Documenting Exhumation in the Central and Northern Menderes Massif (Western Turkey): New Insights from Garnet-Based P-T Estimates and K-Feldspar ⁴⁰Ar/³⁹Ar Geochronology

The Menderes Massif, located in western Turkey (Figure 1), is a large (>40,000 km²), regionally significant, metamorphic core complex made up of both metamorphic and igneous rocks. These rocks contain abundant geochemical and geochronological records of an evolving orogen from collision through to extension-driven exhumation. As such, this area has been the focus of many studies that were aimed at understanding the lithosphere dynamics of syn- to postcollisional tectonics (e.g., [1–6]). Many of these studies make predictions about a lithospheric response to plate tectonic processes such as slab-roll back and orogenic collapse.

However, deformation in the Menderes Massif is a consequence of two major orogenic episodes (i.e., polymetamorphism) from the Late Proterozoic to Eocene, then followed by mylonitization overprinting in the Miocene. This history has made it difficult to adequately study the regional prograde metamorphic history during collision following the closure of the Neo-Tethys Ocean (e.g., [1, 7–10]). Specifically, previous peak thermobarometric estimates are broad, ranging from <450°C to >650°C and ~4 kbar to ~10 kbar [8, 10–12], and reported $P$-$T$ paths are conflicting [10–12]. The broader implication of such uncertainty is an apparent lack in understanding the complete thermal-tectonic evolution of western Anatolia during the Alpine-Himalayan orogenic cycle. Because the Menderes Massif is a primary target for studying postcollisional tectonics, an appropriate understanding of the thermal state of the crust prior to exhumation is critical for understanding anticipated rheological behaviors and for reconstructing the magnitude of heat and mass transfer during extension (e.g., [13]).

Chemically zoned garnets found throughout the Menderes Massif appear to offer a promising insight for extracting their prograde history. For instance, in the Çine Massif, Etzel et al. [14] utilized chemically zoned garnets to deduce that regional Barrovian metamorphism achieved maximum conditions of $7.5±0.5kbar$ and $590±20°C$⁠.

Garnet zoning is the chemical expression of changing environmental conditions such as pressure (⁠$P$⁠) and temperature (⁠$T$⁠) during its growth history. Efficiently modeling high-resolution $P$-$T$ paths by exploiting this chemical variability is a promising approach for reconstructing tectonic histories (e.g., [14–17]). In this study, we explore conditions of garnet growth during metamorphism for a suite of rocks collected in the Central Menderes Massif (CMM) by examining their chemical zoning and propose some of the most informative $P$-$T$ paths related to regional metamorphism in the CMM. Reporting detailed $P$-$T$ paths resolves existing uncertainties regarding the magnitude of burial (and heating), which, in turn, refines tectonic reconstructions and our understanding of lithospheric dynamics in this region of the Alpine-Himalayan Orogenic Belt. Monazite grains (REEThPO₄) within the matrix of garnet schists were dated in thin section (in situ) and provide insight into timing the transition from peak metamorphism to extensional exhumation. Additionally, ⁴⁰Ar/³⁹Ar K-feldspar ages from granites exposed along or near prominent detachments in the central and northern (Gördes) Menderes Massif provide further insight into the thermal and physical history of the crust during exhumation. The data are used to present a conceptual overview of the Menderes Massif’s thermal and tectonic evolution.

The Menderes Massif (Figure 1), located in western Anatolia (Turkey), is situated along the Alpine-Himalayan Orogenic Belt. This massif is the largest of several prominent Aegean core complexes; research across the region has demonstrated that the Menderes Massif and Aegean-proper domains are distinct from one another (e.g., [2, 18–21]). The Menderes Massif is divided into three submassifs: the northern Gördes, central, and southern Çine Massifs. Many have already described the geological and tectonic background of this region (cf., [2, 22–24]), which is briefly summarized here.

The Menderes Massif metamorphic basement was mapped initially as “core” rocks consisting of Precambrian gneisses and schists overlain by Mesozoic-Cenozoic “cover” mica-schists and marbles based upon structural relationships (e.g., [5, 21]). Others mapped this region as four nappes stacked during south-directed thrusting (e.g., [7, 24–26]). The nappes are named, from the structurally lowest to the highest: Bayındır, Bozdağ, Çine, and Selimiye. Preserved prograde mineral assemblages in these metapelites, eclogites, and amphibolites record evidence of Barrovian-style metamorphism during multiple collisions between Gondwana blocks (both continent and microcontinent-derived blocks) and continental Laurasia (e.g., [1, 7, 12, 21, 22, 27]). Overprinting has made estimating conditions of prograde metamorphism difficult, and previous estimates of metamorphic pressures and temperatures are highly variable. Estimates range from 425°C to 650°C and ~4 kbar to ~10 kbar, depending on nappe and location (see review in [28]).

The range of conditions should not be unexpected as at least three Alpine deformation events and one pre-Alpine deformation are reported in the Menderes Massif (cf., [24]). Pre-Alpine upper-amphibolite fabrics, Neoproterozoic in age, are recorded in the Selimiye, Çine, and Bozdağ nappe rocks and are presumably related to Pan-African activity (D_PA in [24]). This interpretation was drawn based upon Neoproterozoic granites (Pb/Pb zircon age) cross-cutting the fabric. Regional structural mapping throughout the Menderes Massif indicates that the first Alpine deformation event observed in all Menderes Massif rocks reflects overall top-to-south shearing developed during prograde metamorphism (referred to as D_A3 in [24]). This event is referred to as the Main Menderes Metamorphism (e.g., [21]). Following this, mylonitic foliation overprinted much of the Bozdağ and Bayındır nappe rocks (D_A4). We note that this overprinting, resulting from the regional extension, is responsible for the dominant foliation observed in Bozdağ and Bayındır nappe rocks [29]. Finally, brittle deformation (D_A5), manifested as regional normal faulting, was underway by the Oligocene-Miocene. This exhumation was facilitated in the CMM by faulting along the E-W trending (100 km long) Alaşehir and Buyuk low-angle detachments and in the Gördes Massif by the high-angle Simav normal fault (e.g., [2, 18, 19, 30], Bozkurt and Park, 1994). Approximately 150 km of N-S extension has occurred in the entire Menderes Massif since Oligo-Miocene times [6]. In the CMM, Mid-Miocene aged granites emplaced at a depth of ~7 km are now exposed in the footwall of the Alaşehir detachment [31, 32]. Gessner et al. [29] estimated that displacement of 10-12 km along the Alaşehir detachment has occurred.

