Source materials for inception stage Hawaiian magmas: Pb-He isotope variations for early Kilauea



[1] New noble gas and radiogenic isotopic compositions are presented for tholeiitic, transitional, and alkalic rocks from the submarine Hilina region on the south flank of Kilauea, Hawaii. The 3He/4He ratios for undegassed glass and olivine separates (11–26 Ra) contrast with those of postshield and rejuvenated alkalic lavas, consistent with the alkalic and transitional basalts at Hilina corresponding to early Kilauea magmas. Most early Kilauea samples contain highly radiogenic Pb isotopes compared with other Hawaiian rocks and therefore derive from a Hawaiian plume end-member source (here referred to as the Hilina component) distinctive in that respect. Besides radiogenic Pb isotopes, the Hilina component has relatively low 3He/4He (<12 Ra) among the Hawaiian magmas. Hawaiian inception stage magmas, including Hilina, Loihi, and deep Hana Ridge (east Maui), define a linear array in 206Pb/204Pb-3He/4He isotope space, indicating that mixing between the Hilina and Loihi components (or their melts) dominates magmatism at the leading edge of the Hawaiian plume. The Hilina component's isotopic characteristics can be derived from young subduction-recycled crust or metasomatised mantle. The isotopic differences between the geographically discriminated Kea and Loa trend volcanic chains, observed in shield stage lavas, are also seen in the inception stage magmas, suggesting that proportions of melts derived from the Hilina and Loihi components were different between the Kea and Loa trend volcanoes.

1. Introduction

[2] Hawaiian volcanoes are fed by a mantle plume that impinges the base of the drifting Pacific plate. As the volcanoes grow, their magma compositions change, beginning with alkalic to tholeiitic basalts in the preshield stage, through voluminous tholeiitic basalts in the shield stage, and ending with alkali basaltic and other highly alkalic suites in the postshield and rejuvenated stages [e.g., Chen and Frey, 1983; Garcia et al., 1993]. These chemical shifts are associated with temporal isotopic variations, indicating that the mantle sources sampled by Hawaiian volcanoes are diverse and change in a roughly systematic progression [e.g., Chen and Frey, 1985; Kurz and Kammer, 1991; Eiler et al., 1996; Lassiter and Hauri, 1998; Tanaka et al., 2008]. The shield and postshield stages are well characterized due to their extensive subaerial exposures on multiple islands, but the voluminous shield stage products, which can constitute more than 95% of an edifice [Clague and Dalrymple, 1987; Frey et al., 1990], conceal the early stage rocks. The only presently active preshield stage volcano is Loihi Seamount where juvenile volcanism built a small submarine edifice that stands 1000–3500 m above the local seafloor southeast of Hawaii Island [e.g., Moore et al., 1982; Staudigel et al., 1984; Garcia et al., 1993].

[3] Additional products of preshield stage volcanism were discovered by the 1998 through 2002 cruises to the Hawaiian Islands by the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) [Takahashi et al., 2002, and references therein]. Dives off the south flank of Kilauea using the ROV Kaiko and manned submersible Shinkai 6500 recovered a diverse suite of submarine-erupted alkalic rocks including nephelinites, basanites, hawaiites, and alkali basalts; volcaniclastic glass grains extend the compositional spectrum to tephrites, phonolites, mugearites, and benmoreites [Sisson et al., 2002, 2009]. Measured eruption ages, eruption depths, geologic setting, and geochemical and isotopic features indicate that this submarine Hilina region exposes magmatic products from Kilauea's inception stage, overlain by tholeiitic pillow lavas that, with higher stratigraphic position and decreasing age, become more similar in composition to Kilauea's historic shield magmas [Lipman et al., 2002, 2006; Sisson et al., 2002; Calvert and Lanphere, 2006; Coombs et al., 2006; Kimura et al., 2006]. Following the initial discovery of early Kilauea's products [Lipman et al., 2000], explorations of the steep submarine flanks of Kohala, Mauna Loa, and Koolau volcanoes failed to find additional inception stage localities. Limited exposures of deep submarine-erupted basalts discovered on the lower Hana Ridge may be inception stage products of Haleakala Volcano, east Maui [Ren et al., 2005, 2006; Hanyu et al., 2007], and volatile rich hawaiitic glasses in the submarine north Kona area, Hawaii Island, may be volcaniclastic products of the inception stage of Hualalai Volcano [Hammer et al., 2006]. Of these four inception stage localities, only Loihi and the Hilina region offshore of Kilauea provide extensive exposures of compositionally diverse and voluminous inception stage magmatic products. These studies were complementary to attempts to drill through and sample the various geochemical stages of Hawaiian magmatism (Hawaii Scientific Drilling Project (HSDP) [e.g., DePaolo et al., 2001]), with the dives providing access to wider areas, and allowing observations of outcrop-scale structural and depositional features, at the expense of discontinuous sampling.

[4] In this paper, we present the first Hf and comprehensive noble gas isotopic measurements for samples from the Hilina region, offshore of Kilauea, as well as Pb, Sr, Nd, and geochemical analyses, augmenting those in previously published reports [Sisson et al., 2002; Kaneoka et al., 2002; Kimura et al., 2006; Lipman et al., 2006]. Most of the analyzed Hilina samples erupted from early Kilauea, but some deep tholeiitic samples distinguished by relatively high SiO2 and low TiO2 and Na2O are products of the Mauna Loa shield on whose flank Kilauea grew, and some modern Kilauea-like tholeiitic samples collected from the base of the Puna Ridge are probably young shield stage magmas that flowed to great water depths [Lipman et al., 2006; Coombs et al., 2006]. Previous work [Kimura et al., 2006; Calvert and Lanphere, 2006] shows that early Kilauea's products were dominated for over 150,000 years by an isotopically nearly uniform high 206Pb/204Pb source. This is the “Hilina” component of Kimura et al. [2006]. The existence of a high 206Pb/204Pb source or sources was previously inferred from studies of shield stage Hawaiian tholeiites and was broadly referred to as a “common radiogenic end-member” [Abouchami et al., 2000] or the “Kea” component [Stille et al., 1986], the latter in reference to its apparent predominance in Hawaii's northeastern Kea geographic volcanic chain [Tatsumoto, 1978; Dana, 1849]. Prior to Kimura et al. [2006], the isotopic characteristics of the high 206Pb/204Pb source were only known by extrapolation, and estimates differed appreciably from study to study [Abouchami et al., 2000; Eiler et al., 1996]. Loihi Seamount also erupts magmas derived mainly from a source with 206Pb/204Pb higher than many shield stage magmas, but that differs from the Hilina component in having higher 208Pb/204Pb and less extreme 206Pb/204Pb. The Loihi component is further notable for high 3He/4He (20–32 Ra), which had been taken as a fingerprint of a less degassed deep mantle source [Honda et al., 1993; Kurz et al., 1983; Valbracht et al., 1997]. We show that high 3He/4He is not a characteristic of the Hilina component, and therefore that the components feeding the inception stage of Hawaiian magmatism are dissimilar, despite the proximity of Loihi and Kilauea (∼30 km projected across the trend of plate motion). The relatively low 3He/4He of the Hilina component compared to that of the Loihi component also helps to reconcile complex variations between He and Pb isotopes in shield stage tholeiites.

2. Geological Background and Samples

[5] The Hilina region is on the south submarine flank of Kilauea. Prominent geomorphic features are a raised midslope bench with a top at about 2500–3000 m depth, a southeast facing lower scarp that drops steeply to 5000 m depth, and a smooth upper slope that rises to Kilauea's southern shoreline (Figure 1). To the east these features give way to the Puna Ridge, which is the submarine extension of Kilauea's east rift zone. To the west the Hilina region ends with Papa'u Seamount, with a summit at ∼1000 m depth, flanked on its west by a north-northwest trending steep escarpment that represents an active structural boundary. The summit vent of Loihi Seamount lies a further 10 km southwest of this structural boundary. Bathymetry in the Hilina region is smooth due to mantling by copious glassy sands shed by the quenching and shattering of shoreline crossing subaerial Kilauea lava flows. Despite this supply of sand, the midslope bench supports a closed basin due to active uplift of the bench front [Phillips et al., 2008].

Figure 1.

Map of Hawaii Island and the Hilina region, south submarine flank of Kilauea. Colored and black dashed lines outline the midslope bench and delimit the southwest structural boundary. The dive sites are shown by white circles annotated with dive numbers. K95, K98, K207, K208, and K209 dives were conducted by ROV Kaiko, and the other dives were performed by manned submersible Shinkai 6500 (“K” and “S” prefixes are omitted from the dive numbers). The inset illustrates the two parallel geographic lineaments of Hawaiian volcanoes after Dana [1849] and Jackson et al. [1972]. The inception stage volcanisms of Hilina (early Kilauea) and Loihi are on the Kea and Loa trends, respectively.

