JOURNAL OF GEOPHYSICAL
RESEARCH, VOL. 98, NO. BI0, PAGES 17,649-17,665, OCTOBER 10, 1993
TemporalandGeochemicalVariability of VolcanicProductsof
the MarquesasHotspot
D. L. DESONIE1 AND R. A. DUNCAN
Collegeof OceanicandAtmospheric
Sciences,OregonStateUniversity,Corvallis
J. H. NATLAND 2
ScrippsInstitutionof Oceanography,
Universityof California,La Jolla
The Marquesas
archipelago
is a short,NW-SE trendingclusterof islandsandseamounts
thatformedas a
resultof volcanicactivity over a weak hotspot. This volcanicchainlies at the northernmarginof a broad
regionof warm and compositionally
diversemantlethat meltsto build severalothersubparallelvolcanic
lineaments.Basaltsdredgedfrom submerged
portionsof volcanoes
alongthe Marquesaslineamentdecrease
in age from northwestto southeast.The new sampleage distributionyields a volcanicmigrationrate
significantlyslowerthanthatexpectedfor Pacificplatemotionovera stationaryMarquesas
hotspot.This
and the aberrantorientationof the chain indicatedeflectionof the plume by westwardupper mantle flow.
The interactionof this weak plumewith uppermantleflow accountsfor the temporaland spatialpatternsin
Marquesan
volcanism.The compositions
of subaerialandsubmarine
basaltsreflectthe mixingof at least
two mantlesources,distinguished
by Sr, Nd, and Pb isotopeand traceelementcompositions.There is a
consistentevolutionarypatternat each volcano,from early tholeiitic to later alkalic basalteruptions.
Tholeiitic and transitionallava compositions
can be derivedby variabledegreesof partial meltingof a
sourcecomposedof depletedmid-oceanridge basalt mantle (DMM) and mantle characterizedby
radiogenic
Pb (HIMU). Alkaliclavacompositions
appearto be dominantly
theresultof smallerdegrees
of
meltingof a moreradiogenic
mantlesource(EMII). Large-scale
meltingof the lowerlithosphere
or upper
mantle (DMM+HIMU) entrainedwithin a sheared,thermallybuoyantplume (EMII)
could producethe
tholeiitic and transitional basaltsfound in the main shields of the volcanoes,while alkalic basaltscould result
frommeltingof mantleof EMII composition
at theedgesof thehotspot.
THE MARQUESASVOLCANICLINEAMENT
In this paperwe reportnew 4øAr-39Arand K-Ar age
The Marquesas lineament, the northernmostof the many
hotspot lineaments of French Polynesia, includes 12 islands
and at least eight seamountsthat extend for nearly 500 km
north of the Marquesas Fracture Zone (Figure 1). These
youthfulvolcanicconesrise up to 5000 m from 50- to 65-m.y.old oceanic lithosphere [Kruse, 1988] that formed at the
ancestral East Pacific Rise (EPR). The Marquesas, and the
other islandsof FrenchPolynesia,are on the westernsideof an
unusually shallow region of thinner than normal lithosphere,
termed the South Pacific isotopic and thermal anomaly
(SOPITA), thought to be a zone of broad-scale mantle
upwelling [McNutt and Fischer, 1987; Smith et al., 1989;
McNutt and Judge, 1990; Staudigelet al., 1991]. Extreme
compositionalvariationsare found for basaltson a variety of
scales: between and among island groups, and within
individual volcanoes[Bishopand Woolley, 1973; Liotard et
al., 1986; Duncan et al., 1986; Vidal et al., 1987]. Volcanism
above the Marquesas hotspot has been intermittent, is
somewhat oblique to volcanism at other south Pacific
lineaments[Croughand Jarrard, 1981], and is the only French
Polynesianhotspottrack for which there is no known site of
active
volcanism.
1Now
atLamont-Doherty
Earth
Observatory,
Palisades,
NewYork.
2Now
atRosenstiel
School
ofMarine
andAtmospheric
Sciences,
Universityof Miami, Miami, Florida.
Copyright1993 by the AmericanGeophysical
Union.
Papernumber93JB01562.
0148-0227/93/93JB-01562505.00
determinationsfrom eight dredgehaulsand the islandof Fatu
Hiva in the Marquesas lineament, which revise the age
progressionpreviously determined from studies of island
samplesonly. Major and traceelementcontentsare consistent
with fractional crystallizationfrom a range of primary basalt
melt compositions. Isotopic ratios identify the parental
mantlecompositionsas havingdepletedmid-oceanridge basalt
mantle plus mantle characterized by radiogenic Pb
(DMM+HIMU) and more radiogenic mantle source (EMII)
signatures,in Zindler and Hart's [1986] terminology. We
discussthese data in terms of possible dynamic models of
mantle plume and lithosphere/asthenospheremelting and
mixing.
Age Relationshipsof MarquesasVolcanoes
Previousresultsof K-Ar age determinations
from the islands
of the Marquesas[Duncanand McDougall,1974;Duncanet al.,
1986; Brousse et al., 1990] have shown a southeastward
migrationof volcanismalongthe Marquesaschainfrom 6 Ma
at Eiao [Brousseet al., 1990; Caroff et al., 1990] through 1.3
Ma at Fatu Hiva [Duncanand McDougall, 1974]. The resultant
volcanicmigrationrate of 104 + 18 km/m.y. closelymatches
those calculated for other French Polynesian volcanic chains
[McDougalland Duncan,1980].
It has been suggestedthat volcanismover the Marquesas
hotspotdid not initiate at Eiao (Figure 1). Volcanismat ridges
to the west of the MarquesasIslands along the Galapagos
FractureZone and the Nova-Cantontrough,with agesof 13 and
30 Ma, respectively, has been proposedto have originated
17,649
17,650
DESONIEET AL.: VOLCANICPRODUCTSOFTHE MARQUESASHOTSPOT
.
ark
Iti
Huka
Pou
Nuku
Huku
Hiv;
O8
ane
Nao
/FatuHiva
12
12
I
I
141'
140'
139'
138'
137'
,6'
Fig. 1. Bathymetricmap of the Marquesaslineament. Arrowsindicatelocationsof dredgehauls;after GEBCO Map 297,
Centre National pour rExploitation des Oceans,Paris, 1973. Contour interval = 500 m, dotted !in, contour= 200 m. Insert
indicatesthe location of the Marquesasbathymetricmap and showsmajor tectonicfeaturesof the Pacific Ocean basin:
GalapagosFractureZone (GFZ), MarquesasFractureZone (MFZ), and the EastPacificRise.
over the Marquesashotspot [McNutt et al., 1989]. A long,
broad topographicand geoid swell extendswestwardfrom the
Marquesas lineament through the center of the north-south
trending Line Islands volcanic lineament. This feature, termed
the Line-Crosstrend, has allowed speculationthat at least some
of the centralLine Islandsformedabovethe Marquesashotspot
[Croughand Jarrard, 1981]. Three of the volcanoes,with ages
between38 and 60 Ma, show an age progressionwhich could
be related to volcanism along the Marquesas-Line trend
[Schlanger et al., 1984]. If so, the Marquesashotspot has
producedvolcanic activity only intermittently.
In contrast to the three other volcanic
chains of the French
Polynesian region (Society, Austral-Cook, Pitcairn-Gambier),
no active volcanism
has been found at the southeastern end of
the Marquesasline [McNutt et al., 1989]. McNutt et al. [1989]
proposedthat the most recent expressionof volcanism above
the Marquesashotspot may be found at the northern ridge
marginof the MarquesasFractureZone (MFZ). Furtherstudyof
this ridge has shown that it is an old feature unrelated to
hotspot activity (M. K. McNutt, personal communication,
1991). In September1991, however,freshglassybasaltswere
dredgedfrom a seamountat the southeastend of the Marquesas
lineament, 20 km north of the MFZ, now thoughtto be the
present location of the hotspot (M. K. McNutt, personal
communication,1991); radiometricages for thesesamplesare
in progress(R. A. Duncan,unpublishedresults,1993).
The duration of volcanism at a single edifice within this
lineament is longer than that measured at other hotspot
locations. Duncan et al. [1986] discoveredthat the 3.8-m.y.
history of volcanism at the island of Ua Pou followed a
petrogeneticsequencesimilar to Hawaiian volcanoes. Shield
lavas exposedin stream valleys, road cuts and wave-cut cliffs
around the perimeter of Ua Pou are part of a tholeiitic basalt
phase of volcanism which occurred from 5.6 to 4.5 Ma.
DESONIEET AL.: VOLCANIC PRODUCTSOF THE MARQUESASHOTSPOT
Volcanism began again after a 1.6 m.y. hiatus with eruption of
alkali basalts from 2.9 to 2.7 Ma, and finally more evolved
lavas from 2.5 to 1.8 Ma [Duncan et al., 1986]. In contrast
with Hawaiian volcanism, the magmatic production rate is
independentof compositionand does not diminish with time,
based on the age relationshipsof the compositionalphaseson
eachof the Marquesanvolcanoes[Brousseet al., 1990].
17,651
Samoa
0.707
0.706
0.705
GeochemicalVariability of SOPITA Volcanic Chains
The south-central Pacific basin presents several remarkable
geochemical and geophysical characteristics. This region
(essentially French Polynesia) has thinner lithosphere
[Nishimura and Forsyth, 1985] and shallowerdepths[McNutt
and Fischer, 1987] than comparablyold lithosphereelsewhere.
Slow lower mantle seismic velocities [Dziewonski and
Woodhouse, 1987] directly beneaththe region correlate with
an unusually high density of hotspot tracks. These volcanoes
exhibit some of the most extreme isotopic compositions
among oceanic basalts[Duncan and Compston,1976], which
Hart [1984] and Castillo [1988] have related to large-scale
convectiveupwelling from the deep mantle. Staudigel et al.
[1991] have shown that the compositional identity of the
volcanism has persistedat least since mid-Cretaceoustime and
speculatedthat the diversity in the island and seamountbasalts
is derived from long-term accumulation and isolation of
subducted lithosphere deep in the mantle. In this
interpretation, recycled material is delivered to the upper
mantlebeneaththe regionvia plumeflow [White and Hofmann,
1982]. The Darwin Rise [McNutt and Judge, 1990] and the
western Pacific Cretaceousatoll and guyot province [Smith et
al., 1989] probably originated over hotspotsin this area. The
term "superswell"has been used to describethis broad, shallow
region of the Pacific correlatedwith deep mantle upwelling and
excessintraplate volcanism[McNutt and Fischer, 1987].
Within the four volcanic lineamentsof French Polynesiaare
some of the most extreme Sr and Pb isotopiccompositionsyet
measured in ocean island basalts (OIB). The mantle
componentsvisualized as endmember source compositionsin
globally distributed oceanic volcanism [White, 1985; Zindler
and Hart, 1986] must have developedfrom long-termisolation
0.704
Pitcairn
0.703
Is.
PREMA
ter Is.
0.702
16.0
17.0
18.0
19.0
20.0
21.0
22.0
206pb/204pb
Fig.2. 87Sr/86Sr
versus
206pb/204pb
for southPacificspreading
ridge
basaltsand hotspotvolcaniclineamentswith mantlecomponents
after
Zindler and Hart [1986] shownin hatchuredboxes. Althoughdata from
most island chains are grouped,data from each volcano in the CookAustral lineamentis shownas a separatefield but with the samedotted
patternandprint. Basaltsfrom the islandof Ua Pou [Duncanet al., 1986]
are tholeiitic basalts(solid circles) and alkalic basalts(solid triangles).
Data sourcesare as follows: East Pacific Rise (EPR) [White et al., 1987];
Easter Island [White and Hofrnann, 1982]; Society Islands [White and
Hofrnann, 1982; Cheng et al., 1987; Devey et al., 1990; W. M. White,
unpublished
data, 1990]; MarquesasIslands[Vidal et al., 1984;Duncanet
al., 1986; Dupuy et al., 1987; Desonie, 1990]; Cook and Austral Islands
[Palacz and Saunders, 1986; Nakarnura and Tatsurnoto,1988]; Pitcairn
Island[Woodheadand McCulloch,1989];andthe SamoanIslands[Wright
and White, 1986;Farley et al., 1990]; afterHart and Zindler [1989].
extreme heterogeneityis found within a single volcano. The
quartz-normative tholeiites from Ua Pou are among the most
depleted
in 87Sr/86Sr
foroceanislands
yetfound,buttheyhave
consistently higher Pb-isotope ratios than MORB. The more
radiogenic Pb-isotope ratios of these tholeiites may be due to
the incorporation of pods of a HIMU component within the
DMM mantle source [Duncan et al., 1986; Hart and Zindler,
of materials with variable U/Pb, Th/Pb, Rb/Sr and Sm/Nd; these
1989]. Alkalic lavas from the same volcano, however, are
intermediate in Pb isotopic compositions and have highly
are well representedin south-centralPacific lavas. Figure 2
shows spreading ridge and hotspot lavas from the region radiogenic Sr isotopic ratios, characteristicof an EMII source
plottedin 87Sr/86Sr-2ø6pb/2ø4pb
spacewith hypothesizedcomposition (Figure 2).
