Origin of Felsic Migmatites by Ductile Shearing and Melt Infiltration

Origin of migmatites by deformation-enhanced melt infiltrationof orthogneiss: a new model based on quantitativemicrostructural analysis

P. HASALOVA,1 , 2 K. SCHULMANN,1 O. LEXA,1 , 2 P . ST IPSKA ,1 F . HROUDA,2 , 3 S . ULRICH,2 , 4

J . HALODA5 AND P. TYCOVA5

1Universite Louis Pasteur, CGS/EOST, UMR 7517, 1 rue Blessig, Strasbourg 67084, France ([email protected])2Institute of Petrology and Structural Geology, Charles University, Albertov 6, 12843 Prague, Czech Republic3AGICO, Jecna 29a, 621 00 Brno, Czech Republic4Institute of Geophysics, Czech Academy of Sciences, Bocnı II/1401, 14131 Praha 4, Czech Republic5Czech Geological Survey, Klarov 3, 118 21 Prague 1, Czech Republic

ABSTRACT A detailed field study reveals a gradual transition from high-grade solid-state banded orthogneiss viastromatic migmatite and schlieren migmatite to irregular, foliation-parallel bodies of nebulitic migmatitewithin the eastern part of the Gfohl Unit (Moldanubian domain, Bohemian Massif). The orthogneiss tonebulitic migmatite sequence is characterized by progressive destruction of well-equilibrated bandedmicrostructure by crystallization of new interstitial phases (Kfs, Pl and Qtz) along feldspar boundariesand by resorption of relict feldspar and biotite. The grain size of all felsic phases decreases continuously,whereas the population density of new phases increases. The new phases preferentially nucleate alonghigh-energy like–like boundaries causing the development of a regular distribution of individual phases.This evolutionary trend is accompanied by a decrease in grain shape preferred orientation of all felsicphases. To explain these data, a new petrogenetic model is proposed for the origin of felsic migmatites bymelt infiltration from an external source into banded orthogneiss during deformation. In this model,infiltrating melt passes pervasively along grain boundaries through the whole-rock volume and changescompletely its macro- and microscopic appearance. It is suggested that the individual migmatite typesrepresent different degrees of equilibration between the host rock and migrating melt duringexhumation. The melt topology mimicked by feldspar in banded orthogneiss forms elongate pocketsoriented at a high angle to the compositional banding, indicating that the melt distribution wascontrolled by the deformation of the solid framework. The microstructure exhibits features compatiblewith a combination of dislocation creep and grain boundary sliding deformation mechanisms. Themigmatite microstructures developed by granular flow accompanied by melt-enhanced diffusion and/ormelt flow. However, an AMS study and quartz microfabrics suggest that the amount of melt present didnot exceed a critical threshold during the deformation to allow free movements of grains.

Key words: crystal size distribution; melt infiltration; melt topology; migmatites; quantitative texturalanalysis.

INTRODUCTION

Movement of a large volume of granitic melt is animportant factor in the compositional differentiationof the continental crust (Fyfe, 1973; Collins & Sawyer,1996; Brown & Rushmer, 2006) and the presence ofmelt in rocks profoundly influences their rheology(Arzi, 1978). The migration of melt through the crust iscontrolled by melt buoyancy and pressure gradientsresulting from the combination of gravity forces anddeformation (Wickham, 1987; Sawyer, 1994). Thereare three major mechanisms controlling melt migrationthrough the continental crust: (i) diapirism resulting inupward motion of low-density magma through higherdensity rocks (Chandrasekhar, 1961; Ramberg, 1981);(ii) dyking that describes melt migration by hydro-

fracturing of the host rock and transport of meltthrough narrow dykes (Lister & Kerr, 1991; Petford,1995); (iii) and migration of a melt through a networkof interconnected pores during deformation or com-paction of solid matrix (McKenzie, 1984; Wickham,1987).

Brown & Solar (1998a) and Weinberg & Searle(1998) proposed that during active deformation meltmoves by pervasive flow and it is essentially pumpedthrough the system parallel to the principal finiteelongation in the form of foliation-parallel veins.Based on a number of field studies, pervasive meltmigration at outcrop scale controlled by regionaldeformation has been suggested by various authors(Collins & Sawyer, 1996; Brown & Solar, 1998b;Vanderhaeghe, 1999; Marchildon & Brown, 2003).

J. metamorphic Geol., 2008, 26, 29–53 doi:10.1111/j.1525-1314.2007.00743.x

� 2007 Blackwell Publishing Ltd 2 9

These authors argued that magma intrudes perva-sively, parallel to the main anisotropy represented byfoliation planes (John & Stunitz, 1997), fold hinges andinterboudin partitions (Brown, 1994; Brown et al.,1995). It is also commonly observed that vein-likeleucosomes are injected into extensional structuresprovided the magma pressure is high enough (Wick-ham, 1987; Lucas & St.Onge, 1995) or parallel to axialsurfaces of folds (Vernon & Paterson, 2001).

Microscopic studies of natural rocks show orienta-tions of former melt microstructures that are inter-preted in terms of grain-scale channel networks(Sawyer, 2001). Melt migration pathways at the grainscale are commonly determined from distribution ofmelt films and pools (now glass) in experimentallyprepared samples or by distribution of minerals sup-posed to preserve the original melt topology in naturalrocks (Brown et al., 1999; Rosenberg & Riller, 2000;Rosenberg, 2001). The melt topology in experiments iscontrolled mainly by differential stress, confiningpressure and the amount of melt in the system(Rosenberg, 2001). At static conditions, the melttopology is characterized by equilibrium dihedral(wetting) angles at triple point junctions (Jurewicz &Watson, 1984; Laporte & Watson, 1995; Laporteet al., 1997; Cmıral et al., 1998; Walte et al., 2003) andthe mobility of the melt remains very low, even if themelt phase forms an interconnected network alongtriple-junction grain edges at dihedral angles lowerthan 60� (Laporte & Watson, 1995; Connolly et al.,1997).

Experimental studies on rock analogues to investi-gate grain-scale melt flow under laboratory conditionsshow that during contemporaneous melting anddeformation melt connection allows the nucleation ofshear bands along which a melt is further segregated(Rosenberg & Handy, 2000, 2001; Barraud et al.,2001). Rosenberg (2001) reviewed the experimentaldata and concluded that the melt migration and meltflow direction are controlled by incremental shorteningand melt pressure gradients between source and areasof melt accumulation.

There have only been a few attempts to quantifymelt distribution in rocks using methods of quantita-tive and computer aided microstructural analysis(Dallain et al., 1999; Tanner, 1999; Marchildon &Brown, 2003). These studies commonly deal with graincontact frequency distributions, grain size evolutionand orientation of former melt films (Dougan, 1983;McLellan, 1983; Rosenberg & Riller, 2000). However,modern quantitative microstructural analysis mayprovide further important information about: (i)reorganization of the rock structure associated withmelt migration in terms of grain contact distributions

(Lexa et al., 2005); (ii) characterization of dynamic orstatic conditions of melt movement through rocksusing analysis of grain boundaries and shape orienta-tions (Rosenberg & Riller, 2000; Marchildon & Brown,2002); and (iii) cooling or heating histories of rocksusing crystal size distribution (CSD) theory (Higgins,1998; Berger & Roselle, 2001).In this work, a sequence of deformed felsic rocks is

studied, ranging from high-grade banded orthogneissto fine-grained isotropic migmatite both at macro- andmicroscale using structural, petrographic and quanti-tative microstructural analyses. It is shown that asequence of banded orthogneiss, stromatic, schlierenand nebulitic migmatites results from progressivedeformation in a crustal-scale shear zone in the pres-ence of melt. The microstructural and fabric modifi-cations connected with disintegration of parentalbanded orthogneiss and development of random min-eral microstructure are quantified. The relationships ofthe individual rocks types and the possible origin ofthis sequence are discussed in terms of deformationand migmatization of different protoliths, meltinfiltration from an external source or in situ melting ofthe same protolith during progressive deformation. Itis argued that banded orthogneiss and nebuliticmigmatites can be interpreted as end-members of acontinuous sequence resulting from melt infiltrationfrom an external source during deformation. Finally,the role of melt for activity of grain-scale deformationmechanisms and bulk rheological behaviour of crustalrocks during melt infiltration is discussed.

GEOLOGICAL SETTING

The Moldanubian zone represents the highest gradeunit of the Bohemian Massif and is interpreted as aninternal zone of the Variscan orogen developed duringthe Variscan convergence (Matte et al., 1990). TheMoldanubian zone is comprised essentially of high-grade gneisses and migmatites containing relicts ofhigh-pressure felsic granulites, eclogites and peridotitesthat are intercalated with mid-crustal rocks (Fig. 1a).Schulmann et al. (2005) described the structural andmetamorphic evolution of high-grade crustal rocks ofthe so-called Gfohl Unit and of the adjacent middlecrustal units. For the mechanism of exhumation, theyproposed a model of vertical extrusion of orogeniclower crust and its lateral spreading in mid-crustallevels due to an indentation of the easterly Bruniapromontory. As a consequence of this process, thehigh-pressure rocks were thrust over adjacent middlecrustal units in conjunction with retrogression of ori-ginal mineral assemblages and partial melting of all therock types (Stıpska et al., 2004).

Fig. 1. (a) Geological map of the eastern margin of the Bohemian Massif (modified after Schulmann et al., 2005) with the location ofthe study area (black rectangle). The upper right inset shows the general location of the Bohemian Massif within the EuropeanVariscides. (b) Schematic block diagram displaying the main structural features in the study area (modified after Schulmann et al.,2005). Dominant S1 and S2 fabrics with their orientations are shown. This block diagram is not vertically scaled.

3 0 P . H A S A L O V A E T A L .

� 2007 Blackwell Publishing Ltd

O R I G I N O F F E L S I C MI GM A T I T E S 3 1


The onset of the exhumation process is dated byzircon U–Pb ages of c. 340 Ma on felsic granulites,migmatites and mantle-derived syn-tectonicallyemplaced plutons (van Breemen et al., 1982; Kroneret al., 1988; Holub et al., 1997; Schulmann et al.,2005). Tajcmanova et al. (2006) assigned metamorphicconditions of 840 �C at 18–19 kbar and 760–790 �C at10–13 kbar to relict steep granulitic fabrics whichoriginated by vertical extrusion of lower crust. Theseauthors also estimated the conditions of re-equilibra-tion of granulites associated with horizontal spreadingstage to 720–770 �C and 4–4.5 kbar. High pressurerocks of the Gfohl Unit are accompanied by largebodies of biotite–sillimanite Gfohl orthogneiss spa-tially associated with K-feldspar–sillimanite parag-neisses and leucocratic migmatites for which P–Tconditions of 7–10 kbar and 750 �C were estimated byRacek et al. (2006).

The area of this study is located at the easternmosttermination of the Gfohl Unit (Fig. 1a). The main rocktype is represented by the Gfohl orthogneiss withprotolith ages 488 ± 6 Ma (U–Pb SHRIMP: Friedlet al., 2004) including small bodies of amphibolite,granulite, eclogite, ultrabasic rock and paragneiss. TheGfohl orthogneiss shows different stages of migmati-

zation characterized by the assemblage of Kfs +Pl + Qtz + Bt ± Grt ± Sill. This migmatizedorthogneiss complex is heterogeneously deformed bytop to the NE shearing along a large-scale, gentlydipping shear zone (Schulmann et al., 1994). Conse-quently, the northern margin of this complex is thrustover a footwall comprised of the Namest� granulitebody and Neoproterozoic metagranites of the north-eastern continental margin (Urban, 1992).

STRUCTURAL EVOLUTION

Mesoscopic structures

Two major deformation events are recognized. Thefirst event (D1) is represented by a steep, west-dippinghigh-grade foliation S1 (Fig. 1b). This fabric is pre-served in banded orthogneisses (type I), as an alter-nation of recrystallized monomineralic K-feldspar,plagioclase and quartz layers, separated by bands ofbiotite ± sillimanite (Fig. 2a). Lineation L1 is locallymarked by alignment of biotite, sillimanite and byelongation of quartz and feldspar aggregates. Thisdeformation is attributed to an early stage of exhu-mation of the lower crust along a vertical channel

1.0 m

a

b

c

d

(b)

1cm

Type II

(c)

1 cm

Type III

2 cm

(d) Type IV

(a)

1cm

Type I

Fig. 2. Schematic representation of the rock relationships at an outcrop scale and photographs of the individual rock types. (a) Bandedorthogneiss (type I); (b) stromatic migmatite (type II); (c) schlieren migmatite (type III); and (d) nebulitic migmatite (type IV). Theposition of this outcrop in the study area is shown in Fig. 1b.

3 2 P . H A S A L O V A E T A L .


during horizontal shortening of the thickened orogenicroot (Schulmann et al., 2005).

The second deformation (D2) is associated withreworking and folding of the S1 compositional layeringin banded orthogneiss, so that S1 is only preservedlocally in elongate relict domains (Fig. 2a). The D2

shearing is attributed to horizontal flow of hot lowercrust in a zone up to 10 km wide at a mid-crustal levelabove the Brunia promontory over distances of severaltens of kilometres (Schulmann et al., 2005). Relicdomains with gently folded S1 fabric are surrounded byhighly deformed zones with tightly folded S1 fabric(Fig. 2). The composite S1)2 fabric is characterized bya banded structure with diffuse boundaries betweenpolymineralic K-feldspar- and plagioclase-richdomains similar to a stromatic migmatite structure(type II) (Fig. 2b). Locally the S1 fabric is completelytransposed and a new S2 foliation is dipping gently tothe SW (Fig. 1b). A sub-horizontal, gently S–SWplunging L2 lineation (Fig. 1b) is mostly defined bypreferred orientation of sillimanite.

