Phase equilibria constraints on melting of stromatic migmatites from Ronda (S Spain): insights on the formation of peritectic garnet.

Phase equilibria constraints on melting of stromatic migmatitesfrom Ronda (S. Spain): insights on the formation of peritecticgarnet

O. BARTOLI ,1 L . TAJ �CMANOV �A,2 B. CESARE1 AND A. ACOSTA-VIGIL3

1Dipartimento di Geoscienze, Universit�a di Padova, Via Gradenigo 6, Padova, 35131, Italy ([email protected])2Department of Earth Sciences, Swiss Federal Institute of Technology, Zurich, 8092, Switzerland3Instituto Andaluz de Ciencias de la Tierra, Consejo Superior de Investigaciones Cient�ıficas-Universidad de Granada,Armilla, Granada, 18100, Spain

ABSTRACT Stromatic metatexites occurring structurally below the contact with the Ronda peridotite (Oj�en nap-pe, Betic Cordillera, S Spain) are characterized by the mineral assemblage Qtz+Pl+Kfs+Bt+Sil+Grt+Ap+Gr+Ilm. Garnet occurs in low modal amount (2–5 vol.%). Very rare muscovite is present asarmoured inclusions, indicating prograde exhaustion. Microstructural evidence of melting in themigmatites includes pseudomorphs after melt films and nanogranite and glassy inclusions hosted ingarnet cores. The latter microstructure demonstrates that garnet crystallized in the presence of melt.Re-melted nanogranites and preserved glassy inclusions show leucogranitic compositions. Phase equi-libria modelling of the stromatic migmatite in the MnO–Na2O–CaO–K2O–FeO–MgO–Al2O3–SiO2–H2–O2–C (MnNCaKFMASHOC) system with graphite-saturated fluid shows P–T conditions ofequilibration of 4.5–5 kbar, 660–700 °C. These results are consistent with the complete experimentalre-melting of nanogranites at 700 °C and indicate that nanogranites represent the anatectic melt gen-erated immediately after entering supersolidus conditions. The P–T estimate for garnet and meltdevelopment does not, however, overlap with the low-temperature tip of the pure melt field in thephase diagram calculated for the composition of preserved glassy inclusions in garnet in the Na2O–CaO–K2O–FeO–MgO–Al2O3–SiO2–H2O (NCKFMASH) system. A comparison of measured meltcompositions formed immediately beyond the solidus with results of phase equilibria modelling pointsto the systematic underestimation of FeO, MgO and CaO in the calculated melt. These discrepanciesare present also when calculated melts are compared with low-T natural and experimental melts fromthe literature. Under such conditions, the available melt model does not perform well. Given thepresence of melt inclusions in garnet cores and the P–T estimates for their formation, we argue thatsmall amounts (<5 vol.%) of peritectic garnet may grow at low temperatures (≤700 °C), as a result ofcontinuous melting reactions consuming biotite.

Key words: melt inclusions; nanogranites; peritectic garnet; phase equilibria modelling with graphite;Ronda migmatites.

INTRODUCTION

During medium- to low-pressure partial melting ofAl-rich metasedimentary rocks, garnet is consideredto be a product of fluid-absent biotite melting reac-tions (see Thompson, 1982; Grant, 1985; Spear et al.,1999; Vielzeuf & Schmidt, 2001; Brown, 2010). Melt-ing experiments on natural rocks and synthetic mix-tures demonstrate that biotite-breakdown reactions atmedium-to-low–P occur mostly at ≥800 °C (seeClemens, 2006 and references therein). During ana-texis, peritectic garnet can trap droplets of melt – i.e.melt inclusions (MI) – produced by the same incon-gruent melting reaction that generates the host min-eral (Cesare et al., 2011). Some MI trapped inperitectic garnet of migmatitic granulites from the

Kerala Khondalite Belt (India) show chemical com-positions very far from a ‘minimum melt’, indicatingmelting conditions well above minimum or eutectictemperatures. Such an observation is in agreementwith a high–T origin of these rocks and presumablyreflects garnet growth during biotite-breakdownmelting (Cesare et al., 2009; Ferrero et al., 2012).However, Acosta-Vigil et al. (2010, 2012) haverecently demonstrated that MI trapped in peritecticgarnet from metapelitic enclaves partially melted at~5–7 kbar have geochemical signatures typical ofmuscovite melting at ~700–750 °C. In addition,Bartoli et al. (2013b) have experimentally re-meltedat 700 °C the nanogranites (i.e. crystallized MI)hosted in garnet of low-P migmatites from Ronda(Spain), indicating the occurrence of low-T anatectic

© 2013 John Wiley & Sons Ltd 775

J. metamorphic Geol., 2013, 31, 775–789 doi:10.1111/jmg.12044

melts within that garnet. This is in contrast with tra-ditional models of pelite melting by which generationof peritectic garnet at conditions of muscovite melt-ing should occur only at high-P (>10 kbar), when thefluid-absent biotite breakdown melting reaction inter-sects the muscovite fluid-absent melting curve (Clem-ens & Vielzeuf, 1987). On the other hand, phaseequilibria modelling based on large, internally consis-tent thermodynamic data sets has improved ourunderstanding of crustal melting in the last decade(see White et al., 2011). Although thermodynamiccalculations on melt-bearing metasedimentary rocksshow that peritectic garnet may be stable along withbiotite and/or muscovite in multivariant fields at low-T, low-P conditions (e.g. White et al., 2004; Johnsonet al., 2008; Groppo et al., 2012; Imayama et al.,2012; Riel et al., 2013), the formation of peritecticgarnet in natural samples under these conditions hasnot been carefully investigated so far.

From the above and considering the continuouslygrowing list of case studies in which the geochemicalvariability of S-type granitic magmas is interpreted asdue to the entrainment of peritectic minerals (seeClemens & Stevens, 2012 and references therein), it isimportant to investigate the role of peritectic garnetduring low-T, low-P melting, integrating field andpetrographic evidence, phase equilibria calculationsand experiments. MI testify for the presence of meltin the rock and, through the analysis of their majorand trace element composition, can be used to relatemelting events to particular anatectic reactions(Cesare et al., 2009, 2011; Acosta-Vigil et al., 2010).To study the generation of garnet during low-T ana-texis, we report the occurrence of garnet with MI ina stromatic metatexite from Ronda (S Spain) andconstruct P–T diagrams using: (i) the bulk-rockcomposition of the graphite bearing migmatite (corre-sponding to a Ca-poor, Si-rich peraluminous meta-greywacke) and, for the first time in the literature, (ii)the major element composition of MI hosted in thegarnet. The modelling allows discussion of the reli-ability of thermodynamic data for granitic melts, andconfirms the petrographic and experimental evidenceof peritectic garnet growth at, or soon after, musco-vite breakdown through the continuous melting reac-tion involving biotite. Therefore, garnet can beconsidered a good candidate for trapping also theearliest formed crustal melts.

GEOLOGICAL SETTING

The study area is situated in the western part of theBetic Cordillera, S Spain (Fig. 1a). During the con-vergence of Africa, Eurasia and the Alboran domainfrom Late Cretaceous to Tertiary times, large bodiesof subcontinental lithospheric mantle (i.e. the Rondaperidotites; Obata, 1980) have been exhumed andemplaced within the continental crust as a tectonicslab, in the Internal Zone of the Betic Cordillera

