      17.7      18.0      18.3      18.6  0   1   2   3   4   5   6   7206Pb/204PbLō‘ihi Mauna KeaMauna LoaMauna Kea206Pb/204Pba. Preshield and Shieldb. Postshield and Rejuvenated     18.30     18.45     18.60  2   3   4   5   6   7206Pb/204Pb   δ7Li   δ7LiKo‘olauEPR MORBc.Site 843      17.8      18.0      18.2      18.4  0   1   2   3   4   5   6   7HualālaiMauna KeaKohalaKaua‘iRejuvenatedEPR MORBSite 843      18.5   δ7LiAverage LCCAverage LCCShieldPostshieldHualālai     17.75     17.90     18.05     18.20   0   1   2   3   4   5206Pb/204Pb   δ7Lid.ShieldPostshieldMauna Kea*KohalaKīlaueaMauna KeaKohalaMauna Loa HualālaiMauna Loa* Kaua‘iKo‘olau* Ka‘ulaHualālai West Ka‘enaWest Ka‘ena Lō‘ihiKaua‘iLoa TrendKea TrendFigure 2.7: Comparison of the Pb and Li isotopic characteristics for preshield, shield, postshield and rejuvenated volcanism on Hawai‘i. (a) Shield and preshield basalts. Aside from Ko‘olau (see discussion in text), Hawaiian shield basalts exhibit a positive correlation. (b) Postshield and rejuvenated basalts. Hualālai postshield samples extend to the lowest δ7Li values on Hawai‘i. For plots (a) and (b) the field for EPR MORB is compiled data from Elliott et al., 2006 and Tomascak et al., 2008. ODP Site 843 basalts are noted by the dark grey field. Plots (c) and (d) show the transition from shield to postshield in 206Pb/204Pb versus δ7Li for Mauna Kea and Hualālai. Note different vertical and horizontal scales for (c) and (d). The lithium isotopic composition lightens with the transition from shield to postshield volcanic stage at individual volcanoes. Data noted in legend with an asterisk is from Chan and Frey (2003). The composition of lower continental crust is indicated by the arrow and is from Teng et al., 2004 and Rudnick and Goldstein, 1990. Symbols for this figure are as in Figure 2.155pyroxenitic melt in these basalts, they should also exhibit lower lithium isotopic signature. However, low-SiO2 Mauna Kea basalts have lower lithium isotopic signature relative to the higher SiO2 basalts (Figure 2.5). The lithium isotopic signature between Mauna Loa and Ko‘olau also conflicts with expected results. According to Sobolev et al. (2005), there is approximately 60% hybrid pyroxenite in Mauna Loa and 80% in Ko‘olau, supported by the higher weight percent SiO2 in Ko‘olau basalts. A higher percentage of pyroxenite in the source of a volcano should produce lighter lithium isotopic signatures, a relationship not observed in this study as Ko‘olau exhibits the heaviest lithium isotopic signature of any Loa trend volcano (Figure 2.5). Thus lithium isotopes do not support the addition of a hybrid pyroxenite in the source of Hawaiian magmas.2.4.2.2 Lō‘ihi: Heterogeneity in Lithium Isotopes The substantial heterogeneity observed in δ7Li for Lō‘ihi is perhaps not extraordinary when considering the overall anomalous and heterogeneous character of this volcano (Moore et al., 1982; Garcia et al., 1993, 1995, 1998, 2006). Lō‘ihi is a very young submarine volcano (between 5-102 ka; Guillou et al., 1997) and is the only Hawaiian volcano that has been sampled during the preshield stage, whose chemistry may be controlled by the first melting of plume heterogeneities with different solidi (Huang et al., 2005b). The initial stages of melting of a new, previously unmelted portion of the mantle source along with the lower magma production of the preshield stage (characterized by mostly alkalic basalts of a lower percentage melt) presents the opportunity to resolve smaller scale mantle heterogeneities. In the shield-stage, such heterogeneities may be fully exhausted from the mantle source or may be overwhelmed by the greater percentage melts and larger magma chambers of the shield-stage (Garcia et al., 2010 and references therein). The heterogeneity in lithium isotope compositions of Lō‘ihi Volcano may reflect very small-scale heterogeneities that are not sampled during the shield-stage (e.g., smaller scale than the heterogeneity observed in shield-stage Mauna Loa and Kīlauea by Marske et al., 2007). Conversely, the heterogeneity56143Nd/144Nd0.512580.512780.512980.513181 2 3 4 5 6 7143Nd/144NdEPR MORBMauna LoaMauna Kea0.512880.512930.512980.513030.513080 1 2 3 4 5 6 7EPR MORBMauna KeaHualālaiRejuvenated   δ7Li0%0.5%1%Ko‘olauAverage LCCAverage LCCb. Postshield and rejuvenateda. Preshield and shield   δ7LiMauna Loa HualālaiMauna Loa* Kaua‘iKo‘olau* Ka‘ulaHualālai West Ka‘enaWest Ka‘ena Lō‘ihiKaua‘iMauna Kea*KohalaKīlaueaMauna KeaKohalaLoa TrendKea Trend0.5090.5100.5110.5120.5130 4 8 12 16143Nd/144Nd   δ7LiAncient Marine CarbonateFigure 2.8:  Nd versus Li isotopes for (a) shield and preshield basalts and (b) postshield and rejuvenated basalts. Note different vertical and horizontal scales for (a) and (b). The line in (a) illustrates the mixing of an ancient marine carbonate with a plume component (Huang et al., 2011b; see Table 2.3 for model parameters). Tick marks are at intervals of 0.5%. The maximum amount of ancient marine carbonate in Ko‘olau basalts constrained by Nd isotopes is 1%. Inset to (a) shows a larger-scale view of the plotted data and the full mixing line to an ancient marine carbonate end-member (see text for discussion). The field for EPR MORB is compiled data from Elliott et al. (2006) and Tomascak et al. (2008). ODP Site 843 basalts are noted by the dark grey field. Data noted in legend with an asterisk is from Chan and Frey (2003). The composition of lower continental crust is indicated by the arrow and is from Teng et al. (2004) and Stosch and Lugmair (1984). Symbols for this figure are as in Figure 2.1.5710.70250.70290.70330.70370.70412 3 4 5EPR MORBMauna KeaMauna Loa0%1%2%3%4%5% Ko‘olauSite 843Average LCC0.70230.70280.70330.70380 1 2 3 4 5 6EPR MORBMauna KeaHualālaiRejuvenated   δ7Li87Sr/86Sr0.70330.70370.70410.70450 4 8 12 1687Sr/86Sr   δ7LiAncient Marine Carbonate1 6Average LCCa. Preshield and ShieldMauna KeaKohalaLō‘ihi Mauna Kea*KohalaKīlaueaKea TrendMauna LoaMauna Loa*Ko‘olau*HualālaiWest Ka‘enaKaua‘iLoa Trend87Sr/86SrHualālaiKaua‘iKa‘ulaWest Ka‘enaSite 843b. Postshield and Rejuvenated   δ7LiFigure 2.9: Sr versus Li isotopes for (a) shield and preshield basalts and (b) postshield and rejuvenated basalts. Note different vertical and horizontal scales for (a) and (b). The line in (a) illustrates the mixing of an ancient marine carbonate with a plume component (Huang et al., 2011a; see Table 2.3 for model parameters). Tick marks are at intervals of 1%. The maximum amount of ancient marine carbonate in Ko‘olau basalts constrained by Sr isotopes is 5%. Inset to (a) shows a larger-scaled view of the plotted data and the full mixing line to an ancient marine carbonate end-member (see text for discussion). For plots (a) and (b) the field for EPR MORB is compiled data from Elliott et al. (2006) and Tomascak et al. (2008) and does not overlap with Hawaiian data. ODP Site 843 basalts are noted by the dark grey field. Data noted in legend with an asterisk is from Chan and Frey (2003). The composition of lower continental crust is indicated by the arrow and is from Teng et al. (2004) and Stosch and Lugmair (1984). Symbols for this figure are as in Figure 2.1.  58of Lō‘ihi may be an artifact of its dynamic hydrothermal system and frequent summit collapses that have been shown to contaminate some Lō‘ihi samples with a seawater-derived component (Kent et al., 1999; Garcia et al., 2006). Incorporation of varying amounts of seawater-altered rocks into Lō‘ihi’s summit magma chamber would create substantial lithium isotopic heterogeneities, although this mechanism would result in heavier lithium isotopic signatures and cannot account for the extremely light lithium isotopic signatures found in some Lō‘ihi samples (Figure 2.5)2.4.2.3 Hualālai: Light Lithium End-Member  Hualālai Volcano is identified here and in previous papers as a perplexing outlier among Loa trend volcanoes, with the lowest documented 206Pb/204Pb at a given 208Pb/204Pb for postshield basalts (Hanano et al., 2010). Hualālai also exhibits the highest Li/Y of almost any oceanic island basalt (Figure 2.6). Li and Y have very similar partition coefficients in basaltic systems, so variations in these elements relative to one another are unlikely to be related to partial melting or fractional crystallization (Ryan and Langmuir, 1987). Because Y is essentially a fluid-immobile element, variations in Li/Y are thus typically attributed to the geochemical behavior of lithium during subduction and, more specifically, to the differences in lithium geochemical behavior based on both type of material being subducted and depth level in the subducting package. Lithium is a fluid mobile element that enters fluids as subducted materials are dewatered, leaving the slab both depleted in lithium and Table 2.3: Input model parameters for Figures 2.8, 2.9, 2.107Li Li (ppm) 143Nd/144Nd Nd (ppm) 87Sr/86Sr Sr (ppm) La (ppm) Nb (ppm) Th (ppm)Hawaiian Plume 4.4a 1.6c 0.512946a 1.354c 0.7034e 18.21e 0.687c 0.713c 0.085cAncient Carbonate 14.5b 0.8d 0.50938d 10.13d 0.7046e 800e 14.22d 0.79d 0.29daAverage of all Mauna Loa and Mauna Kea shield-stage analyses of samples in this studycPrimitive mantle values from McDonough and Sun, 1995dGuatemala Carbonate (Reference Site 495) of Plank and Langmuir, 1998eHuang et al., 2011bbHeavy end-member carbonate of Chan et al. (2006), chosen because subducted carbonates have likely undergone diagenetic recrystallization, a process found to result in heavier lithium isotopic signatures59with a lighter lithium isotopic signature by up to 3‰ (Bouman et al., 2004; Marschall et al., 2007; Tang et al., 2010). If this process is incomplete or heterogeneous, there is the potential to introduce a seawater-lithium enriched reservoir into the deep mantle through recycling oceanic crust (Chan et al., 2009; Vlastélic et al., 2009; Krienitz et al., 2012). Hualālai postshield samples may sample such a source. Conversely, the low Li/Y source exhibited in basalts from Lō‘ihi relative to MORB may arise from a subducted source in which much of the lithium was removed during subduction.  Hualālai postshield basalts also exhibit the most EM-I like signature of any volcano or volcanic stage on Hawai‘i (see Figure 2.11). This signature includes low 206Pb/204Pb and 87Sr/86Sr as well as high abundances of Li, Ba, and Rb and depletions of Pb relative to Ce and Nd (Lustrino and Dallai, 2003; Nishio et al., 2004; Willbold and Stracke, 2006; Hanano et al., 2010). It has also been shown that ultramafic xenoliths with EM-I type geochemical characteristics from southwestern Japan and eastern Russia have light lithium isotopic signatures, down to δ7Li≈ -17‰ (Nishio et al., 2004; it should be noted that this value may be lower than the actual EM-I mantle source due to measurements solely on clinopyroxene in this study). Although Nishio et al. (2004) attributed the light lithium isotopic signature of EM-I mantle to a mixture of the upper, most highly altered layer of oceanic crust with pelagic sediments, the highly unradiogenic 206Pb/204Pb and radiogenic 176Hf/177Hf isotopic ratios of Hualālai postshield basalts rules out pelagic sediment as a source of these basalts (Blichert-Toft et al., 1999; Hanano et al., 2010). Instead, a mixture of altered oceanic crust and either subduction eroded or delaminated lower continental crust may account for the trace element, radiogenic isotopes, and lithium isotopic signature of Hualālai postshield volcanics (see Figures 2.7, 2.8, 2.9; Willbold and Stracke, 2006). Lower continental crust has an average lithium isotopic composition of δ7Li≈2.5‰ (Teng et al., 2008), that when mixed with subduction-altered oceanic crust may be light enough to explain signatures down to the δ7Li=0.8‰ as observed for Hualālai. Because mantle lithosphere, which is difficult to60 separate from overlying lower continental crust in delamination scenarios, carries a distinct osmium isotopic signature that is not observed in oceanic island basalts, it is more likely that the lower continental crust in the source of Hualālai was subduction-eroded rather than delaminated (White, 2010). Finally, although the lithium isotopic signature of lower Mauna Kea*Lō‘ihiMauna LoaMauna Loa*Ko‘olau*HualālaiWest Ka‘enaKaua‘i!"(*10254055 Sr/Nb0%1%2%a.b.30 1 2 4 5 6 7δ7Li30 1 2 4 5 6 7δ7LiKo‘olauMauna KeaHualālaiMauna LoaLō‘ihi0.50.91.31.7 La/Nb0%1%2%3%Ko‘olauMauna KeaHualālaiLō‘ihiMauna LoaFigure 2.10: Plots of shield-stage (a) La/Nb and (b) Sr/Nb versus δ7Li Averages of each volcano are marked with a black-outlined star of the same color. The dashed lines show linear regressions of Loa trend volcano averages. There are correlations on each of these plots between average Lō‘ihi at low La/Nb and Sr/Nb and Ko‘olau at high La/Nb and Sr/Nb. Averages for Mauna Loa and Hualālai plot in between Ko‘olau and Lō‘ihi. Mauna Kea plots entirely off of the trend and is thus not affected by this mixing relationship. The solid black line in each plot illustrates the mixing of an ancient marine carbonate with a plume component (Huang et al., 2011b; see Table 2.3 for model parameters). Tick marks are at intervals of 1%. The maximum amount of marine carbonate in the Makapu‘u section of Ko‘olau constrained by La/Nb and Sr/Nb is 3-4%. Data noted in legend with an asterisk is from Chan and Frey (2003). Symbols for this figure are as in Figure 2.1. 61continental crust (δ7Li≈2.5‰; Teng et al., 2008) and pelagic sediment (δ7Li≈2.25‰; Bouman et al., 2004) are too similar to distinguish at the current level of analytical precision, we propose that the source of heterogeneity in Hualālai postshield samples is more likely the presence of a small amount of recycled lower continental crust rather than pelagic sediments.2.4.2.4 Lithium Isotopes of the Makapu‘u Section of Ko‘olau Volcano: A Sedimentary Component Previous authors have attributed a light lithium isotopic signature in mantle melts to melting of recycled oceanic crust and/or sediments (Brooker et al., 2004; Elliott et al., 2004, 2006; Kobayashi et al., 2004; Tomascak et al., 2004; Tang et al., 2010; Krienitz et al., 2012). A recycled sedimentary source has been proposed for some Hawaiian basalts based on evidence from hafnium, osmium, lead, calcium, and oxygen isotopes as well as trace and major elements (Frey et al., 1994; Roden et al., 1994; Eiler et al., 1996; Lassiter and Hauri, 1998; Blichert-Toft et al., 1999; Jackson et al., 1999; Huang and Frey, 2005; Huang et al., 2009; Huang et al., 2011b; Jackson et al., 2012). Recycled altered oceanic crust is also cited as a source of mantle heterogeneity for Hawai‘i, oceanic island basalts, and MORB (Zindler and Hart, 1986; Rehkämper and Hofmann, 1997; Hofmann, 2003; Salters et al., 2006; Jackson et al., 2012; Gale et al., 2013). The average lithium isotopic signatures for shield basalts from Loa trend volcanoes correlate with geochemical indices of sediment input, i.e., La/Nb and Sr/Nb (Figure 2.10) (Huang and Frey, 2005). The volcanoes at either end of this mixing array are Ko‘olau and Lō‘ihi with Hualālai and Mauna Loa in between (a trend also observed for these volcanoes in Pb-Hf-Nd-Sr isotope systematics; Frey et al., 1994; Tanaka et al., 2008). Analyses of Kea trend volcanoes plot off of these trends, suggesting that Kea trend volcanoes do not have a recycled sedimentary input (Jackson et al., 2012) and correspond to average deep Pacific mantle (Nobre Silva et al., 2013a). Based on Figure 2.10, Loa trend shield basalts plot on a mixing array between the light lithium isotopic composition (e.g., Lō‘ihi and Hualālai) and the heaviest lithium 62isotopes (e.g., Ko‘olau). This may reflect a continuum in the Loa trend source between the recycling of a carbonate sedimentary input and a lower crustal input. Huang and Frey (2005) and Huang et al. (2009; 2011b) have argued for the presence of ancient recycled carbonate in the source of Ko‘olau Volcano based on major and trace element abundances as well as the radiogenic and calcium isotopic signatures of these volcanoes. Lithium isotopes seem to agree with this hypothesis for several reasons.  The basalts from the Makapu‘u section of Ko‘olau Volcano, an end-member volcano in many trace element, radiogenic and stable isotopic plots (e.g., Lassiter and Hauri, 1998; Tanaka et al., 2002; Fekiacova et al., 2007; Tanaka et al., 2008; Huang et al., 2011b; Jackson et al., 2012; Teng et al., 2013), plot consistently off-trend in Sr, Nd, and Pb isotopic plots versus lithium due to their heavier lithium isotopic signature relative to other Loa trend volcanoes. Thus, Ko‘olau is likely sampling an end-member that is not common in Mauna Loa, Hualālai, and Kaua‘i (Figures 2.7, 2.8, 2.9), a conclusion that is also supported by the much steeper slope of Ko‘olau relative to every other Hawaiian volcano in Pb isotopic space (Abouchami et al., 2005; Weis et al., 2011). Pelagic sediments have extremely high lithium abundances (up to 57.3 ppm) and lithium isotopic signatures between δ7Li=1.7 – 4.8‰ with an average signature of 2.25‰ (Bouman et al., 2004). The relatively heavy lithium isotopic signature of the Makapu‘u section of Ko‘olau volcano precludes pelagic sediment as a major source. In contrast, the lithium isotopic signature of carbonate sediments varies between δ7Li=6.4-23.33‰ and is controlled by both equilibrium with the fluids they are in contact with (typically a heavy seawater signature ≈32‰) and sedimentary diagenetic processes (e.g., recrystallization of silica and carbonate in the presence of heavy seawater; Chan et al., 2006). Ancient recycled carbonates have higher 87Sr/86Sr (Huang and Frey, 2005; Ray et al., 2002), and marine carbonates in general have high Sr/Nb and low Th/La (Plank and Langmuir, 1998), characteristics also exhibited by Ko‘olau basalts. The low Lu/Hf values of Ko‘olau relative to other Loa trend volcanoes has been used to argue against a modern 63carbonate source, where the lower concentrations of Hf relative to Lu drive the trace element ratios to slightly higher values. However, ancient carbonates may have had a lower Lu/Hf than modern carbonates. In pre-Phanerozoic time, a high-REE scavenging phosphate component was likely absent from subducting carbonate-rich sediments due to the lack of abundant fish and shell debris in oceanic sediments before the Cambrian Explosion (Plank and Langmuir, 1998; Willbold and Stracke, 2006). Contrary to Huang and Frey (2005), we propose a largely phosphate-poor carbonate component in the source of Makapu‘u section Ko‘olau basalts. Based on a simple two-component mixing model (Langmuir et al., 1978; Huang et al., 2011b; see Table 2.3), we calculate roughly 1-5% carbonate in the Makapu‘u shield stage of Ko‘olau Volcano. Some West Ka‘ena basalts sample a similar end-member as Ko‘olau, but in smaller amounts (Figures 2.7a, 2.8a, 2.9a). The proposed ancient recycled carbonate end-member has been present in the Loa trend Hawaiian mantle from at least West Ka‘ena to Ko‘olau, roughly 1-3 million years (Doell and Dalrymple, 1973; Greene et al., 2010; Huang et al., 2011b). The presence of recycled material in the source of Loa trend volcanoes and the absence of this material in Kea trend volcanoes may be the main reason controlling Loa and Kea trend differences in lithium isotopes.2.4.2.5 Lithium Isotopic Heterogeneity in Loa Trend Volcanoes Loa trend volcanoes are considerably more heterogeneous in lithium isotopes than Kea trend volcanoes (Figure 2.9), as is also observed in radiogenic isotopic systems (e.g., Weis et al., 2011). The Loa trend volcanic source appears to contain recycled components that are carbonate-rich sediment in the Makapu‘u section of Ko‘olau Volcano and lower continental crust and altered oceanic crust rich (with a lighter lithium isotopic signature) in Hualālai. The greater heterogeneity observed in Loa trend volcanoes, and especially Lō‘ihi, is therefore likely due to the fact that the lithium isotopic signature of subducting materials is heterogeneous. For example, the range of lithium isotopes in altered oceanic crust is large (δ7Li=-2.29-13.81‰; Gao et al., 2012) and highly dependent on both the water-rock ratio 64(which changes with depth in the crust) and the highly channelized and flow-dependent nature of hydrothermal systems. In addition, there is a large range in the possible lithium isotopic composition of subducting sediment that may have a large impact on recycled lithium. The four main types of sediment (carbonate-rich, pelagic clay and ooze-rich, terrigeneous, and volcaniclastic) have very different lithium isotopic signatures (δ7Li=13.0-23.3‰, 1.3-14.5‰, -1.7-2.5‰, and ≈6‰, respectively; Bouman et al., 2004). The type of sediment recycled, different dehydration behaviors and processes, and the spatial locations