Journal of Petrology Advance Access originally published online on February 15, 2007
Journal of Petrology 2007 48(4):693-709; doi:10.1093/petrology/egl078
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Do We Really Need Mantle Components to Define Mantle Composition?
Dipartimento Di Scienze Della Terra, Università Di Pisa, 56126 Pisa, Italy
RECEIVED AUGUST 3, 2006; ACCEPTED DECEMBER 18, 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHOD OF PROJECTION OF...GEOCHEMICAL AND ISOTOPIC...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: THE PROJECTION...APPENDIX B: SOME NUMERICAL...REFERENCES |
|---|
We discuss the concept of components in the Earth's mantle starting from a petrological and geochemical approach, but adopting a new method of projection of geochemical and isotopic data. This allows the compositional variability of magmatic associations to be evaluated in multi-dimensional space, thus simultaneously accounting for a large number of compositional variables. We demonstrate that ocean island basalts (OIB) and mid-ocean ridge basalts (MORB) are derived from a marble-cake mantle, in which different degrees of partial melting of recycled lithosphere, which are heterogeneous in age and composition, contribute to the magma genesis. This view is supported by the variability in the geochemical and isotopic signatures of OIB that are observed on the scale of a single ocean island as well as on that of an ocean, mostly varying between two extreme compositions, that are not strictly related to the commonly accepted mantle components (DMM, EMI, EMII, HIMU). Rather they are a distinctive feature of the mantle source sampled at each ocean island and are strongly dependent on the Pb isotope system. We recommend a change in perspective in studies of MORB–OIB geochemistry from one based on physically distinct mantle components to a model based on the existence of a marble-cake-like upper mantle. Although resembling the statistical upper mantle, this model implies that geochemical homogenization can be attained only within the limits of local mantle composition, so that a world-wide uniform depleted reservoir cannot be sampled by simply extending the volume of the region undergoing partial melting.
KEY WORDS: geochemistry; isotope; mantle; OIB
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHOD OF PROJECTION OF...GEOCHEMICAL AND ISOTOPIC...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: THE PROJECTION...APPENDIX B: SOME NUMERICAL...REFERENCES |
|---|
The isotopic heterogeneity of oceanic basalts [mid-ocean ridge basalts (MORB) and ocean island basalts (OIB)] has been described by many researchers in terms of the interaction of a small number of mantle components [Depleted Mantle (DMM); Enriched Mantle I (EMI); Enriched Mantle II (EMII); High-µ (HIMU); White, 1985
; Zindler & Hart, 1986; Allègre et al., 1987; Hart et al., 1992; Hanan & Graham, 1996; Hofmann, 1997; Saal et al., 2005; Stracke et al., 2005]. Although these were not originally envisaged as distinct and physically accessible mantle portions, there has been some confusion as to what these components actually represent. The concept of a mantle component stems from the geochemical differences between MORB and OIB, respectively considered to originate from a degassed (low 3He, 40Ar), incompatible element-poor portion of the shallow mantle, and from mantle plumes sampling undegassed, lithophile element-enriched deep mantle (Jacobsen & Wasserburg, 1979; O'Nions et al., 1979; Allègre et al., 1980; Hofmann, 1997; Graham et al., 2001; Saal et al., 2002). The geochemical evidence for the existence of distinct, relatively homogeneous, upper and lower mantle reservoirs, and hence a two-layer mantle convection regime, was inferred from mass-balance calculations on heat production caused by radioactive decay and/or the isotopic budget of incompatible elements (Allègre et al., 1996; Blichert-Toft & Albarède, 1997; Hauri & Hart, 1997; Albarède, 1998; Moreira & Allègre, 1998; Davies, 1999; Kellogg et al., 1999; Allègre et al., 2001; Helffrich & Wood, 2001; Turcotte et al., 2001). Nevertheless, this model is hard to reconcile with geophysical evidence for whole-mantle convection (Hart et al., 1992; Grand, 1994; Stein & Hofmann, 1994; van der Hilst et al., 1997; Tackley, 2000). In addition, in the last decade geochemical characterization of the mantle components sampled by MORB and OIB has changed following improvements in analytical techniques for both radiogenic and stable isotope systems (e.g. Blichert-Toft et al., 1997; Luais et al., 1997; Galer, 1999; Maréchal et al., 1999; White et al., 2000; Nishio & Nakai, 2002). The new data have provided growing evidence that oceanic basalts are heterogeneous on different scales and do not represent a specific mixture of the well-known mantle components (DMM, EMI, EMII, HIMU). This has led to the conclusion that the correspondence between mantle components and specific mantle reservoirs is highly speculative. As a consequence, it is now more generally accepted that mantle processes, rather than a limited number of physically distinct components or reservoirs, are responsible for variability in the geochemical signatures of mantle-derived partial melts (e.g. Hauri, 2000; Kellogg et al., 2002).
The conventional view on the origin of basaltic magmas (sensu lato) assigns a major role to variable degrees of partial melting of a peridotite source (McKenzie & O'Nions, 1991). At least at high pressures, and for relatively Ol-rich melts, experimental data on peridotite melting could explain the compositional variability of most near-primary OIB lavas (e.g. Keshav et al., 2004). However, there is increasing recognition of the role played by olivine-poor sources in the genesis of mantle partial melts (e.g. Carlson & Nowell, 2001; Sobolev et al., 2005; Fig. 1). Even if the dominant source of basalts is the convecting asthenopheric mantle, the process of subduction provides an efficient mechanism for the continuous recycling of pyroxenite-veined oceanic lithospheric mantle and of its cap of MORB and its differentiates. The Ca-Tschermak–Olivine–Quartz (CaTs–Ol–Q) diagram of Fig. 1 illustrates the compositions of a large dataset of OIBs along with relevant peridotite and pyroxenite melting relations. Following Hirschmann et al. (2003), it seems likely that less alkaline basic melts (tholeiites, picrites, picrobasalts) are mostly linked to high degrees of partial melting of a rather shallow peridotite source, whereas more alkaline melts (nephelinites, basanites, alkali basalts), although reflecting a deeper origin at lower partial melting degrees, may be also linked to an enriched source. Mantle sources sampled by these alkaline melts may be related to melting of metasomatic pyroxenite veins, as well as to the solid residuum of the extraction of silicic melts from an eclogitic source (Hirschmann et al., 2003). Nevertheless, other researchers have denied any important role for pyroxenitic material in the chemical budget of OIB (Keshav et al., 2004; for a recent review on this topic refer to Lustrino, 2006).
|
The wide variety of mantle mineral assemblages that can account for the genesis of the spectrum of MORB–OIB primary melts suggests that individual basic magmas should inherit an isotope signature reflecting the heterogeneity of the specific mantle portion in which partial melting occurred. Meibom & Anderson (2003
) suggested that the upper mantle is characterized by a statistical distribution of small- to moderate-scale heterogeneities (the Statistical Upper Mantle Assemblage; SUMA) that could account for both MORB and OIB characteristics, without the need for any specific mantle reservoir or thermal convection (see also Rudge et al., 2005). The composition of oceanic basalts would thus reflect the complex history of their sources in which partial melting events and recycling by passive convection of oceanic and continental lithosphere cause heterogeneity, whereas homogenization is provided by the extent of sampling allowed by the volume of the source region and the degree of partial melting (Anderson, 2006). From this perspective, it is unlikely that specific compositions related to distinct mantle reservoirs or components will be identifiable. In this paper we use the following definitions:
compositional parameter, a geochemical variable used to define the composition of samples (concentrations or elemental and isotope ratios);
mantle component, an extreme in geochemical composition existing independently of the chosen set of compositional parameters adopted for its identification, hence it should be identifiable in all the sub-spaces defined by the compositional parameters;
end-member, a fixed geochemical composition defined by a set of n compositional parameters; mixing of a given number of end-members is able to generate all the compositions found in a set of analyses of natural samples;
mantle reservoir, a mantle portion, physically distinct and potentially accessible, that may result from mixing of mantle components.
