Stratigraphy of the Mio-Pleistocene sequence in the Göktepe region based on the fossil record (Muğla, SW Turkey)

Özlem YILMAZ*, Talip GÜNGÖR, Mine Sezgül KAYSERİ-ÖZER, Funda AKGÜN, Serdar MAYDA, Tanju KAYA, Atike NAZİK, Yeşim BÜYÜKMERİÇ Department of Geological Engineering, Faculty of Engineering, Dokuz Eylül University, İzmir, Turkey Department of Geological Engineering, Faculty of Engineering, Muğla Sıtkı Koçman University, Muğla, Turkey Institute of Marine Science and Technology, Dokuz Eylül University, İzmir, Turkey Natural History Museum, Ege University, İzmir, Turkey Department of Geological Engineering, Faculty of Engineering, Çukurova University, Adana, Turkey Department of Geological Engineering, Faculty of Engineering, Bülent Ecevit University, Zonguldak, Turkey


Introduction
Extensional basins that develop after the closure of an ocean have a key role in understanding the spatial and temporal evolution from compressional to extensional tectonics. The basins in southwest Turkey are optimal places for studying postcollision events following the closure of the Neotethys Ocean. The development of sedimentary basins, such as the Kale-Tavas, Yatağan, and Ören basins, preserves the evidence of the transition from compression to extension of the regional tectonism (Seyitoğlu et al., 2004, Sözbilir, 2005. These basins in SW Turkey started to develop immediately after the emplacement of the Lycian Nappes over the Menderes Massif (Gürer and Yılmaz, 2002).
The stratigraphy of the Yatağan Basin was documented in detail by Becker-Platen (1970), Atalay (1980), and Hakyemez (1989) (Figure 2). The coal-bearing lacustrine and fluvial deposits in the Yatağan Basin were named as the Muğla Group and divided into the Turgut, Sekköy, Yatağan, and Milet formations from base to top. Due to the rich coal reserves, the Yatağan Basin provides a substantial amount of data to infer the paleoenvironment and paleoclimate during the middle Miocene to Pleistocene. The Turgut, Eskihisar, Bağyaka, Tınaz, and Bayır sectors are the main coal production locations in the Yatağan Basin. These coal deposits were documented in detail by Nebert (1956), Becker-Platen (1970, Benda (1971), Nakoman (1978, Ünal (1988), Seyitoğlu and Scott (1991), Nakoman and İnaner (1996), Querol et al. (1999), and Bouchal et al. (2016Bouchal et al. ( , 2017. These studies were focused on coal geology, coal quality, or coal reserve estimation. Some others are related to the sporomorph contents of the coal layers. Nevertheless, the stratigraphic position of the lignite beds is still under debate. Atalay (1980) suggested that the Turgut Formation can be differentiated from the Sekköy Formation with thick coal layers. Bouchal et al. (2016Bouchal et al. ( , 2017 confirmed the same stratigraphic position of coal layers described by Atalay (1980). However, Querol et al. (1999) and Inaner et al. (2008) put this main coal seam within the Sekköy Formation.
In this paper, we describe the paleontology and stratigraphy of the Mio-Pleistocene sequence in the Göktepe area. The area is located in the west of the Kale-Tavas Basin, in the south of the Bozdoğan Basin, and is currently separated by a mountain range from the Yatağan Basin ( Figure 1a). We present new paleontological and radiometric data that reveal the depositional environment and paleoclimatic conditions of the region. We describe palynomorph assemblages, gastropods, ostracods, and mammalian fossils that allow us to constrain the new age of the Turgut Formation. In addition, we provide the U series dating radiometric age for the Milet Formation. Based on our new data, we discuss the stratigraphy and tectonic history of the Göktepe area.

Materials and methods
We studied the basin-fills throughout detailed geological mapping at 1/25,000 scale and measured sections from the two different outcrops of the Turgut Formation. Pollen assemblages (palynomorphs), ostracods, gastropods, and mammal bones and teeth have been collected from the lignite layers and surrounding clay levels of the Turgut Formation. We also applied U series dating for the travertines of the Milet Formation that were sampled from a travertine quarry at the Kepez Hill.

