Introduction

Juxtaposition of the Samoan island chain north of the Tonga arc-Lau backarc basin provides a fortuitous opportunity, through geophysical and geochemical methods, to study the distribution in space and time of mantle flow patterns, both underlying the basin and to the north in the vicinity of Samoa. The largest, and still volcanically active Samoan islands of Savai’i and Upolu, are now along strike of the axis of the northernmost Tonga arc, separated from the latter by the WNW-trending Tonga Trench (Fig. 1). On the basis of strongly elevated 3He/4He in the basin’s backarc magmas, southward ingress of the mantle plume tapped during formation of Samoa has been proposed1,2,3,4,5,6,7. There is convincing evidence the Samoan chain is formed from the partial melting of a mantle plume8,9,10,11,12. The Samoan islands are remarkable in that all of the five distinctive geochemical components that currently exist in the Earth’s mantle are manifest in the isotopic compositions of the Samoan lavas13. Extensive geochemical studies have revealed a spectrum of geochemically diverse mantle sources must be involved including the archetypal ‘enriched mantle 2’ (EM2) component with the highest 87Sr/86Sr (~0.720) yet discovered in ocean island basalt (OIB)14. Samoan magmatism samples a mixture of mantle sources that range from a ubiquitous low 87Sr/86Sr to high 3He/4He end-member component (Focal Zone–Austral (FOZO-A ref. 15, 16) separately to EM2, but also to enriched mantle 1 (EM1), high-μ (high 238U/204Pb, HIMU) and a depleted mid-ocean ridge basalt (MORB) mantle (DMM;13 see Table 1 for an overview of mantle components).

Figure 1: Overview map of the greater study area.
figure 1

North of the Paleo-Vitiaz Trench is the presumed plume track (orange dashed line) for the Samoan Plume. The Rochambeau Rifts (RR) and parts of the Northwest Lau Spreading Centre (NWLSC) have geochemical signatures akin to oceanic island basalts43, whereas the Central Lau Spreading Centre (CLSC) has ultra-depleted mantle wedge geochemical signatures. A subduction overprint is not observed beyond ca 200 km west of the Tonga-Kermadec arc (blue dashed line). 1: submarine Dugong Volcano; 2: subaerial Niuafo’ou Island. The map is taken from the Global Multi-Resolution Topography (GMRT) synthesis60.

Table 1 Overview of main mantle components.

The Earth’s upper mantle, as sampled by MORBs, is depleted in incompatible elements consequent to the long-term sequestration of the latter in the continental crust17, outgassed in noble gases by melt extraction relative to primitive mantle, and at least locally, re-fertilized by whole-mantle advective processes18. On the other hand, some OIBs, including Samoa, have elevated 3He/4He and other distinct noble gas isotopic systematics compared with MORB19,20,21, requiring a distinctive source reservoir characterized either by relatively low U/He and/or lesser extents of outgassing19,20.

Enriched mantle components (EM1, EM2, HIMU) prevalent in global OIB and characterized by distinctive radiogenic Sr-Pb-Nd-Hf isotope systematics22,23,24,25 converge in multi-isotope space on the high 3He/4He FOZO component equivalent to the ‘primitive He mantle’8 or common (C) component26. Although the 3He/4He of MORB is ~8 × atmospheric, that of FOZO exceeds ~35 × atmospheric21. On the basis of distinction of FOZO and DMM, particularly in terms of noble gases, mantle plumes with OIB as their likely surface expression, have been postulated to originate in the deeper (relatively undegassed) mantle19.

Geochemical mapping of OIB has revealed consistent spatial isotopic distributions in volcanic island chains, first observed for Hawaii27,28, and subsequently for the Samoan, Marquesas, Tristan-Gough and Society islands13,29,30,31,32. A bilateral conduit that inherits its isotope distribution from large-scale heterogeneity in the source of the mantle plume has been invoked to explain these characteristics. Current debate on plume structure is mostly focused on variations relating to distributions between enriched components, whereas the possible role of FOZO and modes of mixing between FOZO and enriched mantle components remains elusive. The remarkable isotopic variations of the Samoan chain13 together with the potential for diversion of the Samoan mantle source(s) southwards into the opening Lau backarc basin permits some further exploration of spatial distribution of plume components.

