On the geoarchaeology of Limyra (SW Anatolia)—new insights into the famous Lycian city and its environs

Abstract Geoarchives in ancient settlement sites and their environs bear valuable information about Holocene landscape evolution and human–environment interactions. During the last six millennia, sea‐level and coastline changes have had a significant impact on coastal settlements, some of which even had to be relocated. This paper reveals new insights into the spatio‐temporal development of the Lycian city of Limyra. Selected sediment cores were analyzed using a multiproxy approach, combining sedimentology, geochemistry, micropaleontology, and 14C dating. When the postglacial sea‐level rise decelerated, a coastal barrier and a deep lake, presumably a lagoon, evolved after the mid‐Holocene. The siltation history of the lake is complex: three coastal peat layers (mid‐4th millennium BC, end of 3rd/beginning of 2nd millennium BC, beginning of 1st millennium BC), indicate periods of semiterrestrial conditions. That they are sandwiched by lake sediments is consistent with new expansion phases of the lake, most likely triggered by coseismic subsidence. There is evidence of a former lakeshore, dated to between 1400 and 1100 BC, with an intentionally deposited layer of anthropogenic remains, now at 5.5 m below the surface. In the mid‐1st millennium BC, the lake silted up, river channels evolved, and people started to settle the area of the later city of Limyra.

The city of Limyra in southwestern Turkey is situated at the foothill of Toçak Dağı, a part of the southern Taurus Mts., today ca. 4 km inland.
Therefore, a geoarchaeological research project of the Austrian Archaeological Institute (ÖAI) and the University of Cologne (Brückner, Stock, & Uncu, 2016a, 2016bStock, Uncu, & Brückner, 2017) aimed at deciphering (a) the environmental changes, especially from the Lycian (ca. 550-330 BC) to the Late Roman (mid-1st millennium AD) periods; (b) the thickness of the settlement layers; (c) the maximum extension of the former lake; (d) the earthquake chronology; and (e) the spatio-temporal shifts in the coastline.

| STUDY AREA
The study area is located in southwestern Turkey, topographically ca. 80 km southwest of the city of Antalya, geologically at the junction between the Hellenic Arc and the Cyprus Arc. The geology to the north and northwest of the Finike plain is dominated by the Bey Dağları unit with Mesozoic to Paleogene limestones. The Antalya nappes to the east and northeast are composed of nonmetamorphic bedrock and ophiolites (Şenel, 1997). Quaternary alluvia of the Finike plain have a thickness of at least 26 m (Öner, 2013). Offshore, Eocene sediments fill the Finike Basin (shelf break at ca. 150 m water depth; then down to 3,000 m water depth) . Several alluvial fans derive from the surrounding mountains (Öner, 2013).
The Finike coastal plain at the foothill of Toçak Dağı (up to ca. 1,200 m; Figure 1) has a W-E extension of max. 31 km and a width of 7-13 km (Ergin et al., 2007; Figure 1). It evolved despite the enhanced subsidence during the Plio-Quaternary (including the offshore basin: in total~5,000 m during the last 5 million years) Hall, Aksu, Yaltırak, & Winsor, 2009). To the west, the Finike plain is bordered by the Burdur-Fethiye fault zone, which is one of the major faults in Turkey  Figure 1) generating earthquakes greater than 5.0 (Richter scale; Glover & Robertson, 1998). This explains the frequent occurrence of earthquakes that have already been reported during antiquity (Akyüz & Altunel, 2001;Karabacak et al., 2013). As for Turkey and the surrounding region, Bayrak et al. (2009) attributed the second highest earthquake hazard level to the study area. Based on Öner's (2013) research, who detected an aquatic environment in the subsurface strata, Uyanık et al. (2013) deduced a high risk of earthquake-generated liquefaction; thus, the 6.8 Mw earthquake of Finike in 1926 would have felt like a magnitude 10 Mw earthquake.
Nowadays, the ruins of Limyra are located ca. 5 km northeast of the modern small town of Finike (Borchhardt, 1993). Several karstic springs rise from the karstified Toçak Dağı immediately adjacent to F I G U R E 1 Location of Limyra in the Finike plain (source: Ch. Kurtze, 2014, modified). The locations of some of the drill cores are indicated (yellow dots) [Color figure can be viewed at wileyonlinelibrary.com] the north of the city (Bayari et al., 2011). Their waters permanently flow like perennial streams through the ruins, before forming the Limyros (Akçay) river (Öner, 2013).

