Kahoolawe Island, Hawaii: The role of an ‘inaccessible’ shield volcano in the petrology of the Hawaiian islands and plume

Kahoolawe Island, Hawaii: The role of an ‘inaccessible’ shield volcano in the petrology of the Hawaiian islands and plume

ARTICLE IN PRESS Chemie der Erde 70 (2010) 101–123 Contents lists available at ScienceDirect Chemie der Erde journal homepage: www.elsevier.de/cheme...

5MB Sizes 1 Downloads 45 Views

ARTICLE IN PRESS Chemie der Erde 70 (2010) 101–123

Contents lists available at ScienceDirect

Chemie der Erde journal homepage: www.elsevier.de/chemer

INVITED REVIEW

Kahoolawe Island, Hawaii: The role of an ‘inaccessible’ shield volcano in the petrology of the Hawaiian islands and plume R.V. Fodor a,n, G.R. Bauer b a b

Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, NC 27695, USA Mink & Yuen, Inc. Honolulu, Hawaii 96826, USA

a r t i c l e in fo

abstract

Article history: Received 8 September 2009 Accepted 4 January 2010

Kahoolawe volcano ( 10  17 km) forms one of the eight major Hawaiian islands. Access for geologic sampling has long been restricted due to military and preservation policies. However, limited visits to Kahoolawe in the 1980s yielded 4 200 samples, many of which have since been used to study the volcano within the framework of Hawaiian shield and mantle source geochemistry, petrology, mineralogy, and igneous processes. Kahoolawe is a tholeiitic shield with tholeiitic caldera-filling lavas, and at least seven postshield vents that erupted tholeiitic and (sparse) alkalic lavas. On smaller scales are a gabbro intrusion and ultramafic and gabbroic xenoliths in some postshield lavas. There is no evidence for rejuvenated volcanism. In its structural setting, Kahoolawe lies along the ‘‘Loa’’ trend of Hawaiian shields. Major element compositions of shield and caldera-filling lavas are similar and cluster at  6–7 wt% MgO, range from  5.5 to 16 wt% MgO, and include  9 wt% MgO examples that can be modeled as parental to the evolved lavas. For example, least squares mass balancing demonstrates that from  15% to 30% crystallization of olivine ( 7 orthopyroxene), clinopyroxene, and plagioclase accounts for the  5.5–6 wt% MgO range of tholeiitic lavas. Greater differentiation occurred in the gabbro (diabasic) intrusive body as a segregation vein with  2.9 wt% MgO, and extreme differentiation produced local, small-volume rhyolitic melts that segregated into lava vesicles. Postshield lavas are mainly tholeiitic, have  5–7 wt% MgO, and overlap shield compositions. Some are alkalic, as low as  3.9 wt% MgO (hawaiite), and can be modeled as liquids after a  9 wt% MgO alkalic magma crystallized  30% olivine, clinopyroxene, plagioclase, and magnetite. Important aspects of Sr, Nd, Hf, and Pb isotopic ratios in Kahoolawe shield and caldera-filling lavas are slightly higher 87Sr/86Sr than in Koolau shield lavas (Oahu island; Makapuu-stage; e.g., Koolau mantle ‘endmember’) when compared at a given 143Nd/144Nd (e.g.,  0.7042 at 0.5128), 206Pb/204Pb largely at the low end of the range for Hawaiian shields (e.g.,  18), and eHf generally comparable to the values of other Hawaiian shields and ocean islands (e.g., eHf 8 at eNd 4). The isotopic ratios overall suggest small-scale source heterogeneity, considering the island size, and that Kahoolawe shield and caldera lavas were derived from a Hawaiian plume source containing recycled oceanic crust of gabbro and sediments. Based on certain geochemical indicators, however, such as Ce/Sr and La/Nb vs. 87Sr/86Sr, the source contained slightly less gabbro component than other shield sources (e.g., Koolau). Isotopic data for Kahoolawe postshield lavas are scarce, but those available generally overlap the shield data. However, ratios among certain alterationresistant incompatible trace elements (e.g., Zr/Nb) discriminate some postshield alkalic from shield lavas. But because the different ratios for those postshield lavas can be explained by smaller partial-melting percentages of the shield source and by differentiation, neither isotopes nor trace elements identify postshield magmas as originating in a source unlike that for the shield lavas. & 2010 Elsevier GmbH. All rights reserved.

Keywords: Shield volcano Hawaii petrology Basalt mineralogy Tholeiitic basalt Hawaiian mantle source Igneous differentiation

1. Introduction Since petrologic studies of the Hawaiian Islands began in earnest about 75 years ago (e.g., Stearns and Vaksvik, 1935, 1938), Kahoolawe – the smallest of the eight main islands (Fig. 1) – has

n

Corresponding author. Tel.: + 1 919 515 7177. E-mail address: [email protected] (R.V. Fodor).

0009-2819/$ - see front matter & 2010 Elsevier GmbH. All rights reserved. doi:10.1016/j.chemer.2010.01.001

been unpopulated by Hawaiian society and one of the most difficult Hawaiian islands to access for geologic studies. Restrictions to Kahoolawe began during World War II when the US Navy took occupation for military purposes. Post-war until 1990, the Navy continued a military program for using Kahoolawe as a bombing target. For several decades, then, as field, sampling, and geochemical programs developed for the other Hawaiian islands, only initial, pre-war Kahoolawe studies by Stearns (1940) and Macdonald (1940) provided any geologic information about

ARTICLE IN PRESS 102

R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

Fig. 1. Map of Kahoolawe Island, Hawaii, showing the locations for three stratigraphic sections sampled for shield lavas (asterisks; Kanapou Bay; Waikahalulu Bay, and southeastern shore) and one for caldera-filling lavas (two small stars), the seven postshield vents identified by Stearns (1940), the Kanapou Bay gabbro intrusion and a gabbro block collected by Macdonald (1940, sample N21), and two ‘‘post-erosional’’ eruptives reported by Stearns (1940) and Macdonald (1940) (our samples 32 and 156). The inset shows Kahoolawe in relationship to other Hawaiian islands, and its position on the Loa trend of volcanoes.

this small, 10  17 km shield volcano. Even today, long after the US Navy relinquished its position on Kahoolawe, the island remains protected and restricted from public access under the state of Hawaii’s Kahoolawe Island Reserve Commission. Sampling Kahoolawe for geologic purposes remains almost impossible, and only about 20 of the original Stearns (1940) and Macdonald (1940) samples remain available for study. There was, however, a window of time in the 1980s when the US Navy’s Pacific Fleet permitted seven visitations for purposes of sampling lavas. Six of these excursions involved at least one of the authors (GRB), and all combined they yielded  190 lava (and some gabbroic) samples; a seventh visit made by M. Garcia provided 25 samples. Since then, various studies of nearly half of the  235 existing Kahoolawe samples have enabled evaluations of the geochemical, petrologic, and geochronological history of Kahoolawe volcano. That limited opportunity to sample Kahoolawe has led, over the past 20 years, to bringing an understanding of its geology up to the levels known for most all other Hawaiian islands. Accordingly, this report is a review and summary of Kahoolawe characteristics that describe its composition, shallow-level igneous processes, and mantle source, and how the volcano fits into the overall geochemical, spatial, and stratigraphic development of Hawaiian volcanoes and magma systems. Nearly 200 samples collected by the authors are now in the public domain collection of the Smithsonian Institution, cataloged, and available for additional study.

2. Previous reports and Kahoolawe general geology Stearns (1940) and Macdonald (1940) conducted the earliest field explorations to establish the volcanic stratigraphy and geology for Kahoolawe and to sample for lava petrography and compositions. Those investigations led Stearns to describe Kahoolawe Island as constructed by lavas that are pre-caldera (shield), caldera-filling, and post caldera (postshield). The shield lavas, he noted, are best exposed along steep southern and southeastern shoreline cliffs, and the caldera-filling lavas form much of the eastern Kanapou Bay shoreline (Fig. 1). Stearns (1940) mapped Kahoolawe as largely capped by postshield lavas

and identified them as originating from seven discrete vents dispersed across the volcano – the largest of which is Makika and occupies the caldera location (Fig. 1). He grouped all of the Kahoolawe lavas as the Kanapou Volcanics. Stearns (1940) and Macdonald (1940) additionally described dikes and cinder eruptives associated with lavas and faults in the caldera lavas along the steep Kanapou Bay shoreline. They categorized these features as post-erosional, or ‘recent’, and interpreted them as emplaced after erosion had carved sea cliffs and developed talus in the caldera region (locations of two are in Fig. 1, noted as 32 and 156). Macdonald (1940) also described a gabbroic block collected from a fault breccia, and a diabasic rock, both from the caldera region. Structurally, Stearns (1940) noted that Kahoolawe’s caldera marked the focal point for the intersection of north, east, and southwest rift zones. In a larger structural framework – namely that of Hawaiian shield locations with respect to the subparallel, curvilinear Loa and Kea trends of volcanoes extending from Loihi hotspot (e.g., Jackson et al., 1972) – Kahoolawe lies along the Loa trend (Fig. 1). Stearns’ (1940) Kahoolawe geologic map and island stratigraphy are now in digital format for internet availability from the US Geological Survey (Sherrod et al., 2007). The website is http://pubs.usgs.gov/of/2007/1089/. In the initial petrographic and compositional presentation for Kahoolawe, Macdonald (1940) provided a single whole-rock major-element composition and several modal mineral assemblages for 47 samples collected in 1939. All characteristics were those of Hawaiian tholeiitic basalt lavas and what he considered to be basaltic andesite. Without benefit of compositional data, and utilizing only field and petrographic observations, Macdonald (1940) concluded that Kahoolawe magma differentiation was limited relative to that for other Hawaiian volcanoes. He based that on ‘andesine andesite’ lavas (later called hawaiite; Macdonald, 1960) representing the most evolved rock type, and on basaltic andesite being Kahoolawe’s most abundant evolved rock having erupted only after caldera collapse (Macdonald’s original rock names are defined in Macdonald, 1949, p. 1544). Prior to caldera formation, Macdonald (1940) interpreted Kahoolawe shield as constructed largely by olivine basalt lavas. In his view, Kahoolawe volcanism concluded with local post-erosional dikes

ARTICLE IN PRESS R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

and eruptions in the caldera region, and that these events were more likely related to a rift zone extending from neighboring Haleakala (Maui) volcano than to Kahoolawe magmatism. The first attempt to radiometrically date Kahoolawe Island was by Naughton et al. (1980) who presented two K–Ar ages of 1.03 m.y. (Note: those samples were collected by H.S. Palmer in 1925.) A more extensive K–Ar assessment by Fodor et al. (1992) reported two Kahoolawe shield lavas to be 1.25 and 1.4 Ma, and the ages of five postshield lavas between 1.08 and 1.2 Ma. Sano et al. (2006) tested Stearns’ (1940) field interpretation for posterosional volcanism on Kahoolawe. They produced instead K–Ar ages comparable to those known for the shield and postshield lavas. Namely, two dikes penetrating caldera-filling lavas and a caldera-filling lava range between 1.4 and 1.0 m.y. (Note: to retain integrity for the history of Kahoolawe studies, and to both honor and examine Stearns’ (1940) and Macdonald’s (1940) first call of what they believed to be genuine post-erosional eruptives, we identify in this paper our samplings of those eruptives as ‘‘posterosional’’ – presenting the name in quotation marks – to distinguish them as not representing post-erosional, or rejuvenated-stage, Hawaiian volcanism in a strict or genuine sense.) A comprehensive presentation of major and trace element compositions for Kahoolawe shield and postshield vent lavas was first offered by Fodor et al. (1992), which followed some introductions to Kahoolawe lava compositions by West et al. (1987) and Fodor et al. (1987). West et al. (1987) also provided the first Sr, Nd, and Pb isotopic compositions for Kahoolawe, and Leeman et al. (1994) contributed the whole-rock compositions for those in West et al. (1987). Later, Blichert-Toft et al. (1999) added Hf isotopic ratios to the West et al. (1987) samples, and Abouchami et al. (2005) made high-precision re-analyses of the Pb isotopes for a reassessment of the West et al. (1987) ratios. Huang et al. (2005) published the largest, most comprehensive examination of Sr, Nd, Pb, and Hf isotopic ratios for 25 shield and two caldera-filling lavas for the purpose of characterizing Kahoolawe’s mantle sources. Other insights to Kahoolawe petrology extended the evaluation of its coarse-grained aspect in reports about gabbroic and ultramafic xenoliths in some postshield lavas (Rudek et al., 1992; Fodor et al., 1993, 1998), and, in this report, new information on diabasic intrusive rock in shield lavas near the caldera boundary at Kanapou Bay. There is also information for differentiation and magma mixing processes at a postshield vent (Fodor et al., 1998), small-scale occurrences of rhyolite glass (Fodor et al., 1993), and for the identification of REE- and Ba-bearing secondary minerals that originated by weathering processes (Fodor et al., 1989, 1994). Finally, in an archaeological-directed study, Collerson and Weisler (2007) used geochemical characteristics of ocean island basalts to help determine the island sources for a collection of basalt-composition adzes. These are tool artifacts from early Hawaiian and other Polynesian societies that are found on various islands of Polynesia. Of 19 studied adze samples that represent Pacific island basalts, Collerson and Weisler (2007) identified one that likely came from Kahoolawe.

