Tephrostratigraphy and tephrochronology of lakes Ohrid and Prespa, Balkans

. Four cores from Balkans lakes Ohrid and Prespa were examined for recognition of tephra layers and cryp-totephras, and the results presented along with the review of data from other two already published cores from Lake Ohrid. The six cores provide a previously unrealised tephros-tratigraphic framework of the two lakes, and provide a new tephrostratigraphic proﬁle (composite) for the Balkans, which spans from the end of the Middle Pleistocene to the AD 472. A total of 12 tephra layers and cryptotephras were recognised in the cores. One is of Middle Pleistocene age (131 ka) and correlated to the marine tephra layer P-11 from Pantelleria Island. Eight volcanic layers are Upper Pleistocene in age, and encompass the period between ca. 107 ka and ca. 31 ka. This interval contains some of the main regional volcanic markers of the central Mediterranean area, including X-6, X-5, Y-5 and Y-3 tephra layers. The other layers of this interval have been related to the marine tephra layers C20, Y-6 and C10, while one was for the ﬁrst time recognised in distal areas and correlated to the Taurano eruption of probable Vesuvian origin. Three cryptotephras were of Holocene age. Two of which have been correlated to Mercato and AD 472 eruptions of Somma-Vesuvius,


Introduction
Owing to the intense explosive volcanic activity that affected the Mediterranean over the last 200 ky (Vezzoli, 1988;Poli et al., 1987;Santacroce, 1987;Rosi and Sbrana, 1987;Keller et al., 1990;Orsi et al., 1996;Pappalardo et al., 1999;Di Vito et al., 2008;Santacroce et al., 2008) application of tephrochronology to volcanology, Quaternary science, paleoceanography, and archaeology has an exceedingly high potential in this area.In the last 30 years, Quaternary tephra layers have been extensively used to develop a high-resolution event stratigraphy for the late Pleistocene and Holocene across the central and eastern Mediterranean (Keller et al., 1978;Paterne et al., 1988Paterne et al., , 1990;;Narcisi and Vezzoli, 1999;Wulf et al., 2004;Margari et al., 2007;Aksu et al., 2008;Giaccio et al., 2009;Zanchetta et al., 2010).While early work focused mainly on samples from marine cores (Keller et al., 1978;Paterne et al., 1988Paterne et al., , 1990;;Calanchi et al., 1998), recent work on terrestrial archives (including Italian, Greek, Turkish and Balkan lakes, and Bulgarian, Greek and Italian cave sites; e.g.St. Seymour and Christianis, 1995;Narcisi and Vezzoli, 1999;St. Seymour et al., 2004;Wulf et al., 2004;Frisia et al., 2008;Giaccio et al., 2008Giaccio et al., , 2009;;Sulpizio et al., 2010;Vogel et al., 2010) have considerably advanced the development of a long, highresolution tephrostratigraphy, which will link marine and terrestrial records of Pleistocene-Holocene-age across the Mediterranean region and mainland Europe.Nevertheless, there are some areas in the Central Mediterranean, in which volcanologic and Quaternary studies are still at an early stage, and the link to the general tephrostratigraphic Mediterranean network is lacking.The Balkans area across Macedonia, Albania and Montenegro is certainly one, in which tephrostratigraphic and tephrochronologic studies are in their infancy, although some studies on lacustrine settings indicate the area is extremely promising for tephrostratigraphic studies (Wagner et al., 2008;Caron et al., 2010;Lézine et al., 2010;Sulpizio et al., 2010;Vogel et al., 2010).For these areas both tephrostratigraphic and tephrochronologic studies can offer invaluable stratigraphic support for sedimentologic, palaeoclimatic, paleoenvironmental, and volcanological studies.Here we present tephrostratigraphic and tephrochronologic data of lakes Ohrid and Prespa (Fig. 1), which encompass the last 131 ka.
The paper reviews some already published data from the Macedonian side of Lake Ohrid, which are presented along with new data from both Ohrid and Prespa lakes.The recognised tephra layers supply a composite tephrostratigraphy of lakes Ohrid and Prespa, which has been correlated to other archives from the Balkans (Albanian side of Lake Ohrid, Caron et al., 2010;Lézine et al., 2010;Lake Shkodra, Sulpizio et al., 2010), and linked to the regional tephrostratigrapic network of the central Mediterranean area.