(Lundeen, 1978; Tub�ıa et al., 1997). The InternalZone of this orogenic belt consists of metamorphicrocks of Palaeozoic and Triassic age organized intotwo main tectonic complexes, from bottom to top:the Alpuj�arride and Mal�aguide (Platt et al., 2013).The Alpuj�arride complex encompasses a stack ofmetamorphosed tectonic units whose number variesalong different traverses (Aldaya et al., 1979). TheRonda peridotites form the lower portion of the LosReales nappe, the highest unit of the Alpuj�arridecomplex in the study area. The mantle rocks are em-placed over the Guadaiza and Ojen nappes and cropout primarily in three massifs: Sierra Bermeja, SierraAlpujata and Carratraca (Fig. 1b). Guadaiza andOjen nappes show a typical Alpuj�arride lithologicalsequence with metapelites at the bottom and marbleson the top (Tub�ıa et al., 1997). The Ojen napperecords high-P conditions with eclogites (Tub�ıa & GilIbarguchi, 1991); conversely, the Guadaiza nappeexhibits only low-P conditions (Esteban et al., 2008).The study area includes two main tectonic units: theLos Reales nappe and the underlying Ojen nappe(Fig. 1c). The emplacement of the Ronda peridotitesover metasedimentary sequences produced high-Tmetamorphism and partial melting in the adjacentcrustal rocks (Torres-Rold�an, 1983; Tub�ıa & Cuevas,1986; Tub�ıa et al., 1997) during the Alpine Orogeny(c. 20 Ma; Platt & Whitehouse, 1999; Esteban et al.,2011a), although zircon U-Pb isotopic analyses haveyielded both Hercynian (c. 300 Ma) and Alpine (c.20 Ma) ages, suggesting the existence of an olderanatectic event (Acosta, 1998; S�anchez-Rodr�ıguez,1998). In this unusual “contact aureole”, the degreeof melting but also the intensity of deformationincrease towards the contact with the peridotites, andthe migmatites (hereafter Ronda migmatites) evolvefrom metatexites, through diatexites, and to myloniticdiatexites (Torres-Rold�an, 1983; Tub�ıa, 1988; Acosta-Vigil et al., 2001; Esteban et al., 2008).The stromatic migmatites were collected in the

metamorphic envelope of the Sierra Alpujata massif(Ojen nappe, Tub�ıa et al., 1997), to the NW of SierraAlpujata and ~400 m below the contact with the peri-dotite (measured perpendicular to the contact,Fig. 1c). The metasedimentary sequence exposedbelow the Sierra Alpujata massif has a maximumthickness of ~700 m, but varies along different tra-verses (Tub�ıa et al., 1997). Based on the Grt-Btexchange thermometer and GASP barometer, Tub�ıaet al. (1997) reported metamorphic peak conditionsof ~800 °C and ~8.3 kbar for the mylonitic migma-tites at the contact with the peridotites. Conversely,there are no previous P–T estimates on the stromaticmigmatites outcropping at the base of the anatecticsequence. Overall, crustal anatexis at Sierra Alpujatais still poorly characterized, especially as concernsmelting reactions and conditions, fluid regimesand melt compositions through the migmatiticsequence.

© 2013 John Wiley & Sons Ltd

776 O. BARTOL I ET AL .

ANALYTICAL METHODS

The compositions of muscovite, biotite, ilmenite, feld-spar and glass were obtained using a Jeol JXA 8200Superprobe at the Dipartimento di Scienze dellaTerra, Universit�a di Milano (Italy). Analyticalparameters for minerals were: 15 kV acceleratingvoltage, 5 nA current, counting time of 30 s on peakand 10 s on background. Analytical parameters forglass were: 15 kV accelerating voltage, 2 nA current,1 lm beam diameter and a counting time of 10 s onpeak and 2 s on background. Na, K, Al and Si wereanalyzed first. Owing to Na loss during electronmicroprobe (EMP) analysis of rhyolitic glasses witheffects also on K, Al and Si (Morgan & London,1996, 2005), concentrations were corrected by analyz-ing leucogranitic glass standards. Details concerningthe application of correction factors and the compo-sition of the standard glasses are given by Ferrero

et al. (2012) and Bartoli et al. (2013a,b). Garnet com-positions were determined using the Cameca SX50microprobe of the C.N.R.-I.G.G. (Consiglio Nazio-nale delle Ricerche-Istituto di Geoscienze e Geori-sorse) at the Dipartimento di Geoscienze, Universit�adi Padova, Italy. Measurements were performedusing 20 kV accelerating voltage, 20 nA beam cur-rent, counting time of 10 s on peak and 5 s on back-ground. Natural and synthetic silicates and oxideswere used as standards.Back-scattered electron (BSE) imaging and semi-

quantitative energy dispersive spectroscopy (EDS)were carried out on crystallized and glassy inclusionsusing a CAM SCAN MX2500, equipped with LaB6

cathode, at the Dipartimento di Geoscienze, Univer-sit�a di Padova (Italy) and a Jeol JSM–6500F thermalField Emission Scanning Electron Microscope (FE-SEM), at INGV (Istituto Nazionale di Geofisica eVulcanologia), Rome, Italy.

(a)

(c)

(b)

Fig. 1. (a) Location map of the study areain S. Spain. (b) Simplified geological mapof the western sector of the Betic Cordillera(modified after Esteban et al., 2011b).(c) Geological map of the Sierra Alpujatamassif. Star shows the location of themetatexites (N 36°36′37.6′, W4°49′15.6′).


CONSTRA INTS ON MELT ING OF RONDA MIGMAT ITES 777

SAMPLE DESCRIPTION

Metatexite petrography

The Ronda metatexites have a stromatic structurewith thin (≤1 cm) discontinuous layers of leucosome

surrounded by a fine-grained mesocratic matrix com-posed of biotite, fibrolitic sillimanite, garnet, graph-ite, quartz, plagioclase, K-feldspar, apatite andilmenite (Fig. 2). The foliation in the rock is definedby abundant oriented biotite grains generally clu-stered with sillimanite (Fig. 3a). Muscovite(~1 vol.%) is very rare, anhedral, and appearsincluded in K-feldspar porphyroblats or associatedwith biotite and fibrolitic sillimanite (Fig. 3b,c).Graphite (<1 vol.%) is randomly distributed in thematrix, whereas ilmenite and apatite are generallyincluded within biotite and sillimanite aggregates(Fig. 3a). Alkali feldspar is often poikiloblastic, con-taining inclusions of quartz, plagioclase, biotite andsillimanite (Fig. 3d). Myrmekite is present along theboundaries of some K-feldspar porphyroblasts.Garnet (2–5 vol.%) occurs as small (50–200 lm indiameter) subhedral to euhedral crystals both in themesocratic matrix (Fig. 3e) and in leucosomes

Fig. 2. (a) Field aspect of the investigated stromatic metatexiteALP1 at Ronda. White arrow: thin leucosomes subparallel tofoliation.

(a) (b)

(e)

(g)

(f)

(d)

(c)

(h)

Fig. 3. Microstructures in the stromaticmetatexite ALP1. (a) Photomicrographshowing the fine-grained mesocraticmatrix. White arrow: graphite. Red arrow:melt inclusions (MI)-bearing garnet.Yellow arrow: ilmenite enclosed in aBt+Sil aggregate. Plane-polarized light.(b) Resorbed muscovite armoured inK-feldspar porphyroblast. Crossed polars.(c) Primary muscovite partially replacedby fibrolite. Plane-polarized light.(d) K-feldspar poikiloblast with inclusionsof quartz, plagioclase and biotite.(e) Euhedral garnet present in themesocratic matrix. Red arrows: MIclusters at the core. PPL. (f) SEM BSEimage showing a MI-bearing garnethosted in leucosome. (g) Meltpseudomorph: K-feldspar with cuspateoutlines that has probably crystallizedfrom a pool of melt (see Holness et al.,2011). The reactant minerals, quartz andplagioclase, are rounded and resorbed.Crossed polars with 530 nm k plate.(h) Euhedral faces of feldspar, suggestingcrystal growth from melt (Vernon, 2011)in a leucosome. Crossed polars.



(Fig. 3f). Most (~90%) garnet crystals contain clus-ters of several MI (Fig. 3e,f). Leucosomes (~5 vol.%)are medium-grained and contain quartz, plagioclase,K-feldspar, biotite and rare garnet. Feldspar in leuco-some may include quartz and garnet, and may showeuhedral shapes (Fig. 3h).

In the metatexites, the most convincing evidence ofthe occurrence of melt throughout the entire pro-grade history are: (i) MI-bearing garnet (Fig. 3e, f),(ii) mineral pseudomorphs after melt films and pools(Fig. 3g) and (iii) leucosomes containing mineralswith euhedral shapes (Fig. 3h).

Melt inclusions

Clusters of MI display a subspherical geometry andare preferentially located at garnet cores (Fig. 3e).Locally, the clusters of MI may be closer to the rimsof the host, but this is observed in those cases wherethe host is partially resorbed (Fig. 3f). MI do notform arrays along linear discontinuities of the hostcrystal, and their zonal arrangement is a strong indi-cator of the primary nature of MI – i.e. that theywere trapped when garnet was growing (Roedder,1984; Frezzotti, 2001). MI range from approximately2 to 10 lm in size and show a variable degree ofcrystallization in the same cluster (Fig. 4a), rangingfrom totally crystallized MI (hereafter nanogranites;Cesare et al., 2009), to partially crystallized MI,down to very rare crystals-absent MI (hereafterglassy MI). Generally, nanogranite inclusions in theserocks are aggregates of quartz, biotite, muscovite andplagioclase (Fig. 4a). K-feldspar has also beenobserved in a few inclusions (Fig. 4b). These MI maydisplay a diffuse micro- to nano-porosity (Fig. 4b),which contains liquid H2O at ambient temperature(Bartoli et al., 2013b). Unlike in other partiallymelted rocks (see Cesare et al., 2007; Ferrero et al.,2011), fluid inclusions cogenetic with MI were notobserved within garnet of the metatexites.

