The open-system geochemical evolution of alkalic cap lavas from Haleakala Crater, Hawaii, USA

The open-system geochemical evolution of alkalic cap lavas from Haleakala Crater, Hawaii, USA

Geochimica Pergamon et Cosmochimica Acta, Vol. 58, No. 2, pp. 773-796, 1994 Copyright 0 1994 Elsevier Science Ltd Printedin theUSA.Allrights reserv...

3MB Sizes 0 Downloads 2 Views

Recommend Documents

No documents
Geochimica

Pergamon

et Cosmochimica

Acta, Vol. 58, No. 2, pp. 773-796, 1994 Copyright 0 1994 Elsevier Science Ltd Printedin theUSA.Allrights reserved 0016-7037194 $6.00 + .OO

The open-system geochemical evolution of alkalic cap lavas from Haleakala Crater, Hawaii, USA H. B. WEST’ and W. P. LEEMAN’ ‘Hawaii Institute of Geophysics, School of Ocean and Earth Science and Technology, University of Hawaii, Honolulu, HI 96822, USA *Keith-W& Geological Laboratories, Rice University, P.O. Box 1892, Houston, TX 7725 1, USA (Received Februury 15, 1993; accepted in revised.form September 7, 1993)

Abstract-Lavas ranging from alkalic basalt to mugearite are exposed in a continuous stratigraphic section in Haleakala Crater on the island of Maui. These lavas define a series of discrete magma batches separated by sharp compositional breaks characterized by upsection increases in incompatible element (e.g., K20, Nb, REE, Th, Ba, Rb) contents, Al203/CaO ratios, and La/Sm ratios and complementary decreases in compatible element (e.g., MgO, Ni, Cr, V, SC) contents. Within magma batches, there are systematic upsection decreases in incompatible element contents, A120,/Ca0 ratios, and La/Sm ratios and systematic increases in compatible element contents. The observed compositional cyclicity can be explained as the result of alternating periods of low and high magma chamber recharge rates. The upsection interbatch compositional breaks represent periods of low or negligible magma chamber recharge, during which eruption is suppressed and the geochemical effects of crystal fractionation dominate those imposed by magma mixing. At the initiation of recharge, the influx of mafic magma into the magma chamber produces an eruption, resulting in partial evacuation of the magma chamber. Following eruption, recharge magma mixes with magma remaining in the magma chamber. Continued recharge and mixing during these eruptive periods cause the compositions of lavas to become more mafic with time. Thus, during a recharge period, the bulk composition of the magma chamber approaches that of the recharge magmas. The most mafic lavas (i.e., the youngest lavas within each magma batch) typically are relatively evolved (Mg# < 48, MgO < 7 wt%) and, therefore, cannot represent primary magmas. Evidence for polybaric fractionation suggests that these lavas represent recharge magmas that have undergone crystal fractionation in a deeper-level reservoir prior to injection into the shallow-level Haleakala magma chamber. The observed fine-scale geochemical cyclicity is superimposed over the larger-scale, systematic changes in isotope ratios and highly incompatible trace element ratios that are attributed to time-dependent changes in the proportions of distinct mantle components in the source of these lavas. These contrasting scales of temporal geochemical variability illustrate how shallow-level magma chamber processes have imposed short time period ( -2000 years) compositional controls on magmas whose compositions were also changing over relatively long time periods (=800,000 years) due to source-related effects. The decrease in magma supply rates during the alkalic cap phase of Hawaiian volcanism probably results in increased magma chamber residence times, thus creating the greater magmatic diversity observed. The specifics of the petrologic and geochemical evolution of individual Hawaiian volcanoes during this period differ at least in part because of differences in magma supply rates. predicted for closed-system magmatic conditions ( O’HARA, 1977; O’HARA and MATTHEWS, 1981). Volcanoes of the Hawaiian hotspot provide excellent natural laboratories for petrologic and geochemical studies of magma chambers because of their relatively simple geologic context. Situated roughly in the middle of the Pacific Ocean, magmas rising from the Hawaiian plume source are not susceptible to contamination by the geochemically extreme continental crust. Much of the compositional variability displayed by Hawaiian lavas can be attributed to the interaction of magmatic process, such as fractional crystallization and magma mixing, that occur within magma chambers and rift zones (MACDONALD, 1949, 1968; MACDONALD and KATSURA, 1964; WRIGHT, 1971, 1973: WRIGHT and FISKE, 1971). Four distinct evolutionary stages have been recognized for volcanoes of the Hawaiian hotspot, although not all exist at anyonevolcano (STEARNS, ~~~~;MACWNALD etal., 1983). The pre-shield stage consists of intercalated basanitic, alkalic,

INTRODUCTION THE IMPORTANCE OF MAGMA chamber

processes in controlling the compositions of erupted lavas has been documented for all major tectonic environments, including mid-ocean ridges (BRYAN et al., 1979; RHODES and DUNCAN, 1979; ELTHON, 1984; STAKES et al., 1984; KARSON and ELTHON, 1987), volcanic arcs (REAGAN et al., 1987; BROPHY, 1990), within-plate continental settings ( WORNER and WRIGHT, 1984; LEEMAN and HAWKESWORTH, 1986; NORMAN and LEEMAN, 1990), and oceanic islands (e.g., Hawaii: MACDONALD, 1942; WRIGHT and FISKE, 197 1; WRIGHT, 1973; CLAGUE, 1987a; FREY et al., 1990; CLAGUE and BOHRSON, 1991; CHEN et al., 1992). Geochemical and petrologic evidence indicates that magma chambers can behave as open systems in which repeated magma recharge, magma mixing, and fractional crystallization lead to more complex and extreme variations in magma compositions (incompatible trace element abundances and ratios, in particular) than those 773

774

H. B. West and W. P. Leeman

and tholeiitic basalts and accounts for perhaps 3% of the total volume of a Hawaiian shield volcano ( CLAGUE, 1987b). This stage is recognized for Loihi (MOORE et al., 1982) and Mahukona (GARCIA et al., 1990) seamounts. The shield-building stage is characterized by voluminous outpourings of tholeiitic basalt, which produce 294-95% of the total shield volume (MACDONALD and KATSURA, 1962). The two most active Hawaiian volcanoes, Kilauea and Mauna Loa, are in this stage. The alkalic cap or post-shield stage forms just O.l3.0% of the total shield volume (MACDONALD, 1963 ). Rock types fall within the alkalic basalt + trachyte series and typically form a thin veneer on the upper slopes of these volcanoes. The transition between shield-building and alkalic cap volcanism probably is gradual, based on results for Kohala, Mauna Kea, and Haleakala ( FEICENSON et al., 1983; SPENGLER and GARCIA, 1988; WEST et al., 1988; FREY et al., 1990; CHEN et al., 1991). The post-erosional or rejuvenated stage is evident on some Hawaiian volcanoes (e.g., Haleakala, Koolau, Kauai) and consists of localized, smallvolume eruptions of highly silica-undersaturated lavas, such as basanites, nephelinites, and melilitites, that follow an extended period (0.2-2.5 m.y.; CLAGUE et al., 1982) of volcanic quiescence and erosion. These lavas compose only a tiny fraction (< 1% ) of the total shield volume (MACDONALD, 1968). Magma storage and migration have been studied extensively at the relatively young and still active Hawaiian shield volcanoes of Kilauea and Mauna Loa (WRIGHT, 197 1, 1973; WRIGHT and FISKE, 1971; RYAN et al., 1981; KLEIN et al., 1987; RHODES, 1988). However, because of high magma recharge rates resulting from high melt production, lavas erupted from these volcanoes exhibit comparatively little compositional diversity (e.g., in incompatible element abundances and ratios) relative to lavas erupted during the alkalic cap stages of other Hawaiian volcanoes. Shield-building tholeiites, to a large extent, fall along or are not far removed from olivine control lines ( POWERS, 1955; MACDONALD and KATSURA, 1964; WRIGHT, 197 1 ). Thus, studies of tholeiitic magmatism provide limited insights into the evolution of magma chambers existent during the alkalic cap stage, a period of waning magma supply and diminishing eruptive frequency. In this paper, we present a geochemical investigation of a nearly continuous stratigraphic section of alkalic cap lavas from Haleakala Crater, located at the summit of Haleakala volcano. This relatively thick exposure of lavas is particularly well-suited for examining in detail geochemical evolution within the Hawaiian alkalic cap stage, because it occurs near the summit, where the eruptive record is more complete than are exposures at the flanks of the volcano. Our results indicate that alkalic cap lavas from Haleakala Crater display finescale cyclic geochemical variations that are imprinted over the larger-scale systematic temporal changes in Sr, Nd, and Pb isotope compositions and some incompatible trace element ratios (CHEN and FREY, 1985; WEST and LEEMAN, 1987; CHEN et al., 1990) associated with evolution of the Haleakala source. We interpret the cyclic, upsection geochemical variations as evidence that these lavas represent mixed magmas erupted from an evolving open-system magma chamber.

GENERAL GEOLOGIC SETTING AND SAMPLING Haleakala isa large shield volcano (29.3 X IO3 km3; BARGARand JACKSON,1974) that forms the eastern half of the island of Maui and is one of six volcanoes that together compose the Maui Volcanic Complex (Fig. 1). Lavas from Haleakala have been divided into four formations (STEARNSand MACDONALD,1942; MACDONALDand POWERS,1946, 1968; MACDONALD,1978). The oldest, the Honomanu Formation (>0.83 Ma; NAUGHTONet al., 1980). contains interbedded tholeiitic and alkalic basalts and, therefore, represents a transitional period between the shield-building (tholeiitic) and alkalic cap stages of Haleakala. The Haleakala alkalic cap stage comprises two formations: the Kumuiliahi Formation and the Kula Formation. The Kumuiliahi Formation (0.70-0.9 I Ma; NAUGHTONet al., 1980) is exposed on11 in Haleakala Crater and consists of interbedded alkalic basalts and hawaiites ( MACDONALD,1978; WESTand LEEMAN,1987). The Kula Formation (0.36-0.88 Ma; MCDOUGALL,1964; NAUGHTONet al.. 1980) unconformably overlies the Honomanu Formation at the distal flanks of the volcano (STEARNSand MACDONALD,1942) and the Kumuiliahi Formation at a few isolated localities in Haleakala Crater (MACDONALD, 1978). The Kula Formation forms the vast bulk of subaerially exposed lavas at Haleakala and consists of alkalic basalt. hawaiite, mugearite, basanitoid, ankaramite, and trachyte. Unconformably overlying the Kula Formation is the post-erosional Hana Formation, which is composed of basanitoids, hawaiites, and alkalic basalts. The time of inception of Hana volcanism is unknown but the youngest eruption occurred in z 1790 (OOSTDAM,1965). Haleakala Crater is an erosional feature (STEARNS, 1942; MACDONALD, 1978) located at the summit of Haleakala (Fig. 1). The floor of the crater is blanketed by Hana lava flows and cinder cones. With the exception of a relatively small section of Kumuiliahi lavas on the south wall, the walls of Haleakala Crater are composed of interbedded Kula flows. Because the thickness ofthe Haleakala alkalic cap thins away from the summit (MACDONALD et al., 1983), lavas exposed in Haleakala Crater provide the most continuous record available of the growth of the volcano during this stage. Moreover. lavas exposed in the summit area generally are fresher than those exposed on the rainy flanks of Haleakala, and magmas erupted near the summit are less likely to have had their compositions modified due to migration and storage in rift zones. For this study, fifty samples of Kula Formation lavas were collected along a continuous 300 meter section exposed on the northwest wall of Haleakala Crater and traversed by the Halemauu Trail. Only fresh flows were collected, and owing to partial coverage of the section by soil zones. vegetation, and talus, not every flow was sampled. However. because the average flow thickness ( -5.8 m, excluding ash and soil zones) obtained for fifty samples in 287 meters of section is virtually identical to that estimated by STEARNS and MACDONALD(1942: ~6. I m), our sample suite is probably an accurate representation of the section as a whole. In addition to flows collected from the Halemauu section, we sampled a younger flow located at Pakaoao, on the upper southwest rim of Haleakala Crater. We also sampled the two flows that directly overlie the Kumuiliahi Formation on the south wall of Haleakala Crater; these are likely the oldest Kula lavas exposed in Haleakala Crater. A cautionary note should be made concerning our interpretations of the data presented in this paper. Detailed stratigraphic studies are the only realistic means of unraveling the complex interactions between the physical and geochemical processes that occur in magma chambers. However, even the most detailed of such studies are fraught with uncertainty. Even assuming, per impossibile, that a volcano’s entire eruptive record could be sampled, a complete magmatic history would still be unattainable, because some magmatic processes probably are significant only during protracted periods of storage. The effects of such processes on magma compositions can be lost entirely if the magmas are never erupted, or may be obliterated or masked by other processes. Our interpretations and conclusions stem from the assumption that our sampling density is sufficient to discern the processes responsible for the temporal geochemical variations observed. We think it unlikely that the systematic geochemical variations found in lavas from the Haleakala Crater section are simply an artifact of incomplete sampling. Given that gaps in the eruptive record are

Petrogenesis

of the alkali-rich

lavas on Haleakala

775

I

157

Haleakala

I

I

I

156”

Crater

(after Macdonald, 1978)

1 mile p:--_:

I

Hana formation

~

Kumuiliahi Formation

~

Kula formation (outside Haleakala Crater) Kula formation (Haleakala Crater walls)

FIG. I. General geologic map of Haleakala Crater showing the distribution of lavas of the Kumuiliahi, Kula, and Hana Formations (after MACDONALD, 1978). Also shown is the location of the Halemauu section. Shaded areas represent Kula Formation lavas. Solid areas represent Kumuiliahi Formation lavas. Light stipled and unmarked areas represent Hana Formation lava flows and cinder cones. Inset map shows location of Haleakala relative to other volcanoes of the Maui Volcanic Complex (from West et al.. 1987; W = west, E = east, H = Haleakala). Solid line around the islands is the 180 meter submarine contour.

likely, chemical vectors still can be discussed meaningfully and it is justifiable to explain them in terms of documentable magmatic processes. PREVIOUS

INVESTIGATIONS ALKALIC CAP

OF HALEAKALA LAVAS

The first chemical analyses of Haleakala lavas were presented by CROSS ( 19 15 ) and WASHINGTON and KEYES ( 1928 ), although the first systematic petrologic study was that of STEARNS and MACDONALD (1942). That seminal work was followed by other petrologic studies led by Macdonald (MACDONALD and POWERS, 1946, 1968; MACDONALD, 1972, 1978). Those studies concluded that Haleakala lavas were erupted from a compositionally zoned magma chamber in which differentiated magmas (hawaiites, mugearites, trachytes) formed near the top of the magma chamber or in isolated, peripheral pockets and mafic magmas resided near the bottom of the chamber. MACIXNALD and POWERS ( 1946, 1968) further suggested that Hana post-erosional lavas were derived from greater depths in the Haleakala magma chamber than were Kula lavas. In essence, these early studies envisaged all Haleakala lavas (shield-building, alkalic cap, and posterosional) to have been derived from a common parental magma composition by fractional crystallization at various depths in the Haleakala magma chamber.

