You are on page 1of 12

Earth and Planetary Science Letters 498 (2018) 203–214

Contents lists available at ScienceDirect

Earth and Planetary Science Letters


www.elsevier.com/locate/epsl

Petrogenesis of boninitic lavas from the Troodos Ophiolite, and


comparison with Izu–Bonin–Mariana fore-arc crust
Dominic Woelki ∗ , Marcel Regelous, Karsten M. Haase, René H.W. Romer, Christoph Beier
GeoZentrum Nordbayern, Friedrich-Alexander-Universität (FAU) Erlangen–Nürnberg, Schlossgarten 5, D-91054 Erlangen, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Supra-subduction zone (SSZ) ophiolites such as the Troodos Ophiolite of Cyprus are thought to have
Received 23 December 2017 formed at spreading centres close to subduction zones. Similarities in the geochemistry between lavas
Received in revised form 25 May 2018 from SSZ ophiolites and fore-arc lavas from active subduction zones, and the presence of boninites in
Accepted 28 June 2018
both, have led to the suggestion that SSZ ophiolites represent fragments of fore-arc crust and mantle,
Available online 10 July 2018
Editor: M. Bickle
formed during subduction initiation. Here we present major and trace element and Sr, Nd and Pb isotope
data for fresh volcanic glasses from a section through the lava pile on the southern margin of the Troodos
Keywords: Ophiolite, and compare these to lavas from the Izu–Bonin–Mariana fore-arc. In Troodos, boninites and
Cyprus tholeiitic basalts are interbedded and were derived from a highly depleted mantle source that was later
boninitic glass enriched by both fluids and melts derived largely from subducted sediment, before melting beneath
ophiolite the spreading axis. Troodos boninites differ from Izu–Bonin–Mariana boninites by their greater source
Izu–Bonin–Mariana depletion, enrichment in Nb by small degree melts, and lack of Zr enrichment relative to Sm. Together
subduction initiation
with the lack of fore-arc basalts in Troodos, our data imply that the Troodos Ophiolite was formed in a
interbedded
fore-arc location at a back-arc spreading centre that propagated into arc crust. The Troodos Ophiolite was
thus not formed during subduction initiation and thus may not be used as analogue for the formation of
fore-arc basalts.
© 2018 Elsevier B.V. All rights reserved.

1. Introduction of hydrous fluids or melts released from the slab (Pearce, 1982;
Plank, 2005; Plank and Langmuir, 1993).
At subduction zones, oceanic crust and sediment is recycled Similar rock types to those found in fore-arcs also charac-
into the mantle, with important consequences for the crust–mantle terise many Supra-subduction zone (SSZ) ophiolites, and it has
system, including mantle melting and arc magmatism, volatile de- been suggested that such ophiolites represent fragments of fore-
gassing, formation of mantle heterogeneity, and the creation of arc crust formed during subduction initiation (Stern et al., 2012).
new continental crust (Spandler and Pirard, 2013). Despite the im- In particular, boninites are present in many western Pacific fore-
portant role of subduction zones in global geochemical cycles, the arcs and many SSZ ophiolites, but are rare or absent in other
processes by which new subduction zones are initiated are not tectonic settings (Cooper et al., 2010; Dilek and Furnes, 2014;
well understood. Current models suggest that rock exposures in Pearce and Robinson, 2010; Whattam and Stern, 2011).
fore-arcs provide insights into the magmatic evolution of the upper Boninites are rare volcanic rocks with high SiO2 and MgO but
plate at the time of subduction initiation (Stern and Gerya, 2017; low TiO2 contents (>8 wt% MgO, >52 wt% SiO2 and <0.5 wt%
Stern et al., 2012). These models are largely based on detailed
TiO2 (Le Bas, 2000)), and can be divided into high- and low-Ca
studies of the Izu–Bonin–Mariana (IBM) fore-arc, where early ‘fore-
types based on their CaO/Al2 O3 ratios. Boninites have been re-
arc basalts’ are overlain by boninites, island arc tholeiites, and
covered from fore-arcs of the IBM and Tonga island arc systems
finally calc-alkaline lavas representing the transition from fore-arc
(Crawford et al., 1981; Ishizuka et al., 2014; Meffre et al., 2012;
spreading to island arc crust (Arculus et al., 2015; Ishizuka et al.,
Reagan et al., 2010), although the only known actively erupting
2014; Reagan et al., 2010, 2017). These lava successions apparently
boninitic lava volcanoes are located in the Lau back-arc and Tonga
record partial melting of an increasingly depleted mantle above
arc (Cooper et al., 2010; Rubin et al., 2009). The petrogenesis of
the subducting plate yielding magmas with an increasing input
boninites may therefore yield important insights into the processes
of subduction initiation. Many boninites display U-shaped Rare
* Corresponding author. Earth Element (REE) patterns which have been attributed to an ad-
E-mail address: dominic.woelki@fau.de (D. Woelki). dition of a Light Rare Earth Element (LREE) enriched component

https://doi.org/10.1016/j.epsl.2018.06.041
0012-821X/© 2018 Elsevier B.V. All rights reserved.
204 D. Woelki et al. / Earth and Planetary Science Letters 498 (2018) 203–214

Fig. 1. Right panel: Simplified geological map of the Troodos Ophiolite (modified from Osozawa et al., 2012), showing location of the sampled area at Parekklisia. Central
panel: Topographic map (from openstreetmap.org) of the Parekklisia section showing sample localities. Left panel: Variation of Zr/Yb with latitude (younging direction to the
south) showing lack of systematic chemical change with time, and that flows containing depleted tholeiitic glass (DTG) and boninitic glass (BG) are interbedded.

to a refractory peridotite mantle source (Crawford et al., 1989). (Robertson, 2002). The E–W trending Arakapas fault zone (AFZ), lo-
Boninites are enriched in fluid mobile elements (e.g. U, Pb, Rb, Ba), cated at the south-east corner of the Troodos ophiolite (Fig. 1), is
relative to REE and High-Field Strength Elements (HFSE) of simi- thought to represent a fossil transform fault (Simonian and Gass,
lar incompatibility. Boninite magmas are apparently derived from 1978).
highly depleted mantle sources to which a fluid is added, and re- The tectonic setting in which the Troodos ophiolite was formed
quire higher mantle potential temperatures (Crawford et al., 1989; is debated, but the occurrence of a thick sheeted dike complex
Falloon and Danyushevsky, 2000). Boninites have also been re- overlain by pillow lavas with similar geochemical composition in-
ported from many SSZ ophiolites, including the Troodos Ophiolite dicate formation at an oceanic spreading centre (Dilek and Furnes,
(Cameron, 1985; König et al., 2008; Osozawa et al., 2012), the New 2009). However, the enrichment of fluid-soluble elements, e.g.
Caledonia ophiolite (Cluzel et al., 2016) and the Koh and Betts Cove Th, U, Rb, Ba and Pb in Troodos lavas (Osozawa et al., 2012;
ophiolites (Bédard, 1999; Meffre et al., 1996). Boninites from ophi- Pearce and Robinson, 2010; Regelous et al., 2014; Sobolev et al.,
olites provide an ideal opportunity to understand boninite magma 1993), requires formation in an SSZ environment, i.e. in a spread-
genesis because the stratigraphic relationships of boninites with
ing environment close to a subducting slab. The origin of the ophi-
other lava compositions are preserved, and the freshest material
olite in an island arc, as first proposed by Miyashiro (1974), has
for geochemical analysis can be selected.
largely been ruled out on the basis of the composition of the lavas
In this paper, we present major element, trace element and
(including a lack of calc-alkaline lavas), and a lack of island arc
Sr–Nd–Pb isotope data for fresh boninitic and tholeiitic volcanic
deposits (Osozawa et al., 2012; Pearce and Robinson, 2010). Previ-
glasses from a >2 km section through the lava sequence on the
ous studies have proposed that the ophiolite formed in a juvenile
southern margin of the Troodos ophiolite. We discuss the petro-
arc, back-arc, fore-arc, plate edge, or ridge–trench–trench or ridge–
genesis of the lavas and compare their geochemistry to that of
similar rocks from the IBM fore-arc. trench–transform triple junction setting (Osozawa et al., 2012;
Pearce and Robinson, 2010; Regelous et al., 2014).
2. Geological setting The Troodos extrusives have previously been divided into two
main series (Pearce and Robinson, 2010); the upper pillow lavas
The Troodos ophiolite is exposed on the island of Cyprus, in (UPL) and the lower pillow lavas (LPL). The LPL are dominantly
the eastern Mediterranean Sea. The ophiolite covers approximately composed of tholeiitic basalts to rhyolites with an increase of SiO2
3000 km2 in the central part of the island, and preserves the typi- towards higher stratigraphic levels. The LPL have been divided into
cal ophiolite sequence of serpentinized upper mantle peridotites, three distinct compositional groups (Pearce and Robinson, 2010),
overlain by layered and isotropic gabbros, a sheeted dike com- based largely on their MgO and TiO2 contents (Fig. 2). The UPL in-
plex and an extrusive volcanic series (Dilek and Furnes, 2009; clude the most depleted lavas, but it is currently unclear whether
Gass, 1968; Fig. 1). The age of the Troodos ophiolite is 90.3 to boninitic lavas of this series are distributed over the entire ophi-
92.4 ± 0.7 Ma based on U–Pb zircon dating (Mukasa and Lud- olite or whether they are restricted to the southern margin and
den, 1987). The central part of the ophiolite was uplifted by to the vicinity of the Arakapas FZ (e.g. Osozawa et al., 2012;
mantle serpentine diapirism (Schuiling, 2011), but the Troodos Pearce and Robinson, 2010). A detailed study of a section through
ophiolite shows almost no evidence for emplacement deformation the lava sequence at Akaki Canyon on the northern margin of the
D. Woelki et al. / Earth and Planetary Science Letters 498 (2018) 203–214 205

