Volume 24, Issue 7 e2022GC010592
Research Article
Open Access

Rapid Formation of the Ellice and Osbourn Basins and Ontong Java Nui Breakup Kinematics

P. C. Davidson

Corresponding Author

P. C. Davidson

College of Earth Ocean and Atmospheric Science, Oregon State University, Corvallis, OR, USA

Correspondence to:

P. C. Davidson,

[email protected]

Search for more papers by this author
A. A. P. Koppers

A. A. P. Koppers

College of Earth Ocean and Atmospheric Science, Oregon State University, Corvallis, OR, USA

Search for more papers by this author
J. G. Konter

J. G. Konter

School of Ocean and Earth Science and Technology, University of Hawai'i at Mānoa, Honolulu, HI, USA

Deceased 3 July 2022

Search for more papers by this author
First published: 22 July 2023

Abstract

Breakup of the proposed greater Ontong Java Nui large igneous province during the Cretaceous Normal Superchron can be constrained by the opening of the Ellice Basin (EB) separating the Ontong Java and Manihiki Plateaus and the Osbourn Basin separating the Manihiki and Hikurangi Plateaus. Dating of recovered dredged samples using plagioclase 40Ar/39Ar and zircon U/Pb geochronology methods indicates that spreading was well underway in the EB by 118 Ma with full spreading rates up to 3X faster than any observed today of 30–45 cm/yr and spreading likely continued until 112-108 Ma. Ellice Basin samples show diverse geochemical affinities ranging from mid-ocean ridge basalt (MORB) or Ontong Java-like to more enriched OIB-like. Pb and Nd isotopes from six samples contain varying influences from Pacific MORB and possibly Ontong Java. The geochemistry shows a lack of a clear mantle plume influence despite EB's close temporal and spatial relationship to Ontong Java, while some data resemble the Louisville Seamounts. This compositional diversity complements morphological differences among dredge sites and shows that both in situ MORB and younger overprinted features related to the nearby Tuvalu Seamounts were sampled. 40Ar/39Ar geochronology confirms the age of International Ocean Discovery Program Site U1365 near the Osbourn Trough (OT) to be 102.60 ± 0.26 Ma (2σ, n = 18). This age constrains the timing of a spreading reorientation event observed in the OB to coincide with a global plate reorganization event around 105 Ma and estimates the cessation of spreading at the OT to 96 Ma.

Key Points

  • New geochronology from the Ellice Basin (EB) suggests it formed between ∼123 and 112–108 Ma with ultra-fast spreading rates up to 45 cm/yr

  • We confirm that the Osbourn Basin continued opening later than the EB and finished by 96 Ma

  • The EB mantle source appears mostly like Pacific mid-ocean ridge basalt with a possible influence from Ontong Java and the Louisville hotspot

Plain Language Summary

Ontong Java Nui is proposed to have been the largest volcanic system on Earth when it erupted about 120 million years ago on the sea floor in the Pacific Ocean. Shortly after it erupted, plate tectonics supposedly split this massive volcanic feature into five fragments that drifted away from each other. We present new age and geochemistry results from samples collected in two ocean basins that separate the three main fragments of this feature still located in the western Pacific Ocean. We suggest that all three fragments separated extremely rapidly, possibly as much as 3X faster than any plate tectonics today and stopped in roughly their current configuration only 20 million years after they began to spread. We also show that the source of magma in one of the basins, the Ellice Basin (EB), differed slightly from the initial volcanism that made up Ontong Java Nui as a whole. These are the first ages and geochemistry data for the EB, and they, along with our new ages from the Osbourn Basin, are critical to further understanding the eruption and breakup history of Ontong Java Nui.

1 Introduction

The proposed Ontong Java Nui (OJN) large igneous province (LIP) consisted of the individual Ontong Java Plateau (OJP), Manihiki Plateau (MP), and Hikurangi Plateau (HP) that are all now separated by thousands of kilometers (Figure 1; Chandler et al., 2012; Taylor, 2006). The timing and kinematics of the breakup and separation of OJN have remained under discussion since Taylor (2006) proposed that the three individual oceanic plateaus formed together as a single LIP in the Pacific Ocean. Both the Ellice Basin (EB), separating OJP and MP (Figure 1), and the Osbourn Basin (OB), separating MP and HP, formed entirely during the Cretaceous Normal Superchron (CNS), so seafloor magnetic lineations cannot be used to estimate the timing for the initiation of separation and subsequent breakup and ocean basin formation. The CNS began at 121.4 Ma and spans nearly 40 Myrs of continuous normal magnetic polarity until 83.7 Ma (Ogg, 2020) and the lack of seafloor magnetic lineations in the EB and OB has led to debates on the timing and rate of formation of both basins (Benyshek et al., 2019; Billen & Stock, 2000; Chandler et al., 2012; Downey et al., 2007; Taylor, 2006; Zhang & Li, 2016). The timing and duration of the formation of these basins have implications for regional tectonics questions such as the docking of HP with the Gondwana margin (Davy et al., 2008; Mortimer et al., 2019) and the potential link between the Louisville mantle plume and the origin of OJN (Chandler et al., 2012). The current paradigm of OJP having formed together with MP and HP as a single LIP (the so-called Taylor hypothesis; Taylor, 2006) is also still a matter of debate. Previous and more recent studies have suggested that OJP and MP are formed by two separate mantle plumes (Golowin et al., 2018; Larson, 1997) and that the age of the EB has direct consequences for evaluating these hypotheses.

Details are in the caption following the image

Overview map of the Southwestern Pacific Ocean basin. The Ontong Java, Manihiki, and Hikurangi Plateaus that make up Ontong Java Nui are outlined in red. Note that we do not include Robbie Ridge as part of the Manihiki Plateau and Ontong Java Nui, consistent with recent work (e.g., Golowin et al., 2018; Hochmuth et al., 2019). Colored lines with labeled numbers are age isochrons (Seton et al., 2020). Thin white lines are selected fracture zones and the thicker white line indicates the Osbourn Trough. Deep Sea Drilling Project/Ocean Drilling Program (ODP)/International Ocean Discovery Program sites are labeled and indicated by white circles. Other relevant features are labeled as West Wishbone Scarp (WWS), East Wishbone Scarp (EWS), East Manihiki Scarp (EMS), Nova-Canton Trough (NCT). The black box around the Ellice Basin outlines the map in Figure 2. Global bathymetry and elevation are from the GEBCO Compilation Group (2022).

The lack of seafloor magnetic lineations during the CNS requires other methods to better delineate the breakup and separation kinematics of OJN and formation of a large part of the Pacific Ocean basin. These include high-resolution multibeam bathymetric mapping, and geochemical and geochronological analyses of recovered in situ oceanic crust samples. Several challenges arise when attempting to employ these methods. Since these basins were emplaced tens of millions of years ago, thick sediment covers exist in places, obscuring the spreading fabric potentially observable by multibeam bathymetry, and also preventing the recovery of in situ basement mid-ocean ridge basalts (MORBs) by dredging. In places where sediment cover is thin or non-existent, the long exposure of basaltic basement to seawater means that MORB samples, which have ultra-low potassium concentrations, have likely undergone extensive seawater alteration. This typically leaches out or adds new potassium to the basalts, preventing robust determination of the eruption ages, and instead often results in discordant ages or records only the age of the alteration events (Baksi, 2007; Kaneoka, 1972; Koppers et al., 2000; Seidemann, 19771978). Submarine alteration of basalts also presents obstacles to geochemical analysis, limiting us to using only elements resistant to alteration as geochemical fingerprints to investigate magmatic origins (e.g., Jackson et al., 2010).

High resolution multibeam bathymetry in the EB reveals primary fabric that is the result of seafloor spreading (Benyshek et al., 2019), mainly abyssal hills from normal faulting at the ridge axis that trend perpendicular to spreading direction (Macdonald et al., 1996; Shaw, 1992), and fracture zones separating spreading ridge segments that trend parallel to spreading (Menard & Atwater, 1969; Wilson, 1965). Similar data are available from the OB (Downey et al., 2007).

In this study, we date samples from International Ocean Discovery Program (IODP) Expedition 329 Hole U1365E in the OB and R/V Kilo Moana cruise KM1609 in the EB to constrain the breakup and separation kinematics of Ontong Java Nui. We also use major, trace, and isotope geochemistry from EB samples to evaluate the mantle source and petrogenesis of these samples. We will show that the EB was formed earlier and more rapidly than previously thought and overlaps with the formation of the OB, and that the history of both basins has connections to broader global plate reorganization events during the Cretaceous.

2 Regional Setting

The Ontong Java, Manihiki, and Hikurangi Plateaus are located in the equatorial and southern western Pacific Ocean (Figure 1). These three individual oceanic plateaus together are proposed to make up the Ontong Java Nui LIP (Taylor, 2006) along with two other fragments that have since been rifted away and subducted along the South American and Antarctic continental margins (Hochmuth & Gohl, 2017; Hochmuth et al., 2015; Viso et al., 2005). Basement ages from OJP span a considerable range of 128.2 to 119.6 Ma with relatively large 2σ uncertainties often >2 Ma (Chambers et al., 2004; Mahoney et al., 1993; Parkinson et al., 2001; Tejada et al., 19962002). Biostratigraphic ages from the sediments directly overlying the basaltic basement are generally Aptian to Albian (Sikora & Bergen, 2004; Tarduno et al., 1991). Ages from MP and HP span some 30 Myrs from 126.0 to 115.8 Ma and 118.4 to 96.3 Ma, respectively, with similarly large uncertainties (Hoernle et al., 2010; Timm et al., 2011).

The EB (Figures 1 and 2) separates OJP and MP and is bound to the north by the Nova-Canton Trough, originally thought to be an abandoned spreading center (Larson, 1997) but has since been shown to be a major fracture zone (Benyshek et al., 2019; Taylor, 2006). R/V Onnuri cruise NAP09-3 traversed the EB and recorded high resolution multibeam bathymetry that revealed preserved primary spreading fabric nearly unobscured by sediment cover (Chandler et al., 2012). Both the EB and OB reside in a part of the South Pacific Gyre that has a low sedimentation rate (Zhang & Smith-Duque, 2014). R/V Kilo Moana cruise KM1609 expanded on the work of NAP09-3 and consisted of multibeam mapping throughout a greater portion of the EB. Analysis of KM1609 and NAP09-3 multibeam data reveals a complex three-stage spreading history in the EB (Benyshek et al., 2019). The first two stages constitute most of the spreading in the EB, with rotations separating each stage manifested as changes in the orientation of the spreading fabric. Formation of the EB took place primarily by strike-slip motion with small MOR spreading segments opening between fracture zones (Benyshek et al., 2019).

Details are in the caption following the image

Multibeam bathymetric map of the Ellice Basin (black outlined box in Figure 1) from R/V Kilo Moana cruise KM1609 with satellite-derived predicted bathymetry (GEBCO Compilation Group, 2022). Black labeled circles are recovered dredges from KM1609 and gray-labeled circles are from R/V Roger Revelle cruise RR1310 from Finlayson et al. (2018). 40Ar/39Ar ages from three Tuvalu Seamounts are indicated (Finlayson et al., 2018). White lines are selected fracture zones and the dotted black box indicates the region of the map in Figure 5.

