A Rapid Fluctuation in the Mantle Source and Melting ...imina.soest.hawaii.edu/GG/FACULTY/garcia/publications/Pietruszka-… · RECEIVED SEPTEMBER 3, 1998; REVISED TYPESCRIPT ACCEPTED

JOURNAL OF PETROLOGY VOLUME 40 NUMBER 8 PAGES 1321–1342 1999

A Rapid Fluctuation in the Mantle Sourceand Melting History of Kilauea VolcanoInferred from the Geochemistry of itsHistorical Summit Lavas (1790–1982)

AARON J. PIETRUSZKA∗ AND MICHAEL O. GARCIAHAWAII CENTER FOR VOLCANOLOGY, DEPARTMENT OF GEOLOGY AND GEOPHYSICS, UNIVERSITY OF HAWAII,

HONOLULU, HI 96822, USA

RECEIVED SEPTEMBER 3, 1998; REVISED TYPESCRIPT ACCEPTED FEBRUARY 25, 1999

KEY WORDS: Kilauea Volcano; geochemistry; mantle melting; sourceThe geochemical variations of Kilauea’s historical summit lavasheterogeneity; crustal assimilation(1790–1982) document the short-term magmatic evolution of one

of the Earth’s most active volcanoes. Most of these lavas are thoughtto have erupted directly from the shallow (2–4 km deep) magmareservoir that underlies the volcano’s summit region. This paperdetails a remarkable variation of lava chemistry that spans nearly INTRODUCTIONthe entire known compositional range of the volcano in only 200 years. Kilauea Volcano is an ideal location for studying theThe Pb, Sr, and Nd isotope and incompatible trace element ratios short-term (tens to hundreds of years) changes in basaltic(e.g. La/Yb or Nb/Y) of the lavas vary systematically over time magma genesis because it is a highly active volcano withwith an abrupt reversal after 1924. This rapid geochemical numerous eruptions during its historical period (1790 tofluctuation records the temporal changes in the parental magma the present). Continuous monitoring of the volcano’scomposition delivered to Kilauea’s summit reservoir since 1790. activity since the establishment of the Hawaiian VolcanoThe isotope and incompatible trace element ratio systematics suggest Observatory in 1912 has made it the ‘best understoodthat the source region of historical Kilauea magma is both isotopically basaltic volcano in the world’ (Tilling & Dvorak, 1993).and chemically heterogeneous. These source variations can be ex- Geophysical studies have delineated the basic plumbingplained by the melting of small-scale heterogeneities within the system of Kilauea, which includes a shallow (2–4 km deep)Hawaiian mantle plume. Model calculations suggest that the degree magma storage reservoir located beneath the volcano’sof partial melting decreased from the early 19th century until the summit region (e.g. Fiske & Kinoshita, 1969; Klein et al.,mid-20th century, which correlates with a lower eruption rate (and 1987). Furthermore, the fundamental magmatic processespresumably a lower magma supply rate) for the volcano between that operate within this plumbing system, such as crystal1840 and 1959. This interval of declining output from the fractionation (mostly olivine controlled) and magma mix-Hawaiian plume culminated with an explosive summit eruption in ing, have been identified by major element studies of the1924 and the longest quiescent period in Kilauea’s historical record volcano’s lavas (e.g. Wright, 1971; Wright & Fiske, 1971).(1934–1952). Lavas erupted just after 1924 are geochemically In addition, these mainly ‘crustal’ magmatic processesanomalous and may have been contaminated by the assimilation of are thought to be superimposed upon changes in thecountry rock into the volcano’s magma reservoir during the explosions. composition of parental magma delivered to Kilauea’sSubsequently, the inferred degree of partial melting and the volcano’s summit reservoir (e.g. Wright, 1971; Tilling et al., 1987).eruption rate have increased, with the highest values since the early The origin and timing of the short-term variations

in parental magma composition at Kilauea are poorly19th century observed during the current Puu Oo rift zone eruption.

∗Corresponding author. Present address: Carnegie Institution of Wash-ington, Department of Terrestrial Magnetism, 5241 Broad BranchRoad, NW, Washington, DC 20015, USA Oxford University Press 1999

JOURNAL OF PETROLOGY VOLUME 40 NUMBER 8 AUGUST 1999

understood. The oldest subaerial Kilauea lavas (from the beyond these single eruptions (except for major elements;Wright, 1971; Wright & Fiske, 1971).~25–100 kyr old Hilina Basalt), exposed along fault

scarps located 10–15 km south of the volcano’s summit, This paper presents the results of the first detailedstudy of the incompatible trace element, and Pb, Sr, anddisplay source- and melting-related fluctuations of in-

compatible trace element and isotope ratios (Chen et al., Nd isotope geochemistry of Kilauea’s historical lavas.Forty-three lavas that erupted within or near Kilauea’s1996), a feature also observed in lavas drilled from the

flank of Mauna Kea Volcano during the Hawaii Scientific summit caldera (Fig. 1) with known eruption dates (1790–1982) were collected. Most of these lavas are thought toDrilling Project (e.g. Lassiter et al., 1996; Yang et al.,have erupted directly from the summit reservoir of the1996). The time scale of the geochemical variations involcano (Wright, 1971). These samples record a nearlythese stratigraphic sections is not well known, but iscontinuous, 200 year history of the changes in lavaprobably limited by the average recurrence interval forchemistry at Kilauea. Rift zone lavas were excludedlava flow emplacement on distal areas of the volcanobecause these magmas intruded into the rift zones at(~500–1100 years for Kilauea, Holcomb, 1987; ~700–unknown times (e.g. Wright & Fiske, 1971). The main1375 years for Mauna Kea, Albarede, 1996). The wallsobjectives of this study are (1) to document the geo-of Kilauea Caldera (Fig. 1) preserve a more completechemical evolution of Kilauea lavas over the last tworecord of the changes in the volcano’s parental magmacenturies and (2) to evaluate the cause(s) of geochemicalcomposition than either the Hilina Basalt or Maunadiversity within these lavas. Our results show that theKea sections. These caldera-wall lavas, which eruptedisotope and incompatible trace element ratios (e.g. La/~0·2–2·8 kyr ago, exhibit cyclic major and trace elementYb or Nb/Y) of Kilauea’s historical summit lavas varyvariations when normalized to a constant MgO to correctsystematically over time with an abrupt reversal followingfor olivine fractionation (Casadevall & Dzurisin, 1987).the explosive 1924 eruption (Fig. 2). The post-1924However, the timing of these geochemical fluctuations iscompositional trends continue through the current Puuuncertain (perhaps hundreds of years?) because severalOo eruption. Overall, this temporal fluctuation in lavaeruptive pauses may have occurred within the strati-chemistry spans almost the entire known geochemicalgraphic section (Casadevall & Dzurisin, 1987).range of the volcano in only 200 years. These short-termThe interplay between crustal and mantle processesparental magma changes can be explained by systematicover time scales of decades to centuries at some ofvariations in the degree of partial melting of small-the Earth’s most active volcanoes such as Piton de lascale heterogeneities within the Hawaiian plume. LavasFournaise, Hekla, and Mt Etna has been inferred fromerupted just after the 1924 eruption are geochemicallythe short-term geochemical evolution of their lavas (e.g.anomalous and may have been contaminated by theAlbarede & Tamagnan, 1988; Sigmarsson et al., 1992;assimilation of country rock into the volcano’s magmaCondomines et al., 1995). Detailed geochemical samplingreservoir during the explosions.of historical lavas from ocean-island volcanoes such as

Kilauea and Mauna Loa can also be used to resolve thelength scales of the compositional heterogeneities in themantle and to distinguish the relative importance of

AN OVERVIEW OF KILAUEA’Ssource heterogeneity versus changes in the degree ofHISTORICAL SUMMIT ERUPTIONSpartial melting. Mauna Loa Volcano lavas erupted from

1843 to 1984 display systematic temporal variations Kilauea’s historical period began with a violently ex-of incompatible trace element and isotope ratios that plosive summit eruption around the year 1790. Theprobably result from melting small-scale source het- events of this eruption have been reconstructed fromerogeneities near the margin of the Hawaiian mantle early 19th century native Hawaiian oral accounts andplume (Rhodes & Hart, 1995). In contrast, historical modern geologic studies (e.g. Swanson & Christiansen,Kilauea lavas present the opportunity to study the dy- 1973; McPhie et al., 1990). During the 1790 eruption,namics of melt generation within an adjacent (central?) a series of phreatomagmatic and culminating phreaticpart of the Hawaiian plume (e.g. Frey & Rhodes, 1993). explosions ejected 0·1 km3 of tephra (~70% juvenile)Previous studies show that lavas from the Mauna Ulu from a vent within the boundaries of the modern caldera(1969–1974) and Puu Oo (1983 to the present) rift zone (McPhie et al., 1990). Between the phreatomagmatic anderuptions of Kilauea record changes in the volcano’s phreatic phases of the eruption, a few small lava flowsparental magma composition over short periods of time issued from a circumferential fissure 1·3 km southwest(<10 years) that can be explained by variations in the of the caldera (e.g. McPhie et al., 1990; Fig. 1). Followingextent of partial melting of a relatively homogeneous these explosions and before the first recorded observationsmantle source region (Hofmann et al., 1984; Garcia et al., in 1823, there are no descriptions of volcanic activity at1996). However, the geochemical variations of historical Kilauea. However, Sharp et al. (1987) reconstructed a

fire-fountaining eruption, presumably emanating from aKilauea lavas have never been examined systematically

1322

PIETRUSZKA AND GARCIA HISTORICAL KILAUEA VOLCANO LAVAS

Fig. 1. Geologic map of Kilauea’s summit region showing the distribution of historically erupted lavas (modified from the unpublished map ofJ. Lockwood and T. Neal, 1999). The boundary of Kilauea Caldera is shown by the heavy continuous line. Pit craters, such as Halemaumau,are marked by the heavy hatchured lines, and eruptive fissures are indicated by straight or curved lines. The small dots and arrows mark thelocations of samples analyzed in this study (the letters or numbers, keyed to the sample names in Table 1, indicate multiple analyses from asingle eruption). Samples not shown on the map were covered by later flows (e.g. 1868), collected molten (e.g. September and April 1982), orobtained from museum collections (e.g. 1931HM). The stippled area includes both prehistorical volcanics and tephra from historical eruptions(e.g. 1790 and 1959). The position of the Hawaiian Volcano Observatory (HVO) is provided for reference.

vent within the caldera between 1820 and 1823, on the was dominated by nearly continuous lava lake activityfrom several vents on the caldera floor before 1868basis of a deposit of reticulite (commonly termed the

‘golden pumice’) found stratigraphically overlying the and from Halemaumau (Fig. 1) alone after the greatearthquake of 1868. Overflows of the lava lake(s) occurred1790 tephra deposits on the southwest rim of the caldera.

Five Kilauea samples that erupted between 1790 and frequently. We sampled the six lava lake overflows thatremain exposed on the caldera floor (Fig. 1). In addition,1823 are included in this study (Fig. 1): two clasts of the

golden pumice, two extracaldera 1790 lava flows, and we analyzed some samples that are no longer exposed:an 1866 caldera lava (Wright, 1971), and lavas thatscoria from the 1790 eruption [layer 6 of McPhie et al.

(1990)]. erupted from Halemaumau in 1912 and 1917. We alsoanalyzed samples from the three brief eruptions on theWritten descriptions of Kilauea’s volcanic activity be-

gan in August 1823 [Macdonald et al. (1983) is used for edge of Kilauea Caldera near Kilauea Iki Crater in 1832,1868, and 1877 (Fig. 1), which coincided with briefthis summary unless otherwise noted]. Over the next

century (1823–1924), volcanism at the summit of Kilauea hiatuses in lava lake activity.

1323


Fig. 2. Temporal variations of Pb, Sr, and Nd isotope and incompatible trace element ratios in Kilauea’s historical summit lavas. Thegeochemical trends defined by the summit lavas have continued through the current Puu Oo rift zone eruption (stippled rectangles; Garcia etal., 1996). Only Puu Oo lavas that erupted after the initial period of mixing with an evolved rift zone magma are shown in this and subsequentfigures. The vertical stippled lines mark the dates of Kilauea’s explosive summit eruptions in 1790 and 1924. The 1868 lava may have beenstored beneath Kilauea Iki Crater for several decades before eruption (dashed arrow). The samples are grouped according to eruption date(down triangle, 1790 tephra; up triangles, 1790 lava; circles, 1820–1921; diamonds, 1924–1954; squares, 1959–1982). The open symbols areliterature data from Tatsumoto (1978), White & Hofmann (1982), Hofmann et al. (1984), Stille et al. (1986), and Chen et al. (1991). The 2r errorbars are given in a corner of each plot.