Detailed field observations reveal that with ongoingdeformation the type II rock gradually pass into moreisotropic rock (type III) composed of K-feldspar–quartz and plagioclase–quartz aggregates (Fig. 2c) andcontaining rootless folds of the deformed S1 fabric.This rock type alternates with irregular bodies orelongate lenses of fine-grained isotropic felsic rock(type IV, Fig. 2d), which in this region traditionallyhas been described as a nebulitic migmatite (Matejo-vska, 1974). Such a structural sequence originatedthrough intense D2 deformation superimposed on anolder steep anisotropy and was identified in outcropscale along several sections. These observations aresupported by the existence of macroscopically visibleleucosomes or granitic veins that are also parallel to S2and form isolated elongate pockets and lock-up shearbands.

This area has been extensively studied by Matejo-vska (1974) and Dudek et al. (1974) who used theclassical migmatite terminology of Mehnert (1971) forthe above-described rock types. These authors identi-fied type I rock as banded orthogneiss, rock type II asstromatic migmatite and rock type IV as nebuliticmigmatite. Rock type III resembles the schlierenmigmatite of Mehnert (1971). Because the Gfohl Unitis considered as one of the largest migmatitic terranesof the Variscan belt, the traditional migmatite termi-nology was adapted to these rocks.

MICROSTRUCTURAL OBSERVATIONS

The microstructural characteristics including grainsize, grain shape and grain boundary geometry werestudied in each of the four rock types and inK-feldspar- and plagioclase-rich domains. Thin sec-tions were cut perpendicular to the foliation andparallel to L2 lineation (XZ section). To discriminateK-feldspar from plagioclase, the thin sections were

stained according to the method of Bailey & Stevens(1960).

Type I: banded orthogneiss

This rock type is a fine-grained orthogneiss with 0.25-to 2.0-mm-thick layers of recrystallized plagioclase(30 modal%), K-feldspar (40 modal%) and quartz(20 modal%), separated by discrete layers of biotite(10 modal%) commonly associated with minor silli-manite and garnet (Table 1, Fig. 3a).

K-feldspar forms completely recrystallized aggre-gates (0.2–0.8 mm grain size) with straight grainboundaries locally meeting in triple point junctions at120� (Fig. 4a). Numerous rounded inclusions of quartz(0.05 mm) occur preferentially at triple points, alongplanar boundaries or in cores of feldspar (Fig. 4a).Plagioclase (An10)20) is present in K-feldspar aggre-gates as small interstitial grains or forms thin filmspreferentially tracing those K-feldspar boundaries thatare oriented at a high angle to the foliation (Fig. 5a).Rarely, tiny interstitial biotite is present in the K-feldspar-rich bands.

Plagioclase aggregates (0.2–0.5 mm) are composedof an equidimensional polygonal mosaic with straightboundaries, and minor interstitial quartz and biotite(Fig. 4b). The plagioclase grains show abundanttwinning and form a foam-like texture with a perfecttriple point network of grain boundaries. Plagioclaseexhibits normal zoning with homogeneous oligoclasecores (An24)28) and more sodic (An10)18), clear,2 to 10 lm-thick rims at boundaries with K-feldspar.Plagioclase grain size continuously decreases from thecentre of an aggregate towards its borders. Quartzoccurs as small (0.01–0.05 mm) rounded inclusions orinterstitial grains, whereas K-feldspar exhibits charac-teristic cuspate shapes (Fig. 5b). Tiny biotite grains(0.1–0.5 mm in length; XFe ¼ 0.42–0.48, Ti ¼ 0.2–0.27 p.f.u.) commonly occur along the plagioclaseboundaries that are sub-parallel to the foliation(Fig. 3a).

Quartz ribbons 0.3–1.0 mm wide are composed ofelongate grains with straight grain boundaries per-pendicular to the ribbon margin (Fig. 3a). Quartz–feldspar boundaries are gently curved, with cusps thatpoint from feldspar to quartz. Biotite-rich layerscommonly show decussate microstructure, which is atextural equivalent of the foam-like texture of the felsicminerals (Vernon, 1976). Contacts between biotite-and plagioclase-rich layers are marked by numerous(<1 modal%) small idiomorphic garnets (0.05–0.10 mm in size; XFe ¼ 0.77–0.85).

Type II: stromatic migmatite

This rock type is composed of plagioclase- and K-feldspar-rich aggregates with subordinate quartz andirregular quartz aggregates (Fig. 3b); modes are givenin Table 1. These aggregates are rimmed by relicts of



biotite-rich layers commonly intergrown with fibroliticsillimanite. The ill-defined and rather diffuse bound-aries between individual aggregates are characteristicfor this textural type (Fig. 3b).

K-feldspar-rich aggregates (0.2–0.8 mm grain size)include abundant small interstitial plagioclase (An17)20in the core and An4)10 at the rim) and quartz. K-feldspar forms irregular grains with lobate boundaries.Most of the K-feldspar grain boundaries are decoratedby thin interstitial plagioclase (An6)8). Quartz is pres-ent as inclusions (0.01–0.06 mm) in K-feldspar orforms irregular grains (0.1–0.2 mm) along the K-feld-spar boundaries. In plagioclase-rich aggregates (0.1–0.3 mm grain size, An17)20), plagioclase grains havestraight to lobate boundaries, whereas irregularly dis-tributed interstitial quartz has irregular shapes. Similarto the type I orthogneiss, interstitial K-feldspar showscuspate shapes.

Biotite (15 modal%; XFe ¼ 0.55–0.59, Ti ¼ 0.18–0.26 p.f.u) is variable in size and appears along pla-gioclase boundaries mostly parallel to the foliation(Fig. 3b). Small idiomorphic garnet (0.07–0.2 mm insize; XFe ¼ 0.84–0.91) occurs in the plagioclase-richaggregates and biotite-rich layers.

Type III: schlieren migmatite

This textural type does not show macroscopically vis-ible feldspar-rich aggregates, but stained thin sectionsreveal the presence of plagioclase–quartz and K-feld-spar–quartz-enriched domains (Fig. 3c); modes aregiven in Table 1. The foliation is marked by preferredorientation of biotite and sillimanite dispersed in therock.K-feldspar forms large irregularly shaped grains

(0.1–0.3 mm in size) (Figs 3c & 4c) or small cuspategrains along plagioclase boundaries. The most char-acteristic feature is the presence of irregular embay-ments of quartz and plagioclase in the K-feldspargrains (Fig. 4d). Myrmekitic aggregates are commonlydeveloped along the K-feldspar boundaries (Fig. 4c).Plagioclase occurs as large irregular twinned grains(An12)16 in the core, An2)4 at the rim) and as films(An1)4) lining the K-feldspar boundaries (Fig. 5c).Entirely dispersed quartz forms large relict grains(0.7–1.0 mm) with undulatory extinction and highlylobate boundaries, abundant irregular interstitialgrains lining the K-feldspar boundaries androunded inclusions (0.02–0.05 mm) in K-feldspar and

Table 1. Representative data for the quantitative textural analysis.

Banded orthogneiss Stromatic migmatite Schlieren migmatite Nebulitic migmatite

Kfs domain P1 domain Kfs domain P1 domain Kfs domain P1 domain

Grain size – Feret diameter (mm)

Median Kfs 0.430 0.121 0.345 0.065 0.172 0.138 0.137

P1 0.134 0.224 0.086 0.225 0.094 0.119 0.110

Qtz 0.079 0.076 0.071 0.079 0.070 0.076 0.074

Ql Kfs 0.120 0.065 0.194 0.042 0.103 0.082 0.085

P1 0.103 0.095 0.055 0.158 0.061 0.084 0.074

Qtz 0.046 0.051 0.044 0.050 0.046 0.047 0.039

Q3 Kfs 0.630 0.170 0.556 0.101 0.263 0.211 0.237

P1 0.257 0.373 0.161 0.350 0.172 0.164 0.160

Qtz 0.105 0.114 0.119 0.127 0.105 0.121 0.127

Q3 ) Q1 Kfs 0.510 0.105 0.362 0.059 0.161 0.129 0.152

P1 0.154 0.278 0.107 0.192 0.111 0.080 0.086

Qtz 0.059 0.063 0.075 0.077 0.060 0.074 0.089

Crystal size distribution (CSD)

N0. (mm)4) Kfs 0.0037 – 0.00487 – 0.053 – 0.2124

P1 – 0.0303 – 0.0733 – 0.06812 0.1857

Qtz 1.008 2.334 1.6448 3.73 2.093 0.6585 2.286

Gt Kfs 0.347 – 0.286 – 0.148 – 0.1127

P1 – 0.15731 – 0.1269 – 0.11 0.0689

Qtz 0.0736 0.0569 0.0669 0.0547 0.0644 0.0813 0.0623

Shape preferred orientation (SPO)

Eigenvalue ratio (Rg) Kfs 1.48 1.42 1.42 1.32 1.21 1.13 1.15

P1 1.17 1.42 1.1 1.23 1.21 1.14 1.13

Qtz 1.25 1.47 1 1.24 1.1 1.33 1.32

Aspect ratio (median) Kfs 1.66 1.69 1.6 1.5 1.59 1.6 1.55

P1 1.51 1.6 1.59 1.44 1.65 1.5 1.61

Qtz 1.5 1.5 1.46 1.5 1.5 1.5 1.49

Bt 2.14 2.7 2 2.2 2.2 2.35 2.2

Grain boundary preferred orientation (GBPO)

Eigenvalue ratio (Rb) Kfs–Kfs 1.34 – 1.25 – 1.06 – 1.5

Kfs–P1 1.15 1.18 1.09 1.13 1.14 1.13 1.12

Kfs–Qtz 1.17 – 1.15 – 1.17 – 1.17

P1–P1 – 1.18 – 1.15 – 1.15 1.36

Modal proportion (%)

Kfs 70–80 10 70–80 5 50 20–25 30

P1 10 60 10 60 20 40 30

Qtz 10–20 20 10–20 25 30 30 30

Bt <1 <10 <1 �10 <1 <10 10

Sill, Grt 0 <1 0 <1 0 <1 <2

3 4 P . H A S A L O V A E T A L .


plagioclase (Fig. 4c). Biotite (10–15 modal%; XFe ¼0.76–0.79, Ti ¼ 0.18–0.19 p.f.u.) is homogeneouslydispersed and is most prevalent in the plagioclase–quartz domains. Atoll-shaped garnet (0.05–0.25 mm insize; XFe ¼ 0.96–0.97) appears inside the felsic aggre-gates, rather than along contacts with biotite.

Type IV: nebulitic migmatite

This type of rock is composed of almost equal amountsof plagioclase, K-feldspar and quartz, and containsminor biotite (XFe ¼ 0.91–0.93, Ti ¼ 0.01–0.04 p.f.u.),sillimanite and garnet (XFe ¼ 0.98–1.00) (Fig. 3d),with a weakly developed preferred orientation of thebiotite and sillimanite; modes are given in Table 1. K-feldspar (0.10–0.25 mm in size) occurs in the form ofirregular grains embayed with quartz and plagioclase.Commonly, the intensity of quartz and plagioclaselobes correlates well with highly cuspate irregularforms of corroded relics of K-feldspar (Fig. 4e). Sim-ilarly, the relics of irregular plagioclase (0.05–0.15 mmin size; An6)10 in the core and An0)4 at the rim) showcuspate boundaries, but with curvature less pro-nounced than that of the corroded relics of K-feldspargrains. An important feature is the presence of newplagioclase (An0)1)–K-feldspar intergrowths embayingcorroded relics of K-feldspar grains (Fig. 4f). Quartz(0.04–0.07 mm) with highly lobate boundaries is uni-formly distributed in the rock. Biotite of low aspectratio shows highly corroded cuspate forms filled withquartz, K-feldspar and plagioclase.

Summary of modal changes

Modal composition of the feldspar aggregates in thetype II migmatite does not change significantly com-pared with the type I orthogneiss. However, the typeIII migmatite is characterized by an important increasein quartz content in feldspar domains (up to 30 mod-al%) associated with a slight increase in interstitialplagioclase in K-feldspar-rich domains and K-feldsparin plagioclase-rich domains. The proportions of thefelsic minerals are equal in the type IV migmatite.

1 mm

(a)

1 mm

(b)

1 mm

(c)

(d)

1 mm

QtzPlKfs

Bt,Sill,Grt

Type II

Type III

Type IV

Type I

Fig. 3. Representative digitalized microstructures (XZ sections)for individual textural types (note differences in scales whenmaking comparisons). (a) Banded orthogneiss (type I) with dis-tinct monomineralic layers composed of a polygonal mosaic ofwell-equilibrated plagioclase, K-feldspar and quartz polycrystal-line ribbons separated by discrete layers of biotite ± sillima-nite ± garnet (sample PH60/B). (b) Stromatic migmatite (type II)composed of K-feldspar-rich, plagioclase-rich and quartz-richaggregates separated by relicts of biotite ± sillimanite-rich layers(sample PH60/A). (c) Schlieren migmatite (type III) showingalternation of K-feldspar- and plagioclase-rich domains inter-preted to correspond to an original spatial distribution (K-feld-spar domain is shown, sample PH90). (d) Isotropic nebuliticmigmatite without any gneissosity (type IV) composed of equalamounts of K-feldspar, plagioclase and quartz (sample PH59/D).