The OB (Figure 1) separates MP and HP, and is marked in its center by the now extinct Osbourn Trough (OT), a segmented and abandoned spreading center (Billen & Stock, 2000; Downey et al., 2007). The OB also has a relatively thin sediment cover, which allows for detailed observations of spreading fabric in this region (Billen & Stock, 2000; Downey et al., 2007; Worthington et al., 2006). Similar to the EB, the OB was formed in two stages as indicated by a change in the abyssal hill fabric (Downey et al., 2007), but the formation timing is poorly constrained. The OT's ridge morphology suggests that spreading rates were slow to intermediate, around 2–6 cm/yr, immediately prior to cessation of spreading (Downey et al., 2007) but was considerably faster throughout the majority of the OB's formation, up to 19 cm/yr (Zhang & Li, 2016). Studies have estimated the initiation of separation to be at 119 Ma (Larson et al., 2002; Taylor, 2006) and the cessation of spreading to have occurred sometime between 101 and 71 Ma (Billen & Stock, 2000; Davy et al., 2008; Downey et al., 2007; Mortimer et al., 2019; Zhang & Li, 2016).

3 Drill and Dredge Samples

Samples used in this study come from one drill site in the OB and 12 dredge sites in the EB. KM1609 dredged 16 locations throughout the EB (Figure 2) for geochemical and geochronological studies. International Ocean Discovery Program Expedition 329 Hole U1365E located in the OB (Figure 1) recovered ∼43 m of basaltic basement (Expedition 329 Scientists, 2011). Samples from Hole U1365E gave a Re-Os age of 103.7 ± 2.3 Ma (Zhang & Li, 2016) and have geochemical affinities closest to Pacific MORB (Zhang et al., 2012). However, a more recent estimate from the adjacent Deep Sea Drilling Project (DSDP) Site 595 gives a 40Ar/39Ar age of 84.4 ± 3.4 Ma (Mortimer et al., 2019) for this location.

Islands and seamounts bisect the EB (Tuvalu seamounts, Figure 2) and were sampled during the dredging cruise RR1310. The majority of these seamounts are attributed to the Rurutu hotspot with ages between 50 and 60 Ma, but three of the seamounts yielded anomalously old 40Ar/39Ar ages between 90 and 95 Ma that are either related to the opening of the EB or to later-stage alkalic OJN volcanism (Finlayson et al., 2018). The presence of these seamounts constrains the minimum age of the EB to 95 Ma or older. Pacific MORB samples from the Nova-Canton Trough north of MP have several 40Ar/39Ar ages between 74.8 and 94.1 Ma (unpublished data, Pyle & Mahoney, 2006) and likely represent alteration ages considering most are younger than the CNS.

4 Materials and Methods

KM1609 mapped the EB using an EM122 multibeam echosounder and dredged 16 locations to recover seafloor samples. Dredge locations were targeted to sample in situ MORB crust with the objective of obtaining rocks emplaced on-axis during the opening of the EB. Overprinted features were avoided but in some places they were chosen when other factors suggested that they may be related to spreading, for example, volcanic features running parallel to the spreading fabric. To maximize the likelihood of basement recovery, dredge locations were selected on the basis of multibeam mapping at locations of thin sediment cover and high relief such as abyssal hills and along the inside of fracture zones. Thirteen of 16 dredges recovered sufficient material for 40Ar/39Ar dating and geochemistry; however, most samples recovered were strongly affected by alteration and were not included in our analytical work.

4.1 40Ar/39Ar Dating

Samples from KM1609 were initially selected based on their apparent freshness from hand-sample observations. Fifty samples from 13 dredge locations were selected to make thin sections. From these 50 thin sections, 20 samples (from 10 dredge locations) were selected for dating based on mineral and groundmass phases being the least altered.

All samples were crushed and sieved to appropriate size fractions for a given mineral phase ideal for 40Ar/39Ar dating (between 180 and 300 μm) and then washed to remove fine dust-sized particles attached to the grains. Plagioclase was separated as the least magnetic fraction in a Franz magnetic separator and groundmass as the most magnetic fraction. The separated fractions were leached in 1 and 6 N hydrochloric acid and 1 and 3 N nitric acid for 1 hour at each strength and type to remove alteration products. Plagioclase separates were leached for an additional 5 min in 5% hydrofluoric acid to further remove alteration. Leached samples were re-sieved to the original size fraction, and 20–30 mg of the least altered groundmass samples, and plagioclase samples with the fewest or no melt inclusions, were picked under a binocular microscope to be irradiated. All samples were irradiated for 6 hours at Oregon State University's TRIGA reactor using Fish Canyon Tuff sanidine as flux monitors.

Irradiated samples were analyzed using a Thermo Fisher ARGUS VI multi-collector mass spectrometer at Oregon State University's Argon Geochronology Lab. Groundmass phases were analyzed in 39–45 incremental heating steps and 26–33 steps for plagioclase. All data were reduced using ArArCALC software (Koppers, 2002). Plagioclase analyses assumed an atmospheric composition for the trapped 40Ar/36Ar component of 298.56 ± 0.31 (Lee et al., 2006). Groundmass analyses with statistically significant inverse isochrons (i.e., MSWDs < confidence interval; Koppers, 2002) that had a non-atmospheric trapped 40Ar/36Ar component that does not overlap with the atmospheric value used their respective trapped value (and associated uncertainty) to calculate their age spectra and total fusion ages, and are indicated as such in Table 1 (Heaton & Koppers, 2019; Schaen et al., 2021).

Twenty samples from 10 dredge locations were analyzed by 40Ar/39Ar dating. Among these, five were groundmass, nine were plagioclase, six had both groundmass and plagioclase, and many had duplicates of each phase analyzed. Duplicate groundmass and plagioclase analyses were combined in ArArCALC for improved statistics when possible. In samples where both duplicates of groundmass and plagioclase were present, all analyses from both phases were combined to yield a single plateau and inverse isochron age. Combined analyses that used non-atmospheric trapped 40Ar/36Ar for age spectrum calculations used the same trapped compositions for the respective steps in the combined analyses. This is necessary when combining plagioclase and groundmass analyses since the trapped component is not always the same for the two phases (Figure 4f).

Twelve samples from the recovered basement in IODP Hole U1365E were chosen for 40Ar/39Ar analyses based on the lowest degrees of alteration. Samples were processed using the same methods described above.

4.2 U/Pb Dating

One sample was analyzed with U/Pb dating. Zircon crystals were extracted from sample KM1609-D2-14 using a jaw crusher and sieving (400 μm mesh) followed by washing and decanting, heavy liquids, and a Frantz magnetic separator. About 60 high-quality grains were selected and mounted into a 1” epoxy mount together with fragments of Sri Lanka, FC-1, and R33 zircon crystals that are used as primary standards. The mounts were sanded down to a depth of ∼20 microns, polished, imaged, and cleaned prior to isotopic analysis. Prior to analysis, zircon grains were imaged with back scatter electron and cathodoluminescence (CL) to provide a guide for locating analysis pits in optimal locations, and to assist in interpreting results. Images were made with a Hitachi S-3000N SEM and a Gatan Mini-CL detector system at Carleton College. U-Pb geochronology of zircons was conducted by laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS) at the Arizona LaserChron Center (Gehrels & Pecha, 2014; Gehrels et al., 20062008). More details of the analyses are given in the supplements.

4.3 Whole-Rock Major and Trace Element Geochemistry

Thirty samples from 12 dredge locations in the EB were chosen for major and trace element analyses. These samples were selected based on their appearance in the hand sample and thin section as being the least altered and the most representative of the compositional diversity among the recovered samples for a given dredge recovery. Samples were sawed to isolate the least altered portions and then chipped in a rock crusher. These chips were pulverized in a ring mill and the powders were melted in a furnace and then dissolved in acid. Solutions were analyzed for major and trace elements at Washington State University's GeoAnalytical Lab.

Geochemical plots comparing our samples to large literature datasets (e.g., Ontong Java, Pacific MORB, etc.) plotted these latter datasets as 2-D kernel density estimations (KDEs) using the Seaborn library in Python. This was done to minimize subjective selection of representative fields that encompass data and instead apply the same objective parameters for each data set. The 80% KDE was used as it appears to encompass the most data without including small pockets of outlier data.

4.4 Pb and Nd Isotopes

Six samples were chosen for Pb and Nd isotopic analyses. Samples were chosen based on their trace element chemistry focused on spanning the observed variability as well as the availability of ages. Samples were crushed and washed, and 0.5–1.0 g was picked, avoiding altered grains. Samples were leached in hot 2M HCl, 6M HCl, 4M HNO3 and then dissolved in 2:1 HF:HNO3. The reagents were double distilled in teflon or quartz glass. Separations followed a modified technique from Konter and Storm (2014): Pb was separated with 3M HNO3 and 6M HCl passing the entire sample through Eichrom Sr resin, followed by a purification step with 1M HBr, 2M HCl, 6M HCl and Eichrom AG1-X8 (Hanan & Schilling, 1989). The REEs were separated from the remaining sample using 16M acetic acid, 10M HCl and Eichrom AG1-X8, following experiments by Van den Winkel et al. (1972) and Hooker et al. (1975). A purification pass was used with 10M HCl and Eichrom AG1-X8. Nd was separated from the REE fraction following Pin and Santos Zalduegui (1997) using 0.18M HCl and Eichrom LN resin. Measurements were made at the University of Hawaii on a Nu Plasma HR MC-ICP-MS using an APEX with Spiro introduction system. Nd values were normalized to JNdi-1 143Nd/144Nd = 0.512115, and the BCR-2 rock standard gave 143Nd/144Nd of 0.512638 ± 4, while BIR-1 gave 143Nd/144Nd of 0.513100 ± 10. Pb values were bracketed against NBS 981 values of Todt et al. (1996), and monitored for fractionation with Tl (NBS 997 Tl). BCR-2 gave 206Pb/204Pb of 18.7465 ± 4, 207Pb/204Pb of 15.6079 ± 4, 208Pb/204Pb of 38.6840 ± 10. All reported errors are 2σ.

5 Results

5.1 Geochronology

5.1.1 Ellice Basin

Two of the 33 total 40Ar/39Ar analyses yielded plateaus interpreted as eruption ages, 12 as alteration ages, and the rest did not yield concordant ages (Table 1). This low success rate was anticipated for dating ultra-low potassium and highly altered submarine MORB samples. Many groundmass samples had similar shaped age spectra that were highly disturbed and discordant. The only reliable 40Ar/39Ar ages were derived from sample KM1609-D11-01, which yielded two plagioclase plateau ages with a combined age of 117.20 ± 0.75 Ma (2σ) from 96% of the released 39Ar(K) with an MSWD of 1.26 and an ultra-low K/Ca of 0.0009 (Figure 4a). These two plagioclase analyses meet all criteria for a reliable age plateau as defined in Schaen et al. (2021) for submarine volcanic incremental step heating experiments.