Paleontological analysis
We studied the pollen assemblages of five samples from the Sungur location and 38 samples from the lignite seams, which were collected from 100 m below the surface in a coal mine and from the drilling cores around Çamoluk village. The pollen analyses were performed at Dokuz Eylül University, İzmir, Turkey, using standard palynological techniques (e.g., Batten, 1999). We treated the samples with HCl and HF to remove the carbonates and silicate minerals, respectively. The separation of the palynomorphs was performed by centrifugal action, using ZnCl 2 as a heavy liquid.
The ostracod analyses were performed at Çukurova University, Adana, Turkey. Claystone samples of 100 g were disaggregated with 10% H 2 O 2 and stored for 24 h. The samples were then washed with pressurized water through 100-and 150-mm mesh sieves and dried at 50 °C. The ostracod contents of residue materials were examined with a binocular microscope. Scanning electron microscope (SEM) images were taken at Muğla Sıtkı Koçman University on a JSM 7600F field emission scanning electron microscope (JEOL, Japan) at an accelerating voltage of 15 kV. The ostracod samples were placed on specimen stubs with double-sided adhesive carbon tape and were coated with gold (Emmitech K550, UK).
The gastropods and the mammal bones were studied at Zonguldak Bülent Ecevit University, Zonguldak, Turkey, and the Ege University Natural History Museum (EUNHM), Bornova, İzmir, Turkey respectively. 2.2. U series dating U series dating was applied to eight micritic, nonporous samples from travertine deposits of the Milet Formation. The analysis was performed at the GEOTOP Research Center of the University of Quebec in Montreal, Canada. The details of the analytical procedure were described by Ghaleb et al. (2019). Briefly, about 1 g of travertine sample was cut using a rotary tool with a diamond disc (Deremel). Then the subsamples were weighed in Teflon beakers, in which an accurately weighed amount of 236 U-233 U-229 Th spike was evaporated at low temperature (~60 °C) to determine U and Th isotopes by isotope dilution technique. The samples were covered with deionized water and dissolved slowly by adding HNO 3 (7 M) drop by drop. After the total dissolution, about 10 mg of the iron carrier was added and the samples were left overnight in order to ensure spike-sample equilibration. U and Th isotopes were coprecipitated with Fe(OH) 3 by adding ammonium hydroxide to the solution until pH 8-9 was reached. The precipitates were recovered by centrifugation, washed with water, and dissolved in HCl (6 M). Chemical separation of U and Th isotopes was done by anion exchange resin (AG1X8, 200-400 mesh). Thorium fractions were recovered with HCl (6 M) while the uranium with iron fractions were eluted with water. Finally, U purification from iron was done using U-Teva resin, the iron being eluted with HNO 3 (3 M) and the uranium with HNO 3 (0.02 M). For the thorium fraction, after evaporation, a purification column of 2 mL of AG1X8 was used in HNO 3 (7 M) and eluted with HCl (6 M).
Measurements of U and Th isotopes were performed using a Nu Instruments Multi-Collector via inductively Becker -Platen, 1970 Yatağan Sekköy Künar Hakyemez, 1989 Conglomerate, mudstone  U-233 U certified ratio of the spike was used to correct for mass bias for U and Th isotopes. Age calculations and detrital corrections were done using the Isoplot 4.15 software program (Ludwig, 2012).

Basement rocks
Metamorphic rocks of the Menderes Massif form the basement and extend throughout the west of the study area ( Figure 1b). The Lycian Nappes are not exposed at Göktepe, but they are found around the villages of Yeşimendere and Özlüce, near Muğla.
The Menderes Massif starts with the fusulinid-bearing Permo-Carboniferous Göktepe Formation (Önay, 1949;Erdoğan and Güngör, 2004). The formation is made up of dark brown to gray phyllites, schists, quartzites, and marbles. Calc schist-chlorite schist intercalations with rare quartzite interlayers show a strong brittle deformation. Foliations are thin and have a shallow dip angle. The Göktepe Formation is unconformably overlain by a Mesozoic succession of marbles (Önay, 1949;Konak et al., 1987;Özer and Sözbilir, 2003;Erdoğan and Güngör, 2004). In the study area, the Mesozoic sequence is composed of white to dark gray marbles and black marbles. The marbles show a prominent foliation with varying dip directions.

