Journal of Petrology | Pages |
© 1998 Oxford University Press |
Pressure-temperature (P-T) paths can be difficult to determine for high-grade metamorphic rocks because high-temperature reactions can obliterate assemblages formed during earlier, lower-temperature phases of metamorphism. This can hinder use of petrologic information to interpret tectonic processes. If reaction textures are preserved, however, rocks may record a range of pressure-temperature conditions from their prograde and/or retrograde paths and a more complete picture of the tectonic evolution of a terrane can be reconstructed. Of particular interest are the mechanisms and timing of transition from contraction to extension and the accompanying exhumation of middle- to lower-crustal rocks. The present study describes a medium-pressure, high-temperature terrane (the Nig[breve]de massif; Fig. 1) in central Anatolia, Turkey, that developed as a metamorphic core complex in response to Alpine crustal thickening. Metamorphic rocks in the central, high-grade part of the massif record in their mineral assemblages, reaction textures, and structures the processes of burial, exhumation, and accompanying magmatism.
Anatolia consists of fragments of oceanic and continental crust exposed in east-west trending tectonic belts (Fig. 1) that formed during closure of Tethyan seaways in the Mesozoic and early Cenozoic (S¸engör & Yllmaz, 1981; Robertson & Dixon, 1984; Robertson & Grasso, 1995). The Anatolide tectonic belt consists of several crystalline massifs, including the Menderes and the Central Anatolian crystalline complex (CACC) (Fig. 1). Both are bounded to the north by oceanic rocks of the Izmir-Ankara-Erzincan suture zone (Fig. 1), along which the Neo-Tethys ocean was consumed in late Cretaceous to early Tertiary time (S¸engör & Yllmaz, 1981lREF>).
Although separated from each other by the Inner-Tauride suture zone, the CACC and the Menderes massif are both considered part of the Anatolide tectonic belt. Both massifs contain metamorphosed Gondwanan platform sediments which range in grade from greenschist to upper amphibolite facies, both were intruded by syn- to post-kinematic granodiorite plutons that have low-pressure minerals in their contact aureoles, and both have been interpreted as metamorphic core complexes (Bozkurt & Park, 1994; Whitney & Dilek, 1997). Because of these similarities, and because we use information for the timing of petrologic and tectonic events in western Turkey to interpret the geological evolution of the CACC, we describe the Menderes massif in the following section, highlighting its similarities to and differences from the CACC. The Menderes massif is a large (200 km * 300 km) structural dome that, despite some differences in timing and style of tectonic events, has long been identified as part of the Aegean (Attic Cycladic) domain (Dürr et al., 1978) and has recently been characterized, in whole or in part, as a metamorphic core complex (Verge, 1993; Bozkurt & Park, 1994; Hetzel et al., 1995). The massif was metamorphosed at upper amphibolite facies conditions during the Paleozoic, then underwent Paleocene-Eocene Barrovian metamorphism that has been attributed to Alpine collisional events and burial of the massif beneath the Lycian nappes (Akkök, 1983; S¸engör et al., 1984; Satlr & Friedrichsen, 1986). A late metamorphic event is attributed to Aegean extension and exhumation of the massif (S¸engör & Yllmaz, 1981; S¸engör et al., 1984; Hetzel et al., 1995). Metamorphic grade decreases from upper amphibolite facies in the core of the massif to greenschist facies at higher structural levels. Metapelitic rocks from the upper units contain garnet, staurolite, staurolite + kyanite, and kyanite zones (Ashworth & Evirgen, 1985a, 1985b). Estimated conditions of 440-550°C, 4-7 kbar for staurolite + kyanite and kyanite schist from the central part of the massif are consistent with estimates for the eastern part of the massif (Akkök, 1981). The central part of the massif contains syntectonic granodiorite intrusions that have andalusite and sillimanite in their contact aureoles. Similar Miocene granodiorite intrusions crop out in the Cycladic Islands and the Nig[breve]de massif (see below). The metamorphic and plutonic rocks east and southeast of Ankara (Figs 1 and 2) have been called the Central Anatolian massif, Klrs¸ehir massif, Klrs¸ehir continent (S¸engör et al., 1984), and the Central Anatolian crystalline complex (CACC) (e.g. Aklman et al., 1993). We adopt the last terminology and use the term `Klrs¸ehir massif' to refer only to the NW part of the CACC. The Akdag[breve] and Nig[breve]de massifs occur in the east and the south, respectively, of the CACC (Figs 1 and 2). The CACC is bounded to the north by the Izmir-Ankara-Erzincan suture, to the west by the dextral Tuz Gölü fault, and to the east by the sinistral Ecemis¸ fault (Figs 1 and 2).
