Peridotite xenoliths from the Polynesian Austral and Samoa ...€¦ · Peridotite xenoliths from...

Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta xxx (2016) xxx–xxx

Peridotite xenoliths from the Polynesian Australand Samoa hotspots: Implications for the destruction of

ancient 187Os and 142Nd isotopic domains and the preservationof Hadean 129Xe in the modern convecting mantle

M.G. Jackson a,⇑, S.B. Shirey b, E.H. Hauri b, M.D. Kurz c, H. Rizo d

aDepartment of Earth Science, University of California, Santa Barbara, USAbDepartment of Terrestrial Magnetism, Carnegie Institution of Washington, Washington, DC 20015, USA

cDepartment of Marine Geology and Geophysics, Woods Hole Oceanographic Institution, 266 Woods Hole Road, Woods Hole,

MA 02543, USAdGeotop, Departement des sciences de la Terre et de l’atmosphere, UQAM, Montreal, Canada

Abstract

The Re–Os systematics in 13 peridotite xenoliths hosted in young (<0.39 myr) rejuvenated lavas from the Samoan island ofSavai’i and 8 peridotite xenoliths from 6 to 10 myr old lavas from the Austral island of Tubuai have been examined to eval-uate the history of the oceanic mantle in this region. Modal mineralogy, trace element compositions and 187Os/188Os ratiossuggest that these peridotites are not cognate or residual to mantle plumes but rather samples of Pacific oceanic lithospherecreated at the ridge. Savai’i and Tubuai islands lie along a flow line in the Pacific plate, and provide two snapshots (separatedby over 40 Ma in time) of Pacific mantle that originated in the same region of the East Pacific rise. Tubuai xenoliths exhibit187Os/188Os from 0.1163 to 0.1304, and Savai’i (Samoa) xenoliths span a smaller range from 0.1173 to 0.1284. The 187Os/188Osratios measured in Tubuai xenoliths are lower than (and show no overlap with) basalts from Tubuai. The 187Os/188Os of theSavai’i xenoliths overlap the isotopic compositions of lavas from the island of Savai’i, but also extend to lower 187Os/188Osthan the lavas. 3He/4He measurements of a subset of the xenoliths range from 2.5 to 6.4 Ra for Tubuai and 10.8 to 12.4 Ra forSavai’i.

Like abyssal peridotites and xenoliths from oceanic hotspots that sample the convecting mantle, Os isotopes from theSavai’i and Tubuai xenolith suites are relatively unradiogenic, but do not preserve a record of depleted early-formed (Hadeanand Archean) mantle domains expected from earlier cycles of ridge-related depletion, continent extraction, or subcontinentallithospheric mantle erosion. The lack of preservation of early-formed, geochemically-depleted Os-isotopic and 142Nd/144Nddomains in the modern convecting mantle contrasts with the preservation of early-formed (early-Hadean) 129Xe/130Xe iso-topic heterogeneities in the convecting mantle. This can be explained if the initial isotopic signatures in Re–Os andSm–Nd systems are erased by recycling because the parent and daughter elements are retained in subducting slabs and moreefficiently returned to the mantle during subduction than Xe. In this way, early-formed Os and Nd-isotopic heterogeneitiescould have been overprinted with, and diluted by, younger isotopic signatures. In contrast, preservation of early-formedheterogeneities in the modern convecting mantle is possible for other elements, such as Xe, that are not as efficiently recycledback into the mantle, owing to greater fluid mobility that concentrates such elements in the near-surface. Differing recyclingefficiencies for Os, Nd and Xe lead to wide differences in the preservation of Hadean isotopic signatures of these elements inthe modern convecting mantle. In general, incompatible elements that are fluid mobile (e.g., Xe) concentrate in surface

http://dx.doi.org/10.1016/j.gca.2016.02.011

0016-7037/� 2016 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: [email protected] (M.G. Jackson).

Please cite this article in press as: Jackson M. G., et al. Peridotite xenoliths from the Polynesian Austral and Samoa hotspots: Implica-tions for the destruction of ancient 187Os and 142Nd isotopic domains and the preservation of Hadean 129Xe in the modern convectingmantle. Geochim. Cosmochim. Acta (2016), http://dx.doi.org/10.1016/j.gca.2016.02.011


mailto:[email protected]



2 M.G. Jackson et al. /Geochimica et Cosmochimica Acta xxx (2016) xxx–xxx

reservoirs and are more likely to preserve Hadean geochemical signatures in the convecting mantle than compatible elements(e.g., Os) and fluid immobile incompatible elements (e.g., Nd).� 2016 Elsevier Ltd. All rights reserved.

1. INTRODUCTION

Lavas erupted at oceanic hotspots can host peridotitexenoliths that originated in the upper mantle and weretransported to the surface in upwelling magmas. Such peri-dotite mantle xenoliths thus offer a means to investigate thecomposition of the oceanic mantle beneath hotspot volca-noes—an approach pioneered in the 1970s and 1980s inclassic papers on the rare earth elements (REE) and othertrace elements by Fred Frey and his coworkers (Reid andFrey, 1971; Frey and Prinz; 1978; Frey, 1980). Whereasthe REE in mantle peridotites are trace element tools thatare sensitive to melt depletion and melt/metasomaticenrichment, the advent of the Re–Os system has providedanother tool that is especially sensitive to prior melt deple-tion of the mantle. The Re–Os system can be used to iden-tify and place time constraints on a wide spectrum ofancient mantle domains in the oceanic mantle that maybe related to melt extraction at the ridge, the aggregatedeffects of continent extraction, or subcontinental litho-spheric mantle erosion that may be preserved in the mantle.

187Re undergoes b-decay to 187Os (t1/2 = 41.6 Ga; Shenet al., 1996; Smoliar et al., 1996), and fractionation of Refrom Os during mantle melting is recorded by time-integrated variability in 187Os/188Os ratios. The Re–Os sys-tem is unlike other radiogenic isotopic systems in that Re isincompatible in silicate minerals and in the absence of resid-ual sulfide is extracted from the mantle during melting,while Os is a chalcophile element and is highly compatibleand retained in the mantle residue. The Re–Os isotopic sys-tem is characterized by extreme parent daughter fractiona-tions: mantle melts have Re/Os ratios that are orders ofmagnitude higher than their mantle sources, which evolvehighly radiogenic 187Os/188Os over time, while depletedmantle residues remaining after melt extraction have verylow Re/Os that preserve low 187Os/188Os over time. There-fore, melt-depleted, refractory peridotites are ideally suitedfor preserving the time-integrated history of mantle deple-tion by melt extraction, and low 187Os/188Os in peridotitescan be used to infer a minimum age for the mantle deple-tion events (e.g., Walker et al., 1989). Depleted mantledomains in oceanic environments that host unusually low187Os/188Os have been identified in xenoliths associatedwith oceanic hotspots (e.g., Bizimis et al., 2007), largeigneous provinces (Hassler and Shimizu, 1998; Ishikawaet al., 2011) and abyssal peridotites (e.g., Harvey et al.,2006; Liu et al., 2008; Lassiter et al., 2014). The depletedmantle domains identified in the oceanic mantle have con-troversial origins, and are suggested to be pieces of subcon-tinental lithospheric mantle (e.g., Hassler and Shimizu,1998), ancient subducted oceanic mantle lithosphere thatis embedded in upwelling mantle plumes (e.g., Bizimiset al., 2007), or ancient domains of depleted, refractory

Please cite this article in press as: Jackson M. G., et al. Peridotite xenotions for the destruction of ancient 187Os and 142Nd isotopic domainsmantle. Geochim. Cosmochim. Acta (2016), http://dx.doi.org/10.1016/j.gc

regions preserved in the convecting mantle (e.g., Liuet al., 2008; Lassiter et al., 2014).

Lavas erupted at islands related to two Polynesianhotspot localities in the south Pacific—Savai’i island atthe Samoan hotspot and Tubuai island at the Macdonald(or Austral) hotspot—host peridotite xenoliths that havebeen the focus of numerous geochemical studies(Wright, 1987; Hauri, 1992; Poreda and Farley, 1992;Hauri et al., 1993; Hauri and Hart, 1994; Farley et al.,1994; Farley, 1995; Burnard et al., 1998). Hauri et al.(1993) characterized two metasomatically-enriched peri-dotite xenoliths each from the islands of Savai’i andTubuai (Fig. 1). Clinopyroxenes separated from thesefour xenoliths exhibit extreme REE (particularly lightREE) enrichment, but concomitant enrichment in thehigh field strength elements (HFSE; Ti, Zr, Hf, Nb)was not observed (Hauri et al., 1993). Therefore, theREE enrichment of the clinopyroxenes was interpretedto result from metasomatism of the xenoliths by carbon-atitic melts, which impart characteristic REE enrichmentand HFSE depletion (e.g., Hauri et al., 1993). The originof the carbonatitic melt is thought to be related tohighly-enriched, small-degree melts of carbonated peri-dotite in the mantle plumes that generated both hotspots(Hauri, 1992; Hauri et al., 1993). Supporting this hypoth-esis, the Sr, Nd and Pb isotopic compositions of clinopy-roxenes isolated from the metasomatized xenoliths werefound to reflect EM2 and HIMU endmember isotopiccompositions associated with lavas erupted at the twohotspots that generated the islands of Savai’i (Samoa hot-spot) and Tubuai (Austral hotspot), respectively. How-ever, the ultra-enriched xenoliths analyzed by Hauriet al. (1993) are rare among the xenolith suites fromSavai’i and Tubuai, and most xenoliths from these local-ities do not exhibit evidence for significant REE enrich-ment (Hauri, 1992; Hauri and Hart, 1994) in keepingwith their origin as oceanic lithosphere created at theridge. To better constrain the geochemical variability ofthe Savai’i and Tubuai xenolith suite and examine theisotopic heterogeneity of the oceanic lithospheric mantle,we have characterized the Os and He isotopic composi-tions of xenoliths from the same localities that exhibit arange of enriched and depleted REE patterns (Hauri,1992; Hauri et al., 1993; Hauri and Hart, 1994).

Re–Os isotopic systematics reveal depleted mantledomains preserved in Savai’i and Tubuai xenoliths, whichhave relatively low 187Os/188Os (down to 0.1163 and0.1173, respectively). Neither the new Os-isotopic datasetpresented here from Savai’i and Tubuai xenoliths, nor theglobal oceanic hotspot xenolith and abyssal peridotite data-set, appear to preserve a record for mantle depletion in theHadean and Archean (e.g., Harvey et al., 2006; Lassiteret al., 2014) because they do not show the low Os isotopic

liths from the Polynesian Austral and Samoa hotspots: Implica-and the preservation of Hadean 129Xe in the modern convectinga.2016.02.011


Savai’i

Macdonald Trend

Samoa Trend

Tubuai

Ontong JavaPlateau

Tong

a tr

ench

25 S o

170 Wo 160 Wo 150 Wo

Austral Is.10

20

30

40

50

0

20 S o

15 S o

Hawai’i

New Zealand

Fig. 1. Map of the south Pacific, showing the locations of Tubuai (Austral hotspot) and Savai’i (Samoan hotspot) islands. Tubuai lies alongthe Macdonald hotspot trace (Chauvel et al., 1997), and Samoa lies along the Samoan hotspot trace. Pacific plate reconstructions (reproducedfrom Jackson et al., 2010) are based on the model reported in Wessel and Kroenke (2008); the Macdonald hotspot trend is anchored toMacdonald seamount, and numbers on the Macdonald hotspot reconstruction are ages in Ma. The hotspot track reconstruction shows thatthe portion of the Pacific lithosphere now occupied by the Samoan islands may have been located over the Macdonald hotspot at 40–50 Ma.

M.G. Jackson et al. /Geochimica et Cosmochimica Acta xxx (2016) xxx–xxx 3

compositions (187Os/188Os < 0.11) that would represent theEarth’s primitive and depleted mantle domains in thesetimes (Fig. 2). This is unexpected because there is evidencefor the existence of early-generated depleted mantledomains from ancient continental crust (e.g., Harrisonet al., 2005, 2008; Blichert-Toft and Albarede, 2008;Kemp et al., 2010; Guitreau et al., 2012; Shirey andHanson, 1986; Stein and Hofmann, 1994; Blichert-Toftet al., 1999; Moorbath et al., 1997; Vervoort and Blichert-Toft, 1999; Caro, 2011; O’Neil et al., 2011; Khanna et al.,2014).

In stark contrast to the Os-isotopic record in oceanicperidotites, 129Xe/130Xe isotopic heterogeneities (generatedby the decay of the short-lived nuclide 129I to 129Xe[t1/2 = 15.7 Ma]) identified in mid-ocean ridge basalts andoceanic hotspots indicate that the convecting mantle pre-serves heterogeneities that were generated in the mantleprior to 4.45 Ga (Staudacher and Allegre, 1982; Allegreet al., 1983; Mukhopadhyay, 2012; Tucker et al., 2012;Parai et al., 2012; Peto et al., 2013). In light of the preserva-tion of early-formed Xe-isotopic heterogeneities in the con-vecting mantle, the lack of preservation of earlydifferentiation in Os-isotopes, and other isotopic systemsthat record early differentiation (like 146Sm–142Nd,t1/2 = 106 Ma), is perplexing. We present a conceptualmodel that shows how different recycling efficiencies ofXe, Os and Nd may explain the preservation of Hadean129Xe/130Xe, but not 187Os/188Or or 142Nd/144Nd, in themodern convecting mantle.

2. METHODS

Twenty-one xenoliths are geochemically-characterizedin this study, 13 from Savai’i (Samoan island chain) and8 from Tubuai (Austral Island chain). The Savai’i sampleswere erupted in a tuffaceous volcanic cone, and the Tubuaisamples were recovered from large basaltic boulders(Hauri, 1992). The xenoliths from both localities are hostedin relatively young lavas: Savai’i subaerial lavas are young


(<0.42 Ma; Workman et al., 2004; McDougall, 2010;Konter and Jackson, 2012) and Tubuai subaerial lavaserupted at 5.7–10 Ma (Bellon et al., 1980). With the excep-tion of one xenolith (a pyroxenite from Savai’i), the xeno-liths examined in this study are peridotites collected byHauri during the 1990 field season. The peridotites gener-ally have low modal clinopyroxene (<5% even in REE-enriched examples; Hauri, 1992; Hauri and Hart, 1994).However, several of the Tubuai xenoliths have highermodal clinopyroxene (Hauri, 1992; Hauri et al., 1993).The single pyroxenite xenolith was collected on Savai’i byJackson during the 2005 field season at the same localityas the Savai’i peridotite xenoliths. Savai’i and Tubuai peri-dotite xenolith sample descriptions and sample locationsare described in Hauri and Hart (1993, 1994) and Hauriet al. (1993).

The xenoliths were crushed in plastic bags and the fresh-est representative rock chips were selected under a binocu-lar microscope. The rock chips were sonicated in milli-Qwater and then powdered by hand in an agate mortar thatwas cleaned by grinding with silica sand followed by rinsingin Milli-Q water. Owing to the small size of the xenoliths,only relatively small quantities of material were availablefor analysis and chips were selected sparingly for makingpowder. Up to 5 g of powder was generated from eachxenolith in the manner described, but additional xenolithpowder was required for duplicate and triplicate Os-isotopic analyses. Therefore, in eight cases, additionalaliquots of powder (labeled ‘‘batch 2” in Table 1) were pre-pared in the same manner from new chips removed fromdifferent portions of the same xenoliths from which the firstbatch of chips (‘‘batch 1” in Table 1) was isolated.

