Earth and Planetary Science Letters 413 (2015) 111–122
Contents lists available at ScienceDirect
Earth and Planetary Science Letters
www.elsevier.com/locate/epsl
Pulsed dehydration and garnet growth during subduction revealed
by zoned garnet geochronology and thermodynamic modeling,
Sifnos, Greece
Besim Dragovic a,b,∗ , Ethan F. Baxter a , Mark J. Caddick b
a
b
Department of Earth and Environment, Boston University, 675 Commonwealth Ave., Boston, MA 02215, USA
Department of Geosciences, Virginia Polytechnic Institute and State University, 4044 Derring Hall, Blacksburg, VA 24061, USA
a r t i c l e
i n f o
Article history:
Received 28 January 2014
Received in revised form 4 December 2014
Accepted 11 December 2014
Available online xxxx
Editor: T. Elliott
Keywords:
garnet
subduction
dehydration
Sifnos
Sm–Nd
a b s t r a c t
We use coupled zoned geochronology and thermodynamic modeling of garnet to elucidate the nature
and scale of metamorphic dehydration during Eocene subduction of a quartzofeldspathic lithology from
Sifnos, Greece. Two large garnet porphyroblasts were microdrilled to sample concentric growth zones,
and these were dated using Sm–Nd geochronology. To put results in a geodynamic context and reveal
the causes and consequences of garnet growth, we constructed thermodynamic forward models for a
series of prescribed pressure–temperature (P –T ) paths. Our data reveal three distinct phases of garnet
growth: initial growth at 53.4 ± 2.6 Ma (∼0.8 GPa and ∼300 ◦ C), followed by a period of very limited
growth until a second phase, at 47.22 ± 0.36 Ma, and then a major pulse of growth, responsible for the
majority of the final garnet volume, at 44.96 ± 0.53 Ma (2.06–2.19 GPa and 490–550 ◦ C). This suggests
a >2 order of magnitude acceleration in volumetric growth rate from crystal core to rim, with the final
growth pulse occurring rapidly (<0.8 My), during a period of nearly isobaric heating at >75 ◦ C/My. This
final pulse was accompanied by net bulk rock dehydration of ∼0.5 wt.%. Rapid heating during early
stages of exhumation in the subduction channel, or by sharp thermal gradients related to slab-mantle
coupling could be causes for this pulsed metamorphism and dehydration. The garnet data thus record
a concentrated pulse of dehydration and heating during the otherwise slow and continuous process of
subduction.
© 2014 Elsevier B.V. All rights reserved.
1. Introduction
Dehydration during subduction zone metamorphism plays an
important role in Earth’s volatile budget. The transport and release
of fluids during metamorphism can have a strong influence on
many geologic processes, including mantle wedge melting and rheology (van Keken et al., 2002; Grove et al., 2006), arc magmatism
(Peacock, 1990; Bebout, 1991), intra-slab earthquakes (Kerrick and
Connolly, 2001; Hacker et al., 2003), and element recycling (Rüpke
et al., 2002). Previously held as structurally bound H2 O or hydroxyl
in hydrous phases, water-rich fluids are released due to metamorphic reactions as pressures and temperatures increase. These
*
Corresponding author at: Department of Geosciences, Virginia Polytechnic Institute and State University, 4044 Derring Hall, Blacksburg, VA 24061, USA. Tel.:
+1 617 869 1118.
E-mail addresses: dragovic@vt.edu (B. Dragovic), efb@bu.edu (E.F. Baxter),
caddick@vt.edu (M.J. Caddick).
http://dx.doi.org/10.1016/j.epsl.2014.12.024
0012-821X/© 2014 Elsevier B.V. All rights reserved.
fluids can migrate through the slab via channelized flow (Zack and
John, 2007; John et al., 2012), through the slab-mantle interface
via ductile shear zones and microfracturing in high-pressure blocks
(Konrad-Schmolke et al., 2011), and can kinetically trigger further
mineral reactions (Austrheim, 1987; John and Schenk, 2003) before ascending to the overlying mantle wedge, where they can
induce partial melting (Ulmer, 2001). The timescales of dehydration reactions may be short, occurring in pulses, rather than continuously during subduction. Estimates for very short timescales
ranging from hundreds to hundreds of thousands of years have recently been made for these reactions (Dragovic et al., 2012) and for
the transient passage of the resulting fluids (Camacho et al., 2005;
John et al., 2012). In order to test geodynamic and thermodynamic model predictions (e.g. Schmidt and Poli, 1998; Kerrick and
Connolly, 2001; Hacker, 2008; van Keken et al., 2011), we employ a field-based method using garnet (Dragovic et al., 2012;
Baxter and Caddick, 2013) to determine the primary metamorphic
flux and timescale of dehydration in a subducting crustal lithology.
112
B. Dragovic et al. / Earth and Planetary Science Letters 413 (2015) 111–122
Fig. 2. Field photograph of sample 09DSF-1A (one euro coin for scale). Gt1 is the
larger ∼5 cm garnet below. Gt2 is the smaller ∼3 cm garnet above.
The second metamorphic stage consisted of Miocene greenschist to amphibolite facies overprinting of the earlier metamorphism in the southern part of the island (Fig. 1), dated
at 24–18 Ma using K–Ar, Rb–Sr, and Ar–Ar on white micas
(Altherr et al., 1979; Wijbrans et al., 1990; Bröcker et al., 2013).
This latter phase of metamorphism occurred during exhumation and infiltration of an ascending fluid phase (Breeding et al.,
2003). Greenschist facies conditions determined for cooling/decompression are 350–450 ◦ C and 0.5–1.0 GPa (Trotet et al., 2001b;
Schmädicke and Will, 2003).
Fig. 1. Map of Sifnos, Greece showing sample location (star). Stratigraphic column
shows relationship between major units on Sifnos. The Cherronissos and Faros units
are separated by a low angle fault. Map is modified after Matthews and Schliestedt
(1984), Trotet et al. (2001b), and Dragovic et al. (2012).
2. Geologic setting
Sifnos is part of the Cycladic Islands located southeast of mainland Greece. These islands form part of a subduction-related accretionary complex. Sifnos lies within the lower unit of this complex,
termed the Cycladic Blueschist Unit (CBU) (Matthews and Schliestedt, 1984; Okrusch and Bröcker, 1990), consisting of a crystalline
basement overlain by thrust sheets of a metamorphosed continental margin sequence (Fig. 1). A well-preserved eclogite-blueschist
unit, located in the northwestern part of the island, is bounded
on the north and south by separate marble complexes (Trotet et
al., 2001a; Ring and Layer, 2003). This unit contains metabasites,
metapelites, quartzites, and acidic gneisses, and is regarded as a
structurally coherent sequence that all experienced essentially the
same metamorphic evolution (Dragovic et al., 2012 and references
therein).
Two stages of metamorphism occurred on Sifnos. The first
stage involved Eocene blueschist to eclogite facies metamorphism.
This is associated with collision of the Apulian microplate and
the Eurasian continent (Avigad, 1993) and has been dated at
48–41 Ma using K–Ar and Rb–Sr in white micas (Altherr et al.,
1979; Wijbrans et al., 1990), ∼41 Ma using Ar–Ar apparent age
spectra (Forster and Lister, 2005), and 46.49 ± 0.36 Ma using Sm–
Nd in garnet (Dragovic et al., 2012). Dragovic et al. (2012) also
determined that garnet growth metamorphism occurred over a
brief period of time spanning just tens to hundreds of thousands
of years (no more than 1 My). Recent estimates for peak P –T conditions of Eocene metamorphism have ranged from 525 to 600 ◦ C
at 2.0–2.2 GPa (Schmädicke and Will, 2003; Groppo et al., 2009;
Dragovic et al., 2012; Ashley et al., 2014).
3. Sample description
Sample 09DSF-1A is a moderately foliated, quartzofeldspathic
gneiss found as float (the primary outcrop was not found, though
similar material was identified in place further west along the sea
cliffs) on the north face of Vroulidia Bay, in the northern section
of the CBU, but structurally lower than the shear zone near the
upper marble unit (Fig. 1). The sample consists of fine-grained,
granoblastic quartz-rich layers, and finer-grained layers of quartz,
plagioclase, magnetite, and paragonite, and is notable because it
contains several large (>2 cm) garnet crystals including one 5 cm
crystal (“gt1”) in contact with a 3.5 cm crystal (“gt2”) (Fig. 2).
The large size of these crystals provides the highest potential for
microsampling of multiple zones for geochronology. All garnet porphyroblasts have a flattened, lenticular shape. Smaller, isolated porphyroblasts are euhedral, while the two larger crystals studied here
are subhedral, with weathered surfaces (Fig. 2). The sample contains quartz, plagioclase, garnet, paragonite, phengite, magnetite,
hematite, and less abundant chlorite, glaucophane, epidote, rutile,
and zircon; the latter three only occurring as inclusions in garnet.
