Mantle Potential Temperature Estimates of Basalt from the East Taiwan Ophiolite

The East Taiwan Ophiolite (ETO) is a fragment of the eastern-most South China Sea that was accreted to the Eurasian margin during the Pliocene and is a member of the Western Pacific and Cordilleran belt ophiolite group. Ophiolites from the Western Pacific and Cordilleran belt are typically subduction-related (i.e., supra-subduction or volcanic-arc) but the ETO is compositionally, isotopically and mineralogically similar to subduction-unrelated ophiolites. The primary melt compositions of ETO basaltic rocks were calculated and range from high-Mg basalt to picrite (i.e., MgO = 10 to 14 wt%). The mantle potential temperature (TP) estimates are within the range of ambient mantle (1300 1400°C) and other mid-ocean ridge ophiolites (i.e., Macquarie Island and Masirah) indicating that it is consistent with a mid-ocean ridge setting. Mantle potential temperature estimates for rocks from a mantle-plume-type ophiolite (i.e., 1620 1630°C) are anomalously high whereas rocks from supra-subduction zone ophiolites show a wider range that extends from ambient (i.e., Troodos and Semail) TP to very high (i.e., Betts Cove and Bay of Islands) TP.

Although all interpretations recognize the oceanic nature of the ETO rock suite, the precise origin is debated Jahn 1986;Chung and Sun 1992;Shao 2015). Early interpretations suggest the ETO was a submarine scree deposit consisting of angular mafic and ultramafic plutonic blocks that formed at a 'leaky' transform fault offsetting (Liou et al. 1977;Suppe et al. 1981). Jahn (1986), Chung and Sun (1992), and Hsieh et al. (2016), based on basalt geochemistry and Cr-spinel data from the peridotites, suggested that the ETO is likely representative of a midoceanic ridge. The nature of the ETO has direct implications on understanding the growth and development of the South China Sea (Taylor and Hayes 1980;Lee and Lawver 1995;Barckhausen and Roeser 2004;Barckhausen et al. , 2015Li et al. 2014;Chang et al. 2015).
The ETO thermal regime has not been estimated before, thus it is uncertain if the conditions were anomalously high or typical of ambient mantle. The primary melt composition and mantle potential temperature of basalt from different tectonic settings (i.e., ocean-islands, MOR, flood basalt provinces) can be deduced from its bulk composition providing it has only experienced olivine loss (Herzberg et al. 2007;Asimow 2008, 2015). Based on forward modeling of dry peridotite, the primary melt composition can constrain the mantle potential temperature (T P ) required to produce the melt (Herzberg et al. 2007;Herzberg and Asimow 2008). Mantle potential temperatures at oceanic spreading centers typically range from 1300 -1400°C whereas some oceanic hotspots and continental flood basalts can be 200 -300°C above ambient temperatures (Herzberg and Asimow 2008; Ali et al. 2010). The higher mantle potential temperatures at within-plate settings are thought to be an artifact of a mantle-plume (Herzberg et al. 2007;Ali et al. 2010). Thus, mantle potential temperature esti-mates for basaltic rocks from different types of ophiolites (i.e., MOR, SSZ, and plume) may show differences in their thermal regimes. For example, basalt from MOR-type ophiolites should have lower mantle potential temperatures than basalt from mantle-plume-type ophiolites.
In this paper, using PRIMELT3 software, we estimate the primary melt compositions and mantle potential temperatures of basaltic rocks from the ETO in order to determine the likely thermal conditions (Herzberg and Asimow 2015). Moreover, we compare the results to estimates from rocks related to MOR-, plume-, and SSZ-type ophiolites so that the tectonic setting of the ETO can be further constrained.

