Palaeogeography, Palaeoclimatology, Palaeoecology

Transcription

1 PALAEO-05474; No of Pages 16 Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx Contents lists available at ScienceDirect Palaeogeography, Palaeoclimatology, Palaeoecology journal homepage: The age of the Sarmatian Pannonian transition in the Transylvanian Basin (Central Paratethys) Iuliana Vasiliev a,, Arjan de Leeuw a, Sorin Filipescu b, Wout Krijgsman a, Klaudia Kuiper c, Marius Stoica d, Andrei Briceag e a Paleomagnetic Laboratory Fort Hoofddijk, Budapestlaan 17, 3584 CD, Utrecht, The Netherlands b Department of Geology, Babeş-Bolyai University, Kogălniceanu St. 1, Cluj-Napoca, Romania c Department of Isotope Geochemistry, Faculty of Earth and Life Sciences, Vrije University, De Boelelaan 1085, 1081 HV, Amsterdam, The Netherlands d Department of Palaeontology, Faculty of Geology and Geophysics, University of Bucharest, Bălcescu Bd. 1, Bucharest, , Romania e National Institute of Marine Geology and Geoecology, GeoEcoMar, Dimitrie Onciul Street 23-25, Bucharest, 70318, Romania article info abstract Article history: Received 3 February 2010 Received in revised form 8 July 2010 Accepted 15 July 2010 Available online xxxx Keywords: Central Paratethys Sarmatian Pannonian Magnetostratigraphy 40 Ar/ 39 Ar dating A marked paleoenvironmental change took place at the beginning of the late Miocene in the Central Paratethys, with dominantly marine Sarmatian successions grading rapidly into mainly brackish Pannonian deposits. A long and excellently exposed section comprising the Sarmatian Pannonian transition has been investigated at Oarba de Mureş in the Transylvanian basin (Romania). In this paper, we focus on both radiometric and magnetostratigraphic dating to provide a chronology for the Sarmatian Pannonian transition in Transylvania. Two volcaniclastic layers, located approximately 40 m below the Sarmatian Pannonian transition, yield excellent 40 Ar/ 39 Ar ages. The weighted mean plateau age for biotite and sanidine separates provided isotopic ages of 11.62±0.12 Ma and 11.65±0.13 Ma. This implies deposition during the magnetic chron C5r.2r, which is in agreement with the magnetostratigraphic results of the Oarba de Mureş composite section. Rock magnetic analyses indicate greigite as the main magnetic carrier, with characteristics very similar to the magnetosomal greigite found in the Carpathian foredeep. The newly obtained chronology at Oarba de Mureş constrains the age of the Sarmatian Pannonian transition in the Transylvanianbasin to 11.3±0.1 Ma, slightly younger than the Ma postulated in the Styrian and Vienna Basins Elsevier B.V. All rights reserved. 1. Introduction The Paratethys represents the large epicontinental sea that extended from central Europe to inner Asia since the Oligocene (Laskarev, 1924). It gradually transformed into an inland sea and finally into a series of giant lakes because of ongoing continental collision that shaped the Alpine Himalayan orogenic belt (Rögl, 1998). Tectonic deformations during the Neogene favored the subdivision of this vast paleogeographic realm in three major sub-domains, namely the Western Paratethys, covering the northern Alpine foredeep, the Central Paratethys occupying the area between the Carpathians, Dinarides and Alps (Fig. 1a), and the Eastern Paratethys extending from the Carpathians to the Caspian Sea. Biota assemblages of these subbasins are characterized by recurrent endemism related to the progressive isolation from the open ocean. Consequently, the chronostratigraphy of the Paratethys regions are based on regional (sub)stages. Chronological ages are generally controversial since direct correlations to the standard Geological Time Scales (GTS) are possible only for restricted intervals. Corresponding author. Tel.: ; fax: address: vasiliev@geo.uu.nl (I. Vasiliev). In the Central Paratethys, a major change from marine to brackish water ecosystems took place in the late Miocene at the transition from the Sarmatian (sensu Suess, 1866) to Pannonian, transforming it into the Pannonian Lake (Magyar et al., 1999a,b; Piller and Harzhauser, 2005). An extinction rate of more than 90% for gastropods and foraminifera characterizes the Sarmatian Pannonian-Extinction- Event (SPEE) (Piller et al., 2007). Only few marine taxa survived the newly established conditions and adapted by their subsequent radiation. Additionally, new forms migrated from fresh water environments in the brackish water Lake Pannon. Several factors, such as eustatic sea level lowering (Vakarcs et al., 1994; Harzhauser and Piller, 2004), astronomically-forced changes in regional climate (Lirer et al., 2009), tectonic events in the Carpathians and Dinarides (Săndulescu, 1988), and intraplate stress might have contributed to the isolation of the Central Paratethys from the open ocean. The age of the Sarmatian Pannonian outcropping transition has not been determined by coupled magnetostratigraphic an radiometric dating techniques so far, but was generally derived from sequence stratigraphic correlations to global sea level curves and estimated to coincide with the Middle Late Miocene boundary at 11.6 Ma (Rögl, 1996a,b; Steininger et al., 1996; Piller and Harzhauser, 2005; Hüsing et al., 2007). This is partly related to the lack of long (in time) and continuously exposed sections in the Vienna basin, where the transition is bounded by either sedimentary /$ see front matter 2010 Elsevier B.V. All rights reserved. doi: /j.palaeo