Nanogranites and partially crystallized MI havebeen completely re-homogenized using a piston cylin-der apparatus at conditions of 700 °C and 5 kbar(Bartoli et al., 2013b). Higher experimental tempera-tures (750 & 800 °C) produced dissolution of the hostgarnet into the melt, and inclusion decrepitation(Bartoli et al., 2013a), suggesting that the trappingtemperature was significantly exceeded during theseexperimental runs (Frezzotti, 2001; Danyushevskyet al., 2002). The temperature of 700 °C can thereforebe considered a maximum value for the entrapmentof MI in garnet.

Phase chemistry and bulk-rock composition

Table 1 reports representative EMP analyses of theminerals and the bulk rock composition, whereas thedescription below refers to the set of EMP data inTables S1–S5.

Biotite shows constant XMg values (0.33–0.35; Mg/Mg+Fe). Ti content ranges from 0.42 to 0.49 atomsper unit formula (apfu). F and Cl contents are low(0.2–0.4 and 0.0–0.1 wt% respectively). Garnet is analmandine-rich solid solution, with modest zoning inthe pyrope and spessartine components. No chemicalvariations between MI-free and MI-bearing garnetwere observed. Garnet cores are characterized byAlm77–78Prp11–13Sps07–09Grs03–04. Towards the rims,garnet displays an increase in spessartine componentand a slight decrease in XMg (from 0.12–0.14 in the

(a)

(b)

Fig. 4. (a) SEM BSE image of coexisting preserved glassy meltinclusions (MI) and nanogranites containing biotite, quartz,muscovite and plagioclase. White arrow: some glassy MIcontain trapped minerals (e.g. zircon or apatite) attached to, orpartially enclosed in, the host garnet. (b) SEM BSE image of ananogranite inclusion composed of quartz, K-feldspar,plagioclase and biotite. White arrows: micro- to nano-porosity.



Table

1.Representativeelectronmicroprobeanalysesofallphasesfrom

Rondametatexiteandbulk

rock

composition.

Bt

Grt

(core)

Grt

(rim

)Pl

Kfs

Ms

Ilm

Melt

Bulk

rock

Matrix

Included

inKfs

MI-brg

MI-brg

MI-free

MI-brg

MI-free

Matrix

Leuco-some

Matrix

Included

inKfs

Matrix

Matrix

GlassyMI

SiO

235.26

35.51

36.83

36.55

36.84

36.34

36.79

60.88

62.20

65.63

46.26

45.61

0.02

68.42

71.86

TiO

24.22

3.88

0.00

0.04

0.01

0.01

0.00

0.01

0.04

0.04

0.29

0.03

52.81

0.00

0.47

Al 2O

319.96

18.98

21.16

20.77

21.46

21.00

21.35

25.06

23.88

18.65

35.06

35.87

0.02

11.57

14.53

Cr 2O

30.11

0.00

n.a.

n.a.

n.a.

n.a.

n.a.

0.00

0.02

0.00

0.04

0.00

0.07

n.a.

n.a.

FeO

22.16

21.29

35.14

35.28

34.72

33.98

33.91

n.a.

n.a.

n.a.

1.84

1.19

45.07

1.26

b3.07

MnO

0.21

0.15

3.29

3.48

3.04

4.11

4.72

0.13

0.00

0.01

0.01

0.00

0.95

0.19

0.05

MgO

6.47

6.33

3.18

2.81

3.13

2.62

2.47

0.01

0.00

0.01

0.82

0.42

0.39

0.06

0.75

CaO

0.04

0.02

0.91

1.26

0.90

1.02

0.92

6.08

5.11

0.07

0.01

0.00

0.02

0.31

1.32

Na2O

0.21

0.19

0.00

0.01

0.00

0.00

0.00

7.90

8.50

2.23

0.53

0.47

0.00

3.37

2.20

K2O

9.26

9.16

n.a.

n.a.

n.a.

n.a.

n.a.

0.36

0.46

13.41

10.41

10.65

0.00

4.05

4.65

BaO

0.03

0.18

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

0.04

0.15

n.a.

n.a.

n.a.

P2O

5n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

0.04

0.24

F0.39

0.22

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

0.00

0.00

n.a.

n.a.

n.a.

Cl

0.07

0.07

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

0.01

0.01

n.a.

n.a.

n.a.

LOI

0.72

Total

98.39

95.98

100.50

100.20

100.08

99.09

100.17

100.43

100.22

100.05

95.33

94.40

99.35

89.27

99.86

O=F,Cl

�0.18

-0.11

0.00

0.00

Total

98.21

95.87

95.33

94.40

22(O

)12(O

)8(O

)22(O

)3(O

)

Si

5.260

5.414

2.967

2.966

2.970

2.972

2.976

2.697

2.754

2.998

6.170

6.133

0.001

Ti

0.474

0.445

0.000

0.002

0.001

0.000

0.000

0.000

0.001

0.001

0.029

0.003

1.004

Al

3.509

3.410

2.009

1.987

2.039

2.024

2.036

1.308

1.246

1.004

5.511

5.685

0.001

Cr

0.013

0.000

0.004

0.000

0.001

Fe2

+2.764

2.714

2.367

2.394

2.340

2.324

2.294

0.005

0.000

0.000

0.205

0.134

0.953

Mn

0.027

0.019

0.224

0.239

0.207

0.285

0.323

0.000

0.001

0.000

0.002

0.000

0.020

Mg

1.439

1.439

0.381

0.340

0.376

0.320

0.298

0.001

0.000

0.001

0.164

0.084

0.015

Ca

0.006

0.004

0.078

0.109

0.078

0.089

0.080

0.289

0.242

0.003

0.001

0.000

0.001

Na

0.062

0.057

0.000

0.002

0.000

0.001

0.000

0.678

0.730

0.198

0.138

0.122

K1.762

1.782

0.020

0.026

0.782

1.771

1.827

Ba

0.002

0.011

0.002

0.008

F0.184

0.026

0.000

0.000

Cl

0.018

0.017

0.002

0.002

XMg

0.34

0.35

0.14

0.12

0.14

0.12

0.12

0.02

0.49

XAlm

0.78

0.78

0.78

0.77

0.77

XPrp

0.12

0.11

0.13

0.11

0.10

XSps

0.07

0.08

0.07

0.09

0.11

XGrs

0.03

0.04

0.03

0.03

0.03

XAb

0.69

0.73

0.20

XAn

0.29

0.24

0.00

XOr

0.02

0.03

0.80

ASI

1.10

1.31

H2O

a3.1–7

.6

MI,meltinclusions;MI-brg,MI-bearinggarnet;MI-free,garnet

free

ofMI;ASI,AluminaSaturationIndex

(=mol.Al 2O3/(CaO

+Na 2O+K2O);n.a.,notanalyzed.

aRangeofH

2O

contents

determined

byRamanspectroscopyonre-homogenized

MI[see

Bartoliet

al.(2013b)].

bFeasFe 2O

3.



core to 0.11–0.12 in the rim). The grossular componentis homogeneous throughout crystals. The rim of garnethas a composition of Alm76–77Prp10–11Sps09–12Grs02–03.

Plagioclase composition varies according to themicrostructural position. The most anorthitic plagio-clase (with composition Ab66–69An29–33Or1–2) is eitherpresent in the matrix or included in K-feldspar por-phyroblasts. Slightly more albitic and orthoclasericher grains are present in leucosomes (Ab73–77An21–24Or2–3). The latter probably represents the plagio-clase component crystallized from the anatectic melt(Sawyer, 2001; Hasalov�a et al., 2008).

Alkali feldspar porphyroblasts have a compositionof Or78–82Ab17–22An0–1. No compositional differencesare observed between these grains and those in leuco-somes (Or78–82Ab18–21An0–1). Ilmenite shows constantcomposition and is characterized by Fetot ~0.95 apfu,low Mn <0.02 apfu and XMg = 0.01–0.02. Muscoviteshows an Si content from 6.13 to 6.27 apfu andcontains little Na (0.10–0.17 apfu) and Mg (0.06–0.16 apfu). Fe ranges from 0.08 to 0.21 apfu (corre-sponding to 0.74–1.84 wt% FeO).