Evidence of melting

Sawyer (1999, 2001) summarized criteria for recogni-tion of former melt at grain scale in metamorphicrocks. The three most important features are: (i) min-

eral pseudomorphs after thin melt films along crystalfaces, a feature typically observed in melting experi-ments under dynamic conditions (Jin et al., 1994); (ii)rounded and corroded reactant minerals embayed bysurrounding mineral pseudomorphs after melt (Busch

(d)

100 µm

Kfs

Qtz

Pl

Qtz

Pl

(f)

0.2 mm

Kfs-Pl intergrowth

Kfs

PlBt

Qtz

Pl

(e)

100 µm

Pl

Qtz

Qtz

Kfs

Pl

Kfs

(a)

0.5 mm

Kfs

Q

(b)

Bt

Pl

Kfs

0.2 mm

(c)

0.5 mm

Kfs

Qtz

Kfs

Pl

PlBt Kfs

Qtz

Kfs

Qtz Pl Irregular embayments of relict feldspar Myrmekites

3 6 P . H A S A L O V A E T A L .


Kfs

100 µm

QtzPl

(b)

Kfs

KfsKfs

Qtz

Pl(An25)

Pl(An25)

0.5 mm

An10-20An10-20

An10-20

(a)

Qtz

Qtz

Qtz

100 µm

Qtz

Kfs

Kfs

Pl relict grain

(c)

Pl rim(An1-4)

Pl rim(An1-4)

Qtz

(An1-4)

(An12-16)(An12-16)

Pl

Pl core(An12-16)

200 µm

Qtz

QtzQtz

Kfs

(An15)

(An15)

(An0-4)(d)

Kfs

Kfs

Qtz

Pl

Pl

QtzQtz

Qtz

Qtz

Qtz

Pl

Pl

Fig. 5. SEM backscatter images showing the inferred former melt topology (note differences in scales when making comparisons). (a)Type I banded orthogneiss: interstitial plagioclase (An10)20), representing the plagioclase component crystallized from the anatecticmelt (grey arrow), tracing the K-feldspar boundaries sub-perpendicular to the foliation (sample PH60/B). Black arrows show smallrounded quartz grains crystallized along feldspar boundaries. (b) Type I banded orthogneiss: inferred former melt pools with cuspatemargins in a plagioclase band (sample PH60/B). The former melt has crystallized to K-feldspar (cuspate melt pools), plagioclase(growing on the old plagioclase grains) and quartz (forming small rounded grains along the feldspar boundaries (black arrow)). (c)Type III schlieren migmatite: more developed interstitial plagioclase (grey arrow) with normal zoning (core ¼ An12)16; rim ¼ An1)4)and distinct albite rims (An1)4) on relict feldspar grains (white arrow) (sample PH90). The interstitial plagioclase is not in opticalcontinuity with any residual plagioclase grains adjacent to it and does not show any preferred orientation, in contrast to plagioclase intypes I and II. (d) Type III schlieren migmatite: new plagioclase inferred to have crystallized from melt (growing on an old plagioclasegrain in the form of the discrete albite rims (white arrow)) and quartz grains that resorb relict K-feldspar grains (sample PH14/D).

Fig. 4. Photomicrographs showing characteristic textures of the rock sequence (note differences in scales when making comparisons).(a) Type I banded orthogneiss: recrystallized K-feldspar aggregate with straight grain boundaries and numerous smaller roundedquartz grains (white triangles) along the boundaries or in the cores of feldspar (sample PH60/B). (b) Type I banded orthogneiss: well-developed plagioclase polygonal foam-like texture with straight grain boundaries, interstitial quartz (white triangles) and biotite(sample PH60/B). (c) Type III schlieren migmatite: typical microstructure with irregularly shaped feldspar and quartz grains withhighly lobate boundaries. Myrmekitic aggregates commonly develop along the K-feldspar boundaries (black arrow). New smallinterstitial plagioclase (grey triangles), K-feldspar and quartz (white triangles) grains trace almost all the relict feldspar boundaries.Interstitial quartz forms preferentially rounded shapes different from plagioclase which forms thin elongated grains/films coatingK-feldspar boundaries (sample PH90). Such a microstructure is typical also for the type IV. (d) Type III schlieren migmatite: irregularcuspate K-feldspar grain embayed with newly crystallized quartz and plagioclase (sample PH90). (e) Type IV nebulitic migmatite:corroded relics of K-feldspar grains (sample PH59/D). (f) Type III nebulitic migmatite: plagioclase-K-feldspar intergrowths embayingrelict K-feldspar grain (sample PH14/D). White arrows in (d), (e) and (f) point to irregular embayments of relict K-feldspar originatedthrough resorption of old K-feldspar grains by newly crystallized material.



et al., 1974); and (iii) cuspate and lobate areas inferredto represent pools of crystallized melt (Jurewicz &Watson, 1984).

The former presence of melt at grain scale was in-ferred from the following microstructures (Figs 4 & 5).(i) Plagioclase films between adjacent K-feldspargrains, inferred to represent a plagioclase componentcrystallized from melt (Fig. 5a, c). This plagioclase ischaracterized by more albitic composition and by dif-ferent topology compared with original grains. (ii) Pl–Kfs–Kfs and Kfs–Kfs–Pl dihedral angles commonlylower than 30� (Fig. 5a, c), as observed in granitic meltcrystallized under experimental conditions (e.g.Laporte et al., 1997). (iii) Cuspate K-feldspar pools inplagioclase aggregates (Fig. 5b), inferred to represent aK-feldspar component crystallized from melt (Jurewicz& Watson, 1984; Sawyer, 1999, 2001). (iv) Normalzoning of plagioclase from An10)30 to An0)15 (Sawyer,1998; Marchildon & Brown, 2001) lining K-feldsparboundaries (Fig. 5c, d). An important feature is thepreferential orientation of plagioclase films coating K-feldspar boundaries in type I orthogneiss and type IImigmatite sub-perpendicular to the foliation (Fig. 5a),in contrast to the types III and IV migmatites, wherethese films are wider and do not show any opticallyvisible preferred orientation. Bulbous myrmekite(Fig. 4d) and new highly irregular lobate grains thatovergrow partially resorbed corroded feldspar grains(e.g. Fig. 4c) are similar to microstructures describedas typical of minerals reacting with melt (Mehnertet al., 1973; Busch et al., 1974; McLellan, 1983).

QUANTITATIVE TEXTURAL ANALYSIS

The quantitative analysis of texture is based onstatistical evaluation of grain size distributions

(Kretz, 1966, 1994; Ashworth, 1976; Ashworth &McLellan, 1985; Cashman & Ferry, 1988; Cashman& Marsh, 1988; Higgins, 1998; Berger & Roselle,2001), spatial distribution of minerals and GBPOs(Panozzo, 1983), and grain contact frequencies(Flinn, 1969; Kretz, 1969; McLellan, 1983; Kruse &Stunitz, 1999). In simple chemical systems, thesetextural parameters are more sensitive to changes ofphysical conditions than compositional characteris-tics. This is due to the high activation energies ofchemical reactions needed to produce new crystalgrowth compared with the small amount of latticestrain energy and grain boundary energy required todrive recrystallization processes (Spry, 1969; Stunitz,1998).In this study, the textures of three samples were

analysed from each rock type, and in each samplemore than 1000 grains were evaluated in thin section.Due to significant textural variations, the individualK-feldspar-rich and plagioclase-rich domains wereanalysed separately. Maps of grains with full topol-ogy were manually traced into the ESRI ArcViewDesktop GIS environment and grain boundaries weregenerated using the ArcView PolyLX extension (Lexa,2003). The �shapefiles� of individual digitalized thinsections are attached in Appendix S1 (Supplementarymaterial). Analysis of grain size, CSD, grain shapepreferred orientation (SPO), grain boundary preferredorientation (GBPO) and grain contact frequencieswere obtained using the MATLABMATLAB

TM PolyLX toolbox(Lexa, 2003; http://petrol.natur.cuni.cz/ ondro/). Thegrain sizes of the minerals were evaluated in termsof Feret diameter (diameter of a circle having thesame area as the grain). Two methods were used todetermine the grain SPO: (1) mean directions usingcircular statistics; and (2) eigenvalue analysis ofScheidegger’s bulk orientation tensor calculated fromindividual long axes weighted by grain size (Lexaet al., 2005), where degree of SPO is expressed as theeigenvalues ratio Rg. GBPO was assessed by similartechniques, but the bulk orientation tensor is formedfrom the decomposed grain boundaries betweenchosen phases (Lexa et al., 2005) and the degree ofGBPO is expressed as the eigenvalues ratio Rb. Graincontact frequency, used to examine statistical devia-tion from a random spatial distribution of contactrelations between the individual minerals, was evalu-ated in a manner similar to the method of Kretz(1969, 1994), except that contact frequencies wereobtained directly from grain map topologies insteadof using line intercepts.Results of the quantitative microstructural analyses

show an evolutionary trend from the banded orthog-neiss, through the migmatite types II and III to thenebulitic migmatite. Therefore, in the following sec-tions the rock types are discussed as a sequence inwhich the type I orthogneiss and type IV nebuliticmigmatite are considered to be end-members of acontinuous microstructural evolution.

Nebulitic migmatite (Type IV)

Banded orthogneiss (Type I)Stromatic migmatite (Type II)Schlieren migmatite (Type III)

Kfs70% 30%

30%

50%

Pl

Qtz

Fig. 6. Modal changes in both plagioclase (open symbols) andK-feldspar (closed symbols) aggregates in different rock typesplotted in a quartz–plagioclase–K-feldspar triangle. Arroweddashed lines indicate evolutionary trend from type I bandedorthogneiss to type IV nebulitic migmatite.

3 8 P . H A S A L O V A E T A L .


Grain size analysis

The CSD is an important tool to estimate residencetime of magmas in magma chambers, cooling rates inrapidly quenched lavas, as well as to quantify texturesrelated to phenocrysts accumulation and fractionation(Cashman & Marsh, 1988; Marsh, 1988; Higgins,1998). In metamorphic petrology, CSD is used toobtain quantitative information concerning crystalnucleation and growth rates and nucleation densityand/or annealing (Randolph & Larson, 1971; Cash-man & Ferry, 1988; Carlson, 1989; Waters & Love-grove, 2002). Hickey & Bell (1996) proposed thatduring dynamic recrystallization decreasing strain rateto temperature ratio ( _e/T) leads to decrease in the ratioof nucleation and growth rate (N/G) and developmentof coarser grain size, whereas increasing _e/T leads toincreasing N/G and therefore to grain size decrease.This hypothesis is well documented in experimentalstudies with steel alloys (Sakai & Jonas, 1984) sup-ported by Azpiroz & Fernandez (2003) and Lexa et al.(2005) in naturally deformed rocks. These authorsevaluate the role of recrystallization mechanisms on N/G ratio of the CSD. The CSD is commonly used informerly partially molten rocks to evaluate combinedprocess of resorption and grain size decrease in react-ing phases in mesosome and nucleation and graingrowth and coarsening of minerals crystallizing inleucosomes (Dougan, 1983; McLellan, 1983; Ashworth& McLellan, 1985; Dallain et al., 1999). Because theprocesses controlling grain size distributions in thecrystallization of partially molten rocks are complexand interpretations uncertain, CSD has only rarelybeen used to describe textural evolution of migmatites(Berger & Roselle, 2001). In this work, the CSDmethods are used as a practical approach to parame-terize grain size frequency histograms and visualizetheir trends in a simple manner.

Grain size statistics were evaluated for the four rocktypes for plagioclase, K-feldspar and quartz and theresults are presented in the form of average grain size,expressed as a median value of the Feret diameter, andgrain size range expressed as the difference between thethird and first quartiles instead of standard deviationbecause of the log-normal distribution of measureddata (Fig. 7a, Table 1). The results are also summa-rized as CSD curves (plot of logarithms of populationdensity against crystal size) that were constructed usingthe method of Peterson (1996); values of the zero-sizeintercept (N0 – population density interpreted as theratio of nucleation rate to growth rate) and negativeinverse of slope (Gt interpreted as a function of growthrate) of the linear parts of the CSD curves are plottedin Fig. 7b, c.

Both plagioclase and K-feldspar in the type Iorthogneiss are characterized by log-normal grain sizedistribution exhibiting average grain size of �0.2 and0.2–0.5 mm respectively. Interstitial quartz yields sig-nificantly smaller average grain size of 0.05–0.1 mm in

both domains. Quartz grains from polycrystalline rib-bons were not evaluated statistically but their grain sizeof 0.5–2.0 mm was estimated using an optical micro-scope. The grain size of new interstitial plagioclase inthe K-feldspar aggregates is close to 0.1 mm. The grainsize distributions from type I orthogneiss to type II,type III and type IV migmatites are characterized bythe following features. The average grain size of pla-gioclase and K-feldspar decreases compared with typeI orthogneiss (Fig. 7a, Table 1). This is accompaniedby a continuous decrease in grain size range for bothfeldspars. The interstitial quartz grain size remainsfairly constant throughout all the stages of texturalevolution, ranging between 0.05 and 0.1 mm, beinglarger in the K-feldspar than in the plagioclasedomains (Fig. 7a). The grain size of minor plagioclasein the K-feldspar domains shows a bimodal distribu-tion that is attributed to the presence of small newlynucleated grains (0.06–0.1 mm) and to larger plagio-clase grains (0.2 mm) already present in the feldsparaggregates.

