The migration of arc magmatism that is a fundamental aspect of plate tectonics may reflect the complex interaction between subduction zone processes and regional tectonics. Here we report new observations on volcanic migration from northwestern Sumatra, in the westernmost Sunda arc, characterized by an oblique convergent boundary between the Indo-Australian and Eurasian plates. Our study indicates that in northwestern Sumatra, volcanism ceased at 15–10 Ma on the southern coast and reignited to form a suite of active volcanoes that erupt exclusively to the north of the trench-parallel Sumatran fault. Younger volcanic rocks from the north are markedly more enriched in K2O and other highly incompatible elements, delineating a geochemical variation over space and time similar to that in Java and reflecting an increase in the Benioff zone depth. We relate this mid-Miocene volcanic migration in northwestern Sumatra to the far-field effect of propagating extrusion tectonics driven by the India-Eurasia collision. The extrusion caused regional deformation southward through Myanmar to northwestern Sumatra and thus transformed the oblique subduction into a dextral motion–governed plate boundary. This tectonic transformation, associated with opening of the Andaman Sea, is suggested to be responsible for the volcanic migration in northwestern Sumatra.
There is considerable evidence that arc magmatism is constant in neither time nor space (Paterson and Ducea, 2015). Arc magma migration, consequently, may reflect changes in convergence rate, subduction geometry, the depth of slab dehydration or the extent of partial melting, and/or their combined effect (Karlstrom et al., 2014). The West Pacific and Sunda subduction zones, circum–East and Southeast Asia (Fig. 1), in particular, have interacted with major tectonic events such as continental deformation and propagating extrusion owing to the collision of India into Eurasia (Tapponnier et al., 1982; Schellart et al., 2019). We report our finding of volcanic migration in the middle Miocene in northwestern Sumatra as part of the outcome of a systematic investigation on the magmatism of the island (Lai et al., 2019; Zhang et al., 2019; Li et al., 2020) (Fig. 1). A geochemical change is associated with the volcanic migration, which we attribute to the interplay between oblique subduction and regional tectonics, with an emphasis on the far-field role of the India-Eurasia collision in Southeast Asia.
The convergent movement of the Indo-Australian plate beneath the Eurasian plate is responsible for the Sunda subduction zone (Fig. 1). Whereas the subduction in Java is nearly perpendicular to the trench, that in Sumatra is highly oblique, resulting in the trench-parallel, strike-slip Sumatran fault system (Malod et al., 1995; McCaffrey et al., 2000). Running the length of Sumatra, this transform fault extends northward into the spreading center of the Andaman Sea (Curray et al., 1979). Active volcanoes typically occur near the fault (on either side), ∼100 km above the Benioff zone, except those from northwestern Sumatra that formed exclusively to the north of the fault system (Fig. 1). Such a volcanic “offset”, first noticed by Page et al. (1979) and attributed to the changing angle of the Benioff zone, however, has attracted little attention. The Toba caldera that occurs in the transition area (Fig. 1A) has been a research focus because of its super-eruptions and environmental impacts (see Chesner  for a review).
As synthesized by Barber et al. (2005), K-Ar ages obtained mainly by Bellon et al. (2004), along with sparse age and geochemical data reported by earlier studies, indicate that magmatism in Sumatra can be divided into several pre-Cenozoic stages and five subsequent stages active from the Paleocene to recent. Magmatic stages from pre-Cenozoic to recent may be affiliated with a change in the subduction system that started operating in the eastern portion of the Paleo-Tethys and persists to the modern Indo-Australian system in Southeast Asia (Hall, 2012; Metcalfe, 2013; Zhang et al., 2019; Li et al., 2020).
