Deformation and alteration of a granodiorite during low-angle normal faulting (Serifos, Greece)

Low-angle normal faults (LANFs) with fault dips of less than 30° are an important class of faults that have been described in continental crust (Axen et al., 1999; Burchfiel et al., 1992; Wernicke, 1981) and at slow-spreading mid-ocean ridges (Garcés and Gee, 2007; MacLeod et al., 2002). In continental extensional settings, LANFs are frequently associated with the exhumation of lower to mid-crustal rocks and syntectonic to late tectonic granitic plutons (Davis and Lister, 1988; Gans et al., 1989; Hill et al., 1995; Lister et al., 1984; Wernicke, 1981). Although such continental LANFs have received considerable attention, their formation remains controversial (see Axen, 2007, and references cited therein) because (1) fault mechanical theory does not predict the formation of extensional faults with orientations of less than 30° in the brittle upper crust (Anderson, 1951), and (2) evidence for large earthquakes on LANFs is rare or absent (Jackson and White, 1989; Wernicke, 1995, Collettini and Sibson, 2001). Although some reported continental LANFs may have rotated from steep to shallow dips (Buck, 1988; Koyi and Skelton, 2001), others clearly originated at low angles, especially where syntectonic igneous intrusion and associated dikes do not allow large rotations due to geometric constraints (Buick, 1991; Faure et al., 1991; Iglseder et al., 2008). LANFs that move at low dip angles require rotation of the stress field around the fault (Yin, 1989; Spencer and Chase, 1989) or low apparent fault friction, either due to weak fault-zone materials (e.g., Collettini and Holdsworth, 2004; Gueydan et al., 2004; Mehl et al., 2005; Niemeijer and Spiers, 2007; Numelin et al., 2007) and/or high fluid pressure within or adjacent to the fault zone, creating tensile fluid overpressures in the foot-wall of the LANFs (Reynolds and Lister, 1987).

Changes in element chemistry and stable isotopes are clear evidence for strong fluid flow in many shear zones (Kerrich et al., 1980; Rolland et al., 2003; Hürzeler and Abart, 2008). Several studies have demonstrated that both brittle faults and ductile shear zones are preferential conduits for fluid flow parallel to the fault but may also act as a barrier perpendicular to the fault or shear zone (McCaig, 1987; Kisters et al., 2000; Sibson, 2000; Micarelli et al., 2003).

In this work, we describe the geochemical, petrological, and structural development of a major LANF that evolved close to the roof of an upper crustal granodiorite intrusion that is part of the Miocene Serifos metamorphic core complex (Western Cyclades, Greece). In the studied location, the LANF developed exclusively in the granodiorite; the limited offset along the fault has led to the special situation that the foot-wall, consisting of granodiorite with increasing magnitude of ductile deformation toward the fault, is juxtaposed along a knife-sharp cataclastic fault against strongly “bleached” granodiorite in the hanging wall. Therefore, the source rock before deformation was, at the scale of the outcrop, a relatively homogeneous granodiorite intrusion with a clearly defined geochemical composition. This makes it ideally suited for studying the processes and effects of deformation on the protolithic microstructure and geochemistry.

Serifos, located ~100 km SSE of Athens, geologically belongs to the Attic-Cycladic massif. The island is dominated by a late Miocene, high-level I-type granodiorite pluton (Ballindas, 1906; Marinos, 1951) that intruded into orthogneisses, amphibolites, greenschists, and marbles (Fig. 1). Intrusion ages of the main granodiorite body and its associated dikes lie between 11.6 and 9.5 Ma (Iglseder et al., 2008) followed by rapid cooling (Altherr et al., 1982; Henjes-Kunst et al., 1988). Apatite fission-track ages from the central parts of the granodiorite plot between 6.7 ± 0.8 and 5.3 ± 0.6 Ma (Hejl et al., 2002). The granodiorite intruded into upper crustal levels (Stouraiti and Mitropoulos, 1999) during ongoing extension. The main granodiorite body discordantly crosscuts the regional foliation of the host rocks and is essentially undeformed or slightly foliated. However, the roof of the granodiorite intrusion, associated dikes, and the country rocks have been deformed by a greenschist-facies, brittle-ductile to brittle LANF during cooling and exhumation. The LANF arches over the island and consistently records a top-to-SSW–directed shear sense. The formation of the LANF has been interpreted as the result of Miocene crustal-scale extension and metamorphic core complex formation (Grasemann and Petrakakis, 2007; Iglseder et al., 2008), similar, although with opposing kinematics, to other islands in the Cyclades (e.g., Avigad and Garfunkel, 1989; Forster and Lister, 1999; Gautier and Brun, 1994; Jolivet et al., 2004; Lister et al., 1984).

At Platis Yialos (ΠλατύςΓιαλός) in the N, at Kavos Kiklopas (ΚάβοςΚύκλωπας) in the SW, and at Tsilipaki (Τσιλιπάκι) in the SE (Fig. 1), excellent exposures of the LANF are preserved. These outcrops have the following characteristics in common (Grasemann and Petrakakis, 2007): (1) The footwall below the low-angle shear zone consists of gneisses, greenschists with marble layers, which record an earlier deformation and metamorphic history, probably related to an earlier high-pressure event typical for the Attic-Cycladic massif (Salemink and Schuiling, 1987). The dominant lineation trends ENE-WSW. (2) Toward the low-angle fault, there is an increase in noncoaxial strain with a NNE-SSW–trending stretching lineation recording top-to-SSW–directed shear. (3) Below the major fault plane, there are several meters of ultrafine-grained (<10 μm), ultramylonitic marbles with top-to-SSW shear-sense indicators. (4) The LANF is marked by a knife-sharp plane associated with polyphase ultracataclasites recording top-to-SSW–directed shear. (5) The greenschists, gneisses, and marbles above the fault are strongly overprinted by protocataclastic deformation and ankeritized by massive fluid infiltration. (6) The LANF predates, interacts with, and postdates sets of WNW-ESE–striking, conjugate high-angle normal faults indicating a localization of the LANF during ongoing NNE-SSW extension of the crust under subvertical maximum principal stresses.