The CSD of plagioclase indicate continuous increasein N0 (nucleation density) values coupled with adecrease in Gt (growth rate) values from type Iorthogneiss towards type IV migmatite (Fig. 7b,Table 1). By contrast, K-feldspar shows a decrease inGt values from type I orthogneiss to type III migmatitewithout significant increase in N0 values, which remainvery low. From type III to type IV migmatite a dra-matic increase in N0 values is observed for K-feldsparat almost constant Gt values (Fig. 7c). This evolutionis clearly shown by steepening of the slopes of the CSDcurves accompanied by increase in their upper inter-cept with the ordinate axis (insets in Fig. 7b, c).

Grain shapes and grain shape preferred orientation

Grain shape or grain aspect ratio together with grainSPO analyses provide important information aboutdeformation during or after leucosome formation(Mehnert, 1971; McLellan, 1983) or about degree ofinheritance of original anisotropy (Ashworth, 1979).Measurements of preferred orientations of inferredmelt-filled grain boundaries in rocks give insights intoprocesses of melt draining and melt transfer (Rosen-berg & Handy, 2000, 2001; Sawyer, 2001).

Grain shape and SPO statistics were evaluated in allthe textural types for plagioclase, K-feldspar, quartzand biotite. The results of SPO statistic are summa-rized in a boxplot-type diagram, where the axial ratiosof the individual minerals are plotted against bulk SPO(Lexa et al., 2005) for the corresponding minerals(Fig. 8).

Aspect ratios for both K-feldspar and plagioclaseshow small median values ranging from 1.5 to 1.7throughout the whole microstructural sequence(Fig. 8, Table 1). Quartz exhibits slightly smaller andstable aspect ratio close to 1.5. An important feature isthe continuous decrease in SPO of K-feldspar and



plagioclase from type I orthogneiss to type IV nebuliticmigmatite (Fig. 8, Table 1). Rose diagrams for therock types I, II and III show that K-feldspar andplagioclase (Fig. 8b) have weakly inclined SPO withrespect to the aggregate elongation direction at anangle of 15�–30�. Biotite shows a high aspect ratio fortype I orthogneiss (Table 1) and strong SPO parallelwith mesoscopic foliation for the types I, II and IIImigmatites. In the type IV migmatite, biotite aspectratio and preferred orientation are lower and the latterparameter shows bimodal distribution with one maxi-mum sub-parallel to the main foliation and a secondone almost perpendicular to it. Interstitial K-feldspar,plagioclase and quartz exhibit always small aspect

ratio and weakly developed SPO maxima at an angleof 40�–60� to the foliation for types I, II and III. Theexception is type IV migmatite, where, in similarfashion to biotite, the interstitial plagioclase shows twomaxima, one sub-parallel and one perpendicular to thefoliation.

Grain contact frequency analysis and grain boundarypreferred orientation

The grain contact frequency method (Kretz, 1969)allows an examination of the statistical deviationfrom the hypothesis of random distribution of phasesin rocks. In random distribution, the number of

(b) (c)

(a)

Fig. 7. Grain size statistics and CSD evolution for the rock sequence. (a) Calculated average grain size (median value of the Feretdiameter) and range (difference of third and first quartiles) for plagioclase, K-feldspar and quartz. (b,c) Plots of crystal size distributionparameters N0 (corresponding to the nucleation density per size per volume) and Gt (non-dimensional value dependent on the growthrate) with examples of linearized CSD curves (upper right insets) used for Gt and N0 estimates. (b) Plagioclase, (c) K-feldspar. The CSDcurves show single lines of four representative samples corresponding to the individual rock types.

4 0 P . H A S A L O V A E T A L .


contacts of given phases depend only on the totalnumber of grains of each phase present. There aretwo possible deviations from random distribution: (i)aggregate distribution, where grains of the samephase tend to occur in aggregates in which contactsbetween grains of the same phase (like–like contacts)predominate; and (ii) regular distribution, where thegrains tend to occur in a regular (chessboard-like)pattern in which contacts between different phases(unlike contacts) are more common. McLellan (1983)reviewed processes responsible for different types ofgrain distributions. A random distribution shouldtheoretically develop during rapid quenching of gra-nitic melt, whereas regular distribution commonly isinterpreted as resulting from extensive solid-stateannealing under very high temperatures (Flinn, 1969;Vernon, 1976; McLellan, 1983; Lexa et al., 2005).These interpretations are based on the assumption ofreducing surface energy (Seng, 1936; DeVore, 1959)by elimination of high-energy contacts (commonlynon-coherent like–like contacts) either by reduction ofgrain boundary area or by nucleation and growth ofnew phases along such a boundary (Kim & Rohrer,2004). In addition, Kruse & Stunitz (1999) and Bar-atoux et al. (2005) proposed that the regular distri-bution was induced by mechanical mixing andheterogeneous nucleation. According to Vernon(1976) and McLellan (1983), an aggregate distributionresults from a solid-state differentiation associatedmostly with dynamic recrystallization where devel-opment of monomineralic layers results from unevenefficiency of deformation mechanisms simultaneouslyoperating in different phases (Jordan, 1988).

Grain contact frequency and the GBPO were eval-uated for K-feldspar, plagioclase and quartz in all thetextural types over the full digitized area of individualK-feldspar and plagioclase domains. The results arepresented in Fig. 9, where the v-value

v ¼ Observed� Expectedffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

Expectedp

or deviation from the random distribution is plottedagainst the ratio of eigenvalues of the orientationtensor or the degree of GBPO (Lexa et al., 2005).Values of expected frequencies are estimated usingLafeber’s method of testing for randomness (Lafeber,1963; Kretz, 1969). This diagram offers a simple visualevaluation of the relationship between degree of devi-ation from expected random distribution of graincontacts and GBPOs of like–like and unlike bound-aries.

The type I orthogneiss is characterized by a rela-tively small proportion of like–like K-feldspar andplagioclase contacts indicating a weak regular dis-tribution (slightly negative v-values for like–like andpositive v-values for unlike contacts; Fig. 9), despitea macroscopically banded texture in which a strongaggregate distribution should be observed. This fea-ture is attributed to a great proportion of minorinterstitial grains (Qtz, Bt, Kfs and Pl) lining theK-feldspar and the plagioclase boundaries. Addi-tionally, the number of like–like K-feldspar andplagioclase contacts continuously decreases from typeI orthogneiss to type IV migmatite, whereas thenumber of Pl–Kfs, Kfs–Qtz and Pl–Qtz unlike

(b)(a)

Fig. 8. Plot of grain shape preferred orientation (SPO) of K-feldspar (a) and plagioclase (b). The results are summarized in a box plotof aspect ratios (characterizing the shape of grains) v. eigenvalue ratios (showing the degree of preferred orientation). Individual boxesshow median, and first and third quartiles of the aspect ratio. The whiskers represent a statistical estimate of the data range whereoutliers are not plotted. Representative rose diagrams for individual rock types show maxima orientation in respect to the aggregateelongation direction. The degree of shading corresponds to the individual rock types.



contact continuously increases (negative like–like v-values and positive unlike v-values) (Fig. 9). This isin a good accordance with the increasing amount ofinterstitial phases towards the type IV migmatite.Quartz exhibits the same strong regular distributionfrom the type I orthogneiss to the type IV migmatitein both feldspar domains.

In K-feldspar-rich aggregates, the degree of GBPOof the K-feldspar like–like boundaries slightly de-creases from type I orthogneiss to type III migmatite,whereas type IV migmatite is characterized by an in-crease in the degree of K-feldspar like–like GBPO(Fig. 9, Table 1). The GBPO of plagioclase–plagio-clase boundaries in the plagioclase-rich aggregates issimilar to the evolution of K-feldspar like–likeboundaries. The GBPO of the K-feldspar–quartzboundaries as well as those of K-feldspar–plagioclaseboundaries are weak, and decrease throughout thetextural evolution (Fig. 9, Table 1).

MINERAL FABRIC

In rocks deformed in the presence of melt, the texturesof quartz and feldspar can be used to evaluate thedeformation mechanisms of the solid fraction as wellas the deformation of crystallizing intragranular melt(Zavada et al., 2007). The mineral fabrics of ferro-magnesian phases can be indirectly assessed usinganisotropy of magnetic susceptibility (AMS). TheAMS method has been recently used to determine thedegree of susceptibility, shape of the fabric ellipsoidand relative contribution of ferro- and para-magneticminerals to the bulk fabric in migmatites (Ferre et al.,2003, 2004).

Anisotropy of magnetic susceptibility

Types III and IV migmatites are macroscopically closeto isotropic, so that the mineral alignment defined bythe orientation of dispersed biotite is poorly defined(Fig. 2c, d). To better characterize the fabric, the AMSmethod was used to determine the internal fabric ofthese rocks. Oriented samples were collected using aportable drill at four sampling sites covering a sectionacross the well-defined structural sequence. The AMSdata were statistically evaluated using the Anisoftsoftware package (Jelınek, 1978; Hrouda et al., 1990).The low values of mean susceptibility (<250 · 10)6

SI) indicate that biotite is the main carrier of themagnetic susceptibility (Ferre et al., 2003). The AMSstudy reveals a homogeneous pattern of a south-dip-ping magnetic foliation (Fig. 10b) and SW sub-hori-zontally plunging magnetic lineation (Fig. 10b) thatare consistent with the mesoscopic D2 structuralpattern.According to the degree of AMS (expressed by

parameter P¢, Jelınek, 1981) and shape of the AMSellipsoid (expressed by parameter T, Jelınek, 1981) thesamples are divided in two groups (Fig. 10a). Thebanded orthogneiss and the type II migmatite exhibit astrong degree of magnetic anisotropy (P¢ ¼ 1.14–1.2)and planar shape of the ellipsoid of magnetic suscep-tibility (T ¼ 0.4–1) (Fig. 10a). These values are typicalfor metamorphic rocks with well-developed composi-tional layering and biotite aligned in planar aggregates.Samples from types III and IV migmatites occur in aregion of lower degree of anisotropy (P¢ ¼ 1.06–1.16)and correspond to a planar–linear fabric (T ¼ 0.1–0.7)(Fig. 10a) marked by more intense magnetic lineation.

Fig. 9. Grain boundary statistics plotted as the deviation from a random spatial distribution (grain contact frequency) v. degree ofgrain boundary preferred orientation (GBPO). For details see text. The degree of shading corresponds to the individual rock types.

4 2 P . H A S A L O V A E T A L .


Lattice preferred orientation

To understand the deformation behaviour of individ-ual phases, we measured and evaluated statistically thelattice preferred orientation (LPO) of aggregate grains(Pl, Kfs and Qtz) and grains apparently crystallizedfrom melt (Pl, Kfs and Qtz) separately (Fig. 12g, h).The LPO of quartz, plagioclase and K-feldspar weremeasured on a scanning electron microscope Cam-Scan3200 in the Czech Geological Survey using theelectron back-scattered diffraction technique (EBSD)and HKL technology (Adams et al., 1993; Bascouet al., 2001). Diffraction patterns were acquired at20 kV of accelerating voltage, 5 nA of probe currentand working distance of �33 mm from the thin sectionprepared from the structural XZ plane. The procedurewas carried out manually due to small differences indiffraction patterns. The chemistry and orientation ofindividual grains was controlled using a forescatterdetector with combination of orientation and chemicalcontrast. Thus, each individual grain is represented byonly one orientation measurement. The resulting polefigures are presented as lower hemisphere equal-areaprojections in which the trace of foliation is orientedalong the equator and the stretching lineation is in theE–W direction.

Old quartz grains in ribbons of the type I orthog-neiss show c-axes distributed in weak sub-maxima ar-ranged along weakly developed small circles close tothe S1 foliation trace. The most intense sub-maximaare developed close to the lineation direction. This typeof c-axis pattern may indicate preferential prism Æcæslip-system activity and dominantly coaxial deforma-tion. The c-axes of large quartz grains in types II, IIIand IV migmatites reveal strong maxima either parallelto the S2 foliation pole or close to the centre of the

diagram. These c-axis patterns indicate mainly activityof basal Æaæ or rhomb Æa + cæ slip-systems and lessfrequently prism Æaæ slip (Fig. 11a). Towards types IIIand IV migmatites, the LPO of the matrix quartz be-came less well developed, preserving activity of thesame slip-systems as in the previous microstructuraltypes (Fig. 11a). New quartz grains crystallized frommelt in type I orthogneiss, and type II migmatite showvery weak LPO and nearly random distribution of allquartz axes (Fig. 11b). Whereas old grains show pro-gressive weakening of the LPO from type II to type IVmigmatite, the new and randomly crystallized grainstends to develop weak crystal preferred orientationduring the same microstructural evolution from type IIto type IV migmatite (Fig. 11). It is difficult to distin-guish old from new quartz grains in the type IV rockand therefore the LPO of quartz in this microstructurelinks LPO evolution between old and new grains in thefinal microstructural type.

K-feldspar and plagioclase commonly show weakLPO in all rock types regardless the origin of grains.K-feldspar shows crystallographic patterns which arecompatible with dominant activity of the 1/2 [110](001)slip system (Willaime & Gandais, 1977; Willaime et al.,1979) (Fig. 12a, c). Contribution of other slip systemsas [100](010) (Fig. 12b) and [100](001) (Fig. 12d) hasalso been identified in both relict K-feldspar grains andin K-feldspar grains apparently crystallized from meltrespectively.