Table 1. 40Ar/39Ar Analyses from the Ellice Basin
Plateau Inverse isochron Total Fusion
Sample Phase Age [Ma] Error MSWD % 39Ar(K) Steps used / Total steps K/Ca Age [Ma] Error MSWD 40Ar/36Ar intercept Error Age [Ma] Error K/Ca
Ellice Basin (Eruption ages)
KM1609-D11-01 PLAG 117.54 ± 1.08 1.09 99% 24 / 27 0.0009 117.64 ± 1.43 1.12 301.91 ± 49.31 117.30 ± 1.18 0.0009
KM1609-D11-01* PLAG 116.97 ± 1.04 1.49 94% 20 / 29 0.0009 117.17 ± 1.29 1.62 312.82 ± 98.10 116.53 ± 0.98 0.0009
KM1609-D11-01 (n = 2) PLAG X2 117.20 ± 0.75 1.26 96% 44 / 56 0.0009 117.29 ± 0.78 1.29 312.18 ± 37.19 116.87 ± 0.77 0.0009
Ellice Basin (Alteration ages)
KM1609-D02-22 PLAG 99.33 ± 1.46 1.59 73% 12 / 20 0.0015 101.99 ± 3.50 1.43 294.98 ± 4.50 93.44 ± 1.05 0.0016
KM1609-D03-12 PLAG 52.19 0.29 13.54 66% 7 / 19 0.0459 52.06 0.61 15.69 302.22 14.14 53.39 ± 0.17 0.0454
KM1609-D07-01 PLAG 70.90 ± 0.27 1.85 65% 15 / 29 0.0054 71.06 ± 0.25 1.16 286.46 ± 7.83 69.83 ± 0.21 0.0050
KM1609-D07-01* PLAG 71.81 ± 0.26 2.32 56% 11 / 27 0.0049 72.07 ± 0.61 2.37 279.93 ± 39.96 70.92 ± 0.18 0.0047
KM1609-D07-01 (n = 2) PLAG X2 71.44 ± 0.27 4.52 61% 26 / 56 0.0052 71.61 ± 0.33 4.26 286.93 ± 14.50 70.32 ± 0.17 0.0049
KM1609-D07-02 PLAG 78.19 ± 0.70 3.44 55% 13 / 31 0.0023 78.74 ± 1.26 3.48 290.08 ± 17.44 78.12 ± 0.36 0.0019
KM1609-D07-02* PLAG 76.86 ± 0.67 3.68 35% 8 / 29 0.0031 78.87 ± 2.03 2.61 247.76 ± 49.08 77.98 ± 0.29 0.0022
KM1609-D07-02 (n = 2) PLAG X2 77.47 ± 0.57 4.73 43% 21 / 60 0.0026 77.62 ± 1.08 5.01 296.33 ± 18.67 78.04 ± 0.25 0.0021
KM1609-D07-03 PLAG 69.80 ± 0.80 8.68 66% 8 / 20 0.0050 68.52 ± 1.93 7.53 310.38 ± 16.30 68.87 ± 0.47 0.0049
KM1609-D07-11 GM 100.36 ± 0.45 2.57 57% 14 / 28 0.0362 100.40 ± 0.59 3.86 296.50 ± 1.29^ 92.56 ± 0.35 0.0295
KM1609-D07-16 PLAG 77.12 ± 0.97 38.34 86% 10 / 20 0.0050 77.32 ± 2.07 43.92 295.79 ± 31.38 76.10 ± 0.30 0.0045
KM1609-D10-09 PLAG 65.01 ± 0.23 8.70 59% 21 / 30 0.3180 65.08 ± 0.32 10.31 297.71 ± 2.88 65.42 ± 0.20 0.5010
KM1609-D16-02 PLAG 67.02 ± 0.19 1.15 63% 18 / 33 0.0100 67.23 ± 0.47 1.16 286.13 ± 25.81 67.05 ± 0.17 0.0074
KM1609-D16-02* PLAG 65.82 ± 0.27 0.60 50% 14 / 31 0.0115 65.75 ± 0.41 0.63 301.07 ± 10.65 66.62 ± 0.23 0.0075
KM1609-D16-02 (n = 2) PLAG 66.76 ± 0.24 3.33 57% 32 / 64 0.0106 67.03 ± 0.38 3.10 284.83 ± 15.33 66.83 ± 0.17 0.0074
Ellice Basin (No ages)
KM1609-D01-05 GM 0 / 44 112.01 ± 0.22 0.4750
KM1609-D01-05* GM 0 / 44 110.54 ± 0.22 0.4600
KM1609-D02-14 PLAG 0 / 20 100.34 ± 0.39 0.0153
KM1609-D02-20 GM 0 / 44 82.71 ± 0.20 0.0317
KM1609-D02-20 PLAG 0 / 21 113.22 ± 1.79 0.0007
KM1609-D02-22 GM 0 / 44 69.66 ± 0.15 0.0630
KM1609-D02-22* GM 0 / 44 67.72 ± 0.15 0.0660
KM1609-D03-13 PLAG 0 / 20 53.76 ± 0.16 0.0585
KM1609-D06-18 GM 0 / 40 93.08 ± 0.24 0.1630
KM1609-D07-03 GM 0 / 28 54.43 ± 0.25 0.0196
KM1609-D11-01 GM 0 / 44 92.77 ± 0.33 0.0075
KM1609-D12-02 GM 0 / 28 57.66 ± 0.49 0.0106
KM1609-D12-12 GM 0 / 44 88.09 ± 0.21 0.0191
KM1609-D12-12* GM 0 / 39 91.43 ± 0.22 0.0182
KM1609-D15-12 PLAG 0 / 27 97.92 ± 0.19 0.0140
KM1609-D15-12* PLAG 0 / 26 97.90 ± 0.20 0.0137
KM1609-D15-20 GM 0 / 45 88.94 ± 0.18 0.3470
KM1609-D15-20* GM 0 / 42 89.85 ± 0.18 0.3340
KM1609-D16-01 PLAG 0 / 28 87.52 ± 0.36 0.0024
  • Note. Duplicate analyses are indicated by a * after the sample name and (n = #) indicates combined analyses. Our single preferred 40Ar/39Ar age from the EB is indicated in bold. Italicized ages under the “Plateau" heading are “pseudo age spectra” since they cannot be considered proper age plateaus (Schaen et al., 2021) and are discussed below.

A U/Pb zircon age was obtained from zircons separated from a gabbroic sample KM1609-D02-14 from a suite of gabbroic rocks recovered at this dredge location. All zircons appear to be single-growth magmatic zircons based on their lack of zonation observed in CL images (Figure 3a). Thirty-three zircons gave a LA-ICPMS age of 117.8 ± 1.2 Ma (2σ) from all 33 spot analyses with an MSWD of 1.4 (Figure 3b). This is interpreted as a robust eruption age due to the zircons single growth character forming within gabbroic rocks associated with MORBs near a ridge axis, and the concordant ages among all zircons within error (Hellebrand et al., 2007).

Details are in the caption following the image

(a) Cathodoluminescence (CL) image of zircon grains from KM1609-D2-14 with laser ablation pit locations (and approximate size). (b) U/Pb ages from 33 laser ablation inductively coupled plasma mass spectrometry analyses and the weighted mean final age (horizontal black line) and its associated error (gray bar), and a probability distribution (black curve). (c) U/Pb concordia diagram of the 33 analyses plotted with their uncertainties. All uncertainties plotted as 2σ.

All ages younger than 83.7 Ma (the end of the CNS) are interpreted as an alteration age because EB's oceanic crust was entirely formed during the CNS, and all samples that are younger than this age cutoff also show signs of alteration or discordance in their degassing patterns. This includes disturbed age spectra that may or may not yield an age plateau due to alteration or 39Ar or 37Ca recoil (Schaen et al., 2021) or variable K/Ca patterns and age spectra in plagioclase analyses likely due to partial sericitization (Jiang et al., 2021; Verati & Jourdan, 2013). Some plagioclase samples appear to yield reliable age spectra with age plateaus containing more than 50% of the released 39Ar(K) (e.g., KM1609-D10-09 or D07-01) but are too young for the region as mentioned above (Figures 4c and 4d), or have ages conflicting with other samples collected at the same dredge location (e.g., KM1609-D7; Table 1). All plagioclase samples with altered ages have considerably elevated K/Ca compared with the unaltered plagioclase from sample KM1609-D11-01 that has a reliable age.

Details are in the caption following the image

Selected 40Ar/39Ar age spectra, K/Ca spectra, and inverse isochrons from samples in the Ellice Basin (a–d) and the Osbourn Basin (e and f). Blue/purple samples are plagioclase and orange/red are groundmass. Ages above colored bars are plateau ages and ages below sample names are combined ages, all with associated 2σ errors.

Details are in the caption following the image

A map of the northeastern Ellice Basin (dotted black box in Figure 2) showing dredge locations. Eruption ages and alteration ages are shown in green and orange, respectively. Colored lines are pseudo-isochrons of spreading stages 1 (red), 2 (blue), and 3 (white) from Benyshek et al. (2019). Note the clear tectonic fabric.

5.1.1.1 Alteration Ages

Our 40Ar/39Ar ages record both the eruption and alteration ages of EB samples. We obtained alteration ages from 12 samples from five dredge locations (KM1609-D2, D3, D7, D10, and D16; Table 1). These ages are not as robust as ages that are interpreted as eruption ages, but they can still inform us about the maturation of oceanic crust in the EB. Some alteration ages have high MSWDs (up to 38.34) reflecting the disturbed nature of their age spectra, but others are flat and contain enough released 39Ar(K) to constitute a plateau age that is geologically meaningful (Schaen et al., 2021).

KM1609-D7-2* (a duplicate, Table 1) does not meet the age plateau criteria in Schaen et al. (2021) since it contains only 35% of the released 39Ar(K). We call this a “pseudo age spectrum” to differentiate it from an age plateau, but consider it meaningful since its age is close to a second analysis from this sample. Sample KM1609-D10-9 (Figure 4c) contains 59% of the released 39Ar(K) in another “pseudo age spectrum” that, because of slightly ascending step ages, also fails the criteria for an age plateau in Schaen et al. (2021). This sample has a high MSWD of 8.70 indicating that this analysis is underestimating actual errors due to a high dispersion between 21 high-precision individual steps that have as little as ± 0.17 Ma 2σ uncertainties. This high dispersion likely results from variations in higher and lower retention sites caused by hydrothermal and/or seawater alteration that stochastically degas throughout the experiment. However, all individual steps in this analysis are between ∼64 and 67 Ma in age, indicating that this sample has remained a closed system since at least 64 Ma and therefore provides a geologically meaningful age estimate of this alteration event.

These alteration ages in the EB range from 100.36 to 52.19 Ma. Eruption ages from KM1609-D2-14 and KM1609-D11-1 of ∼117.5 Ma implies that the oceanic crust in this region was interacting with seawater for up to 65 Myrs. Plagioclase samples that yielded alteration ages show a progressive increase in K/Ca with decreasing age (Figure 6). The K/Ca ratio ranges from 0.0009 in the oldest sample (KM1609-D11-1 = 117.20 Ma) to 0.0459 in the youngest sample (KM1609-D3-12 = 52.19 Ma) with the exception of one sample that has exceptionally high K/Ca ratio of 0.3180 (KM1609-D10-9 = 65.01 Ma). This increase in K relative to Ca in the plagioclases (which probably did not originally vary much among samples) likely reflects a progressive addition of K during sericitization (Jiang et al., 2021; Verati & Jourdan, 2013) over tens of millions of years resulting in increasingly lower 40Ar/39Ar ages. Prolonged fluid flow has been demonstrated by other studies (Fisher & Becker, 2000; Stein & Stein, 1994) and our results further demonstrate the long period over which oceanic crust remains permeable and can undergo alteration. This contrasts with some earlier studies that argued that ocean crustal alteration only lasted several to 10 Myrs after emplacement (Hart & Staudigel, 1978; Staudigel et al., 1981).

Details are in the caption following the image

K/Ca versus age of plagioclase samples from the Ellice Basin. The two oldest samples (points overlap) are KM1609-D11-1 and have the lowest K/Ca. Increases in K/Ca are directly correlated with decreases in age due to progressive alteration of the samples. All age uncertainties are smaller than the symbols.

