Professional Documents
Culture Documents
Zellmer 2012 Adakites Water Content
Zellmer 2012 Adakites Water Content
Geology
Lower crustal H2O controls on the formation of adakitic melts
G.F. Zellmer, Y. Iizuka, M. Miyoshi, Y. Tamura and Y. Tatsumi Geology 2012;40;487-490 doi: 10.1130/G32912.1
click www.gsapubs.org/cgi/alerts to receive free e-mail alerts when new articles cite this article click www.gsapubs.org/subscriptions/ to subscribe to Geology click http://www.geosociety.org/pubs/copyrt.htm#gsa to contact GSA
Copyright not claimed on content prepared wholly by U.S. government employees within scope of their employment. Individual scientists are hereby granted permission, without fees or further requests to GSA, to use a single figure, a single table, and/or a brief paragraph of text in subsequent works and to make unlimited copies of items in GSA's journals for noncommercial use in classrooms to further education and science. This file may not be posted to any Web site, but authors may post the abstracts only of their articles on their own or their organization's Web site providing the posting includes a reference to the article's full citation. GSA provides this and other forums for the presentation of diverse opinions and positions by scientists worldwide, regardless of their race, citizenship, gender, religion, or political viewpoint. Opinions presented in this publication do not reflect official positions of the Society.
Notes
Institute of Earth Sciences, Academia Sinica, 128 Academia Road Section 2, Nankang, Taipei 11529, Taiwan Lamont Doherty Earth Observatory, 61 Route 9W, Palisades, New York 10964, USA 3 Department of Education and Regional Studies, University of Fukui, 3-9-1 Bunkyo, Fukui City, Fukui 910-8507, Japan 4 IFREE, JAMSTEC, 2-15 Natsushima, Yokosuka, Kanagawa 237-0061, Japan ABSTRACT At volcanic arcs, uids released from the subducting slab lower the solidus of the mantle wedge and cause melting. Furthermore, slab melts may inltrate the mantle wedge, and have been suggested to generate adakitic (residual garnet) signatures at some arc volcanoes. However, experimental work indicates that the garnet stability eld will expand in the lower overriding crust in the presence of somewhat less hydrous melts, suggesting that such signatures may also develop at crustal levels. Here we use geothermometry and plagioclase hygrometry of mac eruptives from southwest Japan to demonstrate that the adakitic compositions of associated intermediate magmas are of lower crustal origin due to a decrease in the water content of parental melts, and are not generated by partial melting of the eclogitic subducting slab at elevated temperatures. Lower crustal melt evolution at reduced water contents may represent an important process for generating adakitic signatures in all tectonic settings that have previously been considered to enhance slab melting. Our results demonstrate that magmatic water plays a key role in the differentiation of arc magmas in modern and ancient subduction settings. INTRODUCTION At convergent margins, water plays a pivotal role in melt generation. Hydration of the mantle wedge by slab uids lowers the wedge solidus and causes mantle melting, and this process accounts for many rst-order chemical features in the genesis of arc magmas (Gill, 1981). However, there are enigmatic cases of arc volcanoes erupting intermediate to felsic magmas with unusually high Sr/Y ratios due to low Y (18 ppm) and high Sr (400 ppm) contents, and these have been called adakitic (Defant and Drummond, 1990; Kay, 1978). Adakitic signatures have been linked to garnet being residual during partial melting of the eclogitic subducting slab, as a result of unusually high slab temperatures developing due to subduction of young slabs (Defant and Drummond, 