Geochemistry, Geophysics, Geosystems

Cretaceous-to-recent record of elevated 3He/4He along the Hawaiian-Emperor volcanic chain

Authors


Abstract

[1] Helium isotopes are a robust geochemical tracer of a primordial mantle component in hot spot volcanism. The high 3He/4He (up to 35 RA, where RA is the atmospheric 3He/4He ratio of 1.39 × 10−6) of some Hawaiian Island volcanism is perhaps the classic example. New results for picrites and basalts from the Hawaiian-Emperor seamount chain indicate that the hot spot has produced high 3He/4He lavas for at least the last 76 million years. Picrites erupted at 76 Ma have 3He/4He (10–14 RA), which is at the lower end of the range for the Hawaiian Islands but still above the range of modern mid-ocean ridge basalt (MORB; 6–10 RA). This was at a time when hot spot volcanism was occurring on thin lithosphere close to a spreading ridge and producing lava compositions otherwise nearly indistinguishable from MORB. After the hot spot and spreading center diverged during the Late Cretaceous, the hot spot produced lavas with significantly higher 3He/4He (up to 24 RA). Although 3He/4He ratios stabilized at relatively high values by 65 Ma, other chemical characteristics such as La/Yb and 87Sr/86Sr did not reach and stabilize at Hawaiian-Island-like values until ∼45 Ma. Our limited 3He/4He record for the Hawaiian hot spot shows a poor correlation with plume flux estimates (calculated from bathymetry and residual gravity anomalies [Van Ark and Lin, 2004]). If 3He is a proxy for the quantity of primordial mantle material within the plume, then the lack of correlation between 3He/4He and calculated plume flux suggests that variation in primordial mantle flux is not the primary factor controlling total plume flux.

1. Introduction

[2] High 3He/4He ratios (up to nearly 50 RA [Stuart et al., 2003]) found at ocean islands such as Hawaii and Iceland are fundamental evidence for the existence of mantle plumes rising from regions deep within the Earth [e.g., Kurz et al., 1982; Lupton, 1983; Allègre et al., 1983; O'Nions, 1987], although a few have challenged this viewpoint [e.g., Anderson, 1998; Meibom et al., 2003]. A limitation to the utility of helium isotopes in improving our understanding of mantle plumes has been the lack of studies of long-term 3He/4He variability at ocean island hot spots. Almost all reported analyses from hot spots are of young rocks (<10 Ma), so unlike other radiogenic isotope systems (e.g., Sr-Nd-Pb [White et al., 1990; Weis et al., 1992; Class et al., 1993; Keller et al., 2000; Regelous et al., 2003]), little information exists on the long-term temporal evolution of hot spot 3He/4He ratios. Here we present new helium isotope data for lavas from the Hawaiian-Emperor chain of seamounts and islands which show that the Hawaiian hot spot has been capable of producing lavas with 3He/4He higher than upper mantle (MORB source) values for at least the last 76 million years.

[3] The trace of the Hawaiian hot spot is marked by more than 100 islands and seamounts that make up the Hawaiian-Emperor volcanic chain in the northwest Pacific Ocean (Figure 1). Geochemical studies revealed significant temporal variability of the Hawaiian hot spot over >80 million years of its history [Lanphere et al., 1980; Basu and Faggart, 1996; Keller et al., 2000; Regelous et al., 2003]. The oldest dated Emperor Seamount (Detroit at 76–81 Ma) has 87Sr/86Sr values essentially indistinguishable from MORB [Keller et al., 2000]. Other geochemical features of Detroit Seamount are also MORB-like, although triple-spike Pb isotope data [Regelous et al., 2003] and Ba/Th data from glasses [Huang et al., 2004] are outside of the ranges of modern MORB. With time, the hot spot produced lavas that were less MORB-like. For example, La/Yb and 87Sr/86Sr ratios in Late Cretaceous and Early Tertiary tholeiitic basalts (81–43 Ma) increase steadily southward along the Emperor chain, from MORB-like values (as low as 1.1 and 0.70262, respectively, at ODP Site 884 on Detroit Seamount), to Hawaiian Island-like values (mean of 6.4 and 0.70346) in the southern Emperor Seamounts. These higher values continue southeast of the Hawaiian-Emperor bend all the way to the Hawaiian Islands, although the range of values increases in the well-sampled islands. The temporal trend away from MORB-like compositions in the older Emperor seamounts was attributed to increasing distance between the hot spot and a seafloor spreading center, and the resultant increase in lithospheric thickness [Lanphere et al., 1980; Keller et al., 2000; Regelous et al., 2003]. Younger (thinner) lithosphere permits a taller melting column in the ascending mantle plume, leading to larger extents of partial melting [Ellam, 1992; Keller et al., 2000]. This increases the likelihood that depleted mantle, whether entrained from the surrounding upper mantle [Keller et al., 2000] or intrinsic to the plume [Regelous et al., 2003], would make a greater contribution to hot spot volcanism.