Distribution of the main lattice directions of pla-gioclase revealed slip parallel either to 1/2[1�10] on (001)and (11�1) planes (Fig. 12e) or to 1/2 [110] on (001) and(1�1�1) planes (Fig. 12f) for all types of rocks and bothaggregate grains and plagioclase inferred to havecrystallized from melt (Fig. 12g) (Olsen & Kohlstedt,1984). The textures of plagioclase inferred to have

(a) (b)

Fig. 10. Plots to show the anisotropy of magnetic susceptibility (AMS). (a) P¢–T plot, where the P¢ parameter represents the degree ofmagnetic anisotropy and T is a shape parameter that describes the shape of the ellipsoid of magnetic susceptibility. T can take eitherpositive values (T > 0), characteristic for a planar fabric, or negative values (T < 0), typical for a linear fabric. Dashed ellipses showtwo distinct datasets. For comparison, data obtained by Schulmann K., Edel J.-B., Hasalova P., Lexa O., Jezek J. & Cosgrove J. W.(unpublished data) and Bouchez (1997) are shown. (b) Magnetic foliation (circles), plotted as the minimal susceptibility direction (K3),perpendicular to the magnetic foliation, and magnetic lineation (squares), plotted as the maximal susceptibility direction (K1).



crystallized from melt are commonly weak with theexception of strong LPO of plagioclase in the type Iorthogneiss (Fig. 12f). Such slip-systems are supposedto be secondary and active if grains are in unsuitable(hard) orientation to the dominant slip-system[100](010) (Kruse et al., 2001).

DISCUSSION

This study presents a detailed microstructural andquantitative textural analysis of four types of migmat-itic rocks identified in one of the largest (�5000 km2)migmatitic complex of the eastern Variscan belt. Therock types are interpreted as representing a texturalsequence from banded orthogneiss via stromatic andschlieren migmatites to nebulitic migmatite. The pos-sible mechanisms that could account for the origin ofthis rock sequence involve: (i) genetically unrelatedmigmatites that have originated from distinct proto-liths; (ii) variable degree of in situ partial melting of asingle protolith or different protoliths; and (iii) meltinfiltration from an external source through solid rockin which banded orthogneiss and nebulitic migmatiterepresent genetically linked end-members. Thesehypotheses are discussed further below.

Spatial relationships of individual migmatite types withinthe shear zone

The structural sequence described in this work indi-cates an intimate relationship between types I to IIImigmatites and nebulitic type IV migmatite sheets thatcan be interpreted in terms of a shear zone, which wasexploited by rising magma (Brown et al., 1995; Collins& Sawyer, 1996; Brown & Solar, 1998b). We haveshown that the D2 flat fabrics that cross-cut the steepfoliation S1 developed at high-temperature solid-stateconditions (Fig. 1). Tajcmanova et al. (2006) andRacek et al. (2006) described a similar sequence ofsuperposed fabrics in lower crustal rocks several tensof kilometres to the north and south of the studied arearespectively. These authors proposed that the flat D2

deformation fabrics originated due to thrusting oforogenic lower crust over middle crustal units along alarge-scale retrograde shear zone. In agreement withthese authors, we suggest that the D2 fabrics developedin a thrust related crustal-scale shear zone reportedalready by Urban (1992), Schulmann et al. (1994) andredefined later by Schulmann et al. (2005). The maindifference between other regions is in the degree of D2

reworking, which is so high in the studied area that

(b)

(a)

Fig. 11. Characteristic c-axes preferred orientations of (a) old/relict quartz grains and (b) new quartz grains crystallized from areas ofinferred former melt for all rock types. The c-axis patterns of old/relict quartz grains in type I banded orthogneiss indicate prism Æcæslip system activity whereas in type II, III and IV migmatites basal Æaæ or rhomb Æa + cæ slip systems are dominant with minorprism Æaæ slip. New quartz grains inferred to have crystallized from melt in type I banded orthogneiss to type IV nebulitic migmatiteshow very weak LPO and nearly isotropic distribution of all quartz axes. Equal area projections, lower hemisphere, contoured atinterval of 0.5 times uniform distribution. Foliation is horizontal and lineation is in this plane in the E–W direction. N is the number ofmeasured grains. Maximum densities are marked on the bottom right of each pole figure. The dashed line represents the lowest contourlevel and the grey circle corresponds to the minimum density value.

4 4 P . H A S A L O V A E T A L .


Fig. 12. Characteristic LPO patterns of K-feldspar (a–d) and plagioclase (e,f). Both feldspars commonly show weak LPO in all rocktypes regardless of the origin of the grains. An exception is the strong LPO of new plagioclase grains in the type I banded orthogneiss(f). K-feldspar usually shows activity of 1/2[110](001) (a,c; type I), but also of [100](010) (b; type IV) and [100](001) (d; type IV) slipsystems. Plagioclase reveals activity of secondary slip systems such as 1/2[1�10] on (001) and (11�1) or 1/2 [1�10] on (001) and (1�1�1)(e,f; type I). Equal area projections, lower hemisphere, contoured at intervals of 0.5 times uniform distribution. Foliation is horizontaland lineation is in this plane in the E–W direction. N is the number of measured grains. Maximum densities are marked on thebottom right of each pole figure. The dashed line represents the lowest contour level and the grey circle corresponds to the minimumdensity value. (f,g) BSE images depicting the microstructural appearance of examples of the measured phases (sample PH60/B).Original plagioclase (g) and K-feldspar (h) aggregates with newly crystallized quartz (white arrow), plagioclase (black arrows) andK-feldspar (grey arrows) are shown.



steep D1 fabrics are preserved only as rare relics shownin Fig. 2.

Our structural observations are compatible withprogressive transposition within the ductile shear zoneranging from type I banded orthogneiss to highlyreworked type III schlieren migmatite. This interpre-tation is based on progressive folding of early steeporthogneiss fabric, development of isoclinal folds andcomplete fabric transposition and development of typeIII migmatite in a ductile shear zone (Fig. 2; Turner &Weiss, 1963). In such a context, the elongated bodies oftype IV nebulitic migmatite can be seen as veins ofisotropic granite penetrating parallel to the main S2mylonitic anisotropy (e.g. Cosgrove, 1997; Brown &Solar, 1998a). Alternatively, the type IV nebuliticmigmatites could be interpreted in terms of injectedmelt into hot country rocks (called also magmawedging, Weinberg & Searle, 1998) preventing magmafreezing during D2 shearing. Finally, the nebuliticmigmatite can be also regarded as the most extremeend-member of the structural sequence, i.e. completelydisintegrated parental orthogneiss.

In summary, the type I orthogneiss to type III mi-gmatite show intimate spatial relationships suggestingthat they have originated from the same protolith andthat they are genetically linked. However, the macro-scopic observations alone cannot distinguish the originof type IV migmatite and further arguments arerequired.

Microstructural and petrological arguments for melt–rockinteraction during exhumation

We suggest, in agreement with Sawyer (1999, 2001),that the position and topology of new plagioclase,quartz and K-feldspar grains in type I orthogneiss andtype II migmatite may be interpreted in terms of meltproducts crystallized along boundaries of the feldsparin individual aggregates (Fig. 5a–c). The main differ-ence between type I orthogneiss and type II migmatiteis a more albitic composition of plagioclase and agreater modal content of new phases in the latter.Types III and IV migmatites show development ofhighly corroded shapes of K-feldspar, plagioclase andbiotite (Fig. 4c–f). This indicates that all rock typesexhibit features compatible with the presence of meltand its interaction with the solid rock. Additionally,the degree of melt–rock interaction is inferred toincrease from type I orthogneiss towards type IVnebulitic migmatite (Fig. 4).

The structural sequence exhibits a distinct trend inmodal composition of originally monomineralic layersthat are progressively converted into polymineralicaggregates of granitic composition (Fig. 6). The com-positional paths show evolutionary trends from type Iorthogneiss to type III migmatite, interpreted as beingassociated with crystallization of melt, culminating inthe type IV migmatite, which has equal amounts ofplagioclase, K-feldspar and quartz (Fig. 6).

Plagioclase shows systematic decrease in anorthitecontent for both original plagioclase grains (An30 toAn25), their rims and inter-granular aggregates (An20to An10) towards type IV nebulitic migmatite. Bothgarnet and biotite exhibit systematically increasing XFe

towards type IV migmatite (from 0.7 to 1.00 and from0.4 to 0.9 respectively) coupled with decrease in Ticontent in biotite (from 0.2 to 0.04 p.f.u.). The mineralcompositional data suggest systematic equilibration ofgarnet, biotite and plagioclase compositions in thestability field of sillimanite with decreasing tempera-ture. The full petrological data and P–T estimates formelt–rock interaction are presented in a companionpaper (Hasalova et al., 2008a). Here, we quote the P–Testimates based on thermodynamic modelling usingTHERMOCALC (Powell et al., 1998) to point out thedecrease in temperature from 790 to 850 �C at 7.5 kbarfor type I orthogneiss to 690–770 �C at 4.5 kbar fortype IV nebulitic migmatite.Taken together, the microstructural and petrological

data show paradoxically an increasing degree ofapparent melt–rock interaction coupled with decreas-ing equilibration temperature with textural evolutionfrom type I to type IV rock type. The microstructuraland modal composition data do not exclude eitherpartial melting or infiltration of melt from an externalsource. However, the systematic modification ofchemical composition of minerals across the migmatitesequence is in contradiction with the model of in situpartial melting. Namely, as suggested by many fieldand experimental studies, the composition of plagio-clase would be shifted towards more anorthitic con-tents and the XFe of garnet and biotite would decreaseduring partial melting process (Le Breton & Thomp-son, 1988; Vielzeuf & Holloway, 1988; Gardien et al.,1995; Greenfield et al., 1998; Dallain et al., 1999).

Interpretation of quantitative microstructural data

Microstructural studies of partially molten rock haverevealed systematic changes in grain size, grain SPOand spatial distribution of individual phases duringincreasing degree of partial melting (e.g. Vernon, 1976;McLellan, 1983; Dallain et al., 1999). Here thesetrends are compared with our quantitative micro-structural data, and the alternative origins that mayresult in the observed microstructural sequence arediscussed.

Interpretation of crystal size distributions

The most significant result of this study is the sys-tematic decrease in average grain size (Fig. 7a) andsystematic increase in a population density (nucleationrate) associated with possible decrease in growth ratefor all feldspar from type I orthogneiss to type IVmigmatite (Fig. 7b, c).Results from migmatitic terranes show that the CSD

associated with partial melting is characterized by

4 6 P . H A S A L O V A E T A L .


production of coarse-grained felsic mineral aggregatesresulting from increase in temperature (e.g. Dougan,1983; McLellan, 1983). This process is commonly fol-lowed by textural coarsening (Ashworth & McLellan,1985; Dallain et al., 1999; Berger & Roselle, 2001)explained by two competing approaches: the Lifshitz–Slyozov–Wagner (LSW) model (Lifshitz & Slyozov,1961), and the communicating neighbour theory (CNof DeHoff, 1991). Higgins (1998) showed that texturalcoarsening results in progressive decrease in N0 valueand decrease in the slope of the CSD curve; he inter-preted this trend as a result of rapid undercoolingduring solidification of magma followed by reducedundercooling, suppression of nucleation and texturalcoarsening. However, our textural sequence exhibitsthe opposite trend in evolution of CSD curves, which isinterpreted as indicating that in situ partial melting andtextural coarsening are not responsible for the origin ofobserved CSDs.

The observed CSD trend may be explained by one ofthree different mechanisms: (1) solid-state deformationunder decreasing temperature and or increasing strainrate (Hickey & Bell, 1996; Azpiroz & Fernandez, 2003;Lexa et al., 2005); (2) a different degree of reactionoverstepping (Waters & Lovegrove, 2002; Moazzen &Modjarrad, 2005); and (3) a different degree of und-ercooling (Marsh, 1988).

The grain size for dynamically recrystallized grainsin a power-law creep regime is a function of differentialstress (Twiss, 1977). Such grains are characterized bystrong shape and LPO and commonly solid-state dif-ferentiation (Baratoux et al., 2005; Lexa et al., 2005).However, this microstructural study does not revealany features in quartz, plagioclase and K-feldspar ofrock types II, III and IV which may indicate a dynamicrecrystallization processes operating under decreasingtemperature. Differences in the degree of reactionoverstepping have been documented in contact aure-oles, but may be rejected in this case due to theregional nature of the metamorphism. However, therole of different degrees of undercooling relating to anoverall decrease in equilibration temperature cannot beexcluded.

Our data indicate that the sequence of rock typesreflects the progressive resorption of residual grainsand crystallization of new grains from melt in inter-granular spaces. Moreover, the trend of CSD curvessuggest a progressive increase in nucleation rate anddecrease in growth rate from type I orthogneiss totype IV nebulitic migmatite. This trend could beexplained by an increase in undercooling consistentwith the decreasing equilibration temperature wereport.

The CSD trend is compatible with crystallization ofmelt in a progressively exhuming and rapidly coolingsystem. This is in accordance with exceptionally highcooling rates up to several hundred degrees celsius permillion years estimated for nearby granulites byTajcmanova et al. (2006).

Interpretation of spatial distributions of phases

The quantitative analysis of spatial distributions ofindividual phases shows that the intensification ofregular distribution (increasing amount of unlikecontacts; Fig. 9) correlates with an increasing degree ofhost rock–melt equilibration. The process of meltcrystallization leads to new mineral growth on thesurfaces of residual grains. This is responsible for theincrease in unlike grain boundaries, which commonlyretain melt–solid geometries. Our case study showsthat the development of a regular distribution of felsicphases is not related to solid-state annealing, as sup-posed by some authors (Flinn, 1969; McLellan, 1983;Lexa et al., 2005), but to the process of crystallizationof melt, consistent with precipitation of the minorphase on triple points in granular polygonal aggregatesto achieve lower total interfacial energy (Spry, 1969;Vernon, 1974). This process was documented by Dal-lain et al. (1999), who showed that the predominanceof unlike contacts in polycrystalline aggregates origi-nated through wetting of grain boundaries by fluids ormelt, and subsequent precipitation of other phases onlike–like contacts. However, we cannot exclude thepossibility that a regular distribution reported fromgranulites and high-grade gneisses (Flinn, 1969; Kretz,1994) results from solid-state annealing of rocks wheremelt crystallized. Therefore, the regular distributiondeveloped during melt crystallization may be inheritedand perhaps further accentuated during later thermaland textural re-equilibration.