| ARCHAEOLOGY
The city was first mentioned in Hittite texts (ca. 1350-1300 BC) as zumarri. Zẽmuri, the Lycian name of Limyra, can with some probability be traced back to this term (Keen, 1998). The earliest archaeological finds date to the early 7th century BC. A settlement existed in this area at the latest since the 6th century BC (Konecny & Marksteiner, 2007). The city experienced its first heyday in the 4th century BC when it became the residence city of an aspiring east Lycian dynasty under the reign of Pericles.
By then, an extensive building program took place in Limyra, during which a massive ring of city walls, including a fortification on the acropolis, was erected. The residential quarters were enlarged, including the lower slope of the acropolis hill with houses partially hewn out of the bedrock. At this time the monumental "heroon" of the dynasty was erected on the acropolis (Borchhardt, 1976). This magnificent royal tomb is a counterpart to the famous, slightly older Nereid Monument from Xanthos. It was designed like a Greek temple with four larger than life-sized caryatids at each facade. The tomb was raised up above a high podium in which the tomb chamber was situated. The five necropoleis of the Classical period, located in the direct vicinity or close-by the city, host ca. 400 tombs; this is by far the largest number of tomb buildings of all Lycian cities (Borchhardt & Pekridou-Gorecki, 2012;Kuban, 2012).
The Hellenistic and Early Roman periods are only sporadically attested in Limyra by few structures; nevertheless, due to their monumentality and the high quality of their architecture and sculptural decoration, they attest to the importance of the settlement during these epochs. The so-called Ptolemaion, located in the lower city, belongs to this group of impressive structures ( Figure 2). It was set up by the Egyptian dynasty of the Ptolemies in the first half of the 3rd century BC; due to its excellent craftsmanship, it has to be regarded as one of the most outstanding monuments from this period in Asia Minor (Stanzl, 2012;Stanzl, 2016). The most prominent building from the Early Roman Imperial period is the cenotaph of Gaius Caesar, grandson and adopted son of the Roman Emperor Augustus, who died in Limyra on 21 Febuary, AD 4 (Borchhardt, 2002;Ganzert, 1984). Its massive core of opus caementitium is preserved in the western lower city of Limyra. The pedestal was originally faced with a 60 m long frieze of marble plaques with life-sized figures, displaying scenes from the life of Gaius Caesar.
Besides the theater, built in the 2nd − 1st century BC and partly rebuilt after a devastating earthquake in AD 141, other buildings-such as huge bathing complexes, broad colonnaded streets, bridges, and so forth-also illustrate the flourishing urban life of Limyra in the Roman Imperial period (Seyer, 2016).
During Late Antiquity and the Byzantine period, the city was an episcopal see. Between the 4th and the late 9th centuries AD, the names of six bishops are known. In the late 5th/early 6th century AD, the city was divided into two parts by means of two separate wall circuits. Amongst the monuments of that time, the three church buildings are noteworthy (Jacobek, 1993;Peschlow, 1984;Pülz, 1996), as well as a building discovered in 2012 that might have been the synagogue of the city (Seyer, 2014;Seyer & Lotz, 2014).

| Fieldwork
Our research is based on 21 sediment cores (Lim 1-Lim 21) with a max. length of 12 m. They have been retrieved from Limyra's city center and its environs with a vibracorer Cobra pro (Atlas Copco) in semiopen augerheads (outer diameters: 6, 5, 3.6 cm; Figure 2). Representative samples were taken from different layers. Four master cores were chosen from the study area and analyzed in detail.
All sites were leveled with a DGPS (Topcon) (precision: ±2 cm in all three dimensions). The geoarchaeological research design is according to Brückner and Gerlach (2020).