4. Analytical methods
Two large garnets were selected for Sm–Nd geochronology following the methods of Pollington and Baxter (2011). To properly
isolate a central section of the larger garnet (gt1), a 10 × 9 × 5-cm
subvolume (including both large crystals) was sent for high resolution X-ray computed tomography at The University of Texas at
Austin, Department of Geological Sciences. Based on sixty 1-mm
scans, the depth to the largest diameter of gt1 was determined.
A ∼2.5-mm thick wafer was cut such that the morphologic center of gt1 was the top of the wafer. The morphologic center of the
smaller garnet (gt2) was missed by ∼1 mm.
The major divalent cation (Fe, Mg, Ca, and Mn) concentrations
of each garnet crystal were measured on the polished wafer and
B. Dragovic et al. / Earth and Planetary Science Letters 413 (2015) 111–122
113
Fig. 3. a) Microdrilling of gt1 (left) and gt2 (right) based on MnO wt.% contours. Ten and nine concentric zones were drilled in each garnet, respectively. Drill trenches are
shown in white. b) Rim to core to rim traverse of garnet chemistry in sample 09DSF-1A with radial extent of drilled zones shown as dashed lines.
used to create chemical contour maps (Fig. 3a). MnO wt.% contours, which display a smooth, concentric pattern, (Fig. 3a–b) were
used to guide microsampling for geochronology. The MnO contour
map highlights two crystal growth characteristics, undetectable
without the analysis: (1) the chemical cores of the garnets are not
in their morphologic centers, and (2) garnet growth continued after the two crystals impinged upon one another. Therefore, care
was taken to define the growth zones to be sampled, with positioning of each zone based on; a) extraction of roughly equivalent
volumes of sample, b) expected loss during physical and chemical sample preparation, and c) expected [Nd] in pure garnet after
preparation.
Ten, and nine, growth zones were microsampled, for gt1 and
gt2, respectively (Fig. 3a). The garnets were microdrilled using the
New Wave MicroMill at Boston University, following the procedures of Pollington and Baxter (2011). The zones were individually
collected, hand crushed, and sieved to between 63 and 106 μm.
Samples were then cleaned using a Frantz magnetic separator and
handpicked to remove inclusions and visibly inclusion-rich garnet fragments. In addition to isolating garnet fragments from each
zone, fine (<63 μm) garnet powders from the sieving process were
collected. Several samples of the surrounding rock matrix were
also sampled for analysis.
Inclusion phases in garnet (other silicates such as epidote and
more refractory phases like rutile) were preferentially removed using a partial dissolution technique (see Appendix A.1) modified
after Dragovic et al. (2012). This was performed on the 63–106 μm
fraction. Additionally, several of the fine garnet powders were
treated. Subsequently, all cleansed garnet separates, untreated and
treated fine garnet powders, and rock matrices were completely
dissolved.
All samples (cleansed garnet separates, fine garnet powders,
and rock matrices) were treated with a mixed 147 Sm–150 Nd spike
prior to loading through a three-stage column chromatography
procedure following Harvey and Baxter (2009). Three-column
blanks with in-house, distilled 2-methyl lactic acid, run alongside
all of the samples, ranged from 1–2 pg of Sm and 4–6 pg of Nd,
leading to high sample-to-blank ratios, appropriate for high precision analysis at such low sample sizes.
Samples were analyzed at the Boston University TIMS (thermal
ionization mass spectrometer) facility using a Thermo–Finnigan
TRITON (for Nd) following Harvey and Baxter (2009). Samarium
was loaded in nitric acid onto rhenium double filaments. Over the
span of this study, 4 ng loads of an in-house Nd standard solution (Ames metal) yielded 143 Nd/144 Nd = 0.5121313 ± 0.0000078
(15.2 ppm, 2 RSD, n = 62). The reproducibility in 147 Sm/144 Nd is
0.023% (2 RSD) based on repeat analyses of a mixed gravimetric
normal solution with our calibrated in-house spike. The isotopic
data collected in this study are provided in Table 1.
Standard analytical methods involved in determination of the
bulk rock and mineral compositions are detailed in the appendix
(A.2 and A.3, respectively).
5. Data and observations
5.1. Petrography and mineral chemistry
Garnet inclusions consist of quartz, clinozoisite/epidote, phengite, paragonite, albite, magnetite, rutile, and zircon (Figs. A.1a and
A.1b; in appendix). Inclusion assemblages of clinozoisite/epidote,
phengite, and quartz are presumed to reflect pseudomorphs after
lawsonite (Schmädicke and Will, 2003). Quartz comprises a significant portion of the garnet inclusion population. While the existence of garnet and its inclusions implies a high-pressure assemblage, the matrix displays a strong retrogressive overprint, interpreted to be associated with late fluid influx. This overprint manifests itself as: partly chloritized garnet rims (outer 2–3 mm), matrix adjacent to garnet consisting of coarse-grained quartz, phengite, and paragonite, veined quartz layers penetrating through the
outer edge of the garnet and parallel to compositional banding in
the matrix, and the near absence of high pressure phases remaining in the matrix.
Fig. 3b shows that garnet preserves apparent growth zoning.
XSps decreases from core to rim, with a slight spike in spessartine
content near the “outer mantle/inner rim” of the grain. The geometric center of the garnet crystal was used to date initiation of
garnet growth, so it was intentionally located in the wafer center,
114
B. Dragovic et al. / Earth and Planetary Science Letters 413 (2015) 111–122
Table 1
Sm–Nd isotopic data for 09DSF-1A.
±2 SE
±2 SE
(abs)
(ppm)
0.5128587
0.5128616
0.5128544
0.5128675
0.5128123
0.5128545
0.5143776
0.5128903
0.5129053
0.5128877
0.5129301
0.5129379
0.0000044
0.0000088
0.0000043
0.0000063
0.0000081
0.0000131
0.0000075
0.0000265
0.0000225
0.0000086
0.0000129
0.0000174
8.5
17
8.4
12
16
25
15
52
44
17
25
34
3.351
3.532
7.597
6.356
9.368
9.816
9.686
9.365
8.471
0.9640
0.5139304
0.5139024
0.5151548
0.514720
0.5156004
0.5157870
0.5156784
0.5156318
0.5153298
0.5131254
0.0000883
0.0000098
0.0000093
0.000006
0.0000082
0.0000057
0.0000054
0.0000066
0.0000099
0.0000039
170
19
18
12
16
11
10
13
19
7.5
1.1
52
5.2
1.1
26
26
4.1
4.8
23
2.968
0.4357
9.826
7.593
2.701
2.060
9.682
4.613
2.778
0.5138267
0.5129421
0.5158485
0.5150395
0.5136093
0.5134255
0.5156226
0.5141722
0.5136357
0.0000979
0.0000037
0.0000168
0.0000244
0.0000055
0.0000073
0.0000176
0.0000094
0.0000064
190
7.3
33
47
11
14
34
18
12
5.18
5.05
3.29
3.21
2.95
9.58
3.65
4.13
3.90
2.33
1.18
79.5
17
27
23
24
13
11
14
16
21
14
10
640
0.4126
0.4220
0.4574
0.5880
0.5476
0.7112
0.6025
0.7303
0.7717
0.5928
0.5197
0.1916
0.5129627
0.5129496
0.5129623
0.5130042
0.5129860
0.5130302
0.5129826
0.5130555
0.5130613
0.5129923
0.5129566
0.5128688
0.0000066
0.0000058
0.0000042
0.0000067
0.0000069
0.0000054
0.0000051
0.0000051
0.0000083
0.0000055
0.0000055
0.0000024
13
11
8.2
13
13
11
10
10
16
11
11
4.6
1.46
7.82
280.