3. Analytical methods

In order to avoid the effects of alteration, which particularly


affects the concentrations of fluid-mobile elements, only fresh
glasses were used in this study. Glass samples were crushed and
hand-picked, and single 1–2 mm sized chips were cleaned, em-
bedded in epoxy resin and polished for electron microprobe and
laser-ablation ICP-MS analyses. The fresh, transparent appearance
of the glasses (Fig. S1) indicates that they have undergone insignif-
icant alteration. Ninety-seven fresh glasses from Parekklisia were
analysed for major elements, and 43 representative samples were
selected for trace element analysis.
Analysis of the major element contents of glasses (SiO2 , TiO2 ,
Al2 O3 , FeOt , MnO, CaO, Na2 O, K2 O, P2 O5 , SO3 and Cl) was car-
ried out on a JEOL JXA-8200 Superprobe electron microprobe
at the GeoZentrum Nordbayern, Friedrich-Alexander Universität
Erlangen–Nürnberg, Germany. The microprobe was operated with
an acceleration voltage of 15 kV, a beam current of 15 nA and a
defocused (10 μm) beam. Counting times were set to 20 s and 10
s for peaks and backgrounds for all elements, except for Cl where
peak counting time was 40 s and background times were set to
20 s. Accuracy and precision were determined using international
glass standards VG-2 and VG-A99 measured during each analytical
session (Beier et al., 2018). Accuracy is better than 4.7% for all el-
ements (except for Cl, SO3 and P2 O5 ) and precision is better than
5.5% (except for Cl, SO3 and P2 O5 ). The average values of 10 spot
analyses on each sample are presented in Table A1. Measurements
of olivine and spinel inclusions are described in detail in Woelki et
al. (2018). Data is presented in supplemental Table A1.
Trace element concentrations of the glasses were determined
by laser ablation inductively coupled plasma mass spectrometry
(LA-ICP-MS) on the same glass fragments analysed for major ele-
ments, using an Agilent 7500i quadrupole mass spectrometer cou-
pled with a New Wave Research UP193FX excimer laser at the
GeoZentrum Nordbayern. The beam diameter was set to 35 μm and
the SiO2 contents, previously determined by electron microprobe,
were used as internal standard. External calibration was performed
Fig. 2. Major element variations in volcanic glasses from Parekklisia (this study), and using the NIST614 glass standard. The international rock standard
Akaki Canyon (Regelous et al., 2014), compared with Troodos lava groups defined by BCR-2g was measured to determine accuracy and reproducibility.
Pearce and Robinson (2010); lower series 1 (ls1), lower series 2 (ls2), lower series
Accuracy was better than 12.6% except for Zn and Cu and repro-
3 (ls3), and upper series (us). Shown for comparison are data for MORB glasses
(Jenner and O’Neill, 2012), subducting sediments (Plank and Langmuir, 1998), and ducibility was better than 9.7% except for As and Cr. Data in Table
IBM fore-arc and arc lavas (Arculus, 1992; Dobson et al., 2006; Hickey-Vargas et al., A1 represent averages of four points analysed on each sample or
2018; Ishizuka et al., 2018, 2011; Reagan et al., 2010; Taylor et al., 1994; Yajima standard.
and Fujimaki, 2001). We measured the Pb, Sr and Nd isotope composition of 15
representative samples from the Parekklisia section, and for com-
parison six glass samples from Akaki Canyon previously analysed
ophiolite only determined tholeiitic lavas of basalt to dacite com- for major and trace elements by Regelous et al. (2014). For Sr,
position (Regelous et al., 2014). Nd and Pb isotope analysis, 50–100 mg of handpicked fresh glass
Most previous geochemical studies of Troodos boninites were chips were leached for 10 min at room temperature in a 25:25:50
carried out on variably altered bulk rock samples (Pearce, 1975; mixture of H2 O2 , 2.5M HCl and MQ H2 O, then washed thor-
Portnyagin et al., 1997), or on fresh glasses from localities scattered oughly with MQ H2 O. The samples were dissolved in 12M HF and
widely over the ophiolite (Golowin et al., 2017; König et al., 2008; 15M HNO3 overnight at 100 ◦ C. The solution was evaporated, 15M
Osozawa et al., 2012; Pearce and Robinson, 2010; Rautenschlein HNO3 added, evaporated and redissolved in 6M HCl, dried down
et al., 1985). Our new samples are from a section through the and finally dissolved in 1M HCl. Lead was separated from the rock
lava pile close to the village of Parekklisia, on the SE margin matrix using 100 μl Eichrom Sr-Spec® resin. The sample solution
of the ophiolite south of the AFZ, where fresh glasses are pre- was loaded onto the resin and rinsed with 1M HCl before col-
served in pillow margins. Ninety-seven fresh glass samples were lecting Pb in 6M HCl. The 1M HCl wash from the Pb columns
collected from an approximately 2200 m long section extending was evaporated and redissolved in 3.5M HNO3 , and Sr and the
from Parekklisia village northwards towards the boundary of the REE separated using Sr-Spec® and TRU-Spec® resin in HNO3 . The
volcanic section with the sheeted dyke complex (Fig. 1). The upper REE fraction was evaporated, redissolved in 0.25M HCl, and Nd
1500 m of the lava pile (beneath the lava–sediment boundary) and was separated from the other REE using LN-spec® resin in 0.25M
the lowermost 1000 m were not sampled due to the lack of fresh HCl. All reagents were Teflon distilled (those for Pb were double-
glass. Individual flow units were between 1–5 m thick, but due distilled), and typical procedural blanks for Pb, Sr and Nd were 30
to alteration, limited petrographic variation and intermittent expo- pg, 200 pg and 80 pg, respectively.
sure, we were not able to identify the thickness or lateral extent Sr and Nd isotope measurements were carried out on a
of all flow units. Thermo-Fisher Triton thermal ionisation multi-collector mass spec-
206 D. Woelki et al. / Earth and Planetary Science Letters 498 (2018) 203–214

Fig. 3. Trace element composition of the Parekklisia glasses normalised to N-MORB (Sun and McDonough, 1989). Green field shows composition of glasses from Akaki Canyon
(Rautenschlein et al., 1985; Regelous et al., 2014). Mean fore-arc basalt (FAB) and boninite (BON) for comparison (data sources as in Fig. 2). Note enrichment in fluid soluble
elements in all Troodos lavas compared to MORB, the relative Nb and LREE enrichment in the most depleted Troodos lavas, and lack of Zr enrichment relative to Sm.
(For interpretation of the colours in the figure(s), the reader is referred to the web version of this article.)

