Roadmap | The review process | Home
   Rodinia crust in OJ?
Recycling of the Rodinia supercontinent in the Cretaceous Ontong Java Plateau?

Akira Ishikawa1,2, Takeshi Kuritani1,3, Akio Makishima1 & Eizo Nakamura1

1The Pheasant Memorial Laboratory for Geochemistry and Cosmochemistry, Institute for Study of the Earth's Interior, Okayama University, Tottori, 682-0193, Japan.

2Arthur Holmes Isotope Geology Laboratory, Department of Earth Sciences, Durham University, Durham, DH1 3LE, UK.

3Graduate School of Science, Tohoku University, Sendai, 980-8578, Japan,,,,

Modified from: Ishikawa, A., T. Kuritani, A. Makishima and E. Nakamura, Ancient recycled crust beneath the Ontong Java Plateau: Isotopic evidence from the garnet clinopyroxenite xenoliths, Malaita, Solomon Islands. Earth Planet. Sci. Lett., 259 134-148, 2007.



A rare quartz-garnet clinopyroxenite xenolith recovered from the island of Malaita in the southwest Pacific on the Ontong Java Plateau has lower 206Pb/204Pb-143Nd/144Nd and higher 87Sr/86Sr-207Pb/204Pb ratios than most oceanic basalts, providing the first conclusive evidence for an ancient crustal origin for pyroxenites within the Pacific convective mantle. Constraints from major and trace-element characteristics, the large extent of Hf-Nd isotopic decoupling, and the good agreement of Pb isotopes with the Stacey-Kramers curve, all indicate that contamination of southern Pacific mantle occurred by the subduction or delamination of Neoproterozoic granulite-like lower crust with an age of 0.5-1 Ga. This crustal recycling could have taken place around the suture of the Rodinia supercontinent, a part of which resurfaced during the mantle upwelling responsible for creating the Cretaceous Ontong Java Plateau.

1. Introduction

The heterogeneity of Earth's mantle, revealed by the geochemistry of oceanic basalts such as ocean island basalt (OIB) and mid-ocean ridge basalt (MORB), has been commonly attributed to recycling of crustal materials (e.g., Hofmann, 1997; Stracke et al., 2003) which are thought to be present as eclogitic/pyroxenitic bodies within the convecting peridotite-dominated mantle (Allègre & Turcotte, 1986). During the last decade, the role of eclogitic/pyroxenitic sources in the formation of mantle-derived magmas has received an increasing amount of attention in a number of detailed investigations (e.g., Hauri, 1996; Hirschmann & Stolper, 1996). However, their nature, origin, recycling timescales and even their existence remain controversial because the determination of the magma sources is not straightforward. Mafic layers in orogenic peridotite massifs and eclogite xenoliths in kimberlites have frequently been taken as an analogue for the aging of old oceanic crust (e.g., Jacob et al., 2005; Pearson & Nowell, 2004). The isotopic compositions of these rocks show extremely large variations compared with those observed in oceanic basalts, reflecting their derivation from ancient subcontinental lithosphere. This suggests that their involvement in basalt source regions in the convecting mantle is questionable and underlines the importance of natural samples of pyroxenite/eclogite derived from known oceanic locations for assessing better the origin of oceanic basalts and heterogeneities in the mantle.

Here we present a study of a suite of garnet pyroxenite xenoliths from Malaita, Solomon Islands, as a convincing example of recycled material from within the Pacific convective mantle. This argument is founded on three main lines of evidence:

  • The Malaitan xenoliths were brought up to the surface by a 34-Ma alnöite magma intruded within the southwestern margin of the 120-Ma Ontong Java Plateau (e.g., Fitton & Godard, 2004; Tejada et al., 2004) in an essentially oceanic setting (Figure 1). This is evidence that the xenoliths are fragments of oceanic lithosphere influenced by the plateau emplacement, and not associated with any known subducting slab or slab-related structures.
  • Thermobarometric analyses of the suites of garnet pyroxenite xenoliths reveal derivation from the bottom of the lithosphere (110-120 km depth; Ishikawa et al., 2004) overlain by the pre-existing Jurassic oceanic lithosphere (ca. 160 Ma; Ishikawa et al., 2005). This indicates that isolation of the garnet pyroxenites from the convective mantle occurred between 160 and 34 Ma, probably associated with the 120-Ma plateau emplacement.
  • The extent of petrochemical variation in the garnet pyroxenites is illustrated by the presence of quartz-garnet clinopyroxenite, which cannot be regarded as a high-pressure cumulate or melt from normal peridotitic mantle. This suggests that the rock originated from normative quartz-rich basaltic material, most likely ancient crust stored in the Pacific convective mantle.