2.19
1.12
4.28
49.2
1.10
0.488
3.17
8.5
13
1400
16
8.5
8.9
360
6.4
3.5
1.2
0.4719
0.4978
0.1176
0.5784
0.7442
0.2896
0.1917
0.9403
1.486
0.3392
0.5129458
0.5129813
0.5128408
0.5130071
0.5130629
0.5129718
0.5128729
0.5131111
0.5132246
0.5130341
0.0000058
0.0000117
0.0000033
0.0000079
0.0000104
0.0000364
0.0000036
0.0000105
0.0000244
0.0000598
11
23
6.5
15
20
71
7.1
20
48
120
Sm
(μg/g)
Nd
(μg/g)
ng Nd
loaded
147
09DSF-1A matrices
matrix 1
matrix 2
matrix 3
matrix 4
top matrix
bottom matrix
matrix 1a
matrix 1b
matrix 1c
matrix 1d
matrix 1e
matrix 1f
3.33
7.58
3.12
6.13
0.788
0.532
0.585
0.238
0.352
3.40
0.836
0.534
12.7
26.1
10.8
20.5
1.56
1.05
1.12
0.463
0.571
9.14
1.25
1.23
18
37
24
23
7.9
5.3
11
0.67
0.87
14
1.8
1.9
0.1583
0.1756
0.1758
0.1806
0.3065
0.3083
0.3148
0.3109
0.3731
0.2247
0.4053
0.2638
09DSF-1A gt1
gt1 zone 1
gt1 zone 2
gt1 zone 3
gt1 zone 4
gt1 zone 5
gt1 zone 6
gt1 zone 7
gt1 zone 8
gt1 zone 9
gt1 zone 10
0.111
0.241
0.923
0.751
1.71
1.55
1.89
1.23
0.714
0.876
0.020
0.041
0.074
0.072
0.110
0.096
0.118
0.079
0.051
0.550
1.7
3.0
9.9
9.4
7.5
12
11
10
3.7
65
09DSF-1A gt2
gt2 zone 1
gt2 zone 2
gt2 zone 3
gt2 zone 4
gt2 zone 5
gt2 zone 6
gt2 zone 7
gt2 zone 8
gt2 zone 9
0.237
0.994
1.50
0.791
2.10
2.28
2.13
0.916
0.972
0.048
1.38
0.093
0.063
0.470
0.670
0.133
0.120
0.212
09DSF-1A gt1 pwd
gt1 z1 pwd
gt1 z1 lch pwd
gt1 z2 pwd
gt1 z3 pwd
gt1 z4 pwd
gt1 z5 pwd
gt1 z6 pwd
gt1 z7 pwd
gt1 z7 lch pwd
gt1 z8 pwd
gt1 z9 pwd
gt1 z10 pwd
3.53
3.52
2.49
3.12
2.67
11.3
3.64
4.99
4.98
2.28
1.02
25.2
09DSF-1A gt2 pwd
gt2 z1 pwd
gt2 z1 lch pwd
gt2 z2 lch pwd
gt2 z3 lch pwd
gt2 z4 lch pwd
gt2 z5 lch pwd
gt2 z6 lch pwd
gt2 z7 lch pwd
gt2 z8 lch pwd
gt2 z9 lch pwd
1.14
6.44
54.5
2.09
1.47
2.05
15.6
1.71
1.20
1.78
Sample
rather than at the top surface that was prepared for electron microprobe analysis (see Appendix A.3). Despite the top of the wafer
not sampling the true crystal core, the relatively smooth, decreasing spessartine content still indicates prograde growth zoning, with
zoning in the outermost garnet rim possibly attributed to postgrowth diffusive re-equilibration.
Clinozoisite/epidote occurs exclusively as an inclusion phase in
garnet as pseudomorphs with paragonite and quartz (after lawsonite), forming grains up to 400 μm in diameter (Fig. A.1b). Inclusions in the garnet core (Ep I) are generally more Al-rich, although
Sm/144 Nd
143
Nd/144 Nd
many grains display a Fe3+ -rich overgrowth. Occurrences towards
the rim of garnet (Ep II) are Fe3+ -rich (Table A.1).
Paragonite occurs both as inclusions in the garnet core and in
the matrix surrounding garnet. Large paragonite grains (∼100 s μm)
are found in the coarser-grained matrix adjacent to the garnet rim,
or in the more medium to finer-grained matrix farther from the
garnet rim. Phengite occurs both as inclusions in garnet and as
larger matrix grains near the garnet rim. Phengite inclusions occur
as aggregates with clinozoisite/epidote and quartz. In some grains
phengite is altered to biotite.
B. Dragovic et al. / Earth and Planetary Science Letters 413 (2015) 111–122
Chlorite occurs mainly in fractures around the garnet rim (outer
2–5 mm). Some chlorite also occurs as an alteration product of
biotite in the coarse-grained matrix directly around the garnet rim.
Compositionally, grains can be characterized as high Fe3+ , low Al,
Mg–Fe chlorites.
Rutile is found as both inclusions in garnet and in the finergrained matrix 2–3 cm from the garnet rim. Rutile has a grain size
of 100–200 μm (garnet inclusion) or tens of μm (fine-grained matrix). Rutile is observed to be the primary Ti-bearing phase.
The remainder of the coarser-grained matrix surrounding garnet contains smaller biotite grains, large sodic–calcic and calcic
amphibole grains (∼ hundreds of μm), and abundant granoblastic
quartz. The finer-grained matrix (2–3 cm from the garnet) is dominated by abundant quartz and magnetite, with smaller amounts
of albite and rutile. The matrix away from the garnet is comprised mainly of quartz and albite (∼90% of the matrix) with lesser
amounts of magnetite, hematite, and phengite. Matrix compositional banding exists of coarser grained quartz bands alternating
with finer grained matrix consisting of quartz, albite, magnetite,
hematite, and phengite.
5.2. Geochronology
Sm–Nd data for all analyzed samples are shown in Table 1. Due
to the abundance of inclusions, sample loss during partial dissolution cleansing was high (78–90% by mass). As a result, the amount
of Nd analyzed from each garnet zone was small (as low as 1.7 ng).
Cleansed garnet 147 Sm/144 Nd values range from 2.06–9.82, excluding gt1 zone 10 (0.96), and gt1 zone 2 (0.44). Untreated fine garnet
powders give 147 Sm/144 Nd ranging from 0.19–0.73. Nd concentrations in cleansed garnet range from 0.02–0.21 μg/g, with the exception of gt1 zone 10 (0.55), gt2 zone 2 (1.4), gt2 zone 5 (0.47),
and gt2 zone 6 (0.67). Nd concentrations in untreated fine garnet
powders ranged from 1.2–79 μg/g. This indicates that the partial
dissolution cleansing successfully mitigated contamination by inclusions for most garnets. Indeed, some of these 147 Sm/144 Nd isotopic values are among the highest ever reported for garnet (Baxter
and Scherer, 2013). Gt1 zone 10 and gt2 zone 2 give anomalously
low 147 Sm/144 Nd and anomalously high [Nd] indicating they are
still significantly contaminated by inclusions. Those data will not
be used for geochronologic interpretations.
Additionally, twelve matrix samples were analyzed. Four of
these (matrix 1, 2, 3, 4 in Table 1) are from large chunks (∼15 g
each) taken several cm away from the garnet edges. These give
similar isotopic data, though some separation in 147 Sm/144 Nd
likely reflects small-scale mineralogical variations. The matrix
147
Sm/144 Nd ratios (0.15–0.17) are typical for common crustal
rocks. The other eight matrix samples (matrix 1a–f) are from
smaller representative volumes (∼3 g each) and are much closer
to the garnets themselves. These matrix analyses have anomalously
high 147 Sm/144 Nd (0.22–0.41), far too high for any common crustal
rock type and thus must reflect some localized enrichment in Sm,
perhaps related to late garnet resorption that variably flushed the
local matrix with Sm and elevated 143 Nd/144 Nd. The four larger
matrices (matrix 1, 2, 3, 4), sampled away from the garnet, are
more representative of the matrix in isotopic equilibrium with
garnet during growth. Only these matrix analyses are used in
geochronologic interpretations.
Untreated fine garnet powders from each garnet zone (denoted
“pwd” in Table 1) generally fall off of the clean garnet-matrix
isochron due to inherited inclusions and are thus not included
in age interpretations. Previous work has shown that full cleansing of such fine-grained garnet powders is impossible (Pollington
and Baxter, 2011). However, if the key inherited phase is a phosphate (e.g. monazite) then a less aggressive leach (Appendix A.1),
as performed here, might remove it. Most of these “leached pow-
115
Table 2
Major element analysis of 09DSF-1A matrix, garnet and whole rock.
wt.%
Matrix
Garnetb
Whole rockc
SiO2
TiO2
Al2 O3
FeO(tot)
a
Fe2 O3
a
FeO
MnO
MgO
CaO
Na2 O
K2 O
P2 O5
Total
81.56
0.19
10.00
2.61
2.30
0.54
0.02
0.18
0.14
5.14
0.23
0.05
100.13
46.81
0.40
18.14
26.59
n/a
n/a
0.82
0.49
5.17
0.69
0.89
0.07
100.07
78.43
0.21
10.73
4.77
2.37
2.63
0.09
0.21
0.60
4.74
0.29
0.05
100.12
a
Ferric/ferrous ratio estimated from mineral mode and chemistry.
Garnet chemistry as analyzed by a single whole garnet crystal (garnet + inclusions) by ICP-ES.
c
Whole rock chemistry by adding 9% garnet (from ICP-ES analysis) and 91% of
the matrix chemistry.
b
ders” (denoted “lch pwd’ in Table 1) produced data falling on the
garnet-matrix isochron, lowering the MSWD. When this was the
case, leached powders were included in geochronologic interpretations.