5.1.2 Osbourn Basin

Five of 12 samples yielded 40Ar/39Ar age results interpreted as eruption ages from the OB (Table 2). Among the five samples, all had successful groundmass and duplicate analyses, all within error of each other. Four of the five samples also had plagioclase analyses, again all with duplicates, that agreed with each other and the groundmass analyses within error. Groundmass analyses yielded ages an order of magnitude more precise than plagioclase analyses due to the ultra-low K content of the plagioclase samples. Plagioclase samples have K/Ca ratios between 0.0003 and 0.0005—even lower than the low 0.0009 values from the unaltered plagioclase from EB sample KM1609-D11-01—while groundmass samples ranged from 0.0031 to 0.0142.

Table 2. 40Ar/39Ar Analyses From the Osbourn Basin (OB)
Plateau Inverse isochron Total Fusion
Sample Phase Age [Ma] Error MSWD % 39Ar(K) Steps used / Total steps K/Ca Age [Ma] Error MSWD 40Ar/36Ar intercept Error Age [Ma] Error K/Ca
Osbourn Basin (Preferred Eruption Ages)
329-U1365E-3R-1W 130/133 (n=2) GM X2 103.58 ± 0.23 1.43 73% 19 / 48 0.0126 103.59 ± 0.27 1.65 305.26 ± 2.35 104.96 ± 0.25 0.0113
329-U1365E-3R-3W 73/85 (n=4) GM X2 & PLAG X2 102.86 ± 0.26 1.53 89% 50 / 94 0.0041 102.61 ± 0.40 4.88 297.07 ± 1.34 106.70 ± 1.91 0.0023
329-U1365E-7R-1W 45/57 (n=4) GM X2 & PLAG X2 102.61 ± 0.43 1.69 86% 53 / 93 0.0019 104.84 ± 0.82 6.60 299.44 ± 0.89 118.39 ± 1.94 0.0011
329-U1365E-7R-4W 29/40 (n=4) GM X2 & PLAG X2 102.14 ± 0.44 1.67 72% 49 / 96 0.0021 103.18 ± 1.10 8.71 317.16 ± 3.17 97.44 ± 0.68 0.0013
329-U1365E-8R-2W 95/108 (n=4) GM X2 & PLAG X2 104.15 ± 0.40 1.65 65% 45 / 96 0.0040 105.41 ± 0.59 4.38 298.42 ± 0.88 104.05 ± 1.48 0.0014
Hole U1365E age (n=18) GM X2 & PLAG X2 102.60 ± 0.26 4.24 78% 218 / 427 0.0040 97.95 0.72 0.0020
Osbourn Basin (Eruption Ages)
329-U1365E-3R-1W 130/133 GM 103.77 ± 0.25 1.17 71% 9 / 23 0.0134 103.76 ± 0.32 1.60 302.50 ± 4.28^ 106.06 ± 0.39 0.0116
329-U1365E-3R-1W 130/133* GM 103.28 ± 0.25 1.10 79% 12 / 25 0.0142 103.28 ± 0.34 1.67 308.58 ± 2.43^ 103.67 ± 0.29 0.0110
329-U1365E-3R-3W 73/85 GM 102.95 ± 0.30 1.54 93% 14 / 23 0.0068 102.96 ± 0.36 1.93 288.72 ± 2.12^ 107.14 ± 0.38 0.0062
329-U1365E-3R-3W 73/85* GM 102.70 ± 0.36 1.80 85% 13 / 25 0.0074 102.90 ± 0.42 1.63 297.43 ± 1.36 105.87 ± 0.40 0.0064
329-U1365E-3R-3W 73/85 (n=2) GM X2 102.85 ± 0.27 1.71 89% 27 / 48 0.0071 102.77 ± 0.48 6.54 295.38 ± 2.14 106.52 ± 0.31 0.0063
329-U1365E-3R-3W 73/85 PLAG 105.36 ± 2.45 1.79 86% 11 / 23 0.0005 107.89 ± 3.40 1.51 296.97 ± 1.64 103.22 ± 7.59 0.0005
329-U1365E-3R-3W 73/85* PLAG 103.90 ± 4.12 0.46 89% 12 / 23 0.0005 104.58 ± 5.37 0.50 297.69 ± 4.96 111.93 ± 24.59 0.0005
329-U1365E-3R-3W 73/85 (n=2) PLAG X2 105.12 ± 1.73 1.06 88% 23 / 46 0.0005 107.11 ± 2.43 0.96 297.21 ± 1.25 107.81 ± 13.46 0.0005
329-U1365E-7R-1W 45/57 GM 102.54 ± 0.56 1.24 83% 14 / 29 0.0031 102.55 ± 0.83 1.43 306.89 ± 1.66^ 109.79 ± 2.43 0.0032
329-U1365E-7R-1W 45/57* GM 102.50 ± 0.44 0.84 92% 15 / 25 0.0036 102.52 ± 0.57 1.02 305.34 ± 1.24^ 130.47 ± 3.49 0.0032
329-U1365E-7R-1W 45/57 (n=2) GM X2 102.51 ± 0.38 1.23 88% 29 / 54 0.0033 102.53 ± 0.53 1.38 306.12 ± 1.08 119.67 ± 1.93 0.0032
329-U1365E-7R-1W 45/57 PLAG 105.25 ± 4.75 1.78 61% 8 / 23 0.0003 106.99 ± 5.78 1.92 298.01 ± 1.14 113.57 ± 5.65 0.0003
329-U1365E-7R-1W 45/57* PLAG 109.28 ± 3.73 1.62 100% 16 / 16 0.0003 111.10 ± 3.86 1.49 298.03 ± 0.58 105.47 ± 7.00 0.0003
329-U1365E-7R-1W 45/57 (n=2) PLAG X2 107.65 ± 2.98 1.74 80% 24 / 39 0.0003 109.50 ± 3.18 1.62 298.02 ± 0.52 109.51 ± 4.50 0.0003
329-U1365E-7R-4W 29/40 GM 102.36 ± 0.72 1.77 59% 9 / 25 0.0059 102.53 ± 1.50 5.70 320.31 ± 3.65^ 99.07 ± 0.98 0.0038
329-U1365E-7R-4W 29/40* GM 101.85 ± 0.47 1.24 77% 11 / 25 0.0037 101.92 ± 0.79 2.22 323.51 ± 3.05^ 93.95 ± 1.05 0.0039
329-U1365E-7R-4W 29/40 (n=2) GM X2 102.03 ± 0.43 1.53 68% 20 / 50 0.0047 102.27 ± 0.76 3.68 321.39 ± 2.27 96.52 ± 0.73 0.0038
329-U1365E-7R-4W 29/40 PLAG 105.95 ± 2.69 1.44 93% 14 / 23 0.0003 107.79 ± 3.46 1.43 294.90 ± 5.44 102.63 ± 2.34 0.0003
329-U1365E-7R-4W 29/40* PLAG 104.80 ± 2.87 1.17 96% 15 / 23 0.0003 105.33 ± 3.44 1.25 298.18 ± 3.78 102.85 ± 2.79 0.0003
329-U1365E-7R-4W 29/40 (n=2) PLAG X2 105.47 ± 1.94 1.27 94% 29 / 46 0.0003 106.53 ± 2.36 1.30 296.96 ± 3.10 102.74 ± 1.82 0.0003
329-U1365E-8R-2W 95/108 GM 104.02 ± 0.47 1.76 82% 14 / 25 0.0067 104.09 ± 0.68 2.58 307.86 ± 3.53^ 104.20 ± 0.52 0.0040
329-U1365E-8R-2W 95/108* GM 104.65 ± 0.57 0.99 43% 10 / 24 0.0080 104.62 ± 0.98 1.79 303.37 ± 2.77^ 104.46 ± 0.61 0.0040
329-U1365E-8R-2W 95/108 (n=2) GM X2 104.19 ± 0.40 1.55 63% 24 / 49 0.0073 104.33 ± 0.58 2.51 305.23 ± 2.24 104.32 ± 0.42 0.0040
329-U1365E-8R-2W 95/108 PLAG 101.17 ± 3.57 1.55 76% 12 / 24 0.0003 102.76 ± 4.35 1.65 298.05 ± 0.96 105.40 ± 7.27 0.0003
329-U1365E-8R-2W 95/108* PLAG 101.93 ± 5.74 1.77 76% 9 / 23 0.0003 108.26 ± 9.19 1.66 297.47 ± 1.32 98.68 ± 21.26 0.0002
329-U1365E-8R-2W 95/108 (n=2) PLAG X2 101.41 ± 2.99 1.57 76% 21 / 47 0.0003 103.80 ± 3.86 1.59 297.96 ± 0.71 102.28 ± 10.61 0.0003
Osbourn Basin (Alteration Ages)
329-U1365E-11R-1W 51/71 GM 101.65 ± 0.49 0.97 73% 24 / 39 0.0691 101.66 ± 0.52 1.03 280.43 ± 1.59^ 100.70 ± 0.53 0.0146
329-U1365E-11R-1W 51/71 PLAG 90.65 ± 1.05 1.32 72% 10 / 22 0.0015 91.39 ± 3.47 1.46 287.72 ± 51.12 93.10 ± 0.98 0.0015
Osbourn Basin (No Ages)
329-U1365E-4R-2W 80/83 GM 0 / 25 82.70 ± 0.29 0.0092
329-U1365E-4R-2W 80/83* GM 0 / 25 83.54 ± 0.28 0.0089
329-U1365E-9R-2W 37/40 GM 0 / 23 111.78 ± 0.23 0.0267
329-U1365E-9R-2W 37/40* GM 0 / 25 111.87 ± 0.23 0.0268
329-U1365E-10R-2W 50/55 GM 0 / 25 111.04 ± 0.23 0.0205
329-U1365E-10R-2W 50/55* GM 0 / 23 111.49 ± 0.24 0.0205
329-U1365E-11R-3W 64/74 GM 0 / 30 58.84 ± 0.34 0.0156
329-U1365E-12R-4W 8/11 GM 0 / 23 66.70 ± 0.27 0.0267
329-U1365E-12R-4W 8/11* GM 0 / 24 69.14 ± 0.26 0.0259
329-U1365E-12R-4W 37/47 GM 0 / 31 62.39 ± 0.62 0.0150
  • Note. Duplicate analyses are indicated by a * after the sample name and (n = #) indicates combined analyses. Combined, preferred ages for each sample from the OB are indicated, with an overall combined age (n = 18) for Hole U1365E indicated in bold. 40Ar/36Ar intercept values that were used as the trapped component for age spectra calculations are indicated with a ^.

Many groundmass analyses yield robust inverse isochrons that do not have atmospheric trapped 40Ar/36Ar components, and fall outside the uncertainty of the 298.56 ± 0.31 atmospheric value (Lee et al., 2006). In these cases, the trapped components were used to calculate the age spectra (Heaton & Koppers, 2019). Using these trapped values in all cases increased the concordance of the age spectra and reduced their associated MSWD values. All plagioclase analyses had trapped 40Ar/36Ar components that overlapped within 2σ confidence limits with the above Lee et al. (2006) atmospheric value, so this atmospheric composition was assumed during age calculations for all plagioclase samples.

One groundmass analysis from sample 329-U1365E-8R-2W 95/108 contains <50% of the released 39Ar(K) (italicized in Table 2), but overlaps in age with the other groundmass analysis from this sample. These two analyses result in a combined age of 63% of the released 39Ar(K) and should therefore be considered reliable.