2 2 I. Vasiliev et al. / Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx Fig. 1. (a) The extent of the Central Paratethys basin during the Miocene on the present-day land configuration. (b) Simplified geological map of Transylvanian basin (T.B.) where the southern and central part are covered by Sarmatian and Pannonian sedimentary rocks. The star indicates the location of Oarba de Mureş section. hiatuses or distinct facies changes. Continuous exposures of the transitional interval have been documented from deeper parts of the Panonnian basin in Croatia (Kovačić, 2004; Vasiliev et al., 2007a) (Kovačić and Grizelj, 2006), Serbia (Lesic et al., 2007) and Romania (Sztanó et al., 2005). So far, the magnetostratigraphic record of Central Paratethys is provided by multiple boreholes in the Pannonian Basin (Lantos et al., 1992; Elston et al., 1994; Sacchi et al., 1997; Juhász et al., 1999; Sacchi and Horváth, 2002; Sacchi and Muller, 2004; Magyar et al., 2007). However, the derived polarity pattern is usually not straightforward, permitting several correlations to the geomagnetic time scale (GPTS).The available radiometric dating (Vass et al., 1987; Vass, 1999) is not tieddirectlyto the obtained magnetostratigraphy. In this paper, we present a chronologic framework for the Sarmatian to Pannonian transition of the Oarba de Mureş section in the Transylvanian basin, based on the integration of biostratigraphy, magnetostratigraphy and radio-isotopic data. The results will provide a step forward in the discrimination of global climate (eustatic sea level lowering) versus regional tectonics (orogenic uplift) as the main cause for the isolation of the Central Paratethys at the Sarmatian Pannonian transition. 2. Geological background The present-day Transylvanian depression is bordered by the Eastern and Southern Carpathians and the Apuseni Mountains (Fig. 1b). It developed as a sedimentary basin after the main phase of deformation in the Carpathians (Săndulescu, 1988; Ciulavu et al., 2000). More than 4000 m of sedimentary rocks were deposited in its depocentre during the Late Cretaceous to Neogene. The basin fill can be subdivided into several tectono-stratigraphic mega-sequences (Vancea, 1960; Ciupagea et al., 1970; Broucker et al., 1998; Ciulavu, 1999; Krézsek and Bally, 2006), the latest one being of Middle Miocene age. Here we use the term Transylvanian basin to designate the

3 I. Vasiliev et al. / Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx 3 compressional back-arc basin (Broucker et al., 1998; Ciulavu, 1999; Sanders et al., 1999) that evolved in the hinterland of the Carpathian orogen, filled with Badenian to Pannonian sediments. The Miocene evolution of the Transylvanian Basin was controlled by the final stages of subduction along the Carpathian arc (Săndulescu, 1984; Royden, 1988; Săndulescu, 1988; Ciulavu et al., 2000; Sanders et al., 2002). In contrast to the Pannonian Basin (Tari et al., 1992), no major extensional features can be found, although a very high rate of subsidence persisted from Badenian to Pannonian times (Crânganu and Deming, 1996). Deep-marine siliciclastic turbiditic successions were deposited during the early Badenian in the central part of the basin (typically m thick), while mixed (carbonate and siliciclastic) shelf to coastal deposits are known from the basin margins (Ciupagea et al., 1970; Krézsek and Bally, 2006; Krézsek et al., 2007). In the middle Badenian, the outflow from the basin was closed determining the precipitation of halite in the deep-marine setting and shallow to sabkhatype gypsum in the marginal parts of the basin. Halite deposits can presently be found in the subsurface of almost the entire Transylvanian basin, with thicknesses ranging from a few tens to a maximum of 1800 m (Ciupagea et al., 1970; Krézsek and Bally, 2006). Salt tectonic activity began in the late Miocene and continued until Pliocene, but the main phases of diapirism occurred during the late Sarmatian and Pannonian (Krézsek and Filipescu, 2005; Krézsek and Bally, 2006). The Sarmatian is mainly represented by siliciclastic rocks, but carbonates and evaporites are also present, mainly at the rim of the basin. The thickness of the Sarmatian deposits at the basin centre (in the Tîrnave area Fig. 1b) is more than 1000 m (Krézsek and Filipescu, 2005). In the late Sarmatian, collision between Eurasia and the Tisza Dacia block initiated a tectonic phase associated with nappe emplacements and uplift of the Eastern Carpathians (Săndulescu, 1988; Maţenco, 1997). The low-diversity upper Sarmatian microfossil record reflects progressive changes of water chemistry generated by the closing of connections with the Eastern Paratethys. These environmental changes provided new opportunities for immigrating species to occupy new ecological niches. Ostracods were among the first immigrants, and began endemic radiation under the new conditions, mainly during the Pannonian (Magyar et al., 1999a). Most of the outcropping Pannonian deposits represent deeplacustrine turbidites and hemipelagites from submarine fan systems (Krézsek and Filipescu, 2005). The overlying shallow lacustrine to coastal deposits of Lake Pannon were largely eroded during the Pliocene to Quaternary (Sanders et al., 1999). 3. Integrated stratigraphy of the Oarba de Mureş section A relatively long and excellently exposed section comprising the Sarmatian Pannonian transition is present at Oarba de Mureş in the Transylvanian basin (Fig. 1b). The Oarba section was previously studied for sedimentologic and biostratigraphic purposes focussing on paleoenvironmental and sedimentologic interpretations (Sztanó et al., 2005; Süto and Szego, 2008). The Oarba de Mureş section is a composite of five outcrops (Fig. 2a); four of them (A, B, C and D) have been described by Sztanó et al. (2005), the lowermost extension (Beta) is new (Fig. 2a). Section Beta covers 48 m of Sarmatian deposits and consists of alternations of clayey marls, siltstones and sandstones. It is followed by a non-exposed interval after which outcrop A starts. This part of the section (outcrop A) measures ~51 m and consists of alternating beds of calcareous marls, siltstones, sandstones and thin ash beds (mainly andesitic) and a thick tuff level. According to Vancea (1960), the position of the Sarmatian Pannonian transition at Oarba de Mureş is located at the top of the last significant volcanic ash layer (Oarba Tuff) in section A(Süto and Szego, 2008). Outcrop B consists of ~39 m, outcrop D covers ~22 m, and both are interpreted as representatives of Pannonian deep lacustrine fans (Sztanó et al., 2005). The sections at Oarba de Mureş were sampled for integrated stratigraphic techniques focussing on biostratigraphy (foraminifera and ostracods), magnetostratigraphy and radio-isotopic dating. At least two standard paleomagnetic cores were taken with an electrical drill and a generator as power supply. The upper parts of all sections could only be sampled using climbing gear, resulting in reduced resolution. A total number of 213 levels were