The late quaternary Diego Hernandez Formation, Tenerife: Volcanology of a complex cycle of voluminous explosive phonolitic eruptions

The Diego Hernandez Formation (DHF; 600-ca. 180 ka) represents the products of the most recent complete cycle of phonolitic explosive volcanism, on Tenerife (Canary Islands, Spain). We provide a revised and detailed stratigraphy, new Ar-40/Ar-39 and (U-Th)/He age determinations for major eruptive units, a summary of new chemical data and an overview of the key characteristics of the cycle, including volume estimates, dispersal patterns, eruption styles, phreatomagmatic influences and caldera collapse episodes. The complex stratigraphy of the DHF is divided into 20 named members, each representing a major eruption, as well as numerous unnamed members of limited present-day exposure. The major eruptions are represented by the Fortaleza (370 ka), Roque (347 ka, 3 km(3)), Aldea (319 ka, 3 km(3)), Fasnia (309 ka, 13 km(3)), Poris (268 ka, 3.5 km(3)), Arafo (4 km(3)), Caleta (223 ka, 3.5 km(3)) and Abrigo (between 196 and 171 ka, 20 km(3)) Members. The Aldea, Fasnia and Poris Members consist of highly complex successions of plinian fall, surge and flow deposits and several of the eruptions produced widespread and internally complex ignimbrite sheets. Phreatomagmatism occurred most frequently in the opening phase of the eruptions but also recurred repeatedly throughout many of the sequences. Inferred sources of water include a shallow caldera lake and groundwater, and intermittent phreatomagmatic activity was an important influence on eruption style. Another important factor was conduit and vent instability, which frequently loaded the eruption column with dense lithic debris and occasionally triggered column collapse and ignimbrite formation. Most of the major DHF eruptions were triggered by injection of mafic magma into existing phonolitic magma bodies. Two phonolitic magma types were available for eruption during the lifetime of the DHF, but each was dominant at different times. The results presented here support a caldera collapse rather than a landslide model for the