Magma balloons or bombs?

To the Editor — Detailed analysis of volcanic materials1 allows the study of eruptions that have not been witnessed directly. Rotella et al.2 undertake the challenge of interpreting submarine volcaniclastic deposits around Macauley Volcano, part of the Kermadec Arc in the southwest Pacific Ocean. They propose a ‘Tangaroan’ eruption style, based on the textural characteristics of dredged pumice clasts. However, we argue that the analytical methodology provides an inadequate basis for the identification of a new eruption style. The style of an eruption is determined based on a suite of parameters that define the duration, mass eruption rate and degassing behaviour of the eruption3. To define eruptive style, a single event must be isolated in the volcanic stratigraphy. Additionally, to link vesicle textures observed within individual clasts with magma-degassing behaviour, it must be shown either that all the samples represent a single magma type, or that a variety of magma types are associated with the same distinctive vesicle texture. Otherwise, variations in vesicle textures caused by magma degassing cannot be isolated from those resulting from variations in major element content, volatile composition and temperature. The samples used by Rotella et al.2,4 were dredged along seven transects, each hundreds of metres in length. They probably originate from different eruptions, as demonstrated by the large range in chemical composition and isotopic signatures5. Moreover, clast density varies widely and inconsistently with composition5. Because individual eruptive events cannot be distinguished, and the textural and geochemical variations are relatively large, the degassing histories of clasts within the deposit cannot be attributed to a single style, much less underpin a previously unrecognized eruption type. Rotella et al.2 also cite bimodal clastdensity distributions as evidence for a unique eruption style. However, the distributions are obtained by aggregating data from samples collected from multiple transects, thus probably from separate eruption events. Such data stacking produces a composite picture that masks the styles of individual events. To demonstrate the consequences of this procedure, we apply the same treatment to pyroclast-density data from two well-characterized, subaerial explosive sequences. The magmas have homogeneous bulk compositions and each sequence is composed of discrete eruptive units that exhibit a range of eruptive styles (Fig. 1a,b). The bimodal density distributions in our stacked data are similar to those of the Macauley samples2, illustrating the failure of this stacking procedure to recover the degassing history of a nuanced eruptive sequence. Finally, Rotella et al.2 analysed in detail a single ‘gradient clast’, interpreted to represent the contact between the quenched rim and expanded interior of a pumice clast. They suggest the gradient clast is the remnant of a distinct parcel, or bleb of magmatic foam that detached from the volcanic conduit, but we present an alternative interpretation (Supplementary Information). The bubble number densities (BNDs) calculated for this clast span an order of magnitude, similar to values inferred for highly explosive, high mass eruption rate (MER > 106 kg s–1) subaerial events6–8 (Fig. 1c). However, because BNDs depend on water concentration and diffusivity, the availability of nucleation substrates and magma temperature9, the values are too scattered to allow discrimination of MER. Instead, BNDs can be used to assess eruption style through calculation9 of magma decompression rate, dP/dt (Fig. 1d). Calculated dP/dt for the Macauley pumice varies from about 4 to 35 MPa s–1. These ascent rates impose high levels of shear strain10 and should promote Magma balloons or bombs?