Os-isotopic data for 21 different xenoliths from Savai’iand Tubuai are reported in Table 1 together with previouslypublished data on four xenoliths from these two islands. Reand Os methods are described in Jackson and Shirey (2011).Os extraction followed a Carius tube method (Shirey andWalker, 1995) where a mixed 185Re–190Os spike was addedto the sample with inverse aqua regia. This was followed by



Fig. 2. Relationship between 187Os/188Os and Os concentrationsfor Savai’i and Tubuai xenoliths. Basalts from Tubuai (Hauri andHart, 1993) and Savai’i (Hauri and Hart, 1993; Workman et al.,2004) are plotted for comparison. Additionally, whole rock abyssalperidotites (Snow and Reisberg, 1995; Brandon et al., 2000;Standish et al., 2002; Alard et al., 2005; Becker et al., 2006;Harvey et al., 2006; Liu et al., 2008; Sichel et al., 2008; Lassiteret al., 2014) and whole rock pyroxenite (from Hawaii; Sen et al.,2011) and peridotite (from OJP, Hawaii, Kerguelen, Cape Verde,Canary Islands, and Fernando de Noronha; Hassler and Shimizu,1998; Widom et al., 1999; Meisel et al., 2001; Brugmann et al.,2008; Sen et al., 2003; Becker et al., 2006; Bizimis et al., 2007;Ishikawa et al., 2011) xenoliths from oceanic hotspot localities arealso shown. PUM is from Meisel et al. (2001). Each xenolith fromSavai’i and Samoa is represented by a single datapoint; in caseswhere replicate measurements have been made, the average Osconcentrations and the weighted (by concentration) average Os-isotopic compositions are shown. The Savai’i pyroxenite (SAV-05-25), and the highly metasomatized, enriched xenoliths examined byHauri et al. (1993)—Savai’i (SAV-1-28) and Tubuai (TBA-1-9)—are highlighted separately. TBA-5-4 is a relatively altered xenolith,but plots together with the other Tubuai xenoliths. Rheniumdepletion (TRD) model ages are provided and assume a present-dayPUM (primitive upper mantle) value of 0.1296 (Meisel et al., 2001)and assume that the Re/Os of the samples is 0 following meltdepletion of PUM at the age indicated.


solvent extraction of Os (Cohen and Waters, 1996) withcarbon tetrachloride. A microdistillation method was used


to purify Os (Roy-Barman and Allegre, 1995). Columnexchange chromatography was used to purify Re (Walker,1988; Morgan and Walker, 1989). Os-isotopic compositionswere measured by negative thermal ionization mass spec-trometry (N-TIMS) on the Thermofinnigan Triton at theDepartment of Terrestrial Magnetism (DTM). Osmiumwas loaded on Pt ribbon filaments and measured as OsO3

�.Samples with sufficient Os were measured by static faradaymulticollection, and samples with low beam intensity wereanalyzed by peak hopping using a single collector sec-ondary electron multiplier (SEM). Following oxide correc-tion to the metal composition, all runs were corrected formass bias using the exponential law to a 192Os/188Os ratioof 3.082614 (Creaser et al., 1991). Rhenium concentrationswere measured by isotope dilution and analyzed on eitherthe P54 or the NU plasma MC-ICPMS (multicollectorinductively-coupled plasma mass spectrometer) at DTM.Full chemistry procedural blanks for Re and Os were<6 pg/g and <2.5 pg/g, respectively. The fraction of thetotal Os and Re measured that is blank is also reported inTable 1. Repeated measurements of the Johnson Matthey(J–M) Os standard during the course of this study gave a187Os/188Os value of 0.17396 ± 0.00040 (2r, n = 9) for allSEM runs and 0.174012 ± 0.000039 (n = 6) for Faradayruns. During each analytical session on the Triton, runsof xenolith 187Os/188Os ratios were adjusted for the offsetbetween the measured and preferred (0.17399) J–M Os val-ues, and these ratios are reported in Table 1.

Following light crushing in an agate mortar, the freshestolivine and orthopyroxene grains were removed from a sub-set of the xenolith samples for helium isotopic analysis. Forone sample (SA-3-3), both olivine and orthopyroxene wereseparated for isotopic analysis. For SAV-1-41, three differ-ent olivine separates were made: olivine was separated bycolor (light green versus dark green) and by inclusion con-tent. Helium isotopic measurements were carried out atWoods Hole Oceanographic Institution following methodsdescribed in Kurz et al. (2004). Helium data were obtainedby crushing olivine and orthopyroxene in vacuo, and thenew data are presented in Table 2.

The bulk of the incompatible trace element budget of theperidotite xenoliths is hosted in clinopyroxene. Clinopyrox-ene trace element concentrations for all of the peridotitesamples reported here were measured by ion probe at theWoods Hole Oceanographic Institution and the data arereported elsewhere (Hauri, 1992; Hauri et al., 1993; Hauriand Hart, 1994). The data for the Savai’i orthopyroxeniteare reported here for the first time, and were obtained inthe same laboratory using the same methods as outlinedin Hauri et al. (1993). The relevant trace element data aresummarized in Table 1.

3. DATA AND OBSERVATIONS

3.1. Re and Os concentrations and 187Os/188Os compositions

Replicate Os-isotopic analyses were made on 11 wholerock peridotites (six triplicate and five duplicate analyses)(Table 1). Each replicate was performed on a new aliquotof powder (either from the same batch of powder, or on a



Table 1Os-isotopic compositions and Re–Os concentrations of xenoliths from Savai’i (Samoa) and Tubuai (Australs) Islands.a

Sample ID Powderbatch

Samplemass(g)

187Os/188Os In-runprecisionat 2r

Osconc.(ppb)

Os blank(% of totalOs analyzed)

Reconc.(ppb)

Re blank(% of totalRe analyzed)

TRD

(Ga)c(Zr/Nd)cpx

d (La/Yb)cpx Lacpx Ybcpx

Sava’i Island (Samoa)

SA-3-9 batch 1 2.158 0.12295 0.00003 0.69 0.17 0.76 80.0 1.97 0.26 0.13SA-3-9 rep1 batch 1 0.782 0.12118 0.00004 0.62 0.10 1.00SA-3-9 rep2 batch 1 0.658 0.12196 0.00004 0.77 0.32 0.89

SAV-1-30 batch 1 1.672 0.11928 0.00001 0.86 0.14 1.41 3.3 34.4 7.2 0.21SAV-1-30 rep1 batch 2 0.980 0.12023 0.00006 0.37 0.18 0.018 13.2 1.28SAV-1-30 rep2 batch 2 0.876 0.11999 0.00005 0.42 0.59 0.020 7.4 1.31

SA-3-3 batch 1 1.906 0.12709 0.00002 1.31 0.09 0.35 0.01 7.5 9.0 1.2SA-3-3 rep1 batch 1 0.710 0.12724 0.00010 0.43 0.15 0.005 36.7 0.32


SAV-1-28 batch 1 1.775 0.12537 0.00005 0.25 0.48 0.58 0.62 26.0 39.3 1.51SAV-1-28 rep1 batch 2 0.667 0.12539 0.00003 0.76 0.05 0.58

SA-3-11 batch 1 0.740 0.12663 0.00004 0.98 0.10 0.002 77.5 0.41 62.8 14.4 0.72 0.05


SAV-1-13 batch 1 2.285 0.12542 0.00002 1.27 0.09 0.57 3.0 11.7 7.0 0.60SAV-1-13 rep batch 1 0.541 0.12557 0.00003 0.98 0.10 0.035 14.5 0.55

SAV-1-46 batch 1 1.894 0.11935 0.00005 0.31 0.39 1.40 0.65 51.3 23.0 0.45SAV-1-46 rep batch 2 0.772 0.11936 0.00006 0.12 0.31 1.40

SAV-1-47 batch 1 1.674 0.12649 0.00008 0.053 2.20 0.43 7.6 9.3 18.5 2.0SAV-1-47 rep batch 2 0.542 0.12733 0.00005 0.043 0.86 0.31

SAV-05-29 batch 1 0.349 0.12628 0.00008 0.036 1.02 0.46 15.3 0.47 0.07 0.15

SAV-1-42 batch 1 0.460 0.11976 0.00002 5.35 0.01 1.34 2.0 59.3 10.4 0.18SAV-1-42 rep1 batch 1 2.386 0.11996 0.00002 1.31 0.09 1.31SAV-1-42 rep2 batch 2 0.953 0.12055 0.00008 0.85 0.29 0.002 47.1 1.23

SAV-3-2 batch 1 2.057 0.12479 0.00009 0.025 4.50 0.66 66.7 0.56 0.19 0.34

SAV-1-1b 0.1284 0.17 1.2 185 102 0.55

SAV-1-27b 0.1220 1.04 94.4 0.93(continued on next page)

M.G

.Jack

sonet

al./Geochim

icaet

Cosm

ochim

icaActa

xxx(2016)

xxx–xxx5

Please

citethisarticle

inpress

as:Jack

sonM.G.,et

al.Perid

otite

xenolith

sfro

mthePolyn

esianAustral

andSam

oahotsp

ots:

Implica-

tionsforthedestru

ctionofan

cient187O

san

d142N

diso

topic

domain

san

dthepreservatio

nofHad

ean129X

ein

themodern

convectin

gman

tle.Geochim

.Cosm

ochim

.Acta

(2016),http://dx.doi.org/10.1016/j.gca.2016.02.011


Table 1 (continued)

Sample ID Powderbatch

Samplemass(g)

187Os/188Os In-runprecisionat 2r

Osconc.(ppb)

Os blank(% of totalOs analyzed)

Reconc.(ppb)

Re blank(% of totalRe analyzed)

TRD

(Ga)c(Zr/Nd)cpx

d (La/Yb)cpx Lacpx Ybcpx

Tubuai Island (Cook-Australs)

TBA-4-4 batch 1 0.480 0.12800 0.00003 1.62 0.06 0.047 11.1 0.22 2.18 18.0 12.6 0.70

TBA-5-1 batch 1 0.472 0.12362 0.00002 5.08 0.02 0.354 1.6 0.82 116 0.01 0.01 0.46

TBA-5-2 batch 1 0.605 0.12713 0.00003 4.17 0.02 0.088 6.3 0.34 16.2 0.34 0.35 1.0

TBA-5-4 batch 1 0.495 0.12639 0.00005 5.31 0.02 0.030 16.0 0.44 0.97 21.1 141 6.7

TBA-1-9 batch 1 0.700 0.12440 0.00006 1.44 0.03 0.71 0.48 10.4 39.1 3.8

TBA-4-5 batch 1 0.486 0.11632 0.00004 0.85 0.04 1.80 1.0 12.5 7.6 0.61TBA-4-5 rep 1 batch 2 1.038 0.11861 0.00003 2.80 0.02 0.147 1.8 1.50TBA-4-5 rep 2 batch 2 0.639 0.11844 0.00006 2.53 0.10 0.149 1.1 1.52

TBA-4-8 batch 1 0.514 0.12871 0.00003 4.19 0.02 0.177 1.5 0.12 19.0 0.16 0.09 0.55

TBA-2-1 batch 1 0.696 0.12662 0.00005 3.01 0.02 0.070 3.7 0.41 2.5 2.6 0.62 0.24

TBA-4-11b 0.1285 0.15 1.5 4.6 61.0 13.3

TBA-1-7b 0.1304 �0.11 44.9 0.02 0.01 0.64

a Replicates analyses were made on either the original batch of powder (called ‘‘batch 1”), or were made on a different batch of powder (called ‘‘batch 2”) prepared from a different part of thesame xenolith. All osmium standard and unknown analyses with elevated 185ReO3/

188OsO3 (i.e., >0.0005) were discarded. Re and Os were measured on the same aliquot of powder.b Os-isotopic analyses for two Savai’i xenoliths and two Tubuai xenoliths shown in this table are reported in Hauri et al. (1993) and Hauri (1992).c TRD ages were calculated using the PUM 187Os/188Os value (0.1296) from Becker et al. (2006).d Trace element abundances in cpx are from Hauri (1992), Hauri et al. (1993) and Hauri and Hart (1994). Trace element data on SAV-05-29 are reported here for the first time. Trace element

concentrations that are below the detection limit are not reported. Trace element concentrations measured on more than one (clinopyroxene) cpx grain from a single xenolith are averaged.

6M.G

.Jack

sonet

al./Geochim

icaet

Cosm

ochim

icaActa

xxx(2016)

xxx–xxx

Please

citethisarticle

inpress

as:Jack

sonM.G.,et

al.Perid

otite

xenolith

sfro

mthePolyn

esianAustral

andSam

oahotsp

ots:

Implica-

tionsforthedestru

ctionofan

cient187O

san

d142N

diso

topic

domain

san

dthepreservatio

nofHad

ean129X

ein

themodern

convectin

gman

tle.Geochim

.Cosm

ochim

.Acta

(2016),http://dx.doi.org/10.1016/j.gca.2016.02.011


Table 2Helium isotopic compositions of Samoan and Tubuai xenoliths obtained by crushing in vacuo.

Sample ID Samplemass (g)

Phase analyzed 3He/4He(R/Ra)

In-run precisionat 1r

4He ccSTP/g

SA-3-9 Savai’i xenolith 0.2876 Olivine 11.16 0.05 5.2E�08SA-3-11 Savai’i xenolith 0.2054 Olivine 11.63 0.08 3.6E�08SA-3-3 Savai’i xenolith 0.2982 Olivine 11.85 0.17 1.5E�07SA-3-3 Savai’i xenolith 0.2750 Orthopyroxene 11.24 0.16 2.3E�07SA-1-47 Savai’i xenolith 0.1700 Olivine 12.41 0.06 7.5E�07SAV-1-41 Savai’i xenolith 0.2205 Cloudy olivine 11.71 0.06 4.7E�07SAV-1-41 Savai’i xenolith 0.1573 Clear olivine 11.78 0.08 3.0E�07SAV-1-41 Savai’i xenolith 0.1600 Dark green olivine 11.71 0.05 9.3E�07SAV-1-1 Savai’i xenolith 0.1122 Olivine 11.16 0.04 3.0E�07SAV-1-28 Savai’i xenolith 0.1504 Olivine 11.39 0.05 1.3E�07SAV-05-29 Savai’i xenolith 0.2017 Orthopyroxene 11.98 0.08 3.1E�06SAV-1-18 Savai’i xenolith 0.2988 Olivine 11.85 0.17 4.6E�08SAV-1-30 Savai’i xenolith 0.2986 Olivine 11.58 0.17 1.1E�07SAV-1-46 Savai’i xenolith 0.3035 Olivine 11.70 0.17 2.0E�07SAV-1-27 Savai’i xenolith 0.3012 Olivine 11.43 0.16 3.4E�07SAV-1-13 Savai’i xenolith 0.3014 Olivine 10.81 0.16 1.9E�07SAV-1-42 Savai’i xenolith 0.2972 Olivine 11.53 0.17 6.1E�07TBA-4-5 Tubuai xenolith 0.3007 Olivine 4.87 0.09 8.1E�09TBA-4-8 Tubuai xenolith 0.1868 Olivine 6.29 0.20 1.7E�09TBA-2-1 Tubuai xenolith 0.2036 Olivine 6.43 0.10 2.4E�08TBA-5-1 Tubuai xenolith 0.2677 Olivine 5.20 0.12 3.9E�09TBA-1-9 Tubuai xenolith 0.1372 Olivine 2.45 0.18 1.4E�09


different batch of powder prepared from different portionsof the xenolith) that was processed independently throughthe various steps of wet chemistry and mass spectrometry(and analyzed on a different filament, in a different barrel,and on a different day on the mass spectrometer). The repro-ducibility of the 187Os/188Os replicate measurements is bet-ter than 0.8% for eight samples, but is 1.5% to 2.3% forthree samples (SAV-1-41, SA-3-9, TBA-4-5). This farexceeds the internal precision for all the measurements,which is better than 0.1% (2r), and suggests that there is iso-topic heterogeneity within these peridotites. Two of thethree replicate analyses that have Os-isotopic reproducibil-ity worse than 1.5% represent analyses of different aliquotsof the same batch of powder, which indicates that therecan be significant isotopic heterogeneity within a batch ofpowder. The third example of reproducibility worse than1.5% occurred in a replicate for which different batches ofpowder (sampling different portions of the same xenolith)have different Os-isotopic compositions. The isotopicheterogeneity within a single batch of powder, or betweentwo batches of powder sampling different parts of the samexenolith, may be due to the presence of trace phases, or‘‘nuggets”, that have distinct 187Os/188Os. The different nug-gets might represent different generations of Os in a xenoliththat contains appreciable Os. The observation that peri-dotite xenoliths host isotopically-heterogeneous nuggets iswell established in the literature (Shirey and Walker, 1995,1998; Reisberg and Meisel, 2002; Lorand et al., 2008). Wenote that, while intra-xenolith 187Os/188Os variability is upto 2% (sample TBA-4-5) in the Tubuai xenoliths and up to2.3% (sample SAV-1-41) in the Savai’i xenoliths, the totalinter-xenolith 187Os/188Os variability among differentTubuai (10.8%) and Savai’i (8.6%) xenoliths exceeds thevariability observed within any single xenolith.


