Enclaves provide new insights on the dynamics of magma mingling: A case study from Salina Island (Southern Tyrrhenian Sea, Italy)

Three lava flows (hereafter, flows A, B, and C) from Salina Island (Italy) consist of basaltic andesitic enclaves dispersed in a dacitic matrix. Enclaves represent 8–12 vol.% of the erupted magma. The number of enclaves and the surface covered by the enclaves at each outcrop do not vary significantly with the distance from the vent in the flows A and B. These feature reflect the dynamics of magma mingling within the reservoir and not the kinematics of the lava flow. In the flow C, these parameters vary irregularly. The statistical entropy S(t) of the enclaves, which is a measure of their spatial distribution (dispersion), is estimated in outcrops located at different distance from the vent. The Kolmogorov–Sinai entropy rate k, which describes the variations of S(t) with time, is also determined. In the lava flow A, S(t) increases linearly with time t for 0 < t < 0.4; k is 0.04. For t > 0.5, S(t) attains its maximum value and maintains constant with increasing t. In the lava flow B, S(t) linearly increases with t, and k is 0.01. In the lava flow C, there is not correlation between S(t) and t. The comparison between the results from the analysis of the Porri enclaves and those from numerical experiments on the variation of S(t) in chaotic advective mixing systems and from previous experimental models on magma mixing, allow us to draw some conclusions on dynamics of the basaltic andesite–dacite mingling in the magma chamber. Fully chaotic magma mingling systems show three evolution stages. An initial stage, which is unknown because of the disruption of the initial configuration of the interacting magmas, a second stage characterized by a linear increase of the statistical entropy with time, and a third stage, in which the uniformity of the system is reached, and the entropy does not vary with increasing time. A system in which the uniformity is never attained, is characterized by irregular variations of S(t) with time. In the flows A and B, the relations between S(t) and t are consistent with those of a fully chaotic dynamics possibly associated to convection. The basaltic andesite was uniformly distributed in the dacitic host due to the occurrence of convective movements driven by the injection of the basaltic andesite within the dacitic chamber. The mingling system recorded by the lava flow A evolved with a higher rate with respect to that of the flow B. This suggests that chaotic advection (stirring and folding) is more efficient in the magmatic system A than in B. On the contrary, the mingling system C is characterized by a non-uniform distribution of the basaltic andesite within dacite. This reflects the occurrence of a dynamics in which stirring and folding processes do not operate efficiently and are unable to uniformly distribute the dispersed phase within the continuous one. The decrease of k from A to B, and the lack of a measurable k in C, along with the observation that A and B were emitted before C, indicate that the efficiency of advective movements within the Porri magma chamber declined with decreasing time. Mingled magmas characterized by a homogeneous spatial distribution of enclaves or an initially inhomogeneous distribution evolving towards a homogeneous one are indicative of efficient advection processes that may favor magma mixing. Mingled magmas characterized by an inhomogeneous distribution of enclaves suggest low dynamical interaction between the two end-members. Magma mixing is not allowed.