Origin of microstructural and compositional trends

The sequence from type I orthogneiss to type IVmigmatite exhibit continuous trends in all quantitativeparameters (Table 1). The grain size decreases(Fig. 7a) and there is a progressive development of aregular distribution of all felsic phases (Fig. 9), whichis linked with mineral compositional trends indicatingtemperature decrease. These clear evolutionary trendsare incompatible with a process of partial melting ofdifferent protoliths. Partial melting of the same pro-tolith may develop continuous trends, but these shouldshow increase in grain size of individual felsic phases(Dallain et al., 1999) and different mineral composi-tional evolution (e.g. Gardien et al., 1995; Greenfieldet al., 1998). Additionally, we show that the degree ofregular distribution for K-feldspar- and plagioclase-dominated aggregates evolves in the same mannerthroughout the microstructural sequence (Fig. 9).However, Dallain et al. (1999) reported significantlymore advanced regular distribution of plagioclase-compared with K-feldspar-rich aggregates in themicrostructural sequence originated by partial melting.These authors proposed that this microstructuralcontrast originated due to melting process preferen-tially operating in mica–plagioclase rich aggregates,whereas the K-feldspar-rich aggregates were more



refractory. In the present case, Hasalova et al. (2008b)report continuous trends in whole-rock geochemistryand mineral compositions for the sequence of rocktypes, but different Nd isotopic composition for thetype I orthogneiss compared with the rest of the se-quence, which precludes of in situ anatexis in a closedsystem.

Melt infiltration model

The discrepancies between the evolutionary trends wereport and generally accepted trends for anatecticterranes require an appropriate explanation that isconsistent with the structural, quantitative micro-structural and mineral compositional data. As a pos-sible explanation, we introduce the concept of meltinfiltration from an external source, where melt passespervasively along grain boundaries through the whole-rock volume and changes macroscopic (Fig. 2) andmicroscopic (Fig. 3) appearance of the rock. Thisprocess is characterized by resorption of old phases,nucleation of new phases along high-energy like–likegrain boundaries and modification of mineral andwhole-rock compositions. These gradual changes areaccompanied by grain size reduction (Fig. 7) andprogressive disintegration of former aggregate (lay-ered) distribution of original phases (Fig. 9). We sug-gest that the individual migmatite types representdifferent degrees of equilibration between the host rockand migrating melt. It should be emphasized, that allthese processes occur along a retrograde path duringexhumation of the Gfohl Unit. We are aware that adecrease in P–T conditions during melt infiltration is afundamental and limiting factor for the model pro-posed.

The amount of melt and its connectivity are criticalparameters controlling melt mobility and the rheo-logical behaviour of melt-present rocks. To constrainthese parameters both AMS and EBSD were used.Using AMS, it is possible to distinguish between solid-state dominated deformation mechanisms in themelanosome and free rigid body particle rotation in theleucosome (e.g. Ferre et al., 2003). On the other hand,using the EBSD technique enables us to distinguishdeformation mechanisms in the solid framework andto constrain the mechanical role of melt during thedeformation.

AMS fabric origin: solid framework or melt controlleddeformation

The AMS study shows that the magnetic anisotropy isdominated by biotite. The oblate shape of magneticellipsoid and high degree of anisotropy of type I or-thogneiss and type II migmatite (Fig. 10a) are consis-tent with strong preferred orientation of biotite and thefact that biotite has a intrinsically oblate shape of thesingle-grain magnetic ellipsoid (Zapletal, 1990; Martın-Hernandez & Hirth, 2003). The type III and IV mi-

gmatites reveal partly resorbed biotite flakes uniformlydispersed in the rock marked by slightly weaker degreeof magnetic anisotropy and less oblate fabric ellipsoidcompared with types I and II migmatite (Fig. 10a).This contrasts with common granites and diatexitesfrom other migmatitic terranes which show signifi-cantly lower values of degree of anisotropy and highlyvariable shapes of AMS ellipsoids (Fig. 10a; Bouchez,1997; Ferre et al., 2003).Numerous natural studies supported by numerical

modelling indicate that the magnetic susceptibility inviscously flowing magmas is characterized by a verylow degree of anisotropy, pulsatory fabrics and dom-inantly a plane strain AMS ellipsoid shape (Blumen-feld & Bouchez, 1988; Hrouda et al., 1994; Arbaretet al., 2000). A comparison of the AMS fabrics withthose of diatexites and results of numerical modelsindicate that the intensity of the AMS fabric of typesIII and IV migmatites does not originated from freelyrotated biotite in viscously flowing melt. On the con-trary, we argue that the AMS fabric in all types ofmigmatites resembles fabrics usually acquired throughsolid-state deformation of a load-bearing framework,similar to melanosomes in migmatites (Ferre et al.,2003). To understand the mechanisms responsible fordevelopment of such fabrics the grain-scale deforma-tion mechanisms and melt behaviour in individual rocktypes is discussed.

Deformation mechanisms

Experimental studies of low melt fraction rocks de-formed under high differential stress show that matrixminerals deform by grain boundary migrationaccommodated dislocation creep (Dell� Angelo et al.,1987; Walte et al., 2005). Strong shape and GBPO offeldspar (Figs 8 & 9) as well as LPO of residual quartzgrains (Fig. 11a) in the type I orthogneiss may beinterpreted in terms of plastic deformation consistentwith a dislocation creep deformation mechanism(Rosenberg & Berger, 2001). However, the weak LPOof residual grains of both feldspars (Fig. 12a, e) in thetype I orthogneiss suggests a contribution of grainboundary sliding during the development of themicrostructure. In other words, the type I micro-structure corresponds to a transient microstructure interms of decreasing activity of dislocation creep andenhancement of diffusion controlled processes.Decrease in SPO and GBPO and constantly weak

LPO in feldspar of type II migmatite (Figs 8 & 9) maybe interpreted as a result of melt-enhanced diffusioncreep (Garlick & Gromet, 2004). However, the largequartz grains reveal intense activity of basal Æaæ slipsuggesting important plastic yielding of this mineral(Fig. 11a). Elongate pockets inferred to represent for-mer melt oriented at a high angle to the stretchinglineation in the type I orthogneiss (Fig. 5a) and type IImigmatite indicate that the melt distribution wascontrolled by the deformation. This is supported by the

4 8 P . H A S A L O V A E T A L .


strong LPO of interstitial plagioclase (Fig. 12f).Rosenberg & Riller (2000) reported that pockets withinquartz aggregates inferred to have been former meltare oriented at high angle to the foliation plane, pos-sibly close to r1. Melt distribution in our samples issimilar to their results and also to experiments at highdifferential stresses (Dell� Angelo & Tullis, 1988) andhigh confining pressures. In these experiments, meltaccumulated in pockets along faces of the grains sub-parallel to the main compressional stress direction r1(Dell� Angelo & Tullis, 1988; Daines & Kohlstedt,1997). Such a melt topology is also termed �dynamicwetting� (Jin et al., 1994).

In types III and IV migmatites both residual andnew grains of K-feldspar and plagioclase exhibit lowSPO, GBPO and LPO (Figs 8, 9 & 12), indicatingabsence of dislocation creep, in contrast to quartz,which exhibits relatively strong crystallographic pre-ferred orientation (Fig. 11a). The topology of formermelt is poorly constrained in both types III and IVmigmatite but, the SPO of the minor phases inter-preted to have crystallized from melt shows a bimodaldistribution sub-perpendicular and sub-parallel to theS2 foliation. These observations are neither compatiblewith high differential stress nor low differential stressexperiments, in which the melt occurs primarily intriple point junctions without any SPO (Dell� Angeloet al., 1987; Gleason et al., 1999). However, in somenatural samples, former melt pockets are preferentiallylocated along grain boundaries parallel to the foliation(John & Stunitz, 1997; Sawyer, 1999; Rosenberg &Berger, 2001), indicating that the orientation of meltpocket in nature is not always in agreement withexperimental studies (Rosenberg, 2001). In this study,the melt pocket orientation sub-parallel to the foliationmay indicate low differential stress and high fluid/meltpressure as suggested by Cosgrove (1997).

We conclude that during evolution from type Ibanded orthogneiss to type IV nebulitic migmatite meltwetted a majority of grain contacts. The AMS studyand quartz microfabrics in types II to IV migmatitessuggest that the melt fraction did not exceed the criticalamount to allow free relative movement of grainswithout interference, i.e. the melt fraction is below thecritical threshold (e.g. RCMP of Arzi, 1978; RPT ofVigneresse et al., 1996). Rosenberg & Handy (2005)argued that melt fractions of only / ¼ 0.07 (meltconnectivity threshold, MCT) will enable the forma-tion of interconnected networks of melt under dynamicconditions which will lead to a substantial strengthdrop. These authors suggested that weakening at theMCT probably involves localized, inter- and intra-granular microcracking, as well as limited rigid bodyrotation of grains, without an important contributionof dislocation creep and diffusion processes at grainboundaries. However, we do not observe any strainlocalization associated with brittle failure and there-fore it is suggested that the deformation has to beaccommodated by mechanisms operating homoge-

neously across significant rocks volumes. Materialscience experiments (Mabuchi et al., 1997) show thatweakening due to melt-enhanced grain boundary slid-ing at low melt fraction is an efficient mechanismallowing homogeneous deformation. We suggest thatdeformation of both feldspars and quartz in the type IIto type IV migmatites occurred by melt-enhanced grainboundary sliding with a contribution to the overalldeformation by dislocation creep. These characteristicsare compatible with granular flow as described byPaterson (2001) accompanied by melt-enhanced diffu-sion and/or direct melt flow.

CONCLUSIONS

Based on a detail field and microstructural study, wedistinguish four types of gneiss/migmatite in the Gfohlgneiss complex: (i) banded orthogneiss (type I), withdistinct layers of recrystallized plagioclase, K-feldsparand quartz separated by layers of biotite; (ii) stromaticmigmatite (type II), composed of plagioclase and K-feldspar aggregates with subordinate quartz andirregular quartz aggregates – the boundaries betweenindividual aggregates are ill defined and rather diffuse;(iii) schlieren migmatite (type III), which consists ofplagioclase–quartz- and K-feldspar–quartz-enricheddomains with a foliation marked only by preferredorientation of biotite and sillimanite dispersed in therock; and, (iv) nebulitic migmatite (type IV), with norelicts of gneissosity. It is demonstrated that this is acontinuous sequence developed by melt-presentdeformation, in which the type I banded orthogneissesand type IV nebulitic migmatites are end-members.

The progressive disintegration of the bandedmicrostructure and the development of nebuliticmigmatite is characterized by several systematic tex-tural changes. The grain size of all felsic phasescontinuously decrease whereas the population densityof precipitated phases increases. The new phasespreferentially nucleate along high-energy like–likeboundaries, causing the development of a regular dis-tribution of individual phases. Simultaneously, themodal proportions of felsic phases evolve towards a�granite minimum� composition. Further, this evolu-tionary trend is accompanied by a decrease in grainSPO of all felsic phases. To explain these textural andcompositional changes we introduce a model of meltinfiltration from an external source in which melt isargued to pass pervasively along grain boundariesthrough the whole-rock volume. It is suggested that theindividual migmatite types represent different degreesof equilibration between the host rock and migratingmelt during the retrograde metamorphic evolution.

The inferred melt topology in type I orthogneissexhibits elongated pockets of melt oriented at a highangle to the compositional banding, indicating that themelt distribution was controlled by deformation thesolid framework. Here, the microstructure exhibitsfeatures compatible with a combination of dislocation



creep and grain boundary sliding deformation mech-anisms. The types II–IV microstructures developed bygranular flow accompanied by melt-enhanced diffusionand/or melt flow. However, the amount of melt presentnever exceeded a critical threshold during the defor-mation to allow free rotation of biotite grains.

The model of melt infiltration based on structuraland microstructural observation is supported by ther-modynamic (Hasalova et al., 2008a) and geochemicalmodelling (Hasalova et al., 2008b). Although our dataseem to be consistent with such a model, there are stilla number of issues to be resolved (e.g. time-scale of theprocess, the character of the melt and the grain-scaledeformation mechanisms enabling pervasive flow ofviscous melt). Nevertheless, our model has profoundconsequences for the petrogenesis of migmatites, therheology of anatectic regions during syn-orogenicexhumation and melt transport in the crust.

ACKNOWLEDGEMENTS

This work was financially supported by the GrantAgency of the Czech Academy of Science (Grants No.IAA311140 and No. 205/04/2065) by an internalCNRS funding (CGS/EOST) and by the FrenchNational Agency (No. 06-1148784). Visits by P.Hasalova to ULP Strasbourg were funded by theFrench Government Foundation (BGF). We gratefullyacknowledge A. Langrova from the Institute ofGeology at the Czech Academy of Sciences for oper-ating the microprobe. We also thank the reviewersC. Rosenberg, R. Weinberg and H. Stunitz for theirconstructive comments and suggestions for improvingthis paper, and the Journal editor M. Brown for hiscareful handling of the manuscript. R. Powell andM. Brown are thanked for helpful discussions.