The Klrs¸ehir massif is characterized by a sequence of greenschist to granulite facies rocks. Seymen, (1984) estimated P-T conditions of 400-700°C at low pressure, on the basis of index minerals in metapelitic and metacarbonate rocks located northwest of the city of Klrs¸ehir (Fig. 2). These rocks were intruded by Tertiary granitoids that also intruded the Izmir-Ankara-Erzincan suture zone.
Kocak & Leake, (1994) studied the Ortaköy region of the Klrs¸ehir massif (Fig. 2) and described a sequence of metasedimentary rocks including migmatitic gneiss that contains potassium feldspar, sillimanite, unzoned garnet, and texturally late andalusite and cordierite. Garnet-biotite geothermometry yielded 680 ± 50°C at 4 kbar for metapelitic schist. A pressure of 4 ± 1·5 kbar (at 600°C) was estimated for the assemblage garnet + biotite + sillimanite + cordierite. The Akdag[breve] massif (Figs 1 and 2) is regarded to be the eastern extension of the Klrs¸ehir massif. Garnet-staurolite schist collected during reconnaissance field-work (this study) in the Akdag[breve] massif contains minor amounts of fibrolite. Thermobarometric results from these schists yielded T ~550-600°C, ~4 kbar (D. L. Whitney & Y. Dilek, unpublished data, 1995). Although petrologically similar, the Nig[breve]de massif displays a P-T-t history and tectonic style distinct from these other regions of the CACC. It is an isolated dome of crystalline rocks south of the main exposures of the Klrs¸ehir and Akdag[breve] massifs (Fig. 1), adjacent to the Inner-Tauride suture, and truncated on the east by the Tertiary Ecemis¸ fault zone (Fig. 2). We recently identified the Nig[breve]de massif as a metamorphic core complex that developed as a result of crustal thickening during Alpine collision and subsequent extension in the hanging wall of a north-dipping subduction zone (Whitney & Dilek, 1997). The massif has all the elements of a Cordilleran-style metamorphic core complex: plastically deformed metamorphic and plutonic basement (footwall), sedimentary cover rocks (hanging wall), and a low-angle normal fault zone separating the basement and cover (Fig. 3). The more northern regions of the CACC (Klrs¸ehir and Akdag[breve] massifs) do not appear to be core complexes.
The petrology and stratigraphy of the Nig[breve]de massif have been described by
Göncüog[breve]lu, (1981, , 1982, , 1986). The structurally lowest unit of the massif is the Gümüs¸ler Formation (Fig. 3), which consists primarily of pelitic or psammitic schist and gneiss with interlayered calc-silicate, amphibolite, quartzite, and marble. The Gümüs¸ler Formation is overlain by thinly bedded metaclastic and metacarbonate schist of the Kaleboynu Formation, and finally by the As¸lgedig[breve]i Formation, which is composed dominantly of marble, with amphibolite, quartzite, fine-grained schist, and meta-ultramafic rocks. Some of the quartzite contains Mn andalusite (Göncüog[breve]lu, 1981). As in the cover series of the Menderes massif, metapelitic or semipelitic schist predominates in the structurally lower parts of the Nig[breve]de massif, whereas marble is abundant in the upper parts. The most common assemblage in Gümüs¸ler gneiss is biotite + quartz + plagioclase ± fibrolite-sillimanite ± garnet ± muscovite ± potassium feldspar. Accessory minerals include tourmaline, apatite (up to a millimeter in diameter), rutile, zircon, pyrite, ilmenite, magnetite, and calcite veinlets. Rutile and ilmenite do not coexist. Andalusite-bearing quartz veins are present in highly deformed and retrograded gneiss near the eastern margin of the massif. Metapelitic rocks display a foliation defined by biotite and a lineation defined by sillimanite. Fibrolite is the dominant textural variety of sillimanite (Fig. 4), but coarse grains also occur. Biotite is dark red and contains ~1·3-3·5 wt % TiO2 (Table 1). In many rocks, muscovite (Table 2) is randomly oriented and cuts the foliation defined by biotite, perhaps indicating that it formed late in the metamorphism and deformation of the rocks. Rocks in the central part of the massif lack muscovite or contain texturally late muscovite. Table 1. Representative biotite analyses. Table 2. Representative muscovite analyses.