As the Tonga arc has rotated clockwise, because of an ultra-fast rollback of the subducting Pacific Plate, a number of spreading centres and rift zones have created the Lau Basin and led to the abandonment of the Lau Ridge as a remnant arc. Rapid rotation, resulting in a very high convergence rate with the Pacific Plate (~240 mm a−1), means meridional separation of the northern terminus of the Tonga arc with the primary Samoan hot-spot locus12 (submarine volcano Vailulu’u33) has progressively decreased. Toroidal flow of Pacific mantle around the foundering northern edge of the subducting Pacific Plate and into the widening Lau Basin has been proposed on the basis of He, Ne and Xe isotopic characteristics in the basin’s backarc magmas1,2,4,7,34 and inferred arc-parallel seismic anisotropy by shear wave splitting related to olivine grain alignment in the mantle wedge35,36,37.

A wide range of volcanic rock compositions has been discovered in the Lau Basin. These range from subduction-unaffected ridges and rifts in the northwest (Central Lau Spreading Centre (CLSC); Northwest Lau Spreading Centre (NWLSC); Rochambeau Rifts (RR), Fig. 1) to strongly subduction-overprinted spreading centres (Valu Fa Ridge and Fonualei Rifts) adjacent to the Tonga arc in the Southeast5,38,39,40,41,42,43,44. Superimposed on these geochemical features is the presence of two isotopically distinct mantle domains, identified primarily by Pb isotope compositions in dredged seafloor samples in the north-western and south-eastern Lau Basin, respectively39,45,46. Pacific-type mantle is dominant in the southern and western Lau Basin, and so-called Indian-type-mantle is predominant in the north-west.

In this contribution, we present new high-precision Hf-Nd isotope data for exactly the same samples previously analysed for He and Ne isotopes1,2, and compare these with other published isotopic data for the Lau Basin and the Samoan chain. The combined isotopic data for the northwestern Lau Basin range from a local mantle wedge comprising a DMM source (Indian- or SOPITA-type45) to values generally characteristic of FOZO, and not the EM1, EM2 or HIMU components manifest in surficial expressions of the Samoan Plume. We conclude selective ingress of some part of the Samoan Plume beneath the Lau Basin is occurring.

Results

Sampling and previous analyses

Samples from this study were dredged during a research voyage (SS02/2009; NoLauVE) of Australia’s Marine National Facility, the RV Southern Surveyor47. Aliquots of the same glass samples from the RR, NWLSC, Dugong Volcano (DV) and the CLSC (Fig. 1) have been previously analysed for major and trace element abundances43, He and Ne isotope compositions1,2. The DV is located ~100 km east of the southern end of the RR, 20 km northwest of the subaerial volcano of Niuafo’ou42,48. Separate glass samples from the same dredge hauls have been independently analysed for major and trace elements (including volatiles), Sr-Nd-Pb-Hf isotopes41 and a subset of those for Xe isotopes3.

Hf-Nd isotope compositions of Lau backarc basin samples

The new Hf and Nd isotopic data are presented in Table 2 and shown in Fig. 2. Isotopic data previously obtained in our laboratory using the same methods for the CLSC and tabled in Nebel et al.49 are shown for comparison. Hafnium and Nd isotope compositions for the RR range from DMM-like (εHf=+16.1 and εNd=+9.4) towards less radiogenic values, that is, εHf=+5.8 and εNd=+4.1, indicative of the involvement of time-integrated lower Lu/Hf–Sm/Nd mantle source component(s) than DMM. Samples from the NWLSC have a narrower spread from εHf=+16.3 to +13.8 and εNd=+10.2 to +6.7, respectively. The DV has εHf and εNd similar to the least radiogenic sample from the RR. In Fig. 3, the 143Nd/144Nd versus 3He/4He, a plot previously used to identify primordial components50, determined on the same aliquots for our sample set from northwestern Lau Basin are shown for comparison; the available data from Samoa are also displayed16.