[6] Geologic units conform to the major geomorphic features and preserve a succession from Kilauea's early alkalic to its modern tholeiitic magma types. The crest and lower scarp of the midslope bench, as well as landslide blocks shed from the lower scarp, expose bedded volcanic glass sandstones with interlayered debris flow breccias (dives K91, K93, K98, S505, S508, S509, S708, S710). These sandstones and breccias are cemented by palagonite and zeolites, and are jointed and faulted, but generally have gentle dips. Seismic and geodetic surveys, as well as geologic and geochemical observations indicate that the bench and lower scarp are an uplifting and southward moving fold and thrust wedge driven by seaward spreading of the Kilauea and Mauna Loa edifices [Lipman et al., 2002, 2006; Hills et al., 2002; Phillips et al., 2008]. Clasts within debris flow breccias are predominantly alkalic [Sisson et al., 2002; Kimura et al., 2006], and where glass rims are preserved, are known from their dissolved volatiles to have erupted under submarine conditions [Coombs et al., 2006]. Glass grains in sandstones are mostly degassed tholeiites similar in composition to modern Mauna Loa magmas (higher SiO2, and lower TiO2 and Na2O than Kilauea or Mauna Kea products), but alkalic through transitional glass grains are widespread. The composition and degassed character of the tholeiitic glasses are consistent with their derivation by shattering of shoreline-crossing lavas from the Mauna Loa shield, and possibly also some from Mauna Kea, whereas elevated volatile concentrations in the alkalic through transitional glasses indicate early Kilauea vent locations that were spread across the flank of the Mauna Loa shield from perhaps subaerial to as much as 5000 m water depth [Coombs et al., 2006].

[7] Rock exposures are rare on the smooth slope above (north of) the midslope bench (Figure 1), but a faint ridge protruding through the mantling Kilauea sands consist of in-place alkalic and transitional pillow lavas with interstratified volcanic glass sandstones rich in alkalic grains (dives K208, S709) [Kimura et al., 2006]. The pillow lavas are generally truncated and exposed in cross section due to having shed landslides into adjacent basins, but their presence verifies that alkalic magmas were erupted in the Hilina region, consistent with an early Kilauea source. Dives at the east end of the midslope bench (Figure 1) also discovered in-place pillow lavas and local pillow talus (dives K95, S504, S506), but alkalic compositions are absent. Instead, these are tholeiitic basalts that are distinguished from nearly all modern Kilauea tholeiites by lower concentrations of SiO2 and higher concentrations of TiO2 and Na2O in their whole rocks and glasses [Sisson et al., 2002]. These low-Si tholeiites mark the beginnings of the eastward propagation of Kilauea's east rift zone and submarine Puna Ridge as the volcano first entered its tholeiitic shield stage and magmatic productivity increased sharply [Lipman et al., 2006]. Dive S506 at the foot of the Puna Ridge and one sample in dive S504 recovered a few tholeiites with compositions typical of the modern Kilauea shield, but these samples are degassed and are interpreted to have erupted at much shallower water depths [Sisson et al., 2002]. Overall, the geologic and geochemical relations support a progression of erupted compositions from early alkalic and transitional (midslope bench, lower scarp, and upper slope exposures), then younging upward and eastward through distinctly low-Si tholeiites (dives K95 and S504), to modern shield stage tholeiites (Puna Ridge and dive S506) as eruptive output increased and constructed the East Rift Zone, Puna Ridge, and subaerial shield [Coombs et al., 2006].

[8] To the west the midslope bench ends at Papa'u Seamount and the transverse structural boundary (Figure 1). Papa'u Seamount's steep west face exposes massive to indistinctly bedded breccias (dive K209). Breccia clasts are vesicular, oxidized (reddish), and commonly rounded indicating subaerial eruption and probable littoral reworking. Their major and trace element compositions are typical of Mauna Loa shield tholeiites [Lipman et al., 2006], and Papa'u Seamount represents either a landslide block or an uplifted welt of coarse sediments shed from the subaerial Mauna Loa shield. No alkalic, transitional, or otherwise Kilauea-like clasts were recovered from Papa'u Seamount breccias. Similar Mauna Loa compositions are abundant in breccia exposures at the lower southwest corner of the Hilina frontal scarp (low in dive S710 (Figure 1)). These exposures, including Papa'u Seamount, are taken as Mauna Loa's pre-Kilauea sedimentary apron, subsequently largely buried by Kilauea's products and uplifted and deformed by volcano spreading [Lipman et al., 2006]. Dive K207 at the western end of the Hilina region differed from all others in that it encountered steep exposures of essentially noncemented pillow breccia mainly of transitional basalt [Kimura et al., 2006]. The absence of cementation indicates that these are young landslide deposits shed onto the midslope bench, but the deposits are close to Loihi Seamount, and for isotopic reasons spelled out subsequently, we cannot rule out the possibility that the breccia blocks were shed from Loihi, and subsequently uplifted during creation of the bench and the transverse structural boundary.

[9] The 40Ar/39Ar ages of Hilina lavas, ranging from 130 to 280 ka, are older than the age of subaerial Kilauea, confirming that Hilina is the early expression of Kilauean volcanism [Calvert and Lanphere, 2006]. Furthermore, strongly alkalic rocks (S508, S509, S710; 210–280 ka) are older than weakly alkalic rocks collected at shallower and more easterly locations (S504, S505, K208; 120–230 ka), documenting progressive compositional changes during growth of the volcano and increase in its magmatic productivity.

[10] Major and trace element concentrations, Pb-Sr-Nd-Hf isotope ratios and noble gas isotopic compositions were measured for all the studied samples listed in Table 1, and 40Ar/39Ar ages were determined for selected samples. The analytical details are shown in Appendix A.

Table 1. Sample Description
Sample NameLocationRock TypeCompositional TypeInformation for Eruption DepthaRemarksb
LatitudeLongitudeDepth (m)
  • a

    Saturation pressure of volatiles (Psat; in megapascal) and sulfur concentrations ([S]; in ppm) in glass are from Coombs et al. [2006].

  • b

    Rock descriptions after the cruise report.

R/V Kairei With Kaiko in 1998
K98-319°00.95′N155°00.65′W3771picritic basaltlow-Si tholeiite dark grey picrite with few small vesicles; boulder
R/V Yokosuka With Shinkai 6500 in 1999
S504-2A19°15.27′N154°49.37′W3772olivine basaltlow-Si tholeiitein situ judging from field observationsin situ pillow fragment from an outcrop
S504-2B19°15.27′N154°49.37′W3772olivine basaltlow-Si tholeiitePsat = 75, [S] = 1330, presumably in situin situ pillow fragment from an outcrop
S504-3B19°15.26′N154°49.55′W3596olivine basaltlow-Si tholeiitein situ judging from field observationsin situ pillow fragment from an outcrop that is different from S504-2B
S505-9A19°03.94′N154°55.85′W4346picritic basalthigh-Si tholeiite volcaniclastic breccia from highly fractured outcrop
S505-9B19°03.94′N154°55.85′W4346olivine basalthigh-Si tholeiite volcaniclastic breccia from highly fractured outcrop
S505-10B19°04.10′N154°55.99′W4234olivine basaltalkali basaltPsat = 38, [S] = 1800talus block
S506-119°20.90′N154°33.15′W5448olivine basaltlow-Si tholeiite pillow fragment from talus deposit at the base of the lower scarp
S506-5A19°21.29′N154°33.37′W5148olivine basaltlow-Si tholeiitePsat = 56, [S] = 430pillow fragment from young-appearing talus
S506-6A19°21.32′N154°33.39′W5096picritic basaltlow-Si tholeiite pillow fragment from slope failure outcrop
S506-7B19°21.37′N154°33.40′W5016picritic basaltlow-Si tholeiite pillow fragment from pillow breccia
S506-8A19°21.70′N154°33.54′W4847picritic basaltlow-Si tholeiitePsat = 41, [S] = 1130pillow fragment from outcrop
S506-8C19°21.70′N154°33.54′W4847picritic basaltlow-Si tholeiitePsat = 57, [S] = 1160pillow fragment from outcrop
S509-4B19°12.09′N154°51.68′W4013olivine basaltbasanitePsat = 49, [S] = 1350rounded boulder from breccia outcrop
S509-6C19°12.09′N154°51.99′W3522olivine basaltlow-Si tholeiite basaltic breccia from local small outcrop
R/V Kairei With Kaiko in 2001
K207-1619°01.69′N155°11.88′W2658olivine basalttransitional basaltPsat = 27, [S] = 1030pillow basalt fragments with glass, possibly in situ
K207-1719°01.69′N155°11.88′W2658olivine basalttransitional basaltPsat = 22, [S] = 1030pillow basalt fragments with glass, possibly in situ
K207-1919°01.88′N155°11.99′W2568olivine basalttransitional basaltPsat = 18, [S] = 1040clast of pillow basalt with glass
K208-319°08.00′N155°06.35′W2515aphyric basalttransitional basaltPsat = 35, [S] = 1800pillow lava from an outcrop
K208-719°08.22′N155°06.65′W2372aphyric basaltalkali basalt[S] = 1850, in situ judging from field observationspillow clast, but probably just moved down for 10 m
K208-919°08.56′N155°07.01′W2243pl-ol basalttransitional basaltPsat = 24, [S] = 1480pillow lava from an outcrop
K209-1B19°05.60′N155°16.60′W1715pl-ol basalthigh-Si tholeiitepresumably landslide materialstalus block
K209-8A19°06.62′N155°16.61′W1230picritic basalthigh-Si tholeiitepresumably subaerial rockangular clast from a breccia block
R/V Yokosuka With Shinkai 6500 in 2002
S709-419°09.29′N155°07.77′W2012olivine basalttransitional basaltin situ judging from field observationspillow lava from an outcrop
S709-519°09.47′N155°07.78′W1965aphyric basaltalkali basaltin situ judging from field observationspillow lava from an outcrop
S709-919°09.72′N155°08.13′W1821olivine basaltalkali basaltin situ judging from field observationspillow lava from an outcrop