trometer (MC-ICP-MS) in static mode at the GZN. Strontium iso- (Golowin et al., 2017; König et al., 2008; Osozawa et al., 2012;
tope measurements were corrected for instrumental mass fraction- Pearce and Robinson, 2010; Rautenschlein et al., 1985; Regelous et
ation assuming 88 Sr/86 Sr = 0.1194, and corrected for the contribu- al., 2014). The concentrations of TiO2 , SiO2 (Fig. 2), Al2 O3 , Na2 O
tion of 87 Rb to mass 87. Neodymium isotope data were corrected and FeOt increase, and CaO decreases with decreasing MgO. Sixty-
for mass fractionation using a 146 Nd/144 Nd ratio of 0.7219. Samar- two of the Parekklisia lavas are boninites sensu stricto having MgO
ium interference on masses 144, 148, 150 was corrected by mea- >8 wt%, TiO2 <0.5 wt% and SiO2 >52 wt% (Le Bas, 2000) similar
suring 147 Sm. The average value of the NBS987 Sr standard during to some previously analysed Troodos glasses (König et al., 2008;
this study was 0.710259 ± 0.000013 (n = 8), and an in-house Osozawa et al., 2012). The remainder have TiO2 >0.4 wt% and MgO
Nd standard solution gave 143 Nd/144 Nd of 0.511539 ± 0.000006 <8 wt% and are tholeiitic basalts. We thus divided the Parekklisia
(n = 6), equivalent to a value of 0.511841 for the La Jolla stan- samples into two groups – depleted tholeiitic glasses (DTG) and
dard. boninitic glasses (BG), which also differ in their trace element
Lead isotope measurements were carried out on a Thermo- composition. The DTG and BG glasses contain few mm-sized phe-
Fisher Neptune MC-ICP-MS using a 207 Pb–204 Pb double spike to nocrysts of olivine (Fo: 88–91) with small euhedral inclusions of
correct for instrumental mass fractionation. The double spike, with Cr-rich spinel (Cr#: 64–79). Some samples contain rare microphe-
a 207 Pb/204 Pb ratio of 0.813508, was calibrated against a solu- nocrysts of clinopyroxene and orthopyroxene.
tion of the NBS982 equal atom Pb standard. Samples were diluted The Parekklisia glasses correspond broadly to the UPL and LPL
with 2% HNO3 to a concentration of approximately 20 ppb, and Group 3 of Pearce and Robinson (2010). The Parekklisia boni-
an aliquot of this solution spiked in order to obtain a 208 Pb/204 Pb nite glasses belong to the high-Ca (CaO/Al2 O3 >0.75) group of
ratio of about 1. Spiked and unspiked sample solutions were intro- boninites.
duced into the plasma via a Cetac Aridus desolvating nebulizer, and Compared to average mid-ocean ridge basalt (MORB), the
measured in static mode. Interference of 204 Hg on mass 204 was Parekklisia glasses have lower REE concentrations, but are enriched
corrected by monitoring 202 Hg. An exponential fractionation cor- in La–Sm relative to the middle and Heavy Rare Earth Elements
rection was applied offline using the iterative method of Compston (HREE; Fig. 3). The Parekklisia glasses are enriched in fluid-mobile
and Oversby (1969), the correction was typically 8 permil per elements (Cs, Rb, Ba, U, Pb, Sr) relative to REE and HFSE of similar
amu. Twenty measurements of the NBS981 Pb isotope standard incompatibility (Fig. 3), which is a characteristic of subduction-
(measured as an unknown) over the course of this study gave related magmas. Trace element data for the boninitic samples over-
206
Pb/204 Pb, 207 Pb/204 Pb, 208 Pb/204 Pb ratios of 16.9410 ± 0.0020, lap with published trace element data for Troodos boninites (König
15.4993 ± 0.0019 and 36.7244 ± 0.0046 respectively. Initial Sr, Nd et al., 2008; Osozawa et al., 2012). Compared to the tholeiitic
and Pb isotope ratios were calculated using parent–daughter ratios glasses from the Akaki Canyon section (Rautenschlein et al., 1985;
determined by LA-ICP-MS, and assuming an age of 92 Ma for all Regelous et al., 2014), the Parekklisia lavas extend to lower Sm/Yb
samples. and Zr/Yb ratios of 0.2 to 0.9 and 4.9 to 17.9 respectively (Fig. 4).
The BGs have the lowest Zr/Yb <11 whereas the DTGs have Zr/Yb
4. Results of 10 to 20 and overlap with Akaki Canyon tholeiites and the de-
pleted end of the MORB array (Fig. 4). Ce/Pb ratios (1.0 to 8.8)
The volcanic glasses from Parekklisia have SiO2 contents of 51 of the Parekklisia glasses are lower than those of most oceanic
to 55 wt% (Fig. 2) and range in composition from basalt to basaltic basalts (20–30), and correlate positively with Zr/Yb (Fig. 4) and
andesite (Le Bas et al., 1986). The MgO contents of Parekklisia negatively with Th/Nd and Nb/La, extending to higher Th/Nd com-
glasses range from 7.1 to 9.7 wt% (Fig. 2), overlapping the high- pared to MORB (Fig. 5). The Parekklisia glasses have Ba/Th ratios
MgO end of the Troodos glasses analysed in previous studies between 83 and 208, which is higher than Akaki glasses and most
D. Woelki et al. / Earth and Planetary Science Letters 498 (2018) 203–214 207

Fig. 5. Negative correlation of Ce/Pb with (A) Th/Nd and (B) Nb/La for Parekklisia
and Akaki lavas, extending from the depleted end of the MORB range, towards a
sediment melt component with higher Nb/La and Th/Nd. Note contrasting positive
array defined by IBM lavas in (B). Data sources as in Figs. 2 and 4.

MORB, but lower than many island arc lavas (>300). The lavas
with the lowest Zr/Yb and Sm/Yb ratios also have the lowest Ce/Pb
and highest Th/Nd (Figs. 4, 5).
Age-corrected 87 Sr/86 Sr and 144 Nd/143 Nd ratios of Parekklisia
glasses range between 0.70346 to 0.70439 and 0.51283 to 0.51291,
respectively. Glasses from Akaki Canyon have 87 Sr/86 Sr ratios be-
tween 0.70341 and 0.70447, overlapping with glass data from
Rautenschlein et al. (1985) from the same location. Most of our
samples have lower 87 Sr/86 Sr than the boninitic–tholeiitic Troo-
dos lavas analysed by König et al. (2010). For the dataset as
a whole, 87 Sr/86 Sr correlates positively with Th/Nd and nega-
tively with Ce/Pb (Fig. 6). Age-corrected Pb isotope ratios of
206
Pb/204 Pb, 207 Pb/204 Pb and 208 Pb/204 Pb from Parekklisia range
Fig. 4. Variations of (A) Sm/Yb, (B) Nb/Yb and (C) Ce/Pb with Zr/Yb. Assuming
that Zr, Sm and Yb are immobile in fluids, the low Zr/Yb and Sm/Yb ratios in between 18.555 to 18.908 and 15.445 to 15.605 and 38.117 to
Troodos lavas result from melting a mantle source more depleted than that of 38.787, respectively, and overlap with previously published Pb iso-
MORB. Note lower Zr/Yb in Troodos boninites compared to IBM boninites. Lower tope data for Troodos tholeiites and boninites (Fig. 7). The most ra-
Ce/Pb and higher Nb/Yb ratios in the most depleted Troodos lavas with low
diogenic Pb isotope compositions overlap with the global sediment
Zr/Yb indicate the addition of Pb and Nb from fluids and melts. Note contrast-
ing trends in these diagrams defined by Troodos and IBM lavas. Data sources; FAB values of Plank and Langmuir (1998) (Fig. 7). The BGs have higher
87
(Hickey-Vargas et al., 2018; Ishizuka et al., 2018, 2011; Reagan et al., 2010); tran- Sr/86 Sr and Pb isotope ratios compared to the DTGs (Figs. 6, 7).
sitional IBM (Reagan et al., 2010); boninite IBM as in Fig. 2; arc IBM (Tollstrup et The Akaki glasses have higher 207 Pb/204 Pb and 208 Pb/204 Pb for a
al., 2010); Troodos tholeiites and boninites (Golowin et al., 2017; König et al., 2008;
given 206 Pb/204 Pb compared to the Parekklisia glasses. Lead isotope
Osozawa et al., 2012).
ratios are positively correlated with 87 Sr/86 Sr, Th/Nd and negatively
correlated with Ce/Pb (Fig. 7).
208 D. Woelki et al. / Earth and Planetary Science Letters 498 (2018) 203–214

Fig. 6. Variations of (A) Ce/Pb and (B) Th/Nd with 87 Sr/86 Sr. For Parekklisia glasses,
87
Sr/86 Sr is positively correlated with Th/Nd, and negatively correlated with Ce/Pb,
indicating that the slab component with low Ce/Pb and high Th/Nd has high
87
Sr/86 Sr similar to the subducting sediment averages of Plank and Langmuir
(1998).

5. Discussion

5.1. Stratigraphic relation between the DTGs and BGs, and other Troodos
lavas

Previous studies suggested that Troodos boninitic lavas may


have been erupted after the tholeiitic lavas (Cameron, 1985; König
et al., 2010; Portnyagin et al., 1997; Rautenschlein et al., 1985).
However, our detailed stratigraphic sampling at the Parekklisia sec-
tion shows that the DTGs and BGs are interbedded (Fig. 1), with no
systematic change in composition with stratigraphic age. Our field
observations indicate that the DTG and BG occur together in single
outcrops and were erupted contemporaneously. This implies that Fig. 7. Variations of 206 Pb/204 Pb with (A) 207 Pb/204 Pb, (B) 208 Pb/204 Pb and (C) Ce/Pb.
the UPL and LPL lava groups of Pearce and Robinson (2010) do not The low Ce/Pb slab component inferred from the negative correlation in (C) has
represent distinct stratigraphic units in this area. The Parekklisia a 206 Pb/204 Pb of about 19, and lies within the field of subducting sediments in
(A) and (B). Akaki and Parekklisia glasses define different arrays in (A) and (B),
section contrasts with the lava sequence in Akaki Canyon where suggesting that the Pb isotope composition of the slab component differed between
only tholeiitic lavas were sampled (Regelous et al., 2014). Our data, the northern and southern margins of Troodos. Black arrows indicate age correction
together with published Troodos glass and whole rock data (König for IBM lavas. Data sources as in Fig. 2 and 4.
et al., 2008; Osozawa et al., 2012; Pearce and Robinson, 2010;
Rautenschlein et al., 1985; Regelous et al., 2014) suggest that
boninites may be restricted to the southern margin of the ophi- dos lava stratigraphy also differs from that of the IBM fore-arc
olite. In neither of the (incomplete) sections through the Troodos crust, as discussed further in section 5.5.
lava pile at Akaki and Parekklisia are all of the geochemical lava Crawford et al. (1989) proposed that boninite magmatism com-
groups defined by Pearce and Robinson (2010) present. The Troo- monly starts at late stages or after oceanic crust production and
D. Woelki et al. / Earth and Planetary Science Letters 498 (2018) 203–214 209

overlies tholeiitic magmatism. In the Izu–Bonin fore-arc and the


Betts Cove ophiolite (Bédard et al., 1998; Ishizuka et al., 2006),
tholeiites and boninites appear to have erupted separately. On the
other hand, interbedded tholeiitic and boninitic lavas have been
found in the Izu–Bonin fore-arc region (ODP Leg 125, Site 786;
Haraguchi and Ishii, 2007), and the Mariana fore-arc at DSDP Leg
60, Site 458 (Wood, 1981) and on Guam Island (Hickey-Vargas and
Reagan, 1987). The progression of melting processes, and the evo-
lution of mantle and lava compositions in fore-arcs may therefore
be less systematic than suggested.