and mixing of each type of oceanic crust and/or sediment in the source could impact the sampled heterogeneity of oceanic island basalts. The lower mantle near the core mantle boundary, described by some as a “slab graveyard”, may be sampling many different types of materials subducted over presumably millions of years, reflecting the long-term evolution of elemental and isotope ratios of continental crust, seawater, and biogenic activity (Plank and Langmuir, 1998; Vervoort and Blichert-Toft, 1999; Ray et al., 2002; Valley et al., 2005). The lithium isotopic signature of such a mélange of material is not uniform (Stracke et al., 2003), and the heterogeneity observed in the Loa trend volcanoes on Hawai‘i possibly reflects this.2.4.3 Why Are Postshield Basalts Lighter in Lithium Isotopic Signature Than Shield Basalts? The transition from the shield to postshield stage in Hawaiian volcanoes is accompanied by changes in basalt alkalinity, trace element concentrations, and isotopic ratios (Feigenson et al., 1983; Clague, 1987; Frey et al., 1990; Lassiter et al., 1996; Xu et al., 2005, 2007; Hanano et al., 2010). Because the trace element abundances of shield and postshield basalts typically lie along common fractionation lines (see Figure 4 in Garcia et al., 2010), the changes in basalt geochemistry most likely stem from changes in degree of partial melt, magma supply, and the depth of melt segregation rather than a wholesale change in source as is required for rejuvenated basalts (Frey et al., 1990; Garcia et al., 2010; Hanano et al., 2010). Jeffcoate et al. (2007) found that the lithium isotopic signatures of clinopyroxene in 65fertile peridotite xenoliths were heavier than clinopyroxene in more depleted xenoliths. In addition, 7Li will preferentially partition into the mobile fluid phase in any system (Chan et al., 1992; Tomascak et al., 2004 and references therein; Chan et al., 2006). Any previous depletion of the postshield source rock by shield stage melting should systematically lower the lithium isotopic signature of postshield magmas. This effect is quite small, however, as no fractionation is observed between olivine and melt and the fractionation factor between clinopyroxene and olivine is on the order of 0.9985 (Tomascak et al., 1999; Chan and Frey, 2003; Jeffcoate et al., 2007). Consequently, variable source depletion cannot fully explain the observed reduction in lithium isotopic signatures of postshield basalts. The decrease in lithium isotope ratios from shield to postshield basalts could result from the lower degrees of partial melting that characterize postshield volcanism relative to shield volcanism. Pyroxenite has a lower solidus temperature than peridotite and will be the first mantle lithology to melt (Hirschman and Stolper, 1996; Petermann and Hirschmann, 2003; Lambart et al., 2012). If any pyroxenite was not fully melted during the shield-stage, or exist as vertical “stringers” of heterogeneity in the Hawaiian mantle plume that are consistently migrating into the plume melting zone (Blichert-Toft et al., 2003; Marske et al., 2007; Hanano et al., 2010), then the smaller degrees of melting associated with postshield volcanism would result in a higher proportions of the clinopyroxene-rich source in the melt. Higher modal melting of clinopyroxene, a mineral with a preferentially light lithium isotopes, would result in a lighter lithium isotopic signature in the resulting erupted basalt (Seitz et al., 2004; Magna et al., 2006; Jeffcoate et al., 2007). This effect may be magnified by the higher concentrations of lithium in mantle pyroxenite relative to fertile peridotite (Seitz and Woodland, 2000). The ≈32% higher lithium concentrations of postshield relative to shield basalts support the possibility of a greater role for pyroxenite source in postshield basalts. The signature of the transition from shield to postshield in lithium isotopes remains ambiguous, but preferential partial melting of clinopyroxene accompanied by smaller melt fractions would lead to lighter 66the lithium isotopic signatures of postshield basalts.2.4.4 Is MORB-related Lithosphere or Asthenosphere Assimilated into the Hawaiian Plume? The proposed participation of oceanic lithosphere and crust in Hawaiian volcanism is a topic of continued discussion (Chen and Frey, 1985; Gaffney et al., 2004; Frey et al., 2005; Fekiacova et al., 2007; Garcia et al., 2010; Hanano et al., 2010). In Nd and Pb isotopes, most rejuvenated basalts overlap with East Pacific Rise basalts (Figures 2.8b, 2.9b), although binary mixing trends among Hawaiian basalts do not intersect MORB fields (Abouchami et al., 2000; Fekiacova et al., 2007; Hanano et al., 2010). There are several lines of evidence indicating that young oceanic crust and lithosphere are not melted to form Hawaiian basalts. First, rejuvenated basalts do not overlap with MORB in Sr isotopes (Figure 2.9a). Second, ODP Site 843 basalts, which are more representative of low-temperature altered oceanic crust that may be incorporated and melted in Hawai‘i than fresh MORB, do not overlap with the Hawaiian shield or postshield data in Pb, Sr, or Li isotopes (Figures 2.7, 2.9), and do not follow the same trend in Nd isotopes (Figure 2.8). Although a detailed analysis of the sources of rejuvenated volcanism is beyond the scope of this chapter, lithium isotopes indicate that there is not a significant proportion of “young” oceanic crust and/or lithosphere in the source material of any stage of Hawaiian volcanism.2.4.5 Global Lithium Isotope Systematics In a plot of 206Pb/204Pb versus δ7Li (Figure 2.11a), there is a general positive trend for all oceanic island basalts. The continuity of this trend among data from various sources and for oceanic islands from all oceans (and versus strontium isotopes in Figure 2.11b) indicates mantle-wide coherent behavior of lithium isotopes. Although not as constrained as with radiogenic isotopes, data frequency is now sufficient to identify and classify the lithium isotopic signatures of traditional mantle reservoirs. The mantle end-members EM-I, EM-II, HIMU, and PREMA (Zindler and Hart, 1986; Hofmann, 2003) show defined fields (Figure 670.7020.7030.7040.70587Sr/86SrHIMUPREMAEM-II (high 87Sr/86Sr)EM-IMORB0 2 4 6 8b.206Pb/204Pb17.519.020.522.00 2 4 6 8HIMUEM-IMORBEM-IIPREMAa.c.   δ7LiAll OIBsN=2100 2 4 6 8HIMUEM-IEM-IIPREMA10403020Mauna LoaMauna Loa*Ko‘olau*HualālaiWest Ka‘enaKaua‘iMauna Kea*KohalaKīlaueaMauna KeaKohalaHualālaiKaua‘iKa‘ulaWest Ka‘enaLō‘ihiCook-Austral SCAzoresEaster SCFoundation SCJuan FernandezMarquesasPitcairnRéunionSocietySt HelenaLiterature DataHawai‘iFigure 2.11: A compilation of oceanic island basalts with lithium isotopic data (a) 206Pb/204Pb and (b) 87Sr/86Sr and a histogram of all OIB lithium isotopic data (c). The MORB field is compiled data from Moriguti and Nakamura, 1998; Elliott et al., 2006; Tomascak et al., 2008; Nishio et al., 2007; and Chan et al., 2009. Symbols for Hawaiian data from this study and Chan and Frey (2003) are the same as in previous diagrams. SC is shorthand for seamount chain. Stars represent extreme compositions of the mantle end-members rather than the entire range. The OIB histogram (c) includes all compiled Hawaiian data from this study. The y-axis on this plot indicates the number of samples within 0.5‰ intervals of δ7Li variation. In general, HIMU has lithium isotopic signature in the range of δ7Li ≈ 2.5-8.5‰, EM-II from δ7Li ≈ 2.5-6.0‰, and EM-I from δ7Li ≈ 0.5-4.5‰.  The classification of compiled samples into traditional mantle end-member groups shown in this diagram (HIMU, EM-I, EM-II, PREMA) is a general sorting only as many of the compiled samples are classified as a mix of two end-members. Oceanic island basalt data is from Weaver et al., 1987; Albarède and Tamagnan, 1988; Ryan, 1989; Chauvel et al., 1997; Kogiso et al., 1997; Lassiter et al., 2003; Ryan and Kyle, 2004; Nishio et al., 2005; Chan et al., 2009; Vlastélic et al., 2009; and Krientiz et al., 2012 (see their supplemetary information). 682.11a, b, c). PREMA is the focal point of variations with δ7Li≈4.5‰, and this composition also corresponds to the Kea end-member (see Nobre-Silva et al., 2013a for a discussion of a Kea-like prevalent mantle). HIMU is characterized by lithium isotopic signature in the range of δ7Li ≈ 2.5-8.5‰, EM-II from δ7Li ≈ 2.5-6.0‰, and EM-I from δ7Li ≈ 0.5-4.5‰ (Figure 2.11c). Krienitz et al. (2012) also found that large-scale OIB mantle reservoirs (HIMU, EMI, EMII; Zindler and Hart, 1986) are characterized by distinct lithium isotopic signatures. Our lithium isotopic signatures of mantle end-members are within the range described by Krienitz et al. (2012) with the exception that Hualālai extends the EM-I mantle source down to δ7Li≈0.5‰ and the HIMU mantle source shows a greater range due to the inclusion of lithium isotopic data from the Cook-Austral seamount chain (Chan et al., 2009; Vlastélic et al. 2009).2.5. CONCLUSION The application of lithium isotopes to Hawai‘i provides an additional perspective into the source components of the different stages of Hawaiian volcanism that is not offered by radiogenic isotopes. Our study of Hawaiian basalts using this isotopic system has led to the following interpretations.1. Correlations of lithium isotopes with Pb, Hf, Nd, and Sr isotopes and trace elements indicate that lithium can be used as a source tracer if the data is filtered to remove the effects of fractional crystallization of clinopyroxene and any post-magmatic alteration.2.  Lithium isotopic signatures of 89 Hawaiian basalts extends the range observed to date on Hawai‘i in this isotopic system, with variations from δ7Li=0.8-5.7‰. This is the first study to identify a difference between the Kea and Loa geochemical trends in lithium isotopes, reflecting the presence of different source components for each of these trends.3. Postshield basalts are generally lighter than shield stage basalts in lithium isotopes, a 69result of smaller degrees of partial melting of a clinopyroxene-rich source, which has a lighter lithium isotopic signature. This suggests that pyroxenite heterogeneities in the Hawaiian mantle plume exist as vertical stringers rather than discrete blobs.4. The lithium isotopic signature near PREMA of Kea trend basalts suggests that the source of this component is similar to normal peridotitic mantle.5. The enriched end-member in radiogenic isotopic systems and trace elements is Makapu‘u Ko‘olau, whereas the light end-member in lithium isotopic space is Hualālai. Heavy lithium isotopic signature of Ko‘olau may be explained by incorporation of between 1-5% ancient carbonate in the source.6.  The cause of the heterogeneity observed in lithium isotopes for Loa trend volcanoes is not directly resolvable. This is due to the extreme variation and overlap in the lithium isotopic signature of different possible types of recycled material (i.e., marine sediments, oceanic crust, lower continental crust) as well as uncertainties in how uniformly processes such as subduction de-watering and mantle mixing of subducted material proceeds. The light lithium isotopic signature observed in Hualālai postshield samples may arise from a mix of altered oceanic crust and subduction eroded lower continental crust.70Chapter 3 The Link Between Hawaiian Mantle Plume Composition, Magmatic Flux, and Deep Mantle GeodynamicsChapter 3 The Link Between Hawaiian Mantle Plume Composition, Magmatic Flux, and Deep Mantle Geodynamics3.1. INTRODUCTION Long-lived intraplate oceanic islands and chains such as Iceland, Kerguelen, and Hawai‘i are typically explained as the products of mantle plumes (Morgan, 1972; Boschi et al., 2007). Hawai‘i is one of the most studied examples of such volcanism because it has a well-documented age progression along the chain (O’Connor et al., 2013; Garcia et al., 2015), is far from continent and mid-ocean ridge sources of contamination, has a deep mantle source (French and Romanowicz, 2015), and, on the main islands, exhibits two geographical trends, Kea and Loa, with distinct geochemical signatures (Tatsumoto, 1978; Abouchami et al., 2005; Tanaka et al., 2008; Weis et al., 2011). Hawai‘i, the archetypal mantle plume, is anomalous in many of its dynamic and geochemical features. For example, intraplate basalts worldwide are dominantly alkalic in composition, reflecting lower degrees of partial melting, in comparison to Hawai‘i where ~98% of eruptive products are of tholeiitic composition (Garcia et al., 2015). In addition, mantle plumes typically exhibit higher magmatic production at their initial arrival at the base of the lithosphere that diminishes with time, i.e., from melting the plume head to the plume tail (White, 1993; Jellinek and Manga, 2004; Kumagi et al., 2008). The volume flux of Hawaiian eruptive products, conversely, has increased ~650% along the Northwest Hawaiian Ridge (NWHR) and an additional ~375% during the formation of the Hawaiian Islands to an high at the current locus of volcanism 71centered on Kīlauea and Lō‘ihi volcanoes (Wessel, 2016). At the same time, mantle potential temperature has increased (Tree, 2016), along with the volcano propagation rate (i.e., the frequency that a new volcano is produced as the Pacific plate moves over the Hawaiian mantle plume; O’Connor et al., 2013). Thus, on the basis of most major physical parameters, the Hawaiian mantle plume has strengthened, an unexplained observation that is an unusual feature among mantle plumes worldwide (Tree, 2016). No modern geochemical study has addressed this issue or determined the isotopic variation along the NWHR, to evaluate whether source changes have played a role. The source of the Loa component in Hawaiian basalts is more enriched (higher Th/U, 208Pb*/206Pb*, and Sr isotopic ratios; lower Nd and Hf isotopic ratios) than that of the Kea component. This signature is predominately explained by the presence of subduction recycled material (Weis et al., 2011) or primordial material in the deep source (Williams et al., 2015). Lead isotopes provide the most robust tool for delineating the Loa and Kea components in Hawaiian basalts: at a given 206Pb/204Pb, Loa trend volcanoes will have a higher 208Pb/204Pb than Kea trend volcanoes (Abouchami et al., 2005; Weis et al., 2011). Emperor Seamount basalts (~81-51 Ma) are Kea-like in geochemical affinity, whereas both Kea and Loa compositions are present in basalts from the Hawaiian Islands, and clearly distributed along two geographical trends of volcanoes (Keller et al., 2000; Regelous et al., 2003; Abouchami et al., 2005; Tanaka et al., 2008; Weis et al., 2011). On the basis of the only three Pb isotope analyses previously available for the NWHR, Loa geochemical compositions have been observed at only one seamount, Daikakuji, near the bend in the Hawaiian-Emperor chain (Regelous et al., 2003). It was unknown whether this is an isolated case or represents the first arrival of the Loa geochemical component (Garcia et al., 2015). Because the Loa composition typically is associated with a more fusible source component, its presence may be linked to the increase in volcano volume (lowest at Daikakuji with 30 km3 to highest at Gardner with 540 km3; Bargar and Jackson, 1974), magmatic flux 72(0.4 m3/s at Daikakuji to a maximum of 4 m3/s at Gardner; Wessel, 2016), and volcanic propagation rate (Detroit to Midway 57 ± 2 km/Ma, from which it increases to 80-100 km/Ma afterwards; O’Connor et al., 2013) observed along the ~2800 km long ridge (Pertermann and Hirschmann, 2003; Garcia et al., 2015).  Lead isotopic compositions were measured on 22 shield-stage samples from 13 NWHR volcanoes spanning from ~47 Ma at the bend to ~7 Ma at Nīhoa (Figure 3.1) (Dalrymple et al., 1974; O’Connor et al., 2013). There were no Pb isotopic measurements for NWHR volcanoes younger than Daikakuji Seamount; this study fills this 40 million year gap. The Pb isotopic evolution of the entire Hawaiian-Emperor chain was examined to identify when the enriched Loa component appears and to assess its involvement in magmatic flux 0 km 500 km175˚175˚180˚180˚−175˚−175˚−170˚−170˚−165˚−165˚−160˚−160˚−155˚20˚25˚ 25˚20˚30˚−6000 −4000 −2000 0Bathymetry (m)GardnerFrench FrigateYuryakuDiakakujiBrooksAcademicianBergE. NorthamptonMaro ReefLaysanMidwayKaua‘iNīhoaW. NīhoaMokumanamana (Necker)AbbottColahan HancockUnnamedPearl and HermesNeroHelsleyLaddHawai‘iKo‘olauLō‘ihiMaunaKeaMaunaLoaMiddle BankTwin BanksKeoeaRiataPioneerTownsendCromwell47 Ma141 Ma132 Ma127 Ma1N20 Ma225 Ma112 Ma411 Ma38 Ma3<5 Ma5 170˚170˚180˚180˚−170˚−170˚−160˚−160˚20˚ 20˚30˚ 30˚40˚ 40˚50˚ 50˚500 kmEmperor SeamountsMendocino FZMurray FZMolokai FZBendNW Hawaiian RidgeHawaiian IslandsFigure 3.1: Bathymetric map of the ~51 seamounts and islands of the Northwest Hawaiian Ridge and sample locations.Bathymetry is 2-minute Gridded Global Relief Data ETOPO2v2 satellite altimetry dataset (U.S. Department of Commerce, National Oceanic and Atmospheric Administration, National Geophysical Data Center, 2006; downloaded March 14, 2014) and new multibeam bathymetry (Smith et al., 2014). White circles show sample locations. References for ages are in yellow superscript and are as follows: 1 – O’Connor et al., 2013; 2 – Dalrymple et al., 1974; 3 – Dalrymple et al., 1981; 4 – Garcia et al., 1987; 5 – Garcia et al., 2010. Inset figure is a traverse Mercator projection of the Hawaiian-Emperor chain modified from Garcia et al, 2015. Dashed lines in this figure indicate major Pacific fracture zones (FZ).73variations. Finally, we discuss the implications of these results for the Hawaiian deep mantle source and the potential contribution of material from the Pacific LLSVP to account for flux variations.3.2. GEOLOGICAL SETTING AND SAMPLING Fifty-one volcanoes erupted over ~42 million years between the bend in the Hawaiian-Emperor chain and the Hawaiian Islands constitute the Northwest Hawaiian Ridge (NWHR; Garcia et al., 2015). We focus on shield-stage tholeiitic basalts as they are most likely to record the plume source composition variation (i.e., Loa and Kea geochemical variation), whereas the later postshield and rejuvenated Hawaiian basalts are more uniform and present depleted isotopic compositions regardless of the geochemical affinity of shield-stage basalts at the same volcano (Frey et al., 2005; Hanano et al., 2010). Study of the Hawaiian Islands shows that single islands are composed of several volcanic centers that may erupt different geochemical signatures from as close as ~50 km apart (e.g., Mauna Loa and Mauna Kea on Hawai‘i; Abouchami et al., 2005). One to four samples from selected volcanoes were available for this large-scale study, which provides average isotopic trends along a previously unstudied section of the Hawaiian-Emperor chain. Sampling of the NWHR is currently highly restricted by the Papahānaumokuākea Marine National Monument, a recently expanded ~1,508,870 km2 U.S. Marine Conservation Area and UNESCO World Heritage Site that stretches from Middle Bank to Hancock Seamount (~80% of the NWHR). Because of this access impediment, samples in this study originate from pre-2006 dredge or island sampling expeditions stored at the University of Hawai‘i and Scripps Institution of Oceanography rock collections or from samples collected by the University of Hawai‘i’s human occupied vehicle (HOV) submersible Pisces V reconnaissance survey (2003 and 2007) of the new monument. Table 3.1 summarizes sample locations. Samples from near the bend of the Hawaiian-Emperor chain (Yuryaku and Daikakuji) 74were dredged by the research vessel Thomas Washington in August of 1971 during the Scripps Institution of Oceanography expedition Aries VII. This expedition dredged Koko, Kimmei, Yuryaku and Daikakuji Seamounts. Published data from this cruise includes ages, some major and trace element data, and isotopic compositions (Davies et al., 1971; Clague et al., 1973; Clague, 1974; Clague et al., 1975; Dalrymple and Clague, 1976; Lanphere et al., 1980; Clague and Dalrymple, 1987; Regelous et al., 2003; Sharp and Clague, 2006). Multiple expeditions by the Hawai‘i Institute of Geophysics’ research vessel Ka‘imikai-O-Kanaloa dredge sampled the NWHR from 1972-1984. These samples are Table 3.1: Sample locations and collection methodsA-53-1 Yuryaku b dr 1971 ARES07WT 172.20 32.71 a,b,cA-55-1 Daikakuji b dr 1971 ARES07WT 172.28 32.12 a,b,cA-55-2 Daikakuji b dr 1971 ARES07WT 172.28 32.12 a,b,cA-55-4 Daikakuji b dr 1971 ARES07WT 172.28 32.12 a,b,c84-28E Unnamed b dr 1984 KK840428-5 180.72 29.67 d84-30G Academician Berg b dr 1984 KK840428-5 181.13 28.89 dR-1316 Midway b drill 1965 - 182.63 28.28 e,f,gR-1295.6 Midway b drill 1965 - 182.63 28.28 e,f,gP5-524-42 Pioneer b s 2003 KOK0319 186.56 25.80 h KK-76-5-4DD E. Northampton b dr 1976 KK760806-1 187.98 25.30 d76-5-4A E. Northampton b dr 1976 KK760806-1 187.98 25.30 dP5-530-3 Laysan b s 2003 KOK0319 188.63 25.66 h 76-6-7F Gardner b dr 1984 KK761108 192.30 25.68 i84-40E Gardner b dr 1984 KK840428-5 192.24 25.91 d84-39B Gardner b dr 1984 KK840428-5 192.28 25.93 dNEC-3A Mokumanamana b o 1985 ALCYONE k72-49A Keoea b dr 1972 KK720702-2 196.44 23.35 dP5-688-1 Twin Banks b s 2007 KOK0714 196.85 23.07 h 76-9-11 West Nīhoa T-b dr 1976 KK761108 197.52 23.11 iNIH-D4 N hoa b o 1985 ALCYONE jNIH-W-11-1 N hoa P-b o 1985 ALCYONE jNIH-F9 N hoa b o 1985 ALCYONE jNIH-D-1-2 N hoa b o 1985 ALCYONE jno recordno record1rock type: b=basalt, T-b=transitional basalt, P-b=picro-basalt2collection methods: dr= dredge, drill=drillcore, s=submersible, o=subaerial outcrop3References: a - Davies et al., 1971; b - Clague et al., 1975; c - Dalrymple and Clague, 1976; d - Hawaiian Institute of  Geophysics, U. of Hawai‘i; e - Ladd et al., 1967; f - Dalrymple et al., 1974; g - Dalrymple et al., 1977; h - Hawai‘i Undersea  Research Laboratory; i - Garcia et al., 1987; j - Scripps Institution of Oceanography Long.     (°W)Lat.  (°N) Ref.3no recordno recordno recordSample Seamount Rock Type1Collection Method2Year CollectedResearch Cruise75identified by the prefix KK- followed by the last two digits of the year of the expedition, the dredge number, and the sample letter. Some of these samples have published XRF major and trace elemental analyses (Garcia et al., 1987, 2015). Recently in 2003 and 2007, the Ka‘imikai-O-Kanaloa has been used for several NWHR sampling expeditions to Pioneer, Laysan, and Twin Banks by the Hawai‘i Undersea Research Laboratory (HURL). These samples are identified by the prefix P5- and were collected using the submersible Pisces V.   Two sample cores were drilled at Midway Island, one on Sand Island (Sand Island hole) and one in the lagoon of Midway Atoll (reef hole) in 1965 by the Hawai‘i Institute of Geophysics at the University of Hawai‘i in partnership with the U.S. Geological Survey and the U.S. Office of Naval Research (Ladd et al., 1967, 1970; Macdonald, 1969). Samples analyzed in this study are from the reef hole, which drilled 1,261 feet (~384 meters) of limestone, dolomite, marl, and volcaniclastic sedimentary rock before penetrating approximately 387 feet (~118 meters) of tholeiitic basalt (Ladd et al., 1967, 1970; Macdonald, 1969).  Finally, subaerial samples from Mokumananana (formerly known as Necker Island) and Nīhoa Island were collected by Harmon Craig during the 1985 ALCYONE expedition by Scripps Institution of Oceanography. This same cruise also dredged Nīhoa, recovering manganese crust, volcaniclastic pebblestone, and volcanic rock. Strontium and neodymium isotopes on unleached powders from some of these Nīhoa and Mokumananana samples were previously reported (Clague, 1973; Basu and Faggart, 1996). Some of these samples were reanalyzed in this study.3.3. ANALYTICAL TECHNIQUES AND AGE CORRECTION Lead isotopic compositions of 22 NWHR basalts were analyzed at the Pacific Centre for Isotopic and Geochemical Research (PCIGR) at the University of British Columbia. In addition to isotopic analyses, major elements were analyzed at the University of Massachusetts Amherst and this data is reported in Garcia et al. (2015). 76 Step-wise acid leaching is a necessary procedure to remove possible post-eruption alteration and/or contamination (Weis et al., 2005, 2006; Nobre Silva et al., 2009, 2010). For isotopic analyses, whole rock powders were sequentially leached in 6N HCl in steps of 20 minutes of acid-leaching in an ultrasonic bath followed by immediate decantation before fine particles had time to settle. This leaching step was repeated until the supernatant was clear, typically between 6 and 14 times. After leaching, samples were digested using HF and HNO3 and purified on chromatographic ion exchange columns following the procedures of Fourny et al. (2016). Lead isotopic ratios were determined on a Nu Plasma (Nu Instruments) MC-ICP-MS NP II 214 or Nu 1700 that were cross-calibrated by multiple analysis of standard NIST SRM 981 and reference materials of comparable composition (Kil-93 and BHVO-2). A Tl spike was added to monitor internal mass fractionation (White et al., 2000) and all analyses were corrected on-line to a 205Tl/203Tl=2.3885 (Weis et al., 2005). Analyses were then normalized off-line to NIST SRM 981 triple-spike values of Galer and Abouchami (1998) using the sample-standard bracketing method. For the duration of this study, the measured ratios for the NIST SRM 981 were: 208Pb/204Pb=36.7188 ± 0.0107, 207Pb/204Pb=15.4983 ± 0.0041, and 206Pb/204Pb=16.9418 ± 0.0028 (2SD, n=102). On a daily basis, the standard errors on the analyses of the NIST SRM 981 were smaller, in the range of ~100 ppm. Mercury 204 was monitored for possible interference on mass 204Pb by analysis of the signal on the 202 mass. Total procedural blanks contained 3-88 picograms Pb, negligible compared to the amount of Pb present in samples (on average ~100-300 nanograms Pb). Repeated analyses of U.S Geological Survey (USGS) reference material BHVO-2, a Hawaiian basalt, resulted in average Pb isotopic compositions of 208Pb/204Pb=38.2091 ± 0.0107, 207Pb/204Pb=15.4893 ± 0.0023, and 206Pb/204Pb=18.6482 ± 0.0001 (2SD, n=3), comparable to recent published values (Weis et al., 2005; Fourny et al., 2016). In-house PCIGR reference material Kil-93, a Hawaiian basalt from the 1993 Kīlauea eruption, resulted in average Pb isotopic compositions of 208Pb/204Pb=38.0679 ± 0.0038, 207Pb/204Pb=15.4741 ± 770.0017, and 206Pb/204Pb=18.4090 ± 0.0035 (2SD, n=4), also comparable to recent published values (Nobre Silva et al., 2013b). All data compiled from the literature are normalized to NIST SRM 981 standard values used in this study to ensure comparability: 16.9405 for 206Pb/204Pb, 15.4963 for 207Pb/204Pb, and 36.7219 for 208Pb/204Pb (Weis et al., 2011). Trace element concentrations (e.g., Pb, Th, U, La, Yb) of samples were measured at the PCIGR using a Thermo Finnigan Element2 high-resolution ICP-MS (Schudel et al., 2015; Fourny et al., 2016). Powdered samples were digested in a mixture of 10:1 HNO3:HF with no prior leaching. After digestion, samples were fluxed in 6N HCl to eliminate insoluble fluorides, dried down, and diluted 5000 times in a 1% HNO3-0.05% HF-1 ppb In solution, and analyzed on the Element2 using the USGS basaltic standard BCR-2 for external calibration (Schudel et al., 2015; Fourny et al., 2016). BHVO-2 was analyzed with all samples, and resulting values were within two standard deviations of published values for all elements (Schudel et al., 2015; Fourny et al., 2016). Full procedural duplicates and replicate measurements showed excellent agreement, and most elements were analyzed with one relative standard deviation (RSD) of less than 5%. Radiogenic Pb, a powerful discriminator of the Loa and Kea trends (in this chapter, mantle radiogenic Pb refers to 208Pb*/206Pb*) is a measure of the time integrated Th/U of the source throughout Earth history (Tatsumoto et al., 1973; Galer and O’Nions, 1985; Hofmann, 1997). It is calculated for each sample based on equation 3.1 below where the initial Pb isotopic values are taken as the measured isotopic compositions of the Canyon Diablo troilite, assumed to be representative of the initial isotopic compositions of the Earth.  208Pb*208Pb204Pb208Pb204Pbsamplesample initialwhereinitial initialinitial206Pb204Pb206Pb204Pb206Pb204Pb208Pb204Pb206Pb* == 29.476 = 9.307__ (3.1)78 Identifying differences in Pb isotopic ratios for rocks that are older than a few million years is contingent upon the accurate and precise determination of initial isotopic ratios of Pb at the time of eruption. Samples in this study span an age range from 8.5 to ~47 Ma (O’Connor et al., 2013; Dalrymple et al., 1974, 1981; Garcia et al., 1987, 2010). Age correction of measured isotope ratios is necessary in the older NWHR samples to account for in situ radiogenic decay of parent to daughter isotopes since eruption. This correction is very small for young samples, however, for samples older than ~15-20 Ma age corrections of measured isotopic ratios are outside the range of analytical error. The accuracy of this correction is also affected by the fact that both subaerial weathering while the volcano was above sea level and seawater alteration once the volcano subsided below sea level might have altered many of these samples, disturbing the magmatic elemental concentrations and ratios. All NWHR volcanoes older than Midway Island are currently below sea level. Loss-on-ignition of analyzed NWHR rocks ranges from less than 1 to 5 weight percent (Garcia et al., 2015), not as altered as some samples analyzed from the Emperor Seamounts (Regelous et al., 2003), but greater than what is considered fresh for a Hawaiian basalt (loss on ignition <3 weight percent; Garcia et al., 2015). In the U-Th-Pb decay systems, U and Pb are the most sensitive to secondary alteration and, as a result, the elemental concentrations measured in unleached powders are not always representative of the primary magmatic concentrations. The degree to which samples have been compromised by alteration can be assessed using element ratios with limited variation in oceanic island basalts (OIB). Useful ratios for the U-Th-Pb systems include Ce/Pb (25 ± 5), Nb/U (47 ± 10), and Th/U (4 ± 0.5; Hofmann et al., 1986; O’Nions and McKenzie, 1993; Nobre Silva et al., 2013b). Here, we use the Th/U of samples to estimate primary U concentrations in unleached sample powders following the method of Nobre Silva et al. (2013b) that considers Th to be immobile during basaltic weathering and uses the Th/U to assess U loss or gain. The average Th/U of Hawaiian shield basalts is 79~3 (n=294), not significantly different from NWHR basalts or OIB globally (O’Nions and McKenzie, 1993; Nobre Silva et al., 2013b). Correcting measured U concentration values back to Th/U = 3 (assuming only U has been gained or removed by alteration) provide an estimate of magmatic U concentration and a better approximation of initial Pb isotopic ratios. In addition to using this method, we also re-corrected Emperor Seamount literature data using this technique for overall comparability of the dataset. Previous researchers addressed secondary alteration elemental mobility by measuring concentrations of leached sample powders for age correction purposes (Regelous et al., 2003; Huang et al., 2005a; Shafer et al., 2005). However, this method assumes that secondary alteration was a closed system and did not fractionate elements between the primary and secondary phases (Staudigel et al., 1996). More recently, it has been shown that acid leaching differentially removes some elements (Hanano et al., 2009; Nobre Silva et al., 2009, 2010). The impact of elemental mobility during alteration is particularly well-displayed by the oldest sampled Emperor Seamount, Meiji (>81 Ma), which is significantly altered and has proven problematic for accurate age correction (Regelous et al., 2003; Frey et al., 2005; Huang et al., 2005a). Using the Th/U estimation method described here to age correct previous data provides age corrected Pb isotopic compositions that are more coherent than previously reported for Emperor Seamount data. To illustrate, radiogenic Pb of Meiji ranges from 208Pb*/206Pb*=0.88-0.92 after applying the correction method compared to unrealistically high values of 0.96-1.01 in published data (Regelous et al., 2003). Average 208Pb*/206Pb* of the most enriched Hawaiian volcano Ko‘olau is 0.96 ± 0.02 (n=36) and East Pacific Rise MORB is 0.92 ± 0.03 (n=146). Because Meiji erupted under the thinnest lithosphere of any Hawaiian volcano with high degrees of melting near a mid-ocean ridge, it is highly unlikely that basalts have such highly radiogenic Pb isotopic compositions, overlapping with Ko‘olau Makapu‘u compositions that are only observed on O‘ahu (Keller et al., 2000; Regelous et al., 2003; Frey et al., 2005). The use of measured elemental concentrations on leached sample powders of such altered rocks tends 8015.4015.4415.4815.5215.5617.40 17.90 18.40 18.90207Pb/204Pb206Pb/204PbHawaiian IslandsEmperor SeamountsMeijiDetroit SuikoOjinKokoYuryakuTwin BanksWest NīhoaNW Hawaiian RidgeYuryakuDaikakujiUnnamedAcademician BergMidwayPioneerNorthamptonGardnerKeoeaNīhoaMokumanamanaMauna KeaMauna LoaKo‘olauLō‘ihioldestyounger EPR MORB Figure 3.2: 207Pb/204Pb versus 206Pb/204Pb of Northwest Hawaiian Ridge shield-stage basalts(error bars are smaller than data points for MC-ICP-MS data). Plot of 206Pb/204Pb versus 207Pb/204Pb for NWHR basalts, Emperor Seamounts, EPR MORB glasses, and Ko‘olau, Mauna Loa, and Mauna Kea volcanoes. The black line is a regression line of all NWHR, Emperor Seamount, Mauna Kea, Mauna Loa, Lō‘ihi, and Ko‘olau data (r2=0.59, increases to r2=0.60 if the Emperor Seamounts are excluded). The variation in 207Pb/204Pb in NWHR basalts is limited and generally a similar range for the Emperor Seamounts, NWHR, and Hawaiian Islands. This suggests a common plume source for all of these sections of the Hawaiian-Emperor chain, and that the leaching and age correction procedure used here works to recover primary Pb isotopic ratios. Data sources are as follows: Emperor Seamounts: Keller et al., 2000; Huang et al., 2005a; Regelous et al., 2003; Lō‘ihi: Abouchami et al., 2005; Kīlauea: Pietruszka and Garcia, 1999; Abouchami et al., 2005; Marske et al., 2007; Mauna Loa: Kurz and Kammer, 1991; Kurz et al., 1995; Abouchami et al., 2000; Wanless et al., 2006; Marske et al., 2007; Weis et al., 2011; Hualālai: Yamasaki et al., 2009; Mauna Kea: Eisele et al., 2003; Rhodes and Vollinger, 2004; Nobre Silva et al., 2013a; Kohala: Abouchami et al., 2005; Māhukona: Huang et al., 2009; Garcia et al., 2012; Hāna Ridge: Ren et al., 2006; West Maui: Gaffney et al., 2004; Kaho‘olawe: Abouchami et al., 2005; Huang et al., 2005b; Lāna‘i: Abouchami et al., 2005; Gaffney et al., 2005; East Moloka‘i: Xu et al., 2005; West Moloka‘i: Xu et al., 2007; Penguin Bank: Xu et al., 2014; Ko‘olau: Tanaka et al., 2002, 2008; Fekiacova et al., 2007; Wai‘anae: Coombs et al., 2004; Van der Zander et al., 2010; West Ka‘ena: Greene et al., 2010; Kaua‘i: Mukhopadhyay et al., 2003; Garcia et al., 2010; Cousens and Clague, 2015. EPR MORB Glasses were downloaded from PetDB on November 8, 2012 using the search criteria of East Pacific Rise fresh glasses (www.earthchem.org/petdb). All literature data in this and following figures is normalized to the same NIST SRM 981 values to ensure comparability.81Table 3.2: Pb isotopic composition and Pb, U, and Th concentrations of Northwest Hawaiian Ridge basalts208Pb/204Pb 2 SE 207Pb/204Pb 2 SE 206Pb/204Pb 2 SE 208Pb/204Pb 207Pb/204Pb 206Pb/204Pb 208Pb*/206Pb*A-53-1 Yuryaku 3541 0.4 46.9 1 3.17 38.0527 21 15.4718 8 18.5140 8 0.74 0.71 0.314 2.27 37.904 15.465 18.363 0.9306A-55-1 Daikakuji 3520 0.4 46.8 1 6.86 37.9374 22 15.4244 8 17.9992 7 1.53 1.31 0.436 3.02 37.809 15.418 17.869 0.9732A-55-2 Daikakuji 3520 0.4 46.8 1 8.53 37.9977 26 15.4372 11 18.0915 12 2.22 2.18 0.672 3.24 37.850 15.430 17.942 0.9698A-55-4 Daikakuji 3520 0.4 46.8 1 8.42 38.0094 23 15.4402 14 18.0972 11 2.16 2.22 0.689 3.22 37.855 15.433 17.941 0.970584-28E Unnamed 2801 0.6 31.7 6 5.38 38.2106 27 15.4958 8 18.7045 10 1.70 1.46 0.474 3.08 38.119 15.492 18.612 0.928984-30G Academician Berg 2608 1.3 31.0 6 5.41 38.2009 27 15.4923 9 18.7037 11 1.24 1.42 0.438 3.25 38.079 15.487 18.580 0.9277R-1316 Midway 2447 1.8 27.6 1 5.09 38.2085 29 15.4890 8 18.7484 16 1.06 1.08 0.368 2.93 38.117 15.485 18.656 0.9243R-1295.6 Midway 2447 1.8 27.6 1 4.82 38.2173 22 15.4890 8 18.7601 11 1.03 1.04 0.375 2.77 38.127 15.485 18.669 0.9241P5-524-42 Pioneer 1998 3.4 28.0 7 4.81 38.2475 20 15.4949 5 18.7008 6 2.63 1.20 0.817 1.47 38.206 15.493 18.659 0.933576-5-4DD Northampton 1846 3.6 26.6 3 3.81 38.2028 19 15.4818 7 18.6307 8 0.83 0.83 0.275 3.01 38.126 15.478 18.553 0.935576-5-4A Northampton 1846 3.6 26.6 3 3.50 38.2039 23 15.4813 8 18.6348 8 0.84 0.77 0.277 2.79 38.133 15.478 18.564 0.935376-6-7F Gardner 1449 5.8 12.3 4 3.12 38.1315 68 15.4785 28 18.5903 30 0.61 0.53 0.196 2.69 38.097 15.477 18.555 0.932284-40E Gardner 1449 5.8 12.3 4 3.04 38.0246 29 15.4690 12 18.4651 13 0.75 0.64 0.264 2.44 37.991 15.467 18.431 0.933284-39B Gardner 1449 5.8 12.3 4 3.30 38.1808 33 15.4858 14 18.5857 13 1.91 0.93 0.287 3.24 38.161 15.485 18.566 0.9380NEC-3A Mokumanamana 1080 3.1 10.0 2 2.79 37.9611 28 15.4714 10 18.3017 11 1.31 0.91 0.284 3.22 37.934 15.470 18.274 0.943272-49A Keoea 963 2.0 11.0 7 3.49 38.0024 23 15.4754 8 18.4213 9 0.98 0.85 1.041 0.81 37.972 15.474 18.390 0.9353P5-688-1 Twin Banks 920 2.1 9.6 4 3.59 38.0459 28 15.4792 10 18.4813 9 0.71 0.55 0.854 0.64 38.024 15.478 18.459 0.934076-9-11 West N hoa 825 2.6 8.8 1 6.23 37.9402 25 15.4724 7 18.1919 7 66.2 b 1.38 0.694 1.98 37.940 15.472 18.191 0.9527NIH-D4 N hoa 794 2.6 7.0 2 3.88 38.0444 19 15.4789 7 18.4540 10 0.78 0.70 0.211 3.32 38.020 15.478 18.429 0.9366NIH-W-11-1 N hoa 794 2.6 7.0 2 5.72 38.1702 27 15.4827 9 18.6103 9 0.78 0.74 0.246 3.02 38.144 15.481 18.584 0.9344NIH-F9 N hoa 794 2.6 7.0 2 4.69 38.1109 24 15.4744 8 18.5338 9 0.74 0.73 0.153 4.76 38.084 15.473 18.507 0.9357NIH-D-1-2 N hoa 794 2.6 7.0 2 8.17 37.9256 27 15.4752 9 18.2092 10 2.53 1.16 0.429 2.70 37.913 15.475 18.200 0.94881-8References to ages and melt flux: 1- O'Connor et al., 2013; 2- Dalrymple et al., 1974; 3- Dalrymple et al., 1981; 4- Garcia et al., 1987; 5- Garcia et al., 2010; 6- Sharp and Clague, 2006; 7- Age inferred by interpolating between existing ages and assuming a linear age progression of NWHR volcanoes; 8- Wessel, 2016. aTrace element concentrations (in ppm) were measured on unleached whole rock powders using a Thermo Finnigan Element2 high resolution ICP-MS at the PCIGR at UBC. USGS Reference Material BCR-2 was used for external calibration and a 1 ppb In spike was used as an internal standard. RSDs are less than 5% for all elements and measured concentrations of USGS Standard BHVO-2 are within 2SD of published values.  bThis sample was run in duplicate. The high Pb appears to come from the rock itself, likely contained in secondary minerals that are removed by acid leaching before isotopic analysis. This is supported by Pb isotopic values from leached whole rock powder that plot within the field for Hawaiian shield basalts. cPb isotopes were measured on acid leached whole rock powders (after being purified twice on ion exchange columns) using a Nu Plasma II multicollector ICP-MS at the PCIGR at UBC. Pb data were corrected for fractionation by a Tl spike and 2 SE  error is the absolute error values of the individual sample analysis (internal error) reported to the significant digit. Pb standard NIST SRM-981 values (n=102) during the course of this study were 208Pb/204Pb=36.7188 ± 0.0107, 207Pb/204Pb=15.4983 ± 0.0041, and 206Pb/204Pb=16.9418 ± 0.0028.dTh/U ratio is corrected to a Th/U value of 3 (average Th/U of the mantle, all OIB, and average of Hawaiian Island shield tholeiites) based on the assumption of U mobility only. These adjusted U concentrations are then used to age correct measured Pb isotopic ratios avoiding major complications due to secondary alteration.U and Age Corrected Isotopic Ratios dTh/UdSample SeamountDistance from K lauea (km)Estimated Vol. Flux (m 3/s) 8 Age (Ma)Age Ref.La/YbPMaMeasured RatioscPbICP (ppm)aTh ICP (ppm)aUICP (ppm)a82to over-correct measured isotopic ratios and is not an adequate solution to correct for in situ decay of radioactive parents (Frey et al., 2005). Finally, the limited range of 207Pb/204Pb of NWHR basalts that overlaps with non-age corrected young Hawaiian Island data supports the of effectiveness of the acid leaching and age correction method to recover accurate initial isotopic ratios of Hawaiian basalts (Figure 3.2).3.4. RESULTS Extreme Ko‘olau Makapu‘u-type Pb isotopic signatures were found at only one locality along the NWHR; Daikakuji Seamount has the most unradiogenic 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb among NWHR basalts (17.869, 15.418, and 37.809, respectively) (Figure 3.3a; Table 3.2). The most radiogenic lead isotopic compositions are observed for basalts from Midway in 206Pb/204Pb (18.669) and Pioneer Seamount in 207Pb/204Pb (15.493) and 208Pb/204Pb (38.206) (Figure 3.3a). Samples from Daikakuji, Mokumanamana, and West Nīhoa plot on the Loa side of the Loa-Kea divide line in 208Pb/204Pb-206Pb/204Pb isotopic space (Figure 3.3a). After Daikakuji, there are no Loa-like compositions until the youngest section at Mokumanamana and Nīhoa (less than ~11 Ma; Dalrymple et al., 1974). The Pb isotopic compositions of Nīhoa basalts span about two-thirds of the entire range of 208Pb/204Pb in young Hawaiian basalts (37.913 – 38.144) and define a trend that crosses the Loa-Kea divide (Figure 3.3a). This implies the possibility that Nīhoa may be a composite island with two volcanic centers of distinct compositions. The majority of NWHR basalts are distinctly Kea-like in their geochemical affinity (Figure 3.3a; Table 3.2). These basalts span nearly the entire range in 208Pb/206Pb and 207Pb/206Pb observed in Kea-trend shield basalts on the Hawaiian Islands (Figure 3.3b). The variation in Pb isotopic composition can be explained by mixing 40-91% of the Kea end-member component with an isotopically depleted plume source such as the Depleted Makapu‘u (DMK) component, one of the four end-member components identified in Hawaiian basalts (i.e., Kea, EMK, DMK, Lō‘ihi; Tanaka et al., 2002, 2008) (Figure 3.3b). All832.032.052.072.092.112.130.82 0.83 0.84 0.85 0.86 0.87802/bP602bP207Pb/206PbEPR MORBKEAEMKEMKLō‘ihiLō‘ihiDMMDMKDMKLOAKEA37.7037.8037.9038.0038.1038.20802/bP402bPEPR MORB206Pb/204Pb17.6 17.8 18.0 18.2 18.4 18.6 18.804812162024La/YbPMN=2971 2 3 4 5 6 7 8 9NWHRDaikakujiEmp.Smts MKMLHawaiian IslandsEmperor SeamountsMeijiDetroit SuikoOjinKokoYuryakuTwin BanksWest NīhoaNW Hawaiian RidgeYuryakuDaikakujiUnnamedAcademician BergMidwayPioneerNorthamptonGardnerKeoeaNīhoaMokumanamanaMauna KeaMauna LoaKo‘olauLō‘ihiKEAPb 207Pb/206Pb 208Pb/206PbKEA 0.76 0.82 2.04DMK 0.5 0.86 2.09EMK 1.7 0.87 2.13Compositions of mixing end membersoldestyounger   Averageage corr.Averageage corr.