GEOlOGIcAl BAckGrOund
Taiwan is situated at the junction between the Ryukyuarc and the Luzon-arc and consists of accretionary wedge rocks, island arc rocks and older continental and oceanic lithosphere ( Fig. 1) (Chai 1972;Bowin et al. 1978;Suppe 1984;Tsai 1986;Kao et al. 1998). Westward subduction of the Pacific plate beneath the Eurasia plate during the Mesozoic Era (~150 Ma) was responsible for the Zhejiang-Fujian magmatic arc and the formation of proto-Taiwan. The changing nature of convergence (i.e., slowing subduction and slab roll back) between the paleo-Pacific plate and Eurasia during the Cenozoic allowed for fore-arc sediment accumulation as the Philippine Sea plate rotated clockwise in a northwestward direction (Lee and Lawver 1995). During the middle to late Miocene (~15 -12 Ma), the Luzon-arc shifted towards the Eurasian continental shelf along a left-lateral transform fault system in the east and subduction zone system in the north (Teng 1990;Huang et al. 2006). The obduction between the Luzon-arc and Eurasian continental shelf induced uplift of accretionary wedge sediments and formed the proto-Central Mountain Range of Taiwan during the mid to late Pliocene (Teng 1990;Huang et al. 2006). By the late Pliocene, the northern Luzon-arc collided obliquely with Eurasia and reached its current position. The volcanic rocks of the Luzonarc accreted into the uplifted Pliocene-Pleistocene passive margin sediments of Eurasia and became the Coastal Range whereas the uplifted sedimentary rocks developed into the Central Range. At the same time as the Luzon-Eurasia collision, the tectonic stress in northeast Taiwan transformed from compression to extension as the Philippine Sea plate continued to subduct beneath the Ryukyu-arc that induced back-arc extension and the opening of the Okinawa Trough and Yilan basin (Suppe 1984;Lee and Wang 1987;Teng 1990Teng , 1996Teng , 2007Kao et al. 1998).
The ETO is within the Lichi mélange of the Coastal Range and consists of large blocks and boulders of fossiliferous sea-floor sediments, pillow basalt with glassy margins, mafic dykes, plagiogranite, gabbro, pegmatite, and serpentinized peridotite (Liou et al. 1977;Liou 1979;Liou and Ernst 1979;Suppe and Liou 1979;Page and Suppe 1981;Suppe et al. 1981;Jahn 1986;Chung and Sun 1992). The idealized stratigraphy of the ETO is shown in Fig. 2 and is based on the work by Liou et al. (1977). The lower brecciated sequence consists of plutonic (i.e., gabbros, diabase, plagiogranite) rocks and serpentinized peridotite and is con-tact with the Lichi mélange. Thin layers (i.e., 10 -50 cm) of red shale and siltstone are found at the top of the plutonic rocks followed by the extrusive sequence of massive flows, pillow basalt with glassy rims and pillow breccia. Layers of red shale are reported between some volcanic units. Radioisotope dating (i.e., U/Pb and Ar/Ar) of the ETO yielded results between 17.5 ± 0.2 and 14.1 ± 0.2 Ma (Jahn 1986;Shao 2015;Hsieh et al. 2016). The whole rock geochemistry of the ETO basalts show depleted light-rare earth element patterns, N-MORB-like (normal-mid-ocean ridge basalt) whole rock compositions and Sr-Nd isotopic characteristics of depleted mantle (Jahn 1986;Yui and Yang 1988). Furthermore, the Cr-spinels from the ETO peridotites are compositionally similar to spinels from peridotites dredged from MOR (Dick and Bullen 1984;Hsieh et al. 2016).
Basalts with Mg# ≥ 65 from the ETO and MOR were selected for modeling whereas basalts from other settings were not restricted but generally had Mg# > 60. The selection criteria for the ETO and MOR ophiolites were relatively strict because primitive MORB typically has Mg# ≥ 65 (Kamenetsky et al. 2000). Other modeling parameters such as source composition and relative oxidation state were adjusted based on the type of ophiolite. For example, the FeOt content of mantle peridotite generally ranges between 8.0 and 9.0 wt% and the relative oxidation state of the mantle at MOR and mantle-plume settings tend to be more reducing relative to volcanic-arc-related settings (Herzberg and O'Hara 2002;Bézos and Humler 2005;Cottrell and Kelley 2011;Kelley and Cottrell 2012). We selected the lowest FeOt mantle value that produced a meaningful result whereas MgO was fixed to 38.12 wt%. In order to compensate for the differences in the relative oxidation state between volcanic-arc-unrelated and volcanic-arc-related ophiolites we set the Fe 2 O 3 /TiO 2 = 0.5 (i.e., reducing) for the ETO, Macquarie Island and Masirah whereas the remaining ophiolites are modeled with an Fe 2 O 3 /TiO 2 = 1.0 Asimow 2008, 2015).
The accumulated fractional melt (AFM) results from PRIMELT3 are plotted on a series of FeOt vs. MgO diagrams that show the primary melt composition and the equilibrium melting olivine control line (Figs. 3 and 4). The AFM composition represents the accumulation of melt fractions over the duration of source melting and is probably more representative of the processes that generate melts from the mantle (Herzberg and Asimow 2008). The solidus, melt fraction, and pressure lines shown in the models are derived from Herzberg and O'Hara (2002), Herzberg et al. (2007), and Asimow (2008, 2015). The ETO and MOR ophiolite results are plotted relative to the depleted mantle equilibrium melting curves due to their interpretation as mid-ocean spreading centers whereas the others are plotted relative to fertile peridotite melting curves due to their association with either a mantle plume or subduction zone systems. Pressure estimates for depleted mantle melting are not provided due to limited experimental data (Herzberg and O'Hara 2002). The complete primary melt compositions and temperature estimates are listed in the online supplementary Table S1.