3. Petrography 3.1. Shield Many shield lavas contain olivine phenocrysts, generally o5 vol%, and in some, olivine is attended by trace amounts to a few volume percent of clinopyroxene, orthopyroxene, and/or plagioclase crystals (Fig. 2A, B). Of the three stratigraphic sections sampled, Kanapou Bay and the southeastern section lavas are dominated by olivine, whereas the Waikahalulu section has sparse olivine and mainly clinopyroxene and plagioclase, often

103

in clusters (Fig. 2B). Only a few shield lavas have modal olivine as high as 10–12 vol%, and a few are aphyric. Olivine crystals are commonly o2 mm in size, but there are instances of grains 4–5 mm. These large grains do not appear strained and are generally subhedral, and we therefore do not interpret them as xenocrysts entrained from subvolcanic cumulate dunite. Groundmasses of shield lavas range from glassy, to microcrystalline, to intersertal and fine intergranular (Fig. 2A, B). 3.2. Caldera-filling Some caldera-filling lavas resemble shield lavas by having only a few volume percent olivine, clinopyroxene, or plagioclase, or they are aphyric. But one is olivine-rich,  11 vol%, and a few are ankaramitic (abundant pyroxene plus olivine), and some others have diabasic (Fig. 2C) and micro-diabasic textures. 3.3. Postshield Postshield lavas range from aphyric to highly porphyritic, up to 20 vol% crystals as in a Kealialalo lava, and with a variety of modal mineral combinations involving olivine, clinopyroxene, and plagioclase (Fig. 2D). More commonly, however, postshield lavas have o5 vol% phenocrysts that range in size from 1 to 4 mm. Some lavas have only olivine as phenocrysts (Kealialuna), whereas most others have small amounts ( 1 vol%) of olivine and a few percent of clinopyroxene and plagioclase phenocrysts. Groundmasses in postshield lavas are generally micro-intergranular textures. The lavas from Moaula vent are unusual among all other postshield lavas because the olivine and plagioclase phenocrysts are subrounded to rounded, some with resorption margins, and some moderately large, up to 4 mm in size (Fig. 2E). The hawaiites of Kamama vent can be nearly aphyric or carry plagioclase and olivine grains altered to iddingsite, each up to 2 mm in size (Fig. 2F). 3.4. ‘‘Post-erosional’’ eruptives We sampled two different outcroppings of what Stearns (1940) and Macdonald (1940) believed to represent the final (post-erosional) volcanic activity on Kahoolawe. They have emplacement types and textures that differ from one another. A dike is glassy and with  2 vol% olivine and plagioclase (Fig. 2G), and a tuff cone, also glassy, has  15 vol% glomerocrysts of olivine, clinopyroxene, and plagioclase. The tuff cone additionally contains blocks, r10 cm, of shield and/or caldera-filling lavas. 3.5. Coarse-grained rocks Non-volcanic, coarse-grained rocks on Kahoolawe occur as ultramafic and gabbroic xenoliths and as a gabbroic intrusive body. Ultramafic xenoliths are 1–2 cm in size and comprised of lherzolite, harzburgite, norite, and wehrlite rock types within a postshield picrite lava (Rudek et al., 1992). The principle texture among these xenoliths is poikilitic (Fig. 2H) and it reflects their cumulate origin. Two other postshield lavas contain gabbroic xenoliths. One is Moaula vent lava containing 1–5-cm sized intergranular-textured orthopyroxene gabbro and andesine gabbro (Fodor et al., 1998). The other postshield lava has gabbro xenoliths that are 1–3 cm and mainly intergranular-subophitic plagioclase, clinopyroxene, and orthopyroxene, plus vesicular basaltic glass dispersed throughout and occupying  30–45 vol% of xenolith space (Fodor et al., 1993). The interspersed glass creates open-textured gabbros (Fig. 2I). Vesicles in and around the gabbro xenoliths are lined or filled with glass.

ARTICLE IN PRESS 104

R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

The gabbro intrusion and the gabbro block of Macdonald (1940, sample N21) collected at the opposite side of Kanapou Bay from the intrusion (Fig. 1) are medium- to coarse-sized grains (2–6 mm) of clinopyroxene, plagioclase, and FeTi oxides in largely intergranular relationships (diabasic; Fig. 2J, K). They also include minor orthopyroxene, pigeonite, olivine in some samples, and pockets of interstitial glass, partially devitrified. A segregation vein in the intrusion consists largely of clinopyroxene and plagioclase in a glassy mesostasis (Fig. 2L).

4. Major- and trace-element compositions The data sources for most all Kahoolawe major and trace element compositions are Fodor et al. (1992), Leeman et al. (1994), and Huang et al. (2005). Some additional whole-rock compositions are in Jacobs (1986), Rudek (1988), Rybek (1995), and Fodor et al. (1993, 1998).

4.1. Shield lavas The total alkalis vs. SiO2 diagram in Fig. 3A is for 39 shield lavas, more than half collected from one of three stratigraphic sections along the southern and eastern shorelines. Fig. 3A shows the tholeiitic compositions of the shield lavas, where all plot below the Hawaiian reference line for tholeiitic and alkalic basalts at  48–52 wt% SiO2 and in a 2–3 wt% range for total Na2O + K2O. The total alkali abundances in some lavas, however, are likely to be slightly lower than magmatic values because of their exposure to tropical weathering. Namely, when applying the K2O/P2O5 and K/Rb low-temperature alteration filters for Hawaiian shield lavas (e.g., Chen and Frey, 1985; Frey et al., 1990, 1994; Huang and Frey, 2003), the respective o1 and 4600 ratios illustrated in Fig. 3C suggest that K and Rb were leached from several Kahoolawe lavas. More than half the shield lavas analyzed have K2O/P2O5 o1 (Fig. 3C), and, as observed for fresh Hawaiian lavas such as at Mauna Loa, ratios are  1.4 and certainly 41 (e.g., Rhodes, 1996). Some Kahoolawe shield lavas are within this range, and a few are

Fig. 2. Photomicrographs of Kahoolawe lavas, gabbroic rocks, and xenoliths in postshield lavas. All photos in crossed-nicols except L, and all scale bars (black line at bottom of each photo) are 1 mm long: (A) Shield lava KAH-179, southeastern shore section; olivine (left) and clinopyroxene-plagioclase (right) clusters. (B) Shield lava KAH-67, Waikahalulu section; glomerocryst of clinopyroxene-plagioclase. (C) Caldera-filling lava KAH-29, Kanapou Bay shoreline; diabasic texture of largely clinopyroxene and plagioclase. (D) Postshield lava KAH-115, Kealialalo vent; olivine phenocrysts. (E) Postshield lava KAH-62, Moaula vent; rounded/resorbed andesine and olivine phenocrysts. (F) Postshield hawaiite KAH-130, Kamama vent; trachytic texture of plagioclase crystals. (G) ‘‘Post-erosional’’ dike representing the ‘‘post-erosion eruption E’’ of Stearns (1940) (Fig. 1, marked ‘32’); microphenocrysts of olivine and plagioclase. (H) Xenolith of harzburgite in postshield picrite lava KAH-164; olivine in orthopyroxene oikocryst. (I) Open-textured gabbro xenolith in postshield lava KAH-135; plagioclase and clinopyroxene; some vesicles are glass lined. (J) Gabbro intrusion sample from Kanapou Bay; diabasic texture of largely clinopyroxene and plagioclase and some olivine and Fe–Ti oxides. (K) Gabbro block N21 of Macdonald (1940) found in fault breccia of caldera-filling lavas (Fig. 1); diabasic texture of largely clinopyroxene and plagioclase and some Fe–Ti oxides. (L) Segregation vein in the Kanapou Bay intrusive body, viewed in plane-polarized light; largely plagioclase and clinopyroxene in a light-brown glassy mesostasis. KAH- sample numbers refer to those in Fodor et al. (1992, 1993, 1998), Rudek et al. (1992), Rybek (1995), and Huang et al. (2005).

ARTICLE IN PRESS R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

105

Fig. 2. (Continued)

Fig. 3. A, B. SiO2 vs. total alkalis diagram, with boundary line for alkalic-tholeiitic basalts (after Macdonald and Katsura, 1964). Shield lava compositions encircled in each panel; three postshield vent lavas, alkalic basalt KAH-115, and hawaiite H1440 (Leeman et al., 1994) are identified in panel B. C. K2O/P2O5 vs. K/Rb discrimination to help identify samples from which weathering removed K and Rb as indicated by ratios of o1 and 4600. ‘‘Post-erosional’’ refers to eruptives believed by Stearns (1940) and Macdonald (1940) to represent genuine posterosional events, but are now known to be tholeiitic.

higher. However, due to the weathering of the island and the wide range in K2O/P2O5, it is impossible to know the magmatic K2O/P2O5 ratios. In spite of this, major and trace elements as a whole can still be evaluated with reasonably good confidence, and most major elements are plotted with MgO in the panels composing Fig. 4. The majority of Kahoolawe shield lava compositions are confined to a MgO range of  6–8 wt%, and a large number of them overlap at MgO  7 wt%. Four shield lavas with modal olivine 410 vol% extend the MgO range to 12–16 wt%, creating a ‘tail’ to the compositional field in many of the panels of Fig. 4. Only two lavas, in contrast, are compositionally evolved to o6 wt% MgO, and the most evolved shield lava observed is 5.5 wt% MgO. The wide MgO range, probably due mainly to olivine removal and addition within differentiating parent magmas, affords an opportunity to observe correlations of the other major elements over the range of decreasing MgO. Good examples of correlation are the increasing trends created by TiO2, CaO, Al2O3, and P2O5. But overall, while trends are present, they are weak and largely due to the few samples that extend the compositional range to 412 wt%. In the case of CaO, some samples show a departure at 6.5 wt% MgO from the overall upward trend. Because a corresponding Al2O3/CaO plot shows a generally increasing ratio with decreasing MgO at o6.5 wt%, the slight downward shift in CaO at MgO o6.5 wt% for some shield lavas likely reflects subtraction of more clinopyroxene than plagioclase at MgO o7 wt% during shield magma differentiation.

ARTICLE IN PRESS 106

R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

In the MgO panels of Fig. 4 where most shield lavas overlap at 7 wt%, there are few differences in major element abundance ranges to distinguish the different stratigraphic sections they represent. One example shows that Waikahalulu Bay lavas by and large have higher CaO and lower Al2O3/CaO than other shield sections. As evolved lavas, then, they likely experienced more plagioclase than clinopyroxene removal in their differentiation histories – or, alternatively, contain more modal clinopyroxene than plagioclase (e.g., Fig. 2B). (It is possible that Waikahalulu lavas are derivatives from relatively CaO-rich parentage produced by peridotite partial melting – e.g., Hirose and Kushiro (1993), Herzberg (2006), Huang et al. (2009) – but without a sample of parental, MgO-rich Waikahalulu lava, this cannot be fully assessed.) Additionally, Fodor et al. (1992) plotted compositions of the lavas in the Waikahalulu and Kanapou Bay sections with their sample elevations and reported that correlations between lava compositions and their stratigraphic positions are not apparent. Huang et al. (2005), in like manner, did not observe correlations in the southeastern shore shield section. Relationships between elevations and the Mg#s for the three shield lava stratigraphic sections are in Fig. 5. On a regional scale, plots that compare CaO and SiO2 vs. MgO in Kahoolawe shield lavas to those abundances in the well-documented Koolau, Mauna Loa, and Mauna Kea shields show that Kahoolawe overlaps all these volcanoes (Fig. 6). But while Kahoolawe shield lavas are rather high in SiO2 – many having between 50 and

51 wt% – there is noticeably higher SiO2 in Koolau lavas of Makapuustage (i.e., the uppermost lavas of Koolau shield) (Fig. 6). Among trace elements, abundances of incompatible Zr, Nb, Y, La, Hf, and Th have inverse correlations with decreasing MgO that are apparent largely due to the wide 5.5–16 wt% MgO range among shield lavas (Fig. 7). Trends for Sr and Ba are less defined because a few samples likely reflect the weathering susceptibility of these elements. In the  7 wt% MgO region of the panels where most shield lavas plot, each incompatible trace element is restricted to a relatively small ppm abundance range, demonstrating that there are no clear compositional distinctions among the three stratigraphic sections sampled. The rare-earth elements (REE) are plotted as chondrite-normalized patterns in Fig. 8A. All Kahoolawe shield lavas have light-REE (LREE) relative enrichments that are characteristic of Hawaiian shields, but to varying degrees, where La(n) values extend from  18 to 53 (La(n) =sample La ppm normalized to average chondritic La ppm). A representative primitive-mantle normalized trace-element pattern for shield lavas is in Huang et al. (2005).

4.2. Caldera-filling lavas Of the 18 caldera-filling lavas analyzed, all plot below the tholeiitic-alkalic reference line and overlap SiO2 in shield lavas at  49–52 wt% (Fig. 3A). A few samples, however, have slightly

Fig. 4. MgO variation diagrams for major elements of all Kahoolawe lavas analyzed. Both left- and right-hand panels have shield lava compositions encircled. Several righthand panels have three vents, and samples KAH-115 and H-1440 (hawaiite of Leeman et al., 1994) identified, as these are the most likely to have element abundances different from those of shield and suggest alkalic compositions. ‘‘Post-erosional’’ refers to eruptives believed by Stearns (1940) and Macdonald (1940) to represent genuine post-erosional events, but are now known to be tholeiitic.

ARTICLE IN PRESS R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

107

higher total alkali abundances than shield lavas to distinguish them as possibly representing more extensive differentiation in some caldera-filling magma reservoirs or lower percentages of source melting. But also, as shown in Fig. 3C, many caldera-filling lavas likely lost K and Rb to leaching during weathering. On MgO variation diagrams, most caldera-filling lavas are compositionally evolved, o7.5 wt%, and some overlap the few highly evolved shield lavas that are o6 wt% MgO (Fig. 4). Four caldera-filling lavas have from  11 to 14 wt% MgO. Overall, there is good compositional overlap with shield lavas in the MgO variation diagrams for almost all major elements. Exceptions are for lavas with higher P2O5, K2O, and Ba at MgO o6 wt%. Fig. 5 shows the variation in and the weak correlations between Mg#s and elevations of the caldera-filling lavas where they range from moderately evolved (Mg#  60) to o50 at higher elevations, to the highest Mg#s, 460, at highest elevations. Among trace elements, only Ba is slightly higher than in shield lavas, but this could be an artifact of weathering (Fig. 7). The REE abundances of most caldera-fill lavas are similar to those of the evolved shield lavas, although one lava has a pattern slightly above the compositional range for the shield (Fig. 8B).