5.2. Limited effects of fractional crystallisation and crustal assimilation

The BG samples display well-defined correlations of most ma-


jor elements with MgO whereas the DTGs have a more restricted
range in major and trace element composition (Figs. 2, 3). Both
groups lie at the high-MgO end of the Troodos glasses analysed
by Pearce and Robinson (2010). Small but significant differences
in TiO2 (Fig. 2), Na2 O and CaO at a given MgO between the DTGs
and BGs rule out a common parental melt for both groups. This
is supported by the differences in incompatible element and iso- Fig. 8. Ti8 and Fe8 (TiO2 and FeOt contents of lavas (>5 wt% MgO) corrected for
tope ratios, for example Ce/Pb varies by an order of magnitude. crystal fractionation to MgO = 8 wt% using the observed liquid lines of descent)
of the Parekklisia and Akaki glasses. Shown for comparison are data for Mariana
The Ce/Pb ratios correlate positively with immobile (alteration-
back-arc lavas (Gribble et al., 1998), global MORB (Gale et al., 2013) and fore-arc
resistant) trace element ratios such as Sm/Yb and Zr/Yb, indicating basalts (Reagan et al., 2010; Ishizuka et al., 2011). For Akaki lavas, Fe8 is correlated
that the range in trace element and isotope compositions in the positively with Ti8, and overlaps with back-arc lavas and some IBM fore-arc lavas.
lavas result from mantle melting processes or mantle source het- The more depleted Parekklisia lavas have lower Ti8 for a given Fe8, but higher Ti8
erogeneity. The TiO2 , SiO2 (Fig. 2), Al2 O3 , Na2 O and FeOt contents than IBM fore-arc boninites (data sources as in Fig. 2). Grey symbols for melting
conditions are from Langmuir et al. (2006). Low Fe8 and Ti8 require a wet, relatively
increase, and CaO decreases with decreasing MgO within the BG hot and depleted mantle source.
glasses. Both groups display a negative correlation of Ni with de-
creasing MgO, with slightly higher Ni in the DTGs. The decreasing
Ni, CaO and Sc concentrations with decreasing MgO indicate frac-
tional crystallisation of olivine and clinopyroxene, but high MgO
these lavas to have been even more depleted. This extreme mantle
contents up to 9.7 wt% exclude substantial changes due to frac-
depletion is also evident in the decrease of Ti8 and Na8 (fraction-
tional crystallisation.
ation corrected to 8 wt% MgO) from the DTGs to the BGs samples.
The Troodos glasses generally have higher Cl/K2 O (0.10–0.37)
The lower Ti8, Na8 and Fe8 (Fig. 8) of back-arc and fore-arc vol-
compared to mantle values (average of 0.09), and overlap with val-
canic rocks has been attributed to source depletion by previous
ues for MORB from medium- to fast-spreading ridges (Kendrick et
melt extraction and melting under hydrous conditions (Langmuir
al., 2014; Kovalenko et al., 2006; Michael and Schilling, 1989). High
et al., 2006). The Parekklisia and Akaki samples form an array
Cl/K2 O ratios in MORB are thought to result from assimilation of
extending towards lower values of Ti8 (Fig. 8) and Na8 than all
altered crustal material, but unlike the Akaki Canyon lavas, Cl/K2 O
back-arc lavas, suggesting melting of a highly depleted, hydrous
does not increase with decreasing MgO in the Parekklisia lavas,
mantle source. A highly depleted mantle source is also evident
indicating that they have not experienced extensive assimilation
from the compositions of the Cr-rich spinel inclusions in high-Fo
of hydrothermally altered crust during ascent. Rather, high Cl/K2 O
olivines (Kamenetsky et al., 2001).
may reflect the influence of a slab-derived component affecting the
mantle source of the Parekklisia magmas, as also observed in Lau In order to determine the degree of source depletion recorded
back-arc basin glasses, which have Cl/K2 O ratios of 0.1 to 2.7 (Kent by the DTGs and BGs, we used a similar approach to König et
et al., 2002). al. (2010). The trace element composition of the subduction zone
modified source mantle prior to the melting event which created
5.3. A highly depleted mantle source for the Troodos boninites the DTGs and BGs was calculated using the bulk partition coef-
ficients for mantle melting under hydrous (or anhydrous; Th and
The MREEs and HREEs together with Zr and Hf are moderately U) conditions (McDade et al., 2003b). We assumed that Troodos
incompatible and relatively immobile in subduction components, lavas represent 20% partial melts, which is a typical value for
and can be used as tracers of partial melting conditions in subduc- subduction-related lavas (König et al., 2010; Tatsumi and Eggins,
tion zones, including the degree of melting and source depletion 1995). If the Parekklisia lavas were derived by a smaller degree of
(Pearce and Peate, 1995; Woodhead et al., 2001). The Parekklisia melting, then the degree of source depletion calculated is a mini-
glasses have Sm/Yb and Zr/Yb ratios which overlap with the de- mum value. The HREE (i.e. Y, Er, Yb and Lu) of the calculated man-
pleted (lower) end of the MORB array (Fig. 4) and with Akaki tle source can then be used to estimate the degree of source de-
Canyon lavas, but extend to more depleted values. Lower Sm/Yb pletion, and the concentrations of more incompatible, fluid-soluble
and Zr/Yb in the BGs indicate higher degrees of melt extraction elements in the source prior to fluid addition. We used the bulk
and mantle depletion compared to the DTGs. The low Zr/Yb ratios partition coefficients of McDade et al. (2003b) and the Depleted
of BGs cannot be generated by melting of a MORB-type mantle MORB Mantle (DMM) composition of Workman and Hart (2005)
source, but require melting of a source previously depleted by to calculate the degree of mantle wedge depletion. Comparing the
melt extraction. The glasses from the Akaki Canyon section were composition of the pre-subduction mantle with that of the mantle
derived from a mantle source depleted by 1–2% partial melting source that melted to form the BGs and DTGs, we can then quan-
compared to normal spreading ridges (Regelous et al., 2014), and tify the contribution of the subduction component in both groups,
the lower Zr/Yb ratios of the DTGs and BGs require the sources of as discussed in section 5.4.
210 D. Woelki et al. / Earth and Planetary Science Letters 498 (2018) 203–214