3. Results

3.1. Major and Trace Elements

[11] Rock samples from Hilina are geochemically diverse, ranging from tholeiitic through transitional, to alkalic types (Table 2 and Figure 2) [also Sisson et al., 2002; Kimura et al., 2006]. Tholeiitic basalts can be divided into high-Si (Mauna Loa type), and a spectrum of lower-Si tholeiites that range with decreasing SiO2 and increasing TiO2 and Na2O from typical Kilauean shield stage compositions (restricted to minor degassed samples approaching and along the Puna Ridge) to distinctly low-Si tholeiites (transitional basalts of Sisson et al. [2002] and Lipman et al. [2002]) that are only rarely exposed on the subaerial shield. All samples from Papa'u Seamount (dive K209) are high-Si tholeiites, suggesting that they are derived from subaerial Mauna Loa [Lipman et al., 2006]. Rocks with high-Si tholeiite compositions were also recovered from the lower scarp in dives S505 and S710 and are interpreted as Mauna Loa products transported to deep water depositional sites, buried by subsequent products of Mauna Loa and early Kilauea, and then uplifted during volcano spreading [Sisson et al., 2002; Lipman et al., 2006]. Tholeiites with lower SiO2 than the modern shield predominate along the upper eastern slope (dives S504 and S506) and are interpreted as products of the onset of tholeiitic shield magmatism [Sisson et al., 2002]. Dives K207 and K208 samples are mostly transitional with some low-Si tholeiitic rocks. Mildly to strongly alkalic rocks are ubiquitous at the other sites rather than tholeiites.

Figure 2.

SiO2 versus total alkalis diagram. Hilina samples demonstrate large compositional variations from tholeiite, transitional basalt, and alkali basalts to strongly alkalic rocks. Tholeiitic rocks can be further divided into low-Si and high-Si tholeiites comparable to subaerial Kilauea and Mauna Loa lavas, respectively. The tholeiitic/alkalic boundary is from Macdonald and Katsura [1964]. Transitional basalts plot near the tholeiitic/alkalic boundary. The samples from this study are represented by large symbols. The other data, represented by small symbols, are from Sisson et al. [2002] and Kimura et al. [2006].

Table 2. Major and Trace Element Concentrations and Isotope Ratiosa
Major elements (wt %)                            
Trace elements (ppm)                            
Isotope ratios                            
   2 SE0.0000170.0000150.0000150.0000150.0000150.0000150.0000150.0000150.0000150.0000130.0000150.0000140.0000150.0000150.0000150.0000140.0000140.0000150.0000140.0000140.0000150.0000150.0000150.0000070.0000070.000007 0.000052
   143Nd/144Nd0.5128900.5128820.5129400.5129390.5130160.5128760.5129710.5129300.5128980.5129870.5130600.5129700.5129800.5129290.5129030.5129340.5129320.5129740.5129470.5129580.5129680.5129700.5129220.5129290.5129450.512961La Jolla0.511852
   2 SE0.0000120.0000120.0000110.0000120.0000090.0000110.0000110.0000110.0000120.0000110.0000110.0000110.0000110.0000120.0000110.0000100.0000100.0000000.0000100.0000100.0000100.0000100.0000100.0000110.0000090.000011 0.000036
   2 SE 0.00080.00100.00080.00080.00090.00090.00090.00070.00090.00090.00070.00090.00090.00080.00090.00070.00090.00070.00070.00070.00070.00070.00070.00070.0007 0.0043
   2 SE 0.00080.00070.00080.00070.00070.00080.00100.00080.00080.00080.00070.00080.00090.00080.00100.00070.00100.00080.00070.00090.00070.00080.00070.00070.0007 0.0044
   2 SE 0.00240.00210.00280.00250.00220.00250.00330.00240.00240.00260.00210.00280.00260.00300.00280.00210.00280.00280.00240.00250.00200.00240.00180.00200.0021 0.0117
   2 SE0.0000150.0000120.0000080.0000150.0000140.0000130.0000110.0000100.0000110.0000090.0000090.0000110.0000090.0000100.0000110.0000110.0000110.0000090.0000090.0000100.0000090.0000110.0000100.0000100.0000100.000010 0.000011

[12] The different rock types plot on distinct fields in major element diagrams, such as MgO versus SiO2, TiO2, Na2O, K2O and P2O5 (not shown), as reported previously [Sisson et al., 2002; Kimura et al., 2006]. However, these fields are unrelated to isotopic compositions. For example, transitional basalts from K207 and K208 define a tight trend in major element diagrams despite their isotopic diversity (see below). This suggests that major element variations are controlled by melting and fractionation processes in addition to the difference in source composition.

[13] Concentrations of incompatible trace elements decrease from alkalic, through transitional to tholeiitic rocks at a given MgO (Table 2). The plots of Zr versus other trace elements display linear covariations between elements [see also Sisson et al., 2002; Kimura et al., 2006], again irrespective of isotopic compositions. Primitive upper mantle-normalized spidergrams illustrate coherent trace element abundance patterns with some exceptions (Figure 3). Depletions in Pb and enrichments in U in samples S504-2A, S504-3B and K208-7 may result from low-temperature seawater alteration [e.g., Yokoyama et al., 2003], although clear evidence for alteration was not observed in major elements, such as K2O/P2O5. A negative anomaly at Th is common to all the samples, which is widely observed in Hawaiian samples [Hofmann and Jochum, 1996; Kimura et al., 2006]. Steep slopes from LIL through LREE to HREE in the trace element abundance patterns indicate residual garnet during partial melting [Sisson et al., 2002; Kimura et al., 2006]. Enhanced incompatible element concentrations commensurate with alkalinity are consistent with decreasing melting degrees to produce low-Si tholeiites, transitional and alkalic rocks in that order. The deduced source composition is similar to primitive mantle with minor modifications, such as addition of high field strength elements and light REE with reduction of Rb, Ba, Th and K [Kimura et al., 2006].

Figure 3.

Trace element concentrations of the Hilina samples normalized to primitive upper mantle (PUM) of McDonough and Sun [1995]. (a) Samples with radiogenic Pb isotopic compositions (206Pb/204Pb > 18.62), which are typical in Hilina. (b) Samples with Loihi-like Pb-Sr-Nd-Hf-He isotopic compositions from the K207 site. (c) The other samples associated with relatively unradiogenic Pb isotopic compositions. Deduced source composition of the Hilina basalts is shown by blue lines with dots [Kimura et al., 2006].

3.2. Pb-Sr-Nd-Hf Isotopes

[14] Radiogenic isotope ratios of Hilina samples obtained in this study expand the isotopic ranges beyond those previously reported [Kimura et al., 2006], but the preponderance of high 206Pb/204Pb samples remains. Among the radiogenic isotopes, Pb is the best to classify the rocks in relation to Hawaiian source heterogeneity. The majority of Hilina samples (38 of 52 samples analyzed in this study and by Kimura et al. [2006]) have elevated 206Pb/204Pb greater than 18.62 (Table 2 and Figures 4a and 4b). Such samples include various rock types ranging from low-Si tholeiite, transitional, to alkalic. Slightly anomalous among these are alkali basalts S505-2B, S505-10A and S505-10B distinguished by lower 206Pb/204Pb (between 18.62 and 18.63) with similar 208Pb/204Pb (Figure 4b), which is clearly indicated by 208Pb*/206Pb* (Figure 4f) (XPb* indicates amount of radiogenic XPb produced after formation of the Earth [Galer and O'Nions, 1985]). Although tholeiitic basalts with isotopic compositions similar to this subgroup have been reported in the subaerial Kilauea lavas, alkalic basalts never erupt on subaerial shield stage Kilauea. The 123 ± 55 kyr 40Ar/39Ar age of S505-10B also precludes the possibility that these alkali basalt samples are from modern Kilauea volcano [Calvert and Lanphere, 2006].

Figure 4.