| Laboratory analyses
In the laboratory, the sediment samples (59 in total with a resolution of up to 25 cm intervals per drill core) from the four master cores were dried, carefully pestled, and sieved (2 mm). For grain size analysis, they were treated with hydrogen peroxide (15% H 2 O 2 ) to destroy the organic matter and with sodium pyrophosphate (Na 4 P 2 O 7 , 46 g/l) for dispersion. The samples were measured with a laser diffraction particle size analyzer from Beckman Coulter (LS-13320, 116 classes from 0.04 to 2,000 μm) and statistically analyzed with the Gradistat software (Blott & Pye, 2001).
A Bartington MS2 sensor was used for measuring the magnetic susceptibility (MS). To determine the C/N ratio, the samples were homogenized with a ball grinder (Retsch MM 400), weighed in tin boats and measured with a vario EL cube (Meyers & Teranes, 2001).
After the determination of total carbon (TC) and nitrogen (N), the total inorganic carbon (TOC) was measured after treating each sample with 10% HCl. X-ray fluorescence measurements were carried out with a portable XRF device (Niton™ XL3t GOLDD+).
Selected samples from different layers were chosen for microfaunal species determination (ca. 5 g sediment; sieves with mesh sizes of 63 and 100 μm). Ostracods were determined according to Meisch (2000). However, only one sample each from Lim 4 and Lim 14 contained ostracods.

| Statistical analyses
Principal component analysis (PCA) was done with the Past software (version 3.11) for differentiating the facies (Hammer, Harper, & Ryan, 2001). All values of the four master cores were standardized and standard deviation calculated. Ten components-Fe, K, clay, sand, C/N, sorting, mean, CaCO 3 , Ca/Sr, magnetic susceptibility-were used for the PCA.

| RESULTS
Four master cores (Lim 4, 12, 13, and 14) were chosen from Limyra and its environs to decipher the different facies of the Finike plain.  with abundant ceramic, brick and bone fragments, charcoal, as well as core-filling layers of angular stones (mostly limestone). The C/N ratios differ strongly (0.9-3.3 and 2.2-6.8, respectively). According to the 14 C chronology, the central part of the uppermost layers dates from the 6th to the 3rd centuries BC (Lim 4/19 Hk, 14 C age 2350 ± 35 BP, 536-368 cal BC; Lim 4/14 Hk2, 14 C age 2300 ± 39 BP,

412-208 cal BC).
Lim 12 is situated between the western and the eastern city and a processed stone were found (Figure 4). Then follows peat that is overlain by silts with many stones and ceramic fragments.
At a depth of 1.48 m a.s.l., a plant seed was dated to the beginning of the 1st millennium AD (Lim 12/16 Pf, 14 C age 1917 ± 31 BP, cal AD 4-208).

| Drill core Lim 14 in the western city
Lim 14 is located in the western city close to the cenotaph of Gaius Caesar (surface: 6.71 m a.s.l., max. depth: 10 m; Figures 2, 6 and 7). At its base, alternating organic-rich clayey silts (mean 5-18 μm, C/N 5-8) and calcium carbonate-rich yellowish-brown sands (mean 160-210 μm, C/N 2-5) occur. The transition to the overlying strata dates to 1874-1626 cal BC (Lim 14/8 HK, 14 C age 3417 ± 35 BP). In this layer, few ostracods (juvenile Candona sp., Ilyocypris bradyi, Eucypris sp., Darwinula stevensoni) were found. The uppermost 7 m are characterized by angular limestones (up to 5 cm), fragments of ceramics and bones as well as pieces of charcoal. No analyses were conducted in these anthropogenic layers, since no natural sediment occurred between the core-filling stones.