The concentration of Os also varies within a single pow-der and between different batches of powder prepared fromdifferent portions of the xenolith. Five of the eleven xeno-liths with replicate measurements have Os concentrationsthat are reproducible within 6–29%. The remaining sixxenoliths have Os concentrations that vary by a factor of2.3–7.1, which includes comparison of analyses for differentaliquots of the same powder and different powders pre-pared from the same xenolith. This variability can beexplained by the heterogeneous distribution of Os-rich,trace phases (nuggets), both within a powder and withindifferent regions of the same xenolith. We note that thereis no relationship between variability in Os concentrationsand heterogeneity in 187Os/188Os within a xenolith: xeno-liths with highly variable Os concentrations can haveheterogeneous 187Os/188Os (TBA-4-5, SAV-1-42, SAV-1-30) or relatively homogenous 187Os/188Os (SAV-1-28,SA-3-3, SAV-1-46). Similarly, xenoliths with relativelyhomogeneous Os concentrations can have heterogeneous187Os/188Os (SAV-1-41) or relatively homogenous187Os/188Os (SAV-1-13).

The Savai’i xenoliths generally have lower Re than theTubuai xenoliths: Re concentrations vary from just below0.001 ppb to 0.035 ppb in the Savai’i xenoliths, and from0.030 ppb to 0.354 ppb in the Tubuai xenoliths. Replicatemeasurements for Re were made on four whole rock peri-dotites; like Os, each replicate was performed on a new ali-quot of powder processed independently through thevarious steps of wet chemistry and mass spectrometry. Ofthese four xenoliths, the sample with the highest Re concen-tration (TBA-4-5) is also the most reproducible (the highestand lowest Re concentrations differ by only 1.1%), as is typ-ical with the nugget effect. The xenolith with the lowest Reconcentration (SAV-1-41) has the least reproducible result.




There is no relationship between 187Os/188Os and Osconcentrations (Fig. 2) and Re concentrations (not shown)in the new xenolith dataset. The average 187Os/188Os of theTubuai xenolith suite (0.1262 ± 0.0035, 1r; replicate187Os/188Os analyses for each xenolith were averaged andincluded as a single value in the final average) is higher thanthe average 187Os/188Os of the Savai’i xenoliths (0.1237± 0.0032), but the two xenolith populations overlap at the1r level. While the Os concentrations of the Tubuai andSavai’i xenoliths overlap, the Tubuai xenoliths extend tohigher Os concentrations than the Savai’i xenoliths(Fig. 2). Two peridotite xenoliths from Samoa haveremarkably low Os concentrations: SAV-1-47 has repro-ducibly low Os (0.043 and 0.053 ppb) and SAV-3-2 hasthe lowest Os concentration in the xenolith suite(0.025 ppb). Such low Os concentrations are more typicallyfound in pyroxenites, including the Samoan pyroxenitexenolith sample SAV-05-29, which has 0.036 ppb Os.

The variability in 187Os/188Os and Os in the Tubuai andSamoa xenoliths lies approximately within the rangedefined by previously published data from abyssal peri-dotites and peridotite xenoliths from oceanic hotspots.Excluding a single Tubuai xenolith with 187Os/188Os of0.1304 (TBA-1-7) (not shown in Fig. 2 because Os concen-

Fig. 3. Trace elements ratios (Zr/Nd and La/Yb) and concentrations (element data were measured in clinopyroxene and were obtained by ion prin Fig. 2. The Savai’i pyroxenite (SAV-05-29), the altered Tubuai xenolithreported in Hauri et al. (1993)—two from Savai’i (SAV-1-28 and SAV-1-separately. Rhenium depletion (TRD) model ages are provided as dashed lof 0.1296 (Meisel et al., 2001) and assumes that the Re/Os of the sample


trations are not available for this xenolith; see Hauri, 1992),xenoliths from Savai’i and Tubuai have 187Os/188Os lowerthan the primitive upper mantle (PUM). This contrasts witha subset of abyssal peridotites and peridotite xenoliths fromKerguelen (187Os/188Os up to 0.1383) and Ontong Java Pla-teau (OJP) (187Os/188Os up to 0.1404) with relatively radio-genic 187Os/188Os ratios that exceed the PUM value. Theextent to which these enriched Os isotopic compositionsare due to melt re-enrichment is not clear, but such a pro-cess apparently does not affect the Savai’i and Tubuai xeno-liths studied here.

Peridotites with low Os concentrations (<0.1 ppb) arerelatively uncommon. Xenoliths from Kerguelen (down to0.044 ppb) and OJP (down to 0.068 ppb) can have low Osconcentrations and intermediate 187Os/188Os that approachthe compositions observed in the peridotite xenoliths withthe lowest Os from Savai’i (down to 0.025 ppb). Oceanicperidotites with 187Os/188Os as low as (or lower than) thelowest values found in Samoa and Tubuai are also rela-tively uncommon, but are found at Hawaii (187Os/188Osdown to 0.1138; Bizimis et al., 2007), OJP (down to0.1152; Ishikawa et al., 2011), and a subset of abyssal peri-dotites (down to 0.1139; Liu et al., 2008).

La and Yb) measured in clinopyroxene versus 187Os/188Os. Traceobe. Replicate Os measurements are reported in the same manner as(TBA-5-4), and the four highly metasomatized, enriched xenoliths

1) and two from Tubuai (TBA-1-9 and TBA-4-11)—are highlightedines and assume a present-day PUM (primitive upper mantle) values is 0 following melt depletion of PUM at the age indicated.




Fig. 2 shows that the 187Os/188Os of the Savai’i xenolithsoverlap the isotopic compositions of lavas from the islandof Savai’i, but also extend to lower 187Os/188Os than thelavas. The 187Os/188Os ratios measured in Tubuai xenolithsare lower than (and show no overlap with) basalts fromTubuai.

A subset of the Savai’i and Tubuai xenoliths exhibit dif-ferent degrees of enrichment by a carbonatitic melt. Hauriet al. (1993) argued that clinopyroxene from a subset ofSavai’i and Tubuai peridotite xenoliths grew in the presenceof a carbonatitic melt. The metasomatic clinopyroxeneshave anomalously high REE concentrations, particularlythe light REE’s (LREE, with La/Yb ratios of up to 185).In contrast, Ti and Zr show strong depletions relative tothe REEs. Like the LREE enrichment in clinopyroxene,Zr depletion relative to Nd (low Zr/Nd ratio; Fig. 3) inclinopyroxene suggests carbonatite metasomatism. Neitherindicator of carbonatite metasomatism, low Zr/Nd or highLa/Yb in clinopyroxene, correlates with 187Os/188Os. Forexample, the Savai’i xenolith SAV-1-1 with the strongestcarbonatite metasomatic signature (highest La/Yb and sec-ond lowest Zr/Nd) does have the most radiogenic187Os/188Os, but the Savai’i xenoliths with among the stron-gest carbonatite signatures (SAV-1-42 and SAV-1-46) haveamong the lowest 187Os/188Os in the Savai’i suite. TheTubuai xenolith with the highest 187Os/188Os (TBA-1-7)appears to be one of the least affected by carbonatite (it

Fig. 4. Helium isotopic ratios compared to 4He concentrations.Helium isotopic compositions are reported for a subset of Samoanand Tubuai xenoliths. Additionally, for comparative purposes,helium isotopic measurements for olivine and cpx phenocrysts fromall Samoan lavas (Farley et al., 1992; Workman et al., 2004;Jackson et al., 2007a,b; Jackson et al., 2010a,b; Jackson et al.,2014) and two lavas from the Samoan island of Savai’i (Workmanet al., 2004; Jackson et al., 2014) are shown. Helium isotopicmeasurements for olivine and cpx phenocrysts from Tubuai basalts(Graham et al., 1992; Parai et al., 2009; Hanyu et al., 2011b) arealso shown. Replicate Os measurements are reported in the samemanner as in Fig. 2. Replicate measurements of 3He/4He areaveraged. The Savai’i pyroxenite (SAV-05-29), and the highlymetasomatized xenoliths—two from Savai’i (SAV-1-28 and SAV-1-1) and one from Tubuai (TBA-1-9)—are highlighted separately.Cpx is clinopyroxene.


has the second lowest La/Yb and the second highest Zr/Nd), and Tubuai xenoliths that have with the strongesttrace element signatures for carbonatite metasomatismhave both low (TBA-4-5) and relatively high (TBA-1-9,

Fig. 5. Helium isotopic ratios compared to 187Os/188Os, (La/Yb)cpxand (Zr/Nd)cpx. Replicate Os and He measurements are treated asin Fig. 4. The Savai’i pyroxenite (SAV-05-29), and the highlymetasomatized xenoliths—two from Savai’i (SAV-1-28 and SAV-1-1) and one from Tubuai (TBA-1-9)—are highlighted separately.Cpx is clinopyroxene.




TBA-4-11 and TBA-5-4) 187Os/188Os. In summary, there isno simple relationship between trace element indicators ofmetasomatic enrichment and 187Os/188Os in the xenoliths,suggesting that the Os isotopic compositions have not beenmodified by metasomatism.

3.2. Helium concentrations and isotopic data

Helium isotopes were measured in a subset of the Savai’iand Tubuai xenoliths. 3He/4He in the Savai’i xenoliths var-ies from 10.8 to 12.4 times atmosphere (Ra) (Figs. 4 and 5),lower than values of 12.7 and 13.7 Ra reported for Savai’irejuvenated lavas (Workman et al., 2004; Jackson et al.,2014). The rejuvenated lava measurements are relevant herebecause only rejuvenated lavas host peridotite mantle xeno-liths in Samoa. These values are higher than the globalupper mantle 3He/4He ratio as measured in MORB(approximately 7–8 Ra; Graham, 2002; Georgen et al.,2003). The new peridotite xenolith 3He/4He measurementsfrom Savai’i are similar to previous helium isotopic mea-surements of Savai’i peridotite xenoliths by crushing, whichvaried from 11.3 to 12.0 Ra (Farley, 1995) and from 10.2 to11.9 Ra (Poreda and Farley, 1992). Poreda and Farley(1992) did report lower 3He/4He values for individual hightemperature heating steps, but if the 3He/4He values forheating and fusion of the same sample are combined, the3He/4He falls within the range of 3He/4He reported herefor the crushing experiments. We find a weak positive rela-tionship between 3He/4He and helium concentrations in theSavai’i xenoliths, where 4He concentrations vary from

Table 3Modal abundances of Savai’i and Tubuai xenoliths.

Sample ID Olivine Orthopyroxene Clinopyrox

SA-3-9a 76.9 22.1 0.5SAV-1-30a 62.1 29.6 6.1SA-3-3a 67.6 31.7 0.3SAV-1-41a 76.1 18.9 2.7SAV-1-28a 67.8 27.9 3.1SA-3-11a 80.8 18.5 0.4SAV-1-18a 56.7 38.9 2.6SAV-1-13a 70.9 24.1 3.9SAV-1-46a 76.1 18.6 3.3SAV-1-47a 75.4 20.6 2.6SAV-05-29 99 1SAV-1-42a 82.9 14.3 2.2SAV-3-2a 83.8 15.4 0.4SAV-1-1a 81.4 15 2.1SAV-1-27a 83.6 15.9 0.3TBA-4-4 68 15 11TBA-5-1 66 24 4TBA-5-2 58 24 7TBA-5-4 55 40TBA-1-9 80 10 10TBA-4-5 70 24 4TBA-4-8b NA NA NATBA-2-1 65 28 5TBA-4-11 82 10TBA-1-7 70 24 4

a Modes reported previously (Hauri and Hart, 1994).b Remaining portion of xenolith is too small to accurately estimate mo


3.6 * 10�8 to 3.1 * 10�6 cc STP/g (and the highest concen-trations are found in the Savai’i pyroxenite); the range inconcentrations spans nearly two orders of magnitude, andlikely reflects the variable (but generally high) density offluid inclusions present in the xenoliths. Three visually-distinct (cloudy, clear and dark green olivine) populationsof olivine isolated from SAV-1-41 exhibit a factor of threevariability in helium concentrations (Table 2). In Savai’ixenoliths, helium isotopic ratios exhibit no relationshipwith 187Os/188Os or with trace element indicators of carbon-atite metasomatism (Fig. 5).

Helium isotopic values in the Tubuai xenoliths are con-sistently lower, and show a wider range, than the Samoanxenoliths (Figs. 4 and 5), and are lower than the upper man-tle average 3He/4He ratio measured in MORB. 3He/4He inthe Tubuai xenoliths (2.5–6.4 Ra) are also lower than valuesobserved in Tubuai lavas: 3He/4He ratios measured inTubuai lavas range from 6.4 to 7.2 Ra (Graham et al.,1992; Parai et al., 2009; Hanyu et al., 2011b). However,the 4He concentrations in the Tubuai xenoliths are all lowerthan the Savai’i xenoliths, and range from 1.4 * 10�9 to2.4 * 10�8 cc STP/g. While there is no relationship between3He/4He and 4He concentrations in the Tubuai xenoliths,the xenolith with the lowest 4He concentration has thelowest 3He/4He, which might be explained by radiogenicingrowth of 4He. We do not have Th and U abundancesin these samples to test this hypothesis, but peridotitestypically have very low Th and U concentrations (e.g.,Recanati et al., 2012).

ene Spinel Apatite Melt Albite

0.52.20.42.31.20.31.8 trace1.1 trace21.4

<10.60.41 0.3 0.20.26 trace5 1 <13 8

5

2 <1NA NA NA NA2<1 3 42 <1

dal abundances.




Fig. 5 shows that there is no relationship between3He/4He and 187Os/188Os in the dataset. The samples withpaired 3He/4He and 187Os/188Os from Savai’i and Tubuaispan a similar range of 187Os/188Os, but the Tubuai xeno-liths all have lower 3He/4He. Additionally, there is no rela-tionship between 3He/4He and trace element indicators ofcarbonatite metasomatism in the Tubuai xenolith suite(Fig. 5). Furthermore, the modal abundances of the phasesin the Savai’i and Tubuai xenoliths (Table 3) do not exhibitcorrelations with 187Os/188Os or 3He/4He (not shown).

4. DISCUSSION

4.1. Variable Os concentrations in the Tubuai and Savai’i

xenoliths

A puzzling feature of the Savai’i xenolith suite is theobservation that two peridotite xenoliths have unusuallylow Os concentrations (Os 6 0.053 ppt). Peridotite xeno-liths hosted in oceanic hotspot lavas tend to have lowerOs concentrations than abyssal peridotites (Ishikawaet al., 2011), but the very low Os concentrations in twoSavai’i xenoliths are unusual.

Relatively low Os concentrations (down to 0.068 ppt) inOJP peridotite xenoliths were identified by Ishikawa et al.(2011). Ishikawa et al. (2011) suggested the low Os concen-trations may result from the breakdown of sulfide, which isthe primary host of Os in peridotites (Hart and Ravizza,1996; Burton et al., 1999; Alard et al., 2000). The Savai’ixenoliths are relatively fresh, so breakdown of sulfide bysurface weathering processes is unlikely. The breakdownof sulfide by devolatilization during eruption and exposureto oxygenated conditions (Handler et al., 1999) was alsoconsidered by Ishikawa et al. (2011). We consider such anexplanation unlikely in the Samoan suite because the mech-anism of sulfide breakdown should affect all the xenoliths,but only two Savai’i xenoliths exhibit unusually low Osconcentrations. One possible hypothesis is that sulfideswere exhausted by extensive melt extraction from the peri-dotites, leaving the Os highly depleted in these samples.However, Os concentrations do not exhibit clear correla-tions with indicators of melt extraction (e.g., modal abun-dance of spinel), and the xenoliths with unusually low Osconcentrations do not appear to be more refractory thanthe other Savai’i xenoliths. Another possible explanationfor the low Os concentrations is percolation of sulfur-undersaturated melts through the peridotites, which wouldmobilize sulfide and reduce the Os concentrations (Reisberget al., 2005). However, there is no relationship between Osconcentrations and indicators of metasomatic enrichment(e.g., La/Yb in xenolith clinopyroxene), which arguesagainst this hypothesis. Another possibility is that the xeno-liths’ low Os content is an expression of the nugget effect ona large scale. In other words, the heterogeneous distributionof Os-carrying phases is on a scale much larger than thesmall xenolith carried by the alkali basalt host. The vari-ability of the heterogeneity increases with lower Os concen-trations because the probability of sampling representativeabundances of the Os-carrying phases decreases. In sum-mary, while there are several possible models for the origin


of the low Os concentration in the xenoliths, there is no sin-gle mechanism that clearly explains the low Os concentra-tions identified in a subset of the Savai’i xenoliths.

