Palaeoecological and sedimentological characterisation of Middle Miocene sediments from the Hrvatska Kostajnica area (Croatia)

The Miocene deposits of the Hrvatska Kostajnica (KOS-I) area belong to the south-western marginal part of the Pannonian Basin System (PBS). Investigation of the lithostratigraphical column included: mineralogical, geochemical, sedimentological and integrated palaeontological (calcareous nannofossil, foraminifers, ostracodes, palynomorphs) analyses. Badenian and Sarmatian sediments of this column were deposited in a marine offshore environment with local input of terrigenous material represented by marls and silty marls. Based on palaeontological data, the recorded palaeoclimate was subtropical in the late Badenian changing to a warm temperate climate of the early Sarmatian. Marly sediments predominantly consist of carbonate (calcite and aragonite) and clay minerals, while quartz and plagioclase are less abundant. Most samples contain a small amount of zeolite minerals from the clinoptilolite/heulandite series. Among the clay minerals, smectite and illite/muscovite are the most abundant. Based on provenance analyses we concluded that the Badenian-Sarmatian marls were predominantly formed by the weathering of acidic (Si-rich) source rock derived material from the neighbouring Inner Dinarides. in the offshore environment (Vejalnica formation) and coarsegrained material in the shoreface depositional environment (Trstenik member and Vrapče formation) (AVANIĆ et al., 2018). During the Sarmatian, reduced salinity occurred, while sediments were represented by shallow-water conglomerates, calcarenites and limestones (the Pećinka member) as well as horizontally laminated pelitic sediments (the Dolje formation) deposited in a deeper marine environment (AVANIĆ et al., 2018; PAVELIĆ & KOVAČIĆ, 2018 and references therein). Studies of the provenance of the Middle Miocene sedimentary rocks of the SW margin of NCB are deficient. The only reliable data about the provenance of these sedimentary rocks are from areas N and NE of Hrvatska Kostajnica investigated by KOVAČIĆ et al. (2011) and GRIZELJ et al. (2017). According to these investigations in the area of Medvednica Mt. and the Slavonian Mts., the sources of material for the Middle Miocene clastic sedimentary rocks were located to the south of the investigated area, in the area of the northern Inner Dinarides, while some of the material was also supplied from locally uplifted blocks in the SW part of the PBS. The aim of this paper is to reconstruct the Badenian/Sarmatian transition in the area of the SW marginal part of the NCB and PBS based on a multiproxy approach: mineralogical, geochemical, sedimentological and integrated palaeontological analyses (calcareous nannofossil, foraminifers, ostracodes, palynomorphs determination) and palaeoecological reconstruction putting the results in the frame of new chronostratigraphic divisions within the Middle Miocene. Furthermore, the aim of the present study is also to make a contribution to the discussion about the evolution, palaeogeography and climate history of the marine succession in the south-western PBS. Article history: Manuscript received July 29, 2020 Revised manuscript accepted October 05, 2020 Available online October 23, 2020