Fodor and others (FODOR et al., 1972, 1975, 1977; KEIL et al., 1972; FODOR and KEIL, 1979) reported mineral compositions for a subset of the Haleakala lavas studied by MACDONALD and KATSURA ( 1964) and MACDONALD and POWERS (1968). Those studies were concerned more with establishing compositional differences between the Haleakala formations than with deciphering the petrogenesis of the lavas. CHEN and FREY (1983, 1985) and CHEN et al. (1990) used geochemical and isotopic data for drill-core samples collected from wells on the northeast flank of the volcano to demonstrate that the source of Haleakala lavas changed progressively with time. They proposed that Honomanu, Kula, and Hana lavas represent successively increasing degrees of contamination of ascending primitive, plumederived magmas by small degree melts of lithospheric [ MORB] wallrock (CHEN and FREY, 1983, 1985). Helium isotope studies (KURZ et al., 1985, 1987; KURZ, 1986) have supported the role of an undegassed (primitive) component in the source of Haleakala lavas. Strontium and lead isotope systematics of lavas from Haleakala Crater (WEST and LEEMAN, 1987) combined with isotopic results for other volcanoes of the Maui Volcanic Complex (WEST et al., 1987) show that the source of these lavas consists of at least three distinct components. In those studies, we proposed a model in which Hawaiian shield-building lavas represent relatively uncontaminated melts derived from a Hawaiian mantle plume source that contains

716

H. B. West and W. P. Leeman Table la. Major and trace element data for Kula Formation lavas from the Halemauu section of Haleakala Crater. Major element data are given in weight percent. Trace element data are given in ppm. Rock type is listed below each sample name. Batch 1 HK_I

H

SiO, *‘@a Fez% M8O CaO Na,O KzO pzo5 TiO, MnO LOI Total BP K’ CS’ Rb’ RbX Sri SrX Bai BaX Lax C.? NdX La Ce Nd Sm Eli

Tb HO Yb LU Y Zr Hf Nb Ta

Th Ni V Cr” CrX SC” sex CU Zn Pbl

45.54 14.47 15.59 5.16 10.38 3.23 1.04 0.48 3.94 0.21 100.04 891; 0.110 19.48 19 677 664 410 32 58 39 29.6 73.0 39.0 9.13 3.08 1.32 1.40 2.49 0.33 34 269 6.9 31 2.0 2.3 38 420 15 13 26.3 22 78 119 1.92

Batch 2 HK_2

HK_5

HK_6 H

%?

AB 45.55 16.61 12.97 5.83 11.94 2.22 0.80 0.38 3.13 0.16 0.28 99.87

H 46.38 13.64 15.05 5.08 10.16 3.06 1.05 0.50 3.81 0.20 0.37 99.30

10

19

17

II

693

618

624

62;

64;

31; 18 58 35

372 26 63 46 29.1 69.8 40.9 9.83 3.23 1.48 1.46 2.63 0.34 34 282 7.2 31 2.3 2.0 50 397 31 26 27.5 30 68 122

408 30 64 42

34; 22 60 38

308 28 65 36 25.5 58.9 33.9 8.27 2.79 1.15 1.06 2.28 0.31 30 230 5.9 27 2.0 1.5 61 387 96 73 27.0 29 89 105

27 210 24 8; 331 153 23 91 92

46.78 14.00 15.04 5.21 10.42 2.98 1.04 0.51 3.81 0.21 0.63 100.63

35 281 32 48 394 21 30 78 117

Batch 3 HK_9 HK_IO

HK_7

AB 45.99 14.50 14.05 5.72 11.18 2.55 0.85 0.45 3.21 0.19 0.51 99.26

2:: 26

3:; 8; 27 77 105

H 46.39 13.81 15.05 5.10 10.28 3.30

46.05 14.56 13.84 5.39 11.17 2.65 0.85 0.44 3.27 0.19 0.68 99.09

1.11 0.53 3.84 0.21 0.93 100.55

19

AB 45.56 14.21 14.30 6.56 10.92 2.99 0.94 0.46 3.43 0.21 99.58

HK_II

AB 46.53 14.13 14.00 6.47 11.09 2.76 0.95 0.45 3.39 0.19 0.4 1 100.37

17

17

65;

638

657

39; 33 68 40 29.1 11.8 47.0 9.33 3.08 1.31

350 28 65 35

33; 30 61 35 27.1 63.4 35.5 8.38 2.87 1.16 1.13 2.12 0.29 30 237 6.2 29 2.1 1.9 96 389 178 151 27.1 26 86 109

12

2.10 0.27 33 276 6.6 34 2.4 2.6 40 417 13

29 235 2s

103 391 146

ii.2 23 66 119

25 83 111

-

two components, one with isotopic characteristics approximately equal to those inferred for primitive mantle and the other being an enriched component. It was further proposed that post shield-building lavas (alkalic cap and post-erosional) represent (two component) plume melts that, with time, underwent increasing amounts of contamination by an isotopically depleted component (e.g., depleted asthenosphere or lower lithosphere).

versity Reactor Center using methods described in LEEMAN ( 1988). Precision determined from six replicate analyses of sample HK-46 is better than 5% for La, Ce, Nd, Sm, Eu. Yb, Lu, Hf, and SC, better than 10% for Th and Ta, and better than 15%for Tb. Major element data are plotted after setting Fe2+/(FeZf + Fesf) to 0.85 and recalculating to 100% on an anhydrous basis. RESULTS

ANALYTICAL

METHODS

Major and trace element data are listed in Table 1. Major oxides were determined by x-ray fluorescence [ XRF] at the University of Massachusetts at Amherst and the University of Houston. Rb, Sr, Ba, Y, Zr. Nb, Ni, V. Cr, SC, Cu, Zn, La, and Ce were determined by XRF at the University of Edinburgh. Rare earth elements [ REE], Th, Ta, Hf, Cr, and SC were determined by instrumental neutron activation analysis [ INAA] at MIT using techniques described in ILA and FREY ( 1984) and at Oregon State University using similar methods. Potassium, rubidium, cesium, and barium concentrations were determined by isotope dilution at MIT. Lead concentrations were determined by isotope dilution at MIT and the University of California at Santa Barbara. Boron abundances were determined by prompt gamma neutron activation analysis at the McMaster Uni-

Rock Classification Haleakala Crater Kula lavas fall entirely within the alkalic field on the silica-alkalies diagram of MACDONALD and KATSURA ( 1964) and form a trend typical of Hawaiian alkalic cap suites (Fig. 2). Although Kula lavas generally are less silica-undersaturated than are the post-erosional Hana lavas, the two groups overlap substantially. The rock classification used in this paper (Table 1) is based on normative mineral composition (MUIR and TILLEY, 196 1; COOMBS and WILKINSON, 1969). Normative mineral compositions were calculated on a molecular basis following the

Petrogenesis of the alkali-rich lavas on Haleakala

777

Table la (cant). Major and trace element data for Kula Formation lavas from the Halemauu section of Haleakala Crater. Major element data are given in weight percent. Trace element data are given in ppm. Rock type is listed

SiO, *‘A Fez% MgO CaO Na,O Kzo PA TiO, MnO LO1 Total BP K’ CS’ Rb’ RbX Sri SP Bai BP LP Ce” NdX La Ce Nd Sm EU Tb Ho Yb LU Y Zr Uf Nb Ta Th Ni V Cr” CrX SC” sex CU Zn Pb’

Batch 4 HK_13 H 48.02 14.52 13.04 3.99 8.64 3.97 1.59 0.77 3.85 0.22 0.52 99.13

below each sample Batch 5 HK_14 M 49.60 15.18 12.53 3.51 7.69 4.6 1

1.81

0.77 3.18 0.24 0.39 99.5 1

name.

HK_15 H 48.33 15.09 13.46 4.12 8.53 3.93 I .57 0.70 3.88 0.23 0.05 99.89

HK_IS

H 48.04 14.47 13.19 4.05 8.45 4.08 I .54 0.67 3.79 0.22 0.63 99.13

HK_17

H 48.03 14.53 13.19 3.90 8.50 3.86 1.51 0.68 3.78 0.23 0.99 99.20

HK_18

HK_19

H 4z3 14.98 13.37 4.15 8.59 4.63 1.52 0.7 1 3.81 0.24 99.49

10160 0.226 26.81

47.74 15.25 13.05 4.89 9.20 3.61 1.41 0.67 3.55 0.21 0.27 99.85

Batch 6 HK_21 M 53.65 17.28 8.61 2.52 5.44 5.90 2.39 0.67 1.94 0.24 0.7 1 99.35

HK_22 M 51.84 16.32 11.00 2.96 6.33 4.88 2.33 0.95 2.55 0.25 0.76 100.17

12597 0.45 1 32.18 34

30

29

27

29

6:;

781

784

754

770

778

753

542 52 108 65 42.9 109.8 60.3 13.21 4.25 I .45 I .59 3.10 0.45 44 401 9.5 48 3.8 3.4 6 257 2 3 18.5 17 16 131

590 49 119 67

58; 39 104

576 43 94 55

514 39 96 53

G.8 92.9 58.0 11.70 3.87 1.52

498 34 99 54 39.3 95.1 54.0 11.80 3.82 1.47

3 148

3.10 0.35 42 373 8.2 46 3.4 3.2 5 233

5 216

2.90 0.40 43 372 8.6 46 3.2 3.1 6 222

13 11 134

: 16.0 17 15 130

13 15 125

16.3 14 14 123

48 426 51

42 374 46

approach of JOHANNSON ( 193 1); prior to calculation, major elements were normalized to 100% on an anhydrous basis, with Fe’+/( Fe’+ + Fe3+) set to 0.85. Alkalic basalts (normative plagioclase is labradorite) with greater than 5% normative nepheline are termed basanitoids. Similarly, hawaiites (normative plagioclase is andesine) and mugearites (normative plagioclase is oligoclase) with greater than 5% normative nepheline are termed nepheline hawaiites and nepheline mugearites, respectively. Kula lavas exposed in Haleakala Crater include alkalic basalt, hawaiite, and mugearite, with subordinate ankaramite, basanitoid, and trachyte. The compositional diversity of Kula lavas is large compared to that of alkalic cap lavas from other Hawaiian volcanoes. For example, Hualalai lavas consist mainly of alkalic basalts with minor hawaiite and trachyte (CLAGUE et al., 1980; MOORE et al., 1987), Mauna Kea lavas are composed primarily of hawaiite with rare mugearite (WEST et al., 1988), and Kohala (FEIGENSON et al., 1983; LANPHERE and FREY, 1987; SPENCLER and GARCIA, 1988)

I 1

42 371 46

5 234 1 15 16 132

6:; 749 531 35 89 47 36.4 90.8 49.2 10.90 3.50 1.51 I .72 2.47 0.37 37 325 7.6 42 2.9 3.1 2;: 71 57 19.3 17 52 117

45

45

1002

935

819 51 136 77 55.2 126.0 74.0 13.60 4.42 1.56 i.80 0.45 51 516 9.9 72 4.9 4.5 3 34 1 3 6.0 6 2 124

861 55 140 78

56 515 73

; 58

12 5 135

and West Maui (STEARNS and MACDONALD, 1942; MACDONALD and KATSURA, 1964; MACDONALD, 1968) lavas contain few basalts relative to more evolved rocks (e.g., mugearite, benmoreite, and trachyte). Petrography The petrography of samples in this study is essentially equivalent to that of earlier studied Kula rocks (MACDONALD, 1942, 1978; MACDONALD and POWERS, 1946, 1968; FODOR et al., 1972, 1,975, 1977; KEIL et al., 1972; FODOR and KEIL, 1979). Representative mineral modes are presented in Table 2. Haleakala Crater Kula alkalic basalts contain variable amounts of olivine (up to it 11 ~01% ) , clinopyroxene ( up to ~2 1 ~01% ) , and plagioclase ( up to = 13 ~01% ) as their dominant phenocryst and microphenocryst phases with minor magnetite. Textures range from seriate to strongly porphyritic; some samples are glomeroporphyritic. The groundmass tends

778

H. B. West and W. P. Table la Haleakala

(coat). Crater.

Major Major

and trace element data element data are given

for Kula

Leeman

Formation

lavas

from

the Halemauu section of are given in ppm.

in weight percent. Trace element data

Rock type is listed below each sample name.

SiO, A’,% Fe,% [email protected] CaO Na,O K&J Pps TIO,

MnO LO1 Total BP K’ CS’ Rb’ RbX Sr’ SF Bai BaX Lax CeX NdX La Ce Nd Sm Eu Tb Ho Yb Lu Y Zr Hf Nb Ta Th Ni V Cr” CrX SC” sex Cu Zn Pb’

Batch 6 HK_23 M 52.23 16.19 11.31 3.01 6.43 4.87 2.35 0.99 2.60 0.24 100.82

Batch 7A HK_25 M 54.38 17.26 9.73 2.28 5.20 5.47 2.57 0.70 1.89 0.24 0.45 100.17 4.9

43 955 783 61 139 79 59.2 137.0 63.0 16.20 5.03 2.00 4.10 0.47 56 509 10.6 72 4.2 4.6 3 56 1 1 10.6 10 4 135

48

HK_26 M 53.96 16.89 9.55 2.29 5.10 5.66 2.55 0.67 1.87 0.25 0.66 99.45 4.5

49 938

857 61 149 76

818 63 148 82 68.3 144.4 77.6 16.90 5.15 2.30 2.53 4.68 0.64 63 563 12.5 78 5.0 4.5 4 25

79

3 27 3 s 2 137

99.70

99.16

4.0

1.8 20944 1.598 55.85

50

963

58 567

;:2 17.07 9.43 2.27 5.13 6.36 2.53 0.69 I .94 0.26

HK_28 M 52.76 15.90 10.90 3.09 5.25 5.54 2.44 0.34 2.72 0.22

HK_27

6:; 649

93;

I

s., : 140

856 64 145 7s

59 570 79

4 27

736 57 136 67 61.4 134.4 63.7 14.75 3.27 1.58 1.70 4.81 0.65 6:; 14.0 79 5.1 6.0 3 116 2

3 7 4 136

to be coarse and weakly trachytic. Plagioclase phenocrysts commonly are zoned and partially resorbed. Olivine phenocrysts commonly contain subgrain boundaries and are also partially resorbed. Some samples contain clinopyroxene-rich crystal clots; these clots typically are monomineralic but others include subordinate magnetite or olivine. Haleakala Crater Kula hawaiites and mugearites are nearly aphyric and have well-developed trachytic groundmasses. Plagioclase is the dominant phenocryst and microphenocryst phase, with subordinate olivine (< 1 vol%), clinopyroxene ( < 1 ~01%), and magnetite ( < 1 ~01’36). Clinopyroxene is absent in some of the more evolved samples. Phenocrysts and microphenocrysts in these rocks typically are partially resorbed and strongly zoned. Crystal clots containing clinopyroxene + magnetite * olivine are present in some samples. Apatite is a common groundmass phase and occurs in a few of the most evolved lavas as inclusions (typically in plagioclase) or as rare microphenocrysts. As noted by STEARNS and

10.1 9 4 118 4.42

Batch 7B HK_29 4:02 16.19 13.27 4.68 7.85 4.57 I .32 0.60 3.59 0.22 99.31 3.1

27 918 906 546 37 89 46 38.1 87.3 45.5 9.85 3.12 1.19 2.30 2.59 0.36 35 308 7.0 47 3.1 3.2 3 177 1 14.1 14 8 101

HK_30 H 48.04 16.02 13.23 4.76 7.78 4.03 1.34 0.60 3.57 0.23 0.41 100.01 2.4

26

HK_31 NH 45.02 IS.63 14.86 4.82 9.66 4.23 1.23 0.54 3.68 0.22

%iP 45.93 14.98 14.54 4.56 9.59 3.84

1.25

99.89

0.52 3.59 0.22 0.59 99.61

1.9

1.2

26

27

918

781

783

51;

548 35 84 48 35.4 79.7 39.8 9.58 3.01 I .47 1.90 2.27 0.34 33 264 6.2 44 3.4 3.1 9 337 5 2 16.7 13 14 115 2.21

527 31 86 43 35.5 80.0 46.0 9.24 2.91 1.13

;; 47 38.0 86.6 53.0 9.78 3.24 1.26 2.60 0.29 35 311 6.9 47 3.2 3.5 2 176

I

1 14.1 9 8 99

;.oo 0.26 32 261 64 44 3.0 3.0 34;

I

16.6 14 II 114

MACDONALD ( 1942), biotite occurs as a rare groundmass phase in some Kula lavas. Brown amphibole occurs in the groundmass and as phenocrysts in some of the mugearites. Amphibole phenocrysts typically are strongly resorbed, leaving rims of granular magnetite surrounding fresh cores. Commonly, pseudomorphs of granular magnetite are all that remain to mark former amphibole grains. This feature is prominent in the Pakaoao nepheline mugearite (sample HK-48; Table 1b). In addition, a loose rock fragment whose source could not be located, and whose composition is not reported in this paper, contains abundant phenocrysts of brown amphibole. Geochemistry In basaltic magmas, Nb should behave as a highly incompatible element. Niobium is used later in this paper as a reference incompatible element for the purposes of geochemical

Petrogenesis of the alkali-rich lavas on Haleakala Table la (cant). Haleakala Crater. Rock type is listed

SiO, * ‘~0s Fe,% MU CaO Na,O KsO PZOS X0, MnO LO1 Total BP K’ Cs’ Rb’ RbX Sr’ SrX Bai BaX Lax CeX NdX La Ce Nd Sm Eu Tb Ho Yb LU Y ZI Hf Nb Ta Th Ni V 0” C? SC" sex

Cu Zn Pb’

Batch 7B HK_33 MH 45.04 15.79 15.02 4.84 9.93 3.92

1.20 0.54 3.68 0.21 100.17 1.6

27

Major and trace element data Major element data are given below each sample name.