4.2. Relationship between 187Os/188Os in xenoliths and whole

rock lavas from Samoa and Tubuai

Hauri et al. (1993) and Hauri (1992) report Sr, Nd andPb isotopic compositions on clinopyroxene separates fromthe most incompatible trace element enriched xenolithsfrom Savai’i and Tubuai: SAV-1-1, SAV-1-28, TBA-4-11,and TBA-1-9 (the Os-isotopic data for these xenoliths areshown in Table 1). The Sr, Nd and Pb isotopic composi-tions of the two Savai’i xenoliths (87Sr/86Sr is 0.710969–0.712838; 143Nd/144Nd is 0.512467–0.512516; 206Pb/204Pbis 18.845–18.877) are more geochemically enriched thanthe rejuvenated lavas from Savai’i that host the xenoliths(Wright and White, 1986; Hauri and Hart, 1993;Workman et al., 2004), but they overlap with shield stagelavas from the island (Jackson et al., 2007a,b). The Sr,Nd and Pb isotopic compositions of the Tubuai xenoliths(87Sr/86Sr is 0.702796–0.702755; 143Nd/144Nd is 0.512902–0.512942; 206Pb/204Pb is 20.971–21.313) overlap withbasalts from the volcano (Vidal et al., 1984; Chauvelet al., 1992; Kogiso et al., 1997; Schiano et al., 2001;Hanyu et al., 2011a, 2013). Unfortunately, Sr, Nd and Pbisotopic data were not reported for the xenoliths fromSavai’i and Tubuai that exhibit less incompatible trace ele-ment enrichment, and therefore it is unknown whetherthese xenoliths have isotopic compositions similar tobasalts from the respective islands.

In this study, we report 187Os/188Os in xenoliths thatexhibit incompatible trace element enrichment and xeno-liths that are incompatible trace element depleted. The187Os/188Os of the Tubuai xenoliths (<0.1304) is muchlower than the 187Os/188Os measured in Tubuai basalts(>0.1433) (Fig. 2). Hauri and Hart (1993) noted that the187Os/188Os of Tubuai xenoliths is similar to the rangeexpected for oceanic lithosphere, while the 187Os/188Os ofTubuai basalts is elevated and similar to basalts from otherHIMU localities (e.g., Hauri et al., 1993; Day et al., 2009;Hanyu et al., 2011a). Hauri et al. (1993) argued that, whilethe Sr, Nd and Pb isotopic composition of the two mostincompatible element enriched Tubuai xenoliths reflectsthe composition of the metasomatic fluid, the compatibleelement budget (Os) of these peridotite were not signifi-cantly changed by the metasomatic fluid, and theOs-isotopic composition of the xenoliths preserves a litho-spheric origin. The unradiogenic Os in the Tubuai xenolithsindicates that the xenoliths are not the source of the basalticvolcanism at Tubuai. Instead, while CO2-rich melts associ-ated with the plume overprinted the Sr, Nd and Pb isotopiccomposition of the peridotite xenoliths that were entrainedin the upwelling melts of from the Tubuai mantle source(Hauri et al., 1993), they did not affect the Os isotopic com-position of the xenoliths.

Unlike the case of the Tubuai xenoliths and basalts,the range of 187Os/188Os in lavas from Savai’i (>0.1233;Hauri and Hart, 1993; Workman et al., 2004) overlaps withthe 187Os/188Os xenoliths from this locality (0.1173–0.1284)




(Fig. 2). The Savai’i xenoliths are hosted in rejuvenatedlavas, and rejuvenated lavas from Savai’i have the lowest187Os/188Os (down to 0.1233) among Samoan hotspot lavas(for comparison, the lowest 187Os/188Os in Samoan shieldlavas is 0.1268; Jackson and Shirey, 2011). Indeed, Savai’irejuvenated lavas exhibit some of the lowest 187Os/188Osin the global OIB (ocean island basalt) database. However,xenoliths are abundant in rejuvenated lavas from Savai’i,and Jackson and Shirey (2011) suggested that xenocrysticolivines from disaggregated xenoliths (with low187Os/188Os) might contribute to the low 187Os/188Os inrejuvenated Savai’i lavas. While unradiogenic Os in oceanichotspots is not always a result of the incorporation of xeno-lith material with low 187Os/188Os (Schaefer et al., 2002),this mechanism has been suggested to reduce the187Os/188Os of the host lava at the Azores, Comoros andCanary hotspots (Widom and Shirey, 1996; Widom et al.,1999; Class et al., 2009; Day et al., 2010). We note that halfof the 14 peridotite xenoliths in the Savai’i dataset have187Os/188Os ratios lower than the lowest ratio identified inlavas from the Samoan hotspot. Therefore, even if thelow 187Os/188Os of the Samoan rejuvenated lavas reflectsthe mantle source of rejuvenated lavas, and is not just theresult of incorporation of disaggregated xenoliths, thereare still 7 peridotite xenoliths with 187Os/188Os lower thanSamoan lavas and are derived from a mantle domain thatis unlike that sampled by any Samoan lavas. We note thatthe other seven peridotite xenoliths with 187Os/188Os over-lapping with the range observed in Samoan lavas may alsosample an Os-isotopic domain distinct from that sampledby the lavas, because the 187Os/188Os of the Savai’i rejuve-nated lavas may have been reduced by incorporation ofxenolith material with highly unradiogenic 187Os/188Os.

4.3. Melt–rock reaction in the Tubuai and Savai’i xenoliths

Extensive carbonatite metasomatism related to theSamoa hotspot (which sources Savai’i volcanism) and theMacdonald hotspot (thought to source Tubuai volcanism;Chauvel et al., 1997) is clear in a subset of the Savai’i andTubuai xenoliths, respectively (Hauri et al., 1993). Howeverthere is no simple relationship between indicators of meta-somatism (trace element ratios measured in xenolithclinopyroxenes) and 187Os/188Os of the bulk xenolith. Ifthe Os-isotopic compositions of the xenoliths were modifiedby the metasomatic fluid, we would expect the mostmetasomatically-enriched xenoliths (with the highest La/Yb and lowest Zr/Nd) to exhibit a relationship with wholerock 187Os/188Os, but none is observed. Instead, xenolithswith the weakest metasomatic signatures exhibit both highand low 187Os/188Os (Fig. 3), suggesting that preexistingOs-isotopic heterogeneity dominates the Os-isotopic vari-ability of the Savai’i and Tubuai xenolith suites. It is likelythat the metasomatic fluid did not have sufficiently high Osconcentrations to strongly modify the Os-isotopic composi-tions of the peridotite xenoliths, a hypothesis that is sup-ported by measurements of low Os concentrations inoceanic carbonatites from the Canary islands (0.0047–0.0145 ppb; Widom et al., 1999). In summary, while theincompatible trace elements and the radiogenic isotopes


of incompatible elements (He, Nd, Sr and Pb) show exten-sive evidence that these xenoliths interacted with metaso-matic melts, the data do not support recent metasomaticmodification of the 187Os/188Os of the Savai’i and Tubuaixenoliths. Instead, we argue that the Os-isotopic composi-tions of the peridotite xenoliths reflect long-term mantleheterogeneity generated by ancient melt extraction andare not significantly modified by metasomatism (e.g.,Walker et al., 1989; Shirey and Walker, 1998).

Helium isotopic compositions do not correlate with Os-isotopic compositions. Helium is a strongly incompatibleelement in the mantle (Parman et al., 2005; Heber et al.,2007; Jackson et al., 2013). Therefore, the lack of correla-tion between 3He/4He and 187Os/188Os is not surprising,as other incompatible trace elements fail to show correla-tions with 187Os/188Os (Fig. 3). It is noteworthy that heliumisotopic compositions do not correlate with incompatibletrace element indicators of metasomatism (Fig. 5). Oceaniccarbonatites can be enriched in helium (e.g., Mata et al.,2010), and helium might be expected to impart a clear meta-somatic signature on the metasomatized peridotite in thesame manner as the other incompatible trace elements.Nonetheless, within the population of xenoliths from eachisland, there is no simple relationship between 3He/4Heand trace element signatures.

In the xenoliths from Savai’i and Tubuai, magma–man-tle reaction is reflected in the precipitation of metasomaticclinopyroxene. The extensive overprinting of the incompat-ible elements and their isotopes (He, Sr, Nd, Pb) in a peri-dotite, without correlated shifts in compatible elements andtheir isotopes (Os), can be readily understood by consider-ing how the budgets of incompatible and compatible ele-ments change in a magma that undergoes melt–rockreaction with peridotite. This process was modeled byHauri (1997), who parameterized magma–mantle interac-tion in terms of the solid/melt partition coefficient (K)and the Damkohler number (Da), which is a dimensionlessnumber that is a measure of the extent of magma–mantleexchange. The Damkohler number represents the ratio oftwo timescales, the timescale for melt–solid exchange (tE)versus the timescale of magma transport (tc).

From a simple mixing standpoint, it is relatively easyto overwhelm the compatible element budget of a magmabecause of the much larger concentrations of compatibleelements in peridotite; the converse is true for incompat-ible elements, where the magma overwhelms the peri-dotite. This simple situation becomes complicated whenthe ‘‘mixing” occurs during magma transport throughperidotite. When the Damkohler number is low (analo-gous to a high melt/rock ratio), melt–rock reaction can-not keep up with the high rate of magma ascent, andmagmas can be transported through the mantle at fulldisequilibrium while retaining even the compatible ele-ment (e.g., Os) isotopic compositions of their sources(Hauri and Kurz, 1997). When the Damkohler numberis high, magma ascent is slow enough to allow extensiveequilibration between the ascending magma and sur-rounding peridotite, to the point of complete bufferingof the melt in equilibrium with the mantle. Whether theelements in the magma approach equilibrium with the




surrounding mantle (or not) depend on the partitioncoefficient and the Damkohler number.

Hauri (1997) demonstrated that the wide variety ofhighly-fractionated REE patterns and widely-varyingREE/HFSE ratios in Samoan xenoliths were consistentwith migration of Samoan hotspot melts into the oceaniclithosphere, with the initial melt flux dominated bycarbonatite-like melts that gave way over time to a mixtureof carbonatitic and basaltic trace element compositions.The best-fit Damkohler numbers estimated from the traceelements in xenolith clinopyroxenes (Da = 1412 decreasingto 158 with time) represent values that are high enough tofractionate REE during magma–mantle reaction. However,these Damkohler numbers are all high enough (by severalorders of magnitude) that the compatible element budgetof the magma is entirely buffered by the surrounding man-tle. As a result, such a magma has essentially no capacity toalter significantly the isotopic composition of compatibleelements like osmium in the xenolith, while still retainingmuch of its original budget of highly incompatible elementsderived from its mantle source. This kind of melt–rockinteraction explains the overprinting of Sr, Nd, Pb andHe isotopes in the Savai’i and Tubuai peridotite xenolithsby their respective hotspot magmas, while largely retainingthe Os isotopic composition of the oceanic lithosphere thatwas formed at a mid-ocean ridge.

Another possible explanation for the lack of relationshipbetween 3He/4He and incompatible element indicators ofmetasomatism is that several different pulses of metasoma-tism may have overprinted the Savai’i xenolith suite(Burnard et al., 1998), possibly as the Pacific lithospherecurrently under Savai’i passed over several different hot-spots with different compositions (Jackson et al., 2010a,b;Konter and Jackson, 2012). Multicomponent mixingcaused by multiple pulses of metasomatic fluids with differ-ent 3He/4He ratios and variable trace element compositionsmay generate complicated geochemical patterns thatobscure relationships among incompatible elements. Thissimple model may also help explain the lack of relationshipsbetween 3He/4He and 187Os/188Os and trace elements in theTubuai xenoliths. It is not clear whether the non-atmospheric 129Xe/130Xe reported in a subset of Savai’ixenoliths analyzed by Poreda and Farley (1992) relates tometasomatic enrichment, as trace element and isotopic(i.e., Sr, Nd, Pb) indicators of carbonatite metasomatismwere not analyzed in this suite of xenoliths.

In summary, the unradiogenic Os isotopic signatures ofall of the Tubuai and half of the Savai’i xenoliths are lowerthan the lowest values identified in lavas with mantle-derived Os-isotopic signatures at their respective islands,and this finding echoes similar observations made at theHawaiian hotspot and the OJP. Bizimis et al. (2007) identi-fied highly unradiogenic 187Os/188Os in peridotite xenolithsfrom Hawaii (down to 0.1138) that are lower than the low-est 187Os/188Os ratios identified in lavas from the hotspot.Similarly, the low 187Os/188Os found in peridotite xenolithsfrom the OJP (down to 0.1152; Ishikawa et al., 2011) extendto lower values than identified in lavas from the Plateau(Tejada et al., 2013). The unradiogenic 187Os/188Os identi-fied in xenoliths, but not in lavas, from these Pacific hot-


spots and the OJP shows that they are not mantle residuato the lavas and thus raises a fundamental question: Whatis the origin of the low 187Os/188Os mantle domain sampledby the xenoliths?

4.4. Origin of the low 187Os/188Os signature in Tubuai and

Savai’i xenoliths

4.4.1. Subcontinental lithospheric mantle as the origin of the

low 187Os/188Os in Samoa and Tubuai xenoliths?

The relatively unradiogenic 187Os/188Os ratios reportedin Savai’i (down to 0.1173) and Tubuai (down to 0.1163)xenoliths requires formation by ancient melt extraction fol-lowed by long-term preservation, but the provenance of theancient depleted reservoir contributing to these hotspotxenoliths—SCLM or ancient depleted domains in the ocea-nic mantle—is uncertain. The subcontinental lithosphericmantle (SCLM) is known to host unradiogenic 187Os/188Os,which is the result of ancient melt extraction from the man-tle followed by long-term preservation of the melt-depletedreservoir beneath the continents (e.g., Walker et al., 1989;Carlson and Irving, 1994; Carlson et al., 2005; Pearsonand Wittig, 2008). Peridotite xenoliths from Kerguelen withrelatively low 187Os/188Os (down to 0.1182) were suggestedto sample slivers of SCLM incorporated into the IndianOcean lithosphere following rifting of Gondwanaland(Hassler and Shimizu, 1998). Indeed, continental litho-sphere is known to underlie portions of the Kerguelen pla-teau (Frey et al., 2002; Ingle et al., 2002), which makes theinterpretation of the low 187Os/188Os Os compelling at Ker-guelen and perhaps suggestive at other oceanic localities(e.g., Coltorti et al., 2010; Debaille et al., 2009). However,unlike Kerguelen, the Savai’i and Tubuai Islands formedin an oceanic setting far from continental margins and arenot located at the locus of past continental rifting (e.g.,Taylor, 2006). There is no obvious tectonic mechanism thatcan raft fragments of SCLM to regions of the Pacific platenow occupied by Savai’i and Tubuai, and it is unlikely thatfragments of SCLM reside in this region of the southPacific.

In addition to lacking a clear tectonic mechanism fortransporting SCLM fragments to Savai’i and Tubuai, theSCLM mantle is relatively rare. Bizimis et al. (2007) arguedthat the relative scarcity of SCLM mantle—only 2.5% ofthe mantle’s mass—makes it far less likely to contributeto the unradiogenic 187Os/188Os observed in peridotitexenoliths from the Hawaiian hotspot than other depletedmantle reservoirs. Depleted mantle domains in the convect-ing mantle that may host low 187Os/188Os constitute a muchlarger fraction of the mantle and are therefore more likelyto contribute to the peridotite xenoliths at Savai’i andTubuai (see Sections 4.4.2 and 4.4.3 below). These depletedreservoirs include the ambient upper mantle (as sampled byabyssal peridotites) and ancient subducted oceanic mantlelithosphere, where the latter may comprise 80% (or more)of the mass of the mantle (Salters and Stracke, 2004).How these reservoirs might contribute to the unradiogenic187Os/188Os observed in Savai’i and Tubuai xenolithsbecomes the essential consideration.




4.4.2. Low 187Os/188Os in hotspot xenoliths: Ancient

subducted oceanic lithosphere entrained in mantle plumes?