(a)(b)Figure 3.3: Pb-Pb isotopic diagrams for Northwest Hawaiian Ridge shield-stage basalts(error bars are smaller than data points). (a) Plot of 208Pb/204Pb versus 206Pb/204Pb and (b) 208Pb/206Pb versus 207Pb/206Pb. For reference the Loa-Kea divide line (Abouchami et al., 2005) is shown in black, and the four Hawaiian end-member components are shown by white circles (KEA, DMK, EMK, and Lō‘ihi; Tanaka et al., 2002, 2008). Mixing lines are calculated between the Kea end-member and enriched Makapu‘u (EMK; blue line) and depleted Makapu‘u (DMK; red line). Tick marks are at intervals of 10%. Inset to figure (a) shows a histogram of primitive mantle-normalized (McDonough and Sun, 1995) (La/Yb)PM of the Emperor Seamounts (orange), Mauna Loa (blue), Mauna Kea (red), the Northwest Hawaiian Ridge (yellow), and Daikakuji (purple). The black line is the average of Mauna Loa and Mauna Kea, representative of average Hawaiian shield basalt ((La/Yb)PM ≈ 3.4) and the dashed line is average EPR MORB ((La/Yb)PM ≈ 0.8). Inset to figure (b) gives the composition of the end members used in the calculation of mixing lines (Tanaka et al., 2002). Data sources are the same as in Figure 3.2.84of the Hawaiian geochemical components are thought to be intrinsic to the plume and sourced from the deep mantle (Tanaka et al., 2002, 2008; Regelous et al., 2003; Weis et al., 2011).   Some NHWR basalts (Daikakuji and Nīhoa) have Loa-like Pb isotopic compositions whose range of variations may be explained by mixing 35-40% of the Enriched Makapu‘u (EMK) with the Kea component (Figure 3.3) (Tanaka et al., 2002). The Enriched Makapu‘u end-member, proposed to contain mafic recycled crustal material (Tanaka et al., 2008), is more fusible and thus likely to be observed under lower melting conditions (Figure 3.3a inset) (Pertermann and Hirschmann, 2003). High primitive mantle-normalized (La/Yb)PM values (>5) correspond to lower degrees of partial melting as La is more incompatible and highly sensitive to degree of partial melting, whereas Yb is nearly constant in Hawaiian basalts due to melting in the garnet stability field (Hofmann and Farnetani, 2013). This is not an effect of regional changes in lithospheric thickness as the NWHR erupted on oceanic lithosphere older than ~70 Ma (White, 1993; Caplan-Auerbach et al., 2002), which is the age when oceanic lithospheric thickness reaches a stable maximum of ~90 km based on the plate tectonic model of Parsons and Sclater (1977). The isolated occurrence of enriched Loa-type Pb isotopic compositions at Daikakuji suggests it is not related to a first arrival of a sustained Loa component in the Hawaiian mantle plume as previously proposed (Garcia et al., 2015). Daikakuji is characterized by anomalously high (La/Yb)PM compared to Hawaiian shield basalts ((La/Yb)PM=6.9-8.5 for Daikakuji basalts versus 2.0-5.1 for Mauna Loa and Mauna Kea basalts; primitive mantle values from McDonough and Sun, 1995). Many NWHR Loa- and Kea-type basalts have high (La/Yb)PM compared to Hawaiian Island and Emperor Seamount shield basalts and may reflect, at least locally, lower degrees of melting. Previous authors argued that Emperor Seamount basalts are characterized by higher 206Pb/204Pb at a given 208Pb/204Pb, because they were initially erupted onto thin, young crust that grew progressively thicker with time and increasing distance from the Kula-Pacific-Izanagi 85spreading ridge (Keller et al., 2000; Regelous et al., 2003). The geochemistry of Emperor Seamount basalts reflects this thickening lid effect, with a possible contribution from a Mauna KeaMauna LoaKo‘olauEPR MORBHawaiian IslandsEmperor SeamountMeijiDetroit SuikoOjinKokoYuryakuTwin BanksWest NīhoaNW Hawaiian RidgeYuryakuDaikakujiUnnamedA. BergMidwayPioneerNorthamptonLaysanGardnerKeoeaNīhoaMokumanamana0.880.900.920.940.960.980 20 40 60 80 100208 Pb*/206 Pb*in km)( ssenkcihT cirehpsohtiLAge of Seafloor at Time of EruptionEmperor SeamountsHalf-space cooling modelPlate cooling model020406080100120140Figure 3.4: Radiogenic Pb and lithospheric thickness versus age of the seafloor at time of eruptionPlot of 208Pb*/206Pb* and models of lithospheric thickness versus age of the seafloor at the time of eruption for NWHR basalts, Emperor Seamounts, EPR MORB glasses, and Ko‘olau, Mauna Loa, and Mauna Kea volcanoes (data sources are the same as in Figure 3.2). The lithospheric thickness secondary axis on the right applies to the black and blue dashed curves, which are theoretical models of the rate of lithospheric thickness growth with increasing age of the oceanic crust. The curves on the plot are thoretical models approximating the evolution of the thickness of oceanic crust as it cools. The black curve are the results of a plate model, in which lithospheric thickness increases according to the equation L=11*t1/2 (where L is the lithospheric thickness in kilometers and t is the age of the crust in millions of years) until approximately 75 Ma (dashed vertical line), the age when oceanic crust has theoretically reached a steady state maximum thickness (Parson and Sclater, 1977). The dashed blue curve are the results a half-space cooling model L=10*t1/2 (where L is the lithospheric thickness in kilometers and t is the age of the crust in millions of years) where the crust continues to thicken with age without reaching a steady, constant thickness (Parker and Oldenburg, 1973). Lithospheric thickness determined from different seismic models agrees with the thickness predicted by the half space cooling model when oceanic plate ages are younger than ~110 Myr, after which results of different models diverge (Steinberger and Tecker, 2016). The only time where lithospheric thickness may possibly have an effect on the radiogenic Pb of erupted basalts is when the thickness of the crust is changing rapidly (i.e., during the formation of the oldest Emperor Seamounts; Regelous et al., 2003). All NWHR and Hawaiian Island volcanoes have erupted onto lithosphere with relatively the same thickness and age (Müller et al., 2008), so this cannot account for the variation observed in radiogenic Pb for these volcanoes.86depleted MORB mantle source for the oldest seamounts, Detroit and Meiji (Figure 3.3) (Keller et al., 2000; Regelous et al., 2003; Whittaker et al., 2015). Conversely, all NWHR volcanoes erupted onto oceanic crust older than 70 Ma without interaction with a nearby spreading ridge (Figure 3.4) (White, 1993; Whittaker et al., 2015). Thus, the depleted component (DMK) involved in mixing with the Kea component to produce NWHR Kea-type basalts is different from the one involved in the formation of the Emperor Seamounts.40 904200681010 20 70605030 80802/ *bP602* bP det amit sEVm( xul F emul o3) ces/Age (Ma)Northwest Hawaiian Ridge EmperorSeamounts468MeijiDetroitSuikoOjinKokoKimmeiKaua‘iYuryakuDaikakujiMidwayPioneer0.88Kea TrendLoa TrendEnriched LoaHawaiian Islands Magmatic FluxRadiogenic Pb Van Ark & Lin, 2004  Vidal & Bonneville, 2004 Wessel, 2016  Emperor SeamountMeijiDetroit SuikoOjinKokoYuryakuTwin BanksWest NīhoaNW Hawaiian RidgeYuryakuDaikakujiUnnamedA. BergMidwayPioneerNorthamptonGardnerKeoeaNīhoaMokumanamanaoldestyounger12140.910.930.920.900.890.940.950.960.97GardnerKea!"#$"#$0.9250.9350.9450.9550.9650.9750 2 4 6 8 10Est. Vol. Flux (m3/sec)208 Pb*/206 Pb*Enriched LoaLoaFigure 3.5: Northwest Hawaiian Ridge radiogenic Pb variations with Hawaiian plume estimated volume flux(in m3/s assuming flexural compensation; Van Ark and Lin, 2004; Vidal and Bonneville, 2004; Wessel, 2016). The black curve is the average flux variation with uncertainty shown in red (Wessel, 2016). The flux of the Emperor Seamounts is from Van Ark and Lin (2004; green line) and Vidal and Bonneville (2004; purple line) or not calculated (Wessel, 2016). Radiogenic Pb increases gradually with age along the NWHR (dashed regression lines). Emperor Seamount triangles are averages of all data from each seamount. Hawaiian Island data is the average of each volcano. Data sources are the same as in Figure 3.2. Inset shows in detail 208Pb*/206Pb* versus estimated volume flux (m3/s; Wessel, 2016). 87 Lead isotopes vary systematically in time and space along the NWHR (Figure 3.5). They become more radiogenic (i.e., increase in 208Pb*/206Pb*) from the oldest to the youngest volcanoes. This represents a fundamental difference from the Emperor Seamounts, which on average are uniformly unradiogenic. Radiogenic Pb correlates with magmatic flux along the NWHR, except for Gardner and Daikakuji seamounts, which are respectively local maxima and minima of magmatic flux along the ridge (Figure 3.5 inset). Volcano compositions separate into three compositional trends (Kea, Loa, and Enriched Loa, Figure 3.5 inset) only for the main Hawaiian Islands at a critical threshold of magmatic flux (~4 m3/s based on the most recent magmatic flux model; Wessel, 2016).3.5. DISCUSSION A discussion of the variations of Pb isotopes with magmatic flux must address both lithospheric and mantle source potential origins for the covariations. Both of these topics are addressed below.  3.5.1 Role of the Lithosphere in Magmatic Flux and Pb Isotopic Composition of NWHR Basalts The thickness of the lithosphere is postulated to have an effect on the magmatic flux of mantle plumes because thinner, warmer lithosphere allows for higher degrees of melting than thick, cold lithosphere (White, 1993; Keller et al., 2000; Regelous et al., 2003; Niu et al., 2011). Higher degrees of melting results in magmas that are less enriched in incompatible elements with lower radiogenic Pb due to dilution of the signature of small, enriched heterogeneities by productive melting of large volumes of peridotite (Pertermann and Hirschmann, 2003; Garcia et al., 2010). Thus, if lithospheric thickness did exert a primary control over magmatic flux, we would also expect it to impact the Pb isotopic signature of erupted basalts. Along the NWHR, there is no correlation of radiogenic Pb with lithospheric thickness (Figure 3.4), regardless of the type of lithospheric thickness model employed (e.g., a half space cooling model or a plate tectonic model; Parker and Oldenberg, 1973; Parsons 88and Sclater, 1977). This indicates that lithospheric thickness variations are not as significant as source composition changes in explaining the exponential increase in magmatic flux along the NWHR. Emperor Seamount geochemistry, conversely, correlates with changes in lithospheric thickness, although it is important to note that increases in lithospheric thickness from the oldest to youngest Emperor Seamounts did not result in any significant decrease in magmatic flux (Figure 3.4) (Regelous et al., 2003; Garcia et al., 2015). Both the Emperor Seamounts and all of the Hawaiian chain younger than the bend exhibit the same magnitude of variation in radiogenic Pb. To illustrate, the Emperor Seamounts 208Pb*/206Pb* range from 0.89 at Meiji to 0.94 at Detroit, representing 59.4% of total range of Hawaiian-Emperor chain. Younger than the bend range from 208Pb*/206Pb* 0.92 at Midway to 0.97 at Ko‘olau, 59.6% of total range. This corresponds to significantly different ranges of variations of lithospheric thickness between the two segments (Emperor Seamounts 22.4-92.2 km, 92% of total range of Hawaiian-Emperor chain; younger than the bend 89.4-98.2 km, 10% of the total range) and magmatic flux (Emperor Seamounts ~1% of variation; younger than the bend ~99% of variation). These results are inconsistent with the hypothesis that changes in the geochemical signature of erupted basalts and magmatic flux is principally controlled by lithospheric thickness.  Physical variations of the lithosphere, such as changes in temperature and thickness between the major Pacific plate fracture zones (e.g., Murray, Mendicino, and Molokai fracture zones; Figure 3.1 inset), could have an effect on magmatic flux (Van Ark and Lin, 2004; Wessel, 2016). However, NWHR geochemistry does not change across these structures and instead gradually increases in radiogenic Pb across all fracture zones of the NWHR and Hawaiian Islands (Figure 3.5). It has also been hypothesized that variations between extensional and compressional stress in the Pacific plate by changes in absolute plate motion could have an effect on how much magma is able to pass through the lithosphere and erupt (Wessel and Kronke, 2008; O’Connor et al., 2013; Wessel, 2016). This could have influenced 89the degree of melting along the NWHR, however, both the steady increase in radiogenic Pb and in magmatic flux along the NWHR after Midway Island would require a gradual transition into an increasingly tensional stress regime for the Pacific plate from ~20 Ma to the present. This is unlikely as geologic events that change absolute plate motion around the Pacific are more episodic (Wessel and Kronke, 2008). It has also been hypothesized that small-scale sub-lithospheric rolls caused by sub-lithospheric convection is the cause of the Loa and Kea bilateral trends on the Hawaiian Islands (Ballmer et al., 2011). However, the entire NWHR was erupted onto oceanic crust mature enough to have small-scale sub-lithospheric convection, yet there is only an episodic presence of the Loa component and no evidence for a bilateral plume along the NWHR.  It is possible that smaller timescale fluctuations in magmatic flux (e.g., 1-2 and 10-15 Ma variations; Wessel, 2016; see Figure 3.5) could be caused by plume dynamical variations (e.g., solitary waves; Whitehead and Helfrich, 1988; Ito, 2001) possibly generated by pulses of core-mantle heat transfer (Mjelde et al., 2010). Our sample resolution is not fine enough to assess the potential correlation of geochemistry with such a periodicity. The current resolution of sampling does not show a correlation of radiogenic Pb with these small fluctuations in flux, whereas there is a clear general trend of increasing magmatic flux. The same limitation in sampling resolution applies to assessing the possible effect of local fluctuations in the thickness or stress field of the lithosphere. Below, we propose a deep mantle geodynamic conceptual model that accounts for the variation of Pb isotopes with time along the Hawaiian-Emperor chain and also explains geophysical changes in magmatic flux.3.5.2 Role of the LLSVP in Magmatic Flux and Pb Isotopic Composition of NWHR Basalts The arrangement of compositionally heterogeneous lower mantle structures in the source region of deep-sourced plumes is thought to control the geochemistry of erupted oceanic island basalts (Farnetani and Hofmann, 2009; Weis et al., 2011). The Hawaiian 90mantle plume is located above an ultra-low velocity zone (ULVZ) at the edge of the large low shear velocity province (LLSVP) located in the Pacific lower mantle near the core-mantle boundary (McNamara et al., 2010; Cottaar and Romanowicz, 2012; French and Romanowicz, 2015; Garnero et al., 2016). LLSVP are thousands of kilometers wide, sharp-sided (~400-1200 km thick), chemically heterogeneous, long-lived piles of anomalously seismically slow material in the lower mantle below the Pacific Ocean and continental Africa (Garnero et al., 2016). ULVZ are much smaller, denser features (~10 km thick and <100 km across, except underneath Hawai‘i where it is ~900 km across; Cottaar and Romanowicz, 2012) that may contain partial melts (Jellinek and Manga, 2004; McNamara et al., 2010; Garnero et al., 2016). Mantle plumes may drift across or around these lower mantle structures on long timescales triggered by dynamic re-equilibration from the arrival of subducting slabs (McNamara et al., 2010; Bower et al., 2013; Garnero et al., 2016; Hassan et al., 2016). It has been suggested that the enriched geochemical compositions of some oceanic island basalts originate from these lower mantle reservoirs, which have been proposed to be the remnants of a primordial magma ocean or accumulations of subduction-recycled crustal material (Weis et al., 2011; Williams et al., 2015; Garnero et al., 2016). ULVZ material may be denser than surrounding peridotitic mantle and only small amounts are likely to be entrained by plumes in addition to LLSVP material (Jellinek and Manga, 2004; McNamara et al., 2010; Weis et al., 2011; Cottaar and Romanowicz, 2012; Williams et al., 2015). Increasing radiogenic Pb along the NWHR suggests an increasing contribution from these enriched sources with time. Dynamic changes in the lower mantle may control the amounts of LLSVP, ULVZ, and ambient mantle entrained into the Hawaiian mantle plume. We propose that the Hawaiian plume initiated close to the margin of the Pacific LLSVP, sampling only the Kea component (the deep Pacific mantle) during Emperor Seamount formation, then drifted up and anchored at the edge of the Pacific LLSVP to entrain enriched material in increasing quantities as the NWHR formed (Figure 3.6). 91Pacific LLSVPcorelowermantleAEmperor Seamounts~82 - 47 MaKeaCKeaLoaE Hawaiian Islands ~6.5 - 0 Ma NWHR~47 - 6.5 MaDInter-mittentLoaKeaHawaiian Plume FormationBFigure 3.6: Model evolution of the Hawaiian plume source at the CMBFigure shows the lowermost 800-1000 kilometers of mantle (not to scale; blue lens is an ULVZ). Black and grey arrows represent generalized flow in the mantle (Steinberger, 2000; Davaille et al., 2003; Walker et al., 2011; Hassan et al., 2016) and LLSVP (McNamara et al., 2010; Zhao et al., 2015; Garnero et al., 2016). (A) Pacific LLSVP (B) The Hawaiian plume is created at the LLSVP edge where temperature anomalies are greatest and large viscosity gradients stabilize an initially thermal plume (Jellinek and Manga, 2004; Davies and Davies, 2009; Garnero et al., 2016). (C) The plume will migrate in mantle flow up the sloped northern edge of the LLSVP to a ridge where plumes are dynamically stabilized over geologic time (Steinberger, 2000; Davaille et al., 2003; Burke and Torsvik, 2004; Jellinek and Manga, 2004; Steinberger et al., 2004; Davies and Davies, 2009; Bower et al., 2013; Hassan et al., 2015; Zhao et al., 2015). Flow into thermals perturbs ULVZ material below (Williams et al., 2015). (D) LLSVP material is entrained in the plume and 208Pb*/206Pb* values increase in erupted basalts (Weis et al., 2011; Williams et al., 2015). (E) A significant amount of LLSVP (+/- ULVZ) is consistently and continuously entrained into the southeast side of the Hawaiian plume, resulting in the presence of the bilateral Loa and Kea trends observed on the Hawaiian Islands. 