Based on our calculations a DMM (Workman and Hart, 2005) less affected. Although the estimates of subducting slab contri-
with lherzolite mineral mode (56.5% olivine + 28.7% orthopyrox- bution using this method may not be very precise, they give an
ene + 14.1% clinopyroxene + 4% spinel) (McDade et al., 2003a) indication of the composition of the slab component. The calcu-
approximately 6% batch melting is required to explain the low lated contribution of subducted materials to the budget of most
HREE abundance of the subduction modified mantle wedge of the incompatible elements, in μg per g of mantle, was broadly similar
DTGs (Sample: IEGZN0105; Fig. S2). To explain the HREE abun- for sources of both the BGs and DTGs (Fig. S3). Lower Ce/Pb ratios
dance of the subduction modified mantle wedge of the BG (Sam- in the BGs compared to the DTGs therefore result from the greater
ple: IEGZN0111) approximately 9% batch melting of the same man- depletion of the mantle source of the former, rather than a greater
tle source as the DTG would be required. The interbedded occur- slab flux.
rence of the DTGs and BGs requires melting of variably depleted The BGs and DTGs lavas with the lowest Ce/Pb ratios also have
mantle on a scale smaller than the entire region of melt genera- the highest 87 Sr/86 Sr and 207 Pb/204 Pb ratios (Figs. 6, 7), consis-
tion. tent with the involvement of subducted materials to the Sr and
We note that the mantle source of the Parekklisia lavas was Pb budget of these lavas. The negative correlations between Ce/Pb
considerably more depleted (refractory) than that of the Akaki and Pb isotope ratios indicate that the subduction component had
206
Canyon lavas on the northern margin of the ophiolite. Experiments Pb/204 Pb >18.72, 207 Pb/204 Pb >15.62 and 208 Pb/204 Pb >38.88.
suggest that boninite magmas may require unusually hot man- Both the Parekklisia and the Akaki Canyon lavas extend to higher
207
tle (Crawford et al., 1989; Sobolev et al., 1993), but variations of Pb/204 Pb and 208 Pb/204 Pb isotope ratios for a given 206 Pb/204 Pb
mantle temperature on a scale of <40 km beneath the Troodos than most mid-ocean ridge basalts, similar to previously published
ophiolite appear unlikely because the contemporaneous eruption data for Troodos lavas (Fig. 7). The most radiogenic Pb isotope
of DTGs and BGs indicates that both lava groups likely had sim- compositions overlap with global subducting sediments (Fig. 7),
ilar temperatures. Alternatively, variable initial levels of depletion indicating that the slab component affecting the Parekklisia lavas
of the mantle, together with variable input of a slab component originated at least in part from subducting sediment. In Pb isotope
beneath the northern and southern margins of the ophiolite can space, the Akaki and Parekklisia glasses define separate arrays,
explain much of the trace element variation in Troodos lavas, as indicating that the composition of the sediment component was
discussed below. different in these two regions (Fig. 7).
Trace element ratios can be used to determine whether the
5.4. Addition of the subduction component: fluids or melts from slab component was a hydrous fluid or a melt derived from sedi-
sediments ment. Parekklisia lavas have Ce/Pb ratios that are lower, and Sr/Nd
(Fig. 9) ratios higher than average subducting sediments or sed-
The Parekklisia lavas have trace element compositions charac- iment melts, indicating that a fluid phase was at least partly
teristic of subduction influenced magmas (Pearce, 1982), including responsible for Sr and Pb enrichment in the source of these
enrichment of Pb, U, Rb, Ba and other fluid-mobile elements rela- lavas. In contrast, both Th and Nd are generally immobile in low-
tive to REE and HFSE of similar incompatibility (Fig. 3). The Ce/Pb temperature fluids (Brenan et al., 1995) and thus the Th/Nd ra-
ratios of the Parekklisia lavas are similar to those of many island- tio of the lavas is mainly controlled by mantle composition and
arc lavas, and lower than those of most lavas from back-arc basins melts from subducting sediment, rather than a fluid phase (Plank
(Regelous et al., 2014). The enrichment of fluid-soluble elements and Langmuir, 1993). Sediments have higher Th/Nd than the up-
in subduction-related lavas is generally attributed to the addition per mantle, and experiments under relevant P–T conditions show
of these elements by hydrous melts or fluids, derived from either that due to the higher incompatibility of Th compared to Nd
subducted altered oceanic crust or sediments (Brenan et al., 1995; during melting, Th/Nd is higher in sediment melts (Johnson and
Kessel et al., 2005; Plank, 2005). Plank, 2000). The calculated compositions of melts from global
Trace element ratios indicate that the subduction component subducting sediment of Plank and Langmuir (1998), using the ex-
input is highest in the BGs, lower in the DTGs, and lowest in the perimental determined partition coefficients of Johnson and Plank
Akaki glasses. The Ce/Pb ratios of ∼1 in some Parekklisia lavas (2000) yield Th/Nd ratios of >0.25 for sediment melt and Th/Nd
are as low as in many arc lavas indicating that most of the Pb <0.15 for sediment fluid (Fig. 9). The BGs have elevated Th/Nd
is slab-derived. Assuming an initial Ce/Pb ratio of a DMM of ∼30 ratios up to 0.28 (Fig. 9) indicating the influence of a sediment
(Workman and Hart, 2005), then >96% of the Pb in the sample melt. Re-enrichment of the source of the Troodos lavas by melts is
with the lowest Ce/Pb (Sample: IEGZN0176) was slab-derived, and also evident in their relatively high Nb contents. Niobium, Zr and
this is a minimum estimate if Ce is also in part derived from the Yb are immobile in slab-derived fluids (Pearce and Peate, 1995)
slab. The boninitic Parekklisia glasses with the lowest Zr/Yb (i.e. and so Nb/Yb and Zr/Yb ratios in the Parekklisia glasses would
the lavas from the most depleted source) also have the lowest be expected to lie on an extension of the linear array formed
Ce/Pb ratios (i.e. the highest relative subduction component con- by MORB, as neither ratio is affected by fluid addition. However,
tribution), whereas the lavas of Akaki Canyon have higher Zr/Yb Troodos lavas have higher Nb/Yb for a given Zr/Yb, compared to
and higher Ce/Pb (Fig. 4). The relative subduction component input oceanic basalts (Fig. 4). The negative correlation of Nb/Yb with
thus increases with increasing source depletion. This could reflect Zr/Yb (Fig. 4) is most consistent with addition of Nb by a melt
the fact that the more depleted source of the BGs was more sen- phase, following the depletion event(s) that led to low Zr/Yb. In
sitive to subduction component addition, or a greater subduction contrast, later depletion of an initially fluid- and melt-enriched
flux to the source of the BGs, as discussed below. source would produce positive arrays in Fig. 4, parallel to, but dis-
We can quantify the contribution of the slab-derived compo- placed above the MORB array. The enrichment of Nb as well as Th
nent to the total budget of other trace elements by comparing by a melt component is also evident from the negative correlation
the difference in trace element content of the mantle sources of Nb/La and Th/Nd with Ce/Pb (Fig. 5).
that melted to form the Parekklisia lavas, and the composition of The influence of both hydrous fluids and melts can be seen
the mantle immediately following depletion as calculated in sec- from the distribution of the Troodos data in Fig. 9. Addition of
tion 5.3. Highly incompatible and fluid-mobile trace element abun- fluid decreases the 143 Nd/144 Nd ratio of the mantle, with little ef-
dances in the DTGs and BGs including Th (97–99%), Nb (97–99%), fect on Th/Nd (Fig. 9). Subsequent mixing of a sediment melt leads
La (85–98%) and Sr (92–94%) are almost completely controlled by to the negative correlation of Th/Nd and 143 Nd/144 Nd observed
the slab derived material, whereas Sm (27–34%) and Y (5–6%) are in Troodos lavas towards the estimated composition of sediment
D. Woelki et al. / Earth and Planetary Science Letters 498 (2018) 203–214 211

5.5. Was the Troodos ophiolite formed during subduction initiation in


a fore-arc setting?