In 1924, the lava lake within Halemaumau drained was studied). There have been no eruptions at the summitof the volcano since September 1982.away and major collapses of the floor and walls of

Halemaumau occurred. This engulfment was ac-companied by 3 weeks of phreatic explosions that ejected0·0008 km3 of lithic material (e.g. Decker & Christiansen,

PETROGRAPHY1984; Dvorak, 1992). The volume of Halemaumau in-creased by ~0·2 km3, which is 250 times greater than All of the samples analyzed in this study are pristine,the amount of ejected rock (Dvorak, 1992). After the moderately to strongly vesicular lavas with a glassy to1924 explosions, sporadic eruptions at the summit of microcrystalline groundmass. As in most Kilauea summitKilauea have been of short duration (<1 day to several lavas (e.g. Wright, 1971; Casadevall & Dzurisin, 1987),months) from fissures within and near the caldera (Fig. 1). olivine is the main phenocryst (>0·5 mm across) andWe analyzed samples from 1924 and 1929, and a lava microphenocryst (>0·1 to <0·5 mm across) phase (Fo78–89;from nearly every summit eruption between 1931 and M. Garcia, unpublished data, 1999). Most of the lavas

are aphyric (<0·5% phenocrysts), although some samples1982 (only one of the three phases of the 1961 eruption

1324


are weakly olivine phyric (<5% phenocrysts) and a few THE GEOCHEMISTRY OF KILAUEA’Sare olivine rich (the December 1974 and 1959 eruption

HISTORICAL SUMMIT LAVASlavas). These olivines commonly contain minute Cr-spinel and glass inclusions. Many of the lavas (sampleHM68-15, and most of those erupted between 1868 and1921) contain rare glomerocrysts (mostly micro-phenocrysts) of clinopyroxene ± plagioclase ± olivine.

Incompatible trace element variationsRare, small phenocrysts of clinopyroxene are found inmany of the lavas erupted between 1885 and 1921, The historical summit lavas of Kilauea show small butalthough they are generally present only in glomerocrysts. significant variations of incompatible trace elementPlagioclase phenocrysts do not occur in any of Kilauea’s abundances and ratios. Trace elements with similar de-historical summit lavas. grees of incompatibility produce tight, positive cor-

relations in variation diagrams (e.g. La vs Ce; Fig. 3).The coherence of these trends decreases with increasingcontrast in the incompatibility of the trace elements (e.g.

ANALYTICAL METHODS La vs Sm to La vs Yb; Fig. 3). This scatter resultsThe abundances of Cs, Rb, Ba, Th, U, Nb, Sr, Zr, Hf, from systematic changes in the highly to moderatelyY, and the rare earth elements (REE) were measured on incompatible trace element ratios of the lavas (e.g. Nb/43 samples by inductively coupled plasma mass spec- Y and Nb/Hf vs La/Yb and La/Sm; Fig. 4). Thesetrometry (ICP-MS) at Washington State University incompatible trace element systematics are similar to(Table 1). Before dissolution, samples of glass or ground- those that have been found previously for other Kilaueamass were repeatedly washed in a sonic bath of distilled lavas (e.g. Hofmann et al., 1984; Casadevall & Dzurisin,water until the water was clear, and then hand picked 1987; Tilling et al., 1987; Chen et al., 1996; Garcia et al.,to avoid crystals. Some samples (1924, KL1952S, 1974D- 1996). For comparison, the ranges of incompatible traceE2, and 1974D-E3) and the Hawaiian basalt standard element abundances and highly to moderately in-Kil1919 (from the same 1919 lava flow in Kilauea Caldera compatible trace element ratios for the historical summitas the BHVO-1 standard) were analyzed from whole- lavas are similar to those found in the prehistorical Hilinarock powders. Details of the chemical and analytical Basalt lavas of Kilauea, which erupted over a ~75 kyrprocedures for the ICP-MS analyses have been described period (Chen et al., 1996; Figs 3 and 4).by King et al. (1993). Table 1 includes an estimate of theanalytical precision for each element based on replicateanalyses of samples and Kil1919 (1r errors of ~1–3%for most elements except Cs ~5%). Compared withrecent analyses of Kil1919 for the Hawaii Scientific Pb, Sr, and Nd isotope ratio variationsDrilling Project by neutron activation (Yang et al., 1996)

The total ranges of the Pb, Sr, and Nd isotope ratios inand X-ray fluorescence (Rhodes, 1996), our results forthe historical summit lavas of Kilauea are small butKil1919 agree within 1–5% for all elements except Rbsignificant relative to our analytical uncertainty (Fig. 5).(9% higher in this study), Th (10% higher), Y (12%The 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb isotopehigher), and Nb (7% lower). For comparison with ourratios of the lavas define positive correlations, whereasresults for Kil1919, the trace element abundances forthe Pb and Sr isotope ratios are roughly inversely cor-BHVO-1 (based on a compilation of literature sourcesrelated. The Nd isotope ratios of the lavas vary onlyand unpublished isotope-dilution data) are shown inslightly and correlate poorly with the other isotope ratios.Table 1.The isotopic variability of the historical summit lavas isMeasurements of Pb, Sr, and Nd isotope ratios wereclose to (Pb and Nd isotopes) or larger than (Sr isotopes)made on 17 samples using the University of Hawaii VGthe prehistorical Kilauea lavas from the Hilina BasaltSector thermal ionization mass spectrometer (Table 2).(Chen et al., 1996; Fig. 5). Thus, Kilauea has eruptedHand-picked samples of 50–80 mg of glass or groundmasslavas in only 200 years with compositions that spanchips were cleaned in distilled water (see above) andalmost the entire known isotopic range for this volcano.washed briefly (~10 min) in sonic baths of methanol,Previous studies of Pb, Sr, and Nd isotope ratios in2 N HCl, and ultrapure water before dissolution. OneKilauea’s historical summit lavas analyzed samples fromsample (Kil1919) was analyzed from a whole-rock pow-only a few eruptions (1919, 1921, and 1967–1968) andder. The chemical and mass spectrometric techniquesdid not identify any isotopic variation (Tatsumoto, 1978;used in this study have been described by Mahoney etWhite & Hofmann, 1982; Hofmann et al., 1984; Stille etal. (1991). Additional information on standard meas-

urements and procedural blanks is presented in Table 2. al., 1986; Chen et al., 1991).

1325


Tab

le1:

ICP

-MS

trac

eel

emen

tda

tafo

rth

ehi

stor

ical

sum

mit

lava

sof

Kilau

eaVol

cano

1790

-617

90-1∗

1790

-2∗

1820

-118

20-2

1832∗

1866

1868∗

K18

77-1∗

1885

-118

85-2

1885

-318

85-4

1889

1894

-2

1820

–182

318

20–1

823

( n=

2)

La10

·99·

28·

810

·110

·311

·614

·813

·513

·416

·6±

0·1

16·8

15·7

15·5

15·2

15·4

Ce

27·7

23·2

22·6

25·8

26·4

29·4

36·9

34·0

33·8

40·5±

0·6

41·3

38·0

38·8

37·4

38·2

Pr

4·12

3·51

3·50

3·81

3·89

4·45

5·18

5·15

4·87

5·84±

0·02

5·87

5·44

5·54

5·34

5·39

Nd

18·8

16·6

16·4

17·6

17·9

20·7

23·2

23·3

22·0

25·9±

0·3

26·3

24·9

24·5

24·0

23·8

Sm

5·15

4·58

4·62

4·87

4·94

5·62

5·90

6·28

5·69

6·69±

0·05

6·84

6·28

6·26

6·01

6·09

Eu

1·75

1·61

1·63

1·65

1·70

1·89

2·01

2·10

1·94

2·22±

0·03

2·24

2·15

2·13

2·05

2·10

Gd

5·61

5·17

4·99

5·22

5·23

5·86

6·38

6·60

6·01

6·97±

0·03

7·14

6·68

6·63

6·60

6·41

Tb

0·84

0·75

0·78

0·78

0·79

0·86

0·93

1·00

0·88

1·02±

0·00

1·05

0·98

0·96

0·95

0·94

Dy

5·05

4·69

4·62

4·76

4·71

5·17

5·41

5·86

5·24

5·99±

0·07

6·09

5·74

5·60

5·67

5·50

Ho

0·95

0·88

0·89

0·89

0·88

0·97

1·00

1·13

0·98

1·12±

0·01

1·13

1·08

1·06

1·03

1·01

Er

2·46

2·32

2·28

2·35

2·31

2·50

2·66

2·89

2·53

2·90±

0·03

3·00

2·79

2·75

2·71

2·65

Tm

0·34

40·

329

0·32

20·

321

0·32

50·

346

0·36

60·

413

0·35

40·

406±

0·00

20·

415

0·38

90·

383

0·37

70·

363

Yb

1·97

1·87

1·86

1·88

1·87

2·04

2·12

2·34

2·06

2·33±

0·01

2·36

2·23

2·19

2·11

2·12

Lu0·

283

0·26

10·

271

0·26

50·

262

0·28

50·

297

0·32

10·

285

0·32

5±0·

002

0·33

90·

308

0·30

50·

301

0·28

7

Cs

0·07

10·

055

0·05

80·

061

0·06

20·

082

0·09

60·

088

0·08

40·

101±

0·00

40·

107

0·10

00·

096

0·09

50·

099

Rb

7·8

5·7

5·7

6·5

6·9

8·1

10·2

9·6

9·2

11·2±

0·0

11·6

9·9

9·6

9·8

9·6

Ba

100

7574

9195

109

133

123

127

152±

215

314

514

213

714

1

Th

0·83

0·68

0·70

0·80

0·77

0·88

1·16

1·05

1·03

1·28±

0·04

1·34

1·16

1·19

1·15

1·15

U0·

296

0·23

60·

233

0·27

00·

275

0·30

90·

403

0·36

60·

362

0·45

1±0·

010

0·47

80·

426

0·40

90·

404

0·40

3

Nb

12·2

10·0

10·0

10·9

11·7

12·5

16·8

15·9

15·4

19·2±

0·1

20·0

17·5

17·4

17·2

17·6

Sr

318

——

295

308

——

—35

938

7±2

386

——

——

Zr

146

129

128

135

141

152

174

182

159

194±

119

816

617

217

417

5

Hf

3·71

3·28

3·33

3·53

3·55

3·95

4·32

4·62

4·07

4·87±

0·01

4·97

4·39

4·51

4·45

4·37

Y27

·524

·924

·824

·925

·927

·329

·233

·027

·832

·2±

0·1

33·1

30·3

29·4

29·7

29·5

1326


Tab

le1:

cont

inue

d

95-T

AJ-

319

1719

18–1

919

BH

VO

-1K

il191

919

21-2

1921

-419

2419

2919

31H

M19

34K

L195

2S19

54-2

1959∗

1912

1919

1931

–193

2

( n=

3)C

om

pila

tio

n(n=

11)