There are two main textural types of garnet (Table 3) in the Gümüs¸ler gneiss: (1) small (<1 mm) grains that contain a few inclusions of plagioclase, biotite, and/or quartz and which have been partially resorbed, as indicated by rims of plagioclase ± quartz (Fig. 5), and (2) larger, poikiloblastic grains, 1-3 mm in diameter, that contain abundant inclusions of biotite, plagioclase, quartz, fibrolite, apatite, Li-Al tourmaline, and potassium feldspar, as well as abundant, minute mineral and fluid(?) inclusions (Fig. 6). No kyanite inclusions have been observed in Gümüs¸ler garnet. Large biotite grains as well as composite inclusions of biotite + fibrolite, biotite + quartz ± fibrolite, and biotite + fibrolite + quartz + plagioclase + potassium feldspar ± apatite preferentially replace type 2 garnet cores (Fig. 6). Some samples of Gümüs¸ler gneiss contain rounded clots of sillimanite + biotite ± muscovite ± plagioclase ± quartz (Fig. 5c and d). The former presence of garnet is inferred from the rounded shape and from the occurrence of similar assemblages that have partially replaced garnet. Table 3. Representative garnet analyses.
Garnet grains are typically unzoned except at their outer 10-200 µm (Fig. 7a) and next to biotite inclusions, where Mn increases and Mg decreases (Fig. 7b and c). In grains with numerous biotite inclusions, a significant volume of the garnet has been affected by chemical exchange with biotite (Fig. 7c). Biotite inclusions in garnet are similar in composition to matrix biotite (Table 1).
Garnets of both textural types contain inclusions of plagioclase with an average composition of An20-35 (Table 4), although some of these oligoclase-andesine inclusions contain albitic patches (An0-9). Most plagioclase inclusions in garnet are irregularly zoned and the host garnet is slightly zoned (Ca in garnet decreases) next to some plagioclase inclusions (see
Whitney, 1991). Table 4. Representative plagioclase analyses from metapelitic rocks.
In garnets that contain abundant inclusions, overall grossular content is extremely low except for local regions of slightly higher grossular content (Fig. 7d) that are similar to the overall composition of garnets that lack mineral inclusions. The rest of the garnet has apparently been compositionally modified by reaction with inclusions. Most matrix plagioclase is oligoclase-andesine (up to An45) and grains are typically irregularly zoned. In migmatitic samples, small ( <= 10-15 µm) grains of albite (An0-1) occur in mesosomes and some leucosome plagioclase grains (oligoclase-andesine) have albitic patches. Centimeter- to millimeter-scale granitic leucosomes in these rocks contain approximately equal proportions of plagioclase + potassium feldspar (Table 5) + quartz. In leucosomes, myrmekitic intergrowths of plagioclase and quartz are common along potassium feldspar grain boundaries. Less deformed leucosomes contain plagioclase with euhedral cores. Table 5. Representative potassium feldspar analyses from metasedimentary rocks.
Marble interlayered with metapelitic gneiss typically contains >95-98% calcite with minor quartz and plagioclase. Calc-silicate gneiss interlayered with metapelitic rocks contains the assemblages calcite + plagioclase + quartz + diopside ± grossular ± titanite and calcite + quartz + diopside. Some Si- and Al-rich calcareous rocks contain skeletal, highly resorbed grossular-rich garnet (Grs component 28-34 mol %), clinopyroxene, hornblende, biotite, muscovite, plagioclase, potassium feldspar, epidote, and titanite. Plagioclase is strongly zoned, with calcic cores (up to An83) and sodic rims (An25-29). Amphibolite interlayered with metasedimentary rocks contains hornblende + plagioclase ± clinopyroxene ± texturally late calcite and zeolite minerals (Tables 6 and 7). Plagioclase is zoned, with calcic cores (An30-34) and sodic rims (An21). Table 6. Representative plagioclase analyses from amphibolites. Table 7. Representative amphibole analyses.