The lakes of Ohrid and Prespa
Lake Ohrid (Fig. 1) is a transboundary lake shared by the Republics of Albania and Macedonia.It is located at 693 m above sea level (a.s.l.), and surrounded by high mountain ranges reaching heights up to 2300 m.It is 30 km long, 15 km wide and covers an area of 358 km 2 .The lake basin shows a relatively simple tub-shaped morphology with a maximum water depth of 289 m, an average water depth of 151 m and a total volume of 50.7 km 3 (Popovska and Bonacci, 2007).
Lake Ohrid is mainly fed by inflow from karst springs (55%), while the remaining 45% of the hydrological input includes direct precipitation on the lake surface, river and direct surface runoff (Matzinger et al., 2006).The direct watershed of Lake Ohrid covers an area of 1002 km 2 (Popovska and Bonacci, 2007).Surface outflow (60%) through the River Crn Drim to the north and evaporation (40%) are the main hydrological outputs (Matzinger et al., 2006).The average annual precipitation on the Lake Ohrid watershed is 907 mm (Popovska and Bonacci, 2007).

Core recovery and description
The sediment cores presented here were recovered during field campaigns between 2005 and 2009, using a floating platform equipped with a gravity corer for surface sediments and a piston corer (both UWITEC Co.) for deeper sediments.Cores Lz1120, Co1200, Co1201, and Co1202 are from Lake Ohrid and cores Co1204 and Co1216 from Lake Prespa (Fig. 1).Core description and high-resolution colour scanning was carried out immediately after lengthwise opening of the cores in the laboratory.Tephrostratigraphy of cores Lz1120 and Co1202 was already published (Wagner et al., 2008;Vogel et al., 2010).

Tephra detection and analysis
High-resolution XRF analysis was carried out on the surface of one of the core halves using an ITRAX core scanner (COX Ltd), equipped with a Mo-tube set to 30 kV and 30 mA, and a Si-drift chamber detector.Scanning was performed at 0.5 mm (Co1204), 1 mm (Co1202), 2 mm (Co1216), and 2.5 mm (Co1200, Co1201) resolution and an analysis time of 20 s (Co1200, Co1201, Co1202, Co1204) and 10 s (Co1216) per measurement.The obtained count rates for K, Sr and Zr can be used as estimates of the relative concentrations for these elements.
From horizons which were distinctive because of their macroscopic grain size composition, colour or element count rates derived from XRF scanning, about 1 cm 3 was washed and sieved.The >40 µm fraction was embedded in epoxy resin and screened for glass shards and micro-pumice frag-ments using scanning electron microscopy (SEM).Energydispersive spectrometry (EDS) of glass shards and micropumice fragments was performed using an EDAX-DX micro-analyser mounted on a Philips SEM 515 (operating conditions: 20 kV acceleration voltage, 100 s live time counting, 200-500 nm beam diameter, 2100-2400 shots s −1 , ZAF correction).The ZAF correction procedure does not include natural or synthetic standards for reference, and requires analysis normalisation at a given value (chosen at 100%).Analytical precision is 0.5% for abundances higher than 15 wt%, 1% for abundances around 5 wt%, 5% for abundances of 1 wt%, and less than 20% for abundances close to the detection limit (around 0.5 wt%).Interlaboratory standards are shown in Table 1.
Accuracy of measurements is around 1%, a value analogous to that obtained using wave dispersion spectroscopy (WDS), as tested by Cioni et al. (1998) Cioni et al., 1998;Sulpizio et al., 2010) confirming the full comparability of EDS analyses from the Pisa laboratory and data from WDS microprobes.The concentration of thirty-five trace elements was determined by ICP-MS for four selected samples.About 50-60 mg of bulk sample powders were dissolved in screw-top PFA vessels with a mixture of HF and HNO 3 on a hot-plate at ∼120 • C. The sample solutions were then spiked with Rh, Re and Bi as internal standards (20 ng ml −1 in the final solutions) and diluted to 50 ml in polypropylene flasks.Milli-Q purified water (18.2M cm), and HF and HNO 3 Aristar grade were used in each step of sample preparation.Analyses were performed by external calibration using geochemical reference samples as composition-and matrixmatching calibration solutions.The correction procedure includes (i) blank subtraction; (ii) instrumental drift correction using internal standardization and repeated (every 5 samples) analysis of a drift monitor; (iii) oxide-hydroxide interference correction.At the concentration levels of the studied samples, precisions are better than 5% RSD, except for Sc, Ni and Cu for which the precisions are between 5 and 10% RSD.