92 Motion of the Hawaiian mantle plume has the potential to sample different geochemical sources over its >81 million year lifetime. Plumes migrate slowly through the mantle at horizontal velocities of ~1.1 cm/year (Davies and Davies, 2009), an observation supported by the better fit of dynamic models with moving-source plumes than fixed-source plumes in global seismic tomographic models (Boschi et al., 2007). The lower mantle has been moving south at a rate of 1-4 cm/year (seismic anisotropy studies; Walker et al., 2011) or 1-2 mm/year (plume experiments; Davaille et al., 2003). Dynamic global mantle convection and hotspot track reconstructions of the last 230 Ma identified south-southeast flow with varied strength for the base of Hawaiian mantle plume towards the central Pacific (Steinberger, 2000; Davaille et al., 2003; Burke and Torsvik, 2004; Steinberger et al., 2004; Hassan et al., 2016). Assuming the base of the Hawaiian mantle plume was advected in that flow at the lowest rate (~1 mm/year; Steinberger, 2000; Steinberger et al., 2004; Garnero et al., 2016; Hassan et al., 2016), it could have drifted ~80 kilometers in the lower mantle. Plume movement tends to be episodic (Davies and Davies, 2009; Hassan et al., 2015), allowing the sampling of different lower mantle domains. Dynamic models and plume experiments have established that plumes anchor on topographic features such as ULVZ or LLSVP ridges (Jellinek and Manga, 2004; Steinberger et al., 2004; Boschi et al., 2007; McNamara et al., 2010; Bower et al., 2013; French and Romanowicz, 2015). Finally, the mass conservation law requires compensation of upper mantle wind by equal and opposite movement in the lower mantle (Steinberger, 2000; Farnetani and Hofmann, 2009), likely resulting in plume drift. It has been proposed that the highest temperatures at the core mantle boundary are located along the rim of LLSVP; this is where P-waves begin to slow, many mantle plumes are generated, and ULVZ are detected (Jellinek and Manga, 2004; Kumagi et al., 2008; Davies and Davies, 2009; McNamara et al., 2010; Cottaar and Romanowicz, 2012; Hassan et al., 2015). The northern edge of the Pacific LLSVP underneath Hawai‘i is not steep-sided, 93but has a 25-35° north-northwest-facing gentle slope to a topographic high ~450 km above the core mantle boundary (He and Wen, 2012; Frost and Rost, 2014; Zhao et al., 2015). The Hawaiian mantle plume initiated slightly to the north of the present plume location where the highest temperatures occur (Figure 3.6b). After formation, the flow of ambient mantle south up the edge of the LLSVP would have pushed the plume to the most dynamically stable place, anchored at an LLSVP ridge above a ULVZ (Figure 3.6c, d) (Davaille et al., 2003; Jellinek and Manga, 2004; Zhao et al., 2015; Garnero et al., 2016). During this time, the Hawaiian mantle plume sampled peridotitic mantle from outside the LLSVP (Sobolev et al., 2005; Tanaka et al., 2008), which corresponds to the Kea component, also present in nearly all Pacific oceanic plateaus (e.g., Ontong Java, Shatsky Rise, and Wrangellia) and oceanic islands (Nobre Silva et al., 2013a). It is also the same composition as prevalent mantle (PREMA), an ubiquitous component in the mantle where oceanic island basalt isotope arrays converge (White, 2015 and references therein). The dominance of this material during the initiation of the Hawaiian mantle plume (e.g., Emperor Seamounts) would account for the low magmatic flux during this time as peridotite is less productive than enriched, mafic lithologies (Pertermann and Hirschmann, 2003; Van Ark and Lin, 2004; Vidal and Bonneville, 2004). This is in agreement with the theoretical flux for a fully peridotitic Hawaiian plume (~3.5 m3/s for peridotite only as found by Sobolev et al. [2005] rather than the observed 8 m3/s modeled by Wessel [2016]) and within the range of other Pacific plume fluxes (Courtillot et al., 2003). Conversely, if the Hawaiian plume was both generated and fixed on the Pacific LLSVP, we would expect to observe Loa compositions in basalts from the entirety of the Hawaiian-Emperor chain. Although it is possible that the participation of the Loa component was concealed in Emperor Seamount basalts by dilution from voluminous melting beneath young, thin lithosphere (Figure 3.4) (Regelous et al., 2003; Hofmann and Farnetani, 2013), we would then expect the Emperor Seamount data to exhibit mixing relationships that include the Loa enriched end member (EMK), which is not 94observed (Figure 3.3b).  After the plume was anchored on a LLSVP ridge above the ULVZ, it entrained initially small amounts of enriched material from the LLSVP and/or ULVZ (Figure 3.6d). Incorporation of increasing amounts of this material in NWHR volcanoes gradually elevated radiogenic Pb (Figure 3.5), changed the slope of magmatic flux growth (Van Ark and Lin, 2004; Vidal and Bonneville, 2004; Wessel, 2016), and accelerated the volcanic propagation rate (O’Connor et al., 2013). These changes all commenced ~20 Ma after the bend around Midway Island. Assuming a 15-25 Ma mantle plume transit time, this enriched material was entrained around the time of the bend (Hassan et al., 2015; 2016). Plate reconstruction studies have found that absolute plate motion change was likely 4-31° in the Pacific around the time of the bend (Steinberger et al., 2004; Tarduno et al., 2009; Wright et al., 2015), not enough to account for the ~120° change in bearing of the Hawaiian-Emperor chain at the bend (i.e., a ~60° change in absolute Pacific plate motion; Tarduno et al., 2009; Hassan et al., 2016). Previous studies hypothesized that interaction of the Hawaiian mantle plume with the Kula-Pacific-Izanagi mid-ocean ridge around ~85 Ma tilted the plume. With both increasing distance from and waning in spreading rate of the ridge, the plume moved back to a more vertical position, resulting in hotspot motion (Tarduno et al., 2009). This documented plume movement in the upper mantle coupled with the hypothesized southward plume motion in the lower mantle stabilized around the time of the bend and likely contributed to the sharpness of the bend and to the transition in Hawaiian plume geochemical signature (Steinberger, 2000; Steinberger et al., 2004; Tarduno et al., 2009; Hassan et al., 2016).  We speculate that the longer the plume is fixed on a ridge of the LLSVP, the more sustained is the sampling of LLSVP- and/or ULVZ-enriched material, resulting in both higher radiogenic Pb and magmatic flux as Loa-type source lithologies are more fusible pyroxenites and eclogites (Pertermann and Hirschmann, 2003; Sobolev et al., 2005). The bilateral Kea-Loa trends of the Hawaiian Islands appear once the plume had been anchored 95on the Pacific LLSVP for enough time to provide a sustained supply of both enriched material on the southwest edge of the plume (Loa) and ambient Pacific mantle on the northeast side of the plume (Kea; Farnetani and Hofmann, 2009; Weis et al., 2011; Nobre Silva et al., 2013a) (Figure 3.6e). The ephemeral presence of the Loa composition along the NWHR suggests that a simple concentric or bilateral zonation of the Hawaiian mantle plume is overly simplistic, as concluded for the Hawaiian Islands (Xu et al., 2014). The isolated and highly radiogenic basalts erupted at Daikakuji volcano right after the Hawaiian-Emperor bend may be explained by a combination of unique melting conditions of this volcano coupled with the presence of a small mantle heterogeneity. Daikakuji exhibits the coldest mantle potential temperature of any studied Hawaiian volcano (1335 ± 26 °C, compared to 1632-1690 °C on Hawai‘i Island; Tree, 2016), resulting in low degrees of melting. This is also supported by the fact that Daikakuji has the highest (La/Yb)PM of the entire Hawaiian-Emperor chain (Figure 3.3a inset). These low degrees of melting allowed the signature of a finite and isolated heterogeneity to be expressed for a short period of time. The mantle has been shown to be heterogeneous on scales varying between melt inclusion scale (Ren et al., 2005) to the hundreds of kilometers scale (Garnero et al., 2016), and periodic entrainment of plum-pudding type heterogeneities is not unexpected.3.6. CONCLUSION New high-precision Pb isotopic data closes the 2,700 km (~42 Ma) gap in the Hawaiian-Emperor chain geochemical record and shows a transition in basalt composition commencing ~20 million years after the bend. During this period, the flux of the Hawaiian plume, mantle potential temperature, and volcanic propagation rate all increased. The correlation of radiogenic Pb with magmatic flux indicates that all of these changes are related, and we suggest may be triggered by the movement of the deep source of the Hawaiian plume through different lower mantle compositional domains. First the Hawaiian plume sampled dominantly ambient Pacific mantle outside of the LLSVP, then entraining enriched material 96from the ULVZ and Pacific LLSVP. Mounting geochemical and geodynamical data on oceanic islands and chains suggest the lower mantle is both dynamic and compositionally heterogeneous in space and time. Long-lived mantle plumes such as Hawai‘i provide a window into this fundamental Earth reservoir and future integrated geochemical and geophysical studies of long island chains will provide the key insights needed to decipher the relationship between geochemical components and geophysical anomalies in the deep mantle.97Chapter 4 The Spatial Distribution and Composition of the Loa Mantle Component Along the Northwest Hawaiian RidgeChapter 4 The Spatial Distribution and Composition of the Loa Mantle Component Along the Northwest Hawaiian Ridge4.1. INTRODUCTION The Hawaiian-Emperor chain is the ~6000 kilometer long surface expression of the Hawaiian mantle plume and is composed of >100 volcanoes that erupted over the past >81 million years (Dana, 1890; Jackson et al., 1972; Regelous et al., 2003; Tanaka et al., 2008; Garcia et al., 2015). The chain is divided into three sections: the Emperor Seamounts, the Northwest Hawaiian Ridge (NWHR), and the Hawaiian Islands. The oldest section is composed of the Emperor Seamounts (~85-50 Ma), which extend from the Northwestern Pacific near the Aleutian Islands to the bend in the Hawaiian-Emperor chain (Jackson et al., 1972; Regelous et al., 2003). The Northwest Hawaiian Ridge (NWHR) includes 51 volcanoes spanning ~42 million years and ~2800 kilometers between the bend in the chain and the Hawaiian Islands (Sharp and Clague, 2006; O’Connor et al., 2013; Garcia et al., 2015). The Hawaiian Islands, a chain of subaerial islands beginning ~6.3 Ma at Ni‘ihau and extending to the currently active Kīlauea and Lō‘ihi volcanoes, constitute the youngest section of the chain (Sherrod et al., 2007). Hawaiian Island volcanoes are geographically and geochemically distinct and sort into two parallel chains, named the Kea and Loa trends (Jackson et al., 1972; Tatsumoto, 1978; Abouchami et al., 2005; Weis et al., 2011). The cause of this geochemical dichotomy in the Hawaiian mantle plume is still debated, with proposed hypotheses including variations in lithospheric thickness, plume dynamics, mantle potential 98temperature, or sampling of a deep mantle source with variations in composition introduced by the presence of recycled matrials (e.g., pelagic sediments, lower continental crust, oceanic crust, ancient gabbroic cumulates, or carbonate sediments) (Sleep, 1984; Hauri, 1996; Blichert-Toft et al., 1999; Phipps Morgan and Morgan, 1999; DePaolo et al., 2001; Gaffney et al., 2004; Abouchami et al., 2005; Huang et al., 2005b; 2009; Sobolev et al., 2005; Ballmer et al., 2010, 2011; Hanano et al., 2010; Weis et al., 2011; Jackson et al., 2012; Hofmann and Farnetani, 2013; Pietruszka et al., 2013; Harrison et al., 2015; Frey et al., 2016; Harrison et al., 2017).  Only Kea trend compositions have been observed in Emperor Seamount lavas (>50 Ma), whereas the young Hawaiian Islands have erupted both Kea and Loa-type compositions (Regelous et al., 2003; Frey et al., 2005; Huang et al., 2005a, 2011a; Tanaka et al., 2008; Garcia et al., 2010, 2015; Greene et al., 2010; Cousens and Clague, 2015). High-precision isotopic and elemental analysis of NWHR shield-stage basalts fills the ~42 million year gap between the two sections of the Hawaiian-Emperor chain, and addresses key questions such as what are the mantle source differences between the two trends and why the Kea trend is dominant earlier on while the Loa trend appears episodically along the NWHR and dominates volumetrically on the Hawaiian Islands (Tanaka et al., 2008; Weis et al., 2011; Garcia et al., 2015; Harrison et al., 2017). Any explanation for Loa and Kea geochemical differences on the Hawaiian Islands must also explain Loa and Kea relative proportions and distribution along the NWHR (and vice versa), so characterizing the spatial occurrence of the Loa geochemical trend along the NWHR has the potential to provide time constraints to the Loa and Kea trend origin debate.   In this study, Nd, Sr, and Hf isotope ratios and trace element concentrations of NWHR shield-stage tholeiitic basalts were analyzed to determine the spatial distribution and geochemical fingerprint of the Loa composition along the NWHR. To accomplish this, a statistical model of the geochemical signature of Loa- and Kea-type affinity is developed 99using the well-characterized Hawaiian Islands, where more than ~600 samples of shield basalts have been analyzed for high-precision Pb, Sr, and Nd isotopes and ~400 samples for Hf isotopes. This model is then used to classify NWHR shield-stage samples as either Loa or Kea trend based on their isotopic compositions. Comparison of the resulting distribution and geochemical compositions of NWHR Loa trend basalts to that of the Hawaiian Islands shows which explanations for the origin and distribution of the Loa geochemical trend can stand the “test of time” (here ~47 million years). Recently, variations in Pb isotopic composition of NWHR basalts were shown to vary in tandem with magmatic flux, which was proposed to result from the drifting of the Hawaiian mantle plume through different lower mantle compositional domains and sampling materials with different time-integrated incompatible element enrichments (Harrison et al., 2017). In this model, the spatial orientation of lower mantle geophysically imaged structures has the potential to account for the observed Loa-Kea trend distribution over time if the variations in seismic velocity in the lower mantle are produced by compositional distinctions and if the position of the heterogeneities sampled by the plume are preserved during mantle transit (Weis et al., 2011; Farnetani et al., 2009, 2010, 2012; Garnero et al., 2016 and references therein; Harrison et al., 2017). This study resolves large-scale Loa trend heterogeneities sampled repeatedly by the Hawaiian mantle plume over ~47 Ma. These results support a dynamic, plume-related model of Loa and Kea trend genesis.4.2. GEOLOGICAL SETTING AND SAMPLE LOCATIONS The NWHR comprises ~51 volcanoes spanning ~2800 kilometers (Garcia et al., 2015). This study analyzed 23 samples of Northwest Hawaiian Ridge tholeiitic shield-stage basalt from 13 different seamounts or islands that represent ~42 million years of the geochemical evolution of the Hawaiian mantle plume (Figure 4.1). Sampling of the ridge is currently highly restricted by the Papahānaumokuākea Marine National Monument, a ~1,508,870 km2 U.S. Marine Conservation Area and World Heritage Site that stretches from 100Middle Bank to Academician Berg (~80% of the NWHR). As a result, samples in this study are from pre-2008 dredge or island sampling expeditions stored at the University of Hawai‘i and Scripps Institution of Oceanography rock collections or from the University of Hawai‘i’s HOV (Human Occupied Vehicle) submersible Pisces V reconnaissance survey of the new monument in 2003 and 2007 (Garcia et al., 2015). Harrison et al. (2017) gives detailed sample locations. 4.3. ANALYTICAL TECHNIQUES Neodymium, Sr, and Hf isotopic compositions and trace element concentrations were analyzed at the Pacific Centre for Isotopic and Geochemical Research at the University of British Columbia (PCIGR; Tables 4.1 and Table 4.2). Pb isotopic composition of these 175˚175˚180˚180˚−175˚−175˚−170˚−170˚−165˚−165˚−160˚−155˚−155˚20˚25˚ 25˚30˚−160˚20˚30˚GardnerFrench FrigateYuryakuDiakakujiBrooksAcademicianBergE. NorthamptonMaro ReefLaysanMidwayKaua‘iNīhoaW. NīhoaMokumanamana (Necker)AbbottColahan HancockUnnamedPearl and HermesNeroHelsleyLaddHawai‘iKo‘olauMiddle BankTwin BanksKeoeaRiataPioneerTownsendCromwell47 Ma141 Ma132 Ma127 Ma1N20 Ma225 Ma112 Ma 411 Ma37 Ma2<5 Ma5 0 100KilometersKaua‘iKo‘olauWai‘anaeE. Moloka‘iW. Moloka‘iW. MauiHana RidgeKohalaKīlaueaHualālaiLāna‘iKaho‘olawePenguin BankW. Ka‘enaNi‘ihauMauna KeaMauna LoaHawaiianIslands Lō‘ihiLoa Kea −6000 −4000 −2000 00 km 500 kmBathymetry (m)Figure 4.1: Bathymetric map of the ~51 seamounts and islands of the Northwest Hawaiian Ridge and sample locationsBathymetry is 2-minute Gridded Global Relief Data ETOPO2v2 satellite altimetry dataset (U.S. Department of Commerce, National Oceanic and Atmospheric Administration, National Geophysical Data Center, 2006; downloaded March 14, 2014) and new multibeam bathymetry (Smith et al., 2014). White circles show sample locations. References for ages are in yellow superscript and are as follows: 1 – O’Connor et al., 2013; 2 – Dalrymple et al., 1974; 3 – Dalrymple et al., 1981; 4 – Garcia et al., 1987; 5 – Garcia et al., 2010. Inset figure is a simplified topographic map of the Hawaiian Islands modified from the Main Hawaiian Islands Chart 750-001 Version 17 downloaded from the Hawai‘i Mapping Research Group at the University of Hawai‘i at Mānoa’s website (Main Hawaiian Islands Multibeam Bathymetry Synthesis: www.soest.hawaii.edu).101 samples are reported in Harrison et al. (2017) and major element concentrations in Garcia et al. (2015) and Tree (2016). For isotopic analysis, whole rock powders were sequentially leached in multiple steps using 6 N HCl until the supernatant was clear, followed by two rinsing steps with ultra-pure water. This procedure is necessary to remove post-eruption secondary alteration or contamination from sample preparation, which should be more susceptible to acid attack than the more resistant primary magmatic minerals (Weis and Frey, 1991; Weis et al., 2005; Hanano et al., 2009; Nobre Silva et al., 2009, 2010). After leaching, samples were digestedTable 4.1: Trace element concentrations of Northwest Hawaiian Ridge basaltsVolcano Yuryaku Daikakuji Daikakuji Daikakuji Unnamed A. Berg Midway Midway Pioneer Northampton NorthamptonSample A-53-1 A-55-1 A-55-2 A-55-4 84-28E 84-30G R-1316 R-1295.6 P5-524-42 KK-76-5-4DD 76-5-4ASc 31.3 22.5 20.8 21.8 27.0 27.5 25.2 24.8 33.2 30.3 31.6V 275 207 225 235 263 265 195 191 325 277 286Cr 426 477 256 274 120 131 517 490 440 969 977Co 46.2 49.9 42.2 43.4 39.9 42.6 53.3 57.9 47.6 55.0 55.0Ni 156 303 199 190 83 89 323 362 253 303 284Zn 138 85 79 110 113 113 119 126 125 94 98Ga 22 20 23 23 24 24 22 22 23 20 20Cs 0.123 0.478 0.674 0.678 0.467 0.464 0.086 0.089 0.225 0.153 0.257Rb 5.0 14.2 23.4 21.2 12.3 8.0 2.7 2.8 6.5 4.8 5.1Sr 323 473 573 610 436 435 351 323 360 312 308Ba 76 169 290 284 139 102 57 54 65 86 67Y 28.89 26.16 32.59 34.19 32.28 32.75 33.75 28.01 33.27 26.57 25.97Zr 138 195 282 286 187 184 189 186 192 140 133Hf 3.82 4.75 6.65 6.59 4.94 4.78 4.90 4.88 5.01 3.73 3.71Nb 13.32 15.63 25.56 26.17 18.95 18.38 17.58 17.20 21.24 13.68 13.58Ta 0.76 0.71 1.12 1.13 1.05 1.03 1.00 0.97 1.19 0.78 0.80W 0.78 0.67 0.30 0.30 0.79 0.15 0.18 0.17 0.63 1.40 0.17Li 36.83 23.87 18.00 21.89 12.54 23.89 5.85 6.30 29.81 8.57 19.36La 9.33 17.06 26.44 26.40 17.96 17.35 15.46 13.66 15.28 10.59 9.98Ce 23.73 38.27 57.75 58.15 40.91 40.72 36.66 35.84 35.66 27.58 26.87Pr 3.50 5.04 7.42 7.48 5.61 5.52 5.29 4.87 5.49 3.85 3.86Nd 16.98 23.47 33.12 33.74 26.48 25.54 25.93 23.96 25.80 19.10 19.25Sm 5.03 6.00 8.24 8.36 6.98 6.83 7.14 6.77 6.72 5.04 5.25Eu 1.89 2.10 2.70 2.77 2.27 2.22 2.37 2.18 2.28 1.80 1.72Gd 6.14 6.30 8.11 8.41 7.45 7.24 7.65 7.14 7.36 5.47 5.69Tb 0.95 0.92 1.19 1.23 1.13 1.11 1.16 1.07 1.15 0.91 0.86Dy 5.53 4.93 6.31 6.46 6.29 6.14 6.22 5.81 6.10 5.00 5.23Ho 1.01 0.92 1.14 1.14 1.14 1.11 1.10 1.01 1.06 0.93 0.86Er 2.53 2.34 2.85 2.99 2.91 2.89 2.75 2.55 2.79 2.42 2.38Tm 0.33 0.29 0.36 0.36 0.39 0.39 0.36 0.34 0.37 0.33 0.34Yb 2.00 1.69 2.11 2.13 2.27 2.18 2.07 1.93 2.16 1.89 1.94Lu 0.270 0.220 0.264 0.292 0.307 0.300 0.280 0.262 0.311 0.262 0.276Pb 0.74 1.53 2.22 2.16 1.70 1.24 1.06 1.03 2.63 0.83 0.84Th 0.71 1.31 2.18 2.22 1.46 1.42 1.08 1.04 1.20 0.83 0.77U 0.314 0.436 0.672 0.689 0.474 0.438 0.368 0.375 0.817 0.275 0.2771All abundances in ppm. 102using a combination of HF and HNO3 and purified on chromatographic ion exchange columns (Weis et al., 2005, 2006, 2007). Hf and Nd isotopic ratios were determined on a Nu Plasma (NP214, Nu Instruments Ltd.) MC-ICP-MS and Sr isotopic ratios on a Thermo-Finnigan Triton TIMS (Finnigan) where the accuracy of sample analyses was controlled by repeat analyses of international reference materials and standards and duplicate and replicate analyses of samples. Instrumental mass fractionation was corrected using natural stable isotope ratios and normalized on-line to ratios of 146Nd/144Nd=0.7219, 86Sr/88Sr=0.1194, and 179Hf/177Hf=0.7325 for Nd, Sr, and Hf isotope analyses, respectively. External reproducibility is established based on repeated analyses of full procedural