| The natural sedimentation outside the city area
Lim 13 (surface 4.32 m a.s.l., max. depth: 10 m b.s.) is located 500 m to the southeast of the city area and 50 m to the east of the Limyros river (Figures 1, 8 and 9). The base is composed of homogeneous grayish clayey silts (mean: 3-8 μm, C/N 1-2) with two intercalated peat layers. A sharp contact separates this unit from a sandy layer

| Slope debris facies
This facies occurs in the three cores from the western city close to the mountain before bedrock was reached. It is characterized by many angular stones (mostly limestones), some of which are even bigger than the augerhead (i.e., originally bigger than 5 cm). They derive from the adjacent mountains. In the upper parts of the cores, the slope debris is often mixed with anthropogenic material (pieces of ceramics, bricks, mortar, bones).

| Swampy/amphibic facies
The brown organic-rich silts in Lim 14 are characterized by a high C/N ratio (6-8) and CaCO 3 content (20-40%) as well as a sorting of 5-6 (swampy; Figure 10). The PCA clearly reveals a distinction to all other sediments. Sorting and C/N point to a low-energy deposition in an aquatic environment (Kaushal & Binford, 1999). The high Fe (20-30 k cps) and K (8-12 k cps) values together with freshwater microfossils Candona sp., Ilyocypris bradyi, Eucypris sp., Darwinula stevensoni also refer to an aquatic origin, probably characterized by shallow water with many plants (Meyers & Teranes, 2001).

| Lacustrine facies
The homogeneous gray clayey silts containing the freshwater

| Peat layers
In total, four peat layers occur, which are sandwiched by lacustrine strata. As expected, they are all characterized by high C/N ratios of 8-14. The preservation of organic material points to anoxic conditions (Turney, Canti, Branch, & Clark, 2005

| Anthropogenic layers
The upper meters of the drill cores from the city area consist of relatively badly sorted silts (mean up to 43 µm) with ceramic fragments, bones, charcoal, and angular stones. The heterogeneity of the anthropogenic layer is reflected in the PCA: most of the samples do not cluster well (red color in Figure 10). Due to their small size and weathering effects the ceramic fragments could not be dated.
However, several 14 C age estimates exist (a) for Lim 4 spanning from There is a special layer with strong anthropogenic input, much deeper, however, than the just described ones. It was discovered in several drillings between the two parts of the city at a depth of about 5.50 m b.s., that is ca. 1.10 m b.s.l. (e.g., Lim 12 and 16, Figure 5) with an average thickness of 10-15 cm. A synopsis of the 14 C ages from this layer and the topping peat leads to the conclusion that it was deposited in the second half of the 2nd millennium BC, roughly between 1400 and 1100 BC. This presumably Late Bronze Age layer is at least 500 years older than the first so far detected archaeological finds.