4.3. Postshield lavas Fig. 5. Mg#s of shield lavas from three stratigraphic sections (locations in Fig. 1) plus caldera-filling lavas all plotted against the elevations of sample collection sites. (Mg# =mol MgO/(MgO +FeO)n100), where FeO represents 0.9 of all Fe expressed as FeO. (Elevations for the two lowest lavas in the southeastern shore section are corrected after re-evaluation of field notes, and differ slightly from elevations given in Huang et al., 2005.)

Fig. 6. MgO variation diagrams for SiO2 and CaO in Kahoolawe shield (+ ) and caldera-filling (D) lavas for comparison to three other well-documented shields that have generally different SiO2 and CaO abundances. Kahoolawe lavas overlap all three, but weathering may have modified magmatic SiO2 in some lavas. (after Huang et al., 2005) Makapuu-stage refers to the uppermost shield lavas on Koolau volcano.

Most of the postshield lavas analyzed represent one of seven vents (Fig. 1), and the others are from locations with no identifiable source vent. In general, postshield lava compositions range from overlapping those of shield lavas to more evolved in some major elements, such as MgO and K2O, and to more enriched in some incompatible trace elements (Figs. 3, 4, 7). As examples, most lavas from Makika vent, atop the caldera, and from Kealialuna and Kolekole vents resemble shield lavas in most all compositional aspects. Moiwi vent lava compositions, however, are as evolved as the most evolved shield, having MgO 5.6 wt% and slightly higher Na2O+K2O that distinguishes them from shield lavas (Figs. 3B, 4). Some lavas from Kealialalo vent have 452 wt% SiO2 and corresponding relatively low MgO,  5.2–6.6 wt%. But while these Moiwi and Kealialalo examples are different in these regards from shield lavas, they remain compositionally tholeiitic. At the other compositional extreme for tholeiitic postshield lavas is one reported picrite that has  15 vol% olivine. This mode gives the lava 22 wt% MgO, far more than any other lavas, and we therefore do not plot it in the MgO panels of Fig. 4. This sample, KAH-164 of Rudek et al. (1992), contains the cumulate ultramafic xenoliths noted earlier. In another kind of compositional contrast – that of alkalic vs. tholeiitic – one Kealialalo vent lava (KAH-115; Fodor et al., 1992, and Table 2) has 49 wt% SiO2, 9 wt% MgO, and is alkalic in terms of Na2O + K2O (Figs. 3B, 4). Moaula vent lavas are marginally alkalic, having  1.4–1.6 wt% K2O and straddling the line separating alkalic from tholeiitic (Fig. 3B). Lavas from Kamama vent and three others without identifiable source vents also plot in the alkalic field of Fig. 3B. Additionally, two Kamama lavas and sample H-1440 (Leeman et al., 1994) are hawaiites when evaluated in the LeBas et al. (1986, 1992) total alkali-silica diagram (i.e., 45 wt% Na2O + K2O and  50 wt% SiO2). Lavas from Moaula and the hawaiites of Kamama are distinguished from one another and from all other lavas in the CaO, Al2O3/CaO, K2O, P2O5, and several incompatible element panels of Figs. 4 and 7. On the other hand, Tanaka et al. (2008) refer to normative mineralogy to question whether alkalic lavas are present at all on Kahoolawe, and conclude that there are none. But we point out that one hawaiite, H-1440, has 2.4% normative nepheline (CIPW norm; Fe3 + 10% of total Fe). And while Kealialalo basalt KAH-115 (Fig. 2D) does not have normative nepheline, it has, along with

ARTICLE IN PRESS 108

R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

Fig. 7. MgO variation diagrams for trace elements of all Kahoolawe lavas analyzed. Both left- and right-hand panels have shield lava compositions encircled. Right hand panels have three vents and samples KAH-115 and H-1440 (hawaiite of Leeman et al., 1994) identified, as these are the most likely to have some element abundances different from those of shield and suggest alkalic compositions. ‘‘Post-erosional’’ refers to eruptives believed by Stearns (1940) and Macdonald (1940) to represent genuine post-erosional events, but are now known to be tholeiitic.

four Moaula samples, small amounts of normative olivine,  2–16%, which is consistent with their plotting in the alkalic field of Fig. 3B. They also have enrichments in K2O, P2O5, Zr, and Nb, which are all indicative of alkalic basalt. We accordingly consider postshield lavas from Kamama and H-1440 (hawaiites), some Moaula lavas, and one lava from Kealialalo (KAH-115) as alkalic, and their incompatible element abundances separate them as such from shield lava compositions in the MgO variation diagrams (Fig. 7). Lavas from these three vents are distinguished from one another in MgO expressions that compare, for example, their Zr and Nb abundances (Fig. 7). In some postshield lavas, the incompatible trace elements Ba, La, Y, and total REE are rather strongly elevated, and randomly so (Fig. 7). The REE and Y abundances among the Moaula vent lavas are good examples, where La(n) reaches 4600 in one sample (Fig. 8E). As described later, however, the high concentrations of these elements in some lavas were caused by weathering and secondary mineralization and are not magmatic.

4.4. ‘‘Post-erosional’’ eruptives Four analyses of two basaltic eruptives that Stearns (1940) and Macdonald (1940) believed to represent rejuvenated volcanism – a dike and a tuff cone – compositionally overlap shield lavas in

almost all respects. Their Na2O+K2O of 2.1–2.6 wt% and MgO of  6.4–7.3 wt% places them among the evolved and most common shield lava compositions (Figs. 3A, 4, 7). Similarly, REE patterns for the two samples fall near the middle of the observed range for shield lavas (Fig. 8C) with La(n) 35.

4.5. Coarse-grained rocks Of 12 xenoliths studied, only two gabbros were large enough for whole-rock analyses (Fodor et al., 1998). They are MgO-rich, 10 and 14 wt% – therefore within the MgO range for shield lavas – and different from the shield compositions by lower total alkalis, TiO2, Fe2O3, Zr, Nb, and Hf, and by higher Al2O3 (Figs. 3B, 4, 7). Both xenoliths have REE patterns subparallel to those for shield lavas, and at the low end of the range for shield REE patterns, which represents lavas with the highest MgO (Fig. 8C). The La(n) for the xenoliths is  18. Low incompatible element abundances, such as for Zr and Nb, and relatively high Al2O3 (Figs. 4, 7) suggest that these xenoliths had cumulate origins involving plagioclase and therefore do not represent magmatic liquid. But there are no corroborating positive Eu anomalies (Fig. 8C). For this report, we analyzed four samples of the gabbroic (diabasic) intrusion present at Kanapou Bay (Fig. 1) and the gabbro block of Macdonald (1940, sample N21), and we plot their

ARTICLE IN PRESS R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

109

Fig. 8. Chondrite-normalized rare-earth element (REE) patterns for all analyzed Kahoolawe lavas and coarse-grained rocks expressed in five panels. Two thick lines in panels B, C, and E are the lowest and highest shield lava REE patterns included for comparison of the range in shield REE abundances. Note log scale change in panel E. The majority of data are from Fodor et al. (1992), Fodor et al. (1998), Leeman et al. (1994), and Huang et al. (2005); REE data for gabbro intrusion are unpublished. ‘‘Posterosional’’ refers to eruptives believed by Stearns (1940) and Macdonald (1940) to represent genuine post-erosional events, but are now known to be tholeiitic.

compositions in Figs. 3A, 4, 7, and 8d. Notable is the compositional range represented, where the coarsest varieties have  6–8 wt% MgO, but a segregation vein in the intrusion and the gabbroic block have  2.9 and 4.2 wt% MgO, respectively. These are among the most evolved compositions observed for Kahoolawe in terms of MgO, and some of their incompatible-element abundances, such as TiO2, P2O5, K2O, REE, Ba, Zr, and Nb, are elevated over those in the most evolved shield and caldera-filling lavas. The CaO for the intrusion and gabbro block samples collectively trends lower with decreasing MgO, and Al2O3 remains ‘flat’ (Fig. 4). Accordingly, the Al2O3/CaO ratios are far higher than for any lavas. These characteristics can be explained by more clinopyroxene than plagioclase having separated from their precursor magmas during their differentiation, and/or by the accumulative effects of plagioclase in reservoir(s) for these gabbros. The overall REE pattern for each sample is parallel to those of shield lavas, and their chondrite-normalized values span almost the full range recorded for the shield (e.g., La(n)  22–50; Fig. 8D). Another REE characteristic of the intrusion is the small

complementary Eu anomalies between the most mafic representation and the highly evolved segregation vein within the intrusion – where the relative Eu abundances suggests a cumulate history for the most mafic.

5. Isotopic compositions 5.1. Shield and caldera-filling lavas Because the major and trace element compositions for calderafilling lavas overlap with those for shield lavas, we group together for plotting purposes the Sr, Nd, Pb, and Hf isotopic compositions available for both the shield and caldera-filling. Some fundamental isotopic characteristics for Kahoolawe lavas, first noted by West et al. (1987), are the extreme 87Sr/86Sr and 143Nd/144Nd relative to those ratios for most other Hawaiian islands. The high Sr and low Nd isotopic ratios resemble and partly overlap one of the Hawaiian mantle-source ‘endmembers’ for Sr–Nd ratios – the

ARTICLE IN PRESS 110

R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

Fig. 9. Isotopic ratios for 87Sr/86Sr and 143Nd/144Nd in Kahoolawe shield and caldera-filling lavas ( & ) and two postshield alkalic basalt (filled triangles; one is hawaiite H-1440) compared with those for other Hawaiian shields. Also shown are the general compositions for isotopic endmembers (large circles) identified as Kea and Koolau, believed to represent the two extremes in the Hawaiian mantle source (after Huang et al., 2005). Notable is that Kahoolawe lavas plot to the 87Sr/86Srhigh side of the general trends for Hawaiian shields. Kahoolawe data sources are West et al. (1987), Leeman et al. (1994), and Huang et al. (2005).

Fig. 11. 176Hf/177Hf and 144/Nd/143Nd isotopic ratios expressed as e values for Kahoolawe shield and caldera-filling lavas (&) and two postshield alkalic basalts (filled triangles; one is hawaiite H-1440) compared with other Hawaiian shields and the array of data formed by ocean island basalts (thick line). Kahoolawe compositions generally overlap the ocean island array (after Huang et al., 2005). Kahoolawe data sources are Leeman et al. (1994), Blichert-Toft et al. (1999), and Huang et al. (2005). eHf = ((176Hf/177Hf)sample/(176Hf/177Hf)CHUR 1 )  10,000; eNd = ((143Nd/144Nd)sample/(143Nd/144Nd)CHUR 1)  10,000; where CHUR values are 0.282772 and 0.512638, respectively.

Fig. 10. . 87Sr/86Sr ratios for Kahoolawe shield and caldera-filling lavas (&) compared with incompatible element ratios La/Nb and that relationship observed for other Hawaiian shields. Notable is that many Kahoolawe lavas plot off the general Hawaiian shield trend – implying a source composition with different components than those for some other shields (after Huang et al., 2005).

Makapuu-stage Koolau shield lavas documented by Frey et al. (1994) and Roden et al. (1994). The Koolau isotopic endmember has the highest 87Sr/86Sr for Hawaii,  0.7044, and lowest 143 Nd/144Nd,  0.51255. On the other hand, and pertinent isotopically, is that some Kahoolawe lavas have relatively higher 87 Sr/86Sr than Koolau (Makapuu) when evaluated at a given 143 Nd/144Nd on a Sr–Nd isotope trend line (Fig. 9). This offset to higher 87Sr/86Sr compared with the trend line formed by other Hawaiian shields is also observed in plots for some incompatible element ratios, such as La/Nb vs. 87Sr/86Sr (Fig. 10) (Huang et al., 2005). The 176Hf/177Hf ratios for Kahoolawe shield and caldera-filling lavas are expressed as eHf in Fig. 11. By and large, they overlap the fields created by most other Hawaiian Islands, and lie along the array for ocean island basalts. The Pb–Pb isotope diagrams for Kahoolawe demonstrate that the shield and caldera-filling lavas form 208Pb/204Pb–206Pb/204Pb and 207Pb/204Pb–206Pb/204Pb trends parallel to the collective trends formed by all Hawaiian shields (Fig. 12). However, at a

Fig. 12. Pb isotopic ratios for Kahoolawe shield and caldera-filling lavas (&) and two postshield alkalic basalts (filled triangles; one is hawaiite H-1440) compared with Pb isotope compositions for other Hawaiian shields and the northern hemisphere reference line. Most all lavas plot at the low 206Pb/204Pb end of the range for Hawaiian shields, and some are slightly enriched in 208Pb/204Pb (after Huang et al., 2005). Kahoolawe data sources are West et al. (1987), Leeman et al. (1994), Abouchami et al. (2005), and Huang et al. (2005).

given 206Pb/204Pb ratio, Kahoolawe 208Pb/204Pb ratios are slightly higher, conforming to that expected for Loa-trend Hawaiian shields (Abouchami et al., 2005). Compared with the range for

ARTICLE IN PRESS R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123 206 Pb/204Pb ratios created by all Hawaiian shields, Kahoolawe compositions are at the low end. The range is seemingly large, 17.85–18.35, but only three samples extend it beyond 18.15, essentially doubling the range expressed by the large majority of shield and caldera-filling lavas (Fig. 12). These high 206Pb/204Pb lavas also have the lowest 87Sr/86Sr ratios observed. Also notable about the isotopic ratios of Kahoolawe lavas is that significant proportions of the range in overall isotopic ratios observed (e.g., Figs. 9, 11, 12) can be found within single stratigraphic columns (Fig. 13). Seven samples from the Kanapou Bay section, for example, have 87Sr/86Sr ratios that extend from  0.7040 to 0.7043, and 206Pb/204Pb ratios from 17.86 to  18.06. These ranges occur among lavas representing about 200 m of emplacement thickness (Fig. 13). The isotopic compositions for Kahoolawe shield and caldera-filling lavas accordingly point to heterogeneity on a rather local scale – not only in view of variations within stratigraphic sections, but also when considering the small size of the island compared with most other Hawaiian shields sampled. As noted by Huang et al. (2009), however, small volume shields can have as much geochemical heterogeneity as large volume shields.