Plots of 206Pb/204Pb versus (a) 207Pb/204Pb, (b) 208Pb/204Pb, (c) 87Sr/86Sr, (d) ɛNd, (e) ɛHf, and (f) 208Pb*/206Pb*. The samples from this study are represented by large symbols. Data for the other Hilina samples, represented by small symbols, are from Kimura et al. [2006]. Closed symbols denote the samples from early Kilauea. Samples interpreted not to be early Kilauea products are shown by open symbols. Small orange dots are HSDP samples of Mauna Kea [Lassiter et al., 1996; Blichert-Toft and Albarède, 1999, 2009; Blichert-Toft et al., 1999, 2003; DePaolo et al., 2001; Abouchami et al., 2000, 2005; Eisele et al., 2003; Bryce et al., 2005]. The isotopic ranges of other volcanoes are from Chen and Frey [1983, 1985], Hofmann et al. [1984], Staudigel et al. [1984], Newsom et al. [1986], Stille et al. [1986], West and Leeman [1987], Chen et al. [1990, 1991, 1996], Kennedy et al. [1991], Kurz and Kammer [1991], Garcia et al. [1993, 1995, 1996, 1998, 2000], Frey et al. [1994], Roden et al. [1994], Yang et al. [1994], Kurz et al. [1995], Rhodes and Hart [1995], Bennett et al. [1996], Cohen et al. [1996], Valbracht et al. [1996], Lassiter and Hauri [1998], Blichert-Toft et al. [1999], Norman and Garcia [1999], Pietruszka and Garcia [1999], Stracke et al. [1999], Lassiter et al. [2000], Tanaka et al. [2008], and Ren et al. [2009].

[15] Also distinctive are low-Si tholeiites S506-6A, S506-7B, S509-6C, and degassed tholeiite S506-5A that plot within the Kilauean tholeiitic shield Pb isotopic field (Figures 4a and 4b). The S506 samples originate from low on the Puna Ridge, well to the east of the other Hilina samples, so their shield-like isotopic values are not unexpected. Low-Si tholeiite S509-6C was collected from near the eastern termination of the midslope bench, so it may also be a shield onset product. Additional noteworthy findings are that the transitional basalts K207-16, K207-17 and K207-19 from the open framework pillow breccias overlying the southwest end of the Hilina bench have Pb isotopic compositions similar to Loihi Seamount magmas. Finally, the high-Si tholeiites S505-9A, S505-9B, and K209-8A have Pb isotopic compositions similar to Mauna Loa, as expected from their major and trace element compositions and geologic situations.

[16] Other radiogenic isotopes, such as Sr, Nd and Hf, covary with Pb isotopes (Figures 4c4e). The Hilina samples with radiogenic Pb isotopes (206Pb/204Pb > 18.62) have 87Sr/86Sr, 143Nd/144Nd and 176Hf/177Hf values similar to subaerial Kilauea lavas, although the former have a slightly larger scatter. The Hilina rocks altogether exhibit isotopic trends in Sr-Pb, Nd-Pb and Hf-Pb spaces, overlapping with those for the shield stage tholeiites, and therefore the Hilina samples share the same mantle sources irrespective of their large diversity in alkalinity. The results of this study further confirm that the samples with the most radiogenic Pb isotopes closely represent one of the end-components of Hawaiian magma sources referred to as the Hilina component [Kimura et al., 2006] (see below). Notably, the Hilina samples plot away from the isotopic ranges of the postshield and rejuvenated stage lavas (Figure 4).

3.3. Noble Gases

[17] The samples show various 3He/4He values (Table 3 and Figure 5a). Two glass samples have very low 3He/4He between 2 and 4 Ra. These samples presumably experienced severe degassing and subsequent contamination from an atmospheric component, because they are highly depleted in 4He. This is supported by atmospheric Ne and Ar isotope ratios in association with relatively high Ne and Ar concentrations of these samples. Stepwise crushing extraction of glass samples provides almost constant 3He/4He despite a decrease in 40Ar/36Ar as crushing proceeds (e.g., K98-3, K208-9), because He is not as affected by atmospheric contamination as Ar.

Figure 5.

(a) Plot of 3He/4He versus 4He concentrations. The ranges of 3He/4He for Loihi, typical shield stage, and postshield and rejuvenated stage lavas are shown on the right side of the diagram. The samples with low 3He/4He (<5 Ra) are presumably degassed glass because of low 4He concentrations. (b) Neon three-isotope diagram. The Hilina samples plot on the Hawaiian trend defined by Honda et al. [1991]. The symbols for the Hilina samples are the same as those in Figure 4.

Table 3. Noble Gas Isotope Ratios of Olivine and Glass Samples
SamplePhaseWeight (g)N of CrushingDepth (m)Abundances (cm3 STP/g)Isotope Ratios
4He (× 10−9)20Ne (× 10−12)36Ar (× 10−12)84Kr (×10−12)132Xe (× 10−12)3He/4He20Ne/22Ne21Ne/22Ne38Ar/36Ar40Ar/36Ar
Ratio1 SERatio1 SERatio1 SERatio1 SERatio1 SE
   +50 16.812.110.20.4650.02013.00.310.980.080.03240.00130.18870.0017222542
   +100 3.336.105.290.2540.01112.60.610.420.150.03130.00250.18710.0020789.714.9
   +50 25916366717.30.28517.00.110.450.030.03140.00060.18750.0009147628
   +100 87.959.767.81.740.08717.20.210.380.060.03100.00070.18700.0008156429
   +100 27.725.441.10.9610.06217.60.29.980.040.03150.00110.18740.0008139726
   +100 2.1540.
   +50 3.635622494.260.08024.10.79.870.020.02900.00030.18790.0003296.45.5
   +50 2.2870.67.500.1680.01212.40.510.110.060.02970.00080.18620.0014636.712.0

[18] The transitional basalts from the K207 site have 3He/4He as high as Loihi lavas [e.g., Honda et al., 1993; Valbracht et al., 1997]. The rocks from the other sites have 3He/4He between 10.0 and 18.7 Ra, which overlap with 3He/4He of shield stage lavas from subaerial Kilauea [e.g., Kurz and Kammer, 1991; Kurz, 1993; Kurz et al., 1996]. Samples collected from a single dive site have almost constant 3He/4He. Exceptions are the rocks from dive S506 at the foot of the Puna Ridge, where 3He/4He range from 10.0 to 17.6.

[19] Several samples have elevated 20Ne/22Ne and 21Ne/22Ne ratios compared with atmospheric ratios (Table 3 and Figure 5b). The results plot on the general Hawaiian trend (Loihi-Kilauea trend defined by Honda et al. [1993]), irrespective of their 3He/4He. Ne and Ar isotope ratios are weakly correlated. Atmospheric Ne is always coupled with atmospheric Ar. However, radiogenic Ar isotope ratios are not always associated with elevated 20Ne/22Ne, thus a simple mixing between atmospheric and mantle-derived components does not account for 20Ne/22Ne and 40Ar/36Ar in the samples. Both 4He/21Ne* and 4He/40Ar* ratios (21Ne* and 40Ar* are the nonatmospheric portion of 21Ne and 40Ar, respectively) are lower than the mantle production ratios (2 × 107 for 4He/21Ne* and 2–5 for 4He/40Ar*) [Yatsevich and Honda, 1997], and these two parameters do not correlate well with each other (figures not shown, see data in Table 3). This indicates that He, Ne and Ar in the mantle-derived melt were variously fractionated and degassed [Harrison et al., 2003; Trieloff et al., 2003]. It is difficult to isolate contributions of each process and this subject is not discussed further in this paper.

3.4. Ar-Ar Dating

[20] In addition to previously reported 40Ar/39Ar ages for Hilina rocks [Calvert and Lanphere, 2006], we determined ages for two transitional basalts from the K207 site. K207-1 and K207-4 yielded plateau ages of 67 ± 29 ka and 65 ± 28 ka (2σ), respectively. Isotope correlation (isochron) plots indicate atmospheric argon intercepts with ages indistinguishable from plateau ages (Table 4 and Appendix B). Sample K207-4 yielded an excellent plateau age comprising >99% of the 39Ar released. The slightly declining age spectrum with progressive Ar release is interpreted to be due to 39Ar recoil during irradiation. Sample K207-1 yielded a plateau age comprising 63% of the 39Ar released with discordant steps early and late in the heating schedule. K contents of the sample were considerably lower than K207-4, likely due to more glass being removed during sample preparation.

Table 4. Summary of 40Ar/39Ar Agesa
SampleLocationTypeComposition, Lava TypePlateauPreferred Age (ka)IsochronTotal Gas Age (ka)
LatitudeLongitudeDepth (m)%39Ar (Steps)Age (ka)MSWDAge (ka)MSWD40/36i
  • a

    Analytical uncertainties are 2σ.

K207-119°01.05′N155°11.55′W2935groundmassweakly alkalic basalt, pillow fragment65 (650–800)67 ± 290.0967 ± 2994 ± 1310.07294.0 ± 11.312 ± 43
K207-419°01.13′N155°11.57′W2883groundmassweakly alkalic basalt, pillow fragment99 (550–1050)65 ± 280.9665 ± 2871 ± 661.36295.1 ± 3.291 ± 25

[21] K207 basalt ages are interpreted to be older than modern Kilauea tholeiites, but younger than transitional and alkalic basalts from other Hilina dive sites (Figure 6) [Calvert and Lanphere, 2006]. Notably, K207 basalt ages overlap with the age range of Loihi (100 ka to present [Guillou et al., 1997]).

Figure 6.

Plots of interpreted 40Ar/39Ar ages versus alkalinity of samples. Alkalinity indicates difference in Na2O + K2O (in weight percent) at a given SiO2 between the sample and the alkalic-tholeiitic discrimination boundary line defined by Macdonald and Katsura [1964], calculated by the following formula: alkalinity = (Na2O + K2O) − 0.37 × (SiO2 − 39). Two K207 samples are represented by orange symbols. Data for the other Hilina samples (blue symbols) and Loihi rocks (green symbols) are from Calvert and Lanphere [2006] and Guillou et al. [1997].