REFERENCES

Adams, B. L., Wright, S. I. & Kunze, K., 1993. Orientationimaging: the emergence of a new microscopy. MetallurgicalTransactions A, 24, 819–831.

Arbaret, L., Fernandez, A., Jezek, J., Ildefonse, B., Launeau, P.& Diot, H., 2000. Analogue and numerical modeling of shapefabrics: application to strain and flow determination in mag-ma. Transactions of the Royal Society of Edinburgh: EarthSciences, 90, 97–109.

Arzi, A. A., 1978. Critical phenomena in the rheology of par-tially melted rocks. Tectonophysics, 44, 173–184.

Ashworth, J. R., 1976. Petrogenesis of migmatites in the Huntly-Portsoy area, north-east Scotland. Mineralogical Magazine,40, 661–682.

Ashworth, J. R., 1979. Comparative petrography of deformedand undeformed migmatites from the Grampian Highlands ofScotland. Geological Magazine, 116, 445–456.

Ashworth, J. R. & McLellan, E. L., 1985. Textures. In: Mi-gmatites (ed. Ashworth, J.R.), pp. 180–204, Blackie, Glasgow.

Azpiroz, M. D. & Fernandez, C., 2003. Characterization oftectono-metamorphic events using crystal size distribution(CSD) diagrams. A case study from the Acebuchesmetabasites (SW Spain). Journal of Structural Geology, 25,935–947.

Bailey, E. H. & Stevens, R. H., 1960. Selecting straining of K-feldspar and plagioclase on rock lobes and thin sections.American Mineralogist, 45, 1020–1025.

Baratoux, L., Schulmann, K., Ulrich, S. & Lexa, O., 2005.Contrasting microstructures and deformation mechanisms inmetagabbro mylonites contemporaneously deformed underdifferent temperatures (c. 650 �C and c. 750 �C). In: Defor-mation Mechanisms, Rheology and Tectonics: FromMinerals toLithosphere. (eds Gapais, D., Brun, J.P. & Cobbold, P.R.),Geological Society Special Publications, pp. 97–125, Geologi-cal Society of London, London, UK.

Barraud, J., Gardien, V., Allemand, P. & Grandjean, P., 2001.Analogue modelling of melt segregation and migration duringdeformation. Physics and Chemistry of the Earth (A), 26, 317–323.

Bascou, J., Barruol, G., Vauchez, A., Mainprice, D. & Egydio-Silva, M., 2001. EBSD-measured lattice-preferred orientationsand seismic properties of eclogites. Tectonophysics, 342, 61–80.

Berger, A. & Roselle, G., 2001. Crystallization processes inmigmatites. American Mineralogist, 86, 215–224.

Blumenfeld, P. & Bouchez, J.-L., 1988. Shear criteria and mi-gmatite deformed in the magmatic and solid states. Journal ofStructural Geology, 10, 361–372.

Bouchez, J.-L., 1997. Granite is never isotropic: an introductionto AMS studies of granitic rocks. In: Granite: From Segrega-tion of Melt to Emplacement Fabrics (eds Bouchez, J.-L.,Hutton, D.H.W. & Stephens, W.E.), pp. 95–112. Kluwer,Dortrecht.

van Breemen, O., Aftalion, M., Bowes, D. R. et al., 1982.Geochronological studies of the Bohemian Massif, Czecho-slovakia, and their significance in the evolution of CentralEurope. Transactions of the Royal Society of Edinburgh, EarthSciences, 73, 89–108.

Brown, M., 1994. The generation, segregation, ascent andemplacement of granite magma: the migmatite-to-crustally-derived granite connection in thickened orogens. Earth-Science Reviews, 36, 83–130.

Brown, M. & Rushmer, T., 2006. Evolution and Differentiation ofthe Continental Crust. Cambridge University Press, Cam-bridge, 553 pp.

Brown, M. & Solar, G. S., 1998a. Granite ascent and emplace-ment during contractional deformation in convergent orogens.Journal of Structural Geology, 20, 1365–1393.

Brown, M. & Solar, G. S., 1998b. Shear zone systems and melts:feedback relations and self-organization in orogenic belts.Journal of Structural Geology, 20, 211–227.

Brown, M., Averkin, Y., McLellan, E. L. & Sawyer, E. W., 1995.Melt segregation in migmatites. Journal of Geophysical Re-search, 100, 15655–15679.

Brown, M. A., Brown, M., Carlson, W. D. & Denison, C., 1999.Topology of syntectonic melt-flow networks in the deep crust:inferences from three-dimensional images of leucosomegeometry in migmatites. American Mineralogist, 84, 1793–1818.

Busch, W., Schneider, G. & Mehnert, K. R., 1974. Initial meltingat grain boundaries. Part II. Melting in rocks of granodioritic,quartz dioritic and tonalitic composition. Neues Jahrbuch FurMineralogie Monatshefte, 8, 345–370.

Carlson, W. D., 1989. The significance of intergranular diffusionto the mechanism and kinetics of porphyroblast crystalliza-tion. Contributions to Mineralogy and Petrology, 103, 1–24.

Cashman, K. V. & Ferry, J. M., 1988. Crystal size distribution(CSD) in rocks and the kinetics and dynamics of crystalliza-tion. 3. Metamorphic crystallization. Contributions to Miner-alogy and Petrology, 99, 401–415.

Cashman, K. V. & Marsh, B. D., 1988. Crystal size distribution(CSD) in rocks and the kinetics and dynamics of crystalliza-tion. 2. Makaopulu lava lake. Contributions to Mineralogy andPetrology, 99, 292–305.

Chandrasekhar, S., 1961. Hydrodynamic and HydromagneticStability. Oxford University Press, London.

5 0 P . H A S A L O V A E T A L .


Cmıral, M., Fitz Gerald, J. D., Faul, U. H. & Green, D. H.,1998. A close look at dihedral angles and melt geometry inolivine-basalt aggregates: a TEM study. Contributions toMineralogy and Petrology, 130, 336–345.

Collins, W. J. & Sawyer, E. W., 1996. Pervasive granitoidmagma transfer through the lower-middle crust during non-coaxial compressional deformation. Journal of MetamorphicGeology, 14, 565–579.

Connolly, J. A. D., Holness, M. B., Rubie, D. C. & Rushmer, T.,1997. Reaction-induced microcracking: an experimentalinvestigation of a mechanism for enhancing anatectic meltsextraction. Geology, 25, 591–594.

Cosgrove, J. W., 1997. The influence of mechanical anisotropyon the behaviour of the lower crust. Tectonophysics, 280,1–14.

Daines, M. J. & Kohlstedt, D. L., 1997. Influence of deformationon melt topology in peridotites. Journal of GeophysicalResearch, 102, 10257–10271.

Dallain, C., Schulmann, K. & Ledru, P., 1999. Textural evolu-tion in the transition from subsolidus annealing to meltingprocess, Velay Dome, French Massif Central. Journal ofMetamorphic Geology, 17, 61–74.

DeHoff, R. T., 1991. A geometrically general theory of diffusioncontrolled coarsening. Acta Metallica and Materialia, 39,2349–2360.

Dell� Angelo, L. N. & Tullis, J., 1988. Experimental deformationof partially melted granitic aggregates. Journal of Metamor-phic Geology, 6, 495–515.

Dell� Angelo, L. N., Tullis, J. & Yund, R. A., 1987. Transitionfrom dislocation creep to melt-enhanced diffusion creep infine-grained granitic aggregates. Tectonophysics, 139, 325–332.

DeVore, G. W., 1959. Role of minimum interfacial free energy indetermining the macroscopic features of minerals assemblages.Journal of Geology, 67, 211–227.

Dougan, T. W., 1983. Textural relations in melanosomes of se-lected specimens of migmatitic pelitic schists: implications forleucosome-generating process. Contributions to Mineralogyand Petrology, 83, 82–98.

Dudek, A., Matejovska, O. & Suk, M., 1974. Gfohl orthogneissin the Moldanubicum of Bohemia and Moravia. Krystalini-kum, 10, 67–78.

Ferre, E. C., Teyssier, C., Jackson, M., Thill, J. W. & Rainey,E. S. G., 2003. Magnetic susceptibility anisotropy: a newpetrofabric tool in migmatites. Journal of Geophysical Re-search, 108, B2, doi: 10.1029/2002JB001790

Ferre, E. C., Martın-Hernandez, F., Teyssier, C. & Jackson, M.,2004. Paramagnetic and ferromagnetic anisotropy of magneticsusceptibility in migmatites: measurements in high and lowfields and kinematic implications. Geophysical Journal Inter-national, 157, 1119–1129.

Flinn, D., 1969. Grain contacts in crystalline rocks. Lithos, 2,361–370.

Friedl, G., Finger, F., Paquette, J., Quadt, A., McNaughton, N.& Fletcher, I., 2004. Pre-Variscan geological events in theAustrian part of the Bohemian Massif deduced from U–Pbzircon ages. International Journal of Earth Sciences, 93, 802–823.

Fyfe, W. S., 1973. The granulite facies, partial melting and theArchaean crust. Royal Society of London PhilosophicalTransactions, Series, A, 273, 457–461.

Gardien, V., Thompson, A. B., Grujic, D. & Ulmer, P., 1995.Experimental melting of biotite + plagioclase + quartz +muscovite assemblage and implications for crustal melting.Journal of Geophysical Research, 100, 15581–15591.

Garlick, S. R. & Gromet, L. P., 2004. Diffusion creep and partialmelting in high temperature mylonitic gneisses, Hope Valleyshear zone, New England Appalachians, USA. Journal ofMetamorphic Geology, 22, 45–62.

Gleason, G. C., Bruce, V. & Green, H. W., 1999. Experi-mental investigation of melt topology in partially moltenquartzo-feldspathic aggregates under hydrostatic and non-

hydrostatic stress. Journal of Metamorphic Geology, 17, 705–722.

Greenfield, J. E., Clarke, G. L. & White, R. W., 1998. A se-quence of partial melting reactions at Mt. Stafford, centralAustralia. Journal of Metamorphic Geology, 16, 363–378.

Hasalova, P., Stıpska, P., Powell, R., Schulmann, K., Janousek,V. & Lexa, O., 2008a. Transforming mylonitic metagranite byopen-system interactions during melt flow. Journal of Meta-morphic Geology, doi: 10.1111/j.1525-1314.2007.00744.x

Hasalova, P., Janousek, V., Schulmann, K., Stıpska, P. & Erban,V., 2008b. From orthogneiss to migmatite: geochemicalassessment of the melt infiltration model in the Gfohl Unit(Moldanubian Zone, Bohemian Massif). Lithos, doi: 10.1016/j.lithos.2007.07.021

Hickey, K. A. & Bell, T. H., 1996. Syn-deformational graingrowth: matrix coarsening during foliation development andregional metamorphism rather than by static annealing.European Journal of Mineralogy, 8, 1351–1373.

Higgins, M. D., 1998. Origin of anorthosite by textural coars-ening: quantitative measurements of a natural sequence oftextural development. Journal of Petrology, 39, 1307–1323.

Holub, F. V., Cocherie, A. & Rossi, P., 1997. Radiometric datingof granitic rocks from the Central Bohemian Plutonic Com-plex (Czech Republic): constraints on the chronology ofthermal and tectonic events along the Moldanubian–Barran-dian boundary. Earth and Planetary Sciences, 325, 19–26.

Hrouda, F., Jelınek, V. & Hruskova, L., 1990. A package ofprograms for statistical evaluation of magnetic anisotropydata using IBM-PC computers. EOS, Transactions, AmericanGeophysical Union, 71, 1289.

Hrouda, F., Melka, R. & Schulmann, K., 1994. Periodicalchanges in fabric intensity during simple shear deformationand its implications for magnetic susceptibility anisotropy ofsedimentary and volcanic rocks. Acta Universitatis CarolinaeGeologica, 38, 37–56.

Jelınek, V., 1978. Statistical processing of anisotropy of magneticsusceptibility measured on groups of specimens. Studia Geo-physica et Geodaetica, 22, 50–62.

Jelınek, V., 1981. Characterization of magnetic fabric of rocks.Tectonophysics, 79, 63–67.

Jin, Z., Green, H. W. & Zhou, Y., 1994. Melt topology in par-tially molten mantle peridotite during ductile deformation.Nature, 372, 164–167.

John, B. E. & Stunitz, H., 1997. Evidence for magmatic frac-turing and small-scale melt segregation during plutonemplacement. In: Processes of Melt Segregation (eds Bouchez,J.-L. & Clemens, J.), pp. 55–74, Kluwer Press, Amsterdam.

Jordan, P., 1988. The rheology of polymineralic rocks: an ap-proach. Geologische Rundschau, 77, 285–294.

Jurewicz, S. R. & Watson, E. B., 1984. Distribution of partialmelt in a felsic system – the importance of surface energy.Contributions to Mineralogy and Petrology, 85, 25–29.

Kim, C.-S. & Rohrer, G. S., 2004. Geometric and crystallo-graphic characterization of WC surfaces and grain boundariesin WC–Co composites. Interface Science, 12, 19–27.

Kretz, R., 1966. Grain-size distribution for certain metamorphicminerals in relation to nucleation and growth. Journal ofGeology, 74, 147–173.

Kretz, R., 1969. On the spatial distribution of crystals in rocks.Lithos, 2, 39–66.

Kretz, R., 1994. Metamorphic Crystallization. Wiley, New York.Kroner, A., Wendt, I., Liew, T. C. et al., 1988. U–Pb zircon andSm–Nd model ages of high grade Moldanubian metasedi-ments, Bohemian Massif, Czechoslovakia. Contributions toMineralogy and Petrology, 99, 257–266.