The precise tectonic setting of the Troodos ophiolite is still


debated but there is the consensus that, like other SSZ ophio-
lites, it probably formed at a spreading centre in the vicinity of
a subduction zone (Pearce and Robinson, 2010). The Akaki Canyon
lavas have some similarities to lavas from back-arc spreading cen-
tres (Regelous et al., 2014), whereas boninites are predominantly
found in fore-arc, and less commonly in back-arc and arc loca-
tions (Cooper et al., 2010; Rubin et al., 2009). The Tonga and
Izu–Bonin–Marianas (IBM) fore-arc crust is thought to have formed
at short-lived spreading centres shortly after subduction initia-
tion, and includes boninites as well as other lavas with a sub-
duction input (Ishizuka et al., 2006, 2011; Reagan et al., 2010).
Several authors have suggested that SSZ ophiolites, including Troo-
dos, formed in a fore-arc setting as a result of spreading during
subduction initiation (Dilek and Furnes, 2009; Dilek and Thy, 2009;
Pearce and Robinson, 2010; Stern and Bloomer, 1992; Stern et al.,
2012). Recent sampling of the IBM fore-arc region revealed an ap-
parently progressive change in magma compositions, with increas-
ing source depletion and subduction zone influence with decreas-
ing stratigraphic age. The oldest lavas, so-called ‘fore-arc basalts’
(FAB) have chemical affinities to MORB, with variable depletion of
the LREE, and TiO2 contents >1 wt%. FAB magmatism is followed
by lavas that are more depleted in HFSE, but enriched in the LREE
and fluid-mobile elements and have compositions intermediate be-
tween boninites and FAB (Ishizuka et al., 2006; Reagan et al., 2010;
Whattam and Stern, 2011). Boninites normally overlie the interme-
diate lavas and FAB.
We compiled major and trace element data for boninites, FAB
and transitional lavas from the IBM fore-arc to compare with Troo-
dos lavas. Whole-rock FAB from the IBM fore-arc show slight en-
richment in fluid-mobile elements including Rb, Ba and Pb (Reagan
et al., 2010; Fig. 3), although analysed fresh glasses of FAB show al-
most no enrichment in fluid-mobile elements (Reagan et al., 2010).
Fig. 9. Variations of (A) Sr/Nd and (B) 143 Nd/144 Nd with Th/Nd. Increasing Sr/Nd
with increasing Th/Nd in the DTG samples could result from a contribution of FAB glasses thus show similar degrees of source depletion (e.g. low
hydrous fluid and melt from subducted sediments. Increasing Th/Nd at constant Zr/Yb and Sm/Yb) as the Akaki lavas (Regelous et al., 2014), but
Sr/Nd in the BG results from addition of a sediment melt with high Th/Nd. In (B), are less enriched in fluid-mobile elements. The FAB and transi-
DTGs have lower 143 Nd/144 Nd than DMM (Workman and Hart, 2005) consistent tional lavas show enrichments of fluid mobile elements towards
with addition of a hydrous fluid from sediment. Negative correlation of Th/Nd with
143 the composition of the IBM arc lavas, which are dominated by
Nd/144 Nd for the BGs indicates the contribution of a hydrous melt of sediment.
See text for discussion. Fluid and melt composition of GLOSS (Plank and Langmuir, fluids from altered oceanic crust (AOC), whereas in Troodos, the
1998) were calculated using the experimental determined partition coefficients of Akaki lavas and the DTGs extend towards higher values of Th/Nd
Johnson and Plank (2000). Additional Troodos data from König et al. (2008) and and are overall more enriched in Th (Fig. 5). Lavas with the ex-
Osozawa et al. (2012).
act trace-element composition of FABs are not found in the Akaki
and Parekklisia sections, although some Akaki lavas partly over-
melts (Fig. 9). The high Sr/Nd ratios of the Troodos lavas also indi- lap in composition with FABs from the IBM fore-arc. However,
cate the influence of a hydrous fluid, and for the DTGs and Akaki the compositions of boninites from the IBM region and Troodos
lavas, Th/Nd ratios increase with increasing Sr/Nd, indicating a con- differ in several important respects. Fractionation corrected Fe8
current small enrichment by a hydrous melt (Fig. 9). In contrast, and Ti8 values show distinct differences in major elements be-
Troodos boninites form a horizontal array towards higher Th/Nd at tween Troodos glasses in the north and south. FAB, boninites and
constant Sr/Nd (∼30) indicating the influence of a melt component arc lavas from the IBM region form an array towards lower Ti8
as discussed above. and Fe8 (Fig. 8), suggesting increasing water content and degree
The initially highly depleted mantle source of the Parekklisia of depletion of the mantle source. Parekklisia lavas overlap in
lavas was thus enriched by both fluids and melts, at different Fe8 with IBM arc lavas, boninites and with Akaki Canyon lavas
times, prior to the melting event beneath the former spreading (Fig. 8). Akaki Canyon lavas tend to higher Ti8 similar to Mari-
ridge. Our data confirm previous conclusions, that the source of ana back-arc basalts (Fig. 8). Compared to the boninitic glasses of
Troodos lavas was affected by both hydrous fluids and melts from Parekklisia, the boninites of the IBM fore-arc have higher Sm/Yb
subducting materials, including sediment (Fonseca et al., 2017; and Zr/Yb ratios, overlapping in Sm/Yb with Akaki Canyon lavas
Golowin et al., 2017; König et al., 2008; Portnyagin et al., 1997; and the DTGs. The Zr/Yb ratios are distinctively higher in IBM
Rautenschlein et al., 1985; Regelous et al., 2014). Re-enrichment boninites and overlap with enriched MORB (Sun and McDonough,
of a highly depleted mantle source by small degree melts derived 1989) and are displaced from the MORB array in Fig. 4, indicat-
from subducting materials has also been reported for boninites ing an enrichment of Zr. This enrichment in Zr relative to Sm is
from other localities worldwide (Crawford et al., 1981; Hickey and also present in some IBM transitional lavas. Lead isotope ratios of
Frey, 1982; König et al., 2010; Reagan et al., 2010; Yajima and Fu- the IBM boninites and FAB overlap in part with the Parekklisia
jimaki, 2001). DTGs and BGs indicating a role for subducted sediment in both
212 D. Woelki et al. / Earth and Planetary Science Letters 498 (2018) 203–214

mapping out the geographic chemical variations within the Troo-


dos lava pile with data from additional sections.

6. Conclusions

Fresh volcanic glasses from the Parekklisia section on the south-


ern margin of the Troodos ophiolite can be divided into two dis-
tinct groups (boninites and depleted tholeiites) on the basis of
their geochemistry. Boninites were derived from a mantle source
depleted by about 9% previous melt extraction, whereas the source
of the tholeiites was less depleted (6%). The sources of all lavas
were subsequently enriched by both fluids and small-degree melts,
derived in part from subducted sediment. Combined with previ-
ously published data for Troodos volcanic glasses, the Parekklisia
lavas form arrays in trace element diagrams that differ from those
for IBM fore-arc lavas. The differences in the compositions of
boninites from Troodos and IBM, the lack of ‘fore-arc basalt’ com-
positions in Troodos, and the lack of any systematic chemical evo-
lution with time within the Troodos lava pile, suggests that this
ophiolite was not formed as a result of subduction initiation.

Acknowledgements

This work was funded by the Deutsche Forschungsgemein-


schaft, grant RE 3020/11-1. We acknowledge the Smithsonian In-
stitution for providing electron microprobe standards. We thank
Helene Brätz and Reiner Klemd for their support during the LA-
ICP-MS analysis, and Patrick Hoyer for help with sample collection
and isotope analysis.

Appendix A. Supplementary material

Fig. 10. Modified schematic stratigraphic section through Bonin fore-arc crust Supplementary material related to this article can be found on-
(Ishizuka et al., 2011) compared to the lava stratigraphy observed in the Akaki line at https://doi.org/10.1016/j.epsl.2018.06.041.
Canyon (Regelous et al., 2014) and Parekklisia sections (this study). The Troodos
stratigraphy lacks early FAB and later arc tholeiites and calc-alkaline lavas. References

Arculus, R., 1992. Major-element geochemistry of ashes from Sites 782, 784, and 786
settings, but the IBM lavas extend to less radiogenic 207 Pb/204 Pb
in the Bonin Forearc. In: Proceedings of the Ocean Drilling Program, Scientific
and 208 Pb/204 Pb ratios (Fig. 7). Akaki lavas have similar 206 Pb/204 Pb Results. College Station, TX, USA, vol. 125, pp. 277–292.
and 207 Pb/204 Pb ratios to IBM arc lavas (Fig. 7). The enrichment of Arculus, R.J., Ishizuka, O., Bogus, K.A., Gurnis, M., Hickey-Vargas, R., Aljahdali, M.H.,
fluid-mobile elements (including Pb) overlaps in the Troodos and Bandini-Maeder, A.N., Barth, A.P., Brandl, P.A., Drab, L., 2015. A record of spon-
taneous subduction initiation in the Izu–Bonin–Mariana arc. Nat. Geosci. 8 (9),
IBM boninites, whereas Nb/La ratios of IBM boninites are much
728–733.
lower, extending towards the composition of IBM arc lavas (Fig. 5). Bédard, J., 1999. Petrogenesis of boninites from the Betts Cove ophiolite, Newfound-
The Troodos BGs are unique among boninites worldwide in their land, Canada: identification of subducted source components. J. Petrol. 40 (12),
extremely low Zr/Yb (source depletion), coupled with both low 1853–1889.
Ce/Pb (fluid enrichment) and high Nb/Yb (melt enrichment) ra- Bédard, J., Lauziere, K., Tremblay, A., Sangster, A., 1998. Evidence for forearc
seafloor-spreading from the Betts Cove ophiolite, Newfoundland: oceanic crust
tios. In contrast, IBM boninites are enriched in Zr but not Nb, of boninitic affinity. Tectonophysics 284 (3–4), 233–245.
and the high Zr/Sm and Zr/Yb in IBM boninites have been at- Beier, C., Brandl, P.A., Lima, S.M., Haase, K.M., 2018. Tectonic control on the genesis
tributed to residual amphibole and titanite (Hickey and Frey, 1982; of magmas in the New Hebrides arc (Vanuatu). Lithos 312, 290–307.
König et al., 2010). Brenan, J., Shaw, H., Ryerson, F., Phinney, D., 1995. Mineral-aqueous fluid partition-
ing of trace elements at 900 ◦ C and 2.0 GPa: constraints on the trace element
In summary, Troodos DTGs, BGs and Akaki glasses have dif-
chemistry of mantle and deep crustal fluids. Geochim. Cosmochim. Acta 59 (16),
ferent compositions to IBM fore-arc lavas, and the geochemical 3331–3350.
stratigraphy of Troodos lavas also differs from that proposed for Cameron, W., 1985. Petrology and origin of primitive lavas from the Troodos ophio-
the IBM fore-arc crust (Fig. 10). In Troodos there are no lavas re- lite, Cyprus. Contrib. Mineral. Petrol. 89 (2–3), 239–255.
Cluzel, D., Ulrich, M., Jourdan, F., Meffre, S., Paquette, J.-L., Audet, M.-A., Secchiari,
sembling FAB, and Troodos and IBM fore-arc lavas define different
A., Maurizot, P., 2016. Early Eocene clinoenstatite boninite and boninite-series
arrays in incompatible trace element diagrams, indicating a differ- dikes of the ophiolite of New Caledonia; a witness of slab-derived enrichment
ent depletion and slab-enrichment history. The Troodos boninitic of the mantle wedge in a nascent volcanic arc. Lithos 260, 429–442.
lavas were derived from a more depleted source, suggesting that Compston, W., Oversby, V., 1969. Lead isotopic analysis using a double spike. J. Geo-
a subduction zone already existed prior to spreading. We sug- phys. Res. 74 (17), 4338–4348.
Cooper, L.B., Plank, T., Arculus, R.J., Hauri, E.H., Hall, P.S., Parman, S.W., 2010. High-Ca
gest that the apparent geographical variations in Troodos lavas, boninites from the active Tonga Arc. J. Geophys. Res., Solid Earth 115 (B10).
from back-arc like lavas on the northern margin of the ophio- Crawford, A.J., Beccaluva, L., Serri, G., 1981. Tectono-magmatic evolution of the West
lite, to highly depleted boninitic lavas with a stronger subduction Philippine-Mariana region and the origin of boninites. Earth Planet. Sci. Lett. 54
component in the south, could be explained if the Troodos ophi- (2), 346–356.
Crawford, A.J., Falloon, T., Green, D., 1989. Classification, petrogenesis and tectonic
olite formed at a ridge–trench–trench/transform triple junction
setting of boninites. In: Boninites and Related Rocks, pp. 1–49.
(Regelous et al., 2014), with the trench located south of the present Dilek, Y., Furnes, H., 2009. Structure and geochemistry of Tethyan ophiolites and
southern margin of Troodos. This hypothesis could be tested by their petrogenesis in subduction rollback systems. Lithos 113 (1), 1–20.
D. Woelki et al. / Earth and Planetary Science Letters 498 (2018) 203–214 213