4. Discussion

4.1. Lineage of the Hilina Magmas

[22] Prior to discussing the geochemical features of early Kilauea, we examine where the studied rocks were derived from, mainly using field observations and glass volatile contents [after Coombs et al., 2006], as summarized in Table 1. Coombs et al. [2006] verified that Hilina magmas were vapor saturated and underwent equilibrium degassing prior to eruption. Among the alkalic and transitional rocks, K208-7 and all the S709 samples are considered to be in situ lavas from field observations. Other samples have high sulfur and carbon dioxide contents in the glass, the latter being close to the level of saturation at the recovery depths [Coombs et al., 2006]. Consequently, all the alkalic and transitional rocks used in this study were recovered from near eruption depths, and thus they are products of early Kilauea. The low-Si tholeiites may have various origins. Low-Si tholeiites from the S504 site are in situ pillow lavas or pillow fragments that also erupted near the sampling locations (Table 1). Among low-Si tholeiites from the S506 site near the base of the Puna Ridge, S506-5A, S506-8A and S506-8C have reasonably high sulfur and carbon dioxide contents in glasses, which allows us to infer that they are in situ rocks. However, low sulfur contents in the other tholeiites from the S506 site indicate these were shield stage Kilauean magmas that erupted at shallower depths, or that underwent an episode of shallow rift zone degassing, before quenching in deep water [Lipman et al., 2002]. The origin of S509-6C remains uncertain due to the lack of evidence from field observations or volatile contents.

[23] The high-Si tholeiites have straightforward interpretations, as follows. Dive K209 clasts from Papa'u Seamount are vesiculated, depleted in sulfur [Coombs et al., 2006], and commonly rounded, indicating that they erupted above sea level. Indeed, this seamount consists of sediments shed from Mauna Loa that then either slumped or were upwarped to their present location [Lipman et al., 2006]. High-Si tholeiite clasts from the foot of the lower scarp (S505-9A, S505-9B) also have major element compositions like Mauna Loa with low sulfur content (S505-9B), and are interpreted to be coarse clasts shed from that volcano [Lipman et al., 2002; Sisson et al., 2002]. Consequently, high-Si tholeiites collected in the Hilina region are interpreted to be products of Mauna Loa volcanism and thus do not contribute to understanding Kilauea's magmatic evolution.

[24] Following these arguments, samples unlikely to be early Kilauea products are shown as open symbols in Figures 4, 5, and 7. However, some questions remain regarding the source of the transitional basaltic K207 samples collected near the transverse structural boundary. These have Pb-Sr-Nd-Hf isotopic compositions similar to Loihi rocks, and their ages overlap with the age range of Loihi [Guillou et al., 1997]. Although they might be landslide debris shed from Loihi and subsequently uplifted during growth of the Hilina bench, this seems unlikely because their volatile saturation pressures are close to their recovery depth [Coombs et al., 2006], which would be improbable if they were distal sediments. Instead, we interpret that these were either true Loihi magmas that invaded and erupted through Kilauea's preshield plumbing system, or that Loihi-type source domains in part underlay and fed early Kilauea.

Figure 7.

Mixing relationships between the components are illustrated in (a) 206Pb/204Pb-3He/4He space and (b) 206Pb/204Pb-208Pb/204Pb space. The symbols for the Hilina samples are the same as those in Figure 4. The black dotted lines indicate the mixing trend of the inception stage magmas, consisting of Hilina, Loihi, and Hana Ridge (submarine ridge of Haleakala) samples as well as the Kea-hi8 samples of HSDP. Note that the samples with less radiogenic Pb and lower 3He/4He among Hana Ridge are vesiculated floating rocks collected from the upper portion of the submarine ridge, which are presumably products of subaerial eruption [Hanyu et al., 2007]. The two components of the inception stage magmas, the Loihi and Hilina components, should plot on the extension of this mixing trend. The shield stage rocks have lower 206Pb/204Pb and 3He/4He than the inception stage rocks, indicating involvement of some unradiogenic Pb components (Koolau, DM, and/or Loa components) in the shield stage. The isotopic trends of Mauna Loa (blue arrow) and Mauna Kea (two red arrows that correspond to Kea-mid8 and Kea-lo8 trends of Eisele et al. [2003]) indicate distinct source composition on the radiogenic Pb side, suggesting different proportions of the Hilina component against the Loihi component in the source between Loa and Kea trend volcanoes. Compiled data are from Kurz et al. [1983, 1987, 1996, 2004], Valbracht et al. [1997], and references in the caption of Figure 4.

[25] Extensive geological, geochemical, and geochronological evidence lead to the interpretation that the transitional and alkalic samples from the Hilina region derived from early Kilauea, which started to grow across a broad depth range on the shoulder of the extant Mauna Loa shield [Lipman et al., 2000, 2002, 2006; Sisson et al., 2002; Coombs et al., 2006; Calvert and Lanphere, 2006; Kimura et al., 2006]. The subsequent onset of Kilauean shield volcanism was marked by the loss of alkalic and transitional compositions, and by a great increase in magmatic productivity that built the tholeiitic East Rift Zone, Puna Ridge, and the subaerial shield [Lipman et al., 2006]. The moderately elevated 3He/4He and generally radiogenic Pb isotopic compositions of the Hilina alkalic and transitional rock differ from postshield and rejuvenated stage magmas, consistent with the preshield origin. Consequently, Hilina magmas, in combination with Loihi, provide information on the early geochemical evolution of Hawaiian volcanoes. Moreover, Hilina and Loihi represent the inception stage leading tips of the Kea and Loa trend volcanic chains, and therefore provide information on the structure of the Hawaiian mantle plume [Kimura et al., 2006].

4.2. Source Components in Inception Stage Magmatism

[26] Isotopic variations in Hilina region samples require at least three distinct source components (exclusive of the rare Mauna Loa rocks). Most of rocks have highly radiogenic Pb isotopic compositions, in association with radiogenic Nd and Hf and nonradiogenic Sr, among Hawaiian magmas (Figures 4 and 7). These samples closely represent one of the dominant end-member sources in Hawaiian magmatism. As noted, the existence of high 206Pb/204Pb source materials had been interpreted previously, with various estimates of isotopic composition and assigned names, under the general assumption that this constitutes a single component type rather than a blend of isotopically more diverse materials [Stille et al., 1986; Eiler et al., 1996; Abouchami et al., 2000]. The situation has changed in that it is now known that early Kilauea erupted isotopically nearly uniform high 206Pb/204Pb materials for more than 150,000 years, and therefore that a true end-member type probably underwent melting [Kimura et al., 2006]; however, trace amounts of melt with even more extreme radiogenic (HIMU-influenced) Pb isotopic characteristics also contributed to early Kilauea magmatism [Shimizu et al., 2001]. Because of this real complexity, and the conflicting estimates and names in the literature, we follow Kimura et al. [2006] in referring to the predominant and persistent high 206Pb/204Pb source for early Kilauea as the “Hilina component,” after the Hilina region where its melting products can be sampled in abundance. Kimura et al. [2006] showed that the Hilina component appears to have contributed preferentially to the volcanoes of the northeasterly Kea volcanic chain, and thus the Hilina component meets the original intent of Stille et al.'s [1986] more loosely defined Kea component. Similarly, the Loihi component is regarded as the leading edge source with the most radiogenic Pb isotopes that contributes predominantly to Hawaii's southwestern Loa volcanic chain. In this study, rocks collected from the Hilina region vary somewhat more widely in isotopic compositions than those analyzed by Kimura et al. [2006], but the conclusion that the majority of Hilina samples have the most radiogenic Pb isotopes among Hawaiian lavas is unchanged.

[27] Inception stage magmas provided by the Loihi source component are enriched in 208Pb and particularly in 3He/4He, as represented by Loihi itself, and in the Hilina region by the K207 transitional basalts. The S508 nephelinites and the S505 alkali basalts are isotopically intermediate between the Hilina and Loihi components. Finally, degassed sample S506-5A, previously interpreted as a Kilauea shield tholeiite [Sisson et al., 2002], documents the appearance of a third, depleted component once Kilauea entered its voluminous growth stage.

4.2.1. Two Geochemical Components in the Inception Stage: Pb-He Isotope Systematics

[28] By combining He with radiogenic isotopes, new and important differences between the Loihi and Hilina components are revealed. In 206Pb/204Pb-3He/4He space (Figure 7a), the Hilina component defines one corner of a roughly triangularly shaped data distribution and is characterized by relatively low 3He/4He (<12 Ra) and high 206Pb/204Pb among the Hawaiian magmas. Loihi samples plot with progressively higher 3He/4He and lower 206Pb/204Pb defining a linear array between the Loihi and Hilina components.