Peridotite xenoliths with exceptionally low 187Os/188Os(down to 0.1138) hosted in Hawaiian lavas were suggestedto be derived directly from ancient subducted, depletedmantle lithosphere that is embedded in the upwellingHawaiian mantle plume (Bizimis et al., 2007). The Hawaiianxenoliths were not considered to be peridotite fragmentsfrom the modern Pacific oceanic lithosphere because, atthe time of the study (Bizimis et al., 2007), abyssalperidotites were not known to host 187Os/188Os as low asidentified in the Hawaiian xenolith suite (down to 0.1138).Therefore, a different (deeper?) reservoir composed ofancient subducted oceanic lithosphere was suggested asthe reservoir supplying material with lower 187Os/188Os tothe Hawaiian xenolith suite. In this model, subducted ocea-nic mantle lithosphere, now residing in the mantle, experi-enced melt extraction and Re-depletion at an ancient mid-ocean ridge, and domains of mantle comprised of this mate-rial may preserve low 187Os/188Os. Upwelling mantleplumes sourcing oceanic hotspot volcanism may entrain asubducted reservoir comprised of ancient subducted ocea-nic mantle lithosphere which itself becomes embedded inthe plume (Bizimis et al., 2007) and attaches itself to thebase of the preexisting oceanic lithosphere (Ishikawaet al., 2011). In this model, the subset of peridotite mantlexenoliths at hotspots that have very low 187Os/188Os aresimply portions of the refractory substrate attached to thebase of the oceanic lithosphere that were entrained inupwelling hotspot lavas (Bizimis et al., 2007; Ishikawaet al., 2011).

However, the discovery of low 187Os/188Os in abyssalperidotites (down to 0.1139; Liu et al., 2008), within errorof the lowest values from the Hawaiian xenolith suite,makes a xenolith origin in the modern Pacific mantle litho-sphere feasible. Indeed, like Hawaiian xenoliths, the187Os/188Os ratios measured in Savai’i and Tubuai xenolithsare also well-bracketed by the range of 187Os/188Os found inthe oceanic mantle lithosphere (Fig. 2), and the peridotitexenoliths from all three hotspots may originate in the ocea-nic lithosphere reservoir as sampled by abyssal peridotites.

Ishikawa et al. (2011) also argued that a subset of peri-dotite xenoliths recovered from the island of Malaita areresidues from the OJP mantle plume. This subset of peri-dotite xenoliths with the highest mantle temperatures (andinferred greatest depths) sample fragments of the previouslyrecycled ancient oceanic mantle lithosphere that makes uppart of the OJP mantle plume. These deeper xenoliths arefragments of the plume that underplated and formed arefractory substrate attached to the base of the Pacific man-tle lithosphere at �120 Ma when the OJP erupted. Ishikawaet al. (2007) maintained that a second population of lowertemperature (shallower) peridotite mantle xenoliths fromthe same locality are fragments of the 160 Ma Pacific litho-sphere upon which the OJP was constructed, and theselower temperature xenoliths are not fragments of the under-plated OJP mantle plume. The shallow and deep xenolithswere argued to exhibit different 187Os/188Os distributions,thereby supporting a distinct plume origin for the deeperxenoliths. However, there is substantial overlap in the Os-


isotopic composition of the shallow and deep OJP xenolithpopulations: the deep xenolith population (n = 12 xeno-liths) spans a range of 187Os/188Os (0.1152–0.1270) thatoverlaps considerably with the shallow xenolith population(0.1161–0.1388; n = 60). Additionally, the differentOs-isotopic distributions between the deep and shallowxenoliths may be related to undersampling the deeper pop-ulation, which consists of just 12 samples (which may notprovide a representative sampling of the Os-isotopic distri-bution of the deeper xenoliths). It is thus not clear whetherthe Os-isotopic dataset of Ishikawa et al. (2007) fully sup-ports a model where the deep OJP lithosphere has distinct187Os/188Os from the shallow lithosphere, and if this is true,then the deep and shallow OJP xenoliths may simply sam-ple typical oceanic mantle lithosphere. Indeed, abyssal peri-dotite 187Os/188Os ratios extend to values lower than thelowest identified in the OJP (Liu et al., 2008).

In summary, we cannot exclude the possibility that peri-dotite xenoliths from Hawaii and the OJP represent piecesof ancient subducted, depleted oceanic mantle lithosphereembedded in the respective mantle plumes that underplatedthe Pacific lithosphere. However, echoing Lassiter et al.(2014), we note that xenoliths from the hotspots with thelowest 187Os/188Os—Hawaii, OJP, Tubuai and Savai’i—allexhibit 187Os/188Os that falls within the range (or withinerror) of 187Os/188Os ratios measured in abyssal peridotites(Fig. 2).

4.4.3. Peridotite xenoliths at hotspots from the same

heterogeneous oceanic mantle lithosphere as abyssal

peridotites?

The depleted upper mantle sampled by abyssal peri-dotites and fracture zones has been found to host domainswith low 187Os/188Os. The low 187Os/188Os in many abyssalperidotite samples is too low to result from melt extractionat modern ridges, but instead requires depletion to haveoccurred at up to 2 Ga (Harvey et al., 2006; Liu et al.,2008). This observation suggests that ancient depleteddomains that have survived in the convecting mantle areincorporated into the oceanic mantle lithosphere at ridgesand appear in the abyssal peridotite data set. Portions ofthe oceanic mantle lithosphere (including portions withlow 187Os/188Os) beneath hotspots may be dislodged by(and entrained in) upwelling hotspot lavas. This modelargues that the peridotite xenoliths hosted in hotspots lavassample the same mantle reservoir (and have the sameheterogeneous Os-isotopic composition) as abyssal peri-dotites (e.g., Lassiter et al., 2014).

The depleted mantle sampled by abyssal peridotites is anear-perfect match for the Os-isotopic composition of theperidotite xenoliths sampled by Tubuai and Savai’i.Lassiter et al. (2014) noted that, globally, the peridotitexenolith population hosted in oceanic hotspot lavas (aver-age of 0.1244 ± 0.0043, 1SD) and abyssal peridotites (aver-age of 0.1245 ± 0.0043) have indistinguishable average187Os/188Os and nearly identical distributions. Like the glo-bal OIB peridotite xenolith average, the average187Os/188Os of Savai’i (0.1237 ± 0.0032, 1SD) and Tubuaixenoliths (0.1262 ± 0.0035) are indistinguishable from theglobal abyssal peridotite average compiled by Lassiter




et al. (2014). Additionally, the most unradiogenic187Os/188Os in Savai’i (down to 0.1173) and Tubuai(0.1163) are bracketed by the lowest values found in abyssalperidotites, as the depleted oceanic mantle sampled byabyssal peridotites also hosts unradiogenic 187Os/188Os(down to 0.1139; Harvey et al., 2006; Liu et al., 2008). Stud-ies of abyssal peridotite 187Os/188Os (Snow and Reisberg,1995; Standish et al., 2002; Harvey et al., 2006; Liu et al.,2008; Lassiter et al., 2014) demonstrate that the shallowoceanic mantle encompasses the range of Os-isotopic com-positions observed in oceanic hotspot peridotite xenoliths.The similarity in Os isotopic composition between the peri-dotite xenoliths in this study, other peridotite xenoliths athotspots globally, and abyssal peridotites permits an inter-pretation whereby the xenoliths sample an ancient depletedcomponent in the convecting mantle that has been incorpo-rated into the oceanic lithosphere at a mid-ocean ridge. Asdid Lassiter et al. (2014), we conclude that the similarity of187Os/188Os of ocean hotspot mantle xenoliths and abyssalperidotites suggests that they sample the same heteroge-neous mantle reservoir.

For Tubuai and Savai’i peridotite xenoliths, it is reason-able to adopt a model in which they also sample the samemantle reservoir as global average abyssal peridotites. Wefavor a model where the unradiogenic Os in xenoliths fromSamoa and Tubuai are simply portions of depleted domainsof ancient, heterogeneous oceanic mantle that have sur-vived intact in a convecting mantle for up to 1.8 Ga (seeTable 1). These depleted domains could be regions of uppermantle that experienced melt extraction in the past, andthey are incorporated into the oceanic mantle lithosphereduring mantle upwelling beneath ridges. Once capturedand incorporated into the oceanic lithosphere, thesedomains are sampled as abyssal peridotites or they aretransported from mantle lithospheric depths to the surfacein erupting ocean island basalts.

4.5. The paradox of the survival of early-Hadean 129Xe/130Xe

signatures, but not Hadean 187Os/188Os or 142Nd/144Nd, in

the convecting mantle

The low 187Os/188Os observed in Savai’i and Tubuaiperidotite xenoliths, and in abyssal peridotites and peri-dotite xenoliths from other hotspot localities, indicates thatancient, depleted mantle domains are preserved at multiplelocations in the ocean basins. While depletion ages of up to2 Ga are recorded at several localities, an important ques-tion is why even older depletion events, dating from thetime of the Archean and even Hadean, are not recordedin the Os-isotopic compositions measured in peridotitesfrom the oceanic mantle?

The geologic record provides evidence that mantle melt-ing that led to extraction of continental and oceanic crusthas occurred throughout much of Earth’s history.143Nd/144Nd and 176Hf/177Hf isotopes in Archean crustalrocks (e.g., Shirey and Hanson, 1986; Stein andHofmann, 1994; Moorbath et al., 1997; Blichert-Toftet al., 1999; Vervoort and Blichert-Toft, 1999; Caro, 2011;Khanna et al., 2014) and Hf isotopes in Hadean zircons(e.g., Harrison et al., 2005, 2008; Blichert-Toft and


Albarede, 2008; Kemp et al., 2010; Guitreau et al., 2012)preserve evidence for early mantle depletion by melt extrac-tion. Concomitant melting and Re-depletion will generatereservoirs that preserve low 187Os/188Os over time. If allRe is assumed to be extracted from the mantle during melt-ing, a model age—a Re-depletion age, called TRD—can becalculated for the depletion event by calculating when themeasured Os-isotopic composition of the depleted peri-dotite intersects the time evolution of mantle 187Os/188Os(Walker et al., 1989). Continental xenoliths from the SCLMpreserve the lowest known terrestrial 187Os/188Os, and haveRe-depletion model ages that indicate melt extraction asearly as the Mesoarchean (e.g., Walker et al., 1989; Shireyand Walker, 1998; Pearson et al., 2003; Carlson et al.,2005). Abyssal peridotites can exhibit lower 187Os/188Osthan chondritic compositions, thus preserving evidence ofpast melt depletion (e.g., Snow and Reisberg, 1995;Brandon et al., 2000; Standish et al., 2002; Harvey et al.,2006; Liu et al., 2008; Lassiter et al., 2014), but abyssal peri-dotites and mantle xenoliths sampling the oceanic mantledo not preserve 187Os/188Os as low as identified in theSCLM. The oldest TRD model ages in the oceanic mantleare only slightly older than 2 Ga (Fig. 2) (Harvey et al.,2006; Bizimis et al., 2007; Liu et al., 2008).

It is paradoxical that the oceanic mantle does not pre-serve the Os-isotopic record of earlier depletion events, asthe mechanisms responsible for mantle depletion aredirectly tied to plate tectonics, which, by some estimates,has operated for at least the past �3 Ga (VanKranendonk, 2010; Shirey and Richardson, 2011). Duringthe time period over which modern plate tectonics isthought to have operated, the upper mantle has been con-tinuously depleted by partial melting and melt extractionat ridges, and it is inferred that hotter mantle temperaturesat 3 Ga would have resulted in higher degrees of mantlemelting (and the generation of larger volumes of depletedmantle domains) compared to the more recent geologicalpast. While larger volumes of depleted mantle productionprior to 2 Ga might be expected, TRD model ages in oceanisland xenoliths and abyssal peridotites provide little recordof mantle depletion before �2 Ga, and give average modelages of 0.8 Ga and 0.7 Ga, respectively. Employing a globalcompilation of Os-isotopic data from abyssal peridotitesand ocean island peridotite xenoliths, Lassiter et al.(2014) found that various model ages based on the Re–Ossystem, including TMA (mantle extraction model age;Shirey and Walker, 1998) and TAl2O3 (‘‘aluminachron”model age; Reisberg and Lorand, 1996), do yield olderaverage ages for hotspot mantle xenoliths (1.3 Ga and1.6 Ga, respectively) and abyssal peridotites (1.7 Ga and1.5 Ga, respectively). These older ages are not unexpected,as the assumptions upon which the model ages are basedtend to yield older model depletion events than TRD ages(which yield minimum ages for the depletion event, as Reis unlikely to be completely extracted from the peridotiteand Re may be reintroduced into the peridotite by meltinfiltration, or ‘‘metasomatism”, following the depletionevent). Nonetheless, the various Os model ages still fail tocapture the large volumes of depleted mantle that musthave formed prior to 2 Ga (Fig. 2), and depleted reservoirs




with older ages are underrepresented or absent in the ocea-nic hotspot mantle xenolith and abyssal peridotite datasets.If the rate of production of oceanic lithosphere and the gen-eration of depleted mantle reservoirs was higher when themantle was hotter, then ancient Os depletion ages mightbe expected to be prevalent in the geologic record. Whyare old depletion ages rare or absent in the abyssal peri-dotite and oceanic hotspot xenolith Os-isotopic record?

The lack of very old Os-isotopic depletion ages in theoceanic mantle is a long-standing problem. A commonsolution is to suggest that isotopic homogenization duringconvective mixing has erased old Os-isotopic ages (Hauri,2002; Harvey et al., 2006), perhaps as Re-rich crust is recy-cled into the mantle and rehomogenized with Re-depletedhartzburgitic restite that comprises recycled oceanic mantlelithosphere (Lassiter et al., 2014) or perhaps simply becausemost of the Os is returned back into the mantle with theperidotitic portion of the oceanic lithosphere. If Os-isotopic evidence for early depletion events was erased bysuch a mixing process, the relatively young average Os-isotopic model ages available in peridotites sampling theconvecting mantle may reflect mixing timescales as theoceanic mantle is homogenized by mixing.

Convective mixing also likely explains the lack of142Nd/144Nd variability in the modern mantle. Early-formed silicate-Earth reservoirs exhibited 142Nd/144Nd vari-ability that was generated by decay of short-lived 146Sm to142Nd (146Sm half life is 68–103 Ma; Kinoshita et al., 2012)following early silicate Earth differentiation, and all terres-trial 142Nd/144Nd variability was generated in the lifetime of146Sm (the first 300–500 Ma of Earth’s history) (e.g.,Harper and Jacobsen, 1992; Caro et al., 2006; Boyet andCarlson, 2006; Bennett et al., 2007; O’Neil et al., 2008;Rizo et al., 2013). Variability in 142Nd/144Nd diminishedsince the Hadean but persisted in the mantle until 2.7 Ga(e.g., Bennett et al., 2007; Debaille et al., 2013), and142Nd/144Nd anomalies have not been observed in youngerrocks. Like early (Hadean and Archean) Re-depletionevents that are no longer recorded in the Os-isotopic com-position of the oceanic mantle, early-formed 142Nd/144Ndheterogeneity has not been detected in the modern convect-ing oceanic mantle (Caro et al., 2006; Boyet and Carlson,2006; Andreasen et al., 2008; Murphy et al., 2010;Cipriani et al., 2011; Caro, 2011; Jackson and Carlson,2012). A simple model for the erasure of early-formed142Nd/144Nd and 187Os/188Os isotopic variability is that vig-orous convection in the hotter Hadean and Archean mantlewas less conducive to preserving early-formed 142Nd/144Ndand Os-isotopic heterogeneities.