We discuss the concept of mantle component with respect to the evaluation of some key processes responsible for MORB–OIB heterogeneity. We assess the range of isotope and trace element variability of oceanic basalts and bind them simultaneously to a large number of compositional parameters using a novel multi-dimensional data plotting approach. This systematic and simultaneous evaluation of both trace element and isotopic data provides a more robust insight into the chemical structure of the mantle.
|
METHOD OF PROJECTION OF GEOCHEMICAL DATA |
|---|
|
TOPABSTRACTINTRODUCTIONMETHOD OF PROJECTION OF...GEOCHEMICAL AND ISOTOPIC...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: THE PROJECTION...APPENDIX B: SOME NUMERICAL...REFERENCES |
|---|
An n-dimensional space (Rn) relative to n (n
3) independent variables can be projected into three-dimensional (3D) space (R3) by choosing four points (hereafter referred to as ‘end-members’; Table 1) that play the role of ‘generating’ the R3 space into which the dataset in Rn is to be projected. End-members can assume an arbitrary position in R3 and represent the image of four points in Rn: they must not be linearly dependent (i.e. non co-planar), both in R3 and in Rn, and allow rescaling of the position of all data points in Rn into R3 by projection (Albarède, 1995). For the sake of simplicity, end-members can be constrained to plot at the vertices of a regular tetrahedron (see Appendix A). Relationships in Rn are maintained in R3; thus planar relationships in n-dimensional space are also planar in 3D space. Trace element and isotope ratios can be plotted simultaneously to represent the data. Such data share the characteristic not to be constrained to be closed to a constant sum. Mixing lines are not necessarily segments of straight lines. On the other hand, plots based on major element analyses are such that their sum is a constant. For major element data the processes of mixing or subtraction of a phase(s) with a constant composition are represented by straight line segments (Pearce, 1987; Russel & Stanley, 1990). Performing some simulations, it is easy to check how a distribution of points, generated by adding random amounts of the end-members to an initial composition [e.g. the reference composition (RC); see Appendix B], generates a cluster of data points that projects within the tetrahedron, without any particular symmetry. The prevalent addition of a specific composition generates a trend lying on a mixing line. Distributions generated by random variations of each analytical compositional parameter show a symmetric elongation that is linked to the intrinsic variability of each component in the dataset (see Appendix B).
|
To allow evaluation of this new method of projection, two Excel files (anamorfosis.xls and data.xls) are provided for downloading from http://www.petrology.oxfordjournals.org. The use of these files is straightforward and self-explanatory. Details of the method of projection are given in Appendix A. The Excel files are set for the Italian language by default. Provide in cells A2-E2 the denomination for the graph window in the language your computer is set: it is required for the proper setting of the basic code. It is a requirement to assign the same set of n compositional parameters to describe the geochemical compositions of the four end-members, represented at each tetrahedron vertex (cells Q39–T48 in anamorfosis.xls), as well as to provide the data to process in data.xls. The data file of interest should be formatted exactly as in data.xls (e.g. using the copy/paste function of Windows and overwriting the analyses in the sample file), and its pathname has to be reported in cell N3 of the ‘tetrahedron’ spreadsheet of anamorfosis.xls. The number (n) of the compositional parameters to be plotted can vary between three and ten, and each parameter can be included or excluded from the evaluation of the tetrahedral coordinates (
1, 2, 3, 4) of each point, by simply setting to one or zero the values given in cells U39–U48 in anamorfosis.xls. Macros in anamorfosi.xls allow the program to run all the calculations automatically to process the data for projection, by using the coloured buttons in the ‘tetrahedron’ spreadsheet. The comments in this sheet may be checked to view the defined values and obtain more detailed information. Buttons are pressed to load (‘Load data’ button), then project (‘ANAMORFOSIS’) and finally plot (‘Plot data’) data, as well as to rotate (‘vertical rotation’ and ‘horizontal rotation’) the tetrahedron. The addition or suppression of one or more compositional parameters, as well as the choice of a different end-member, requires recasting of the projection by pressing again the ‘ANAMORFOSIS’ button in anamorfosis.xls. The spreadsheet can be saved with a new name but, in this case, the new name must be updated in cell N1 in sheet ‘tetraedro’. |
GEOCHEMICAL AND ISOTOPIC VARIABILITY OF MORB AND OIB |
|---|
|
TOPABSTRACTINTRODUCTIONMETHOD OF PROJECTION OF...GEOCHEMICAL AND ISOTOPIC...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: THE PROJECTION...APPENDIX B: SOME NUMERICAL...REFERENCES |
|---|
We compared trace element and isotopic (Sr, Nd, Pb) data for oceanic alkali basalts, basanites, nephelinites, picrites and tholeiites, representative of OIB magmatism, and MORB data, from the Pacific, Atlantic and Indian oceans from the GEOROC database (http://georoc.mpch-mainz.gwdg.de/Start.asp). Hf isotope data are too sparse to be processed together with the Sr–Nd–Pb isotope data. We selected about 1500 analyses for which loss on ignition is less than 2· 2 wt % and mg-number is >63. In our selection, the composition of olivine in equilibrium with the bulk-rock ranges between Fo 63 and 85, assuming Kd(Fe/Mg) = 0· 30 (Roeder & Emslie, 1970
). Recent OIB magmatism is comprehensively represented on a worldwide scale (Fig. 2). We adopted the original classification of mantle components based on DMM, EMI, EMII and HIMU, as defined by Hart et al. (1992), by projecting their Sr, Nd and Pb (206Pb/204Pb) isotopic compositions (n = 3) onto the four vertices of a regular tetrahedron (Fig. 2). Once MORB and OIB data, the latter classified on the basis of their mantle component affinity (e.g. Pitcairn-EMI, Society-EMII, etc.), are plotted, more than 90% of the data fall within the tetrahedron, suggesting that the range of compositions could result from the physical mixture of the selected end-members (See Appendix B). OIB mixing trends converge towards a focal zone (e.g. the FOZO, Focal ZOne; Hart et al., 1992) on the DMM–HIMU edge (Fig. 2). Because the tetrahedron can rotate around two of its axes, trends relative to each ocean or archipelago are easily identifiable.