Sumatra consists of three geologic units: namely, from southwest to northeast, the Woyla terrane, the West Sumatra block, and the East Sumatra block (Barber et al., 2005). The Woyla terrane is an intra-oceanic arc complex formed in eastern Tethys and accreted to its present location in the Early Cretaceous (Hall, 2012; Advokaat et al., 2018). West Sumatra was conventionally correlated with Cathaysia, whereas East Sumatra was thought to be a part of Sibumasu, which belongs to East Gondwana (Metcalfe, 2013). Zhang et al. (2018), however, used new detrital zircon evidence to argue that divides Sibumasu and integrates West and East Sumatra to correlate with the West Burma block, or part of the Irrawaddy block, a pre-Cenozoic tectonic element renamed by Ridd (2016).
SAMPLES AND METHODS
We collected a total of 23 basalt and andesite samples from northwestern Sumatra (Fig. 1B), including 6 from the Tertiary volcanicsalong the southern coast, 7 from Seulawah Agam, and 10 from Geureudong, the latter two of which are principal volcanoes that formed north of the Sumatran fault. In situ zircon U-Pb age dating of six samples was carried out using laser ablation–inductively coupled plasma mass spectrometry (LA-ICPMS) at the Department of Geosciences, National Taiwan University (see Table S1 in the Supplemental Material1). Whole-rock major and trace element determinations for all studied samples, along with Sr-Nd isotopic analyses of selected samples, were performed at the same institution (Table S2, and supplemental text in the Supplemental Material). Note that relevant zircon age data from our counterpart analyses of Cenozoic arc volcanism from the entire Sumatra island (Lai et al., 2019), including those from Toba and adjacent areas, are summarized in Figure S1. Additional U-Pb ages of detrital zircons (Fig. 2A) from sample sites in northern Sumatra (Fig. 1A) are also presented here (Zhang et al., 2018, 2019; Table S3).
RESULTS AND DISCUSSION
Volcanic Migration in Space and Time
The two samples of Tertiary volcanics from the southern coast gave early Miocene mean 206Pb/238U ages at 16.5 ± 0.5 Ma (sample 13SU01) and 20.1 ± 0.3 Ma (sample 13SU04) (Figs. 2B and 2C). Magmatic zircon separates from both samples are mostly euhedral to subhedral, with those from basaltic andesite sample 13SU01 (SiO2 = 53.6 wt%) having lower uranium (80–471 ppm) than those from andesite sample 13SU04 (SiO2 = 58.5 wt%; U = 687–1742 ppm) (Table S1). In contrast, zircon separates from four other andesite samples from the Seulawah Agam and Geureudong volcanoes are all too young (<0.3 Ma) to be dated precisely by the LA-ICPMS method (Lai et al., 2019; Fig. S1). Their ages are therefore referred to as Quaternary.
Figure 2A synthesizes all available age data from northwestern Sumatra (Fig. S1; Lai et al., 2019), including young detrital zircon ages from back-arc basins and riverbanks (Fig. 1A; Table S3). These age data indicate a migration of the volcanic arc between 15 and 5 Ma. More specifically, in northwestern Sumatra, arc volcanism previously occurred along the southern coast as a linear extension of the volcanic front in central and southeastern Sumatra (Fig. S1), where Quaternary volcanic centers were limited to the southern coast along the Sumatran fault (Page et al., 1979; Barber et al., 2005). In this portion of the arc, volcanism ceased south of the Sumatran fault at ca. 15–10 Ma and then resumed at ca. 10–5 Ma to the north of the fault, forming a suite of active volcanoes. The Toba caldera complex developed in the volcanic “offset” or transitional area between the southern and northern portions of the Cenozoic volcanic arc (Fig. S1). We note that despite the abundance of Quaternary ages from Toba in the age histogram (Fig. 2A), volcanic reinitiation north of the Sumatran fault at ca. 10–5 Ma is documented by a late Miocene sandstone that yielded U-Pb ages of ca. 10–5 Ma on 48 of 94 detrital zircons (sample 13SU26; Table S3). These detrital zircons are presumably sourced from nearby volcanic deposits.