In contrast to the 187Os/188Os and 142Nd/144Nd isotopicrecords, which do not preserve early-formed isotopic signa-tures in the modern oceanic mantle, xenon isotopes recordheterogeneities in the modern oceanic mantle that formedprior to 4.45 Ga, indicating that Xe is more resistant toconvective mixing than Os and Nd. 129Xe is the decay pro-duct of the short lived isotope 129I (T1/2 = 15.7 Ma), and all129Xe/130Xe variability in the Earth was generated by frac-tionation of I from Xe during the first �100 million years ofEarth’s history. Early work on mantle-derived samplesshowed that the 129Xe/130Xe composition of the mantle is


different from the atmospheric composition (Staudacherand Allegre, 1982; Allegre et al., 1983; Poreda and Farley,1992). A recent series of papers have demonstrated thatheterogeneous 129Xe/130Xe signatures are preserved in themodern oceanic mantle and sampled at mid-ocean ridgesand plume influenced lavas from Iceland and RochambeauRift (Tucker et al., 2012; Parai et al., 2012; Mukhopadhyay,2012; Peto et al., 2013; Parai and Mukhopadhyay, 2015).This 129Xe/130Xe heterogeneity in the oceanic mantle wasgenerated in the early Hadean, and has survived for4.5 Ga in the convecting mantle to be sampled by recentMORB and OIB volcanism. Therefore, it is critical tounderstand how the early-formed 187Os/188Os and142Nd/144Nd isotopic record of mantle depletion was erasedwhile 129Xe/130Xe heterogeneities from the earliest Hadeanpersist in the modern oceanic mantle. Mukhopadhyay(2012) suggested that the erasure of Hadean 142Nd/144Ndsignatures, but not Hadean 129Xe/130Xe, from the convect-ing mantle is due to the small magnitude of the 142Nd/144Ndanomalies (<0.004% variability in 142Nd/144Nd has beenobserved in terrestrial Archean terraines; e.g., Caro et al.,2006; O’Neil et al., 2008; Rizo et al., 2013) compared tothe large range of 129Xe/130Xe signatures (>10% variabilityis observed in the modern mantle; Mukhopadhyay, 2012),and the smaller 142Nd/144Nd anomalies are more easilyerased than the larger magnitude 129Xe/130Xe hetero-geneities. One problem with using a ‘‘small magnitudeanomaly” explanation for the Os data is that Os-isotopicheterogeneities in the mantle are large, yet early-formedOs-isotopic signatures still have been erased. For example,the 187Os/188Os of early-formed depleted mantle at 4.55 Gawas �0.095, which is >25% lower than the modern mantle(0.1296; Meisel et al., 2001). Therefore, the absence ofearly-formed, large magnitude 187Os/188Os excursions inthe modern convecting mantle is evidence that the preserva-tion of early geochemical heterogeneities is not necessarily afunction of the magnitude of the original geochemicalanomaly. However, we cannot exclude the possibility that142Nd/144Nd heterogeneity exists in the modern mantle,but it is not detected because either (1) existing instrumen-tation does not yet provide the necessary precision toresolve such anomalies or (2) lavas sampling the modernmantle that have resolvable 142Nd/144Nd anomalies havenot been analyzed. Nonetheless, given the existing dataset,the 142Nd/144Nd in the modern mantle is homogeneous.

The variable fluxes of trace elements between differentterrestrial geochemical reservoirs—including the crust,mantle and atmosphere—must also be an important con-sideration when evaluating the survival of Hadean isotopicsignatures in the modern mantle (e.g., Galer and O’Nions,1985; Albarede, 2001; Kellogg et al., 2002).Mukhopadhyay (2012) proposed an alternative explanationfor the preservation of Hadean 129Xe/130Xe, but not142Nd/144Nd, isotopic signatures in the modern mantle,and this model may explain the lack of abundant, early-formed Os-isotopic reservoirs in the mantle as well.Mukhopadhyay (2012) argued that the lower recycling effi-ciency of xenon into the mantle compared to Nd helps pre-serve early-formed Xe-isotopic signatures. The mantledegasses xenon to the Earth’s atmosphere during mantle




melting, and noble gases are not efficiently returned to themantle during recycling (e.g., Staudacher and Allegre,1988). As a result, during partial melting at ridges followedby lithospheric return to the mantle, most of the mantlexenon enters the atmosphere. A small portion of the xenonis thought to be returned to the mantle during recycling(Porcelli and Wasserburg, 1995; Holland and Ballentine,2006; Holland et al., 2009; Kendrick et al., 2011; Moreira,2013). However, xenon is not as efficiently recycled backinto the deep Earth as Nd and Os (Mukhopadhyay, 2012)because xenon is incompatible, fluid mobile and volatile,and largely degassed to the atmosphere (Fig. 6). Therefore,early-formed Hadean 129Xe signatures in the mantle are notcompletely overprinted by young, recycled atmospheric Xe.

Using Xe-isotopic constraints, Parai andMukhopadhyay (2015) argue that 85–92% of the Xe inthe convecting mantle sampled by MORB and OIB lavasis recycled atmospheric Xe that was transported into themantle by subduction over geologic time. However, whilethe Xe budget of the mantle is actually dominated by recy-cled atmospheric Xe, the fraction of recycled Xe is not suf-ficient to erase Hadean 129Xe/130Xe signatures: much of theremaining Xe is primitive mantle Xe (even after accounting

Mantle Wedge

Oceanic lithosphere

Mantle Wedge

Arc crust

Arc crust

Xe

Nd, Re, Os

eX elements (Xe). Hadean mantle signatures more likely to be preserved.

Oceanic lithosphere

Xe degassing

compatible elements (Os)

mobile elements (Nd). Hadean mantle signatures more likely to be erased.

Fig. 6. The differing recycling efficiencies of the elements maydetermine whether Hadean isotopic signatures are preserved in themodern convecting mantle. Compatible elements (like Os, whichare retained in the mantle lithosphere) and incompatible elementsthat are not fluid mobile (like Nd) are efficiently returned to themantle during subduction, and Hadean isotopic signatures forthese elements are efficiently overprinted in the mantle. However,incompatible fluid-mobile elements, both volatile (e.g., Xe) andrefractory, are more efficiently extracted from the downgoing slaband accumulate in shallow geochemical reservoirs (e.g., thecontinental crust and atmosphere). Thus, incompatible fluid-mobile elements are not efficiently subducted into the mantle,and Hadean isotopic signatures of these elements are more likely tobe preserved in the modern convecting mantle.


for Xe produced by U and Pu fission). Unlike the case forXe, the fractions of the Nd and Os budgets in the mantlethat have been recycled are poorly constrained, but thisfraction must exceed the fraction of Xe in the mantle thatis recycled (i.e., >85–92%) such that the Hadean andArchean Nd and Os isotopic signatures have been largelyoverprinted in the convecting mantle.

For example, Nd is refractory and fluid-immobile andretained in the crustal portion of the oceanic lithosphere.Thus, Nd is more efficiently returned to the mantle duringlithospheric recycling, and recycled Nd (meaning Nd thathas gone through the plate tectonic cycle at least once) con-stitutes a large fraction of the mantle’s Nd inventory.Assuming modern subduction rates, ancient oceanic litho-sphere (including crust and mantle) recycled into theEarth’s mantle since the inception of plate tectonics mayreach �80% of the mantle’s mass (Salters and Stracke,2004; Brandenburg et al., 2008). Therefore, if Nd in oceaniclithosphere is immobile during subduction and quantita-tively returned to the mantle in subducting slabs, then�80% of the Nd in the mantle is recycled. This is likely aminimum value: if oceanic lithosphere formation and sub-duction operated at higher rates in the geologic past, thefraction of the Nd in the mantle that is recycled exceeds80% (and it is possible that nearly the entire Nd budgetof the mantle may be recycled). Thus, Hadean 142Nd signa-tures in the mantle are more efficiently overprinted by recy-cled material than Hadean 129Xe signatures.

Like Nd, Os is also efficiently returned to the mantleduring recycling. While Os is almost certainly mobile in flu-ids and some can be lost from the oceanic crustal portion ofa downgoing slab, little Os resides in the oceanic crust.Instead, Os is compatible and chalcophile and >99% ofthe Os in the oceanic lithosphere (assuming <100 ppt Osin the oceanic crustal lithosphere and 3000 ppt Os in thedepleted mantle lithosphere) is retained in the depletedoceanic mantle lithosphere portion of the slab, and is thusreturned to the mantle during recycling. If the concentra-tions of osmium in the convecting mantle and oceanic man-tle lithosphere are similar, and if the mantle is comprised of80% recycled mantle lithosphere, then �80% of the osmiumin the convecting mantle has been recycled and thus haspassed through the plate tectonic cycle as least once sincethe onset of plate tectonics. Again, this is a minimum value,and the fraction of Os in the convecting mantle that is recy-cled may approach 100%. Thus, early-formed, Hadean187Os/188Os signatures in the mantle have been more effi-ciently overprinted by recycled material than Hadean129Xe signatures.

Ultimately, the isotopic evolution of different radiogenicisotopic systems in the mantle depends on the relative recy-cling efficiencies of parent and daughter elements. However,unlike the 146Sm–142Nd and the 129I–129Xe systems, wherethe parent isotopes have been effectively extinct since theHadean, the parent in the 187Re–187Os system, 187Re, is stilldecaying and thus can modify the 187Os/188Os of themodern mantle. Like Os, Re is also efficiently returned tothe mantle during recycling of oceanic crust, and any Rereturned to the mantle will serve to overprint ancient(low) 187Os/188Os reservoirs via generation of radiogenic



Fig. 7. The conceptual model that relates survival of early-formed(Hadean) isotopic heterogeneities in the modern convecting mantleto the recycling efficiency of an element. Hadean 129Xe/130Xeheterogeneities are preserved in the modern convecting mantle(upper left portion of the figure) because xenon is not efficientlyrecycled back into the mantle (and a significant fraction of thexenon in the modern mantle reflects its primordial budget).Therefore, early-formed Hadean 129Xe/130Xe heterogeneities sur-vive in the modern mantle because they have not been diluted withrecycled xenon. In contrast, Nd and Os are efficiently recycled backinto the mantle at subduction zones. Therefore, 187Os/188Os and142Nd/144Nd signatures formed in the Hadean have been dilutedand ultimately overprinted by continuous recycling of young Osand Nd into the mantle over geologic time.


187Os in the mantle over time. Thus, long-term recycling of187Re also has played a supporting role with Os recycling inoverprinting ancient (low) 187Os/188Os domains in themantle.

The long-term preservation of early-formed isotopicsignatures of a particular element is a function of not justmantle convective stirring, but also the recycling efficiencyof the element back into the mantle at subduction zones(Gonnermann and Mukhopadhyay, 2009; Mukhopadhyay,2012). This conceptual model permits a more generaldescription of which elements are (e.g., Xe), and arenot (e.g., Nd and Os), likely to preserve Hadean geo-chemical signatures in the Earth’s mantle: incompatibleelements that are fluid mobile are more likely to preserveHadean geochemical signatures in the mantle than com-patible elements and fluid immobile incompatible ele-ments. Incompatible, fluid-mobile elements (like Xe) areefficiently extracted from the downgoing slab and accu-mulate in shallow reservoirs (e.g., the crust and atmo-sphere) instead of being returned to the mantle. Suchelements are less efficiently returned to the mantle thanOs and Nd, and are therefore more likely to preserveearly-formed, primordial signatures in the mantle. Theincompatible, fluid-mobile element need not be volatileto preserve Hadean geochemical signatures in the mantle:incompatible, fluid-mobile elements that are refractoryalso will be extracted from the downgoing slab, concen-trated in the continental crust, and not efficientlyreturned to the mantle. For example, W is both fluid-mobile and incompatible (Arevalo and McDonough,2008; Konig et al., 2008, 2011) and the long-term preser-vation of early (<30 Ma) formed 182W/184W variability in2.8 Ga komatiites (Touboul et al., 2012) and relativelymodern flood basalts (Rizo et al., 2015) would be favoredin the model presented here. However, non-fluid mobileincompatible elements (like Nd) and compatible elements(which, like Os, are preserved in the mantle lithosphereduring recycling) are efficiently returned to the mantleduring recycling and mix with, dilute, and ultimatelyoverprint primordial signatures associated with the iso-topes of such elements (Fig. 7). While high recycling effi-ciencies (for Nd and Os) may not be conducive to thepreservation of early-formed heterogeneities, somewhatlower recycling efficiencies for incompatible fluid-mobileelements (e.g., Xe) may allow long-term preservation ofearly-formed Hadean signatures in the convecting mantle.

This conceptual model may be extended to many iso-topic systems, including the preservation of primordial3He/4He isotopic signatures in the deep mantle resultingfrom the return of noble gas depleted slabs (Gonnermannand Mukhopadhyay, 2009), to build an understanding forthe role of plate tectonics and subduction in determiningthe preservation of early-formed, primordial reservoirs inthe Earth’s mantle (Figs. 6 and 7). Several radiogenic iso-topic systems based on incompatible lithophile elements,including 87Rb–87Sr, 147Sm–143Nd, 176Lu–176Hf and238,235U–232Th–208,207,206Pb, are unlikely to preserveHadean isotopic signatures in the convecting mantle: owingto variable fluid mobility during subduction, either the par-ent (e.g., U and Th), the daughter (e.g., Sr), or both (e.g.,


Sm–Nd and Lu–Hf) are likely to be efficiently returned tothe mantle. Thus, these radiogenic isotopic systems are lesslikely to preserve memories of the Hadean in the convectingmantle. However, some deep mantle reservoirs, possiblylocated in the large low shear wave velocity provinces(LLSVPs), may remain relatively unmodified by the addi-tion of subducted material over geologic time (White,2015). These relatively unmodified reservoirs, characterizedby hosting primitive noble gas signatures (White, 2015),provide an exception where Hadean isotopic signaturescan persist even for elements (e.g., Pb) that may be effi-ciently recycled into the mantle (Jackson et al., 2010a,b;Jackson and Carlson, 2011).

ACKNOWLEDGEMENTS

We acknowledge discussion with Roberta Rudnick, SujoyMukhopadhyay, Bernhard Peucker-Ehrenbrink, AndrewReinhard and Jasper Konter. We thank Alison Price for helpimproving a figure. MGJ thanks Vincent Salters for askingwhy xenon-isotopic heterogeneities are preserved in the mantle,but early-formed Os-isotopic signatures are not. We thankJessica Warren and Graham Pearson for sharing Os-isotopicdatabases for peridotites. Constructive and helpful reviews fromDave Graham, Bill White and Barry Hanan are acknowledged.MGJ thanks Fred Frey for the introduction to trace elementgeochemistry during coursework at MIT. Jackson acknowledgessupport from NSF grants EAR-1348082, EAR-1347377,EAR-1145202 and OCE-1153894. Kurz acknowledges NSFgrants OCE-1259218 and OCE-1232985.




REFERENCES

Alard O., Griffin W. L., Lorand J.-P., Jackson S. E. and O’Reilly S.Y. (2000) Non-chondritic distribution of the highly siderophileelements in mantle sulphides. Nature 407, 891–894.

Alard O., Luguet A., Pearson N. J., Griffin W. L., Lorand J.-P.,Gannoun A., Burton K. W. and O’Reilly S. Y. (2005) In-situ Osisotopes in abyssal peridotites bridge the isotopic gap betweenMORB and their source mantle. Nature 436, 1005–1008.

Albarede F. (2001) Radiogenic ingrowth in systems with multiplereservoirs: applications to the differentiation of the mantle-crustsystem. Earth Planet. Sci. Lett. 189, 59–73.

Allegre C. J., Staudacher T., Sarda P. and Kurz M. D. (1983)Constraints on evolution of Earth’s mantle from rare gassystematics. Nature 303, 762–766.

Andreasen R., Sharma M., Subbarao K. V. and Viladkar S. G.(2008) Where on Earth is the enriched Hadean reservoir? EarthPlanet. Sci. Lett. 266, 14–28.

Arevalo R. and McDonough W. F. (2008) Tungsten geochemistryand implications for understanding the Earth’s interior. EarthPlanet. Sci. Lett. 272, 656–665.

Becker H., Horan M. F., Walker R. J., Gao S., Lorand J.-P. andRudnick R. L. (2006) Highly siderophile element compositionof the Earth’s primitive upper mantle: constraints from newdata on peridotite massifs and xenoliths. Geochim. Cosmochim.

Acta 70, 4528–4550.Bellon H., Brousse R. and Pantaloni A. (1980) Ages de l’ile de

Tubuai: l’alignement des Australes et des Cook. Cah. Pac. 2.Bennett V. C., Brandon A. D. and Nutman A. P. (2007) Coupled

142Nd–143Nd isotopic evidence for Hadean mantle dynamics.Science 318, 1907–1910.

Bizimis M., Griselin M., Lassiter J. C., Salters V. J. M. and Sen G.(2007) Ancient recycled mantle lithosphere in the Hawaiianplume: osmium–hafnium isotopic evidence from peridotitemantle xenoliths. Earth Planet. Sci. Lett. 257, 259–273.

Blichert-Toft J. and Albarede F. (2008) Hafnium isotopes in JackHills zircons and the formation of the Hadean crust. Earth

Planet. Sci. Lett. 265, 686–702.Blichert-Toft J., Albarede F., Rosing M., Frei R. and Bridgwater

D. (1999) The Nd and Hf isotopic evolution of the mantlethrough the Archean. Results from the Isua supracrustals, WestGreenland, and from the Birimian terranes of West Africa.Geochim. Cosmochim. Acta 63, 3901–3914.