|
For a variety of basalts from MOR (Atlantic, Pacific, Indian) and EMI (Pitcairn), EMII (Society and Samoa) and HIMU (Tubuai, Rurutu and St. Helena) oceanic islands (http://georoc.mpch-mainz.gwdg.de/Start.asp), we calculated the average arithmetic values for the 208Pb/204Pb ratio and several incompatible trace element abundances, characterized by different ratios of charge/ionic radii, and geochemical behaviour in igneous processes. We selected suitable trace element ratios (e.g. Ce/Pb, Nb/U, La/Sm, Rb/Ba, etc.) to discriminate OIB on the basis of source- and/or process-related geochemical characteristics (Table 2). We constrained the geochemical compositions of the four mantle components of Hart et al. (1992
) to n compositional parameters (3 < n 
10), thus using up to six extra trace element and/or isotope ratios. We note that even the addition of only one variable (n = 4) to the three considered by Hart et al. (1992), either elemental or isotopic, significantly affects the distribution of data points. The combination of more than three geochemical parameters causes clustering of points towards the DMM–HIMU–EMI face of the tetrahedron, on a roughly even surface (here hence ‘pseudo-plane’), as highlighted by an opportune rotation of the tetrahedron (e.g. Fig. 3 and supporting file anamorfosis.xls). Because most of points fall outside the tetrahedron, such a distribution seems not to be generated by a physical mixture of the selected end-members (Appendix B, Fig. B1) and clearly recalls the OIB ‘plane’ described by Zindler et al. (1982).
|
|
For 4
n 10, samples from each oceanic island (e.g. Canary Islands, St. Helena, Hawaii, etc.) describe sub-parallel alignments in the pseudo-plane, showing geochemical variability that may cover the entire compositional range observed for the OIB dataset as a whole (e.g. Zindler et al., 1984). On average, such geochemical heterogeneity varies between two extreme compositions that: (1) are not constrained to any geographical location, and (2) never match or even converge to a unique end-member, as each extreme is distinctive of magmas feeding each ocean island or archipelago [e.g. the Canary end-members are not the same as the Iceland end-members or the Hart et al. (1992) end-members; Figs 3 and 4]; this holds even if they share similar geochemical characteristics, marked by clustering in neighbouring positions in R3.
|
Each ocean island may show multiple extreme compositions, especially those characterized by complex petrogenetic history, e.g. note the various alignments described by Hawaii magmas (Gaffney et al., 2005
, and references therein; inset in Fig. 4a).
With the new choice of compositional parameters, samples with geochemical characteristics close to the Hart et al. (1992) mantle components (e.g. St. Helena = HIMU sensu stricto, Society = EMII sensu stricto, etc.) do not cluster close to the vertices of the tetrahedron but form sub-parallel elongated trends in the pseudo-plane, independent of the position of the Hart et al. (1992) end-members (Fig. 3). Significantly, MORB also spread out from the DMM vertex, revealing the intrinsic variability of their mantle source (Fig. 3; e.g. Hofmann, 2003; Salters & Stracke, 2004; Workman & Hart, 2005). The so-called FOZO (Hart et al., 1992) and EAR (Enriched Asthenospheric Reservoir; Hart et al., 1992; Granet et al., 1995; Stracke et al., 2005; Lustrino & Wilson, 2006) do not hold any key position in this new projection (Fig. 3). Consequently, samples with the isotopic compositions of the Hart et al. (1992) components cannot properly describe the compositional variability of OIB and MORB, unless this is restricted to the isotopic ratios of Sr, Nd and Pb (206Pb/204Pb). This picture is not significantly affected by redefinition of the compositions of Hart et al. (1992) end-members; for example, adopting isotope ratios as suggested in the recent literature for DMM, EMI, EMII and HIMU (e.g. Workman et al., 2004; Stracke et al., 2005).
As the tetrahedron is simply a tool with which to ‘visualize’ the R3 space for the projection of points from Rn, it may be dropped, with the focus being moved to the point distributions and clusterings. Thus it is possible to explore a more appropriate choice of end-members to obtain projections that better highlight the trends of interest; for example, allowing the dataset to reveal its planar distribution for n 4 (Table 3; Figs 4 and 5), regardless of arrays’ positions with respect to the tetrahedron. From this perspective, if samples are tagged on the basis of their major element composition (e.g. classified on the TAS diagram; Figs 1 and 4a and b), picrites, along with OIB tholeiites (most of the tholeiite samples are from the Hawaii database) and some alkali basalts, independent of their provenance, mainly cluster to form a well-defined alignment (see inset in Fig. 4a for Hawaii). The alkali basalts, which form part of this trend, have major element characteristics that are similar to those of picrites and fall on the picrite array (P < 2 GPa) in the CaTs–Ol–Q diagram shown in Fig. 1. Nephelinites, basanites and the other alkali basalts spread on OIB arrays that diverge from the picrite–tholeiite trend, pointing to different end-members (Figs 4 and 5). The two different alignments are not an artefact of the choice of adopting both process-related and source-related compositional parameters. Actually, the trends described are evident also if only tracers of source composition are selected as variables (e.g. isotopes, Th/U, Nb/U, etc.; Fig. 4b). This would suggest that the range of tholeiite–picrite vs basanite–nephelinite trends does not depend upon different degrees of partial melting.
|
|
Data for modern oceanic sediments (from Plank & Langmuir, 1998
) plot mainly apart from the OIB trends, scattering far from the pseudo-plane together with some alkali basalt (± picrite) samples [Fig. 4a; see also sediment distribution with respect to the ‘plane’ of Zindler et al. (1982)]. Although incorporation of modern sediments from the sea floor appears to modify the geochemistry of a minority of OIB mafic magmas, most of the OIB magmas thought to be characterized by recycling of sedimentary material in their mantle source (e.g. EMII) still plot on the pseudo-plane.
Also in the projection of Fig. 5, strong correlations between Sr and Nd isotopes and/or 207Pb/204Pb–208Pb/204Pb and 206Pb/204Pb, such as the mantle–crust array or the Northern Hemisphere Reference Line (NHRL; Hart, 1984), do not affect the distribution of points in Figs 4 and 5 and do not define linear trends on the pseudo-plane.
Mixing equations involving trace element ratios in Rn (Langmuir et al., 1978) can be used to calculate mixing curves in n-dimensional space, and the resulting curves can also be projected in R3. Physical mixing between the pairs of distinct end-members of each oceanic island or archipelago can account for its compositional variability (Fig. 5). Consequently, in this diagram OIB lie on a family of sub-parallel curves.