Kruse, R. & Stunitz, H., 1999. Deformation mechanisms andphase distribution in mafic high-temperature mylonites fromthe Jotun Nappe, southern Norway. Tectonophysics, 303, 223–249.

Kruse, R., Stunitz, H. & Kunze, A., 2001. Dynamic recrystalli-zation processes in plagioclase porphyroclasts. Journal ofStructural Geology, 24, 587–589.



Lafeber, D., 1963. On the spatial distribution of fabric elementsin rock and soil fabric. Proceedings of the 4th Australia-NewZealand Conference on Soil Mechanics and Foundation Engi-neering, University of Adelaide, Adelaide, 11, 185–199.

Laporte, D. & Watson, E. B., 1995. Experimental and theoreticalconstraints on melt distribution in crustal sources: the effect ofcrystalline anisotropy on melt interconnectivity. ChemicalGeology, 124, 161–184.

Laporte, D., Rapaille, C. & Provost, A., 1997. Wetting angle,equilibrium melt geometry, and the permeability threshold ofpartially molten crustal protoliths. In: Granite: From Segre-gation of Melt to Emplacement Fabrics (eds Bouchez, J.L.,Hutton, D.H.W. & Stephens, W.E.), pp. 31–54. Kluwer,Dordrecht.

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

Lexa, O., 2003. Numerical approaches in structural and micro-structural analyses, Unpublished PhD Thesis. CharlesUniversity, Prague.

Lexa, O., Stıpska, P., Schulmann, K., Baratoux, L. & Kroner,A., 2005. Contrasting textural record of two distinct meta-morphic events of similar P–T conditions and different dura-tions. Journal of Metamorphic Geology, 23, 649–666.

Lifshitz, I. M. & Slyozov, V. V., 1961. The kinetics of precipi-tation from supersaturated solid solutions. Journal of Physicsand Chemistry of Solids, 19, 35–50.

Lister, J. R. & Kerr, R. C., 1991. Fluid-mechanical models ofcrack propagation and their application to magma-transportin dykes. Journal of Geophysical Research, 96, 10049–10077.

Lucas, S. B. & St.Onge, M. R., 1995. Syn-tectonic magmatismand the development of compositional layering, UngavaOrogen (northern Quebec, Canada). Journal of StructuralGeology, 17, 475–491.

Mabuchi, M., Jeong, H. G., Hiraga, K. & Higashi, K., 1997.Partial melting at interfaces and grain boundaries for high-strain-rate superplastic materials. Interface Science, 4, 357–368.

Marchildon, N. & Brown, M., 2001. Melt Segregation in latesyn-tectonic anatectic migmatites: an example from the Ona-wa Contact Aureole, Maine, USA. Physics and Chemistry ofthe Earth, 26, 225–229.

Marchildon, N. & Brown, M., 2002. Grain-scale melt distribu-tion in two contact aureole rocks: implications for controls onmelt localization and deformation. Journal of MetamorphicGeology, 20, 381–396.

Marchildon, N. & Brown, M., 2003. Spatial distribution of melt-bearing structures in anatectic rocks from Southern Brittany,France: implications for melt transfer at grain-to orogen-scale.Tectonophysics, 364, 215–235.

Marsh, B. D., 1988. Crystal size distribution (CSD) in rocks andthe kinetics and dynamics of crystallization; 1, Theory. Con-tributions to Mineralogy and Petrology, 99, 277–291.

Martın-Hernandez, F. & Hirth, A. M., 2003. Paramagneticanisotropy of magnetic susceptibility in biotite, muscovite andchlorite single crystals. Tectonophysics, 367, 13–28.

Matejovska, O., 1974. Petrographic Characteristic of GneissSeries between Mohelno and Bıtov, SWMoravia. MS Geofond,Prague (in Czech).

Matte, P., Maluski, H., Rajlich, P. & Franke, W., 1990. Terraneboundaries in the Bohemian Massif: Result of large-scaleVariscan shearing. Tectonophysics, 177, 151–170.

McKenzie, D., 1984. The generation and compaction of partiallymolten rock. Journal of Petrology, 25, 713–765.

McLellan, E. L., 1983. Contrasting textures in metamorphic andanatectic migmatites – an example from the Scottish Caledo-nides. Journal of Metamorphic Geology, 1, 241–262.

Mehnert, K. R., 1971. Migmatites and the Origin of GraniticRocks. Elsevier, Amsterdam.

Mehnert, K. R., Busch, W. & Schneider, G., 1973. Initial meltingat grain boundary of quartz and feldspar in gneisses and

granulites. Neues Jahrbuch Fur Mineralogie Monatshefte, 4,165–183.

Moazzen, M. & Modjarrad, M., 2005. Contact metamorphismand crystal size distribution studies in the Shivar aureole, NWIran. Geological Journal, 40, 499–517.

Olsen, T. S. & Kohlstedt, D. L., 1984. Analysis of dislocations insome naturally deformed plagioclase feldspars. Physics of TheEarth and Planetary Interiors, 11, 153–160.

Panozzo, R., 1983. The two-dimensional analysis of shape fabricusing projections of lines in a plane. Tectonophysics, 95, 279–295.

Paterson, M. S., 2001. A granular flow theory for the deforma-tion of partially molten rock. Tectonophysics, 335, 51–61.

Peterson, T. D., 1996. A refined technique for measuring crystalsize distributions in thin section. Contributions to Mineralogyand Petrology, 124, 395–405.

Petford, N., 1995. Segregation of tonalitic–trondhjemitic melts inthe continental crust: the mantle connection. Journal of Geo-physical Research, 100, 15735–15743.

Powell, R., Holland, T. & Worley, B., 1998. Calculating phasediagrams involving solid solutions via non-linear equations,with examples using THERMOCALC. Journal of Metamor-phic Geology, 16, 577–588.

Racek, M., Stıpska, P., Pitra, P., Schulmann, K. & Lexa, O.,2006. Metamorphic record of burial and exhumation of oro-genic lower and middle crust: a new tectonothermal model forthe Drosendorf window (Bohemian Massif, Austria). Miner-alogy and Petrology, 86, 221–251.

Ramberg, H., 1981. Gravity, Deformation and the Earth’s Crustin Theory, Experiments and Geological Applications. AcademicPress, London.

Randolph, A. D. & Larson, M. A., 1971. Theory of ParticleProcesses. Academic Press, New York.

Rosenberg, C., 2001. Deformation of partially molten granite: areview and comparison of experimental and natural casestudies. International Journal of Earth Sciences, 90, 60–76.

Rosenberg, C. & Berger, A., 2001. Syntectonic melt pathways ingranitic gneisses, and melt-induced transitions in deformationmechanisms. Physics and Chemistry of the Earth (A), 26, 287–293.

Rosenberg, C. L. & Handy, M. R., 2000. Syntectonic meltpathways during simple shearing of a partially molten rockanalogue (norcamphor-benzamide). Journal of GeophysicalResearch, 105, 3135–3149.

Rosenberg, C. & Handy, M. R., 2001. Mechanisms and orien-tation of melt segregation paths during pure shearing of apartially molten rock analogue (norcamphor-benzamide).Journal of Structural Geology, 23, 1917–1932.

Rosenberg, C. L. & Riller, U., 2000. Partial-melt topology instatically and dynamically recrystallized granite. Geology, 28,7–10.

Rosenberg, C. L. & Handy, M. R., 2005. Experimentaldeformation of partially melted granite revisited: implicationsfor the continental crust. Journal of Metamorphic Geology, 23,19–28.

Sakai, T. & Jonas, J. J., 1984. Dynamic recrystallization:mechanical and microstructural consideration. Acta Metal-lurgica, 32, 198–209.

Sawyer, E. W., 1994. Melt segregation in the continental crust.Geology, 22, 1019–1022.

Sawyer, E. W., 1998. Formation and evolution of granite mag-mas during crustal reworking: the significance of diatexites.Journal of Petrology, 39, 1147–1167.

Sawyer, E. W., 1999. Criteria for the recognition of partialmelting. Physics and Chemistry of the Earth, 24, 269–279.

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

Schulmann, K., Melka, R., Lobkowicz, M., Ledru, P., Lar-deaux, J. M. & Autran, A., 1994. Contrasting styles ofdeformation during progressive nappe stacking at the south-

5 2 P . H A S A L O V A E T A L .


eastern margin of the Bohemian Massif (Thaya Dome).Journal of Structural Geology, 16, 355–370.

Schulmann, K., Kroner, A., Hegner, E. et al., 2005. Chrono-logical constraints on the pre-orogenic history, burial andexhumation of deep-seated rocks along the eastern margin ofthe Variscan orogen, Bohemian Massif, Czech Republic.American Journal of Science, 305, 407–448.

Seng, H., 1936. Die Migmatitfrage und der Mechanismus para-kristalliner Pragung. Geologische Rundschau, 27, 471–492.

Spry, A. H., 1969. Metamorphic Textures. Pergamon Press,Oxford.

Stıpska, P., Schulmann, K. & Kroner, A., 2004. Vertical extru-sion and middle crustal spreading of omphacite granulite: amodel of syn-convergent exhumation (Bohemian Massif,Czech Republic). Journal of Metamorphic Geology, 22, 179–198.

Stunitz, H., 1998. Syndeformational recrystallization – dynamicor compositionally induced? Contributions to Mineralogy andPetrology, 131, 219–236.

Tajcmanova, L., Konopasek, J. & Schulmann, K., 2006. Thermalevolution of the orogenic lower crust during exhumation withina thickened Moldanubian root of the Variscan belt of CentralEurope. Journal of Metamorphic Geology, 24, 119–134.

Tanner, D. C., 1999. The scale-invariant nature of migmatitefrom the Oberpfalz, NE Bavaria and its significance for melttransport. Tectonophysics, 302, 297–305.

Turner, F. J. & Weiss, L., 1963. Structural Analysis of Meta-morphic Tectonites. McGraw Hill, New York.

Twiss, R. J., 1977. Theory and applicability of a recrystallizedgrain size paleopiezometer. Pure and Applied Geophysics, 115,227–244.

Urban, M., 1992. Kinematics of the Variscan thrusting in theEastern Moldanubicum (Bohemian Massif, Czechoslovakia):evidence from the Namest� granulite massif. Tectonophysics,201, 371–391.

Vanderhaeghe, O., 1999. Pervasive melt migration from mi-gmatites to leucogranite in the Shuswap metamorphic corecomplex, Canada: control of regional deformation. Tectono-physics, 312, 35–55.

Vernon, R. H., 1974. Controls of mylonitic compositional lay-ering during non-cataclastic ductile deformation. GeologicalMagazine, 111, 121–123.

Vernon, R. H., 1976. Metamorphic Processes: Reactions andMicrostructure Development. Wiley, New York.

Vernon, R. H. & Paterson, S. R., 2001. Axial-surface leucosomesin anatectic migmatites. Tectonophysics, 335, 183–192.

Vielzeuf, D. & Holloway, J. R., 1988. Experimental determina-tion of the fluid-absent melting relations in the pelitic system-consequences for crustal differentiation. Contributions toMineralogy and Petrology, 98, 257–276.

Vigneresse, J.-L., Barbey, P. & Cuney, M., 1996. Rheologicaltransitions during partial melting and crystallization withapplication to felsic magma segregation and transfer. Journalof Petrology, 37, 1579–1600.

Walte, N. P., Bons, P. D., Passchier, C. W. & Koehn, D., 2003.Disequilibrium melt distribution during static recrystalliza-tion. Geology, 31, 1009–1012.

Walte, N. P., Bons, P. D. & Passchier, C. W., 2005. Deformationof melt-bearing systems – insight from in situ grain-scaleanalogue experiments. Journal of Structural Geology, 27,1666–1679.

Waters, D. J. & Lovegrove, D. P., 2002. Assessing the extent ofdisequilibrium and overstepping of prograde metamorphicreactions in metapelites from the Bushveld aureole. Journal ofMetamorphic Geology, 20, 135–149.

Weinberg, R. F. & Searle, M. P., 1998. The Pangong InjectionComplex, Indian Karakoram: a case of pervasive granite flowthrough hot viscous crust. Journal of Geological Society, 155,883–891.

Wickham, S. M., 1987. The segregation and emplacement ofgranitic magmas. Journal of the Geological Society of London,144, 281–297.

Willaime, C. & Gandais, M., 1977. Electron microscope study ofplastic defects in experimentally deformed alkali feldspars.Bulletin de la Societe Francaise de Mineralogie et de Cristal-lographie, 100, 263–271.

Willaime, C., Christie, J. M. & Kovacs, M.-P., 1979. Experi-mental deformation of K-feldspar single crystals. Bulletin deMineralogie, 102, 168–177.

Zapletal, K., 1990. Low-field susceptibility anisotropy of somebiotite crystals. Physics of the Earth and Planetary Interiors,63, 85–97.

Zavada, P., Schulmann, K., Konopasek, J., Ulrich, S. & Lexa,O., 2007. Extreme ductility of feldspar aggregates – meltenhanced grain boundary sliding (GBS) and creep failure:rheological implications for felsic lower crust. Journal ofGeophysical Research, doi: 10.1029/2006JB004820.2007

Received 14 February 2006; revision accepted 25 September 2007.

SUPPLEMENTARY MATERIAL

The following material is available online at http://www.blackwell-synergy.com:

Appendix S1 Digitalized thin sections (separatedplagioclase and K-feldspar domains) that have beenused for the quantitative analyses.



Origin of Felsic Migmatites by Ductile Shearing and Melt Infiltration

Documents