The composition of melt trapped in preservedglassy MI is leucogranitic (SiO2 ~68–71 wt%;FeO+MgO+MnO+TiO2 <1.5) and peraluminous[ASI = 1.08–1.25; ASI =mol. Al2O3/(CaO+Na2O+K2O)].Re-homogenized nanogranites have bulk composi-tions similar to those of glassy inclusions (cf. fig. 5 inBartoli et al., 2013a). The H2O content of the glassin the re-homogenized MI after experiments at700 °C ranges from 3.1 to 7.6 wt%. The occurrencewithin the same crystal of MI showing comparablecompositions but variable degrees of crystallizationhas been explained as due to the effect of several fac-tors that affect the kinetics of nucleation, such as thesize of MI, and the presence of impurities within theinclusions and/or irregularities on the MI wall(Cesare et al., 2011; Ferrero et al., 2012). The bulk-rock composition corresponds to that of Ca-poor,Si-rich peraluminous greywacke with lower aluminacontent (Al2O3 = 14.5 wt%) and aluminium satura-tion index (ASI = 1.3) than typical pelitic rocks.

PHASE EQUILIBRIA MODELLING

In the following section, the P–T conditions of equili-bration for the metatexite and the P–T conditions atwhich the melt inclusions were trapped are investi-gated. For the migmatite, the model chemical systemMnNCaKFMASHOC was used with the bulk-rockcomposition obtained from XRF analysis. Field andpetrographic observations (i.e. lack of leucosome net-works) and chemical data (i.e. lack of residual bulk-rock compositions) suggest no melt extraction andloss from the metatexites (Solar & Brown, 2001;Brown, 2007). Although the Mn content of this rockis low (<0.1 wt%), MnO was included in the model-ling due to its influence on the stability of garnet atlow temperatures (Spear, 1993; Tinkham et al.,

2001). TiO2 was not considered due to its detrimentaleffect on modal proportions of phases at the P–Tconditions of interest (see Discussion for details). Theoccurrence of graphite requires the system to containa graphite saturated C–O–H (COH) fluid, where theH2O activity must be lowered owing to the presenceof diluting carbonic species such as CH4 and/or CO2

(Connolly & Cesare, 1993). Under the assumptionthat the fluid in the rock is essentially produced byH2O release from phyllosilicates, the amount of H2

and O2 components in the fluid is constrained at aratio 2:1. At this initial condition (i.e. XO = 1/3 ofConnolly, 1995), the fluid composition contains themaximum activity of water for a graphite saturatedCOH fluid. However, given the common partitioningof some ferric iron in biotite (Guidotti & Dyar,1991), redox equilibria in the present model systemlead, as long as biotite is present, to the depletion inoxygen of the fluid and to the attainment of the con-dition XO < 1/3.For inferring the P–T condition of melt inclusion

entrapment, the NCKFMASH chemical system wasused. All calculations were done by the Gibbs energyminimization (Connolly, 2009) with the thermo-dynamic database of Holland & Powell (1998), asrevised in 2003). The solution model of melt fromWhite et al. (2007) was used, garnet from Holland &Powell (2001), biotite from Taj�cmanov�a et al. (2009),white mica from Coggon & Holland (2002), plagio-clase from Newton et al. (1980) and K-feldspar fromThompson & Hovis (1979). An ideal model was usedto account for the solution of Mn in ilmenite andcordierite.The amount of H2O component involved in the

calculation for the bulk-rock composition wasassumed as the loss of ignition of XRF analysis andthus represents the water content available for equili-bration of the observed mineral assemblage. Theamount of carbon for the migmatite was estimatedfrom the modal proportion of graphite in the rock.For the modelling of generation of MI, the meanmajor element composition of preserved glassy MIwas used. Melt H2O content was measured byRaman spectroscopy analysis of re-homogenized MI.The resulting bulk-rock compositions (in mol.%)used for calculation are indicated in the upper leftinset of calculated P–T phase diagram sections(Figs 5 & 6).

Stromatic migmatite

In the phase diagram section for the migmatite(Fig. 5a), the phase assemblage Grt–Bt–Sil–Pl–Kfs–Qtz–Gr–COH–Liq (abbreviations after Kretz, 1983),observed in the sample, corresponds to a quadrivari-ant field in the middle of the diagram. Remarkably,the relevant compositional isopleths of XMg, grossu-lar and spessartine content in garnet cores cloudedwith MI (XMg = 0.14; XGrs = 0.04; XSps = 0.08)



(a) (b)

(c) (d)

Fig. 5. (a) P–T section for stromatic migmatite calculated in MnNCaKFMASHOC system. The observed mineral assemblagecorresponds to the Grt–Bt–Sil–Pl–Kfs–Qtz–Gr–COH–Liq stability field. See text for details. (b) Contours for grossular andspessartine components and XMg value of garnet. (c) Isopleths of modal proportions of garnet and melt. Zoomed in modalproportions of biotite, muscovite, melt and garnet between 650 and 700 °C at 4.7 kbar. (d) Inferred P–T conditions. See text fordetails.



appear consistently in this quadrivariant field at 660–680 °C and 4.5–5 kbar (Fig. 5b), indicating that themigmatite equilibrated at these conditions (Fig. 5d).The Liq-in curve that corresponds to the H2O-richfluid saturated solidus of the rock coincides with theMs-out curve at ~4.5–5 kbar (Fig. 5a). Therefore, themuscovite melting reaction has a discontinuousnature only in this narrow pressure range. At theseconditions and owing to the absence of Ti in thechemical system (see Discussion for details), thecontinuous fluid-absent melting reaction consumingbiotite takes place over an interval of ~20 °C (see thelocation of the Bt-out curve in Fig. 5a). The modalproportions of garnet and melt increase rapidly aftercrossing the melt-in line (Fig. 5c), indicating a conti-nuous melting reaction consuming biotite and Fe-bearing muscovite and producing peritectic garnetand melt at >5 kbar. At the pressure of interest, gar-net growth results from the continuous reactioninvolving biotite dehydration soon after muscovitebreakdown (see the modal proportions in the inset ofFig. 5c). Graphite is stable in the whole P–T range,but its modal abundance is low (~ 0.15 vol.%). The

composition of the COH fluid is related to that ofthe other phase containing O2 component, which isbiotite. Moreover, any H2O-bearing phase such asmelt or cordierite must have aH2O < 1 in order tocoexist with graphite.The displacement of the wet solidus in graphite-

bearing systems is around 10–20 °C towards highertemperatures and the ‘invariant’ melting point (theequivalent of the KNASH Ms–Qtz–Ab–Kfs–Sil–Liq–fluid ‘I3’ of Thompson & Algor, 1977), shifts ofabout 0.5 kbar towards higher pressures. The topologyin the suprasolidus part of the diagram does notdiffer significantly in graphite-bearing or graphite freesystems.

Melt inclusion in garnet

Modelling of the bulk composition for the melt inclu-sion as obtained by electron microprobe shows thepresence of melt starting at about 650 °C and4.5 kbar (Fig. 6). The lack of H2O-rich fluid inclu-sions coexisting with MI suggests that the meltswere not H2O-saturated at the time of trapping and,

Fig. 6. P–T section for melt inclusioncalculated in NCKFMASH system. Seetext for details.



therefore, that MI were trapped at pressure condi-tions above the H2O-out line. The one phase field forLiq starts at ~750 °C extending to higher tempera-tures in the 4–5 kbar range (Fig. 6). The water con-tent used in the calculation (7.6 wt%) was thehighest concentration measured by Raman spectros-copy in the completely re-homogenized nanogranite(Bartoli et al., 2013b), which is believed to give thebest estimate. However, if the water content washigher, the assemblages above the H2O-out linewould shift towards higher pressures. For 10 wt%H2O, the Liq-bearing fields are stable only at >6 kbar(not shown). Free H2O in the subsolidus part ofFig. 6 represents the H2O exsolved from the meltduring its crystallization. In the studied rock, thisresulted in the formation of the H2O-bearing micro-and nano-bubbles observed within nanogranites (cf.fig. 4 in Bartoli et al., 2013b). In addition, modellingfor melt inclusion predicts that the phase assemblageMs–Bt–Pl–Kfs–Qtz–H2O should be present withintotally crystallized MI at subsolidus conditions.