111

5.2. Postshield lavas Published isotopic data are available for only two postshield alkalic lavas (West et al., 1987; Leeman et al., 1994; Blichert-Toft et al., 1999; e.g., hawaiite H-1440), and are illustrated in the plots for the shield and caldera-filling lavas (Figs. 9, 11, 12). Perhaps the most notable isotopic characteristic is that 87Sr/86Sr ratios are offset from Koolau (Makapuu-stage) ratios by equal to slightly greater amounts than are the shield lavas when plotted against 143 Nd/144Nd ratios (Fig. 9). In a similar distinction, 207Pb/204Pb ratios are more radiogenic than most of those representing the shield (Fig. 12), but the 208Pb/204Pb and eHf expressions (Fig. 11) do not separate these postshield lavas from shield lavas.

6. Mineral and glass compositions 6.1. Olivine The olivine core-area Fo (forsterite) compositions observed among olivine phenocrysts in shield lavas are presented in

Fig. 13. Variations in isotopic compositions with elevation for three shield lava stratigraphic sections (locations in Fig. 1). No trends can be defined, and samples within a stratigraphic section can range widely, and, in general, the stratigraphic sections have overlapping compositions. To serve as reference to Kahoolawe isotopic heterogeneity, even though it is a small volcano, the range in 143Nd/144Nd ratios for all Hawaiian shields is shown (Huang et al., 2009).

ARTICLE IN PRESS 112

R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

Fig. 14. A. Olivine forsterite (Fo) compositions for phenocryst cores in all categories of Kahoolawe lavas, the gabbro intrusion, and a postshield picrite (Rudek et al., 1992). Three shield stratigraphic sections are identified (Fig. 1). Each symbol represents a grain within the category identified. B. FeO/MgO partitioning between olivine phenocryst cores and their host rocks. A KD of  0.3070.03 identifies olivine/liquid equilibrium (Roeder and Emslie, 1970) and only a few lavas meet that criterion, the others implying disequilibrium between olivine cores and their host lavas. Symbols used are from panel A. For explanation of ‘‘post-erosional’’, see captions for Figs. 3, 4, 7, and 8.

Fig. 14B and some representative compositions are in Table 1. The range is from a high Fo89.6 core to a low Fo79. Olivine phenocryst cores in caldera-filling lavas range from an identical high of Fo89.6 to slightly more evolved, Fo78. The Fo-rich cores observed in postshield lavas largely overlap those in shield lavas, as  Fo87.8 extending to Fo75, except in a hawaiite where Fo64 is present. Olivine in the ‘‘post-erosional’’ eruptives overlap olivine in shield lavas with Fo81 86, and the gabbro intrusion has Fo86. A postshield picrite (Rudek et al., 1992) has olivine phenocrysts with cores of Fo82 86. Phenocryst core compositions are evaluated as to whether they are in equilibrium with their host rocks by applying a FeO/ of 0.30 ( 70.03) partition coefficient (Fig. 14B). MgO Kolivine/liquid D Only some of the olivine cores plots within the field designating crystal-liquid equilibrium, indicating that many samples have phenocrysts that crystallized in liquids other than represented by their host rocks. The straightforward explanation is that olivine phenocrysts were mobile in their reservoirs, both adding to and subtracting from magmas during mixing and crystallization processes. Notable is that some olivine cores suggest equilibrium with liquids having  15 wt% MgO. Olivine in postshield lavas and ‘‘post-erosional’’ eruptives show approximately the same distributions as shield lavas with regard to Fo cores and olivine-liquid equilibrium relationships. Olivine in the ultramafic xenoliths have compositions Fo82 86. (Rudek et al., 1992). This range overlaps with the primitive olivine cores in shield lavas, but does not match the Fo89 present in some of those lavas. Kahoolawe ultramafic xenoliths appear, then, to represent cumulates from liquids that had undergone some differentiation from a parental (e.g., o9 wt% MgO) magma.

6.2. Pyroxene Average clinopyroxene and orthopyroxene compositions are illustrated in Fig. 15, and some representative compositions are in Table 1. Shield and caldera-filling lavas (Fig. 15A) have clinopyroxene compositions characteristic of Hawaiian tholeiite magmas, namely Wo37 39 (e.g., Fodor et al., 1975; Fodor and Moore, 1994; Weinstein et al., 2004; McCarter et al., 2006), and with Mg#s 75–84. Coexisting orthopyroxene has Fs8 12Wo4.5. One exception is relatively Fe-enriched

clinopyroxene in a diabasic caldera-filling lava, Fs20Wo35 and Mg# 69, and coexisting orthopyroxene with Fs22Wo9. Clinopyroxene in postshield lavas overlap those in shield lavas as well as some having slightly higher Wo,  40 mol%, to distinguish them as occurring in slightly more alkalic lavas (Fig. 15A). Pyroxenes in the ‘‘post-erosional’’ eruptives generally overlap those in shield lavas (Fig. 15A). While compositions of both clinopyroxene and orthopyroxene in the gabbro intrusion and gabbro block have some overlap with those in shield lavas, most are more evolved to where Fs values are commonly  20–25 mol% (Fig. 15B). The wide range in pyroxene compositions reflects the protracted crystallization in a slowly cooling confined magma reservoir. Fig. 15C shows the compositions of pyroxenes in the Kahoolawe xenoliths. The ultramafic xenoliths have slightly more primitive pyroxene ( Fs9Wo38) than shield lavas – the reverse relationship observed for olivine, possibly due to re-equilibrium between clinopyroxene and olivine in their cumulate environment. But the gabbro xenoliths have pyroxene compositions overlapping those in shield lavas.

6.3. Plagioclase Plagioclase endmembers for average compositions of individual grains are illustrated in Fig. 16A, and some full compositions are in Table 1. Shield lavas have bytownite–labradorite compositions, and caldera-filling lavas are slightly more sodic, An64 68. Postshield lava plagioclase is mainly bytownite and labradorite, but also andesine-oligoclase in the moderately alkalic Moaula vent lavas. The ‘‘post-erosional’’ eruptive samples have bytownite (Fig. 16A). All of these compositions are expected for Hawaiian tholeiitic and alkalic lavas (e.g., Keil et al., 1972; Fodor and Moore, 1994; Weinstein et al., 2004; McCarter et al., 2006). Plagioclase in the gabbro intrusion and gabbro block are plotted as point analyses acquired on many grains (Fig. 16B). Compositions are labradorite and andesine, An70 45, more evolved than plagioclase in shield lavas. Among ultramafic xenoliths, the sparse plagioclase present is bytownite, An75 80, and in the gabbro xenoliths, plagioclase is labradorite, An58 68 (Rudek et al., 1992; Fodor et al., 1993). The

ARTICLE IN PRESS R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

Table 1 Representative compositions of olivine, clinopyroxene, orthopyroxene, and plagioclase in shield, caldera-filling, postshield, and ‘‘post-erosional’’ lavasa of Kahoolawe Island, Hawaii.

exception is the andesine gabbro xenoliths in Moaula vent lavas, which have An35 50 (Fodor et al., 1998). 6.4. Rhyolite glass

Olivine Shield 182

Caldera

Postshield

117

115

158

Sample

18

SiO2 FeO MnO MgO CaO NiO Total

39.7 13.4 0.15 46.2 0.22 0.50 100.17

38.0 19.2 0.25 41.4 0.24 0.29 99.38

38.7 16.0 0.19 44.1 0.20 0.46 99.65

37.4 21.5 0.31 40.4 0.21 0.22 100.04

39.9 12.3 0.24 47.5 0.25 0.44 100.63

86.0

79.4

83.1

77.0

87.3

Fo

113

Some vesicles contained within the open-textured gabbro xenoliths in a postshield lava contain glassy linings (Fig. 17A) with  76 wt% SiO2 (Fodor et al., 1993) and overall rhyolite composition. Additionally, some samples of the gabbro intrusion have rhyolite glass,  78 wt% SiO2 (Fodor et al., 1993), in spaces interstitial to grains that otherwise create an intergranular diabasic texture (Fig. 17B, C). Beyond these high-SiO2 glass occurrences in gabbro, rhyolite glass lines or fills vesicles in some postshield lavas, creating what Anderson et al. (1984) call segregation vesicles (Fig. 17D).

Pyroxene Shield Sample

178

Caldera

Postshield

‘‘Post-erosional’’

178

117

192

156

SiO2 TiO2 Al2O3 Cr2O3 FeO MnO MgO CaO Na2O Total

51.9 0.65 2.2 0.37 8.9 0.18 17.7 17.9 0.16 99.96

54.6 0.34 1.1 0.23 14.6 0.18 27.2 2.3 0.06 100.61

52.6 0.56 2.6 0.68 7.6 0.27 18.8 16.8 0.18 100.09

52.5 0.70 2.8 0.55 6.3 0.19 17.8 19.3 0.29 100.41

51.8 0.62 2.7 0.60 7.0 0.20 17.8 18.1 0.21 99.03

Mg# Fs Wo

78.0 14.0 36.2

76.9 22.1 4.5

81.5 12.1 34.4

83.4 10.0 39.4

81.9 11.3 37.5

Shield

Caldera

Postshield

178

117

192

7. Magmatic differentiation: tholeiitic The wide range in MgO and La(n) observed for Kahoolawe shield and caldera-filling lavas reflects the crystallization of parental magmas and subtraction of the primary Mg minerals,

Plagioclase

Sample

‘‘Post-erosional’’ 111

32

SiO2 Al2O3 FeO CaO Na2O K2O Total

51.9 29.2 0.72 13.7 3.4 0.16 99.08

50.4 30.7 0.54 14.3 3.3 0.10 99.34

48.8 31.8 0.67 15.5 2.5 0.11 99.38

51.7 30.5 0.85 13.7 3.1 0.17 100.02

51.8 30.3 0.63 13.7 3.4 0.15 99.98

An Or

68.4 0.9

70.2 0.6

76.9 0.6

70.2 1.0

68.4 0.9

a Sample numbers are from Fodor et al. (1992), but presented here without KAH prefix. All analyses by electron microprobe; each column is an average of 5–10 points/grain.

Fig. 16. Plagioclase compositions expressed as molecular An, Ab, and Or endmembers for all lava categories and gabbro. Each symbol for lavas (panel A) represents an average composition of a grain; each symbol for gabbros (panel B) represents a single point-analysis. For explanation of ‘‘post-erosional’’, see captions for Figs. 3, 4, 7, and 8.

Fig. 15. A quadrilateral for clinopyroxene and orthopyroxene in all categories of Kahoolawe lava, gabbros, and xenoliths. Each symbol for lavas and xenoliths represents the average composition of a grain, determined after 5–10 microprobe point analyses. Each symbol for gabbros (panel B) is for a single point-analysis. Gabbro block refers to Macdonald’s (1940) sample N21. For explanation of ‘‘post-erosional’’, see captions for Figs. 3, 4, 7, and 8.

ARTICLE IN PRESS 114

R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

Fig. 17. Photomicrographs of rhyolite glass occurrences in Kahoolawe lavas and gabbro. Scale bars= 1 mm: (A) Vesicle lining (top, center) in open-textured gabbro, postshield lava KAH-135 (Fodor et al., 1993), with microlites of plagioclase, clinopyroxene, Fe–Ti oxides (plane-polarized light). (B) Interstitial glass in gabbro intrusion of Kanapou Bay (arrow; plane-polarized light). (C) Same as B, but crossed-nicols. (D) Vesicle-filling in postshield lava KAH-186 with microlites of plagioclase, clinopyroxene, and Fe–Ti oxides (plane-polarized light).

olivine, orthopyroxene, and clinopyroxene to chemically differentiate the tholeiitic magma system. Moreover, the wide range in Al2O3/CaO ratios within the MgO range of  6–6.5 wt% (Fig. 4) suggests that crystallization of plagioclase in various proportions with clinopyroxene also influenced the compositional evolution of Kahoolawe shield. Accordingly, magma differentiation was one of the important igneous processes occurring during volcano development and warrants quantitative examination. Processes presented in this section first address the tholeiitic magmas represented by shield lavas. But because many postshield lavas are also tholeiitic as based on Na2O+K2O abundances (Fig. 3B) – namely, those lavas from the vents Makika, Kealialuna, Kolekole, Moiwi, and some from Kealialalo (Fig. 1) – we additionally evaluate differentiation for one (Moiwi) postshield lava. All postshield tholeiitic lavas apparently erupted near the conclusion of the shield-building stage, during time of waning volcanism and could have their origins in shield magmas. Finally, we also evaluate differentiation in the gabbro intrusion and its differentiation to create rhyolite glass. In the differentiation models we present for shield lavas, parent magmas of  9 wt% MgO are used because that value is the highest observed among shield lavas likely to have magmatic compositions. While some shield lavas have 410 wt% MgO (Fig. 4), they may have experienced olivine accumulation during their reservoir times and therefore are not appropriate to model as parental. The Fo89.7 olivine core in at least one shield lava (Fig. 14A) is consistent with crystallization from a magma having an FeO/MgO ratio of  0.68, or MgO 15 wt%. We did not, however, observe such high MgO lavas that could be unequivocally interpreted as magmatic and not olivine-cumulative. The models we use as candidates for parent–daughter relationships are based only on major element compositions, and, in the case of shield lavas, we do not consider their stratigraphic positions within their respective sections. Also, while the isotopic compositions of the shield lavas may be slightly different from one another (e.g., Fig. 13) or are

undetermined, the intent here is to evaluate the general qualities of chemical differentiation within stratigraphic sections, at postshield vents, and within a closed system intrusion, and not to identify exact parent–daughter lavas.