Description of tephra layers
The studied sediment cores contain a variable number of tephras and cryptotephras, which are described here on the basis of their lithology and composition.Tephra and cryptotephra numbering follows the origin of the core (OT for Ohrid Tephra and PT for Prespa Tephra), the year of core recovery (2005)(2006)(2007)(2008)(2009) and the last two identification numbers of the respective core.Glass shards and micropumice fragments from the different cores were compositionally classified using the silica vs. total alkali diagram (TAS; Le Bas et al., 1986;Fig. 3).Fig. 3. Total alkali vs. silica diagrams for the tephra layers recognised in the six cores studied or reviewed in this paper.

Core Co1200
Core Co1200 contains two discrete tephra layers visible at naked eye.Tephra layer OT0700-1 (40-38 cm) contains coarse to fine ash, appears light-brown to ochre in colour, and contains highly vesicular, aphyric micro-pumice and glass shards (Fig. 4a).Most of the glass shards show a trachytic composition with only few samples that plot close to the phonolitic field on the TAS diagram (Fig. 3a).Tephra layer OT0700-2 (120.5-85.5 cm) comprises coarse to fine ash, has a light red colour, and contains aphyric, vesicular micro-pumices and aphyric glass shards with thick septa (Fig. 4b).Glass composition ranges from trachyte to phono-trachyte (Fig. 3a), with three different alkali ratios (Table 2).

Core Co1201
Core Co1201 contains seven discrete tephra layers visible at naked eye inspection and with different thicknesses.
Tephra layer  comprises coarse ash, is light-brown to red in colour, and contains both aphyric glass shards and poorly vesicular fragments with small pyroxene crystals in the groundmass (Fig. 4d).The glass composition continuously spans between phonotephryte/shoshonite and phonolite/trachyte (Fig. 3b).Three main compositional groups can be identified on the basis of the alkali ratio and SiO 2 content (Table 2).