[29] We also add data from the Hana Ridge, which is the submarine extension of the east rift zone of Haleakala volcano, to the data set of the inception stage volcanism. Most of the tholeiitic samples collected at the tip of this elongated ridge predate the subaerial Honomanu shield stage tholeiites by 0.3–0.9 Myr, and therefore they reflect an early stage of that volcano, despite their tholeiitic compositions [Hanyu et al., 2007]. The Hana Ridge tholeiites have higher 3He/4He and 206Pb/204Pb ratios than the subaerial tholeiites of Haleakala, and they plot in the middle of the Loihi-Hilina array (Figure 7a). Some Hilina samples also plot along the array toward the Loihi component, attaining elevated 3He/4He (18 Ra), which are similar to those in the Hana Ridge samples (Figure 7a). This may indicate that the inception stages of Haleakala and Kilauea both received contributions from the Loihi component despite being Kea geographic trend volcanoes.

[30] In 206Pb/204Pb-208Pb/204Pb space (Figure 7b), the inception stage magmas again form a linear trend from the Loihi component to the Hilina component. This trend has a shallower slope than the general Hawaiian trend and bounds the high 208Pb/204Pb side of the data distribution of all the Hawaiian samples [Eisele et al., 2003; Abouchami et al., 2005]. For the HSDP core samples of Mauna Kea volcano, Eisele et al. [2003] defined a similar trend, Kea-hi8 array, that consists of a group of samples showing systematically elevated 208Pb/204Pb at a given 206Pb/204Pb compared to Kea-mid8 and Kea-lo8 arrays (Figure 7b). The rocks belonging to Kea-hi8 array possess elevated 3He/4He coupled with high 208Pb*/206Pb* (or high Δ8/6) [Kurz et al., 2004] and the Kea-hi8 suite ends at the mixing array between the Hilina and Loihi components. The more abundant Kea-mid8 and Kea-lo8 suites intersect the mixing array close to the Hilina component (Figure 7), showing that the component was also a predominant and sustained contributor to Mauna Kea shield magmatism.

4.2.2. Origin of the Hilina Component

[31] Previous studies have proposed several hypotheses for the origin of the Hilina component (or Kea component) with its radiogenic Pb isotopic compositions. Acceptable hypotheses for the Hilina component include recycled oceanic crust or lithosphere [Stille et al., 1986; Thirlwall, 1997; Lassiter and Hauri, 1998; Eisele et al., 2003; Hanyu et al., 2007], FOZO (i.e., Focus Zone [Hart et al., 1992]) [Ren et al., 2006] and peridotite depleted or metasomatised by intramantle processes such as sulfide melt extraction and silicate melt addition [Gaffney et al., 2005; Kimura et al., 2006]. The discovery that He isotopes covary with Pb isotopes in inception stage magmas helps to discriminate between the source models. Linear trends defined by Hawaiian inception stage magmas in Pb-He isotope space document lower 3He/4He (<12 Ra) for the Hilina component than for the Loihi component (Figure 7).

[32] HIMU-type recycled crust was proposed based on Pb-He relationships of deep Hana Ridge and Loihi rocks, as shown in Figure 7 [Hanyu et al., 2007]. Hanyu et al. [2007] argued that the Kea (Hilina) component can be explained by mixing of the Loihi component and HIMU that is characterized by very radiogenic Pb isotopes (206Pb/204Pb > 22) and low 3He/4He (5–7 Ra) by assuming a curved mixing line in 206Pb/204Pb-3He/4He space. The newly obtained Hilina data, however, indicates that the inception stage magmas form an almost straight mixing array in 206Pb/204Pb-3He/4He space (Figure 7a). Extrapolating this linear trend toward lower 3He/4He indicates that 206Pb/204Pb of the radiogenic Pb Hilina component should be more than 18.7, and less than 19.5 for a nonzero 3He/4He. For this reason, old recycled crust of HIMU is ruled out as a precursor of the Hilina component. Alternatively, young recycled material (e.g., young HIMU [Eisele et al., 2003]), is a likely explanation for the Hilina component. Simulations of Pb isotopic evolution indicate that 206Pb/204Pb would not exceed 18.8 for recycled oceanic crust younger than 1.5 Ga [Eisele et al., 2003], consistent with the 206Pb/204Pb value of the Hilina component (between 18.7 and 19.5). Although 3He/4He of young HIMU has not been estimated, young recycled oceanic crust would have lower 3He/4He than MORB due to degassing at mid-ocean ridge and in the subduction zone, followed by radiogenic ingrowth of 4He by U and Th decay [Graham et al., 1992].

[33] FOZO is a mantle end-member common to many ocean island basalts, having relatively radiogenic Pb isotopic compositions [Hart et al., 1992; Stracke et al., 2005; Jackson et al., 2007], and therefore it may be a candidate of the Hilina component [Ren et al., 2006]. However, estimated 206Pb/204Pb of FOZO is between 19.5 and 20.5 [Stracke et al., 2005], inconsistent with moderately radiogenic Pb isotopes of the Hilina component. Furthermore, modest 3He/4He of the Hilina component is inconsistent with a relatively primitive, less degassed FOZO-type source.

[34] Kimura et al. [2006] proposed that mantle metasomatism might establish the necessary geochemical characteristics of the Hilina component. They invoked metasomatism of the proto-Hilina source by melts that had left residual sulfide in their own sources. Those metasomatic liquids would therefore have had high U/Pb, and with time, the regions they metasomatised would develop high 206Pb/204Pb. Because Rb and Sr are typically incompatible during peridotite melting, the metasomatic melt may have had only moderately elevated Rb/Sr, so the regions it metasomatised need not have developed high 87Sr/86Sr. Upon later melting, the aged metasomatised regions would yield liquids with high 206Pb/204Pb and moderate 87Sr/86Sr like the Hilina component. The relatively low 3He/4He identified for the Hilina component might be consistent with a metasomatic scenario. Partitioning experiments indicate that He might be more compatible than U in olivine [Parman et al., 2005]. Low-degree melts that leave behind an olivine-rich residue could therefore have U/He greater than their source, and metasomatism by such melts could create source regions with high U/He, although it is still under debate [Kurz et al., 2009; Starkey et al., 2009]. Subsequent radioactive decay of U would produce 4He, leading to low 3He/4He over time, consistent with the Hilina component.

4.3. Implications for Magmatism in the Hawaiian Mantle Plume

4.3.1. Temporal and Spatial Variations in Magma Sources

[35] Our observations, in combination with previous studies on shield stage magmas, provide information on the early stage geochemical evolution of Hawaiian volcanoes. Isotopic differences between Hilina, shield stage Kilauea, and shield stage Mauna Kea (HSDP) document temporal compositional changes in the source of the Kea trend volcanoes. The majority of Mauna Kea HSDP samples (Type-1 of Rhodes and Vollinger [2004] and Kea-mid8 of Eisele et al. [2003]) have very similar petrologic and geochemical characteristics to modern Kilauea tholeiites. Provided that the two neighboring volcanoes on the same geographical trend experienced similar magmatic evolutions, compositional differences between the early Kilauea products from the Hilina region and the mature shield products of the HSDP samples reflect temporal development of the source of Kea trend volcanoes. The striking difference in Sr, Nd and Hf isotope ratios between the Hilina and HSDP samples is that the Hilina samples show limited variation in isotopic composition and are at the high 206Pb/204Pb extreme for Hawaiian magmas. With the onset of Kilauea's shield stage volcanism, the Kilauean magmas become isotopically indistinguishable from the majority of Mauna Kea tholeiites. Subsets of Mauna Kea tholeiites define arrays (Kea-lo8, Kea-mid8) that project close to the Hilina component in Pb-He isotope space (Figure 7a) with the baseline 3He/4He value of the deep HSDP samples (12–14 Ra) [Kurz et al., 2004]. This suggests that the Hilina component is a common source for the inception and shield stages of Kea trend volcanoes, but that the relative contribution of the Hilina component diminishes, while that of low-206Pb/204Pb components (e.g., DM component) increases, as the volcanoes enter their highly productive shield stages (Figure 7).

[36] It is widely documented that the geographically defined Loa trend and Kea trend volcanoes differ in major and trace element compositions on average, as well as in isotopes [e.g., Stille et al., 1986; Frey and Rhodes, 1993; Xu et al., 2007], and these differences are well signaled by Pb isotope ratios [Tanaka et al., 2002; Abouchami et al., 2005]. Kea trend (including Kilauea and Mauna Kea) and Loa trend (including Mauna Loa and Loihi) volcanoes define subparallel arrays in 206Pb/204Pb-208Pb/204Pb space. Compilation of He and Pb isotopic data also demonstrate similar elongate arrays for the Kea and Loa trend volcanoes (Figure 7a). Loa trend volcanoes, as typified by Mauna Loa, project to high proportions of the Loihi inception stage component, whereas Kea trend volcanoes (Kilauea, Mauna Kea) project to high proportions of the Hilina component. Though measurements are limited, some volcanoes (e.g., Hana Ridge) that are intermediate in the isotopic compositions of their shield stages products appear to also have been intermediate in their inception stage (Figure 7). These observations suggest that (1) the inception and shield stages of a particular volcano share a similar magma source, which is potentially formed by mixing of the Loihi and Hilina components; (2) the Kea trend volcanoes receive more material from the Hilina inception stage component, whereas the Loa trend volcanoes receive enhanced proportions of the Loihi component; and (3) the depleted components do not contribute appreciably to magmatism until the volcanoes enter their highly productive shield stages.