Figure 1.

The Hawaiian hot spot track in the North Pacific as shown by the Hawaiian-Emperor chain of islands and seamounts. Names are shown for the islands and seamounts sampled for this study. Ages are shown for the three Emperor seamounts from which picrites have been recovered. Modified from Shipboard Scientific Party [2002]. Suiko Seamount age is from Dalrymple et al. [1980]; Detroit and Koko ages are from Duncan and Keller [2004].

[4] Individual ocean islands often show spatial and temporal variability in 3He/4He that is usually explained by intrinsic isotopic heterogeneity within the plume, or by mixing between high-3He/4He plume-derived material and lower-3He/4He material (∼6–10 RA) derived from the MORB source region in the upper mantle [Graham, 2002]. However, little is known about the long-term 3He/4He evolution of mantle plumes because of difficulties in unambiguously measuring the initial 3He/4He in old volcanic rocks, due to factors such as posteruptive radiogenic production of 4He, low 3He contents resulting from diffusive loss, and a lack of unaltered helium-retentive minerals such as olivine or pyroxene. Tertiary lavas from the Iceland hot spot have been successfully studied [Graham et al., 1998; Marty et al., 1998; Hilton et al., 1999; Breddam and Kurz, 2001; Stuart et al., 2003], but we are not aware of published data from pre-Tertiary hot spot lavas.

[5] Suitably fresh samples from three Late Cretaceous-Early Tertiary seamounts along the Emperor Seamount chain (Figure 1) now make it possible to study the 3He/4He evolution of the Hawaiian hot spot. Samples used in this study include tholeiitic picrites recovered during Leg 197 of the Ocean Drilling Program from the shield-building phase of Detroit Seamount at Site 1203 (76 Ma), and from Koko Seamount at Site 1206 (49 Ma) (see Tarduno et al. [2003] for a summary). We also obtained picrite recovered from the shield-building phase of Suiko Seamount at Site 433 (65 Ma [Dalrymple et al., 1980]) by Leg 55 of the Deep Sea Drilling Project. Combined with alkalic and tholeiitic basalt samples from the northwest Hawaiian Islands (including Niihau, Kaula, Nihoa, and Necker islands, and La Perouse and Gardner pinnacles, as well as Northampton Bank), and previously published data from the main Hawaiian Islands (compiled by Farley and Neroda [1998]), a sufficient number of olivine-phyric lavas is now available to allow a first-order determination of the 3He/4He history of the Hawaiian hot spot since the Cretaceous. We recognize that a few samples from a few seamounts does not fully characterize the nature of volcanism at the Hawaiian hot spot, but until a full suite of samples is available from several of the Emperor Seamounts, the best comparison we can make is using similar rock types from similar stages in the volcanic development of the seamounts. We therefore concentrated on tholeiitic picrites from late in the shield-building stage of the Emperor Seamounts.