44.15 14.83 15.26 6.44 10.21 3.30 I .06 0.46 3.55 0.22 0.03 99.51 1.1

25

HK_35 Bas 44.01 14.56 15.43 6.70 10.46 3.04 1.oo 0.50 3.60 0.21 0.56 100.07 2.4

21

for Kula in weight

42.82 15.18 15.64 6.76 10.39 3.40 0.99 0.50 3.68 0.22 99.58

96.83

1.8

4.8 8284 0.557 50.34 49 1016 1013

81;

69;

719

760

516 34 70 40 35.2 78.3 40.3 9.08 2.98 1.62 1.60 2.36 0.36 31 259 6.2 43 3.2 2.6 8 345 4 1 17.2 13 17 116 2.08

407 29 68 34 29.7 68.4 36.0 8.18 2.57 1.12

451 35 71 36 29.1 68.2 40.0 8.24 2.72

1.80 0.33 29 227 5.7 36 2.7 2.8 54 407 42 34 23.5 20 53 112

I .60 0.24 28 222 5.2 36 2.6 2.5 57 420 60 54 24.4 26 51 Ill

453 25 76 33 28.9 67.6 35.2 7.92 2.69 0.84 1.37 2.02 0.31 28 226 5.2 37 2.6 2.6 61 409 61 50 24.4 24 55 Ill

1.05

Formation lavas from the Halemauu section of percent. Trace element data are given in ppm.

Batch 8 HK_37 M 51.19 17.67 10.11 2.84 6.57 5.73 2.17 0.99 2.31 0.25

21

modeling because there are data for all samples reported in this study, whereas data for most other highly incompatible trace elements (e.g., Th, Ta, REE) exist for only a subset of samples. The incompatibility of Nb in Haleakala Crater Kula lavas is supported by its strong negative correlation with MgO and strong positive correlation with other presumably highly incompatible elements (e.g., Th, Ta, light rare earth elements [ LREE], K, Rb, Ba). With increasing Nb abundance, SiOz, Zr, Hf, Y, middle rare earth element [ MREE], and heavy rare earth element [ HREE] contents increase, and thus also behave incompatibly. A1203/Ca0 and A1203/Ti02 ratios also increase with increasing incompatible element abundances. In contrast, Mg# [ (Mg X lOO)/( Mg + Fe2’)] and CaO, SC, Cr, and Ni contents decrease with increasing Nb and other incompatible element abundances. For Kula basalts, Sr behaves incompatibly even though plagioclase phenocrysts and microphenocrysts are common. This behavior could result from inefficient segregation of less-

119

84; 57 139 66 64.4 141.1 65.0 13.15 3.97 1.37 1.50 3.21 0.45 43 445 9.4 72 4.5 6.0 4 58 2 5.1 3 7 107 4.23

HK_38 NH+NM 50.38 17.58 10.52 3.04 6.80 5.39 2.11 1.04 2.39 0.23 0.12 99.61

49.05 17.33 11.04 3.70 7.43 5.56 1.81 0.82 2.87 0.24 99.85

46

39

956

1014

770 59 139 64 59.4 131.0 69.0 12.80 3.83 1.16 2.20 0.42 41 408 8.4 67 4.2 5.3 4 72 1 1 5.7 6 4 110

49.22 17.52 11.34 3.73 7.41 5.02 1.83 0.84 2.85 0.22

Batch 9 HK_43 NM 52.52 17.83 9.31 2.24 5.38 7.05 2.64 0.72 2.0 I 0.24

99.98

99.94

II&$!2

722 48 121 63 52.5 118.5 56.1 11.82 3.55 1.37 1.80 3.01 0.40 38 369 8.0 63 3.7 4.8 3 103 2 2 7.7 7 7 110

15443 0.434 41.97 38 1029 984

60 740

228.0 113.8 56.0 11.81 3.78 1.59 I .48 2.86 0.41 38 353 8.2 61 4.1 3.9 3 105

971 66 144 64 62.5 139.6 63.5 13.31 3.93 1.50 2.10 3.75 0.49 46 484 10.0 83 4.9 6.3 4 34

: 8.2 7 6 107 3.24

I 4.6 3 5 128 4.38

664 51 123

I

dense plagioclase crystals from denser surrounding basaltic magma (cf. CAMPBELL et al., 1978; ELTHON, 1984). Strontium abundances increase with increasing Nb, from 580-700 ppm Sr (at 24-31 ppm Nb) up to 1240 ppm Sr (at ~70 ppm Nb), and then decrease to ~750 ppm Sr (at 79-83 ppm Nb). This decrease in Sr at higher Nb abundances is accompanied by increases in Ba/Sr and Rb/Sr ratios. P205 behaves similarly to Sr. With increasing incompatible element abundances, Ti02 and V contents increase at low incompatible element contents (~30 ppm Nb) and then decrease substantially at higher Nb abundances. The major and trace element compositions of Kula lavas from Haleakala Crater span significantly larger ranges than do those exposed on the flanks of the volcano. In addition, ratios of highly vs. moderately incompatible elements (e.g., La/Sm, La/Yb) in these lavas span relatively large ranges and several vary systematically with incompatible element abundances. La/Sm ratios in these lavas range from 3.0 to

H. B. West and W. P. Leeman

Table

Major and trace element data Major element data are given Rock type is listed below each sample name. la (coat). Haleakala Crater.

SiO A&, Fez% [email protected] CaO NasO GO p205

TiO, MnO LO1 Total BP K’ Cs’ Rb’ RbX

Sri SrX Ba’ BaX Lax CeX NdX La Ce Nd Sm Eli Tb Ho Yb LU Y Zr Hf Nb Ta Th Ni V Cr” CF SC” sex Co Zo Pb’

6.4 (cf.

Batch 9 HK_44 NM 48.46 16.36 11.86 3.55 7.13 5.44 1.80

1.05 3.14 0.25 99.64

HK_45 NH 46.04 14.96 14.42 4.83 10.06 3.58 1.14 0.50 3.68 0.23 0.23 99.61

HK_49 Bas 45.22 16.19 14.04 4.16 10.24 3.59 1.01 0.49 3.55 0.21 0.09 99.39

39

24

846

781

85;

129 53 125 62 52.5 121.0 67.0 13.20 4.02 1.45

535 31 82 42 33.3 14.8 38.8 9.21 2.86 0.88 1.20 2.39 0.31 31 243 5.9 38 2.6 2.8 11 334 8 4 17.5 15 18 124

437 26 72 36 29.6 67.5 38.0 8.15 2.64 0.99

2.70 0.46 44 366 7.4 63 3.9 4.5 4 133 1 1 10.1 12 12 125

20

2.40 0.31 29 213 5.1 35 2.6 2.7 7 286 13 13 15.4 15 12 102

for Kula in weight

1.8 8415 0.242 22.69 21 719 691

-

Batch E&t

11

HK_52 NH 48.25 16.53 11.49 3.50 8.09 5.14 2.08 0.69 2.91 0.24 0.16 99.14

HK_53 NH 46.23 16.55 12.89 4.31 9.72 4.24 1.53 0.63 3.15 0.21 0.21 99.67

47

48

34

37

21

886

939

876

734

744

747 53 119 56 52.8 116.0

805 51 129 63 52.5 113.0

611 39 95 47 40.6 93.2 43.0 9.93 3.09 1.15

67; 45 108 54 48.2 104.0

474 32

12.00 3.69 1.52

:: 30.7 70.5 43.0 8.23 2.74 1.21

2.40 0.25 33 293 6.3 50 3.5 3.8 7 264 9 5 13.9 15 18 111

;.oo 0.43 43 346 7.5 57 3.9 4.6 4 166 2 1 11.2 11 10 122

1.60 0.27 29 230 56 31 2.6 25 12 319 9 9 16.9 15 21 113

100.18

99.63

range in La/Sm for flank lavas: 3.0-5.1; CHEN and 1985) and increase with increasing Nb up to =77 ppm; at higher Nb abundances, La/Sm ratios decrease slightly. La/Yb ratios in these lavas range from 11.2 to 27.0 (cf. range in La/Yb for flank lavas: 10.3-18.9; CHEN and FREY, 1985) and also increase with increasing Nb abundances up to a maximum and then decrease at higher Nb abundances. Certain ratios of elements of similar geochemical affinity span relatively large ranges and are correlated with incompatible element abundances but not with isotopic compositions. For example, La/Cc (range: 0.391-0.493) and Zr/Hf (range: 37.8-52.1) ratios generally increase with increasing Nb abundances. Similarly Ba/Rb ratios span a large range ( 13.4-3 1.6) and are negatively correlated with incompatible element abundances. In contrast, other incompatible element ratios (e.g., Zr/Nb, La/Hf, Ba/Zr, Ba/Y) are not correlated FREY,

Batch 10 HK_51 NH 48.18 17.00 11.91 3.47 8.15 5.32 2.14 0.75 3.03 0.23

HK_50 Bas 41.68 14.63 15.77 6.16 12.90 3.08 0.96 0.43 3.82 0.20

454 22 60 31 27.4 63.5 32.9 7.66 2.49 0.91 1.06 1.50 0.2 I 24 202 5.2 33 2.3 2.3 64 548 30 25 27.5 24 120 110 1.80

Formation lavas from the Halemauu section of oercent. Trace element data are given in ppm.

11.90 3.58 1.16

12.00 3.62 1.42

3.10 0.36 40 380 7.9 67 4.3 5.4 4 174 2

3.70 0.35 39 384 7.8 69 4.3 4.8 3 179 2 2 6.2

5.9 5 4 126 3.30

; 126

47.87 15.94 12.93 3.52 8.14 4.87 1.68 0.78 3.12 0.23 0.30 99.38

HK_55 AB 44.89 16.03 15.63 5.12 9.81 3.23 1.05 0.48 3.48 0.21 0.05 99.99

with absolute abundances of these elements correlated with isotopic ratios (Fig. 3).

but are strongly

Temporal Compositional Variations and Magma Batches The isotopic compositions of Kula lavas from both Haleakala Crater (WEST and LEEMAN, 1987) and the flanks of the volcano ( CHEN and FREY, 1985) are roughly correlated with relative stratigraphic position, indicating a general, systematic shift in source composition with time (Fig. 4a). This evolution of the Haleakala source composition is also reflected in some incompatible trace element ratios (e.g., Zr/Nb; Fig. 4b). The isotope and incompatible trace element ratios that, for these lavas, are indicative of source processes are not correlated with elemental abundances (Fig. 3). Thus, these ratios appear to have been unaffected by magma chamber processes, such as fractional crystallization and magma mixing.

Petrogenesis of the alkali-rich lavas on Haleakala Table la (cant). Haleakala Crater. Rock type is listed

Major and trace element data for Kula Major element data are given in weight below each sample name.

SiO *Isa, Fe& MgO CaO Na,O K-0

b*os TiO, Ma6 LO1 Total BP K’ Cs’ Rb’ RbX Sri SF Ba’ BaX LaX CeX NdX La Ce Nd Sm EU Tb Ho Yb LU V

ir

Hf Nb Ta Th Ni V Cf” CrX SC” SCX CU Za Pb’

Formation lavas from the Halemauu section of percent. Trace element data are given in ppm.

Batch 12 HK_56 NH 48.71 17.42 11.51 3.47 7.18 5.35 1.86 0.80 2.82 0.24

HK_57 Bas 41.72 15.03 16.94 1.25 11.81 2.52 0.79 0.42 3.51 0.19

HK_58 Bas/Ank 42.91 11.58 14.83 11.27 12.30 2.11 0.75 0.43 3.09 0.18

99.36

100.19

99.51

42

18

28

1092

665

580

829 49 117 55 48.1 104.0 47.0 10.90 3.42 1.18

437 19

2.70 0.28 36 330 6.9 61 3.9 4.6 3 70 4 1 6.4 4 7 114

1.80 0.27 26 174 4.5 29 2.3 2.5 59 503 76 67 28.5 26 58 105

503 22 49 28 25.8 58.5 29.8 6.81 2.38 0.66 0.96 1.34 0.17 22 174 4.6 31 2.3 2.5 217 408 610 552 35.1 32 102 96 2.05

AB = alkalic basalt, Bas = basanitoid, mugearite, NM = nepheline mugearite. x = XRF, n = INAA, i = isotope dilution,

:: 26.1 60.1 33.0 7.31 2.46 1.28

Ank

= ankaramite,

p = prompt

In contrast to the systematic temporal isotopic variations displayed by Haleakala Crater Kula lavas, the absolute major and trace element contents of these lavas did not change progressively with time. A general, progressive temporal change in lava compositions with respect to absolute abundances might result if compositional variations were strictly a function of increasing residence time in a single closedsystem magma chamber or of progressively decreasing amounts of partial melting in their source. Instead, major and trace element contents, and many trace element ratios, in these lavas vary upsection in a striking cyclic fashion, irrespective of isotopic variations or variations in incompatible trace element ratios that are indicative of changes in source composition. Plotted against relative stratigraphic position, distinct compositional breaks are observed that form coherent sub-

781

gamma

Batch 13 HK_59 4::4 11.96 11.64 2.71 7.62 6.17 1.98 0.89 2.47 0.22

HK_60 Bas 43.97 15.63 15.56 6.38 10.06 3.59 1.16 0.53 3.53 0.19

99.30

100.61

1284 0.573 52.20 50 1284 1260 865 941 47 125 55 55.9 126.2 55.2 11.36 3.36

747 531 25 ::

1.28 1.oo 2.09 0.30 31 325 6.5 70 4.5 5.9 9 66 18 9 4.1 5 14 124 3.30

H = hawaiite, neutron

28

activation

27 223 39

315: 63 19 65 111

NH = nepheline

hawaiite,

M =

analysis.