5.3. Garnet growth ages
Combinations of cleansed garnet separates from two different
garnets, fine garnet powders (treated or untreated), and four matrices (matrix 1, matrix 2, matrix 3, and matrix 4; Table 1) were
used to create 19 multi-point isochron ages (ranging from 5 to 6
points); 10 from gt1 and 9 from gt2. These are shown in Table 2
and Fig. 4a (isochrons are shown Figs. A.2 and A.3 in appendix).
For both garnets, zone 1 is significantly older than the remaining zones, with initiation of growth at 52.60 ± 3.30 Ma and
54.60 ± 3.90 Ma, for gt1 and gt2, respectively. The cores of both
garnets, if assembled on the same isochron, are shown to be concordant in age, at 53.4 ± 2.6 Ma (MSWD = 1.3). Zones 2 and 3
of both garnets (excluding gt2 zone 2 as described above) are significantly younger, at 47.40 ± 0.47 Ma and 47.24 ± 0.20 Ma (gt1),
and 47.32 ± 0.27 Ma (gt2). Similar to the treatment of ages in the
core between gt1 and gt2, an 8-point isochron age, including three
garnet separates (zone 2 from gt1; zone 3 from gt1 and gt2), one
leached garnet powder, and four matrices, gives a garnet growth
age of 47.22 ± 0.36 Ma (MSWD = 4.7) for zones 2 and 3.
The remainder of ages for both crystals is significantly younger.
The ages of zones 4 through 9 are similar, with ages of zone 4 in
gt1 (45.98 ± 0.21 Ma) almost statistically within error (2σ ) of zone
9 in gt1 (45.49 ± 0.19 Ma). Zone 4 in gt2 (44.80 ± 1.2 Ma) is statistically within error of zone 9 in gt2 (45.52 ± 0.50 Ma). Note that
zones 6 and 8 from gt1 are slightly (though significantly) older.
During full dissolution, zones 6 and 8 for gt1 spent considerably
longer in acid (several weeks vs. a few days) than the other garnet
separates, perhaps incorporating some Nd from restitic inherited
phases, resulting in falsely older ages. As a result, ages from zones
6 and 8 from gt1 are excluded from subsequent interpretations.
5.4. Growth spans of gt1 and gt2
The difference in age between the core and rim multi-point
isochron ages for gt1 and gt2 is 7.1 ± 3.3 Ma and 9.1 ± 3.9 Ma
(2σ ), respectively. Using combined multi-point isochron ages, the
garnet growth duration is 7.9 ± 2.6 Ma (2σ ). However, the most
noteworthy aspect of the geochronologic results presented here are
the multiple stages of garnet growth. The majority of the garnet
growth duration is between growth of zones 1 and 2 (Fig. 4a). This
116
B. Dragovic et al. / Earth and Planetary Science Letters 413 (2015) 111–122
logic center but rather in opposing “horseshoe” shapes; Fig. 3a). To
calculate average volumetric growth rates, a spherical geometry is
assumed for ease of calculation; however, as stated earlier, the garnets are flattened in one axis. Uncertainties in these growth rates
are calculated from the radial span (difference in inner and outer
radius) of each microsampled growth zone.
Considering gt1, the average volumetric growth rate of garnet
between zones 1 and 2 is 0.38 ± 0.30 cm3 /My. Subsequently, the
growth rate increases between zones 2 and 4, to 4.5 ± 2.0 cm3 /My
for gt1. Finally, growth of zones 4 to 9 was very rapid, with an
average volumetric garnet growth rate of 92 ± 54 cm3 /My for gt1.
Since zone 2 of gt2 failed to produce a robust age, we can calculate an average volumetric growth rate from zone 1 to zone 4 in
gt2 of 0.30 ± 0.14 cm3 /My. The ages of zone 4 and 9 in gt2 overlap
within error, providing a minimum (2σ ) volumetric growth rate of
50 cm3 /My.
Combining geochronologic data with calculated volumetric
growth rates reveals multi-stage garnet growth. Initiation of
growth, at 53.4 ± 2.6 Ma, is either rapid and followed by slow
or no growth, or is continuously slow over a ∼6 My period. Additional growth occurs, forming zones 2 and 3 of both garnets, at
47.22 ± 0.36 Ma (combined isochron age of zones 2–3). This is followed by a period of slow (or no) growth until an extremely rapid
“pulse” of garnet growth at 44.96 ± 0.53 Ma. Indeed, comparing
calculated time-averaged garnet growth rates between zones 1–4
and zones 4–9 for both garnets (Fig. 4b), acceleration in garnet
growth is at least 2 orders of magnitude. Thermodynamic analysis
of garnet-forming reactions and P –T trajectories helps elucidate
causes and consequences of this acceleration in the net reaction
rate.
6. Thermodynamic analysis
Fig. 4. a) Summary of Sm–Nd ages for gt1 (blue diamonds) and gt2 (red squares).
The combined data include includes both garnets (green triangles). b) Percent garnet
grown (of total) in both garnets over time. Note the slow early growth, shown by
low time-averaged minimum volumetric growth rate between zones 1 and 4, and
accelerated growth between zones 4 and 9. Growth rate values for gt1 in blue and
gt2 in red. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
can either be interpreted as a) short-lived growth of the garnet
core before a period of relative quiescence in growth then subsequent growth of zones 2 and 3, or b) a period of sustained slow
core growth followed by accelerated growth of zones 2 and 3.
The ages of zones 4 through 9 are nearly statistically significant
as a single population, with a 14-point isochron age, consisting of
zones 4, 5, 7, and 9 from gt1, zones 4–9 from gt2, and four matrices gives 44.96 ± 0.53 Ma (MSWD = 10.6). This confirms that the
growth interval of latter stages was rapid. The difference in ages
between zones 4 and 9 in gt1 and gt2 are 0.49 ± 0.28 Ma and
−0.69 ± 1.30 Ma (2σ ), respectively. This gives a maximum (2σ )
growth duration of 0.77 My (gt1) and 0.61 My (gt2) during this
stage. Instantaneous growth is statistically allowable within our
uncertainties for this span of garnet growth in gt2.
5.5. Accelerated volumetric garnet growth
Combining garnet growth durations with observed spatial sampling of the zones, volumetric growth rates are calculated. Since
the chemical core of each garnet is not in the morphologic center,
an average radial distance of each zone from four axes was calculated, where permitted (zones 7, 8, 9, and 10 for gt1 and zones 7,
8, and 9 from gt2 do not grow concentrically from the morpho-
Thermodynamic forward modeling was performed for an appropriate bulk rock composition to constrain the amount of garnet growth along likely P –T trajectories during subduction of this
rock. Models were initiated at fluid-saturated conditions, with both
garnet and free fluid then progressively removed at each model
increment (following procedures from Baxter and Caddick, 2013),
as these phases have important effects on the chemical composition of the effective rock volume (e.g. Marmo et al., 2002;
Konrad-Schmolke et al., 2006; Gaidies et al., 2008; Dragovic et
al., 2012). Model calculations used version 6.6.7 of the program
Perple_X (Connolly, 2009), and the ‘ds55’ update of the internally consistent dataset of Holland and Powell (1998), describing the rock with the system SiO2 –TiO2 –Al2 O3 –FeO–Fe2 O3 –MnO–
MgO–CaO–Na2 O–K2 O–H2 O. Activity-composition models used in
this study are shown in the appendix (A.4). Fluid was considered
as comprising both H2 O and CO2 . All models assume equilibrium
(minimized Gibbs’ free energy) at every calculated increment along
discretized input P –T paths, and assume a closed chemical system
with the exception of the H2 O, CO2 , and garnet removed at each
increment.
6.1. Initial bulk rock composition
Direct measurement of initial whole rock chemistry was not
possible due to the large size of the garnet with respect to the
hand sample size. Therefore, an analyzed whole garnet composition (with its inclusions) was reintegrated with the analyzed matrix composition in 9%/91% proportions (see discussion in A.2 of
appendix) to arrive at the initial bulk rock composition used in
all modeling presented herein (Table 3). Sensitivity analyses show
that uncertainty in observed garnet mode (within a few percent)
yields little difference to model results. Additionally, a range of initial volatile contents, from a bulk rock that was saturated solely in
B. Dragovic et al. / Earth and Planetary Science Letters 413 (2015) 111–122
117
Table 3
Sm–Nd garnet ages. Combined zonal ages are multipoint isochrons with garnets, matrices, and associated garnet powders.