Boyet M. and Carlson R. W. (2006) A new geochemical model forthe Earth’s mantle inferred from 146Sm–144Nd systematics.Earth Planet. Sci. Lett. 250, 254–268.

Brandenburg J. P., Hauri E. K., van Keken P. E. and Ballentine C.J. (2008) A multiple system study of the geochemical evolutionof the mantle with force-balanced plates and thermochemicaleffects. Earth Planet. Sci. Lett. 276, 1–13.

Brandon A. D., Snow J. E., Walker R. J., Morgan J. W. and MockT. D. (2000) 190Pt–186Os and 187Re–187Os systematics of abyssalperidotites. Earth Planet. Sci. Lett. 177, 319–355.

Brugmann G. E., Ryabchikov I. D. and Kogarko L. N. (2008) Osisotope geochemistry of mantle peridotites from Sal Island,Cape Verde Archipelago. Doklady Earth Sci. 419, 325–328.

Burnard P. G., Farley K. A. and Turner G. (1998) Multiple fluidpulses in a Samoan harzburgite. Chem. Geol. 147, 99–114.

Burton K. W., Schiano P., Birck J.-L. and Allegre C. J. (1999)Osmium isotope disequilibrium between mantle minerals in aspinel–lherzolite. Earth Planet. Sci. Lett. 172, 311–322.

Carlson R. W. and Irving A. J. (1994) Depletion and enrichmenthistory of subcontinental lithospheric mantle: an Os, Sr, Nd, andPb isotopic study of ultramafic xenoliths from the northwesternWyoming Craton. Earth Planet. Sci. Lett. 126, 457–472.


Carlson R. W., Pearson D. G. and James D. E. (2005) Physical,chemical, and chronological characteristics of continentalmantle. Rev. Geophys. 43, 1–24.

Caro G., Bourdon B., Birck J. L. and Moorbath S. (2006) High-precision 142Nd/144Nd measurements in terrestrial rocks: con-straints on the early differentiation of the Earth’s mantle.Geochim. Cosmochim. Acta 70, 164–191.

Caro G. (2011) Early silicate Earth differentiation. Annu. Rev.

Earth Planet. Sci. 39, 31–58.Chauvel C., Hofmann A. W. and Vidal P. (1992) HIMU-EM: the

French Polynesian connection. Earth Planet. Sci. Lett. 110, 99–119.

Chauvel C., McDonough W., Guille G., Maury R. and Duncan R.(1997) Contrasting old and young volcanism in Rurutu Island,Austral chain. Chem. Geol. 139, 125–143.

Cipriani A., Bonatti E. and Carlson R. W. (2011) Nonchondritic142Nd in suboceanic mantle peridotites. Geochem. Geophys.

Geosyst. 12, Q03006. http://dx.doi.org/10.1029/2010GC003415.Class C., Goldstein S. L. and Shirey S. B. (2009) Osmium isotopes

in Grande Comore lavas: a new extreme among a spectrum ofEM-type mantle endmembers. Earth Planet. Sci. Lett. 284, 219–227.

Cohen A. S. and Waters F. G. (1996) Separation of osmium fromgeological materials by solvent extraction for analysis bythermal ionisation mass spectrometry. Anal. Chim. Acta 332,269–275.

Coltorti M., Bonadiman C., O’Reilly S. Y., Griffin W. L. andPearson N. J. (2010) Buoyant ancient continental mantleembedded in oceanic lithosphere (Sal Island, Cape VerdeArchipelago). Lithos 120, 223–233.

Creaser R. A., Papanastassiou D. A. and Wasserburg G. J. (1991)Negative thermal ion mass spectrometry of osmium, rhenium,and iridium. Geochim. Cosmochim. Acta 55, 397–401.

Day J. M. D., Pearson D. G., Macpherson C., Lowry D. andCarracedo J.-C. (2009) Pyroxenite-rich mantle formed byrecycled oceanic lithosphere: oxygen–osmium isotope evidencefrom Canary Island lavas. Geology 37, 555–558.

Day J. M. D., Pearson D. G., Macpherson C. G., Lowry D. andCarracedo J. C. (2010) Evidence for distinct proportions ofsubducted oceanic crust and lithosphere in HIMU-type mantlebeneath El Hierro and La Palma, Canary Islands. Geochim.

Cosmochim. Acta 74, 6565–6589.Debaille V., Trønnes R. G., Brandon A. D., Waight T. E., Graham

D. W. and Lee C.-T. A. (2009) Primitive off-rift basalts fromIceland and Jan Mayen: Os-isotopic evidence for a mantlesource containing enriched subcontinental lithosphere. Geo-

chim. Cosmochim. Acta 73, 3423–3449.Debaille V., O’Neill C., Brandon A. D., Haenecour P., Yin Q.-Z.,

Mattielli N. and Treiman A. H. (2013) Stagnant-lid tectonics inearly Earth revealed by 142Nd variations in late Archean rocks.Earth Planet. Sci. Lett. 373, 83–92.

Farley K. (1995) Rapid cycling of subducted sediments into theSamoan mantle plume. Geology 23, 531–534.

Farley K., Natland J. and Craig H. (1992) Binary mixing ofenriched and undegassed (primitive) components (He, Sr, Nd,Pb) in Samoan lavas. Earth Planet. Sci. Lett. 111, 183–199.

Farley K. A., Poreda R. J. and Onstott T. C. (1994) Noble gases indeformed xenoliths from an ocean island: characterization of ametasomatic fluid. In Noble Gas Geochemistry and Cosmochem-

istry (ed. J.-I. Matsuda), pp. 159–178.Frey F. A. (1980) The origin of pyroxenites and garnet pyroxenites

from Salt Crater, Oahu, Hawaii; trace element evidence. Am. J.

Sci. 280-A, 427–449.Frey F. A. and Prinz M. (1978) Ultramafic inclusions from San

Carlos, Arizona: petrologic and geochemical data bearing ontheir petrogenesis. Earth Planet. Sci. Lett. 38, 129–176.


http://refhub.elsevier.com/S0016-7037(16)30048-5/h0005
























































































http://dx.doi.org/10.1029/2010GC003415




















































Frey F. A., Weis D., Borisova A. Y. and Xu G. (2002) Involvementof continental crust in the formation of the CretaceousKerguelen plateau: new perspectives from ODP Let 120 sites.J. Petrol. 43, 1207–1239.

Galer S. J. G. and O’Nions R. K. (1985) Residence time of Th, Uand Pb in the mantle with implications for mantle convection.Nature 316, 778–782.

Georgen J. E., Kurz M. D., Dick H. J. B. and Lin J. (2003) Low3He/4He ratios in basalt glasses from the western SouthwestIndian Ridge (10�–24�E). Earth Planet. Sci. Lett. 206, 509–528.

Gonnermann H. M. and Mukhopadhyay S. (2009) Preservingnoble gases in a convecting mantle. Nature 459, 560–563.

Graham D. W., Humphris S. E., Jenkins W. J. and Kurz M. D.(1992) Helium isotope geochemistry of some volcanic rocksfrom Saint Helena. Earth Planet. Sci. Lett. 110, 121–131.

Graham D. W. (2002) Noble gas isotope geochemistry of mid-ocean ridge and ocean island basalts; characterization of mantlesource reservoirs. In Noble Gases in Geochemistry and Cosmo-

chemistry, vol. 47 (eds. D. Porcelli, C. J. Ballentine and R.Wieler), pp. 247–318. Reviews in Mineralogy andGeochemistry.

Guitreau M., Blichert-Toft J., Martin H., Mojzsis S. J. andAlbarede F. (2012) Hafnium isotope evidence from Archeangranitic rocks for deep-mantle origin of continental crust. EarthPlanet. Sci. Lett. 337–338, 211–223.

Handler M. R., Bennett V. C. and Dreibus G. (1999) Evidencefrom correlated Ir/Os and Cu/S for late-stage Os mobility inperidotite xenoliths: implications for Re–Os systematics. Geol-ogy 27, 75–78.

Hanyu T., Tatsumi Y., Senda R., Miyazaki T., Chang Q., HiraharaY., Takahashi T., Kawabata H., Suzuki K. and Kimura J.-I.(2011a) Geochemical characteristics and origin of the HIMUreservoir: a possible mantle plume source in the lower mantle.Geochem. Geophys. Geosyst. 12. http://dx.doi.org/10.1029/2010GC003252 Q0AC09.

Hanyu T., Tatsumi Y. and Kimura J.-I. (2011b) Constraints on theorigin of the HIMU reservoir from He–Ne–Ar isotope system-atics. Earth Planet. Sci. Lett. 307, 377–386.

Hanyu T., Dosso L., Ishizuka O., Tani K., Hanan B. B., Adam C.,Nakai S., Senda R., Chang Q. and Tatsumi Y. (2013)Geochemical diversity in submarine HIMU basalts fromAustral Islands, French Polynesia. Contrib. Mineral Petrol.

166, 1285–1304.Harper C. L. and Jacobsen S. B. (1992) Evidence from coupled

147Sm–143Nd and 146Sm–142Nd systematics for very early(4.5 Gyr) differentiation of the Earth’s mantle. Nature 360,728–732.

Harrison T. M., Blichert-Toft J., Muller W., Albarede F., HoldenP. and Mojzsis S. J. (2005) Heterogeneous Hadean hafnium:evidence of continental crust at 4.4 to 4.5 Ga. Science 310,1947–1950.

Harrison T. M., Schmitt A. K., McCulloch M. T. and Lovera O.M. (2008) Early (P4.5 Ga) formation of terrestrial crust: Lu–Hf, d18O, and Ti thermometry results for Hadean zircons. EarthPlanet. Sci. Lett. 268, 476–486.

Hart S. R. and Ravizza G. (1996) Os partitioning between phasesin lherzolite and basalt. In Earth Processes: Reading the Isotopic

Code, vol. 95 (eds. A. Basu and S. R. Hart), pp. 123–134.Geophys. Monogr. Am. Geophys. Union.

Harvey J., Gannoun A., Burton K. W., Rogers N. W., Alard O.and Parkinson I. J. (2006) Ancient melt extraction from theoceanic upper mantle revealed by Re–Os isotopes in abyssalperidotites from the Mid-Atlantic Ridge. Earth Planet. Sci.

Lett. 244, 606–621.


Hassler D. R. and Shimizu N. (1998) Osmium isotopic evidence forancient subcontinen-tal lithospheric mantle beneath the Ker-guelen Islands, southern Indian Ocean. Science 280, 418–421.

Hauri E. H. (1992) Geochemical and fluid dynamic investigations

into the nature of chemical heterogeneity in the Earth’s mantle

(PhD thesis). Massachusetts Institute of Technology/WoodsHole Oceanographic Institution Joint Program inOceanography.

Hauri H. (2002) Osmium isotopes and mantle convection. Philos.Trans. R. Soc. Lond. A 360, 2371–2382.

Hauri E. H. (1997) Melt migration and mantle chromatography, 1:simplified theory and conditions for chemical and isotopicdecoupling. Earth Planet. Sci. Lett. 153, 1–19.

Hauri E. H. and Hart S. R. (1993) Re–Os isotope systematic ofHIMU and EMII oceanic island basalts from the south PacificOcean. Earth Planet. Sci. Lett. 114, 353–371.

Hauri E. and Hart S. (1994) Constraints on melt migration frommantle plumes: a trace element study of peridotite xenolithsfrom Savai’i, Western Samoa. J. Geophys. Res. 99, 24301.

Hauri E. H. and Kurz M. D. (1997) Melt migration and mantlechromatography, 2: a time-series Os isotope study of MaunaLoa volcano, Hawaii. Earth Planet. Sci. Lett. 153, 21–36.

Hauri E. H., Shimizu N., Dieu J. J. and Hart S. R. (1993) _Evidence for hotspot-related carbonatite metasomatism in theoceanic upper mantle. Nature 365, 221–227.

Heber V. S., Brooker R. A., Kelley S. P. and Wood B. J. (2007)Crystal melt partitioning of noble gases (helium, neon, argon,krypton, and xenon) for olivine and clinopyroxene. Geochim.

Cosmochim. Acta 71, 1041–1061.Holland G. and Ballentine C. J. (2006) Seawater subduction

controls the heavy noble gas composition of the mantle. Nature

441, 186–191.Holland G., Cassidy M. and Ballentine C. J. (2009) Meteorite Kr in

Earth’s mantle suggests a late accretionary source for theatmosphere. Science 326, 1522–1525.

Ingle S., Weis D. and Frey F. A. (2002) Indian continental crustrecovered from Elan Bank, Kerguelen Plateau (ODP Leg 183,Site 1137). J. Petrol. 43, 1241–1257.

Ishikawa A., Kuritani T., Makishima A. and Nakamura E. (2007)Ancient recycled crust beneath theOntong JavaPlateau: Isotopicevidence from the garnet clinopyroxenite xenoliths, Malaita,Solomon Islands. Earth Planet. Sci. Lett. 259, 134–148.

Ishikawa A., Pearson D. G. and Dale C. W. (2011) Ancient Osisotope signatures from the Ontong Java Plateau lithosphere:tracing lithospheric accretion history. Earth Planet. Sci. Lett.

301, 159–170.Jackson M. G. and Shirey S. B. (2011) Re–Os isotope systematics

in Samoan shield lavas and the use of Os-isotopes in olivinephenocrysts to determine primary magmatic compositions.Earth Planet. Sci. Lett. 312, 91–101.

Jackson M. G. and Carlson R. (2011) An ancient recipe for floodbasalt genesis. Nature 476, 316–319.

Jackson M. G. and Carlson R. W. (2012) Homogeneous super-chondritic 142Nd/144Nd in the mid-ocean ridge basalt and oceanisland basalt mantle. Geochem. Geophys. Geosyst. 13. http://dx.doi.org/10.1029/2012GC004114.

Jackson C. R. M., Parman S. W., Kelley S. P. and Kelley R. F.(2013) Constraints on light noble gas partitioning at theconditions of spinel-peridotite melting. Earth Planet. Sci. Lett.

384, 178–187.Jackson M. G., Carlson R., Kurz M. D., Kempton P. D., Francis

D. and Blusztajn J. (2010a) Evidence for the survival of theoldest terrestrial mantle reservoir. Nature 466, 853–856.

Jackson M. G., Hart S. R., Koppers A. A. P., Staudigel H., KonterJ., Blusztajn J., Kurz M. D. and Russell J. A. (2007a) The


































http://dx.doi.org/10.1029/2010GC003252

http://dx.doi.org/10.1029/2010GC003252



























































































http://dx.doi.org/10.1029/2012GC004114

http://dx.doi.org/10.1029/2012GC004114












return of subducted continental crust in Samoan lavas. Nature

448, 684–697.Jackson M. G., Kurz M. D., Hart S. R. and Workman R. K.

(2007b) New Samoan lavas from Ofu Island reveal a hemi-spherically heterogeneous high 3He/4He mantle. Earth Planet.

Sci. Lett. 264, 360–374.Jackson M. G., Hart S. R., Konter J. G., Koppers A. A. P.,

Staudigel H., Kurz M. D., Blusztajn J. and Sinton J. M. (2010b)Samoan hot spot track on a ‘‘hot spot highway” Implicationsfor mantle plumes and a deep Samoan mantle source. Geochem.

Geophys. Geosyst. 11. http://dx.doi.org/10.1029/2010GC003232Q12009.

JacksonM.G., Hart S. R., Konter J. G., KurzM.D., Blusztajn J. andFarley K. (2014) Helium and lead isotopes reveal the geochemicalgeometry of the Samoan plume. Nature 514, 355–358.

Kellogg J. B., Jacobsen S. B. and O’Connell R. J. (2002) Modelingthe distribution of isotopic tracers in geochemical reservoirs.Earth Planet. Sci. Lett. 204, 183–202.

Kemp A. I. S., Wilde S. A., Hawkesworth C. J., Coath C. D.,Nemchin A., Pidgeon R. T., Vervoort J. D. and DuFrane S. A.(2010) Hadean crustal evolution revisited: new constraints fromPb–Hf isotope systematics of the Jack Hills zircons. Earth

Planet. Sci. Lett. 296, 45–56.Kendrick M. A., Scambelluri M., Honda M. and Phillips D. (2011)

High abundances of noble gas and chlorine delivered to themantle by serpentine subduction. Nat. Geosci. 4, 807–812.