The transition from ductile to brittle deformation mechanism suggests that the low-angle fault operated at the brittle-ductile transition zone. This is supported by the 12–8 km granodiorite intrusion depth, which was emplaced during extensional movement of the fault (Stouraiti and Mitropoulos, 1999). Furthermore, quartz grains in the marble mylonites record microstructural evidence for dislocation glide and brittle fracturing but no evidence for dislocation creep. This transition is inferred to be at temperatures of ~280 °C for rocks deforming at strain rates within the commonly inferred range of 10⁻¹¹ s⁻¹ to 10⁻¹⁴ s⁻¹ (Stipp et al., 2002).

In the SE part of Serifos, N of Tsilipaki, the LANF cuts through the granodiorite pluton at the bay of Aghios Sostis (Αγιος Σώστης), recording progressive deformation of the undeformed granodiorite at lower structural levels to mylonitically deformed granodiorite within the shear zone. Due to the relatively homogeneous lithology of the plutonic body, the geometry, kinematics, and associated chemical and mineralogical changes of the shear zone, which are the focus of this work, could be studied in detail.

Structural field observations are shown in Figures 2, 3, and 4.

Figure 2 shows a panoramic view toward the SE, onto the headland at the bay of Aghios Sostis. Here, a complete section of the undeformed or weakly foliated granodiorite (W of the bay), the strongly foliated granodiorite (lower structural levels, E of the bay) to the mylonitized granodiorite (central part of the picture), has been exposed. Above the mylonitic granodiorite, which represents the LANF, the precursor granodiorite is completely metasomatized by infiltrating fluids above a knife-sharp brittle fault (eastern end and topographic top of the headland). Following the designation of Salemink (1985), this altered type of rock will be called bleached granodiorite in the subsequent text. The main foliation (sm) generally dips toward SE (Fig. 3A). The LANF indicated in Figure 2 continues to the point where the panorama picture was taken. The structural distance between the investigated footwall samples and the LANF decreases with the order given in the tables, while at the same time the intensity of the foliation increases. Sample SB8-07 is located in the hanging wall of the LANF (Fig. 2).

In the following sections, the main structural characteristics of the rocks are described, starting from the undeformed granodiorite in the footwall, passing up into the mylonitic granodiorite, and thence to the bleached granodiorite at the top, in the hanging wall.

The protogranodiorite SB1-07 at the lowermost structural level is macroscopically essentially undeformed; locally a vague foliation defined by preferred orientation of biotite can be discerned (Fig. 3A). The rocks are cut by three different joint systems (Fig. 4B). The oldest system (sjf₁) consists of isolated joints that dip moderately steeply toward SE. These joints are extremely difficult to recognize in the undeformed granodiorite and have been only occasionally observed. No bleaching or any evidence of diffusive mass transfer has been observed along these joints. Because sjf₁ occurs in the undeformed granodiorite but locally also in the foliated granodiorite, where the joints were reactivated as ductile shear zones, this joint system formed during shearing along the LANF and cooling and exhumation of the pluton. Two, almost vertical, joint systems (sjf₂ and sjf₃) are striking roughly WNW-ESE and NE-SW, respectively. Both systems form dense networks of a few cm- to several m-spaced joints, which are partly filled with chlorite. The joints are almost invariably surrounded by bleached haloes, generally only a few centimeters thick and symmetrically developed to either side of the joint. In the outcrop section, they are segmented, with lengths of several tens of centimeters up to meters. The segments have en echelon geometry with a slight overlap at their terminations (Fig. 3B), but no systematic left- or right-stepping pattern has been observed (Segall and Pollard, 1980).

Toward higher structural levels, there is a progressive transition from “undeformed” granodiorite to macroscopically, weakly foliated granodiorite SB4-07 and into a clearly foliated granodiorite SB6-07 (Fig. 2). With increasing strain toward higher structural levels, isolated ductile shear zones (szf), localized on brittle precursor joints sjf₁ (Mancktelow and Pennacchioni, 2005), formed a network of shear zones separating mylonitic granodiorite from less strongly deformed granodiorite (Fig. 3B; compare also Figs. 4A and 4B). Note, however, that locally observed deflection and/or offset of sm along sjf₁ suggest that localization of strain along sjf₁ occurred after the formation of an initial main foliation (sm). The initial shear zones, recording an offset of a few millimeters to centimeters, are sharp planes with extremely localized shear deformation. However, with increasing offset, the shear zones broadened, forming an intense mylonitic foliation and transform toward higher structural levels into homogeneously deformed (ultra)mylonites without any evidences of further localization of deformation. Along sections on the NE side of Aghios Sostis, the transition from weakly deformed to mylonitic granodiorite is more gradational, and no strain localization along shear zones has been observed. Two joint systems sjf₂ and sjf₃ crosscut the main foliation (sm) and clearly post-date the ductile (mylonitic) deformation. Neither system was reactivated by ductile shear, and both record the same characteristics (orientation, en echelon geometry, spacing, and bleaching; Fig. 3D) as the corresponding systems in the undeformed granodiorite.

The mylonitic granodiorite SB2-07 and the ultramylonite SB3-07, which gradually form the highest structural level below the bleached granodiorite of the hanging wall, have a structural thickness of several meters, with a strong foliation and stretching lineation. Similar to the less deformed and undeformed granodiorite below, the rocks are cut by joint systems sjf₂ and sjf₃, both with bleached halos.