If incompatible trace element ratios alone (n = 5, ... , 10) are adopted, or Pb isotope ratios are not included among the compositional parameters (e.g. selecting Rb/Ba, Th/Yb and La/Nb and/or 87Sr/86Sr and 143Nd/144Nd ratios as in Fig. 6), the pseudo-plane defined by most of the OIB data collapses into a single alignment. The main trend lies on a mixing curve on which the various extreme compositions tend to merge into apparently only two end-members. Consequently, OIB compositional variability seems to be strongly controlled by two main end-members, in turn dependent on Pb isotope ratios. Even if data overlap is evident along this alignment, nephelinites–basanites–most alkali basalts tend to cluster near one of the two extreme compositions, whereas tholeiites–picrites–few alkali basalts are more representative of the composition of the other end-member. The basalts and sediments that in Fig. 4a plot away from the pseudo-plane spread perpendicularly to the trend of the main data array.
|
We emphasize that these conclusions can be verified for any sets of compositional parameters not including Pb isotopes (e.g. Ce/Pb, La/Sm, Nb/U and 87Sr/86Sr, 143Nd/144Nd). It should be noted that this ‘Pb-effect’ appears to be little influenced by the specific choice of the Pb isotope parameter plotted, as such spreading on the pseudo-plane is induced by any kind of Pb isotope ratio included among the compositional parameters. Otherwise, because of the scarcity of Hf isotope data in OIB databases, the effect of these isotopic ratios on the arrangement of data points in the tetrahedron cannot be strictly constrained. The limited Hf isotope data available confirm the similar behaviour of the Nd and Hf isotope systems, making it possible to switch from one compositional parameter to the other, without affecting the multi-dimensional distribution of data.
We made several attempts to verify this observation by replacing the arithmetic average values with extreme (minimum and maximum) values of 208Pb/204Pb and trace element abundances for the Hart et al. (1992) end-members, without noticing any significant deviation from the above conclusion. We further considered the possibility that the use of compositional parameters involving U/Pb and/or Th/Pb ratios might account for spurious correlations, because of the presence of reliant variables, or distort intra-oceanic correlations because of the scarce correspondence between Pb and Sr–Nd isotope ratios. Nevertheless, even if we replace 206Pb/204Pb with 208Pb*/206Pb* (the ratio of the radiogenic additions to the initial terrestrial lead, defined as [(208Pb/204Pb) – (208Pb/204Pb)init]/[(206Pb/204Pb) – (206Pb/204Pb)init]; Tatsumoto et al., 1973; Hofmann, 1997), which linearly correlates with Sr and Nd isotope composition, or 
7/4 Pb with 207Pb/204Pb (Hart, 1984), the distribution of data in multi-dimensional space (n 4) still matches the above-mentioned correlations. We also verified that such relationships are still evident if only the most primitive OIB samples are selected (mg-number >70).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHOD OF PROJECTION OF...GEOCHEMICAL AND ISOTOPIC...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: THE PROJECTION...APPENDIX B: SOME NUMERICAL...REFERENCES |
|---|
The proposed method of plotting multi-parameter datasets from n-dimensional space to R3 allows a comprehensive analysis of the effects of a large number of independent compositional parameters at the same time, once a proper set of end-members at the vertices of the tetrahedron has been defined. This represents a substantial improvement with respect to the adoption of a fixed reference system tied to a maximum of three independent variables (e.g. Hart et al., 1992
) that, as a matter of fact, is a projection in a 3D space that excludes the effect of (geochemical) data that were not taken in consideration. Such a drastic reduction of components to allow 3D or 2D projections may result in fictitious alignments or clusters that disappear when the complexity of the system is taken into account. In particular, for projections in R3 based on four end-members, data fall inside the tetrahedron defined by the end-members only when the compositional parameters adopted to describe MORB and OIB variability are the isotopes of Sr, Nd and Pb (206Pb/204Pb) (e.g. Hart et al., 1992). However, when the compositional parameters are increased to four and include 207Pb/204Pb and/or 208Pb/204Pb ratios, the distribution within the volume collapses into an approximately planar surface. This plane extends outside the tetrahedron (Figs 3–7
). When the number of compositional parameters plotted is progressively increased to 10, this pseudo-plane becomes better defined, with the exception of the presence of a few OIB that seem to be significantly affected by incorporation of modern sediments from the sea floor (Fig. 4a). Conversely, exclusion of any compositional parameter related to Pb isotopes leads to the collapse of the pseudo-plane into a linear array, provided that n > 4 (Figs 6 and 7b).
|
Even if we have explored a large number of compositional parameters, in principle we cannot exclude that we may have missed some key geochemical tracers. Mathematically speaking, if any significant chemical species is responsible for the point spreading, its neglect is equivalent to perform a projection of the dataset onto an R(n–1) space that is perpendicular to the spread in Rn. This implies loss of the dataset variability in the chosen projection (i.e. it occurs when Pb isotopes are excluded).
The planar distribution of data in R3 suggests that OIB compositional variability is characterized by two degrees of freedom. As three points are enough to define a planar distribution in R3, we might be tempted to relate the observed variability to three distinct end-members. Three mantle components have been previously invoked to explain the isotopic variability of OIB, distributed on a ‘plane’ in any 3–5-dimensional isotope space (Zindler et al., 1982). However, the three end-members are not geographically located, as extreme compositions on the pseudo-plane are not associated with any specific archipelago or, more generally, with the ocean of origin; they cannot be defined as ‘mantle components’, as the components necessary to define each magmatic suite are distinct and different, and depend on the choice of the compositional parameters (e.g. including or excluding Pb isotopes). The Hart et al. (1992) end-members do not represent extreme geochemical compositions within the entire MORB–OIB dataset and they are not appropriate to explain the observed variability of OIB sources by physical mixing, even if they can apparently satisfactorily describe their Sr–Nd–Pb isotopic characteristics.
Different pairs of compositional extremes could account for most of the compositional variability recorded in each archipelago (Fig. 7a). If Pb isotopes are not included in the set of compositional parameters, the distribution on the pseudo-plane collapses into a single array, thus the OIB geochemical variability can be related to only two distinct end-members. Three general considerations can help constrain the nature of the extreme components in MORB–OIB, by linking petrological and geochemical evidence:
- picrites–tholeiites and some alkali basalts mostly cluster towards one of the two end-members and are probably characterized by shallow depths of origin and/or large degrees of partial melting of peridotitic, eventually pyroxenitic, sources (Figs 1, 4 and 5);
- the sources of basanites–nephelinites and most alkali basalts show affinities with the other extreme composition and seem to be mostly related to deep mantle sources (Figs 1, 4 and 5);
- The effect of including Pb isotope ratios seems to enhance the heterogeneity of OIB sources (Fig. 7a).
End-members in OIB and MORB mantle sources do not show a ‘fixed’ geochemical composition and are unlikely to reflect physically accessible mantle reservoirs. The complex variety of compositions along the array(s) may result from the various processes that cause mantle heterogeneity, such as depletion by melt extraction or enrichment by metasomatism, or recycling of subducted or delaminated lithospheric segments. If Pb isotopes are not among the compositional parameters, any set of end-members seems to generate a single linear distribution of OIB–MORB data (Fig. 7b). Magmas generated under largely different partial melting degrees mostly lie at the opposite extremes of this alignment, although both nephelinite–basanite–alkali basalt and tholeiite–picrite magmas overlap in composition. Therefore, diverse mantle processes (e.g. enrichment by recycling and depletion by melt extraction) appear to pervasively affect mantle sources independently of the depth at which they are located.