Antimony-mercury-tungsten deposits in the massif consist of veins of stibnite, stibnite-cinnabar, stibnite-cinnabar-scheelite ± barite ± gold and abundant tourmaline (Akçay et al., 1995). Tourmaline compositions and fluid inclusion data suggest a magmatic origin for the tourmaline and possibly for the Sb-Hg-W mineralization as well (Akçay et al., 1995). The main body of the Üçkaplll granite crops out in the central and NE part of the massif (Fig. 3), although there are many smaller exposures of mineralogically similar granite throughout the massif (Fig. 3). The granite is peraluminous, ranges from monzonite to syenitic granite (Aklman et al., 1993), and has an Sr initial ratio of 0·7104 (Göncüog[breve]lu, 1986). The most common assemblage in the Üçkaplll granite is quartz, plagioclase (An8-40), potassium feldspar, biotite, and muscovite, with accessory zircon, apatite, tourmaline, and magnetite. Plagioclase is zoned from euhedral calcic cores to sodic rims. A small intrusion near the southern end of the main body contains abundant, coarse garnet grains. The Üçkaplll granite intruded all formations of the Nig[breve]de Group. It contains metasedimentary xenoliths of various sizes, including a large raft of gneiss near its southern margin. The contact aureole is centimeters to decimeters thick (Göncüog[breve]lu, 1986) and contains cordierite and andalusite in metapelitic Gümüs¸ler rocks. The Gümüs¸ler gneiss is cut by two suites of granitic dikes. The first occurs near the margins of the Üçkaplll two-mica granite, is mineralogically identical to the pluton, and cuts the mineral foliation and migmatitic layering of the host rocks at steep to moderate angles. The second consists of centimeter- to meter-scale peraluminous dikes. Some of these are concordant to the host gneiss foliation and others cross-cut foliation and migmatitic layering. The most common assemblage in the peraluminous dikes is potassium feldspar + plagioclase + quartz + tourmaline ± garnet ± fibrolite-sillimanite ± andalusite ± xenotime ± muscovite. Similar peraluminous dikes intruded the Menderes massif (Bozkurt et al., 1995). Both dike suites contain centimeter-scale xenoliths of the host metapelitic gneiss and most dikes produced a 1-3 cm thick contact zone in the host rock. Both xenoliths and contact zones are depleted in biotite relative to the rest of the country rock and consist of granular, recrystallized plagioclase and quartz. The xenoliths and some of the contact zones also contain abundant, coarse-grained, texturally late muscovite. Unmetamorphosed to greenschist facies basalt dikes intruded all formations of the massif, including the Üçkaplll granite. Some of these dikes contain angular fragments of their host rocks. The Kaleboynu Formation (Fig. 3) consists of weathered, thinly interbedded calc-silicate, marble, metasiltstone, quartzite, and fine-grained schist that are cut by deformed granitic dikes and sills. The stratigraphically youngest formation of the Nig[breve]de Group is composed largely of monomineralic calcite marble with interlayered quartzite and amphibolite.Bedded marble and quartzite are deformed into folds of tens to hundreds of meters in amplitude. Deformed, meter-scale greenschist facies basaltic dikes and sills cut marble and quartzite near the contact between these lithologies. Meta-ultramafic rocks are infolded with the metasedimentary rocks. This poorly outcropping mafic unit consists of variably deformed and metamorphosed gabbro. Samples from individual outcrops contain both apparently unmetamorphosed hornblende gabbro with igneous-appearing, randomly oriented, lath-shaped plagioclase grains and no metamorphic minerals, as well as highly deformed metagabbro with greenschist facies assemblages. Some of the gabbro bodies are intruded by fine-grained granitic dikes. Because contacts between gabbro and country rock are not exposed, it is unclear whether the gabbro intruded the Nig[breve]de massif or was tectonically emplaced. Kocak & Leake, (1994) interpreted metagabbro in the Ortaköy region (Fig. 2) of the CACC as tectonically emplaced ophiolite fragments. The Sineksizyayla gabbro, however, is not spatially associated with ultramafic rocks. An andalusite + sillimanite-bearing dike provides additional information about the P-T path of the central part of the massif. The 1·5 m thick dike cuts a large raft of gneiss near the main body of the Üçkaplll granite (Fig. 3), and also cuts layer-parallel leucosomes. The dike contains plagioclase + potassium feldspar + quartz + garnet + tourmaline + andalusite + sillimanite + ilmenite + apatite + euhedral xenotime. Coarse, prismatic sillimanite clearly replaces andalusite (Fig. 8a, b). This assemblage differs from those of other peraluminous granitic dikes and leucosomes in the Nig[breve]de massif in that it contains Al2SiO5 phases and lacks primary muscovite. The aluminosilicate phases are not evenly distributed throughout the dike, but occur in clusters.