DISCUSSION

P–T estimates and formation of peritectic garnet

The use of the most complex chemical system is gen-erally recommended for phase equilibria modelling(e.g. White et al., 2001, 2007). The recent progress inthis field enabled to involve minor components suchas Mn, Ti, Fe3+, which increase the stability ofimportant minerals such as garnet and biotite (e.g.Tinkham et al., 2001; White et al., 2007; Taj�cmanov�aet al., 2009). The result of modelling in suchexpanded system is then apparently closer to experi-mental predictions as well as to natural observations.However, the number of available solid solutions inthe database involving these minor components isstill restricted, and the quality of the calculation maybe also hampered by the lack of good thermody-namic data for given, mostly hypothetical, mineralend-members. This may lead to the unrealistic prefer-ential stabilization of a phase hosting the minor com-ponent, and to the increase of its modal proportionsat the expense of other important phases in theassemblage. This aspect plays a major role especiallyin cases where a key phase such as garnet has alreadylow modal proportion in the rock. When an inappro-priate combination of models with minor componentsis used, such a modally minor phase can be artifi-cially suppressed in the calculation.

The model P–T section for the metatexite of Fig. 5has been calculated in a Ti-free system. The involve-ment of Ti in the chemical system results in phaserelationships which differ from the one previouslydiscussed in that the biotite-out line shifts about70 °C towards higher temperatures (Fig. 7), in agree-ment with the experimental observations (e.g. Stevenset al., 1997). However, it should be noted that these

discrepancies are maximized at high temperatures, sothat for the studied sample Ti has only a minor influ-ence (~20 °C; Fig. 7) on the final estimate of the P–Tconditions (680–700 °C). Ti is involved only in biotiteand ilmenite solid solution models. This makes thesephases more stable and leads to a significantly lowermodal proportion of garnet (<1 vol.%) at the P–Tconditions of interest. As a consequence, garnet com-position has also unrealistically high concentrationsof spessartine component. When Ti is excluded fromthe system, the garnet modal proportions and com-positions correspond to the actual values. Further-more, also the calculated muscovite composition fitsbetter in terms of Fe–Mg content. These observationsindicate that it is very important to compare thecalculation results not only with the measured com-positions but also with the modal proportions ofphases.Given the presence of MI in the cores of garnet

crystals, this mineral is interpreted as a peritecticphase produced during incongruent melting of themigmatite (Bartoli et al., 2013b). This is supportedalso by the phase equilibria modelling where garnetmode rapidly increases after crossing the melt-in line(Fig. 5c). These relationships account for the growthof garnet in the presence of melt and the possibilityfor it to entrap primary MI. An inferred equilibration

Fig. 7. Topology changes after involvement of TiO2

component into the system. Shift of the estimated P–Tconditions in the TiO2 bearing system is identified by thedashed ellipse.



temperature of ~700 °C is consistent with the temper-ature at which garnet-hosted nanogranites have beentotally re-homogenized (Bartoli et al., 2013b). How-ever, it is in contrast with results from melting experi-ments that predict peritectic garnet generallydeveloping above 800 °C by the fluid-absent meltingof biotite (see Clemens, 2006 and references therein)and not by muscovite breakdown (e.g. Pati~no Douce& Harris, 1998). Growth of peritectic garnet stablycoexisting with biotite at 700–750 °C has beenrecently demonstrated by Acosta-Vigil et al. (2010,2012) and is also supported by the discovery of gar-net-hosted nanogranites in the amphibolite-facies me-tapelitic migmatites (‘kinzigites’) of the Ivrea Zone(A. Turina, pers. comm.).

The direct experimental confirmation of the for-mation of peritectic garnet at <750 °C in metapelitic/psammitic bulk compostions has not been possibleso far owing to technical difficulties such as sluggishkinetics at low temperatures (Le Breton & Thomp-son, 1988; Spear & Kohn, 1996) and the lack ofequilibration of H2O-poor experimental runs at<800 °C (Brearley & Rubie, 1990). In addition,detecting and analyzing the melts produced at lowmelting degrees remains a challenging task (e.g. Vie-lzeuf & Montel, 1994; Stevens et al., 1997). Peritecticgarnet was experimentally produced at 750 °C,7 kbar only under excess H2O conditions whichlikely favoured its nucleation (Ward et al., 2008).Therefore, the width of the multivariant fieldBt+Sil+Pl+Qtz+Grt+Kfs+melt in metasedimentaryrocks is rather difficult to constrain by experiments,because ‘it is very difficult to determine where itbegins’ (Vielzeuf & Montel, 1994). In this respect,the results of thermodynamic modelling are useful infilling the gap. Several phase equilibria calculationshave already shown that peritectic garnet may bestable with melt, biotite and sometimes muscovite inmultivariant fields at ≥650 °C and ≥3–4 kbar (e.g.White et al., 2004; Johnson et al., 2008; Groppoet al., 2012; Imayama et al., 2012; Riel et al., 2013).Our calculations support that continuous meltingreactions consuming biotite may produce peritecticgarnet starting from as low as 660–670 °C, wellbelow Bt-out conditions, but also that some peritec-tic garnet may form from Fe-bearing muscovitemelting at pressures above the ‘invariant’ meltingpoint, explaining the formation of garnet in metapel-ites at conditions corresponding to muscovite meltingat medium-P (<10 kbar; e.g. Acosta-Vigil et al.,2010).

Phase equilibrium modelling for melt inclusions

The phase diagram of Fig. 6 has been calculated toevaluate the potential of using thermodynamic mod-elling of phase equilibria for MI compositions todetermine the P–T conditions of melt entrapmentinto the growing peritectic phase.

The calculation predicts that the one-phase field ofliquid is stable at >750 °C, in contrast with the fullexperimental re-melting of nanogranite at 700 °C,that indicates that the melt alone should be stablealready at 700 °C. A possibility is that this inconsi-stency is related to an overestimation of the Fe andMg content of the melt bulk composition used forthe calculation that would stabilize garnet excessively.In this respect, special care was taken in the EMPanalysis of the melt inclusion composition: targetswere selected by using careful petrographic andSEM-BSE image analysis in order to avoid contami-nation from the adjacent garnet host. Nonetheless,we cannot exclude the potential contribution of gar-net from the bottom of the inclusion, and this mighthave artificially increased the FeO content of the meltto the average analyzed value of 1.20 wt%. However,considering that similar or even higher FeO contentshave been measured in MI enclosed in Fe-free hostssuch as andalusite and plagioclase at the beginning ofanatexis (respectively 1.3–1.7 wt%, Cesare et al.,2003; and 1.15 � 0.33 wt%, Acosta-Vigil et al.,2007) it is clear that FeO contents ≤1.5 wt% shouldbe considered regular melt concentrations in anatecticpelitic-psammitic systems at 700–750 °C. Therefore,the possibility of contamination of melt by the adja-cent host garnet, which would alter the resultingphase relationships in the P–T diagram, seems unli-kely.When the analyzed compositions of MI are com-

pared with the calculated melt compositions obtainedfrom the thermodynamic modelling of the bulk rockat the P–T conditions of interest (just above the soli-dus; Fig. 5a), the calculated melt compositions havelower FeO, MgO and CaO, and higher SiO2