If OIB–MORB mantle sources were generated only from sources plotting at the two extremes, a marked bimodal arrangement should characterize the observed distributions. Similarly, if the source had an intermediate composition, a cluster of data points would be evident. In this case, each ocean island might be expected to record the fingerprint of a mantle component (or mix of mantle components), thus inheriting a well-defined and restricted geochemical composition. Moreover, this scenario would suggest the existence of ‘enriched’ sensu stricto (e.g. basanites–nephelinites–alkali basalts) or ‘depleted’ sensu stricto (e.g. tholeiites–picrites) magmas, thus calling for the presence of distinct mantle reservoirs. Indeed, the continuous distribution of data along the OIB mixing trends testifies to the existence of a wide spectrum of melts in each ocean island. Both the collection of melts from a definite mantle source region and mixing of magmas from distinct mantle sources are able to explain the observed arrays of OIB data. Thus each ocean island reflects the scale of local mantle heterogeneity by showing its typical range of geochemical variability (Fig. 7b). Therefore, the concept of mantle component is strictly related to the mantle portion(s) sampled at any specific ocean island (Figs 4 and 5). Extreme geochemical compositions and the whole range of intermediate compositions may coexist independently of the degree of partial melting (e.g. as commonly observed at Hawai), supporting the hypothesis of a marble-cake mantle. However, we consider that the model of the statistical distribution of mantle heterogeneity (Meibom & Anderson, 2003; Anderson, 2006) should produce geochemical homogenization within the limits of local mantle variability, so that a uniform depleted reservoir cannot be sampled by simply extending the volume of the region undergoing partial melting (Fig. 7b).
The inclusion of Pb isotopic ratios among the compositional parameters plotted, and the consequent spreading of trends in the pseudo-plane, reveals that each oceanic island does not simply differ in terms of the compositional range between each extreme composition (Fig. 7b). Because Nd and Sr isotope systematics, or elemental Pb in trace element ratios, do not induce differences in the data point distribution, we argue that such behaviour may be strictly related to the variation of U/Pb, Th/Pb and Th/U, and associated Pb isotope ratios, with respect to Sm/Nd or Rb/Sr in OIB mantle sources. This is confirmed by the use of 208Pb*/206Pb* and 7/4 as compositional parameters that do not cause the data to collapse into a single array, suggesting that the spreading of the OIB trends on the pseudo-plane is not a geometric effect of the scarce correlation between Sr–Nd and Pb isotopes.
The discriminant power of Pb isotopes clearly seems to indicate a key role for geological time in characterizing OIB–MORB geochemistry. Recycled oceanic crust, with its sediment veneer, is thought to be a common component in the mantle source of OIB, largely responsible for their chemical heterogeneity (Hofmann & White, 1982; Chauvel et al., 1992; Lassiter & Hauri, 1998; Blichert-Toft et al., 1999; Chauvel & Hemond, 2000; Eisele et al., 2002; Workman et al., 2004; Stracke et al., 2005). U/Pb variability and variability in the age of different portions of subducted lithospheric slabs adds a distinctive Pb isotope fingerprint to the recycling process. Recycling of geochemically diverse subducted plates and delaminated lithospheric keels (e.g. Lustrino, 2005), each with unique U/Pb and Th/Pb values and residence time in the mantle, confers distinctive flavours to OIB mantle sources (e.g. Kelley et al., 2005; Stracke et al., 2005). These are reflected by characteristic Pb isotope ratios in OIB from different oceanic islands, and are thus probably responsible for the observed range of sub-parallel trends in the pseudo-plane (Fig. 7). Thus, each ocean island appears to be characterized by its own range of compositional variability between its own geochemical end-members, with respect to recycling of peculiar lithospheric segments occurring in its local mantle source region. Sr and Nd isotopes in OIB do not seem to be such effective tracers of recycling processes, as differences in these isotope compositions in diverse subducted plates are not an effective discriminant (i.e. the fields of terrigenous and pelagic sediments largely overlap in the Sr–Nd isotope space). This might be related to the fact that Rb/Sr and Sm/Nd do not fractionate as strongly as U/Pb and Th/Pb during the recycling process, thus leading to the more homogeneous Sr–Nd isotope compositions of the enrichment agents. On the other hand, Pb isotope variability in OIB has traditionally been the basis on which the enriched mantle components or end-members (HIMU, EMI, EMII) required to describe mantle heterogeneity were defined (e.g. Weaver, 1991; Hofmann, 1997; Lustrino & Dallai, 2003; Stracke et al., 2005), reflecting the age and composition of recycled lithospheric materials in their mantle source.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHOD OF PROJECTION OF...GEOCHEMICAL AND ISOTOPIC...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: THE PROJECTION...APPENDIX B: SOME NUMERICAL...REFERENCES |
|---|
The chemical structure of the mantle can be constrained by the contemporaneous evaluation of the trace element and isotopic compositions of MORB and OIB. The choice of a particular set of geochemical parameters represents a filter through which mantle composition can be visualized, allowing a more independent perspective of the causes and nature of mantle heterogeneity. Multi-dimensional projections allow a broad view of large-scale mantle heterogeneities, accounting for several possible perspectives at once. The apparently complex projection scheme allows a change of coordinates to simplify the description of mantle compositions. This comprehensive evaluation of several compositional parameters reveals that each ocean island has its own local geochemical end-members, thus the so-called mantle components (DMM, EMI, EMII, HIMU; Hart et al., 1992
) are significant only in the Sr, Nd and Pb isotope perspective. The proposed method allows reconstruction of a relatively simple scenario in which mantle heterogeneity results from a range of processes that cause incompatible element enrichment and depletion, where time plays a major role. Different degrees of partial melting of peridotite and recycled eclogite could contribute to OIB petrogenesis. The geochemical signature of the enriching component, often deriving from recycled lithosphere with a distinctive age and Pb isotope signature, is crucial to characterize specific OIB mantle sources. The multi-dimensional evaluation of OIB–MORB mantle sources allows us to move away from a perspective based on physically distinct mantle reservoirs to a model based on a marble-cake structure for the mantle. Although this model is consistent with the SUMA model of Meibom & Anderson (2003), the highest degrees of partial melting of the mantle source of each ocean island can reveal only the distinctive local depleted source composition, precluding the existence of a mantle-wide, common and uniform depleted DMM reservoir. |
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONMETHOD OF PROJECTION OF...GEOCHEMICAL AND ISOTOPIC...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: THE PROJECTION...APPENDIX B: SOME NUMERICAL...REFERENCES |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
APPENDIX A: THE PROJECTION SCHEME |
|---|
|
TOPABSTRACTINTRODUCTIONMETHOD OF PROJECTION OF...GEOCHEMICAL AND ISOTOPIC...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: THE PROJECTION...APPENDIX B: SOME NUMERICAL...REFERENCES |
|---|
The chemical composition of a system can be reported in terms of a set of n variables corresponding to a point in n-dimensional space. For a system in which four end-members, defined by a given set of components, represent a suitable basis (a basis for a vectorial space is a set of vectors that can generate the whole compositional space Rn to be projected into R3) the projection scheme is based on well-known procedures reported in detail in basic textbooks of linear algebra. In the projection scheme an n-tuple of values that defines the composition of system E
(E1 , E2, ... , En) is recast choosing the four end-members: |
|
The orthogonal projection of the point E* in the 3D space through the end-members is obtained minimizing the n-dimensional Euclidean distance from the given point E.