7.1. Shield lavas Among the shield lavas comprising the stratigraphic sections sampled at Kanapou Bay and the southeastern shore (Fig. 1), there are ranges in wt% MgO large enough to apply least squares mass balancing for exploring fractional crystallization relationships. For example, by selecting KAH-18 (Fodor et al., 1992) of Kanapou Bay,  9 wt% MgO and 51.2 wt% SiO2, as a parent for the most evolved shield lava KAH-28,  5.5 wt% MgO and 48.9 wt% SiO2, about 32% crystallization is required (Table 2). But reasonable results are achieved only when orthopyroxene, not olivine, and a rather evolved plagioclase, An51, are in the fractionating assemblage. On the other hand, because of the large decrease in MgO from 9 to 5.5 wt% in this parent–daughter model, an intermediatecomposition plagioclase may be the best representation for the average mol% An fractionated. Also, orthopyroxene and a plagioclase with relatively high SiO2 are required to produce a daughter with o50 wt% SiO2 from the selected parent composition (SiO2 451 wt%). For comparison, a less evolved daughter composition, KAH-13, with 6.1 wt% MgO and SiO2 contents comparable to the parent, requires only  16% removal of plagioclase (An76) and both orthopyroxene plus olivine to yield a fractionation relationship with parent KAH-18 (Table 2). To a general extent, these least-squares crystallization models are confirmed by MELTS software (Ghiorso and Sack, 1995). MELTS thermodynamics-based calculations corroborate the extent of crystallization to produce the evolved MgO values, and also demonstrate that pressure influences the proportions of orthopyroxene to olivine that crystallize. For example, MELTS crystallization of KAH-18 composition (Kanapou Bay section; K2O

ARTICLE IN PRESS R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

115

Table 2 Least-squares linear regression (mass balancing) models for crystallizing parent Kahoolawe shield basaltsa KAH-18 and KAH-182 to yield shield and postshield daughter basalts.a Kanapou Bay

SiO2 TiO2 Al2O3 FeO MnO MgO CaO Na2O K2O

Southeastern shore

Parent

Daughters

KAH-18 shield

KAH-28 shield

KAH-13 shield

51.20 2.25 12.83 10.98 0.16 9.06 9.10 2.21 0.61

48.90 3.18 14.30 13.28 0.21 5.52 9.99 2.59 0.28

51.80 2.62 13.60 11.61 0.18 6.14 10.00 2.15 0.40

Parent

Daughter

KAH-39 Moiwi postshield

KAH-182 shield

KAH-179 shield

51.79 2.75 14.71 10.99 0.16 5.60 9.58 3.11 0.80

49.6 2.30 13.60 11.34 0.18 9.09 9.79 2.24 0.18

51.70 2.29 14.0 10.89 0.18 6.82 10.20 2.18 0.26

These mineralsb removed (in %) from parent magmas (KAH-18; KAH-182) to yield daughter magmas 7.0(Fo79.6) ol – 4.7(Fo83.5) opx(Mg#82) 14.4 6.4 – cpx 6.7(Mg#82) – 6.9(Mg#84) 4.7(An76) – pl 11.1(An51) – – – mt(Usp70) (Residual liquid; daughter) 67.8 84.2 86.1 Sr2 0.27 0.32 0.59

7.3(Fo83.5) – 2.2(Mg#79) 7.3(An76) 1.8 81.4 0.15

These crystallizing minerals and residual liquids (in %), and conditions based on MELTSc (fO2 = FMQ; 0.1 wt% H2O) Crystallization pressure 1.5 kbar 0.15 kbar – 6.1 ol(Fo81.7) opx 18.4(Mg#82) 4.7(Mg#81) pl 14.0(An63) 6.1(An65) (Residual liquid) 67.6 83.1 Derivation at T (1C) 1150 1150

Sr2 = differences between calculated parent magma composition by linear regression and actual parent magma compositions, examined for each oxide expression (e.g., SiO2, TiO2, etc.), squared and all summed (closer to zero, the better the ‘fit’; e.g., Bryan et al. (1969). a b c

See Fodor et al. (1992) and Huang et al. (2005). Based on compositions in shield lavas, analyzed by microprobe (e.g., Rybek, 1995; Fodor, unpublished). Ghiorso and Sack (1995).

adjusted for weathering to equal P2O5) under both 1.5 and 0.15 kbar pressures at FMQ fO2 (0.10% H2O) conditions produces 5.67 and 6.19 wt% MgO derivative liquids (i.e., the KAH-28 and KAH-13 equivalent daughter compositions) after  32% and 17% crystallization, respectively – similar to the mass balancing results (Table 2) – but orthopyroxene (  18%) and An63 plagioclase, and no olivine, are in the higher pressure crystallization assemblage;  6% olivine and 5% orthopyroxene form the 0.15 kbar assemblage (Table 2). Similar least-squares mass balancing results are reached for lavas in the southeastern shoreline stratigraphic section where a 9.1 wt% MgO parent, KAH-182 (Huang et al., 2005), yields a daughter KAH-179 with  6.8 wt% MgO after  19% crystallization (Table 2). But in this case, olivine and no orthopyroxene fractionates as the principle Mg mineral. All in all, then, least squares modeling demonstrates that lavas making up particular stratigraphic sections are generally related by fractional crystallization of up to  30% separation of the principle basalt phases from parent magmas having 9 wt% MgO. Because no compositions of o5 wt% MgO were observed for shield lavas as they are for the Kanapou Bay gabbro intrusion, crystallization percentages are less than what occurs in a closed-system hypabyssal reservoir where, as noted below, o5 wt% MgO was produced.

7.2. Gabbro intrusion Discovery of the sill-like tholeiitic gabbroic intrusion along Kanapou Bay shoreline provides an example for tholeiitic differentiation of a thick lens of magma. This is manifested by

the diabasic rock having an MgO range of 7.9 wt% as the highest observed, to 2.9 wt% in a vesicular, partly glassy segregation vein,  5 cm thick (Fig. 2J, L). This MgO decrease is attended by La doubling from  6.8 to 13.1 ppm, and the complementary Eu anomalies in the REE patterns, noted earlier (Fig. 8D), can be interpreted as magma differentiation processes involving accumulation and subtraction of plagioclase. A Sr/Nd ratio of  25 for the MgO-rich samples with the positive Eu anomaly is twice those of the other samples ( 10–12), and supports accumulation of plagioclase. These characteristics resemble those recorded in magma solidification-zone samples of other Hawaiian closed-systems, such as lava lakes (e.g., McCarter et al., 2006) and sub-shield reservoirs such as Uwekahuna laccolith (e.g., Murata and Richter, 1961; Fodor and Galar, 1997). The Uwekahuna laccolith closely resembles the Kahoolawe intrusion in physical emplacement, but its composition is more mafic and there are no examples of segregation as evolved as that in the Kahoolawe intrusion. The mineral compositions of the Kanapou Bay intrusion reflect the compositional whole-rock differentiation represented among the samples (Figs. 14A, 15B, 16B). The most MgO-rich sample has Fo86 olivine and An72 62 plagioclase, whereas the segregation vein lacks olivine and has plagioclase zoned to andesine composition. These two samples of compositional extremes have clinopyroxene  Fs15 and Fs25 35, respectively. To evaluate the differentiation history that produced the highly evolved segregation vein within in the gabbroic hypabyssal emplacement on Kahoolawe, we refer to a crystallization model for a similar Hawaiian closed-system. It is for a  9 wt% MgO

ARTICLE IN PRESS 116

R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

Mauna Loa lava-lake parent magma that yielded melts having o4.5 wt% MgO after  67% crystallization of a plagioclase (An63)dominated assemblage that includes clinopyroxene4olivine4 Fe–Ti magnetite (e.g., McCarter et al., 2006) – and where the most evolved rocks are also segregation veins, but in a lava lake rather than a sill. To infer, then, the Kahoolawe gabbro body underwent at least 70% crystallization to yield the segregation vein. Finally, Macdonald’s (1940) gabbro block (N21) has mineral and bulk compositions (Figs. 4, 7, 15B, 16B), including REE abundances (Fig. 8D), that resemble those of the segregation vein in the Kanapou Bay gabbro intrusion. Whether or not this gabbro block is a portion of that intrusive body is uncertain – especially in view of its intergranular texture (Fig. 2K) attending the evolved composition (  4.4 wt% MgO) and their separate collection sites (Fig. 1). But gabbro N21 is nonetheless among the most compositionally evolved rocks observed for Hawaii and could represent evolved liquid in an intrusive environment allowed to cool at a rate appropriate for intergranular texture.

7.3. Postshield lavas Notable about postshield tholeiitic lavas when compared with shield lavas is their greater differentiation as expressed not only by several examples having MgO o6 wt%, but by higher P2O5, Al2O3/CaO, and K2O at o6.5 wt% MgO, and lower CaO, Fe2O3, and to some extent Al2O3 (Fig. 4). Removal of plagioclase and relatively Fe-rich clinopyroxene (cpx4pl) can generally explain these trends, where extended differentiation could have been in response to diminishing lava production and replenishment after the shield-building stage was completed. The evolved Moiwi vent lavas ( 5.6 wt% MgO) are appropriate postshield representatives to examine for lineage with shield magmas because their Na2O + K2O is higher than in shield lavas (Fig. 3B) but they remain tholeiitic. Indeed, least squares calculations show that Moiwi compositions can generally be produced after 14% crystallization of largely olivine and clinopyroxene from KAH-18 (Table 2). Alternatives to the least squares model are that Moiwi lavas, which have K2O and some incompatible element abundances (e.g., La, Th, Nb) higher than shield lavas (Fig. 7), represent smaller degrees of partial melting than do shield lavas, or they had a source different in some incompatible-element abundances from those in the shield source. But these alternatives are not compelling as Moiwi lavas compositional differences from shield lavas are almost insignificant. Preliminary isotopic data for Moiwi vent lavas provided by Shichun Huang (personal comm.) show 87Sr/86Sr and 143Nd/144Nd ratios to plot at the least enriched end – low 87Sr/86Sr and high 143 Nd/144Nd – of the range for shield lavas (e.g., 0.7037), and 206 Pb/204Pb ratios (  18.27) that place Moiwi lavas at the most radiogenic end of the shield range. The isotope ratios do not rule out a shield source for Moiwi postshield lavas because of the wide variation in shield isotope ratios (e.g., Fig. 9), and therefore their relationship to shield lavas is better done by evaluating ratios of incompatible elements in both postshield and shield lavas. Owing to alteration, only selected incompatible elements are reliable for evaluating Kahoolawe compositions. Eliminated are REE, Y, Rb, Ti, and Ba, all of which were modified in many samples by secondary mineralization (see section 9) or weathering – leaving those we believe to have resisted alterations and remaining at magmatic values to be Zr, P, Th, Nb, and probably Sr. Ratios among these elements are illustrated in Fig. 18, where they are plotted against MgO for assessing any possible influence from differentiation.

In all panels of Fig. 18, Moiwi lavas appear to be extensions of evolved shield and/or caldera-filling lava compositions. Their relationship to shield/caldera magmas include slight enrichment in Zr (Zr/TiO2) compared with the o6 wt% MgO shield and caldera-filling lavas, probably owing to prior magnetite crystallization. Their Nb/P2O5 ratios are consistent only with shield lavas, not caldera-filling, as they do not overlap the downward trend of caldera-filling lavas, perhaps due to apatite crystallization in parental magmas. 7.4. Small-scale: rhyolitic melts The rhyolite glass in the open-textured gabbro xenoliths and in the interstitial areas of the gabbro intrusion samples (Fig. 2J) represents the extreme in melt fractionation from basaltic magma, albeit on micro-scales. Such local occurrences of SiO2rich glasses have been described by Anderson et al. (1984) as ‘pressed’ by gas effervescence into relatively low pressure environments during crystallization of matrices. In an attempt to quantify the fractionation required to produce rhyolitic melt from basaltic magma, Fodor et al. (1993) used least squares mass balancing to model SiO2-rich melt segregation from the groundmass composition of a Kahoolawe lava. Their results show that  88% crystallization of clinopyroxene, plagioclase, and Fe–Ti oxides of compositions like those in the lava groundmass is required to yield a residual rhyolite melt of composition resembling glass in vesicles of the gabbro xenoliths. While extremely small in rhyolite melt volume, this level of differentiation to high-SiO2 liquids within the Hawaiian system has significance as a first step toward producing silicic melts on a lava scale, which is present in the Waianae range as rhyodacite (Bauer et al., 1973) and on Kilauea as dacite (Marsh et al., 2008). 7.5. Gabbro and ultramafic xenoliths formation The rare occurrences of 1–2 cm-sized fragments of gabbroic and ultramafic rock (Fig. 2H, I) represent magma differentiation in solidification zones of tholeiitic magma reservoirs. In some examples, particularly the ultramafic xenoliths, poikilitic textures corroborate this origin whereby early-formed cumulus olivine (  Fo86) gathering along the margins of a reservoir were enveloped by intercumulus liquids that crystallized to orthopyroxene and plagioclase. The largely intergranular gabbro xenoliths – some open-textured (Fig. 2I) – and the low incompatible element abundances (Fig. 7) are also consistent with solidification zones. A high Sr/Nd of  27 for one of the MgO-rich orthopyroxene-gabbro xenoliths (Fodor et al., 1998) is consistent with a cumulate origin, although no Eu anomaly is evident (Fig. 8C). The entrainment of xenoliths by their host lavas likely involved fragmentation of solidification zones along reservoir walls by magma scouring during eruptions (e.g., Rudek et al., 1992). In each example of xenoliths, the host is postshield lava, so we infer that the xenoliths largely represent shield building. Mineral compositions are consistent with tholeiitic parentage, with possible exception of the andesine-bearing xenoliths in Moaula lavas (Fodor et al., 1998).