La15

·915

·4±

0·3

15·4

15·4

15·4±

0·4

16·6

16·0

16·8

15·6

17·0

15·0

16·2

16·6

15·0

Ce

40·0

38·4±

0·7

38·1

38·1

38·2±

0·7

41·1

39·7

41·7

37·8

41·3

37·0

39·3

40·3

37·1

Pr

5·71

5·52±

0·12

5·42

5·9

5·51±

0·12

5·82

5·72

5·96

5·38

5·82

5·25

5·55

5·65

5·36

Nd

25·6

24·6±

0·5

24·6

24·7

24·5±

0·4

25·8

24·8

26·3

23·0

24·6

22·9

24·2

24·8

23·4

Sm

6·42

6·15±

0·10

6·17

6·13

6·12±

0·08

6·46

6·17

6·29

5·60

6·01

5·54

5·93

5·92

5·80

Eu

2·14

2·08±

0·07

2·08

2·09

2·08±

0·03

2·16

2·10

2·16

1·85

2·08

1·88

1·98

2·04

1·95

Gd

6·57

6·13±

0·12

6·28

6·31

6·26±

0·12

6·50

6·35

6·35

5·66

6·27

5·62

6·06

6·38

5·90

Tb

0·95

0·89±

0·01

0·92

0·98

0·91±

0·02

0·95

0·91

0·91

0·83

0·91

0·82

0·89

0·89

0·86

Dy

5·57

5·27±

0·06

5·39

5·33

5·29±

0·07

5·68

5·36

5·31

4·87

5·24

4·83

5·22

5·21

5·05

Ho

1·03

0·97±

0·02

0·98

1·03

0·98±

0·02

1·05

0·98

1·00

0·91

0·97

0·90

0·99

0·98

0·93

Er

2·73

2·51±

0·06

2·59

2·56

2·53±

0·05

2·63

2·56

2·59

2·42

2·52

2·32

2·53

2·54

2·37

Tm

0·37

20·

351±

0·00

40·

358

0·33

0·35

2±0·

007

0·36

30·

348

0·36

20·

332

0·35

00·

321

0·35

20·

352

0·33

3

Yb

2·13

1·99±

0·03

2·02

2·01

2·02±

0·05

2·13

2·02

2·10

1·92

1·99

1·85

2·04

2·07

1·89

Lu0·

295

0·27

7±0·

005

0·28

50·

280·

283±

0·01

00·

292

0·28

10·

287

0·26

10·

283

0·25

30·

289

0·28

10·

260

Cs

0·09

40·

086±

0·00

10·

092

0·13

0·09

3±0·

005

0·10

00·

092

0·09

90·

087

0·10

10·

086

0·09

80·

098

0·09

5

Rb

10·2

9·5±

0·3

9·8

9·35

9·8±

0·5

10·2

9·5

10·9

10·2

11·5

9·4

11·0

10·4

10·1

Ba

140

131±

113

413

1·7

131±

213

913

614

813

714

313

214

414

414

0

Th

1·26

1·23±

0·02

1·26

1·10

1·25±

0·03

1·22

1·28

1·36

1·33

1·36

1·18

1·34

1·33

1·21

U0·

461

0·43

6±0·

009

0·46

20·

420·

441±

0·00

50·

438

0·45

90·

463

0·43

00·

458

0·39

20·

461

0·44

60·

424

Nb

18·9

17·9±

0·2

18·4

18·0

18·2±

0·2

18·9

18·8

19·9

18·2

19·7

17·5

19·7

18·6

17·6

Sr

385

389±

12—

396

——

—41

337

3—

343

395

——

Zr

185

174±

318

117

917

8±1

163

180

187

166

176

153

175

168

161

Hf

4·68

4·41±

0·02

4·48

4·39

4·47±

0·09

4·42

4·52

4·60

4·10

4·41

3·94

4·34

4·28

4·15

Y29

·327

·4±

0·5

28·3

28·2

28·0±

0·4

28·4

28·0

28·5

25·9

27·1

24·5

28·5

27·6

25·5

1327


Tab

le1:

cont

inue

d

K61

-22

HM

68-1

519

71A

-119

71A

-2∗

1971

S19

74J

1974

S19

74D

-E2∗

1974

D-E

3∗19

7519

82A

-20

1982

S-1

019

82S

-12

1982

S-1

4

1961

1968

Au

gA

ug

Sep

Jul

Sep

Dec

Dec

Ap

rS

epS

epS

ep

( n=

3)(n=

2)

La16

·316

·014

·211

·313

·813

·113

·411

·3±

0·1

11·3±

0·1

13·0

12·4

12·6

12·9

13·6

Ce

40·3

39·4

35·4

28·1

34·3

32·8

33·4

27·9±

0·5

28·3±

0·3

32·1

30·6

31·3

32·1

34·3

Pr

5·67

5·64

5·08

4·07

5·07

4·74

4·80

4·08±

0·03

4·09±

0·09

4·64

4·54

4·74

4·72

5·14

Nd

24·9

25·0

22·7

18·7

22·8

21·5

21·4

18·2±

0·1

18·4±

0·4

21·2

20·0

21·1

21·3

23·1

Sm

6·34

6·47

5·79

4·92

5·78

5·64

5·55

4·74±

0·02

4·78±

0·14

5·49

5·40

5·62

5·52

5·93

Eu

2·15

2·19

1·98

1·71

2·04

1·92

1·93

1·65±

0·02

1·69±

0·06

1·93

1·87

1·93

1·96

2·06

Gd

6·52

6·80

6·13

5·29

6·12

5·97

6·05

5·12±

0·02

5·22±

0·23

5·84

5·82

5·97

5·95

6·30

Tb

0·93

0·97

0·89

0·77

0·91

0·87

0·86

0·75±

0·01

0·78±

0·04

0·86

0·87

0·90

0·90

0·96

Dy

5·50

5·67

5·19

4·70

5·20

5·15

5·31

4·42±

0·10

4·44±

0·05

5·17

5·06

5·38

5·35

5·59

Ho

1·03

1·05

0·95

0·88

0·95

0·97

0·97

0·82±

0·01

0·84±

0·00

0·95

0·95

1·01

0·98

1·06

Er

2·63

2·77

2·49

2·29

2·51

2·54

2·49

2·18±

0·04

2·21±

0·01

2·49

2·48

2·62

2·61

2·75

Tm

0·35

90·

384

0·34

90·

324

0·35

50·

351

0·34

40·

303±

0·00

30·

311±

0·00

30·

340

0·34

50·

368

0·36

50·

379

Yb

2·06

2·17

2·01

1·86

2·03

2·04

1·99

1·78±

0·00

1·77±

0·00

1·98

1·96

2·13

2·05

2·20

Lu0·

285

0·31

10·

274

0·26

20·

279

0·28

30·

279

0·24

7±0·

002

0·24

6±0·

000

0·27

70·

283

0·28

50·

289

0·31

3

Cs

0·10

40·

103

0·09

80·

076

0·08

70·

087

0·07

90·

073±

0·00

60·

081±

0·00

30·

080

0·08

30·

076

0·08

00·

087

Rb

10·8

10·7

9·6

7·8

9·4

9·0

8·4

7·7±

0·1

8·1±

0·1

7·9

8·0

8·8

8·7

9·3

Ba

146

143

129

103

128

120

120

103±

110

7±1

115

113

117

119

123

Th

1·29

1·24

1·13

0·90

1·08

1·05

1·02

0·85±

0·01

0·89±

0·04

0·99

0·92

0·97

1·00

1·06

U0·

443

0·42

00·

380

0·31

00·

375

0·35

00·

356

0·29

4±0·

005

0·29

2±0·

012

0·34

20·

315

0·33

70·

334

0·37

5

Nb

18·2

17·4

15·4

12·8

15·3

14·6

14·4

12·6±

0·2

12·6±

0·1

14·1

13·4

14·3

14·1

14·6

Sr

384

393

367

318

—36

7—

305±

430

1±3

——

——

—

Zr

178

188

169

140

173

159

160

137±

113

8±0

158

145

160

162

158

Hf

4·48

4·64

4·07

3·56

4·20

4·07

4·01

3·50±

0·07

3·50±

0·02

3·94

3·79

4·10

4·06

4·23

Y28

·630

·127

·625

·328

·727

·627

·323

·8±

0·0

23·4±

0·0

27·1

27·2

29·0

28·7

30·0

Mo

sto

fth

ese

lava

ser

up

ted

fro

mve

nts

wit

hin

Kila

uea

Cal

der

a(∗

exce

pti

on

sto

this

).(S

eeFi

g.1

for

sam

ple

loca

tio

ns.

)Tr

ace

elem

ent

abu

nd

ance

sar

ein

par

tsp

erm

illio

n.

Th

ean

alyt

ical

un

cert

ain

ties

(1r

)fo

rea

chel

emen

tar

eb

ased

on

rep

licat

ean

alys

eso

fsa

mp

les

and

the

stan

dar

dK

il191

9(e

xpre

ssed

asp

erce

nt)

:La

(1·7

),C

e(1

·8),

Pr

(1·7

),N

d(1

·4),

Sm

(1·1

),E

u(1

·9),

Gd

(1·4

),T

b(1

·8),

Dy

(1·6

),H

o(1

·7),

Er

(2·0

),T

m(1

·4),

Yb

(1·3

),Lu

(2·0

),C

s(5

·0),

Rb

(3·3

),B

a(1

·1),

Th

(2·0

),U

(1·7

),N

b(1

·2),

Zr

(1·3

),H

f(1

·5),

Y(1

·2).

1328


Table 2: Pb, Sr, and Nd isotope data for the historical summit lavas of Kilauea Volcano

206Pb/204Pb 207Pb/204Pb 208Pb/204Pb 87Sr/86Sr 143Nd/144Nd eNd

1790-6 18·394 15·463 38·023 0·703563 0·512970 6·4

0·703564

1790-1 18·508 15·472 38·130 0·703595 0·512945 5·9

1820-1 18·400 15·467 38·050 0·703673 0·512953 6·1

0·512952 6·1

0·512956 6·2

1832 18·429 15·463 38·034 0·703649 0·512948 6·0

1866 18·559 15·475 38·121 0·703534 0·512952 6·1

0·703529

1885-4 18·540 15·486 38·128 0·703563 0·512969 6·4

0·512980 6·6

1894-2 18·552 15·477 38·114 0·703528 0·512970 6·4

0·512978 6·6

0·512958 6·2

95-TAJ-3 18·599 15·487 38·132 0·703522 0·512968 6·4

1917 18·648 15·487 38·168 0·703482 0·512975 6·5

Kil1919 18·671 15·499 38·244 0·703471 0·512968 6·4

0·703471

1929 18·572 15·490 38·219 0·703565 0·512940 5·9

1931HM 18·549 15·472 38·148 0·703577 0·512943 5·9

18·562 15·486 38·197

1954-2 18·573 15·494 38·228 0·703611 0·512940 5·9

K61-22 18·540 15·469 38·157 0·703550 0·512969 6·4

1971S 18·509 15·478 38·132 0·703585 0·512961 6·3

0·512959 6·2

1982A-1 18·455 15·461 38·056 0·703568 0·512971 6·5

1982A-20 18·454 15·465 38·079 0·703579 0·512975 6·5

Isotopic measurements were made using the single collector mode of the mass spectrometer for most samples. Nd and Srisotopic fractionation corrections are 148NdO/144NdO = 0·242436 (148Nd/144Nd = 0·241572) and 86Sr/88Sr = 0·1194. Thedata are reported relative to University of Hawaii standard values (maximum uncertainties based on repeated standardmeasurements): La Jolla Nd, 143Nd/144Nd = 0·511844 ± 0·000009 (1r), and NBS 987 Sr, 87Sr/86Sr = 0·710258 ± 0·000013(1r). Pb isotope ratios are corrected for fractionation using the NBS 981 Pb standard values of Todt et al. (1996). Theestimated maximum uncertainties (1r) for the Pb isotope ratios are 206Pb/204Pb ±0·007, 207Pb/204Pb ±0·007, and 208Pb/204Pb±0·018, based on repeated measurements of NBS 981. Within-run uncertainties on individual sample measurements of Pb,Sr, and Nd isotope ratios are less than the 1r mean external uncertainties of the La Jolla Nd, NBS 987, and NBS 981standards in all cases. Values in italics represent replicate analyses of separate sample dissolutions (all other repeats arealiquots of the same solution or multiple measurements of a single load). Total procedural blanks are negligible: 10–50 pgfor Pb, <120 pg for Sr, and <20 pg for Nd. It should be noted that eNd is calculated relative to 143Nd/144Nd = 0·512640 foreNd = 0. Sample 1982A-1 erupted in April 1982.

THE TEMPORAL GEOCHEMICAL 1790–1921The highly to moderately incompatible trace elementVARIATIONS OF HISTORICALratios (e.g. La/Yb or Nb/Y) of Kilauea’s historical summitKILAUEA LAVAS lavas increased from 1790 to 1921 (Fig. 2), when the

The historical summit lavas of Kilauea display a sys- eruptive style of the volcano was dominated by nearlytematic temporal fluctuation of incompatible trace ele- continuous lava lake activity. The only lava that plotsment, and Pb, Sr, and Nd isotope ratios (Fig. 2). In this significantly off these temporal trends (1868) eruptedsection, we describe these geochemical variations and outside the boundaries of the caldera near Kilauea Ikidiscuss the changes in the parental magma composition Crater (Fig. 1). This moderately evolved lava (~6·7 wt

% MgO; M. Garcia, unpublished data, 1999) may havedelivered to the volcano’s summit reservoir since 1790.