Core Co1202
Core Co1202 contains four visible tephra layers and six cryptotephras, which were already described in detail by Vogel et al. (2010a).
Cryptotephra OT0702-2 (145.5-144cm) consists of nonvesicular and blocky fragments with a porphyritic texture.Fragments exhibit mineral inclusions of plagioclase, clinopyroxene and olivine up to some tens of microns in size (Fig. 4f) and frequent occurrences of Fe-Ti oxides.The glass composition is mostly benmoreitic, with few analyses plotting in the trachytic and mugearitic fields on the TAS diagram (Fig. 3c).
Tephra layer OT0702-4 (620-617 cm) is light brown in colour, and is characterised by normally graded coarse to fine ash.Glass shards are mixed with lacustrine sediment up to 2 cm above the tephra layer.The tephra comprises highly vesicular, aphyric micro-pumice and cuspate glass shards.Most of the glass shards show a trachytic composition with only few samples that plot close to the phonolitic field on the TAS diagram (Fig. 3c).Trace element distribution (sample 523; Table 3, Fig. 5) shows the less enriched pattern in the four analysed samples, with small negative anomalies in Ba and Sr and a marked enrichment in Pb (Fig. 5a).The rare earth element (REE) pattern shows a more or less regular decrease passing from light REE (LREE) to heavy REE (HREE; Fig. 5b).Cryptotephra OT0702-5 (696-689 cm) comprises tachilitic particles with a crystal-rich groundmass containing acicular clinopyroxene, plagioclase and sanidine (Fig. 4h).The few glass compositions available range from latite to phonolite when plotted on the TAS diagram (Table 2 and Fig. 3c).
Tephra layer OT0702-6 (752-743 cm) comprises coarse to fine ash, is reddish-brown to light-brown in colour, and glass shards are mixed with lacustrine sediments in the overlying 10 cm.Volcanic particles mainly comprise aphyric, vesicular micro-pumices and aphyric glass shards with thick septa.Glass composition ranges from trachyte to phono-trachyte (Fig. 3c), with two different alkali ratio (Table 2).Trace element distribution (sample 565; Table 3, Fig. 5) shows an intermediate enrichment in the analysed samples, with a pronounced negative anomalies in Sr and Eu, moderate anomaly in Ba, and moderate positive anomalies in Th, U and Pb (Fig. 5a).The REE pattern shows a regular decrease in  enrichment passing from LREE to HREE, with a pronounced negative Eu anomaly (Fig. 5b).Cryptotephra OT0702-7 (825-822 cm) comprises aphyric, cuspate glass shards with thin septa and glassy groundmass.Glass composition ranges from rhyolitic (main) to trachytic (Table 2 and Fig. 3c).
Tephra layer OT0702-8 (1146.5-1140cm) is rusty-red in colour and comprises coarse ash and shows sharp basal and top contacts.Volcanic particles are aphyric micro-pumices and glass shards.Glass composition is mainly phonolitic (Table 2 and Fig. 3c).
Tephra layer OT0702-9 (1232.5-1229cm) is light-brown in colour, and comprises coarse to fine ash with sharp basal and top boundaries.Grain size mainly comprises fine to coarse ash.Volcanic particles are highly vesicular, aphyric micro-pumices and glass shards with a large variability in shape and size (Fig. 4i).Glass composition is mainly trachytic, with few analyses that plot into the phonolitic field, and two different alkali ratios (Table 2 and Fig. 3c).The trace element (sample 1003; Fig. 5, Table 3) distribution shows the most enriched pattern in the four analysed samples, with marked negative anomalies in Ba, Sr and Eu that testify for a feldspar-dominated magma fractionation (Fig. 5a).The REE pattern shows a regular decrease from LREE to HREE, with a marked negative Eu anomaly (Fig. 5b).Cryptotephra OT0702-10 (1447-1440 cm) comprises mainly aphyric cuspate glass shards.When plotted on the TAS diagram the glass shards reveal a bimodal chemical composition, which comprises a trachyte and a rhyolite without any compositional trend between (Fig. 3c).

Core Lz1120
Core Lz1120 contains two visible tephra layers and one cryptotephra, which were already described in detail by Wagner et al. (2008).
Tephra layer OT0520-2 (897-896 cm) comprises coarse to fine ash, and contains light coloured, elongated, highly vesicular fragments.Vesicles are mainly tubular and form channels throughout the entire length of the pyroclastic fragments.Groundmass is glassy, with very few elongated microcrystals of sanidine.Composition of single shards shows a limited variability within the trachytic field (Fig. 3d).
Tephra OT0520-3 (1075-1070 cm) is of unknown thickness, as the resistance of this layer prevented penetration of the coring equipment through it.It comprises coarse to fine ash, and volcanic fragments comprise mainly light coloured, highly vesicular fragments, with minor dark coloured glass shards and micropumice fragments.In both typologies of volcanic fragments the vesicles are spherical or ovoidal, separated by thin, glassy sets.Groundmass is almost aphiric, even if small sanidine crystals sometimes occur on larger sets among bubbles.Single shard composition show a trend from the trachytic to the phonolitic fields (Fig. 3d), and can be arranged into three groups on the basis of different alkali ratios (Table 2).

Core Co1204
Core Co1204 was recovered from the north-western part of Lake Prespa (Fig. 1).It contains two discrete tephra layers and one cryptotephra.
Cryptotephra PT0704-2 (767.2-764.2cm) was identified by high Sr count rates through high-resolution XRFscanning, and contains tachilitic fragments with highly crystalline groundmass.Glass is rare and shows a compositional trend from shoshonites to trachytes (Fig. 3e).Three different compositional groups were identified on the basis of SiO 2 , CaO, and total alkali contents (Table 2).
Tephra layer PT0704-3 (879.3-863.3cm) comprises coarse to fine ash, and contains aphyric, highly vesicular micro-pumice and cuspate to convolute glass shards with glassy groundmass.Glass composition straddles the phonolitic and trachytic fields (Fig. 3e), it can be split into two or three groups depending on the different alkali ratio (Table 2).
The trace element distribution and the REE pattern (sample PR628; Table 3, Fig. 5) are identical to the sample 565 from core Co1202.