The spatial variability of melt compositions across the
mantle endmember compositions. DMM, depleted mid-ocean
ridge basalt (MORB) mantle, is the low radiogenicSr and Pb, hotspots can be examined through sampling of
globally quite uniform, endmember that gives rise to normal contemporaneous volcanoes within the region of current
spreadingridge lavas. HIMU, the radiogenic Pb, unradiogenic activity. In the group of young volcanoesat the southeastend
Sr component, was defined in part by basaltic rocks from of the Societieschain, betweenTahiti and Mehetia, an isotopic
in 87Sr/86Sr
perpendicular
to thehotspot
trackis found
Mangaia and Tubuai in the Cook-AustralIslands,and appearsto gradient
be present in Easter Island lavas and seafloor basalts from the in which values of 0.7060 at Rocard seamount to the north,
EPR.
EMII,
with highly radiogenic Sr but moderate decrease to values of 0.7031 at Moua Pihaa seamount to the
2ø6pb/2ø4pb,
is mostevidentin samples
fromSamoabut is south [Devey et al., 1990]. Isotopic gradientsin the direction
also strongly expressed in lavas from the Society and the of plate motion are also observed. For example, as mentioned
=
Marquesas Islands. EM I component, with moderately high above,at the islandof Ua Pou,tholeiiticbasalts(87Sr/86Sr
= 0.7055)by> 1.5
87Sr/86Sr
butlow2ø6pb/2ø4pb
is represented
by samples
from 0.7027)predatealkalicbasalts(87Sr/86Sr
Pitcairn Island and Rarotongain the Cook Islands.
Compositional diversity is observed along and across
lineaments, and on the scale of individual volcanoes. In the
Cook-Austral
lineament
islands
exhibit
rather
uniform
compositions, whereas great isotopic diversity is found
between volcanoes (Figure 2). In the Marquesas lineament
m.y. Given the high plate velocities inferred from Pacific
volcano migration rates [Duncan and Clague, 1985], this
implies a sample spacingof 100 to 150 km acrossthe hotspot.
Duncan et al. [1986] suggestedthat the composition of the
eruptive products at Ua Pou was related to position of the
volcano above a compositionally segregatedhotspot.
17,652
DESONIE
ETAL.:VOLCANICPRODUCTS
OFTHEMARQUESAS
HOTSPOT
NEW RESULTSFROMTHE MARQUESAS
ISLANDSAND SWELL
19-2), alkali basalts (19-3, 19-11), and one basanite (19-8).
The diversity of compositionsfrom this samplesite is similar
to that observedon the island of Ua Pou. Hyaloclastiteswere
Samplesand Analytical Methods
dredgedfrom threelocations(DH 17, DH 18, and DH 21).
Basalt samplesexamined in this study were dredgedfrom
Samples used for age determinations and compositional
submarineportions of the Marquesasvolcanic chain in March
analyses were the freshest available, with slight to moderate
and April of 1987 by the R/V ThomasWashington,expedition alteration and little or no interstitial glass. All CRGN 02
Crossgrain(CRGN 02). The objectivesof the programwere samples, except the potassic trachyandesites,are highly
twofold: to study the thermo-mechanicalproperties of the vesicular. Tholeiitic basaltsinclude olivine (some embayed)
lithosphere underlying the Marquesas lineament and to
and abundant clinopyroxene phenocrysts. Fe-Ti oxides,
determine the chemical composition of the early phase of
clinopyroxene, and small amounts of interstitial glass are
constructionof the volcanoes. A portion of the resultsof Sea found in the groundmass. Basalts (transitional series) are
Beam swath mapping, a single channel seismic study, and similar to olivine tholeiites but include small laths of
gravity and magneticsurveyswere discussedby McNutt et al.
groundmassplagioclase. Abundant embayedolivine crystals
[1989] and Filmer and McNutt [1990].
are found as phenocrystsand in the groundmassof alkali
In this paper we report analysesof rocks recoveredin nine olivine basalts. Augitic clinopyroxene is abundant as well.
dredgehauls of seamounts,banks,and the submarineportions Although DH 21 samples are classified chemically as alkali
of islands from CRGN 02. Dredge locations from which olivine basalts, they contain no olivine, little clinopyroxene
basalticrocks were recoveredare shownin Figure 1 and brief and have abundant small plagioclase laths in a trachytic
descriptions of the dredge targets are given in Table 1. matrix. The highly vesicular tephrites contain abundant
Previousstudiesof Marquesanvolcanismhave relied solely on clinopyroxene (rhomboid phenocrysts), and some olivine.
samplesfrom the emergent portions of the volcanoeswhich Little or no plagioclase is found in the tephritic rocks.
representlessthan 2% of the massof any edifice.
Basanites are fine grained with abundant olivine,
For petrologiccomparisonof dredgedlavas and glasseswith clinopyroxene,
and plagioclase
phenocrysts and
those from emergent portions of the islands, initial basalt microphenocrysts.Vesicles in the basanitoidsare extremely
classifications are those used by Liotard et al. [1986],
round, and up to 1 mm in diameter. Highly alkalic samples
following Feigenson et al. [1983] for Hawaiian basalts. contain abundanttitanaugitewhich is sometimessectorzoned
Hypersthene-normative basalts are classified as tholeiitic if
with normally and reversely zoned rims.
Potassic
they lie in the tholeiitic field on an alkalis versussilica plot trachyandesitescontain fine grained plagioclaselaths, minor
and as transitional if they lie in the alkalic field. All
clinopyroxeneand Fe-Ti oxides in a trachyticmatrix.
nepheline-normativesamplesare alkalic basalts. Although all
Age determinations
by 4øAr-39Arincremental
heatingand
three compositional types are representedin the CRGN 02
total fusion experimentsand conventionalK-Ar radiometric
sample suite (Tables 4 and 5), highly alkalic rocks (>30%
datingmethodswere performedon 23 freshbasaltsamplesfrom
normative nepheline), which are abundant among subaerial the CRGN 02 dredgecollectionand six samplesfrom the island
samples,were not recovered. Rock nameswere then given to
of Fatu Hiva. Samplepreparationand analysisof samplesby
membersof each of these lava types using the International K-Ar and 4øAr-39Ar total fusion methods followed the
Union of Geological Sciences classification based on their
techniques outlined by Desonie and Duncan [1990].
major and trace element compositions, and normative and Experimental
methods
for 4øAr-39Ar
incremental
heating
are
modal mineralogies. Glasses were named based on their
describedby Duncanand Hargraves[ 1990]. Air measurements
to determine the mass discrimination factor, critical to
positionon an alkalies versussilica diagram.
The dredge hauls contained glassesand well-crystallized correction for atmosphericargon, were performed after each
rocks. In general, the dredgedmaterial providedthe samerock two sample runs for total fusion analyses and before each
types found on the islands;most individual dredgescollected stepwiseheating experiment.
rocks of essentially uniform composition. For example,
Becauseof the potential for potassiumaddition and argon
nearly identical ages and compositions from all analyzed loss in seawater, K-Ar analyses on crystalline submarine
samplesfrom DH 12 indicatethat only one flow sequencewas volcanicrocksprovideonly a minimumage. 4øAr-39Ar
sampled. A few dredgehaulscontaineda greatervarietyof rock incremental heating experiments can provide a reliable
types. The mostdiverse,DH 19, from a fault scarpon the south crystallizationage if a plateauin the age versustemperature
side of the island of Tahuata, returned olivine tholeiites (19-1,
spectrumis formed and appropriatestatisticalcriteria are met.
TABLE 1. LocationandDescriptionof CRGN 02 DredgeHaulsin WhichBasalticRocksWere Recovered
Station
DH
DH
DH
DH
DH
DH
DH
DH
DH
12
14
16
17
18
19
21
22
24
Latitudes,
øS
10ø49.1 '
10ø32.3 '
10o26.0 '
10o31.5'
10o23.8 '
10o05.9 '
9o39.6 '
9ø13.6'
9o25.2 '
Longitudes,
øW
138ø25.9 '
138ø50.8 '
138ø45.8 '
138o48.7 '
138o32.2 '
139o08.8 '
139o43.7 '
140o07.6 '
140o14.6 '
Depth,
m
2985-3136
2442-2077
2307-1936
2437-2348
1992-1758
1936-1717
2524-2296
2036-1692
2780-2006
Feature
small seamount 50 km southeast of Fatu Hiva
scarpin basaldomewestof Fatu Hiva
N-S trendingscarpbasaldomewestof FatuHiva
scarpin basaldomesouthwestof FatuHiva
southwest flank of Motu Nao
flank of N-S trendingridge6 km southof Tahuata
west flank Dumont D'Urville
Nord rift zone
smallconeon northslopeof Ua Pou
riftzoneonsteep
slope
westof UaPou
DESONIE
ETAL.:VOLCANIC
PRODUCTS
OFTHEMARQUESAS
HOTS
POT
For a plateau to be defined, two contiguous steps with
calculatedages within lc• and representingat least 50% of the
17,653
a VG Sector thermal ionization mass spectrometerat Cornell
University, following the procedureof White et al. [ 1990].
total39Arreleased
fromthesample
wererequired.If threeor
more steps were used in the calculation an isochron age
(36Ar/4øAr versus39Ar/4øAr) was also calculated. The
4øAr/36Arinterceptin eachof thesecalculations
(Table3)
shows no evidence for nonatmospheric, initial argon. (By
virtue of the distribution of the few points used in the
Age Determinations
Analytical
dataandagecalculations
fromtheK-Ar and4øAr39Ar total fusionand incremental
heatingexperiments
are
given in Tables 2 and 3. Figure 3 showsthe variationin age of
regressions,
threeisochrons
yielded4øAr/36Ar
intercepts
less 12 islands and eight dredge sites with distance along the
than the atmospheric value, 295.5, and resulting calculated
agesare too old. In thesecaseswe acceptthe plateauage as the
best estimate of the crystallization age.) Because of the
expected young ages and only moderatepotassiumcontentsin
lineament,measuredfrom the positionof DH 12. Samplesfrom
CRGN 02 expand the age range previously seen at some
volcanoes. Because the current location of the Marquesas
hotspot relative to the Pacific plate is unknown, we use the
someof theselavas,andthuslowamounts
of radiogenic
4øAr, location of DH 12 as the youngest point in the Marquesas
4øAr-39Ar
totalfusionexperiments
weredoneonsomedredged lineament for plate motion calculations. DH 12 is the most
southeasterlylocation from which ages have been determined;
samples.
the age of basalts from these seamounts is significantly
Major and trace elementanalysesof the dredgedbasaltswere
youngerthan thosefrom Fatu Hiva.
carried out by X ray fluorescencespectroscopy (XRF) using
Becausevolcanism at a single volcano can last for as much
fused glass disks, as describedby Hooper [ 1981]. Rare earth
as 3-4 m.y. in the Marquesas [Duncan et al., 1986] and the
element (REE) and Hf concentrations were determined by
instrumental neutron activation analysis (INAA) following the
volcanic record at the various sites has been incompletely
method of Laul [ 1979]. Five basalt fragments(14-1, 16-1, 16sampled, we have calculated the rate of plate motion over the
Marquesas hotspot in three ways. For all calculations, the
2, 16-3, and 19-11), too small for analysis by XRF, were
mostreliableage determinations
available(4øAr-39Artotal
powdered, fused under controlled atmosphere (QFM) and
analyzedfor major elementsby electronmicroprobeat Oregon fusion or incremental heating for submergedbasalts, K-Ar for
island samples) were used in a least squareslinear regression.
State University. Compositionsof 13 Marquesanglasseswere
In one calculation,the oldest age from each island and dredge
analyzed by electron microprobe at Scripps Institution of
location was used. In another analysis, only basaltsknown to
Oceanography(Table 5).
be of tholeiitic or transitional composition, and thus
For analyses of radiogenic isotope ratios, 0.5- to 1.0-mm
representing a common phase in volcano development, were
chips of all sampleswere leachedin warm, distilled 6 N HCI for
1 hour to attempt to eliminate effects of seawater alteration. used. Regressionrates for both of these cases,which we take
Sr, Nd, and Pb concentrates were then obtained from dissolved
to estimate the time when the individual
powdersby ion exchange,and isotopicratios were measuredon
over the center of the hotspot,are identical at 74 + 6 km/m.y.
volcanoes were located
TABLE2. K-Arand40Ar-39Ar
TotalFusion
AgesforBasaltsFromCRGN02 Dredges
andtheIslandof FatuHiva,
MarquesasSwell
Sample
%K
40ArRadiogenic,
x10-8mol/•
% 40Ar
Age_+1 o, Ma
Radio[[enic
CRGN 02
DH
12-1
DH 12-10
DH 12-11
DH
17-1
DH 17-2
DH 18-26
DH 18-40
DH 19-1
2.11
3.2
[n.d.]
[2.02]
[2.07]
1.49
4.9
4.4
[n.d.]
DH
19-8
DH 21-8
DH 21-14
DH 22-1
2.31
15.8
[1.12]
[n.d.]
DH 24-4
DH 24-7
1.54
2.57
2.62
DH 24-8
[n.d.]
0.39 + 0.04
4.3
4.5
5.2
0.35 + 0.04
0.45 + 0.10
0.48 + 0.10
18.7
[1.14]
[0.71]
[n.d.]
0.73
4.4
15.0
29.6
30.2
0.84 + 0.03
9.4
4.1
4.9
1.31 _+ 0.12
0.74 + 0.12
1.27 + 0.03
6.9
1.57 + 0.05
4.9
1.58 + 0.08
26.1
1.76 _+ 0.10
11.4
25.1
3.13 + 0.14
2.81 + 0.04
32.8
27.4
31.9
2.49 _+ 0.04
2.97 _+ 0.04
2.96 + 0.03
7.8
2.45 + 0.14
Fatu Hiva
FT01-02
FT 01-23
FT 01-39
FT 01-42
FT 01-61
FT 01-65
0.95
1.41
1.47
1.47
1.02
1.07
4.4
10.0
8.2
7.2
7.6
7.3
6.6
5.2
5.5
17.5
7.1
5.7
1.18
1.84
1.43
1.24
1.91
1.76
+ 0.06
_+ 0.14
+ 0.05
+ 0.02
+ 0.09
+ 0.12
Agescalculated
using
thefollowing
decay
andabundance
constants:
)•e= 0.581x10'-10yr1; )•[5= 4.963x10-10yr
-1;
40K/K= 1.167x10-4mol/mol.
