Magma evolution revealed by anorthite-rich plagioclase cumulate xenoliths from the Ontong Java Plateau: Insights into LIP magma dynamics and melt evolution

Petrologic studies of igneous rocks often rely on whole-rock analyses alone to deduce source and melting histories. Whole-rock compositions represent a mixture of the processes responsible for the final texture and composition of the rock, but do not emphasize the evolutionary processes that the rock may have experienced. We examine the major, minor, and trace element compositions of plagioclase phenocrysts and plagioclase-rich cumulate xenoliths from a sequence of Kwaimbaita pillow basalt and massive flows recovered from Ocean Drilling Program (ODP) Site 1183 on the expansive ∼122Ma Ontong Java Plateau (OJP) and from subaerial outcrops on the island of Malaita, Solomon Islands. These data are used to reconstruct equilibrium liquid compositions to understand magma dynamics, magma evolution, and the distribution basalt types across the OJP. Results reveal that crystals frequently are out of equilibrium with their host basalt. Two distinct compositional zones are present in plagioclase phenocrysts and xenolith crystals — An80–86 and An65–79. Equilibrium magma compositions derived from each of these zones are distinct. In relative terms, the following distinctions are made: An80–86: Sr, Y, and Eu richer, Ba and LREE depleted, higher Sr /Ti, lower Ba/Sr and La /Y; An65–79: Sr, Y, and Eu poorer, Ba and LREE enriched, lower Sr /Ti, higher Ba/Sr and La /Y. We observed several An65–79 parent magmas with higher La /Y, lower Sr, and higher LREE than the majority of An65–79 zones, which suggests that they grew from highly fractionated magmas. The An80–86 zones of both xenolith crystals and phenocrysts appear to have crystallized relatively early and are attributed to growth in magmas more primitive than parent magmas of An65–79 zones or the host magma. Using MELTS modeling we demonstrate a significant role for H2O is not necessary to form An-rich crystals from known OJP magma types. These types of zones can instead form when relatively primitive OJP magmas ascend and partially crystallize at low pressures. Crystal mush layers are rheological elements of multiply saturated solidification fronts, and conditions for extreme fractionation are met deep with thermally and mechanically insulated mush-solidification fronts. Conditions in a homogenous magma body alone cannot generate the range of magma compositions and textures recorded in OJP plagioclase crystals. We suggest that the OJP magma chamber system consisted of crystal mush dominated regions and liquid dominated regions in a laterally and vertically extensive system of interconnected dikes and sills. Solidification front disruption was common during magma recharge and transport, which freed debris to mix with OJP magmas. The volume of crystalline debris from disrupted solidification fronts was small relative to the overall magma volume, which leads to little change of bulk magma chemistry. This can explain the dominance of the Kwaimbaita basalt type across the OJP. Evidence of solidification front disruption ⁎ Corresponding author. E-mail address: [email protected] (W.S. Kinman). 0377-0273/$ see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jvolgeores.2005.09.024 132 W.S. Kinman, C.R. Neal / Journal of Volcanology and Geothermal Research 154 (2006) 131–157 is best preserved as allochthonous crystals both more and less primitive than bulk magma chemistry, both of which were observed in OJP basalts. Coincident with a zone of neutral buoyancy, the shallow OJP magma chamber system existed at a depth of 0–7km. The OJP magma chamber system was thinner near the plateau margins, which provided less density filtration of magmas. This allowed more diverse magma types to reach the surface along the margins. It is envisaged that as the OJP formed, the zone of neutral buoyancy slowly migrated upward leading to slow yet pervasive assimilation of overlying seawater-altered basalt at the base of the lava pile. © 2006 Elsevier B.V. All rights reserved.