Associated Change in Magma Composition
The volcanic migration in northwestern Sumatra is associated with changes in magma composition. The samples studied vary from basalt to andesite in composition (SiO2 = 48.2–62.4 wt%;Table S2), with K2O contents increasing from south (Tertiary volcanics, 0.30–1.26 wt%; low-K tholeiitic series) to north (Seulawah Agam, 0.83–1.71 wt%; Geureudong, 1.68–2.84 wt%; high-K calc-alkaline series) (Fig. 3A). Associated geochemical variations are depicted by the rare earth element (REE) patterns and multi-element spidergrams (Fig. S2), with all samples showing marked depletion in high field strength elements (Ti, Nb, and Ta) and enrichment in large ion lithophile elements (LILEs; Cs, Rb, Ba, and Sr), typical of subduction-related magmas.
The compositional change from south to north, moreover, is correlated with Benioff zone depths (Fig. 1B). Early Miocene volcanics to the south crop out ∼80 km above the present subducting slab, whereas the active volcanoes in the north (Seulawah Agam and Geureudong) erupted from ∼120 km and ∼180 km depths. According to the tectonic setting of this area, we argue this variation can be compared to the K-h correlation (enrichment of K and other LILEs as a function of depth [h]) in Java, the eastern Sunda subduction zone (Fig. 1A). A “widened” magmatic arc occurs here, with the volcanic front ∼100 km above the Benioff zone, consisting of a calc-alkaline main arc (Whitford et al., 1981) and more K-rich alkaline rocks in the back-arc, such as Muria volcano (Edwards et al., 1991) situated ∼300 km above the subducting slab (Hall and Spakman, 2015). Comparison of these two magma suites from Sumatra and Java is illustrated using a K2O versus SiO2 plot (Fig. 3A) and Sr-Nd isotopic correlation diagram (Fig. 3B). Note that nearly all the Sunda arc volcanics, except for Muria highly potassic rocks, exhibit a “juvenile” isotopic nature marked by positive εNd values (Fig. 3B).
Causes of Volcanic Cessation in the South
Island arcs commonly undergo magmatic flare-ups and lulls relating to regional tectonics (Paterson and Ducea, 2015; Zhang et al., 2019). In northwestern Sumatra, we consider two principal tectonic controls on termination of the arc volcanism at ca. 15–10 Ma along the southern coast. One possible cause is the resistance of slab subduction beneath northern Sumatra, where ca. 35–50 Ma oceanic lithospheric associated with the Wharton Fossil Ridge (Whittaker et al., 2007; Jacob et al., 2014) is now being subducted (Fig. 1). In contrast, east of the Investigator fracture zone, older Cretaceous-age crust is currently subducting beneath central and southern Sumatra (Jacob et al., 2014). Consequently, beneath northern Sumatra, younger and more buoyant oceanic lithosphere resists subduction, and this manifests as complexly folded subducted lithosphere in the mantle transition zone, as imaged by tomographic studies (Pesicek et al., 2008; Hall and Spakman, 2015).
The other, and arguably more significant, control is progressive transition of the tectonic setting from oblique subduction to predominantly strike-slip movement that we argue to have begun during Miocene times as a result of the far-field effect of propagating block extrusion driven by the India-Eurasia collision (Tapponnier et al., 1982). We propose a conceptual model (Fig. 4) to illustrate how such propagating extrusion may have interacted with regional tectonics in the western part of Southeast Asia, confined by dextral movement of the Sagaing-Sumatran fault system, initiating at ca. 20–15 Ma on the Sagaing fault (Bertrand et al., 1999) and ca. 23–15 Ma on the Sumatran fault (Curray et al., 1979). The propagating extrusion that caused crustal deformation from the eastern Himalayas southward through the West Burma block to northwestern Sumatra would have progressively transformed the oblique subduction there into a dextral motion–governed plate boundary in response to the continuing Indian collision (Figs. 4A and 4B). As noted specifically in the Burmese-Popa arc (Rao and Kalpna, 2005) or more widely over this highly oblique section of the Indo-Australian convergent zone (Richards et al., 2007), the stress field in the upper part (<90 km) of the subducted lithosphere is governed by horizontal plate motion, albeit its lower part (>90 km) is still controlled by gravitational loading of the slab. We therefore suggest such a tectonic transition from oblique convergence to dextral motion of the subducted slab to have played a controlling role in the magmatism all the way from the West Burma block to northwestern Sumatra during the middle Miocene (Fig. 4A), not only resulting in the volcanic gap and/or migration observed from the Popa arc (Lee et al., 2016), but also terminating the volcanism along the southern coast in northwestern Sumatra.