1329


geochemical variations (see Rhodes & Hart, 1995). Thus,the composition of the parental magma delivered to thevolcano’s summit reservoir must have changed over time.Parental magma changes at Kilauea have been proposedfor this period on the basis of major and limited traceelement data [the ‘magma batches’ of Wright (1971)],which show a temporal increase of MgO-normalizedincompatible element abundances (e.g. K2O and La) inlavas erupted during the 18th, 19th, and 20th centuries(Wright, 1971; Tilling et al., 1987).

1924–1982After the collapse of Halemaumau lava lake and sub-sequent phreatic explosions in 1924, the Pb, Sr, and Ndisotope ratios of Kilauea’s summit lavas display suddenshifts (Fig. 2). For the next several decades, lavas withrelatively high ratios of highly to moderately incompatibletrace elements were erupted at the volcano’s summit(diamonds in Figs 2 and 4). This time period was alsocharacterized by a major change in eruptive style atKilauea from nearly continuous lava lake activity tosporadic summit eruptions, and the longest quiescentperiod in the volcano’s historical record (1934–1952;Macdonald et al., 1983). During the second half of the20th century, the highly to moderately incompatible traceelement and Pb isotope ratios of the lavas decreased overtime, whereas the Sr and Nd isotope ratios remainedessentially constant (Fig. 2). Crystal fractionation from asingle parental magma cannot explain a temporal de-crease in the highly to moderately incompatible traceelement or Pb isotope ratios (see Rhodes & Hart, 1995).Therefore, the geochemical variations during the secondhalf of the 20th century record a previously undetectedreversal in the parental magma composition supplied tothe volcano’s summit reservoir. This change has con-tinued through the current Puu Oo rift zone eruptionwith lavas that are geochemically similar to those eruptedat the volcano’s summit in the early 19th century (Fig. 2).

Fig. 3. Incompatible trace element variation diagrams for Kilauea’sThus, the lavas erupted at Kilauea over the last 200 yearshistorical summit lavas. For comparison, the dashed fields enclose therecord a rapid fluctuation in the composition of thegeochemical range of lavas from the prehistorical Hilina Basalt of

Kilauea (~25–100 kyr old; Chen et al., 1996). The samples are grouped parental magma delivered to this volcano.according to eruption date (see inset for symbols). The 2r error barsare given at the corner of each plot. All trace element concentrationsare in ppm.

A HAWAIIAN PLUME SOURCE FORKILAUEA LAVASbeen stored below Kilauea Iki for several decades before

eruption. The increase of the highly to moderately in- Kilauea lavas form an end member in the isotopic arraycompatible trace element ratios from 1790 to 1921 also of Hawaiian volcanoes (e.g. West et al., 1987; Chen etcorrelates with a systematic temporal change in the Pb, al., 1996). This end member is characterized by relativelySr, and Nd isotope ratios of the lavas (87Sr/86Sr varies high 206Pb/204Pb and eNd, and low 87Sr/86Sr (Fig. 6), andinversely; Fig. 2), except for the lavas erupted during the is best represented historically by the early 20th century1790 explosive summit eruption. Crystal fractionation summit lavas of Kilauea (Fig. 2). Several models have

been proposed for the origin of this unique isotopicfrom a single parental magma cannot explain these

1330


Fig. 4. Highly to moderately incompatible trace element ratio variations for Kilauea’s historical summit lavas. For comparison, the dashedfields enclose the geochemical range of prehistorical Hilina Basalt lavas of Kilauea (Chen et al., 1996). The Nb and Y concentrations from Chenet al. (1996) have been adjusted to account for the interlaboratory differences described in the text. The 2r error bars are given in a corner ofeach plot. The other symbols are the same as in Fig. 3.

signature in Hawaiian lavas, including (1) assimilation of (King et al., 1993; Fig. 6). However, enormous amountsof bulk assimilation are required to match the relativelyhydrothermally altered oceanic crust into plume-derived

melts (Eiler et al., 1996), (2) melting the upper mantle high 206Pb/204Pb ratios of the early 20th century Kilauealavas (~89 or 95%) assuming plume-derived, parentalbelow Hawaii (e.g. Tatsumoto, 1978; Stille et al., 1986;

Hauri, 1996; Lassiter et al., 1996), and (3) melting a magmas of either historical Mauna Loa or Koolau com-position, respectively (Fig. 6). Even if the historical Ki-long-term depleted component within a heterogeneous

Hawaiian plume (e.g. West et al., 1987; Bennett et al., lauea lava with the lowest 206Pb/204Pb ratio (sample 1820-1, using 0·7 ppm Pb) is assumed to represent a plume1996; Chen et al., 1996; Lassiter & Hauri, 1998). Because

our interpretations depend critically on the source of the magma, the required amount of assimilation would stillbe unreasonably large (~82%).‘Kilauea’ end member (oceanic crust, ambient upper

mantle, or plume?), we briefly evaluate these models by Melting the upper mantle (lithosphere or astheno-sphere) below Hawaii also probably cannot explain thecomparing the isotopic composition of Kilauea’s historical

summit lavas (and the early 20th century lavas in par- Pb, Sr, and Nd isotope ratios of Kilauea lavas. Mantlexenoliths from Salt Lake Crater on the Island of Oahuticular) with inferred Pb, Sr, and Nd isotope ratios for

the oceanic crust and upper mantle below Hawaii. are fragments of the lithosphere below Hawaii (Okano& Tatsumoto, 1996). Although these xenoliths have SrCretaceous basalts collected from Ocean Drilling Pro-

gram Site 843 are the best available samples of the and Nd isotope ratios that could account for the relativelyhigh eNd and low 87Sr/86Sr ratios of Kilauea lavas fromoceanic crust near Hawaii. Assimilation of oceanic crust

with Pb, Sr, and Nd isotope ratios between those of the early 20th century, their 206Pb/204Pb ratios are toolow (Fig. 6). An alternative upper-mantle compositionseawater-altered (unleached analyses) and ‘fresh’ (leached

analyses) Site 843 basalts into a hypothetical plume- inferred from the Site 843 basalts (which may representthe mantle below Hawaii that was previously melted toderived magma (of either historical Mauna Loa or Koolau

composition) might explain Kilauea’s isotopic signature form mid-ocean ridge basalts) also cannot explain the

1331


Fig. 5. Pb, Sr, and Nd isotope ratio variations for Kilauea’s historical summit lavas. The open symbols are for previous analyses of Kilauea’shistorical summit lavas from Tatsumoto (1978), White & Hofmann (1982), Hofmann et al. (1984), Stille et al. (1986), and Chen et al. (1991). Thestippled field shows the isotopic range of lavas from the Puu Oo eruption (Garcia et al., 1996). For comparison, the dashed fields enclose theisotopic range of lavas from the prehistorical Hilina Basalt of Kilauea (Chen et al., 1996). All of the isotopic data from literature sources discussedin the text and shown in the figures are corrected relative to the University of Hawaii standard values (Table 2). The 2r error bars are givenin a corner of each plot. The other symbols are the same as in Fig. 3.

isotopic composition of Kilauea’s early 20th century lavas time scale of decades to centuries that are similar inbecause the calculated 206Pb/204Pb and 87Sr/86Sr ratios magnitude to the volcano’s overall compositional range.are too low (similar to modern-day MORB from the Because no long-term geochemical evolution has beenEast Pacific Rise; Fig. 6). Instead, the most likely in- recognized for Kilauea (Chen et al., 1996), these short-terpretation is that the isotopic signature of Kilauea’s term changes are probably the dominant mode of par-historical summit lavas from the early 20th century (and ental magma variation at this volcano (see Casadevall &the ‘Kilauea’ end member of Hawaiian volcanoes) results Dzurisin, 1987; Chen et al., 1996). A similar pattern offrom melting a long-term depleted component within the relatively large, fast parental magma changes (includingHawaiian plume [model (3) above], rather than part of both isotope and incompatible trace element ratios) isthe oceanic crust or upper mantle below Hawaii. A emerging from studies of shield lavas from other Hawaiianplume source for Kilauea lavas is also consistent with volcanoes such as Mauna Loa (Kurz et al., 1995; Rhodestheir relatively high 3He/4He ratios (Kurz, 1993). & Hart, 1995) and Mauna Kea (Lassiter et al., 1996;

Yang et al., 1996), although these volcanoes also displaya long-term geochemical evolution.THE ORIGIN AND TIMING OF

The systematic variations of Pb, Sr, and Nd isotopeKILAUEA’S PARENTAL MAGMA ratios in lavas erupted at Kilauea since 1790 requireVARIATIONS an isotopically heterogeneous source for the volcano’s

parental magma. The rapid changes in the source (orThe parental magma changes in historical Kilauea lavasare characterized by rapid geochemical variations on a proportions of the source components) for Kilauea (and

1332


other Hawaiian shield volcanoes) suggest that the length ratio (for example) and vice versa for the early 19th andlate 20th century lavas. In detail, however, a simplescale of source heterogeneity in the Hawaiian plume is

small relative to the scale of the partial melting region mixture of two chemically and isotopically distinct sourcecomponents cannot explain the combined geochemical(see Rhodes & Hart, 1995). This small scale of source

heterogeneity must be superimposed on any possible variations observed historically at Kilauea (Fig. 8). Short-term changes in the partial melting process are alsolarge-scale plume heterogeneity, such as a radial zonation

(e.g. Frey & Rhodes, 1993; Hauri, 1996; Lassiter et al., required. In the next section, we discuss the geochemicaleffects of source vs melting processes on the composition1996).

Chen et al. (1996) proposed that the source of Kilauea of Kilauea’s parental magma for the last 200 years.magma is also chemically heterogeneous (in ratios ofincompatible trace elements such as Nb/Zr), on the basisof a correlation between Nb/Zr and 206Pb/204Pb for the

A PARTIAL MELTING MODEL FORprehistorical Hilina Basalt lavas. The Pb and Sr isotoperatios of Kilauea’s historical summit lavas also correlate KILAUEA’S HETEROGENEOUSwith highly to moderately incompatible trace element SOURCEratios (e.g. La/Yb; Fig. 7), except for the lavas that

At least two source components are required to explainerupted during the phreatomagmatic 1790 eruption andthe combined Pb, Sr, and Nd isotope ratio variations inΖ40 years after the phreatic explosions of HalemaumauKilauea’s historical summit lavas (Fig. 2). One of thesein 1924. This correlation might be explained in thecomponents has relatively high 206Pb/204Pb and eNd, andcontext of chemical heterogeneity if the early 20th centurylow 87Sr/86Sr (lavas erupted in the early 20th century),lavas (with relatively high 206Pb/204Pb and low 87Sr/86Sr;and the other has relatively low 206Pb/204Pb and eNd, andFig. 2) were derived from a source with a higher La/Ybhigh 87Sr/86Sr (early 19th and late 20th century lavas).

Fig. 6. Pb, Sr, and Nd isotope ratios of Kilauea’s historical summitlavas compared with Koolau and Mauna Loa lavas, and possibleexplanations for the high 206Pb/204Pb and eNd, and low 87Sr/86Sr isotopicend member of Hawaiian volcanoes (i.e. the ‘Kilauea’ component).Assimilation of oceanic crust (represented by the Ocean Drilling Pro-gram Site 843 basalts) into a plume-derived, parental magma of eitherhistorical Mauna Loa or Koolau isotopic composition cannot explainKilauea’s isotopic signature because enormous amounts of assimilationare required (continuous lines). The composition of the oceanic crustused in this calculation (filled star) is constrained to lie in the field ofthe Site 843 basalts so that the mixing trends pass through the Kilaueadata (assuming average unleached isotope dilution Pb, Sr, and Ndconcentrations; King et al., 1993). Two possible isotopic compositionsfor the Hawaiian plume source are represented by historical MaunaLoa lavas (ML-85, Φ; Kurz et al., 1995) and Koolau lavas (Β, 69-TAN-2; Roden et al., 1994). The Pb, Sr, and Nd concentrations in theassumed plume-derived magmas are from the picrite data of Norman& Garcia (1999) for Koolau (average of KOO-CF and KOO-17A) andRhodes & Hart (1995) for Mauna Loa (ML-85, assuming 0·6 ppm Pb).To estimate the present-day isotopic composition of the upper mantlebelow Hawaii today (filled rectangle), we age corrected the least altered(for Pb and Nd) and/or leached (for Sr and Nd) Site 843 lavas (110my old; King et al., 1993) to obtain the Pb, Sr, and Nd isotope ratiosof the mid-ocean ridge basalt (MORB) source at the time of eruption,and then assumed parent/daughter ratios for depleted mantle [Rb/Sr= 0·0045 and 238U/204Pb= 6·3 from White (1993); 147Sm/144Nd=0·2148 from Zindler & Hart (1986)] to evolve the isotopic compositionof the source to the present day. The calculated 206Pb/204Pb ratio forthis source is only approximate because the U and Pb concentrationsused for the calculation were measured on different sample splits thatwere not acid washed (M. Garcia, unpublished data, 1999). Additionaldata sources are Roden et al. (1984), Mahoney et al. (1994), and Okano& Tatsumoto (1996). Only the least altered Koolau samples with K2O/P2O5 [ 1·4 (e.g. Frey et al., 1994) or leached analyses are shown. Theprehistorical Mauna Loa field includes samples from only the volcano’ssubmarine southwest rift zone (Kurz et al., 1995). The other symbolsare the same as in Fig. 3.