PrIMAry MElT cOMPOsITIOns And MAnTlE POTEnTIAl TEMPErATurE EsTIMATEs
The primary melt compositions and olivine control lines for ETO basaltic rocks are shown in Fig. 3 and listed in Table 1. The calculations indicate the primary magmas are picritic to high-Mg basalt (MgO = 9.9 to 13.4 wt%) that experienced ~3 to ~6% olivine loss (Le Bas et al. 2000). The calculated initial olivine had forsterite values between 89 and 91 (Fig. 3b). The eruptive temperatures (T) and mantle potential temperatures (T P ) are estimated to be 1220 -1310°C and 1300 -1390°C respectively.
The melt compositions and T P for basaltic glass from the Macquarie Island ophiolite and rocks from the Masirah ophiolite show the primary magmas of both ophiolites were picritic to high-Mg basalt (MgO = 10.8 to 13.4 wt%) (Fig. 3c). The Masirah primary melt composition experienced 8 -12% olivine loss and the Macquarie Island melt experienced 7 -11% loss. The T P of the Masirah ophiolite ranges from 1320 -1400°C whereas the Macquarie Island ophiolite ranges from 1340 -1390°C (Table 2).
The primary melt compositions of rocks from the Mino-Tamba belt are shown in Fig. 4a. The calculations indicate the melt compositions are picritic (MgO = 22 wt%, Na 2 O + K 2 O > 1 wt%) and experienced ~22% olivine loss from a garnet peridotite. The estimated eruptive temperatures (T) and mantle potential temperatures (T P ) are 1475 -1490°C and 1620 -1630°C ( Table 2).
The rocks from supra-subduction-zone ophiolites produced high-Mg basalt to picritic (MgO = 10.9 to 17.1 wt%) primary melts (Fig. 4b). The estimated eruptive and mantle potential temperatures of the Troodos (upper pillow sequence at the Kythreotis quarry) and Semail (Lower extrusive Geotimes unit) ophiolites are 1250 -1330°C and 1325 -1420°C. In comparison the Bay of Islands and Betts Cove ophiolites have higher estimated eruptive and mantle potential temperatures of 1320 -1390°C and 1410 -1490°C (Table 2).