Causes of Volcanic Reinitiation in the North
Renewal of magmatism requires thermal perturbation in the source region, similar to that proposed for the Burmese-Popa arc where volcanism renewed in the Quaternary (<1 Ma) due to onset of a transtensional regime triggered by roll-back of the sinking slab (Lee et al., 2016). The cause of Toba super-eruptions, for another example, has been accepted by most workers (Page et al., 1979; Barber et al., 2005; Hall and Spakman, 2015) to have been extra heat from mantle upflow through a slab tear confined by subduction of the Investigator fracture zone (Fig. 4B). The slab tear, if indeed responsible for Toba super-eruptions from a gigantic silicic magma reservoir in the upper continental crust (Chesner, 2012), however, would have been way too restricted to also account for the volcanic reignition in northwestern Sumatra across its northern coast and extending into the southeastern part of the Andaman Sea (Fig. 1). Notably, our northwestern Sumatran samples have Sr-Nd isotopic compositions (Fig. 3B) that are distinct from those of any Toba-related volcanics, also not supportive of a link with the slab tear for the volcanic reinitiation.
Resistant subduction of the Wharton Fossil Ridge in the west of Toba may have led to change in the subduction angle and/or rate, thus causing not only a magmatic lull but also migration of the volcanic arc with changing composition. If so, the northward volcanic migration could be an indication of subduction flattening in northwestern Sumatra, owing to presence of the younger and thus more buoyant Wharton Fossil Ridge. However, this is not the case because no apparent shift of the Benioff zone contours to the west and east of Toba is observed at magma-generating depths (∼100 km), despite the complex folding of the slab in the mantle transition zone (Pesicek et al., 2008; Hall and Spakman, 2015). Therefore, an interpretation of local subduction forcing for the volcanic migration is unfavored.
We relate the volcanic renewal also to the extrusion tectonics that played a role in the opening of the Andaman Sea, a late Cenozoic pull-apart basin formed by dextral shear of the Sagaing-Sumatran fault system (Curray et al., 1979). While the Central Andaman Basin has been the site of active seafloor spreading since the early Pliocene (ca. 4 Ma) or as early as the middle Miocene (Morley, 2017), its adjacent regions are mostly extended continental crust resulting from the interplay between the western Sunda back-arc extension and the differential motion of India with respect to Southeast Asia (Curray, 2005; Morley and Alvey, 2015). Such an extensional regime, in particular with the seafloor spreading, could have sufficiently enhanced the regional geotherm to initiate magma generation from the trench side (Narcondam and Barren Islands) to the back-arc side on and off northwestern Sumatra (Fig. 4B). The geochemical and Sr-Nd isotopic constraints enable us to argue that, relative to high-Al basalts of Barren volcano that belong to the low-K tholeiitic series produced by shallow melting in the mantle wedge (Luhr and Haldar, 2006), the volcanic reinitiation in northwestern Sumatra involves small-degree initial melting at greater depths of an enriched lithospheric mantle source that, as suggested by Edwards et al. (1991) for generation of the Muria potassic magmas, was metasomatized by pre-existing subduction zone processes.
We thank M.H. Roselee for help in fieldwork; C.-H. Hung, J.-T. Tang, and Allie Honda for help with experiments; T.-T. Lo for drawing the oceanic crust map; and D. Brown, C. Morley, and two anonymous reviewers for very helpful and constructive comments. This study was financially supported by Ministry of Science and Technology (MOST) grants 106-2116-M-003-006 and 107-2116-M-003-001 to Y.-M. Lai, and Academia Sinica grant ASIA-IVA-M01 to S.-L. Chung.