% K determined
by atomicabsorption
spectrophotometry,
or calculated
frommeasured
39Ar(in brackets;
n.d.= notdetermined).
17,654
DESONIE ET AL.: VOLCANIC PRODUCTSOF THE MARQUESASHOTSPOT
TABLE3. 40Ar-39ArPlateauandIsochron
AgesFromCRGN02 Basalts
Sample
Plateau
Age+ 1c•.Ma
39Ar,
Integrated
Isochron
Age,
Wei[[ht
by1/o2 Wei[[ht
b•,%39Ar%ofTotalAl•e,Ma
DH
DH
DH
DH
DH
DH
DH
DH
12-17
16-1
17-1
17-5
18-13
19-2
19-3
19-8
0.76+_0.10
1.31 + 0.04
1.70 + 0.07
1.69+0.11
1.27 +0.10
1.74 +-0.08
1.97 +- 0.12
1.87+-0.05
DH 19-11
nonedeveloped
DH
DH
DH
DH
DH
2.63
2.69
3.50
3.24
3.28
21-14
22-1
22-2
24-4
24-7
+ 0.19
+ 0.07
+ 0.08
+0.09
+ 0.07
N SUMS/(N-2) 40Ar/36Ar
+ lo,Ma
0.76+0.24
1.28 + 0.13
1.60 + 0.20
1.60+0.38
1.29+-0.20
1.58+0.23
1.86 + 0.19
1.83+0.06
100.0
94.7
87.0
100.0
67.0
96.2
99.4
99.4
0.76
1.22
1.73
1.60
1.92
1.58
1.82
1.82
0.60+0.05
1.22 + 0.03
1.54 + 0.37
1.25+ 1.52
2.28+-0.36
1.85+-0.12
1.75+0.18
1.60
....
2.60 + 0.28
2.69 +- 0.24
3.43 + 0.16
3.36+0.20
3.27 + 0.17
73.8
92.3
75.2
81.8
94.6
2.61
2.87
3.85
3.12
3.23
3.57 + 0.12
3.17 +0.14
3.03 + 0.20
Intercept
6
4
5
6
4
3
2
5
0.3
0.0
0.8
3.1
1.0
0.7
2.4
295.6+2.4
296.5 + 0.7
283.4 + 9.9
298.1 + 11.0
286.2+2.5
291.6+0.5
293.4+-2.8
2
2
3
5
5
0.2
1.0
0.4
280.9 + 4.4
290.1 + 4.4
296.1 + 11.9
Corrected
for 37Ardecaysinceirradiation,
half-life= 35.1 days.Agescalculated
usingthefollowingdecayandabundance
constants:
)•œ=0.581x10
-10yr-1;
)•[J=4.963x10
-10yr-1;
40K/K=l.167x10
-4mol/mol.
Neutron
fluxmonitored
withhornblende
standardMMhb-1 (520.4 Ma).
These
estimates
are low
relative
to those
calculated
from
averagevolcanoagescompiledfrom all samplesfor which age
determinationshave been performed. A third regressionof all
140 age determinations
[Duncanand McDougall, 1974;Duncan
available to them in their regression. Regressionlines were
extrapolatedto zero age to determinethe currentlocationof the
Marquesashotspot. For the regressionrate calculatedfrom the
tholeiitic/transitional basalts the current hotspot location is
et al., 1986; Katao et al., 1988; Brousse et al., 1990; Desonie,
37 km southeast
1990; D. L. Desonie,unpublisheddata, 1992] leadsto a rate of
93 + 7 krn/m.y. This rate is lower than (but within statistical
uncertainty) the rate of 104 + 18 km/m.y. calculated by
McDougall and Duncan [1980] who used all island data
Marquesan basalts the current hotspot location is 49 km
southeastof DH 12. In view of the more completesamplingof
these volcanoes and the evidence for a common compositional
sequenceat each center,we concludethat the most appropriate
of DH
12.
For the rate calculated
for all
Distance from DH 12 (km)
100
0
200
I
300
I
400
I
I
7j +Katao
et
al.,
1988
6
o Brousse
etal.,1990
O DuncanandMcDougall,1974
and Duncan et al., 1986
5
,•,
A Desonie,
K-Ar
total
fusion
(unpub.)
o
[] 40Ar-39Ar
andK-Artotalfusion
• 4 ß 40Ar-39Ar
incremental
heating
o
II
[]
•
o• •:•o
o
74 63
Fig. 3. Basaltageversusdistance
fromDH 12 for seamounts
andislandsof theMarquesas
lineament.Distances
aremeasured
alongthetrendof theMarquesas
chain(N40øW).Solidsymbols
represent
rocksanalyzed
by the40Ar-39Ar
incremental
heating
technique
forwhicha plateau
agewascalculated.
Opensymbols
represent
datagenerated
by40Ar-39Ar
orK-Artotal
fusionexperiments.
Two leastsquares
regression
linesarepictured:onewitha slopeof 74 km/m.y.calculated
fromoldestages
at eachvolcaniccenterandonewith a slopeof 93 km/m.y.calculatedfrom all availableMarquesas
agedeterminations.
A third
regression
line for all tholeiiticand transitional
basaltsamplesfor whichageshavebeendetermined,
alsowith a slope74
km/my,is not pictured. The interceptof eachregression
line with thex axisshowsthe predictedpresentlocationof the
Marquesas
hotspot
southeast
of DH 12. Islandlocations
in thisfigureandin Figure12 areindicated
by thefollowingsymbols:
Bc J. Goguel(BJ), Hatutu(H), Eiao (E), NukuHiva (NH), Ua Huka(UH), Ua Pou(UP), FatuHuku(FH), Hiva Oa (HO),
Tahuata(T), Motane (M), and Fatu Hiva (FH).
DESONIEET AL.: VOLCANICPRODUCTSOFTHE MARQUESASHOTSPOT
velocity for Pacific plate motion over the Marquesashotspotis
our new estimate of 74 km/m.y. We later discussimplications
for motion of the Marquesas hotspot with respect to other
Pacific hotspots.
DH 19 is the only dredge haul that containsrepresentatives
of tholeiitic, transitional, and alkalic magma types. Lack of
stratigraphicinformationfrom dredgesamplingdoesnot allow
relative position of the three compositional phases of
volcanism to be known. Ages of each of the three phasesof
volcanismcan be comparedwith the sequencefoundon Ua Pou
where tholeiitic shield lavas predatealkalic cap rocks [Duncan
et al., 1986]. The youngestsamplesfrom DH 19 are tholeiitic
(19-1:1.58 + 0.08 Ma; 19-2:1.70 + 0.08 Ma), but the agesare
similar to ages from transitional(19-3:1.97 + 0.12 Ma) and
alkalic (19-8:1.87 + 0.03 Ma) basalts. Although at Ua Pou a
distinct age versuscompositionrelationshipis seen, samples
from DH 19 show that the two compositionaltypes sometimes
overlap in time, perhaps especially during the submarine
construction
of the volcanoes.
New sampleswere also collectedfrom the emergentportion
of the islandof Fatu Hiva duringthe CRGN 02 expedition. Age
determinations, performed by K-Ar methods, expand the age
rangepreviouslyseenat that edifice and are includedin Table 2.
Compositional analysesfor this sample suite are planned [K.
Johnson,personalcommunication,1992]. Twenty new K-Ar
age determinationswere also performedfor emergentsamples
from the island of Ua Huka [H. Barsczus, personal
communication, 1991 ].
ChemicalVariability of MarquesanBasalts
Previousstudiesof MarquesasIslands samples[Bishop and
Woolley, 1973; Liotard et al., 1986;Duncan et al., 1986; Vidal
et al., 1987] have documenteda range in compositionfrom
tholeiitic to alkalic basalts, including compositions as
evolved as trachytes and phonolites. Although quartz
normative tholeiitic basalts (-1% quartz) are found on the
islandsof Ua Pou [Liotard et al., 1986] and Eiao [Caroff et al.,
1990], most tholeiitic basalts from the Marquesas have
olivine-normative compositions [Liotard et al., 1986]. The
tholeiitic basalts can be distinguishedfrom alkalic basaltsby
lower rare earth element (REE) concentrations,especially in
the light REEs, and by lower total incompatible element
concentrations and ratios [Liotard et al., 1986]. Isotopic
compositiondoesnot uniformly correlatewith major and trace
element composition. In general, however, alkali basalts
show more radiogenicSr and moderatelyradiogenicPb, and
tholeiitic/transitional basalts have relatively unradiogenic Sr
and variablePb, indicatingderivationfrom mantlesourceswith
distinctSr/Rb and U/Pb histories[Dupuy et al., 1987].
The distributionof basalt compositionsalong the lineament
gives a picture of the general petrologic developmentof a
Marquesanvolcano. Drill holes throughthe volcanicpile on
the northwesternmost(oldest) island of Eiao (Figure 1) have
sampled moderately alkalic basalts in the upper 900 m and
quartz-bearingtholeiites in the lower 100 m [Caroff et al.,
1990]. Tholeiitic basalts have also been collected from the
exposedshield areasthat underliethe alkalic cappinglavas of
the central islands of Ua Pou [Liotard et al., 1986], Nuku Hiva
and Hiva Oa [Duncan, 1975]. In dredge hauls on two younger
volcanoes, tholeiitic basalts have been recovered where only
alkalic basaltshad been found subaerially:at Tahuata (DH 19)
and Fatu Hiva (DH 14) (Figure 1). Transitionalbasaltsare often
17,655
found within the volcanic shield formed principally of
tholeiitic
basalts.
A
closer
look
at
Fatu
Hiva,
the
southeasternmost(youngest) island (Figure 1), shows that a
full quarter of the volcanic edifice has collapsed below sea
level. Deep submarinefault scarpsboundingthe flanks of this
volcano have exposed tholeiitic (DH 14) and transitional
basalts (DH 16). We combine these observationsto conclude
that the bulk of the submergedshieldof a Marquesanvolcanois
composedof hypersthene-normative
tholeiitic and transitional
basalts that grade into, or are overlain after some break in
volcanic activity by alkalic basalts. Sea Beam swath mapping
[McNutt et al., 1989] defined the location of several small
cones southeast of Fatu Hiva, close to the inferred location of
the hotspot. We cannot be certain if the compositionof these
cones (DH 12) representsthe compositionof the initial phase
of Marquesanvolcanismor if that initial phaseis buried by DH
12 volcanism. Ages and chemical compositionssuggestthat
DH 12 sampledonly one flow.
Substantial crystal fractionation is indicated by the wide
rangeof MgO content(11.16-2.52 wt %) and Mg numbers(67-
37) [=100 Mg+2/(Mg+2+Fe+2)](Tables4 and 5) of the
submarine lavas. None of the lavas, even the most primitive
amongthe dredgedsamples,appearto have fractionatedolivine
alone. The more primitive lavas (Mg numbers >60) can be
divided into three groups based on major and incompatible
trace element concentrations,and Sr and Nd isotoperatios. The
liquidsparentalto each of thesethree groupshave resultedfrom
variable melting conditions (depth, percent melting) of
compositionallydistinct sources. More evolved rocks can be
related to these parental compositions by fractional
crystallization of primarily olivine, clinopyroxene, and
relatively less orthopyroxeneand plagioclase(Figure 4). In
Group I, the least fractionatedlavas are tholeiitic basalts(DH
14, 19-1, 19-2); they are relatively depleted in incompatible
elementsand show unradiogenicisotope ratios, especiallySr.
Based on major element compositions(49% SiO2, 9.5% FeO,
2.0% Na20), the parentalliquidsfor theselavaswere generated
by about 20% partial melting of somewhatdepletedperidotite
at 12 to 16 kbar pressure(40 to 60 km depth) [Falloon et al.,
1988]. The most primitive lavas of Group II are mildly alkalic
or transitional in composition (DH 12, 17, 18, 19-3)and
exhibit relatively unradiogenicisotopic ratios. Parental lavas
for the strongly alkalic group, Group III (DH 22, 24) are
relatively enriched in incompatible elements and have
relatively radiogenic isotopic ratios. Parental liquids (43%
SiO2, 11.5% FeO, 2.8% Na20) for these lavas must have come
from deeper than 60 km, after rather small degreesof partial
melting (<5%) [Falloon et al., 1988].
The drop in CaO/A1203 with decreasingMg number (Figure
4) seen in the compositionaltrends for each group, indicates
clinopyroxene fractionation. Although similar fractional
crystallization paths are observed in all lava groups, the
parental compositionsin each group cannot be derived from
one another by that process because of the diversity of
CaO/A1203 at a given Mg number. Instead,variabledegreesof
partial melting of a mantle source with rather constantmajor
element composition could produce the observed trend of
decreasingCaO/A1203 with increasingMg-number within the
parentallavas within each group [e.g., Falloon et al., 1988].