1333


Fig. 8. A partial melting model for Kilauea’s historical summit lavas.A simple mixture (dashed line) of two chemically and isotopicallydistinct sources (calculated from the early 20th century, and the early19th–late 20th century isotopic end-member lavas assuming they wereproduced by a constant degree of partial melting) fails to explain thecombined 206Pb/204Pb vs Ba/Sm ratio variations (and other com-binations of trace element and isotope ratios as well; Fig. 7). Changesin the degree of partial melting of an isotopically and chemicallyheterogeneous source are required (grid). The large open circle andopen square represent the assumed compositions of the early 19th–20thcentury and the early 20th century sources, respectively. The melt–source variation grid was calculated by mixing the two end-membersources in variable proportions and partially melting these mixturesfrom 4 to 11%. We assume non-modal batch partial melting (Shaw,1970) with a source mineralogy of olivine (60%), clinopyroxene (15%),orthopyroxene (15%) and garnet (10%) which enter the melt in 3:4:2:Fig. 7. Geochemical variations of Kilauea’s historical summit lavas. 1 proportions. The mineral–liquid partition coefficients are given inThe compositional range of the prehistorical Hilina Basalt lavas of Table 3. The 2r error bar is given on the left side of the plot. TheKilauea (Chen et al., 1996) is shown to emphasize the relatively other symbols are the same as in Fig. 3.large geochemical range of historical Kilauea lavas. Lavas from other

Hawaiian volcanoes are shown for comparison: Koolau (Frey et al.,1994; Roden et al., 1994), prehistorical Mauna Loa (Garcia et al., 1995;Kurz et al., 1995), and historical Mauna Loa (Rhodes & Hart, 1995). The source of historical Kilauea lavas must also beOnly the least altered Koolau samples with K2O/P2O5 [ 1·4 (e.g. chemically heterogeneous because some highly in-Frey et al., 1994) are shown. The 2r error bars are given in a corner

compatible trace element ratios vary slightly over time.of each plot. The other symbols are the same as in Fig. 3.For example, the early 20th century lavas have lowerBa/Nb ratios than would be expected from their relatively

For our partial melting model, we assume that the isotopic high La/Yb and Nb/Y ratios (Fig. 2) on the basis of thecompositions of these two lava groups represent the end- order of trace element incompatibility during partialmember values in Kilauea’s heterogeneous source. The melting (Ba > Nb > La > Y > Yb; Sun & McDonough,Pb, Sr, and Nd isotope ratios of Kil1919 and the average 1989). This observation can be explained if the early1990 Puu Oo lavas (Garcia et al., 1996) are used for the 20th century lavas were derived from a source that hasearly 20th century and the early 19th–late 20th century relatively low concentrations of highly incompatible tracesources, respectively (i.e. the historical Kilauea lavas with elements (compared with the source of earlier and laterthe maximum and minimum Pb isotope ratios). Of course, erupted lavas). Thus, we constrain the early 20th centurythe actual source components probably have more ex- source to be slightly more depleted than the early 19th–treme isotope ratios, but it is noteworthy that the early late 20th century source (primitive mantle normalized20th century Kilauea lavas define (87Sr/86Sr = 0·70347) La/Sm ratios of 0·8 and 0·9, respectively; see Fig. 9 and

Table 3 for more details). A depleted source for Kilaueaor closely approach (206Pb/204Pb and eNd) the ‘Kilauea’isotopic end member for Hawaiian volcanoes (compare lavas is consistent with the incompatible trace element

inversion model of Hofmann et al. (1984) for Mauna Ulumaximum 206Pb/204Pb = 18·86 and eNd = 7·2; Chen etal., 1996; Lassiter et al., 1996). eruption lavas, and similar results for other Hawaiian

1334


Table 3: Mineral–liquid partition coefficients used in

this paper

Olivine Clinopyroxene Orthopyroxene Garnet Plagioclase

Rb 0 0·0005 0 0·00001 —

Ba 0 0·0005 0 0·00001 —

Th 0 0·014 0 0·001 —

U 0 0·013 0 0·006 —

Nb 0 0·008 0 0·05 —

La 0 0·07 0·02 0·01 0·036

Ce 0 0·10 0·03 0·02 0·031

Pr 0 0·18 0·035 0·05 0·028

Nd 0 0·25 0·04 0·09 0·025

Sm 0 0·35 0·05 0·22 0·020

Eu 0 0·40 0·06 0·32 0·33

Gd 0 0·45 0·07 0·5 0·016

Dy 0 0·48 0·11 1·1 0·013

Er 0 0·49 0·13 2·0 0·010

Yb 0 0·50 0·15 3·0 0·007Fig. 9. Primitive-mantle normalized trace element patterns of historical

Pb 0 0·01 0 0·0001 0·26 Kilauea lavas and the inferred source end members used in the partialmelting model. It should be noted that the patterns are plotted on aSr 0 0·16 0 0·01 1·8linear scale with the elements in the order of decreasing incompatibilityduring partial melting (e.g. Sun & McDonough, 1989). (a) Average

Partition coefficients (D) were estimated as follows: incompatible trace element abundances for summit lavas eruptedOlivine. The D values for these incompatible trace elements between 1918 and 1921 (circles) and 1990 Puu Oo lavas (squares;are very low in olivine (e.g. Green, 1994) and are assumed Garcia et al., 1996), which are used to calculate the incompatible traceto be zero. element concentrations in the source end members for the model.Clinopyroxene. The Th, U, Nb, Pb, and Sr D values are from Patterns for the historical summit lavas with the highest and lowestHauri et al. (1994). DBa is from Beattie (1993a). DLa, DSm, DGd, concentrations of highly incompatible trace elements (1931HM andand DYb (= DLu) are average values of experiments using 1790-2, respectively) are shown for comparison (dashed lines). (b) SourceKilauea starting materials from Gallahan & Nielsen (1992). compositions calculated from the average incompatible trace elementOrthopyroxene. DLa, DSm, and DGd are average values from concentrations of the 1990 Puu Oo lavas and 1918–1921 summit lavas.Nielsen et al. (1992). The non-REE D values are assumed to The lavas are normalized to 16 wt % MgO by incremental additionbe zero. of equilibrium olivine [with an olivine–liquid Fe–Mg KD = 0·3 fromGarnet. The Th, U, Nb, Pb, and Sr D values are from Hauri Roeder & Emslie (1970)] using the major element compositions ofet al. (1994). DBa is from Beattie (1993b). DCe, DNd, DSm, DEu, Garcia et al. (1996) for the 1990 Puu Oo lavas, Rhodes (1996) forDGd, DDy, and DEr are from Shimizu & Kushiro (1975). Kil1919, and M. Garcia (unpublished data, 1999) for the other 1918–Plagioclase. DLa, DCe, DSm, DGd, and DYb are from the ex- 1921 summit lavas. The source compositions were calculated fromperimental run of Phinney & Morrison (1990). DEu is calculated these primary magmas assuming that they were produced by non-using the relationship between oxygen fugacity and the modal batch partial melting (Shaw, 1970) of a source containing olivineoxidation state of europium from Drake (1975) assuming (60%), clinopyroxene (15%), orthopyroxene (15%) and garnet (10%),DEu(II) = DSr and DEu(III) is the average of DSm and DGd. An which entered the melt in 3:4:2:1 proportions. The partition coefficientsoxygen fugacity of 1·4 × 10–8 is calculated from Kilinc et are given in Table 3. The degrees of partial melting are arbitrarilyal. (1983) using the major element composition of sample constrained to 5% for the 1918–1921 lavas and 10% for the 1990 Puu1931HM from Wright (1971) and a magmatic temperature of Oo lavas, to obtain a pattern for the early 20th century source (circles)1164°C estimated from the MgO geothermometer of Helz & that is slightly more depleted in highly incompatible trace elementsThornber (1987). DSr is calculated from the expression of than the early 19th–late 20th century source (squares). The Sr and PbBlundy & Wood (1991) that relates DSr to plagioclase com- concentrations of the end-member sources were calculated using theposition and magmatic temperature assuming XAn = 0·75 above references for Sr and assumed values for Pb: 1990 Puu Oo (0·8and T = 1164°C. DPb is the average value for basic rocks ppm Pb) and 1918–1921 (1.1 ppm Pb).from Henderson (1982).All other partition coefficients are interpolated or ex-trapolated from these values, or are estimated by comparison (Fig. 7) because the lavas with the highest La/Yb ratioswith the average D profiles of Green (1994). (for example) are assumed to be derived from a source

with relatively low La/Yb (and vice versa). A morechemically depleted source for the early 20th centurylavas is consistent with their relatively low 87Sr/86Sr andshield volcanoes (e.g. Budahn & Schmitt, 1985). However,

the inferred chemical heterogeneity is opposite to that high eNd.We model the variations in the partial melting processexpected from a simple comparison of the highly to

moderately incompatible trace element and isotope ratios at Kilauea since 1790 as follows. The incompatible trace

1335


element abundances in the source for each of the historical 1840 and 1959 (Dvorak & Dzurisin, 1993). The lowestsummit lavas are calculated as a two-component mixture model melt fractions began a few years before and lastedon the basis of the observed 206Pb/204Pb ratio compared Ζ40 years after the collapse of Halemaumau lava lakewith the assumed Pb isotope ratios in the early 20th and subsequent phreatic explosions in 1924, which wascentury, and the early 19th–late 20th century end-mem- also a period of greatly reduced eruptive activity thatber sources. This mixed source is then partially melted included the longest quiescent period in the volcano’sto a degree necessary to reproduce the observed La/Yb historical record (1934–1952; Macdonald et al., 1983).ratio of the lava (see Fig. 8 for more details). Finally, the Subsequently, the calculated degree of partial meltingcalculated primary magma compositions were evolved increased (again by a factor of ~2). The higher modelby olivine control to match the observed Yb concentration melt fractions after 1959 correlate with a greater averageof the lavas. Although the absolute degree of partial magma supply rate for Kilauea of ~0·06 km3/yr frommelting calculated using this method is model dependent, 1959 to 1990 (Dvorak & Dzurisin, 1993). The eruptionthe relative differences between samples are significant. rate during the Puu Oo eruption (which has the highest