(Table 2). The compositional departure gets smallertowards higher pressures, but still the model nevermatches the measured compositions. This inconsi-stency was already pointed out by Grant (2009) whodocumented the remarkable underestimation (up toone-tenth) in MgO, FeO and CaO of calculated meltcompositions with respect to actual melt composi-tions produced by experiments at temperatures up to780 °C. Conversely, the comparison between calcu-lated and analyzed melts is generally good at>780 °C (see Grant, 2009). Low-T melt compositionspredicted from thermodynamic modelling do notmatch not only those of the corresponding analyzedmelts, but also those of low-T natural and experi-mental melts from the literature (Fig. 8), indicatingthat calculated compositions are unrealistic undersuch conditions. It is important to emphasize thatanatectic MI hosted in Fe-, Mg-free phases (i.e. pla-gioclase and andalusite) and formed at low-T(≤700 °C) show the same Fe+Mg contents as anatec-tic MI hosted in garnet (Fig. 8). This refutes concernsabout the reliability of studied MI to reflect the com-position of natural anatectic melts and reinforces theinference of the available melt model not performing



well at low-T. Underestimating FeO, MgO and CaOcontents of melt enlarges the stability field of garnetand plagioclase towards high temperatures and thusdoes not allow a correct assessment of the P–T con-ditions of melt entrapment. It follows that in orderto become a useful tool in petrology, thermodynamicmodelling of phase equilibria for the composition of(re-melted) melt inclusions needs refining of the meltmodel in order to reproduce the compositions of

natural low-T melts. Such refinement may takeadvantage of the analytical data set that is beingobtained from melt inclusions in peritectic minerals(e.g. Cesare et al., 2011).The P–T diagram for the melt inclusion indicates

that muscovite, biotite, plagioclase, K-feldspar andquartz should crystallize from the trapped melt dur-ing natural cooling (Fig. 6). This assemblage hasbeen observed in a few inclusions (Fig. 4b). However,most of nanogranite inclusions in the metatexite con-sist of aggregates of Ms–Bt–Pl–Qtz and the lack ofK-feldspar is not predicted in the subsolidus assem-blages (Fig. 6). This inconsistency may be partiallyrelated to pitfalls regarding melt model (see above)and to the two-dimensional analysis of these smallinclusions that may indicate incomplete phase assem-blages (Cesare et al., 2011). However, it is importantto note that the K2O contents measured in remeltednanogranites (4–5 wt%; Bartoli et al., 2013a) areunlikely to be provided entirely by biotite and musco-vite, suggesting the occurrence of another K2O-bear-ing phase (e.g. K-feldspar and/or K-bearing glass)not detected during microstructural study. Glasspresent in partially crystallized MI coexisting withnanogranites may show K2O-rich compositions(Table S5). This would support the hypothesis ofsome (undetectable ?) amounts of K2O-rich glassoccurring with quartz, biotite, muscovite and plagio-clase within nanogranite inclusions.

CONCLUSIONS

From this study, we can conclude that:

(i) Small amounts of peritectic garnet may be pro-duced in metasedimentary rocks at low tempera-tures (≤700 °C) by continuous melting reactionsinvolving biotite and muscovite. In the specificrock under investigation, garnet was produced bythe continuous melting reaction consuming biotiteat 660–700 °C and 4.5–5 kbar, soon after musco-vite breakdown.

(ii) Peritectic garnet can trap droplets of melt pro-duced immediately after crossing the melt-in line.

(iii) At lower temperature conditions, the melt com-position calculated from phase equilibria model-ling does not match the measured meltcompositions. Therefore, the current melt modelneeds to be improved, as recently suggested byWhite et al. (2011).

(iv) P–T diagrams constructed using MI composi-tions may become an additional useful tool forthe thermobarometry of low-T anatectic rocks,provided that a better thermodynamic model forsilicate melt is developed.

Furthermore, in the framework of melting equili-bria calculations, the reintegration of calculated meltcompositions into the analyzed bulk composition is aroutine approach to investigate the prograde P–T

Table 2. Comparison between melt composition measured byelectron microprobe and those calculated just above thesolidus. Measured composition recast to 100%.

Melt

Analyzed Calculated

T (°C) 670 670

P (kbar) 4 5

SiO2 70.85 76.54 74.17

TiO2 0.08 0.00 0.00

Al2O3 11.98 9.88 10.62

FeO 1.22 0.17 0.12

MnO 0.09 0.00 0.00

MgO 0.07 0.03 0.02

CaO 0.40 0.20 0.21

Na2O 3.14 2.04 2.37

K2O 4.26 4.82 4.87

P2O5 0.18 0.00 0.00

H2O 7.73 6.33 7.62

Total 100 100 100

Fig. 8. CaO v. FeO + MgO (wt%) diagram comparingcompositions of glassy melt inclusions, experimental low-Tmelts and calculated melts. Trace element thermometryindicates that melt inclusion in andalusite and plagioclase ofpartially melted metapelitic enclaves represent melts formed at~615–720 °C (Cesare et al., 2003; Acosta-Vigil et al., 2010).Low-T experimental melts were produced at 650–700 °C and2–3 kbar using metapelites and quartzofeldspathic rocks asstarting materials (Holtz & Johannes, 1991; Icenhower &London, 1995; Acosta-Vigil et al., 2006). Experimental meltsfrom Grant (2009) were produced at 720–740 °C and 2 kbar.Lines connect calculated compositions with the correspondingmeasured compositions (for melt inclusions from this study weconsider the mean composition of preserved glassy inclusions).



evolution of anatectic rocks that have undergonemelt loss (see White et al., 2004; Indares et al., 2008;Groppo et al., 2012). In view of the present research,we suggest that a more precise protolith composition(and therefore more constrained P–T estimations)may be obtained by reintegrating composition of MIhosted in peritectic phases.

ACKNOWLEDGEMENTS

Financial support for this project came from the Ital-ian Ministry of Education, University, Research(grant PRIN 2010TT22SC) and from the Universityof Padua (Progetto di Ateneo CPDA107188/10) toBC, a research contract from the University of Paduato OB, a support of the European Commission underthe Marie Curie Intra-European Fellowship Programto LT and a Ram�on y Cajal research contract andgrants CGL2007-62992, CTM2005-08071-C03-01,CSD2006-0041 to AAV. The authors thank A. Ri-splendente and R. Carampin for assistance duringEMP analyses. J.A.D. Connolly is also acknowledgedfor his helpful discussion. The authors are deeplygrateful to S. Poli for his stimulating discussions onpitfalls of the phase equilibria modelling, to E. W.Sawyer and an anonymous reviewer for their con-structive comments which improved the clarity andthe quality of the manuscript, and to M. Brown forthe careful editorial handling.

REFERENCES

Acosta, A., 1998. Estudio de los fen�omenos de fusi�on corticaly generaci�on de granitoides asociados a las peridotitas deRonda. Unpublished PhD Thesis, Universidad de Granada,Granada, 305 pp. (in Spanish).

Acosta-Vigil, A., Pereira, M.D., Shaw, D.M. & London, D.,2001. Contrasting behaviour of boron during crustal anatex-is. Lithos, 56, 15–31.

Acosta-Vigil, A., London, D.f. & Morgan, G.B., 2006.Experiments on the kinetics of partial melting of a leucog-ranite at 200 MPa H2O and 690-800°C: compositional vari-ability of melts during the onset of H2O-saturated crustalanatexis. Contributions to Mineralogy and Petrology, 151,539–557.

Acosta-Vigil, A., Cesare, B., London, D. & Morgan, G.B.,2007. Microstructures and composition of melt inclusions ina crustal anatectic environment, represented by metapeliticenclaves within El Hoyazo dacites, SE Spain. Chemical Geol-ogy, 235, 450–465.

Acosta-Vigil, A., Buick, I., Hermann, J., et al., 2010. Mecha-nisms of crustal anatexis: a geochemical study of partiallymelted metapelitic enclaves and host dacite, SE Spain.Journal of Petrology, 51, 785–821.

Acosta-Vigil, A., Buick, I., Cesare, B., London, D. & Morgan,G.B., 2012. The extent of equilibration between melt andresiduum during regional anatexis and its implications fordifferentiation of the continental crust: a study of partiallymelted metapelitic enclaves. Journal of Petrology, 53, 1319–1356.

Aldaya, F., Guarc�ıa-Due~nas, V. & Navarro-Vil�a, F., 1979. LosMantos Alpuj�arrides del tercio central de las CordillerasB�eticas. Ensayo de correlaci�on tect�onica de los Alpuj�arrides.Geologica Acta, 14, 154–166. (in Spanish).

Bartoli, O., Cesare, B., Poli, S., Acosta-Vigil, A., Esposito, R.,Turina, A., Bodnar, R.J., Angel, R.J. & Hunter, J., 2013a.Nanogranite inclusions in migmatitic garnet: behavior duringpiston cylinder re-melting experiments. Geofluids. doi:10.1111/gfl.12038.

Bartoli, O., Cesare, B., Poli, S., Bodnar, R.J., Acosta-Vigil, A.,Frezzotti, M.L. & Meli, S., 2013b. Recovering the composi-tion of melt and the fluid regime at the onset of crustal anat-exis and S-type granite formation. Geology, 41, 115–118.

Brearley, A.J. & Rubie, D.C., 1990. Effects of H2O on the dis-equilibrium breakdown of muscovite + quartz. Journal ofPetrology, 31, 925–956.

Brown, M., 2007. Crustal melting and melt extraction, ascentand emplacement in orogens: mechanisms and consequences.Journal of the Geological Society, 164, 709–730.