sampled, 34 from outcrop Beta, 117 from A, 49 levels from B and 13 samples from D. Additional samples have been collected for rock-magnetic, biostratigraphic, sedimentological and geochemical studies Biostratigraphy Foraminifera are usually not very abundant and assemblages are dominated by juveniles, mainly because of deep, poorly oxygenated environments in the Transylvanian basin. Most specimens were probably transported from shallower depth. Biostratigraphically, the foraminifera in section A belong to the late Sarmatian Porosononion aragviensis Biozone (Popescu, 1995). The assemblage includes very rare agglutinated taxa (Ammodiscus sp., Glomospira charoides, Trochammina kibleri) and globigerinids (Globigerina tarchanensis, Streptochilus spp.). The common species belong to the rotaliids (Ammonia beccarii, A. tepida, Discorbis effusus, Elphidium flexuosum, E. nataliae, E. subangulatum, Nonion biporus, N. bogdanowiczi, N. commune, Porosononion aragviensis, P. hyalinum, P. martkobi, P. subgranosum), miliolids (Articulina bidentata costata, A. problema, A. sarmatica, A. vermicularis, Cycloforina predkarpatica, Flintina georgii, Pseudotriloculina consobrina, Quinqueloculina gracilis, Q. fluviata, Sarmatiella moldaviensis, S. prima, Varidentella pseudocostata, V. reussi, V. sarmatica), and buliminids (Bolivina arta, B. moldavica, B. pseudoplicata, B. sarmatica, Buliminella elegantissima, Caucasina sp., Fissurina bessarabica). The composition and abundance of foraminifera assemblages allow the identification of several periodicities in the migration of depositional environments, in relation to climatic and base-level changes (Filipescu et al., 2009). The Sarmatian ostracod assemblages are very scarce in taxa and most of them (Aurila merita, Hemicytheria omphalodes omphalodes, Callistocythere egregia, Leptocythere tenuis, Loxoconcha sp., Argiloecia sarmatica) are in the juvenile stage and probably reworked from the marginal areas. In contrast, the Pannonian deposits (outcrops B, C and D) contain a rich ostracod association specific for low salinities), represented by: Amplocypris abscissa, Candona (Lineocypris) transilvanica, Candona (Lineocypris) sp., Candona (Typhlocypris) ornata, Candona (Serbiella) sp., Candona (Caspiolla) alsi, Candona (Caspiolla) spamplocypris recta, Hungarocypris hieroglyphica, Cypria sp., Leptocythere sp., Leptocythere (Amnicythere) monotuberculata, Callistocythere egregia, Loxoconcha kolubarae, Loxoconcha djaffarovi, Loxoconcha granifera, Hemicytheria dubokensis, andcyprideis pannonica. Other Sarmatian microfossils belong to the group of diatoms (Coscinodiscus sp.), dasyclad algae (Halicoryne moreletti), mysids (Sarmysis sarmaticus, S. vancouveringi)andfish (bones, teeth, and otoliths). Our results thus indicate a very clear difference between the Sarmatian and the Pannonian micropaleontological assemblages. The last Sarmatian foraminifera (Prosononion aragviensis Biozone) occur near the top of the section A (sample 11, approx 2.5 m below the top), in an assemblage characterised by miliolids (Sarmatiella spp., Articulina spp.) The assemblages in the samples above 56 m, up to the top of the section A, contain typical Pannonian deep-water ostracods (Amplocypris Candona group). Based on our biostratigraphic observations, the Sarmatian Pannonian transition is located 2.5 m below the top of section A (Fig. 2b, c). This is in excellent agreement with earlier results that located the transition about 3.4 m from the top of section A, based on the massive occurrence of the acritarch Mecsekia ultima (Süto and Szego, 2008). However, Süto and Szego (2008) also observed that the index taxon for the base of the Pannonian (Spiniferites bentonii pannonicus) occurs only at 1.4 m below the top of section A. Foraminifera assemblages in the outcrop Beta from Oarba de Mure have a quite similar paleoecologic behavior compared to section A. Most of the species are also very common for the same biostratigraphic unit

4 4 I. Vasiliev et al. / Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx Fig. 2. (a) Photograph of the five outcrops (Beta, A, B, C and D) from Oarba de Mureş with the acknowledged position of the Sarmatian Pannonian transition in outcrop A. (b) Photograph of the four outcrops (A, B, C and D) from Oarba de Mureş with the acknowledged position of the Sarmatian Pannonian transition in the outcrop A. (c) Outcrop A with the position of the sampled and analyzed volcaniclastic layers indicated with arrows. (Porosononion aragviensis Biozone of the late Sarmatian). This suggests a high rate of sedimentation for the late Sarmatian Magnetostratigraphy Rock magnetism Several rock-magnetic experiments were carried out on bulk samples to identify the carriers of the magnetisation. First, thermomagnetic runs were measured in air with a modified horizontal translation type Curie balance with a sensitivity of approximately Am 2 (Mullender et al., 1993). Approximately mg of powdered samples were put into a quartz glass sample holder and were held in place by quartz wool; heating and cooling rateswere10 C/min.Thethermomagneticrunsshowinallcasesthe alteration of an iron sulphide, transforming above ~400 C to a more magnetic phase (pyrrhotite and magnetite) and finally above C, to hematite (Fig. 3). The majority of the samples shows an irreversible decrease in magnetisation with increasing temperature up to ~410 C which is typical of greigite (Fig. 3a, b). In some cases (Fig. 3c, d) a reversible decrease in magnetisation is observed, which is characteristic for pyrrhotite and/or magnetite. Next, an alternating gradient magnetometer (MicroMag Model 2900 Princeton, noise level Am 2 ) was used to successively measure (at room temperature) hysteretic loops, isothermal remanent magnetisation (IRM) acquisition, and first order reversal curves (FORC) diagrams. Samples masses ranged mg. The hysteresis loops are generally of good quality and indicate the presence of a low-coercivity component (Fig. 4a, d, h). They were performed up to maximum±1 T and are (almost) closed in fields of 300 mt. The isothermal remanent magnetisations show saturation below 300 mt (Fig. 4b, f, i). The quality of the FORC diagrams varies (Fig. 4c, g, j) but most examples show contours closing around a peak of ~55 mt. The