Turgut Formation
The unconformable Mio-Pleistocene sequence starts with the Turgut Formation (Figures 2 and 3). The Turgut Formation crops out at the lower elevations of the study area, particularly along both sides of the Sarhoş Stream and Akçay River ( Figure 1b). The Turgut Formation is made up of yellowish-gray and green, thin-to mediumbedded claystone, mudstone, sandstone, and conglomerate alternations including some thin limestone levels (Figures 4a and 4b). Conglomerate levels contain small-scale channel fills. Sandstones show mostly cross-bedding ( Figure 4b).
The lower part of the formation is exposed in a small area near Sungur village (Figures 1b and 5a). The detailed section of this outcrop is given in Figure 5b. The section starts with yellowish-orange, laminated, marl-mudstone, very thin pale green claystone and thin laminated brownish mudstone. The section continues upwards with yellowish beige mudstone, gray claystone, and siltstone. The upper levels of the section are composed of gray-colored, poorly lithified, medium-to coarse-grained sandstones and matrix-supported granule-and pebble-sized conglomerates. The sandstones and conglomerate levels show both parallel and cross-bedding. The section ends with sandstones >10 m thick. Abundant broken gastropod shells, bivalve fossils, and mammal bones are found in this outcrop (Figures 5c, 5d, and 5e). These mammal bones and gastropods have been collected and the claystone levels were also sampled for palynological analysis (Figure 5f). The fossil records give the age of late early-early middle Miocene. The details are given in Section 4.
The upper parts of the formation are made up of claystone, limey claystone, and clayey limestone alternations (Figures 6a-6f). To the north of Çamoluk village, this alternation includes two seams of lignite (Figures 6a and 6c). The thickness of the lignite seams varies along its strike and wedges out. The main lignite seam is approximately 2 m thick and contains mammal bones and teeth. This lignite layer grades upward into beige-colored, laminated claystone and dark green claystone with abundant broken gastropod shells and ostracod fossils ( Figure 6b). Dark green claystone levels include some plant remnants. On top of this layer, there are gray-beige, clayey limestone and beige claystone levels with broken gastropod shells and abundant ostracods. The above-lying second seam has been measured from drilling cores and is approximately 30 cm thick ( Figure 6c). These coal seams and surrounding clay levels were sampled for palynological analysis and gastropods, ostracods, and mammal bones have been collected from these layers. The fossil record gives the age of middle-late middle Miocene. The details are given in Section 4.

Sekköy Formation
The Sekköy Formation is poorly exposed in the west of Çamoluk village with elevations of approximately 400 m ( Figure 1b). The formation is composed of gastropodbearing, lacustrine clayey limestones. There are no lignite layers observed. The lignite layers are located in the upper part of the Turgut Formation and these coal layers gradationally pass upward to the lacustrine deposits of the Sekköy Formation ( Figure 6a).
The age of the formation is late middle Miocene (Atalay, 1980). The Sekköy Formation is conformably overlain by the Yatağan Formation ( Figure 3).

Yatağan Formation
The Yatağan Formation crops out along the hillslopes of the Çongara, Yarılgan, and Kepez hills, at elevations of above ~400 m ( Figure 1b). This formation has a distinct orange-red color and is composed of conglomerate, sandstone, siltstone, and mudstone alternation with some clayey limestone and limestone lenses (Figure 7). The formation starts with matrix-supported, poorly sorted conglomerates. These conglomerates contain subrounded, subangular clasts and alternate with medium-to coarsegrained cross-bedded sandstones. These fluvial clastics change upwards to lacustrine limestone beds.   Atalay, 1980 andHakyemez, 1989). Mio-Pleistocene sedimentary sequence covers the tectonic boundary of the Menderes Massif and Lycian Nappes along an angular unconformity.
The age of the Yatağan Formation is Turolian (upper Miocene) based on vertebrate fossils described by Atalay (1980) and Hakyemez (1989). The Yatağan Formation is conformably overlain by the Milet Formation.