Two samples yielded ages that are not likely eruption ages but alteration ages. One plagioclase sample is 90.65 ± 1.05 Ma and has a higher K/Ca of 0.0015 compared to the other plagioclase analyses. A groundmass analysis from the same sample yielded an age of 101.65 ± 0.49 Ma and does not agree with the stratigraphy, and also has a higher K/Ca than all other groundmass samples, so both are likely alteration ages.

Combined groundmass ages all have plateau MSWDs under the 2σ confidence limit and are generally better than the individual analyses. This is less true for inverse isochron MSWDs where only one combined analysis is within the 2σ confidence limit. Three of the four combined plagioclase ages have acceptable age plateaus with MSWDs within the 2σ confidence limit. Combined eruption ages (from both groundmass and plagioclase) range from 102.14 ± 0.44 Ma to 104.15 ± 0.40 Ma. These are all within the Re/Os age uncertainty of Zhang and Li (2016) and come from both upper N-MORB and lower D-MORB geochemical groups (Figure 7). A combined age from all 18 of the analyses interpreted as eruption ages yields a single plateau age of 102.60 ± 0.26 Ma for the entire hole.

Details are in the caption following the image

Preferred 40Ar/39Ar ages (black circles with 2σ errors) and Zr/Nb (red diamonds) with depth from International Ocean Discovery Program Hole U1365E in the Osbourn Basin. Zr/Nb includes labels for the three geochemical zones identified by Zhang et al. (2012). The light gray area shows the age uncertainty of the Zhang and Li (2016) Re/Os isochron age and the darker gray area shows the age uncertainty for our combined (n = 18) plateau age for the entire Hole U1365E. The dark gray curve at the top is a probability density function (PDF) for the suite of our new 40Ar/39Ar analyses.

5.2 Ellice Basin Whole-Rock Major and Trace Element Geochemistry

Thirty samples from 12 dredge locations had major element analyses performed by XRF and trace element analyses by ICP-MS (Table 3). Samples are mostly basalts and trachy-basalts, but also include four basaltic trachy-andesites and one basanite (although as an intrusive rock it is technically a foid-gabbro). Most samples plot as alkali-basalts in a TAS diagram (Figure S1 in Supporting Information S1), though the least-altered samples are tholeiitic and most alkali basalts are probably due to their alteration (discussed below). Samples show a wide range of loss on ignition (LOI) values ranging from 0.64% to 6.99%, and major element abundances with K2O contents of 0.18%–3.69%, MgO from 1.18% to 7.73%, FeOt from 8.11% to 25.04%, and TiO2 from 0.87% to 4.72%. Much of this compositional diversity is the result of a single sample (the foid-gabbro), which is an outlier; it has the lowest SiO2, K2O, and LOI contents and highest TiO2 and FeOt contents of any sample. This sample is from the suite of gabbros that were recovered from dredge KM1609-D2 and contains a high abundance of large iron and titanium oxides. This sample is also uniquely important among the samples as it was the only one found to contain zircons from which we were able to determine the U/Pb age. Among the remainder of the samples, much of the compositional heterogeneity in major elements such as K2O and LOI can be attributed to alteration.

Table 3. Major and Trace Element Geochemistry for the Ellice Basin
Sample SiO2 TiO2 FeOt MgO Al2O3 CaO Na2O K2O P2O5 MnO LOI Total
KM1609-D01-05 46.66 2.66 10.14 2.22 17.64 6.52 2.92 2.24 1.20 0.11 6.99 99.30
KM1609-D01-20 48.46 1.40 8.58 7.73 16.75 10.47 2.20 0.93 0.14 0.13 2.82 99.60
KM1609-D02-03 44.75 2.01 12.81 1.70 16.49 8.67 2.66 2.26 2.66 0.21 5.23 99.46
KM1609-D02-14 47.13 2.38 13.77 1.98 17.17 5.43 2.72 2.05 0.72 0.28 5.77 99.40
KM1609-D02-20 44.76 2.75 14.58 1.58 16.47 6.61 2.97 1.56 1.03 0.38 6.56 99.24
KM1609-D02-21 47.92 2.33 12.41 1.80 17.10 5.80 2.67 2.43 0.78 0.14 5.90 99.27
KM1609-D02-25b 49.34 1.64 11.54 3.25 16.81 7.52 3.03 1.78 0.16 0.20 4.11 99.37
KM1609-D03-12 46.98 3.02 12.50 1.18 16.47 6.45 3.54 1.65 1.70 0.18 5.91 99.58
KM1609-D03-13b 49.73 2.26 11.58 1.31 17.26 4.65 3.03 2.95 0.44 0.18 5.88 99.29
KM1609-D04-04 49.08 1.44 8.39 7.12 16.39 10.98 2.75 0.33 0.13 0.15 2.82 99.58
KM1609-D04-06 47.69 1.40 11.50 3.11 18.24 7.22 2.64 1.60 0.21 0.18 5.52 99.31
KM1609-D06-13b 47.84 1.15 10.21 2.08 18.62 7.53 2.58 2.34 1.49 0.22 5.49 99.56
KM1609-D06-15 48.03 2.84 9.34 1.55 16.96 6.82 3.24 3.69 2.19 0.19 4.56 99.41
KM1609-D07-02 48.74 1.49 8.11 4.94 18.26 10.84 2.70 0.78 0.15 0.13 3.33 99.48
KM1609-D07-03 50.44 2.70 9.19 2.80 18.15 5.42 3.59 2.81 0.69 0.08 3.75 99.61
KM1609-D07-11 49.47 0.87 9.38 6.37 16.31 11.96 2.69 0.49 0.08 0.19 1.75 99.57
KM1609-D07-12b 42.66 4.72 25.04 3.98 12.42 5.80 3.11 0.18 0.41 0.44 0.64 99.40
KM1609-D08-01c 46.89 1.85 10.57 6.67 17.66 8.17 2.54 0.94 0.17 0.23 3.69 99.38
KM1609-D08-07 48.55 2.40 14.01 2.08 17.66 5.21 2.92 2.16 0.39 0.20 4.14 99.71
KM1609-D10-03b 45.88 2.77 14.74 2.11 15.91 5.45 3.10 1.86 0.71 0.16 6.97 99.65
KM1609-D10-09 48.38 2.30 12.58 1.61 17.46 6.72 3.01 2.12 1.28 0.06 3.96 99.47
KM1609-D11-01 47.96 1.82 11.03 1.89 19.27 7.22 3.27 1.71 0.27 0.12 5.12 99.67
KM1609-D11-03b 50.37 1.49 9.33 2.84 19.18 8.25 2.95 1.90 0.36 0.16 2.57 99.41
KM1609-D12-04b 49.01 2.02 9.65 5.08 16.78 10.85 3.24 0.77 0.30 0.13 1.73 99.56
KM1609-D15-01c 49.06 2.71 14.30 1.43 17.05 5.23 3.47 2.26 0.49 0.23 3.49 99.71
KM1609-D15-12 50.77 2.28 12.49 1.32 18.02 4.81 3.33 2.60 0.39 0.12 3.45 99.60
KM1609-D15-20 46.34 2.49 8.78 1.39 18.43 9.39 3.54 2.71 2.43 0.07 3.95 99.53
KM1609-D16-01b 46.10 3.13 10.75 1.42 17.33 6.30 3.11 3.10 1.74 0.22 6.37 99.57
KM1609-D16-04 49.57 1.83 10.26 3.85 17.75 8.31 3.22 1.39 0.30 0.15 3.03 99.65
KM1609-D16-05 48.74 1.56 9.25 2.57 20.17 8.61 2.99 1.48 0.60 0.12 3.38 99.48
Sample Rb Sr Y Zr Nb Cs Ba La Ce Pr Nd Sm Eu Dy Yb Lu Hf Pb Th U
KM1609-D01-05 30.8 379 44.30 240 27.74 1.32 145 30.15 50.82 7.43 31.54 7.37 2.46 7.67 3.49 0.53 5.41 1.62 2.08 0.86
KM1609-D01-20 39.2 359 45.31 246 28.62 1.56 159 36.33 53.11 8.25 34.97 8.03 2.60 7.94 3.26 0.49 5.68 1.84 2.20 0.78
KM1609-D02-03 9.0 108 22.63 58 0.91 0.34 24 3.25 5.64 1.21 6.61 2.56 1.11 4.90 2.64 0.40 1.70 0.26 0.10 0.18
KM1609-D02-14 28.1 134 92.85 334 13.28 1.20 28 9.03 30.30 5.59 32.12 11.52 4.95 18.57 10.40 1.68 8.64 0.87 0.22 0.16
KM1609-D02-20 10.9 100 28.68 79 0.38 0.30 15 2.05 7.24 1.42 8.35 3.35 1.25 5.43 2.71 0.40 2.29 0.55 0.08 0.10
KM1609-D02-21 14.1 118 30.91 122 1.22 0.58 22 4.28 11.63 2.24 12.12 4.17 1.46 6.29 2.97 0.44 3.23 0.50 0.11 0.16
KM1609-D02-25b 42.8 205 136.48 142 2.65 1.71 65 38.29 15.40 7.46 35.73 9.85 3.18 16.84 9.53 1.55 3.88 4.66 0.22 0.98
KM1609-D03-12 44.4 139 41.95 153 6.56 1.73 60 7.76 19.50 3.20 16.24 5.51 1.91 8.39 4.25 0.66 4.16 0.58 0.52 0.42
KM1609-D03-13b 46.4 155 63.29 151 6.54 1.87 60 14.08 25.04 4.17 20.47 6.33 2.17 10.40 5.65 0.91 4.05 6.69 0.85 0.49
KM1609-D04-04 46.8 151 87.73 192 7.59 2.08 50 25.40 22.69 6.10 30.05 8.72 2.89 14.02 6.92 1.08 5.11 1.21 0.62 0.73
KM1609-D04-06 34.0 189 121.49 197 8.01 1.38 104 44.40 35.27 8.60 40.43 10.92 3.54 17.36 9.00 1.45 5.26 11.29 1.09 1.11
KM1609-D06-13b 56.2 143 75.02 146 4.86 2.03 40 30.54 16.96 5.40 25.78 7.11 2.34 11.04 5.50 0.86 4.02 1.36 0.38 0.66
KM1609-D06-15 58.4 158 137.31 145 4.85 2.35 24 82.05 16.56 13.09 58.36 12.89 3.82 16.67 7.33 1.16 3.98 3.01 0.37 0.69
KM1609-D07-02 35.2 177 33.41 122 4.54 1.30 62 7.28 13.73 2.78 13.89 4.52 1.66 6.86 3.24 0.50 3.35 1.12 0.40 0.34
KM1609-D07-03 39.2 154 37.24 91 3.47 1.44 38 8.13 11.12 2.40 12.10 3.90 1.49 6.24 3.16 0.50 2.56 0.89 0.30 0.31
KM1609-D07-11 13.9 249 38.86 161 13.45 0.91 57 12.96 27.53 4.25 19.50 5.49 1.94 7.23 3.55 0.56 3.93 1.04 1.07 0.34
KM1609-D07-12b 31.8 145 26.94 93 3.52 1.08 58 4.71 12.73 1.94 10.09 3.56 1.41 5.77 2.71 0.39 2.64 1.80 0.32 0.29
KM1609-D08-01c 40.6 208 125.55 204 5.73 1.75 57 47.52 24.42 9.50 43.57 11.55 3.64 17.09 8.85 1.40 5.36 3.65 0.44 1.09
KM1609-D08-07 52.8 190 65.15 186 5.03 1.98 91 14.57 21.10 4.72 24.02 7.38 2.61 11.75 5.88 0.92 4.98 1.26 0.37 0.86
KM1609-D10-03b 51.5 148 54.02 149 4.80 1.93 47 20.05 17.19 4.69 22.36 6.27 2.16 9.34 4.46 0.70 3.89 5.63 0.33 0.76
KM1609-D10-09 52.5 156 52.86 150 4.02 2.19 49 23.53 16.29 6.36 28.95 7.62 2.50 9.32 4.04 0.62 3.89 5.18 0.33 0.68
KM1609-D11-01 7.5 141 30.44 98 1.59 0.41 5 3.31 10.07 1.87 10.42 3.55 1.43 5.74 2.85 0.43 2.59 0.44 0.13 0.14
KM1609-D11-03b 27.4 93 36.81 73 1.12 1.00 24 3.36 7.09 1.47 8.63 3.41 1.33 6.76 3.67 0.56 2.25 0.24 0.10 0.26
KM1609-D12-04b 40.3 168 109.24 68 2.75 1.54 64 56.90 11.32 8.54 37.66 8.41 2.60 12.21 5.83 0.94 1.85 3.79 0.27 0.59
KM1609-D15-01c 49.0 387 158.27 259 39.91 1.38 225 131.18 70.67 22.73 95.33 18.90 5.47 19.32 7.86 1.22 6.06 10.29 3.08 1.26
KM1609-D15-12 53.5 494 152.63 234 37.74 2.79 213 124.63 62.58 19.58 81.12 15.73 4.75 17.16 7.54 1.20 5.46 5.74 3.17 1.36
KM1609-D15-20 40.9 413 135.53 280 41.21 1.62 234 86.96 69.12 17.57 75.70 16.18 4.89 17.10 7.23 1.15 6.52 5.22 3.15 1.35
KM1609-D16-01b 16.7 210 24.11 100 6.16 1.07 29 6.43 15.74 2.40 11.57 3.51 1.34 4.79 2.19 0.33 2.61 1.29 0.46 0.34
KM1609-D16-04 28.9 222 36.57 124 7.74 1.53 41 11.38 19.22 3.44 16.26 4.73 1.75 6.69 3.13 0.49 3.23 0.77 0.60 0.42
KM1609-D16-05 30.6 234 50.52 107 6.55 1.57 43 25.20 16.12 5.43 24.76 5.94 2.04 7.72 3.40 0.53 2.73 2.04 0.52 0.49
  • Note. All major elements in un-normalized percent and all trace elements in ppm.