Syntectonic pluton emplacement occurred in proximity to the prominent detachments and normal faults (Figure 1). Along the Alaşehir detachment, two such S-type peraluminous granites (Salihli and Turgutlu granites, Figure 1) are thought to have crystallized between $21.7±4.5Ma$ and $15.0±0.3Ma$ [32–35]. A monazite inclusion in plagioclase from the Turgutlu granite has a $14.3±0.8Ma$ core region and $11.5±0.8Ma$ rim, whereas the youngest Salihli monazite grain (⁠$9.6±1.6Ma$⁠) is located on the outer edge of an altered plagioclase crystal. These younger ages may record fluid-rock interaction driven by episodes of deformation as the granites were subjected to extension along the AD [33, 34]. Hornblende from the Salihli granite yields a ⁴⁰Ar/³⁹Ar age, although clearly affected by excess argon, of $19.5±1.5Ma$⁠, whereas biotite is $12.2±1.5Ma$ [18, 19]. Rare-earth-element (REE) poor overgrowth rims on REE-rich titanite cores in the Salihli granite are 14-15 Ma, and are thought to time the onset of ductile extension [32]. A single biotite ⁴⁰Ar/³⁹Ar age of $13.1±0.2Ma$ is reported from the Turgutlu granite [19]. Zircon from the Salihli granite yields (U-Th)/He ages of 2 Ma to 4 Ma. Assuming a geothermal gradient of 40°C/km, these ages time exhumation from 4 to 5 km depth [36, 37].

The Koyunoba, Eğrigöz, and Alaçam granitoids are exposed in the Gördes Massif (Figure 1(b)). The crystallization age of these three peraluminous transitionary granite to granodiorite plutons is reported between $30.0±3.9Ma$ and $14.7±2.6Ma$ and based on zircon U-Pb ages [38–40]. Development of a mylonitic fabric has been estimated by dating biotite using Rb-Sr (⁠$18.77±0.19Ma$ to $20.17±0.2Ma$⁠; [41]), K-Ar (⁠$19.9±0.7Ma$ to $20.4±0.6Ma$⁠; [42]), and ⁴⁰Ar/³⁹Ar (⁠$20.19±0.28Ma$⁠; [43, 44]). Dating muscovite using K-Ar and ⁴⁰Ar/³⁹Ar ages have reported similar time frames (⁠$18.6±0.7Ma$⁠; [42];$22.86±0.47Ma$⁠, [44]). Zircon and apatite fission-track and apatite (U-Th)/He methods time passage of these rocks into the He retention zone at between ~25 Ma and ~19 Ma [45]. Two Gordes Massif monazites in allanite-monazite reaction relationships are $29.6±1.1Ma$ and $27.9±1.0Ma$ and suggest that Cenozoic extension in the Gördes Massif, and possibly the entire Menderes Massif, might have begun in the Late Oligocene [22].

This study focuses on nine garnet-bearing amphibolite-facies metapelitic schists from the CMM, two granitoid samples from the Alaşehir detachment [33, 34], and two granitoid samples from the Gördes Massif [39] (Figure 1, Table 1). These samples were chosen because they are considered strong candidates to explore the Cenozoic metamorphic history of the CMM (which is poorly constrained) using a garnet chemical composition-based approach. In addition, we aim to determine trends in Neogene extension/exhumation along two prominent E-W detachment faults mapped across the CMM and Gördes Massifs. Of the metamorphic samples, six are from the Bayındır nappe (MM03-22, 23, 32, 33, 38, and 48; note that MM = Menderes Massif, ## being the specific sample number). Three samples were collected from the Bozdağ nappe (MM03-26, 27, and 28). The protolith is likely Precambrian age sediments first metamorphosed during Pan-African activity. However, the Bozdağ nappe also reportedly contains Triassic granitoid intrusions which were metamorphosed during Alpine collision (c.f., [23]).

Zircon or monazite ages from two of the granitoid samples have been previously reported. Gördes Massif sample WA12 from the Eğrigöz Pluton yields five zircons that range in age from $22.3±1.1Ma$ to $19.0±1.4Ma$ (²³⁸U-²⁰⁶Pb, [39]). Sample AT17 from the Koyunoba Pluton was not dated, but a sample collected nearby yields zircon ages from $26.6±2.5Ma$ to $14.7±2.6Ma$ [39]. Note that the youngest zircons in both rocks have deformation textures, including an annealed fracture in CL. Turgutlu granite sample EB06 yields three Th-Pb monazite ages that average $15.2±1.2Ma$ [34]. Salihli sample EB05 has not been previously dated using monazite or zircon, but another sample from the pluton has eight monazite ages that average $15.0±2.8Ma$ (Th-Pb, [34]).

Thermobarometric data for the MM03-## samples are reported in Ozerdem [46] and briefly discussed here. Chemical characterization of garnet, plagioclase, biotite, and muscovite was performed using electron probe microanalysis (EPMA) at Oklahoma State University. X-ray maps of garnet porphyroblasts were obtained for Mg, Ca, Mn, Fe, and Y using wavelength-dispersive spectrometers attached to a JEOL 733 electron microprobe operated at an accelerating voltage of 20 kV, a current of 150 nA on brass, and a run time of ~12 hours. Mineral chemical compositions were determined using an accelerating voltage of 15 kV and a beam current of 15 nA (details regarding standards are reported in [46]). Data were collected from grains both adjacent to and away from the chosen garnet (up to 2 mm) to gauge the range of compositional variations. The locations of each of the quantitative EPMA traverses spanning the core to the rim of each garnet were selected by referencing X-ray maps to avoid inclusions and cracks. Accompanying BSE maps collected by SEM are provided in Supplementary File S1.