Table 2 Hf-Nd isotope data for northern Lau backarc basin samples.
Figure 2: Hf-Nd isotope co-variations in Northern Lau samples.
figure 2

A well-constrained mixing trend between local mantle wedge compositions (represented by Central Lau Spreading Centre (CLSC) samples61) and a Samoan Focal Zone Austral (FOZO-A) but not FOZO-B (Borealis) component is illustrated for the data presented here. Samples from the Northwest Lau Spreading Centre (NWLSC) and Dugong Volcano (DV) fall on this trend. Other samples from the Lau Basin (including illustrated literature data)41,45,54 fall off this trend but may include older samples that are not related to the current plate tectonic setting, that is, may relate to the Vitiaz subduction zone, or some inter-laboratory inconsistency. Samples from the northernmost Lau Basin at the Rochambeau Rifts (RR) exhibit OIB-type isotope compositions that plot on a mixing trend with Samoan shield lavas55, which is isotopically identical to FOZO-A16. There is no apparent evidence for enriched mantle-2-type (EM2) signatures, that is, archetype for the Samoan plume. The mixing line between local depleted Lau mantle wedge is defined by: (176Hf/177Hf=0.28324, 143Nd/144Nd=0.51313, Nd/Hf=5) and the Samoan FOZO-A mantle (176Hf/177Hf=0.282945, 143Nd/144Nd=0.512838, Nd/Hf=13). The EM2 component is assumed to plot on the lower (that is, crustal) end of the Hf-Nf isotope array for Samoa55 (176Hf/177Hf=0.28277, 143Nd/144Nd=0.5125)

Figure 3: Nd-He isotope composition of Lau Basin rocks in comparison to Samoan shield lavas.
figure 3

The grey field marks primitive mantle (PM) or Focal Zone Borealis (FOZO-B)62. Samoan samples63 are characterized by a lower 143Nd/144Nd, typified as Focal Zone Austral (FOZO-A)16. Mixing proportions between end-members are unconstrained because of unknown concentrations in those end-members, especially for He. End-member compositions estimated here are FOZO-Borealis (FOZO-B)50: 143Nd/144Nd~0.5130, FOZO-A55: 143Nd/144Nd~0.512838 as shown in Fig. 2; both with [Nd]~13 p.p.m., R/RA~35 with [He]~1 μg per c.c.m.; depleted MORB mantle (DMM)64: 143Nd/144Nd~0.5132 with [Nd]~10 p.p.m., R/RA~8 with [He]~0.5 μg per c.c.m.; representing the local mantle; EM214: 143Nd/144Nd~0.5118 with [Nd]~27 p.p.m., R/RA~0.05 with [He]~0.1 μg per c.c.m.; The two individual mixing lines represent different He/Nd (as indicated) with a 20 × lower He concentration of the lower mixing line in FOZO-A. CLSC, Central Lau Spreading Centre; DV, Dugong Volcano; NWLSC, Northwest Lau Spreading Centre; RR, Rochambeau Rifts; the arrow indicates a hypothetical mixing trend with enriched mantle components.

Discussion

To detect invasion of the Samoan Plume into the northern Lau Basin, the composition of the sub-Lau mantle free of ingress has to be determined. Geochemical studies41,43 of latitudinal variations northwards from the CLSC through the NWLSC to the RR have demonstrated significant compositional changes. It should be noted that none of these backarc rifts and spreading centres bear any detectable contributions from the subducted Pacific Plate in terms of arc-like low Nb/U or Ce/Pb (refs 41, 43). We conclude that neither the Vitiaz arc (pre-Lau Basin opening), nor the current plate tectonic setting has modified or is modifying the supra-subducted Pacific Plate mantle wedge currently serving as a magma source below the northwestern Lau Basin. Although substantial overlap exists, the abundances of the light-middle rare earth elements (REEs) in basalt samples increase from south to north in the northwestern Lau Basin. Strikingly, basalts from the CLSC are among the most depleted in terms of REE among the full spectrum of global MORB51,52, and have consistently and markedly light δ57Fe compared with global MORB, likely related to the refractory nature of the sub-CLSC mantle source49. The RR have slightly convex-upward, chondrite-normalized REE abundance patterns, consistent with some degree of partial melting in the presence of residual garnet41,43. The key question that remains therefore is: ‘do the south-to-north variations in the geochemical characteristics of magmas erupted in the northwestern Lau Basin reflect mantle compositional variations independent of the Samoan Plume, or has ingress of some or all of the Plume components from the latter been superimposed on a prevailing CLSC-type mantle?’

