Strontium isotope evidence for mantle events in the continental lithosphere R. J. PANKHURST SUMMARY Following current interpretation of isotope data or undersaturated basic volcanics do show from oceanic basalts, the case for inheritance of remarkable co-variance of Rb/Sr and initial Rb/Sr and 8~Sr/86Sr ratios in continental STSr/SeSr ratios, indicating ages far in excess of igneous rocks from long-lived heterogeneities in extrusion. These and other arguments suggest the mantle is reviewed, both from the point of that heterogeneous mantle, both enriched and view of necessary conditions and the evidence of depleted in lithophile trace elements, is incor- published 'mantle isochrons'. Such inheritance porated into the continental lithosphere during seems unlikely for magmas undergoing plagio- major crust-mantle differentiation events. Re- clase fractionation and can hardly be claimed distribution of isotopes and incompatible for rocks where crustal contamination (es- elements in fluid phases or small partial melts pecially selective contamination with trace may be an important mechanism, although elements or 8~Sr alone) is a feasible alternative. others have suggested addition of enriched Nevertheless, some unfractionated transitional material from below the asthenosphere. x. Introduction MANTLE-DERIVED BASIC VOLCANIC ROCKS erupted in continental areas show a wider range of initial s vSr/SsSr ratios than comparable rocks in an oceanic environment (Faure & Powell I972 , ch. IV). This observation, together with an overall inverse relationship between initial 8vSr/SeSr and Sr content (o10. tit. fig. IV.I), has frequently been taken to imply that contamination with radiogenic 8~Sr from crustal rocks through which the magmas have passed is primarily responsible. Indeed, strong evidence in support of the contamination hypothesis has sometimes been inferred from other geochemical data, for example for the Jurassic volcanic province of Tasmania and Antarctica (Faure et al. 1972, 1974) which exhibits some of the highest initial 87Sr/SeSr ratios reported for basic or intermediate igneous rocks--up to o.714 . In general there has been considerable uncertainty over the degree and mechanism of contamination, some authors favouring bulk assimilation of sialic material at depth (e.g. Hedge I966 , Pushkar 1967, Faure et al. I972), others invoking selective contamination by incompatible trace elements (Green & Ringwood i968), or even by 87Sr alone (Heier I964, Pankhurst 1969). Tests for selective contamination are very difficult to devise and this explanation is usually taken as a last resort when bulk assimilation is shown to be impossible. Nevertheless, the majority of continental basic volcanics have initial svSr/SeSr ratios within, or only a little above, the range oftheir oceanic counterparts (o. 7o25 - c. o.7o7). One of the most important conclusions drawn from modern high- precision data is that real variation in the 8~Sr/SeSr ratios for ocean basalts is fundamental and that it must reflect considerable inhomogeneity of the mantle dTl geol. Soc. Lond. vol. x34, 1977, pp. 255-268, 5 figs., x table. Printed in Great Britain. Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/134/2/255/4885586/gsjgs.134.2.0255.pdf by guest on 26 September 2021 256 R. J. Pankhurst with respect to trace element ratios (Gast 1968, Peterman & Hedge 1971, Hofmann & Hart 1975, O'Nions et al. 1976). Recently, much progress has been made in the detailed interpretation of these heterogeneities (Sun & Hanson 1975, Brooks et al. 1976a) and reappraisal is required of the significance of the continental data. 2. Inheritance of isotope systematics from the mantle Gast (1968) first developed the idea that ocean ridge tholeiites were derived from a source region which was depleted in lithophile elements relative to that of alkali basalts, probably as a result of a previous partial melting event (see also Hanson, this vol.). Pb-isotope studies of oceanic basalts have consistently indicated a major differentiation of the mantle (with respect to U/Pb ratios) at about i ooo-i5oo Ma ago (Tatsumoto I966, Oversby 1972 , Sun & Jahn 1975, Sun & Hanson 1975). Similar treatment of Rb-Sr data has, until recently, been prevented by lack of sufficiently precise data and concern over possible fractionation of Rb/Sr ratios at various stages during the production and evolution of erupted magmas. Sr-isotope geologists now seem to have overcome any reticence. Sun & Hanson (I975) reviewed Rb-Sr data for worldwide oceanic alkali basalts, selected on the basis of high Ni contents as a limitation on olivine fractionation, and showed that in an isochron plot these are consistent with major differentiation in their source regions between I ooo and 3000 Ma ago. They proposed that the parent magmas originated from sources in the deep interior of mantle convection cells or in the underlying mesosphere which they considered to have been essentially isolated as closed systems since these