The trace element geochemistry shows samples are generally enriched in (fluid) mobile, incompatible trace elements such as Cs, Rb, Ba, and U, though great variation exists among samples. LREEs vary widely in primitive mantle-normalized (PM) La values from 191 to 3 with a somewhat even spread between. HREE values show much less variation with Yb values ranging from 21 to 4. A considerable negative Ce anomaly (Ce/Ce* = (2 × Cech)/(Lach + Prch) where ch is chondrite-normalized) is present in some samples to values as low as 0.1 (Figure 8b). Unaltered samples should have Ce/Ce* = 1, whereas “negative” values (<1) have been observed in seawater-altered basalts (Ludden & Thompson, 1979; Masuda & Nagasawa, 1975; Menzies et al., 1977). This Ce/Ce* is the most indicative of the degree of alteration and is inversely correlated with La/Sm in the samples that vary from PM normalized values of 5.1 to 0.4. The La/Sm ratio is also broadly correlated with LOI. LOI is most closely correlated with samples K2O and implies K2O variation is mainly alteration-based.

Details are in the caption following the image

(a) Primitive mantle (PM) normalized La/Sm versus Zr/Nb discrimination diagram showing Ellice Basin (EB) samples compared to Ontong Java (Fitton & Godard, 2004; Mahoney et al., 1993; Shafer et al., 2004; Tejada et al., 19962002), Manihiki (Golowin et al., 20172018; Hoernle et al., 2010; Ingle et al., 2007; Timm et al., 2011), Louisville (Beier et al., 2011; Hawkins et al., 2013; Nichols et al., 2014; Vanderkluysen et al., 2014; Zhong et al., 2017), the Tuvalu Seamounts (Finlayson et al., 2018), and Pacific mid-ocean ridge basalt (MORB) (Gale et al., 2013 and references therein). Comparison fields are 80% of the 2-D kernel density estimation. EB symbols represent their designated degree of alteration (see discussion). (b) REE diagram of EB samples compared to the Tuvalu Seamounts, the Ontong Java Plateau, and Pacific MORB (same references as above). In both panels, warmer colors indicate higher Ce/Ce* anomaly (a proxy for alteration) while greener colors indicate less.

5.2.1 Geochemical Alteration

Samples highly affected by alteration do not retain the same geochemistry as the erupted lavas, making it difficult to assess the contributing mantle source. Thin section and hand sample observations and the variation in LOI suggest a high variability in the degree of alteration among EB samples, but estimating the degree of alteration for any individual sample is difficult. LOI (i.e., H2O and CO2) is sensitive to and typically a good indication of alteration (Staudigel, 2003), but its lack of a strong correlation with other commonly used alteration indicators such as alkali and soluble incompatible elements makes these metrics difficult to use. A stronger correlation between Ce/Ce* with La/Sm, K2O, P2O5, and LOI suggests that Ce anomaly is a better indicator of alteration in these samples. This correlation is demonstrated in a REE diagram and by comparing Ce/Ce* to a sample's Zr/Nb (Figure 8), a mantle source fingerprint. Both Zr and Nb are highly incompatible and immobile elements and are some of the least likely elements to be affected by seawater alteration (Bienvenu et al., 1990; Ludden & Thompson, 1979) and therefore serve as more reliable source indicators. Zr/Nb values among samples from a single dredge location are remarkably homogeneous compared to the variation among all EB samples and in general do not overlap with other locations (except for dredge KM1609-D2). The uniformity in Zr/Nb among samples from a single dredge location relative to other locations would be highly unlikely if this ratio were affected by alteration and reinforces the reliability of this ratio as an unchanged geochemical fingerprint of a mantle source.

The light REEs (LREEs) are relatively mobile compared to the heavy REEs in the oceanic crust, with varying enrichments or depletions depending on the specific location (Bienvenu et al., 1990; Ludden & Thompson, 1979; Verma, 1992). Ce anomalies (Ce/Ce*) along with changes in the other REEs have been observed in altered submarine basalts in various other studies (Ludden & Thompson, 1979; Masuda & Nagasawa, 1975; Menzies et al., 1977; Staudigel et al., 1996; Verma, 1992; Worthington et al., 2006). Ce only exists in a +3-valence state in the mantle, but at surface conditions it is oxidized to its insoluble +4 state (Neal & Taylor, 1989; Schreiber et al., 1980). This results in seawater being highly depleted in Ce relative to the other LREEs (Schijf et al., 1991). Therefore, during submarine alteration of EB lavas, all LREEs except Ce are added to the basalts from seawater, as was also recognized in lavas dredged from the OT (Worthington et al., 2006). This results in increasingly negative Ce anomalies and increasing La/Sm ratios with increasing levels of alteration (Figures 8a and 8b).

If we use the Ce anomaly of a sample to discern between fresh and altered samples, some threshold value needs to be used, and there are two obvious breaks in EB data providing three reasonable categories: Ce/Ce* > 0.85 are fresh, 0.85 > Ce/Ce* > 0.60 are moderately altered, and Ce/Ce* < 0.60 are significantly altered. Samples falling into the significantly altered category (circles in Figure 8a) likely have been highly disturbed in both their major and trace element geochemistry and are not considered in our discussion on the EB mantle source (see below). We posit that samples with Ce/Ce* > 0.85 (squares in Figure 8a) closely reflect primary geochemical signatures and can be used to understand the ocean crust genesis in the EB and in the remainder of this paper we focus on those samples. Samples with Ce/Ce* between 0.60 and 0.85 (triangles in Figure 8a) are considered somewhat altered and are approached with caution, and in general not considered in the rest of this discussion.

Most EB samples with low Ce anomalies (Ce/Ce* ∼ 1) are depleted with Pacific MORB-like or possibly OJP/MP affinities (Figure 8). Only five samples, all from dredges KM1609-D1 and D15, appear similar to the more enriched Louisville or Tuvalu Seamounts.

The KM1609-D11-01 plagioclase sample that provides the only 40Ar/39Ar eruption age falls decidedly in the unaltered group with a Ce anomaly of 0.98. The samples chosen for Nd and Pb isotopic analyses have Ce/Ce* between 0.81 and 1.02 indicating that they are all in the unaltered to somewhat altered groups, with the exception of sample KM1609-D2-14 for which we have a zircon U/Pb age, and has a Ce/Ce* of 0.28 that indicates this sample is altered and so should be considered with caution.

5.3 Ellice Basin Pb and Nd Isotopes

We compare measured and initial Pb and Nd isotopic values (Table 4) with Pacific MORB, Ontong Java, Manihiki, Louisville Seamounts, and the Tuvalu Seamounts (Figure 9). The three samples with the lowest radiogenic Pb appear to most closely resemble Pacific MORB in Pb isotope space. Two of these samples also plot similarly in 143Nd/144Nd(i) versus 206Pb/204Pb(i), but the third appears more similar to Ontong Java with lower radiogenic Nd for its low radiogenic Pb compared to Pacific MORB. All three of these samples have slightly lower 208Pb/204Pb(i) and 207Pb/204Pb(i) values for their respective 206Pb/204Pb(i) compared to OJP samples. The other three samples all have higher radiogenic Pb values and appear more closely related to the Tuvalu or Louisville Seamounts in 207Pb/204Pb(i) versus 206Pb/204Pb(i). The sample KM1609-D15-12 with the highest radiogenic Pb also has the highest 143Nd/144Nd(i) and appears to be an outlier, possibly due to alteration. The other two samples plot close to the Louisville Seamounts, except in 208Pb/204Pb(i) versus 206Pb/204Pb(i), both of which have lower 208Pb/204Pb(i).

Table 4. Nd and Pb Isotopic Data for the Ellice Basin
Sample Age (Ma) 143Nd/144Nd 143Nd/144Nd(i) 206Pb/204Pb 207Pb/204Pb 208Pb/204Pb 206Pb/204Pb(i) 207Pb/204Pb(i) 208Pb/204Pb(i)
KM1609-D01-05 105.92 0.512926 0.512824 19.555 15.624 38.876 18.929 15.594 38.394
KM1609-D02-14 117.80 0.513024 0.512850 18.360 15.496 37.777 18.114 15.484 37.669
KM1609-D02-20 117.80 0.513165 0.512970 18.857 15.519 37.801 18.614 15.507 37.743
KM1609-D07-11 100.29 0.513026 0.512910 19.185 15.593 38.677 18.825 15.576 38.313
KM1609-D11-01 117.01 0.513186 0.513022 19.018 15.530 38.095 18.617 15.511 37.972
KM1609-D15-12 90.21 0.513337 0.513265 20.038 15.629 39.235 19.802 15.618 39.059
  • Note. Time corrections were made using the age in the second column.
Details are in the caption following the image

Time-corrected initial (a) 208Pb/204Pb(i), (b) 207Pb/204Pb(i), and (c) 144Nd/143Nd(i) versus 206Pb/204Pb(i) for Ellice Basin (EB) samples (black circles). Data from Ontong Java (Shafer et al., 2004; Tejada et al., 199620022004), Manihiki (Golowin et al., 2018; Hoernle et al., 2010; Ingle et al., 2007; Timm et al., 2011), Louisville (Beier et al., 2011; Vanderkluysen et al., 2014), and Pacific mid-ocean ridge basalt (Gale et al., 2013 and references therein) are indicated as fields that represent 80% of the 2-D kernel density estimation and EB seamounts are open diamonds (Finlayson et al., 2018). Mantle isotopic endmembers include enriched mantle I (EMI), enriched mantle II (EMII), high radiogenic Pb (HIMU), and the depleted mantle (DMM).