4.3.2. Source Region Heterogeneities Sampled by Minor Melts

[37] Although the mixing trends shown above explain most of the interrelationships between the major components, we recognize that minor melts also participate during the inception and shield stages. Notable is the degassed tholeiitic sample from the base of the Puna Ridge (S506-5A) that has intermediate 3He/4He around 17 Ra and Pb-Sr-Nd-Hf isotopic compositions approaching those of Mauna Loa, demonstrating that the Koolau component contributed to its production (Figure 7). The degassed character and position far from Mauna Loa and from Kilauea's summit vent lead to the interpretation that it is a Kilauean shield stage product erupted through the East Rift Zone, and its isotopic character indicates that Mauna Loa–like melts occasionally enter Kilauea's plumbing system, as has also been found on the subaerial shield [Rhodes et al., 1989; Kurz, 1993]. Similarly variable 3He/4He among subaerial Kilauea and Hilina rocks may signify injection of such melts into small-scale magma chambers during inception and shield stages [Kurz, 1993]. Kilauea-like melts also occasionally exploit Mauna Loa's plumbing system [Pietruszka and Garcia, 1999; Marske et al., 2007, 2008], and samples with Mauna Loa–like isotopic compositions have erupted from Loihi [Garcia et al., 1993; Abouchami et al., 2005]. Heterogeneity is also documented by diverse Pb isotopic compositions recorded in early Kilauea melt inclusions and glass fragments [Shimizu et al., 2001]. These observations suggest that highly heterogeneous melts are introduced, albeit in minor amounts, to the magmatic conduit system, reflecting small-scale heterogeneity of plural components in the source region or complex plumbing. However, given the clear correspondence between Kea and Loa geographic trends and isotopic fields, major source mixing between the Loihi and Koolau/DM components accounts for the Loa trend volcanoes, whereas mixing between Hilina and Koolau/DM components predominates in the Kea trend volcanoes.

5. Summary

[38] The Hilina region offshore of the active Kilauea shield is a unique place where inception stage Hawaiian rocks are exposed in abundance. Alkalic and transitional rocks are geochemically and isotopically distinct from modern Kilauea lavas, indicating temporal source changes during growth of the volcano. In combination with previous geochemical studies for subaerial shield stage lavas, submarine samples and drill cores by HSDP, the present results imply:

[39] 1. Inception stage magmas, including early Kilauea (Hilina), Loihi, and early Haleakala (deep Hana Ridge), were produced mainly by two-component mixing between the Hilina and Loihi type source materials with negligible contributions from the components that are ubiquitous in the shield, postshield, and rejuvenated stages (Koolau/EM1 and DM).

[40] 2. The Hilina component has the highest 206Pb/204Pb of any common Hawaiian igneous material, but does not have elevated 3He/4He. This distinguishes it from the Loihi component and shows that Hawaiian magmatism is not characterized by a monotonic decrease in 3He/4He with volcano evolution (with superimposed low-level noise). Instead, the existence of two inception stage components that differ in He and Pb clarifies complex mixing arrays documented in shield stage tholeiites from various volcanoes.

[41] 3. The present Hawaiian magmatic source is roughly bilaterally divided. The Hilina and Loihi components are ubiquitous, but the Hilina component predominates on the northeast and the Loihi component predominates on the southwest.

Appendix A:: Analytical Methods

[42] The rock chips were soaked in warm water for 1–2 weeks for desalting, then rinsed by acetone and deionized water in an ultrasonic bath. Dried chips were further crushed in an iron pestle and powdered in an alumina mill. Major element compositions were measured by an X-ray fluorescence spectrometer (XRF: RIGAKU Simultix 12) with fused glass beads at Institute for Research on Earth Evolution (IFREE), JAMSTEC, following the method described by Tani et al. [2005]. Concentrations of trace elements, including rare earth elements (REEs), were determined by solution based inductively coupled plasma mass spectrometer (ICP-MS: Thermo ELEMENTAL VG PQ3 at Shimane University [Kimura et al., 2006]). The samples were digested with a HNO3, HF and HClO4 mixture of analytical grade, dried, then dissolved in diluted mixed acid. Standard addition method was employed using diluted SPEX multielement mixed standard.

[43] Sr, Nd and Pb isotopic ratios were determined by a multiple-collector (MC)-ICP-MS (VG ELEMENTAL Plasma-54) at Shimane University. Prior to isotopic analyses, the powdered sample splits were leached with 6M-HCl at 150°C for 2 h, then rinsed with deionized water at least three times until the supernatant liquid was clear, and dried. Acid reagents were Ultrapur grade HF and HNO3 (Merck), and precise measurement grade HCl (Wako Chemicals). Distilled ion exchanged water and HCl were simmered before use, and procedural blanks were <1 pg/g for both Nd and Sr. Element separation was conducted by conventional cation exchange column using DOWEX AG50W-X8 for Sr and REEs separation and further HDEHP separation for Nd [Kimura et al., 2006]. NIST SRM987 Sr standard and La Jolla Nd standard were analyzed before and after each 3 unknowns. Standard values during the analyses were 87Sr/86Sr = 0.710248 ± 0.000052 (n = 3, errors in 2σ) and 143Nd/144Nd = 0.511852 ± 0.000036 (n = 3, errors in 2σ). Typical internal 2 standard errors for sample analyses were ±0.000010 for Sr and ±0.000015 for Nd. Pb was separated using DOWEX 1X8 anion exchange resin following a single column method. Acids used were Tama Chemicals TAMA PURE AA 10 grade HCl, HBr, and HF. Procedural blank for Pb was typically <50 pg. Pb isotopes were also analyzed by the MC-ICP-MS. Mass fractionation factors were corrected using a Tl external standard. Additional mass-dependent interelement fractionations were also corrected [Kimura et al., 2006]. Normalizing values for the Pb isotope ratios were from Baker et al. [2004] using 206Pb/204Pb = 16.9416, 207Pb/204Pb = 15.4999 and 208Pb/204Pb = 36.7259. NIST SRM981 was monitored during analyses and errors were ±0.0043 for 206Pb/204Pb, ±0.0044 for 207Pb/204Pb and ±0.0117 for 208Pb/204Pb (n = 3). Typical in-run precisions were ±0.0009 for 206Pb/204Pb, ±0.0008 for 207Pb/204Pb and ±0.0025 for 208Pb/204Pb. Note that Kimura et al. [2006] used SRM981 normalization value of Todt et al. [1996] and Pb isotope ratios from the literature data were recalibrated to the values of Baker et al. [2004].

[44] Hf isotopic ratios were also determined by an MC-ICP-MS (IsoProbe; GV Instruments) at Earthquake Research Institute, The University of Tokyo. First, the samples were leached as described previously, digested with HF and HClO4, dried, and then dissolved in HCl using Ultrapur grade reagents (Kanto Chemicals). Hf was separated by a single-column method using Ln-Spec resin (Eichrom) following Münker et al. [2001]. The total procedural blank was <30 pg. During measurement of isotope ratios, 173Yb and 175Lu peaks were monitored to correct for the interference of 176Yb and 176Lu on the 176Hf peak. Mass fractionation factor was determined by the 179Hf/177Hf ratio and isotope ratios were normalized to a 179Hf/177Hf value of 0.7325. The JMC475 Hf standard showed a mean 176Hf/177Hf value of 0.282149 ± 0.000011(2σ, n = 6) during measurements. Reported 176Hf/177Hf ratios are adjusted to a reference JMC475 value of 0.28216. Detailed analytical methods are described by Hanyu et al. [2005].

[45] Glass margins of pillow lavas and separated olivine phenocrysts were analyzed for noble gases. The samples were soaked in diluted HNO3 in ultrasonic bath, then washed in acetone, ethanol and deionized water. After drying, the samples were loaded in tubes for in vacuo crushing, then baked out for 1 day to reduce gas blanks. Gases were extracted by crushing 20–100 times for each sample. After purification of gases, abundances and isotopic ratios of He, Ne and Ar were measured separately on a GV5400 (GV Instruments) mass spectrometer at IFREE, JAMSTEC. Abundances of Kr and Xe were determined simultaneously with Ar. Average blanks were 2 pcm3STP and 0.8 ncm3STP for 20Ne and 40Ar, respectively, with atmospheric isotopic compositions. 4He blank was too low to be measured by Faraday cup, but is estimated to be less than 10 pcm3STP. Blanks of 84Kr and 132Xe were less than 0.001 pcm3STP. All the reported abundance and isotopic data were calibrated by repeated measurements of air and in-house He standards. See Hanyu et al. [2007] for detailed methods.