In a plot of Y versusNb (Figure 5) parentallavasclusterinto
three groups, each similar to the other in Y content but with
variable Nb content. Because isotopic ratios for each of the
groups are different, .variationsin Nb, the more incompatible
17,656
DESONIE ET AL.: VOLCANIC PRODUCTSOF THE MARQUESASHOTSPOT
TABLE 4. Major ElementAnalysesandCIPW-NormativeCompositions
for CRGN 02 Basalts
12-1
FTI
12-6
12-10
12-11
12-12
12-17
14-1
16-1
16-2
16-3
17-1
17-2
FTI
FTI
FTI
FTI
FTI
FTI
FTI
FTI
FTI
FTI
FTI
18-13
17-5
FTI
MTN
Major ElementAnalyses
SiO2
TiO2
AI20 3
FeO
51.73
2.26
17.44
8.29
Fe203
1.38
MnO
0.20
MgO
2.84
CaO
7.19
Na20
K20
P205
4.98
2.54
1.28
Total
100.13
Mg number 38
51.53
2.20
17.48
8.16
1.36
0.20
2.93
7.01
4.74
2.82
1.25
99.69
39
51.64
2.27
17.56
8.30
1.38
0.20
2.52
7.20
5.23
2.55
1.32
100.17
39
16.67
35.75
17.98
2.36
0
7.24
0
10.60
1.97
4.18
2.96
15.07
36.68
16.91
4.10
0
8.59
0
9.45
2.00
4.31
3.12
51.67 51.60
2.27
2.27
17.51 17.68
8.24
8.25
1.37
1.38
0.20
0.20
2.77
2.89
7.19
7.17
4.98
4.92
2.58
2.56
1.29
1.29
100.07 100.11
37
38
51.23
2.26
17.41
8.20
1.37
0.20
2.89
7.17
4.97
2.85
1.29
99.85
39
49.10
1.97
13.26
9.63
1.44
0.14
11.16
10.01
2.42
0.57
0.21
99.76
65
48.49
3.84
14.67
10.23
1.53
0.15
6.54
8.79
3.18
1.55
0.69
99.50
50
48.90
3.78
14.44
8.95
1.34
0.12
7.57
9.90
2.72
1.16
0.41
99.16
48
48.17
4.14
14.69
9.97
1.50
0.16
5.73
10.17
2.82
1.26
0.50
98.96
57
46.37
3.87
12.31
10.31
1.72
0.16
9.63
9.53
2.53
1.80
0.53
98.76
62
43.57
3.66
11.28
9.88
1.65
0.16
9.59
8.98
2.33
1.66
0.50
93.26
63
46.36
3.85
12.25
10.48
1.75
6.86
23.02
23.77
0
0
18.35
11.84
5.39
1.94
7.18
0.97
7.45
23.86
23.70
0
0
19.24
9.02
4.64
2.17
7.86
1.18
10.64
19.14
16.92
1.23
0
21.75
0
18.01
2.49
7.35
1.25
9.82
17.27
15.42
1.32
0
20.92
0
18.21
2.18
6.95
1.18
10.52
18.37
16.63
1.83
0
21.87
0
19.12
2.13
7.31
II
AOB
II
AOB
0.16
9.99
9.50
2.57
1.78
0.53
99.22
60
51.21
2.21
17.28
8.13
1.36
0.22
2.91
7.00
4.92
2.79
1.27
99.29
39
CIPW-Normative Compositions
or
ab
an
ne
lc
di
hy
ol
mt
il
ap
Group
Type
15.01
37.01
17.73
2.78
0
8.05
0
10.45
1.83
4.29
3.03
II
KT
18-26
MTN
II
KT
18-40
MTN
II
KT
15.25
36.89
17.81
2.85
0
7.93
0
10.07
1.99
4.31
3.05
15.13
36.84
18.60
2.60
0
7.17
0
10.58
2.00
4.31
3.05
II
KT
II
KT
19-3
THU
16.87
33.77
16.81
4.53
0
8.72
0
10.20
1.81
4.30
3.06
3.37
20.48
23.64
0
0
19.95
11.79
14.38
2.09
3.74
0.50
9.16
26.91
21.18
0
0
14.66
6.00
10.64
2.22
7.29
1.63
II
KT
I
OT
II
B
II
B
19-8
THU
19-11
THU
21-8
DMD
21-14
DMD
19-1
THU
19-2
THU
47.40
3.37
13.06
9.96
1.49
48.25
2.93
12.02
10.47
1.57
48.34
2.91
11.81
10.08
1.51
47.48
3.19
11.30
9.91
1.49
45.07
3.73
13.74
10.92
1.64
48.33
3.82
14.74
10.01
1.50
48.68
3.57
15.50
10.38
1.56
0.30
9.48
9.24
2.99
1.77
0.56
0.20
10.06
10.56
2.09
0.88
0.34
0.16
9.93
10.53
2.08
0.95
0.34
0.16
11.30
10.40
2.32
1.19
0.40
0.18
7.55
10.91
3.20
2.78
0.65
0.18
5.28
10.03
2.69
1.95
0.55
99.62
63
99.37
63
98.64
64
99.14
67
100.37
56
98.92
46
10.46
20.76
16.99
2.46
0
20.41
0
18.68
2.16
6.40
1.33
II
AOB
5.20
17.69
20.82
0
0
23.81
13.34
9.88
2.28
5.56
0.80
I
OT
5.61
17.60
20.08
0
0
24.26
14.28
8.30
2.19
5.53
0.80
I
OT
7.03
19.63
16.91
0
0
25.86
2.80
17.97
2.16
6.06
0.95
II
B
II
B
1.25
II
AOB
16.49
34.84
16.83
3.68
0
8.06
0
10.46
1.79
4.20
3.01
II
KT
22-1
UAP
22-2
UAP
22-3
UAP
24-4
UAP
24-7
UAP
24-8
UAP
47.59
4.09
15.30
11.39
1.71
43.43
3.46
12.92
11.26
1.69
42.88
3.31
12.57
11.48
1.72
43.43
3.33
12.81
11.55
1.73
44.58
44.35
44.29
3.37
15.12
3.42
15.04
3.38
15.47
10.41
1.56
10.56
1.58
10.21
1.53
0.19
4.05
8.42
4.07
1.75
0.78
0.19
4.58
9.26
3.58
1.37
0.77
0.23
9.92
12.30
2.53
1.86
0.45
0.20
10.26
11.70
2.69
1.93
0.43
0.26
9.95
11.91
2.82
1.96
0.43
0.22
6.10
10.02
4.07
3.09
0.75
0.22
5.87
9.93
4.52
3.16
0.20
5.56
9.77
4.20
3.25
0.76
0.78
98.95
41
99.83
42
100.05
61
99.17
62
100.17
61
99.29
51
99.41
50
98.65
50
9.71
0
16.52
12.33
1.33
31.46
0
18.04
2.50
6.29
1.02
III
TE
10.04
0
16.51
12.93
1.21
32.37
0
17.30
2.51
6.32
1.02
Ill
TE
18.26
3.08
13.86
16.99
0
25.46
0
11.25
2.26
6.40
1.78
Ill
BT
18.67
1.32
11.42
20.01
0
27.09
0
10.36
2.29
6.50
1.80
III
BT
19.21
2.68
13.76
17.80
0
24.41
0
10.34
2.22
6.42
1.85
III
BT
Major ElementAnalysis
SiO2
TiO2
AI20 3
47.92
3.76
12.23
FeO
10.15
Fe20 3
MnO
1.52
0.17
MgO
10.00
CaO
10.42
Na20
K20
P205
Total
2.56
1.22
0.52
100.47
Mg number 64
CIPW-Normative Compositions
or
7.21
ab
an
21.66
18.28
ne
0
lc
0
24.25
di
hy
ol
mt
il
2.41
16.10
2.21
7.14
ap
Group
1.23
II
Type
B
16.43
4.14
14.92
12.43
0
28.58
0
12.92
2.37
7.08
1.54
III
BT
11.52
22.76
22.39
0
0
19.57
5.12
7.01
2.17
7.26
1.30
II
B
10.34
31.35
18.86
1.68
0
14.84
0
11.04
2.26
6.78
1.85
I
AOB
8.10
29.28
21.63
0.55
0
16.03
0
12.22
2.48
7.77
1.82
I
AOB
10.99
0.04
18.40
11.58
0
32.24
0
16.73
2.45
6.57
1.07
Ill
TE
,!
FTI = Fatu Hiva, MTN = Motu Nao, THU = Tahuata,DMD = DumontD'Urville, UAP = Ua Pou, TH = tholeiitiicseriesbasalt,TR =
transitionalseriesbasalt,ALK = alkali seriesbasalt,OT = olivinetholeiite,B = basalt,AOB = alkali olivinebasalt,KT = potassic
trachyandesite,
TE = tephrite,BT = basanite.
element, cannot be explained by only different degrees of
partial melting of a single sourceand must reflect a range of
source compositions for the primitive lavas. Linear trends
exist betweenthe primitive and more evolved lavas within each
group that reflect fractionation of minerals in which both
elementsare equally incompatible.
Isotopic and Trace Element Compositions
Sourceheterogeneityhas been implicatedfrom trace element
and isotopic compositions of island samples from the
Marquesaslineament[Liotard et al., 1986;Duncan et al., 1986;
Dupuy et al., 1987; Vidal et al., 1987]. The diversity of
DESONIEET AL.: VOLCANICPRODUCTSOFTHE MARQUESASHOTSPOT
17,657
TABLE 5. ElectronMicroprobeAnalysesandCIPW-NormativeCompositions
for CRGN 02 DredgedBasalts
12-10
FTI
14-1
FTI
16-1
FTI
17-1/5
FTI
18-20
FTI
18-60
FTI
19-1
FTI
54.53
2.14
17.00
48.99
2.17
14.00
51.12
4.22
14.11
46.39
4.31
14.02
45.12
3.25
17.51
47.81
3.75
15.24
48.79
3.31
13.00
8.26
0.20
9.65
0.14
11.23
0.15
11.27
0.16
10.46
0.24
10.51
0.14
11.42
0.14
19-7
FTI
21-1C
DMD
21-1,B
DMD
22-3
UAP
22-5
UAP
46.86
3.20
17.45
46.91
3.80
15.14
48.12
3.66
15.50
48.24
3.77
15.43
44.67
3.27
17.51
10.05
0.13
11.88
0.20
11.81
0.19
10.70
0.15
10.28
0.20
24-1/7
UAP
ElectronMicroprobeAnalyses
SiO2
TiO2
AI20 3
FeO*
MnO
MgO
3.14
CaO
6.36
Na20
K20
P205
4.60
2.67
0.91
Total
6.43
10.07
2.82
0.56
0.23
96.93
Mg number 44
Q
3.90
10.20
3.11
1.43
0.51
3.97
2.98
0.60
4.69
11.13
3.37
2.00
0.50
5.86
3.35
4.69
4.55
4.19
8.57
8.57
8.88
8.73
9.64
2.53
0.88
0.34
3.75
3.20
0.88
3.63
1.37
0.58
3.12
1.54
0.58
3.37
2.06
0.56
3.88
3.63
10.11
9.30
4.20
3.17
0.59
4.43
3.74
0.72
97.30
98.70
98.23
97.38
98.33
98.14
97.08
97.80
98.10
97.88
96.93
50
41
49
44
48
41
44
45
44
45
44
44
0.00
15.76
38.73
18.07
0.09
6.23
2.79
2.43
0.91
5.27
10.91
9.63
0.21
96.99
0.00
Or
Ab
An
Ne
Di
3.82
7.92
44.95
3.07
17.17
4.24
3.31
23.84
26.06
0.00
21.46
14.34
23.59
18.79
0.00
11.99
CIPW-normativecompositions
0.00
0.00
0.22
0.00
0.00
0.44
22.17
20.06
2.24
25.08
17.59
7.34
21.15
14.21
21.06
11.80
23.01
20.53
2.97
21.67
5.19
21.39
23.69
0.00
23.96
23.02
13.06
19.26
10.10
14.5
Hy
0.00
12.54
13.08
0.00
0.00
0.00
14.00
0.00
Ol
Mt
I1
9.94
1.85
4.07
2.59
2.17
4.13
0.00
2.53
8.03
7.50
2.53
8.20
6.93
2.36
6.18
7.51
2.36
7.13
0.00
2.57
6.30
7.82
2.26
6.09
Ap
Group
Type
2.10
II
TA
0.23
I
OT
2.10
II
B
1.18
II
AOB
1.34
II
AOB
0.79
I
QT
2.03
III
HAW
1.39
II
BT
0.00
8.09
28.87
20.96
0.99
15.90
0.00
0.00
0.00
9.09
26.38
23.72
0.00
13.01
12.16
26.02
20.88
1.34
19.19
0.00
18.71
4.64
19.56
16.73
22.05
0.00
22.07
4.27
15.91
17.99
21.38
10.25
0.00
0.00
0.00
2.13
2.26
6.96
7.67
2.4
7.17
0.00
2.31
6.22
5.62
2.17
5.84
1.27
III
AOB
1.36
III
BT
1.66
III
BT
11.05
2.68
7.23
1.34
II
AOB
1.34
II
B
FTI = Fatu Hiva, THU = Tahuata,DMD = Dumont D'Urville, UAP = Ua Pou, TH = tholeiitiic seriesbasalt,TR = transitionalseriesbasalt,
ALK = alkali seriesbasalt,QT = quartztholeiite,OT = olivinetholeiite,B = basalt,AOB = alkali olivinebasalt,TA = trachyandesite,
BT =
basanite, HAW = hawaiite.
70
1.0
60
"-7
•
0.5
0.4
0.3
30
•
.
