Are an early Byzantine seismic event (recorded in Manyas-Ulubat Lake sediment, N-W Turkey) and the end of the Beysehir Occupation Phase linked?
by
Leroy, Suzanne
Department of Geography and Earth Sciences, Brunel University, Uxbridge UB8 3PH, UK
Coauthors: Nizamettin Kazanci (Gebze Institute of Technology, 41400 Gebze, Kocaeli, Turkey and Department of Geological Engineering, Ankara University, 06100 Besevler, Ankara, Turkey), Antje Schwalb (Institut für Geowissenschaften, Technische Universität Braunschweig, Pockelsstrasse 3, D-38106 Braunschweig, Germany), Pedro Costa (Department of Geography and Earth Sciences, Brunel University, Uxbridge UB8 3PH, UK), Salim Önçel (Gebze Institute of Technology, 41400 Gebze, Kocaeli, Turkey), Ozden Ileri (MTA, 06520 Ankara, Turkey), Ozlem Toprak (Gebze Institute of Technology, 41400 Gebze, Kocaeli, Turkey)

Introduction: Multidisciplinary palaeolimnological studies are used to explore the possible impact of a cluster of earthquakes on the North Anatolian Fault (NAF) on the collapse of an arboricultural civilisation (the Beysehir Occupation Phase, BOP) at the beginning of the Byzantine Empire against a background of deteriorating climatic conditions.

Geological setting The north-western region of Turkey is bisected by the NAF, a plate boundary structure with a long history of devastating earthquakes. The most recent of these were the catastrophic earthquakes of Aug. and Nov. 1999, the first of which may have produced a 4 m-high seismic sea wave (tsunami) in the Gulf of Izmit. Written records of past great earthquakes extend back for two millennia and reveal a cluster of devastating earthquakes in the Early Byzantine period from the fourth to the sixth century (the Early Byzantine Tectonic Paroxysm, EBTP; Stiros 2001).

A belt of lakes that straddle the western strands of the NAF offer additional possibilities for establishing Holocene earthquake histories (Kazanci305; et al., in press; Leroy et al. 2002b; fig. 1). Several of these lakes are in the epicentral areas of recent large earthquakes: (from west to east) Lake Manyas in 1964, Lake Ulubat in 1850 and 1855 and Lakes Sapanca and Eften in 1999. Eye-witness accounts report various lake disturbances, including seiches, landslides, sediment resuspension, sand burst, liquefaction, river and spring flow changes. These phenomena can leave microstructural, geochemical and/or palaeoecological signatures in lacustrine sediment.

The Beysehir Occupation Phase The BOP, from roughly 3300 yr BC to AD 200-800, is a cultural phase seen in palynological diagrams of S-W Turkey, but poorly indicated in the archaeological record itself (Vermoere et al., 2002). It is marked by a rich arboriculture when the environment was favourable to olive trees, manna-ashes, walnuts, sweet chestnuts, vines, cereals a well as pastoralism.

The end of the BOP is shown to be relatively sudden and dramatic in palaeoenvironmental terms and coincides with changes in the distribution and lifestyles of people. The end of the BOP is however poorly dated. Hence it is still impossible to establish if this is synchronous across western Turkey. Similar collapses of arboriculture have been observed in diagrams from Greece (Bottema, 1993) and Israel (Heim et al., 1997). From the palynological point of view, the time period covered by the BOP in western Turkey corresponds to similar changes in vegetation in the Dead Sea with the development of Olea, Vitis and Juglans (S. Leroy in Heim et al., 1997).

Deteriorating climatic conditions In the Middle East the Greco-Roman time is considered as being a time when the climate is more humid. For Turkey, the BOP has been shown to follow the same pattern (Bottema et al. 1986; Eastwood et al., 1998). In the Dead Sea, the arboricultural phase is associated with the formation of a laminated sediment (rather than massive salts) in the lake indicating more freshwater input, hence more rainfall (Heim et al., 1997). A wealth of other proxies from the Dead Sea area confirms the climatic change: high lake levels (terraces formed from ca 100 BC to 40 AD), by karst formation in salt cave studies (around 50 to 250 AD), on the base of isotopic studies (from the Persian period to the Byzantine one), by stream channel sedimentation (Roman period), by d13C and d18O of wood from the Roman siege ramparts in Masada (AD 70-73). Shifts of the weather patterns associated with changes in summer monsoonal rains and north-easterly airflow have been suggested for older hydrological Holocene changes (see references in Heim et al., 1997).