Figure 1: (A) Location map of Ontong Java Plateau (OJP) and its hypothetical hot spot trail (Kroenke et al., 2004). The bathymetric map was reproduced from the GEBCO Digital Atlas published by the British Oceanographic Data Centre. (B) Inferred stratigraphic succession beneath the Ontong Java Plateau based on a thermobarometric study of Malaitan xenoliths (Ishikawa et al., 2004). Click here or on Figure for enlargement.

2. Results

2.1 Introduction

We analyzed Sr-Nd-Hf-Pb isotope compositions and trace-element concentrations of pure clinopyroxene and garnet separates (Pb isotopes are only for clinopyroxene) from one quartz-garnet clinopyroxenite sample (QGC) and four bimineralic garnet clinopyroxenite samples (BGC). The mineral compositions of these rocks reflect their high-temperature equilibration (1300-1350°C) as indicated by an apparent lack of compositional gradient either in major- or trace-element concentrations coherently partitioned between garnet and clinopyroxene. The attainment of mineral equilibrium is strikingly demonstrated by two-point Sm-Nd and Lu-Hf inter-mineral isochron ages, displaying good agreement with the 34-Ma host eruption (Figure 2). These data suggest that isotopic equilibrium within individual xenoliths was preserved at the time of host alnöite eruption as ambient conditions exceeded the blocking temperatures of these isotopic systems. Thus whole-rock compositions reconstructed from the mineral data and modal estimates are meaningful in terms of representing their compositional characteristics.

Figure 2: Sm-Nd (A) and Lu-Hf (B) isotope systematics of Malaita garnet clinopyroxenites. Lines connect the clinopyroxene and garnet pair from the same sample.

2.2 Isotopic variability in garnet clinopyroxenite

In contrast to homogeneous features on the scale of individual xenolith specimens, the reconstructed whole-rock data display significant variations as shown in Sr-Nd-Hf-Pb diagrams (Figure 3). It is clear that the QGC has a distinctive composition relative to the BGC. While the Sr-Nd-Hf-Pb isotopic variations of the BGC are within the MORB-OIB array, the QGC is characterized by remarkably lower 206Pb/204Pb-143Nd/144Nd and higher 87Sr/86Sr-207Pb/204Pb ratios, showing some affinities with EM-1 or DUPAL-type isotopic signatures (e.g., Hofmann, 1997; Stracke et al., 2003). The most striking feature of the QGC can be seen in an εHfNd plot, displaying great deviation above and to the left of the terrestrial array defined by the Hf-Nd isotopic values of a wide variety of crust-mantle samples (Vervoort et al., 1999).

Figure 3: The calculated whole-rock Sr-Nd (A), Hf-Nd (B, present-day εNdHf (C), initial εNdHf) and Pb (D, 207Pb/204Pb-206Pb/204Pb; E, 208Pb/204Pb-206Pb/204Pb) isotope compositions of BGC (circles) and QGC (diamonds) compared with 120 Ma Ontong Java Plateau (OJP) lavas and 34 Ma alnöite. References for the background plots are given in the original paper. Inset in (B) shows an enlargement of the evolutionary path for BGC based on present-day parent-daughter ratios. Bold lines in (C) represent evolutionary paths of Hf-Nd isotopes (0-1 Ga) for QGC, which yield different trajectories before 120 Ma according to fractions of extracted melt (0-100 %). Bold curves in (D) and (E) represent the evolution of Stacey-Kramers Pb. Inset in (D) and (E) shows the enlargement of the data of QGC, which fits well in ca. 0.5 Ga of Stacey-Kramers Pb. Error bars (2σ of triplicate analyses) indicate that the present-day and 120 Ma values are almost identical due to very low present μ and ω values. Note that the large uncertainty of clinopyroxene-melt U, Th and Pb partitioning impedes the assessment of the Pb isotope evolution before 120 Ma. If the 120-Ma melt extraction event resulted in preferential removal of U relative to Pb according to ΔU/ΔPb=0.25-0.3, maximum primary μ=6-7 make the initial Pb composition approach the 1- to 1.5-Ga point of the Stacey-Kramers curve in the appropriate time interval (open stars with dashed lines). However, μ=6-7 is too high relative to those of lower crustal mafic granulites with anhydrous mineral assemblages. Click here or on image to enlarge.

