Authors' reply

Shea et al. raise three issues pertaining to our work1. First, they argue that our pyroclasts were potentially from different eruptions or magma types with different degassing histories. However, we do not require the Tangaroan pyroclasts to be from a single eruption; indeed, we propose that this style can apply to magmas of diverse compositions for eruptions at submarine volcanoes worldwide. At Macauley Volcano, glass chemistries for Tangaroan dredged pyroclasts are dacitic1 and the clasts lack microlites, indicating a common history without significant degassing2. Furthermore, we do not claim that all dredged pyroclasts are Tangaroan in origin3. Some high-density microlite-bearing clasts, for example, have contrasting textures interpreted to reflect dome-forming eruptions3.

Second, Shea et al. argue that our stacked density data provide a misleading representation of the density distributions for individual eruption events. However, as discussed in ref. 3, irrespective of whether the data are derived from individual stratigraphic levels, single or multiple eruption sequences or dredge hauls, the pyroclast density characteristics from the four volcanoes we have studied are consistent within and distinctive between the volcanoes and eruptive settings. We chose only one representative Tangaroan clast for detailed discussion, but descriptions and analyses of more clasts are presented elsewhere4. The stacking of density data presented in Fig. 1 from Shea et al. is misleading. The Taupo eruption density bimodality (Fig. 1a) is caused by data from differing eruption styles, recognisable from textural characteristics. The high-density mode represents microlite-rich, degassed clasts from phreatoplinian (Unit 4) and sub-lacustrine effusive (Unit 7) phases, whereas their low-density mode is caused by microlite-poor, highly explosive plinian eruptions5. In contrast, the Tangaroan-style pyroclasts we have studied span both density modes. Individual fragments are linked by the density values and textures across the gradient clasts1.

Third, Shea et al. compare our data to selected data from pyroclasts with differing magma compositions and crystal contents (Fig. 1c). They assert that the Tangaroan discharge rate was equally high and the activity explosive. This comparison is misplaced because explosively erupted magmas with similar compositions to ours show higher bubble number density (BND) values. For example, the Mount St Helens eruption in Washington in 1980 generated clasts with BND values of 8.2 × 108 cm−3 (ref. 6) and the Mount Mazama eruption in Oregon around 7,700 years ago generated clasts with BND values of 6.0 × 109 cm−3 (ref. 7). In contrast to Shea et al., we conclude that BND values from natural pyroclasts are often higher than those obtained through experiments8 or numerical simulations9. The equations9 on which Shea et al. construct their comparative argument are based on a single, homogenous nucleation event that produces bubbles with a narrow size range. Such conditions are more easily replicated in experimental simulations. Natural pyroclasts, however, may result from multiple nucleation events or continuous nucleation before fragmentation (for example, ref. 10 and references therein).

Comparison between the denser, quenched rims of the Tangaroan clasts from Macauley Volcano and subaerially erupted pyroclasts with similar chemistries and crystal contents taken from Raoul Volcano (also part of the Kermadec Arc in the southwest Pacific Ocean) shows that the BND values of the Tangaroan clast rims are significantly lower than the BND values of 2.6 × 109 to 1.9 × 1010 cm−3 measured for the Raoul clasts that were erupted in explosive events4. We therefore conclude that when relevant data are compared on an equal basis, our proposal for the Tangaroan eruption style remains fully justified and open to further application.