central peak has considerable spread and is centred below B u =0 indicating strong magnetic interaction among the particles. These FORC diagrams indicate high similarities with the greigite bearing samples including those from the Paratethys (Roberts and Weaver, 2005; Rowan and Roberts, 2006; Vasiliev et al., 2007b; Vasiliev et al., 2008). The rock magnetic properties indicate that most probably the magnetic carrier is pseudo-single domain greigite Demagnetisation results Stepwise thermal demagnetisation has been applied to at least one sample per stratigraphic level to determine characteristic remanent magnetisation (ChRM) directions. The demagnetisation was performed with temperature increments of 5 30 C up to a maximum temperature of 420 C in a magnetically shielded, laboratory-built, furnace with a residual field less than 10 nt. The natural remanent magnetisation (NRM) was measured on a 2G horizontal Enterprises DC SQUID cryogenic magnetometer (noise level Am 2 ). To monitor the possible mineralogical changes during the thermal treatment the bulk susceptibility was measured on a Kappabridge KLY-2 after the 20, 180, 240 and 390 C

5 I. Vasiliev et al. / Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx 5 Fig. 3. Representative thermomagnetic runs for selected samples. Heating (red solid lines) and cooling (blue dashed lines) were performed with rates of 10 C/min. Total magnetisation is plotted in a series of runs to increasingly higher temperatures. Individual data points have been omitted for clarity. Cycling field varied between 150 and 300 mt. Panels a, b and c show important alteration after heating above 400 C most probably as the effect of the pyrite transformation to magnetite. Panel d represents the detailed part of the run until ~425 C; the outof-scale trend from the panel c is limiting the view of the shape of the heat cool cycle. The measurements have been performed in air with a modified horizontal translation-type Curie balance (Mullender et al., 1993). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) temperature steps. Additional alternating field demagnetisation was performed with small field increments, up to a maximum of 100 mt, using an in-house built robotized sample handler, attached to a horizontal 2G Enterprises DC SQUID cryogenic magnetometer. The thermal demagnetisation diagrams for the outcrops Beta, B and D are in general of good quality and the polarity of these subsections could be convincingly interpreted. The thermal demagnetisation diagrams from outcrop A showed complex NRM behaviour, with up to four differently directed and partially overlapping components (Fig. 5). Consequently, the isolation of the primary component was rather challenging and it was decided to distinguish the directions of all components and perform a thorough analysis of the temperature range in which they were depicted. In our interpretation, four temperature components were distinguished and we coded them as: T1 for the interval C, T2 for C, T3 for C, and T4 for C. T1, T2 and T3 are the highest temperature components in our samples. T4 was the low temperature component and is interpreted as a secondary overprint. This T4 had in several cases clearly reversed directions thus we can conclude that reversed geomagnetic fields were overprinting the original directions as well. In several cases no clear decay to the origin was observed and therefore we plotted the directions using the great circle method of McFadden and McElhinny (1988) (Fig. 5g). We assume that the primary component is partially overprinted by a secondary component caused by overlapping blocking temperature spectra. When plotted on the equal-area diagram the remanence vectors progressed along a great circle toward a northerly (or southerly) downward (upward) direction indicating removal of a reversed (normal) phase from a reversed (normal) primary component. In these cases a best fitting great circle plane was used to estimate the characteristic directions according to the method of McFadden and McElhinny (1988). On the basis of the Zijderveld diagrams and rock magnetic analysis we divided the samples in two main groups Group 1. The thermal demagnetisation diagrams are described by a linear decay of the NRM to a temperature of C (or maximum of 390 C), when the NRM was entirely removed (Fig. 5a, c j). Heating these samples to higher temperatures resulted in an increase in susceptibility and intensity. This behaviour is the result of the oxidation of a non-magnetic iron sulphide transforming it into a magnetic phase and was also observed during the Curie balance measurements. A part of these diagrams show a large secondary (possibly near primary) overprint, which is removed at 240 C (Fig. 5a, b). In some cases the low temperature (later acquired) component could be removed only after heating above 300 C (Fig. 5g, i). The samples of this group are characterised by the presence of the higher temperature components: T1 ( C) and T2 ( C). For some samples the lower temperature component T3 ( C) was the last direction we could obtain since heating to further steps resulted in increasing susceptibilities. The samples of group 1 were interpreted as carrying the characteristic remanent magnetisation and ultimately provided the primary NRM directions used for the magnetostratigraphic record Group 2. Thermal diagrams show a (pseudo-) linear behaviour that can be observed until 240 C (Fig. 5j, k, q). Even though the NRM is not entirely removed in these samples, an increase of intensity and susceptibility is observed after 240 C (Fig. 5j, q) indicating the presence of an iron sulphide. Further demagnetisation at higher temperatures (N240 C) results in mainly randomly directed components (Fig. 5j, q) or in a cluster around the origin (Fig. 5k). Usually, these demagnetisation diagrams either fail to reach/or pass the origin (Fig. 5j) and we assume that the primary component is totally overprinted by a secondary component. These demagnetisation diagrams were mostly found in outcrop A. For this group of samples we could distinguish only the presence of the lower temperature component T4 ( C). For our magnetostratigraphic interpretation we used the highest temperature component found in the Zijderveld plots of each sample.