10
•
decreeing
degre
•
paaial
melting
'
'
g0
'
ß
GroupI
ß
GroupII
ß
GroupIII
•
o
[]
Group I
Group II
a
Group III
50
40
70
Mg-number
I
II
30
Fig. 4. CaO/AI20 3 versusMg numberfor CRGN 02 lavas. Cross
indicates composition of glasses,other symbols representwhole rock
compositions. Lines are shown connecting whole rock and glass
compositions
from the sine samplewith samplenumberon the connecting
line. Whe• the mostp•mitive compositionswithin a trend •e tholeiitic,
thoselavasand their possibledifferentiates•e called Group I; where the
most p•mitive compositionswithin a trend •e •ldly •k•ic, thoselavas
and their possibledifferentiates•e called Group II. Group III refers to
p•mitive alk•ic lavas and their possibledifferentiates. Directionsof the
changeof liquid compositionswith olivine (ol), clinopyroxene(cpx), and
plagioclase(plag) •moval •e shown. A•ow iMicates the composition•
ch•ge of liquids with decreasingdegreeof p•ial melting of a single
homogeneoussource. Rocks and their glasses•e related by c•stal
fractionation. All lavas appear to have undergoneol and some cpx
remov• exceptDH 17-5 which hashad olivine •mov• only.
20
20
.....
do
do
Nb (ppm)
Fig. 5. Y (ppm) versusNb (ppm) for CRGN 02 basalts. Arrows show
fractionation relationships within the three compositional groups as
describedin Figure4. Solid symbolsrepresentlavaswith Mg number>60.
Marquesaslineament is nearly as great as in all OIB globally.
A similarlylargevariation
is foundin 143Nd/144Nd,
buta much
smaller range is found in Pb isotope ratios. A plot of eNd
versus87Sr/86Sr for Marquesansamples(Figure 6a)
incompatible trace elements, reported in Table 6, and
radiogenicisotoperatios (Table 7) found in CRGN 02 dredged
samples supports this idea [Desonie, 1990]. Within the
submarine
samplesuite87Sr/86Srrangesfrom 0.70328to
0.70551, and the total Sr isotopic variation within the
demonstratesthat the tholeiitic basalts have less radiogenic $r
than the transitional and alkalic basalts. Although tholeiitic
and alkalic basalt isotopic compositions are distinct, mixing
betweentwo sourcesis prevalent. At Ua Pou and Tahuata (DH
19) tholeiitic/transitional and alkalic basalts are clearly
separated in isotopic composition space, emphasizing the
17,658
DESONIEET AL.' VOLCANIC PRODUCTSOFTHE MARQUESASHOTSPOT
TABLE 6. Trace ElementAnalysesfor SamplesDredgedon CRGN 02
12-1
12-6
FTI
FTI
12-10 i2-11 12-12 12-17 16-2 17-1 17-2 17-5 18-13 18-26
FTI
FTI
FTI
FTI
FTI
FTI
FTI
FTI
18-40
FTI
FTI
FTI
567
0
0
22
116
6
10
60
13
1225
11
94
44
133
464
254
537
79
24
59
234
6
29
30
605
6
292
31
112
276
327
380
46
22
60
228
10
7.1
44.9
85.2
43.9
9.2
3.02
1.01
1.91
0.26
8.4
48.8
103.0
64.2
10.6
XRF
Ba
Cr
Cu
Ga
Nb
Ni
Pb
Rb
Sc
Sr
Th
V
Y
Zn
Zr
569
1
14
20
112
3
8
57
12
1231
9
106
44
132
439
577
2
17
24
113
5
10
49
7
1222
12
94
44
137
460
584
0
36
27
114
4
7
58
13
1247
10
115
43
155
452
592
0
20
22
110
3
8
55
10
1235
10
119
43
132
439
565
1
14
23
114
2
7
56
10
1225
8
108
45
133
448
555
0
6
24
114
2
8
56
8
1227
9
106
44
128
459
246
376
66
23
64
236
7
29
22
673
6
259
30
120
289
34
249
390
65
21
62
269
8
29
25
655
6
243
31
123
282
249
391
71
23
63
257
7
30
24
670
5
275
34
122
289
34
25
783
4
217
30
128
316
INAA
Hf
La
Ce
Nd
Sm
Eu
Tb
Yb
Lu
12.2
93.9
198.6
98.2
16.5
5.07
1.55
3.39
0.43
(La/Sm)
N 3.51
Group
Type
II
M
12.4
91.8
199.9
103.0
15.2
4.96
1.94
3.10
0.40
11.5
92.6
187.4
98.9
16.3
4.90
1.61
3.57
0.45
12.3
95.2
190.8
97.2
16.4
5.07
1.58
2.71
0.42
12.5
93.5
196.2
96.6
16.1
5.17
1.43
2.86
0.44
3.72
3.51
3.58
3.58
II
M
II
M
II
M
II
M
12.2
91.0
192.7
93.5
15.5
5.02
1.68
2.84
0.46
7.7
36.7
85.3
47.7
8.1
2.94
0.88
1.94
0.18
3.63
19-2
THU
19-3
THU
19-8
THU
21-8
DMN
21-14
DMN
133
522
75
18
32
234
6
19
32
436
3
259
29
104
198
144
522
73
17
35
224
4
19
28
435
2
256
29
96
199
262
711
82
20
43
296
6
29
31
504
6
267
27
108
221
616
147
49
22
88
110
8
77
27
820
10
307
35
120
331
311
3
41
30
69
8
6
27
16
877
2
238
50
169
447
263
4
14
26
60
0
8
21
22
873
5
280
45
154
394
9.5
81.2
151.0
76.5
10.9
3.54
1.51
2.57
0.26
12.4
49.4
109.8
81.6
16.0
5.33
1.94
3.09
0.42
8.1
44.0
92.9
47.1
9.3
3.14
1.18
1.68
0.23
7.5
43.5
92.0
50.4
9.4
3.35
1.27
2.05
0.25
12.1
94.8
188.8
89.8
16.0
4.90
1.59
3.32
0.39
2.93
2.85
3.66
3.93
3.85
3.66
II
B
II
AOB
II
AOB
II
AOB
II
M
II
B
II
M
19-1
THU
6.6
34.6
69.8
37.9
8.3
2.72
0.95
1.96
0.25
22-1
UAP
22-2
UAP
22-3
UAP
24-4
UAP
24-7
UAP
24-8
UAP
557
406
11
20
72
145
7
55
35
710
8
353
30
103
201
525
392
16
18
71
148
7
59
31
685
8
339
28
102
199
562
355
31
21
70
153
8
61
29
684
9
338
26
107
199
709
117
28
23
100
54
7
85
23
947
10
267
32
114
302
743
124
28
21
100
51
12
88
19
952
12
268
32
116
301
755
99
28
23
103
40
10
91
21
1002
11
266
35
118
315
5.7
67.6
111.1
54.4
8.1
2.64
1.03
2.55
0.29
5.5
59.4
103.7
49.2
8.2
2.48
0.92
2.20
0.29
5.9
61.8
102.5
47.8
8.3
2.33
1.06
2.55
0.28
7.5
77.2
145.5
72.9
10.0
2.96
1.05
2.40
0.29
7.7
86.0
154.7
67.1
10.4
3.22
1.00
2.31
0.30
5.09
III
BT
3.37
1.12
2.45
0.42
3.01
II
AOB
XRF
Ba
Cr
Cu
Ga
Nb
Ni
Pb
Rb
Sc
Sr
Th
V
Y
Zn
Zr
INAA
Hf
La
Ce
Nd
Sm
Eu
Tb
Yb
Lu
(La/Sm)
N
6.7
32.7
69.3
41.6
8.1
2.52
1.01
1.90
0.28
2.48
5.0
25.4
49.7
29.3
6.5
2.16
0.97
1.91
0.29
2.42
6.0
33.6
67.7
43.3
7.2
2.46
0.98
1.55
0.18
2.86
4.58
Group
I
I
II
III
Type
OT
OT
B
BT
1.91
I
AOB
11.3
43.2
98.1
77.4
14.2
4.80
1.64
2.95
0.38
1.88
I
AOB
4.47
4.58
III
5.12
III
III
4.76
III
TE
TE
TE
BT
8.0
78.2
153.8
62.6
10.1
3.12
1.20
2.92
0.39
4.76
III
BT
FTI = Fatu Hiva, THU = Tahuata,DMD = DumontD'Urville, UAP = Ua Pou, OT = olivine tholeiite,B = basalt,AOB = alkali
olivinebasalt,M = mugeariticbasalt,TE = tephriticbasalt,BT = basanatic
basalt.
close physical proximity of the sourcesfor the two magma
types, and their availability during constructionof individual
volcanoes. Closer inspection of the diagram reveals a
significant geographic pattern to the compositions. At
constant
end, the87Sr/86Sr
becomes
moreradiogenic
across
the Marquesaslineament,from southwestto northeast(Figure
6b). The presenceof tholeiitic samplesat all positionsshows
that this isotopic gradient is not strictly a function of the
degree of partial mantle melting. Intriguingly, the cross
lineament St-isotopic gradient at the southeast end of the
Society chain [Devey et al., 1990] is also increasingly
radiogenicfrom southwestto northeast.
The same separationseenin Sr and Nd isotopiccomposition
for Ua Pou and DH 19 alsoappearsin 87Sr/86Srversus
2ø6pb/2ø4pb
(Figure7a). In thesediagrams,
tholeiiticand
some transitional samples plot near a mixing line between
DESONIE
ETAL.:VOLCANIC
PRODUCTS
OFTHEMARQUESAS
HOTSPOT
17,659
TABLE 7. Sr, Nd andPb IsotopicCompositions
for CRGN 02 Basalts
Group
12-1
12-17
14-1
16-1
16-3
17-1
17-5
18-13
19-1
19-3
19-8
19-11
21-14
22-1
22-2
24-4
24-7
24-8
II
II
I
II
II
II
II
II
I
II
III
II
I
III
III
III
III
III
Type
87Sr/86Sr143Nd/144Nd
end
M
M
OT
B
B
AOB
AOB
B
OT
B
BT
B
AOB
TE
BT
BT
BT
BT
0.703737
0.703753
0.703490
0.703909
0.703473
0.703650
0.703628
0.703759
0.703673
0.704180
0.705506
0.705198
0.703276
0.704920
0.704915
0.705042
0.704955
0.705052
0.512835
3.8
0.512886
0.512842
0.512869
0.512864
0.512860
0.512840
0.512875
0.512818
0.512690
0.512744
0.512895
0.512700
0.512727
0.512703
4.8
3.9
4.5
4.4
4.3
3.9
4.6
3.5
1.0
2.0
5.0
1.2
1.7
2.9
0.512714
1.5
206pb/204pb
207pb/204pb
208pb/204pb
19.448
19.444
19.522
19.651
19.677
19.530
19.542
19.485
19.334
19.260
19.133
19.286
19.323
19.146
15.590
15.554
15.592
15.621
15.598
15.602
15.615
15.605
15.535
15.571
15.632
15.614
15.562
15.620
39.119
39.025
39.163
39.283
39.331
39.158
39.202
39.193
38.867
39.257
39.183
39.294
38.977
39.086
19.203
19.167
19.218
15.606
15.558
15.624
39.175
39.015
39.240
TH= tholeiitiic
series
•asalt,
TR= transitional
series
basalt,
ALK= alkaliseries
basalt,
OT= olivine
tholeiite,
• = basalt,
AOB = alkali olivinebasalt,M = mugeariticbasalt,TE = tephriticbasalt,BT = basaniticbasalt. Valuesarerelativeto the
following
standards
with2oerrorinparentheses:
E&A= 0.708020
(3.093x 10-5),
LaJolla= 0.511840
(6.693x 10-5),
NBS
981' 206pb/204pb
= 16.906(0.016),206pb/204pb
= 15.448(0.020),208pb/204pb
= 36.558(0.109).
Vidal et al., 1984
Duncan et al., 1986
et
al.,
1986
I ooDuncan
Vidal
et
al.,
1984
0.707
•
• CRGN 02 dredges
Dupuy et al., 1987
CRGN 02 dredges
,-,'_-
Ss
&
0.705
1
'"S S•
0.7041
0.703(a)
•DH19•
.?.?.:................................?.::
(a)
18.0
18.5
19.0
19.5
20.0
center
northeast
0.707 -
center
0.706
southwest
0.705
southwest
0.704
(b)
(b)
0.703
0.702 0.•03 0.704 0.•05 0.•06 0.•07 0.708
87Sr/86Sr
Fig. 6. EpsilonNd versus87Sr/86Srfor Marquesan
basaltswith (a)
compositionalfields for Ua Pou and DH 19 and (b) geographictrends
acrossthe volcanic lineament. Solid symbolsindicate tholeiitic basalts.
Major elementdata were not availablefor analysesof Vidal et al. [1984]
so normative compositionscould not be determinedfor these samples.
Islands in the southwesterngroup of the lineament include Ua Pou,
Dumont D'Urville, Tahuata, Motu Nao, Fatu Hiva, and the seamount at
DH 12. The centralgroupincludesEiao, Hatutu, Hiva Oa, and Motane;
the northeastern
groupincludesBc J.Goguel,Ua Huka, andFatuHuku.
0.702
18.0
18.5
19.0
19.5
20.0
2o6pb
/ 2o4pb
Fig.7. 87Sr/86Sr
versus206pb/204pb
for Marquesan
basalts,
with (a)
compositionaltrends for Ua Pou and DH 19 and (b) geographictrends
acrossthe volcanic lineament. Tholeiitic basalts are shown by solid
symbols. Islandsare groupedgeographicallyas describedin Figure 5.
