Petrological evolution of the crust and mantle

In its simplest form crust‐mantle evolution can be thought of as a study of the processes by which material is lost from the mantle and forms new crust. The overall goal is achievement of mass balance between the average composition of material known to have been added to the crust and the estimated average composition of the entire crust. Attainment of this goal for the Earth has been elusive primarily because the materials available for study have generally been involved in several stages of evolution including fractionation within both the crust and mantle and recycling of material from crust to mantle, thereby complicating the interpretation of available data. Although the overall composition of the continental crust generally has been modelled as andesitic [Taylor and McLennan, 1981a] the present upper continental crust is granodioritic and requires a more mafic lower crust. It has been proposed [Taylor and McLennan, 1981a] that originally the Archean crust had an andesitic composition throughout its volume and that partial melting in the lower crust developed more felsic melts that rose to enrich the upper crust leaving a more mafic residuum of lower crust. This concept has been challenged by Weaver and Tarney [1981a] who argue that the granulites of presumed lower crustal origin do not have the correct compositions for this model. If one considers more recent additions to the crust from the mantle over the past several hundred million years it appears that these latest additions are basaltic with very little, if any, granitic additions. In fact, Barker [1981], in his introduction to a special issue of JGR on granites and rhyolites, states, “To date no one has demonstrated that such magmas may form in the mantle. Indeed, much experimental and geochemical work indicates that such siliceous liquids cannot originate there. Therefore, such magmas must be generated in the crust.” In island arc environments where the average composition of added material normally is thought to be andesitic A. Anderson [1982] has shown that the parent melt is basaltic not andesitic. In addition, Kay [1980] has shown that the K content of the newly added material is less than the average K content of the previously existing crust. In the oceanic crust the additions from the mantle are basaltic and according to O'Hara [1982] probably are derived from very mafic picritic melts as parents. Continental flood basalts which have also added material from mantle to crust also are clearly derived from basaltic parental melts [Basaltic Volcanism Study Project, 1981]. Thus the evidence for the nature of material subtracted from the mantle and added to the crust for the past several hundred million, or perhaps a billion, years points to very mafic melts. This contrasts significantly with the overall andesitic composition of continental crust. It implies, furthermore, that if the overall Archean crustal composition was andesitic as estimated by Taylor and McLennan [1981a] then additions to the crust have become more mafic with time: a concept in direct conflict with most models in which secular trends are from mafic to felsic with time [Veizer and Jansen, 1979 and McLennan et al., 198O]. A question which is beginning to appear either explicitly or implicitly is whether crustal volcanic rocks or the crust as a whole really represents the parent melt that formed in the mantle [A. Anderson, 1982; O'Hara, 1982 and Nisbet and Walker, 1982].