8. Magmatic processes: alkalic 8.1. Magma differentiation Kahoolawe alkalic basalts are rare. What is known about their occurrence is that Kealialalo vent erupted alkalic lavas (KAH-115) along with tholeiitic, Kamama vent produced hawaiites (and

ARTICLE IN PRESS R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

117

Fig. 18. Trace element ratios for selected incompatible elements (weathering resistant) plotted against differentiation index MgO to help evaluate source composition relationships among Kahoolawe shield (+ ), caldera-filling (D), and postshield lavas (&). Lavas from three alkalic vents, alkalic KAH-115, and hawaiite H-1440 (Leeman et al., 1994) are identified for comparison to shield and caldera-filling lavas (encircled in each panel). Interpretations of the plots are in the text.

another undetermined source erupted hawaiite H-1440), and Moaula vent erupted marginally alkalic lavas (Fig. 3B). The 9 wt% MgO of KAH-115 plotted against its trace element abundances and ratios in Figs. 7 and 18 justify modeling it as the parent composition for hawaiites and for Moaula alkalic basalts. Least-squares mass balancing (Table 3) calculates that from  28% to 34% crystallization of olivine, clinopyroxene, plagioclase, and titaniferrous magnetite is required for these relationships. This crystallization assemblage can explain why some trace element ratios in Fig. 18 differ between modeled parent KAH-115 and the modeled daughter hawaiites and Moaula lavas. For example, Zr/Sr ratio differences can be interpreted as KAH-115 yielding both hawaiite and Moaula lavas by plagioclase subtracplag/liq for tion to increase Zr/Sr in the daughter derivatives (e.g., KD Sr  2). In other panels of Fig. 18, the KAH-115 trace element

ratios can be related to those for hawaiites and Moaula basalts by combinations of fractionating plagioclase and Fe–Ti oxide, which can modify both Ti and Nb values during fractional crystallization for Nb can be from 1 to 1.8; Neilson and Beard, 2000). (e.g., Kmt/liq D Additionally, apatite crystallization could affect Nb/P2O5 between parent and daughter magmas. The pressures at which fractional crystallization yields hawaiite daughter magmas are worthy to investigate. The reason is that Frey et al. (1990) point out that the hawaiite lavas capping Mauna Kea shield resulted from differentiation at moderate to high pressures as opposed to occurring in shallow level processes expected for shield building. It probably occurred at the crustmantle boundary (Frey et al., 1990) where clinopyroxene is the principle segregating mineral and where the relatively deep crystallization environment was in response to diminishing magma production at the end of shield building magmatism.

ARTICLE IN PRESS 118

R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

Table 3 Least-squares linear regression (mass balancing) models for crystallizing parent Kahoolawe alkalic basalta KAH-115 to yield alkalic postshield daughter basaltic lavas.a Alkalic lavas

SiO2 TiO2 Al2O3 FeO MnO MgO CaO Na2O K2O

Parent

Daughters

KAH-115 bKealialalo

KAH-128 Kamama

H-1440 hawaiite

KAH-108 Moaula

48.88 2.55 14.16 11.27 0.15 9.01 8.87 3.02 1.01

50.15 3.04 16.64 10.53 0.1 3.85 7.86 4.05 1.68

49.43 3.02 16.32 10.62 0.17 4.62 6.70 4.65 1.83

51.25 2.59 15.03 11.54 0.16 6.02 7.40 3.23 1.40

7.0(Fo80) 17.3(Mg#78) 8.6 1.6 65.5 0.66

6.7(Fo82) 11.3(Mg#84) 11.9 2.5 67.7 0.45

These minerals removed (in %) from parent KAH-115 to yield daughter magmas ol 8.6(Fo83.5) cpx 12.7(Mg#82) 5.4 pl (An76) 1.7 mt (Usp70) (residual liquid; daughter) 71.6 Sr2 0.37

These crystallizing minerals and residual liquids (in %), and conditions based on MELTSc crystallizing KAH-115 (fO2 = FMQ; 0.8 wt% H2O) crystallization pressure 8 kbar cpx (Mg#84) 20.2 opx (Mg#78) 11.3 (Residual liquid) 68.5 Derivation at T (1C) 1185

Sr2 = differences between calculated parent magma composition by linear regression and actual parent magma compositions, examined for each oxide expression (e.g., SiO2, TiO2, etc.), squared and all summed (closer to zero, the better the ‘fit’; e.g., Bryan et al., 1969). a b c

Fodor et al. (1992) for KAH samples; Leeman et al. (1994) for H-1440. Source vent names (see Fig. 1); source vent unknown for H-1440 hawaiite. Ghiorso and Sack (1995).

A similar high-P origin by clinopyroxene fractionation is also reported for some rare alkalic lavas from Mauna Loa (Wanless et al., 2006). The same relatively high pressure origin may hold for Kahoolawe hawaiite, and is suggested by the results of MELTS modeling to crystallize KAH-115 composition at 8 kbar P (FMQ; 0.8 wt% H2O). It yields a 2% ne-normative hawaiite composition that has Na2O+K2O 5.7 wt% and MgO wt 3.7% after  31% pyroxene crystallization (Table 3). When Kahoolawe hawaiite compositions are plotted in the normative Ne–Ol–Di system that evaluates pressures for origins (Sack et al., 1987), only one, H-1440, plots comparable to Mauna Kea hawaiites (Fig. 19). The other two are not ne-normative and plot between the areas for high-P and the field for 1-atmosphere experimental data. The sparse hawaiite lavas of Kahoolawe, then, apparently originated over a range of pressures, with only one example exemplifying a deep crustal origin in the Ne–Ol–Di plot (Fig. 19). Also in Fig. 19 is the melt composition calculated by MELTS crystallizing KAH-115 at 8 kbar (Table 3), and one example of a Mauna Loa alkalic basalt (Wanless et al., 2006). For Moaula lavas, which we demonstrate by mass balancing (Table 3) and trace element ratios to be consistent with fractionation from KAH-115, the wide range in the trace element ratios among five samples (Fig. 18) makes this relationship less clear as one involving only fractionation. Accordingly, we describe below the likelihood that an additional process of magma mixing was involved.

8.2. Magma mixing Moaula vent lavas have K2O 1.1–1.6 wt%, P2O5 0.4–0.5 wt%, and Nb  30 ppm – all comparatively high abundances in view of

Fig. 19. A Ne–Di–Ol ternary plot, projected from plagioclase and based on experimental work of Sack et al. (1987), to illustrate how some Kahoolawe lavas, particularly hawaiite H-1440 (Leeman et al., 1994), probably represent relatively high-pressures magmas, as do hawaiites of Mauna Kea (Laupahoehoe; Frey et al., 1990) and alkalic basalts of Mauna Loa (J2-15-13; Wanless et al., 2006). KAH-115 is a possible parent magma for hawaiites of Kamama vent (KAH-128; 130) and for H-1440. The star is the MELTS-calculated liquid for KAH-115 at 8 kbar P (details in text and Table 3).

their  6–6.8 wt% MgO when compared with shield lavas in that MgO range (Figs. 4, 7). As alkalic lavas, however, they have rather high SiO2,  51–52 wt% (Fig. 3B). And petrographically, these lavas have subrounded andesine and olivine phenocrysts (Fig. 2E), and the olivine core compositions vary widely,  Fo86-71, (Fig. 14A).

ARTICLE IN PRESS R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

One reconciliation presented for these various Moaula lava characteristics vis-a-vis shield lavas was by Fodor et al. (1998) who proposed that Kahoolawe tholeiitic magmas evolved in shallow reservoirs with repeated replenishments by primitive tholeiitic magmas to create a continuum of hybrid compositions. Such hybrid-magmas would precipitate phenocrysts with wide and evolved compositional ranges and would acquire resorbed rims by interactions with hot replenishments. Also, the transition from Kahoolawe shield building to postshield activity likely included reduced production rates for replenishment magmatism (e.g., Clague, 1987), and hybrid-magmas were able to evolve to o7 wt% MgO and SiO2 451 wt%. That hybridization model can explain the mineral compositions and petrographic features in Moaula lavas, but an additional component is required to place Moaula lavas into its marginally alkalic classification. This can be accomplished by introducing Kamama vent hawaiite into the reservoir(s) of stored tholeiitic hybrids (Fodor et al., 1998). However, as stated in the previous section, Moaula lavas can instead be interpreted as derivatives from alkalic parent KAH-115 (Table 3) whereby fractionation from and magma-mixing replenishments by parent alkalic magma account for the range of their bulk compositions observed – and introducing hawaiite is therefore not necessary to invoke as an explanation for the alkalic character of Moaula lavas (Fig. 3B).

9. Secondary mineralizations And assortment of secondary minerals has been observed in the groundmasses of some Kahoolawe lavas. The Ba–Mn oxide mineral hollandite is in a Kealealalo postshield lava, and is a secondary product related to tropical weathering and consequent mobilization of labile elements (Fodor et al., 1994). The hollandite contains nearly 10 wt% BaO,  1.1 wt% CeO2, and small amounts of La, Nd, Y, K, Na, P, Cl, and Cu. Another Mn-oxide, vernadite, and barite are also present in the same lava as secondary minerals. The vernadite hosts small amounts of REE and Ba. The postshield lavas of Moaula vent manifest secondary mineralization as the phosphate rhabdophane (Fodor et al., 1989), which has La2O3 6.8 wt%, Ce2O3  16 wt%, and Y2O3 7.6 wt%. Most notable about this weathering product is that it creates high whole-rock concentrations of LREE, such that both La(n) and Ce(n) can be 400–600 (Fig. 8E), and Y up to three times magmatic abundances (Fig. 7). Accordingly, the REE, Y abundances in Moaula lavas are highly variable, and can also be in some Moaula lavas at or close to magmatic values (Fig. 7). Besides Moaula lavas, there are other postshield examples that likely contain rhabdophane as suggested by anomalously high La and/or Y abundances observed in some postshield lavas, such as those at Moiwi vent (Fig. 7) (e.g., West et al., 1987; Fodor et al., 1992). Beyond Kahoolowe, a similar occurrence of secondary mineralization is suggested for some Koolau lavas, as Frey et al. (1994) reported anomalously high REE.

10. Discussion 10.1. Attributes of the shield The preceding descriptive and quantitative presentations about Kahoolawe – namely, the volcanic stratigraphy, petrography, and compositions of whole-rock major- and trace-elements, isotopes, and minerals – highlight the overall development, characterization, and igneous processes special to this 1–1.4 Ma volcano. Among highlights are the various ways by which Kahoolawe manifests shield magma differentiation: (i) olivine control from

119

parental tholeiitic magma at least 9 wt% MgO to create both MgO-rich ( 412 wt%) and evolved liquids 6.5–7 wt% MgO; (ii) additional fractionation of up to 30% of largely clinopyroxene, plagioclase, and orthopyroxene to produce liquids o6 wt%; and (iii) extreme fractionation leading to liquids having o5 wt% MgO within an intrusive body, and to rhyolitic melts segregated into vesicles. Additionally, there are several shield lava olivine core Fo values (Fig. 14) that identify that primary magmas up to  15 wt% MgO were also part of the shield building. Finally, among the three shield stratigraphic sections collected, no systematic compositional variations – elemental or isotopic – with elevation are apparent. All of these processes and features, including reduction of magmatic K2O and Rb abundances by weathering, have been identified for other Hawaiian shields as well. On the other hand, Kahoolawe shield has a unique place relative to the other islands when evaluating its origin in terms of mantle source composition.

10.2. Mantle source components for the Kahoolawe shield Central to characterizing the source for Kahoolawe volcano is recognizing that the collective isotopic and trace-element compositions for all Hawaiian shields point to a heterogeneous mantle associated with the Hawaiian plume source (e.g., Frey and Rhodes, 1993; Lassiter et al., 1996; Lassiter and Hauri, 1998; Blichert-Toft et al., 1999; Frey et al., 2005; Ren et al., 2005), and that the source differed in detail from one time to another when providing magmas to construct a volcano or a string of volcanoes (e.g., Kurz et al., 1995; Rhodes and Hart, 1995; Huang and Frey, 2005; Bryce et al., 2005; Yamasaki et al., 2009). That is, while magmas formed as partial melts of asthenospheric mantle, largely the plume, and possibly lithospheric peridotite and lower primitive mantle, melts of varying amounts and melt-percentages of other geologic materials contributed as well (e.g., Huang and Frey, 2005). These additional components were to a certain extent partial melts of ancient, recycled subducted oceanic crust (e.g., Lassiter and Hauri, 1998; Huang et al., 2009). But crust-derived domain in the Hawaiian mantle source can be divided into several discrete components that contributed to Hawaiian shield magmas in various proportions: fresh and altered basalt, gabbro, either plagioclase-rich or not; garnet-pyroxenite (eclogite) that represents high-P subduction of ocean crust; dacitic partial melts from garnet-pyroxenite; pelagic, terrestrial, and carbonate-rich sediments associated with subducted oceanic crust; and ancient seawater. A detailed assessment of components believed to have contributed to Hawaiian shields and a list of references is in Tanaka et al. (2008). Because of the great number of possible variables contributing to the Hawaiian mantle source, determining which ones and their proportions that were involved in the construction of a particular shield or stratigraphic section requires meticulous deciphering of shield compositional data (e.g., relatively high lava La/Nb and Sr/Nb suggest recycled sediments in the source, perhaps carbonate-rich; e.g., Huang et al., 2009; relatively high eHf can originate from pelagic sediments; relatively high SiO2 suggests a dacitic component or eclogite; e.g., Hauri, 1996; Jackson et al., 1999; Blichert-Toft et al., 1999; Bryce et al., 2005; Ren et al., 2005, 2009; Huang and Frey, 2005; Huang et al., 2009). Such assessments for defining the Kahoolawe source and magma origins for comparison to those for other Hawaiian shields, and especially to the Kea and Koolau mantle-composition endmembers, has been achieved by Huang et al. (2005), and somewhat modified by Tanaka et al. (2008). We offer them here in an abridged version. Among lava composition attributes that provide Kahoolawe source information are 87Sr/86Sr, 206Pb/204Pb, eHf, and Sr/Nb,