$P$-$T$ estimates were also made using a garnet chemical zoning method (GZM) initially developed by Moynihan and Pattison [16]. We follow the procedure outlined in detail by Kelly et al. [47], Catlos et al. [15], and Etzel et al. [14]. This approach requires a representative bulk-rock chemical composition, here determined by fusion-inductively coupled plasma spectrometry (FUS-ICP) (Table 2), and garnet compositions. An ideal workflow for choosing a garnet containing the most complete growth history would be to first analytically determine the geometric center of the largest garnet in a sample (e.g., by CT scanning) then cut a central section exposing this point for chemical characterization. An alternative approach, which we took due to limited sample availability, was to diligently study thin sections of each sample and select the largest garnets within each. This approach does not guarantee that the geometric center is analyzed, but it, in part, helps ensure that we study garnets theoretically possessing the most complete growth history available to us. Previous studies using the GZM approach following this method produced plausible and explainable results (e.g., [14–16, 47, 48]).

The GZM approach involves first estimating core $P$-$T$ conditions by computing an isochemical phase diagram using the bulk-rock composition (Table 2) and overlying garnet isopleths. Uncertainty reported in Table 3 for core and rim $P$-$T$ estimates made by isopleth thermobarometry corresponds to the region of $P$-$T$ space over which garnet isopleths overlap; we note that these values are likely an underestimation. Modeling and construction of isochemical phase diagrams were performed using Theriak-Domino and the Holland and Powell [49] dataset with updates through 2010 in the system MnO-Na₂O-CaO-K₂O-FeO-MgO-Al₂O₃-SiO₂-H₂O-TiO₂ (MnNCKFMASHT). Although other data sets could be applied, these parameters appear appropriate for our assemblages [16]. Herein, we report core and rim $P$-$T$ estimates using isopleth thermobarometry for nine samples (MM03-22, 23, 26, 27, 28, 32, 33, 38, and 48), and estimate $P$-$T$ paths for two (MM03-22 and MM03-23) using the GZM approach.

High-resolution garnet $P$-$T$ paths were produced using the automated routine of Moynihan and Pattison [16]. Garnet compositional transects used as input into the modeling program are smoothed. As the modeling itself predicts how the garnet zoning profiles would evolve within the bulk composition, we compare the predicted $P$-$T$ path with both the smoothed and raw electron microprobe analyses to establish confidence in the $P$-$T$ estimates (e.g., [14, 15]). After the final step of the $P$-$T$ path is created, we use the final effective bulk composition and measured garnet rim composition to estimate a rim $P$-$T$ condition. Herein, effective bulk composition refers to the evolved chemical composition modified in response to chemical fractionation during garnet growth. We then compare that condition with the $P$-$T$ estimated by conventional thermobarometry and the final estimate from the GZM $P$-$T$ path.

We provide all the modeling data in the manuscript as supplementary files (S2 and S3). This data includes raw isochemical phase diagrams with complete reaction labels, the smoothed composition fit profiles for specific garnet transects, and effective bulk compositions generated by the Theriak-Domino program [50].

Monazite grains both in the rock matrix and as inclusions in garnet were dated (Th-Pb) in samples MM03-28 (Bayındır nappe), MM03-33, and MM03-48 (both from the Bozdağ nappe) using a CAMECA IMS 1270 Secondary Ion Mass Spectrometer (SIMS) at the University of California Los Angeles. The portion of thin sections containing the grains was cut out and mounted in epoxy with monazite age standard 554 (⁠$45±1Ma$⁠, [51]). Overall, we measured 48 spots on the monazite age standard, and the calibration (⁠$ThO2+/Th+=0.095208Pb/Th++0.973±0.116$⁠) reproduced its age to $45.1±2.3Ma$ (±1σ).

K-feldspar was dated using ⁴⁰Ar/³⁹Ar laser geochronology from granitoid samples WA12, AT17, EB05, and EB06 at Oregon State University. In this approach, mineral separates (180-250 μm grain size) are heated using a continuous 25 W Synrad CO₂ laser that incrementally increased from ~0.2% to 13.0% total intensity over 31-36 steps. Released gas fractions were analyzed using a Thermo ARGUS VI multicollector mass spectrometer. Samples and the Fish Canyon sanidine standard [52] were irradiated in the Oregon State University TRIGA reactor ICIT facility. Irradiation parameter $J$ for the individual samples was calculated by parabolic interpolation between measure monitors. The estimated uncertainty of this value is between 0.2% and 0.3%. All experiment data, including increment step size heating schedules, are included in the supplementary materials (Table S4). The age calculations are handled using ArArCALC v.25.2 [53].

Mineralogy is generally consistent amongst samples within each nappe (Table 1). Because of the petrological similarity within each nappe, we present descriptions by nappe instead of a sample-by-sample review; however, we do refer to specific samples that best illustrate our observations.

Bayındır nappe samples (MM03-22, 23, 32, 33, 38, and 48) are metapelites containing quartz + biotite + muscovite + plagioclase + garnet (Figures 2(a)–2(f)). Two generations of biotite exist. The more prominent biotite population is finer grained and defines the prominently observed foliation (Figures 2(b), 2(d)–2(f)). In contrast, the second population is composed of rare larger biotite lathes running orthogonal to the cross-cutting main biotite foliation (Figure 2(b)). Chloritized biotite is only rarely observed in MM03-23, MM03-33, and MM03-48. Trace rutile is visible in every sample, and apatite is present in all but two samples (MM03-33 and 48). Interestingly, ilmenite is also present in a subset of samples (MM03-33, 38, and 48). Of interest to this study, monazite was found as a matrix accessory mineral in multiple samples (MM03-33 and 48) and along cracks within garnet (MM03-33). Matrix monazites observed here appear to be associated with the foliation-defining biotite population.

In every Bayındır sample, disjunctive foliation is observed with microlithons of primarily quartz + feldspar and cleavage domains with biotite + white mica (Figure 2). The presence of two distinct generations of biotite in our samples may imply the finer-grained, prominent foliation-defining population formed during extensional deformation. Gessner et al. [29] pointed out that the dominant foliation observed in CMM rocks formed from Miocene extension-related deformation. Another plausible support of this interpretation is that garnet inclusion dispersal patterns and strain shadows suggest that garnet growth predates deformation responsible for the development of the main biotite foliation (Figures 2(b), 2(d), and 2(e)).