These isotopic differences between BGC and QGC can be ascribed to the different extent of chemical reaction against ambient peridotite in the context of melting of eclogite/pyroxenite bearing mantle. The petrochemical features of the QGC suggests that its precursor was a high-pressure liquidus clinopyroxene produced by melting of basaltic material after extraction of siliceous melt, and thus the influence of ambient peridotite is apparently negligible. The BGC may also represent basaltic residues, but the rocks apparently underwent extensive chemical mass exchange with ambient peridotite through liquid percolation and diffusion in order to reach chemical equilibrium. It should be noted that the most depleted BGC records very similar Sr-Nd-Hf-Pb isotopic compositions to those of 120-Ma Ontong Java Plateau lavas (Tejada et al., 2004), suggesting that the abovementioned reaction process operated in the framework of the plateau generation. This in turn implies that the magmatism resulted in melting of eclogite/pyroxenite-bearing mantle as proposed by several previous workers (Ishikawa et al., 2004; Korenaga, 2005; Tejada et al., 2002). Thus, the isotopic characteristics of the BGC are best explained if the primary enriched signatures of the rocks were largely obscured by the melt-mediated reaction against ambient peridotite with depleted isotopic characteristics.

2.3 Origin of quartz-garnet clinopyroxenite

The isotopic characteristics of QGC reflect a long time-integrated evolution with high Rb/Sr-Lu/Hf and low Sm/Nd-U/Pb ratios without recent interaction with ambient peridotite, inferred from the preservation of the normative quartz-rich composition. When the Hf-Nd data are viewed at 120 Ma, they approach the terrestrial array because of the subchondritic Sm/Nd and superchondritic Lu/Hf ratios (Figure 3C). This shows that present-day parent-daughter ratios predominantly reflect ancient fractionation. Assuming that the 120-Ma magmatism produced the present whole-rock composition by accumulation of residual clinopyroxene after extraction of equilibrated batch melt, the series of isotopic trajectories earlier than 120 Ma can be approximated by varying the fractions of extracted melt, whose trace element concentrations can be assessed by clinopyroxene/melt partitioning data (Johnson, 1998; Pertermann & Hirschmann, 2002; Watson et al., 1987). Interestingly, the data devolve back to the terrestrial array in the upper right quadrant at approximately 1 Ga regardless of extracted melt fraction.

Hf-Nd decoupling in OIB magmas is frequently attributed to the involvement of a pelagic sediment component (e.g., Blichert-Toft et al., 1999; Eisele et al., 2002), but similar isotopic evolution and larger extents of Hf-Nd decoupling among crustal samples have been recognized also in mafic granulite xenoliths from the lower crust beneath southern Africa (Schmitz et al., 2004) and northern Queensland, Australia (Vervoort et al., 2000). Despite the fact that these mafic granulites could be generated through complex multiple melting episodes, they all consistently possess normative quartz-rich basaltic compositions, depletion in HFSE relative to the MREE, and enrichment of LREE relative to the HREE (Figure 4; Rudnick & Taylor, 1987; Schmitz et al., 2004). These characteristics can be matched with potential precursor compositions of QGC estimated by adding equilibrated batch melt. This leads to the suggestion that the QGC protolith formed through a similar lower-crustal differentiation event at about 1 Ga.