ARTICLE IN PRESS 120

R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

La/Nb, and Ce/Sr ratios. For example, Kahoolawe’s high 87Sr/86Sr ratios, which are similar to the high endmember ratios of the Koolau (Makapuu-stage) lavas, become distinct from Koolau when plotted against given values for eNd, eHf, 206Pb/204Pb, and Ce/Sr. The 87Sr/86Sr vs. Ce/Sr relationship (Fig. 20) is a distinguishing feature for the Kahoolawe source because Sr (ppm) and 87Sr/86Sr can be tied to contributions from ancient, recycled plagioclase-rich gabbroic oceanic crust in the source (believed to be present in sources for all Hawaiian shields; e.g., Lassiter and Hauri, 1998; Sobolev et al., 2000). That is, by Kahoolawe lavas having higher 87Sr/86Sr attended by lower Sr than Koolau (Makapuu-stage) lavas, and because ancient, recycled plagioclase-rich gabbro would have relatively high Sr and low 87 Sr/86Sr (i.e., there is little Rb in gabbro to age-generate higher 87 Sr/86Sr), the Kahoolawe source is believed to have had lower proportions of gabbroic component than the source for Koolau (Figs. 10, 20). Put another way, all Hawaiian shield lavas appear to have had recycled oceanic crust in their sources, but the amounts for each shield varied, and some shields varied during their stages of growth (e.g., Koolau; Huang and Frey, 2005). For Kahoolawe, the Sr characteristics of the source calculate as having  1% less plagioclase-rich gabbro component than did the Koolau mantle source (Huang et al., 2005). Tanaka et al. (2008) elaborate on the Koolau mantle component that is prevalent in Kahoolawe (e.g., high 87Sr/86Sr). They identify it as EMK source component, or the enriched Makapuu-stage of the Koolau shield, and emphasize that it included a large fraction of silica-saturated mafic material representing recycled oceanic crust (e.g., relatively high SiO2 lavas; Figs. 3A, 6) with sediment (higher 87Sr/86Sr); they do not describe the kind of sediment. To extend Kahoolawe compositional characteristics to their places among the Loa and Kea trends and mantle endmembers, Huang et al. (2005) plot 87Sr/86Sr, 176Hf/177Hf, 143Nd/144Nd vs. 208 Pbn/206Pbn, where this latter ratio is an expression that evaluates time-integrated 232Th/238U (e.g., Galer and O’Nions, 1985) (Fig. 21). First, Fig. 21 shows that Loa-trend shields extend along steep slopes from the Koolau endmember – or high 87Sr/86Sr and recycled sediment characterizations – toward the Kea endmember. Secondly, the Loa trend joins the flat Mauna Kea field (with Kea endmember characteristics) at plot coordinates that coincide with those for Loihi lavas. This suggests that yet a third endmember composition identified as Loihi is present in the

Fig. 21. Sr and Nd isotopic ratios in Kahoolawe shield and caldera-filling lavas (&) plotted against 208Pbn/206Pbn ratios (which convey time integrated 232Th/238U) to illustrate the suggestion of a third mantle component, Loihi, in the Kahoolawe source. Kahoolawe lavas plot midway along a somewhat curved line between Hawaiian shields lying between Kea and Koolau mantle components (large circles) and nearly overlap Loihi lava compositions (after Huang et al., 2005; calculation for 208Pbn/206Pbn presented therein).

mantle sources for Hawaiian shields and for Kahoolawe in particular. Kahoolawe lavas in effect unite the discrete Kea and Koolau endmembers in the isotopic ratio plots of Fig. 21. Tanaka et al. (2008) propose that the Loihi component is largely fertile peridotite from lowermost mantle. Huang et al. (2005) apply Kahoolawe compositions and those of other shields to describe how the heterogeneous Hawaiian plume probably melted to yield the various compositions observed among the many volcanoes. Their model invokes the three mantle endmembers – Koolau (EMK), Loihi, and Kea – where the Koolau component has the lowest solidus and is therefore first to melt in the upwelling plume. With further upwelling, the Loihi component partially melts, and the melts of the two endmembers mix to form the steep trend lines of Fig. 21. Finally, with continued upwelling, the Kea component partially melts and mixes with Loihi component melts.

10.3. Attributes of caldera-filling lavas Fig. 20. 87Sr/86Sr ratios for Kahoolawe shield and caldera-filling lavas (&) plotted against trace element ratio index Ce/Sr to help evaluate relative amounts of a gabbroic component in Hawaiian mantle sources. Most Kahoolawe lavas have compositions consistent with lesser gabbro component in their sources compared with other Hawaiian shields (see also Fig. 10). A general composition for sediment components in the mantle source is shown (after Huang et al., 2005).

The lavas that filled the caldera have the same overall compositions as shield lavas, and appear to be a continuum of the last of the magmas that constructed the shield. Because of similarity to shield lavas in bulk and mineral compositions, we

ARTICLE IN PRESS R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

121

infer that the differentiation histories of caldera-filling lavas were like those reported here for shield lavas. Also, there is a loose correlation between the compositionally evolved state of calderafilling lavas (e.g., Mg#s) at their emplacement times as marked by stratigraphic positions (Fig. 5). The trend to higher Mg#s observed for the upper section may reflect eruptions of the last of the stored shield and caldera-filling magmas before alkalic postshield magmatism began at some of the vents (e.g., Kamama, Kealialalo).

also be due to a source with certain trace element abundances slightly different from that for the shield. Also, owing to the relatively high SiO2 content of many of the postshield lavas (Fig. 3B), Tanaka et al. (2008) emphasize that the SiO2-rich source evident for shield lavas was sustained into the period of postshield lava generation.

10.4. Attributes of postshield lavas and their source

There is no geochemical evidence for genuine post-erosional, or rejuvenated, volcanism on Kahoolawe in spite of the interpretations made by Stearns (1940) and Macdonald (1940) from their field observations. Namely, there are no examples of SiO2undersaturated lavas in the strict sense of rejuvenated lavas as on, for example, Oahu (e.g., Clague and Frey, 1982). Also, the K–Ar ages and the geochemical and mineralogical characteristics of two post-erosional ‘candidate’ samples overlap with those of the Kahoolawe shield lavas.

Highlights for the Kahoolawe postshield  1.1–1.25 Ma history include the following: (i) eruption of tholeiitic lavas compositionally similar to shield basalts, some slightly enriched in certain incompatible elements over those of the shield but products of differentiation from tholeiitic shield lava compositions; (ii) sparse eruption of alkalic lavas, and differentiation of  9 wt% MgO alkalic parent magma to yield evolved alkalic basalt 6 wt% MgO and hawaiite magma; (iii) mixing between replenishment primitive and evolved magmas to create hybrid alkalic magma; (iv) incorporation of cumulate gabbro and ultramafic rocks as xenoliths from magma reservoir solidification zones; (v) formation of picrite magma; and (vi) secondary mineralization of REE, Y- and Ba-bearing groundmass minerals by trace-element leaching and subsequent concentration of labile elements. In the absence of isotopic compositions for a large suite of postshield lavas – although data are published for two samples and are similar to those of shield lavas (Figs. 9, 11, 12) – we cannot fully evaluate how alkalic lavas on Kahoolawe relate to those of the shield. On the other hand, there are generally similar isotope compositions between shield and postshield lavas of Loatrend shields (Moore and Clague, 1992), and we therefore extrapolate to infer that the alkalic magmas have generally the same isotopic ratios as the shield building magmas. This inference is supported by the data for two postshield samples (Figs. 9, 11, 12), by the unpublished Moiwi vent lava compositions noted earlier, and also by another unpublished composition for postshield alkalic basalt KAH-115 (provided by S. Huang), which we modeled as parental for major elements. This latter isotopic contribution shows that, overall, KAH-115 shares isotopic ratios (e.g., 0.70409; 18.078) with shield lavas except for slightly elevated 207Pb/204Pb (15.47) relative to the available shield data. As a parent composition expressed in all the panels of Fig. 18, KAH-115 shares trace element ratios with shield and calderafilling lavas for all except lower Zr/Nb and higher Zr/TiO2. And these ratios for KAH-115, when compared with those for shield and caldera-filling lavas, can each be explained by smaller percentages of partial melting of essentially the same source as that for shield magmas. The basis for this interpretation lies in partitioning coefficients that have relative values Nb oZr oTi, indicating that ratios among these elements can be sensitive to more or less percentages of source melting (e.g., Kamber and Collerson, 2000). While partial-melting percentage differences provide the straightforward explanation for trace-element ratio differences between a postshield alkalic and shield tholeiitic lavas (Fig. 18), we cannot rule out that a slightly different trace-element composition was in the source material, or that combinations of each account for the ratios observed. To summarize, sparse isotopic data for postshield lavas and the trace element ratios point to a postshield mantle source essentially similar in composition to that for the shield, which is generally the relationship between shield and postshield lavas at Loa-trend Hawaiian islands. Small differences in trace element ratios suggest lesser partial melting of that same source to yield alkalic parental ( 49 wt% MgO) magmas, but the differences could

10.5. Attributes of the ‘‘post-erosional’’ eruptives

10.6. Summary remarks Over the 25 years since the ‘window of opportunity’ opened for extensive sampling of Kahoolawe Island, a variety of studies on about 75 samples have shown it to be an integral part of the characterization of shield compositions and magma reservoir processes, and for the acquisition of compositional details about the Hawaiian mantle source. The studies also verify that the works and observations of the earliest investigators, Stearns (1940) and Macdonald (1940), were remarkably accurate and detailed in view of the technology available in their times for petrologic interpretations – to the extent that Macdonald (1940) recognized that Kahoolawe was evolved to only ‘andesine andesite’, and that basaltic andesite was the most abundant lava type after caldera collapse. Over 100 samples remain to be examined in detail and are available in the Smithsonian Institution of Washington, DC. With the enduring interest in Hawaiian petrology, mineralogy, and geochemistry, samples from the Kahoolawe collection are certain to add yet more to the understanding of plume-related ocean islands and the dynamics of the mantle and reservoir processes that produce their lavas.

Acknowledgements We greatly appreciate the invitation to us by Editor and Professor Klaus Keil to submit this review paper on the geology of an island that has occupied much of our thoughts and research over the past two decades. We are also thankful for the help and support from the graduate students who collected whole-rock and mineral compositions on Kahoolawe lavas as contributions to their Masters theses: E.A. Rudek, I. Rybek, and R.S. Jacobs – the last (Rob Jacobs) additionally having since provided financial support for furthering our petrologic studies. Other financial support has come from National Science Foundation Grant EAR-8903704 (RVF), and logistical support from the US Navy’s Pacific 3rd Fleet, Pearl Harbor, Hawaii. Some analytical support was provided by Oregon State University Radiation Center for neutron activation services funded by the US Department of Energy. Special thanks to Prof. F.A. Frey for his collaboration and the thoughtful insights he has provided about Kahoolawe geochemistry, and we thank Shichun Huang for generous access to unpublished isotopic compositions for Kahoolawe postshield lavas. S. Huang was also a courteous and meticulous reviewer of this paper, and M. Garcia

ARTICLE IN PRESS 122

R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

was a contentious one. Finally, we are grateful to the Smithsonian Institution for housing our collection of Kahoolawe samples.