6 6 I. Vasiliev et al. / Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx Fig. 4. Rock magnetic experiments. Hysteresis loops, (a, d and h), IRM and back-field curves (b, f and i) and FORC (c, g and j) for characteristic samples. The hysteresis loops, IRM and back-field curves were measured for 1 T B 1 T. The hysteresis figures show the results up to ±300 mt (the important part of the loop) with applied paramagnetic contribution and mass correction. The sample codes are indicated in the left-down part and are followed by the hysteresis loops parameters. Also displayed are: saturation magnetisation (M s ), the remanent saturation magnetisation (M rs ) and coercive force (B c ). The (remanent) saturation magnetisations were determined after correction for the paramagnetic contribution and displayed on mass-specific basis. The FORC diagrams have indicated the smoothing factors (SF); they are presented at 10 contours levels. For instance, in the diagrams clearly showing the presence of both T1 ( C) and T3 ( C) components, T1 is considered as the ChRM (Fig. 5b, i). In the cases where, between 240 and 390 C, a single direction was found, we coded it as T2, obviously being the highest temperature component and was interpreted as the ChRM one (Fig. 5c, d, f, l, m, o, p, r, s). Temperature component T4 ( C) is not plotted on the declination/inclination record despite the fact that, in some cases, its direction is similar to the ChRM of samples located upwards or downwards in stratigraphy, seemingly not affected by any overprint (Fig. 5q) Ar/ 39 Ar dating Six volcaniclastic layers were sampled for 40 Ar/ 39 Ar dating; four were collected from outcrop A and two from outcrop B. The bulk samples were washed and sieved. Standard heavy liquid and magnetic separation techniques were applied to obtain K-feldspar and mica separates. Crystal sizes were mostly between 300 and 500 μm, while some were in the range of μm. After careful hand picking, the biotite fraction was leached using HNO 3 while the K-feldspar fraction was leached with HF solution. After this final cleaning step, the crystal separates were wrapped in Al-foil and loaded in a 6 mm ID quartz vial. Samples were irradiated for 7 h in the Cd-lined RODEO facility of the EU-Petten High Flux Reactor (The Netherlands). Fish Canyon Tuff (FCT) sanidine, used as neutron fluence monitor, was loaded at the top and bottom of the vial and between each set of 3 unknowns. After irradiation, samples and standards were loaded in holes, 2 mm in diameter, of a copper tray and placed in an ultra-high vacuum extraction line. 40 Ar/ 39 Ar analyses were performed at the Vrije Universiteit Amsterdam. Samples and standards were fused with a 20 W argonion continuous wave laser beam and the resulting gas was analyzed with a Mass Analyzer Products LTD noble gas mass spectrometer. Beam intensities were measured in peak-jumping mode in 0.5 mass intervals over the mass range on a Balzers secondary electron multiplier. System blanks were measured every 3 steps. Mass discrimination was monitored by frequent analysis of aliquots of air. The irradiation parameter J for each unknown was determined by interpolation using a 2nd order weighted polynomial fit between the individually measured standards. Ages are calculated using the in-house developed ArArCalc software (Koppers, 2002) based on the recently astronomically calibrated age of ±0.28 Ma for FCT (Kuiper et al., 2008) and the decay constant values of Steiger and Jäger (1977). Correction factors for neutron interference reactions are for ( 36 Ar/ 37 Ar) Ca, for ( 39 Ar/ 37 Ar) Ca, for ( 38 Ar/ 39 Ar) K and for ( 40 Ar/ 39 Ar) K. Errors are quoted at the 1σ level and include the analytical error and the analytical error in J. The external error additionally includes decay constant uncertainties (as reported in Steiger and Jäger, 1977) and the uncertainty in standard age.

7 I. Vasiliev et al. / Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx 7 Fig. 5. Representative thermal demagnetisation diagrams (after tilt correction) with the directions of the different temperature components: T1 red solid lines, T2 orange solid line, T3 blue dashed lines, T4 green dashed lines. (b). The selected examples are displayed in stratigraphic order from old (a) to young (s). Solid (open) circles denote projection on the horizontal (vertical) plane and the attached numbers indicate temperatures in C. The samples code (in capital letters) is located in the right-top corner; stratigraphic levels (in metres) are indicated in the left-down corner. The panels a to d are examples of diagrams from of section Beta, e to l from section A, m to p from B and q to t are from D. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

8 8 I. Vasiliev et al. / Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx From the six analyzed samples two yielded reproducible ages. The two thinner ash levels, OM-C and OM-D located 40 and 10 cm below the distinct white tuff (OM5-C level) respectively (Fig. 2b) provided good age information. Repeated total fusion of multiple grain biotite experiments yield a weighted mean age of 11.62±0.12 Ma for OM-C (Table 1) while experiments on OM-C feldspar crystals yield an age of 11.64±0.14 Ma (Table 2). Similar multiple grain total fusion experiments on feldspar separates of OM-D provided a weighted mean age of 11.65±0.13 Ma (Table 3). Age probability diagrams (Fig. 6) do not show a unimodal age distribution, but the major peak occurs around Ma for OM-C biotite (Fig. 6a) and 11.7 Ma for OM-C feldspar (Fig. 6b). For OM-D feldspar, the peak occurs at Ma. These small differences might be the result of a minor detrital component in the fused multiple grain samples. Therefore, the age of sedimentation of these thin volcaniclastic layers might be slightly younger. The two main volcaniclastic levels from outcrop A indicated in Fig. 2 as OM-B ( Oarba Tuff ) and OM5-C (the white volcaniclastic level, 1 m thick) provided extremely heterogeneous ages and consist of reworked volcaniclastic material. The volcaniclastic layers from outcrop B did not provide sufficient crystals for 40 Ar/ 39 Ar dating. 