Figure 2 shows photomicrographs of select regions from the thin sections and demonstrate the variety of observed garnet textures. Figures 3–5 show major element X-ray maps and the position of EPMA compositional transects across selected garnet crystals. Supplementary file S1 contains BSE images associated with element X-ray maps for each sample. Observed garnets are variable in size, shape, and preservation. Garnets in Bayındır sample MM03-22 are large (>2 mm in diameter), subidioblastic, and contain abundant inclusions (quartz, plagioclase, and rare ilmenite). In contrast, those in sample MM03-23 are smaller (1-2 mm diameter), subidioblastic, and have a complicated inclusion pattern not consistent with the observed phyllosilicate foliation surrounding the grains (Figure 2(b)). Observed inclusions are primarily quartz, but apatite, zircon, and ilmenite are present. Clearly visible strain shadows containing quartz adjacent to garnet is also observed (Figure 2(b)). Garnets in Bayındır nappe samples MM03-32 and MM03-33 are subidioblastic with a $graindiameter<1mm$⁠. Sample MM03-32 garnets contain large core inclusions of quartz + biotite and are generally skeletal with only rims present (i.e., atoll textures). Inclusions within MM03-33 (quartz, biotite, and rare ilmenite) do not align with the D_A4 phyllosilicate foliation surrounding the grains (Figure 2(d)); strain shadows of quartz + biotite trending with the dominant foliation are observed. As seen in Figure 2(e), garnet in sample MM03-38 is inclusion-rich containing quartz, plagioclase, biotite, apatite, and zircon. Again, inclusions do not trend with the dominant foliation. Sample MM03-48 contains garnets no larger than ~1.5 mm, with most <1 mm in diameter. These subidioblastic (observed in mica-rich domains) to idioblastic garnets (exclusively observed in quartz-rich domains) contain inclusions of quartz, zircon, biotite, and ilmenite near the rims.

Samples from the Bozdağ nappe (MM03-26, 27, and 28) are metapelites containing quartz + muscovite + garnet + staurolite + ilmenite + chlorite + apatite but lack biotite and plagioclase. Rarely, chlorite was observed in each sample, adjacent to garnet. Other accessory minerals observed include rutile (MM03-27 and 28), hematite (MM03-26 and 28), tourmaline (MM03-26), and monazite (MM03-28). Monazite is found as a matrix accessory mineral, as an inclusion within staurolite, and as a fracture filler in garnet. The dominant foliation of Bozdağ nappe samples is defined by muscovite and matrix staurolite. We note that rutile rims on ilmenite are occasionally observed. Garnets here are subidioblastic to xenoblastic, 2-4 mm in diameter, and contain an abundance of quartz inclusions, but also zircon, monazite, and ilmenite inclusions. MM03-27 and MM03-28 garnets contain large (0.5-1 mm) inclusions of staurolite (Figures 2(h) and 2(i)). Texturally, matrix staurolite has irregular edges and contain inclusions or quartz, monazite, and ilmenite. The occurrence of staurolite and its tectonic implications is discussed below.

Using Theriak-Domino, isochemical phase diagrams were produced for all samples (Figure 6; Supplementary File S3). Garnet “core” $P$-$T$ conditions were estimated for eight samples (Table 3; Figures 6 and 7), and high-resolution $P$-$T$ paths were modeled for two samples (Figure 8). Note, GZM core and rim conditions reported in Table 3 are estimates made by isopleth thermobarometry and not an estimate computed by the automated routine of Moynihan and Pattison [16]. Isopleth intersections of ±0.01 mole fraction of almandine, spessartine, pyrope, and grossular are used to infer approximate conditions of garnet growth. Results in Figures 6 and 7 are not likely the real garnet core, but the $P$-$T$ results we report from this region should be considered our closest approximation. All Bayındır nappe samples, except MM03-22, reach the garnet’s central section. All transects across Bozdağ nappe garnets and MM03-22 extend from the rim to the midcrystal region. Midcrystal sampling was done for garnets with exceptionally large grains (>2 mm diameter), as it was problematic to determine an appropriate transect due to large inclusions and fractures (see Figure 2). The X-ray element maps helped guide transect positioning (Figures 3–5). Although isopleths failed to overlap for sample MM03-32 (Figure 6(c)), core isopleth intersections did occur for the other rocks. Paths were attempted for every sample, but only those generated by Theriak-Domino are within a ±0.01 mole fraction of the observed compositions reported (sample MM03-22 and MM03-23).

The Bayındır nappe samples record an average core $P$ of $6.8±0.3kbar$⁠. Modeled $T$ core conditions from the Bayındır nappe range from $530±9°C$ to $610±20°C$ (Figures 6(a)–6(f); Table 3). Note that errors are underestimated, as this represents only the area covered by the polygon in the $P$-$T$ diagram. Garnet isopleths from Bayındır nappe sample MM03-32 never overlap, but appear to converge at higher $P$-$T$ conditions compared to the other rocks (8-9 kbar and 650-700°C; Figure 6(c)). This sample has skeletal garnets or garnets with large core inclusions (Figures 2(c) and 4(a)), so this result should not be unexpected. Bozdağ nappe samples record an average $P$ of $6.1±0.2kbar$⁠, and similar $T$ conditions as the Bayındır rocks (580-625°C). All core conditions match the observed mineral assemblages. For example, isopleths in the Bozdağ samples overlap within the staurolite stability field, consistent with petrographic observations (Figures 2(g)–2(i), Figure 7).

Garnet $P$-$T$ paths were completed for Bayındır rocks MM03-22 (⁠$n=1$ path) and MM03-23 (⁠$n=2$ paths from the same crystal). The quality of each path is evaluated by comparing the modeled garnet chemical zoning to the observed chemical zoning (Figures 3(c) and 3(d)) and constructing an isochemical phase diagram using the final (garnet fractionated) effective bulk composition to gauge if rim isopleths intersect (Figure 8). For sample MM03-22, the garnet grew during a continuous increase in pressure with increasing temperature from ~6.5 kbar and 570°C to 7.5 kbar and 590°C. While minor pressure fluctuations are observed, they are <100 bars, and therefore, within uncertainty. This path cannot be considered a complete record of the garnet’s growth history as the microprobe transect may not have intersected the actual core. Sample MM03-23 produced two paths from the same garnet crystal. Both are nearly identical isobaric (7.1-7.3 kbar) paths that record growth over a 40°C increase in temperature from ~530°C to ~570°C.

Although we could not model the $P$-$T$ paths for the remaining rocks, we used the rock bulk composition (Table 2) and rim compositions to generate garnet rim isopleth intersections and $P$-$T$ conditions. Garnet rim compositions yield isopleth intersections for all samples, except MM03-28 (Table 3). These rim conditions are reported in Table 3.