Milet Formation
The Milet Formation was described by Becker-Platen (1970) as the Milet beds and revised by Hakyemez (1989) to the Milet Formation ( Figure 2). In the study area and further east towards the town of Kale (Denizli), the Milet Formation is exposed on top of plateaus around Göktepe village ( Figure 8). The Milet Formation is composed of yellowish cream claystone, marl, and travertine-type lacustrine limestones. The lower parts of the formation are graded vertically into the orange-colored conglomerates of the Yatağan Formation and pass through upward travertine-type lacustrine limestones. The thickness of the travertine is approximately 80 m and it includes plant remains and gastropod and ostracod fossils. The Milet Formation is almost horizontal in the south of the study area, but it shallowly dips to the NW (5-10°) north of Göktepe village. Atalay (1980) determined the age of the Milet Formation as upper Miocene (Turolian)-Pliocene according to vertebrate fossils from Bozarmut village (Muğla). Hakyemez (1989) also stated the age of the formation as Pliocene because stratigraphically the Milet Formation lies on top of sequences that contain Turolian mammal fossils. In this study, U series dating was used to determine the age of the travertine-type lacustrine limestones of the Milet Formation at the uppermost part of the formation. These results were first presented by Yılmaz et al. (2016). The radiometric dating results show a Middle Pleistocene age for the travertines. The details are given in Section 4.
The Milet Formation is unconformably overlain by the Sungur Formation.

Sungur Formation
The Sungur Formation is exposed in a small area between Göktepe and Sungur villages (Figure 1b). The Sungur Formation consists of brownish-red, matrix-supported, very poorly sorted, poorly lithified conglomerates and sandstones ( Figure 9). Conglomerates are made up of angular pebble-and cobble-sized clasts of schist and marble. The matrix is composed of fine-grained sand and silt.
The alluviums of the Akçay and Sarhoş rivers are the Quaternary deposits of the study area that unconformably overlay the sequence (Figures 1b and 3).

Late Early Miocene-Early Middle Miocene
The Turgut Formation is exposed along a road-cut outcrop and a drainage channel near Sungur village (Figures 1b and  5a). We collected mammal bones and gastropods from this outcrop. In addition to these fossils, five claystone samples were also taken for palynological analysis. Fossil contents of this formation are given below.

Mammalian records
The Sungur locality yielded a rhino mandible fragment referred to Brachypotherium brachypus (Figure 5e). The material is biometrically larger than Brachypotherium brachypus remains from the middle Miocene locality of Paşalar (NW Turkey) (Fortelius, 1990) as well as those from European ones. It is also larger than B. minor records from Africa (Geraads and Miller, 2013)   from Eppelsheim (MN6-9). The Sungur specimen is morphologically and metrically similar to those from the early middle Miocene (MN5) locality of Kultak (Muğla, Turkey) .

Mammalian records
Despite the small number of specimens, mammal remnants were extremely well preserved. We recorded dental remains of two proboscidean taxa at Çamoluk: a fragmentary astragal of a primitive equid, Anchitherium sp., and a single lower molar of a giant suid. For proboscideans, a lower second molar and a portion of an upper tusk can be referred to Gomphotherium angustidens, which is the most common gomphothere of Europe during the middle Miocene (Figure 6e, inset 1). Some additional molar fragments that we found also display the typical morphology of this taxon. This species is known from Anatolia, mainly from the early middle Miocene localities of Bursa-Pașalar (MN5-6), Muğla-Kultak (MN5), and late middle Miocene localities of Muğla-Sarıçay and Çatakbağyaka (MN7/8). Metrically, G. angustidens from Çamoluk falls in the size range of G. angustidens from the middle Miocene (Astaracian, MN6-8) localities of France and Germany and it is larger than the Paşalar samples, though these taxa did not show a tendency towards increasing molar size over time (Tassy, 2014). However, the well-developed posterior cingulum and bunodonty clearly differentiate the Çamoluk specimens from the Orleanian (MN4-5) forms, such as at Corcoles (Spain, MN4) and Quinta Grande (Portugal, MN4) (Bergounioux et al., 1953;Mazo, 1977Mazo, , 1996Maldonado et al., 1983;Tassy, 1985). The occurrence of a second proboscidean species at Çamoluk is supported by an upper second molar. This specimen is tentatively referred to the amebelodontid genus Protanancus based on its generally well-pronounced pseudo-anancoidy, cementodonty, and posttrite conules.
Besides the Anatolian records from the middle Miocene localities, such as İzmir-Mordoğan, Bursa-Paşalar, and Ankara-Çandır, the Çamoluk specimen mostly resembles the Protanancus material from Bulgaria-Vetren, which was also dated to the middle Miocene (Markov and Vergiev, 2010). A second lower molar represents a giant Kubanochoerine suid and confirms the presence of the enigmatic genus Megalochoerus at Çamoluk. The kubanochoerines migrated into Eurasia during MN5 and survived until MN7/8 (Pickford, 2001). The previous record from Turkey, an upper molar from Ankara-Kazan-İnönü (MN6), was thought to be an M2 of Kubanochoerus mancharensis (junior synonym of Megalochoerus humungous) (Pickford and Ertürk, 1979;van der Made, 1996) (Figure 6e, inset 2). Since the Kubanochoerine suids underwent a rapid increase in size from early to middle Miocene times, they can be used for biochronology (Pickford, 2001;Pickford and Morales, 2003). In light of this, the Çamoluk specimen fits well with the morphologic and metrical features of the Astaracian species M. humungous, although it has slightly lower range values compared to the youngest records of this taxa, collected from MN8 localities (late Astaracian) of Pakistan. This could also be interpreted as a sign of an older age than MN8 for Çamoluk. The fragmentary astragal of anchithere remains from Çamoluk is morphologically identical to the Turkish and European middle Miocene Anchitherium samples but distinguished from the early Miocene forms by its relatively larger size.