Sillimanite also occurs in the matrix of the dike, primarily as fibrolite (Fig. 8c and d; Fig. 9). Andalusite and sillimanite-fibrolite define a strong lineation parallel to the trend of the dike, although some fibrolite occurs as abundant, randomly oriented grains in quartz and feldspar.
Some andalusite grains display patchy zoning. Pleochroic pink to colorless regions that contain up to 1 wt % Fe2O3 (Table 8) generally occur as grain cores surrounded by colorless, Fe2O3-free rims. Andalusite grains are surrounded by An11-20 plagioclase (Fig. 8c and d; Table 8), similar in composition to matrix plagioclase grains. Plagioclase rimming andalusite typically occurs as blocky crystals that display grain boundaries perpendicular to the andalusite margins. Rounded, colorless andalusite grains occur in the plagioclase coronas (Fig. 8c and d). Table 8. Representative mineral analyses from the Al2SiO5-bearing dike.
In addition to plagioclase, the region around andalusite also contains quartz, which is separated from andalusite by a plagioclase rim. The plagioclase ± quartz zone is surrounded by fibrolite-sillimanite (Fig. 8c and d). Matrix plagioclase is in the range An11-20, but is typically An13-15. The average composition of K-feldspar in the dike is Or87-88Ab10-12An0-01 (Table 8). Almandine-spessartine garnet (~1 mm diameter)occurs as euhedral to slightly rounded, largely inclusion-free grains (Fig. 9) that are essentially unzoned at XAlm = 0·70, XSps = 0·21, with minor amounts of pyrope and grossular (Table 8). Some of these garnets contain inclusions of fibrolite that are typically aligned with matrix fibrolite (Fig. 9). The Gümüs¸ler Formation is the only unit of the massif that contains assemblages suitable for both pressure and temperature determination. Metaclastic rocks in younger formations apparently equilibrated at lower temperatures and pressures than the Gümüs¸ler Formation: the Kaleboynu Formation contains very fine-grained rocks that lack high-grade mineral assemblages and the overlying As¸lgedig[breve]i Formation contains Mn-andalusite (Göncüog[breve]lu, 1981). Metamorphic P-T conditions were calculated for Gümüs¸ler metapelitic rocks using garnet-biotite geothermometry combined with the garnet-plagioclase-aluminosilicate-quartz (GPAQ) geobarometer (Table 9). For some rocks that contain muscovite, the garnet-plagioclase-muscovite-biotite (GPMB) geobarometer was applied using garnet rim compositions (Table 9). Temperatures were also determined by hornblende-plagioclase thermometry for an amphibolite (Table 9). Table 9. Thermobarometric results.
Temperatures were estimated for nine garnet-sillimanite gneiss samples collected from a NNW-SSE traverse across the massif (Fig. 3). `Peak' conditions were determined using near-rim compositions of garnet in regions of the garnet not affected by reaction with mineral inclusions (determined from X-ray maps and microprobe analyses). Neighboring matrix biotite was used for thermometric calculations. Oligoclase-andesine inclusions combined with adjacent garnet compositions yielded similar pressures to those determined with garnet outer core-matrix (adjacent garnet) plagioclase compositions. Estimated temperatures (730-780°C) and pressures (~5-6 kbar) are consistent with the mineral assemblages of the Gümüs¸ler gneiss, but these temperatures may not correspond to the peak temperatures attained because minerals may continue to react with each other during decompression and cooling. Pressure-temperature determinations for high-grade rocks are difficult because Fe-Mg exchange geothermometers tend to have different closure temperatures from geobarometers based on net transfer equilibria (e.g. Frost & Chacko, 1989). In addition, the presence of garnet pseudomorphs and resorbed garnet rims suggests that garnets may not record maximum P-T conditions. Garnets are slightly zoned at their rims (Fig. 7). Increase of spessartine contents at garnet rims is probably a retrograde feature, and temperature estimates based on garnet rim + adjacent biotite (550-580°C) are reported as retrograde temperatures in Table 9. The closure of Tethyan seaways between the Pontide and Anatolide tectonic belts in the north and between the Tauride and Anatolide belts in the south caused Alpine orogenesis in Turkey. Crustal thickening, metamorphism, and magmatism in the Anatolide tectonic belt (Menderes massif and CACC) followed collision in the Early Eocene (S¸engör & Yllmaz, 