[6] The Emperor samples are picrites containing 20–30% olivine, and whole rock analyses gave 24–25 wt.% MgO for the Detroit samples [Shipboard Scientific Party, 2002] and 22–26 wt.% MgO for the Suiko samples [Shipboard Scientific Party, 1980]. The Koko core intervals analyzed for He have not been analyzed for major elements, but nearby core intervals contain 16.5–18 wt.% MgO (P. Thompson, personal communication, 2003), which suggests that these lavas may not be true picrites (i.e., >18 wt.% MgO), although we refer to them as picrites here for convenience. During hand-picking of olivine for the He isotope analyses, we found that the Detroit samples contained two distinct subpopulations of olivine distinguished by their light and dark color. We therefore separated each Detroit picrite into two aliquots for He isotopic analysis (light olivine versus dark olivine in Table 1). Representative olivine crystals from all three of the Emperor seamounts were analyzed for major element compositions by electron microprobe at Oregon State University, and their forsterite contents are included in Table 1. Olivine from Koko Seamount ranged between Fo82–89, and that from Suiko Seamount between Fo85–89. Olivine from Detroit Seamount formed two groups with virtually no overlap, the light olivine being Fo86–90 and the dark olivine being Fo78–87.

Table 1. Helium Analyses of Olivines From Emperor Seamount Picritesa
LocationAge, MaDepth, mbsfSamplePhaseFo of OlivineCrushedMelted Powder
3He/4He[He], ccSTP/g3He/4He[He], ccSTP/g
  • a

    The 3He/4He values are reported relative to the atmospheric ratio of 1.39 × 10−6. Analytical uncertainties are the 2-sigma quadrature sum of uncertainties in analyses of the sample, standards, and blanks. Detroit and Koko ages are from Duncan and Keller [2004]; Suiko age is from Dalrymple et al. [1980]; and other ages are from numerous sources compiled by Clague and Dalrymple [1987].

Detroit Seamount76591.9197-1203A-32R-3,100-105cmlt olivine89–9010.350.423.03 × 10−92.100.081.15 × 10−8
(50°57.00′N, 167°44.40′E)   dk olivine80–8210.680.553.00 × 10−90.490.071.27 × 10−8
  593.6197-1203A-32R-4,50-65cmlt olivine86–9011.080.241.91 × 10−81.380.071.44 × 10−8
    dk olivine78–87(7.30)0.612.18 × 10−90.290.041.90 × 10−8
  637.5197-1203A-38R-1,43-48cmlt olivine86–9010.910.278.68 × 10−91.640.052.29 × 10−8
    dk olivine84–8613.780.254.07 × 10−87.910.163.37 × 10−8
Suiko Seamount65315.855-433C-24-7,138-145cmolivine85–8723.520.415.78 × 10−85.540.151.05 × 10−8
(44°46.63′N, 170°1.23′E) 338.955-433C-27-2,138-144cmolivine85–8916.700.458.88 × 10−92.280.121.45 × 10−8
Koko Seamount49104.5197-1206A-10R-1,13-16cmolivine82–8724.130.541.21 × 10−80.380.051.31 × 10−8
(34°55.55′N, 172°8.75′E) 306.6197-1206A-40R-1,10-17cmolivine87–8919.850.394.22 × 10−83.100.122.10 × 10−8

2. Analytical Procedures

[7] Olivine was separated from basaltic groundmass by crushing and sieving to 0.85–1.7 mm size fraction, and hand-picking under a binocular microscope for fresh appearance and lack of attached groundmass. Olivine separates were then ultrasonically cleaned in deionized water, rinsed in acetone and air-dried before hand-picking again for the freshest grains. The ODP and DSDP samples were analyzed using procedures outlined by Graham et al. [1998]. To ensure that we were looking at the magmatic component of the helium, gas extractions were first performed by crushing the olivine in vacuum and analyzing the released helium. The crushed powders were then retrieved and melted in a high-temperature vacuum furnace and the helium released during melting was also analyzed. During the period in which these analyses were performed, line blanks were run before and after all samples. The crushing blanks were <1 × 10−10 ccSTP for 4He, and the furnace blanks at 1800°C were <5 × 10−10 ccSTP 4He. Samples from the ≤27 Ma section of the Hawaiian-Emperor chain were analyzed by in vacuo crushing at Scripps Institution of Oceanography in 1987 (A. R. Basu and H. Craig, manuscript in preparation, 2004). Analytical procedures used for those samples are as described by Farley et al. [1992, 1993].