To explain the Hf-Nd-He isotope signals in the northern Lau Basin, we make use of the generally coupled behaviour of the Hf-Nd isotope systems in mantle-derived magmas. We emphasize a number of features of the data in Fig. 2. Although there is overlap between the three spreading centres and rifts, the northwestern Lau Basin suites become progressively more radiogenic in terms of Hf-Nd isotopic character from the RR through the NWLSC to the CLSC. Variable degrees of partial melting of a single common source cannot explain this diversity, and the strongly radiogenic character of the CLSC (εNd≤+13) is consistent with a mantle source that has experienced an ancient melt depletion episode(s). A range of mantle sources is required to account for the spread of Hf-Nd isotopic data, adequately explained by mixing between two mantle components. We note further that our data for the RR and the NWLSC suites are well correlated in εHf versus εNd with a curvature that is deviant from the global MORB-OIB array (Fig. 2). Such curvature can be attributed to mixing between two components with differing Hf/Nd (ref. 53) and distinct isotope compositions. Assuming a mantle source mixing process is occurring beneath the Lau Basin, the radiogenic component is plausibly the local depleted wedge-mantle, represented by samples from the CLSC. Progressive decrease in εHf versus εNd requires mixing between such ‘ambient’ Lau mantle and a less-radiogenic Hf-Nd component (that is, εHf~+7; εNd~+4), consistent with FOZO. Although published data for the northwestern Lau Basin include some samples that are slightly less radiogenic in terms of Hf-Nd isotopic character than FOZO, a critical point is there are none that extend very far towards the strongly unradiogenic, archetypal EM2 end-member represented by the Malu portion of the Samoan islands13. We note further that some of the other literature data apply to dredged samples recovered before the era of multi-beam swath mapping and targeting of morphologically young volcanic structures, unlike those of the NoLauVE voyage47. It is possible these samples are significantly older than those recovered during NoLauVE47, and their compositions may reflect different conditions of magma generation.

The comparison in Fig. 3 of our Nd data with 3He/4He for the same samples and published values for Samoa is instructive. First, a value of 143Nd/144Nd=+0.51285 (εNd~4) divides >95% of the samples from the northwestern Lau Basin from Samoa. Only the DV sample has a lower εNd at R/RA<10. Second, although considerable scatter at any given 143Nd/144Nd exists in terms of 3He/4He, both for the northwestern Lau Basin and Samoa, the majority of the former are encompassed by an envelope of mixing trajectories defined by DMM (CLSC) and FOZO end-members, whereas Samoa fans between DMM, EM2 and FOZO. The mixing lines are a function of absolute abundances of Nd and He, 143Nd/144Nd and 3He/4He of the end-members. We note further the high 3He/4He samples are also characterized by Ne (same aliquots) and Xe (samples from ref. 41) isotopic values consistent with a FOZO end-member1,2,3.

Jackson et al.16 present evidence that the high 3He/4He component in Samoan lavas has other isotopic similarities with a group of OIB suites that occur in the southern hemisphere. These are isotopically distinct from high 3He/4He OIB in the northern hemisphere, and stimulated the hypothesis of austral (FOZO-A) and boreal (FOZO-B) components in the mantle. Although the Hf-Nd isotopic data for the Lau Basin extend to the FOZO-A component, the mixing envelopes shown in Fig. 3 for current data do not exclude the possibility of FOZO-B involvement. Published Nd-Pb-Sr isotopic data for the Lau Basin in comparison with Samoa are shown in Supplementary Figs 1 and 2; the great majority are consistent with mixing between a DMM and FOZO component, and do not extend towards the EM2, HIMU and EM1 components seen in the Samoan data arrays.

Decompressional mantle melting during upwelling has previously been proposed to explain OIB-type volcanism in the northern Lau Basin42,54. OIB components are assumed to have lower solidus temperatures at any given pressure than DMM sources, and preferentially melt but mix with and trigger partial melting of the relatively refractory DMM matrix42,54. The two-component mixing invoked here to create the trend observed in Fig. 2 requires the specific plume (OIB) end-member has a very narrow Hf-Nd isotope composition, unlike the total range documented for the Samoan Plume13,55.

On the basis of geographic isotopic provinciality in the subaerial and submerged portions of the Samoan chain, Jackson et al.13 have recently proposed a geometric model for the structure of the Samoan Plume: a sheath of FOZO encloses stringers of DMM, EM2 and HIMU. The location of the EM1 component manifest in mixing arrays with FOZO, characteristic of the ‘rejuvenated’ volcanism on Savai’I, Upolu and Tutuila, remains uncertain.