Our results suggest a relatively large (factor of ~2) model melt fraction in the 20th century; Fig. 10) haschange in the degree of partial melting at Kilauea since generally exceeded this average supply rate (~0·1 km3/1790 (Fig. 8). For comparison, this is an order of mag- yr; Dvorak & Dzurisin, 1993; Heliker et al., 1998). Thus,nitude larger then the range inferred for the lavas from when the melt output of the Hawaiian plume is inferredthe Mauna Ulu rift zone eruption of Kilauea (an ~20% to be the lowest, Kilauea’s eruption rate (and presumablyrelative increase in the extent of partial melting from its magma supply rate) is correspondingly low, and vice1969 to 1971; Hofmann et al., 1984). Our partial melting versa. Similar observations were made for historicalmodel successfully accounts for the overall geochemical Mauna Loa lavas, where the most chemically depletedvariations of Kilauea’s historical summit lavas (the results parental magmas (which are thought to be produced byare summarized in Table 4). However, some aspects of greater extents of partial melting) erupt during periodsthe data are not well explained: (1) the calculated 87Sr/ of relatively high eruption rate (Rhodes & Hart, 1995).86Sr ratios are generally slightly lower than observed in Historical Kilauea lavas also display an inverse cor-the lavas; (2) the lavas erupted just after the 1924 ex- relation between the source (expressed as a percentageplosions have the lowest eNd even though their Pb and of the early 20th century end member) and the modelSr isotope ratios are intermediate; (3) the residuals are melt fraction (except for the lavas erupted during thegenerally higher for isotopically intermediate lavas de- 1790 and Ζ40 years after the 1924 explosive summitrived by relatively low melt fractions (such as those from

eruptions; Fig. 10). This explains the observation thatlate 19th century and just after 1924). These discrepanciesthe lavas with greater ratios of highly to moderatelysuggest that Kilauea’s source is more isotopically andincompatible trace elements (such as those that eruptedchemically heterogeneous than can be explained by ain the early 20th century with high La/Yb; Fig. 5) seemsimple two-component mixture. The Pb and Sr isotopeto be derived from a more chemically depleted sourceratio variations are consistent with this interpretation(with relatively high 206Pb/204Pb and eNd, and low 87Sr/because lavas from the Puu Oo eruption have somewhat 86Sr). In contrast, historical Mauna Loa lavas with higherlower 87Sr/86Sr ratios for a given 206Pb/204Pb (similar toLa/Yb ratios tap a source with relatively high 87Sr/86Sr,the 1790 tephra and late 20th century summit lavas)and low eNd and 206Pb/204Pb (Rhodes & Hart, 1995),compared with early 19th century summit lavas (Fig. 5).which is opposite to the relationship observed at Kilauea(Fig. 7). This is further evidence for fundamental differ-ences in the source composition and/or melting processbetween Kilauea and Mauna Loa (see Frey & Rhodes,THE SOURCE, MELTING, AND1993). However, there are geochemical similarities be-

ERUPTIVE HISTORY OF KILAUEA tween historical Kilauea lavas and the oldest analyzedCOMPARED WITH MAUNA LOA Mauna Loa lavas from the volcano’s submarine southwest

rift zone (~100–300 kyr old; Garcia et al., 1995; Kurz etIn addition to the source changes recorded by the fluc-al., 1995). These prehistorical Mauna Loa lavas plottuations in the Pb, Sr, and Nd isotope ratios (Fig. 2), ouralong the extension of the trend between highly tomodel results suggest that there have also been systematicmoderately incompatible trace element ratios (e.g. La/variations in the partial melting process at Kilauea overYb) and Pb and Sr isotope ratios for Kilauea’s historicalthe last 200 years (Fig. 10). The inferred degree of partialsummit lavas (Fig. 7), and may have been formed bymelting decreased by a factor of ~2 from the early 19threlatively high degrees of partial melting of a source thatcentury to the mid-20th century. This correlates with ais similar to the source of Kilauea’s early 19th and latechange in the volcano’s eruption rate from [0·1 km3/

yr before 1840 (Mastin, 1997) toΖ0·01 km3/yr between 20th century summit lavas.

1336


Table 4: Representative results of the partial melting model for Kilauea’s historical summit lavas

2r error (%) 1790-1 1820-1 1885-4 95-TAJ-3 1917 1931HM K61-22 1982A-20

(% Diff) (% Diff) (% Diff) (% Diff) (% Diff) (% Diff) (% Diff) (% Diff)

Rb (6·6) 5·6 (–1·9) 6·6 (1·6) 10·0 (3·5) 10·1 (–0·6) 9·7 (2·2) 11·5 (0·1) 10·8 (–0·2) 8·2 (2·1)

Ba (2·2) 77 (2·3) 91 (–0·6) 137 (–3·4) 139 (–1·1) 133 (1·5) 143 (9·8) 149 (1·7) 112 (–0·8)

Th (4·0) 0·67 (–1·4) 0·75 (–6·9) 1·20 (0·8) 1·25 (–0·9) 1·22 (–0·4) 1·36 (1·6) 1·29 (0·4) 0·94 (2·8)

U (3·4) 0·236 (–0·3) 0·257 (–5·0) 0·423 (3·2) 0·443 (–4·1) 0·437 (0·2) 0·458 (5·9) 0·455 (2·7) 0·327 (3·6)

Nb (2·4) 9·9 (–1·1) 10·7 (–1·9) 17·5 (0·2) 18·3 (–3·5) 18·0 (0·4) 19·7 (0·4) 18·6 (2·2) 13·5 (1·1)

La (3·4) 9·2 (0·0) 10·1 (0·0) 15·5 (0·0) 15·9 (0·0) 15·4 (0·0) 17·0 (0·0) 16·3 (0·0) 12·4 (0·0)

Ce (3·6) 23·7 (2·3) 25·6 (–1·0) 38·7 (–0·3) 39·6 (–1·1) 38·3 (–0·3) 41·3 (0·7) 40·1 (–0·3) 31·2 (1·9)

Pr (3·4) 3·59 (2·2) 3·79 (–0·6) 5·58 (0·7) 5·67 (–0·6) 5·46 (–1·1) 5·82 (–0·3) 5·67 (0·0) 4·54 (0·1)

Nd (2·8) 16·9 (2·3) 17·7 (0·6) 25·1 (2·7) 25·3 (–0·8) 24·2 (–1·4) 24·6 (3·4) 25·2 (1·1) 20·8 (4·0)

Sm (2·2) 4·61 (0·7) 4·78 (–1·9) 6·42 (2·5) 6·39 (–0·5) 6·06 (–1·5) 6·01 (4·5) 6·31 (–0·5) 5·45 (0·8)

Eu (3·8) 1·62 (0·6) 1·67 (1·5) 2·19 (2·6) 2·17 (1·3) 2·05 (–1·3) 2·08 (1·7) 2·13 (–0·8) 1·88 (0·2)

Gd (2·8) 5·01 (–3·4) 5·09 (–2·6) 6·59 (–0·6) 6·53 (–0·7) 6·17 (0·6) 6·27 (0·1) 6·36 (–2·6) 5·65 (–2·9)

Dy (3·2) 4·64 (–1·1) 4·74 (–0·6) 5·78 (3·1) 5·65 (1·4) 5·29 (0·4) 5·24 (2·4) 5·50 (0·1) 5·10 (0·7)

Er (4·0) 2·30 (–0·6) 2·33 (–0·7) 2·76 (0·5) 2·69 (–1·7) 2·51 (–0·2) 2·52 (0·3) 2·61 (–0·8) 2·46 (–0·6)

Yb (2·6) 1·87 (0·0) 1·88 (0·0) 2·19 (0·0) 2·13 (0·0) 1·99 (0·0) 1·99 (0·0) 2·06 (0·0) 1·96 (0·0)

RR2 (186) (40·5) (95·0) (65·5) (41·4) (15·5) (174) (24·5) (56·8)

87Sr/86Sr (0·0037) 0·70354 0·70358 0·70352 0·70350 0·70348 0·70352 0·70352 0·70356

(–0·0085) (–0·0131) (–0·0058) (–0·0033) (–0·0003) (–0·0087) (–0·0039) (–0·0030)

% Early 20th 53·3 7·8 64·2 81·8 94·5 69·1 64·2 32·5

century source

Model melt % 10·3 10·2 6·2 5·5 5·1 4·7 5·4 7·8

The first column lists the analytical uncertainties from Tables 1 and 2, and the RR2 value (= the sum of the squared percentuncertainties) based on the incompatible trace element uncertainties. Subsequent columns show representative modelresults (the incompatible trace element concentrations are in ppm) with the residuals (% Diff) expressed as 100× [(calculated/observed) – 1]. The RR2 value for each sample is the sum of the squared percent residuals for each element. The calculateddegree of partial melting and the source mixing proportions used to model each sample are shown (see text for moreinformation).

and (2) these post-1924 lavas (for up to 40 years) areTHE GEOCHEMICAL EFFECTS OFgeochemically anomalous (e.g. elevated La/Yb for a

KILAUEA’S EXPLOSIVE 1924 given Pb or Sr isotope ratio; Fig. 7). There are twoSUMMIT ERUPTION possible eruption-related mechanisms for the post-1924

geochemical shift and subsequent anomaly: influx ofIn February 1924, the long-standing lava lake withina compositionally distinct parental magma or crustalHalemaumau pit crater drained away, and less than 3contamination.months later, major collapses of the floor and walls of

The post-1924 lava compositions might simply rep-Halemaumau were accompanied by 3 weeks of phreaticresent a new parental magma that formed by relativelyexplosions (Dvorak, 1992). Although this eruption waslow degrees of partial melting of a source that is iso-probably caused primarily by near-surface interactiontopically distinct from the source of lavas erupted earlierbetween hot rock and groundwater (e.g. Decker & Chris-in the 20th century (Fig. 10). The high rate of geochemicaltiansen, 1984), ~0·4 km3 of magma was withdrawn fromvariation around the time of the explosive eruption inthe volcano’s summit reservoir and Halemaumau was1924 (e.g. an implied rate of 206Pb/204Pb change ~5 timesenlarged by 0·2 km3 through collapse (Dvorak, 1992).faster than the average for the previous century; Fig. 2)This event profoundly affected the magma stored in themay be explained in this context if the intrusion ofvolcano’s summit reservoir because (1) the compositionthis compositionally distinct parental magma followed aof lavas erupted just after 1924 abruptly shifted to lower

206Pb/204Pb and 143Nd/144Nd, and higher 87Sr/86Sr (Fig. 2), drastic reduction in the volume of Kilauea’s summit

1337


reservoir. If the volume of magma in this reservoir is As the lowest melting temperature component of the twotypes of country rock would be the interstitial glasssmall (~2–3 km3; Pietruszka & Garcia, 1999), then the

0·4 km3 of magma withdrawn in 1924 (Dvorak, 1992) (<1000°C; Helz, 1987), we assume that only this materialwas assimilated by the reservoir magma in 1924.would have been volumetrically significant (~10–20%).

Such a substantial decrease would reduce the buffering This calculation (Table 5) indicates that the observedPb and Sr isotope, and La/Yb ratios of sample 1931HMcapacity of Kilauea’s summit reservoir and cause a greater

rate of geochemical variation in subsequent lavas (e.g. can be reproduced by mixing 3·5% interstitial glass,46·0% resident magma, and 50·5% input magma. Al-Albarede, 1993).

Crustal assimilation as a result of stoping of country though this assimilation model does not explain the loweNd of the post-1924 lavas (Fig. 2), the calculation doesrock during the engulfment of Halemaumau in 1924 is

another possible explanation for the geochemical shift reproduce the observed REE concentrations of sample1931HM within analytical uncertainty for most elements.and subsequent anomaly. We model the effects of as-

similation for the most geochemically anomalous post- The model also predicts negative Eu (~2%) and Sr(~17%) anomalies (which are not observed in the post-1924 eruption lava (sample 1931HM) using a mass-1924 lavas) as a result of the large amount of plagioclasebalance approach (see Table 5 for more details). Thefractionation assumed for the calculation of the interstitialisotopic (Pb and Sr) and incompatible trace elementglass composition. However, Kilauea’s parental magma(REE) composition of this sample is modeled by mixingmay contain positive Eu and Sr anomalies, which mightthe three components likely to be found in the shallowoffset this effect (Hofmann et al., 1984; Hofmann &plumbing system after the 1924 explosions: un-Jochum, 1996).contaminated resident magma, input magma, and melted

The assimilation model results are geologically reason-country rock (basalt and gabbro). The gabbros are as-able. First, only a small portion of Kilauea’s summitsumed to contain a small amount of highly evolvedmagma reservoir must have been contaminated. Theinterstitial glass, similar to the residual dacitic to rhyolitictotal volume of lava erupted at Kilauea’s summit betweenliquids observed in Kilauea Iki lava lake (Helz, 1987).