sections of lavas (Fig. 5). These subsections were initially delineated using highly incompatible elements, such as K and Nb. For the most part, compatible elements and elemental ratios sensitive to the effects of crystal fractionation are consistent with the locations of these divisions. Not all elements and ratios conform wholly to these subsections, which is not surprising considering that the behavior of these geochemical parameters during crystal fractionation depends on the crystallizing phase assemblage. To some extent, these subdivisions are arbitrary, owing to incomplete sampling or incomplete deposition of flow units. In general, the breaks between these geochemical subsections correspond to parts of the Haleakala Crater section covered by vegetation. These covered areas vary in thickness but typically they are thin (< 10 feet). Because extensive vegetative cover is usually confined to soil zones or to ashy or

782

H. B. West and W. P. Leeman Table lb. Major and trace element data for Kula Formation lavas Haleakala Crater. Major element data are given in weight percent. type is listed below each sample name. South HK_46 SiO, *U’s Fez% MgO CaO Na,O K2O ‘2’5

TiO, MnO LO1 Total BP K’ Cs’ Rb’ RbX Sr’ S? Ba’ BaX LaX Ce” NdX

Wall HK_47

H

H

46.56 16.43 13.29 4.69 8.28 4.38 1.32 0.58 3.74 0.24

48.12 16.66 11.94 3.92 1.84 5.21 1.62 0.79 3.33 0.26

99.51

99.69

3.8 11093 0.342 30.55 29 844 833 488 518 32 90 49

% LU Y Zr Hf Nb Ta Th Ni V Cr” CIX SC” sex cu Zn Pb’

100.04 3.1 19442 0.631 65.46

3;

68; 48 Ill 57

1022 60 136 54

data

H

1.62

9;; 918

element

La Ce Nd Sm Eli Tb

0.23

972

Trace

South HK_46

Pakaoao HK_48 NM 51.08 17.71 10.03 3.47 6.39 6.61 2.31 0.59

H = hawaiite, NM = nepheline mugearite. x = XRF, n = INAA, i = isotope dilution,

p = prompt

rubbly deposits between lava flows, it is likely that these compositional breaks represent periods during which lavas were either not erupted or not deposited at that location. Based on these observations, we interpret the geochemical subsections as representing discrete eruptive episodes separated by periods of volcanic quiescence. Thus, these separate lava packages appear to form a series of individual magma batches. It is possible that these gaps simply represent periods during which eruptions occurred elsewhere on the volcano. Even

Haleakala

from the south wall and

Crater Holua Trachyte aA

FIG. 2. Si02 vs. total alkalies ( NazO + KlO) for Kula Formation lavas from Haleakala Crater. Diagonal line represents the division between Hawaiian alkalic and tholeiitic lavas (MACDONALD and KATSURA, 1964). The solid triangle represents the Holua trachyte (WESTand LEEMAN,unpubl. data), which is exposed at the base of the northwest wall of Haleakala Crater. Shown for comparison are fields for Honomanu Formation (MACDONALD and POWERS, 1968; CHEN et al., 1991; H. B. West and W. P. Leeman, unpubl. data), Kumuiliahi Formation (H. B. West and W. P. Leeman, unpubl. data; MACDONALD,1978), and Hana Formation (H. B. West and W. P. Leeman, unpubl. data; CHEN et al., 1991) lavas.

gamma

neutron

activation

36.9 85.8 43.4 9.52 3.09 1.08 1.69 2.47 0.35 33 280 6.4 46 3.3 3.3 3 196 2 1 13.1 13 9 109

west rim (Pakaoao) of are given in ppm. Rock

Wall HK_47

H

39 325 58

3 145 I 8 1 124

Pakaoao HK_48 NM 61.2 124.2 49.9 9.56 2.97 1.27 1.60 2.51 0.37 33 401 8.1 77 5.1 6.9 45 64 135 Ill 6.4 6 13 123

analysis.

so, the inference that these subsections represent individual, distinct magma batches is not invalidated. Interbatch compositional breaks are characterized by substantial and sharp upsection increases in incompatible element contents (e.g., I&O, Nb, Rb, Th. LREE) and in some ratios (e.g., A1203/Ca0, La/Sm) and complementary sharp decreases in compatible element contents (e.g., Cr, SC, V, Ni) (Fig. 5). In contrast, compositional variations within magma batches are characterized by systematic upsection decreases in incompatible element contents, A1203/Ca0 ratios, and La/Sm ratios and systematic increases in compatible element contents. The magnitudes of interbatch compositional breaks vary, but in some instances the changes in composition span more than half of the entire compositional range of the Haleakala Crater section (e.g., batches 5 + 6, 7 + 8. and 9 -+ IO). In the batch 7 + 8 break, KZO contents increase from I .O to 2.2 wt%, equivalent to 63% of the total observed range. lntrabatch compositional variations can span even greater ranges; for example, within batch 9, K20 contents decrease from 2.7 to I .O wt%. equivalent to 88% of the total observed range. Note that batch 7 contains an internal compositional gap evident for certain elements and elemental ratios (Fig. 5 ); we have, therefore, subdivided it into two parts, 7A (four samples, HK-25 + HK-28) and 7B (eight samples, HK-29 + HK-36). This compositional gap corresponds to a part of the Haleakala Crater section that was not sampled because of poor exposure; therefore, it is possible that batch 7 represents two independent batches that are incompletely exposed. We have identified thirteen potential magma batches in the Haleakala Crater section. The compositional breaks between these batches are significant and appear to signify major

Petrogenesis of the alkali-rich lavas on Haleakala modes of representative Kula Formation Table 2. Mineral Crater. Modes are based on a minimum of 1000 counts.

CPX Amph Apat gm

95.2 97.0 99.7 99.3 99.3 >99.9 99.9 99.2 90.1 98.4 95.1 75.8 99.4 99.9 99.8 99.2 94.7 93.0 86.1 86.0 68.3 94.7

the

0.2 0.1


0.6


0.15 0.1 0.2 0.1 0.3 6.4 1.3

0.6 0.05 0.5 3.0 3.8

0.2 -

Magnetite & D!I

3.1 0.6 0.2

0.7

-

co.1 0.5 0.4
I -

-

0.3 -

0.1 0.3 20.0 2.6

co.1 co.1 0.6


0.3 0.05


0.3 2.2 10.6 0.5
1.6 0.1

0.3 1.0 5.75 6.2

2.7 0.7

clinopyroxene amphibole apatite groundmass

compositional changes in the Haleakala magma reservoir. Within eight of the thirteen batches, the first lavas erupted (i.e., oldest or lowermost stratigraphically within each magma batch) are hawaiites or mugearites and the last lavas erupted (i.e., youngest or highest stratigraphically) are alkalic basalts, basanitoids, or ankaramites (Table 3). DISCUSSION The existence of a well-established magma chamber during the alkalic cap stage at Haleakala is supported by four basic observations. First, even the most mafic lavas are relatively evolved; with the exception of ankaramite HK-58, all Haleakala Crater Kula basalts collected contain <7 wt% MgO and have Mg# ~48. Therefore, these lavas, along with the associated hawaiites and mugearites, must have undergone significant fractional crystallization. The relatively extensive crystal fractionation required to produce the most evolved Haleakala lava compositions is likely to have occurred in a magma reservoir. Second, the ubiquitous presence of partially resorbed phenocrysts is consistent with their evolution in a magma body whose composition evolved over a significant time period. Third, xenoliths are exceedingly rare in Haleakala Crater Kula lavas, which suggests that any such materials entrained in these magmas were filtered or settled out prior to eruption. A magma chamber would provide an effective filtering environment. Fourth, the geochemical cyclicity observed in the Haleakala Crater section must have resulted from a reproducible set of conditions that operated over a significantly long time frame. An established, long-lived magma chamber would provide a suitable environment. We rule out short-term variations in either source com-

Halemauu

Planioclase r.Ue!?

ti

m HK-I HK-13 HK-21 HK-22 HK-23 HK-25 HK-26 HK-27 HK-28 HK-29 HK-33 HK-36 HK-37 HK-39 HK-40 HK-43 HK-44 HK-45 HK-49 HK-50 HK-58 HK-48

lavas from

783

-

0.1 0.1 0.1 0.1

ph mph #

0.3
section

of Haleakala

AmDh ph/mDh

&&t &


-

-


-

-

0.2

:

-


phenocryst microphenocryst groundmass phase

position or partial melting as having produced the cyclic geochemical variations observed in the Haleakala Crater section. Cyclical variations in source composition seem unlikely, because temporal variations in the isotopic compositions of Kula lavas that are attributed to source variations (CHEN and FREY, 1985; WEST and LEEMAN, 1987) are continuous. Similarly, there is little evidence to support the existence of cyclical variations in the degree of partial melting over the duration of Hawaiian alkalic cap magmatism. A progressive decrease in the extent of partial melting during this period is consistent with the hypothesis that the heat source for these magmas decreased systematically as the focus of the Hawaiian hotspot became increasingly more distal. Moreover, simple partial melting variations cannot account for the drastic decreases in Ni, Cr, and SC contents that accompany interbatch compositional breaks. To establish the specifics of how the Haleakala magma chamber evolved and operated during the alkalic cap phase of the volcano’s evolution, two basic observations must be explained. What process or processes produced the sharp compositional breaks separating magma batches? What is the cause of the systematic compositional changes that occur within magma batches, in particular, the tendency to change from relatively evolved to less evolved compositions? Origin of Interbatch Compositional Fractionating

Breaks

phases and general,fiactionation

efects

Interbatch compositional breaks involve sharp decreases in compatible element abundances and concomitant sharp increases in incompatible element abundances and highly/

784

H. B. West and W. P. Leeman

Haleakala

60.

l:



.

a 50. _I 40

Although amphibole is present as rare microphenocrysts in a few of the mugearites, its role in the petrogenesis of these lavas is uncertain. For most elements, the effects of amphibole fractionation should be comparable to those of clinopyroxene. Thus, the removal of a small amount of amphibole relative to clinopyroxene probably would have produced no discernible geochemical effects. These lavas do not possess concave REE patterns which, based on distribution coefficients ( HIGUCHI and NAGASAWA, 1969; NICHOLLS and HARRIS, 1980; IRVING and FREY, 1984; GREEN and PEARSON, 1985), should be indicative of significant amphibole fractionation.

Crater

Least-squares modeling c&fractional crystallization

B

.

. .

.-.

.

.

.’

30

: .l * I

204

5

11.0

6

Zr;Nb

6

l

C

9.0 1

0, 7.0

+ .l .

I

5.0 3.0 “‘4

‘-yi

. I

9

P0,

.* ..-:. .‘: I

,,,,I

5

6

Zr/Nb a



lo

FIG. 3. (A) *‘Sr/*“Sr, (B) La, and (C) MgO vs. Zr/Nb for Haleakala Crater Kula Formation lavas. Zr/Nb ratios are strongly correlated with isotope ratios but not with absolute major and trace element contents or with ratios of these elements that are indicative of crystal fractionation or variations in the degree of partial melting.

moderately incompatible element ratios (Fig. 5 ). These geochemical effects are consistent with magmatic evolution that was dominated by fractional crystallization. The sharp upsection decreases in MgO, Ni, SC, and Cr contents probably reflect the fractionation of mafic phases such as olivine, clinopyroxene, and Cr-spinel. The sharp upsection increases in A1203/Ca0 ratios also support significant clinopyroxene fractionation. Upsection decreases in TiOz and V contents and increases in A1203/Ti02 ratios for most interbatch breaks are indicative of significant Fe-Ti oxide fractionation in the formation of the hawaiites and mugearites. Although Sr abundances increase upsection for most interbatch breaks, these breaks are also characterized by significant increases in Rb/Sr and Ba/Sr ratios. These effects, along with the decrease in Sr abundances at higher incompatible element contents are indicative of plagioclase fractionation. In general, P205 behaves incompatibly and increases upsection for the majority of interbatch breaks. However, some breaks are characterized by constant (e.g., batch 5 + 6) or decreased (e.g., batches 6 + 7,8 + 9) PzOs contents (Fig. 5), and significant increases in K20/P205 and Ba/P20S ratios. These geochemical effects, along with petrographic evidence, show that apatite fractionation must have played a role in the petrogenesis of the mugearites.

We used the least-squares approximation method to model quantitatively the role of fractional crystallization in producing the upsection geochemical breaks between magma batches. The fractionating phase assemblage used consists of olivine + clinopyroxene + plagioclase + Fe-Ti oxide ? apatite, which is consistent with the phenocrysts observed in these rocks (with the exception of amphibole) and with overall geochemical variations. Mineral compositions were taken from the literature and, in most cases, are from rocks collected from Haleakala Crater (Table 4). In addition to actual mineral compositions, both calculated equilibrium mineral compositions and a range of stoichiometric olivine and plagioclase compositions were tested. Although mineral endmember compositions for olivine (Fo and Fa) and plagioclase (An and Ab) were tested, their use resulted in either higher total residuals or contradictory results (e.g., subtraction of Fo and addition of Fa)

Haleakala Crater L 0.7034 * (III0.7033

5

A

.

l

* .

fi 0.7032 030.7031

.* .* ‘. l. 5

l

4

0.7030 I

Stratigraphic

Position-

Stratigraphic

Position-

FIG. 4. (A) *‘Sr/*% and (B) Zr/ Nb vs. relative stratigraphic position for Kula Formation lavas from the Halemauu section of Haleakala Crater. Arrow on x-axis points upsection. The systematic upsection decrease in *‘Sr/%r and Zr/Nb ratios reflects a systematic change in the composition of the source of these lavas. Isotopic data are from WEST and LEEMAN( I987 )

Petrogenesis of the alkali-rich lavas on Haleakala

80 70 60 p 50 40 30 20

Stratigraphic Position-

Stratigraphic Position-

Stratigraphic Position-

40

2 3.0 Q 2.5 62.0

-0.6

0

CO.6

z1.5 1.o

0.4 0

Stratigraphic Position-

Stratigraphic Position-

56 54

8 FE.0

N52 g :;

g6.0

a

46 44 42

4.0 -,

Stratigraphic Position-

Stratigraphic Position-

Stratigraphic Position-

Stratigraphic Position-

Stratigraphic Position-

Stratigraphic Position-

12 10 06 $6 4 2

FIG. 5. Major and trace element contents and ratios vs. relative stratigraphic position for Kula Formation lavas from the Halemauu section of Haleakala Crater. Arrow on x-axis points upsection. Vertical lines delineate geochemical discontinuities separating individual magma batches. Between batches, there are sham increases in incompatible element contents and in major and trace element ratios sensitive to crystal fractionation and complementary sharp decreases in compatible element contents. In contrast, within batches incompatible element contents systematically decrease upsection and compatible element contents systematically increase upsection.