Given that A, B, C and D 
Rn, we compute the vectors
|
|
Set
|
|
Lastly, set
|
|
For each point, we also compute the vector
|
|
1w1 + 2w2 + 3w3, where |
|
E' is the linear combination of w1, w2, w3 nearest E' in Rn. Please note the different apices appended to E and 
definition. As the distance is translation-invariant, it follows that E' = D + 1v1 + 2v2 + 3v3 is the point in the 3D space passing through A, B, C, D nearest D + E' = E.
E' can be rearranged as
|
|
i of E* with respect to the four vertices A, B, C, D: |
|
From 
i = 1 it follows that each i 1, thus the corresponding point lies within the array of the four end-members A, B, C, D and represents a physical mixture of the end-members. If some coefficient i > 1 the constant sum imposes that one or more coefficient has to attain negative values; thus the point has to lie outside the envelope, a condition that suggests a different choice of end-members in order to express the data as physical mixtures of compositions plotted on the tetrahedron vertices.
The projection E' coincides with E if and only if E itself is a linear combination in Rn of the end-members with the sum of coefficients equal to one. Finally, we can map in R3 the four end-members A, B, C, D, by choosing a set of four points T1 , T2 , T3 , T4 R3. The choice T1 = (0, 0, 1); T2 = (2 x 
2/3, 0, –1/3); T3 = (–2/3, 2/3, –1/3); T4 = (–2/3, –2/3, –1/3) corresponds to a regular tetrahedron.
Finally, E can be mapped in R3 onto E** by the relation
|
|
For a generic orientation of the tetrahedron, it is possible to express its vertices, Ti = {xi, yi, zi}, as a function of the angle 
of rotation around the z axis as
|
|
. The rotation coordinates of its vertices are provided by |
|
1, 2, 3, 4} of a generic point T, from the relation |
|
|
|
|
|
Then
|
|
Analogously,
|
|
|
APPENDIX B: SOME NUMERICAL TESTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHOD OF PROJECTION OF...GEOCHEMICAL AND ISOTOPIC...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: THE PROJECTION...APPENDIX B: SOME NUMERICAL...REFERENCES |
|---|
Adopting four arbitrary end-members EM1, EM2, EM3, EM4 (their compositions with respect to the 10 compositional parameters are given in Table B1), we generated a reference composition RC to plot in the centre of the tetrahedron. RC is thus the average composition for each compositional parameter C1–C10 of the four end-members. RC is then manipulated to obtain four distinct datasets.
|
Dataset A is generated by adding random amounts of the end-members to RC, the resulting data point distribution is completely contained within the tetrahedron and does not show any symmetry (Fig. B1). Conversely, datasets B, C and D are generated by adding variable proportions of each compositional parameter C1–C10 to RC, through a random number generator. In set B, the same range of variability affects each compositional parameter and the resulting distribution is symmetric and perpendicular to C1 (the one with the maximum absolute values); moreover, points can plot outside the tetrahedron (Fig. B1). In sets C and D, different ranges of variability are allowed for each compositional parameter, the standard deviation of each one in set D is reduced by a factor of two with respect to set C. Specifically we took care to allow the largest standard deviation for C10 (the one with the minimum absolute value; Table B1).
Figure B1 shows the different symmetries and positions held by the four datasets. The resulting data arrays show a different spread; the wider distribution is attained by dataset C, which is the one characterized by the largest standard deviation (Table B1). Figure B1 also reports the projections of the versors corresponding to some components. Each versor Cj is drawn by linking RC to a composition for which Cj
j = 0 and Cj = 1. Alignments of the data array in sets C and D are controlled by the versor of the component with the largest absolute values. It is evident that the alignment of the data array in sets C and D are controlled by the direction (versor) assumed for the compositional parameter with the largest absolute values.
|
|
ACKNOWLEDGEMENTS |
|---|
We thank D. Longo for assistance during the development of the new method of projection, and P. Rossi for helpful discussion on its implications in linear geometry. Constructive revisions by J.-M. Cebria, M. Lustrino and A. Stracke, and by two anonymous reviewers of an earlier version of this manuscript, helped us to clarify our ideas and are greatly appreciated. Financial support was provided by MIUR (2003–2005, prot. n° 2003041389_003; Poli, G.: ‘Reattori caotiche e frattali: dinamiche non lineari e invarianza di scala nei sistemi magmatici’).
*Corresponding author. E-mail: d.gasperini{at}dst.unipi.it
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHOD OF PROJECTION OF...GEOCHEMICAL AND ISOTOPIC...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: THE PROJECTION...APPENDIX B: SOME NUMERICAL...REFERENCES |
|---|
Albarède F. (1995) Introduction to Geochemical Modeling(Cambridge University Press, Cambridge) pp. 554.
Albarède F. (1998) The growth of continental crust. Tectonophysics 296:1–14.[CrossRef][Web of Science]
Allègre CJ, Brévart O, Dupré B, Minster JF. (1980) Isotopic and chemical effects produced in a continuously differentiating convecting earth mantle. Philosophical Transactions of the Royal Society, London, Series A 297:447–477.
Allègre CJ, Hamelin B, Provost A, Dupré B. (1987) Topology in isotopic multispace and origin of the mantle chemical heterogeneities. Earth and Planetary Science Letters 81:319–337.[CrossRef][Web of Science]
Allègre CJ, Hofmann AW, O'Nions RK. (1996) The argon constraints on mantle structure. Geophysical Research Letters 23:3555–3557.[CrossRef][Web of Science]
Allègre CJ, Manhes G, Lewin E. (2001) Chemical composition of the Earth and the volatility control on planetary genetics. Earth and Planetary Science Letters 185:49–69.[CrossRef][Web of Science]
Anderson DL. (2006) Speculations on the nature and cause of mantle heterogeneity. Tectonophysics 416:7–22.[CrossRef][Web of Science]
Blichert-Toft J and Albarède F. (1997) The Lu–Hf isotope geochemistry of chondrites and the evolution of the mantle–crust system. Earth and Planetary Science Letters 148:243–258.[CrossRef][Web of Science]
Blichert-Toft J, Chauvel C, Albarède F. (1997) Separation of Hf and Lu for high-precision isotope analysis of rock samples by magnetic sector–multiple collector ICP-MS. Contributions to Mineralogy and Petrology 127:248–260.[CrossRef][Web of Science]
Blichert-Toft J, Frey FA, Albarède F. (1999) Hf isotope evidence for pelagic sediments in the source of Hawaiian basalts. Science 285:879–882.[CrossRef][Web of Science][Medline]
Carlson R and Nowell G. (2001) Olivine-poor sources for mantle-derived magmas: Os and Hf isotopic evidence from potassic magmas of the Colorado Plateau. Geochemistry, Geophysics, Geosystems 2:6 doi:10.1029/2000GC000128.