U series dating
U series analysis was carried out with eight hand samples that were taken from a travertine quarry located on Kepez Hill (Figure 1b and 8a-8c). This quarry has several benches that cut an 80-m-thick horizontal travertine sequence (Figures 8b and 8c). Eight micritic and nonporous samples were collected from the benches. The results are given in the Table. The basis of U series dating is that the CaCO 3 phase incorporates only uranium (soluble in oxidizing conditions) and not 230 Th (insoluble) during the precipitation. The uranium then produces the authigenic 230 Th by radioactive decay and the activity ratio of 230 Th/U increases from zero at the time of precipitation until reaching the secular equilibrium value around 500-600 ka. If the radioactive system remains closed (no loss or gain of any U and Th isotopes except by radioactivity), then the activity ratio of 230 Th/U allows calculation of the time since the precipitation of the carbonate phases.
Stratigraphically, the ages of our samples taken from the uppermost levels of the sequence must be younger than those from the bottom. However, the results are incompatible with the stratigraphy (Table). There may be several potential reasons for this inversion in the stratigraphic order. The first one is that our travertine samples have incorporated variable amounts of detrital contamination. This is indicated by the presence of variable amounts of 232 Th (and all other thorium isotopes). In such a case, the measured 230 Th in our samples corresponds to the sum of authigenic 230 Th that is produced by incorporation with both uranium and the nonauthigenic 230 Th related to detrital contamination. Thus, we should subtract the amount of detrital 230 Th from the total 230 Th before calculating the age of a sample. We used an average crustal correction model similar to the one used by Ludwig and Paces (2002) to obtain the corrected ages that are given in the Table. Specifically, 232 Th was used as an index and typical crustal Th/U was used for the detrital component. The activity ratio values used are 232 Th/ 238 U = 1.21 ± 50%, 230 Th/ 238 U = 1 ± 10%, and 234 U/ 238 U = 1 ± 10%. We believe that with such assigned errors for the detritus component the detrital correction should correct our sample results in a manner that covers most of the natural detritus observed on the earth's surface. As seen in the Table, the corrected ages are still not compatible with the stratigraphic order, therefore indicating that detrital contamination is not the primary cause for the incompatibility with the stratigraphic order and another process should be evoked. As mentioned above, U series dating is based on the assumption of a closed radioactive system (no loss or gain of any U and Th isotopes except by radioactivity). We think that this condition is not met in our samples, thus preventing the obtaining of reliable ages. In fact, our samples show clear evidence that a recrystallization process has occurred. This indicates a potential opening of the radioactive system and this opening will essentially affect the uranium isotopes that can be transferred without necessarily fractionation of 234 U-238 U. Deschamps et al. (2004) observed a disequilibrium of 234 U/ 238 U in Mesozoic limestone formations in the eastern Paris basin and they suggested discrete uranium mobility/transfer as a result of pressure dissolution structures (stylolites or dissolution of seams). The uranium mobility/transfer mechanism can be evoked to induce the slight disequilibrium observed in our samples. On the other hand, and given the old ages of the travertines, the U series activity ratios should be near the secular equilibrium (limit of the method: 500-600 ka). Thus, any small mobility or transfer of uranium will result in huge differences in the calculated ages due to the exponential character of the radioactivity phenomena. The precision of the dating decreases with increasing age (Geyh and Schleicher, 1990 and references therein). Whatever the process responsible for the disequilibrium observed in our samples, and given the limited number of analyzed samples, we suggest that from a theoretical point of view the samples that lost a part of their uranium will yield calculated ages older than the true age, while those that gained uranium will yield a younger age than the true one. From this perspective, one can consider that the youngest calculated age (~346 ± 19 ka) can be considered as a minimum age for this formation.
Consequently, the lacustrine carbonates of the Milet Formation were deposited in the Middle Pleistocene, most likely until ~346 ka (Table).