4. Chronologic framework for the Oarba de Mureş section The Zijderveld diagrams are of mixed quality for outcrop Beta (Fig. 5a d) and A (Fig. 5e k), but good for section B (Fig. 5l, m, o) and D (Fig. 5p, r s). The polarity pattern of outcrop Beta (Fig. 7) startswitha normal polarity interval followed by a reversed interval and ends with normal polarity again. For outcrop A (Fig. 7) the polarity of the samples is changing frequently. The distinction between the characteristic magnetisation and the overprint was delicate and difficult. The majority of the samples show reversed polarity, but several short intervals of normal polarity are also present. Section B (Fig. 8) starts with a normal polarity interval, followed by a reversed polarity one and ends with a normal interval. In outcrop D (Fig. 8) all samples are of normal polarity. The combined weighted mean of the plateau ages of the volcaniclastic layers of section A, located approximately 40 m below the Sarmatian Pannonian transition, provide isotopic ages of 11.62± 0.12 Ma for OM-C and 11.65±0.13 Ma for OM-D. The radio-isotopic ages indicate that deposition must have taken place within magnetic chron C5r, which is known to comprise several short normal sub-chrons (Abdul Aziz and Langereis, 2004; Krijgsman and Kent, 2004). The dated volcaniclastic layers correspond to a normal polarity located below a long reverse polarity interval. The most logical correlation of this normal polarity interval is thus to chron C5r.2n (Fig. 9). The Zijderveld diagrams at the base of Oarba A show frequent changes of demagnetisation behavior and the reliability of the polarity pattern is not straightforward. Diagenetic processes, such as delayed acquisition and lock-in-depth mechanisms may have generated secondary iron sulphides which are difficult to separate from the original signal. The dominantly reversed polarity interval of the main outcrop A correlates very well to chron C5r, with the short normal interval at m to C5r.2r-2n ( Ma). The polarity pattern of the downward extension of Oarba Beta correlates then to chron C5An between 12.4 and 12.0 Ma. The normal polarity interval at the upper part of Oarba B and Oarba D are interpreted to represent the long normal polarity chron C5n.2n (Fig. 9). The dated volcaniclastic layers are located ~40 m below the Sarmatian Pannonian transition. Since the section was deposited in a deep marine environment the expected sedimentation rate is likely to be low and, the 40 m interval would cover significant amount of time. The combination of our magnetostratigraphy and radiometric data indicates that the Sarmatian Pannonian transition in the Transylvanian basin has an age of 11.3±0.1 Ma (Fig. 9). 5. Discussion 5.1. Greigite as the main magnetic carrier in Oarba de Mureş In the Central Paratethys, magnetostratigraphic dating was commonly considered a challenge. The polarity pattern provided by multiple boreholes in the Pannonian Basin (Lantos et al., 1992; Elston et al., 1994; Sacchi et al., 1997; Juhász et al., 1999; Sacchi and Horváth, 2002; Sacchi and Muller, 2004; Magyar et al., 2007) is usually not straightforward, permitting several correlations to the GPTS. The inferences are nonunequivocal and a comprehensive correlation to the GPTS for the Table 1 40 Ar/ 39 Ar data derived from biotites of OM-C volcaniclastic layers (Oarba de Mureş). Errors are reported including the analytical uncertainty in the sample (I), the analytical uncertainty in sample and standard. All errors are reported at the 1σ level. The not-marked values were excluded from the age calculation due to the big errors (1σ). Incremental heating 36 Ar(a) 37 Ar(ca) 38 Ar(cI) 39 Ar(k) 40 Ar(r) Age±lσ 40 Ar(r) 39 Ar(k) K/Ca±lσ (Ma) (%) (%) 06MX164E Fuse ± ± MX164F Fuse ± ± MX164H Fuse ± ± MX1641 Fuse ± ± MX164K Fuse ± ± MX164L Fuse ± ± MX164M Fuse ± ± MX148E Fuse ± ± MX148F Fuse ± ± MX148H Fuse ± ± MX148I Fuse ± ± MX148K Fuse ± ± MX148L Fuse ± ±0.648 Σ Information on analysis Results 40(r)/39(k) ±1σ Age±1σ MSWD 39 Ar(k) K/Ca±1σ (Ma) (%, n) vu59-5 Error plateau ± ± ±1.945 Biotite ±0.18% ±0.35% 11 OM-C External error ± Statistical T ratio IV/KK Analytical error ± Error magnification Project=WPSP Total fusion age ± ± ±0.016 Irradiation=VU59 ±0.14% ±0.33% J = ± External Error ±0.12 FC= ±0.161 Ma Analytical Error±0.02

9 I. Vasiliev et al. / Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx 9 Table 2 40 Ar/ 39 Ar data derived from feldspars of OM-C volcaniclastic layers (Oarba de Mureş). See also figure captions from Table 1. Incremental heating 36 Ar(a) 37 Ar(ca) 38 Ar(cl) 39 Ar(k) 40 Ar(r) Age±1σ 40 Ar(r) 39 Ar(k) K/Ca±lσ (Ma) (%) (%) 06MX145A Fuse ± ± MX145B Fuse ± ± MX145C Fuse ± ± MX145E Fuse ± ± MX145F Fuse ± ± MX145G Fuse ± ± MX145L Fuse ± ± MX145M Fuse ± ± MX145N Fuse ± ± MX145P Fuse ± ± MX162A Fuse ± ± MX162B Fuse ± ± MX162C Fuse ± ± MX162E Fuse ± ± MX162F Fuse ± ± MX162G Fuse ± ± MX163F Fuse ± ±0.020 Σ Information on analysis Results 40(r)/39(k)±1σ Age±1σ MSWD (Ma) 39 Ar(k) K/Ca ±1σ (%,n) vu59-1 Error Plateau ± ± ±0.153 Feldspar ±0.59% ±0.66% 10 OM-C External error ± Statistical T ratio IV/KK Analytical error± Error magnification Project=WPSP Total Fusion Age ± ± ±0.000 Irradiation=VU59 ±0.65% ±0.71% J = ± External error ±0.14 FC-2= ±0.161 Ma Analytical error ±0.07 Sarmatian Pannonian transition interval remains weak (Vakarcs et al., 1994; Magyar et al., 1999b). The magnetostratigraphic dating of Sarmatian Pannonian transition in Našice quarry (Croatia) was unsuccessful, almost the entire record indicating present day field overprinted directions (Vasiliev et al., 2007a). The Central Paratethys magnetostratigraphies are mainly based on greigite bearing samples and, as a rule, the demagnetisation was performed using alternating fields. Greigite reliability as primary magnetic carrier has been regularly questioned because it forms generally after deposition, during diagenesis, while the timing of the NRM acquisition was often considered not well-constrained. Recent greigite-based magnetostratigraphies straightforwardly correlate