Proposedcompositionsof mantle componentsfrom Zindler and Hart
[1986] are shown in cross-hatchedboxes.
17,660
DESONIE ET AL.: VOLCANIC PRODUCTSOF THE MARQUESASHOTSPOT
intermediate,and Group III lavas show the converse. Such a
relationship is consistent with either melting of two
compositionally distinct mantle sources (one generally
DMM+HIMU-like, the other EM II-like), followed by mixing
of the melts, or variable melting of a single source that is
heterogeneous
on a small length scale (kilometer or less). A
general trend in incompatible element ratios across the
lineament, complementaryto the isotopic gradients, is also
seen. For example,in an incompatibleelementratio-ratioplot
(Figure 9) the southwestand northeastgroups of volcanoes
show separateand distinct compositions,varying from a low
Ba/La, high K/Rb component(easterngroup) to a high Ba/La,
low K/Rb component (western group). The central region
ubiquitousand becomesdiluted by the higher degreesof appearsto be a zone of mixing betweenthe southwesternand
melting involved in producingthe tholeiitic basalts,or it is northeastern compositions.
The covariation of isotopic and trace element ratios with
derived from a spatially separatepart of the hotspotthat is
tappedprincipallyduring the later volcanicstageof volcano mixing and degree of partial melting is further explored in
construction. Both endmembers seem to be of regional
Figure 10, where œNdis plotted against (La/Sm)N. Isotopic
significance.
The first (DMM + HIMU) composition
is seenat ratios are not affected by variable melting or crystallization,
Easter Island, some of the Cook-Austral islands, the
whereas incompatible element ratios may be controlled,
posterosionallavas at Tahiti [Cheng et al., 1987; R. A. especially by melting. œNd values for hypothetical source
Duncan, unpublisheddata, 1987], as well as the Ua Pou endmembersfor melting were chosento match the composition
tholeiites; the second (EMII) appears in the shield lavas at of the most EM II-like basalt (DH 19-8) and an arbitrary
Samoa, the Society, the Marquesas,and some of the Cook- compositionbetweenDMM and the mostDMM+HIMU-like Ua
Austral Islands(Figure 2).
Pou tholeiite [Duncan et al., 1986]. The (La/Sm)N values for
A geographicpatternto the isotopicvariationscan also be these endmembers were estimated from the same basalt
recognizedin Figure7b. Samplesfrom the southwestern
sideof compositionsby consideringratios necessaryto generatethe
the Marquesaslineamentshowa largerHIMU componentwhile basalts with less than 25% melting of spinel peridotire
the central samplesare more EM II-like and the northeastern [Falloon et al., 1988]. The (La/Sm)N ratios for hypothetical2
DMM and HIMU while more alkalic compositionsexhibit a
strongEMII influence[Dupuyet al., 1987]. It hasbeennoted
recently [Hart et al., 1992] that lavas from many hotspot
provinces define sublinear Sr-Nd-Pb isotopic arrays that
converge toward depleted compositions,implying mixing
between mantle plumes and the depleted upper mantle
(asthenosphere
and/or lithosphere).Hart et al. [1992] found
that the depletedmaterialin thesemixing arrays,however,was
not strictly DMM, but a compositionfalling betweenDMM
and HIMU (which they termed FOZO, for "Focus Zone"),
exactlyas we seefor the Marquesasarray. In the presentcase
the EMII componentis more strongly expressedat lower
degreesof partial melting. Thus either this componentis
Samplesshow more of a DMM component. Samplingin the to 25% equilibriummeltsof theseendmembersourceswerethen
Marquesaschain,however,hasbeenbiasedtowardthe western calculatedand displayedas mixing lines in Figure 10. The
region and the full compositionalvariability of the eastern resultsshow that the Marquesanmelt compositions(corrected
region must await further studies.
for low pressurecrystalfractionation)can be accommodated
by
Basalts from the CRGN 02 sample collection can also be simplebinary mixing of the two sourcesif degreeof melting
distinguishedin large-ion lithophile element ratio diagrams, varies from 20% (tholeiites) to less than 2% (basanites).
suchas Zr/Nb versus(La/Sm)N(Figure8), whereGroupsI, II and
Chondrite-normalized REE patterns show enrichments
III are clearly separated. The trace element and isotopic relative to MORB in all Marquesan basalts (Figure 11).
variability is remarkablyhighly correlated:GroupI lavashave Compositionsvary rathersmoothlyfrom moderatelylight REE
highZr/Nb,lowLa/Smandlow 87Sr/86Sr,
GroupII lavasare enriched(Group I, [La/Sm]N=I.9 to 2.4) to stronglylight REE
enriched(GroupIII, [La/Sm]N=4.5to 5.1). GroupI basaltshave
14
6
'.703
2-i•55....-:••
.70367
.7
0365
'"••.
"'
'"/"
7
03
7
4.70363....
'":':----'•.70418
12
10
8
6
...:•....•0551
ß70504[."•.....!.••iil..•_
70505
ß70492.70504
•n•":'•::....-•...
-:...
;'.-:,:.ii:.•'
70492
'
Vidaletal., 1984
Duncan et al., 1986
Dupuyetal., 1987
thisstudy
I
•
•
•
(La/Sm)
N
}
6
Fig.8. Zr/Nbversus
(La/Sm)
N with87Sr/86Sr
indicated
forCRGN02
lavas. Basaltsof the three trends(GroupsI, II, and III) clusterbetween
EMII and DMM/HIMU components.
0
' 160' 260' 360' 4(•0' 560' 61•0700
K/Rb
Fig. 9. Ba/La versusK/Rb with geographictrendsacrossthe volcanic
lineament. Islandsare groupedgeographically
as described
in Figure6.
Tholeiiticbasaltsare shownby solidsymbols.
DESONIE ET AL.' VOLCANIC PRODUCTSOF THE MARQUESASHOTSPOT
lO
17,661
1000 -
ß GroupI &
Group
I
UAP tholeiites
DMM+HIMU
c•19-1
ß 19-2
ß GroupII
,•,1
'i iI Ii
•
ß
:.
ß 21-8
GroupIII
•..,•.,,..
:••________•...•
•.,,.,..
100
o 21-14
.
•*•'"'"•"
• "[3
....
?4x•.,,.•
10=
_
_
_
(a)
_
_
1000
ß
Group
II
•_•,'*•
+ 18-13
'•'•,'.:•
•,'....•••,•'•,q•_•
EMII 20%10% 5%
'
i ' k'
'
'••••r•
2%
'
' 6
DMM+HIMU
and EMII
components are approximately end-
[] 18-40
1000
_
[]
Group
III
19-8
ß 22-1
8,
(La/Sm)N - 1, andend - 1, (La/Sm)N -1.7, respectively.A mixingline
for the componentsis shown. The horizontallines indicatepercentof
partial meltingof a DMM+HIMU or an EMII compositionwith curved
linesconnecting
equivalentdegreesof melting. Tholeiiticsamplesare the
resultof higher degreesof meltingof a more dominantlyDMM+HIMU
source;alkalic samplesderive from lower degreesof meltingof a more
EM II-like
_
x 18-26
(b)
(a/Sm)N8.0
Fig. 10. EpsilonNd versus(La/Sm)N for CRGN 02 basalts.To adjustfor
shallowlevel fractionationeffects(La/Sm)N composition
was calculated
for parental liquids (Mg number = 65-70).
Compositions for
12-1
o 12-17
100_
.......••,•1,••--•,,
,,.,,..,....,
......
,,,...,
.,,,,..
o 24-7
_.1
.....
+ 24-8
•t
source.
. 24-4
...........
•g-•
--•m,,,.•
Ua
Pou
tholerites
ns
"••,........,•
10-:
(c)
patterns very similar to the Ua Pou tholeiites (field shown by
Liotard et al. [1986]); the slightly concave-downwardlight
REE portion is reminiscent of Hawaiian tholeiites [e.g.,
Budahn and Schmitt, 1985]. The progressivefanning of these
patterns to higher light REE contents at approximately
constant heavy REE contents from Group I to II to III is
consistentwith decreasingpartial melting of peridotitc within
the spinel to garnet stability fields. A few of the more
primitive Group II lavas (e.g., 19-3) show lower heavy REE
contents than the Ua Pou tholeiites, and the Group III lavas
show patterns with slight concave-upwardcurvature, indicative
of garnet control during melting. All the Group III lavas,
largely basanites, are evolved (Mg number 50-62) and their
primitive melts would have had significantlylower heavy REE
contents, reflecting residual garnet in the region of melting.
Within each group patterns are easily related by fractional
crystallization, dominantly of olivine and clinopyroxene,
consistentwith the major element fractionationpaths (Figure
4). The lack of any significant Eu anomaly suggeststhat
plagioclaseaddition or removal was not a major factor in the
petrogenesisof these suites.
DISCUSSION
Volcanism that has built the Marquesas lineament is
understandablein terms of Pacific plate motion acrossa quasistationary hotspot, maintained by a plume of upwardly
convecting mantle. The individual islands and seamounts of
this chain show a clear age progression and orientation
La
Ce
Nd
Sm Eu
Tb
Yb
Lu
Fig. 11. REE patternsfor all Marquesasdredgedsamples:(a) patternfor
Group I trend lavas, (b) patternfor GroupII trendlavas,(c) patternfor
Group III trend lavas. Chondrite normalized after Nakarnura [1974].
REE patternsfor tholeiiticbasaltsfrom the islandof Ua Pou are shownas
a dotted field.
generally similar to other Pacific hotspot lineaments, and
Marquesas basalt compositions fall within the range for
hotspot-related volcanoes elsewhere in the Pacific Ocean
basin, particularly the French Polynesian region. Several
aspectsof this volcanic province are remarkable,however, and
provide new clues about the dynamic behavior of plumes and
the thermal structureof hotspots.
First, the Marquesas chain is short and began to grow only
about 6 m.y. ago, in contrast with longer-lived and more
vigorous activity over the Hawaiian and Easter hotspots,and,
to a lesser extent, the Cook-Austral and Society hotspots.
Some evidence suggeststhat the Marquesas hotspot was more
active earlier (60 to 30 Ma) and producedvolcanoesthroughthe
region of the central Line Islands [Crough and Jarrard, 1981;
Schlangeret al., 1984], on strike from the MarquesasIslandsin
the direction of Pacific plate motion. In this case, the plume
flux must have varied considerably, perhaps in conjunction
with changing vulnerability of the overlying lithosphere to
melt penetration [Pollack et al., 1981; McNutt et al., 1989].
Buoyancyflux calculationsof Sleep [1990] would suggestthat
17,662
DESONIEETAL.:VOLCANICPRODUCTS
OFTHEMARQUESAS
HOTSPOT
the Marquesasplume is one of the most vigorousin the ocean
basins. His estimatewas made by determiningthe buoyancy
flux for the entire South Pacific Superswell and dividing the
value by the number of hotspots (5) found within the
Superswell [Sleep, 1990]. However, the Superswell is not
caused by the coalescing of the individual hotspot swells
within the region; abundanthotspotvolcanism is found within
the Superswell possibly because the lithosphere was
previously warmed, allowing it to be penetrated by the
hotspots[McNutt and Fischer, 1987]. Therefore calculating
the buoyancy flux of a single hotspot in this manner would
seriously overestimate the value for that hotspot. The
Superswell may, in fact, be a longer-lived phenomenonthan
any of the hotspotscurrently found in the region [Staudigel et
al., 1991; Larson, 1991].
Second, the orientation of the chain is rotated clockwise
about 30ø from the directionexpectedfrom rigid Pacific plate
motion over a hotspotthat is fixed with respectto other Pacific
basin hotspots[Duncan and Clague, 1985; Gripp and Gordon,
1990]. Coupled with this aberrantgeometry,our best estimate
of plate velocity acrossthe hotspot (volcano migration rate) is
significantlyslower (by 30 mm/yr) than the rate expectedfrom
age determinationsin other Pacific lineaments. McNutt et al.
[1989] have proposed that the orientation of the Marquesas
chain results from the coincidence of a limited region of
weakness in the oceanic lithosphere located between the
Marquesasand Galapagosfracture zones (Figure 1) and aligned
parallel with the ancient spreadingaxis (NNW), and the path of
a weak Marquesashotspot. What could cause such a proposed
"window" of vulnerability is unspecified. Because of the
putative low plume flux and small melt productionrates at this
hotspot, magmas normally cannot penetrate the old, thick
oceanic lithosphere. However, eastward sublithosphericflow
from the hotspot as the more vulnerable section of Pacific
lithosphereapproached(6 Ma), direct vertical supply when the
two coincided,and westwardflow from the hotspotas the plate
window receded(today) could producea volcanicchain rotated
clockwise relative to the plate motion vector (Figure 12). The
orientation would depend on the width of the "window" in the
lithosphereand on how far plume material could flow from the
hotspot to the vulnerable lithosphere. An alternative
possibility (which we explore below) is that the weak
Marquesas plume is more susceptibleto shear by horizontal
flow in the upper mantle and has been displaced to the
southwestrelative to strongerPacific hotspots.
Third, we observein the Marquesasbasaltsan unusuallylarge
range in isotopic compositions,which is strongly correlated
with major and trace element contents.
A common
petrogenetic sequence in volcano construction is seen
throughout the lineament, from tholeiitic and transitional
lavas that consistentlyform the main shield-building volcanic
phase, to terminal alkalic and basanitic lavas.