Göktepe fault
The Göktepe fault is a 6-km-long fault that extends along the western part of the study area. The fault is an eastward dipping high angle normal fault that strikes towards the northwest (Figures 1b and 10). The Göktepe normal fault is divided into two segments; each of them is approximately 2 km long and extends toward the north. These branches separate the Mio-Pleistocene sedimentary units from the metamorphic basement. This fault shows a growth fault geometry because the thickness of the sequences on the footwall block is distinctly thicker than those on the hanging wall block. The Millet Formation is highly inclined due to fault dragging to the north of Göktepe village. However, the uppermost levels of the Milet Formation overlap the Göktepe fault (Figures 10 and 11). Our field observations show that the limestone beds in the upper part of the formation are undisturbed at elevations of ~800 m. This location is shown with a red square in Figure 1b. Our U/Th dating results of the uppermost levels of the travertine-type lacustrine carbonates (Milet Formation) yield a minimum radiometric age of ~346 ka B.P. (Middle Pleistocene). Since these carbonates overlapped the Göktepe fault, it may imply that this fault has been not active since at least ~346 ka (Figure 11).

Paleoenvironmental evolution
Our stratigraphic, biostratigraphic, and structural data provide a basis to evaluate the depositional environment and the climatic conditions of i) late early Miocene-early middle Miocene, ii) middle-late middle Miocene, and iii) Middle Pleistocene terrestrial deposits, which have implications for the paleoenvironmental and tectonic evolution of the Göktepe area.

Late Early Miocene-Early Middle Miocene (MN4-5)
The oldest deposits of the study area are fluvial clastics (alternation of conglomerates, sandstones, and mudstones) with some lacustrine interbeds (claystone, carbonated mudstone, and marl). These lacustrine layers include some gastropod fossils, Sungur palynomorph assemblages, mammalian bones, and a mandible, which provide an age constraint of late early-early middle Miocene (MN4-5). Additionally, the gastropods and palynomorph assemblages indicate a lacustrine environment that was surrounded by broad-leaved, deciduous forests under a warm subtropical climate. Some of the palynomorphs (e.g., Botroyococcus, Acacia, Calamus, and Anemia) and gastropods tolerant to brackish conditions suggest that this lake may have had brackish characteristics.
The late early-early middle Miocene (Burdigalian (20.44 Ma)-Langhian (13.82 Ma), MN4-5) corresponds to the deposition period of the lower part of the Turgut Formation in the Yatağan Basin, which should chronologically correspond with the Eskihisar and the lower part of the Yeni-Eskihisar sporomorph associations (Benda and Meulenkamp, 1990). Studies on the base of the Turgut Formation are very limited and most studies attribute a middle Miocene age for the formation (Atalay, 1980; Akgün and Sözbilir, 2001). However, Alçiçek (2010 in fig. 2 and fig. 3) suggested an early to middle Miocene (MN4-6) age for the Turgut Formation by using fossil findings of mollusks by Becker-Platen (1970) and vertebrates by Saraç (2003). Our results provide the age constraint of MN4-5 for the deposition time of the lower part of the Turgut Formation.
The early Miocene succession in the Kale-Tavas Basin consists of the Yenidere, Künar, and Kale formations (Hakyemez, 1989) (Figure 2). Among these formations, the age of the Künar Formation is the least constrained because it lacks any paleontological findings. Hakyemez (1989) considered the Künar Formation as the fluvial continental equivalent of the marine Kale Formation and therefore suggested its age as Burdigalian. Özcan et al. (2009) described two foraminiferal horizons in the middle part of the Kale Formation and indicated a transgression referring to the Burdigalian age. Büyükmeriç (2017) used stratigraphic relations between the Aquitanian Yenidere Formation and late Burdigalian Kale Formation and proposed the age of the Künar Formation as early Burdigalian. Büyükmeriç (2017) also stated that in the late Burdigalian there was a transgression in the Kale-Tavas Basin characterized by the thick carbonate sequence of the Kale Formation. Moreover, De Graciansky (1972) proposed that a branch of the early Miocene Neotethys Ocean extended towards the Kemer barrage, i.e. the Göktepe area ( Figure 12). This transgression may have affected the deposition of the Turgut Formation in the Göktepe area and would explain the existence of the brackish-tolerant palynomorph assemblage and gastropods.
When the lower part of the Turgut Formation started to deposit in the Yatağan Basin, the Künar/Kale formations were deposited in the Kale-Tavas Basin (Akgün and Sözbilir, 2002) and the Kultak Formation was deposited in the Ören Basin .