Isochron
Age
(Ma)
2 SD Age
uncertainty
MSWD
mtx + 1 lch pwd)
mtx)
mtx)
mtx)
mtx)
mtx)
mtx + 1 lch pwd)
mtx)
mtx)
52.60
47.41
47.24
45.98
45.55
46.39
45.23
46.09
45.49
No age reported
3.30
0.47
0.20
0.21
0.15
0.13
0.56
0.14
0.19
1.5
1.7
1.7
1.7
1.7
2.2
7.7
1.7
1.7
(gt2 zone 1 + 4 mtx + 1 lch pwd)
54.60
No age reported
47.27
44.80
45.28
45.65
44.36
45.14
45.49
3.90
1.6
0.60
1.20
0.52
0.68
0.73
0.34
0.50
6.9
11
1.7
1.8
7.3
2.2
1.7
53.40
Not applicable
47.22
45.30
45.55
46.37
44.77
46.00
45.51
2.6
1.3
0.41
1.00
0.15
0.27
0.66
0.67
0.19
5.6
10
1.5
2.4
13
7.6
1.6
Gt1
Gt1
Gt1
Gt1
Gt1
Gt1
Gt1
Gt1
Gt1
Gt1
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
1 (gt1
2 (gt1
3 (gt1
4 (gt1
5 (gt1
6 (gt1
7 (gt1
8 (gt1
9 (gt1
10
Gt2
Gt2
Gt2
Gt2
Gt2
Gt2
Gt2
Gt2
Gt2
zone
zone
zone
zone
zone
zone
zone
zone
zone
1
2
3
4
5
6
7
8
9
Combined
Combined
Combined
Combined
Combined
Combined
Combined
Combined
Combined
(gt2
(gt2
(gt2
(gt2
(gt2
(gt2
(gt2
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
zone
1
2
3
4
5
6
7
8
9
1+4
2+4
3+4
4+4
5+4
6+4
7+4
8+4
9+4
3+4
4+4
5+4
6+4
7+4
8+4
9+4
mtx +
mtx +
mtx)
mtx +
mtx +
mtx +
mtx)
1 lch pwd)
1 lch pwd)
1 lch pwd)
1 lch pwd)
1 lch pwd)
(gt1 + gt2 + 4 mtx + 2 lch pwd)
(gt1
(gt1
(gt1
(gt1
(gt1
(gt1
(gt1
+
+
+
+
+
+
+
gt2
gt2
gt2
gt2
gt2
gt2
gt2
+
+
+
+
+
+
+
4
4
4
4
4
4
4
mtx +
mtx +
mtx)
mtx +
mtx +
mtx +
mtx +
1 lch pwd)
1 lch pwd)
1
2
1
1
lch
lch
lch
lch
pwd)
pwd)
pwd)
pwd)
0.5 ◦ C increments, and the mineral assemblage was calculated at
each.
6.3. Calculated garnet growth along subduction P –T paths
Fig. 5. Conceptual diagram showing that P –T paths for the thermodynamic forward models. Note the change in the angle of the subduction geotherm, and that
all paths include a stage of nearly isobaric heating. Shaded polygons represent core
(left) and rim (right) pressures and temperatures for a Sifnos garnet from Dragovic
et al. (2012). See text for path constraints.
H2 O to a bulk rock with both H2 O and CO2 (in various ratios), was
also explored to test the sensitivity of results to initial volatile content.
6.2. Model pressure–temperature paths
Evolving mineral assemblages, garnet proportion and composition were modeled along eight possible subduction zone P –T
trajectories (Fig. 5) consistent with previous constraints from Sifnos
(e.g. Dragovic et al., 2012). The early part of Path 1 is based upon
inferred Sifnos history from Groppo et al. (2009) while Path 8
represents a geodynamical model for the top of the Aegean slab
(specifically the D80 model of Syracuse et al., 2010). Paths 2–7
represent intermediate conditions. P –T paths were discretized at
Fig. 6a shows the calculated modal abundance of garnet along
prescribed P –T paths. Common characteristics can be observed in
all modeled subduction geotherms. Most significantly, all models
predict the majority of garnet production during a narrow span of
near isobaric heating, between ∼490 ◦ C and ∼550 ◦ C at ∼2.1 GPa.
An early small pulse of garnet growth is predicted at very low
temperature for paths 2 and 3 (at ∼350 ◦ C and ∼290 ◦ C, respectively). The colder, deeper paths (5–8) predict initial garnet growth
at much higher temperatures (from ∼430–460 ◦ C). The final modal
abundances of garnet at the peak T of each path are also broadly
similar (ranging from 4.0–4.7 vol.%, Fig. 6a). This is in good agreement with the observed 4.5% modal abundance of garnet (see
Appendix A.3), an important validation of the modeling. Differences in the final proportion of garnet for each path reflect the
path-dependent nature of these calculations, due to progressive
fractionation of calculated fluid and garnet.
Predicted garnet modal abundance patterns (Fig. 6a) may be
compared to the observed pattern of volumetric garnet growth
rates from zoned geochronology (Fig. 6c). For this, each microsampled (and dated) growth zone corresponds to a whole rock garnet
modal abundance at the time of its growth, calculated as a proportion of the observed 4.5% in the natural sample (assuming spherical geometry and knowing the radial size of each sampled zone).
Garnet growth zone 1 thus represents a modal proportion of just
0.03 and 0.01 vol.% garnet in gt1 and gt2, respectively (Fig. 6c). A
second pulse of garnet growth (zones 2 and 3) formed 0.27 and
0.12 vol.% garnet. A final period of growth produced the remaining garnet, with the three pulses combining to produce a total
4.5% by volume garnet. The major growth pulse (at ∼2.1 GPa in
all models, and at 44.96 ± 0.53 Ma in the zoned geochronology) is
the most important – and robust – feature corroborated by both
118
B. Dragovic et al. / Earth and Planetary Science Letters 413 (2015) 111–122
Fig. 6. Dehydration and garnet growth history during subduction. a) Modeled modal abundance of garnet along selected modeled P –T paths. Path 3 (in bold red) is highlighted
as the path most consistent with observed garnet growth b) Modes of all phases along P –T progress for the modeled path 3 (left axis). Note the significant growth of garnet
at the expense of lawsonite, white mica, and glaucophane at ∼490 ◦ C. Superimposed on the plot is the predicted modal abundance of garnet from 6a (right axis). c) Modal
abundance of garnet vs. age of growth for gt1 (blue), gt2 (red). Combined zone 1 age shown with a green symbol. Numbers in parentheses denote the distinct garnet growth
period, or “pulse.” Note the similarity between observed garnet growth and predicted growth (shown directly above in 6a). d) Change in garnet modal abundance and the
wt.% water stored in the rock along the modeled P –T Path 3. Note the three distinct stages of garnet that can be correlated to those shown in 6c. e) Cumulative weight%
water released from the rock over time for gt1 (blue) and gt2 (red). f) Summary of P –T history for sample 09DSF-1A. Modeled P –T Path 3 shown in black. Also shown are
P –T estimates (red polygons) representing pressure and temperature span of garnet growth from Dragovic et al. (2012), in addition to a prograde path from Groppo et al.
(2009). Blue symbols denote the weight% water stored in the rock prior to, and after rapid heating. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
B. Dragovic et al. / Earth and Planetary Science Letters 413 (2015) 111–122
119
(reaction 2)
2.1 glaucophane + 4.8 white mica (Na-rich) + 7.7 aragonite =
8.8 omphacite + 3.3 quartz + 0.28 white mica (K-rich) + 3.3
lawsonite + 5.6 dolomite + 2.1 magnesite + 1.0 garnet (water conserving reaction)
Limited garnet growth is calculated along Path 3 until a second
growth pulse at ∼ 1.44 GPa and 375 ◦ C. The net reaction for this
second pulse and the subsequent phase of garnet growth, cast for
the temperature interval from 374 to 489 ◦ C, is:
(reaction 3)
2.2 white mica (Na-rich) + 2.1 sphene + 0.96 glaucophane + 1.2
dolomite = 4.0 omphacite + 2.1 rutile + 1.5 lawsonite + 1.5 quartz
+ 0.09 white mica (K-rich) + 1.2 magnesite + 1.0 garnet (water
conserving reaction)
Fig. 7. Comparison of observed and modeled (using Path 3 and 0.15 wt.% CO2 ) garnet
zonation patterns in sample 09DSF-1A, from core (left) to rim (right). Solid lines
represent modeled chemistry, with Xalm in red, Xgrs in blue, Xsps in gray, and Xpyr
in black. Symbols (in equivalent colors) represent the observed garnet chemistry by
electron microprobe. Note that the outer volumetric ∼90% of the garnet is predicted
reasonably well. (For interpretation of the references to color in this figure legend,
the reader is referred to the web version of this article.)
thermodynamic modeling (for all modeled P –T paths and fluid
compositions) and zoned geochronology. The observation of two
additional, much smaller, pulses of garnet growth during the initial growth stages (Fig. 6c) is most consistent with results from
modeled Path 3 (Fig. 6a).