to the GPTS supporting the early (near-primary) formation and preservation of greigite in sedimentary rocks (Vasiliev et al., 2007b; Hüsing et al., 2009). New transmission electron microscopy studies on samples from the Carpathian foredeep indicate the preservation of greigite Table 3 40 Ar/ 39 Ar data derived from biotites of OM-D volcaniclastic layers (Oarba de Mureş). See also figure captions from Table 1. Incremental Heating 36 Ar(a) 37 Ar(ca) 38 Ar(cl) 39 Ar(k) 40 Ar(r) Age±lσ 40 Ar(r) 39 Ar(k) K/Ca±1σ (Ma) (%) (%) 06JX153L Fuse ± ± MX154F Fuse ± ± MX154G Fuse ± ±0.353 O6MX1 541 Fuse ± ± MX154J Fuse ± ± MX154K Fuse ± ± MX154M Fuse ± ± MX154N Fuse ± ± MX154O Fuse ± ± MX154T Fuse ± ±1.980 Σ Information on analysis Results 40(r)/39(k) ±1σ Age±1σ MSWD 39 Ar(k) K/Ca±1σ (Ma) (%, n) vu59-a3 Error Plateau ± ± ±5.484 Feldspar ±0.34% ±0.45% 6 OM-D External error ± Statistical T ratio IV/XK Analytical error± Error magnification Project=WPSP Total Fusion Age ± ± ±0.019 Irradiation=VU59 ±0.22% ±0.37% J = ± External error ±0.13 FC-2= ±0.161 Ma Analytical error ±0.03

10 10 I. Vasiliev et al. / Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx Fig. 6. Cumulative probability distributions for 40 Ar/ 39 Ar biotite and sanidine ages of Oarba de Mureş volcaniclastic layers. (a) OM-C with the probability distributions for 40 Ar/ 39 Ar ages derived from biotite. (b) OM-C with the probability distributions for 40 Ar/ 39 Ar ages derived from K-feldspar. (c) OM-D with the probability distributions for 40 Ar/ 39 Ar K-feldspar ages. magnetofossil crystals (Vasiliev et al., 2008), a direct product of bacterial magnetotactic sensing (Blakemore, 1975; Frankel et al., 1979; Mann et al., 1990; Bazylinski et al., 1993). Greigite seems to be a characteristic feature for the Upper Miocene Pliocene succession of Paratethys. It was identified as the main carrier of the magnetisation in the Miocene successions of the Eastern Paratethys (Vasiliev et al., 2005, 2007b, 2008; Vasiliev, 2006; Krijgsman et al., 2010). Babinszki et al. (2007) monitored the presence of greigite in a multitude of locations in the Central Paratethys. Examining the demagnetisation plots from Magyar et al. (2007) the observed acquisition of gyroremanent magnetisation (GRM) strongly suggests the presence of greigite in the Pannonian to recent sedimentary rocks from Hungary. We argue that both Central and Eastern Paratethys waters favoured the formation of greigite, because the paleoenvironment was the end-result of a combination of oxygen-depleted bottom waters, bacterial sulphate reduction (Berner, 1984; Horng et al., 1998), clay availability, higher concentration of reactive iron, higher total organic carbon associated with organic matter decomposition and low total sulphur content (Kao et al., 2004). The current study adds to the list of greigite as the main magnetic carrier in Paratethys. The rock magnetic properties performed on samples of group 1 from Oarba de Mureş clearly indicate greigite as main magnetic carrier. Similar to the Carpathian foredeep (Vasiliev et al., 2008), the thermal demagnetisation diagrams indicate the presence of (at least) two different (even antipodal) NRM components in the same specimen close to polarity reversals (Figs. 10a, b and 4b, g, i). Unpredictably, the hysteresis loops (Fig. 4a, d) and the FORC diagrams (Fig. 4c, g) indicate only a unimodal coercive force distribution, usually not characteristic for a specimen recording two different directions. In this case we fitted the isothermal remanent magnetisation (IRM) acquisition curve, with two magnetic components having approximately the same mean acquisition field distribution (B 1/2 ) but significantly different dispersion parameters (Fig. 10c, d). We interpret the IRM acquisition curve by two different grain-size distributions having the same mean coercivity (Vasiliev et al., 2008). The wider distribution we attributed to the authigenic phase of greigite, formed some time after deposition (Fig. 10). The high-temperature component, with the narrowest grain size distribution from our record is interpreted as evidence for greigite magnetosomes, formed at the time of deposition and thus recording the magnetic field without (significant) delay. The similarity between the rock-magnetic properties of the Transylvanian samples and the ones from the Carpathian Foredeep (Vasiliev et al., 2008) may indicate that for some samples the earliest NRM was acquired by magnetotactic bacteria. Therefore, we argue that in our section (for the outcrops B and D and parts of outcrops Beta and A), the magnetisation of greigite was acquired at latest during early-diagenesis of the sediments. Moreover, at the levels where we could identify the component between 300 and 390 C, we interpreted this as being carried by magnetosomal greigite and therefore recording the primary direction of the magnetic field at the time of sediment deposition. A full confirmation of this assumption requires extensive transmission electron microscopy. However, all rock magnetic characteristics are strikingly similar to the magnetosomal greigite of the Carpathian foredeep (Vasiliev et al., 2008). The paleomagnetic directions obtained from demagnetisation diagrams of the interval around the dated volcaniclastic layer from Oarba A outcrop are complex and difficult to interpret. Only three of them show clear and consistent reversed directions but two of these indicate normal field components at higher temperatures. Chron C5r is known to comprise two well established subchrons and three additional short subchrons (C5r.2r-1n, C5r.2r-2n and C5r.3r-1n) (Abdul Aziz and Langereis, 2004; Krijgsman and Kent, 2004). The positions of C5r.2r-2 and C5r.3r-1 in the geomagnetic time scale are marked in Fig. 9. The fluctuations recorded in our sections could be related to the high rate of reversals but diagenetic processes, the delayed acquisition and the lock in depth mechanisms are other possibilities for the observed complex magnetic behaviour. The association of gypsum with weak zones (fractures, layers surfaces) indicates a diagenetic origin