Multiple reservoirs and different magma interaction processes in the feeding system of the 1650 CE eruption of Kolumbo submarine volcano, Greece

Abstract The Kolumbo submarine volcano, located NE of Santorini (Greece), represents the major source of volcanic hazard in the region, with a historical explosive eruption in 1650 CE that caused extensive damage and multiple deaths. To understand its behaviour and related hazard, we conducted a micro-analytical petrological study on the 1650 CE juvenile products to reconstruct the feeding system architecture and magma dynamics. The samples collected are heterogenous and consist of variably vesiculated white pumices , grey banded/convoluted pumices and dense juveniles , with prevalent homogeneous rhyolitic composition. Mafic enclaves are also present, mainly in the grey and denser lithotypes, ranging from basalts to dacite and showing different Sr-Nd-Pb isotopic compositions than rhyolites. Distinct mineral parageneses characterise the enclaves and host rhyolites with mainly plagioclase (An15-25), biotite and orthopyroxene in the latter and plagioclase (mostly An80-95), amphibole, clinopyroxene and olivine in the former. Among rhyolites two distinct groups (Group-A and Group-B), representative of two spatially close magma reservoirs feeding the eruption, are recognised. They differ for crystal contents and compositions and for enclave type and abundance. Group-A rhyolites show evidence of late mingling with more mafic magmas, whereas Group-B magmas were more significantly affected by early mixing with rhyodacitic melts. A deeper (~ 7–13 km), mafic system of multiple magma bodies sequentially fed the shallower (~ 3–4.5 km) rhyolitic reservoirs prior to (and during) the 1650 CE eruption and, in turn, repeatedly received new inputs of more primitive magmas, with enclave formation already in the deeper system. Our proposed dynamics and architecture of the 1650 CE plumbing system well reflect those hypothesized by geophysical imaging for the present-day volcano, also providing further, more direct evidence, with significant implications for volcanic risk assessment.