Figure 4: Primitive mantle-normalized (McDonough & Sun, 1995) trace-element patterns for Hf-Nd decoupled mafic granulite xenoliths (Rudnick & Taylor, 1987; Schmitz et al., 2004). Pink field is defined by calculated whole-rock data of QGC (lower bound) and coexisting melt compositions estimated by using clinopyroxene/melt partitioning data (upper bound), which represents the likely compositional range of the QGC protolith.

Pb isotopic compositions contribute another important piece of information about the nature and differentiation age of QGC. In Pb isotope diagrams (Figure 3D, E), the QGC data plot close to the 0.5-Ga point on the Stacey-Kramers Pb evolution curve, which has been regarded as an evolutionary trace for crustal Pb (Stacey & Kramers, 1975). The most straightforward explanation is that the protolith lost U and Th and/or gained crustal Pb at about 0.5 Ga, and subsequent isotopic evolution was hampered by very low μ (238U/204Pb) and ω (232Th/204Pb) values. This is analogous to the typical Pb isotopic evolution history of Proterozoic mafic granulites, where their Pb isotope compositions may reflect the age of the last tectonothermal or orogenic event that affected the area (e.g., Rudnick & Goldstein, 1990). If this hypothesis is valid, the timing of removal of the lower crust into the underlying mantle can be bound by the minimum isolation age of 0.5 Ga because the subduction of crustal material is an inherent process in active orogenic zones. Even if lower crustal delamination facilitated by gravitational instability is considered, the situation is unchanged because the process is thought to be restricted to the high-geothermal-gradient region accompanied by volcanic activity (e.g., Jull & Kelemen, 2001), which may rejuvenate the Pb isotope composition of delaminating lower crust. From these considerations, we envisage that protolith formation and subsequent subduction or delamination of the QGC occurred during the Neoproterozoic era, at 0.5-1 Ga.

3. Discussion

3.1 Recycling history of the garnet clinopyroxenite

The recycling timescale and material documented by the chemical and isotopic compositions of the QGC are consistent with the presence of Neoproterozoic lower-crust-like material within the Cretaceous Pacific mantle. This raises the question of how such material may have been recycled in the context of the geodynamic history of the Pacific region. Based on available geologic records and paleomagnetic data, it has been suggested that a supercontinent Rodinia formed at ca. 1.1 Ga and broke apart at about 0.7-0.8 Ga creating the Proto-Pacific Ocean (Figure 5A; Hoffman, 1991; Weil et al., 1998). Despite considerable uncertainty about the paleogeographic position and configuration of Rodinia (e.g., Meert & Torsvik, 2003), most of the 1.0 to 1.1-Ga orogenic belts are proposed to mark the sutures between various elements of the supercontinent. Thus, one possible scenario is that Rodinia supercontinent breakup induced detachment and dispersal of the overthickened suture zone between accreted fragments of itself. This is analogous to recent models suggesting that the breakup of Gondwana played a major role in the contamination of Atlantic-Indian oceanic mantle via lower continental crust delamination (Escrig et al., 2004; Hanan et al., 2004; Kamenetsky, 2001). This scenario is only likely if the present southern Pacific is the site of Rodinia fragmentation. It remains a tentative model requiring further testing.

Figure 5: Paleogeographic reconstruction models of Rodinia at 1010 Ma (A; Weil et al., 1998) and Cretaceous oceanic plateaus during the period 90-120 Ma (B; Lawver et al., 2003). Present locations of continents and Pacific superswell region are encompassed by thin and thick dashed lines, respectively. A=Australia; B=Baltica; C=Congo; EA=East Antarctica; G=Greenland; I=India; K=Kalahari; L=Laurentia; RP=Rio de la Plata; S=Siberia; SF=São Francisco; W=West Africa. EM=East Mariana Basin; HR=Hess Rise; M=Manihiki Plateau; MP=Mid-Pacific Mountains; N=Nauru Basin; OJP=Ontong Java Plateau; P=Pigafetta Basin.