Our primary conclusion from the sum of all the available trace element and isotopic data for the northwestern Lau Basin is that mixing is occurring in the underlying upper mantle between a local DMM component, similar to that sampled at the CLSC and a FOZO component that diminishes in volumetric proportion southward from the RR through the NWLSC towards the CLSC. The most plausible source of the FOZO component is through advection of a portion of the Samoan Plume southwards through the tear in the Pacific Plate. If so, based on primordial Xe isotope signatures for the northern Lau Basin3, the FOZO-A component of the Samoan mantle plume is also of primordial origin, despite its Nd isotope excursion from a non-chondritic primitive mantle (FOZO-B). There is no evidence that the other mantle components identified in the Samoan Plume are accompanying the FOZO-A component in the Lau backarc basin. These features are consistent with the model of Jackson et al.13, which invokes an external sheath of FOZO hosting stringers of other mantle components, whereby only the southern portion of the external FOZO sheath is advected southwards. Alternatively, it may be the stringers have already been melted out of the rising plume, with only the FOZO fraction remaining to be advected at shallow mantle depths. The slight residual garnet signature in the RR samples suggests some melting of the ingressing plume component is occurring deeper in the northern Lau Basin than to the south.

There are nevertheless some difficulties with our general conclusion. For example, Jenner et al.43 emphasize the high Cu contents of the NWLSC, the markedly low Cu abundances of Samoan lavas and the possibility that three-component mixing between a DMM source (CLSC), a distinctively high-Cu mantle source of the NWLSC and a low-Cu Samoan Plume component is required to account for the abundances of this element in the different spreading centres and rifts of the northwestern Lau Basin. Jenner et al.43 also note despite the lack of a typical subduction zone signature, marked enrichments of hydrothermally mobile As and Tl in the RR, which points to local magma source complexity. The presence of another mantle plume in the region has also been suggested; for example, the post-subduction alkalic volcanism of Fiji became OIB-like following the initial opening of the Lau Basin56. Resolution of a northern (Samoan) versus potential southern (Fijian) source could be resolved with a systematic study of the numerous spreading centres and rifts that lie to the north of Fiji between the North Fiji and Lau basins. Nevertheless, the sum of the available data for the northwestern Lau Basin is most straightforwardly and currently explained by selective ingress of the FOZO component of the Samoan Plume, and variable extents of mixing of this component with an ambient wedge mantle sampled in its purest form along the CLSC.

Methods

Sample dissolution and purification

Twenty-five dredged seafloor samples were analysed for their Hf and Nd isotope compositions. The exact sample powders were used from the RR and NWLSC and a sample from the DV that have previously been analysed for their major and trace elements43 and their He isotope compositions2. Approximately 100 mg of powdered sample material was dissolved in 15 ml Teflon vials in a HNO3-HF acid mixture on a hotplate at ~150 °C for 24 h. Afterwards samples were subsequently dried down and repeatedly re-dissolved and dried in concentrated nitric acid with traces of HCl and HF until a clear sample solution was achieved. Both, Hf and Nd were purified from the same dissolution batch, including Fe for subsequent Fe isotope analyses, and sequentially extracted from the rock matrix using a combination of Biorad MX1 resin (for Fe), Eichrom Ln-Spec and Tru-Spec a resin. Details of the chemical procedures are similar to those reported for Hf (ref. 57) and for Nd (ref. 58).

Hafnium and Neodymium isotope analyses

Hafnium and Nd isotope ratios were analysed using a Thermo-Fisher Neptune plus multi-collector mass spectrometer with inductively coupled plasma source, housed at the ANU. Hafnium was introduced using a mixture of weak nitric and fluoric acids via a Cetac Aridus II introduction system for dry plasma conditions. Neodymium was introduced with weak nitric acid using a cyclone spray chamber and analysed in wet plasma conditions. All samples were normalized to a JMC-475 Hf standard isotope value of 176Hf/177Hf=0.282160 and a La Jolla Nd standard isotope value of 143Nd/144Nd=0.511859 from standard values obtained during the analytical sessions with 176Hf/177Hf=0.282145±9 (JMC-475) and 143Nd/144Nd=0.511829±20 (La Jolla), respectively. Blanks for both elements were constantly low during the course of the study with Hf<10 pg and Nd<30 pg. Samples are reported in the epsilon notation, defined as the deviation from the chondritic uniform/unfractionated reservoir times 10,000 using values by ref. 59.

Additional information

How to cite this article: Nebel, O. and Arculus, R. J. Selective ingress of a Samoan plume component into the northern Lau backarc basin. Nat. Commun. 6:6554 doi: 10.1038/ncomms7554 (2015).