Brown, M., 2010. Melting of the continental crust during oro-genesis: the thermal, rheological, and compositional conse-quences of melt transport from lower to upper continentalcrust. Canadian Journal of Earth Sciences, 47, 655–694.

Cesare, B., Marchesi, C., Hermann, J. & Gomez-Pugnaire,M.T., 2003. Primary melt inclusions in andalusite from ana-tectic graphitic metapelites: implications for the position ofthe Al2SiO5 triple point. Geology, 31, 573–576.

Cesare, B., Maineri, C., Baron Toaldo, A., Pedron, D. & Aco-sta-Vigil, A., 2007. Immiscibility between carbonic fluids andgranitic melts during crustal anatexis: a fluid and melt inclu-sion study in the enclaves of the Neogene Volcanic Provinceof SE Spain. Chemical Geology, 237, 433–449.

Cesare, B., Ferrero, S., Salvioli-Mariani, E., Pedron, D. &Cavallo, A., 2009. Nanogranite and glassy inclusions: theanatectic melt in migmatites and granulites. Geology, 37,627–630.

Cesare, B., Acosta-Vigil, A., Ferrero, S. & Bartoli, O., 2011.Melt inclusions in migmatites and granulites. In: The Scienceof Microstructure – Part II, Journal of the Virtual Explorer(eds Forster, M.A. & Fitz Gerald, J.D.), The VirtualExplorer Pty Ltd, Canberra, Electronic Edition, ISSN 1441-8142. 38, paper 2.

Clemens, J.D., 2006. Melting of the continental crust: fluidregimes, melting reactions and source-rock fertility. In:Evolution and Differentiation of the Continental Crust (edsBrown, M. & Rushmer, T.), pp. 297–331. Cambridge Uni-versity Press, Cambridge.

Clemens, J.D. & Stevens, G., 2012. What controls chemicalvariation in granitic magmas? Lithos, 134–135, 317–329.

Clemens, J.D. & Vielzeuf, D., 1987. Constraints on meltingand magma production in the crust Earth and PlanetaryScience Letters, 86, 287–306.

Coggon, R. & Holland, T.J.B., 2002. Mixing properties ofphengitic micas and revised garnet–phengite thermobarome-ters. Journal of Metamorphic Geology, 20, 683–696.

Connolly, J.A.D., 1995. Phase diagrams for graphitic rocks.Contributions to Mineralogy and Petrology, 119, 94–116.

Connolly, J.A.D., 2009. The geodynamic equation of state:what and how. Geochemistry, Geophysics, Geosystems, 10,Q10014.

Connolly, J.A.D. & Cesare, B., 1993. C-O-H-S fluid composi-tion and oxygen fugacity in graphitic metapelites. Journal ofMetamorphic Geology, 11, 379–388.

Danyushevsky, L.V., McNeill, A.W. & Sobolev, A.V., 2002.Experimental and petrological studies of melt inclusions inphenocrysts from mantle-derived magmas: an overview oftechniques, advantages and complications. Chemical Geol-ogy, 183, 5–24.

Esteban, J.J., Cuevas, J., Vegas, N. & Tub�ıa, J.M., 2008.Deformation and kinematics in a melt-bearing shear zonefrom the Western Betic Cordilleras (Southern Spain). Journalof Structural Geology, 30, 380–393.

Esteban, J.J., Cuevas, J., Tub�ıa, J.M., Sergeev, S. & Larionov,A., 2011a. A revised Aquitanian age for the emplacement ofthe Ronda peridotites (Betic Cordilleras, southern Spain).Geological Magazine, 148, 183–187.



Esteban, J.J., Tub�ıa, J.M., Cuevas, J., Vegas, N., Sergeev, S. &Larionov, A., 2011b. Peri-Gondwanan provenance of pre-Triassic metamorphic sequences in the western Alpujarridenappes (Betic Cordillera, southern Spain). GondwanaResearch, 20, 443–449.

Ferrero, S., Bodnar, R.J., Cesare, B. & Viti, C., 2011. Reequili-bration of primary fluid inclusions in peritectic garnetfrom metapelitic enclaves, El Hoyazo, Spain. Lithos, 124,117–131.

Ferrero, S., Bartoli, O., Cesare, B., Salvioli-Mariani, E., Aco-sta-Vigil, A., Cavallo, A., Groppo, C. & Battiston, S., 2012.Microstructures of melt inclusions in anatectic metasedimen-tary rocks. Journal of Metamorphic Geology, 30, 303–322.

Frezzotti, M.L., 2001. Silicate melt inclusions in magmaticrocks: applications to petrology. Lithos, 55, 273–299.

Grant, J.A., 1985. Phase equilibria in partial melting of politicrocks. In: Migmatites (ed Ashworth, J.R.), pp. 86–144. Blac-kie and Son, Glasgow.

Grant, J.A., 2009. THERMOCALC and experimental modeling ofmelting of pelite, Morton Pass, Wyoming. Journal of Meta-morphic Geology, 27, 571–578.

Groppo, C., Rolfo, F. & Indares, A., 2012. Partial melting inthe Higher Himalayan Crystallines of Eastern Nepal: theeffect of decompression and implications for the “channelflow” model. Journal of Petrology, 53, 1057–1088.

Guidotti, C.V. & Dyar, M.D., 1991. Ferric iron in metamor-phic biotite and its petrologic and crystallochemical implica-tions. American Mineralogist, 76, 161–175.

Hasalov�a, P., Janou�sek, V., Schulmann, K., �St�ıpsk�a, P. & Er-ban, V., 2008. From orthogneiss to migmatite: geochemicalassessment of the melt infiltration model in the Gf€ohl Unit(Moldanubian Zone, Bohemian Massif). Lithos, 102, 508–537.

Holland, T.J.B. & Powell, R., 1998. An internally consistentthermodynamic data set for phases of petrological interest.Journal of Metamorphic Geology, 16, 309–343.

Holland, T.J.B. & Powell, R., 2001. Calculation of phaserelations involving haplogranitic melts using an internallyconsistent thermodynamic data set. Journal of Petrology, 42,673–683.

Holland, T.J.B. & Powell, R., 2003. Activity-composition rela-tions for phases in petrological calculations: an asymmetricmulticomponent formulation. Contributions to Mineralogyand Petrology, 145, 492–501.

Holness, M.B., Cesare, B. & Sawyer, E.W., 2011. Melted rocksunder the microscope: microstructures and their interpreta-tion. Elements, 7, 245–250.

Holtz, F. & Johannes, W., 1991. Genesis of peraluminousgranites I. Experimental investigation of melt compositionsat 3 and 5 kb and various H2O activities. Journal of Petrol-ogy, 32, 935–958.

Johnson, T.E., White, R.W. & Powell, R., 2008. Partial melt-ing of metagreywacke: a calculated mineral equilbria study.Journal of Metamorphic Geology, 26, 837–853.

Kretz, R., 1983. Symbols for rock-forming minerals. AmericanMineralogist, 68, 277–279.

Icenhower, J. & London, D., 1995. An experimental study ofelement partitioning among biotite, muscovite and coexistingperaluminous silicic melt at 200 MPa (H2O). American Min-eralogist, 80, 1229–1251.

Imayama, T., Takeshita, T., Yi, K. et al., 2012. Two-stagepartial melting and contrasting cooling history within theHigher Himalayan Crystalline Sequence in the far-easternNepal Himalaya. Lithos, 134–135, 1–22.

Indares, A., White, R.W. & Powell, R., 2008. Phase equilibriamodelling of kyanite-bearing anatectic paragneiss from thecentral Grenville Province. Journal of Metamorphic Geology,26, 815–836.

Le Breton, N. & Thompson, A.B., 1988. Fluid-absent (dehy-dration) melting of biotite in metapelites in the early stagesof crustal anatexis. Contributions to Mineralogy and Petrol-ogy, 99, 226–237.

Lundeen, M.T., 1978. Emplacement of the Ronda peridotite,Sierra Bermeja, Spain. Geological Society of America Bulle-tin, 89, 172–180.

Morgan, G.B. & London, D., 1996. Optimizing the electronmicroprobe analysis of hydrous alkali aluminosilicate glasses.American Mineralogist, 81, 1176–1185.

Morgan, G.B. & London, D., 2005. Effect of current densityon the electron microprobe analysis of alkali aluminosilicateglasses. American Mineralogist, 90, 1131–1138.