(The
composition of initial submarine eruptions is less clear, but
mildly alkalic lavas are found at the site of most recent
activity, DH 12.) Marquesantholeiitic lavas are among the
least radiogenicin Sr in Pacific islands,but show moderately
radiogenicPb, and reflect a mantle sourcecompositionlying
on a mixing line betweenDMM and HIMU. Alkalic lavas,by
contrast,show very radiogenicSr and manifestmelting of EM
II mantle.
The major element variationsresult largely from the thermal
structure and conditions of melting across the hotspot:
tholeiitic magmas develop from relatively large degrees of
mantle melting at shallow depths,while alkalic magmasderive
from smaller degreesof melting at deeperlevels [e.g., Falloon
et al., 1988]. REE patternsare consistentwith this sequence
and indicate melting progressing from the spinel to garnet
stability fields with time. This presumablycorrespondsto the
location of any individual volcano over the center and trailing
edge, respectively, of the hotspot [Duncan et al., 1986]. But
while changesin the major element contentsof magmaswith
time are controlled by phase equilibria of rather uniform
peridotite composition, isotopic and, to a lesser extent, trace
element compositions require melting of compositionally
BJ
BC
MI
UH
/ Zone
a •Fr,
acture
Fig. 12. Plan view of the Marquesasvolcaniclineamentshowingthe hypothetical
effectof plumeshearby westwardupper
mantleflow. The orientation
of the chainis rotated30ø clockwise(74 km/m.y.)andthe volcanicageprogression
is slowerby
30 km/m.y.,compared
withthatexpected
fromfixedhotspots
(105 km/m.y.). Thisfeature,plusbroadening
of themeltingregion
andinferredcompositional
segregation
withinthe hotspot,couldresultfromwestwarddeflectionof the plumeandentrainment
of uppermantlealongthe plume axis. Small circle is the locationof the deepmantleroot of the plume, dashedlines show
directionof uppermantleshear,andshadedcircleshowspresenthotspot,segregated
by flow into plume(hachured)
andupper
mantle(graystippling)compositions,
aspredictedfromexperiments
by Griffithsand Campbell[1991]. The locationof DR 12 is
indicatedby "12";islandlocationsshownas symbolsasin Figure3 with the additionof Bc Clark (BC), Matu Iti (MI), Dumont
D'Urville (DD), and Motu Nao (MN).
DESONIEET AL.:VOLCANICPRODUCTS
OFTHEMARQUESAS
HOTSPOT
distinct mantle sources. These sources may be physically
separate regions of the hotspot (center, periphery) or be
intimately mixed and expressedby variable degreesof melting.
How can we accountfor theseaspectsof the temporal,spatial
and compositional variability of Marquesan volcanism? We
considerthree possiblemodels:
1. All compositional variations are contained within a
heterogeneous,but well-mixed plume. These heterogeneities
may be primordial or acquired through subductionof altered
oceanic lithosphere and sedimentswith variable Rb/Sr, U/Pb
and Sm/Nd, and accumulationin the deep mantle over very long
time scales[White and Hofmann, 1982]. Magma compositions
are then controlled through variable partial melting of the
heterogeneous
plume. That is, the lower melting fractions(EM
II-like, alkalic) dominate at low degrees of partial melting
(hotspot edge), while the more residual material (DMM-like,
tholeiitic) dominates at higher degrees of partial melting
(hotspot center), following the "plum-pudding mantle"
rationale for melting-compositioncorrelationsfor EPR basalts
[Zindler et al., 1984]. We cannot,however,proposeany strong
argumentwhy the HIMU component(long-term enrichmentof
U over Pb) should be associatedwith DMM and larger degrees
of partial melting.
2. The plume has a uniform composition(EMII) and, over
cooler parts of the hotspot,this compositiondominatesmelts.
Over the center of the hotspot higher temperatures lead to
assimilation and melting of the lower lithosphere, which
dominates magma compositions. The lower lithospheremust
then have DMM+HIMU composition. Some supportfor this
idea comes from detailed studies of EPR basalts [White et al.,
1987; Graham et al., 1988], where Pb isotopic compositions
appearto streak from DMM toward HIMU (Figure 2). Hence
material added to the lower oceanic lithosphere as the plate
cools
and
thickens
could
contain
both
DMM
and
HIMU
components. However, unlike Hawaii, where posterosional
lavas are isotopically depleted but LREE-enriched so small
degrees of melting of oceanic lithosphere can be invoked to
explain their origin [Chen and Frey, 1983], the amount of
lithosphere necessary to produce the LREE-depleted and
isotopically depleted shield lavas in the Marquesas would be
enormous. Geophysical evidence regarding significant
lithospheric melting is equivocal. Fischer et al. [1986]
estimateda compensationdepthof 45 + 5 km for loadingof the
Marquesas volcanoes, implying convective thinning of the
base of the underlying lithosphere. Calmant and Cazenave
[1987] calculated an elastic thicknessfor the lithosphereunder
the Marquesas somewhat thinner than the global trend, and
concludedthat weakeningby thermal erosion had occurred. In
the most recent analysis of gravity and bathymetric data,
however, McNutt et al. [1991] claim to detect no thinning or
weakening of the oceanic lithospherebeneath the Marquesas
region.
3. The plume is EMII composition,but is weak (relatively
cool and low buoyancy flux). Horizontal flow in the upper
mantle, generally from spreading ridges toward subduction
zones,will shearthe plume [Richardsand Griffiths, 1988] and
in the processupper mantle (DMM+HIMU) will be warmed
arounda boundarylayer and entrainedinto the axis of the plume
[Richards and Griffiths, 1988; Griffiths and Campbell, 1991].
This producesa compositionalsegregationat the hotspotsuch
that plume material (EMII) resides at the edges and upper
mantle (DMM+HIMU) occupiesthe center. Temperatureswill
still be highest in the center of the hotspot, leading to the
17,663
largest degrees of melting (tholeiites), and cooler toward the
edges, producing smaller degrees of melting (alkalic basalts
and basanites). Plume flux may actually be irregular: the
Marquesaschain could representa small pulse of material with
waning flux toward the present.
We can dismiss the first model, that all compositional
diversity results from variable melting of an intrinsically
heterogeneous plume. The sequence of melt compositions
repeated at each volcano along the lineament seemsto require
too regular a distribution of heterogeneitieswithin the plume;
the spatial scale and relative proportion of components
available for melting must have been nearly constantthrough
the history of Marquesan volcanism. This seemsunlikely in
view of the difficulty of thorough subsolidus mixing of
subducted material, even on billion-year time scales [Gurnis
and Davies, 1986].
Also, we would expect isotopic
compositions to lie on mixing lines between low melting
fraction material (EMII
and HIMU) and residual material
(DMM). Instead, the apparentmixing line identifies EMII and
DMM+HIMU
as melting endmembers. The DMM+HIMU
componentreflects involvement of the upper mantle sourcefor
spreadingridge melting, as seen in the streaking from DMM
toward radiogenicPb compositionsalong the East Pacific Rise
[White et al., 1987] and in Easter Island lavas [White and
Hofmann, 1982]. This compositionis a common endmember
for hotspot chains distributed throughout the Pacific basin
(Figure 2) and points to an upper mantle contribution to
melting in all cases[White, 1985; Hart et al., 1992].
The remaining two models both acknowledge mixing of
plume and upper mantle material. Assimilation and melting of
the lower lithosphere over the center of the hotspot, and
entrainment of upper mantle into the axis of the plume are
difficult to distinguish because they produce identical spatial
patterns in melt compositions. To explain the Ua Pou
tholeiites by assimilation of lower lithosphere, however,
would require greater than 1:1 proportions of oceanic
lithosphere (composition of primary melts for EPR lavas
[White et al., 1987] and plume melts (primary melt
compositionfrom DH 19-8 + 15% olivine added). Given the
more refractory composition(and higher melting temperatures)
of the lower lithospherecomparedwith the plume material, it
does not seem possible to dilute the plume melts to such an
extent. This level of wholesale heating and erosion of the
lower lithosphereshould also be manifest in weakeningof the
plate under the volcanic chain [Fischer et al., 1986; Calmant
and Cazenave, 1987] but latest evidence denies this [McNutt et
al., 1991]. Large amounts of lithosphere melting are not
consistentwith the physical evidencefor the intermittentand
relatively small volcanic production over the Marquesas
hotspot. Woodhead [1992] has proposed substantial
lithospheremelting and mixing with plume melts to explain
geochemical variability seen within the Marquesas Islands.
Lithospheric melting is favored becauseof the very regular
successionof distinct compositionalchangesat nearly every
island along the chain, and an apparentcorrelationbetweenthe
elapsedtime betweenvolcanicphasesand the magnitudeof the
compositionalshifts. Accordingto Woodhead [1992], longer
repose times should allow greater pooling of plume melts,
which overcomes the assimilation of lithospheric melts.
However, the long reposetimes indicate a waning plume flux
which should lead to proportionally greater dilution by
lithosphericmelts. There-is certainly no evidencethat longer
repose times lead to greater volumes of posterosionallavas.
17,664
DESONIEET AL.: VOLCANICPRODUCTS
OFTHE MARQUESAS
HOTSPOT
Finally, we note that rocks from DH 19 show as great a
compositional
separation as at Ua Pou, yet are
contemporaneous(i.e., no repose time), so melts of extreme
compositionswere available for eruption at the same time, at
REFERENCES
Bishop, A. C., and A. R. Woolley, A basalt-trachyte-phonolite
series
from Ua Pu, MarquesasIslands, Pacific Ocean, Contrib. Mineral.
Petrol., 39, 309-326, 1973.
Brousse,R., H. G. Barsczus,H. Bellon, J.-M. Cantagrel,C. Diraison,H.
Guillou, and C. Leotot, Les Marquises (Polynrsie franqaise):
Volcanologie,g6ochronologie,
discussion
d'unmod61ede pointchaud,
least at Tahuata.
On the other hand, upper mantle material stirred into the axis
of the plume would occupythe hottestpart of the hotspotand is
Bull. Soc. Geol. Fr., 8, 933-949, 1990.
expected to be associated with the largest degree of partial
Budahn, J. R., and R. A. Schmitt, Petrogeneticmodelling of Hawaiian
tholeiitic basalts: A geochemicalapproach,Geochim. Cosmochim.
melting. The Griffiths and Campbell[1991] analysisof sheared
Acta, 49, 67-87, 1985.
plumes is also instructive with regard to the orientation and
Calmant, S., and A. Cazenave, Worldwide estimates of the oceanic
timing of Marquesan volcanism. We have noted the 30ø
lithosphereelastic thicknessunder volcanoes,Nature, 328, 236-238,
departureof the lineament from the orientationexpectedif the
1987.
Caroff, M., R. C. Maury, P. Vidal, G. Guille, and C. Dupuy, Eiao Island
hotspotwere fixed with respectto other Pacific hotspots.What
(Marquesas): Interactionsbetween EM and HIMU reservoirs,paper
if the buoyancyflux in the Marquesasplume hasbeen weak and
presentedat Volcanismeintraplaque. Le point chaudde la Reunion,
intermittent, so that local upper mantle flow has bent the
Universit6 de la Rrunion, St. Denis-de-la-Reunion, November 12-19,
plume toward the southwest? The required flow direction is
1990.
roughly parallel with the old fracture zones, i.e., relative plate Castillo,P., The Dupal anomalyas a traceof the upwellinglowermantle,
Nature, 336, 667-670, 1988.
motion. In this case the plume flux has waned since the time of
the beginning of the Marquesanchain (6 Ma) and the plume has Chen, C. -Y., and F. A. Frey, Origin of Hawaiian tholeiite and alkalic
basalt, Nature, 302, 785-789, 1983.
been progressivelydeflectedto the southwest(Figure 12). This
Cheng,Q., J. D. MacDougall,
G. W. Lugmair,andJ. Natland,Temporal
would account for the orientation and smaller apparent plate
variationin isotopiccompositionat Tahiti, SocietyIslands(abstract),
Eos Trans. AGU, 68, 1521, 1987.
velocity over the hotspot.
Crough,S. T., and R. D. Jarrard,The Marquesas-LineSwell, J. Geophys.
The magnitude of plume shearing and upper mantle
Res., 86, 11,763-11,771, 1981.
entrainment is dependent on plume buoyancy flux. For a
moderate
buoyancy
fluxof 104Ns-1, plumebending
of 20øfrom
vertical, and reasonable initial (lower mantle) temperature
anomaly (AT = 200-400øC), Griffith and Campbell's
calculationsshow that the ratio of entrainedto plume material
at the hotspot is about 2:1, the diameter of the hotspotshould
be 300-400 km, and the temperatureanomaly over the hotspot
is 70-140øC.
Thus entrainment
increases the volume
of the
Desonie,D. L., Geochemicalexpressionof volcanismin an on-axisand
intraplatehotspot: Cobb and Marquesas,Ph.D. thesis,145 pp., Oreg.
State Univ., Corvallis, 1990.
Desonie, D. L., and R. A. Duncan, The Cobb-Eickelbergseamount
chain: Hotspotvolcanismwith MORB affinity, J. Geophys.Res., 95,
12,697-12,711, 1990.
Devey, C. W., F. Albarede,J. -L. Cheminee,A. Michard, R. Muhe, and
P. Stoffers,Active submarinevolcanismon the Societyhotspotswell
(West Pacific): A geochemicalstudy, J. Geophys.Res., 95, 50495066, 1990.
plume and broadens the region of melting, but cools
Duncan, R. A., Linear volcanismin French Polynesia,Ph.D. thesis, 150
temperatures over the hotspot. In the Marquesas the long
pp., Aust. Nat. Univ., Canberra,1975.
duration of volcanic activity at individual islands implies a
Duncan,R. A., and D. A. Clague,Pacific plate motionrecordedby linear
volcanicchains,The OceanBasinsand Margins, vol. 7A, editedby A.
broad hotspot;e.g., the 2-3 m.y. period from shield to capping
E. Nairn et al., pp. 89-121, New York, 1985.
phases(Ua Pou) indicatesa 150-200 km radius melting region.