Above the mylonitic granodiorite, a knife-sharp, brittle, low-angle fault dips gently eastwards and separates the footwall mylonites from weakly foliated, totally bleached, hanging-wall granodiorite SB8-07 (Figs. 2 and 3E). This fault corresponds and links with the LANF mapped to the S, on Tsilipaki (Fig. 1). Although dm-thick lenses of non-cohesive ultracataclasites and fault gouge have been observed along the fault, the contact is in places a sharp, brittle fault. Isolated, several tens of meters long, and up to several centimeters thick localized ductile shear zones (szh) with consistent SW-directed kinematics have been observed, mainly on the S side of the headland of Aghios Sostis (Fig. 4A). These shear zones clearly have a different orientation than the localized shear zones in the granodiorite (szf). The bleached granodiorite and the shear zones are cut by a decimeter- to meter-spaced, subvertical joint system striking roughly both N-S (sjh₁) and W-E (sjh₂; Fig. 4B). This joint system is restricted to the bleached granodiorite in the hanging wall and has not been observed in the footwall granodiorite.

All lithologies and structures are cut by brittle normal faults associated with slickensides and pseudotachylites (Figs. 3F and 4C). The faults have a normal offset in the order of several meters and dip steeply to the NE or SW (Fig. 4C). A major, high-angle normal fault, which separates the NNE part of the headland from the rest of Aghios Sostis (see Fig. 2) dips NNE and juxtaposes bleached hanging-wall granodiorites to the NNE against mylonitic footwall granodiorites to the SSW. As a result, the LANF that is exposed in the footwall at a maximum altitude of 20 m above sea level has been downthrown below sea level in the hanging wall.

Microstructural features of all investigated samples are summarized in Figure 5,⁵⁰², and representative mineral chemistries of potassium feldspars, plagioclases, as well as biotites, are listed in Tables 1–3.

In thin sections, the lowest granodioritic sample collected from the footwall shows a typical protogranular microstructure with quartz grains, perthitic K-feldspars, and plagioclases >500 μm (Fig. 4A). Accessorial mineral phases are represented by biotite (>400 μm) and hornblende as well as mostly idiomorphic magnetite, Ti-magnetite, apatite, titanite, magmatic epidote, and zircon. K-feldspars have orthoclase composition (Or₉₄) with BaO contents of 0.45 wt%, often showing sutured grain boundaries with plagioclases (see Table 1). The plagioclases show compositional magmatic zonation from anorthitic (An₆₁ Ab₃₈ Or₁) cores to albitic (An₄₁ Ab₅₈ Or₁) rims (labradorite to andesine, respectively; Table 2). Biotite clasts in SB1-07 are characterized by Mg# of 0.45 and TiO₂ contents of 4.35 wt% (see Table 3); amphiboles are typical magnesio-hornblende (TiO₂: 1.22 wt%, Al₂O₃: 7.3 wt%, Fe₂O₃: 3.3 wt%, FeO: 11.4 wt%, MgO: 13.6 wt%, CaO: 11.8 wt%; see  Appendix A for analytical techniques). Generally, the quartz grains are strain free or have a patchy undulatory extinction. Feldspars are locally fractured, and biotite packages are frequently kinked.

The microstructures of the foliated to mylonitized granodiorite record the following characteristics. In the slightly foliated granodiorite, a considerable grain-size reduction of all phases has been observed (Fig. 5B). Dynamically recrystallized quartz grains partly form elongated ribbons that define the foliation. The recrystallized grains record a strong shape and lattice preferred orientation consistent with the overall SSW-directed shear. Igneous perthitic K-feldspar relics remain as large clasts, whereas orthoclase partly recrystallized as interstitial grains within the main foliation. Neither clasts nor newly formed orthoclase show any compositional changes compared to the protorock (see Table 1). Fractured plagioclase grains rotated during noncoaxial flow, but their magmatic zonation, from An₆₀ Ab₃₉ Or₁ to An₃₇ Ab₆₁ Or₂, has been preserved. Biotite clasts do not indicate chemical alteration.

Structurally upwards, the foliated granodiorite shows further noticeable grain-size reduction (Fig. 5C). The foliation consists of fine-grained, elongated, dynamically recrystallized quartz grains deformed by subgrain rotation (Regime 2 of Hirth and Tullis, 1992) and elongated aggregates of recrystallized K-feldspar showing no compositional variations. Relic large perthitic K-feldspar do not record any preferred orientation of Ab lamellae nor do they show chemical alteration (see Table 1). Fragmented, fractured, tilted, and boudinaged plagioclase grains show interstitial recrystallization of K-feldspar and biotite, with preferred growth of these minerals in the strain shadow of plagioclase (see Fig. 5G). Plagioclase porphyroclasts have preserved a magmatic zonation but indicate subgrain rotation and dynamic recrystallization to andesine (An₄₄ Ab₅₄ Or₂; see Table 2). Recrystallized biotites indicate an increasing amount of FeO, from 16.7 wt% in biotite grains in the protogranodiorite to 17.3 wt% in recrystallized biotite along the foliation, with a simultaneous decrease of TiO₂ from 4.35 wt% to 3.14 wt% (see Table 3). This grain downsizing and lowering of Ti in biotites at increasing deformation has been described elsewhere (e.g., Kerrich et al., 1980).

In thin sections, the mylonitic granodiorite indicates strongly elongated quartz layers, residual perthitic K-feldspar phenocrysts in thin recrystallized K-feldspar layers and plagioclase porphyroclasts, connected with interstitially growing K-feldspar, albite, biotite, and ore minerals, together forming the foliation (Fig. 5D). The mineral composition of K-feldspar clasts has not changed compared to that in the protogranodiorite; recrystallized K-feldspar grains also do not show any chemical alteration (Table 1). Although plagioclases apparently lost the magmatic zoning, mainly through grain size reduction, the compositions do not alter. Strain shadows of plagioclase clasts or necks of boudinaged plagioclases consist of dynamically recrystallized K-feldspar (Fig. 5G) or albitic subgrains of the porphyroclast, often intergrown with recrystallized K-feldspar and biotite (see Fig. 5H and Tables 1–3).