Sixteen monazite grains from three garnet schist samples (MM03-28, 33, and 48) were dated in situ (Table 4; Figures 8 and 9). Results are consistent with other Oligocene-Miocene monazite ages reported throughout the Menderes Massif (e.g., [14, 22]). The average monazite age for Bayindir nappe sample MM03-33 is $24.7±4.3Ma$ with an MSWD of 0.9, whereas Bayindir nappe sample MM03-48 averages $27.3±2.4Ma$ with an MSWD of 3.0 (Table 4). Three monazite grains in MM03-33 are found as inclusions in garnet, and two of these are in contact with quartz and biotite and are adjacent to cracks (Figures 9(a)–9(f)). The oldest inclusion in garnet is $28.8±6.4Ma$⁠, and the youngest inclusion is $22.0±5.0Ma$⁠. One matrix grain dated in MM03-33 is found adjacent to biotite and is $26.5±1.5Ma$ (Figure 9(h)). A second matrix monazite associated with the dominant foliation is $20.6±3.3Ma$ (Figure 9(g)). Only three monazite grains were dated in sample MM03-48, which range in age from $31.5±2.7Ma$ to $22.8±2.4Ma$⁠. All of these grains are found in the matrix associated with biotite lathes (Figures 9(i)–9(k)).

In Bozdağ sample MM03-28 (average $31.2±2.5Ma$⁠, $MSWD=0.9$⁠), monazite grains are found primarily as inclusions in or adjacent to staurolite. Inclusion ages within staurolite range from $35.8±3.0Ma$ to $28.0±2.5Ma$ (Figures 10(a)–10(e)). One inclusion in garnet is $30.7±1.9Ma$ (Figure 10(e)). This monazite is part of a linearly arranged group of grains that extends from the garnet rim to the matrix quartz and staurolite (Figure 10). The pattern of grains closely parallels a rutile lathe that also impinges on the same garnet rim. The monazite inclusion in garnet overlaps within uncertainty with the oldest matrix monazite in the sample (⁠$35.8±3.0Ma$⁠). Overall, Bozdağ nappe monazite ages are statistically indistinguishable from Bayındır nappe monazite ages, but we do point out that the oldest inclusion within staurolite is 21.8-5.8 m.y. older than the youngest Bayındır nappe monazite, depending on uncertainty.

K-feldspar extracted from the Gördes Massif (samples: WA12, Eğrigöz Pluton; AT17, Koyunoba Pluton) and CMM (samples: EB05, Salihli granite; EB06, Turgutlu granite) were dated. The mineralogy, geochemistry, and textures of granitoids from the CMM and Gördes Massif have been described in detail by others ([34, 39] for these samples; [54, 55]). Igneous rocks from the CMM are peraluminous granodiorites (determined from the analysis of 20 igneous samples), whereas Gördes Massif plutons are peraluminous granodiorites and granites. Based on the modified alkaline lime index (MALI), the Salihli granite is calcic, and the Turgutlu Granite is more alkali-calcic. This is consistent with the observed mineralogy in that the Turgutlu granite has a higher proportion of K-feldspar relative to plagioclase when compared to the Salihli granite. Based on their MALI values, both the Koyunoba and Eğrigöz Plutons are dominantly calc-alkaline.

Figure 11 reports the K-feldspar ⁴⁰Ar/³⁹Ar ages and inverse isochron ages from the four Menderes Massif granitoids. Both Gördes Massif samples have similar degassing behavior, returning effectively flat spectra, excluding the initial 2-3 steps. Both Gördes Massif granites are Miocene age (~20 Ma) and yield well-behaved inverse isochron data that shows mixing between radiogenic and atmospheric components (Figures 11(b)–11(e)). The CMM granites are more complex, with ages that steadily increase as heating progressed. K-feldspar from sample EB06 (Turgutlu granite) steadily increases in age from $10.62±0.03Ma$ to a maximum weighted mean average age of $14.06±0.03Ma$⁠, with similar inverse isochron suggesting an age of $13.66±0.29Ma$⁠. The total gas age for this sample is $13.36±0.2Ma$ (Figures 11(a) and 11(d)). Sample EB05 (Salihli granite) increases in age from $3.27±0.10Ma$ (step 3, 0.5% ³⁹Ar released) to a maximum of $6.05±0.09Ma$ (step 33, 96.6% ³⁹Ar released). A plateau age could not be estimated for this sample, but two inverse isochron ages from different degassing steps are calculated (⁠$3.02±0.09Ma$ for the initial 19 steps and $3.29±0.22Ma$for the final 20-31 steps).

Etzel et al. [14] reported high-resolution garnet-growth $P$-$T$ paths for six garnets from the Çine and Selimiye nappes in the Çine Massif (Figure 1) using the same GZM approach (Figure 12). In the Çine Massif, two distinct $P$-$T$ shapes emerge, a burial-shaped path for Selimiye nappe rocks (from 520°C to 585°C and 6 to 7.5 kbar), and an N-shaped path (i.e., buried, partially exhumed, then reburied) for Çine nappe rocks. The N-shaped paths suggest that garnet growth occurred from ~520°C and ~5.5 kbar to ~575°C and ~7.0 kbar, but an ~1 kbar drop in pressure develops between 550°C and 565°C. To explore the driver for garnet growth along such a path, thermal-kinematic models were ran and suggested that the N-shaped path is consistent with an episode of denudation between two periods of thrust faulting.