Dilek, Y., Furnes, H., 2014. Ophiolites and their origins. Elements 10 (2), 93–100. Le Bas, M., 2000. IUGS reclassification of the high-Mg and picritic volcanic rocks.
Dilek, Y., Thy, P., 2009. Island arc tholeiite to boninitic melt evolution of the Creta- J. Petrol. 41 (10), 1467–1470.
ceous Kizildag (Turkey) ophiolite: model for multi-stage early arc–forearc mag- Le Bas, M.J., Le Maitre, R., Streckeisen, A., Zanettin, B., 1986. A chemical classifica-
matism in Tethyan subduction factories. Lithos 113 (1), 68–87. tion of volcanic rocks based on the total alkali-silica diagram. J. Petrol. 27 (3),
Dobson, P.F., Blank, J.G., Maruyama, S., Liou, J., 2006. Petrology and geochemistry 745–750.
of boninite-series volcanic rocks, Chichi-Jima, Bonin Islands, Japan. Int. Geol. McDade, P., Blundy, J.D., Wood, B.J., 2003a. Trace element partitioning between man-
Rev. 48 (8), 669–701. tle wedge peridotite and hydrous MgO-rich melt. Am. Mineral. 88 (11–12),
Falloon, T.J., Danyushevsky, L.V., 2000. Melting of refractory mantle at 1·5, 2 and 2·5 1825–1831.
GPa under anhydrous and H2 O-undersaturated conditions: implications for the McDade, P., Blundy, J.D., Wood, B.J., 2003b. Trace element partitioning on the
petrogenesis of high-Ca boninites and the influence of subduction components Tinaquillo Lherzolite solidus at 1.5 GPa. Phys. Earth Planet. Inter. 139 (1),
on mantle melting. J. Petrol. 41 (2), 257–283. 129–147.
Fonseca, R.O., Kirchenbaur, M., Ballhaus, C., Münker, C., Zirner, A., Gerdes, A., Heuser, Meffre, S., Aitchison, J.C., Crawford, A.J., 1996. Geochemical evolution and tectonic
A., Botcharnikov, R., Lenting, C., 2017. Fingerprinting fluid sources in Troodos significance of boninites and tholeiites from the Koh ophiolite, New Caledonia.
ophiolite complex orbicular glasses using high spatial resolution isotope and Tectonics 15 (1), 67–83.
trace element geochemistry. Geochim. Cosmochim. Acta 200, 145–166. Meffre, S., Falloon, T.J., Crawford, T.J., Hoernle, K., Hauff, F., Duncan, R.A., Bloomer,
Gale, A., Dalton, C.A., Langmuir, C.H., Su, Y., Schilling, J.G., 2013. The mean composi- S.H., Wright, D.J., 2012. Basalts erupted along the Tongan fore arc during sub-
tion of ocean ridge basalts. Geochem. Geophys. Geosyst. 14 (3), 489–518. duction initiation: evidence from geochronology of dredged rocks from the
Gass, I., 1968. Is the Troodos massif of Cyprus a fragment of Mesozoic ocean floor? Tonga fore arc and trench. Geochem. Geophys. Geosyst. 13 (12).
Nature 220, 39–42. Michael, P.J., Schilling, J.-G., 1989. Chlorine in mid-ocean ridge magmas: evidence for
Golowin, R., Portnyagin, M., Hoernle, K., Sobolev, A., Kuzmin, D., Werner, R., 2017. assimilation of seawater-influenced components. Geochim. Cosmochim. Acta 53
The role and conditions of second-stage mantle melting in the generation of (12), 3131–3143.
low-Ti tholeiites and boninites: the case of the Manihiki Plateau and the Troodos Miyashiro, A., 1974. Volcanic rock series in island arcs and active continental mar-
ophiolite. Contrib. Mineral. Petrol. 172 (11–12), 104. gins. Am. J. Sci. 274 (4), 321–355.
Gribble, R.F., Stern, R.J., Newman, S., Bloomer, S.H., O’Hearn, T., 1998. Chemical and Mukasa, S.B., Ludden, J.N., 1987. Uranium-lead isotopic ages of plagiogranites from
isotopic composition of lavas from the northern Mariana Trough: implications the Troodos ophiolite, Cyprus, and their tectonic significance. Geology 15 (9),
for magmagenesis in back-arc basins. J. Petrol. 39 (1), 125–154. 825–828.
Haraguchi, S., Ishii, T., 2007. Simultaneous boninitic and arc-tholeiitic volcanisms in Osozawa, S., Shinjo, R., Lo, C.-H., Jahn, B.-m., Hoang, N., Sasaki, M., Ishikawa, K.i.,
the Izu forearc region during early arc volcanism, based on ODP Leg 125 Site Kano, H., Hoshi, H., Xenophontos, C., 2012. Geochemistry and geochronology of
786. Contrib. Mineral. Petrol. 153 (5), 509–531. the Troodos ophiolite: an SSZ ophiolite generated by subduction initiation and
Hickey-Vargas, R., Reagan, M.K., 1987. Temporal variation of isotope and rare earth an extended episode of ridge subduction? Lithosphere 4 (6), 497–510.
element abundances in volcanic rocks from Guam: implications for the evolu- Pearce, J., 1975. Basalt geochemistry used to investigate past tectonic environments
tion of the Mariana Arc. Contrib. Mineral. Petrol. 97 (4), 497–508. on Cyprus. Tectonophysics 25 (1–2), 41–67.
Hickey-Vargas, R., Yogodzinski, G., Ishizuka, O., McCarthy, A., Bizimis, M., Kusano, Y.,
Pearce, J., Robinson, P., 2010. The Troodos ophiolitic complex probably formed in a
Savov, I., Arculus, R., 2018. Origin of depleted basalts during subduction initia-
subduction initiation, slab edge setting. Gondwana Res. 18 (1), 60–81.
tion and early development of the Izu–Bonin–Mariana island arc: evidence from
Pearce, J.A., 1982. Trace element characteristics of lavas from destructive plate
IODP expedition 351 site U1438, Amami-Sankaku basin. Geochim. Cosmochim.
boundaries. Andesites 8, 525–548.
Acta 229, 85–111.
Pearce, J.A., Peate, D.W., 1995. Tectonic implications of the composition of volcanic
Hickey, R.L., Frey, F.A., 1982. Geochemical characteristics of boninite series volcanics:
arc magmas. Annu. Rev. Earth Planet. Sci. 23, 251–286.
implications for their source. Geochim. Cosmochim. Acta 46 (11), 2099–2115.
Plank, T., 2005. Constraints from thorium/lanthanum on sediment recycling at sub-
Ishizuka, O., Hickey-Vargas, R., Arculus, R.J., Yogodzinski, G.M., Savov, I.P., Kusano, Y.,
duction zones and the evolution of the continents. J. Petrol. 46 (5), 921–944.
McCarthy, A., Brandl, P.A., Sudo, M., 2018. Age of Izu–Bonin–Mariana arc base-
Plank, T., Langmuir, C.H., 1993. Tracing trace elements from sediment input to vol-
ment. Earth Planet. Sci. Lett. 481, 80–90.
canic output at subduction zones. Nature 362 (6422), 739–743.
Ishizuka, O., Kimura, J.-I., Li, Y.B., Stern, R.J., Reagan, M.K., Taylor, R.N., Ohara, Y.,
Plank, T., Langmuir, C.H., 1998. The chemical composition of subducting sediment
Bloomer, S.H., Ishii, T., Hargrove, U.S., 2006. Early stages in the evolution of Izu–
and its consequences for the crust and mantle. Chem. Geol. 145 (3), 325–394.
Bonin arc volcanism: new age, chemical, and isotopic constraints. Earth Planet.
Portnyagin, M., Danyushevsky, L., Kamenetsky, V., 1997. Coexistence of two distinct
Sci. Lett. 250 (1), 385–401.
mantle sources during formation of ophiolites: a case study of primitive pillow-
Ishizuka, O., Tani, K., Reagan, M.K., 2014. Izu–Bonin–Mariana forearc crust as a mod-
lavas from the lowest part of the volcanic section of the Troodos Ophiolite,
ern ophiolite analogue. Elements 10 (2), 115–120.
Cyprus. Contrib. Mineral. Petrol. 128 (2–3), 287–301.
Ishizuka, O., Tani, K., Reagan, M.K., Kanayama, K., Umino, S., Harigane, Y., Sakamoto,
Rautenschlein, M., Jenner, G., Hertogen, J., Hofmann, A., Kerrich, R., Schmincke, H.-U.,
I., Miyajima, Y., Yuasa, M., Dunkley, D.J., 2011. The timescales of subduction
White, W., 1985. Isotopic and trace element composition of volcanic glasses
initiation and subsequent evolution of an oceanic island arc. Earth Planet. Sci.
from the Akaki Canyon, Cyprus: implications for the origin of the Troodos ophi-
Lett. 306 (3), 229–240.
Jenner, F.E., O’Neill, H.S.C., 2012. Analysis of 60 elements in 616 ocean floor basaltic olite. Earth Planet. Sci. Lett. 75 (4), 369–383.
glasses. Geochem. Geophys. Geosyst. 13 (2). Reagan, M.K., Ishizuka, O., Stern, R.J., Kelley, K.A., Ohara, Y., Blichert-Toft, J., Bloomer,
Johnson, M.C., Plank, T., 2000. Dehydration and melting experiments constrain the S.H., Cash, J., Fryer, P., Hanan, B.B., 2010. Fore-arc basalts and subduction initia-
fate of subducted sediments. Geochem. Geophys. Geosyst. 1 (12). tion in the Izu–Bonin–Mariana system. Geochem. Geophys. Geosyst. 11 (3).
Kamenetsky, V.S., Crawford, A.J., Meffre, S., 2001. Factors controlling chemistry of Reagan, M.K., Pearce, J.A., Petronotis, K., Almeev, R.R., Avery, A.J., Carvallo, C., Chap-
magmatic spinel: an empirical study of associated olivine, Cr-spinel and melt man, T., Christeson, G.L., Ferré, E.C., Godard, M., 2017. Subduction initiation
inclusions from primitive rocks. J. Petrol. 42 (4), 655–671. and ophiolite crust: new insights from IODP drilling. Int. Geol. Rev. 59 (11),
Kendrick, M.A., Arculus, R.J., Danyushevsky, L.V., Kamenetsky, V.S., Woodhead, J.D., 1439–1450.
Honda, M., 2014. Subduction-related halogens (Cl, Br and I) and H2 O in mag- Regelous, M., Haase, K., Freund, S., Keith, M., Weinzierl, C., Beier, C., Brandl, P., En-
matic glasses from Southwest Pacific Backarc Basins. Earth Planet. Sci. Lett. 400, dres, T., Schmidt, H., 2014. Formation of the Troodos Ophiolite at a triple junc-
165–176. tion: evidence from trace elements in volcanic glass. Chem. Geol. 386, 66–79.
Kent, A.J., Peate, D.W., Newman, S., Stolper, E.M., Pearce, J.A., 2002. Chlorine in sub- Robertson, A.H., 2002. Overview of the genesis and emplacement of Mesozoic ophi-
marine glasses from the Lau Basin: seawater contamination and constraints on olites in the Eastern Mediterranean Tethyan region. Lithos 65 (1), 1–67.
the composition of slab-derived fluids. Earth Planet. Sci. Lett. 202 (2), 361–377. Rubin, K., Embley, R., Clague, D., Resing, J., Michael, P., Keller, N., Baker, E., 2009.
Kessel, R., Schmidt, M.W., Ulmer, P., Pettke, T., 2005. Trace element signature of Lavas from active boninite and very recent basalt eruptions at two submarine
subduction-zone fluids, melts and supercritical liquids at 120–180 km depth. NE Lau Basin sites. In: Proceedings AGU Fall Meeting Abstracts, vol. 1, p. 05.
Nature 437 (7059), 724–727. Schuiling, R.D., 2011. Troodos: a giant serpentinite diapir. Int. J. Geosci. 2 (2), 98.
König, S., Münker, C., Schuth, S., Garbe-Schönberg, D., 2008. Mobility of tungsten in Simonian, K.O., Gass, I.G., 1978. Arakapas fault belt, Cyprus: a fossil transform fault.
subduction zones. Earth Planet. Sci. Lett. 274 (1), 82–92. Geol. Soc. Am. Bull. 89 (8), 1220–1230.
König, S., Münker, C., Schuth, S., Luguet, A., Hoffmann, J.E., Kuduon, J., 2010. Sobolev, A., Portnyagin, M., Dmitriev, L., Tsameryan, O., Danyushevsky, L., Kononkova,
Boninites as windows into trace element mobility in subduction zones. N., Shimizu, N., Robinson, P., 1993. Petrology of ultramafic lavas and associated
Geochim. Cosmochim. Acta 74 (2), 684–704. rocks of the Troodos massif, Cyprus. Petrology 1, 331–361.
Kovalenko, V., Naumov, V., Girnis, A., Dorofeeva, V., Yarmolyuk, V., 2006. Estimation Spandler, C., Pirard, C., 2013. Element recycling from subducting slabs to arc crust:
of the average contents of H2 O, Cl, F, and S in the depleted mantle on the basis a review. Lithos 170, 208–223.
of the compositions of melt inclusions and quenched glasses of mid-ocean ridge Stern, R.J., Bloomer, S.H., 1992. Subduction zone infancy: examples from the Eocene
basalts. Geochem. Int. 44 (3), 209–231. Izu–Bonin–Mariana and Jurassic California arcs. Geol. Soc. Am. Bull. 104 (12),
Langmuir, C., Bezos, A., Escrig, S., Parman, S., 2006. Chemical systematics and hy- 1621–1636.
drous melting of the mantle in back-arc basins. In: Back-Arc Spreading Systems: Stern, R.J., Gerya, T., 2017. Subduction initiation in nature and models: a review.
Geological, Biological, Chemical, and Physical Interactions, pp. 87–146. Tectonophysics. https://doi.org/10.1016/j.tecto.2017.10.014.
214 D. Woelki et al. / Earth and Planetary Science Letters 498 (2018) 203–214