1990), at subduction (Gutscher et al., 2000), or mantle upwelling along slab edges (Morris, 1995; Yogodzinski et al., 2001). However, recent work has indicated that slab surface temperatures may generally become high enough to cause partial melting of the hydrous slab (Plank et al., 2009), followed by melt inltration of the wedge (Schiano et al., 1995), and that slabderived silicic components may therefore be an integral part of the genesis of typical andesitic arc magmas (Schiano et al., 1995; Straub et al., 2011), rather than generating atypical adakitic volcanic products. Garnet, a high-pressure mineral, may also impose its enigmatic signature at the base of the overriding crust (Garrison and Davidson, 2003; Macpherson et al., 2006; Richards, 2011). Although amphibole typically dominates the phase assemblage (Davidson et al., 2007), recent experimental work has shown that at moderate pressures (12 kbar), garnet stability increases at the expense of amphibole with decreasing H2O contents (Alonso-Perez et al., 2009; Mntener et al., 2001). It has therefore been suggested that H2O variations may modulate adakitic signatures at normal arc crustal thicknesses, i.e., pressures of ~1 GPa (Zellmer, 2009). Our study sets out to test this hypothesis. GEOLOGICAL BACKGROUND AND APPROACH We have chosen southwest Japan, where the crustal thickness is ~35 km (Oda and Ushio, 2007; Ueno et al., 2008), as a suitable natural laboratory (Fig. 1). Here the Philippine Sea slab consists of ca. 40 Ma or older West Philippine Basin lithosphere subducting steeply under central Kyushu, and younger (ca. 2615 Ma), more buoyant Shikoku Basin lithosphere subducting shallowly beneath western Honshu (Mahony et al., 2011), where it terminates with its leading edge at a depth of ~200 km, approximately below the active volcanic arc (Abdelwahed and Zhao, 2007). Based on the occurrence of deep long-period tremors, the young slab is thought to dehydrate beneath Shikoku in the forearc (Obara, 2002). Chemically, the arc changes northeastward from the typical calcalkaline Aso volcano in central Kyushu to eruptives yielding increasingly high Sr/Y ratios at Kuju and Yufu volcanoes in northern Kyushu (Kita et al., 2001; Sugimoto et al., 2006), and Aonoyama, Sanbe, and Daisen volcanoes in western Honshu (Morris, 1995). Residual garnet signatures have previously been ascribed to partial melting at the leading edge of the young subducting slab, bolstered by hot mantle upwelling from depth (Morris, 1995; Sugimoto et al., 2006). Our experiment involves the determination of temperatures and water contents of the mac melts that fed mac volcanism in this area. An origin of adakitic melts in the subducting slab is favored if spatially associated (but not necessarily cogenetic) mac magmas are hot (~11501450 C)
131E
132E
35
133E
134E Daisen
30 35
50 km
35N 35 Abu 40 Aonoyama 34N 38 36 80 60 50 36
80
30 Mengame
Sanbe 60
ern West
30
Hons
50 40 30
hu
Kyushu
34 Oninomi Yufu Kuju o 33 N 38
ik Sh
30
ok
20
40
100
120
Aso 34
10 40 20
*E-mail: gzellmer@earth.sinica.edu.tw.
Figure 1. Regional location map of southwest Japan volcanoes referred to in text. Black contour lines show depth distribution (in km) of upper boundary of subducting Philippine Sea slab estimated from distribution of intermediate-depth earthquakes (after Katagi et al., 2008), accurate to <5 km. In western Honshu, subduction beyond ~60 km depth occurs aseismically (Abdelwahed and Zhao, 2007). Colored contour lines show Moho depth estimates for Kyushu (in blue; Oda and Ushio, 2007) and western Honshu (in green; Ueno et al., 2008).
2012 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or editing@geosociety.org. GEOLOGY, June 2012 Geology , June 2012; v. 40; no. 6; p. 487490; doi:10.1130/G32912.1; 3 gures; Data Repository item 2012143.