duplicates and is less than 87 ppm for Nd, 30 ppm for Sr, and 32 ppm for Hf. Total procedural blanks contained 0.08 ng  Table 4.1 (continued): Trace element concentrations of Northwest Hawaiian Ridge basaltsVolcano Gardner Gardner Gardner Mokumanamana Keoea Twin Banks W. N hoa N hoa N hoa N hoa N hoa Middle BankSample 76-6-7F 84-40E 84-39B NEC-3A 72-49A P5-688-1 76-9-11 NIH-D4 NIH-W-11-1 NIH-F9 NIH-D-1-2 303-04Sc 27.5 36.1 33.7 26.6 32.5 26.7 20.2 24.4 17.0 19.2 21.0 35.0V 233 312 344 221 311 257 239 221 166 180 225 337Cr 1012 886 92 1440 253 976 1026 770 1160 996 815 532Co 72.7 51.8 36.0 88.3 46.7 65.7 257.1 62.5 105.0 91.7 72.2 56.0Ni 693 281 50 772 101 662 774 576 1425 1337 707 193Zn 100 103 149 138 124 123 173 106 113 106 110 130Ga 16 20 24 15 22 16 17 18 12 13 18 18Cs 0.039 0.171 0.560 0.037 0.137 0.152 0.110 0.036 0.024 0.017 0.114  - Rb 0.9 5.5 6.1 5.0 3.8 4.8 7.2 3.5 1.8 1.7 11.6 3.0Sr 219 294 345 225 353 305 410 348 280 266 495 320Ba 60 46 196 108 44 24 145 56 66 55 166 37Y 21.80 30.97 38.06 47.15 31.43 26.80 34.18 25.72 16.58 18.66 26.09 29.00Zr 94 126 153 146 166 110 179 135 114 109 180 139Hf 2.61 3.46 4.32 3.89 4.58 2.99 4.45 3.60 2.96 2.85 4.54 3.70Nb 9.38 11.81 15.93 16.25 15.09 9.62 19.29 11.66 12.44 10.47 20.66 11.80Ta 0.56 0.71 0.96 0.96 0.91 0.46 1.09 0.68 0.76 0.65 1.14 0.90W 0.58 0.56 0.10 0.28 0.43 0.18 1.61 0.13 0.73 0.61 0.36  - Li 6.07 8.89 23.96 4.02 12.97 5.95 12.96 3.30 4.33 3.48 3.76  - La 7.27 9.92 14.10 13.42 11.94 8.91 24.68 9.54 9.92 9.01 19.29 11.90Ce 18.85 24.65 30.56 35.68 30.51 20.03 62.80 25.87 25.68 23.29 46.63 27.30Pr 2.66 3.51 4.71 5.10 4.44 3.03 6.85 3.71 3.42 3.04 6.29 4.10Nd 13.06 17.23 23.28 25.82 22.01 15.35 31.50 17.76 15.94 14.58 27.70 18.60Sm 3.65 4.79 6.34 6.96 6.26 4.32 7.62 4.92 4.11 3.90 6.45 5.30Eu 1.31 1.61 2.10 2.32 2.13 1.43 2.51 1.75 1.31 1.26 2.07 1.70Gd 4.04 5.35 7.13 8.46 6.99 4.75 8.23 5.68 4.16 4.11 6.49 5.20Tb 0.68 0.86 1.13 1.28 1.06 0.73 1.19 0.86 0.62 0.63 0.90 0.90Dy 3.85 5.04 6.58 7.57 6.29 4.23 6.80 4.82 3.48 3.51 4.99 5.20Ho 0.73 0.99 1.30 1.47 1.11 0.81 1.28 0.87 0.61 0.65 0.86 1.00Er 2.03 2.66 3.48 3.99 2.80 2.22 3.26 2.24 1.55 1.72 2.24 2.50Tm 0.27 0.37 0.49 0.56 0.38 0.29 0.45 0.29 0.21 0.23 0.28 0.30Yb 1.59 2.22 2.91 3.27 2.33 1.69 2.70 1.67 1.18 1.31 1.61 2.00Lu 0.248 0.321 0.423 0.481 0.319 0.242 0.386 0.239 0.166 0.179 0.220 0.300Pb 0.61 0.75 1.91 1.31 0.98 0.71 66.21 0.78 0.78 0.74 2.53 0.90Th 0.53 0.64 0.93 0.91 0.85 0.55 1.38 0.70 0.74 0.73 1.16 1.00U 0.196 0.264 0.287 0.284 1.041 0.854 0.694 0.211 0.246 0.153 0.429 0.2001All abundances in ppm. 103Nd, 0.2 ng Sr, and 0.003 ng Hf, negligible compared to the amount of Nd, Sr, and Hf present in samples (on average ~70-90 ng Nd, 0.04 mg Sr, and ~60-80 ng Hf). For the duration of this study, the measured isotope ratios for standards were as follows: 143Nd/144Nd=0.512071 ± 0.000034 for Nd standard JNdi (n=63), 176Hf/177Hf=0.282164 ± 0.000031 for Hf standard JMC 475 (n=99), and 87Sr/86Sr=0.710243 ± 0.000020 for Sr standard SRM 987 (n=19). Repeated analyses of U.S. Geological Survey (USGS) reference material BHVO-2, a Hawaiian basalt, resulted in average isotope compositions of 87Sr/86Sr=0.703466 ± 0.000011 (2SD, n=2), 143Nd/144Nd=0.512991 ± 0.000016 (2SD, n=3), and 176Hf/177Hf=0.283103 ± 0.000001 (2SD, n=3). Analysis of in-house PCIGR reference material Kil-93 sampled from the 1993 eruption of Kīlauea volcano yielded average isotope compositions of 87Sr/86Sr=0.703570 ± 0.000087 (2SD, n=4), 143Nd/144Nd=0.512974 ± 0.000008 (2SD, n=4), and 176Hf/177Hf=0.283101 ± 0.000011 (2SD, n=4). The isotope ratios measured on both of these reference materials are within the range of published values (Weis et al., 2005, 2007; Nobre Silva et al., 2013a; Fourny et al., 2016). Trace element concentrations of samples were measured at PCIGR using a Thermo Finnigan Element2 high-resolution ICP-MS and following the method reported in Carpentier et al. (2013) and Fourny et al. (2016). Powdered samples were digested in a mixture of 10:1 HNO3:HF with no prior leaching. After digestion, samples were fluxed in 6N HCl to eliminate insoluble fluorides, diluted 5000 times, and analyzed on the Element2 using a 1 ppb In spike as an internal standard and the USGS basaltic standard BCR-2 for external calibration. BHVO-2 was analyzed with all samples, and measured concentration values were within two standard deviations of published values for all elements (Schudel et al., 2015; Fourny et al., 2016). Full procedural duplicates and replicate measurements showed excellent agreement, and most elements were analyzed with one relative standard deviation (RSD) of less than 5%.  To place these results in the context of the larger Hawaiian volcano dataset, existing 104 Table 4.2: Sr-Nd-Hf isotopic compositions of Northwest Hawaiian Ridge basaltsA-53-1 Yuryaku 3541 46.9 1 0.703510 8 0.02 0.70348 0.513065 6 0.30 0.51301 8.4 0.283118 6 0.07 0.28311 13.0A-55-1 Daikakuji 3520 46.8 1 0.703860 8 0.03 0.70380 0.512920 6 0.26 0.51287 5.8 0.283009 8 0.05 0.28300 9.2A-55-2 Daikakuji 3520 46.8 1 0.703862 7 0.04 0.70378 0.512917 6 0.25 0.51287 5.7 0.283011 6 0.04 0.28301 9.4A-55-4 Daikakuji 3520 46.8 1 0.703862 8 0.03 0.70380 0.512931 8 0.25 0.51288 6.0 0.283011 6 0.04 0.28301 9.384-28E Unnamed 2801 31.7 6 0.703639 8 0.03 0.70360 0.513038 7 0.26 0.51300 8.0 0.283118 5 0.06 0.28311 12.884-30G Academician Berg 2608 31.0 6 0.703619 10 0.02 0.70359 0.513047 8 0.27 0.51301 8.1 0.283124 6 0.06 0.28312 13.0R-1316 Midway 2447 27.6 1 0.703644 8 0.01 0.70363 0.513059 7 0.28 0.51303 8.3 0.283109 5 0.06 0.28310 12.4R-1295.6 Midway 2447 27.6 1 0.703575 10 0.01 0.70357 0.513061 7 0.28 0.51303 8.3 0.283111 6 0.05 0.28311 12.5P5-524-42 Pioneer 1998 28.0 7 0.703619 8 0.02 0.70360 0.512975 9 0.26 0.51295 6.7 0.283072 3 0.06 0.28307 11.176-5-4DD Northampton 1846 26.6 3 0.703630 8 0.02 0.70362 0.513004 8 0.26 0.51298 7.3 0.283101 5 0.07 0.28310 12.076-5-4A Northampton 1846 26.6 3 0.703625 8 0.02 0.70361 0.513004 8 0.27 0.51298 7.2 0.283101 4 0.07 0.28310 12.076-6-7F Gardner 1449 12.3 4 0.703511 8 0.00 0.70351 0.513035 9 0.28 0.51302 7.8 0.283145 5 0.09 0.28314 13.484-40E Gardner 1449 12.3 4 0.703385 8 0.02 0.70338 0.513065 7 0.28 0.51305 8.4 0.283186 5 0.09 0.28318 14.884-39B Gardner 1449 12.3 4 0.703449 10 0.02 0.70344 0.513039 7 0.27 0.51303 7.9 0.283169 5 0.10 0.28317 14.2NEC-3A Mokumanamana 1080 10.0 2 0.703486 8 0.02 0.70348 0.513012 7 0.27 0.51300 7.4 0.283089 3 0.12 0.28309 11.472-49A Keoea 963 11.0 7 0.703492 8 0.01 0.70349 0.513039 8 0.28 0.51303 7.9 0.283118 4 0.07 0.28312 12.4P5-688-1 Twin Banks 920 9.6 4 0.703494 8 0.02 0.70349 0.513051 7 0.28 0.51304 8.1 0.283130 5 0.08 0.28313 12.876-9-11 West N hoa 825 8.8 1 0.703593 8 0.02 0.70359 0.512939 7 0.24 0.51293 5.9 0.283080 5 0.09 0.28308 11.0NIH-D4 N hoa 794 7.0 2 0.703496 8 0.01 0.70349 0.513024 7 0.28 0.51301 7.6 0.283120 5 0.07 0.28312 12.4NIH-W-11-1 N hoa 794 7.0 2 0.703499 8 0.01 0.70350 0.513016 7 0.26 0.51301 7.4 0.283115 5 0.06 0.28311 12.3NIH-F9 N hoa 794 7.0 2 0.703546 10 0.01 0.70354 0.513013 7 0.27 0.51300 7.3 0.283123 6 0.06 0.28312 12.6NIH-D-1-2 N hoa 794 7.0 2 0.703587 8 0.02 0.70358 0.512959 7 0.23 0.51295 6.3 0.283052 5 0.05 0.28305 10.1303-04 Middle Bank 702 6.5 7 0.703691 10 0.01 0.70369 0.512898 6 0.28 0.51289 5.1 0.283087 6 0.08 0.28309 11.31-7References to ages: 1- O'Connor et al., 2013; 2- Dalrymple et al., 1974; 3- Dalrymple et al., 1981; 4- Garcia et al., 1987; 5- Garcia et al., 2010; 6- Sharp and Clague, 2006; 7- Age inferred by interpolating between existing ages and assuming a linear age progression of NWHR volcanoes. aTrace element concentrations (in ppm) were measured on unleached whole rock powders using a Thermo Finnigan Element2 high resolution ICP-MS at the PCIGR at UBC. USGS Reference Material BCR-2 was used for external calibration and a 1 ppb In spike was used as an internal standard. RSDs are less than 5% for all elements and measured concentrations of USGS Standard BHVO-2 are within 2SD of published values (Fourny et al., 2016).  bHf and Nd isotopes were analyzed on a Nu Plasma II MC-ICP-MS and Sr isotopes were analysed on a Thermo-Finnigan Triton TIMS at the PCIGR. The 2 SE is the absolute error of the individual sample analysis (internal error) and applies to the last decimal place(s). εHf ccεNd and εHf are calculated using both sample and CHUR 143Nd/144Nd and 176Hf/177Hf values age corrected to the age of eruption (Nobre Silva et al., 2013a; Harrison et al., 2017). 143Nd/144Ndi εNd c 176Hf/177Hfm 2 SE Lu/Hf 176Hf/177Hfi2 SE Rb/Sr 87Sr/86Sri 143Nd/144Ndm 2 SE Sm/NdSample SeamountDistance from K lauea Age (Ma)Age Ref.87Sr/86Srm105high-precision radiogenic isotopic data were compiled for Hawaiian Island basalts, along with supporting major and trace element data. All isotope data from this study and data compiled from the literature are normalized to the same standard values to ensure comparability (Weis et al., 2011). Nd isotopes are normalized to a 143Nd/144Nd La Jolla value of 0.511858; Sr isotopes to a 87Sr/86Sr SRM 987 value of 0.710248; Hf to a 176Hf/177Hf JMC 475 value of 0.282160; and Pb isotopes to SRM 981 values of 16.9405 for 206Pb/204Pb, 15.4963 for 207Pb/204Pb, and 36.7219 for 208Pb/204Pb (Weis et al., 2005, 2006, 2007). 4.4. ALTERATION OF SAMPLES AND AGE CORRECTION OF ISOTOPIC RATIOS Selection of NWHR samples was limited by availability and care was taken to select the least altered samples for geochemical analysis as alteration of basaltic rocks exposed to seawater can modify the original magmatic elemental and isotopic compositions (Staudigel et al., 1995, 1996). Many of the trace element systematics of NWHR samples have been disturbed by both subaerial and submarine post-eruption alteration. Fluid-mobile elements, such as Si, Mg, Rb, K, Na, Sr, U, Ba, and Pb, are particularly susceptible to this process and concentrations of these elements in unleached powders cannot be considered primary (Staudigel et al., 1995; Révillon et al., 2007). Loss-on-ignition is a quantification of the volatiles (H2O, CO2, F, Cl, S) present in a rock, used here as an estimation of the degree of hydration or calcitization mafic minerals have undergone (Lechler and Desilets, 1987). Loss-on-ignition values of analyzed NWHR rocks are within the range of mild to moderately altered (LOI <2-4 wt%; Midway volcano is strongly altered with LOI ≈ 10-12 wt%). Because of the rarity of available Midway samples and the relatively low resolution of sampling along the NWHR near Midway, these samples are included here. For isotopic analyses, systematic leaching removes the effect of alteration (Nobre Silva et al., 2009, 2010). The major and trace element analyses were made on unleached rock powders and thus caution should be used in interpreting the compositions of more mobile elements. 106 The degree of alteration can be assessed using plots of fluid mobile elements (U, Sr, Rb, K) versus non-mobile elements (Nb, Ta, Zr, Th) (Figure 4.2). Samples undisturbed by post-eruption alteration show tight, linear correlations between elements with similar partition coefficients as they are not fractionated through partial melting and crystal fractionation processes (Arevalo, 2010). This relationship is apparent when immobile elements such as Zr and Nb are plotted (Figure 4.2a). Conversely, a plot of a mobile element (U) versus an immobile element (Nb) can be used to assess the degree to which a sample has GardnerKeoeaTwin BanksWest NihoaNihoaMiddle BankMokumananaYuryakuUnnamedA. BergMidwayNorthamptonPioneerDiakakuji200400600800100012000 1 2 3K/RbK2O/P2O5Ko‘olauMaunaKeaLō‘ihiMaunaLoa0.00.20.40.60.81.010 20 30UNbU addition during weatheringr2 = 0.2350100150200250300 Zrr2 = 0.89(b)(c)(a)Samples with some alterationFigure 4.2: Alteration indices of NWHR shield basalts Panels (a) and (b) show the distribution of an immobile trace element (Nb) versus another immobile trace element (Zr, plot a) and a mobile trace element (U, plot b). The tight trend in plot (a) indicates that these samples are related by a similar source with varying degrees of partial melting or crystal fractionation. The scatter exhibited in plot (b) identifies the samples that have undergone chemical weathering. Plot (c) is the classic alteration plot of Frey et al. (1991) where fresh samples are characterized by K2O/P2O5 > 1 and K/Rb < 650. The fields for Mauna Kea, Mauna Loa, Lō‘ihi, and Ko‘olau shield stage basalts are shown for comparison.107been altered and identifies NWHR samples that have experienced U gain during alteration (Figure 4.2b). The alteration plot of Frey et al. (1994) is also useful to assess the degree of alteration of samples (Figure 4.2c). Fresh Hawaiian shield basalts have K/Rb=308-630 and K2O/P2O5=1.5-2.2 (Figure 4.2c) (Wright, 1971). Most NWHR basalts that experienced U gain also experienced some degree of K loss during tropical and/or seawater weathering and P gain likely from bird guano contamination (Harris, 1985). This mild to moderate alteration accounts for the scatter observed in the alteration susceptible major and trace elements (U, K, Rb, Si, P, Na, Fe, Cs, Ba, Pb). The measured isotopic ratios of NWHR lavas require age corrections to account for in-situ decay since eruption. Knowledge of parent and daughter element concentrations is necessary on unleached powders. However, the accuracy of these measurements is limited by the extensive post-eruption alteration that many of the old submarine samples have experienced. U-Pb and Rb-Sr decay systems are most susceptible to post-eruption perturbation of the isotopic systematics (Hart et al., 1974; Staudigel et al., 1995; Nobre-Silva et al., 2009). Rare earth elements, such as those involved in the Lu-Hf and Sm-Nd isotopic decay systems, are much less sensitive to these processes as these elements are not fluid mobile (Staudigel et al., 1995; Nobre-Silva et al., 2010). It is possible, however, to disturb primary 143Nd/144Nd by contamination from Fe-Mn precipitates in vesicles or on the exterior of rocks (Staudigel et al., 1995; Regelous et al., 2003). To minimize this problem, only the most unaltered interiors of samples were crushed using the clean crushing setup at PCIGR and all samples for isotopic analysis were sequentially leached (Nobre Silva et al., 2009, 2010). Parent-daughter ratios were measured on unleached sample powders: for the Rb-Sr, Lu-Hf, and Sm-Nd isotopic systems, the measured parent-daughter ratios are consistent with typical values of global OIB, and thus we consider removing visible alteration during sample preparation sufficient (Nobre Silva et al., 2013b).1084.5. RESULTS One of the main goals of analyzing NWHR samples is to determine the emergence point, prevalence, and longevity of the Loa geochemical trend in the Hawaiian mantle plume. With this goal in mind, lavas are classified as having geochemical affinities with the Loa or Kea trends. The most enriched Loa trend lavas from Ko‘olau volcano and Lāna‘i and Kaho‘olawe islands are typically characterized by low 206Pb/204Pb, 143Nd/144Nd, 176Hf/177Hf, Th/La, Nb/La, CaO, CaO/Al2O3, TiO2, Na2O, K2O, and by high 87Sr/86Sr, Sr/Nd, Zr/Nb, SiO2, Na2O/TiO2, and a high ratio of thorogenic to uranogenic Pb (i.e., radiogenic Pb [208Pb*/206Pb*]) (Hauri et al., 1996; Jackson et al., 2012; Frey et al., 2016). We explore the geochemical characteristics of the NWHR lavas below, and compare them with those from Hawaiian Island volcanoes. Northwest Hawaiian Ridge isotopic compositions are reported in Table 4.2 and trace element concentrations in Table 4.1.4.5.1 Major Elements MgO content varies widely in the analyzed NWHR samples (6-23 weight percent; Garcia et al., 2015; Tree, 2016). Most samples are basalts and a few are picro-basalts (Figure 4.3). All samples are tholeiitic or transitional (i.e. plot near the TAS alkalic-tholeiitic divide line of Macdonald and Katsura, 1964) (Figure 4.3) in composition and therefore belong to the shield stage. NWHR basalts are comparable to younger Hawaiian Island lavas in major element oxide plots for samples with LOI < 3 weight percent. Samples from Midway consistently plot off the fractionation trends for other NWHR samples due to a higher degree of alteration (Figure 4.4). The trends of oxides versus MgO are consistent with olivine fractionation from an average Loa-type parental magma. Aluminum contents follow tight linear arrays, consistent with the near-uniform aluminum oxide values at a given MgO in Hawaiian Island lavas (Rhodes, 2015; 2016). Oxides such as SiO2, CaO, and TiO2 vary between volcanoes, whereas Al2O3 is constant in Hawaiian tholeiitic basalt, and K2O, Na2O, and P2O5 are 109susceptible to secondary alteration (Rhodes, 2015). Contents of P2O5 can be anomalously high from contamination from bird droppings, which explains P2O5 values up to 1.38 weight percent at Twin Banks and high values at other NWHR volcanoes (Figure 4.4) (Harris, 1985; Tree, 2016). The major element compositional variation of NWHR basalts is primarily controlled by olivine fractionation (Figure 4.4). Magmas did not fractionate to low enough MgO values to reach clinopyroxene fraction, which typically occurs at < 7 wt% MgO. The lowest Sc concentrations of any NWHR basalt are at Daikakuji, and only these samples and one from Mokumanamana contain rare phenocrysts of clinopyroxene (Tree, 2016). When MgO is corrected for olivine fractionation to 16 weight percent (considered a potential primary composition for Hawaiian basalts; Jackson et al., 2012), NWHR samples show major element variations caused by source differences (Figure 4.5) (Pietruszka et al., 2013). The majority of NWHR basalts studied here overlap with Mauna Kea compositions in MgO-normalized CaO/Al2O3 versus Na2O/TiO2 and in MgO-normalized SiO2 versus CaO (Figure 4.5). West Nīhoa, Daikakuji, Unnamed seamount, and Academician Berg all overlap012345640 45 50 55Na2O+K2O (wt. %)SiO  (wt. %)2BasaltPicro-BasaltHawaiite Basaltic AndesiteAlkalicTholeiiticYuryakuDaikakujiUnnamedAcademician BergMidwayPioneerNorthamptonGardnerMokumanamanaKeoeaTwin BanksWest NīhoaNīhoaMiddle BankFigure 4.3: Total alkali versus silica plot of NWHR shield basaltsNWHR samples range in composition between picro-basalt to basalt, and all samples are very near or below the tholeiitic-alkalic divide line of Macdonald and Katsura (1964).110with shield stage Ni‘ihau compositions (Cousens and Clague, 2015), a Kea trend volcano, in CaO/Al2O3 versus Na2O/TiO2; aside from West Nīhoa, these are all a part of the segment of the NWHR chain near the bend. The high normalized SiO2 and low CaO of Dakakuji is characteristic of Loa-trend volcanoes. The samples with low normalized SiO2 that plot GardnerKeoeaTwin BanksWest NihoaNihoaMiddle BankMokumananaYuryakuUnnamedA. BergMidwayNorthamptonPioneerDiakakuji1.01.52.02.53.03.5Na2O!"#$"%$2025303540Sc020040060080010001200140016000 5 10 15 20 25 30NiMgO4681012CaO57911131517 Al2O342444648505254SiO20.51.01.52.02.53.03.5TiO2891011121314150 5 10 15 20 25 30Fe2O3MgO0.00.10.20.30.40.50 5 10 15 20 25 30P2O5MgOMaunaKeaMaunaKeaMaunaKeaMaunaKeaMaunaKeaMaunaKeaMaunaKeaMaunaKeaMaunaKeaMaunaLoaMaunaLoaMaunaLoaMaunaLoaMaunaLoaMaunaLoaMaunaLoaMaunaLoaMaunaLoaFigure 4.4: MgO variation plots of selected NWHR shield basalt major and minor elementsFor comparison, the fields of Mauna Loa and Mauna Kea shield-stage basalts are shown and the Mauna Loa olivine control regression lines from Rhodes (2015) are included in the SiO2, Al2O3, CaO, and TiO2 plots. The major element patterns of NWHR volcanoes are controlled by olivine fractionation. Midway is consistently the most altered sample and as a result, plots off of the fractionation trends.111outside of the fields defined by the Hawaiian Islands likely lost silica during alteration. 4.5.2 Trace Elements Fluid-mobile element (Cs, Rb, K, U, Sr, Ba) concentrations vary in NWHR basalts due to mild to moderate alteration (Figure 4.6). Immobile elements are undisturbed as all samples display similar rare earth element and extended trace element patterns (Figures 4.6, 4.7).   Most NWHR lavas exhibit trace element characteristics similar to typical Hawaiian shield tholeiites, such as enrichments in Ba and Sr and depletion of Pb and Th (Pietruszka et al., 2013) (Figure 4.6). Basalts from Daikakuji, followed by Nīhoa, West Nīhoa, Academician Berg, and Unnamed Seamount, contain the highest levels of Sr (348-610 ppm). The heavy rare earth elements (HREE) are similar concentrations along the entire NWHR due to the presence of residual garnet in the source of these lavas (Rhodes, 2016) (Figure  GardnerKeoeaTwin BanksWest NihoaNihoaMiddle BankMokumananaYuryakuUnnamedA. BergMidwayNorthamptonPioneerDiakakujiW. Maui0.50.60.70.80.91.01.10.