Newton, R.C., Charlu, T.V. & Kleppa, O.J., 1980. Thermo-chemistry of high structural state plagioclases. Geochimica etCosmochimica Acta, 44, 933–941.

Obata, M., 1980. The Ronda peridotite: garnet-, spinel-, andplagioclase-lherzolite facies and the P-T trajectories of a hightemperature mantle intrusion. Journal of Petrology, 21, 533–572.

Pati~no Douce, A.E. & Harris, N., 1998. Experimental con-straints on Himalayan anatexis. Journal of Petrology, 39,689–710.

Platt, J.P. & Whitehouse, M.J., 1999. Early Miocene high-temperature metamorphism and rapid exhumation in theBetic Cordillera (Spain): evidence from U-Pb zircon ages.Earth and Planetary Science Letters, 171, 591–605.

Platt, J.P., Behr, W.M., Johanesen, K. & Williams, J.R., 2013.The Betic-Rif Arc and its orogenic hinterland: a review.Annual Reviews of Earth and Planetary Sciences, 41, doi:10.1146/annurev-earth-050212-123951.

Riel, N., Guillot, S., Jaillard, E. et al., 2013. Metamorphicand geochronological study of the Triassic El Oro metamor-phic complex, Ecuador: implications for high-temperaturemetamorphism in a forearc zone. Lithos, 156–159, 41–68.

Roedder, E., 1984. Fluid inclusions. Mineralogical Society ofAmerica, Reviews in Mineralogy, 12, 644.

S�anchez-Rodr�ıguez, L., 1998. Pre-Alpine and Alpine evolutionof the Ronda Ultramafic Complex and its country-rocks(Betic chain, southern Spain): U-Pb SHRIMP zircon and fis-sion-track dating. PhD Thesis, ETH, Z€urich, Switzerland,170 pp.

Sawyer, E.W., 2001. Melt segregation in the continental crust:distribution and movement of melt in anatectic rocks.Journal of Metamorphic Geology, 19, 291–309.

Solar, G.S. & Brown, M., 2001. Petrogenesis of migmatites inMaine, USA: possible source of peraluminous leucogranitein plutons? Journal of Petrology, 42, 789–823.

Spear, F.S., 1993. Metamorphic Phase Equilibria and Pressure–Temperature–Time Paths. Mineralogical Society of America,Washington, DC, pp. 353–354.

Spear, F. & Kohn, M.J., 1996. Trace element zoning ingarnet as a monitor of crustal melting. Geology, 24, 1099–1102.

Spear, F., Kohn, M.J. & Cheney, J.T., 1999. P-T paths fromanatectic pelites. Contributions to Mineralogy and Petrology,134, 17–32.

Stevens, G., Clemens, J.D. & Droop, T.R., 1997. Melt produc-tion during granulite-facies anatexis: experimental datafrom’’primitive’’ metasedimentary protoliths. Contributionsto Mineralogy and Petrology, 128, 352–370.

Taj�cmanov�a, L., Conolly, J.A.D. & Cesare, B., 2009. A ther-modynamic model for titanium and ferric iron solution inbiotite. Journal of Metamorphic Geology, 27, 153–165.

Thomas, R., 2000. Determination of water contents of granitemelt inclusions by confocal laser Raman microprobe spec-troscopy. American Mineralogist, 85, 868–872.

Thompson, A.B., 1982. Dehydration melting of pelitic rocksand the generation of H2O-undersaturated granitic liquids.American Journal of Science, 282, 1567–1595.

Thompson, A.B. & Algor, J.R., 1977. Model systems foranatexis of pelitic rocks. I. Theory of melting reactions inthe system KAlO2–NaAlO2–Al2O3–SiO2–H2O. Contributionsto Mineralogy and Petrology, 63, 247–269.

Thompson, A.B. & Connolly, J.A.D., 1995. Melting of thecontinental crust: some thermal and petrological constraints



on anatexis in continental collision zones and other tectonicsettings. Journal of Geophysical Research, 100, 15565–15579.

Thompson, J.B. & Hovis, G.L., 1979. Entropy of mixing insanidine. American Mineralogist, 64, 57–65.

Tinkham, D.K., Zuluaga, C.A. & Stowell, H.H., 2001. Metap-elite phase equilibria modeling in MnNCKFMASH: theeffect of variable Al2O3 and MgO⁄ (MgO + FeO) on mineralstability. Geological Materials Research, 3, 1–42.

Torres-Rold�an, R.L., 1983. Fractionated melting of metapeliteand further crystal-melt equillibria. The example of theBlanca Unit migmatite complex, north of Estepona (south-ern Spain). Tectonophysics, 96, 95–123.

Tub�ıa, J.M., 1988. Estructura de los Alpuj�arrides occidentales:Cinem�atica y condiciones de emplazamiento de las peridotitasde Ronda. Publicaciones Especiales del Bolet�ın Geol�ogico yMinero de Espa~na, Madrid, 124 pp. (in Spanish).

Tub�ıa, J.M. & Cuevas, J., 1986. High-temperature emplace-ment of the Los Reales peridotite nappe, Betic Cordillera,Spain. Journal of Structural Geology, 8, 473–482.

Tub�ıa, J.M. & Gil-Ibarguchi, J.I., 1991. Eclogites of the Ojennappe: a record of subduction in the Alpujarride complex(Betic Cordilleras, southern Spain). Journal of the GeologicalSociety, 148, 801–804.

Tub�ıa, J.M., Cuevas, J. & Gil-Ibarguchi, J.I., 1997. Sequentialdevelopment of the metamorphic aureole beneath the Rondaperidotites and its bearing on the tectonic evolution of theBetic Cordillera. Tectonophysics, 279, 227–252.

Vernon, R.H., 2011. Microstructures of melt-bearing regionalmetamorphic rocks. In: Origin and Evolution of PrecambrianHigh-Grade Gneiss Terranes, with Special Emphasis on theLimpopo Complex of Southern Africa (eds Van Reenen,D.D., Kramers, J.D., McCourt, S. & Perchuk, L.L.), Geolo-gical Society of America Memoir, 207, 1–11.

Vielzeuf, D. & Montel, J.M., 1994. Partial melting of meta-greywackes. Part I. Fluid absent experiments and phase rela-tionships. Contributions to Mineralogy and Petrology, 117,375–393.

Vielzeuf, D. & Schmidt, M.W., 2001. Melting relations inhydrous systems revisited: application to metapelites, meta-greywackes and metabasalts. Contributions to Mineralogyand Petrology, 141, 251–267.

Ward, R., Stevens, G. & Kisters, A., 2008. Fluid and defor-mation induced partial melting and melt volumes in low-

temperature granulite-facies metasediments, Damara Belt,Namibia. Lithos, 105, 253–271.

Westerhof, A.B., 1977. On the contact relations of high-tem-perature peridotites in the Serrania de Ronda, southernSpain. Tectonophysics, 39, 579–591.

White, R.W., Powell, R. & Holland, T.J.B., 2001. Calculationof partial melting equilibria in the system Na2O–CaO–K2O–FeO–MgO–Al2O3–SiO2–H2O (NCKFMASH). Journal ofMetamorphic Geology, 19, 139–154.

White, R.W., Powell, R. & Halpin, A., 2004. Spatially-focussedmelt formation in aluminous metapelites from BrokenHill, Australia. Journal of Metamorphic Geology, 22, 825–845.

White, R.W., Powell, R. & Holland, T.J.B., 2007. Progressrelating to calculation of partial melting equilibria formetapelites. Journal of Metamorphic Geology, 25, 511–527.

White, R.W., Stevens, G. & Johnson, T.E., 2011. Is thecrucible reproducible? Reconciling melting experiments withthermodynamic calculations. Elements, 7, 241–246.

SUPPORTING INFORMATION

Additional Supporting Information may be found inthe online version of this article at the publisher’sweb site:Table S1. Representative electron microprobe

analyses of biotite and muscovite.Table S2. Representative electron microprobe anal-

yses of garnet and ilmenite.Table S3. Representative electron microprobe

analyses of plagioclase.Table S4. Representative electron microprobe

analyses of K-feldspar.Table S5. Representative electron microprobe

analyses of glass in preserved glassy melt inclusions(MI) and partially crystallized MI.

Received 5 February 2013; revision accepted 29 May 2013.



Phase equilibria constraints on melting of stromatic migmatites from Ronda (S Spain): insights on the formation of peritectic garnet.

Documents