Fig. 10. The short-term geochemical evolution of Kilauea Volcano.(a) Eruptive history. The vertical stippled lines mark the dates ofKilauea’s explosive summit eruptions in 1790 and 1924. The upperpanel shows the temporal variation of Kilauea’s average eruption rate.The eruption rate from 1823 to 1840 is calculated using informationfrom Finch (1940a, 1940b), Macdonald et al. (1983), and Mastin (1997).Our result for this period (~0·1 km3/yr) is lower than the ~0·1–0·5km3/yr estimates of Mastin (1997) because we assume a shallowerdepth for Kilauea Caldera at this time [compare with Finch (1940a)].The 1840–1959 eruption rates are estimated using the approach anddata of Dvorak & Dzurisin (1993) except that we divide the analysisinto two time periods (1840–1924 and 1924–1959) to account for thedecrease in eruptive activity at Kilauea after the 1924 explosive eruption(Macdonald et al., 1983). Between 1959 and 1990, the magma supplyrate was ~0·06 km3/yr (Dvorak & Dzurisin, 1993), which is lower thanthe average eruption rate for the Puu Oo eruption (~0·1 km3/yr;Dvorak & Dzurisin, 1993; Heliker et al., 1998). The lower panel showsthe eruptive activity of Kilauea’s summit region (black bars) sinceAugust 1823 determined from Macdonald et al. (1983) and referencestherein. Eruptive activity during the late 20th century has been con-centrated along the volcano’s east rift zone with the sustained MaunaUlu (1969–1974) and Puu Oo (1983 to the present) eruptions (blackbars labeled MU and PO, respectively). (b) Source history. As predictedfrom the variation in 206Pb/204Pb, the source (or proportions of thesource components expressed as a percentage of the early 20th centuryend member) of Kilauea lavas has changed over time. The large opencircle and open square represent the average 1918–1921 summit and1990 Puu Oo lavas, respectively, which are used to define the isotopiccompositions in the partial melting model. The other symbols are thesame as in Fig. 3. (c) Melting history. The results of the partial meltingmodel suggest that the degree of partial melting has varied systematicallyby a factor of ~2 over the last 200 years. The large open circle andopen square represent the amount of partial melting assumed to producethe 1918–1921 summit and 1990 Puu Oo lavas, respectively. The othersymbols are the same as in Fig. 3.

1338


Table 5: Assimilation model results for a geochemically anomalous post-1924 eruption lava

Mixing components Model results

Resident + Input + Country = Anomalous

magma magma rock lava

Pre-1924 16% MgO Seawater Interstitial

reservoir parental altered glass of

magma magma basalt–gabbro gabbro

(1918–1921 av.) (1990 Puu Oo (Av. → (95% crystal 1931HM (% Diff) 2r error (%)

av. + equilibrium Kilauea) fractionation

olivine) of av. Kilauea)

Mixing % 46·0 50·5 0 3·5 100·0

La 15·8 9·6 14·0 152 17·4 (2·4) (3·4)

Ce 39·3 24·3 34·6 332 41·9 (1·4) (3·6)

Pr 5·62 3·58 5·00 36·8 5·68 (–2·5) (3·4)

Nd 24·9 16·7 22·3 133 24·6 (–0·1) (2·8)

Sm 6·23 4·50 5·74 27·3 6·09 (1·4) (2·2)

Eu 2·11 1·57 1·96 6·13 1·98 (–5·0) (3·8)

Gd 6·35 4·76 6·03 23·8 6·15 (–1·8) (2·8)

Dy 5·43 4·43 5·21 19·4 5·41 (3·3) (3·2)

Er 2·58 2·17 2·53 9·22 2·61 (3·5) (4·0)

Yb 2·05 1·75 2·02 7·23 2·08 (4·4) (2·6)

Sr 382 264 361 582

Pb 1·0 0·6 0·9 7·0

RR2 (85·4) (104)

La/Yb 7·7 5·5 6·9 21 8·4 (–1·9) (4·3)

87Sr/86Sr 0·70347 0·70359 0·70450 0·70450 0·70358 (0·0007) (0·0037)206Pb/204Pb 18·671 18·385 18·529 18·529 18·549 (–0·037) (0·083)

The compositions (the incompatible trace element concentrations are in ppm) and mixing proportions of the componentsused to model sample 1931HM are summarized in the first four columns (see text for more information). The model resultsfor 1931HM are shown with the residuals (% Diff) expressed as 100× [(calculated/observed) – 1]. The RR2 value for 1931HMis the sum of the squared percent residuals for each element. The final column lists the analytical uncertainties from Tables1 and 2, and the RR2 value (= the sum of the squared percent uncertainties) based on the incompatible trace elementuncertainties. The composition of the magma in the summit reservoir before 1924 is assumed to equal the average of lavaserupted from 1918 to 1921. For the country rock, we use an average historical Kilauea trace element and Pb isotopiccomposition, but assume an elevated 87Sr/86Sr ratio (as a result of isotopic exchange with seawater) to explain the Sr isotoperatio shift to higher values after 1924 (Fig. 2). Altered lavas from the bottom of the 2 km deep SOH-4 drill hole in Kilauea’seast rift zone have 87Sr/86Sr ratios as high as 0·7043 (H. West, personal communication, 1999). The trace element compositionof the interstitial glass is calculated from 95% equilibrium crystal fractionation of olivine, clinopyroxene, plagioclase, andorthopyroxene in 1:3:2:0·67 proportions (estimated from the modal mineralogy of gabbroic xenoliths from Kilauea’s 1960rift zone eruption lavas; Fodor & Moore, 1994) from the average Kilauea composition. Finally, we assume that the magmareservoir is recharged with an input magma equal to the composition of the average of the 1990 Puu Oo lavas (Garcia etal., 1996) with the incompatible element abundances diluted by equilibrium olivine addition (assuming an olivine–liquidFe–Mg KD = 0·3; Roeder & Emslie, 1970) in small steps to 16 wt % MgO for a presumed parental magma composition. Allcalculations use the mineral–liquid partition coefficients in Table 3.

1924 and 1961 is 0·13 km3 (Macdonald et al., 1983). If Second, the volume of rock that collapsed into theplumbing system in 1924 is ample to account for thethis represents the volume of contaminated magma and

Kilauea’s magma reservoir is assumed to be ~2–3 km3, volume of contaminated magma. Our calculation suggeststhat only a small percentage of melted interstitial glassthen only ~4–7% of the reservoir needs to be affected.

1339


from gabbros is required (Ζ3·5%). If the volume of isotope laboratories, and F. Frey, A. Hofmann, M. Kurz,M. Rhodes and D. Swanson for stimulating discussionscontaminated magma is 0·13 km3, this corresponds to

0·005 km3 of interstitial glass. Because this represents 5% about Hawaiian volcanoes. Helpful and constructive re-views by J. Baker, M. Dungan and M. Roden improvedresidual liquid, the required amount of gabbroic country

rock would be 0·09 km3, which is much less than the 0·2 this paper. This work was supported by National ScienceFoundation Grants EAR-9405573 and EAR-9614247 (tokm3 of rock that collapsed into Halemaumau during the

1924 explosions (Dvorak, 1992). M.G.). This is SOEST Contribution 4766.

REFERENCESCONCLUSIONSAlbarede, F. (1993). Residence time analysis of geochemical fluctuationsHistorical Kilauea lavas display a remarkable variation

in volcanic series. Geochimica et Cosmochimica Acta 57, 615–621.in lava chemistry that spans nearly the entire known Albarede, F. (1996). High-resolution geochemical stratigraphy of Maunageochemical range of the volcano in only 200 years. The Kea flows from the Hawaii Scientific Drilling Project core. JournalPb, Sr, and Nd isotope and incompatible trace element of Geophysical Research 101, 11841–11853.

Albarede, F. & Tamagnan, V. (1998). Modeling the recent geochemicalratios (e.g. La/Yb and Nb/Y) of Kilauea’s historicalevolution of the Piton de la Fournaise Volcano, Reunion Island,summit lavas vary systematically over time with an abrupt1931–1986. Journal of Petrology 29, 997–1030.reversal after 1924. The post-1924 temporal geochemical

Beattie, P. (1993a). The generation of uranium series disequilibria bytrends continue through the current Puu Oo rift zonepartial melting of spinel peridotite: constraints from partitioning

eruption. This rapid fluctuation in lava chemistry records studies. Earth and Planetary Science Letters 117, 379–391.the changes in the composition of parental magma de- Beattie, P. (1993b). Uranium–thorium disequilibria and partitioning onlivered to Kilauea’s summit reservoir since 1790. The melting of garnet peridotite. Nature 363, 63–65.

Bennett, V. C., Esat, T. M. & Norman, M. D. (1996). Two mantle-isotope and incompatible trace element ratio systematicsplume components in Hawaiian picrites inferred from correlatedsuggest that the source of historical Kilauea magma isOs–Pb isotopes. Nature 381, 221–224.both isotopically and chemically heterogeneous. These

Blundy, J. D. & Wood, B. J. (1991). Crystal-chemical controls on thesource variations can be explained by melting small-partitioning of Sr and Ba between plagioclase feldspar, silicate

scale heterogeneities within the Hawaiian mantle plume. melts, and hydrothermal solutions. Geochimica et Cosmochimica Acta 55,Model calculations suggest that the degree of partial 193–209.melting decreased from the early 19th century until the Budahn, J. R. & Schmitt, R. A. (1985). Petrogenetic modeling of

Hawaiian tholeiitic basalts: a geochemical approach. Geochimica etmid-20th century by a factor of ~2, which correlatesCosmochimica Acta 49, 67–87.with a lower eruption rate (and presumably magma

Casadevall, T. J. & Dzurisin, D. (1987). Stratigraphy and petrology ofsupply rate) for the volcano between 1840 and 1959.the Uwekahuna Bluff Section, Kilauea Caldera. US Geological Survey,This interval of declining output from the HawaiianProfessional Paper 1350, 351–375.

plume culminated with an explosive summit eruption Chen, C.-Y., Frey, F. A., Garcia, M. O., Dalrymple, G. B. & Hart,in 1924 and the longest quiescent period in Kilauea’s S. R. (1991). The tholeiite to alkalic basalt transition at Haleakala

Volcano, Maui, Hawaii. Contributions to Mineralogy and Petrology 106,historical record (1934–1952). Lavas erupted just after183–200.1924 are geochemically anomalous and may have been

Chen, C.-Y., Frey, F. A., Rhodes, J. M. & Easton, R. M. (1996).contaminated by the assimilation of country rock intoTemporal geochemical evolution of Kilauea Volcano: comparisonthe volcano’s magma reservoir during the explosions.of Hilina and Puna Basalt. Geophysical Monograph, American Geophysical

Subsequently, the inferred degree of partial melting andUnion 95, 161–181.

the volcano’s eruption rate have increased (again by a Condomines, M., Tanguy, J.-C. & Michaud, V. (1995). Magma dy-factor of ~2), with the highest values since the early 19th namics at Mt. Etna: constraints from U–Th–Ra–Pb radioactive

disequilibria and Sr isotope in historical lavas. Earth and Planetarycentury observed during the Puu Oo eruption.Science Letters 132, 25–41.

Decker, R. W. & Christiansen, R. L. (1984). Explosive eruptions ofKilauea Volcano, Hawaii. In: Explosive Volcanism: Inception, Evolution,

and Hazards. Washington, DC: National Academy Press, pp. 122–132.ACKNOWLEDGEMENTS Drake, M. J. (1975). The oxidation state of europium as an indicatorWe are grateful to T. Wright, M. Rhodes and S. Sorenson of oxygen fugacity. Geochimica et Cosmochimica Acta 39, 55–64.

Dvorak, J. J. (1992). Mechanism of explosive eruptions of Kilaueafor providing several samples for this study, N. Baker forVolcano, Hawaii. Bulletin of Volcanology 54, 638–645.collecting many of the lavas, and J. Lockwood and

Dvorak, J. J. & Dzurisin, D. (1993). Variations in magma supplyT. Neal for allowing us to utilize their unpublished maprate at Kilauea Volcano, Hawaii. Journal of Geophysical Research 98,of Kilauea’s summit region. We also thank J. Parker,22255–22268.

J. Rayray and K. Kolysko-Rose for helping to prepare Eiler, J. M., Farley, K. A., Valley, J. W., Hofmann, A. W. & Stolper,samples for chemical analysis, C. Knaack and K. Spencer E. M. (1996). Oxygen isotope constraints on the sources of Hawaiian

volcanism. Earth and Planetary Science Letters 144, 453–468.for providing technical assistance in the ICP-MS and

1340


Finch, R. H. (1940a). Engulfment at Kilauea Volcano. In: Fiske, R. S., King, A. J., Waggoner, D. G. & Garcia, M. O. (1993). GeochemistrySimkin, T. & Nielsen, E. A. (eds) The Volcano Letter 470. Washington, and petrology of basalts from Leg 136, central Pacific Ocean. In:DC: Smithsonian Institution Press, pp. 1–4. Wilkens, R. H., Firth, J., Bender, J., et al. (eds) Proceedings of the Ocean

Finch, R. H. (1940b). The filling in of Kilauea Crater. In: Fiske, R. S., Drilling Program, Scientific Results 136. College Station, TX: OceanSimkin, T. & Nielsen, E. A. (eds) The Volcano Letter 471. Washington, Drilling Program, pp. 107–118.DC: Smithsonian Institution Press, pp. 1–6. Klein, F. W., Koyanagi, R. Y., Nakata, J. S. & Tanigawa, W. R.