Lake Manyas seismites and changes in water quality Analysis of the sediment structure and biological material within the Manyas sediment cores revealed two potential seismites: at the bottom of the sequence (11 m, between 2290 and 2030 calibrated years BC) and at c. 4 m in the lake centre (station 3, core 11) and at c. 3.20 m in the station near the northern shore (station 1, core 1) (Leroy et al., 2002 b). The age of the last corresponds to the Late Roman-Early Byzantine periods (fig. 5).

The lake was suddenly inundated with some salt water as indicated by these biota which contain elements that may withstand a wide range of salinities (fig. 1): a peak of Amaranthaceae-Chenopodiaceae in the palynological diagram and by seed and ostracod assemblages. Two main explanations have been proposed for this spike in salinity: a tsunami or an expulsion of hydrothermal fluids along the fault (Leroy et al. 2002b, c and Oncel et al., 2002).

In the tsunami hypothesis, the salt water would be deposited by a massive tsunami generated by an earthquake on the Sea of Marmara. Deposits of an undated tsunami were recently found near the mouth of the river in a gorge that connects the lake to the sea, indicating that tsunamis have penetrated far inland in the past. However the distance between Manyas and the sea is much longer.

The alternative is a movement of the Manyas Fault running just south of the lake. Nowadays many hot springs are still active along that fault and some are located on the main inflow (river Kocaçay). In addition, it is known that during earthquake activity, hot springs may cease to be active or, on the contrary, become hyperactive.

The local impact of the earthquake has been reconstructed as follows (fig. 2): · Manyas Fault moves; 1 - main inflowing river goes back to an ancient channel and bypasses the lake. Lake starved of its main source of freshwater. 2 - consolidated lake sediment cracks and soft lake sediment gets mixed up.

· At the same time, a seiche (oscillation of the water surface) develops on the lake => turbidity and increase of nutrient input.

· Hot springs briefly more active, increased salinity and temperature of the waters reaching the lake (as indicated by euryhaline ostracods and seeds) · Later nutrient and salinity levels in the lake water become further concentrated in as lake level drops through evaporation.

· Aquatic vegetation, phytoplankton and fish bloom, until anoxic conditions => formation of siderite and death of plants and animals.

· With a sedimentation rate of 0.44 cm yr-1 (based on 210Pb and 137Cs dating), the time required for partial recovery and for the return of the main river is c. > 25 years.

Lake Ulubat changes in sediment and terrestrial vegetation The sediment of lake Ulubat, which has been cored in its eastern part, is relatively homogenous on its top 5 to 7 m (Leroy et al., 2002a). The bottom meter of several cores (especially station 11, cores LV10 and 11) has however gone through:

1- a stiff silt layer (> 20 cm) with a significant lower amount of clay size particles, organic matter and carbonate than higher in the sequence. It is also characterised by high magnetic susceptibility. Preliminary palynological analyses show an open oak (Quercus) forest and marshy conditions with macrophytes. X-ray photography show the presence of centimetric clasts. Some heavy and trace metal elements (As, Ba, Cu, Ni, Mn) reach maximum values here in contrast to minimum amount of Fe and Al (<10 ppm).

2- a sharp transition well marked in visual descriptions, X-rays and many proxies (fig. 3).

3- followed by a c. 50 cm clayey layer richer in organics. The mean grain size of this layer is medium to coarse silt and it typically has a low carbonate content. These are the only depths rich in macrofossils. They have provide 2 radiocarbon dates. Palynomorphs indicate a nutrient rich environment (bloom of Tetraedron and Pediastrum). Terrestrial vegetation is replaced by a open Pinus woodland (similar to the transition observed in Manyas after the 4 m event).

4- From here to the top of the sequence, the sedimentary constituents (grain size, carbonate content, heavy metals, clay minerals, etc) of the sediment sequence show only slight oscillations. However the organic content displays two peaks at 600-450 cm and 300-150 cm (Fig. 4). At the depth of 275-255 cm the amount of Fe, Al, Mn and carbonate content suddenly decrease while grain-size and the organic content increase. The Boron content of the sediment is already higher than that of an ordinary lake as the watershed includes boron-rich rocks even a sharp rise is noticeable at 4 m depth.