Chauvel C and Hemond C. (2000) Melting of a complete section of recycled oceanic crust: trace element and Pb isotopic evidence in Iceland. Geochemistry, Geophysics, Geosystems 1:2 doi:10.29/1999GC000002.
Chauvel C, Hofmann AW, Vidal P. (1992) HIMU–EM: the French Polynesian connection. Earth and Planetary Science Letters 110:99–119.[CrossRef][Web of Science]
Davies GF. (1999) Geophysically constrained mantle mass flows and the 40Ar budget: a degassed lower mantle? Earth and Planetary Science Letters 166:149–162.[CrossRef][Web of Science]
Eisele J, Sharma M, Galer SJG, Blichert-Toft J, Devey CW, Hofmann AW. (2002) The role of sediment recycling in EM-1 inferred from Os, Pb, Hf, Nd, Sr isotope and trace element systematics of the Pitcairn hotspot. Earth and Planetary Science Letters 196:197–212.[CrossRef][Web of Science]
Falloon TJ, Green DH, O'Neill H St. C, Hibberson WO. (1997) Experimental tests of low degree peridotite partial melt compositions; implications for the nature of anhydrous near-solidus peridotite melts at 1 GPa. Earth and Planetary Science Letters 152:149–162.[CrossRef][Web of Science]
Falloon TJ, Danyushevsky LV, Green DH. (2001) Peridotite melting at 1 GPa; reversal experiments on partial melt compositions produced by peridotite–basalt sandwich experiments. Journal of Petrology 42:2363–2390.
Gaffney AM, Nelson BK, Blichert-Toft J. (2005) Melting in the Hawaiian plume at 1–2 Ma as recorded at Maui Nui: the role of eclogite, peridotite, and source mixing. Geochemistry, Geophysics, Geosystems 6: Q10L11, doi:10.1029/2005GC000927.
Graham DW, Lupton JE, Spera FJ, Christie DM. (2001) Upper-mantle dynamics revealed by helium isotope variation along the southeast Indian Ridge. Nature 409:701–703.[CrossRef][Medline]
Grand SP. (1994) Mantle shear structure beneath the Americas and surrounding oceans. Journal of Geophysical Research 99:11591–11621.
Granet M, Wilson M, Achauer U. (1995) Imaging a mantle plume beneath the French Massif Central. Earth and Planetary Science Letters 136:281–296.[CrossRef][Web of Science]
Hanan B and Graham DW. (1996) Lead and helium isotope evidence from oceanic basalts for a common deep source of mantle plumes. Science 272:991–995.[Web of Science][Medline]
Hart SR. (1984) A large-scale isotope anomaly in the Southern Hemisphere mantle. Nature 309:753–757.[CrossRef]
Hart SR, Hauri EH, Oschmann LA, Whitehead JA. (1992) Mantle plumes and entrainment: isotopic evidence. Science 256:517–520.
Hauri EH. (2000) Mantle components and mantle reservoirs: bridging the disconnect. Goldschmidt Conference 2000. Journal of Conference Abstracts 5:2495.
Hauri EH and Hart SR. (1997) Rhenium abundances and systematics in oceanic basalts. Chemical Geology 139:185–205.[CrossRef][Web of Science]
Helffrich GR and Wood BJ. (2001) The Earth's mantle. Nature 412:501–507.[CrossRef][Medline]
Hirose K and Kushiro I. (1993) Partial melting of dry peridotites at high pressures; determination of compositions of melts segregated from peridotite using aggregates of diamond. Earth and Planetary Science Letters 114:477–489.[CrossRef][Web of Science]
Hirschmann MM, Kogiso T, Baker MB, Stolper EM. (2003) Alkalic magmas generated by partial melting of garnet pyroxenite. Geology 6:481–484.
Hofmann AW. (1997) Mantle geochemistry: the message from oceanic volcanism. Nature 385:219–229.[CrossRef]
Hofmann AW. (2003) Sampling mantle heterogeneity through oceanic basalts: Isotopes and trace elements. In Carlson RW (Ed.). The Mantle and Core. (Treatise on Geochemistry, Vol. 2.)(Elsevier–Pergamon, Oxford) pp. 61–101.
Hofmann AW and White WM. (1982) Mantle plumes from ancient oceanic crust. Earth and Planetary Science Letters 57:421–436.[CrossRef][Web of Science]
Galer SJG. (1999) Optimal double and triple spiking for high precision lead isotopic measurement. Chemical Geology 157:255–274.[CrossRef][Web of Science]
Jacobsen SB and Wasserburg GJ. (1979) The mean age of mantle and crustal reservoir. Journal of Geophysical Research 84:7411–7427.
Kelley KA, Plank T, Farr L, Ludden J, Staudigel H. (2005) Subduction cycling of U, Th and Pb. Earth and Planetary Science Letters 234:369–383.[CrossRef][Web of Science]
Kellogg LK, Hager BH, van der Hilst RD. (1999) Compositional stratification in the deep mantle. Science 283:1881–1884.[CrossRef][Web of Science][Medline]
Kellogg LK, Jacobsen SB, O'Connell RJ. (2002) Modeling the distribution of isotopic ratios in geochemical reservoirs. Earth and Planetary Science Letters 204:183–202.[CrossRef][Web of Science]
Keshav S, Gudfinnsson GH, Sen G, Fei Y. (2004) High-pressure melting experiments on garnet clinopyroxenite and the alkalic to tholeiitic transition in ocean-island basalts. Earth and Planetary Science Letters 223:365–379.[CrossRef][Web of Science]
Langmuir CH, Vocke RD Jr, Hanson GN, Hart SR. (1978) A general mixing equation with applications to Icelandic basalts. Earth and Planetary Science Letters 37:380–392.[CrossRef][Web of Science]
Lassiter JC and Hauri EH. (1998) Osmium-isotope variations in Hawaiian lavas: evidence for recycled oceanic lithosphere in the Hawaiian plume. Earth and Planetary Science Letters 164:483–496.[CrossRef][Web of Science]
Le Maitre RW. (1989) A Classification of Igneous Rocks and Glossary of Terms(Blackwell, Oxford) pp. 253.
Luais B, Télouk P, Albarède F. (1997) High-precision Nd isotopic measurements using plasma-source mass spectrometry. Geochimica et Cosmochimica Acta 61:4847–4854.[CrossRef][Web of Science]
Lustrino M. (2005) How the delamination and detachment of lower crust can influence balsaltic magmatism. Earth-Science Reviews 72:21–38.