6 Discussion

6.1 Morphology of Ellice Basin Dredge Locations

The morphology of the features that were dredged can help us to understand how the recovered samples from each site might help to constrain the spreading history of the EB. Since our main goal is to constrain the timing of EB formation, the recovery of samples that erupted at mid-ocean ridges is essential. Overprinted features such as seamounts that erupted after MOR crustal formation cannot be used to constrain the source and timing of EB formation. We have the most confidence in having recovered in situ MORB at dredge locations where clear abyssal hill and fracture zone fabric associated with on-axis spreading are recognized in the bathymetry. The best examples of these come from dredges KM1609-D4 through D10 along the northern boundary of the main cohesive spreading segment of the northeastern EB (Figures 2 and 5). Overprinted features such as small (mono-genetic?) seamounts are abundant in the region between dredges KM1609-D10 and D11 among several larger composite seamounts (Figure 5). It is unclear whether these smaller edifices were emplaced during the opening of the EB as near-axial volcanism or whether they are associated with the off-ridge emplacement of later intra-plate seamounts in the region, such as the Rurutu or Tuvalu Seamounts. Of the 16 dredge locations, at least 10 appear not to be influenced by these overprinted features.

Five dredge locations are important for this study: KM1609-D1, D2, D7, D11, and D15 (Figure 10). These locations include those with the most reliable major and trace element results as well as the most conclusive ages and isotopic results. Of these five locations, dredges KM1609-D1 and D15 appear to have morphologically overprinted features. The feature dredged for KM1609-D1 (Figure 10a) is symmetric and elongated in the ENE/WSW direction, the same as the regional fracture zones, and is broad in overall shape. KM1609-D15 (Figure 10e) dredged a similarly broad and elongated feature but in the NNE/SSW direction, which is parallel to the regional abyssal hill fabric. For these reasons, as well as the geochemistry of these samples discussed in the results, we interpret these features to reflect later-stage volcanism taking advantage of pre-existing crustal structures and not primary axial features formed at the spreading center. While these features help to better understand this later-stage volcanism in the EB, they do not help in constraining the formation of the EB.

Details are in the caption following the image

Detailed bathymetric maps of five important dredge locations (a) KM1609-D1, (b) KM1609-D2, (c) KM1609-D7, (d) KM1609-D11, and (e) KM1609-D15. White stippling indicates sediment-filled basins, thick white dashed lines represent fracture zones, thinner dashed lines are abyssal hill fabric, and thick red lines are approximate dredge line traces.

The features dredged from KM1609-D2, D7, and D11 show a significantly different character (Figures 10b–10d, respectively). Abyssal hill fabric from faulting on-axis during emplacement at an MOR is clearly observed at all three locations. In the case of KM1609-D2, they are less pronounced and located east and southeast of the dredge site. All three locations are along the fault scarps of fracture zones. Dredges KM1609-D2 and D11 are located near the bend in the fracture zones at the transition between the two main spreading stages (Benysheck et al., 2019). Both are examples of inside corner highs during this transition period and feature low sediment filled basins on the downslope side of the fault scarp that was dredged (Benyshek et al., 2019). These locations likely sampled primary axial features formed at the spreading center and should reflect the age and geochemistry of in situ MOR crust, thereby helping to constrain the formation of the EB itself.

6.2 Spreading History of the Ellice Basin

The timing of separation between OJP and MP after forming as part of the OJN LIP is still up for debate (e.g., Golowin et al., 2018). Our older than expected age for the EB calls into question the validity of the Taylor hypothesis, but if we assume it to be correct, our new ages enable a schematic reconstruction for the breakup of Ontong Java Nui (Figure 11). Dredge locations KM1609-D2 and D11 have the strongest evidence for in situ MORB emplaced at the time of opening. These dredges are also located at the transition between spreading stages 1 and 2 (Benyshek et al., 2019) as indicated by their locations at the bend in the fracture zones (Figure 5). This reorientation in the spreading direction must have occurred at the same time throughout the EB; therefore, despite dredges KM1609-D2 and D11 being located along two different spreading center segments (flowlines), they should have the same crustal formation age. Our overlapping 40Ar/39Ar plagioclase and U-Pb zircon ages support this conclusion and improve the confidence in each of these ages, with crustal formation occurring at ∼117.5 Ma. This age is older by ∼15–20 Myrs than Benysheck et al.’s (2019) estimation of 98–102 Ma for the spreading stages 1 and 2 transition in the EB. In fact, this is very shortly after the estimated timing of the onset of breakup between OJP and MP of 120 Ma (Chandler et al., 2012). Using this breakup initiation age of 120 Ma implies an unprecedentedly fast spreading rate for stage 1 opening of the EB between ∼60 and 100 cm/yr. This is at least 4X faster than the current fastest spreading rate globally of ∼14.5 cm/yr at the East Pacific Rise (DeMets et al., 2010) and at least 3X faster than the fastest spreading rate known globally in the geologic past of ∼21 cm/yr (Wilson, 1996).

Details are in the caption following the image

Schematic reconstruction of the breakup of Ontong Java Nui at four critical times. (a) 123 Ma when breakup initiated. (b) 119 Ma spreading is well underway in the Ellice Basin (EB) and has begun in the Osbourn Basin (OB). (c) 115 Ma, shortly after the reorientation event in the EB and before spreading finished, and before the reorientation in OB. (d) 105 Ma, after spreading has finished in the EB and during the reorientation event in the OB. The Ontong Java Plateau is green, Manihiki is dark red, Hikurangi is blue, and the northeastern and southeastern Manihiki sub-fragments are orange and purple, respectively (subducted/obducted since). Lighter shaded regions show parts of the plateaus that were likely present but have been subducted. Black lines show fracture zones, gray lines are abyssal hill fabric, and red lines are active spreading centers. M series magnetic isochrons based on (Seton et al., 2020) are shown by colored lines with labeled ages. The locations of our two eruption ages in the EB are shown as stars.

This ultra-fast spreading rate, implied by our ages in the EB, may also suggest an alternative situation, namely that OJP and MP may have rifted, broke up and began to separate before 120 Ma and EB formation thus may have started (and possibly even finished) during the formation of OJP and MP and/or Ontong Java Nui. The existing 40Ar/39Ar ages for MP span some 10 Ma from 126.0 to 115.8 Ma (Hoernle et al., 2010; Ingle et al., 2007; Timm et al., 2011). The youngest ages come from DSDP Hole 317A in the southeastern Manihiki high plateau range from 115.8 to 117.3 Ma with large uncertainties up to ±8.0 Ma (Hoernle et al., 2010). The older MP ages come from samples dredged from within the Danger Islands Trough that bisect MP and are all older than DSDP Site 317, between 122.9 and 126.0 Ma, and are more precise, with uncertainties as low as ±0.8 Ma (Timm et al., 2011). The Danger Islands Trough is estimated to have formed around 125 Ma when predominantly strike-slip motion thinned the northern and northwestern portions of MP (Nakanishi et al., 2015). Larson et al. (2002) and others have used the onset of carbonate sedimentation at ∼119 Ma at DSDP Site 317 as the onset of rifting for the Tongareva Triple Junction, and others have adopted it for the breakup of MP and HP (Taylor, 2006; Zhang & Li, 2016). Separation between OJP and MP need not have occurred after the cessation of volcanism at OJP and MP. If separation between OJP and MP began before the cessation of volcanism at these locations, one might expect to see evidence of lava flows over-printing rift features. This interpretation appears possible, as Benyshek et al. (2019) recognized an asymmetry in distances between the margins of OJP and MP and the transition between stages 1 and 2 spreading in the EB. They attributed the increased distance to MP as caused by extensional stretching of the western MP margin. Continued volcanism at the OJP, including lava flows that obscure and flow into the western region of the EB, could also explain this asymmetry. Seismic reflection surveys from OJP's Eastern Salient into the EB show a smooth transition from overthickened oceanic plateau crust to normal oceanic crust (Coffin et al., 2006) and support this scenario.

Seismic refraction/wide-angle reflection data in the northwestern MP show a gradual thinning of oceanic plateau crust from ∼15 near the Danger Islands Trough to ∼10 km toward the Tokelau Basin east of the EB (Hochmuth et al., 2019). Hochmuth et al. (2019) provide evidence for a more prolonged emplacement of the thicker southeastern MP compared with the northwestern region, with secondary construction of the southeastern region continuing after crustal stretching had occurred in the northwest. If this secondary phase in the southeastern MP (where the youngest ages at DSDP Site 317 are located) was emplaced after an older main phase, it is feasible that the estimate of 120 Ma suggested by Chandler et al. (2012) for opening of the EB is too young. While this estimate does lie between the ages for the older and younger phases of MP volcanism, that is, between 123 and 126 Ma and 118 Ma, there is little other evidence to estimate the onset of separation.

If an older age of 123 Ma is instead used for the beginning of separation between OJP and MP (as is suggested by Nakanishi et al., 2015) and the transition from stage 1 and 2 spreading in the EB, this results in a full spreading rate of ∼30–45 cm/yr, which is about double the fastest rate in the geologic record. This estimate is a more reasonable one but would still imply an ultra-fast spreading rate beyond what has been recognized in geologic history. However, using an age of 123 Ma is potentially problematic in that it occurs before the current age at the beginning of the CNS (Ogg, 2020), and there are no recognized pre-CNS magnetic anomalies in the EB. These pre-CNS anomalies may be difficult to identify in the EB since crustal formation must have occurred during M1n time, whereby the M0r anomaly that separates M1n from the CNS spans only 0.4 Myr (Ogg, 2020). It is therefore possible that some of these shorter reversals have not been identified in the region, especially if complex tectonics occurred. The Nakanishi et al. (1992) data set that is used for the M-series isochrons in this area was created when the Nova-Canton Trough was still thought to be an extinct spreading center instead of a fracture zone, which was later recognized by Taylor (2006). A revisit of ship-based magnetic data in the Tokelau Basin could evaluate this argument for an earlier initiation of separation between OJP and MP.

Sample KM1609-D07-11 is located ∼43 km from the transition between EB spreading stages 1 and 2. If our 40Ar/39Ar groundmass alteration age of 100.36 Ma is instead to be considered an emplacement age at this location, it would mean that spreading slowed substantially after the reorientation event to about 0.25 cm/yr between the transition and that location. At this rate, it would take another ∼34 Ma to fully open the EB through stage 3, implying that spreading ceased around 66 Ma. Since the EB was opened entirely during the CNS—which ended at 83.7 Ma (Ogg, 2020)—this scenario must be excluded. Although seemingly close to an expected crustal age, it is therefore likely that this groundmass age is younger and has been partially reset during alteration or is reflecting off-axis volcanism. If we instead use the spreading estimates of Benyshek et al. (2019) for stages 2 and 3 of 5 to 3.3 cm/yr full spreading rate, then spreading likely ceased between 112 and 108 Ma respectively. After this point, both Ontong Java and Manihiki were then incorporated into the Pacific Plate.