[8] The posteruptive addition of 4He by radioactive decay of U and Th, and its potential for biasing the measured 3He/4He to lower values, is a concern for such old samples. The initial 3He/4He ratio (i.e., the magmatic value at the time of eruption) can, in principle, be distinguished by comparing the crushing and melting results for the ODP and DSDP samples. Most of the melting experiments released an amount of 4He that was similar to or greater than the amount released by crushing of the same sample (Table 1). More importantly, melting always released gas with a lower 3He/4He ratio, from about 1%–50% of the crush 3He/4He ratio (Figure 2a). If a significant amount of posteruptive radiogenic 4He had been transferred into the crushable fraction, then samples with lower [He] during crushing would display lower 3He/4He. Our results demonstrate that significant amounts of radiogenic He, while present in the crystal lattice, were not released until the crushed powder was melted. The fact that the three Fo86–90 olivine samples from Detroit gave very similar 3He/4He values despite nearly an order of magnitude difference in [He] also bears this out. The crushed ODP and DSDP analyses in Table 1 therefore accurately represent original magmatic values at the time of eruption, and have not been significantly modified by 4He ingrowth. The only exception appears to be the lowest 3He/4He value (7.3 RA) obtained during crushing of the dark olivine from Detroit sample 1203A-38R-1. In this case, almost 90% of the 4He sample budget (crushed + melted) was present as radiogenic helium in the melted fraction (Figure 2b). This very high proportion of radiogenic 4He, coupled with the relatively low 3He/4He during crushing, suggests that a significant fraction of radiogenic 4He within the crystal lattice or dissolved in melt inclusions may have also been released during the crushing. While we note that the measured value of 7.3 RA for this crushing analysis is within the range of modern MORB, we do not assign petrogenetic significance to this value given the melting results, and we do not include this sample further in our discussion.

Figure 2.

(a) Reciprocal of radiogenic 4He concentration versus 3He/4He (relative to the atmospheric ratio RA) for Emperor seamount samples. This diagram shows that the 3He/4He upon melting the crushed powder was always lower then the crushed 3He/4He ratio. This indicates that significant radiogenic 4He was released in the melted fraction but not in the crushed fraction. Data points are plotted at the crushed 3He/4He ratio and are labeled with the melting 3He/4He ratio. The crushing 3He/4He values thus appear to closely represent the original 3He/4He of the magmas (with one exception; see below). (b) The proportion of radiogenic helium, 4He*/4He (where 4He* is the amount of radiogenic 4He in a sample and 4He is the total crushed+melted 4He), versus 3He/4He (relative to the atmospheric ratio RA) for Emperor seamount samples. This diagram shows that Detroit samples consistently have the lowest 3He/4He despite a wide range of 4He*/4He. 4He*, the amount of radiogenic helium, is given by 4He* = 4Hemelt (1 − Rcrush/Rmelt), where R is 3He/4He [Graham et al., 1987]. Tie lines connect olivine separates that came from the same rock sample but were split into light (Fo86–90) and dark (Fo78–87) olivine fractions prior to isotopic analyses. All three of the Detroit samples contained these two populations of olivine, but there were no systematic isotopic differences between the light and the dark olivine. The sample with the lowest 3He/4He (7.3 RA) contains the highest proportion of radiogenic He. We suspect that that sample has been disturbed by posteruptive alteration and/or ingrowth of radiogenic 4He and do not use it in our plots or our interpretations. The four Detroit samples in the 10.3–11.1 RA range have a range of He concentrations, which is especially compelling evidence that these are reliable magmatic ratios.