Fiske, R. S. & Kinoshita, W. T. (1969). Inflation of Kilauea Volcano (1987). The seismicity of Kilauea’s magma system. US Geologicalprior to its 1967–1968 eruption. Science 165, 341–349. Survey, Professional Paper 1350, 1019–1185.

Fodor, R. V. & Moore, R. B. (1994). Petrology of gabbroic xenoliths in Kurz, M. D. (1993). Mantle heterogeneity beneath oceanic islands:1960 Kilauea basalt: crystalline remnants of prior (1955) magmatism. some inferences from isotopes. Philosophical Transactions of the RoyalBulletin of Volcanology 56, 62–74. Society of London 342, 91–103.

Frey, F. A. & Rhodes, J. M. (1993). Intershield geochemical differences Kurz, M. D., Kenna, T. C., Kammer, D. P., Rhodes, J. M. & Garcia,among Hawaiian volcanoes: implications for source compositions, M. O. (1995). Isotopic evolution of Mauna Loa Volcano: a view frommelting process and magma ascent paths. Philosophical Transactions of the submarine southwest rift zone. Geophysical Monograph, Americanthe Royal Society of London 342, 121–136. Geophysical Union 92, 289–306.

Frey, F. A., Garcia, M. O. & Roden, M. F. (1994). Geochemical Lassiter, J. C. & Hauri, E. H. (1998). Osmium-isotope variations incharacteristic of Koolau Volcano: implications of intershield geo- Hawaiian lavas and evidence for recycled oceanic lithosphere in thechemical differences among Hawaiian volcanoes. Geochimica et Hawaiian plume. Earth and Planetary Science Letters 164, 483–496.Cosmochimica Acta 58, 1441–1462.

Lassiter, J. C., DePaolo, D. J. & Tatsumoto, M. (1996). IsotopicGallahan, W. E. & Nielsen, R. L. (1992). The partitioning of Sc, Y

evolution of Mauna Kea volcano: results from the initial phase ofand the rare earth elements between high-Ca pyroxene and natural

the Hawaii Scientific Drilling Project. Journal of Geophysical Researchmafic to intermediate lavas at 1 atmosphere. Geochimica et Cosmochimica101, 11769–11780.

Acta 56, 2387–2404.Macdonald, G. A., Abbott, A. T. & Peterson, F. L. (1983). Volcanoes inGarcia, M. O., Hulsebosch, T. P. & Rhodes, J. M. (1995). Olivine-

the Sea: the Geology of Hawaii. Honolulu, HI: University of Hawaiirich submarine basalts from the southwest rift zone of Mauna LoaPress.Volcano: implications for magmatic processes and geochemical

Mahoney, J., Nicollet, C. & Dupuy, C. (1991). Madagascar basalts:evolution. Geophysical Monograph, American Geophysical Union 92, 219–tracking oceanic and continental sources. Earth and Planetary Science239.Letters 104, 350–363.Garcia, M. O., Rhodes, J. M., Trusdell, F. A. & Pietruszka, A. J.

Mahoney, J. J., Sinton, J. M., Kurz, M. D., Macdougall, J. D.,(1996). Petrology of lavas from the Puu Oo eruption of KilaueaSpencer, K. J. & Lugmair, G. W. (1994). Isotope and trace elementVolcano: III. The Kupaianaha episode (1986–1992). Bulletin ofcharacteristics of a super-fast spreading ridge: East Pacific rise,Volcanology 58, 359–379.13–23°S. Earth and Planetary Science Letters 121, 173–193.Green, T. H. (1994). Experimental studies of trace-element partitioning

Mastin, L. G. (1997). Evidence for water influx from a caldera lakeapplicable to igneous petrogenesis—Sedona 16 years later. Chemical

Geology 117, 1–36. during the explosive hydromagmatic eruption of 1790, KilaueaHauri, E. H. (1996). Major-element variability in the Hawaiian mantle Volcano, Hawaii. Journal of Geophysical Research 102, 20093–20109.

plume. Nature 382, 415–419. McPhie, J., Walker, G. P. L. & Christiansen, R. L. (1990). Phreato-Hauri, E. H., Wagner, T. P. & Grove, T. L. (1994). Experimental and magmatic and phreatic fall and surge deposits from explosions at

natural partitioning of Th, U, Pb and other trace elements between Kilauea volcano, Hawaii, 1790 A.D.: Keanakakoi Ash Member.garnet, clinopyroxene and basaltic melts. Chemical Geology 117, 149– Bulletin of Volcanology 52, 334–354.166. Nielsen, R. L., Gallahan, W. E. & Newberger, F. (1992). Experimentally

Heliker, C. C., Mangan, M. T., Mattox, T. N., Kauahikaua, J. P. & determined mineral–melt partition coefficients for Sc, Y, and REEHelz, R. T. (1998). The character of long-term eruptions: inferences for olivine, orthopyroxene, pigeonite, magnetite and ilmenite. Con-from episodes 50–53 of the Puu Oo–Kupaianaha eruption of Kilauea tributions to Mineralogy and Petrology 110, 488–499.Volcano. Bulletin of Volcanology 59, 381–393. Norman, M. D. & Garcia, M. O. (1999). Primitive magma compositions

Helz, R. T. (1987). Differentiation behavior of Kilauea Iki lava lake, and source characteristics of the Hawaiian plume: constraints fromKilauea Volcano, Hawaii: an overview of past and current work. tholeiitic picrites. Earth and Planetary Science Letters 168, 27–44.Geochemical Society, Special Publication 1, 241–258.

Okano, O. & Tatsumoto, M. (1996). Petrogenesis of ultramafic xenolithsHelz, R. T. & Thornber, C. R. (1987). Geothermometry of Kilauea

from Hawaii inferred from Sr, Nd, and Pb isotopes. GeophysicalIki lava lake, Hawaii. Bulletin of Volcanology 49, 651–668.

Monograph, American Geophysical Union 95, 135–147.Henderson, P. (1982). Inorganic Geochemistry. New York: Pergamon Press.Phinney, W. C. & Morrison, D. A. (1990). Partition coefficients forHofmann, A. W. & Jochum, K. P. (1996). Source characteristics derived

calcic plagioclase: implications for Archean anorthosites. Geochimicafrom very incompatible trace elements in Mauna Loa and Maunaet Cosmochimica Acta 54, 1639–1654.Kea basalts, Hawaii Scientific Drilling Project. Journal of Geophysical

Pietruszka, A. J. & Garcia, M. O. (1999). The size and shape of KilaueaResearch 101, 11831–11839.Volcano’s summit magma storage reservoir: a geochemical probe.Hofmann, A. W., Feigenson, M. D. & Raczek, I. (1984). Case studiesEarth and Planetary Science Letters 167, 311–320.on the origin of basalt: III. Petrogenesis of the Mauna Ulu eruption,

Rhodes, J. M. (1996). Geochemical stratigraphy of lava flows sampledKilauea, 1969–(1971). Contributions to Mineralogy and Petrology 88,by the Hawaii Scientific Drilling Project. Journal of Geophysical Research24–35.101, 11729–11746.Holcomb, R. T. (1987). Eruptive history and long-term behavior of

Rhodes, J. M. & Hart, S. R. (1995). Episodic trace element and isotopicKilauea Volcano. US Geological Survey, Professional Paper 1350, 261–350.variations in historical Mauna Loa lavas: implications for magmaKilinc, A., Carmichael, I. S. E., Rivers, M. L. & Sack, R. O. (1983).and plume dynamics. Geophysical Monograph, American Geophysical UnionThe ferric–ferrous ratio of natural silicate liquids equilibrated in air.

Contributions to Mineralogy and Petrology 83, 136–140. 92, 263–286.

1341


Roden, M. F., Frey, F. A. & Clague, D. A. (1984). Geochemistry of Swanson, D. A. & Christiansen, R. L. (1973). Tragic base surge in1790 at Kilauea Volcano. Geology 1, 83–86.tholeiitic and alkalic lavas from the Koolau Range, Oahu, Hawaii:

Tatsumoto, M. (1978). Isotopic composition of lead in oceanic basaltimplications for Hawaiian volcanism. Earth and Planetary Science Lettersand its implication to mantle evolution. Earth and Planetary Science69, 141–158.Letters 38, 63–87.Roden, M. F., Trull, T., Hart, S. R. & Frey, F. A. (1994). New He,

Tilling, R. I. & Dvorak, J. J. (1993). Anatomy of a basaltic volcano.Nd, Pb, and Sr isotopic constraints on the constitution of theNature 363, 125–133.Hawaiian plume: results from Koolau Volcano, Oahu, Hawaii,

Tilling, R. I., Wright, T. L. & Millard, H. T., Jr (1987). Trace-elementUSA. Geochimica et Cosmochimica Acta 58, 1431–1440.chemistry of Kilauea and Mauna Loa lava in space and time: aRoeder, P. L. & Emslie, R. F. (1970). Olivine–liquid equilibrium.reconnaissance. US Geological Survey, Professional Paper 1350, 641–689.Contributions to Mineralogy and Petrology 29, 275–289.

Todt, W., Cliff, R. A., Hanser, A. & Hofmann, A. W. (1996). EvaluationSharp, R. P., Dzurisin, D. & Malin, M. C. (1987). An early 19thof a 202Pb–205Pb double spike for high-precision lead isotope analysis.century reticulite pumice from Kilauea Volcano. US Geological Survey,Geophysical Monograph, American Geophysical Union 95, 429–437.

Professional Paper 1350, 395–404.West, H. B., Gerlach, D. C., Leeman, W. P. & Garcia, M. O. (1987).Shaw, D. M. (1970). Trace element fractionation during anatexis.

Isotopic constraints on the origin of Hawaiian lavas from the MauiGeochimica et Cosmochimica Acta 34, 237–243.

Volcanic Complex, Hawaii. Nature 330, 216–220.Shimizu, N. & Kushiro, I. (1975). The partitioning of rare earth

White, W. M. (1993). 238U/204Pb in MORB and open system evolutionelements between garnet and liquid at high pressures: preliminary

of the depleted mantle. Earth and Planetary Science Letters 115, 211–226.experiments. Geophysical Research Letters 2, 413–416. White, W. M. & Hofmann, A. W. (1982). Sr and Nd isotope geo-

Sigmarsson, O., Condomines, M. & Fourcade, S. (1992). A detailed chemistry of oceanic basalts and mantle evolution. Nature 296,Th, Sr, and O isotope study of Hekla: differentiation processes in 821–825.an Icelandic volcano. Contributions to Mineralogy and Petrology 112, Wright, T. L. (1971). Chemistry of Kilauea and Mauna Loa in space20–34. and time. US Geological Survey, Professional Paper 735, 1–40.

Stille, P., Unruh, D. M. & Tatsumoto, M. (1986). Pb, Sr, Nd, and Hf Wright, T. L. & Fiske, R. S. (1971). Origin of differentiated and hybridisotopic constraints on the origin of Hawaiian basalts and evidence lavas of Kilauea Volcano, Hawaii. Journal of Petrology 12, 1–65.for a unique mantle source. Geochimica et Cosmochimica Acta 50, Yang, H. J., Frey, F. A., Rhodes, J. M. & Garcia, M. O. (1996).2303–2319. Evolution of Mauna Kea volcano: inferences from lava compositions

Sun, S.-S. & McDonough, W. F. (1989). Chemical and isotopic sys- recovered in the Hawaii Scientific Drilling Project. Journal of Geo-tematics of oceanic basalts: implications for mantle composition and physical Research 101, 11747–11767.processes. In: Saunders, A. D. & Norry, M. J. (eds) Magmatism in the Zindler, A. & Hart, S. (1986). Chemical geodynamics. Annual Review of

Earth and Planetary Sciences 14, 493–571.Ocean Basins. Geological Society, London, Special Publication 42, 313–345.

1342

Documents

A Rapid Fluctuation in the Mantle Source and Melting ...imina.soest.hawaii.edu/GG/FACULTY/garcia/publications/Pietruszka-… · RECEIVED SEPTEMBER 3, 1998; REVISED TYPESCRIPT ACCEPTED