[46] Using 40Ar/39Ar incremental heating techniques, we analyzed crystalline groundmass from basalt pillow fragments. Analytical techniques are summarized here; for more detail, see Calvert and Lanphere [2006]. The two most crystalline samples from the Hilina transitional basalt suite were selected for analysis, crushed and ultrasonicated for several hours to remove glass and fine-grained (submicron) components. Dense, clean groundmass was concentrated using a Frantz magnetic separator and careful handpicking under a binocular microscope. Samples were irradiated with 27.87 Ma TCR-2 sanidine monitors in Cd-shielded vials at the U.S. Geological Survey TRIGA reactor in Denver, Colorado [Dalrymple et al., 1981]. Argon was extracted from groundmass by resistance furnace, purified continuously during extraction using two SAES ST-172 getters operated at 4A and 0A, and analyzed on an MAP216 mass spectrometer with a Baur-Signer source and Johnston multiplier. Mass spectrometer discrimination was determined by measuring atmospheric argon before (40Ar/36Ar = 288.69 ± 0.15) and after (40Ar/36Ar = 288.76 ± 0.15) analysis of unknowns. Weighted mean plateau ages (WMPA) and isotope correlation (isochron) ages and isotopic ratios are reported with 2σ errors. Full analytical results including age spectra, K/Ca, K/Cl, radiogenic yield, and isochron plots are available in Appendix B.

Appendix B:: Data for 40Ar/39Ar Dating

[47] Data for each stepwise temperature fraction are shown in Table B1. Isotopic intensities are blank, background, mass discrimination (D1 = 1.005761 ± 0.000127), and decay corrected. Analytical errors in Table B1 are 2σ. Plateau and isochron diagrams are shown in Figure B1.

Figure B1.

The 40Ar/39Ar age spectra and isochron plots for K207-1 and K207-4. Weighted mean plateau ages (WMPA) were obtained using data for the medium temperature fractions for both samples and for all the temperature fractions except for the highest temperature step for K207-1. Boxes for apparent ages and isotopic ratios display errors in 2σ level. Errors of interpreted ages correspond to 2σ.

Table B1. The 40Ar/39Ar Tabulated Data
Temperature (°C)Age (ka)%40Ar*K/CaK/ClMoles 40Ar*Σ39Ar40Ar39Ar38Ar37Ar36Ar
  • a

    Packet IRR267-KL, experiment #irr267-kl, 0.1632 g basalt, and all errors ±2 sigma. J = 0.000136893 ± 1.75E-06. 40Ar* is radiogenic argon. Isotopes are in volts (1.91e-14 moles/volt), corrected for blank, background, discrimination, and decay. Calculated K2O = 0.10 wt %, calculated CaO = 6.55 wt %, and calculated Cl = 0.2 ppm. Total gas age = 12.0 ± 42.6 ka. Weighted mean plateau age = 66.8 ± 28.6 ka (including ±J), 62.9% 39Ar released. MSWD = 0.089 (good fit, MSWD < 3.12). Steps 4 of 11 (650, 700, 750, and 800°C). Isochron age = 94.2 ± 131.1 ka (95% confidence, including ±J). MSWD = 0.07 (good fit, MSWD < 3.69). 40Ar/36Ar intercept = 294.0 ± 11.3 (95% confidence). Steps 4 of 11 (650, 700, 750, and 800°C).

  • b

    Packet IRR267-KM, experiment #09z0036, 0.0906 g basalt, and all errors ±2 sigma. J = 0.000131256 ± 2.03E-06. 40Ar* is radiogenic argon. Isotopes are in volts (1.91e-14 moles/volt), corrected for blank, background, discrimination, and decay. Calculated K2O = 0.32 wt %, calculated CaO = 7.29 wt %, and calculated Cl = 0.3 ppm. Total gas age = 90.6 ± 25.3 ka. Weighted mean plateau age = 65.2 ± 27.7 ka (including ±J), 99.3% 39Ar released. MSWD = 0.96 (good fit, MSWD < 2.05). Steps 11 of 12 (550, 600, 650, 700, 750, 800, 850, 900, 950, 1000, and 1050°C). Isochron age = 71.4 ± 65.8 ka (95% confidence, including ±J). MSWD = 1.36 (good fit, MSWD < 2.11). 40Ar/36Ar intercept = 295.1 ± 3.2 (95% confidence). Steps 11 of 12 (550, 600, 650, 700, 750, 800, 850, 900, 950, 1000, and 1050°C).

K207-1 Basalta
55075.3 ± 193.50.960.0511802.998e-170.020.162518 ± 0.0003710.005176 ± 0.0000330.000188 ± 0.0000220.059106 ± 0.0006530.000561 ± 0.000013
600226.7 ± 74.63.380.0411073.137e-160.080.485188 ± 0.0010130.018038 ± 0.0000640.000602 ± 0.0000260.254879 ± 0.0011250.001658 ± 0.000016
65061.6 ± ± 0.0018160.042271 ± 0.0001030.001173 ± 0.0000250.682324 ± 0.0016070.003152 ± 0.000033
70069.3 ± 31.61.670.0314292.959e-160.400.924545 ± 0.0018940.055769 ± 0.0001130.001474 ± 0.0000250.900655 ± 0.0019320.003330 ± 0.000021
75075.5 ± ± 0.0026070.053625 ± 0.0001090.001683 ± 0.0000360.802884 ± 0.0028110.004505 ± 0.000027
80032.4 ± 75.20.320.0411629.188e-170.711.518286 ± 0.0030820.036995 ± 0.0000960.001578 ± 0.0000290.518347 ± 0.0012950.005267 ± 0.000031
850282.4 ± 94.92.620.044714.842e-160.780.965791 ± 0.0020390.022316 ± 0.0000920.001088 ± 0.0000330.269858 ± 0.0010680.003259 ± 0.000021
900−221.2 ± 101.0−2.760.04326−2.976e-160.840.564297 ± 0.0012380.017524 ± 0.0000900.000825 ± 0.0000320.231552 ± 0.0014060.002027 ± 0.000019
975−267.3 ± 126.7−1.770.02131−4.331e-160.911.279030 ± 0.0026650.021387 ± 0.0000860.001798 ± 0.0000400.676840 ± 0.0025080.004595 ± 0.000026
105035.4 ± ± 0.0037500.023734 ± 0.0000830.001906 ± 0.0000363.812471 ± 0.0078380.007227 ± 0.000030
1150−1713.4 ± 433.2−17.380.01222−7.768e-161.000.233556 ± 0.0011230.006068 ± 0.0001460.000371 ± 0.0000200.303998 ± 0.0013880.001013 ± 0.000018
K207-4 Basaltb
550172.0 ± 70.82.730.2211553.380e-160.050.646626 ± 0.0013370.024353 ± 0.0000760.000801 ± 0.0000310.059278 ± 0.0006560.002145 ± 0.000022
60071.1 ± 32.71.390.1619033.514e-160.161.322949 ± 0.0027020.061263 ± 0.0001240.001750 ± 0.0000370.198162 ± 0.0014400.004471 ± 0.000024
65059.5 ± ± 0.0033500.082598 ± 0.0001670.002251 ± 0.0000550.372581 ± 0.0016990.005609 ± 0.000030
700104.4 ± 32.71.900.0918506.659e-160.471.835753 ± 0.0037270.079240 ± 0.0001600.002345 ± 0.0000440.456050 ± 0.0015020.006223 ± 0.000030
75060.4 ± 29.11.370.0814873.559e-160.611.356987 ± 0.0027700.073181 ± 0.0001480.002002 ± 0.0000400.457273 ± 0.0010980.004658 ± 0.000026
80075.3 ± 32.81.800.0916663.458e-160.721.004130 ± 0.0020640.057026 ± 0.0001160.001507 ± 0.0000250.345268 ± 0.0015940.003434 ± 0.000024
85033.3 ± 38.60.660.129731.326e-160.811.047603 ± 0.0022010.049491 ± 0.0001150.001513 ± 0.0000270.219608 ± 0.0010860.003583 ± 0.000023
900−32.2 ± 50.0−0.380.151062−1.239e-160.901.712959 ± 0.0035320.047662 ± 0.0000970.001892 ± 0.0000290.162654 ± 0.0009550.005864 ± 0.000027
950−13.3 ± 115.8−0.070.06493−2.788e-170.951.993466 ± 0.0040920.026165 ± 0.0000780.001828 ± 0.0000290.220534 ± 0.0011210.006813 ± 0.000030
1000187.1 ± 216.70.670.011511.978e-160.981.551310 ± 0.0032080.013557 ± 0.0000520.001536 ± 0.0000330.676414 ± 0.0029390.005405 ± 0.000028
1050259.7 ± 284.60.700.003441.607e-160.991.194080 ± 0.0024790.008733 ± 0.0000290.000993 ± 0.0000341.518849 ± 0.0039770.004439 ± 0.000028
11503479.3 ± 396.013.250.013271.042e-151.000.410832 ± 0.0009150.003953 ± 0.0000310.000333 ± 0.0000230.354138 ± 0.0016350.001306 ± 0.000017


[48] We thank the scientific party, the crews on the R/V Yokosuka and Kairei, and the operation teams of submersibles Shinkai 6500 and Kaiko of JAMSTEC Hawaiian cruises in 1998–2002. We thank K. Sato for her help with noble gas analyses. K. Tani is acknowledged for major element analyses by XRF. The USGS 40Ar/39Ar facility operates with the able assistance of J. Saburomaru and B. Ito. GEOROC database was helpful to collect compiled geochemical data. We are grateful to M. L. Coombs and A. R. L. Nichols for constructive comments. Careful reviews by M. D. Kurz, an anonymous reviewer, and the Editor V. J. M. Salters have greatly improved the manuscript.