Duncan,R. A., and W. Compston,Sr isotopicevidencefor an old mantle
Excess temperaturesof 100øC are more than sufficient to
sourceregion for FrenchPolynesianvolcanism,Geology,4, 728-732,
1976.
generate tholeiitic melts over the center of such a diluted
Duncan,R. A., andR. B. Hargraves,
40Ar-39Argeochronology
of
hotspot [e.g., Falloon et al., 1988].
basement rocks from the Mascarene Plateau, Chagos Bank, and
While some small amount of lithospheric assimilationmust
Maldive Ridge, Proc. Ocean Drill. Program,Sci. Results,115, 43-51,
occur as melts migrate from the hotspotthrough the plate, we
1990.
conclude that this effect is small in the generation of these
Duncan, R. A., and I. McDougall, Migration of volcanismwith time in
melts and that compositional segregationwithin the hotspot,
the MarquesasIslands,French Polynesia,Earth Planet. Sci. Lett, 21,
414-420, 1974.
derived from interaction of the weak plume with horizontal
Duncan,R. A., and I. McDougall,Linearvolcanismin FrenchPolynesia,
flow in the upper mantle, best accountsfor the temporal and
J. Volcanol. Geotherm. Res., 1,197-227, 1976.
spatial variability of Marquesan volcanism. High resolution, Duncan, R. A., M. T. McCulloch, H.G. Barsczus, and D.R. Nelson,
time-space studies of the compositionalevolution of volcanic
Plume versuslithosphericsourcesfor melts at Ua Pou, Marquesas
Islands, Nature, 322, 534-538, 1986.
chainsproducedover plumeswith different buoyancyfluxes, on
lithosphereof different ages and compositions,and at different Dupuy, C., P. Vidal, H. G. Barsczus,and C. Chauvel,Origin of basalts
from the Marquesas Archipelago (south central Pacific Ocean):
plate velocities (variable shear) should allow a comprehensive
Isotopeand traceelementconstraints,Earth Planet. Sci. Lett., 82, 145assessment
of
the
conditions
of
plume
and
152, 1987.
asthenosphere/lithosphere interaction.
Dziewonski,A.M., and J. H. Woodhouse,Global imagesof the Earth's
interior, Science, 236, 37-48, 1987.
Acknowledgments.We thankthe crew of R/V ThomasWashington
for superbtechnicalassistancein collectingdredgedsamples,K. Johnson
for accessto samplesfrom the island of Fatu Hiva, L. G. Hogan for
analytical supportof radiometricdating, and W. M. White for accessto
massspectrometryfacilities at Cornell University and unpublisheddata
from the Society lineament.J. D. Woodheadkindly provideda preprint
including new Sr, Nd isotopic data for MarquesasIslands samples.
Discussions
with H. Barsczusthroughoutthis studyhavebeeninvaluable.
We are gratefulfor the carefulreviewsof C. Chauvel,C. Devey, and H.
Staudigel. Our work has been supported by the National Science
Foundation(grant OCE-8610250).
Falloon, T. J., D. H. Green, C. J. Hatton, and K. J. Harris, Anhydrous
partial melting of a fertile and depletedperidotitefrom 2 to 30 kb and
applicationto basaltpetrogenesis,
J. Petrol., 29, 1257-1282,1988.
Farley, K., J. Natland, J. D. Macdougall,and H. Craig, He, Sr, and Nd
isotopesin Samoanbasalts: Evidence for enrichedand undepleted
mantlecompositions(abstract),Eos Trans.AGU, 71, 1669, 1990.
Feigenson,M.D., A. W. Hofmann,and F. J. Spera,Casestudieson the
origin of basalt,II, The transitionfrom tholeiiticto alkalic volcanism
on Kohala volcano, Hawaii, Contrib. Mineral. Petrol., 84, 390-405,
1983.
Filmer, P. E., and M. K. McNutt, Gravity and flexurein the Marquesas
Islands(abstract),Eos, Trans. AGU, 71, 1598, 1990.
DESONIEET AL.: VOLCANICPRODUCTSOFTHE MARQUESASHOTSPOT
Fischer, K. M., M. K. McNutt, and L. Shure, Thermal and mechanical
constraintson the lithospherebeneaththe Marquesasswell, Nature,
322, 733-736, 1986.
Graham, D. W., A. Zindler, M.D. Kurz, W. J. Jenkins,R. Batiza, and H.
Staudigel,He, Pb, Sr andNd isotopeconstraints
on magmagenesisand
mantle heterogeneitybeneath young Pacific seamounts,Contrib.
Mineral. Petrol., 99, 446-463, 1988.
Griffiths, R. W., and I. Campbell,On the dynamicsof long-livedplume
conduitsin the convectingmantle, Earth Planet. Sci. Lett. 103, 214227, 1991.
Gripp, A. E., and R. G. Gordon, Current plate velocitiesrelative to
hotspotsincorporatingthe NUVEL-1 global plate model, Geophys.
Res. Lett., 17, 1109-1112, 1990.
Gurnis, M., and G. F. Davies, Mixing in numbericalmodelsof mantle
convectionincorporating
platekinematics,J. Geophys.Res.,91, 63756395, 1986.
Hart, S. R., A large-scaleisotopeanomalyin the SouthernHemisphere
mantle, Nature, 309, 753-757, 1984.
Hart, S., and A. Zindler, Constraintson the nature and developmentof
chemical heterogeneitiesin the mantle, Mantle Convection: Plate
Tectonicsand GlobalDynamics,editedby W. R. Peltier,pp. 261-387,
Gordon and Breach, New York, 1989.
Hart, S. R., E. H. Hauri, L. A. Oschmann, and J. A. Whitehead, Mantle
plumesand entrainment: Isotopicevidence,Science,256, 517-520,
1992.
Hooper, P. R., The role of magneticpolarity and chemicalanalysisin
establishingthe stratigraphy,tectonicevolution,and petrogenesisof
the Columbia River Basalts,Mere. Geol. Soc. India, 3, 362-376, 1981.
Katao, H., H. Morinaga, M. Hyodo, H. Inokuchi, J. Matsuca, and K.
Yaskawa,Geomagneticpaleosecularvariationand K-Ar agesin Hiva
Oa island,Marquesas,FrenchPolynesia,J. Geomagn.Geoelectr.,40,
703-714, 1988.
Kruse, S. E., Magneticlineationson the flanks of the Marquesasswell:
Implicationsfor the age of the seafloor,Geophys.Res.Lett., 15, 573576, 1988.
Larson, R. L., Latest pulse of Earth: Evidence for a mid Cretaceous
superplume,Geology,19, 547-550, 1991.
Laul, J. C., Neutron activationanalysisof geologicalmaterials,Atomic
EnergyReview,17, 603-695, 1979.
Liotard, J. M., H. G. Barsczus,C. Dupuy, and J. Dostal, Geochemistry
and origin of basaltic lavas from MarquesasArchipelago, French
Polynesia,Contrib.Mineral. Petrol.,92,260-268, 1986.
McDougall, I., and R. A. Duncan,Linear volcanicchains- Recording
plate motions?Tectonophysics,
63, 275-295, 1980.
McNutt, M. K., and K. M. Fischer, The South Pacific superswell,in
Seamounts,Islands, and Atolls, Geophys.Monogr. Ser., vol. 43, edited
by B. H. Keatinget al., pp. 25-34, AGU, Washington,D.C., 1987.
McNutt, M. K., K. M. Fischer,S. Kruse, and J. Natland, The origin of the
Marquesasfracturezone ridge and its implicationfor the natureof hot
spots,Earth Planet. Sci.Lett., 91,381-393, 1989.
McNutt, M., D. Caress, J. Mutter, and R. Detrick, Thicknessof the elastic
plate beneaththe MarquesasIslands:Why is it so much greaterin
value than elsewhere in French Polynesia? (abstract), Eos Trans.
AGU, 72, 437, 1991.
McNutt, M. K., and A. V. Judge,The superswelland mantle dynamics
17,665
enrichment in the Cook-Austral-Samoa islands, southwest Pacific,
Earth Planet. Sci. Lett., 79, 270-280, 1986.
Pollack, H. N., I. G. Gass, R. S. Thorpe, and D. S. Chapman, On the
vulnerabilityof lithosphericplatesto mid-platevolcanism: Reply to
commentsby P. R. Vogt, J. Geophys.Res.,86, 961-966, 1981.
Richards, M. A., and R. W. Griffiths, Deflection of plumes by mantle
shearflow: Experimentalresultsand a simpletheory,Geophys.J., 94,
367-376, 1988.
Richards,M. A., and R. W. Griffiths, Thermal entrainmentby deflected
mantleplumes,Nature, 342, 900-902, 1989.
Schlanger,S. O., M. O. Garcia, B. H. Keating, J. J. Naughton,W. W.
Sager,J. A. Haggerty,J. A. Philpotts,and R. A. Duncan,Geologyand
geochronologyof the Line Islands, J. GeophysRes., 89, 11,26111,272, 1984.
Sleep, N.H., Hotspotsand mantle plumes: Some phenomenology,J.
Geophys.Res., 95, 6715-6736, 1990.
Smith, W. H. F., H. Staudigel, A. B. Watts, and M. S. Pringle, The
Magellan seamounts,Early Cretaceousrecord of the South Pacific
isotopic and thermal anomaly, J. Geophys.Res., 94, 10,501-10,523,
1989.
Staudigel,H., K.-H. Park, M. Pringle, J. L. Rubenstone,W. H. F. Smith,
and A. Zindler, The longevityof the SouthPacificisotopicand thermal
anomaly,Earth Planet. Sci. Lett., 102, 24-44, 1991.
Vidal, P., C. Chauvel, and R. Brousse, Large mantle heterogeneity
beneathFrenchPolynesia,Nature, 307, 536-538, 1984.
Vidal, P., C. Dupuy, H. G. Barsczus,and C. Chauvel.,H6t6rog6neti6sdu
maneauet origine des basaltesdes Marquises(Polyn6sie),Bull. Soc.
Geol. Fr., 8, 633-642, 1987.
White, W. M., Sourcesof oceanicbasalts: Radiogenicisotopeevidence,
Geology,13, 115-118, 1985.
White, W. M., and A. W. Hofmann, Sr and Nd isotopegeochemistryof
oceanic basalts and mantle evolution, Nature, 296, 821-825, 1982.
White, W. M., A. W. Hofmann, and H. Puchelt,Isotopegeochemistryof
Pacific mid-ocean ridge basalt, J. Geophys.Res., 92, 4881-4893,
1987.
White, W. M., M. M. Cheatham, and R. A. Duncan, Isotope
geochemistryof Leg 115 basaltsand inferenceson the history of the
Reunion mantle plume, Proc. Ocean Drill. Program Sci. Results,115,
53-61, 1990.
Woodhead,J. D., Temporal geochemicalevolutionin oceanicintra-plate
volcanics: A case study from the Marquesas(French Polynesia)and
comparisonwith other hotspots,Contrib. Mineral. Petrol., 111,458467, 1992.
Woodhead, J. D., and M. T. McCulloch, Ancient seafloor signals in
Pitcairn Island lavas and evidence for large amplitude, small lengthscale mantle heterogeneities,Earth Planet. Sci. Lett., 94, 257-273,
1989.
Wright, E., and W. M. White, The origin of Samoa: New evidencefrom
Sr, Nd, andPb isotopes,Earth Planet.Sci.Lett., 81, 151-162, 1986.
Zindler, A., and S. Hart, Chemical geodynamics,Annu. Rev. Earth
Planet. Sci., 14, 493-571, 1986.
Zindler, A., H. Staudigel, and R. Batiza, Isotope and trace element
geochemistryof young Pacific seamounts:Implicationsfor the scale
of upper mantle heterogeneity,Earth Planet. Sci. Lett., 70, 175-195,
1984.
beneath the South Pacific, Science, 248, 969-975, 1990.
Nakamura, N., Determination of REE, Ba, Mg, Na and K in
carbonaceousand ordinary chondrites,Geochim. Cosmochim.Acta,
38, 757-775, 1974.
Nakamura, Y., and M. Tatsumoto,Pb, Nd, and Sr isotopicevidencefor a
multi-component source for rocks of Cook-Austral Islands and
heterogeneitiesof mantle plumes, Geochim. Cosmochim.Acta, 52,
D. L. Desonie,GeochemistryBuilding, Lamont-DohertyEarth
Observatory,Palisades,NY 10964.
R. A. Duncan,Collegeof Oceanicand AtmosphericSciences,
OregonStateUniversity,Corvallis,OR 97331.
J. H. Natland, RosenstielSchoolof Marine and Atmospheric
Sciences,Universityof Miami, Miami, FL 33149.
2909-2924, 1988.
Nishimura, C. E., and D. W. Forsyth, Anomalous Love-wave phase
velocitiesin the Pacific: Sequentialpure-pathand sphericalharmonic
inversion,Geophys.J. R. Astr. Soc.,81,389-407, 1985.
Palacz, Z. A., and A.D. Saunders,Coupled trace element and isotope
(Received January10, 1992;
revisedMay 27, 1993;
acceptedJune9, 1993.)