References Abouchami, W., Hofmann, A.W., Galer, S.J.G., Frey, F.A., Eisele, J., Feigenson, M., 2005. Lead isotopes reveal bilateral asymmetry and vertical continuity in the Hawaiian mantle plume. Nature 434, 851–856. Anderson, A.T., Swihart, G.H., Atiol, G., Geiger, C.A., 1984. Segregation vesicles, gas filter-pressing, and igneous differentiation. J. Geol. 92, 55–72. Bauer, G.R., Fodor, R.V., Husler, J.W., Keil, K., 1973. Contributions to the mineral chemistry of Hawaiian rocks: III. Composition and mineralogy of a new rhyodacite occurrence on Oahu, Hawaii. Contrib. Mineral. Petrol. 40, 183–194. Blichert-Toft, J., Frey, F.A., Albarede, F., 1999. Hf isotope evidence for pelagic sediments in the source of Hawaiian basalts. Science 285, 879–882. Bryan, W.B., Finger, L.W., Chayes, F., 1969. Estimating proportions in petrographic mixing equations by least-squares approximation. Science 28, 926–927. Bryce, J.G., DePaulo, D.J., Lassiter, J.C., 2005. Geochemical structure of the Hawaiian plume: Sr, Nd, and Os isotopes in the 2.8 km HSDP-2 sections of Mauna Kea volcano. Geochem. Geophys. Geosyst. 6, Q09G18, doi:10.1029/2004GC000809. Chen, C.-Y., Frey, F.A., 1985. Trace element and isotope geochemistry of lavas from Haleakala volcano, East Maui: implications for the origin of Hawaiian basalts. J. Geophys. Res. 90, 8743–8768. Clague, D.A., 1987. Hawaiian xenolith populations, magma supply rates, and development of magma chambers. Bull. Volcanol. 49, 577–587. Clague, D.A., Frey, F.A., 1982. Petrology and trace element geochemistry of the Honolulu volcanics, Oahu: implications for the oceanic mantle below Hawaii. J. Petrol. 23, 447–504. Collerson, K.D., Weisler, M.I., 2007. Stone adze compositions and the extent of ancient Polynesian voyaging and trade. Science 317, 1907–1911. Fodor, R.V., Galar, P., 1997. A view into the subsurface of Mauna Kea volcano, Hawaii: crystallization processes interpreted through the petrology and petrography of gabbroic and ultramafic xenoliths. J. Petrol. 38, 581–624. Fodor, R.V., Moore, R.B., 1994. Petrology of gabbroic xenoliths in 1960 Kilauea basalt: crystalline remnants of prior (1955) magmatism. Bull. Volcanol. 56, 62–74. Fodor, R.V., Keil, K., Bunch, T.E., 1975. Contributions to the mineral chemistry of Hawaiian rocks: IV. Pyroxenes in rocks from Haleakala and West Maui volcanoes, Maui, Hawaii. Contrib. Miner. Petrol. 50, 173–195. Fodor, R.V., Bauer, G.R., Jacobs, R.S., Bornhorst, T.J., 1987. Kahoolawe Island, Hawaii: tholeiitic, alkalic, and unusual hydrothermal (?) ‘enrichment’ characteristics. J. Volcanol. Geotherm. Res. 31, 171–176. Fodor, R.V., Malta, D.P., Bauer, G.R., Jacobs, R.S., 1989. Microbeam analyses of rareearth element phosphate in basalt from Kahoolawe Island, Hawaii. In: Russell, P.E. (Ed.), Proceedings of the 24th Annual Conference of the Microbeam Analytical Society. San Francisco Press, San Francisco, pp. 554–558. Fodor, R.V., Frey, F.A., Bauer, G.R., Clague, D.A., 1992. Ages, rare-earth element enrichment, and petrogenesis of tholeiitic and alkalic basalts from Kahoolawe Island, Hawaii. Contrib. Miner. Petrol. 110, 442–462. Fodor, R.V., Rudek, E.A., Bauer, G.R., 1993. Hawaiian magma-reservoir processes as inferred from the petrology of gabbro xenoliths in basalt, Kahoolawe Island. Bull. Volcanol. 55, 204–218. Fodor, R.V., Jacobs, R.S., Bauer, G.R., 1994. Hollandite in Hawaiian basalt: a relocation site for weathering-mobilized elements. Miner. Mag. 58, 589–596. Fodor, R.V., Bauer, G.R., Jacobs, R.S., 1998. Alkalic magma modified by incorporation of diverse tholeiitic components: ‘complex’ hybridization on Kahoolawe Island, Hawaii. Miner. Petrol. 63, 73–94. Frey, F.A., Rhodes, J.M., 1993. Intershield geochemical differences among Hawaiian volcanoes: implications for source compositions, melting process, and magma ascent paths. Philos. Trans. R. Soc. London A 342, 121–136. Frey, F.A., Wise, W.S., Garcia, M.O., West, H., Kwon, S.-T., Kennedy, A., 1990. Evolution of Mauna Kea volcano, Hawaii: petrologic and geochemical constraints on postshield volcanism. J. Geophys. Res. 95, 1271–1300. Frey, F.A., Garcia, M.O., Roden, M.F., 1994. Geochemical characteristics of Koolau volcano: implications of intershield geochemical differences among Hawaiian volcanoes. Geochim. Cosmochim. Acta 58, 1441–1462. Frey, F.A., Huang, S., Blichert-Toft, J., Regelous, M., Boyet, M., 2005. Origin of depleted components in basalt related to the Hawaiian hot spot: evidence from isotopic and incompatible element ratios. Geochem. Geophys. Geosyst., 6, doi:10.1029/2004GC000757. Galer, S.J.G., O’Nions, R.K., 1985. Residence time of thorium, uranium and lead in the mantle with implications for mantle convection. Nature 316, 778–782. Ghiorso, M.S., Sack, R.O., 1995. Chemical mass transfer in magmatic processes: IV. A revised and internally consistent thermodynamic equilibria in magmatic systems at elevated temperatures and pressures. Contrib. Miner. Petrol. 119, 197–212. Hauri, E.H., 1996. Major-element variability in the Hawaiian mantle plume. Nature 338, 415–419. Herzberg, C., 2006. Petrology and thermal structure of the Hawaiian plume from Mauna Kea volcano. Nature 444, 605–609. Hirose, K., Kushiro, I., 1993. Partial melting of dry peridotites at high pressures: determination of compositions of melts segregated from peridotite using aggregates of diamond. Earth Planet. Sci. Lett. 114, 77–489. Huang, S., Frey, F.A., 2003. Trace element abundances of Mauna Kea basalt from phase 2 of the Hawaii Scientific Drilling Project: petrogenetic implications of

correlations with major element content and isotopic ratios. Geochem. Geophys. Geosyst. 4, 8711, doi:10.1029/2002GC000322. Huang, S., Frey, F.A., 2005. Recycled oceanic crust in the Hawaiian plume: evidence from temporal geochemical variations within the Koolau shield. Contrib. Miner. Petrol. 149, 556–575. Huang, S., Frey, F.A., Bichert-Toft, J., Fodor, R.V., Bauer, G.R., Xu, G., 2005. Enriched components in the Hawaiian plume: evidence from Kahoolawe volcano, Hawaii. Geochem. Geophys. Geosyst. 6, Q11006, doi:10.1029/2005GC001012. Huang, S., Abouchami, W., Blichert-Toft, J., Clague, D.A., Cousens, B.L., Frey, F.A., Humayun, M., 2009. Ancient carbonate sedimentary signature in the Hawaiian plume: evidence from Mahukona volcano, Hawaii. Geochem. Geophys. Geosyst., 10, doi:10.1029/2009GC002418. Jackson, E.D., Silver, E.A., Dalrymple, G.B., 1972. Hawaiian–Emperor chain and its relation to Cenozoic circumpacific tectonics. Geol. Soc. Am. Bull. 83, 601–618. Jackson, M.C., Frey, F.A., Garcia, M.O., Wilmoth, R.A., 1999. Geology and geochemistry of basaltic lava flows and dikes from the trans-Koolau tunnel, Oahu, Hawaii. Bull. Volcanol. 60, 381–401. Jacobs, R.S., 1986. Geochemistry and petrology of basalts from Kahoolawe Island, Hawaii. MS Thesis, North Carolina State University, Raleigh, 101 pp. Kamber, B.S., Collerson, K.D., 2000. Zr/Nb systematics of ocean island basalts reassessed – the case for binary mixing. J. Petrol. 41, 1007–1021. Keil, K., Fodor, R.V., Bunch, T.E., 1972. Contributions to the mineral chemistry of Hawaiian rocks II. Feldspars and interstitial material in rocks from Haleakala and West Maui volcanoes, Maui, Hawaii. Contrib. Miner. Petrol. 37, 237–276. Kurz, M.D., Kenna, T.C., Kammer, D.P., Rhodes, J.M., Garcia, M.O., 1995. Isotope evolution of Mauna Loa volcano: a view from submarine southwest rift zone. In Mauna Loa Revealed: Structure, Composition, History, and Hazards. In: Rhodes, J.M., Lockwood, J.P. (Eds.), American Geophysical Monograph, vol. 92; 1995, pp. 289–306. Lassiter, J.C., Hauri, E.H., 1998. Osmium isotope variation in Hawaiian lavas: evidence for recycled oceanic lithosphere in the Hawaiian plume. Earth Planet. Sci. Lett. 164, 483–496. Lassiter, J.C., DePaolo, D.J., Tatsumoto, M., 1996. Isotopic evolution of Mauna Kea volcano: results from the initial phase of the Hawaii Scientific Drilling Project. J. Geophys. Res. 101, 11769–11780. LeBas, M.J., LeMaitre, R.W., Streckeisen, A., Zanettin, B., 1986. A chemical classification of volcanic rocks based on the total alkali-silica diagram. J. Petrol. 27, 745–750. LeBas, M.J., LeMaitre, R.W., Woolley, A.R., 1992. The construction of the total alkalisilica chemical classification of volcanic rocks. Miner. Petrol. 46, 1–22. Leeman, W.P., Gerlach, D.C., Garcia, M.O., West, H.B., 1994. Geochemical variation in lavas from Kahoolawe volcano, Hawaii: evidence for open system evolution of plume-derived magmas. Contrib. Miner. Petrol. 116, 62–77. Macdonald, G.A., 1940. Petrography of Kahoolawe. Hawaii Div. Hydrogr. Bull. 6, 149–173. Macdonald, G.A., 1949. Hawaiian petrographic province. Geol. Soc. Am. Bull. 60, 1541–1596. Macdonald, G.A., 1960. Dissimilarity of continental and oceanic rock types. J. Petrol. 1, 172–177. Macdonald, G.A., Katsura, T., 1964. Chemical composition of Hawaiian lavas. J. Petrol. 5, 82–113. Marsh, B.D., Teplow, W., Reagan, M., Sims, K., 2008. Puna dacite: likely temperature, viscosity, origin, size, and parent body nature. EOS Trans. Am. Geophys. Union 89(53), Fall Meet. Suppl., Abs., V23A, 2129. McCarter, R.L., Fodor, R.V., Trusdell, F., 2006. Perspectives on basaltic magma crystallization and differentiation: lava-lake blocks erupted at Mauna Loa volcano summit, Hawaii. Lithos 90, 187–213. Moore, J.G., Clague, D.A., 1992. Volcano growth and evolution of the island of Hawaii. Geol. Soc. Am. Bull. 104, 1471–1484. Murata, K.J., Richter, D.H., 1961. Magmatic differentiation in the Uwekahuna laccolith, Kilauea caldera, Hawaii. J. Petrol. 2, 424–437. Naughton, J.J., Macdonald, G.A., Greenberg, V.A., 1980. Some additional potassium– argon ages of Hawaiian rocks: the Maui volcanic complex of Molokai, Maui, Lanai, and Kahoolawe. J. Volcano Geothermal Res. 7, 339–355. Neilson, R.L., Beard, J.S., 2000. Magnetite-melt HFSE partitioning. Chem. Geol. 164, 21–34. Ren, Z.-Y., Ingle, S.S., Takahashi, E., Hirano, N., Hirata, T., 2005. The chemical structure of the Hawaiian mantle plume. Nature 436, 837–840. Ren, Z.-Y., Hanyu, T., Miyazaki, T., Chang, Q., Kawbata, H., Takahasi, T., Hirahara, T., Nichols, A.R., Tatsumi, Y., 2009. Geochemical differences of the Hawaiian shield lavas: implications for melting process in the heterogeneous Hawaiian plume. J. Petrol. 50, 1553–1573. Rhodes, J.M., 1996. The geochemical stratigraphy of lava flows sampled by the Hawaii Scientific Drilling Project. J. Geopys. Res. 101, 11,729–11,746. Rhodes, J.M., Hart, S.R., 1995. Episodic trace element and isotopic variations in historical Mauna Loa lavas: implications for magma and plume dynamics. In: Mauna Loa Revealed: Structure, Composition, History, and Hazards. In: Rhodes, J.M., Lockwood, J.P. (Eds.), American Geophysical Monograph, vol. 92; 1995, pp. 263–288. Roden, M.F., Trull, T., Hart, S.R., Frey, F.A., 1994. New He, Sr, Nd, and Pb isotopic constraints on the constitution of the Hawaiian plume: results from Koolau volcano, Oahu. Hawaii Geochim. Cosmochim. Acta 58, 1431–1440. Roeder, P.L., Emslie, R.F., 1970. Olivine-liquid equilibrium. Contrib. Miner. Petrol. 29, 275–291.

ARTICLE IN PRESS R.V. Fodor, G.R. Bauer / Chemie der Erde 70 (2010) 101–123

Rudek, E.A., 1988. Petrology of ultramafic an gabbroic inclusions in basaltic rocks of Kahoolawe Island, Hawaii. MS Thesis, North Carolina State University, Raleigh, 52 pp. Rudek, E.A., Fodor, R.V., Bauer, G.R., 1992. Petrology of ultramafic and mafic xenoliths in picrite of Kahoolawe Island, Hawaii. Bull. Volcanol. 55, 74–84. Rybek, I., 1995. Petrology of shield-building, caldera-filling, and postshield basaltic rocks, Kahoolawe Island, Hawaii. MS thesis, North Carolina State University, Raleigh, 56 pp. Sack, R.O., Walker, D., Carmichael, I.S.E., 1987. Experimental petrology of alkalic lavas: constraints on cotectics of multiple saturation in natural basic liquids. Contrib. Miner. Petrol. 96, 1–23. Sano, H., Sherrod, D.R., Tagami, T., 2006. Youngest volcanism about 1 million years ago at Kahoolawe Island, Hawaii. J. Volcanol. Geothermal Res. 152, 91– 96. Sherrod, D.R., Sinton, J.M., Watkins, S.E., Brunt, K.M., 2007,. Geologic map of the State of Hawai’i: US Geological Survey Open-File Report 2007-1089. Available from: /http://pubs.usgs.gov/of/2007/1089/S. Sobolev, A.V., Hofmann, A.W., Nikogosian, I.K., 2000. Recycled oceanic crust observed in ghost plagioclase within the source of Mauna Loa lavas. Nature 404, 986–990.

123

Stearns, H.T., 1940. Geology and groundwater resources of the islands of Lanai and Kahoolawe, Hawaii. Hawaii Div. Hydrogr. Bull. 6 (3–95), 119–147. Stearns, H.T., Vaksvik, K.N., 1935. Geology and groundwater resources of the Island of Oahu, Hawaii. Hawaii (Territory): Div. Hydrogr. Bull. 1, 479. Stearns, H.T., Vaksvik, K.N., 1938. Records of the drilled wells on the Island of Oahu. Hawaii (Territory) Div. Hydrogr. Bull. 4, 213. Tanaka, R., Makishima, A., Makamura, E., 2008. Hawaiian double volcanic chain triggered by an episodic involvement of recycled material: constraints from temporal Sr–Nd–Hf–Pb isotopic trend of the Loa-type volcanoes. Earth Planet. Sci. Lett. 265, 450–465. Wanless, V.D., Garcia, M.O., Rhodes, .M., Weis, D., Norman, M.D., 2006. Shield-stage alkalic volcanism on Mauna Loa volcano, Hawaii. Weinstein, J.P., Fodor, R.V., Bauer, G.R., 2004. Koolau shield basalt as xenoliths entrained during rejuvenated-stage eruptions: perspectives on magma mixing. Bull. Volcanol. 66, 182–199. West, H.B., Gerlach, D.C., Leeman, W.P., Garcia, M.O., 1987. Isotopic constraints on the origin of Hawaiian lavas from the Maui volcanic complex, Hawaii. Nature 330, 216–220. Yamasaki, S., Kani, R., Hanan, B.B., Tagami, T., 2009. Isotopic geochemistry of Hualalai shield-stage tholeiitic basalts from submarine North Kona region, Hawaii. J. Volcanol. Geothermal Res. 185, 223–230.