Lustrino M. (2006) Comment on ‘High-pressure melting experiments on garnet clinopyroxenite and the alkalic to tholeiitic transition in ocean-island basalts’ by Keshav et al. [Earth Planet. Sci. Lett. 223 (2004) 365–379]. Earth and Planetary Science Letters 241:993–996.[CrossRef][Web of Science]
Lustrino M and Dallai L. (2003) On the origin of EM-I end-member. Neues Jahrbuch für Mineralogie, Abhandlungen 179:85–100.
Lustrino M and Wilson M. (2006) The Circum-Mediterranean Cenozoic Igneous Province. Earth-Science Reviews (in press).
Maréchal C, Télouk P, Albarède F. (1999) Precise analysis of copper and zinc isotopic compositions by plasma-source mass spectrometry. Chemical Geology 156:251–273.[CrossRef][Web of Science]
McKenzie D and O'Nions RK. (1991) Partial melt distributions from inversion of rare earth element concentration. Journal of Petrology 32:1021–1091.
Meibom A and Anderson DL. (2003) The statistical upper mantle assemblage. Earth and Planetary Science Letters 217:123–139.
Moreira M and Allègre CJ. (1998) Helium–neon systematics and the structure of the mantle. Chemical Geology 157:53–59.
Nishio Y and Nakai S. (2002) Accurate and precise lithium isotopic determinations of igneous rock samples using multi-collector inductively coupled plasma mass spectrometry. Analytica Chimica Acta 456:271–281.
O'Hara MJ. (1968) The bearing of phase equilibria studies in synthetic and natural systems on the origin and evolution of basic and ultrabasic rocks. Earth-Science Reviews 4:69–133.
O'Nions RK, Evensen NM, Hamilton PJ. (1979) Geochemical modeling of mantle differentiation and crustal growth. Journal of Geophysical Research 84:6091–6101.
Pearce TH. (1987) The identification and assessment of spurious trends in Pearce-type ratio variation diagram: a discussion of some statistical arguments. Contributions to Mineralogy and Petrology 97:529–553.[CrossRef][Web of Science]
Plank T and Langmuir CH. (1998) The chemical composition of subducting sediment and its consequences for the crust and mantle. Chemical Geology 145:325–394.[CrossRef][Web of Science]
Roeder PL and Emslie RF. (1970) Olivine–liquid equilibrium. Contributions to Mineralogy and Petrology 29:275–289.[CrossRef][Web of Science]
Rudge JF, McKenzie D, Haynes PH. (2005) A theoretical approach to understanding the isotopic heterogeneity of mid-ocean ridge basalt. Geochimica et Cosmochimica Acta 69:3873–3887.[CrossRef][Web of Science]
Russel JK and Stanley CR. (1990) A theoretical basis for the development and use of chemical variation diagrams. Geochimica et Cosmochimica Acta 54:2419–2431.
Saal AE, Hauri EH, Langmuir CH, Perfit MR. (2002) Vapour undersaturation in primitive mid-ocean-ridge basalt and the volatile content of Earth's upper mantle. Nature 419:451–455.[CrossRef][Medline]
Saal AE, Hart SR, Shimizu N, Hauri EH, Lainer GD, Eiler GM. (2005) Pb isotopic variability in melt inclusions from the EMI–EMII–HIMU mantle end-members and the role of the oceanic lithosphere. Earth and Planetary Science Letters 240:605–620.[CrossRef][Web of Science]
Salters V and Stracke A. (2004) Composition of the depleted mantle. Geochemistry, Geophysics, Geosystems 5:5 doi: 10.1029/2003GC000597.
Sobolev AV, Hofmann AW, Sobolev SV, Nikogosian IK. (2005) An olivine-free mantle source of Hawaiian shield basalts. Nature 434:590–597.[CrossRef][Medline]
Stein M and Hofmann AW. (1994) Mantle plumes and episodic crustal growth. Nature 372:63–68.[CrossRef]
Stracke A, Hofmann AW, Hart SR. (2005) FOZO, HIMU, and the rest of the mantle zoo. Geochemistry, Geophysics, Geosystems 6: doi: 10.1029/2004GC000824.
Tackley PJ. (2000) Mantle convection and plate tectonics: toward an integrated physical and chemical theory. Science 288:2002–2007.[CrossRef][Web of Science][Medline]
Tatsumoto M, Knight RJ, Allègre CJ. (1973) Time differences in the formation of meteorites as determined from the ratio of Lead-207 to Lead-206. Science 180:1279–1283.
Turcotte DL, Paul D, White WM. (2001) Thorium–uranium systematics require layered mantle convection. Journal of Geophysical Research 106:4265–4276.[CrossRef]
van der Hilst RD, Widiyantoro S, Engdahl ER. (1997) Evidence for deep mantle circulation from global tomography. Nature 386:578–584.[CrossRef]
Weaver BL. (1991) The origin of ocean island basalt end-member compositions: trace element and isotopic constraints. Earth and Planetary Science Letters 104:381–397.[CrossRef][Web of Science]
White WM. (1985) Sources of oceanic basalts: radiogenic isotope evidence. Geology 13:115–118.
White WM, Albarède F, Télouk P. (2000) High-precision analysis of Pb isotope ratios by multi-collector ICP-MS. Chemical Geology 167:257–270.[CrossRef][Web of Science]
Workman RK and Hart SR. (2005) Major and trace element composition of the depleted MORB mantle (DMM). Earth and Planetary Science Letters 231:53–72.[CrossRef][Web of Science]
Workman RK, Hart SR, Jackson M, Regelous M, Farley KA, Blusztajn J, Kurz M, Staudigel H. (2004) Recycled metasomatized lithosphere as the origin of the Enriched Mantle II (EM2) end-member: evidence from the Samoan volcanic chain. Geochemistry, Geophysics, Geosystems 5: doi: 10.1029/2003GC000623.
Zindler AW and Hart SR. (1986) Chemical geodynamics. Annual Review of Earth and Planetary Sciences 14:493–571.[CrossRef][Web of Science]
Zindler AW, Jagoutz E, Goldstein S. (1982) Nd, Sr and Pb isotopic systematics in a three-component mantle; a new perspective. Nature 298:519–523.[CrossRef]
Zindler AW, Staudigel H, Batiza R. (1984) Isotope and trace element geochemistry of young Pacific seamounts: implications for the scale of upper mantle heterogeneity. Earth and Planetary Science Letters 70:175–195.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  I. Nardini, P. Armienti, S. Rocchi, L. Dallai, and D. Harrison Sr-Nd-Pb-He-O Isotope and Geochemical Constraints on the Genesis of Cenozoic Magmas from the West Antarctic Rift J. Petrology, July 1, 2009; 50(7): 1359 - 1375. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. S. C. Simon, E.-R. Neumann, C. Bonadiman, M. Coltorti, G. Delpech, M. Gregoire, and E. Widom Ultra-refractory Domains in the Oceanic Mantle Lithosphere Sampled as Mantle Xenoliths at Ocean Islands J. Petrology, June 1, 2008; 49(6): 1223 - 1251. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||