1981; Dewey et al., 1986). Metamorphic rocks in the Aegean region (Cycladic Islands) record an Eocene high-pressure-low-temperature metamorphism at 40-50 Ma (Altherr et al., 1979) and a Barrovian overprint at 25 Ma (Andriessen et al., 1979). In Turkey, Cretaceous high-pressure-low-temperature metamorphism has been recognized insuture zones (Okay, 1984, , 1986) and a Tertiary high-pressure event has recently been documented in the Menderes massif (Oberhänsli, 1997; Oberhänsli et al., 1997). High-pressure metamorphism has not been identified in the CACC. In the Menderes massif, north-south contraction probably ended by late Oligocene, and major north-south extension accompanied by unroofing of the massif and widespread basin formation occurred in the early Miocene (Seyitog[breve]lu & Scott, 1996). These events are constrained by K-Ar ages from volcanic and volcaniclastic rocks and biostratigraphic data (S¸engör et al., 1984; Seyitog[breve]lu & Scott, 1991, , 1996; Seyitog[breve]lu et al., 1992) and by the early Miocene 40Ar-39Ar (biotite, hornblende) age of a syntectonic granodiorite (Hetzel et al., 1995) emplaced during plastic deformation of the core of the massif. Similarly, the Attic Cycladic complex was exhumed between the late Oligocene and Miocene-Pliocene time (e.g. Lee & Lister, 1992; Gautier & Brun, 1994). The timing of regional metamorphism of the Nig[breve]de massif is broadly constrained by biostratigraphic information from overlying sedimentary rocks and the age of the late-metamorphic Üçkaplll granite. The minimum depositional age of the inferred sedimentary protoliths is early Late Cretaceous, based on foraminifera from unmetamorphosed to slightly metamorphosed rocks in the Bolkar Mountains that are correlated with the Nig[breve]de Group rocks (Göncüog[breve]lu, 1986). The Üçkaplll granite has been determined by a four-point Rb-Sr whole-rock isochron to have intruded at 95 ± 11 Ma (Göncüog[breve]lu, 1986). A U-Pb monazite date for crystallization of the pluton, however, indicates that the granitoid intruded in the Miocene (13·7-20 Ma; Whitney & Dilek, 1997). We attempted to determine the age of the granite by analyzing zircon from the pluton and xenotime from a peraluminous dike, but the apparently young age of the intrusion (i.e. Miocene rather than Cretaceous), combined with small sample size, precluded obtaining a date from either mineral. It is difficult to reconcile these very different ages for the Üçkaplll granite, both of which are interpreted to represent the crystallization age of the pluton. We adopt the Miocene date because it is consistent with regional age constraints and because Rb-Sr whole-rock isochrons may not represent crystallization ages. Regional metamorphism of the massif is therefore broadly constrained to be Eocene-Early Miocene, similar to the timing of Barrovian metamorphism in the Menderes massif and Attic Cycladic complexes. Although the prograde history of the metasedimentary rocks is unconstrained, the near-peak and post-peak path can be inferred from mineral assemblages, reaction textures, and results of thermobarometry. Several lines of evidence (discussed in more detail in the following paragraphs) indicate that Gümüs¸ler gneiss in the central part of the massif experienced high temperatures during regional metamorphism. These include: (1) the presence of migmatites interpreted to be anatectic; (2) the coexistence of potassium feldspar and sillimanite in rocks that lack primary muscovite; (3) the absence of zoning in garnet; and (4) the results of garnet-biotite geothermometry. Abundant migmatite is present in the central part of the massif. Granitic leucosomes and relict igneous textures (e.g. euhedral plagioclase cores) in peraluminous leucosomes suggest that migmatization occurred by partial melting of the metapelitic gneiss, although some leucosomes may be injection features related to intrusion of the Üçkaplll granite. If the migmatites formed by anatexis at moderate pressure (~6 kbar; Fig. 10), the assemblage potassium feldspar + sillimanite indicates that migmatites probably formed by dehydration melting during decompression at T > ~650-720°C (Fig. 10). At pressures below 6 kbar, the presence of potassium feldspar and the absence of texturally early muscovite in the matrix of sillimanite-bearing Gümüs¸ler gneiss implies the equilibrium
which signifies relatively high-temperature conditions in metapelitic rocks (>600°C; Fig. 10).