Kinoshita N., Paul M., Kashiv Y., Collon P., Deibel C. M.,DiGiovine B., Greene J. P., Henderson D. J., Jiang C. L.,Marley S. T., Nakinishi T., Pardo R. C., Rehm K. E.,Robertson D., Scott R., Schmitt C., Tang X. D., VondrasekR. and Yokoyama A. (2012) A shorter 146Sm half-life measuredand implications for 146Sm–142Nd chronology of the solarsystem. Science 335, 1614–1617.

Khanna T. C., Bizimis M., Yogodzinski G. M. and Mallick S.(2014) Hafnium–neodymium isotope systematics of the 2.7 GaGadwal greenstone terrane, Eastern Dharwar craton, India:implications for the evolution of the Archean depleted mantle.Geochim. Cosmochim. Acta 127, 10–24.

Kogiso T., Tatsumi Y., Shimoda G. and Barsczus H. G. (1997)High l (HIMU) ocean island basalts in southern Polynesia: newevidence for whole mantle scale recycling of subducted oceaniccrust. J. Geophys. Res. 102, 8085–8103.

Konig S., Munker C., Schuth S. and Garbe-Schonberg D. (2008)Mobility of tungsten in subduction zones. Earth Planet. Sci.

Lett. 274, 82–92.Konig S., Munker C., Hohl S., Paulick H., Barth A. R., Lagos M.,

Pfander J. and Buchl A. (2011) The Earth’s tungsten budgetduring mantle melting and crust formation. Geochim. Cos-

mochim. Acta 75, 2119–2136.Konter J. G. and Jackson M. G. (2012) Large volumes of

rejuvenated volcanism in Samoa: Evidence supporting atectonic influence on late-stage volcanism. Geochem. Geophys.

Geosyst. 13. http://dx.doi.org/10.1029/2011GC003974.Kurz M. D., Curtice J., Lott, III, D. E. and Solow A. (2004)

Rapid helium isotopic variability in Mauna Kea shield lavasfrom the Hawaiian Scientific Drilling Project. Geochem.

Geophys. Geosyst. 5. http://dx.doi.org/10.1029/2002GC000439 Q04G14.

Lassiter J. C., Byerly B. L., Snow J. E. and Hellebrand E. (2014)Constraints from Os-isotope variations on the origin of LenaTrough abyssal peridotites and implications for the composi-tion and evolution of the depleted upper mantle. Earth Planet.

Sci. Lett. 403, 178–187.Lorand J., Luguet A., Alard O., Bezos A. and Meisel T. (2008)

Abundance and distribution of platinum-group elements in


orogenic lherzolites: a case study in a Fontete Rouge lherzolite(French Pyrenees). Chem. Geol. 248, 174–194.

Liu C.-Z., Snow J. E., Hellebrand E., Brugmann G., von der HandtA., Buchl A. and Hofmann A. W. (2008) Ancient, highlyheterogeneous mantle beneath Gakkel ridge, Arctic Ocean.Nature 452, 311–316.

Mata J., Moreira M., Doucelance R., Ader M. and Silva L. C.(2010) Noble gas and carbon isotopic signatures of Cape Verdeoceanic carbonatites: implications for carbon provenance.Earth Planet. Sci. Lett. 291, 70–83.

McDougall I. (2010) Age of volcanism and its migration in theSamoa Islands. Geol. Mag. 147, 705–717.

Meisel T., Walker R. J., Irving A. J. and Lorand J. P. (2001)Osmium isotopic compositions of mantle xenoliths: a globalperspective. Geochim. Cosmochim. Acta 65, 1311–1323.

Moorbath S., Whitehouse M. J. and Kamber B. S. (1997) ExtremeNd-isotope heterogeneity in the early Archaean—fact orfiction? Case histories from northern Canada and WestGreenland. Chem. Geol. 135, 213–231.

Moreira M. (2013) Noble gas constraints on the origin andevolution of Earth’s volatiles. Geochem. Perspect. 2, 229–403.

Morgan J. W. and Walker R. J. (1989) Isotopic determinationsof rhenium and osmium in meteorites by using fusion,distillation and ion-exchange separations. Anal. Chim. Acta

222, 291–300.Mukhopadhyay S. (2012) Early differentiation and volatile accre-

tion recorded in deep-mantle neon and xenon. Nature 486, 101–104.

Murphy D. T., Brandon A. D., Debaille V., Burgess R. andBallentine C. (2010) 142Nd/144Nd reservoir in the deep mantle:Implications for the Nd isotope systematics of the Earth.Geochim. Cosmochim. Acta 74, 738–750.

O’Neil J., Carlson R. W., Francis D. and Stevenson R. K. (2008)Neodymium-142 evidence for Hadean mafic crust. Science 321,1828–1831.

O’Neil J., Francis D. and Carlson R. W. (2011) Implications of theNuvvuagittuq greenstone belt for the formation of Earth’s earlycrust. J. Pet. 52, 985–1009.

Parai R., Mukhopadhyay S. and Lassiter J. C. (2009) Newconstraints on the HIMU mantle from neon and heliumisotopic compositions of basalts from the Cook-AustralIslands. Earth Planet. Sci. Lett. 277, 253–261.

Parai R., Mukhopadhyay S. and Standish J. (2012) Heterogeneousupper mantle Ne, Ar and Xe isotopic compositions and apossible Dupal noble gas signature recorded in basalts from theSouthwest Indian Ridge. Earth Planet. Sci. Lett. 359–360, 227–239.

Parai R. and Mukhopadhyay S. (2015) The evolution of MORBand plume mantle volatile budgets: Constraints from fission Xeisotopes in Southwest Indian Ridge basalts. Geochem. Geophys.

Geosyst. 16, 719–735.Parman S. W., Kurz M. D., Hart S. R. and Grove T. L. (2005)

Helium solubility in olivine and implications for high 3He/4Hein ocean island basalts. Nature 437, 1140–1143.

Pearson D. G. and Wittig N. (2008) Formation of Archaeancontinental lithosphere and its diamonds: the root of theproblem. J. Geol. Soc. London 165, 895–914.

Pearson D. G., Canil D. and Shirey S. B. (2003) Mantle samplesincluded in volcanic rocks: Xenoliths and diamonds. In The

Mantle, vol. 2 (ed. R. W. Carlson). Elsevier, New York, pp.171–277.

Peto M., Mukhopadhyay S. and Kelley K. (2013) Heterogeneitiesfrom the first 100 million years recorded in deep mantle noblegases from the Northern Lau Back-arc Basin. Earth Planet. Sci.

Lett. 369–370, 13–23.










http://dx.doi.org/10.1029/2010GC003232









































http://dx.doi.org/10.1029/2011GC003974

http://dx.doi.org/10.1029/2002GC000439

http://dx.doi.org/10.1029/2002GC000439















































































Porcelli D. and Wasserburg G. J. (1995) Mass transfer of xenonthrough a steady-state upper mantle. Geochim. Cosmochim.

Acta 59, 1991–2007.Poreda R. J. and Farley K. (1992) Rare gases in Samoan xenoliths.

Earth Planet. Sci. Lett. 113, 129–144.Recanati A., Kurz M. D., Warren J. M. and Curtice J. (2012)

Helium distribution in a mantle shear zone from the JosephinePeridotite. Earth Planet. Sci. Lett. 359–360, 162–172.

Reid J. B. J. and Frey F. A. (1971) Rare Earth distributions inlherzolite and garnet pyroxenite xenoliths and the constitutionof the upper mantle. J. Geophys. Res. 76, 1184–1196.

Reisberg L. and Lorand P.-P. (1996) Longevity of sub-continentalmantle lithosphere from osmium isotope systematics in oro-genic peridotite massifs. Nature 376, 159–162.

Reisberg L. and Meisel T. (2002) The Re–Os isotopic system: areview of analytical techniques. Geostandards Newslett. 26, 249–267.

Reisberg L., Zhi X. C., Lorand J.-P., Wagner C., Peng Z. C. andZimmermann C. (2005) Re–Os and S systematics of spinelperidotite xenoliths from east central China: evidence forcontrasting effects of melt percolation. Earth Planet. Sci. Lett.

239, 286–308.Rizo H., Boyet M., Blichert-Toft J., O’Neil J., Rosing M. T. and

Paquette J. L. (2013) The elusive Hadean enriched reservoirrevealed by 142Nd deficits in Isua Archaean rocks. Nature 491,96–100.

Rizo H., Walker R. J., Carlson R. W., Mukhopadhyay S., FrancisD. and Jackson M. (2015) Memories of Earth formation in themodern mantle: W isotopic composition of flood basalt lavas.EOS (Transactions of AGU) 96 DI43B-05.

Roy-Barman M. and Allegre C. J. (1995) 187Os/186Os in oceanicisland basalts: tracing oceanic crust recycling in the mantle.Earth Planet. Sci. Lett. 129, 145–161.

Salters V. J. M. and Stracke A. (2004) Composition of the depletedmantle. Geochem. Geophys. Geosyst. 5. http://dx.doi.org/10.1029/2003GC000597.

Schaefer B. F., Turner S., Parkinson I., Rogers N. andHawkesworth Chris (2002) Evidence for recycled Archaeanoceanic mantle lithosphere in the Azores plume. Nature 420,304–307.

Schiano P., Burton K. W., Dupre B., Birck J.-L., Guille G. andAllegre C. J. (2001) Correlated Os–Pb–Nd–Sr isotopes in theAustral-Cook chain basalts: the nature of mantle componentsin plume sources. Earth Planet. Sci. Lett. 186, 527–537.

Sen G., Yang H.-J. and Ducea M. (2003) Anomalous isotopes andtrace element zoning in plagioclase peridotite xenoliths of Oahu(Hawaii): implications for the Hawaiian plume. Earth Planet.

Sci. Lett. 207, 23–38.Sen I. S., Bizimis M., Sen G. and Huang S. (2011) A radiogenic Os

component in the oceanic lithosphere? Constraints fromHawaiian pyroxenite xenoliths. Geochim. Cosmochim. Acta

75, 4899–4916.Shen J. J., Papanastassiou D. A. and Wasserburg G. J. (1996)

Precise Re–Os determinations and systematic of iron mete-orites. Geochim. Cosmochim. Acta 60, 2887–2900.

Shirey S. B. and Hanson G. N. (1986) Mantle heterogeneity andcrustal recycling in Archean granite-greenstone belts: evidencefrom Nd isotopes and trace elements in the Rainy Lake area,Superior Province, Ontario, Canada. Geochim. Cosmochim.

Acta 50, 2631–2651.Shirey S. B. and Walker R. J. (1995) Carius tube digestions for low-

blank rhenium–osmium analysis. Anal. Chem. 67, 2136–2141.Shirey S. B. and Walker R. J. (1998) The Re–Os isotope system in

cosmochemistry and high-temperature geochemistry. Annu.

Rev. Earth Planet. Sci. 26, 423–500.


Shirey S. B. and Richardson S. H. (2011) Start of the Wilson Cycleat 3 Ga shown by diamonds from subcontinental mantle.Science 333, 434–436.

Sichel S. E., Esperanca S., Motoki A., Maia M., Horan M. F.,Szatmari P., Alves E. C. and Mello S. L. M. (2008) Geophysicaland geochemical evidence for cold upper mantle beneath theEquatorial Atlantic Ocean. Rev. Bras. Geofıs. 26, 69–86.

Smoliar M. I., Walker R. J. and Morgan J. W. (1996) Re–Os agesof Group IIA, IIIA, IVA, and IVB iron meteorites. Science 271,1099–1102.

Snow J. E. and Reisberg L. (1995) Os isotopic systematics of theMORB mantle: results from altered abyssal peridotites. EarthPlanet. Sci. Lett. 133, 411–421.

Standish J., Hart S. R., Blusztajn J., Dick H. J. B. and Lee K. L.(2002) Abyssal peridotite osmium isotopic composition fromCr-spinel. Geochem. Geophys. Geosyst. 3. http://dx.doi.org/10.1029/2001GC000161.

Staudacher T. and Allegre C. (1982) Terrestrial xenology. EarthPlanet. Sci. Lett. 60, 389–406.

Staudacher T. and Allegre C. (1988) Recycling of oceanic crust andsediments: the noble gas subduction barrier. Earth Planet. Sci.

Lett. 89, 173–183.Stein M. and Hofmann A. W. (1994) Mantle plumes and episodic

crustal growth. Nature 372, 63–68.Taylor B. (2006) The single largest oceanic plateau: Ontong Java-

Manihiki-Hikurangi. Earth Planet. Sci. Lett. 241, 372–380.Tejada M. L. G., Suzukia K., Hanyu T., Mahoney J. J., Ishikawad

A., Tatsumi Y., Chang Q. and Nakai S. (2013) Cryptic lowercrustal signature in the source of the Ontong Java Plateaurevealed by Os and Hf isotopes. Earth Planet. Sci. Lett. 377–

378, 84–96.Touboul M., Puchtel I. S. and Walker R. J. (2012) 182W evidence

for long-term preservation of early mantle differentiationproducts. Science 335, 1065–1069.

Tucker J. M., Mukhopadhyay S. and Schilling J.-G. (2012) Theheavy noble gas composition of the depleted MORB mantle(DMM) and its implications for the preservation of hetero-geneities in the mantle. Earth Planet. Sci. Lett. 355–356, 244–254.

Van Kranendonk M. (2010) Two types of Archean continentalcrust: plume and plate tectonics on early Earth. Am. J. Sci. 310,1187–1209.

Vervoort J. D. and Blichert-Toft J. (1999) Evolution of thedepleted mantle: Hf isotope evidence from juvenile rocksthrough time. Geochim. Cosmochim. Acta 63, 533–556.

Vidal P., Chauvel C. and Brousse R. (1984) Large mantleheterogeneity beneath French Polynesia. Nature 307, 536–538.

Walker R. J. (1988) Low-blank chemical separation of rheniumand osmium from gram quantities of silicate rock for measure-ment by resonance ionization mass spectrometry. Anal. Chem.

60, 1231–1234.Walker R. J., Carlson R. W., Shirey S. B. and Boyd F. R. (1989)

Os, Sr, Nd, and Pb isotope systematics of southern Africanperidotite xenoliths: implications for the chemical evolution ofsubcontinental mantle. Geochim. Cosmochim. Acta 53, 1583–1595.

Wessel P. and Kroenke L. W. (2008) Pacific absolute plate motionsince 145 Ma: an assessment of the fixed hot spot hypothesis. J.Geophys. Res. 113, B06101. http://dx.doi.org/10.1029/2007JB005499.

White W. M. (2015) Isotopes, DUPAL, LLSVPs, and Anekan-tavada. Chem. Geol. 419, 10–28.

Widom E. and Shirey S. B. (1996) Os isotope systematics in theAzores: implications for mantle plume sources. Earth Planet.

Sci. Lett. 142, 45l–465.






































http://dx.doi.org/10.1029/2003GC000597

http://dx.doi.org/10.1029/2003GC000597












































http://dx.doi.org/10.1029/2001GC000161

http://dx.doi.org/10.1029/2001GC000161









































http://dx.doi.org/10.1029/2007JB005499

http://dx.doi.org/10.1029/2007JB005499








Widom E., Hoernle K. A. and Shirey S. B. (1999) Os isotopesystematics in the Canary Islands and Madeira: lithospherecontamination and mantle plume signatures. J. Petrol. 40, 279–296.

Workman R. K., Hart S. R., Jackson M. G., Regelous M., FarleyK. A., Blusztajn J., Kurz M. and Staudigel H. (2004) Recycledmetasomatized lithosphere as the origin of the enriched mantleII (EM2) endmember: evidence from the Samoan volcanicchain. Geochem. Geophys. Geosyst. 5. http://dx.doi.org/10.1029/2003GC000623 Q04008.


Wright E. (1987) Mineralogical studies of Samoan ultramaficxenoliths: implications for upper mantle processes. In Sea-

mounts, Islands, and Atolls (ed. B. H. Keating et al.). AmericanGeophysical Union, Washington, DC, pp. 221–233.

Wright E. and White W. M. (1986) The origin of Samoa: newevidence from Sr, Nd and Pb isotopes. Earth Planet. Sci. Lett.

81, 151–162.

Associate editor: Shichun Huang






http://dx.doi.org/10.1029/2003GC000623

http://dx.doi.org/10.1029/2003GC000623









Peridotite xenoliths from the Polynesian Austral and Samoa ...€¦ · Peridotite xenoliths from...

Documents

Transcript of Peridotite xenoliths from the Polynesian Austral and Samoa ...€¦ · Peridotite xenoliths from...