Numerous crystal fractionation models were tested for each of the three interbatch breaks representing the largest compositional changes (i.e., batch 7 --t 8, batch 9 + 10, and batch 12 + 13). For the batch 7 + 8 break, the stratigraphically highest and least evolved lava from batch 7 (alkalic basalt HK-36) was chosen as representing the parent magma, and the stratigraphically lowest and most evolved lava from batch 8 (mugearite HK-37) was chosen as representing the daughter magma. The model that produced the lowest sum of squares of the residuals (B R* = 0.039; Table 5) indicates that the mugearite could have formed by x53% crystallization of the alkalic basalt parent magma by removal of clinopyroxene (23%) + plagioclase ( 15%) + magnetite (9%) + olivine (6Y0) + apatite (2%). In order to produce an acceptable fit (i.e., B R2 < 0.1; WRIGHT, 1974), it was necessary to use a more evolved olivine composition (FoeO) than the calculated equilibrium olivine composition of either the parent ( FoT6) or daughter ( Fo6,) magma. Similar crystal frac-

tionation models tested for the batch 9 + 10 and 12 + 13 breaks produced essentially equivalent results, although total residuals were higher (i.e., Z R2 > 0.3). The consistency of the least-squares major element modeling was tested by using the calculated extent of fractionation to estimate the trace element composition of the daughter magma. Representative results are included in Table 5. Calculated daughter concentrations for many trace elements are close to observed concentrations (Fig. 6), which suggests that fractional crystallization was the dominant process operating during periods of volcanic quiescence. However, discrepancies exist between observed and calculated daughter magma compositions, particularly for highly incompatible trace elements and highly/ moderately incompatible element ratios (e.g., La/Sm). For example, in the batch 7 + 8 break, the observed enrichment of highly over moderately incompatible trace elements for the daughter magma composition cannot be reproduced using the 53% crystallization estimate derived

H. B. West and W. P. Leeman

786

3. Range in rock compositions in magma batches from the Halemauu section of Haleakala Crater. Batch numbers are the same as shown in Fig. 5. Oldest lava refers to first lava erupted in each batch (lowermost, stratigraphically) and Youngest lava refers to the last lava erupted in each batch (uppermost, stratigraphically). Table

Batch

# Samoles 2 4 3

2 3 4 5 6

1

: 9 10 11 12 13

6 3 12 4 5 3

Oldest lava

Younaest lava

hawaiite hawaiite hawaiite hawaiite mugearite mugearite mugearite mugearite nepheline nepheline nepheline nepheline nepheline

alkalic basalt alkalic basalt alkalic basalt hawaiite mugearite basanitoid nepheline hawaiite basanitoid nepheline hawaiite alkalic basalt basanitoid/ankaramite basanitoid

mugearite hawaiite hawaiite hawaiite hawaiite

Summary of total number OJ batches exhibiting the observed variations in rock type Jor the Halemauu section. Rock Tvue Variation

# batches

hawaiite mugearite mugearite hawaiite mugearite

6 2

1

2

I

-+ alkalic basalt/basanitoid/ankaramite - alkalic basalt/basanitoid/ankaramite + hawaiite -t hawaiite -t mugearite

from major element modeling. This result holds even for removal of an equivalent amount of a monomineralic cumulate consisting of clinopyroxene, the mineral phase in these lavas most capable of increasing La/Sm ratios during crystal fractionation (Fig. 7). In addition, the batch 5 -+ 6 and batch 9 + 10 breaks involve roughly equivalent enrichments in highly ( Rb, Th, Nb, LREE) and moderately (Y, HREE) incompatible elements (Fig. 8), which is inconsistent with the predicted behavior of those elements during fractional crystallization of the observed phenocryst phases.

Additional processes operating during periods Qferuptive quiescence The inability of models based on major element variations to account for observed incompatible trace element emichments, as in the case for Haleakala Crater lavas, has also been found for alkalic cap lavas from Mauna Kea (WEST et al., 1988 ) and indicates the operation of processes other than or in addition to simple closed-system fractional crystallization. Potential processes that could account for the apparently ex-

Table 4. Parent-daughter lava compositions and mineral compositions used in least-squares calculation for the batch 7 -t 8 transition, given in Table 5. Major element compositions were first normalized to 100% on an anhydrous basis with all iron converted to FeO.

SiO, Al&‘, Fe0 MgO CaO Na,O KG’ PZOS TiO, MnO crso,

44.17 14.63 14.16 6.66 10.53 3.04 1.00 0.49 3.62 0.21 -

01 Plag

Olivine Foe,. Plagioclase phenocryst in hawaiite from the Halemauu section of Haleakala Crater (Macdonald and Powers, 1968). Clinopyroxene phenocryst in picritic basalt from upper west wall of Haleakala Crater (Macdonald and Powers, 1946). Magnetite from Hualalai alkalic basalt (BVSP, 1981). Apatite (Deer et al., 1966).

CPX Mt AP

37.42 28.20 34.37 -

47.40 32.80 0.59 16.50 2.10 0.10

47.70 6.82 7.45 13.34 21.35 0.65 0.03 1.89 0.16 0.23

1.44 68.30 3.18

0.21 0.54 52.40

1 22.70 0.76 I .56

40.98 1.52

51.80 17.40 9.24 2.87 6.52 5.44 2.19 0.96 2.30 0.26

Petrogenesis of the alkali-rich lavas on Haleakala

787

Table 5. Results of major element least-squares modeling for the batch 7 + 8 compositional break. Parental magma = HK-36, daughter = HK-37. Also shown are observed and calculated concentrations for selected trace elements. Trace element mineral partition coefficients are taken from LeRoex and Erlank (1982). Compositions of subtracted mineral phases are given in Table 4. ca!.G

Qb

SiO, A’,% Fe0 M8O CaO Na,O KzO PsO5 TiO,

51.80 17.40 9.24 2.87 6.52 5.44 2.19 0.96 2.30

51.78 17.40 9.27 2.87 6.55 5.58 2.11 0.92 2.22

Th La Sm Yb Ba Zr Nb Y SC

6.0 64.4 13.15 3.21 849 445 72 43 5.1

5.5 56.1

Residuals

ComoonentProoor.

0.01603 0.00327 -0.02740 0.00504 -0.03328 -0.14761 0.08327 0.04262 0.08154

HK-36

1.ooooo

Olivine Plug CPX

0.05607

0.14830 0.23279

Mf

Apatite

0.09477 0.00 167

HK-37

0.46641

C R2 = 0.03934

13.81 3.31 964 434 51 42 6.3

cessive incompatible element enrichments associated with upsection interbatch breaks include the following. ( 1) During periods of eruptive quiescence, the Haleakala magma chamber was not a closed-system. (2) A process such as periodic mixing of evolved boundary layer liquids back into the main magma chamber operated. (3) Magma stored in the magma chamber assimilated or accumulated phenocrysts. Open-&stern Fractionation. Periodic replenishment of an evolving magma chamber by mafic recharge magmas potentially can decouple major and trace elements ( O’HARA, 1977; O’HARA and MATTHEWS, 198 1). We quantitatively tested the possibility that incompatible element enrichments accompanying interbatch breaks result from open-system frac-

tional crystallization. Closed- and open-system liquid lines of descent were calculated using the method of NIELSEN ( 1988) for four potential parental magma compositions, represented by the most mafic (and stratigraphically highest) basalts from four batches (HK-2, HK-36, HK-50, and HK58 ) (Fig. 9 ) . If these basalts represent suitable parental magma compositions for the hawaiites and mugearites, then the compositions of potential daughter magmas (i.e., the stratigraphically lowest and most evolved lavas in the batches di-

. 5 .o -

Batch 7 to 8 Break

HK-37 0

4 .5 -

o HK-36 0 HK-37 *

Calculated

E ul 34

.o -

3 .5 10-

Rb

K

Th

Ba

Ta

Nb

FIG. 6. Chondrite-normalized

La

Ce

Nd

Sr

Sm i!r P Y Lu Eu Hf Ti Yb

trace element diagram for the batch

7 -+ 8 upsection break. The potential parent magma, basalt sample HK-36 (stratigraphically highest in batch 7B), and potential daughter magma, mugearite sample HK-37 (stratigraphically lowest in batch 8). are indicated by solid lines. The calculated trace element abundance pattern is based on the amount of fractionation derived from least-squares major element modeling is indicated by the dashed line. Distribution coefficients used in the trace element calculations are from DRAKE and WEILL, 1975; DUNN and MCCALLUM, 1982; FuJIMAKI et al., 1984; GREEN and PEARSON, 1985; GRIFFIN and MURTHY, 1969; HIGUCHI and NAGASAWA,1969; HENDERSON,1982; IRVING and FREY, 1984; LEEMAN et al., 1978; LEROEX and ERLANK, 1982; LINDSTROM, 1976; MCKAY and WEILL, 1977; MCKAY et al., 1986; ONUMA et al., 1968: PASTER et al., 1974; PEARCE and NORRY, 1979; WATSON and GREEN, 198 1.

3 .O30

I

40

I

I

50

60

I

70

I

Nb

FIG. 7. Calculated fractionation paths for the batch 7 --* 8 upsection compositional break. Alkalic basalt HK-36 was chosen as the potential parent magma and mugearite HK-37 as the potential daughter magma. The shaded field encompasses the range of calculated daughter compositions using the mineral proportions and degree of crystallization estimated by least-squares major element modeling. Two sets of partition coefficients were used that represent extremes in DcP” values for Nb, La, and Sm. Solid curves are monomineralic fractionation paths. Small dots on the curves represent crystallization increments of I%, 5%, IO%, 20%, etc. The larger than predicted La/ Sm ratio of HK-37 indicates that other processes capable of fractionating these elements operated during the periods of eruptive quiescence that separate magma batches.

H. B. West and W. P. Leeman

788

Batch

c i: I “;

5 -* Batch 6

1.5

1.0

z 0.5

A HK-21 /;-;‘----a’



IIIIlliI Rb

K

P

Th

Nb

La

Ce

i Zr

I I I 11 II 1 Sm Sr HI Y Ba Eu Ti Yb

FIG. 8. Relative enrichment factors for the batch 5 + 6 and batch 9 + 10 upsection interbatch breaks. Plotted are the compositions of the stratigraphically lowest lavas in batches 6 and 10 (potential daughter magmas) relative to the compositions of the stratigraphically highest lavas in the underlying batches 5 and 9 (potential parental magmas). Similar enrichments in highly (e.g., Th, Nb, LREE) and

extents than can magma in the interior of the chamber. Also, the fractionating phase assemblage in the cooling boundary layer can differ substantially from that in the interior of the magma chamber. Periodic remixing of evolved boundary layer liquids back into the interior of the chamber potentially can produce a decoupling between major and trace elements, resulting in higher incompatible element abundances and highly/moderately incompatible element ratios than normally would be produced. Perhaps the amphibole found in some Haleakala Crater mugearites formed in a volatile-rich boundary layer. We quantitatively tested the possibility that in situ crystallization produced the upsection increases in incompatible element abundances and highly/moderately incompatible element ratios. The three most evolved rocks collected from Haleakala Crater (mugearites HK-43 and HK-26, and the Holua trachyte) were used as potential boundary layer liquid compositions. A representative result for the batch 7 + 8 break is shown in Fig. 10.

Open System Fractionation 55 -

o

8 0

0

8”

0

moderately (e.g.. Y. HREE) incompatible elements indicate that simple closed-system crystal fractionation could not have produced the interbatch compositional breaks.

overlying these basalts: HK-5, HK-37, HK-5 1, HK59) should lie along the calculated liquid lines of descent. Both open- and closed-system liquid lines of descent for all four potential parental compositions resulted in excessively high MgO and excessively low Si02 contents (Fig. 9). Similar discrepancies are obtained for other major (e.g., A1203) and trace elements (e.g., SC). These results suggest that the basalts do not represent magmas parental to the hawaiites and mugearites. Three caveats to this interpretation are worth mentioning. It is possible that processes operating in the magma chamber were considerably more complex than those modeled. Also, the method used to calculate the liquid lines of descent is based on one-atmosphere experimental results, which may not reflect the actual pressure-temperature conditions of fractionation for these magmas. Unfortunately, existing data for the partitioning of elements in basaltic liquids at high pressures are insufficient to allow liquid lines ofdescent to be calculated accurately for a full range of pressure-temperature conditions. In addition, the starting compositions used in the 1-atm experiments may be inappropriate for Kula lavas. Mixing of Boundary Layer Liquids. LANCMUIR (1989) showed that in situ crystallization within the boundary layer of a convecting magma chamber potentially can produce geochemical effects significantly different from those imposed by simple closed- or open-system fractional crystallization. Boundary layer liquids are isolated from the main convecting mass of the magma chamber and can evolve to much greater rectly

HK-58

\

6-

t 3-

20

I

I

40

I

I

60

I

I

60

Nb FIG. 9. Open- and closed-system liquid lines of descent for four potential parental magma compositions represented by the stratigraphically highest basalts in magma batches I (HK-2), 7 (HK-36), 9 (HK-50), and 12 (HK-58). Equilibrium phase assemblages and liquid lines of descent were calculated using the method of NIELSEN ( 1988 ). Solid curves represent I-O-O (recharge-assimilation-eruption relative to fractionation) steady-state mass open-system models in which recharge is equal to fractionation. Dashed curves represent closed-system fractional crystallization (O-O-O model). Inflections in the curves reflect the disappearance and appearance of phases in the calculated equilibrium mineral assemblage. The initial increase in MgO for sample HK-2 reflects the absence of olivine in the calculated equilibrium phase assemblage. The poor fit between observed data and calculated liquid lines of descent indicates that these basahs probably do not represent magmas parental to the hawaiites and mugearites.

Petrogenesis of the alkali-rich lavas on Haleakala

In situ

Crystallization

5.0 -

Boundary Layer

HK-37 0

Liquids

Holuatrachyte

-

,/.

HK-26 mugearite HK-43 mugearite

-------

,/

4.5 E c/Y

,’ ,’

2

,’

1

,’

,’

,’

,/;;K-43

HK-26 ,I

,/

4.0 -

Holua 3.5 -

HK-36

30

I

I

40

50

I 60

I

789

dant petrographic evidence to support partial resorption of plagioclase, olivine, and amphibole in Haleakala Crater mugearites. However, resorption of phenocryst phases could not have produced the upsection increases in incompatible element contents and highly/moderately incompatible element ratios or the comparable enrichments in highly and some moderately incompatible elements. None of the Haleakala Crater Kula basalts, hawaiites, or mugearites have Eu anomalies, thus precluding significant resorption of plagioclase. Resorption of mafic phenocryst phases (e.g., olivine, clinopyroxene, amphibole) would increase compatible element contents (e.g., Ni, SC, MgO). However, the mugearites have low MgO ( <3 wt%) SC, and Ni ( -c IO ppm) contents. Clearly, phenocryst accumulation was insignificant in these magmas because evolved Haleakala Crater lavas are essentially aphyric.

70

Nb FIG. 10. In situ crystallization paths for the batch 7 4 8 upsection compositional break calculated after the method of LANGMUIR( 1989) for three potential boundary layer liquids: mugearites HK-43 and HK-26 and the Holua trachyte (Holua data from WEST and LEEMAN, in prep.). The potential parent magma (basalt HK-36) is indicated by the upside-down triangle; the potential daughter magma (mugearite HK-37) is indicated by the open square. Shown for comparison is the closed-system fractionation path for the extent of fractionation derived from least squares major element modeling (open circle).