Garnet compositional zonation was also predicted in all model
runs, as shown for path 3 in Fig. 7. At the low temperature of
initial garnet growth, predicted and observed chemistries do not
match well. This could be related to 1) thermodynamic overstepping, 2) limitations of low-T thermodynamic phase parameters
chosen here, and 3) the fact that the microprobe traverse did not
analyze the exact chemical core because the garnet wafer was intentionally cut such that the crystal core was located at the wafer’s
center (to optimize core geochronology).
The predicted garnet composition at later growth stages compares well to the observed microprobe profile in terms of the
relative change in almandine (steadily decreasing), grossular and
pyrope (steadily increasing). Near the garnet rim (zone 10 in gt1,
zone 9 in gt2), a gradual increase in almandine and decrease in
grossular is not reflected by model results. This feature may be a
result of late stage retrogressive metamorphism or dissolution/recrystallization at lower pressures due to a fluid and/or thermal
pulse during decompression.
Overall, the predicted garnet composition (for the outer volumetric 90% of garnet) and the changes in garnet modal abundance, predicted by thermodynamic modeling and observed by
the geochronology, were best fit using input Path 3 (and assuming 0.15 wt.% CO2 ). For this model, the first calculated garnet
growth begins at ∼0.84 GPa and 290 ◦ C, consistent with reports of
∼300 ◦ C garnet growth in subduction zone settings (e.g. Tsujimori
et al., 2006). Very little garnet would be grown at this temperature,
as highlighted by a reaction calculated from the thermodynamic
analysis to describe the net reaction stoichiometry between 286
and 374 ◦ C (normalized for one mole of garnet produced):
(reaction 1: hydrous phases in bold, carbonate phases italicized)
220 feldspar + 2.0 glaucophane + 1.1 white mica (Na-rich) +
0.05 white mica (K-rich) + 0.03 epidote + 5.0 aragonite = 220
omphacite + 220 quartz + 1.6 lawsonite + 3.8 dolomite + 1.3
magnesite + 1.0 garnet (water conserving reaction)
The breakdown of albite in the earliest stages of this reaction sequence dominates the stoichiometry. Recasting over the temperature interval 303–374 ◦ C, following albite breakdown, results in:
The final pulse of garnet growth is predicted to occur along a relatively isobaric heating trajectory. The net garnet-forming reaction,
calculated between 2.06 GPa, 490 ◦ C and 2.19 GPa, 550 ◦ C is:
(reaction 4)
0.75 glaucophane + 0.70 lawsonite + 0.33 white mica (Na-rich)
+ 0.28 magnesite = 1.8 quartz + 1.8 omphacite + 0.01 white mica
(K-rich) + 0.27 dolomite + 1.0 garnet + 2.5 H2 O + 0.01 CO2
Calculations for Path 3 also accurately predict the observed
modal abundances of other matrix phases upon subsequent decompression of the rock to ∼1.0 GPa. The preserved assemblage
now includes ∼90% quartz and plagioclase, with lesser amounts of
magnetite, hematite, white mica, rutile (in garnet), and epidote (in
garnet, after lawsonite). This suggests significant re-equilibration
(and continued dehydration) during exhumation until approximately 1 GPa, with the majority of these preserved phases being produced via breakdown of higher pressure phases such as
omphacite and glaucophane. Subsequently, later-stage infiltration
by aqueous fluids chloritized garnet rims and possibly mediated
reactions that removed any remaining carbonate minerals (Ague
and Nicolescu, 2014). Overall, there is strong agreement between
predictions for Path 3 (with 0.15 wt.% CO2 ) and observed zoned
geochronology and matrix mineralogy in terms of the (i) pattern
of progressive garnet growth, (ii) overall mineralogy and phase
abundances, and (iii) composition and zoning of 90% of the garnet produced.
7. Discussion
7.1. “Pulsed” garnet growth and dehydration during subduction
Similar to that of regional metamorphic studies (Christensen
et al., 1989; Pollington and Baxter, 2010), this study yields total garnet growth durations spanning several million years. Like
Pollington and Baxter (2010) and Dragovic et al. (2012), this study
shows vivid evidence for brief pulses of garnet growth within prolonged tectonic processes. Below we discuss possible causes, and
implications for pulsed garnet growth in this sample.
For preferred P –T Path 3, initiation of garnet growth occurred
early (53.4 ± 2.6 Ma), at ∼0.83 GPa and ∼300 ◦ C. Incipient garnet
growth at such low temperatures has been documented in nature previously (e.g. Tsujimori et al., 2006). In sample 09DSF-1A,
this can be attributed to breakdown of primarily glaucophane
and white mica, producing ∼0.1 vol.% garnet (Fig. 6b; reactions 1
and 2). H2 O produced by this reaction was incorporated into lawsonite, with no net loss as a fluid. A prolonged phase of limited
garnet growth then accompanied continued subduction. This is
consistent with equilibrium thermodynamic modeling of Path 3,
probably related to sequestration of available Mn into the early
120
B. Dragovic et al. / Earth and Planetary Science Letters 413 (2015) 111–122
garnet core. Growth hiatuses are a common feature of recent models that include the progressive effects of porphyroblast crystallization on the reacting bulk matrix (e.g. Konrad-Schmolke et al., 2006,
2008; Caddick et al., 2010; Baxter and Caddick, 2013). Here, the
effect is mediated by continued, but minor, liberation of cations
upon progressive breakdown of carbonate phases. A second pulse
of garnet growth occurred during the later stages of burial (beginning at ∼1.44 GPa and 375 ◦ C) at 47.22 ± 0.36 Ma. This instance of
garnet growth is correlated with breakdown of Na-rich white mica
and glaucophane (reaction 3).
During subsequent, nearly isobaric heating (∼490 ◦ C at 2.06 GPa
to 550 ◦ C at 2.19 GPa), a third garnet growth pulse grew the majority of garnet now in the sample at 44.96 ± 0.53 Ma. This period
of rapid garnet growth is very brief, lasting no longer than 0.8 My.
The resolution of the geochronologic data even allows for effectively instantaneous garnet growth (and thus dehydration), though
this is unlikely. At this stage, garnet growth is mainly produced by
breakdown of lawsonite, glaucophane, and white mica (reaction 4).
A garnet:water molar production ratio of 1:2.5 is associated with
this garnet growth pulse, corresponding to a net bulk rock dehydration of 0.52 wt.%. The final water content predicted in the rock
at this stage is 0.56 wt.%, in line with the observation of 0.47% LOI
in the rock today.
7.2. Burial and heating during Aegean subduction
Coupling geochronologic and P –T data permits calculation of
burial and heating rates, which can be compared to models for
Aegean subduction. Using the average pressure estimated for the
1st and 3rd pulses, and reasonable approximations of uncertainties
on pressure estimates (±0.1 GPa) and geochronology, the descent
rate during garnet growth is calculated to be 4.2 ± 1.0 km/My.
This estimate is similar, albeit slower, than the modern-day Aegean
descent rate of 7.6 km/My (Syracuse et al., 2010). It is also consistent with a range of descent rates (∼2–10 km/My) estimated using
kinematic models for Eocene subduction in the Aegean (Dewey et
al., 1989; Rosenbaum et al., 2002). Indeed, assuming subduction
initiation of the CBU at ∼60 Ma (Ring et al., 2010), and using our
calculated descent rate, high pressure conditions would have been
reached at ∼44 Ma, in agreement with the 3rd pulse of garnet
growth, at 44.96 ± 0.53 Ma. Finally, an average heating rate for the
entire span of garnet growth (using a temperature uncertainty of
±40 ◦ C) is calculated to be 22.3 ± 8.0 ◦ C/My, though this accelerates rapidly during the final pulse of garnet growth to >75 ◦ C/My
(see below).
7.3. Acceleration of garnet growth
Garnet growth accelerated by at least two orders of magnitude during the nearly isobaric heating stage of subduction on
Sifnos. Rapid garnet growth in an interval spanning <1 My has
previously been determined for a blueschist from Sifnos (Dragovic
et al., 2012), occurring during the same period of isobaric heating described here. This brief period of high garnet production
(here constrained at <0.8 My) accompanied rapid heating from
∼490 to ∼550 ◦ C (>75 ◦ C/My), in agreement with previous studies on Sifnos (Groppo et al., 2009; Dragovic et al., 2012; Ashley
et al., 2014) (Fig. 6f). Thermodynamic predictions reveal that the
P –T path crosses a series of closely spaced garnet-growth isopleths at this time, explaining the observed pulse of growth. This
isobaric phase involved an increase in heating rate (from ∼22
to >75 ◦ C/My). Such isobaric heating during subduction can occur either by 1) slow detachment of the rock from the slab into
the subduction channel, causing heating from the mantle wedge
(Gerya et al., 2002), or 2) passage through sharp thermal gradients, as the slab crosses the MDD (maximum depth of decoupling),
and encounters hot mantle material (Wada and Wang, 2009;
van Keken et al., 2011; Wada et al., 2012). While the exact
depth of the transition from partial to full coupling is dependent on several considerations (Syracuse et al., 2010), a range of
depths (∼70–100 km) associated with the MDD falls roughly in
line with maximum pressures obtained in this study, as well as
those previous studies (Groppo et al., 2009; Dragovic et al., 2012;
Ashley et al., 2014).