High temperatures are also consistent with the absence of zoning in garnet. Lack of zoning is probably due to homogenization of pre-existing growth zoning at high temperature. The temperature at which garnet growth zoning homogenizes depends on various factors such as the size of the garnet and duration of heating, but is likely to be >600-650°C. Because growth zoning is not preserved, lower-grade equivalents of the Gümüs¸ler gneiss are not exposed, and garnet does not contain relict inclusions that are useful for reconstructing prograde reactions, the early history of garnet growth is not recorded. It is therefore not possible to determine if garnet experienced any episodes of partial consumption during prograde metamorphism. Evidence for the high-grade history of garnet growth and partial consumption can, however, be inferred from reactions that occurred between garnet and mineral inclusions in garnet. For example, garnets contain isolated, apparently primary inclusions of potassium feldspar, indicating that garnet growth continued to high temperatures; i.e. beyond the second sillimanite isograd [reaction (1)] or during migmatization. In the latter case, growth of garnet + potassium feldspar may have occurred by means of a partial melting reaction such as
at high temperature (>750°C; Spear & Parrish, 1996; Fig. 10). Additional information can be obtained from the compositions of plagioclase inclusions in garnet. Albite and oligoclase-andesine inclusions in the same garnet (and, in some cases, together in a composite inclusion) may represent a relict greenschist-facies assemblage if the lack of compositions between Ab0-1 and An22 indicates the peristerite gap. Ashworth & Evirgen, (1985a) reported the preservation of plagioclase from either side of the peristerite gap in kyanite-bearing rocks from the Menderes massif. This explanation seems implausible for the Gümüs¸ler rocks, however, because peak temperatures were very high and because garnets are highly fractured and their interiors have therefore not remained armored, isolated systems (Whitney, 1996). Migmatitic Gümüs¸ler gneiss contains albite, however, so alternatively, the albite inclusions in garnet may represent the products of partial melting of the gneiss before or during garnet growth. This indicates that garnet growth continued at relatively high temperatures, consistent with the interpretation of the primary potassium feldspar inclusions discussed above. As noted above, many Gümüs¸ler garnets contain abundant inclusions of biotite that in some cases are accompanied by other phases such as sillimanite (fibrolite), quartz, plagioclase, potassium feldspar, and apatite (Fig. 5). Although some biotite inclusions may represent former matrix phases entrapped during garnet growth, textural evidence (including the inferred complete pseudomorphing of garnet by the above-listed phases; Fig. 5c and d) suggests that most of the large biotite inclusions and all of the composite inclusions have grown at the expense of the host garnet (Figs 5 and 7). Replacement of garnet by K- and Na-bearing phases suggests either dissolution and reprecipitation of pre-existing inclusions (micas, feldspars) or introduction of components from the matrix via fluids that enter the garnet along fractures. Back-reaction of the partial melting reaction (2) could account for growth of biotite and sillimanite at the expense of garnet during decompression and cooling (Fig. 10; see also
Spear & Parrish, 1996), or a subsolidus equilibrium such as (Fig. 10) could also account for the replacement of garnet by biotite and sillimanite. Reaction (3), a hydration reaction, would be encountered during decompression after a substantial amount of cooling (Fig. 10). Preserved textural and compositional features therefore indicate that garnet growth continued to high temperatures and that partial consumption occurred at selected sites in garnet interiors, probably during decompression. The andalusite + sillimanite-bearing dike provides information about the metamorphic history of the high-grade central part of the Nig[breve]de massif. The field relations of the dike indicate that it intruded late in the metamorphic history of the Nig[breve]de rocks, consistent with its low-pressure mineral assemblage. Texturally early andalusite in the dike either crystallized from the magma or represents the incorporation of pre-existing grains. The highly resorbed texture of the andalusite and the presence of plagioclase + quartz coronas suggest that the grains are xenocrysts. The proximity of the dike to the Üçkaplll granite may be significant. The presence of cordierite in the contact aureole (Göncüog[breve]lu, 1986) indicates that the pluton was intruded at shallow crustal levels. Andalusite and fibrolite may have also formed during contact metamorphism and become entrained in the magma. Plagioclase-rich coronas around andalusite may have formed during partial melting during or after incorporation of andalusite in