The calculated fractionation path using the HK-43 mugearite boundary layer liquid approaches the postulated HK-37 daughter magma composition for the amount of fractionation predicted by major element modeling (Fig. IO). However, in situ crystallization models produced inconsistent results. For example, the increase in La/Sm for the batch 7 + 8 break (for the HK-43 boundary layer liquid) requires 40-50% of the boundary layer to be mixed back into the main magma chamber. In contrast, calculated compatible element abundances (e.g., SC) for all three modeled boundary layer liquids are consistent with remixing -70% of the boundary layer. Mixing involving the HK-26 boundary layer liquid results in increased La/Sm but decreased Nb/Sm, a result that clearly is contradictory because Nb is more incompatible than is Sm for basaltic magmas. Mixing of boundary layer liquids having the modeled compositions cannot account entirely for the upsection relative enrichments in incompatible elements. However, such a process could potentially have contributed to the observed enrichment and may account for the increase in La/Sm beyond that predicted by fractional crystallization. It should be noted that all of the potential boundary layer liquids modeled can produce the observed increase in La/Sm at some value off, (where fais the fraction of boundary layer liquid mixed back into the main magma chamber). However, HK-43 produces the most consistent results, i.e., the samef, for several trace elements and ratios. Finally, it is possible that none of the compositions modeled are representative of a Haleakala magma chamber boundary layer liquid. There are no obvious constraints on the compositions of boundary layer liquids as none have been documented in basaltic magmatism. Phenocryst Re.wrption and Accumulation. There is abun-

The Origin of Intrabatch Compositional

Variations

In contrast to the sharp upsection compositional breaks between magma batches, geochemical variations within batches generally change systematically upsection (Fig. 5). During the eruptive periods represented by these magma batches, lavas typically became increasingly more mafic with time, following a mugearite + hawaiite + basalt sequence. One possible explanation for the observed eruptive sequence and the systematic upsection geochemical variations within magma batches is that these lavas represent mixed magmas. Magma mixing is supported by the ubiquitous presence of partially resorbed phenocrysts in the most evolved Haleakala Crater Kula lavas. Trace element enrichment patterns for lavas from batches 7 and 9 are depicted in Fig. I I. Lavas from these batches span large compositional ranges compared to most other batches. Crossing trace element patterns for the most evolved (oldest) and least evolved (youngest) lavas within individual batches preclude the formation of the intermediate composition lavas by simple mixing between compositionally extreme endmember magmas. Other batches display similar crossing patterns. The points where the trace element patterns cross typically occur at elements that behave compatibly during fractional crystallization of clinopyroxene, plagioclase, Fe-Ti oxides, and apatite (e.g., Sr, Eu, Ti, P). This suggests either that fractional crystallization was concurrent with magma mixing during eruptive periods, or that recharge magmas were considerably more mafic than are the basalts in these batches. To test the relative effects of crystal fractionation and magma mixing during eruptive periods, compositional trends were calculated for several batches for both simple magma mixing and more complex cyclic, open-system evolution (i.e., recharge + eruption + magma mixing + crystal fractionation). A representative result is shown in Fig. 12. Because trace element and isotope systematics show that in general the source composition for Kula lavas changed systematically with time (CHEN and FREY, 1985; WEST and LEEMAN, 1987), we did not use a single recharge magma composition to model all batches. Instead, the least evolved (i.e., youngest) basalts from each batch were chosen as potential recharge magmas, with the exception of batch 8. To model batch 8, we used

H. B. West and W. P. Leeman

790

I/II

I

'Rb

P K

I Nb

I

Th

I II I CeZrSmgaSr

La

I

I Eu

I Hf

I TI

I Y

I

I

Yb

FIG. 1 I. Enrichment factors for lavas in batches 7 and 9. Lava compositions are normalized to the stratigraphically highest basalt in each batch (HK-36 and HK-50, respectively). Sample numbers are indicated next to each pattern. (A) Crossing patterns for the mugearites and basalts in batch 7 indicate that the mugearites could not have formed by mixing of an HK-36 composition recharge magma with a more evolved magma composition. Crossing patterns in (A) and (B) show that intermediate composition lavas could not have been produced by simple mixing of potential recharge magmas (HK36 and HK-50) with resident magmas having the compositions corresponding to the earliest erupted mugearites (HK-26 and HK-43).

the most mafic batch because

the

least

7 basalt

evolved

as the potential

lava

from

batch

parent

where C,, = final concentration in the magma chamber, Xrc = mass fraction of mixed magma crystallized relative to the mixed magma chamber mass, and D = bulk partition coefficient. The cyclic, open system model depicted in Fig. 12 represents 10% recharge, 10% eruption, and 5% fractional crystallization per cycle, and is meant to illustrate one possible solution. In reality, there is no control on the relative proportions of recharge, eruption, and crystallization, and these proportions undoubtedly varied during eruptive (recharge) periods. Results show that calculated liquid compositions for cyclic, open-system evolution differ significantly from those for simple binary mixing. Batch 8 lava compositions and those for other batches do not fall along a binary mixing curve. but instead deviate towards the cyclic, open-system path. We interpret these results as indicating that during eruptive periods recharge was episodic. Between individual recharge events, there may have been sufficient time for fractional crystallization to have operated to some extent. A caveat to this interpretation is that some crystal fractionation could have occurred during ascent from the magma chamber via plating along conduit walls. Haleakala Crater Primary and Parental Magma Compositions None of the Haleakala Crater Kula basalts are likely representative of primary magmas (i.e.. Mg# > 70, MgO > 12 wt%; cf. BVSP, 198 I ). With the exception of ankaramite HK-58 (Mg# = 60. MgO = I 1.3 wt%), which appears to be

magma,

8 is a hawaiite.

However, because these two batches are adjacent, recharge magma compositions probably were similar. For the purposes of modeling an evolving magma chamber undergoing recharge, we assumed that eruptive periods are characterized by recharge-eruption-mixing-fractional crystallization cycles. At the start of an eruptive period, the composition of the magma chamber is taken to be that of the oldest lava erupted (i.e., lowermost stratigraphically). Following expulsion of this magma, basalt recharge magmas mix with the remaining magma in the chamber. At this point, the concentration of any element in the mixed magma can be calculated using the binary mixing equation

5.0-

4.5-

Model .lO-.lO-.05 HK-37

v Rechargemagma 0 Residentmagma ---Open systempa - MlXl"Q paw

E c? ? 4.0-

3530

I 40

I 50

I 60

1 70

Nb

in the mixed magma, C;. = conwhere C,,, = concentration centration in the recharge magma, C’i = concentration in the magma chamber prior to a recharge episode, X, = mass fraction of recharge magma relative to the mixed magma chamber mass, and X,,,, = mass fraction of the magma chamber prior to recharge relative to the mixed magma chamber mass = ( 1 - X,). As a simplifying assumption, C, and X, are assumed to be constant. The mixed magma then evolves by fractional crystallization until the next influx of recharge magma, at which point the concentration of an element in the magma chamber at the end of a cycle is given by

FIG. 12. Calculated intrabatch evolutionary paths for batch 8 for magma mixing vs. open-system behavior. Shown is a curve representing simple binary mixing between a potential recharge magma composition (basalt HK-36) and an evolved resident magma composition (mugearite HK-37). Also shown is a path representing opensystem evolution during a period of magma chamber recharge. The open-system path consists of a series of interspersed curves that depict magma chamber recharge as a series of cycles of recharge-eruptionmagma mixing-crystal fractionation, The open-system path represents 10% recharge, 10% eruption, and 5% fractional crystallization per cycle relative to the volume of the magma chamber. Batch 8 lava compositions (diamonds) deviate from the binary mixing path towards the open-system path, implying that during eruptive periods recharge is not continuous. but is instead episodic to some extent. allowing time for a small amount of fractional crystallization to occur.

Petrogenesis of the alkali-rich lavas on Haleakala accumulative, the most mafic basalts have Mg# i= 48 and MgO < 7 wt%, despite the presence of olivine phenocrysts. Because these lavas also contain relatively low Ni, SC, and Cr abundances, it is apparent that even the most mafic basalts have undergone significant olivine + clinopyroxene F spine1 fractionation. Given the relatively large number of Kula lavas that have been sampled and characterized geochemically (MACDONALD and POWERS, 1946, 1968; CHEN and FREY, 1985; CHEN et al., 1990; this study), it is possible that Kula primary magmas were never erupted. If the most mafic Haleakala Crater Kula basaits represent recharge magmas, then they must have undergone crystal fractionation prior to injection into the shallow-level Haleakala magma chamber. Kula lavas define a broad field intermediate to the low ( 1 bar) and high (8-30 kbar) pressure three-phase cotectics for the olivine-diopside-nepheline-plagioclase system (Fig. 13) and, therefore, appear to contain components of magmas that underwent crystal fractionation over a range of pressures. It is possible that the least evolved basalts were derived from primary magmas that underwent an earlier episode of crystal fractionation, probably in a deeper-level magma chamber. Similar results were found for Kula lavas from the flanks of Haleakala ( CHEN et al., 1990). In contrast, aikalic cap lavas from Mauna Kea (dominantly hawaiites) all lie near the 8-30 kb cotectic and probably formed by clinopyroxene-dominated crystal fractionation at moderate (2-5 kb) pressures (WEST et al., 1988). A Model for the Compositional Alkalic Cap Magma Chamber

791

Di

+ Pl

1

01

V



V

\

Nf?

FIG. 13. Major oxide compositions of Haleakala Crater Kula Formation lavas projected from plagioclase onto the olivine-diopsidenepheline plane following the convention OfSACKet al. ( 1987). The low pressure ( 1 bar) cotectic is based on experimental data from SACK et al. ( 1987). The high pressure (8-10 kbar) cotectic is based on high pressure experimental results for liquids coexisting with the assemblage olivine J- clinopyroxene + orthopyroxene (data sources listed in SACK et al., 1987). These favas form a field intermediate between the low and high pressure cotectics, indicating that these lavas could have formed by polybaric fractionation,

Evolution of the Haleakala

Relative to the Hawaiian shield-building stage, the alkalic cap/post-shield alkalic stage corresponds to a period of diminished and sporadic eruptions (MAONALD et al., 1983). The lower eruptive frequency is likely a direct result of a decrease in magma flux from the source due to decreasing amounts of melting as the lithosphere moves away from the focus of the Hawaiian hotspot. The reduction in the volume of generated melts probably resulted in longer residence times for magma in the Haleakala magma chamber. The increase in magma chamber residence times appears to be responsible for the wider spectrum of lava compositions during alkalic cap magmatism than generally is produced during tholeiitic shield-building activity. The cyclic geochemical variations delineated by Kula lavas from Haleakala Crater reflect reduced magma recharge rates associated with the waning magma flux characteristic of the Hawaiian alkalic cap stage. An idealized model for magmatic evolution within the Haleakala magma reservoir system follows. Periods of eruptive quiescence during the alkalic cap stage of Haleakala occurred when magma chamber recharge rates were negligible (Fig. 14a). Prolonged residence in the Haleakala magma chamber allowed magmas to undergo extensive fractional crystallization, during which time the magma chamber evolved from that of the recharge magma composition (i.e., basalt) to a more evolved composition (e.g., mugearite). During this period, A1203/Ca0 ratios increased because of crystal fractionation of clinopyroxene (Fig. 15). When a new period of substantial magma chamber recharge

was initiated, eruption resumed (Fig. 14b). The earliest lavas erupted during these periods were hawaiites or mugearites. Following partial evacuation of the magma chamber, recharge magmas mixed with the remaining resident magma (Fig. 14~). Continued recharge and mixing caused the compositions of erupted lavas to become systematicaIly more mafic with time as the proportion of evolved resident magma decreased (Figs. 14d-e ). During this period, the A1203/CaO ratios of erupted lavas decreased because of the increasingly larger proportion of recharge magma in the magma chamber (Fig. 15). As recharge and mixing proceeded, the bulk composition of the magma chamber approached that of the recharge magma (Figs. I4f, 15). Thus, the youngest (and most mafic) lavas erupted within each magma batch provide the closest available approximations of recharge magma compositions for Haleakala Crater. Together, these events led to the formation of both the sharp upsection compositional breaks between batches and the systematic upsection compositional variations within batches. Differences in the magnitude of interbatch breaks may be related to variations in recharge rates throughout the Haleakala alkalic cap stage. This resulted because recharge rates ultimately control the extent of fractionation. Thus, the periodicity of major changes in recharge rates during the alkalic cap phase of magmatism could have created the geochemical cyclicity exhibited by lavas exposed in the Haleakala Crater section. The results of this study suggest a conceptual model that is applicable to the life cycle of any volcano. The processes

792

H. B. West and W. P. Leeman

(

i

FIG. 14. Schematic depicting the cyclic, open-system evolution of the Haleakala magma chamber. (A) Prior to the initiation of a recharge period, the magma chamber contains evolved magma (e.g., mugearite). (B) The influx of basalt recharge magma results in evacuation of a portion of the magma chamber, followed by (C) mixing between recharge magma and evolved magma remaining in the chamber. (D) Continued recharge by mafic magma results in the eruption of systematically less evolved hybrid magmas. (E) The proportion of the evolved component decreases progressively with time as the result of continual recharge by mafic recharge magma. (F) Eventually, the bulk composition of the magma chamber approaches the composition of the recharge magma.

sioned that at any given point in time, depending on prevailing magma recharge rates, a magma chamber can follow either a fractionation cycle (recharge absent or negligible) or an eruption cycle (high recharge rates) (Fig. 16). During periods of low or negligible magma recharge rates, magma chambers can either undergo protracted crystal fractionation (i.e., insignificant recharge; the FC cycle in Fig. 16), or follow a fractionation cycle (i.e., recharge + magma mixing + fractional crystallization + recharge). As long as recharge rates remain low enough that magmatic pressure does not exceed lithostatic pressure, the magma chamber would remain in the fractionation cycle. Note that in the absence of magma chamber recharge, eruption potentially could occur if the density of resident magma falls below a critical density threshold due to extensive fractional crystallization or volatile exsolution. For example, BARDINTZEFF and BONIN ( 1987) suggested that volatile exsolution following amphibole resorption in evolved magmas can trigger eruptions. During periods of high magma recharge rates, a magma chamber can follow either an eruption cycle (recharge + eruption + magma mixing + recharge), if the recharge rate is high enough that fractional crystallization is negligible, or a recharge + eruption + magma mixing + fractional crystallization -+ recharge path, if the periods between individual recharge events are long enough that fractional crystallization effects are imposed (Fig. 16). Within an eruption cycle, recharge rates could decrease temporarily to a point where the

identified at Haleakala (source variation, crystal fractionation, magma recharge, and eruption) are common processes that can interact to various degrees to control the compositional evolution of any magma chamber (Fig. 16). It can be envi-

Eruption

lnterbatch

lntrabatch

Cycle

0

Fractionation

0

s :: 7

Stratigraphic

Position

+

FIG. 15. Generalized diagram showing Al,OJCaO vs. relative stratigraphic position for a single magma batch. During periods of eruptive quiescence, magma chamber recharge rates are negligible and magma chamber evolution is dominated by fractional crystallization (FC). During this period, the composition of the magma chamber evolves from that of the recharge magma (basalt) to that of the most evolved lavas erupted (e.g., mugearite). During periods of high magma chamber recharge rates, an eruption cycle is initiated in which mafic recharge magmas mix continuously with evolved magma in the chamber. During eruptive periods, magma mixing effects dominate those imposed by fractional crystallization, and with time the compositions of erupted lavas approach that of the recharge magma.

Low

Magma

Flux

Regime

High

Magma

Flux

Regime

FIG. 16. Conceptual diagram illustrating possible evolutionary pathways between periods of low and high magma chamber recharge rates. In the low magma flux regime, recharge rates are either negligible or very low and fractional crystallization (the fractionation cycle) dominates over magma mixing. During these periods, magma resident in a magma chamber may undergo protracted fractional crystallization in the absence of recharge, or may undergo a series of mixing-fractionation events if a small amount of recharge occurs. In the high magma flux regime, recharge rates are high and an eruption cycle is initiated. During these periods, magma mixing dominates over fractional crystallization and a series of recharge + eruption + mixing events may occur. If magma recharge temporarily ebbs, fractional .^ crystalhzation may become more sigmhcant.

