Vertically layered mafic cumulates in the root zone of an ocean island volcano (Fuerteventura, Canary Islands); implications for magma differentiation

The study of the internal structure of basaltic oceanic intraplate volcanoes is essential to better understand

how volcanic systems work, as well as to define and locate magmatic processes occurring between magma

generation in the mantle and eruption. In Fuerteventura, Canary Islands, a Miocene dismembered volcano

provides an excellent opportunity to study the magmatic processes which occurred in the volcanic conduit.

It enables us to link the cumulate formation in vertical channels in the root zone of the volcano with

contemporaneous lava flows.

The mafic – ultramafic intrusion, named PX1, consists of a complex network of amalgamated centimetric to

plurimetric feeder dikes. These are mostly constituted of cumulative rocks represented by wehrlites,

clinopyroxenites and gabbros. The genetic link between plutonic lithologies and lavas was established

based on compositional and textural similarities between plutonic clinopyroxene and phenocrysts in lava.

Semi-quantitative modeling shows that fractional crystallization can explain the evolution of the lavas from

basalts to basaltic trachyandesites. This model reproduces the progressive fractionation of mineral

assemblages corresponding to the crystallization sequence deduced from the plutonic rocks. However,

complex core-rim chemical zoning observed in PX1 clinopyroxene suggest more complicated differentiation

processes than simple crystal segregation and accumulation in vertical conduits. Primitive and resorbed

cores suggest that they represent preexisting crystals entrained from deeper levels or crystallized during

early stages of magma evolution. Reverse zoning and external resorption features in clinopyroxene mantles

suggest progressive crystal coarsening by interaction with successive mafic to moderately differentiated

melt pulses. Asymmetry in chemical zoning suggests crystallization in a confined environment. These

observations support the hypothesis of magma differentiation by fractional crystallization. But unlike the

traditional representation which implies a period of magma stagnation in a reservoir, this process could take

place dynamically during vertical transport of magma in feeder conduits.

The plutonic lithological variability is controlled by three main factors: the efficiency of crystal-melt

segregation; the degree of differentiation and modal proportions of the magmas at time of intrusion and the

extraction efficiency of residual melts from the mineral cumulates, distinguishing typically a clinopyroxenite

from a gabbro. It is assumed that cumulate formation in a dike requires the preservation of high

temperatures in the conduit keeping it partially molten until a subsequent dike may cause the residual melt

extraction by compaction. Based on this assumption, a thermal model was developed simulating

incremental pluton growth by random magma channel emplacement in a confined area. The model

reproduces petrological observations of long-lasting crystal-melt interaction by simulating magma flow

inside the channels for a certain amount of time. The addition of a prolonged magma flow (1 to 3 months)

combined with short intervals between dike injections (< 25 years) allows to reproduce the thermal

constraints required for the PX1 cumulate formation, and also for the development of a broad migmatitic

aureole surrounding the intrusion. The timescales predicted by the model are in agreement with data of

historical eruptive frequencies and durations recorded for some oceanic island volcanoes, supposed

comparable to the Miocene volcanoes of Fuerteventura.