At the top of the footwall, the ultramylonitic sample consists of quartz ribbons, layers with extremely fine recrystallized albitic plagioclase (<10 μm) intergrown with quartz, K-feldspar, and biotite, as well as layers consisting of mainly K-feldspar with minor amounts of biotite (Fig. 5E). In the bright layers in Figure 5E, consisting mainly of recrystallized K-feldspar and biotite, the minerals have similar compositions to those in the less deformed granodiorite (see Tables 1 and 3). The dark layers in Figure 5E are mainly composed of recrystallized andesine (An₄₁ Ab₅₈ Or₁), quartz, and biotite neoblast flakes. Intergrown K-feldspar blasts that are related to this reaction matrix have orthoclase compositions (Or₉₅) with exceedingly high BaO contents (0.53 wt%) compared to the other ribbon or interstitially recrystallized grains of K-feldspar (see Table 1). Plagioclase phenocrysts up to 200 μm in size are homogeneously dispersed. A plethora of unequivocal shear-sense criteria (Fig. 5E) such as shear bands, S-C-C fabrics, σ- and δ-Fsp clasts, and shape- and lattice-preferred orientations of quartz (see Passchier and Trouw, 2005 and references cited therein) suggests a high component of noncoaxial deformation.

Microscopically, the bleached granodiorite in the hanging wall shows a weak foliation defined by an interconnected weak layer of dynamically recrystallized quartz (Fig. 5F). A few old, larger grains have patchy undulatory extinction and irregular grain boundaries. New grains form aggregates of small, equally sized grains and developed at the expense of old grains by low-temperature grain boundary migration. Oligoclase (An₂₂ Ab₇₅ Or₃) grains have been strongly altered through fluid interaction and sericitized. They have lost any magmatic zonation and commonly show extremely fine grained (<10 μm) subgrain formation at mineral boundaries or along intracrystalline fractures. K-feldspars (Or₉₃) show flame-like perthitic morphologies. The mylonitized rocks within the localized shear zones record an extreme grain size reduction, mainly by fracturing of the feldspars. Quartz was deformed by dislocation glide and bulging of grain boundaries (Regime 1 of Hirth and Tullis, 1992). Shear-sense criteria σ- and δ-feldspar clasts, and shape- and lattice-preferred orientations of quartz, indicate top-to-SW–directed kinematics.

Despite increasing deformation from the protogranodiorite to the ultra-mylonitic state, the mineral chemistry of K-feldspar, even in a recrystallized nonperthitic state, has not systematically changed (see Table 1 and Fig. 6G). Only BaO shows a slight but progressive depletion in the porphyroclasts at increasing strains. Despite having the lowest BaO values in K-feldspar clasts, the ultramylonite shows the highest concentrations of barium in K-feldspar grains within the Pl-Qtz-Kfs-Bt aggregates. The composition of magmatic plagioclase has also not been significantly altered (see Table 2 and Fig. 6H). Subgrains of recrystallized plagioclase in the strain shadow of plagioclase porphyroclasts or in necks of boudinaged plagioclases consistently show andesine composition, very similar to the rim composition of magmatic plagioclases. Biotite recrystallization into fine flakes is obvious from SB6-07 to SB3-07 reducing the Mg# from 0.45 to 0.42–0.43 wt% and TiO₂ from 4.32–4.35 wt% to 3.14–3.40 wt% (see Fig. 5F). Except along the bleached haloes adjacent to sjf₂ and sjf₃, which clearly formed under brittle conditions after ductile mylonitic deformation, no evidence for chloritization of biotite or sericitization of feldspars has been observed.

Major, trace, and rare-earth element (REE) compositions are summarized in Table 4, arranged by increasing state of deformation. The data are illustrated in Figures 6A and 6B. Bulk compositional comparisons of the essentially undeformed magmatic protogranodiorite with the progressively strained rock sequence are shown in Figures 6C–6E.

Major element compositions indicate that all the samples are subalkaline (calc-alkaline) granitoid rocks (cf. Iglseder et al., 2008). Despite the increasing deformation, the rocks hosting the shear zone have been classified as granodiorites, except for the ultramylonitic sample SB3-07 and the bleached granodiorite SB8-07, which, due to an increased potassium and decreased calcium concentration, have a granite composition. The same classification has been obtained from mass-balance calculations that were used to estimate the mineral proportions of the samples (see Table 5).

Table 4 and Figure 6 indicate that the variations of major element contents along the sampled traverse through the shear zone, from the protogranodiorite to the ultramylonite, are minor for SiO₂, Al₂O₃, TiO₂, and Fe₂O₃, while the CaO, Na₂O, and K₂O amounts show significant differences. Figure 6A illustrates the similarities of the samples to continental arc granitoids. The hanging-wall sample reflects a highly altered composition with respect to the footwall samples with a much higher content of SiO₂, compensated by lower Al₂O₃ and CaO. This reflects the lack of Fe- and Ti-bearing and mafic minerals as well as the modal decrease of plagioclase. To investigate sequential changes in major element geochemistry during progressive deformation, the analyses were normalized by the precursor granodiorite composition (SB1-07; Fig. 6C). For MgO, this shows an essentially constant relative depletion of −30% to −40% for all samples. A gradual depletion with progressive strain has been observed only for CaO. For the composition of the ultramylonite, the relative losses of Fe₂O₃ (−38%), CaO (−59%), and Na₂O (−22%) are remarkable, as is the more than counterbalancing gain of K₂O. The loss of CaO and Na₂O, as well as the gain of K₂O, respectively, is reflected by the modal mineral proportions in the ultramylonite, which indicate increased K-feldspar contents from 14% to 40%, with decreased plagioclase contents from 40% to 21%, compared to the protorock (see Table 5). The loss of Fe₂O₃ is consistent with decreasing amounts of modal accessorial Fe-bearing minerals.