Petrogenesis of the alkali-rich lavas on Haleakala magma chamber passes into the low magma flux regime. This could result in the interspersion of a series of fractionation cycles within an eruption cycle. It is possible that the small compositional subsections delineated by certain major and trace elements in batch 7 were created by such fluctuations in magma recharge rates (Fig. 5a). Eruption Rate Estimates Based on Magma Batches Assuming that the thirteen magma batches exposed in the Haleakala Crater section represent discrete eruptive intervals, a minimum eruption rate can be calculated if the time span represented by the section can be estimated. Although age dates are unavailable for the lower part of the section, the maximum age must be less than that of the stratigraphically lower Kumuiliahi Formation. We used two estimates for the upper age limit of the section: ( 1) 0.56 + 0.14 Ma, the age of a lava located at Kalahaku on the upper rim of the west wall of Haleakala Crater, and (2) 0.49 f 0.15 Ma, the age of the Holua dike, which cuts the northwest wall of Haleakala Crater ( K-Ar ages from NAUGHTON et al., 1980). Based on these ages, the Halemauu section could represent as much as 140.000-2 10,000 years of eruptive activity. This time interval yields an average of 1 eruption per N 11 ,OOO16,000 years, which is a significantly lower eruptive frequency than are estimates made for the alkalic cap stages of Mauna Kea ( 1 eruption per 1,250-1,500 years; WEST et al., 1988) and Kohala ( 1 eruption per 1,900 years; SPENCLER and GARCIA, 1988). One explanation for this apparent discrepancy is that the maximum age estimate used for the base of the Haleakala Crater section is considerably too old. A paleomagnetic survey of this section (R. Coe, pers. commun.) suggests that the time span represented by these lavas is shorter than that inferred by existing K-Ar data, perhaps by as much as an order of magnitude. Based on the number of observed reversals (produced by secular variations) in the section and estimates of the time required for such changes in the geomagnetic field, the Haleakala Crater section spans ~26,400 years. The calculated minimum eruption rate using this age estimate is 1 eruption per -2,000 years, which is comparable to, although still slightly lower than, those estimated for Mauna Kea and Kohala. Eruption rate estimates for alkalic cap magmatism at Haleakala, Mauna Kea, and Kohala are significantly lower than the alkalic post-shield stage estimate of 1 eruption every 501000 years suggested by CLAGUE ( 1987a). In all likelihood, this is not a discrepancy; the estimated eruption rate for Haleakala (as well as Mauna Kea) is a minimum value, because it is based on the eruptive frequency at a specific area of the volcano and, therefore, is not a volcano-wide eruption rate. Predictably, these estimated alkalic cap eruption rates also are considerably lower than those inferred for early alkalic cap volcanism on Hualalai ( 1 eruption per 50 years; CLAGUE, 1987a; MOORE et al., 1987) and for active shield-building volcanism on Kilauea ( 1 eruption per 1-4 years; KLEIN, 1982; DZURISIN et al., 1984). Haleakala alkalic cap lavas span a larger compositional range (alkalic basalt to trachyte) than do alkalic cap lavas from Kohala (mugearite to trachyte; SPENGLER and GARCIA,

793

1988) or Mauna Kea (hawaiite with very little mugearite; WEST et al., 1988), and they contain a larger proportion of evolved lava compositions. It may be significant that, in addition to these characteristics, Haleakala has the lowest calculated minimum eruption rate of these three volcanoes. Notably, Mauna Kea, which contains the smallest proportion of evolved lavas of the three alkalic caps, has the highest eruption rate. Thus, the proportion of evolved rock types within a particular alkahc cap sequence may be to some extent a function of magma supply rate. However, additional and more precise age measurements for Haleakala are necessary to better test of the prediction that lower eruption rates (and, by extension, lower magma chamber recharge rates) produce larger proportions of more evolved rocks. Implications for Hawaiian Alkalic Cap Magmatism The magnitude of upsection interbatch compositional breaks is larger for the upper (younger) part of the Haleakala Crater section than for those in the lowermost (older) part. For example, in the lower six interbatch breaks, K,O increases by an average of x38%, whereas in the upper six breaks, K20 increases by an average of -89%. Similar increases in magnitude for lower- vs. upper-section interbatch breaks occur for compatible trace element abundances and many trace element ratios. Because the magnitudes of these breaks ultimately are controlled by the extent of crystal fractionation, these increases imply that the periods of eruptive quiescence separating eruptive periods became progressively longer, which was most likely the result of a progressive overall decrease in magma chamber recharge rates. This overall decrease in magma flux probably was a direct consequence of waning heat production as the Pacific plate moved away from the underlying Hawaiian hotspot. The waning magma flux associated with evolution during Hawaiian alkalic cap magmatism has led to speculation that the depth of magma storage during this period also undergoes a significant change. High magma production rates during the Hawaiian shield-building stage are capable of sustaining shallow-level magma chambers because reservoirs are continually replenished. In contrast, reduced magma production rates during the alkahc cap stage may result in the stagnation of melts at deeper levels (e.g., at the crust-mantle interface) and the solidification of shallow-level magma chambers (CLAGUE, 1987a; FREY et al., 1990). For example, at Mauna Kea, the older post-shield basalts appear to have evolved at lower pressures (FREY et al., 1990), which suggests they resided in a shallow-level magma reservoir. In contrast, the younger Mauna Kea hawaiites evolved at moderate pressures (~2-5 kb; WEST et al., 1988), reflecting evolution in a deeperlevel magma reservoir. Strontium and lead isotope ratios undergo the greatest changes in the upper part of the Haleakala Crater section relative to the lower part (Fig. 4). This may have resulted from an increasingly protracted stagnation of melts at depth, which enhanced assimilation of MORB composition wallrock. Such a process could have occurred in a deeper-level magma chamber (e.g., at the crust-mantle boundary) or possibly at the lithosphere-asthenosphere boundary where plume

H. B. West and W. P. Leeman

194

melts may collect prior to ascent through fractures in the lithosphere. In addition, because of waning magma supply rates, the upper part of the Haleakala Crater section probably represents a longer time interval than does a comparable thickness of lavas exposed in the lower part of the section. Thus, the apparent enhanced rate of change in isotopic compositions for the upper part of the section must also reflect to some extent the longer time period involved. Evidence from Haleakala Crater lavas for magma evolution over a range of pressures, the absence of primitive basalts, and the alternating sequences of basalt + mugearite are consistent with the existence of both shallow- and deeper-level magma chambers during the Haleakala alkalic cap stage. Primitive magmas may have undergone crystal fractionation in a deep-level magma chamber. Slightly differentiated basalts derived from these primitive magmas may then have ascended to a shallow-level magma chamber, within which the alternating sequences of basalt --t mugearite formed. The differences in alkalic cap evolution observed between Hawaiian volcanoes ultimately may be traced to differences in magma supply rates. At Haleakala, the decrease in magma supply rates during the alkalic cap stage may have been more protracted than the supply rate decreases at Mauna Kea and Kohala during comparable periods in their evolution. CONCLUSIONS Alkalic cap lavas from a continuous stratigraphic section in Haleakala Crater consist of inter-bedded alkalic basalts, hawaiites, and mugearites. These lavas display cyclic, upsection geochemical variations that delineate a series of discrete compositional subsections. These subsections are interpreted as representing distinct magma batches erupted from a dynamic, evolving, open-system magma chamber during periods of high magma recharge rates. The fine-scale cyclic (short time period) compositional variations associated with magma batches are superimposed on the large-scale (long time period) variations related to changes the source of these magmas. The sharp, upsection geochemical breaks between magma batches represent periods of eruptive quiescence and probably formed during periods of low or negligible magma chamber recharge rates. Magma evolution during these periods was dominated by crystal fractionation of the observed phenocryst phases. However, increases in highly/moderately incompatible trace element ratios associated with interbatch breaks exceed those predicted by simple closed system fractionation or low pressure open-system fractionation models. These effects indicate that additional processes capable of further fractionating these ratios operated during these periods and that the magma chamber was an open-system as these lavas formed. The upsection compositional variations within magma batches correspond to eruptive periods during which lava compositions became systematically more mafic with time. These intrabatch geochemical variations formed by mixing of mafic recharge magmas with evolved magma stored in the magma chamber. However, the deviation of lava compositions for any single batch from simple magma mixing curves

suggests that during eruptive periods the magma chamber evolved by a series of recharge-eruption-mixing-fractionation cycles. Thus, during eruptive periods, recharge appears to have been episodic. The evidence for magma evolution over a range of pressures, the absence of primitive basalts, and the alternating sequences of basalt + mugearite are consistent with the existence of both shallow- and deeper-level magma chambers during the Haleakala alkalic cap stage. The open-system, cyclic geochemical variations displayed by Haleakala alkalic cap lavas are considered to be a natural consequence of a waning magma budget. Decreasing magma supply rates during the alkalic cap stage of Hawaiian volcanism can result in increased magma chamber residence times and create greater magmatic diversity. Differences in petrologic and geochemical evolution between alkalic cap suites from different Hawaiian volcanoes may have resulted from variable rates of magma recharge into evolving magma chambers. Acknowledgments-Mahalo nui loa to Fred T. Mullins for his invaluable field assistance and to Ron Nagata and Hugo Huntzinger of the Haleakala National Park Service for their kokua. We thank F. A. Frey, J. M. Rhodes, D. Elthon, and G. Fitton for access to their laboratories. We are indebted to D. C. Gerlach for lab assistance and constructive advice on disciplinary matters. Holistic reviews of various versions of this paper by Marc D. Norman and Darryl and Wanda Kennedy substantially improved this paper. We thank M. 0. Garcia and T. P. Hulsebosch for their comments. We also thank J. M. Rhodes and B. K. Nelson for their constructive and philosophical reviews. This work was supported in part through GSA. Penrose grant 35 1585 to HBW, National Science Foundation grants EAR85-12167 and EAR83-20358 to WPL, and the Rice University Keith-Wiess Geological Laboratories. SOEST contribution No. 3305. Editorinl

handling. M. S. Ghiorso REFERENCES

BARDINTZEFF J.-M. and BONINB. ( 1987) The Amphibole

Effect: A Possible Mechanism for Triggering Explosive Eruptions. J. Vok. Geoih. Res. 33, 255-262. BARGARK. E. and JACKSONE. D. (1974) Calculated Volumes of Individual Shield Volcanoes Along the Hawaiian-Emperor Chain. USGS J. Rex 2, 545-550. BROPHYJ. G. ( 1990) Andesites from Northeastern Kanaga Island, Aleutians: Implications for Calc-Alkaline Fractionation Mechanisms and Magma Chamber Development. Contrib. Mineral. Peirol. 104, 568-58 I. BRYANW. B., THOMPSONG., and MICHAELP. J. (1979) Compositional Variation in a Steady-State Zoned Magma Chamber: MidAtlantic Ridge at 36”50’N. Tecfonophy. S&63-85. BVSP (Basaltic Volcanism Study Project) ( 198 I ) Basaltic Volcanism on the Terresirial Planets. Pergamon Press. CAMPBELL1. H., RODERP. L., and DIXON J. M. ( 1978) Plagioclase Buoyancy in Basaltic Liquids as Determined with a Centrifuge Furnace. Contrib. Mineral. Petrol. 67, 369-377. CHEN C.-H., PRESNALLD. C., and STERNR. J. ( 1992) Petrogenesis

of Ultramafic Xenoliths from the 1800 Kaupulehu Flow, Hualalai Volcano, Hawaii. J. Petrol. 33, 163-202. CHEN C.-Y. and FREY F. A. ( 1983) Origin of Hawaiian Tholeiites and Alkalic Basalt. Nature 302, 785-789. CHEN C.-Y. and FREY F. A. ( 1985) Trace Element and Isotopic Geochemistrv of Lavas from Haleakala Volcano, East Maui, Hawaii: Implicaiions for the Origin of Hawaiian Basalts. J. Geophys. Rex 90, 8743-8768.

CHEN C.-Y., FREY F. A., and GARCIAM. 0. ( 1990) Evolution of Alkalic Lavas at Haleakala Volcano. East Maui. Hawaii. Contnh. Minerul. Petrol. 105, 197-2 18

Petrogenesis

of the alkali-rich

CHEN C.-Y., FREY F. A., GARCIAM. O., DALRYMPLE G. B., and HART S. R. ( 199 1 ) The Tholeiite to Alkalic Basalt Transition at Haleakala Volcano, Maui, Hawaii. Contrib. Mineral. Petrol. 106, 183-200.

CLAGUE D. A. ( 1987a) Hawaiian Supply Rates, and Development

Xenolith Populations, Magma of Magma Chambers. Bull. Vol-

canal. 49, 577-587.

CLAGUE D. A.

( 1987b) Hawaiian Alkaline Volcanism.

G.S.A. Spec.

CLAGUE D. A. and BOHRSON W. A. ( 199 I ) Origin of Xenoliths in the Trachyte at Puu Waawaa, Hualalai Volcano, Hawaii. Contrib. Petrol.

108, 439-452.

CLAGUE D. A., JACKSON E. D., and WRIGHT T. L. ( 1980) Petrology of Hualalai Volcano, Hawaii: Implications for Mantle Composition. Bull.

VolcanoI.

43, 64 1-656.

CLAGUE D. A., DAO-GONG C.. MURNANE R., BEESON M. H., LANPHERE M. A., DALRYMPLE G. B., FRIESEN W., and HOLCOMB R. T. ( 1982) Age and Petrology of the Kalaupapa Basalt, Molokai. Hawaii. Pac. Sci. 36, 4 I I-420. COOMBS D. S. and WILKINSON J. F. G. ( 1969) Lineages and Fractionation Trends in Undersaturated Volcanic Rocks from the East Otago Province (New Zealand) and Related Rocks. J. Petrol. 10, 440-50 I. CROSS W. ( I9 15) Lavas of Hawaii and their Relations. USGS Prqf: Paper 99.

DEER W. A., HOWIE R. A., and ZUSSMAN J. ( 1966) An Introduction to the Rock Forming Minerals. Longman. DRAKE M. J. and WEILL D. F. (1975) Partition of Sr, Ba, Ca. Y. ELI*+, Eu3+, and Other REE Between Plagioclase Feldspar and Magmatic Liquid: An Experimental Study. Geochrm. Cosmoc,him. Acta 39, 689-7 12. DUNN T. and MCCALLUM 1. S. ( 1982) The Partitioning of Zr and Nb Between Diopside and Melts in the System Diopside-AlbiteAnorthite. Geochim. Cosmochim. Acta 46, 623-629. DZURISIN D.. KOYANAGI R. Y.. and ENGLISH T. T. (1984) Magma Supply and Storage at Kilauea Volcano, Hawaii. 1956-1983. J l/o/c. Geoth. Res. 21, 171-206. ELTHON D. ( 1984) Plagioclase Buoyancy in Oceanic Basalts: Chemical Effects. Geochim. Cosmochim. Acta 48, 753-768. FEIGENSON M. D., HOFMANN A. W., and SPERA F. J. (1983) Case Studies on the Origin of Basalt II. The Transition from Tholeiitic to Alkalic Volcanism on Kohala Volcano, Hawaii. Contrib. Mineral

Petrol. 84, 390-405.