the magma. Growth of sillimanite after andalusite requires an increase in temperature and/or pressure. The absence of cordierite and the stable presence of almandine-rich garnet, sillimanite, and quartz in the dike may indicate pressures of 2·8-3·0 kbar (Fig. 10). The maximum temperature is less well constrained but must have been >550°C based on the mineralogy of the dike (Fig. 10). The late thermal spike was clearly associated with magmatism, but because magmatism was voluminous during the late stages of core complex development, the thermal effects were regional in extent. Simple one-dimensional modeling of the thermal effects of a granitic pluton the size of the main body of the Üçkaplll granite, assuming a country rock temperature in the andalusite stability field (Fig. 10), indicates that most of the high-grade core of the Nig[breve]de massif would have experienced temperatures >550°C on a time scale of several million years. Barrovian metamorphism occurred at mid-crustal pressures of 5-6 kbar but at high temperatures (>700°C), implying a geothermal gradient of 36-50°C/km. Decompression to low pressures (3-4 kbar) at temperatures of >550°C is recorded in the vicinity of the Üçkaplll granite. The peraluminous dike lacks cordierite but contains almandine in apparent equilibrium with sillimanite, which suggests that the dP/dT slope of the late thermal spike may have been slightly positive (Fig. 10), perhaps owing to small amounts of crustal thickening related to magmatism. A P-T-t path for the Gümüs¸ler gneiss that integrates pressure-temperature conditions with inferences about the age of the Barrovian metamorphism (Eocene-Oligocene) and the Miocene crystallization age of the Üçkaplll granite (Whitney & Dilek, 1997) has a clockwise trajectory with a Miocene high-temperature excursion following several kilometers of decompression of the high-grade core. The history of the Nig[breve]de massif more closely resembles that of the Menderes massif than it does the rest of the CACC in estimated P-T-t paths, the timing and composition of magmatism, and tectonic history. Because the entire CACC consists of similar protoliths, the difference between the Nig[breve]de massif and the more northern regions of the CACC is probably related to tectonic position (Whitney & Dilek, 1997). We conclude that both the Nig[breve]de and Menderes massifs developed as core complexes in the vicinity of a southern subduction zone, the Inner-Tauride suture. In contrast, the more northern parts of the CACC underwent lesser amounts of crustal thickening and did not undergo major extension following Alpine contraction. The Nig[breve]de metasedimentary protoliths were buried and exhumed during a single orogenic episode, but the rocks in the high-grade core of the massif did not follow a simple continuous P-T path. Multi-peak P-T paths are expected in polymetamorphic terranes, but few multi-peak paths have been proposed for single orogenic events (Whitney & Dilek, 1998). P-T paths with multiple thermal peaks might be expected in metamorphic terranes in which late magmatism followed crustal thickening. They may, however, only be preserved in rapidly exhumed terranes such as metamorphic core complexes. Post-thickening extension in orogens has been increasingly recognized as a major geodynamic process affecting the evolution of continental lithosphere. The eastern Mediterranean region, including central Anatolia, experienced Alpine crustal thickening and metamorphism, closely followed by extension and magmatism. The Nig[breve]de massif provides useful information about the P-T-t history of the crust during compressional and extensional phases of orogenesis, and the timing and thermal effects of magmatism during core complex development. We would like to acknowledge the logistical support provided by Dr Evren Yazgan, Head of the Geological Research Department at the General Directorate of Mineral Research & Exploration of Turkey (MTA), the Adana Regional Office of MTA, and Professor Cavit Demirkol of the Çukurova University (Adana) during our field-work in Nig[breve]de. S. M. Kuehner provided invaluable assistance with X-ray maps. Reviews by D. Pattison, M. Kohn, and R. Oberhänsli improved the paper. This work was supported by NSF Grant EAR-9317100 to D.L.W. and EAR-9219064 to Y.D.INTRODUCTION
REGIONAL GEOLOGY
Menderes massif
Central Anatolian crystalline complex (CACC)
PETROLOGY AND MINERAL CHEMISTRY OF THE NIG[Breve]DE MASSIF
Gümüs¸ler Formation
Üçkaplll granite
Dike suites
Kaleboynu Formation
As¸lgedig[breve]i Formation
Sineksizyayla Gabbro
PETROLOGY OF AN ANDALUSITE + SILLIMANITE-BEARING DIKE
THERMOBAROMETRY
TIMING OF EVENTS
PRESSURE-TEMPERATURE HISTORY
muscovite + quartz
potassium feldspar + sillimanite + H2O(1)
biotite + silmanite + plagioclase + quartz =
garnet + potassium feldspar + liquid(2)
garnet + potassium feldspar + H2O =
biotite + sillimanite + quartz(3)
Late heating event
Summary of P-T path
REGIONAL SYNTHESIS
ACKNOWLEDGEMENTS
REFERENCES