| Fluvial facies
Medium to coarse sands with pebbles were encountered in all drill cores with a thickness varying from 0.50 to 2.50 m (fluvial I; Lacustrine sediments with a maximum thickness of 7 m were detected in the cores between the western and the eastern city and to the east ( Figure 11). The oldest 14 C age estimate dates to the mid-4th millennium BC (Lim 12, Figure 4). From our research in other coastal areas of Anatolia (e.g., Brückner et al., 2006;Brückner et al., 2014), we presume that the Holocene stratigraphy starts with deposits from the late Pleistocene-early Holocene marine transgression, that is, shallow marine facies, which then turned into lagoonal facies as evidence of a beach barrier-lagoon complex.
However, in the Finike plain marine strata have not yet been encountered. Due to the steep slope gradient of the adjacent mountains (Toçak Dağı), they are obviously at a much deeper depth than the range we can cover with our equipment. Moreover, the strong longshore drift of the sea will have closed the lagoon, which then turned into a freshwater lake due to the strong perennial discharge of the karstic springs. This is why we only encountered limnic freshwater facies. It is thus most probable that the lake originates from a lagoon that later turned into a freshwater lake.
Examples of this sort are the lakes to the north and east of Ephesos (Stock et al., 2015) and Lake Azap in the hinterland of Miletos (Müllenhoff, Handl, Knipping, & Brückner, 2004 (Meyers & Lallier-Vergès, 1999;Meyers & Teranes, 2001), and anoxic conditions (Killops & Killops, 2005). Thus, the lake level must have considerably changed several times between the 4th and the 1st millennium BC, probably due to tectonics (see below). The lowest peat dates to the mid-4th millennium BC, the medium peat originates from the end of the 3rd millennium to the beginning of the 2nd millennium BC, the upper peat dates to ca. 1000-500 BC. In the center of the Finike plain, Öner (2013) describes two peat layers that are separated in the western part where the lower one has a thickness of up to 5 m. A peat layer is a sign that the lake started to silt up; its covering by limnic strata documents another expansion of the lake.
In the cores between the eastern and the western city, a cultural layer was detected grosso modo at the same depth of 1.15-1.00 m b.s.l. It will never be unearthed since the water table starts already at a depth of 0.50 m b.s. due to the perennial discharge of the karstic springs. The finds of a sharp, angular silex, processed stones, and nonlocal big rounded stones are evidence of an intentional setting.
This layer is on top of sediments which can be interpreted as representing the shore of a lake. The coring evidence is not sufficient to interpret this situation as the place of an early settlement in the second half of the 2nd millennium BC (dating to 1389-1115 BC and before 1230-1028 BC; Figure 5), but it certainly indicates a temporary use by humans.
Fluvial strata with sand and pebbles support the assumption that the lake started to silt up (the coring sites of Lim 12 and 16-19 are located at the western edge of the lake; Figure 11). At the latest between 1000 and 500 BC, the uppermost peat and once again lacustrine sediments were deposited. By then, the direct surroundings of the later city of Limyra must have been abandoned. However, the few drill cores can only give a first insight into the landscape formation.
Due to the fact that no drill cores have been retrieved from the eastern part of the Finike plain, the size of the lake could not be reconstructed. However, older scenarios of Öner (2013) can be taken into account. Many of his drill cores also reveal several meters of lake sediments. Uyanık et al. (2013), who studied the effects of liquefaction in the area of Kumluca in the eastern part of the Finike plain (see Figure 1 for location), confirm that lacustrine sediments are also present in that area. Therefore, the maximum lateral extension of the lake must have been at least 8-10 km.
The sediment cores close to the present coastline show fluvial sediments at the base, which are overlain by homogeneous sands that may originate from dunes ( Figure 9). Öner (2013) proves the coastline with the beach barrier close to the location of Lim 20 until the mid-2nd F I G U R E 9 Drill cores Lim 13 and 20 outside the city area. Lim 13 is located ca. 800 m southeast of Limyra, Lim 20 ca. 1 km inland from the present coastline F I G U R E 1 0 Principal component analysis (PCA) of sediment cores Lim 4, 12, 13, and 14 with ten PCs (mean, sorting, clay, sand, C/N, CaCO 3 , Fe, K, Ca/Sr, magnetic susceptibility). It is based on components 1 and 2 with a variance of 39.7% and 25.5%, respectively (PC 3: 11.5%, PC 4: 7.7%, PC 5: 6.4%, PC 6: 3.8%, PC 7: 2.8%, PC 8: 1.7%, PC 9: 0.9%, PC 10: 0.5%) STOCK ET AL. changes, have to be taken into account. The Finike plain with the city of Limyra is located in a tectonically active region between the Hellenic and the Cyprus arcs and-as stated above-has been formed due to subsidence during the Pliocene-Quaternary Hall et al., 2009). This could also be proven by Desruelles et al. (2009) who quantified the relative sea-level evolution in this region. On the other hand, riverine and littoral deposition overcompensated the subsidence effect so that the shoreline prograded seawards.
The subsidence still continues and can directly be connected to the development of the relative sea and groundwater levels. The sudden rise in the lake level, which occurred two to three times during the last 5000 years as evidenced by the limnic sediments covering the peat layers, may go along with subsidence, which may have been triggered by strong earthquakes. The Finike earthquake of 1926 with 6.8 Mw clearly shows the tectonic activity of the region (Uyanık et al., 2013 It is obvious that one factor for the evolution of the lake was the subsidence of the plain. Even before the Roman period, earthquakes must have occurred and influenced the local landscape formation. Therefore, the events which are documented in this study by the abrupt covering of the peat layers by lake sediments may be related to earthquake-induced subsidence, which triggered a sudden expansion of the lake. Another example of a Holocene lake evolution in an active tectonic region is Lake Marmara in the Gediz Graben (Bulkan, Yalçın, & Wilkes, 2018) For the environs of Finike, only few studies exist concerning the local sea-level evolution. Desruelles et al. (2009)  Under the assumption that the peat layers of the former lake reflect the former lake level which in turn was closely connected to sea level, one can state the following regarding the paleogeography in the environs of Limyra: when these peats grew their position mirrored more or less the sea level. Terrestrial limits of these kinds of peats are dominated by freshwater fauna . The abrupt covering of the peats by lake sediments indicates sudden subsidence-most probably triggered by earthquakes. Considering the present position of the three peats and their 14 C age estimates in coring Lim 12, the relative sea level was at −5.90 m around 3700-3500 BC, at −2.30 m around 2100-1800 BC and at −1.00 m around 1050-900 BC (see Figure 5). This is clear evidence of the strong influence of subsidence tectonics in the Limyra area. A systematic study and dating of the peat layers will decipher both the sea-level and the earthquake histories in more detail.
Another factor that might be taken into account is the climate.
During the Holocene, the three rapid climate change events (RCC) 8.2, 6.2, and 3 ka BP (Mayewski et al., 2004;Rohling, Mayewski, Abu-Zied, Casford, & Hayes, 2002)  As for Limyra, the uppermost peat layer may correlate with the 3 ka BP event. However, in the Finike plain, we deal with a karstic water regime, which is not sensitive to climate fluctuations. The rivers, which are fed by karstic springs, flow all year long with a balanced discharge, even after the dry summers. Thus, the factor climate cannot have been of any importance.