FODOR R. V. and KEIL K. ( 1979) Review of the Mineral Chemistry of Volcanic Rocks from Maui, Hawaii. In Field Trip Guide to the Haw’aiian Island.7 (ed. M. GARCIA and J. M. SINTON). pp. 93106. H.I.G. Spec. Pub. FODOR R. V. and VANDERMEYDEN H. J. ( 1988) Petrology of Gabbroic Xenoliths from Mauna Kea Volcano, Hawaii. J. Geophys. Res. 93, 4435-4452.

FODOR R. V.. KEIL K., and BUNCH T. E. (1972) Mineral Chemistry of Volcanic Rocks from Maui. Hawaii: Fe-Ti Oxides. G.S.A. Absfr. fi~/Progr.

Andesitic

4, 507.

FODOR R. V., KEIL. K.. and BUNCH T. E. ( 1975) Contributions to the Mineral Chemistry of Hawaiian Rocks IV. Pvroxenes in Rocks from Haleakala and west Maui Volcanoes. Maui, Hawaii. Contrib. Mineral. Petrol. 50, 173- 195. FODOR R. V.. KEIL K., and BUNCH T. E. (1977) Contributions to the Mineral Chemistry of Hawaiian Rocks. VI. Olivines in Rocks from Haleakala and West Maui Volcanoes, Maui, Hawaii. Pac. Sci. 31, 299-308. FREY F. A., WISE W. S.. GARCIA M. O., WEST H., KWON S.-T.. and KENNEDY A. ( 1990) Evolution of Mauna Kea Volcano, Hawaii: Petrologic and Geochemical Constraints on Postshield Volcanism. J. Geophyr. Rer. 95, 127 l-1 300. FUJIMAKI H., TA~SUMOTO M., and AOKI K.-I. (1984) Partition Coefficients of Hf, Zr, and REE Between Phenocrysts and Groundmass. Proc. 14”’ Lunar Planet. Sri. Co&. 662-672. GARCIA M. O., KURZ M. D., and MUENOW D. W. (-1990) Mahukona: The Missing Hawaiian Volcano. Geology 18, 111 I-1 114. GREEN T. H. and PEARSON N. J. ( 1985) Experimental Determination of REE Partition Coefficients Between Amphibole and Basaltic to

Liquids

795

at High Pressure.

Geochim.

Cosmochim.

Acta

49, 1465-1468.

GRIFFIN W. L. and MURTHY V. R. (1969) Distribution of K, Rb. Sr and Ba in Some Minerals Relevant to Basalt Genesis. Geochim. Cosmochim.

.4cfa 33, 1389-1414.

HENDERSON P. ( 1982) Inorganic Geochemistry. Pergamon. HIGUCHI H. and NAGASAWA H. ( 1969) Partition of Trace Elements Between Rock Forming Minerals and the Host Volcanic Rocks. Earth

Pub. 30, 227-252.

Mineral.

Iavas on Haleakala

Planet.

Sci. Lett. 7, 28 l-287.

ILA P. and FREY F. A. (1984) Utilization of Neutron Activation Analysis in the Study of Geologic Materials. Atomkernenergw Kemtectnik 44, 7 IO-7 16. IRVING A. J. and FREY F. A. ( 1984) Trace Element Abundance in Megacrysts and their Host Basalts: Constraints on Partition Coefficients and Megacryst Genesis. Geochim. Cosmochim. Acta 48, 1201-1221.

JOHANNSEN A. ( 193 I ) A Descriptive Petrography Volume

I

Introduction,

Univ. Chicago

Tewuw,

((the Igneous Rocks Classific~ations. and Glossary.

Press.

( 1987) Evidence for Variations in Magma Production Along Oceanic Spreading Centers: A Critical Appraisal. Geo1og.v 15, 127- I3 I. KEIL K.. FODOR R. V., and 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. Mineral. Petrol. 37, 253-275. KLEIN F. W. ( 1982) Patterns of Historical Eruptions at Hawaiian Volcanoes. J. 1hk. Geoth. Re?. 12, l-35. KLEIN F. W., KOYANAGI R. Y.. NAKATA J. S.. and TANIGAWA W. R. ( 1987) The Seismicity of Kilauea’s Magma System. L’.SGS Pro/: Paper 1350, 1019-l 185. KURZ M. D. (1986) Cosmogenic Helium in a Terrestrial Igneous Rock. Nature 320, 435-439. KURZ M. D., O’BRIEN P. A., GARCIA M. 0.. and FREY F. A. ( 1985) Isotopic Evolution of Haleakala Volcano: Primordial, Radiogenic and Cosmogenic Helium. EOS 66, 1 120. KLJRZ M. D., GARCIA M. O., FREY F. A.. and O’BRIEN P. A. ( 1987) Temporal Helium Isotopic Variations Within Hawaiian Volcanoes: Basalts from Mauna Loa and Haleakala. Geochim. Cosmochim. KARSON J. A. and ELTHON D.

Acta 51, 2905-2914.

LANGMUIR C. H. ( 1989) Geochemical Consequences of In Situ Crystallization. Nature 340, 199-205. LANPHERE M. A. and FREY F. A. ( 1987) Geochemical Evolution of Kohala Volcano, Hawaii. Con/rib. Mineral. Pe/rol. 95, lOO- 113. LEEMAN W. P. ( 1988) Boron Contents in Selected International Geochemical Reference Samples. Geost. NeM~.~/e/ter 12, 6 I-62. LEEMAN W. P. and HAWKESWORTH C. J. ( 1986) Open Magma Svsterns: Trace Element and Isotopic Constraints. j. Gcoph!!\. Re.y. 91,5901-5912. LEEMAN W. P.. MA M.-S., MURAL] A. V., and SCHMITT R. A. (1978) Empirical Estimation of Magnetite/Liquid Distribution Coefficients for Some Transition Elements: A Correction. Contrrb. Mineral. Petrol. 66, 429.

LEROEX A. R. and ERLANK A. J. (1982) Quantitative Evaluation of Fractional Crystallization in Bouvet Island Lavas. J. l’olc. Geoth. Res. 13, 309-338. LINDSTROM D. J. ( 1976) Experimental Study of the Partitioning of the Transition Metals Between Clinopyroxene and Coexisting Silicate Liquids. Ph.D. Dissertation. Univ. Oregon. MACDONALD G. A. ( 1942) Petrography of Maui. 1fawaii Div. Hj,drogr.

Bull. 7, 275-334.

MACDONALD G. A. ( 1949) Hawaiian Petrographic Province. GSA Blrll. 60. 1541-1596. MACDONALD G. A. ( 1963) Relative Abundance of Intermediate Members of the Oceanic Basalt-Trachyte Association-A Discussion. J Gcopl~~x Rec. 68, 5 100-5 102. MACDONALD G. A. ( 1968) Composition and Origin of Hawaiian Lavas. G.S.A. Mem. 116, 477-522. MACDONALD G. A. ( 1972) Composite Lava Flows on Haleakala Volcano, Hawaii. GSA Bull. 83, 297 l-2974. MACDONALD G. A. ( 1978) Geologic Map of the Crater Section of

796

H. B. West and W. P. Leeman

Haleakala National Park, Maui, Hawaii. USGS Misc. Inv. Ser. Map l-1088. MACDONALD G. A. and KATSURA T. (1962) Relationship of Petrographic Suites in Hawaii. A.G.U. Monugr. 6, 187-195. MACDONALD G. A. and KATSURA T. ( 1964) Chemical Composition of Hawaiian Lavas. J. Petrol. 5, 82-133. MACDONALD G. A. and POWERS H. A. ( 1946) Contribution to the Petrography of Haleakala Volcano, Hawaii. GSA Bull. 57, I 15124. MACDONALD G. A. and POWERS H. A. (1968) A Further Contribution to the Petrology of Haleakala Volcano, Hawaii. GSA Bdl. 79, 877-888. MACDONALD Cl. A.. ABBOTT A. T.. and PETERSON F. L. (1983) Volcanoes in the Sea: The Geo1og.v ofHawii. Univ. Hawaii Press. MCDOUGALL I. (1964) Potassium-Argon Ages from Lavas of the Hawaiian Islands. G.S.A. Bull. 75, 107-128. MCKAY G. A. and WEILL D. F. (1977) KREEP Petrogenesis Revisited. Proc. Sfh Lunar Sci. ConJI, 2339-2355. MCKAY G., WAGSTAFF J., and YANG S.-R. ( 1986) Zirconium, Hafnium, and Rare Earth Element Partition Coefficients for Ilmenite and Other Minerals in High-Ti Lunar Mare Basalts: An Experimental Study. J. Geophy.y. Res. 91, D229-D237. MOORE J. Cl., CLAGUE D. A., and NORMARK W. R. ( 1982) Diverse Basalt Types from Loihi Seamount, Hawaii. Geology 10, 88-92. MOORE R. B., CLACUE D.A., RUBIN M., and BOHRSON W. A. (1987) Hualalai Volcano: A Preliminary Summary of Geologic, Petrologic, and Geophysical Data. C’SGS Prqf: Paper 1350, 571-585. MUIR I. D. and TILLEY C. E. ( 1961) Mugearites and Their Place in Alkali Igneous Rock Series. J. Geol. 69, 186-203. NAUCHTON J. J., MACDONALD G. A., and GREENBERG V. A. (1980) Some Additional Potassium-Argon Ages of Hawaiian Rocks: The Maui Volcanic Complex of Molokai, Maui, Lanai and Kahoolawe. .I. l’olc Geolh. Rcs. 7, 339-355. NICHOLLS I. A. and HARRIS K. L. ( 1980) Experimental Rare Earth Element Partition Coefficients for Garnet, Clinopyroxene and Amphibole Coexisting with Andesitic and Basaltic Liquids. Gcochim. Cosmochim Ada 44, 287-308. NIELSEN R. L. (1988) A Model for the Simulation of Combined Major and Trace Element Liquid Lines of Descent. Geochim. Cosmochirn. dcta 52, 27-38. NORMAN M. D. and LEEMAN W. P. ( 1990) Open-System Magmatic Evolution of Andesites and Basalts from the Salmon Creek Volcanics. Southwestern Idaho, USA. Chem. Geol. 81, 167- 189. O’HARA M. J. (1977) Geochemical Evolution During Fractional Crystallization of a Periodically Refilled Magma Chamber. Nature 266, 503-507. O’HARA M. J. and MATHEWS R. E. ( 1981) Geochemical Evolution in an Advancing, Periodically Replenished, Periodically Tapped, Continuously Fractionated Magma Chamber. J. Geol. Sot. London 138,237-277. ONUMA N., HIGUCHI H., WAKITA H., and NAGASAWA H. (1968) Trace Element Partition Between Two Pyroxenes and the Host Lava. Earth Planet. Sci. Left. 5, 47-5 I. OOSTDAM B. L. ( 1965) Age of Lava Flows on Haleakala, Maui. Hawaii. G.%i Bull. 76, 393-394. PASTER T. P.. SCHAUWECKER D. S., and HASKIN L. A. ( 1974) The Behavior of Some Trace Elements During Solidification of the Skaergaard Layered Series. Geochim. Cosmochim. Acta 38, 15491577. PEARCE J. A. and NORRYM. J. ( 1979) Petrogenetic Implications of Ti. Zr. Y, and Nb Variations in Volcanic Rocks. Contrih. Mineral. Peirol. 69, 33-47.

POWERS H. A. ( 1955) Composition and Origin of Basaltic Magma of the Hawaiian Islands. Geochim. Cosmochim. Acta 7,77-107. REAGAN M. K., GILL J. B., MALAVASSI E., and GARCIA M. 0. ( 1987) Changes in Magma Composition at Arenal Volcano, Costa Rica, 1968- 1985: Real-Time Monitoring of Open-System Differentiation. Bull. VolcanoI. 49, 4 15-434. RHODES J. M. ( 1988) Geochemistry ofthe 1984 Mauna Loa Eruption: Implications for Magma Storage and Supply. J. Geophvs. Rex 93, 4453-4466. RHODES J. M. and DUNCAN M. A. ( 1979) The Evolution of OceanFloor Basaltic Magmas. In Deep Drilling Results in the Atlantic Ocean: Ocean Crust (ed. M. TALWANI et al.), pp. 262-272. A.G.U. RYAN M. P., KOYANAGI R. Y., and FISKE R. S. (1981) Modeling the Three-Dimensional Structure of Macroscopic Magma Transport Systems: Application to Kilauea Volcano, Hawaii. J. Geophpp. Re.s. 86, 7 I 1 l-7 129. SACK R. O., WALKER D., and CARMICHAEL 1. S. E. (1987) Experimental Petrology of Alkalic Lavas: Constraints on Cotectics of Multiple Saturation in Natural Basic Liquids. Confrih. Mineral. Petrol. 96, 1-23. SPENGLER S. R. and GARCIA M. 0. ( 1988) Geochemistry of the Hawi Lavas, Kohala Volcano, Hawaii. Contrib. Mineral. Petrol. 99,90- 104. STAKES D. S., SHERVAIS J. W., and HOPSON C. A. ( 1984) The Volcanic-Tectonic Cycle of the FAMOUS and AMAR Valleys, MidAtlantic Ridge (36”47’N): Evidence from Basalt Glass and Phenocryst Compositional Variations for a Steady State Magma Chamber Beneath the Valley Midsections, AMAR 3. J. Geuphys Rec. 89, 6995-7028. STEARNS H. T. ( 1940) Four-Phase Volcanism in Hawaii. G.S.A. Bull. 51, 1947-1948. STEARNS H. T. ( 1942) Origin of Haleakala Crater, Island of Maui. Hawaii. GSA. Bull. 53, 1-14. STEARNS H. T. and MACDONALD G. A. ( 1942) Geology and GroundWater Resources of the Island of Maui, Hawaii. Huwaii Div. Hy drogr. Bull. 7. WASHINGTON H. S. and KEYES M. G. ( 1928) Petrology of the Hawaiian Islands: IV. Maui. Amer. J. Sci. 15, 199-220. WATSON E. B. and GREEN T. H. ( 1981) Apatite/Liquid Partition Coefficients for the Rare Earth Elements and Strontium. Earth Planet. Sci. Let/. 56, 405-42 1. WEST H. B. and LEEMAN W. P. ( 1987) Isotopic Evolution of Lavas from Haleakala Crater, Hawaii. Earth Plane/. Sci. Left. 84, 2 I I225. WEST H. B.. GERLACH D. C., LEEMAN W. P., and GARCIA M. 0. ( 1987) Isotopic Constraints on the Origin of Hawaiian Magmas from the Maui Volcanic Complex, Hawaii. Nuture 330,2 16-220. WEST H. B.. GARCIA M. O., FREY F. A.. and KENNEDY A. (1988) Nature and Cause of Compositional Variation Among the Alkalic Cap Lavas of Mauna Kea Volcano, Hawaii. Confrih. Mineral. Pe/rd. 100, 383-397. WORNER G. and WRIGHT T. L. ( 1984) Evidence for Magma Mixing Within the Laacher See Magma Chamber (East Eifel, Germany). J. I’&. Gcolh. Rex 22, 301-327. WRIGHT T. L. ( I97 1) Chemistry of Kilauea and Mauna Loa in Space and Time. C:SGS Prof: Puper 735, I-40. WRIGHT T. L. (1973) Magma Mixing as Illustrated by the 1959 Eruption, Kilauea Volcano, Hawaii. GSA. Bull. 84, 849-858. WRIGHT T. L. ( 1974) Presentation and Interpretation of Chemical Data for Igneous Rocks. Contrih. Mineral. Petrol. 48, 233-248. WRIGHT T. L. and FISKE R. S. ( 197 I ) Origin of Differentiated and Hybrid Lavas of Kilauea Volcano, Hawaii. _/. Petrol. 12, l-56.