Unlike the Çine and Selimiye nappe garnets, garnet-bearing rocks from the CMM do not show N-shaped $P$-$T$ paths, and GZM $P$-$T$ paths could not be generated for the majority of the CMM garnets. Only two samples produced $P$-$T$ paths (MM03-22 and 23). One notable difference between CMM and Çine Massif garnets is the lack of compositional zoning from the core to the rim. Selimiye and Çine nappe garnets observed in the Çine Massif retain zoning (e.g., [14]). Only garnets in CMM samples MM03-22 and MM03-23 have noticeable compositional differences from the core to the rim. Based on this observation, garnets in the Bayındır nappe samples, excluding MM03-22 and 23, nucleated near the end of metamorphism and/or had a short growth duration. Alternatively, these garnets’ initial chemical gradients relaxed via intracrystalline diffusion because of an exposure to $T>600°C$ for a protracted period (>5 m.y.) [56, 57]. The lack of composition zoning in most samples, particularly those with $radii<0.5mm$⁠, and given observations made by others that minor to significant diffusion is anticipated in $crystals<2mm$ experiencing a maximum $T>600°C$ suggests that the latter explanation is likely responsible for the lack of chemical zoning in CMM garnets (e.g., [56, 58–60]). While alternative explanations for the lack of compositional zoning exist (i.e., late-stage nucleation or nutrient availability), our modeling results strongly support intracrystalline diffusion as the driving cause for homogeneous chemical zoning. Thus, this is our favored explanation. If major cation diffusion has modified compositional zoning to the point of flattened profiles, as is the case for most CMM garnets (Figures 4 and 5), it is expected that the GZM approach would fail to generate $P$-$T$ paths.

Ultimately, the high-resolution Theriak-Domino $P$-$T$ paths generated here only approximate how a garnet with a specific type of compositional zoning would behave in a system of a specified bulk composition that evolves during its growth [15]. Garnets with complex zoning profiles, those with sharp chemical gradients over short distances, those modified by diffusion, or rocks that experienced significant changes in bulk composition over their growth history are not useful candidates for the procedure. Ideal samples are those with garnets that preserve prograde and gradational core-to-rim zoning profiles. Chemical variability in zoned garnets is a consequence of changes in $P$ and $T$ during a crystal’s growth, and incremental modification of chemical zonation would lead to small changes (10-20°C and <500 bars) in $P$-$T$ estimates and, potentially, a deviation from prograde growth conditions. Despite these modifications, conditions are still obtainable, but what they represent in terms of growth history is unclear. Figures 4 and 5 show that the majority of garnets analyzed here have flat compositional profiles. Although they could not be modeled to generate high-resolution $P$-$T$ paths, these types of samples are still useful in that by exploring the reason for their failure, they may provide clues into their broader tectonic history.

We estimate $P$-$T$ conditions near the rim and center portions of the CMM garnets (Figures 6 and 12(a)). By comparing the core and rim conditions for each sample, we find that they generally fall within a similar range of ~575°C to 640°C and 6 kbar to 7.5 kbar, considering the uncertainty in conditions. However, the shape of the $P$-$T$ path from sample MM03-22 parallels that of Selimiye nappe garnets (Figure 12(b)). If we connect the isopleth intersections from the core to the rim (arrows in Figure 12(a)), trajectories for three of the five samples that failed to yield high-resolution $P$-$T$ paths (MM03-38, 27, and 33) also parallel this path. Due to the level of uncertainty inherent in all thermobarometric approaches, the physical significance of the reported conditions at the crystal rims remains to be explored. However, the general agreement with other $P$-$T$ path estimates implies that the GZM method may be capable of estimating the relative conditions that a partially chemically modified crystal experienced at some point during its growth history.

The fact that diffusion has occurred provides clues into the physical environment in which the samples resided. CMM metamorphic rocks were likely heated to a $T>600°C$ and remained at these temperatures for a protracted length of time (>5 m.y.) and/or slowly cooled. Crystal size may be an influencing factor in some cases. For example, garnet analyzed in Bayındır nappe sample MM03-23 is >2 mm in diameter and retains compositional zoning, whereas Bayındır sample MM03-32 is <0.5 mm in diameter and shows no zoning (Figures 2–4).

Despite the significant diffusional modification, core (midcrystal) garnet isopleths intersected for all samples except MM03-32 (Figure 6). We do not anticipate a core isopleth intersection for this sample because this garnet’s core is occupied by large inclusions of quartz and biotite (Figures 2 and 4). Garnet core and rim isopleths for the other samples intersect in stability fields consistent with observed mineral assemblages (Figures 6 and 7). The intersections also yield conditions similar to those reported for other garnet-bearing rocks from these nappes in the CMM [7, 10, 12]. Bayındır nappe samples have the highest estimated average GZM $P$ (⁠$6.8±0.3kbar$⁠), consistent with their structural location beneath the Bozdağ nappe. The average GZM $T$ for Bayındır rocks (⁠$586±16°C$⁠) is less than the structurally higher Bozdağ samples but is within uncertainty (±12°C) typically seen with thermodynamic approaches.

Because of diffusional modification, the GZM $P$-$T$ conditions we report should not be considered as representing “true” peak $P$ or $T$ garnet-growth conditions experienced by the samples (e.g., [56–58]). Ultimately, the GZM results predict what a garnet of specific composition and zoning would record as it grew within a rock of a particular bulk composition using the solution models we selected. The $P$ values are our closest approximation to the actual conditions, as Ca cation diffusion is slower than Mg and Fe (e.g., [57, 61]).

In the study of Etzel et al. [14], confidence in the Selimiye and Çine nappe $P$-$T$ conditions and paths could be achieved as garnets from the same tectonic nappe collected >20 km apart yield similar $P$-$T$ paths in both shape, magnitude, and conditions. However, in this study, samples MM03-22 and MM03-23, collected within 50 meters of one another, yield different shapes (Figures 7 and 10). The isobaric path shown by sample MM03-23 is more consistent with growth at a constant depth, whereas an adjacent sample suggests that garnet growth occurred during burial.

We speculate that several options may explain the disparity. These garnets may record distinct garnet-growth events separated in space and time. Sample MM03-48 contains garnets of distinctly different shapes and zoning patterns [33], and the garnet analyzed in this rock is the only sample that shows a decrease in $T$ from the core to the rim (Figure 12(a)). Garnets from the CMM have Cambro-Ordovician monazite inclusions [22]. All of these factors are consistent with the hypothesis that MM03-48 represents growth in a different environment compared to the other rocks. A possibility exists that some garnets in the region grew during an entirely different orogeny but were juxtaposed during thrusting associated with Main Menderes Metamorphism. Previously reported geochronological evidence strongly supports the hypothesis that rocks in the CMM experienced two distinct metamorphic events (e.g., [22, 62]). Although the rocks have similar mineralogy, garnets from samples MM03-22 and 23 differ in shape, size, and texture (Figures 2(a) and 2(b)), also consistent with this scenario. This is difficult to demonstrate texturally, however, as the dominant foliation in both rocks formed during Miocene extension (see geological background).