487
Downloaded from geology.gsapubs.org on July 13, 2012 and essentially dry (0.7 wt% H2O), as would be expected from melts generated by adiabatic decompression of mantle (Kohut et al., 2006) hot enough to bolster melting of an eclogitic slab that has largely dehydrated in the forearc (Obara, 2002). Conversely, if mac melts in western Honshu have temperatures more similar to the typical mac arc magmas of central Kyushu (<1150 C), but carry less H2O than typical for primitive arc melts (~3.56.5 wt%; Sisson and Layne, 1993), this would imply the absence of enhanced melting of the slab, and that the adakitic signatures are instead linked to melt evolution of somewhat less hydrous mantle melts in the lower overriding crust in the presence of residual garnet (Zellmer, 2009). RESULTS We have sampled Quaternary mac rocks from Aso, Kuju, Yufu, Oninomi, Abu, Menkame, and Daisen (Fig. 1), and have constrained magmatic temperatures of the least-evolved samples by pyroxene thermometry (Lindsley, 1983), and two-pyroxene thermobarometry and melt inclusion-in-olivine thermometry (Putirka, 2008; cf. Fig. 2A, and Table DR1 in the GSA Data Repository1). For pyroxene thermometry, we assumed a pressure of 10 kbar appropriate for parental melts crossing the Moho, from which the most magnesian pyroxenes that yield the highest temperatures would have formed. Within each sample, temperatures vary due to crustal melt evolution, including signicant crustal assimilation (Hunter, 1998; Ohta et al., 1992). Pressures of <10 kbar might be more appropriate for some of the more calcic pyroxenes that yield lower temperatures, but we note that (1) the effect of pressure on the obtained temperatures is comparatively small (<5% difference for temperatures >900 C; cf. Lindsley, 1983), and (2) it is the maximum temperatures returned by the least evolved melts that are of interest to this investigation. These maximum temperatures vary little between Kyushu and western Honshu, and at <1150 C are typical of mac arc magmas derived by ux melting of the mantle wedge. Furthermore, these temperature estimates are consistent with those obtained by the more recent thermodynamic parameterizations of the two-pyroxene thermobarometer at lower crustal pressures, and by the melt inclusionin-olivine thermometer (Fig. 2A). In order to constrain along-arc variations in water content of parental melts, we utilize plagioclase feldspar as a melt hygrometer (Lange et al., 2009), because melt inclusions in magnesian olivines from this area are generally too small (<20 m) for direct H2O analysis. Plagioclase compositions are highly variable within each individual sample (Fig. 2B; Table DR1), and scatter to low anorthite contents due to shallow-level degassing and melt differentiation, crustal assimilation, and magma-mixing processes that have been well documented at some of the volcanoes investigated here (Hunter, 1998; Koyaguchi, 1986; Ohta et al., 1992). Nevertheless, it is evident that anorthite contents in the mac magmas systematically vary along the arc: maximum anorthite contents in our samples, representative for their earliest differentiation history, monotonically decrease northeastward (Fig. 2B). We note also that there is a broad anticorrelation between maximum anorthite contents and bulk-rock Mg# (Fig. DR1 in the Data Repository). Here we use the maximum anorthite contents for estimating water contents in the least evolved melts entering the lower overriding crust, and discuss the implications of the observed covariation. MAGMATIC WATER CONTENTS The plagioclase hygrometer (Lange et al., 2009) is calibrated using data on metaluminous samples spanning a wide range of liquid compositions, plagioclase compositions, temperatures, pressures (to 3 kbar), and dissolved melt water concentrations appropriate for our samples, although early crystallization in the parental mac magmas likely started in the lower crust at the plagioclase saturation boundary (the Moho; cf. Mntener and Ulmer, 2006). The effect of pressure on calculated H2O contents is small (Fig. 3A), and extrapolation of the hygrometer to lower crustal pressures is unlikely to introduce signicant error (Lange et al., 2009; see also Appendix DR1). We have used the whole-rock compositions of our samples as proxies for the primitive melt compositions. One caveat might be the modication of mac melts through contamination with felsic crustal components (Hunter, 1998; Koyaguchi, 1986; Ohta et al., 1992) or crystal fractionation, suggested by the low Mg# of some of our samples. However, the hygrometer we employed is quite insensitive to variations in liquid composition, and is thus very suitable even for samples that have undergone signicant open-system crustal evolution. For example, the hygrometer may only slightly overestimate the water content of the parental melt of our andesite sample from Kuju; assuming that the parental melt composition for Kuju was in fact similar to the bulk rock composition of the Aso basalt we studied, the H2O content calculated on basis of the actual andesitic bulk rock composition would overestimate the melt H2O content by only 0.6 wt%. All other samples we used for hygrometry are of basaltic to basaltic andesitic bulk compositions. In this case, any deviations from plagioclase-liquid equilibria will be negligible in terms of derived melt H2O contents (~0.1 wt%), which for the most
Menkame
Yufu Oninomi
1150
T (C) mafics
Kyushu
Crustal melt evolution (e.g., mixing, degassing)
Western Honshu
Most primitive feld spar
form ed f r
Phenocrysts Groundmass
om m
X An mafics
agma
34
34.5
35
Daisen 35.5
Kuju
Latitude (N)
Figure 2. A: Along-arc temperatures (T) estimated on basis of pyroxene thermometry (O; Lindsley, 1983), orthopyroxene-clinopyroxene equilibrium thermobarometry (red dots; Putirka, 2008), and melt inclusion olivine-liquid geothermometry (blue dots; Putirka, 2008). Mac enclaves are in gray. Temperature precision is estimated to 25 C. Bracketed Kuju samples have equilibrated at mantle pressures. Horizontal bar indicates range of highest mineral thermometry estimates used for calculation of lower crustal magma water contents. B: Along-arc variations in anorthite contents (XAn) of plagioclase from mac eruptives of the southwest Japan volcanic arc. Uncertainties in XAn are estimated to be <1.5% for the most calcic crystals. Note higher XAn contents of groundmass plagioclase in Menkame, indicative of open system processes. Daisen basalts do not contain plagioclase phenocrysts.