7.4. Implications for “pulsed” dehydration
Based on thermodynamic forward modeling of the stability
of metamorphic assemblages along a prescribed P –T subduction path, we calculate that 0.52 wt.% water was released during
garnet-forming reactions, with the entirety occurring during a brief
period (<0.8 My) of near isobaric heating. This amount of water release from a subducting lithology is not unusual (Kerrick
and Connolly, 2001; Hacker, 2008; van Keken et al., 2011) but the
“pulsed” nature of its release has not previously been confirmed by
natural observation of the rock record. Dragovic et al. (2012) did
document garnet growth and dehydration spanning <1 My during
subduction but could not resolve the acceleration in growth and
dehydration seen here.
Fig. 6e shows constraints on the changing rate of water production from this particular lithology during subduction. Until
the rock subducts past about ∼2.1 GPa the system is effectively
water-conserving, owing to the formation of lawsonite (see garnetforming reactions 1–3) and its ability to reincorporate water at
high pressure conditions (Vitale-Brovarone and Beyssac, 2014).
Then, in a short span of time and depth, the rock heats, garnet
grows, and significant water is released (at a rate of 0.65 wt.% water per My). Several variables can alter this estimated dehydration
flux during subduction. Subduction rate, subduction dip, and the
thermal state of the slab will determine the P –T path of metamorphic evolution, allowing for differences in the stability of hydrous
phases during burial. The bulk composition and initial hydration
state of the rock will also change dehydration estimates. Many
studies estimated the dehydration of metabasaltic rocks (Kerrick
and Connolly, 2001; Hacker et al., 2003; Dragovic et al., 2012;
Baxter and Caddick, 2013). Hacker (2008) used the composition
of granitic gneiss to model continental subduction, predicting water loss of <0.5 wt.% at the temperature and pressure interval for
garnet growth on Sifnos. This appears to be the most direct comparison that can be made with this Sifnos lithology.
The initial water content of our rock at saturation is low
(∼1 wt.%), and variability in the initial hydration state of subducted rocks can have profound effects on the dehydration flux
(it is likely that more mafic lithologies could contain more water
prior to metamorphic devolatilization). The eclogite-blueschist unit
on Sifnos contains a variety of lithologies, spanning a wide range
of bulk compositions and initial hydration states. Non-uniform initial hydration of a subducted unit may change the depth and flux
of dehydration (Wada et al., 2012). More work on other Aegean
lithologies is thus required in order to determine the effect of
changing bulk composition (and initial water content) on both
dehydration fluxes and rates of garnet growth across a broader
swathe of the subducted slab.
8. Conclusions
Garnet growth is used as a proxy for water release during subduction, with determination of the rates and durations of crystal
growth linked to that of dehydration of a subducting lithology. Precise geochronologic data using Sm–Nd on two very large garnet
porphyroblasts from a quartzofeldspathic gneiss yielded 16 concentric zonal growth ages (9 zones from one garnet, 7 from the
B. Dragovic et al. / Earth and Planetary Science Letters 413 (2015) 111–122
other). Garnet growth occurred in three, distinct phases; initiation
of growth at 53.4 ± 2.6 Ma, a second phase at 47.22 ± 0.36 Ma,
and a major pulse at 44.96 ± 0.53 Ma. The majority of garnet grew
in this final pulse, which occurred over a very brief time span,
with a maximum (2σ ) growth duration of 0.8 My. Thermodynamic
modeling shows that garnet growth during this final pulse was accompanied by water loss of 0.52 wt.%. This study, like other recent
contributions (Camacho et al., 2005; John et al., 2012) shows that
dehydration during subduction can occur in focused pulses, rather
than slowly and continuously, happening on the order of hundreds
of thousands of years or less. Rapid heating during early stages of
exhumation in the subduction channel, or by sharp thermal gradients related to slab-mantle coupling (Wada and Wang, 2009;
van Keken et al., 2011; Wada et al., 2012) could be causes for
pulsed metamorphism and rapid dehydration.
Acknowledgements
We thank Louise Roy for field assistance and XRF preparation,
Jeremy Inglis and Denise Honn for their assistance with the mass
spectrometry, Neel Chatterjee and Eric Reusser for their help with
electron microprobe analyses, Richard Ketcham and Jessie Maisano
for the X-ray computed tomography, and Michael Bröcker for suggestions about field sampling on Sifnos. Finally, we thank Horst
Marschall and Matthias Konrad-Schmolke for insightful and helpful reviews and Tim Elliott for careful and constructive editorial
handling. We gratefully acknowledge NSF Grants EAR-0547999 and
EAR-1250497 (to EFB) and EAR-1250470 (to MJC). The Boston
University TIMS Facility is funded by NSF-0521266 and NSF
EAR-0949390.
Appendix A. Supplementary material
Supplementary material related to this article can be found online at http://dx.doi.org/10.1016/j.epsl.2014.12.024.
References
Altherr, R., Schliestedt, M., Okrusch, M., Seidel, E., Kreuzer, H., Harre, W., Lenz, H.,
Wendt, I., Wagner, G., 1979. Geochronology of high-pressure rocks on Sifnos
(Cyclades, Greece). Contrib. Mineral. Petrol. 70, 245–255.
Ague, J.J., Nicolescu, S., 2014. Carbon dioxide released from subduction zones by
fluid-mediated reactions. Nat. Geosci. 7, 355–360.
Ashley, K.T., Caddick, M.J., Steele-MacInnis, M.J., Bodnar, R.J., Dragovic, B., 2014.
Geothermobarometric history of subduction recorded by quartz inclusions in
garnet. Geochem. Geophys. Geosyst. 15, 350–360.
Austrheim, H., 1987. Eclogitization of lower crustal granulites by fluid migration
through shear zones. Earth Planet. Sci. Lett. 81, 221–232.
Avigad, D., 1993. Tectonic juxtaposition of blueschists and greenschists in Sifnos Island (Aegean Sea) – implications for the structure of the Cycladic blueschist
belt. J. Struct. Geol. 15, 1459–1469.
Baxter, E.F., Caddick, M.J., 2013. Garnet growth as a proxy for progressive subduction
zone dehydration. Geology 9, 427–432.
Baxter, E.F., Scherer, E., 2013. Garnet geochronology: timekeeper of tectonometamorphic processes. Elements 9, 433–438.
Bebout, G.E., 1991. Field-based evidence for devolatilization in subduction zones:
implications for arc magmatism. Science 251, 413–415.
Breeding, C.M., Ague, J.J., Bröcker, M., Bolton, E., 2003. Blueschist preservation
in a retrograded, high-pressure, low-temperature metamorphic terrane, Tinos,
Greece: implications for fluid flow paths in subduction zones. Geochem. Geophys. Geosyst. 4, 1–11.
Bröcker, M., Baldwin, S., Arkudas, R., 2013. The geological significance of 40 Ar/39 Ar
and Rb–Sr white mica ages from Syros and Sifnos, Greece: a record of continuous (re)crystallization during exhumation? J. Metamorph. Geol. 31, 629–646.
Caddick, M.J., Konopásek, J., Thompson, A.B., 2010. Preservation of garnet growth
zoning and the duration of prograde metamorphism. J. Petrol. 51, 2327–2347.
Camacho, A., Lee, J.K., Hensen, B.J., Braun, J., 2005. Short-lived orogenic cycles and
the eclogitization of cold crust by spasmodic hot fluids. Nature 435, 1191–1196.
Christensen, J.N., Rosenfeld, J.L., DePaolo, D.J., 1989. Rates of tectonometamorphic processes from rubidium and strontium isotopes in garnet. Science 244,
1465–1469.
Connolly, J.A.D., 2009. The geodynamic equation of state: what and how. Geochem.
Geophys. Geosyst. 10, Q10014. http://dx.doi.org/10.1029/2009GC002540.
121
Dewey, J.F., Helman, M.L., Torco, E., Hutton, D.H.W., Knott, S.D., 1989. Kinematics of
the western Mediterranean. Geol. Soc. (Lond.) Spec. Publ. 45, 265–283.
Dragovic, B., Samanta, L.M., Baxter, E.F., Selverstone, J., 2012. Using garnet to constrain the duration and rate of water-releasing metamorphic reactions during
subduction: an example from Sifnos, Greece. Chem. Geol. 314–317, 9–22.
