Subduction initiation is a hot topic in earth science (Arculus et al., 2015; Gerya et al., 2015; Zhou et al., 2020; Zhong and Li, 2021). The style of subduction initiation has been classified into two kinds: induced and spontaneous (Gurnis et al., 2004; Stern, 2004). According to previous studies, the former is associated with compression and uplift of the overriding plate, whereas the latter is controlled by subsidence of the old dense lithosphere that eventually converted into the subducting plate, characterized by extension and magmatism (Guilmette et al., 2018; Stern and Gerya, 2018; Zhong and Li, 2021). However, in reality, it seems not easy to distinguish these two kinds of subduction initiations.

How to clarify ancient subduction initiations? This question is recently tackled using thermomechanical models and metamorphic soles (Zhou and Wada, 2021). Metamorphic soles are highly deformed high-temperature low-pressure metamorphic rocks of a few to several hundred meters thick found in the overriding forearc lithosphere of supra-subduction zone ophiolite complexes (Guilmette et al., 2018; Zhou and Wada, 2021). The thermomechanical modeling results show that the slab surface temperature reaches 800–900 ℃ at ~1 GPa for spontaneous subduction initiation, assuming that the hot asthenospheric mantle flows into the forearc immediately after the older slab starts to sink. For induced subduction initiation, the modeling assumed no intrusion of the hot asthenosphere due to compression, the temperatures of metamorphic soles are much lower, with the exception of a very young overriding plate of < 5 Ma. Clearly, more work is needed to test these assumptions, first. Nevertheless, according to their modeling results, most of the metamorphic soles reported to date are the results of induced subduction initiations that involved a young overriding plate based on the synthesis modeling results and geological records (Zhou and Wada, 2021). For an example, Guilmette et al. (2018) found that the age of lower metamorphic sole is older than that of upper crust of the Semail ophiolite of Oman, suggesting forced subduction initiation. Unfortunately, metamorphic soles are not identified in the Izu-Bonin-Mariana (IBM) subduction belt. Therefore, the thermodynamic modeling results of Zhou and Wada (2021) cannot give direct constraints on the classification of IBM subduction initiation.

Forearc basalts are the most abundant and oldest igneous rocks in the IBM subduction zone, followed by boninites. Based on the assumption that induced subduction initiation does not allow intrusion of hot asthenospheric mantle into the subduction interface, it is argued that forearc basalts cannot be formed during induced subduction initiation (Maunder et al., 2020). However, both forearc basalts and boninites are popularly reported in supra subduction zone ophiolites (Ishizuka et al., 2014). This contradicts the modeling result suggesting that most of the metamorphic soles actually formed during induced subduction initiation. If the modeling results of Zhou and Wada (2021) are correct, then we need to reconsider the characterization of induced subduction initiation, and even the classification of subduction initiation.

Cenozoic subduction initiations in the west Pacific are well preserved but hotly debated examples (Arculus et al., 2015; Li et al., 2019; Reagan et al., 2019; Sun et al., 2020). It was taken as the best example of spontaneous subduction initiation (Stern, 2004; Arculus et al., 2015). The northwestward subduction of the Pacific Plate initiated in the Cenozoic. Previous studies found that the IBM subduction initiation was controlled by the subsidence of the old Pacific lithosphere in the Cenozoic, which is taken as the key evidence supporting spontaneous initiation (Arculus et al., 2015). However, plate reconstruction and geodynamic analyses indicate that the Cenozoic subduction initiation was induced by the hard collisions of the Indian and Australian Plate with the Eurasia continent along the Neo-Tethys suture (Sun et al., 2020). Consistently, the P-T conditions of Amami Sankaku Basin basalts indicate that the magmas formed during subduction initiation were transferred rapidly from ~30–60 km to the surface (Li et al., 2021), implying that they formed in a compressional environment.

Moreover, in contrast to the IBM trench, IODP Expedition 371 reveals that the Lord Howe Rise, northern Zealandia, was uplifted from depths of ~1 km underwater to above sea level and subsided back later on, which was attributed to crustal delamination and mantle flow that led to slab formation and subduction initiation (Sutherland et al., 2020).

If all the Cenozoic subduction initiations in the west Pacific were induced, why was the IBM associated with subsidence, whereas the northern Zealandia was associated with uplift and then subsidence? We propose that compression can result in subsidence as well as uplift, depending on the preexisting conditions (Fig. 1). Subduction initiation associated with the old oceanic crust usually has preexisting subsidence, which is amplified by compression. In contrast, subduction initiation associated with young oceanic plates may experience uplift during compression, followed by subsidence after subduction commenced. In this case, the current classification of subduction initiation should be reconsidered.

All data generated and/or analyzed during this study are available from the corresponding author upon reasonable request.