1 GSA Data Repository item 2012143, Appendix DR1, Table DR1, and Figures DR1 and DR2, is available online at www.geosociety.org/pubs/ft2012.htm, or on request from editing@geosociety.org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.
488
Abu
Aso
1200
Mafic enclaves
10 kbar 3 kbar
5 4 3
Western Honshu
Shallow subduction, younger slab (dehydration in forarc)
Kyushu
2 1 Yufu Oninomi Kuju
a ual g Resid
n t sig rne
n di ire qu c a re ( atu
t us cr er w lo
32.5
33.0
33.5
34.0
34.5
to residual garnet, rather than residual amphibole (which would deplete the melt in middle REEs over heavy REEs; cf. Davidson et al., 2007) or plagioclase accumulation (cf. Dessimoz et al., 2011). An increasing garnet signature with decreasing water contents is consistent with residual garnet in the lower overriding crust during early differentiation of more primitive, parental melts (Alonso-Perez et al., 2009; Mntener et al., 2001). Although rare, magmatic garnets have been described from a variety of volcanic rocks in southwest Japan (Itoh, 1990) and elsewhere (e.g., Day et al., 1992; Harangi et al., 2001; Scheibner et al., 2008), and from deep roots of island arc rocks (e.g., Ringuette et al., 1999). These garnets are typically richer in almandine than pyrope (i.e., have Fe > Mg), and are therefore ideal as lower crustal residuals during the genesis of adakitic intermediates while retaining elevated Mg# (such as those observed in some other adakite localities, where residual garnet has been invoked). Figure DR1 shows that samples from Kyushu with the most calcic plagioclase and thus highest inferred water contents have the lowest bulk-rock Mg#, consistent with lower crustal fractionation of Mg-rich amphibole. Conversely, samples from western Honshu with lower plagioclase anorthite contents and hence lower calculated water contents retain high bulk-rock Mg#, consistent with magmatic garnet as a lower crustal residual phase. These observations indicate that in addition to amphibole (Davidson et al., 2007), garnet is an important phase in controlling the geochemistry of evolving magmas in the lower crust. CONCLUDING REMARKS Our results corroborate previous notions that variations in H2O content have a profound inuence on the chemical evolution of mac arc magmas at crustal levels (Shellnutt and Zellmer, 2010; Sisson and Grove, 1993; Zellmer, 2009). Based on our ndings from southwest Japan, we propose that in some adakitic arc magmas associated with young subducting lithosphere (Defant and Drummond, 1990), at subduction (Gutscher et al., 2000), and mantle upwelling (Yogodzinski et al., 2001), residual garnet signatures may not be generated by increasing amounts of slab melting at higher temperatures, but by crustal differentiation of mac parental melts with lower than typical arc H2O contents expected in these settings. Given the control of magmatic water contents on the crustal differentiation of arc magmas, we speculate that residual garnet at the base of the overriding continental crust was more common in the Archean, when higher geothermal gradients may have led to shallow slab dehydration (Bjrnerud and Austrheim, 2004) and thus the production of less hydrous primitive melts. Reconstructing the H2O content of mac components in bimodal tonalite-trondhjemite-granodiorite suites would be a promising approach for testing this hypothesis.