Upper continental crust (UCC; McLennan, 2001) normalized REE patterns and protogranodiorite normalized patterns are illustrated in Figures 6B and 5E, respectively. The analyzed REEs show patterns very similar to the composition of the UCC. Both light rare-earth elements (LREEs; La-Sm) and heavy rare-earth elements (HREEs; Eu-Lu) show relatively flat trends (La_n/Sm_n for SB3-07 is 1.26; Gd_n/Yb_n for SB3-07 is 1.14). The HREE patterns, however, indicate a slight enrichment (~1.2 × UCC for Lu) for SB1-07 to SB6-07, whereas the mylonitic samples SB2-07 and SB3-07 show a slight depletion (~0.8 × UCC for Lu) with respect to the UCC composition. The REE pattern of SB8-07 is generally depleted compared to the UCC but mainly in the LREE (~0.3 × UCC for La, 0.5 × UCC for Lu). The Eu anomaly, however, points to a modal depletion of plagioclase. Comparing the deformed rock sequence to the REE composition of the protogranodiorite (Fig. 6E), the LREE patterns are hard to differentiate, whereas the HREEs show two clearly separate patterns, with a slight enrichment in SB4-07 and SB6-07 (average enrichment of 4.5% from Eu to Lu) and a depletion in SB2-07 and SB3-07 (average depletion of 26% from Eu to Lu). La to Nd contents of the ultramylonite SB3-07 suffered a significant enrichment of these elements in contrast to the depleted HREE content.

In the normalized trace-element plot (Fig. 6D), the relative gain of large ion lithophile elements (LILE) Rb (+77%) and Ba (+147%) in the ultramylonite is associated with the marked increase of potassium, whereas the trends of other LILEs, Cs, Th, and U, are variable, and Sr is quite stable. In general, the transition trace elements (TTE) Sc–Zn show a tendency of depletion with increasing deformation, similar to Fe₂O₃, MnO, and MgO. The differences, however, are not that significant. The amounts of high field strength elements (HFSE) Y–Nb are relatively constant during progressing strain.

To compare the whole-rock data of the progressively deformed rocks to the initial nondeformed granodiorite composition, and to study the compositional changes that resulted due to the shearing, isocon diagrams were plotted (Gresens, 1967; Grant, 1986, 2005). In addition to the analyzed rock densities, changes of single elements as well as whole-rock mass and volume changes were calculated. Elements that remained immobile during deformation were identified mathematically and graphically following the procedure of Grant (2005). SiO₂ and Al₂O₃ varied least during deformation, reflecting their immobile behavior and were, therefore, chosen as the isocon in all plots. In addition to the immobility of Al, which has been confirmed elsewhere, Ti, P, and Zr, which have been shown to be immobile in shear zones, show good linear correlations with the isocons (Kerrich et al., 1980; Hippertt, 1998, and references therein; Sturm and Steyrer, 2003; Rolland et al., 2003, and references therein). Because many elements and oxides have concentrations in the same range, the gained data were rescaled to plot clearly (Grant, 2005). Whole-rock mass changes, as well as whole-rock volume changes were calculated taking the considerations of Coelho (2006) into account. Figure 7 illustrates the isocon diagrams, comparing the composition of the protolith (C⁰) to the chemical composition of the four progressing states of deformation (C^D). Elements plotting above these defined reference lines were therefore gained during the deformation, whereas elements plotting below it indicate depletion.

The tight scattering of elements around the defined isocon in Figures 7A and 7B reveals relatively immobile elements participated in the deformation process, as was also shown in Figures 6C–6E. Rare-earth elements, which are considered to be mobile in shear zones (Rolland et al., 2003), also show perfect correlations with the defined isocon. Lines of constant mass and constant volume hardly show differences in slope to the isocons, pinpointing a more or less constant mass and volume deformation at this state. This comparability of isocon-based element gains and/or losses and relative abundances of elements under assumed constant mass and volume substantiates the significance of the element patterns in Figure 6. Whole-rock mass losses of 1% and volume gains of 0.5% were assessed for SB4-07 and SB6-07.

In the mylonitic sample (SB2-07), the scattering of elements is still minor, but a linear correlation and systematic slight depletion of REE has been observed, with the exception of La, Ce, and Nd, which indicate slight enrichment (Fig. 7C). CaO and MgO indicate depletion, which has also been observed in all other states of deformation. A whole-rock mass loss of 1.8% and volume gain of 1.7%, respectively, have been calculated for this sample. In the ultramylonitic state (Fig. 7D), a fractionation of LREE and HREE has clearly occurred, in which mainly La and Ce, lying on a linear correlation through the origin, have been considerably enriched. Pr and Nd also plot above the isocon, whereas for the rare-earth elements Sm to Lu, depletion is evident. Under assumed constant mass and volume conditions, K, Rb, Ba, and U indicate major gains in the ultramylonite, whereas Fe, Mn, Ca, Na, and the transition metals are reflected by systematic losses. A whole-rock mass loss of 1.9% as well as a volume change of +1.9% can be assessed.

The isocons are all characterized by a slope very close to one, indicating that the chosen immobile elements as well as all other elements plotting along this reference frame kept almost constant concentrations despite the increasing finite strain. Therefore, the graphically obtained gains and losses of elements are nearly equal to the calculated ones under assumed constant mass and volume conditions. Nevertheless, the samples reveal a slight decrease of mass with increasing strain at simultaneous slight increase of volume.

In spite of the increasing finite ductile strain toward higher structural levels, samples from the undeformed to the mylonitic granodiorite in the footwall of the LANF remain compositionally relatively homogeneous. In the ultramylonitic sample, however, some major and trace elements indicate bulk chemical alteration with respect to the protorock. The most significant changes are the increasing K₂O, Rb, and Ba at simultaneous decreasing CaO and Na₂O, as well as selective gains of LREEs and losses of HREEs (Table 4; Figs. 6C–6E and 7D). Mass-transfer mechanisms causing such heterogeneities are likely controlled by deformation and/or the flow of both externally derived infiltrating and internal fluids. However, due to the absence of vein development in the shear zone, nonpervasive fluid flow can be a priori excluded (McCaig and Knipe, 1990).