3. Results

[9] Olivines from 76 Ma Detroit Seamount picrites have 3He/4He ratios of 10.4–13.8 RA, which is at the low end of the observed range for young Hawaiian lavas (7–35 RA), but higher than the modern MORB range (6–10 RA; Figure 3). Exactly how these Detroit 3He/4He values compare to those of the Cretaceous MORB upper mantle is unknown, but on the basis of detailed modeling of coupled He-N-Ne-Ar isotopes [Tolstikhin and Marty, 1998], the 3He/4He of the upper mantle does not appear to have decreased significantly over the past 500 m.y., suggesting that values >10 RA are not appropriate for Cretaceous MORB mantle. 3He/4He ratios of 16.7–23.5 RA for picritic lavas from 65 Ma Suiko Seamount, and 19.8–24.1 RA for 49 Ma Koko Seamount (Table 1) are much higher than MORB 3He/4He values (Figure 3), but in the middle part of the range of Hawaiian Island values. Lavas from the 27 to 4 Ma part of the Hawaiian-Emperor chain (including the northwest Hawaiian Islands) have 3He/4He from 8.6–17.5 RA, which overlaps with the lower half of the main Hawaiian Islands range. These northwest Hawaiian Islands values are included in Figure 3 for comparison, but will be published elsewhere (A. R. Basu and H. Craig, manuscript in preparation, 2004). The lower limit of 3He/4He for the northwest Hawaiian Island lavas appears to decrease with decreasing age, with only Kaula and Nihoa samples extending below 10 RA and approaching the low 3He/4He values that characterize the end stage of shield building volcanism on Hawaii [Kurz et al., 1983, 1996; Kurz and Kammer, 1991]. The lack of very high 3He/4He (>17.5 RA) in the northwest Hawaiian Islands samples is not surprising given that even on the well-studied main Hawaiian Islands, 3He/4He >20 RA are rare.

Figure 3.

Temporal trends in 87Sr/86Sr and 3He/4He (relative to atmospheric ratio RA) for the Hawaiian-Emperor chain, compared to plume volume flux. Age differences between the seamounts and their underlying crust are shown by the dotted green line in Figure 3b. The 87Sr/86Sr of tholeiitic basalts (shaded field in Figure 3b) from Detroit Seamount are indistinguishable from MORB but by about 45 Ma 87Sr/86Sr transitioned to values equivalent to modern Hawaii. In contrast, all 3He/4He values are distinct from MORB, even at 76 Ma, when the 87Sr/86Sr was MORB-like. Modeling suggests that Cretaceous MORB mantle also had 3He/4He values <10 RA [Tolstikhin and Marty, 1998]. Hawaiian-Emperor 87Sr/86Sr data are from Lanphere et al. [1980], Basu and Faggart [1996], Keller et al. [2000], Regelous et al. [2003], and Huang et al. [2004], plus the Hawaiian Islands data compilation by Farley and Neroda [1998]. Helium isotope data points from the <27 Ma part of the Hawaiian Ridge are unpublished data of A. R. Basu and H. Craig (A. R. Basu, personal communication, 2003). Calculated plume crustal volume flux based on bathymetry and residual gravity anomalies is from Van Ark and Lin [2004].

4. Discussion

[10] In contrast to the MORB-like La/Yb and 87Sr/86Sr values previously found at Detroit Seamount [Keller et al., 2000, Regelous et al., 2003], there is a clear plume signal in the elevated 3He/4He ratios at Detroit (although Detroit 3He/4He values are subdued compared to the highest Hawaiian Island values; Figure 3). As with 87Sr/86Sr [Lanphere et al., 1980; Keller et al., 2000; Regelous et al., 2003], the 3He/4He signal increased with time along the Emperor Seamounts, although unlike for 87Sr/86Sr, almost all of the increase in 3He/4He occurred between 76 Ma and 65 Ma as the hot spot and a seafloor spreading center diverged. While it is possible that the intrinsic geochemical and isotopic characteristics of the plume coincidentally changed as a spreading ridge passed over it, such a model is untestable and therefore unsatisfactory. Besides, several other hot spots also show less-radiogenic 87Sr/86Sr values early in their history (Kerguelen: Class et al. [1993]; Reunion: White et al. [1990]; Iceland: Stuart et al. [2003]). Iceland is the only other hot spot where the amount of He isotope data is sufficient to reveal trends in its long-term history. In contrast to the Emperors, the highest 3He/4He values for the Iceland hot spot occur very early and are accompanied by low 87Sr/86Sr [Stuart et al., 2003] (Figure 4). However, the high-3He/4He samples from the Iceland hot spot are associated with the initiation of the Iceland plume (West Greenland: Graham et al. [1998]; Baffin Island: Stuart et al. [2003]). The time of initiation of the Hawaiian plume is unknown because our volcanic record from the Hawaiian hot spot only goes as far back as the oldest remaining Emperor Seamounts. If there were Emperor Seamounts older than the ones we see today, or any evidence for the initiation of the Hawaiian plume, those features have apparently been subducted into the Kamchatka-Aleutian Trench.