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7CaO/Al2O3 (16 wt % MgO-Normalized)Na2O/TiO2 (16 wt % MgO-Normalized)Lō‘ihiMaunaKeaKo‘olauMaunaLoaKīlaueaNi‘ihauMidway6789101143 45 47 49 51CaO (16 wt % MgO-Normalized)SiO2 (16 wt % MgO-Normalized)MidwayW. MauiLō‘ihiMaunaKeaKo‘olauMaunaLoaKīlaueaNi‘ihau(a) (b)Figure 4.5: Normalized major element abundances of NWHR and Hawaiian Island shield basaltsMajor element oxide abundances were normalized to 16 wt% MgO by adding or subtracting olivine in 0.01 wt% increments following Jackson et al. (2012) and Langmuir et al. (1992). We use an olivine-melt Kd of 0.3 (Roeder and Emslie, 1970; Ford et al., 1983) and assuming 10% Fe3+. Also shown for comparison are shield samples from Lō‘ihi, Mauna Loa, Mauna Kea, Kīlauea, West Maui, and Ni‘ihau, all corrected to 16 wt% MgO (Jackson et al., 2012).112110100110100 CsRbBaTh U NbTa K LaCePbPrNdSrZrHfSmEuTiGdTbDyHoY ErYb LuCsRbBaThU NbTa K LaCePbPrNdSrZrHfSmEuTiGdTbDyHoY ErYb LuYuryakuDaikakujiUnnamedAcademician BergMidwaySample/Primitive Mantle110100PioneerNorthamptonGardnerKeoeaTwin BanksWest NīhoaMiddle BankNīhoaMokumanamana~47-27 MaMauna KeaMauna KeaMauna Kea~27-11 Ma~11-7 MaFigure 4.6: Primitive mantle-normalized extended trace element patterns of NWHR shield basaltsNormalizing values are from McDonough and Sun (1995) and the grey field in the background is the range of compositions from the shield-stage of Mauna Kea. Element incompatibility increases towards the left.1134.7). Most NWHR samples are more enriched in REE than those from Mauna Kea (shown by grey field in Figure 4.7), except for samples from Gardner, Northampton, Twin Banks, and Nīhoa. Basalts from Daikakuji, West Nīhoa, and Nīhoa (NIH-D-1-2) show the most en-richment in REE and incompatible trace elements. The highly enriched LREE of these lavas are higher than is typically observed in Hawaiian tholeiites and are more characteristic of the postshield and rejuvenated volcanic stages (Frey et al., 1994, 2016; Hanano et al., 2010; Pietruszka et al., 2013). However, the low degrees of melting of these lavas (supported by the highest La/Yb ratios of the entire Hawaiian-Emperor chain; Harrison et al., 2017), would result in higher concentrations of REE. Ni and Cr concentrations show a wide range  1010010100La Ce Pr Nd PmSm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd PmSm Eu Gd Tb Dy Ho Er Tm Yb Lu10100Sample/C1 ChondriteYuryakuDaikakujiUnnamedA. BergMidwayPioneerNorthamptonGardnerKeoeaTwin BanksWest NīhoaMiddle BankNīhoaMokumanamanaMauna KeaMauna KeaMauna KeaFigure 4.7: Chondrite-normalized rare earth element patterns of NWHR shield basaltsNormalizing values are from McDonough and Sun (1995) and the grey field in the background is the range of compositions from the shield-stage of Mauna Kea.114of variation in NWHR lavas (Ni 50-1425 ppm; Cr 92-1440 ppm) (Figure 4.4). MgO-rich samples with high Ni and Cr tend to be olivine-rich, as observed in thin section (Tree, 2016). Samples with high Ni contents (e.g., Gardner, Nīhoa, West Nīhoa, Twin Banks) generally YuryakuUnnamedA. BergMidwayNorthamptonPioneerDiakakujiGardnerKeoeaTwin BanksWest NihoaNihoaMiddle BankMokumanana0.70.80.91.01.11.21.31.45 10 15 20La/NbZr/NbLō‘ihiMaunaKeaKo‘olauMaunaLoaKīlaueaNi‘ihau10152025303540455055 Sr/NbLō‘ihi MaunaKeaKo‘olauMaunaLoaKīlaueaNi‘ihau0.400.450.500.550.600.650.700.750.800.85(Th/Ce)PMW. MauiLō‘ihiMaunaKeaKo‘olauMaunaLoaNi‘ihauKohala0.30.81.31.82.32.80.92 0.94 0.96 0.98(Ba/Th)PM208Pb*/206Pb*W. MauiLō‘ihiMaunaKeaKo‘olauMaunaLoaNi‘ihauKaho‘olaweHanaRidgeKohala(a)(b) (d)(c)Figure 4.8: Trace element ratio variation plots of NWHR shield basaltsTrace element ratios versus Zr/Nb (a, b) and radiogenic Pb (c, d). The Pb isotopic ratios of NWHR samples are from Harrison et al. (2017). The Th/Ce and Ba/Th in plots (c) and (d) are normalized to primitive mantle values of McDonough and Sun (1995). Fields from Mauna Loa, Ko‘olau, Ni‘ihau, Lō‘ihi, Kīlauea, Mauna Kea, West Maui, Kohala and Hana Ridge shield-stage basalts are shown for comparison.115erupt at the locations of higher magmatic flux along the NWHR and are only found on the youngest half of the NWHR (<12 Ma; Garcia et al., 1987). The oldest NWHR volcanoes (Yuryaku, Daikakuji, Unnamed, Academician Berg, Midway) are more enriched in incompatible trace elements than the younger NWHR volcanoes (Figure 4.8). West Nīhoa, Daikakuji, and Nīhoa (NIH-D-1-2) exhibit the highest (La/Sm)PM, with Academician Berg, the rest of Nīhoa, and Unnamed also exhibiting higher-than Mauna Kea (La/Sm)PM. Daikakuji, Academician Berg, Unnamed Seamount, Middle Bank, and Pioneer all have higher (Th/Ce)PM than the rest of the NWHR (Figure 4.8). The higher Sr/Nb and La/Nb of West Nīhoa, Nīhoa (NIH-D-1-2), Daikakuji, Twin Banks, and Middle Bank illustrate the higher Sr and La concentrations in these samples in comparison to other NWHR samples. Most other NWHR basalts have Sr/Nb values that overlap with Kīlauea compositions.  4.5.3 Nd-Sr-Hf Isotopic Compositions The new isotopic data extends the range of Nd and Sr isotopic compositions found in NWHR basalts from previous studies (Lanphere et al., 1980; Basu and Faggart, 1996; Regelous et al., 2003). When compared to the Hawaiian Islands, Sr and Nd isotopic compositions are within previously defined literature ranges. The Sr isotope ratios of the new analyses are less radiogenic than previous analyses as a result of the acid leaching procedure that removes the secondary radiogenic seawater 87Sr/86Sr isotope signature (Staudigel et al., 1995, 1996; Nobre Silva et al., 2009; 2010). Nd isotopes are not typically affected by secondary alteration due to the immobile geochemical behavior of the rare earth elements; previous NWHR data are within the range of new 143Nd/144Nd analyses. This is the first published Hf isotope data for the NWHR, and extends the range of epsilon Hf defined by Hawaiian Island shield-stage tholeiites to higher values.  Nd-Sr-Hf isotopes show coherent trends when plotted versus distance from Kīlauea (a proxy for time) (Figure 4.9) that follow comparable trends as those observed in new1160.70320.70340.70360.70380.70400.70420.704487Sr/86Sr345678ɛNd5791113150 1,000 2,000 3,000 4,000ɛHfEndmembersKea trend volcanoLit. data1,2Loa trend volcanoGardnerKeoeaTwin BanksWest NihoaNihoaMiddle BankMokumananaYuryakuUnnamedA. BergMidwayNorthamptonPioneerDiakakujiLKLKLKEDDDMKDistance from Kīlauea (km)EMKEMK(a)(b)(c)KK117NWHR Pb isotopic data (Harrison et al., 2017). Sr isotope ratios decrease for most of the NWHR from ~47 to ~11 Ma before beginning to increase at West Nīhoa and Nīhoa. One of the oldest NWHR volcanoes, Daikakuji, has the most radiogenic Sr isotope ratios (87Sr/86Sr = 0.70380) that corresponds to an apparent maxima in Sr isotopic composition of sampled NWHR lavas, which decrease from that point until West Nīhoa. The least radiogenic 87Sr/86Sr (0.70338) measured is from Gardner Seamount. Nd isotope ratios, after an unradiogenic minima at Daikakuji (εNd = +5.7, overlapping with Mauna Loa values) are Kea-like until after Midway where Nd isotope ratios begin to decrease. West Nīhoa, Middle Bank and Nīhoa (NIH-D-1-2) also approach low epsilon Nd values that are within the Loa trend field. Hf isotopes do not vary much and are Kea-like for most of the NWHR. They show a minima at Daikakuji, a maxima at Gardner and a general trend to lower values after Midway. Daikakuji exhibits the least radiogenic Hf isotopic signature along the NWHR, which at ɛHf = +9.3 overlaps with compositions observed in the most enriched Loa volcano, i.e., Ko‘olau Makapu‘u (Tanaka et al., 2008). Hf isotopic ratios are highest at Gardner Volcano, and have the highest values of any Hawaiian shield-stage basalt younger than the Hawaiian-Emperor bend.  Analyses of basalts from the NWHR and the Hawaiian Islands form negative correlations in epsilon Nd and epsilon Hf versus 87Sr/86Sr plots and positive correlations between Hf and Nd isotopes, which is typical for oceanic basalts (White, 2015) (Figure 4.10b). Most NWHR basalts overlap with Mauna Kea, West Maui, Hana Ridge, and Kīlauea Figure 4.9: Isotopic ratios of Sr, Nd, and Hf versus distance from Kīlauea for NWHR shield basalts(Figure above); 2 sigma error bars are smaller than data points). (a) 87Sr/86Sr (b) epsilon Nd and (c) epsilon Hf versus distance from Kīlauea. For comparison, all Hawaiian Island shield-stage isotopic analyses are presented, with Loa trend volcanoes indicated by blue circles and Kea trend volcanoes by red circles (Data sources and standard normalization values are the same as in Figure 4.10). White diamonds show literature data of NWHR basalts (Basu and Faggart, 1996; Lanphere et al., 1980; Regelous et al., 2003). Circles marked with capital letters designate the location of Hawaiian end-member components (Tanaka et al., 2002; E = EMK, K = KEA, L = Lō‘ihi, D = DMK). A proposed adjustment of the Kea end-member component to ɛNd = 8.8 and ɛHf = 15.4 to encompass the full range of Hf and Nd observed in NWHR lavas is marked with the red circle marked “K.”118  5791113150 2 4 6 8 10KohalaW. Ka‘enaNWHR ArrayOIB ArrayNew Hawaiian ArrayW. MauiE. MolokaiW. MolokaiMaunaKeaKo‘olauLana‘iKaho‘olaweKaua‘iHualalaiMaunaLoaPenguin BankLō‘ihiLOIHIDMKKEAɛNdɛHf10.70350.7030 0.7040 0.704523456789 ɛNd87Sr/86SrLō‘ihiKohalaW. Ka‘enaKīlaueaHana RidgeW. MauiE. MolokaiW. MolokaiNi‘ihauMaunaKeaKo‘olauLana‘iKaho‘olaweKaua‘iHualalaiMaunaLoaPenguin BankEMKLOIHIDMKKEAGardnerKeoeaTwin BanksWest NihoaNihoaMiddle BankMokumananaYuryakuUnnamedA. BergMidwayNorthamptonLit. data1,2PioneerDiakakuji(a)(b)56780.7033 0.7036 0.703987Sr/86SrɛNd91011121314155 6 7 8 9ɛNdɛHfNew KEANew KEAOriginal Hawaiian ArrayKoolauɛNd=-2EMK ɛNd=0.23119isotopic compositions near the Kea depleted end-member side of the Hawaiian arrays (Figures 4.10, 4.11). Daikakuji, Middle Bank, West Nīhoa, and Nīhoa (NIH-D-1-2) consistently overlap with Loa trend volcano fields such as W. Moloka‘i, Mauna Loa, Kaua‘i, West Ka‘ena, and Hualālai. No NWHR volcanoes are similar in composition to Lō’ihi-type compositions. In epsilon Hf versus epsilon Nd, the NWHR forms an array that expands the range previously defined by Hawaiian Island shield basalts towards higher Hf isotope ratios at a given Nd isotope ratio on the high Hf and Nd end of the array (Figure 4.10b). The NWHR array is not parallel to the Hawaiian array, indicating a transition in mantle sources from a lower time-integrated Lu/Hf mantle source during the formation of the NWHR to a higher time-integrated Lu/Hf ratio during formation the Hawaiian Islands (Figure 4.10b). The NWHR shield basalts fit the highly hyperbolic mixing trends defined by the Hawaiian Islands between 87Sr/86Sr and 206Pb/204Pb and epsilon Hf and 208Pb*/206Pb* (Figure 4.11). Most of the NWHR basalts overlap with Hana Ridge, Mauna Kea, Kīlauea,  Figure 4.10: Nd, Sr, and Hf isotopic variation diagrams for NWHR shield basalts(Figure above); 2 sigma error bars are smaller than data points). (a) Epsilon Nd versus 87Sr/86Sr (b) Epsilon Hf versus epsilon Nd for NWHR shield stage basalts. For reference the four Hawaiian end-member components are shown by white circles (KEA, DMK, EMK, and Lō‘ihi, Tanaka et al., 2002, 2008 and Ko‘olau, Mukhopadhyay et al., 2003) and in (b) the OIB array and original Hawaiian array are from Blichert-Toft et al. (1999). The new Hawaiian array is calculated here from an up-to-date compiled database of 452 Hawaiian Island shield-stage samples re-normalized to the same isotopic standards (Weis et al., 2011). The new Hawaiian array has a slope of 1.10 ± 0.04 and a y-intercept of 4.58 ± 0.23 (95% confidence). The NWHR array is a regression of all NWHR shield-stage analyses in this study. White diamonds show literature data of NWHR basalts (Basu and Faggart, 1996; Lanphere et al., 1980; Regelous et al., 2003). An adjustment of the Kea end-member component is proposed to ɛNd = 8.8 and ɛHf = 15.4 to encompass the full range of Hf and Nd observed in NWHR basalts and is marked with a red circle marked “New KEA.” Inset plots show the variation of only NWHR basalts analyzed in this study. Fields for Hawaiian Island volcanoes are shown and have all been re-normalized to the same standard values. Data for these fields are as follows: Lō‘ihi: Garcia et al., 1993, 1995, 1998; Abouchami et al., 2005; Kīlauea: Abouchami et al., 2005; Marske et al., 2007; Pietruszka and Garcia, 1999; Mauna Loa: Weis et al., 2011; Kurz et al., 1995; Abouchami et al., 2000; Marske et al., 2007;  Kurz and Kammer, 1991; Wanless et al., 2006; Mauna Kea: Eisele et al., 2003; Rhodes and Vollinger, 2004; Blichert-Toft et al., 2003; Nobre-Silva et al., 2013a; Bryce et al., 2005; Hualālai: Yamasaki et al., 2009; Kohala: Abouchami et al., 2005; Hofmann et al., 1987; Māhukona: Huang et al., 2009; Garcia et al., 2012; Hāna Ridge: Ren et al., 2005; West Maui: Gaffney et al., 2004; Kahoolawe: Huang and Frey, 2005; Blichert-Toft et al., 1999; Abouchami et al., 2005; West et al., 1987; Lāna‘i: Gaffney et al., 2005; Abouchami et al., 2005; East Moloka‘i: Xu et al., 2005; West Moloka‘i: Xu et al., 2007; Penguin Bank: Xu et al., 2014; Ko‘olau: Fekiacova et al., 2007; Tanaka et al., 2002, 2008; Wai‘anae: Coombs et al., 2004; Van der Zander et al., 2010; West Ka‘ena: Greene et al., 2010; Sinton et al., 2014; Kaua‘i: Garcia et al., 2010; Mukhopadhyay et al., 2003; Cousens and Clague, 2015; Ni‘ihau: Cousens and Clague, 2015.120   GardnerKeoeaTwin BanksWest NihoaNihoaMiddle BankMokumananaYuryakuUnnamedA. BergMidwayNorthamptonLit. data1,2PioneerDiakakuji24681012140.90 0.92 0.94 0.96 0.98208Pb*/206Pb*KohalaW. Ka‘enaW. MauiE. Molokai W. MolokaiMaunaKeaKo‘olauLana‘iKaho‘olaweHualalaiMaunaLoaPenguin BankLō‘ihiEMKLOIHIKEADMKɛHf0.70330.70360.70390.704217.8 18.1 18.4 18.787Sr/86Sr206Pb/204PbLō‘ihiKohalaW. Ka‘enaKīlaueaW. MauiMaunaKeaKo‘olauLana‘iKaho‘olaweKaua‘iMaunaLoaHualalai HanaRidgeNi‘ihauW. MolokaiE. MolokaiPenguin BankKīlaueaEMKLOIHIDMKKEA(a)(b)0.70330.70360.703917.8 18.2 18.687Sr/86Sr206Pb/204Pb812160.92 0.94 0.96 0.98ɛHf208Pb*/206Pb*New KEA121and West Maui compositions in Figure 4.11. Mauna Loa, West Ka‘ena, Hualālai, Ni‘ihau, West Moloka‘i, and Kaua‘i overlap with Daikakuji, Middle Bank, West Nīhoa, and Nīhoa (NIH-D-1-2). Although Daikakuji exhibits the highest values in radiogenic Pb observed on the NWHR and from the Hawaiian Islands, it does not exhibit similarly enriched 87Sr/86Sr and depleted 176Hf/177Hf and 143Nd/144Nd isotopic signatures, indicating a partial decoupling between Pb and the other radiogenic isotopic systems.4.6. DISCUSSION In the following subsections we discuss the spatial distribution and compositional variation of the Loa geochemical trend along the NWHR, with implications for Hawaiian mantle source variation with time.4.6.1 The Spatial Distribution of the Loa Geochemical Trend Along the NWHR: A Logistic Regression Analysis Isotope analyses provides the most robust method for classifying Hawaiian basalts as Loa or Kea trend (Abouchami et al., 2005; Weis et al., 2011). The main goal of this study was to determine whether the Loa geochemical composition is present along the NWHR. To accomplish this goal, a logistic regression model was fit to the compiled and normalized isotopic database of ~733 shield-stage samples of Hawaiian Island basalts (data sources the same as those in Figure 4.10). This model is then used to predict the probability that NWHR basalts are Loa- or Kea-type based on their isotopic signature. This is the first time a statistical model has been used to predict the geochemical affinity of Hawaiian basalts.  Logistic regression uses independent variables (here the isotope ratios 206Pb/204Pb, Figure 4.11: Pb, Sr, and Hf isotopic variation diagrams for NWHR shield basalts(Figure above); 2 sigma error bars are smaller than data points). (a) 87Sr/86Sr versus 206Pb/204Pb (b) Epsilon Hf versus 208Pb*/206Pb* for NWHR shield stage basalts (Pb isotopic analyses for NWHR basalts are from Harrison et al., 2017). For reference the four Hawaiian end-member components are shown by white circles (KEA, DMK, EMK, and Lō‘ihi; Tanaka et al., 2002, 2008). The proposed adjustment of the Kea end-member component to ɛNd = 8.8 is marked with a red circle marked “New KEA.” White diamonds show literature data of NWHR basalts (Basu and Faggart, 1996; Lanphere et al., 1980; Regelous et al., 2003). Inset plots show the variation of only NWHR basalts analyzed in this study. Data sources and standard normalization values are the same as in Figure 4.10 for the Hawaiian Island volcano fields.122207Pb/204Pb, 208Pb/204Pb, 87Sr/86Sr, 143Nd/144Nd, and 176Hf/177Hf) to sort samples into a dichotomous categorical dependent variable (i.e., Loa or Kea trend compositions). This is mathematically achieved by fitting a function that optimizes the odds of classifying data into the correct category by calculating a coefficient for each independent variable based on the weight of that variable in sorting the data (equation 4.1; Hosmer and Lemeshow, 1989 and references therein). Logistic regression does not require that independent variables be normally distributed with equal variances and covariances, or that the independent and dependent variables be linearly correlated. logfor i 1,2, ... , n samples and whereβ0 + β1X1i + β2X2i+ β3X3i + β4X4i LoaLoaPP 1 ==P probability a case is in a catagory=β0 the constant of the equation=β1,2,3, ... ,n coefficient of predictor variable X=_     (4.1) The model can be iteratively optimized by adding and removing independent variables until only the variables that contribute to a response in model fit are included. This can be evaluated for each variable using the Wald test, which assesses whether or not the coefficient that the logistic regression model fitted for each variable is significantly different from zero (Wald, 1943). If a variable fails this test, it may safely be left out of future models without impacting model fit. Once the simplest logistic model is built, the goodness of fit is verified using the likelihood-ratio test, receiver operation characteristic, and area-under-curve estimates and plots (Hosmer and Lemeshow, 1989) (Figure 4.12). The goal is to fit a model that describes the data as well as possible without overfitting, which is a term to describe the outcome when a model is tailored to the noise in the dataset rather than fit to the overall 123population. Potential pitfalls include independent variables that are too highly correlated introducing collinearity problems into the model, or undetected interactions between independent variables. We used R, a free software environment for statistical computing, to calculate the logistic regression model and goodness-of-fit tests. After a parsimonious, well-fitting model is achieved, it was used to predict the Loa or Kea membership of new NWHR samples. The first concern when building a logistic regression model of Hawaiian basalt data is whether the Loa and Kea trends are significantly different populations when all four isotope systems (Pb, Hf, Nd, Sr) are considered simultaneously. In other words, is our dichotomous variable for building the logistic model statistically mutually exclusive? A Hotelling t-squared test was used to determine if the Loa and Kea trends have significantly different mean vectors. The dataset used for this test was a compilation of 362 Hawaiian Island basalts with all isotopic systems (87Sr/86Sr, 143Nd/144Nd, 176Hf/177Hf, 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb) measured on the same sample split and renormalized to the same standard values. The result was a T2 statistic of 840 at six degrees of freedom and a p-value < 0.001. Our F-ratio is 138, significantly above the F critical value at six degrees of freedom (4.28; F =0 when there is no difference between sample mean vectors), and thus the Loa and Kea trends are significantly different at the 95% confidence level.  The second concern when building a statistical model is to assess how representative the dataset is of the population being analyzed. Here, we use data from all of the Hawaiian Islands less than ~6 million years old: Kaua‘i, O‘ahu (Wai‘anae and Ko‘olau volcanoes), Lāna‘i, Kaho‘olawe, Moloka‘i (east and west volcanoes), Maui (West Maui Volcano and Hāna Ridge), and Hawai‘i Island (Māhukona, Kohala, Mauna Loa, Mauna Kea, Hualālai, and Kīlauea volcanoes). The number of samples from the Loa and Kea trends are not equal, nor is the number of observations between different volcanoes included in the model. For example, recent high-precision analyses have been heavily skewed towards Hawai‘i Island 124with the completion of the Hawai‘i Scientific Drilling Project, a high-resolution sampling venture that drilled into Mauna Loa and Mauna Kea volcanoes (HSPD contributed all 61 Mauna Kea samples, but HSDP only contributed 13 of 144 total Mauna Loa samples; Blichert-Toft et al., 2003; Eisele et al., 2003; Kurz et al., 2004; Rhodes and Vollinger, 2004; Bryce et al., 2005; Nobre Silva et al., 2013a).  