The highest deformation, in the ultramylonite, served as the driving force for the observed lateral mass transfer within the shear zone (McCaig and Knipe, 1990). This transfer caused a spatially limited geochemical redistribution of mineral phases and associated elements within the ultra-mylonite that resulted in small-scale geochemical anomalies. Dissolution, transfer, and precipitation mechanisms, mainly affecting feldspars, occurred in reaction domains of quartz, plagioclase, K-feldspar, and biotite that have only been observed in the ultramylonite. Element redistribution was, therefore, caused by breakdown reactions of K-feldspar clasts and plagioclases.

Lateral transport and accumulation of K-feldspar constituents in the ultramylonite resulted in a spatially heterogeneous recrystallization giving the present-day altered assemblage. Similar strain-triggered, mass-transfer processes involving feldspars were described by Ishii et al. (2007) and Menegon et al. (2008). Further, a non–mass-conservative exchange of Na and Ca with K in granitoid rocks was mentioned by Hippertt (1998). The increase in modal K-feldspar from 14% in the protorock to almost 40% in the ultramylonite, together with the loss of modal plagioclase from 40% to 21%, as well as the loss of biotite and Fe-bearing accessorial minerals (Table 5), was, therefore, a local ultramylonitic deformation-induced heterogenization. K-feldspar compositions from the reaction matrix indicate that it was not a primary recrystallization product, as at lower strains, but grew from additional Ba and K, forming a second generation of recrystallized K-feldspars that potentially increased the bulk mode. The apparent fractionated REE patterns were caused mainly by fluid-composition–dependent alteration, dissolution, and precipitation of magmatic REE-bearing minerals (Rolland et al., 2003), also induced by local fluid flow during ultramylonitic deformation.

Despite change in the mineralogy and associated compositional changes, the samples only record marginal mass and volume changes from protogranodiorite to ultramylonite (see Fig. 7D). The minor bulk geochemical changes are reflected in microstructural and mineral chemical observations. Plagioclase clasts maintained their initial labradorite core composition until ultramylonitization, despite intense fracturing and dismantling. Dynamically recrystallized plagioclases have the same chemistry as the andesine rim composition of magmatically zoned minerals (see Table 2 and Figures 5G–5I and 6H). Apart from a slight systematic decrease of BaO, K-feldspars remained compositionally homogeneous; the recrystallized equivalents that appear first in the slightly foliated granodiorite do not indicate any significant variations (see Table 1 and Fig. 5G). Recrystallized K-feldspar that grew within the reaction matrix of the ultramylonite appears to vary slightly geochemically in terms of composition and accounts for the modal enrichment of K-feldspar compared to the protogranodiorite. Biotite grains recrystallized with progressive deformation and show an expected Ti decrease (Table 3 and Fig. 6F).

In summary, the results suggest that, although increasing strains led to progressive microstructural changes, these did not influence the bulk geochemical compositions. In the ultramylonitic granodiorite, however, iso-choric chemical mass transfer caused changes in the reaction mechanisms, resulting in the observed heterogeneity in the mineral mode, although these did not greatly affect either the bulk mass or volume.

Since ductile shear zones contain important information about the deformation history of an area, numerous studies have addressed the question of how strain became localized in essentially homogeneous and isotropic materials. Localization has been successfully modeled, either by applying damage rheologies (e.g., Sleep, 2002; Simakin and Ghassemi, 2005) or by using rheologies that combine the effects of shear heating, elasticity, diffusion creep, dislocation creep, and low-temperature plasticity (e.g., Regenauer-Lieb and Yuen, 2003; Kaus and Podladchikov, 2006). Furthermore, rheologically controlled localization of shear zones has been modeled to occur at preexisting discontinuities (e.g., Mancktelow, 2002; Mandal et al., 2004). Thus, although some authors suggest that shear zones are unrelated to preexisting heterogeneities and have propagated into the undeformed host rock from discrete localization sites (Poirier, 1980; White et al., 1980; Ingles et al., 1999), other studies present clear field evidence for nucleation of ductile shear zones on preexisting brittle joints or fractures (Segall and Pollard, 1983; Takagi et al., 2000; Mancktelow and Pennacchioni, 2005; Pennacchioni et al., 2006; Pennacchioni and Mancktelow, 2007).