Figure 4.

The 87Sr/86Sr versus the 3He/4He of samples from the Hawaiian Islands compared to data from Detroit, Suiko, and Koko seamounts. The field for Baffin Island picritic basalts (the proto-Iceland plume) is shown for comparison [from Stuart et al., 2003]. Hawaiian Islands data fields are modified from Mukhopadhyay et al. [2003]. Emperor Seamounts data are plotted as boxes using Sr data from Keller et al. [2000] and Regelous et al. [2003] because the same samples have not been analyzed for Sr and He, although the high-87Sr/86Sr end of the Detroit field includes data from Huang et al. [2004] for ODP Site 1203 samples very close to the picrites we analyzed for He.

[11] The obvious explanation for the MORB-like La/Yb, 87Sr/86Sr, and 3He/4He of Detroit Seamount is the proximity of the hot spot to a seafloor spreading center at the time Detroit formed in the Late Cretaceous central Pacific basin [Keller et al., 2000; Regelous et al., 2003]. The exact mechanism may relate to the increased decompressional melting within the plume as it was able to ascend to shallower depths beneath very young and thin lithosphere (<10 m.y. old at the time Detroit Seamount formed [Caplan-Auerbach et al., 2000]). The greater extent of melting would have exhausted the fertile, more readily fusible parts of the plume first, and then began melting the depleted (lower La/Yb, 87Sr/86Sr, and 3He/4He), normally refractory portions of the ascending plume. These normally refractory portions may have been intrinsic to the plume [Regelous et al., 2003] or they may have been derived from depleted material entrained from the upper mantle [Keller et al., 2000]. The 17–24 RA and 20–24 RA values for 3He/4He at Suiko and Koko seamounts, respectively, also fit with this thin-lithospheric lid model, as the slightly lower 3He/4He for Suiko compared to Koko is consistent with the fact that Suiko was erupted onto 35 m.y. old crust, whereas Koko erupted onto 72 m.y. old crust (Figure 3b). Alternatively, these minor differences in 3He/4He could be an artifact of our very limited sampling for the Emperor Seamounts.

[12] While the helium isotope systematics of the Hawaiian hot spot were clearly affected by some process related to the proximity of the hot spot to a spreading center, this tectonic setting had an even stronger influence on other geochemical characteristics of the hot spot. For example, the REE and 87Sr/86Sr of Detroit Seamount are nearly indistinguishable from MORB, and recovered more gradually than the 3He/4He as the spreading center and the hot spot diverged (Figure 3b). The cause of this difference in behavior is unclear, but on the basis of a comparison of Figures 3a and 3b, 3He/4He appears nearly immune to the affects of thin lithosphere once that lithosphere is older than about 10 million years. In contrast, the 87Sr/86Sr field in Figure 3b is steepest at the old end of the Emperor Seamounts, but then slowly levels off at recent Hawaiian Islands values. The suggestion that He can be a more sensitive plume tracer than Sr is not surprising given results from the Mid-Atlantic Ridge where it interacts with the Iceland plume. South of Iceland, the high 3He/4He anomaly on the Mid-Atlantic Ridge extends much farther south than the Sr isotope anomaly [Schilling, 1973; Poreda et al., 1986; Taylor et al., 1997; Schilling et al., 1998; Hilton et al., 2000]. Thus, in places on the northern MAR there is evidence for a plume-derived He isotope signal in the absence of a plume-derived Sr isotope signal (similar to what is observed at Detroit Seamount). Alternatively, the depleted upper mantle component entrained by the plume beneath Detroit Seamount may have been recently degassed during melting beneath the nearby ridge. In this case, the entrained upper mantle component, given its low [He], would have little affect on the overall He isotopes of the plume. Such ridge-related melting would deplete Sr as well, but to a lesser degree because of the greater compatibility of Sr during melting (higher solid-melt partition coefficient) as compared to He.