times. Brooks et al. (x 976a) extended this conclusion with a sur- vey of all high-precision data for oceanic basalts, They found that, when 8~Sr/S6Sr and Rb/Sr ratios are averaged for each separate island group, the ocean tholeiites also plot convincingly about a mantle isochron of I6OO 4- 200 Ma. Thus the general picture is that the upper part of the sub-oceanic mantle underwent its principal geochemical differentiation into depleted and non-depleted regions, accompanied by isotopic homogenization, some 15oo Ma ago. Since then it has remained broadly speaking a closed system, presumably because of geophysical constraints on mixing. It should be noted, however, that Sr-isotope variations within individual island groups are either negligible, reflecting derivation from a uniform mantle segment (O'Nions & Pankhurst 1975), or else indicate ages some- what younger than the 1600 Ma event, perhaps due to subsequent local equilibra- tion (Brooks et al. 1976a). An overall treatment of world-wide Rb-Sr data for continental volcanics is not justifiable since, unlike the small ocean island groups, there is frequently significant isotopic variation within a single igneous province. In many of these cases a correlation between initial sTSr/86Sr and Rb/Sr has been observed, but before discussing these it is worth recalling the implicit assumptions made if they are considered to have time significance: (i) All rocks plotted must have been derived from a mantle source region which, though possibly heterogeneous at the time of volcanism, was previously once homogeneous with respect to isotopic composition. Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/134/2/255/4885586/gsjgs.134.2.0255.pdf by guest on 26 September 2021 Mantle events in the continental lithosphere 257 No subsequent open system behaviour has occurred on the scale of the segments of the source from which each erupted magma is drawn, possibly of the order of io km 3. Hoffman & Hart (I975) have argued that diffusion and convective mixing are unlikely to be effective in maintaining continuous equilibration over distances of more than i km even at temperatures close to those of magma production. (iii) The s 7Sr/SeSr and Rb/Sr ratios of all magmas must be representative of their individual source segments. The former is usually taken for granted but the latter is not so obvious, since fractionation of Rb/Sr could occur both during partial melting and as a result of magmatic differentiation prior to eruption. Although the solid-liquid partition coefficients of Rb and Sr differ by a factor often or more for common mineral phases, their absolute values for olivine, pyroxene and garnet are so low that separation of large amounts of these will cause only a very small shift in the Rb/Sr ratio of the liquid. Using typical values it may be calculated that, for example, a 5 per cent melt of garnet peridotite undergoing variable crystal frac- tion of up to 9 ° per cent of olivine, orthopyroxene and clinopyroxene would produce a range of Rb/Sr in the evolved liquids which varies only from I. i to 1.2 times that in the source material. Even for such an extreme case, this is not significant in relation to the precision with which Rb/Sr ratios for basalts are normally reported. Only the involve- ment of plagioclase or minor phases with a very strong preference for Rb or Sr could produce large fractionations. The former can be dis- counted for many basalt sequences by the lack of Eu-anomalies in their rare earth patterns. (iv) No contamination occurs to alter the primary Rb-Sr parameters of erupted magmas, except by mixing with other magmas from the same source region. Provided these conditions are met, an isochron plot for the volcanic rocks in an igneous province will yield an age, not for their extrusion, but for the last time at which their overall source region was isotopically homogeneous. This could be a discrete mantle ditrerentiation event, or alternatively the time of cessation of rapid mixing processes. There seems to be a growing concensus that these conditions may be met more often than previously supposed, at least as regards young basaltic volcanics. 3. Application to continental igneous rocks A philosophy similar to that outlined above forms the basis of an extensive survey of the Rb-Sr literature for continental igneous rocks by Brooks et al. (1976b). These authors cited thirty cases of positive correlation between Rb/Sr and initial 87Sr/ SeSr ratios which they considered sufficiently strong to suggest time significance. Interpreted as mantle isochrons these yield ages between 7 ° Ma and 3340 Ma greater than the known ages of crystallization. The more convincing examples Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/134/2/255/4885586/gsjgs.134.2.0255.pdf by guest on 26 September 2021 ~5 8 R. J. Pankhurst have been recalculated using measured present-day ratios since absolute age is liable to be geologically more significant than age difference (Table I).