of the gypsum. The presence of secondary gypsum in the section may indicate late diagenetic remagnetisation involving greigite formation after gypsum (Roberts and Weaver, 2005). Fig. 7. Polarity zones and schematic lithological columns for the sections Beta (left) and A (right). The position of the Sarmatian Pannonian transition is marked in section A. In the polarity columns black (white) denotes the normal (reversed) polarity intervals. Solid dots represent reliable direction of demagnetisation (ChRM). We show only the highest temperature component (T1, T2 or T3) identified in each sample. In the cases where only T4 is present the direction is not plotted. The black signal line connects only the straightforward directions. The lithological columns display variations of clayey-marly and silty layers (darker grey), sandy units (lighter grey), yellow sandy units (yellowish) and volcaniclastic layers (red). Note the very sandy interval at the top of outcrop A (marked in yellow) having no reliable magnetostratigraphy. The position of the sampled and dated volcaniclastic layers is indicated with arrows (the age is highlighted). The volcaniclastic layers of which the age was not datable (black arrows) are marked with n.d. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

11 I. Vasiliev et al. / Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx 11

12 12 I. Vasiliev et al. / Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx Fig. 8. Polarity zones and the schematic lithological columns for the outcrops B (left) and D (right). For further explanation see caption in Fig Correlation of the Sarmatian Pannonian transition within the Paratethys and to the standard chronostratigraphy The confusion within the chronostratigraphy in the Paratethys in general is well recognized. The best examples are: the unfortunate and misleading (Piller and Harzhauser, 2005) use of the Sarmatian name in the Eastern Paratethys; but also the use of the Pontian in the Central Paratethys for the deposits corresponding to Transdanubian (between ca 9.0 and 7.4 Ma) (Sacchi et al., 1999); some of the papers sustain the idea of completely excluding the term Pontian from Central Paratethys terminology (Sacchi and Horváth, 2002). These disagreements lead to confusion in the definition of the stages and also harden the dating work. Up to date, not a single boundary stratotype has been defined in the Central Paratethys and the stages are all bounded by either sedimentary hiatuses or distinct facies changes, inferred to mark lowstands in sea level (Piller et al., 2007). Generally, the Sarmatian Pannonian transition was considered to be synchronous with the Serravallian Tortonian transition of the Mediterranean domain (Harzhauser and Piller, 2004; Piller et al., 2007) placednowat11.62ma(hüsing et al., 2009). Lately, Lirer et al. (2009) derived an age of Ma for the Sarmatian Pannonian transition, suggesting a relation with oxygen isotope event Mi5. Despite the increasing effort invested in tuning the upper Miocene Central Paratethys sedimentary successions to the APTS (Sprovieri and Sacchi, 2003; Harzhauser and Piller, 2004; Harzhauser and Piller, 2007; Lirer et al., 2009), the absence of well-dated Sarmatian Pannonian rocks is still suppressing the potential advantages of astrochronology. The combined results of magnetostratigraphy and 40 Ar/ 39 Ar dating presented in this study indicate that the Sarmatian Pannonian transition in the Transylvanian basin has an age of 11.3±0.1 Ma, ~300 kyr younger than the age derived at Sarmatian Pannonian transition at the typical locality (Nexing section) in northern Vienna basin. However, the age in Vienna basin has not been determined by magnetostratigraphic or radiometric dating techniques so far, but was derived from sequence stratigraphic correlations to global sea level curves and estimated to

13 I. Vasiliev et al. / Palaeogeography, Palaeoclimatology, Palaeoecology xxx (2010) xxx xxx 13 Fig. 9. Correlation of the polarity sequence of Oarba de Mureş section to the APTS (Lourens et al., 2004). The solid lines between the section record and APTS connect (interpretative) simultaneous polarity boundaries. The names of the subchrons are in the column attached to the APTS. The global deep-sea record of Zachos et al. (2001) is provided for comparison. The light grey band marks the uncertainty in the age of the Sarmatian Pannonian transition. Main trends in eustatic sea level are generalised from Hardenbol et al., coincide with the Middle Late Miocene boundary at 11.6 Ma (Rögl, 1996a,b; Steininger et al., 1996; Piller and Harzhauser, 2005; Hüsing et al., 2007). Regardless the age discrepancy between the two estimates for the Sarmatian/Pannonian boundary, the 11.3±0.1 Ma from Transylvanian basin had probably no direct relation with the global climatic signals Tectonic control for the Sarmatian Pannonian isolation event At the Sarmatian Pannonian transition, the Central Paratethys suffered a major water circulation restriction expressed in changing biota as a response to the total isolation from the open marine system. Two main mechanisms are used to explain the isolation of Central Paratethys: 1) glacio-eustatic sea-level lowering (cycle TB3.1) (Vakarcs et al., 1994; Harzhauser and Piller, 2004; Piller et al., 2007; Lirer et al., 2009) and 2) tectonic uplift by thrusting of the Carpathians (Săndulescu, 1988; Horváth and Cloetingh, 1996). Our newly obtained age of 11.3±0.1 Ma for the Sarmatian Pannonian transition in the Transylvanian basin is ~300 kyr younger than the Serravallian Tortonian boundary (Hüsing et al., 2007). Therefore we argue that correlation of the Sarmatian Pannonian transition recorded in Transylvanian basin to Mediterranean events and/or eustatic sea-level falls cannot be confirmed. The age of 11.3±0.1 Ma obtained for the Sarmatian Pannonian transition has also no equivalent isotope event occurring in the oceanic domain (Fig. 9). However, the global climatic signals cannot be excluded from the geological history of the Central Paratethys but the tectonics seemed to be the primary cause for the