All isotopic analyses included in our dataset have been renormalized to the same standard values to ensure comparability between isotopic compositions analyzed at different laboratories. However, isotopic normalization does not correct for differences in reproducibility introduced by different methods. Pb isotope analysis by triple spike TIMS or MC-ICP-MS is up to ten times more precise than by double spike TIMS (Albarède et al., 2004) and the precision of Sr isotopic data measured on samples that were not acid leached will be lower (Nobre Silva et al., 2009, 2010). Although the dataset is generally high-quality and important to include in the model for minimum number-of-samples requirements in statistic calculations, it is important to keep these constraints in mind when interpreting results as they may explain some scatter and standard error in the model.  Multiple logistic regression models were fit, testing the significance of the different isotopic systems and ensuring pitfalls such as more Loa trend observations are not biasing the coefficients. In the dataset, all compiled Hawaiian observations for basalts older than East Moloka‘i belong to the Loa trend (>2 Ma; Sherrod et al., 2007), which may introduce separation into the model and bias the coefficients. To check if this was an issue, Firth’s bias-adjusted estimates were computed in R software and compared with the standard logistic model outputs (Firth, 1993). We also computed standard and Firth’s bias-adjusted logistic regression models for limited Hawaiian datasets that include only samples from Maui Nui (Moloka‘i, Maui, Kaho‘olawe, and Lāna‘i) and Hawai‘i Island, where the number of Loa and Kea trend observations are more equivalent. The models with less data (i.e., taking subsets of the Hawaiian Islands) had much higher Akaike Information Criterions (AIC; Akaike, 1970). 125AIC is a measure of the relative quality of statistical models for a given set of data, with lower values signifying a better fitting model. Firth’s bias-adjusted logistic regression models did not significantly improve model standard error, suggesting our model was not impacted by bias or separation.  The results of the two strongest logistic regression models are presented in Table 4.3 and Figure 4.12. One model includes all isotopic systems (87Sr/86Sr, 143Nd/144Nd, 176Hf/177Hf, 206Pb/204Pb, 208Pb/204Pb, and 207Pb/204Pb) and the other only the Pb isotope systems (206Pb/204Pb, 208Pb/204Pb, and 207Pb/204Pb). The prediction of Loa- or Kea-type affinity for new NWHR samples is the same for both models. Inclusion in the model required all four isotopic systems be measured on the same sample powder, so the original database of 733 Hawaiian shield basalts was limited to 619 samples for the Pb model and 362 samples for the Nd-Sr- . Table 4.3: Pb-Hf-Nd-Sr logistic regression analysis of NWHR geochemical affinityEstimates SE p-value z-value odds ratio Estimates SE p-value z-value odds ratioIntercept -34370 16910 0.042 -2.032   -  337 300 0.262 1.123   -  87Sr/86Sr 1.083 0.795 0.173 1.362 2.95   -    -    -    -    -  143Nd/144Nd 3.401 2.798 1.216 1.216 29.99   -    -    -    -    -  208Pb/204Pb 0.003 0.001 0.013 2.495 1.00 0.0016 0.0003 <0.001 4.877 1.002207Pb/204Pb -0.011 0.004 0.010 -2.594 0.99 -0.0034 0.0020 0.094 -1.676 0.997206Pb/204Pb -0.005 0.001 <0.001 -3.511 1.00 -0.0024 0.0003 <0.001 -7.675 0.998176Hf/177Hf 3.757 3.552 0.290 1.058 42.82   -    -    -    -    -  NLog Likelihood TestArea under ROC curveAICNWHR Volcano Classification ClassificationYuryaku Kea KeaDaikakuji Loa LoaDaikakuji Loa LoaDaikakuji Loa LoaUnnamed Kea KeaAcademician Berg Kea KeaMidway Kea KeaMidway Kea KeaPioneer Kea KeaNorthampton Kea KeaNorthampton Kea KeaGardner Kea KeaGardner Kea KeaGardner Kea KeaMokumanamana Loa LoaKeoea Kea KeaTwin Banks Kea KeaWest N hoa Loa LoaN hoa Kea KeaN hoa Kea KeaN hoa Kea KeaN hoa Loa LoaMiddle Bank Loa Loa303-04 0.989 303-04 0.971NIH-F9 0.016 NIH-F9 0.069NIH-D-1-2 0.672 NIH-D-1-2 0.872NIH-D4 0.036 NIH-D4 0.125NIH-W-11-1 0.001 NIH-W-11-1 0.023P5-688-1 0.034 P5-688-1 0.069  76-9-11 0.927   76-9-11 0.933NEC-3A 0.542 NEC-3A 0.65372-49A 0.095 72-49A 0.15984-40E 0.323 84-40E 0.10884-39B 0.011 84-39B 0.04176-5-4A 0.001 76-5-4A 0.03576-6-7-F 0.004 76-6-7-F 0.025P5-524-42 <0.001 P5-524-42 0.00776-5-4 DD 0.001 76-5-4 DD 0.039R-1316 <0.001 R-1316 0.002R-1295.6 <0.001 R-1295.6 0.00284-28E <0.001 84-28E 0.00684-30G <0.001 84-30G 0.007A-55-2 1.000 A-55-2 1.000A-55-4 1.000 A-55-4 1.000A-53-1 0.044 A-53-1 0.140A-55-1 1.000 A-55-1 1.00066.0 219.9Sample Probability (Loa > 0.5) Sample Probability (Loa > 0.5)Explanatory Variables Nd-Sr-Hf-Pb Isotope Model Pb Isotope Model362 619=66.7, df=7, p<0.001 =107.2, df=4, p<0.0010.9731 0.9683126 Hf-Pb isotope model. In order to be able to mathematically compute odds ratios, the dataset was scaled by 10,000, a practice that only moves the model relatively in mathematical space without changing the relationship between variables or the resulting prediction. The Hawaiian Island dataset was also partitioned into randomized subsets of 80% and 20%. Eighty percent of the data was used to compute a training logistic model. The remaining 20% dataset was used to test the training logistic model by constructing a receiver operation curve and computing area under the curve, sensitivity (true positive rate) and specificity (false positive rate) of the models (Figure 4.12). The area under the curve (AUC) was a robust 0.9731 for the Nd-Sr-Hf-Pb isotope model and 0.9683 for the Pb isotope model (the closer the AUC measure is to 1 the better the model does at classifying testing samples). Finally, a False positive rateTrue positive rate1.01.00.80.80.60.60.40.40.20.200Nd, Sr, Hf, Pb Isotope ModelPb Isotope Model Figure 4.12: Receiver operating characteristic curve of the Hawaiian Island logistic regression testing dataset The receiver operating characteristic (ROC) curve examines the error rates when the testing portion (20% of the total Hawaiian Island database) is sorted into Loa and Kea trends based on the logistic model fit to the training dataset (the remaining 80% of the total Hawaiian Island database). The green curve is the Pb only model and the red curve is the Nd-Sr-Hf-Pb isotope model. An area under the curve (AUC) is a measure of discrimination and is generally a tradeoff between the false positive rate (specificity) and the true positive rate (1-sensitivity). It is measured relative to a line through unity and a 45-degree slope (shown in black), which represents a completely random guess based on a constant-only model. AUC of 1 is a perfect prediction and AUC of 0.5 is no better than random guessing; here the AUC of the Nd-Sr-Hf-Pb isotope model is 0.9731 and the Pb only model is 0.9683.127likelihood ratio test is used to test for the overall significance of the models by comparing them to a model with no predictor variables. A chi squared value of 66.7 for the four-isotope systems model (the χ² critical value at seven degrees of freedom is 14.067) and 107.2 for the Pb isotope model (the χ² critical value at four degrees of freedom is 9.488) indicate that both models are significantly better than a model with only a constant coefficient (β0) and no predictor variable coefficients (β1,2,3,…,n; equation 4.1). The logistic regression model confirms that Pb isotopes are the best indicators of Loa- or Kea-type affinity for Hawaiian basalts as shown by the more significant p-values of Pb compared to the other isotopic systems in the Nd-Sr-Hf-Pb isotope model (Table 4.3) (Abouchami et al., 2005; Weis et al., 2011). When the logistic regression model is used to predict Loa- or Kea-type affinity of unknown samples (e.g., NWHR samples), Daikakuji, Nīhoa (NIH-D-1-2), and Middle Bank are confirmed as Loa-type volcanoes, whereas Mokumanamana, West Nīhoa, and other Nīhoa samples are close to the cutoff for Loa-type affinity (probabilities >0.5). These results confirm the ephemeral presence of the Loa trend along the NWHR and the Loa-type compositions observed in the major and trace element concentrations and isotopic compositions of these samples. The agreement shown here between a robust statistical model and defined geochemical characteristics of Hawaiian basalts documents the promising application of statistical models in geochemical studies where there is a significant number of reference data to support population statistics. The use of statistical methods increases the confidence of classifying unknown samples into geochemical groups, and is a useful tool that should be considered in future geochemical studies where enough high-quality data exists to fit a model. 4.6.2 Elemental and Isotope Systematics of NWHR basalts Major and trace element characteristics of NWHR basalts show a wide range of variation that encompasses both Loa and Kea compositions in NWHR basalts (Figures 4.5, 4.8). Daikakuji is confirmed as the most extreme Loa trend volcano sampled to date along 128the NWHR, with the telltale high normalized SiO2, Sr/Nb, Zr/Nb, La/Nb, and (Ba/Th)PM and low normalized CaO of Loa trend volcanoes (Jackson et al., 2012; Pietruszka et al., 2013; Frey et al., 2016). Daikakuji is the only volcano with high SiO2 concentrations (43-52 wt% SiO2; high for Hawaiian basalts is SiO2 > 51 wt%; Greene et al., 2010) (Figure 4.4) that overlap with Mauna Loa values (44-52 wt% SiO2). Daikakuji also has low CaO, overlapping with Ko‘olau compositions, although its TiO2 content is systematically higher than that of Loa trend volcanoes. The high TiO2 and SiO2 of Daikakuji is somewhat enigmatic, as high SiO2 is a Loa-type indicator, but high TiO2 is more a Kea-type characteristic. These authors suggested that the differences in Ti content between the Ko‘olau end-member (low TiO2) and the Lō‘ihi end-member (high TiO2) originate from different source lithologies with unique geological histories (Jackson et al., 2012). The high TiO2 Lō‘ihi source has been proposed to be a subduction modified pyroxenite component where the presence of rutile maintains high TiO2 concentrations while other elements such as Si and Pb are removed (Jackson et al., 2012). The low TiO2 Ko‘olau source corresponds to either a high SiO2 eclogite component that has not lost incompatible elements and SiO2 during subduction zone processing, or a particularly enriched peridotite (Jackson et al., 2012; Rhodes, 2015). However, at Daikakuji, the TiO2 conundrum may be a result of unusually low degrees of melting, as illustrated by extremely high (La/Yb)PM values, the highest observed along the entire Hawaiian-Emperor chain (La/Yb = 2.8-8.5; where average of Mauna Loa and Mauna Kea is 3.2 and EPR MORB is 0.8; Harrison et al., 2017). Small degrees of melting (up to ~10%) of a garnet peridotite will yield higher wt % TiO2 in the melt than similar degrees of melting of a spinel peridotite (Prytulak and Elliott, 2007; Le Roux et al., 2015). Similarly, melting of a MORB-like eclogite will yield higher concentrations of TiO2 in the melt up to partial melting fractions of 80% (Le Roux et al., 2015). Several samples from Nīhoa and West Nīhoa also exhibit high (La/Yb)PM and systematically higher TiO2 compositions (Figures 4.4, 4.8), possibly reflecting low degrees of melting. However, interpretations based on NWHR major element 129compositions are ambiguous due to the moderate alteration of these samples; trace element ratios for immobile elements and isotopes measured on leached sample powders provide a much more reliable picture.  Middle Bank, West Nīhoa, and Nīhoa (NIH-D-1-2) also overlap with Mauna Loa compositions in La/Nb, Zr/Nb, Sr/Nb, (Th/Ce)PM, and (Ba/Th)PM (Figure 4.8). In Figure 4.5, many NWHR volcanoes overlap with the field defined by Ni‘ihau volcano (Cousens and Clague, 2015). Lavas from Ni‘ihau are known to be transitional in isotopic and trace element signature between the Loa and Kea trends (Cousens and Clauge, 2015). For example, in Pb isotopic compositions these lavas straddle the Loa-Kea divide line of Abouchami et al. (2005). In trace elements, Ni‘ihau lavas have Zr/Nb in the range of 10-12, closer to a Loa trend source signature, but are low in (Sr/Nd)PM and Ba/La, which are more characteristic of Kea trend lavas (Cousens and Clague, 2015). This transitional geochemical composition and the lack of en echelon Loa and Kea trend volcanoes for the Northern Hawaiian islands (e.g., Ni‘ihau and Kaua‘i) has been used to demonstrate that the classic paradigm of Loa and Kea double volcanic chains of the youngest Hawaiian Islands (i.e., all islands southeast of the Moloka‘i fracture zone; Abouchami et al., 2005) breaks down on volcanoes older than those on O‘ahu (~4 Ma; Cousens and Clague, 2015). Likewise, NWHR volcanoes do not occur in paired volcano chains (Figure 4.1) (Garcia et al., 2015; Kelley et al., 2015; Harrison et al., 2017). There are single volcano segments composed of seamounts with star-shaped multiple rift zones (e.g., Pioneer and Northampton), segments composed of large seamounts with highly developed long rift zones (e.g., St. Rogatien Bank and Gardner), and some areas with as many as three or four volcanoes abreast (e.g., Midway, Ladd, and Nero; Kelley et al., 2015) (Figure 4.1). The transitional nature of NWHR geochemical signatures similar to Ni‘ihau and the absence of a volcanic double chain along the NWHR suggests that the bilateral Loa and Kea trends of the Hawaiian Islands do not continue back in time along the NWHR. This was also observed in Pb isotopes, and has been interpreted as the intermittent 130arrival and progressively greater participation of Loa trend geochemical heterogeneities from the lower mantle and mixed with the Kea-type source with time (Harrison et al.; 2017). The trace and major element characteristics of NWHR volcanoes support this interpretation. In isotopic space, the Hawaiian Islands define well-known trends of binary or pseudo binary mixing of up to four sources (Eiler et al., 1996; Hauri, 1996; Eisele et al., 2003; Blichert-Toft et al., 2003). Mixing between the traditional four end-member components that define the Hawaiian Island geochemical arrays (KEA, EMK, DMK or Kalihi, and LOIHI; Tanaka et al., 2002; 2008; Fekiacova et al., 2007) accounts for most of the NWHR isotopic variation. In Hf and Nd isotope space, the Hawaiian array is better fit by the Koolau end-member ɛNd = -2 of Mukhopadhyay et al. (2003) rather than the ɛNd = +0.2 proposed by Tanaka et al. (2002), who only specified that EMK must be below 0.2 based on Sr, Pb, and Nd isotope systematics. When the lower epsilon Nd enriched end-member of Mukhopadhyay et al. (2003) is used in mixing calculations, the new Hawaiian Islands shield array in Hf-Nd isotope space is more adequately approximated. We propose a revision of the KEA component to ɛHf = +15.4  and ɛNd = +8.8  isotopic compositions to account for the depleted compositions observed on Gardner, Unnamed, and Midway volcanoes (shown as the “New Kea” circle in Figures 4.9, 4.10, 4.11). Conversely, if the depleted Nd and Hf isotopic compositions at Gardner, Unnamed, and Midway are interpreted as mixing between the Kea component of Tanaka et al. (2002) and a more depleted source, this unknown source would require high 143Nd/144Nd and 176Hf/177Hf, and low 87Sr/86Sr, 206Pb/204Pb, and 208Pb/204Pb. This source would have to be more depleted in 87Sr/86Sr and 176Hf/177Hf than rejuvenated Hawaiian lavas, the most isotopically depleted basalts observed on the Hawaiian Islands. Finally, it is possible that the Kea end-member is not as homogeneous as previously thought. If Loa trend volcanoes can sample small-scale heterogeneities that participate in volcanism only for finite periods of time, then it would also be logical to assume the same processes may occur in Kea trend volcanoes. Gardner is the location of both the highest magmatic flux and 131mantle potential temperature along the NWHR; these conditions may melt highly refractory depleted mantle components that are invisible under the typical melting conditions of the Hawaiian mantle plume (Salters et al., 2011; Tree, 2016; Wessel, 2016).4.6.3 A New Hf-Nd Hawaiian Array The Hawaiian Hf-Nd array was first calculated by Blichert-Toft et al. (1999) based on <100 samples from Ko‘olau, Lāna‘i, Kaho‘olawe, Haleakalā, Mauna Kea, Mauna Loa, Kīlauea, and Lō‘ihi volcanoes. Since that time, the number of high precision shield-stage analyses of Hawaiian Island Hf and Nd isotopes has more than quadrupled. The Hawaiian array is recalculated with 452 high-precision analyses, all normalized to the same standard values, from 16 Hawaiian volcanoes (Kaua‘i, West Ka‘ena, Ko‘olau, Penguin Bank, East and West Moloka‘i, Lāna‘i, Kaho‘olawe, West Maui, Māhukona, Kohala, Hualālai, Mauna Kea, Mauna Loa, Kīlauea, and Lō‘ihi; data sources are the same as listed in Figure 4.9 and are all renormalized to the same standard values). Aside from Lō‘ihi, which is in the preshield stage, all analyses included in the new calculation are from the shield-stage (the previous calculation included some postshield samples). The new Hawaiian array has a slope of 1.10 ± 0.04 and a y-intercept of 4.58 ± 0.23 (95% confidence) (Figure 4.10b). Although the new Hawaiian array is similar to the original Hawaiian array (the original Hawaiian array had a slope of 1.00 ± 0.03 and a y-intercept of 5.22 ± 0.17; Blichert-Toft et al., 1999), it is steeper due to the lower epsilon Hf at a given epsilon Nd of some Kaho‘olawe and West Ka‘ena samples (Huang et al., 2005b; Greene et al., 2010).  NWHR lavas define a linear array in Hf-Nd isotopic space that is parallel to the OIB array, rather than the new Hawaiian array (Figure 4.10b). The original Hawaiian array has a higher epsilon Hf at a given epsilon Nd, originally interpreted as addition of pelagic sediment, a source with higher time integrated Lu/Hf, in the source of Hawaiian basalts (Blichert-Toft et al., 1999; Vervoort et al., 1999). Other authors argue that the lack of an Nb anomaly in Ko‘olau lavas is evidence against any sediment input in the source of Hawaiian 132basalts (Pietruszka et al., 2013). Modeling of Ko‘olau compositions has shown that the trace element and isotopic characteristics are not explicable by simple mixing combinations of sediment, recycled gabbro, recycled basalt, and depleted lithosphere, although the high Hf isotope ratios may be reproduced by addition of ancient depleted lithosphere (Salters et al., 2006).  The new NWHR data does not contribute new significant information to this debate on the source of high Hf in the Hawaiian plume aside from the observation that this component, either pelagic sediment or old depleted lithosphere, did not participate in the formation of NWHR shield-stage basalts and is a recent addition for only the Hawaiian Islands starting at Middle Bank. This is reflected in the change in the slope of the Hawaiian array that becomes progressively shallower with time, starting steep and parallel to the OIB array for the NWHR and eventually shallowing into the Hawaiian array for the Hawaiian Islands. The volcanoes that shallow the trend for the Hawaiian Islands are West Ka‘ena, Ko‘olau, Lāna‘i, and Kaho‘olawe, all which erupt the most enriched Loa-type compositions (Tanaka et al., 2002; Gaffney et al., 2005; Huang et al., 2005b; Greene et al., 2010). The Hawaiian and OIB arrays intersect, causing some ambiguity as to when this high Hf component began to contribute to Hawaiian basalts; based on the intersection of NWHR array and new Hawaiian array slopes in Hf-Nd isotope space it is a possible contribution after Gardner (Figure 4.10b).4.6.4 The Lō‘ihi End-Member Component Along the NWHR No NWHR basalts exhibit the high 208Pb/204Pb at a given 206Pb/204Pb (Harrison et al., 2017) or high Hf isotopic ratios at moderately depleted Nd isotopic ratios that are characteristic of the Lō‘ihi end-member (Garcia et al., 1993, 1995, 1998; Tanaka et al., 2008; Jackson et al., 2012) (Figure 4.10b). All NWHR Pb isotope ratios plot along the Kea mid-8 and Kea lo-8 Pb-Pb mixing lines defined by the variations observed during the geochemical evolution of Mauna Kea (Eisele et al., 2003; Rhodes and Vollinger, 2004; Nobre Silva et 133al., 2013a) (Figure 4.13). This suggests Lō‘ihi is not a common source component in most Hawaiian volcanism as noted by Jackson et al. (2012). In fact, Lō‘ihi only participates as a significant end-member in Hawai‘i Island volcanoes, as a main component in Lō‘ihi Seamount basalts and as a contribution in some Māhukona, Mauna Loa, and Mauna Kea basalts (e.g., increasing the 208Pb/204Pb of some Māhukona and Mauna Loa lavas at a given 206Pb/204Pb) and some Mauna Kea lavas where it lowe