Structural field work at Aghios Sostis showed a diffuse deformation gradient, with the intensity of the bulk foliation increasing from an essentially undeformed granodiorite to a penetratively foliated mylonite directly below the LANF. Sibson (1977) suggested that a gradual decrease of strain and foliation into the footwall block is typical for normal faults, and this has been described from many crustal-scale extensional faults (e.g., Herren, 1987; Selverstone, 1988; Mancktelow, 1992). On Serifos, this zone of diffuse increase in foliation intensity toward the LANF has a structural thickness of ~20–30 m. Progressive development of monoclinic microstructures (i.e., shear-sense criteria) with increasing deformation clearly link this foliation to SSW-directed shearing along the LANF. However, isolated and localized shear zones also developed in the strongly foliated granodiorite, with a heterogeneous distribution. These shear zones nucleated on brittle precursor fractures that have the same orientation as the joint system sjf₁ in the undeformed granodiorite. With increasing strain, these shear zones offset the foliated granodiorite with a strong displacement gradient, resulting in a deflection of the foliation (Fig. 3B) and change of normal and reverse drag from the center to the tip of the shear zone (Grasemann et al., 2005). This is an important observation because the brittle fractures cannot be related to cooling of the pluton or thermal contraction and differential solidification of magmatic bodies (Bergbauer and Martel, 1999) since they offset the strongly foliated granodiorite and therefore formed during solid-state deformation. Brittle deformation followed by ductile flow has been observed not only under low-grade metamorphic conditions but also under amphibolite-, granulite-, and eclogite-facies conditions (e.g., Austrheim et al., 1996; Kisters et al., 2000; Mancktelow and Pennacchioni, 2005). The transition from brittle to viscous behavior in rocks can be achieved by an increase in temperature or a change in confining and fluid pressure (Tullis and Yund, 1977; Kisters et al., 2000; Kolb et al., 2004). However, the Serifos pluton was intruded into shallow crustal levels (Stouraiti and Mitropoulos, 1999) and experienced rapid cooling (Altherr et al., 1982; Henjes-Kunst et al., 1988; Hejl et al., 2002; Iglseder et al., 2008). Therefore, cooling followed by reheating can be excluded. Furthermore, our geochemical investigations exclude a major component of rock-fluid interaction and/or mass and volume change during deformation in the footwall of the LANF. Note that fluid-rock interactions played a major role in skarn mineralization (Salemink and Schuiling, 1987), and the hanging wall of the LANF on Serifos, like the bleached granodiorite, has been strongly altered by hydrothermal fluid flow. Additionally, protocataclastic marbles and calc-rich schists above the LANFs at Platis Yialos and Kavos Kiklopas were strongly altered by ankeritic fluids. However, ductile shearing of the granodiorite and localization of shear zones on brittle precursors occurred with almost no bulk geochemical changes. This has been also described from other ductilely deformed plutons (Pennacchioni, 2005). The absence of fluid flow into the ductile shear zones in the footwall of the LANF and the strong evidence for a massive fluid-driven alteration of brittle and cataclastic faulting cutting the hanging wall and the footwall after movement of the LANF probably directly reflects the fundamental difference between brittle and ductile shear zones. The mean stress in brittle fault zones is lower compared to the surrounding rocks, but the pressure in viscous shear zones is higher than in the surrounding material (Mancktelow, 2006).

Neither an increase in temperature nor a change in confining and fluid pressure is supported by the geological data as mechanisms to explain the brittle fractures that formed during ductile shearing of the Serifos pluton. Alternative mechanisms, such as stress inhomogeneities and/or a strong control of the strain rates including a more general rheology of the ductile crust combining elastic, viscous, and plastic mechanisms (Bürgmann and Pollard, 1994; Regenauer-Lieb and Yuen, 2003; Mancktelow, 2006), must be considered in order to explain the ductile localization of shear zone on brittle fractures.

The structural evolution of the LANF in the granodiorite at Aghios Sostis is illustrated in Figures 8A and 8B. For clarity, later doming of the whole island (Grasemann and Petrakakis, 2007) has been removed by rotating the block diagram by 20° around a rotation axis parallel to the NNE-SSW–striking stretching lineation (see also Figs. 4A and 4C).

Between ~11.5 and 9.5 Ma, the granodioritic pluton and associated dikes intruded into a SSW-dipping LANF that accommodated upper crustal extension by top-to-SSW–directed shearing (Iglseder et al., 2008). At Tsilipaki, Platis Yialos, and Kavos Kiklopas, the LANF was active at the ductile-brittle transition, juxtaposing calcitic marbles deformed at lower greenschist-facies conditions in the footwall against cataclasites in the hanging wall. This observation agrees with the suggested intrusion depth of 12–8 km (Stouraiti and Mitropoulos, 1999). However, dislocation climb recrystallization in K-feldspar and isochemical nucleation and growth of new grains of K-feldspar in the strongly deformed granodiorite in the footwall of the LANF at Aghios Sostis suggest deformation temperatures above 450–500 °C at geologically relevant strain rates (Passchier and Trouw, 2005, and references cited therein). Since these clearly higher temperatures during deformation along the LANF were restricted to the shear zones within the granodiorite, we suggest that locally high paleogeothermal gradients were related to advection of heat by the intrusive body and to the residual heat of the pluton. In accord with models of the spatial and temporal evolution of normal faults (Sibson, 1977), the intensity of deformation increased from the undeformed granodiorite in the footwall toward the ultramylonites below the LANF. Ductile deformation was closely associated with brittle fracturing (see discussion above). In the undeformed granodiorites, fracture segments sjf₁ do not show any macroscopic offset (i.e., joints) or alteration and/or bleaching and are therefore difficult to recognize in the field. However, the same orientation of brittle fractures developed in the strongly foliated granodiorite, where ongoing deformation eventually reactivated these precursor joints generating heterogeneous ductile shear zones with displacement gradients and deflections of the preexisting foliation (e.g., Pennacchioni and Mancktelow, 2007). Note, however, that there is no field evidence such as veins or alteration of the fractures and/or shear zones to suggest pervasive fluid activity during ductile deformation in the footwall of the LANF. This observation is in good agreement with the geochemical data and with similar studies of shear zones in plutons, where ductile flow was shown to deform rocks essentially isochemically (Pennacchioni et al., 2006; see discussion above).

The hanging-wall granodiorite above the LANF records a completely different metamorphic, structural, and geochemical evolution. This totally bleached granodiorite was strongly altered by pervasive fluid flow and, as a result, essentially lacks any mafic minerals. An orthogonal bleached fracture set (sjh_{1, 2}) is associated with en echelon quartz veins. Intense metasomatic-hydrothermal activity along the roof and the margins of the pluton, associated with leaching of the mafic components (mainly Fe, Mg, and Mn) and deposition of the same components as ore bodies in the immediate host rocks, was investigated in detail by Salemink (1985). Fracturing of feldspar grains, dislocation glide, and low-temperature grain boundary migration (bulging) in quartz suggest temperatures during deformation of <350 °C (for a review, see Stipp et al., 2002). Therefore, a minimum of 100 °C difference in deformation temperature has been recorded by the granodiorite above and below the LANF. Using a simple analytical solution of the heat diffusion equation (Jaeger, 1964) and modeling a cylindrical intrusion (6 km radius) into an infinite halfspace (10⁻⁶m²s⁻¹ thermal diffusivity; for a detailed description of the parameters, see Salemink, 1985), a magma with a temperature between 700 and 800 °C intruding into host rocks at ~200–300 °C would need ~500,000 years to cool below 300–400 °C. Assuming that the distance from Aghios Sostis to the extrapolated margin of the pluton toward NNE is ~2–3 km would result in maximum displacement rates in the order of 5 mm/a, if the total offset has been accommodated by ductile shearing. However, in order to explain the temperature difference of ~100 °C across the LANF, a significant displacement distance must have been accomplished by brittle deformation. Given that in such tectonic settings the geothermal gradient can be >>50 °C/km (Vanderhaeghe et al., 2003), even reaching values of ~100 °C/km, the slip in the brittle regime along a LANF dipping between 20°–30° must exceed ~1500 m, reducing the calculated ductile slip rates to less than 2.5 mm/a.