| CONCLUSION
For this research, 21 drill cores were carried out in the ancient city of Limyra and its environs. The study gives new insights into the landscape changes of the Finike plain since the Mid-Holocene. The results reveal that a lake, which had probably evolved from a former lagoon, dominated the Finike plain for several millennia. More research is needed to decipher its origin and size. Embedded in the limnic sediments are up to three peat layers showing the repeated onset of the siltation of the lake. That they are all covered by lacustrine strata can be explained by rapid coseismic subsidence.
Between the middle and the end of the 2nd millennium BC, a layer with anthropogenic remains (silex, processed stone, nonlocal big pebbles) at a depth of about 1.15-1.00 m b.s.l. was deposited in cores located between the western and the eastern city. Obviously, people made use of the lakefront. However, an event must have occurred leading to another rise in the lake level. During the 1st millennium BC, the lake completely silted up, and fluvial accumulation with shifting river channels began to dominate. People started to settle this area in the 7th century BC. As far as we know, the eastern city was abandoned between AD 1000 and 1200 due to the rising groundwater table, possibly caused by another earthquake. Then, peat grew and alluvium was accumulated. The formation of the Limyra area is a complex interplay between tectonics, lake evolution, river dynamics, sea-level changes, and human-environment interactions. Additional studies are needed to better understand and date STOCK ET AL.

| 499
the paleogeographic evolution of the Finike plain in general and the environs of Limyra in particular.

ACKNOWLEDGMENTS
The authors thank the Austrian Archaeological Institute for the continued financial and logistic support of our studies. This research project was funded by the FWF Austrian Science Fund (P29027-G25).
The Turkish government authorities kindly granted the research permits. We thank Anna Pint, Cologne for determining the microfossils, and two anonymous reviewers for helpful comments.