ACKNOWLEDGMENTS Zellmer thanks J.-I. Kimura, R.A. Lange, O. Mntener, A. Nichols, and J.G. Shellnutt for discussions. H.-H. Hsieh and Y.-T. Hsu helped with microprobe analyses. L.-M. Cioec, M. Shimono, and H. Shukuno assisted during eld work. The comments of several anonymous reviewers improved earlier versions of this paper. We acknowledge funding by the National Science Council of Taiwan (grants 98-2918I-001-014 and 99-2116-M-001-010 to Zellmer, and 99-2116-M-001-013 to Iizuka). REFERENCES CITED Abdelwahed, M.F., and Zhao, D., 2007, Deep structure of the Japan subduction zone: Physics of the Earth and Planetary Interiors, v. 162, p. 3252, doi:10.1016/j.pepi.2007.03.001. Alonso-Perez, R., Mntener, O., and Ulmer, P., 2009, Igneous garnet and amphibole fractionation in the roots of island arcs: Experimental constraints on andesitic liquids: Contributions to Mineralogy and Petrology, v. 157, p. 541558, doi:10.1007/s00410-008-0351-8. Bjrnerud, M.G., and Austrheim, H., 2004, Inhibited eclogite formation: The key to the rapid growth of strong and buoyant Archean continental crust: Geology, v. 32, p. 765768, doi:10.1130/G20590.1. Davidson, J., Turner, S., Handley, H., Macpherson, C., and Dosseto, A., 2007, Amphibole sponge in arc crust?: Geology, v. 35, p. 787790, doi:10.1130 /G23637A.1.
35.0
35.5
Latitude (N)
Figure 3. A: Mac melt H2O contents of southwest Japan at pressures of 3 and 10 kbar, at the conservative temperature range indicated by gray bar in Figure 2A, calculated using the latest plagioclase hygrometer (Lange et al., 2009). H2O contents at Kuju may be slightly overestimated (by as much as 0.6 wt%; see text for discussion). B: Sr/Y ratios in intermediate to felsic eruptives broadly increase as water contents in associated mac rocks decrease along southwest Japan volcanic arc (data compiled from Hunter, 1998; Kita et al., 2001; Morris, 1995, and references therein; and Sugimoto et al., 2006).
calcic cores vary little with the major element composition of the coexisting melt from basalts to basaltic andesites. Using 10 kbar as a lower crustal pressure estimate, a temperature of 1080 30 C (and 970 30 C for Yufu; Fig. 2A), and the maximum anorthite contents of plagioclase phenocrysts (Fig. 2B), we derive parental melt H2O contents of 3.66 wt% (0.4 wt%) in Kyushu, typical for mac arc magmas (Sisson and Layne, 1993). Melt H2O contents decrease signicantly to 1.82 wt% (0.4 wt%) in western Honshu (Fig. 3A). Concomitantly, Sr/Y ratios increase in spatially associated intermediate compositions (Fig. 3B). The somewhat lower water content of parental mac melts in western Honshu may be related to loss of water in the forearc due to shallow subduction of the young Philippine Sea slab (Obara, 2002), dilution of hydrous wedge mantle with dry mantle derived from the backarc (Zellmer, 2009), or a combination of these processes. However, mac magmas erupted from the volcanic front in western Honshu are clearly more hydrous than melts generated by adiabatic upwelling of hot mantle (<0.7 wt% H2O; cf. Kohut et al., 2006).
DISCUSSION Our geothermometric data are difcult to reconcile with variable contributions of slab melts along the southwest Japan arc as a major source of adakitic volcanic rocks. Increasing Sr/Y ratios in intermediate to felsic melts are instead related to decreasing water contents in associated mac magmas (Fig. 3). The few available rare earth element (REE) data indicate that the adakitic signature is correlated with signicantly increasing Tb/Yb ratios (Fig. DR2), and therefore unequivocally linked
489
490