Along and across arc geochemical variations in NW Central America: Evidence for involvement of lithospheric pyroxenite

The Central American Volcanic Arc (CAVA) has been the subject of intensive research over the past few years, leading to a variety of distinct models for the origin of CAVA lavas with various source components. We present a new model for the NW Central American Volcanic Arc based on a comprehensive new geochemical data set (major and trace element and Sr–Nd–Pb–Hf–O isotope ratios) of mafic volcanic front (VF), behind the volcanic front (BVF) and back-arc (BA) lava and tephra samples from NW Nicaragua, Honduras, El Salvador and Guatemala. Additionally we present data on subducting Cocos Plate sediments (from DSDP Leg 67 Sites 495 and 499) and igneous oceanic crust (from DSDP Leg 67 Site 495), and Guatemalan (Chortis Block) granitic and metamorphic continental basement. We observe systematic variations in trace element and isotopic compositions both along and across the arc. The data require at least three different endmembers for the volcanism in NW Central America. (1) The NW Nicaragua VF lavas require an endmember with very high Ba/(La, Th) and U/Th, relatively radiogenic Sr, Nd and Hf but unradiogenic Pb and low δ18O, reflecting a largely serpentinite-derived fluid/hydrous melt flux from the subducting slab into a depleted N-MORB type of mantle wedge. (2) The Guatemala VF and BVF mafic lavas require an enriched endmember with low Ba/(La, Th), U/Th, high δ18O and radiogenic Sr and Pb but unradiogenic Nd and Hf isotope ratios. Correlations of Hf with both Nd and Pb isotopic compositions are not consistent with this endmember being subducted sediments. Granitic samples from the Chiquimula Plutonic Complex in Guatemala have the appropriate isotopic composition to serve as this endmember, but the large amounts of assimilation required to explain the isotope data are not consistent with the basaltic compositions of the volcanic rocks. In addition, mixing regressions on Nd vs. Hf and the Sr and O isotope plots do not go through the data. Therefore, we propose that this endmember could represent pyroxenites in the lithosphere (mantle and possibly lower crust), derived from parental magmas for the plutonic rocks. (3) The Honduras and Caribbean BA lavas define an isotopically depleted endmember (with unradiogenic Sr but radiogenic Nd, Hf and Pb isotope ratios), having OIB-like major and trace element compositions (e.g. low Ba/(La, Th) and U/Th, high La/Yb). This endmember is possibly derived from melting of young, recycled oceanic crust in the asthenosphere upwelli