6.3 Mantle Character Producing Ellice Basin Volcanism

The mantle source for MORB in the EB can help us understand how this basin formed relative to OJP, MP, and the broader Pacific Ocean basin. We focus on samples with “fresh” geochemistry based on the criteria outlined in the results (Ce/Ce* > 0.85). The morphology of dredge sites suggests at least two sources: primary MORB and a later overprinting source. Dredges KM1609-D1 and D15 morphologically appear to have overprinted features and have among the highest La/Sm while also having low Zr/Nb (Figure 8a—the five points that have the lowest Zr/Nb) and moderate Ce anomalies. Of the five samples from these two locations, all five plot away from any other samples and appear most closely related to ∼90 Ma Tuvalu or the younger Louisville Seamounts (Figure 8). These two EB locations also show the highest radiogenic Pb component compared to the others and again appear most likely related to the Tuvalu or Louisville Seamounts (Figure 9).

The geochemical similarities between the Tuvalu and Louisville Seamounts and the overprinted features dredged in KM1609-D1 and -D15 suggest that all three may be related to a common mantle plume source. While we were not able to determine reliable eruption ages from any samples from dredges KM1609-D1 and -D15, the 90–95 Ma ages of the Tuvalu seamounts suggest that another source for the overprinting volcanism exists in the EB in addition to the 30–40 Myr younger Rurutu hotspot volcanism (which has a distinct HIMU mantle signature) that also bisects EB (Finlayson et al., 2018). Our updated spreading history of the EB suggests that its oceanic crust was fully emplaced no later than ∼108 Ma and thus before the 90 to 95 Ma Tuvalu seamounts erupted, making them unlikely to be related to EB crustal formation. Given the similarities between the 90 to 95 Ma Tuvalu Seamounts and our overprinted features with the Louisville Seamounts, it is possible that the source of the former could be an early expression of the Louisville mantle plume, and the oldest expression of the plume yet recognized.

Besides the two overprinted features that were dredged, all other samples in the EB have a higher Zr/Nb that would indicate a Pacific MORB or potentially OJP affinity (Figure 8). Given that this basin resides between OJP and MP, and presumably formed shortly after their emplacement by one or more mantle plumes (Chandler et al., 2012; Golowin et al., 2018; Larson, 1997; Tarduno et al., 1991), it follows that the (residual) OJP plume signature would remain in the upper mantle and was large enough so that plate motions would not have moved the EB away from this region of the mantle. EB samples have greater compositional diversity than the entirety of OJP with several samples having considerably higher Zr/Nb values (Figure 8a). The higher Zr/Nb values suggest a more typical Pacific MORB affinity, potentially due to the lower degrees of partial melting compared to those during the emplacement of OJN. The residual upper mantle could have also become depleted after the high degrees of melting to create OJN, leading to more depleted MORB-like affinities generating the magma afterward.

Isotopes provide some additional insight but do not indicate one preferred mantle source. The three samples from dredges KM1609-D2 and D11 appear to show intermediate isotopic affinities between Pacific MORB and OJP. Pb space shows more MORB-like affinities, while Nd versus Pb space shows more OJP-like affinities (Figure 9). However, sample KM1609-D2-14 is considered “altered” (see above) and has the lowest radiogenic Pb and looks most like OJP in Pb versus Nd space; therefore, it is possible this sample's isotopic signature has been altered. Discounting this sample and the two KM1609-D1 and D15 (Tuvalu/Louisville-affiliated) samples, all others appear most similar to Pacific MORB. It is possible that the isotopic character of the mantle experienced a similar change through time as it did with incompatible elements as the plume component of OJN was replaced by a more typical Pacific upper mantle.

6.4 Spreading History of the Osbourn Basin

The timing and spreading kinematics of the OB are still not fully understood, with estimates for the cessation of spreading at the OT ranging from 101 to 71 Ma (Billen & Stock, 2000; Downey et al., 2007; Mortimer et al., 2019; Worthington et al., 2006; Zhang & Li, 2016). Our 40Ar/39Ar results from IODP Hole U1365E address this uncertainty directly. A combined age from 18 analyses from Hole U1365E of 102.60 ± 0.26 Ma gives the best estimate of an emplacement age for this location. This age confirms and further constrains the Re/Os isochron age of Zhang and Li (2016) from this Hole (Figure 7) and implies that the Mortimer et al. (2019) 40Ar/39Ar plateau age of 84.4 ± 3.5 Ma (n = 16 steps) from the adjacent DSDP Site 595 is an alteration age. This younger age comes from a single plagioclase analysis whose age plateau appears disturbed, while our results are consistent among the 18 successful analyses that produced eruption age plateaus (Figures 4e and 4f, and 6). Our age implies that the OB was still spreading after EB ceased spreading, and therefore, OB has a more prolonged spreading history than EB.

IODP Hole U1365E is located at a flowline distance of ∼1150 km south of the southern margin of the Manihiki Plateau and 245 km north of the extinct OT spreading center (Figure 1). We use the initiation age of spreading in the OB of Larson et al. (2002) and others of 119 Ma. Assuming that the spreading rate is constant results in a full spreading rate of 14 cm/yr between the initiation and the time that the crust at Site U1365 was emplaced at the paleo-Osbourn spreading center. Using the more recent estimate for the initiation of spreading by Nakanishi et al. (2015) of 123 Ma results in a full spreading rate of ∼11.5 cm/yr. These rates are consistent with estimates from Zhang et al. (2012) but slower than the 19 cm/yr of Zhang and Li (2016) who used a larger distance of 1500 km between Hole U1365E and MP. If our faster spreading rate is extended to the OT, spreading would have ceased around 99.5 Ma. However, the morphology of the extinct ridge suggests that spreading slowed to between 2 and 6 cm/yr at a full spreading rate prior to cessation of spreading (Downey et al., 2007). A more reasonable estimate accounting for this slower rate is a cessation age of ∼96 Ma, which is consistent with estimates based on observed Gondwana margin features (Davy, 2014; Davy et al., 2008).

A change in abyssal hill fabric trends located 300–400 km north and south of the OT indicates that the spreading orientation changed during the formation of the OB (Downey et al., 2007). The 10–15° change from NNE/SSW closest to the Manihiki and Hikurangi Plateaus to N/S spreading closest to the OT suggests that reorientation occurred shortly before spreading ceased at the OT. Based on our 14 cm/yr rate, this reorientation event occurred at ∼105 Ma (Figure 11d), coinciding with a global reorganization event (Matthews et al., 2012) and shortly after our younger estimate for cessation of spreading in the EB. The cause of this reorientation could be the result of the onset of Hikurangi docking with the Chatham Rise (Davy, 2014; Davy et al., 2008; Sutherland & Hollis, 2001), or we speculate it could be the incorporation of the Manihiki microplate with the Pacific plate, or a combination of the two.

The West Wishbone Scarp and the Manihiki Scarp—along the eastern margins of Hikurangi and Manihiki respectively—likely formed together as a single fracture zone during the initial formation of the OB and Pacific-Phoenix spreading to the east (Downey et al., 2007). The Manihiki Plateau was almost certainly larger at the onset of spreading with two missing fragments to the northeast and east of the current plateau that were rifted away via the Tongareva Triple Junction on the Farallon and Phoenix plates, respectively (Hochmuth & Gohl, 2017; Larson et al., 2002; Taylor, 2006; Viso et al., 2005). Pacific-Phoenix and OB spreading likely began around the same time, with motion between the two semi-parallel spreading centers separated by some 750 km (the length of the eastern margin of modern Manihiki) being accommodated via the Manihiki Scarp. The average spreading rate for Pacific-Phoenix spreading was 18 cm/yr between the time of initiation and Chron 34 (Larson et al., 2002), slightly faster than our estimated rate of 14 cm/yr for the OB. The disappearance of the Manihiki Scarp roughly 650 km south of the southern Manihiki margin suggests the northern OB (i.e., the Manihiki microplate) became coupled to the Pacific Plate around 105 Ma based on Pacific-Phoenix spreading rates, which is the same time as our proposed OB reorientation event and the global plate reorganization event (Matthews et al., 2012).

7 Conclusions

We provide three new ages to constrain the breakup history of the proposed Ontong Java Nui. Our two ages from the EB of ∼117.5 Ma give the age of the spreading reorientation event there and can be used to estimate the breakup history of OJP and MP and the spreading kinematics in the EB. Our ages from the OB confirm an older age of 102.6 Ma for IODP Site U1365 and help constrain the breakup history of MP and HP and the spreading kinematics for the OB. Our geochemistry data from the EB offer a first look into the mantle signature in this region at the time of its formation. We provide several main takeaways:
  1. Stage 1 spreading in the EB was extremely rapid, with the transition between stages 1 and 2 occurring much earlier than previously thought. We estimate this transition at ∼117.5 Ma with full spreading rates for stage one between 30 and 45 cm/yr, assuming an older onset of separation of 123 Ma between Ontong Java and Manihiki.

  2. The entire EB likely finished forming and was incorporated into the Pacific Plate between 108 and 112 Ma, assuming a slow-down in full spreading rates to 3.3–5 cm/yr for stages 2 and 3.

  3. Over-printed features in the EB show geochemical similarities to the nearby 90 to 95 Ma Tuvalu Seamounts. The Tuvalu Seamounts and our overprinted features both show Louisville Seamount geochemical affinities, suggesting that some of the seamount construction in the region may be an early expression of the Louisville mantle plume.

  4. The geochemistry of EB lavas shows mantle affinities closest to Pacific MORB, with small possible OJP influences. This suggests that the upper mantle beneath the EB was not dominated by the source of Ontong Java as might be expected if a mantle plume head were the source of the OJP.

  5. The OB likely began to form during the early formation of the EB, but the former continued to spread for 12–16 Myrs after spreading ceased in the EB. A reorientation event at 105 Ma in the OB coincides with the global 105-100 Ma plate reorganization event.

  6. Spreading at the OT terminated around 96 Ma, at which time all three Ontong Java Nui sub-plateaus (OJP, MP, and HP) were incorporated into the Pacific Plate in their current configurations.

Acknowledgments

Our friend and co-author Jasper Konter passed away unexpectedly on 3 July 2022. His unique insights in helping us better understand mantle geodynamics and geochemistry, in particular when deciphering the complexity and interplay of all types of Cretaceous volcanism in the Western Pacific, will be sorely missed and never forgotten. We thank Mike Coffin and two anonymous reviewers for their thoughtful and constructive reviews. Thanks to Rob Harris, Bob Duncan, and Dave Graham for their helpful comments. Thanks are owed to the captain, crew, and science team of KM1609 including the marine technicians who were instrumental in the recovery of dredge samples. The lead author would also like to thank his father Cameron Davidson who separated and dated the zircons, which are an important age in this paper. This project was funded by the National Science Foundation (Grant OCE-1458444). This paper was included as a chapter towards the lead author's Ph.D. dissertation at Oregon State University.

    Data Availability Statement

    The 40Ar/39Ar data are available at earthref.org/ERDA/ and the U/Pb data in this paper are available at geochron.org. Major and trace element geochemistry data are available at (Davidson et al., 2023b) and the Pb and Nd isotope data are available at (Davidson et al., 2023a). All data are also available to download in the supplements.