Middle-Late Middle Miocene (MN6-7)
The middle Miocene period is characterized by fluvial and lacustrine sediments of the Turgut and Sekköy formations. The debated stratigraphic setting of the coal (lignite) in the Yatağan Basin was described in Section 1. Our fossil findings (mammalian bones, ostracods, and Çamoluk palynomorph assemblage correlated with the upper part of the Eskihisar and Yeni-Eskihisar palynomorph associations (Benda and Meulenkamp, 1990)) show that the lignite was deposited during the middle-late middle Miocene (MN6-7) time interval in the freshwater lacustrine environment. Coal-bearing sediments of the Turgut Formation were deposited under warm temperate climatic conditions. It could be said that cooling in the palaeoclimatic conditions (from subtropical to warm temperate) is observed during the deposition of the Turgut Formation in the study area (Benda and Meulenkamp, 1990). Hence, the lignite stratigraphically forms the uppermost levels of the Turgut Formation as proposed by Atalay (1980) and is followed by the marls of the Sekköy Formation Güner et al., 2017).

Middle Pleistocene
The Göktepe area is also located in the south of the Bozdoğan Graben. The N-S running Bozdoğan Graben is a product of E-W extension before the late Miocene (Ocakoğlu et al., 2014). The NW-SE striking Göktepe fault lies at the southeastern end of the graben and may therefore be related to this extension. Moreover, the fault appears to be extinct by ~364 ka (Middle Pleistocene), which is compatible with the direction change of the extension in the Early Pliocene to the NE-SW, and then to N-S extension in the Late Pliocene-Quaternary (Ocakoğlu et al., 2014).

Conclusions
The preceding description of the stratigraphy and biostratigraphy allows evaluating the paleoenvironmental conditions in the basin. The fossil evidence of the late early to early middle Miocene indicates a warm, subtropical brackish/freshwater lacustrine environment that was surrounded by broad-leaved deciduous forests that were poor in diversity but rich in conifers. The lacustrine condition implies a transgression caused by tectonic subsidence in the basin. In the case of a brackish lake, the saline conditions could be explained by a marine transgression. The fossil record of the middle-late middle Miocene lignite deposits presents a freshwater lacustrine environment under warm temperate climatic conditions. The sedimentary sequences on the hanging-wall block of the Göktepe fault are significantly thicker than those on the footwall block. This may indicate continuous subsidence in the basin until ~364 ka.

Acknowledgments
This study was carried out as a PhD thesis at Dokuz Eylül University and was funded by the university's BAP research unit (Project Number: 2014 KB FEN 026). We are very grateful to Bassam Ghaleb for helping to interpret the U/Th analysis results and we also appreciate the reviewers for their constructive comments.