Forster, M.A., Lister, G.S., 2005. Several distinct tectono-metamorphic slices in the
Cycladic eclogite-blueschist belt, Greece. Contrib. Mineral. Petrol. 150, 523–545.
Gaidies, F., de Capitani, C., Abart, R., Schuster, R., 2008. Prograde garnet growth along
complex P –T –t paths: results from numerical experiments on polyphase garnet
from the Wölz complex (Austroalpine basement). Contrib. Mineral. Petrol. 155,
673–688.
Gerya, T.V., Stöckhert, B., Perchuk, A.L., 2002. Exhumation of high-pressure metamorphic rocks in a subduction channel: a numerical simulation. Tectonics 21,
6-1–6-15.
Groppo, C., Forster, M., Lister, G., Compagnoni, R., 2009. Glaucophane schists and
associated rocks from Sifnos (Cyclades, Greece): new constraints on the P –T
evolution from oxidized systems. Lithos 109, 254–273.
Grove, T.L., Chatterjee, N., Parman, S.W., Medard, E., 2006. The influence of H2 O on
mantle wedge melting. Earth Planet. Sci. Lett. 249, 74–89.
Hacker, B.R., Abers, G.A., Peacock, S.M., 2003. Subduction factory 1. Theoretical mineralogy, densities, seismic wave speeds, and H2 O contents. J. Geophys. Res. 108,
2029.
Hacker, B.R., 2008. H2 O subduction beyond arcs. Geochem. Geophys. Geosyst. 9,
1–24.
Harvey, J., Baxter, E.F., 2009. An improved method for TIMS high precision
neodymium isotope analysis of very small aliquots (1–10 ng). Chem. Geol. 258,
251–257.
Holland, T., Powell, R., 1998. An internally consistent thermodynamic dataset for
phases of petrologic interest. J. Metamorph. Geol. 16, 309–343.
John, T., Schenk, V., 2003. Partial eclogitisation of gabbroic rocks in a late Precambrian subduction zone (Zambia): prograde metamorphism triggered by fluid
infiltration. Contrib. Mineral. Petrol. 146, 174–191.
John, T., Gussone, N., Podladchikov, Y.Y., Bebout, G.E., Dohmen, R., Halama, R., Klemd,
R., Magna, T., Seitz, H.-M., 2012. Volcanic arcs fed by rapid pulsed fluid flow
through subducting slabs. Nat. Geosci.. http://dx.doi.org/10.1038/NGEO1482.
Kerrick, D.M., Connolly, J.A.D., 2001. Metamorphic devolatilization of subducted
oceanic metabasalts: implications for seismicity, arc magmatism and volatile recycling. Earth Planet. Sci. Lett. 189, 19–29.
Konrad-Schmolke, M., Babist, J., Handy, M.R., O’Brien, P.J., 2006. The physicochemical properties of a subducted slab from garnet zonation patterns (Sesia
Zone, Western Alps). J. Petrol. 47, 2123–2148.
Konrad-Schmolke, M., O’Brien, P.J., De Capitani, C., Carswell, D.A., 2008. Garnet
growth at high- and ultra-high pressure conditions and the effect of element
fractionation on mineral modes and composition. Lithos 103, 309–332.
Konrad-Schmolke, M., O’Brien, P.J., Zack, T., 2011. Fluid migration above a subducted
slab – Constraints on amount, pathways and major element mobility from partially overprinted eclogite-facies rocks (Sesia Zone, Western Alps). J. Petrol. 52,
457–486.
Marmo, B.A., Clarke, G.L., Powell, R., 2002. Fractionation of bulk rock composition
due to porphyroblast growth: effects on eclogite facies mineral equilibria, Pam
Peninsula, New Caledonia. J. Metamorph. Geol. 20, 151–165.
Matthews, A., Schliestedt, M., 1984. Evolution of the blueschist and greenschist
rocks of Sifnos, Cyclades, Greece. Contrib. Mineral. Petrol. 88, 150–163.
Okrusch, M., Bröcker, M., 1990. Eclogites associated with high-grade blueschists in
the Cyclades archipelago, Greece: a review. Eur. J. Mineral. 2, 451–478.
Peacock, S.M., 1990. Numerical simulation of metamorphic pressure–temperature–
time paths and fluid production in subducting slabs. Tectonics 9, 1197–1211.
Pollington, A.D., Baxter, E.F., 2010. High resolution Sm–Nd garnet geochronology
reveals the uneven pace of tectonometamorphic processes. Earth Planet. Sci.
Lett. 293, 251–257.
Pollington, A.D., Baxter, E.F., 2011. High precision microsampling and preparation
of zoned garnet porphyroblasts for Sm–Nd geochronology. Chem. Geol. 281,
270–282.
Ring, U., Layer, P.W., 2003. High-pressure metamorphism in the Aegean, eastern
Mediterranean: underplating and exhumation from the Late Cretaceous until
the Miocene to Recent above the retreating Hellenic subduction zone. Tectonics 22, 1022.
Ring, U., Glodny, J., Will, T., Thomson, S., 2010. The Hellenic subduction system:
high-pressure metamorphism, exhumation, normal faulting, and large-scale extension. Annu. Rev. Earth Planet. Sci. 38, 45–76.
Rosenbaum, G., Lister, G.S., Duboz, C., 2002. Relative motion of Africa, Iberia, and
Europe during Alpine orogeny. Tectonophysics 359, 117–129.
Rüpke, L.H., Morgan, J.P., Hort, M., Connolly, J.A., 2002. Are the regional variations
in Central American arc lavas due to differing basaltic versus peridotitic slab
sources of fluids? Geology 30, 1035–1038.
Schmädicke, E., Will, T.M., 2003. Pressure–temperature evolution of blueschist facies
rocks from Sifnos, Greece, and implications for the exhumation of high-pressure
rocks in the Central Aegean. J. Metamorph. Geol. 21, 799–811.
122
B. Dragovic et al. / Earth and Planetary Science Letters 413 (2015) 111–122
Schmidt, M.W., Poli, S., 1998. Experimentally based water budgets for dehydrating
slabs and consequences for arc magma generation. Earth Planet. Sci. Lett. 163,
361–379.
Syracuse, E.M., van Keken, P.E., Abers, G.A., 2010. The global range of subduction
zone thermal models. Phys. Earth Planet. Inter. 183, 73–90.
Trotet, F., Jolivet, L., Vidal, O., 2001a. Tectono–metamorphic evolution of Syros and
Sifnos Islands (Cyclades, Greece). Tectonophysics 338, 179–206.
Trotet, F., Jolivet, L., Vidal, O., 2001b. Exhumation of Syros and Sifnos metamorphic
rocks (Cyclades, Greece). New constraints on the P –T paths. Eur. J. Mineral. 13,
901–920.
Tsujimori, T., Sisson, V.B., Liou, J.G., Harlow, G.E., Sorensen, S.S., 2006. Very-lowtemperature record of the subduction process: a review of worldwide lawsonite
eclogites. Lithos 92, 609–624.
Ulmer, P., 2001. Partial melting in the mantle wedge – the role of H2 O in the genesis
of mantle-derived ‘arc-related’ magmas. Phys. Earth Planet. Inter. 127, 215–232.
van Keken, P.E., Kiefer, B., Peacock, S., 2002. High-resolution models of subduction
zones: implications for mineral dehydration reactions and the transport of water
into the deep mantle. Geochem. Geophys. Geosyst. 3, 1–20.
van Keken, P.E., Hacker, B.R., Syracuse, E.M., Abers, G.A., 2011. Subduction factory:
4. Depth-dependent flux of H2 O from subducting slabs worldwide. J. Geophys.
Res. 116, B01401. http://dx.doi.org/10.1029/2010JB007922.
Vitale-Brovarone, A., Beyssac, O., 2014. Lawsonite metasomatism: a new route for
water to the deep Earth. Earth Planet. Sci. Lett. 393, 275–284.
Wada, I., Wang, K., 2009. Common depth of slab-mantle decoupling: reconciling
diversity and uniformity of subduction zones. Geochem. Geophys. Geosyst. 10,
Q10009. http://dx.doi.org/10.1029/2009GC002570.
Wada, I., Behn, M.D., Shaw, A.M., 2012. Effects of heterogeneous hydration
in the incoming plate, slab rehydration, and mantle wedge hydration on
slab-derived H2 O flux in subduction zones. Earth Planet. Sci. 353–354,
60–71.
Wijbrans, J.R., Schliestedt, M., York, D., 1990. Single grain argon laser probe dating
of phengites from the blueschist to greenschist transition on Sifnos (Cyclades,
Greece). Contrib. Mineral. Petrol. 104, 582–593.
Zack, T., John, T., 2007. An evaluation of reactive fluid flow and trace element mobility in subducting slabs. Chem. Geol. 239, 199–216.