Stern, R.J., Reagan, M., Ishizuka, O., Ohara, Y., Whattam, S., 2012. To understand Woelki, D., Haase, K.M., Schoenhofen, M.V., Beier, C., Regelous, M., Krumm, S.H.,
subduction initiation, study forearc crust: to understand forearc crust, study Günther, T., 2018. Evidence for melting of subducting carbonate-rich sediments
ophiolites. Lithosphere 4 (6), 469–483. in the western Aegean Arc. Chem. Geol. 483, 463–473.
Sun, S.-s., McDonough, W.F., 1989. Chemical and isotopic systematics of oceanic Wood, D., 1981. Geochemistry of Igneous Rocks Recovered from a Transect across
basalts: implications for mantle composition and processes. Geol. Soc. (Lond.) the Mariana Trough, Arc, Fore-Arc and Trench, Sites 453 through 461, Deep
Spec. Publ. 42 (1), 313–345. Sea Drilling Project Leg 60. Initial Report of Deep Sea Drilling Project, vol. 60,
Tatsumi, Y., Eggins, S., 1995. Subduction Zone Magmatism. Wiley. pp. 611–646.
Taylor, R.N., Nesbitt, R.W., Vidal, P., Harmon, R.S., Auvray, B., Croudace, I.W., 1994. Woodhead, J., Hergt, J., Davidson, J., Eggins, S., 2001. Hafnium isotope evidence for
Mineralogy, chemistry, and genesis of the boninite series volcanics, Chichijima, ‘conservative’ element mobility during subduction zone processes. Earth Planet.
Bonin Islands, Japan. J. Petrol. 35 (3), 577–617. Sci. Lett. 192 (3), 331–346.
Tollstrup, D., Gill, J., Kent, A., Prinkey, D., Williams, R., Tamura, Y., Ishizuka, O., 2010. Workman, R.K., Hart, S.R., 2005. Major and trace element composition of the de-
Across-arc geochemical trends in the Izu-Bonin arc: contributions from the sub- pleted MORB mantle (DMM). Earth Planet. Sci. Lett. 231 (1), 53–72.
ducting slab, revisited. Geochem. Geophys. Geosyst. 11 (1). Yajima, K., Fujimaki, H., 2001. High-Ca and low-Ca boninites from Chichijima, Bonin
Whattam, S.A., Stern, R.J., 2011. The ‘subduction initiation rule’: a key for linking (Ogasawara) archipelago. Jpn. Mag. Mineral. Petrol. Sci. 30, 217–236.
ophiolites, intra-oceanic forearcs, and subduction initiation. Contrib. Mineral.
Petrol. 162 (5), 1031–1045.

You might also like