Core Co1216
Core Co1216 war recorevered from the north-western side of Lake Prespa, close to the location of core Co 1204 (Fig. 1).It contains only one cryptotephra, labelled PT0916-1 (Fig. 2).Volcanic fragments comprise tachilitic particles with a crystal-rich groundmass containing acicular clinopyroxene, plagioclase and sanidine, and aphyric glass shards.The glass compositions range from tephri-phonolite/shoshonite to phonolite when plotted on the TAS diagram (Table 2 and Fig. 3f).

Correlation to proximal deposits and other distal archives
The correlation of a distal tephra layer with proximal counterparts is a critical process, which in many cases implies the contemporaneous use of different data, such as glass and mineral composition, chronology, lithology, and stratigraphic position.This is because pyroclastic deposits from the same source, with few exceptions, show closely similar major element composition.Furthermore, pyroclastic deposits from different sources but originating from magmas with similar degree of evolution (e.g.trachytes and rhyolites) are barely distinguishable on the basis of the sole major element composition.In this study, cores Lz1120 and Co1202, which contain correlated tephra layers (Wagner et al., 2008;Vogel et al., 2010), can be used as a reference, particularly because core Co1202 contains the largest number of tephra layers in all studied sediment cores from this region (Fig. 2).The youngest volcanic deposit was correlated to the AD 472 (1478 cal.yr BP) eruption of Somma-Vesuvius.It occurs as a cryptotephra in core Co1202 (OT0702-1), but has not been recognised in any of the other studied cores (Fig. 2).
The Codola ash (inferred age 33 cal.kyr BP; Giaccio et al., 2008 or 34.2 cal. kyr BP;Vogel et al., 2010) occurs as a cryptotephra in core Co1202 (OT0702-5; Fig. 2; Vogel et al., 2010).The Codola fragments have a highly microcrystalline groundmass and a glass composition that straddles the tephri-phonolitic/latitic and the phonolitic/trachytic (Fig. 6a; Di Vito et al., 2008;Giaccio et al., 2008;Santacroce et al., 2008).Based on their stratigraphic position and lithology the tephra layers PT0704-2 in core Co1204 and PT0916-1 in core Co1216 from Lake Prespa are good candidates for correlation to the Codola eruption.The inspection of glass composition illustrates a more complex situation, with some of the analyses from the PT0704-2 and the PT0916-1 samples that plot outside the Codola field (Fig. 6a).In particular, the PT0916-1 analyses define two compositional groups separated by a broad gap in SiO 2 and total alkali content (PT0916-1a and PT0916-1b; Fig. 6a; Table 2).Both groups plot outside the Codola field, either in TAS or SiO 2 vs. CaO diagrams (Fig. 6a, b), being more and less evolved, respectively.The less evolved analyses correlate well to the Taurano composition, which defines an evolutionary trend with the Codola samples in the SiO 2 vs. CaO diagram (Fig. 6b).The Taurano eruption was tentatively assigned to the activity of the Somma-Vesuvius volcano, and approximately dated 36-33 cal.kyr BP (Di Vito et al., 2008).Proximal deposits comprise porphyritic dark scoriae of shoshonitic/phono-tephritic composition, with a groundmass rich in microlites of clinopyroxene, sanidine and leucite (Di Vito et al., 2008).
The exact stratigraphic position of Taurano tephra with respect to the Codola deposits in proximal areas is still under debate (Di Vito et al., 2008;Santacroce et al., 2008), because the two deposits have never been described in stratigraphic succession in proximal areas.The tephrostratigraphy of Lago Grande di Monticchio (Wulf et al., 2004) can help in unravelling the stratigraphic position of Taurano deposits.In particular, most of the TM-17 layers (previously attributed to the late activity of Alban Hills volcano; Wulf et al., 2004) show compositional and petrographic similarity with the Taurano deposits (Fig. 6), with glass composition of TM17-d and TM17-e tephra layers that closely matches that of sample PT0916-1a (Fig. 6; Table 4).The TM-17 cluster is splitted into two different groups on the basis of stratigraphic position and glass composition.The first one (TM-17c, TM-17d and TM-17e, hereafter Taurano-a) preceeds the deposition the Codola tephra and has a lower total alkali content of the second group (TM17-a and TM17-b, hereafter  Taurano-b; Fig. 6), which is younger than Codola tephra layer (Wulf et al., 2004).Based on these considerations, the correlation of the tephra OT0916-1a to TM17-d and TM17e tephra layers and to the Taurano-a deposits (age between 34.2 and 36 cal kyr BP) is here proposed.
The evolved group (sample PT0916-1b; Table 2) plots outside the Taurano-Codola trend, and coincides with the Campanian Ignimbrite field in SiO 2 vs. CaO diagram (pinkish area of Fig. 6b).The lithology of the glass shards supports the correlation of this group to the Campanian Ignimbrite deposits, being formed by convolute and cuspate shards with glassy groundmass.The analyses of cryptotephra PT0704-2 from core Co1204 can be divided into three groups, on the basis of different contents of SiO 2 , total alkali, and CaO (Fig. 6a, b; Table 2).Among them, only the PT0704-2b samples plot in the Codola field in both TAS and SiO 2 vs. CaO diagram (Fig. 6a, b).The two analyses of group PT0704-2c are the most evolved, and plot in the Campanian Ignimbrite field in the SiO 2 vs. CaO diagram (Fig. 6b).The PT0704-2a group have shoshonitic/latitic composition (Fig. 6a), and shows a separate compositional trend in the SiO 2 vs. CaO diagram.The correlation of this group with the regional tephrostratigraphy is at present not possible, since tephra layers or cryptotephras with similar lithology, glass composition and age have never been described in the Balkans, in Adriatic/Ionian marine cores, and continental Italy (e.g.Lago Grande di Monticchio succession; Wulf et al., 2004).Based on glass composition, a generic correlation to the activity of Vulcano Island between 53 and 21 ka (De Astis et al., 2000, 2006;Lucchi et al., 2008) is here proposed, although the link to a specific dated eruption is, as yet, impossible.
Cryptotephra OT0702-7 occurs below the CI/Y-5 tephra layer in core Co1202, and was correlated to the Green Tuff eruption from Pantelleria island (Vogel et al., 2010), which corresponds to the Y-6 marine tephra layer (Keller et al., 1978).
Tephra layer OT0702-9 in core Co1202 was correlated to the marine X-6 tephra layer (Keller et al., 1978), of generic Campanian origin.Brauer et al. (2007) quoted an 40 Ar/ 39 Ar age of 107 ± 2 ka for the X-6 tephra, which is in good agreement with the suggested age of 108.43 ka obtained from the varve-supported chronology of the Lago Grande di Monticchio record.The X-6 tephra layer has not been recognised in the other studied cores of lakes Ohrid and Prespa (Fig. 2).The evolved trace element composition of this tephra layer (Fig. 5; Table 3) can help in its discrimination from the mess of trachytic tephra layers recognised in the Upper Pleistocene successions.