[13] Differences between the Emperor samples and the main Hawaiian Islands are most dramatic on a 87Sr/86Sr versus 3He/4He plot (Figure 4, modified from Mukhopadhyay et al. [2003]). Taken together, the Hawaiian Islands shield-stage lavas (Kauai, Mauna Loa, and Mauna Kea shield) form a negatively sloping array. In contrast, our picritic lavas from the Emperor Seamounts form a positively sloping array directed toward the MORB field. Thus both the Hawaiian and Emperor samples appear to share a high-3He/4He-moderate-87Sr/86Sr component similar to Loihi. The other dominant component in the Hawaiian samples is low-3He/4He-high-87Sr/86Sr, whereas the other dominant component in our Emperor samples is low-3He/4He-low-87Sr/86Sr. Although the postshield Hawaiian samples show evidence for a low-3He/4He-low-87Sr/86Sr component similar to that in our Emperor samples, the high-3He/4He component is largely subdued in the postshield samples. The data field for Baffin Island picritic basalts from the proto-Iceland plume [Stuart et al., 2003] has a negative slope, and the inferred mixing components are more akin to the Hawaiian Islands data array than to the Emperor Seamounts array, despite the fact that the Emperor samples, like the Baffin Island samples, are from early in the history of their plume (although for the Emperor samples we do not know how early). We therefore conclude that the life-stage of the plume does not appear to be the controlling factor for the Emperor data, but rather the melting regime permitted by thin, young lithosphere.

[14] Plume buoyancy flux is an additional variable that could have influenced the 3He/4He signal of the Hawaiian hot spot over time. A positive correlation between plume flux and 3He/4He might be expected because the plume carries helium with high 3He/4He from relatively undegassed deep mantle. The long-term crustal volume flux of the Hawaiian plume, on the basis of bathymetry and residual mantle Bouguer anomalies (Figure 3c) [Van Ark and Lin, 2004], shows a general increase over time. These plume flux calculations are similar to calculations based on bathymetry alone [Vidal and Bonneville, 2004], but also incorporate crustal thicknesses constrained by gravity data. There does not appear to be a simple correlation between 3He/4He and plume flux, except there may be a relationship between times of lower 3He/4He and decreasing plume flux. Although our rock sampling is obviously limited, the only two time periods when we have no 3He/4He values higher than 14 RA are times of decreasing plume flux (76 Ma at Detroit and 12–10 Ma at Gardner-to-Necker; Figure 3). Shortly after Necker formed at 10 Ma, the plume flux began to increase, and Nihoa and younger locations all have 3He/4He > 17 RA. Northampton Bank, and Koko and Suiko Seamounts all formed at times when the plume flux was relatively stable or increasing, and these times all include samples with 3He/4He ≥ 17 RA. An obvious caveat of these observations is that all of our samples are from times of moderate to high plume flux, at least in part because times of low flux are much less likely to have produced features conducive to sampling.

[15] Our study shows that it is possible to determine magmatic 3He/4He values from submarine volcanic rocks as old as 76 Ma. This demonstrates the potential to study the long-term helium isotopic behavior of hot spots, and perhaps mid-ocean ridges. In the case of the Hawaiian hot spot, elevated 3He/4He values can be found at least as far back as 76 Ma, and support the hypothesis that the Hawaiian hot spot originates as a deeply rooted mantle plume that taps an undegassed source of primordial helium. Of the many geochemical indicators associated with the Hawaiian mantle plume, helium appears to be one of the most sensitive, as it was the least affected by depletion processes associated with proximity of the plume to a seafloor spreading center during the Late Cretaceous.

Acknowledgments

[16] We are grateful to Asish Basu for use of unpublished helium data and to Marcel Regelous for assistance in selecting the Suiko Seamount samples. We are also grateful to Emily Van Ark and Jian Lin for sharing their plume flux calculations prior to publication. Analytical work on the ODP and DSDP samples was supported by NSF funds managed through the JOI/USSSP. The helium isotope laboratory at the Newport PMEL is supported by the NOAA Vents Program.

Ancillary