Local and regional variation of MORB parent magmas: evidence from melt inclusions from the Endeavour Segment of the Juan de Fuca Ridge

Abstract

The development of petrogenetic models of igneous processes in the mantle is dependent on a detailed knowledge of the diversity of magmas produced in the melting regime. These primary magmas, however, undergo significant mixing and fractionation during transport to the surface, destroying much of the evidence of their primary diversity. To circumvent this problem and to determine the diversity of melts produced in the mantle, we used melt inclusions hosted in primitive plagioclase phenocrysts from eight mid-ocean ridge basalts from the axial and West Valleys of the Endeavour Segment, Juan de Fuca Ridge. This area was selected for study because of the demonstrated close association of enriched (E-MORB) lavas and incompatible element enriched depleted (N-MORB) lavas. Rehomogenized melt inclusions from E-MORB, T-MORB, and N-MORB lavas have been analyzed by electron and ion microprobe for major and trace elements. The depleted and enriched lavas, as well as their melt inclusions, have very similar compatible element concentrations (major elements, Sr, Ni and Cr). Inclusion compositions are more primitive than, yet collinear with, the host lava suites. In contrast, the minor and trace element characteristics of melt inclusions from depleted and enriched lavas are different both in range and absolute concentration. N-MORB lavas contain both depleted and enriched melt inclusions, and therefore exhibit the largest compositional range (K2O: 0.01 to 0.4 oxide wt%, P2O5: <0.01 to 0.2 oxide wt%, LaN: 7 to 35, YbN: 1 to 13, and Ti/Zr: <100 to 1300). E-MORB lavas contain only enriched inclusions, and are therefore relatively homogeneous (K2O: 0.32 to 0.9 oxide wt %, P2O5: 0.02 to 0.35 oxide wt%, LaN: 11 to 60, YbN: 4 to 21, and Ti/Zr: ∼100). In addition, the most primitive E-32 inclusions are similar in composition to the most enriched inclusions from the depleted hosts. Major element data for melt inclusions from both N-MORB and E-MORB lavas suggest that the magmas lie on a low pressure cotectic, consistent with a petrogenesis including fractional crystallization. However, the minor and trace element compositions in melt inclusions vary independently of the major element composition implying an alternative history. When fractionation-corrected, inclusion compositions correlate with their host glass composition. Hence, the degree of enrichment of the lavas is a function of the composition of aggregated melts, not of processing in the upper mantle or lower crust. Based on this fact, the lava suites are not produced from a single parent magma, but from a suite of primary magmas. The chemistry of the melt inclusions from the enriched lavas is consistent with a derivation from variable percentages of partial melting within the spinel stability field by a process of open system (continuous or critical) melting assuming a depleted lherzolite source veined with clinopyroxenite. The low percentage melts are dominantly enriched melts of the clinopyroxenite. In contrast, the depleted lavas were created by melting of a harzburgite source, possibly fluxed with a fluid enriched in K, Ba and the LREE. Such a source was likely melted up to or past the point at which all of its clinopyroxene was consumed. This set of characteristics is consistent with a scenario by which diverse melts produced at different depths travel through the melting regime to the base of the crust without homogenizing en route. The homogeneous major element characteristics are created in the lower crust by fractional crystallization and reaction with lower crustal gabbros. Therefore, the degree of decoupling between major and trace element characteristics of the melt inclusions (and lavas) is dictated by the reaction rate of the melts with the materials in the conduit walls, as well as the residence times and flux rate, in the upper mantle and lower crust.