Composite tephrochronological record and regional correlations
The correlation of the 12 recognised tephra and cryptotephra layers with known proximal deposits allows the reconstruction of a composite stratigraphic and chronological framework for lakes Ohrid and Prespa during the past 131 kyr (Fig. 7).Three cryptotephras occur during Holocene, with the Mercato layer marking the temperature maximum of the Early Holocene.The following seven tephra layers and cryptotephras punctuate the Upper Pleistocene, encompassing about 77 kyr (from 30 to 107 ka; Fig. 7).Four markers cluster between 30 and 40 ka (i.e. between the Y-3 and the Y-5 tephra layers), detailing the stratigraphic succession between Heinrich events 3 and 4 (Ton-That et al., 2001;Zanchetta et al., 2008;Wagner et al., 2010).Particularly significant is the first recognition of the Taurano deposits in a very distal succession, which so far has never been described in nearby Adriatic and Ionian sea cores.The Y-6 cryptotephra occurs below the CI, at about 49 ka (Fig. 7), linking the Balkans to the Ionian sea archives.The sedimentary record below the Y-6 spans about 30 kyr without occurrence of any tephra record, being the following tephra layer correlated to the SA3-a eruption and the C-20 marine tephra layer (79-80 ka; Paterne et al., 1988;Fig. 7).The last two tephra layers of the Upper

Table 2 .
Average EDs analyses of glass shards of tephra layers and cryptotephras from the studued and reviewed cores.