dIscussIOn
The calculated primary melt compositions of the ETO basaltic rocks suggest that the thermal regime of the ETO mantle was not anomalously hot (i.e., > 1550°C) and that it was similar to ambient mantle (Fig. 5). It seems that many of the ETO basaltic rocks are very close to primary melts as the model predicts ≤ 6% olivine loss and initial olivine compositions with Mg# 89 to 91 (Fig. 3b). The reported Mg# from ETO gabbro, mafic dykes, and basalt are between 83 and 87 with Mg# of 87 to 91 for ETO peridotites (Liou et al. 1977). In comparison, the T P estimates of mafic rocks from MOR-type ophiolites (i.e., Macquarie Island and Masirah) are indistinguishable from the ETO (Fig. 5).
An examination of the mantle potential temperatures of plume-type and SSZ-type ophiolites reveals that there are, in general, differences between the thermal regimes of MOR-type ophiolites and other types (Fig. 5). The calculated mantle potential temperatures from the Mino-Tamba plume-type ophiolite from SW Japan are very high (i.e., > 1550°C) and supportive of a hot mantle regime expected from a mantle-plume setting Asimow 2008, 2015). The presence of ultramafic volcanic rocks within the Mino-Tamba belt is consistent with the expectation of a mantle-plume derived large igneous province (Campbell 2007;Ali et al. 2010).
The T P estimates for the SSZ ophiolites clearly indicate that the Troodos and Semail ophiolites have T P (i.e., 1325 and 1420°C) similar to ambient mantle whereas the Betts Cove and Bay of Islands ophiolites have higher thermal conditions (i.e., 1410 -1490°C). It is uncertain why there  Juan et al. (1976Juan et al. ( , 1980, Liou et al. (1977), Chou et al. (1978), Suppe et al. (1981), and Jahn ( Liou et al. (1977), Juan et al. (1980), and Suppe et al. (1981). would be a difference between the thermal regimes of the Troodos and Semail ophiolites and the Betts Cove-Bay of Islands ophiolites but it could be related to their specific formation conditions. The Troodos and Semail ophiolites are classic examples of the supra-subduction "Tethyan ophiolites" (Pearce et al. 1984;Searle and Cox 1999;Dilek and Furnes 2011;Whattam and Stern 2011). In both cases the main ophiolite sequences were formed at 'subduction initiation centers' (i.e., extension above a subduction zone) and were obducted above the same subduction zone that created them (Stern and Bloomer 1992;Whattam and Stern 2011).

MOR
In the case of the Early Ordovician Bay of Islands and Betts Cove ophiolites, they likely formed at a fore-arc ridge axis normal to the arc trench trend (Dewey and Casey 2013). In-termediate mantle potential temperature estimates (i.e., 1400 and 1500°C) of basaltic rocks are interpreted to be related to heat incubation beneath moderately thickened crust in the case of some continental large igneous provinces (e.g., Ferrar and Central Atlantic Magmatic Province) or, in the case of some Cenozoic oceanic islands, as evidence of cool, lowmelt-fraction magmas from the periphery of a mantle plume (Coltice et al. 2007;Herzberg and Gazel 2009;Hole 2015 cooling plume (Herzberg and Asimow 2008;Herzberg and Gazel 2009). Alternatively it could be that some basalt from the Bay of Islands and Betts Cove ophiolites formed within slightly thicker oceanic crust as the ridge was intersecting the arc trench. Regardless, in spite of their similar classification as SSZ ophiolites, the thermal regimes of the Troodos and Semail ophiolites and the Bay of Islands and Betts Cove ophiolites appear to be different implying they may represent distinctly different types of SSZ ophiolites. The similarity in mantle potential temperatures between the ETO and MOR ophiolites, absence of arc-related (i.e., island-arc tholeiites, calc-alkaline basalt, boninite) rocks and the depleted Nd isotopic composition of peridotite is compelling evidence that the ETO was an open ocean-ridge environment (Jahn 1986;Chung and Sun 1992;Hsieh et al. 2016). The ETO thermal regime is definitively dissimilar to mantle-plume ophiolites and thus it is unlikely that it was associated with that setting. There is overlap between the T P estimates for the MOR ophiolites and the Troodos and Semail ophiolites suggesting the thermal regime of some SSZtype ophiolites and MOR-type ophiolites are similar.

cOnclusIOn
The results from this study suggest the primary compositions of ETO basalts are similar to high-Mg basalt to picrite. The mantle potential temperature estimates required to generate the primary melt composition are typical of MOR rather than plume-or some SSZ-type ophiolites. The PRIMELT3 results are additional evidence in support of a MOR interpretation for the ETO.