An alternative scenario is that the pyroxenites might have been derived from oceanic crustal protoliths comprising basaltic igneous crust and overlying sediment, because such materials would be the volumetrically dominant crustal input. A major drawback of this scenario is that inferred granulite-grade differentiation is apparently incompatible with the low geotherms typical of subduction zones, where elemental fractionation via dehydration can take place. The low-HFSE signature in the QGC may preclude its derivation from a dehydrated slab because it selectively retains HFSE due to their lower mobility in slab-derived fluids and/or strong affinity for a residual Ti-rich phase (e.g., Pearce & Peate, 1995). However, it has been suggested that the higher geothermal gradients in the Precambrian were not conducive to generating the stability field for blueschists even in a subduction zone setting (Grambling, 1981; Maruyama & Liou, 1998). If the hypothesis of secular cooling is valid, Proterozoic subducting slabs with high geothermal gradients might have frequently undergone elemental redistribution by partial anatexis.

Lower continental crust-like pyroxenite signatures therefore do not allow us to rule out strictly their derivation from ancient subducted oceanic crust. Nevertheless, they suggest that long-term recycled eclogite/pyroxenite heterogeneity was introduced in Pacific deep mantle through Neoproterozoic amalgamation/breakup of the Rodinia supercontinent, and that this heterogeneity played a substantial role in the formation of the Ontong Java Plateau as the surface expression of a large melting anomaly in the Cretaceous Pacific mantle.

3.2 Implications for Ontong Java Plateau magmatism

The origin of the Cretaceous oceanic plateaus (Figure 5B) that are widespread in the present-day western Pacific, is the subject of a major debate in terms of whether or not the melting was associated with high-temperature mantle plumes. In this regard, two "endmember scenarios" have been invoked to explain the generation of the largest–the Ontong Java Plateau:

  • High-temperature melting of a peridotite mantle plume (>1500°C TP; Fitton & Godard, 2004), or;
  • Entrainment of eclogite-bearing mantle by passive upwelling (~1300°C TP; Korenaga, 2005)

The presence of recycled eclogite/pyroxenite in the magma source demonstrated here is compatible with the second scenario. However, trace-element and isotopic data support a composite source for the magmas rather than a pure eclogite/pyroxenite source. Given the voluminous emplacement of magma in off-axis mature seafloor, it is likely that mantle hotter than normal was implicated in entrainment of recycled eclogite/pyroxenite. Thus, we envisage generation of the Ontong Java Plateau as due to upwelling of a composite mantle plume, although a quantitative estimate of the TP is clearly required before a decisive conclusion can be reached. For this purpose, geochemical modeling should be revisited, taking into account the contribution of recycled materials combined with reasonable extents of heterogeneity.

Finally, it is interesting to consider whether such a plume initiated from the core-mantle boundary as theoretically predicted. Experimental investigations suggest that sinking basaltic crust, except for the coldest slabs, are likely to be gravitationally trapped at the base of the mantle transition zone (e.g., Litasov et al., 2004). It therefore appears that the granulite-like precursor of the QGC is incompatible with its storage in the lower mantle. Moreover, the notion of a primitive mantle-like source previously invoked to explain the geochemical signatures of the Ontong Java Plateau lavas (Fitton & Godard, 2004; Parkinson et al., 2002; Tejada et al., 2004) must be revaluated because the contribution of recycled eclogite/pyroxenite is critical for this argument. Thus, our results lend no support to the hypothesis of plume initiation at the core-mantle boundary. However, the plume hypothesis may explain the temporal coincidence of surges in the creation of oceanic plateaus with the Cretaceous magnetic superchron (Larson & Olson, 1991). Furthermore, mounting evidence from seismic methods clearly indicates the occurrence of a large-scale low-velocity anomaly in the lower mantle under the superswell region of the south-central Pacific (e.g., Hager et al., 1985; Romanowicz & Gung, 2002), which is spatially related to the birth place of Cretaceous oceanic plateaus (Figure 5B; Courtillot et al., 2003; Larson, 1991). Thus, on-going research for geochemical evidence of lower mantle and core interaction in plume-related materials is critical to obtaining a complete understanding of the origin of the low-velocity anomaly and its role in present-day hotspots and ancient oceanic plateaus.


We thank Gillian Foulger for her invitation and help to present this webpage.

last updated 3rd September, 2007