Halmahera (East Molucca)

The Halmahera SZI event - associated with the subduction of the Molucca Sea plate eastwards beneath the Philippine Sea plate - was previously estimated to have initiated at around 17-15 Ma (Baker & Malaihollo 1996; Hall, 1996), but more recent studies have alternatively suggested a younger age of ~10-7 Ma for SZI (Chandra & Hall, 2016; Hall & Spakman, 2015; Zhang et al. 2017). Given the evidence presented below, SZI must be older than 11 Ma. Considering a ~5 Myr delay (in line with Baker & Malaihollo 1996) we suggest ~15 Ma as our best-estimate for Halmahera SZI, but emphasise that this preference is likely to be controversial and may be revised downward if additional evidence demonstrates a disconnect between the oldest arc volcanism (on Obi Island) and later subduction beneath Halmahera. The type of SZI is also unclear, but could be ascribed to episodic subduction, as there is evidence of subduction along Halmahera from the Mesozoic to the Oligocene. That subduction ceased by the time of a regional plate boundary reorganisation at about 25 Ma, whereafter subduction initiated on the western side of the Molucca Sea plate, along the Sangihe arc (Hall and Smyth, 2008). Hall and Smyth (2008) suggested that Halmahera SZI was caused by locking of the Sorong fault zone along the southern edge of the Molucca Sea. Considering the broader geodynamic picture, Halmahera SZI occurred perhaps within ~20° of the Manus plume (Wu et al. 2016), and approximately above the edge of the Pacific LLSVP (i.e., along the edge of its surface projection, according to its present-day shape).

The older (17-15 Ma) age estimate for SZI is based on the presence of ~11 Ma calc-alkaline volcanic rocks on Obi Island at the southern end of the Halmahera system, which are recognised as the oldest arc-related Neogene volcanic products of the system (Baker & Malaihollo 1996). However, Hall & Spakman (2015) have suggested that the Obi Island volcanic rocks may be due to “to minor subduction within the northern New Guinea strike-slip zone” (a strike-slip bounding the Halmahera system to the south; also called Sorong fault zone), and thus potentially unrelated to SZI of the main Halmahera subduction zone. They further point out that arc-related volcanism on the island of Halmahera does not appear until 8-7 Ma, and therefore argue for a younger (~10 Ma) initiation of the Halmahera subduction zone. While acknowledging that the Obi Island volcanic rocks may be in part due to strike-slip processes, we note that if “minor subduction” began along Obi Island and later propagated northward, our definition of SZI would still recognise that earliest phase of subduction to be the SZI event itself. We further note that Baker & Malaihollo (1996) report the occurrence of 9.5-8.8 Ma acidic volcanic rocks from Pulau Bisa (north of Obi Island), which seem to reinforce the notion of northward propagating arc activity, and the occurrence of ~15 and ~9 Ma diorites on Bacan Island (between Obi and Halmahera islands). The latter intrusive rocks could be arc-related, although no coeval volcanic counterparts have been recognised.

SZI along the Sorong fault zone could have produced the calc-alkaline volcanic rocks found on Obi Island and then subduction (and arc magmatism) may have propagated northward to Pulau Bisa, Bacan and then Halmahera islands by ~10-7 Ma (Baker & Malaihollo 1996). Notably, the Neogene Halmahera subduction zone succeeded an earlier Mesozoic to Eocene subduction zone along Halmahera (Hall et al. 1988; Hall & Smyth, 2008), and so the younger subduction zone may have exploited a pre-existing structure.

In the model of Müller et al. (2016), the Halmahera SZI event occurs at 14 Ma, in conjunction with the Philippine SZI event (the subduction zones are contiguous but of opposite polarity). Halmahera SZI immediately follows the cessation of a pre-existing subduction zone (the ‘East Philippine’ subduction zone) within ~250 km of the Halmahera trench; to the east the trace of these two subduction zones become progressively closer until they merge. Coincident with the SZI event (at 14 Ma), the motion of the Philippine Sea plate changes significantly, as does the neighbouring Caroline plate to the immediate east of the Halmahera trench, whereas the motion of the Australian plate remains unchanged.

Subducted slabs of the Molucca Sea plate are well-imaged by seismic tomography, and resemble an inverted “U-shape” due to the occurrence of both west-dipping subduction beneath the Sangihe arc and east-dipping subduction beneath Halmahera. However, in the case of the seismic anomalies beneath Halmahera, it is not straightforward to differentiate the slab of (younger) Neogene Halmahera subduction from lower-mantle anomalies that likely represent the remnants of earlier (Mesozoic-Eocene) subduction. van der Meer et al. (2018) interpret the Halmahera slab to reach ~760 km depth, whereas other interpretations consider the anomalies below 440 km to be separated from the Halmahera slab and associated with older subduction (Zhang et al. 2017). The former interpretation is more consistent with an older (~17-15 Ma) timing for SZI, whereas the latter is more consistent with a younger (~10-7 Ma) initiation.