The whole island of Serifos, including the granodiorite and the LANF, is cut by WNW-ESE–striking, conjugate, high-angle normal faults, which are typically associated with the formation of pseudotachylites (sfp in Figs. 4C and 8), suggesting ongoing NNE-SSW extension after cessation of movement along the LANF. Interestingly, no pseudotachylites have been found along the LANF, probably suggesting that the LANF slipped by aseismic creep.

Based on geometric arguments constrained by the extent of the granodiorite pluton, the possibility that the LANF on Serifos rotated from high to low angles during slip can be ruled out. Conjugate high-angle normal faults, suggesting a vertical maximum principal stress direction, both predate and postdate the formation of the LANF on Serifos (Grasemann and Petrakakis, 2007), and therefore models suggesting rotation of the stress field around the LANF are also inapplicable. Since our field observations and geochemical data do not indicate high fluid pressures either within or adjacent to the fault zone attaining tensile fluid overpressures in the footwall of the LANFs, we suggest reduction of the apparent fault strength by weak fault zone material and/or a change in the deformation mechanisms (Rutter et al., 2001) that facilitated movement, which will be subjects for further studies.

Structural fieldwork, together with petrological and geochemical investigations, constrain the following conditions for a LANF cutting the roof of a ca. 11.5–9.5 Ma granodiorite pluton:

(1) A brittle, knife-sharp, low-angle fault plane separates mylonitic granodiorite, deformed under upper greenschist-facies conditions in the footwall, from a weakly foliated granodiorite deformed under lower greenschist-facies conditions in the hanging wall.

(2) Although strongly deformed under upper greenschist-facies conditions, the bulk geochemistry of the progressively deformed footwall granodiorite recorded almost no whole-rock compositional, mass, or volume changes up to the mylonitic deformation state. Within the ultramylonitic state, however, there was a change in deformation-triggered reaction mechanisms that affected both the bulk and mineral geochemistry of the ultramylonitic granodiorite compared to the precursor protogranodiorite. This is reflected mainly by a remarkable modal increase in K-feldspar and decrease in plagioclase and mafic mineral contents. Nevertheless, due to the absence of extrinsic fluid infiltration, observed bulk mass and volume alterations are marginal, also at the highest state of deformation.

(3) In contrast to the footwall, extensive leaching of the mafic components, causing bleaching of the granodiorite, and strong field evidence for hydrothermal fluid activities have been found in the hanging wall of the LANF. A weak foliation and quartz deformation mechanism suggest minor deformation at lower greenschist-facies conditions.

(4) Isolated shear zones formed in both the hanging wall and the foot-wall. In the footwall at least, these ductile shear zones exploited preexisting brittle fractures. Since some of these shear zones record deflection of foliated granodiorite, the fractures must have formed under ductile conditions.

(5) In both the footwall and the hanging wall, all ductile and brittle shear-sense indicators show a very consistent top-to-SSW–directed shearing of the LANF.

(6) WNW-ESE–striking, conjugate, high-angle normal faults, which crosscut all other structures, indicate ongoing brittle crustal extension after cessation of the movement along the LANF. In contrast to the LANF, the high-angle faults are typically characterized by the formation of pseudotachylites.

A Cameca SX-100 electron probe microanalyzer (EPMA) equipped with four wavelength-dispersive spectrometers (Department of Lithospheric Research, University of Vienna) at working conditions of 15 kV acceleration voltage and 20 nA beam current was used to determine mineral compositions. All analyses were made against natural standards with ZAF corrections. For feldspar, mica, and amphibole, a defocused (3 μm) beam technique was applied. All measurements were carried out in multiples to check the reproducibility of the results. High-resolution, backscattered-electron (BSE) images with dimensions of 4 × 3 mm were acquired from a matrix of 10 × 10 pictures automatically stitched using instrument integrated software.

For whole-rock, major element compositions, fused beads were made from a 1:5 mixture of fired specimens and flux; for trace element compositions, pressed powder pellets were prepared. The analyses were made on a Phillips PW 2400 sequential, wavelength-dispersive, X-ray fluorescence (XRF) spectrometer equipped with an Rh-excitation source (Department of Lithospheric Research, University of Vienna). Loss on ignition (LOI) was determined by heating to 850 °C. Analyses of further trace elements and REEs were performed on a Perkin Elmer ELAN 6100 DRC inductively coupled plasma mass-spectrometer (ICP-MS) (Department of Lithospheric Research, University of Vienna), following the digestion method and the analytical procedure described in Tschegg et al. (2008). Densities were determined using the method of Hutchison (1974), with a Mettler P2010N.

This work was supported by the Austrian Science Fund (FWF) grant numbers P18823-N19 (Project ACCEL) and P18908-N19. Fruitful discussions with Kostas Petrakakis, Theo Ntaflos, Gerlinde Habler, and Hugh Rice are gratefully acknowledged. S. Hrabe, C. Beybel, and L. Slawek are thanked for high-quality thin-section preparation. We want to express our sincere gratitude to Hugh Rice for revision of the manuscript, to Jafar Hadizadeh and an anonymous reviewer for their constructive reviews, as well as to James Evans for editorial handling.