Ostracod shells are not well preserved; but, where present in core LV11, their assemblages contain salt-tolerant species in facies 1 and 3 and flowing freshwater at 275-75 cm depth.

Common age and cause for Manyas-Ulubat event Three radiocarbon dates are available now for the lake Manyas event at 4 m and two for just above the event at the base of the Ulubat sequence (fig. 5). They point at a synchronous phenomenon occurring at AD c. 400. Historical documents on earthquakes (Ambraseys and Finkel 1991) abound for the time period which belongs to the Early Byzantine Tectonic Paroxysm. The local events of AD 460 at Cyzicus (Erdek) and of AD 368 at Germe (M. Kemalpasa, on the Manyas fault) are those most likely responsible for the event, unless it was caused by a more regional earthquake: the large Marmara seism at AD 447 or the eastern Mediterranean seism at AD 365. The last two events produced tsunamis.

Whatever hypothesis confirmed by further studies, it is already clear that events of a similar type/amplitude were rare (1 or 2 in c. 4100 years) and that the effect of the earthquake was major and more long-lasting than a catastrophic event and its immediate side effects. The crucial element is the permanent change in terrestrial vegetation in both sites. Pollen diagrams of terrestrial plants record vegetation on a relatively wide geographical area, certainly here on an area extending outside of the drainage basin of lakes Manyas-Ulubat.

A link between the end of the BOP and a cluster of earthquakes? The high-resolution pollen diagram obtained across the 4 m seismite in Lake Manyas shows that the BOP end occurred immediately after the earthquake (fig. 2a). A statistical analysis (DCA) shows that Fagus, Fraxinus ornus and Juglans (agriculture in the low lands and beech forest in altitude) are replaced definitively by Pinus and Juniperus (degraded open woodland). This opens the question of a potential link between the two.

Although one single earthquake could not have influenced the whole of western Turkey (and some neighbouring countries), there is a strong possibility that the end of the BO-like period in various places is linked to individual earthquakes that are part of the EBTP. This occurred with the unfavourable background of a rainfall deficit over the same time period. A civilisation, that is under stress because of a climatic deterioration, may not be able to survive the additional effects of a cluster of earthquakes.

It is interesting to see that the area affected by the Early Byzantine cluster of earthquakes largely corresponds to the area where the BOP has been recorded. Pollen diagrams in Greece (Bottema 1993) and the Dead Sea (Heim et al., 1997) also show the same collapse of arboriculture and return to fallow land followed by recovery of natural vegetation although degraded. Turkey, Greece and the Middle-East as well as Egypt and Libya have been affected by earthquakes and tsunamis linked to the AD 365 cluster of earthquakes.

Discussion At the site of Sagalassos, Vermoere et al. (2002) used pollen analyses and archaeological data to show a link between an earthquake around the middle of the seventh century and the abandonment of Sagalassos. Arboriculture collapse has been however dated between AD540 and 810.

Although other studies have been able to show clear links between pollen analyses, earthquakes and settlement abandonment, none have yet linked the BOP end and the EBTP. Now the link between a catastrophic event, probably an earthquake and the definitive collapse of arboriculture has been shown for 2 neighbouring lakes in N-W Turkey. Further work on the Dead Sea sediment will focus on the seismite of the Jerusalem earthquake in AD 368 and its relation to the agriculture collapse seen in palynological diagrams roughly around that time (Heim et al., 1997; Baruch, 1990).

Implication of our work In modern Lake Manyas basin, the water quality and the volume of the reservoir may change dramatically due to lake bounding by an active fault. In consequence the present dams erected on the lake shores need to be able to resist further seisms and the outflows of the numerous regional hot-springs need to be monitored.

Acknowledgements NATO grants (CRG 97 3155 and CLG 97 8645) are gratefully acknowledged. All our collaborators, M. Kibar (Ankara University), O. Emre (MTA, Turkey) E. McGee (University of Dublin), I. Stewart (Brunel and Glasgow Universities), D. Bowman (Ben-Gurion University, Israel) and H. I. Griffiths (University of Hull, deceased), are warmly thanked.

