The existence of depleted components in mantle plumes has been used as evidence that plumes and N-MORB share a common source (Anderson, 1994), or that a depleted component is an intrinsic feature of the lower mantle (Kerr et al., 1995). Pb-isotope data from Iceland suggest that the depleted component is not entrained upper (MORB-source) mantle since Icelandic basalt and Atlantic N-MORB plot on separate, sub-parallel arrays (Thirlwall, 1995). Here we show that depleted Icelandic basalt is also chemically distinct from
N-MORB and that the axial part of the Iceland plume cannot contain a significant amount of entrained MORB-source
mantle.
The two depleted components can be distinguished through the unusual behaviour of Nb during the mantle melting processes responsible for the depletion of the upper mantle reservoir. The abundance of most elements in the MORB source can be modelled adequately by mass-balance
calculations in which average continental crust is subtracted from primitive mantle. This procedure, however, fails to account for the low abundance of Nb in the MORB source because both continental crust and N-MORB are depleted in Nb with respect to other incompatible elements (both have higher La/Nb than does primitive mantle, for example). Thus no mixture of crust and MORB source can reproduce the Nb concentration in primitive mantle. The missing Nb is probably stored in subducted oceanic crust, removed from the upper mantle circulation, and ultimately recycled through mantle plumes (Saunders et al., 1988). One effect of this process is that N-MORB are more depleted in Nb, compared with other incompatible elements, than are even the most depleted Icelandic basalts.
Depletion of Nb in N-MORB, compared with Icelandic basalt and primitive mantle, is illustrated in Fig. 1. The Iceland data cover the whole range of Icelandic basalt, from the most enriched off-axis alkaline basalts from Snaefellsjökull to the most depleted picrites from the Reykjanes Peninsula, and include samples collected from the whole of the neovolcanic zone. The Icelandic data form a remarkably linear array on this plot with upper and lower bounds defined by log(Nb/Y) = 1.92 log(Zr/Y)-1.176 (upper bound), and log(Nb/Y) = 1.92 log(Zr/Y)-1.740 (lower bound). Primitive mantle plots within the Iceland array, but all the N-MORB data plot below the lower bound. The diagram is insensitive to low pressure fractional crystallization, and all the variation among Icelandic basalt due to differences in degree and depth of partial melting are contained within the array. Good correlations between indices of chemical depletion and isotope ratios (Hémond et al., 1993; Hardarson and Fitton, 1994) suggest that the Iceland array reflects the compositional range of the mantle source and is not a result of depletion of a homogeneous source through recent melt extraction. The Nb/Y vs. Zr/Y diagram provides a very useful discriminant between basalt samples with MORB and depleted Icelandic mantle sources, and strengthens the
Pb-isotope argument that depleted plume mantle is fundamentally distinct from the depleted upper mantle.
Anderson, D.L., Earth Planet. Sci. Lett. 128, 303 (1994).
Hardarson, B.S. & Fitton, J.G., Mineralog. Mag. 58A, 372 (1994).
Hémond, C. et al., J. Geophys. Res. 98, 15833 (1993).
Hofmann, A.W., Earth Planet. Sci. Lett. 90, 297-314 (1988).
Kerr, A.C. et al., Geology 23, 843 (1995).
Saunders, A.D. et al., J. Petrology, Special Lithosphere Issue, 415 (1988).
Sun, S.-S. & McDonough, W.F., Geol. Soc., London, Spec. Publ. 42, 313 (1989).
Thirlwall, M.F., J. Geol. Soc. 152, 991 (1995).
Fig. 1: Nb/Y and Zr/Y variation in basalt from the neovolcanic zones of Iceland, compared with primitive mantle (+), typical N-MORB, and ocean island basalt. The parallel lines mark the limits of the Iceland array. Data sources: Iceland, BSH and JGF (unpublished); Reykjanes Ridge (squares), R.N. Taylor, M.F. Thirlwall and B.J. Murton (unpublished); Pacific MORB (circles), ADS (unpublished); average
N-MORB (triangles), Hofmann (1988) Sun and McDonough (1989).