Another option is that the rocks experienced the same metamorphic event but recorded different portions of their tectonic history. Here, the garnet $P$-$T$ path for MM03-23 only captures a later growth history over a narrow $P$-$T$ interval, while garnet in MM03-22 records a more complete metamorphic history because it nucleated earlier in metamorphism. Samples MM03-22 and 23 differ in bulk-rock composition (mainly by ~3 wt % FeO, Table 2), and would have initiated garnet growth at different conditions. Alternatively, the garnet in MM03-23 is not a true exposure of the crystal core, that is, the thin section cuts obliquely to the geometric center, thus only exposing the midrim to the rim. However, this is less likely given that its crystal diameter is comparable to the other largest crystals in the sample.

Garnet nucleation and growth sufficiently overstepping the thermodynamically modeled garnet-in line should also be considered here as our estimates suggest overstepping may have occurred (Figure 6). Although a complete discussion is beyond the scope of the contribution, estimating $P$-$T$ paths for samples nucleated after considerable overstepping has been shown to produce results at odds with true growth conditions [63]. Therefore, overstepping may, in part, explain why our paths take different shapes. Ultimately, however, the relative range of the$P$-$T$conditions that we report generally agree with one another and with other estimates made in the Menderes Massif by GZM [14].

Staurolite is observed in all three Bozdağ nappe samples both in the matrix and as inclusions within garnet. Additionally, biotite is conspicuously absent in these three samples. Such observations could potentially be explained by the following reaction: $biotite+aluminiumsilicate+garnet+H2O=staurolite+muscovite+quartz$ (e.g., [61, 64]). Given the presence of staurolite inclusions within garnet, however, a later period of garnet growth must have occurred. Further, the observation of staurolite grains with irregular edges also implies that the mineral was partially dissolved. The alternating stability between garnet and staurolite has been noted in other metamorphic terranes; for example, see[47, 65]. Both studies were able to use this observation to deduce the maximum $P$ of metamorphism and metamorphic reaction histories. Isochemical phase diagrams for each sample here demonstrate that the consumption of staurolite could have led to a later period of garnet growth at 640-650°C and 8.5-10 kbar via the following reaction: $garnet+staurolite+muscovite+kyanite+quartz+rutile+H2O=garnet+muscovite+kyanite+quartz+rutile+H2O$⁠. The products of this reaction are generally consistent with petrological observations (staurolite appears to be partially consumed and as inclusions within garnet), thus suggesting that metamorphism may have achieved a maximum $T$ of at least 650°C and a maximum $P$ of between 8.5 and 9.5 kbar.

The age of prograde Cenozoic metamorphism for Menderes Massif rocks is speculated to be Eocene to earliest Oligocene, based on Lu-Hf garnet ages from schists from the Çine Massif (⁠$42.6±1.9Ma$ and $34.8±3.1Ma$⁠; [66]) and accessory mineral ages from Gördes Massif gneisses (⁠$42.3±9.1Ma$ (allanite U-Pb) and $33.4±2.6Ma$ (monazite U-Pb); [67]). Some of the oldest monazite Th-Pb ages from Bozdag nappe sample MM03-28, which are found as inclusions within staurolite, partially overlap these estimates (⁠$35.8±3.0Ma$⁠, $34.4±2.6Ma$⁠) (Table 4). Although allanite is not observed in our samples, it has been suggested that monazite in the Menderes Massif precipitated at the expense of allanite (a rare-earth element silicate) based on reaction textures (e.g., [14, 22, 67]). The appearance of monazite has been interpreted as occurring during Early Oligocene peak metamorphism or due to Miocene-Pliocene regional extension [22, 67].

Monazite inclusions in garnet can be armored from Pb loss and help constrain the timing of garnet growth (e.g., [68]). However, many monazite grains in these samples occur along cracks in garnet, and they yield some of the youngest ages (Figures 8). These ages are likely postgarnet growth and are consistent with the timing of motion along the Alaşehir detachment. The youngest grains attest to this process, as some of the Miocene ages (e.g., $20.6±3.3Ma$⁠, $22.0±5.0Ma$⁠, and $22.8±2.4Ma$⁠) are consistent with the timing of extension throughout the Menderes Massif (e.g., [11, 22, 32, 35, 69]). In sample MM03-33, the $22.0±5.0Ma$ grain is found as an inclusion in garnet but is in contact with quartz and biotite near a microcrack (Figure 8). The textural relationship suggests that it may have been initially an allanite inclusion, but fluid-induced retrograde alteration involving the allanite, and plausibly apatite, formed monazite during extension. The average age of all monazite grains in this sample is Miocene (⁠$24.7±4.3Ma$⁠), a time also consistent with extensional tectonics, as opposed to the shortening associated with Main Menderes Metamorphism. The average monazite in sample MM03-28 is early Oligocene, and a $30.7±1.9Ma$ grain is present as a rim inclusion in garnet. The age may record the transition from regional shortening to Oligocene-Miocene extension. Within this same sample, the oldest monazite associated with staurolite is $35.8±3Ma$ and may possibly time late prograde metamorphism. In sample MM03-48, we see a range of monazite ages from the early Oligocene to Miocene (Table 4, Figure 8). The monazite ages may time peak prograde metamorphism, but overall appear to reflect postgarnet growth, consistent with the onset of extension.

The Eğrigöz and Koyunoba Plutons in the Gördes Massif experienced a rapid Miocene cooling history, as suggested by their flat K-feldspar ⁴⁰Ar/³⁹Ar age spectra and total gas ages of $20.37±0.04Ma$ and $20.02±0.03Ma$ (Figure 11). Catlos et al. [39] report ²³⁸U-²⁰⁶Pb zircon ages from WA12 that range from $22.3±1.1Ma$ to $19.0±1.4Ma$⁠, consistent with rapid Miocene crystallization. Zircons from a Koyunoba Pluton sample collected near the one dated in this study are $26.6±2.5Ma$ to $14.7±2.6Ma$⁠. If the Koyunoba Pluton crystallized overall between 30 and 20 Ma, as has been previously suggested [38, 39], then it cooled to temperatures below the K-feldspar argon retention zone (i.e., closure temperature) within 10 m.y. of the onset of crystallization. Assuming a closure temperature of K-feldspar within the range of 200-400°C [70] and a crystallization temperature between 750°C and 790°C (zircon saturation temperatures; [39]), the plutons rapidly cooled at rates as high as 275°C/m.y. This rapid cooling was likely a response to shallow emplacement of plutons [54] and/or extension-driven exhumation along the Simav detachment (Figure 1). This proposed rapid cooling/exhumation history in the Gördes Massif is further supported by apatite fission-track ages ranging between ~22 Ma and 14 Ma (see review by [23]). With respect to a regional tectonic model, rapid Early Miocene cooling is the first pulse of exhumation in the region related to swift denudation along the Simav detachment [11, 23].