References: Ambraseys, N., Finkel, C., 1991. Long-term seismicity of Istanbul and the Marmara Sea region. Terra Nova 3, 527-539.

Baruch U., 1990 - Palynological evidence of human impact on the vegetation as recorded in Late Holocene lake sediments in Israel. in: Man's role in the shaping of the eastern Mediterranean landscape, Bottema, Entjes-Nieborg and Van Zeist (ed.), Balkema, Rotterdam: 283-293.

Bottema S. 1993 – The palaeoenvironment of prehistoric man in the Near east: some aspects of palynological research. Japan review 4: 129-140.

Bottema, S., Woldring, H., Aytug, B., 1986. Palynological investigations on the relation between prehistoric man and vegetation in Turkey: the Beysehir occupation phase. In: Demiriz, H., Özhatay, N. (Eds), Proceedings of the fifth meeting of OPTIMA (Istanbul, 8-15 Sept. 1986). Istanbul, pp. 315-328.

Eastwood, W. J., Roberts, C. N., Lamb, H. F., 1998. Palaeoecological and archaeological evidence for human occupance in southwest Turkey: the Beysehir Occupation Phase. Anatolian Studies 48, 69-86.

Heim C., Nowaczyk N., Negendank J., Leroy S., and Ben-Avraham Z., 1997 - Middle-East desertification: evidence from the Dead Sea. Naturwissenschaften: 84, 9: 398-401.

Kazanci; N., Leroy S., Ileri Ö., Emre O., Kibar, M. and Öncel, S. in press.. Late Holocene erosion in NW Anatolia from sediments of Lake Manyas, Lake Ulubat and the southern shelf of the Marmara Sea, Turkey. Catena.

Leroy S., Bowman D., Costa P., Emre O., Ileri O., Kazanci; N., Önçel S. and Stewart I., 2002 a - Preliminary results of a coring campaign on lake Ulubat (N-W Turkey) in search of an early Byzantine seismite. “Environmental catastrophes and recovery in the Holocene” an INQUA-PAGES-Brunel conference, London 29 Aug.-2 Sept. 02.

Leroy S., Kazanci; N., Ileri Ö., Kibar M., Emre O., McGee E. and Griffiths H., 2002 b. Abrupt environmental changes within a late Holocene lacustrine sequence south of the Marmara Sea (Lake Manyas, N-W Turkey): possible links with seismic events Special issue of Marine Geology on "Quaternary Tectonic and Paleoclimatic-Paleoceanographic Evolution of the Marmara Sea and Environs: a collection of papers dedicated to the memories of Drs. Ihsan Ketin and Aykut Barka". Edited by A. Aksu. 190/1-2: 531-552 Leroy S., Kazanci; N., Ileri Ö., Önçel S., Emre O., Løvlie R., 2002 c - Is there a link between an early Byzantine seismic event (recorded in Lake Manyas sediment, N-W Turkey) and the end of the Beysehir Occupation Phase? “Environmental catastrophes and recovery in the Holocene” an INQUA-PAGES-Brunel conference, London 29 Aug.-2 Sept. 02.

Önçel S. M., Leroy S., Ileri Ö., Kazanci; N. and Emre Ö., 2002 - In search of the source a sudden and short-lived salinity peak linked to an early Byzantine seismic event recorded in Lake Manyas sediment (N-W Turkey) “Environmental catastrophes and recovery in the Holocene” an INQUA-PAGES-Brunel conference, London 29 Aug.-2 Sept. 02.

Stiros, S., 2001. The AD 365 Crete earthquake and possible seismic clustering during the fourth and sixth centuries AD in the eastern Mediterranean: a review of historical and archaeological data. J. Struct. Geol. 23, 545-602.

Vermoere M., Bottema S., Vanhecke L., Waelkens M., Paulissen E. and Smets E., 2002 – Palynological evidence for Late Holocene human occupation recorded in two wetlands in SW Turkey. The Holocene 12, 5: 569-584.

Date received: January 27, 2004

Copyright © 2004 by the author(s). The author(s) of this document and the organizers of the conference have granted their consent to include this abstract in Atlas Conferences Inc. Document # camu-34.