In the CMM, the Salihli and Turgutlu granites exposed along the Alaşehir detachment (Figure 1) appear to have undergone a more protracted cooling history based on the shape of their age spectra (Figure 11). Catlos et al. [34] report monazite ages from Turgutlu sample EB06 that average $15.0±1.7Ma$⁠, which is 1-2 m.y. older than its maximum age, and ~5 m.y. older than its youngest ⁴⁰Ar/³⁹Ar age of $10.62±0.03Ma$⁠. Monazite from Salihli granite dated in that study (sample CC20) also averages $15.0±2.8Ma$⁠, but the K-feldspar ⁴⁰Ar/³⁹Ar ages from sample EB05 are ~10 m.y. younger with a total gas age of $5.05±0.02Ma$⁠. This K-feldspar also shows two distinct inverse isochron ages at ~3 Ma, consistent with its youngest step age.

The K-feldspar ages reported here correlate well with a pulsed cooling/exhumation model of the Alaşehir detachment [32]. In this model, the Turgutlu K-feldspar ⁴⁰Ar/³⁹Ar age represents a 12-14 Ma pulse of exhumation following crystallization at ~15 Ma. The Salihli K-feldspar ages are consistent with renewed cooling at 5-3 Ma. A pulsed model of exhumation is consistent with that proposed by Ring et al. [11] and further refined by Gessner et al. [23]. This model is based upon a bimodal distribution of apatite fission-track ages that show a northward younging trend from the core of the CMM toward the Alaşehir detachment.

Based on our new $P$-$T$ estimates, monazite Th-Pb ages, and K-feldspar ⁴⁰Ar/³⁹Ar ages, we propose the following tectonic evolution of the Menderes Massif from Eocene to present. During the Cenozoic, Barrovian-style prograde metamorphism in the CMM reached peak conditions by late Eocene-early Oligocene. The time frame is based on the oldest monazite Th-Pb ages and garnet Lu-Hf ages, which were reported in the Gördes and Çine Massifs (e.g., [66, 67]). Collision between Eurasia and a Gondwana continental crustal block resulted in peak Barrovian metamorphism of pelitic and crystalline rocks in each of the four Menderes nappes by the late Eocene. Two garnet $P$-$T$ paths and six garnet-growth estimates made by isopleth thermobarometry provide the best insight into the probable magnitude of relative conditions experienced during this event. Although some uncertainty exists with these estimates, garnet growth presumably began at ~525-600°C and 6-7 kbar, regardless of nappe, based upon reliable GZM $P$-$T$ estimates made elsewhere in the Menderes Massif [14]. When considering GZM rim $P$-$T$ estimates and $P$-$T$ paths, garnet growth would have terminated at ~625-650°C and 7.2-7.5 kbar. However, given the presence of staurolite within Bozdağ nappe samples, peak $P$ may have even been as high as 8.5 to 9.5 kbar.

The region transitioned from shortening to extension by the Mid-Oligocene (<30 Ma). At this stage, fluid-mediated alteration led to the partial decomposition of allanite and/or preexisting monazite along cracks in garnet, and precipitation of monazite grains that are younger than ~30 Ma. This process is documented elsewhere in the Menderes Massif as occurring at approximately the same time [14, 22]. Extension was likely driven by rollback of the subducting oceanic lithosphere in the Aegean, given the correlation between it and timing of extensional faulting in the CMM (see Geological Background and Previous Work). Delamination of the lithospheric mantle in the Menderes Massif led to heating of the continental crust, which was responsible for melt generation and emplacement of Miocene granites exposed in the Central and Gördes Massifs.

The Eğrigöz and Koyunoba Plutons were soon exhumed and rapidly cooled below ~250°C by 20 Ma during the first pulse of regional cooling driven by rapid denudation as N-NNE movement along the regional Simav detachment occurred [11, 23]. In a broader regional tectonic context, this movement along the Simav detachment not only exposed the Eğrigöz and Koyunoba Plutons and much of the Gördes Massif but also facilitated differential uplift of the South Menderes Monocline which eventually led to the exposure of Çine and Selimiye nappe rocks in the Çine Massif [71].

In the CMM, the Turgutlu and Salihli granites experienced a more prolonged cooling/exhumation history. The Turgutlu granite cooled below 250°C over a 3 m.y. period between 13.5 Ma and 10.5 Ma, likely driven by tectonic extension. The Salihli granite was also partially exhumed during this period. However, our ⁴⁰Ar/³⁹Ar data suggest that it did not cool below 250°C until ~5 Ma. At this time, a combination of tectonic forces and erosional denudation acted in concert to exhume this granitic body along the Alaşehir detachment. The results suggest that the detachment did not have a homogeneous extensional history along strike and support the interpretation that this region experienced a pulse cooling/exhumation history (e.g., [11, 23, 32]).

All data generated or analyzed during this study are included in this published article (and its supplementary information files).

The authors declare that they have no conflicts of interest.

AT samples were obtained with the help of Drs. Yasar Kibici, Mehmet Demirbilek, and Okan Yildiz. We thank Andrew Laskowski, Kathryn Cutts, Claus Gessner, Uwe Ring, and two anonymous experts for insightful and constructive comments that significantly enhanced the quality of this manuscript. We especially thank Mark Cloos and Claudio Faccenna for their insightful comments and shared perspectives on our research. The authors thank the UCLA National Ion Microprobe Facility, which is partially supported by funding from the National Science Foundation’s Instrumentation and Facilities Program. The authors also thank Tim O’Brien for his assistance with argon geochronology sample preparation and data collection. This material is based upon work supported by the National Science Foundation under Grant No. 0937254. Additional funding was provided by the Jackson School of Geosciences.