in the offshore environment (Vejalnica formation) and coarsegrained material in the shoreface depositional environment (Trstenik member and Vrapče formation) (AVANIĆ et al., 2018). During the Sarmatian, reduced salinity occurred, while sediments were represented by shallow-water conglomerates, calcarenites and limestones (the Pećinka member) as well as horizontally laminated pelitic sediments (the Dolje formation) deposited in a deeper marine environment (AVANIĆ et al., 2018;PAVELIĆ & KOVAČIĆ, 2018 and references therein). Studies of the provenance of the Middle Miocene sedimentary rocks of the SW margin of NCB are deficient. The only reliable data about the provenance of these sedimentary rocks are from areas N and NE of Hrvatska Kostajnica investigated by KOVAČIĆ et al. (2011) andGRIZELJ et al. (2017). According to these investigations in the area of Medvednica Mt. and the Slavonian Mts., the sources of material for the Middle Miocene clastic sedimentary rocks were located to the south of the investigated area, in the area of the northern Inner Dinarides, while some of the material was also supplied from locally uplifted blocks in the SW part of the PBS.
The aim of this paper is to reconstruct the Badenian/Sarmatian transition in the area of the SW marginal part of the NCB and PBS based on a multiproxy approach: mineralogical, geochemical, sedimentological and integrated palaeontological analyses (calcareous nannofossil, foraminifers, ostracodes, palynomorphs determination) and palaeoecological reconstruction putting the results in the frame of new chronostratigraphic divisions within the Middle Miocene. Furthermore, the aim of the present study is also to make a contribution to the discussion about the evolution, palaeogeography and climate history of the marine succession in the south-western PBS. realm, a sedimentation area that, during the Miocene lost and reestablished connections with the Mediterranean and the Indo-Pacific Ocean on several occasions (HARZHAUSER & PILER, 2007;KOVÁČ et al., 2017;. Thus, the Early/Late Badenian boundary was marked by a gradual weakening in the connection with the Mediterranean Sea, leading to the interruption of the water exchange between the western and eastern Central Paratethys (KOVÁČ et al., 2017). This event is called the Badenian Salinity Crisis (BSC) which began at 13.81± 0.08 Ma (De LEEUW et al., 2010). Moreover, GRADSTEIN et al. (2012) observed a causal relationship between cooling during the glacial event Mi-3b (13.81 Ma) and evaporite formation. Formation of the PBS back-arc basin began in the Early Miocene with continental collision and subduction of the European Plate beneath the Apulian Plate. In the area of North Croatia during the Early Miocene two basins with different depositional histories evolved: the Hrvatsko Zagorje Basin (HZB), which occupied a small area in north-western Croatia, and the North Croatian Basin (NCB), that covered almost the entire area of north Croatia . The syn-rift phase of basin development was characterized by tectonic thinning of the crust and isostatic subsidence, while the post-rift phase was marked by subsidence caused by cooling of the lithosphere (HORVÁTH & ROYDEN, 1981;ROYDEN et al., 1983;ROYDEN, 1988). In the south-western part of the PBS, the syn-rift phase lasted from the Ottnangian to the Middle Badenian, while the post-rift phase extended from the Late Badenian to the end of the Quaternary .
The oldest deposits of the NCB are Early Miocene fresh-water, alluvial deposits (AVANIĆ et al., 2003;, PAVELIĆ, 2001, PAVELIĆ & KOVAČIĆ 1999. These deposits were covered by Lower Badenian marls and sandstones deposited in a hydrologically open lake (PAVELIĆ et al., 1998 and references therein). The lacustrine environment was replaced by marine environments with a transitional brackish interval (PAVELIĆ et al., 1998;HAJEK-TADESSE et al., 2009). At the beginning of the Middle Badenian, due to the marine connection, lithothamnium limestones, sandstones and marls intercalated with pyroclastic rocks (caused by occasional volcanic activity) were deposited (AVANIĆ et al., 2003, PAVELIĆ et al., 1998. The Badenian/Sarmatian transition is correlated with the top of polarity Chron C5Ar2n at 12.829 Ma (HOHENEGGER et al., 2014 and references therein). In the Sarmatian, the connections of the PBS with marine areas were significantly reduced which caused changes in depositional environments from marine to marine with reduced salinity and stratification. Sarmatian marls and sandstones were deposited mostly conformable on the Badenian deposits (PAVELIĆ, 2001;PAVELIĆ & KOVAČIĆ, 2018 and references therein).

Fieldwork
Fieldwork included the following: outcrop rigging and sampling via Single rope technique, in situ determination of sediment type, observation of the type of contacts between intervals, measurement of the dimensions of the sedimentary bodies, sampling for various analyses and finally construction of a lithostratigraphic column.

Mineralogical and geochemical analysis
Mineralogical and geochemical analysis included the following: X-ray powder diffraction (XRPD) analyses (6 samples), chemical analysis of major and trace elements (7 samples) and measurement of CaCO 3 using Scheiblers calcimeter (15 samples).
Preparation for XRPD analyses included: sieving the samples through the 63 µm sieve, carbonate fraction dissolution by acetic acid with ammonium acetate (1 mol dm -3 ) buffer of pH 5 (JACKSON, 1956) and separation of the < 2 µm fraction from the insoluble residue of the sample. XRPD patterns were recorded on random mounts of bulk samples, fraction < 63 µm and oriented mounts of the < 2 µm fraction using the Philips vertical goniometer (type X`Pert) equipped with a Cu-tube (HGI-CGS), and using the following experimental conditions: 45 kV, 40 mA, PW 3018/00 PIXcel detector, primary beam divergence 1/4° and continuous scan (step 0.02 °2θ/s). Oriented mounts of the < 2 µm fraction were recorded after the following treatments: a) air drying, b) ethylene-glycol solvation, c) heating to 400°C and 550°C. The X-ray interpretation was performed using the HIGH SCORE PLUS (2016) calculation and PDF-4 / MINERALS 4.5 (2020) databases. Quantitative analysis was performed according to SCHULTZ (1964).
Chemical analyses of samples were undertaken by the Bureau Veritas Commodities Canada Ltd., (www.acmelab.com) in Vancouver (Canada). After lithium borate (LiBO 2 ) fusion, the major elements content was determined by inductively coupled plasma emission spectroscopy (ICP-ES), while trace elements were measured on an inductively coupled plasma mass spectrometer (ICP-MS).
Provenance analyses were performed following the results of chemical analysis of major and trace elements using oxide or elemental ratios, ternary diagrams and diagrams based on the ele mental ratios.
Reconstruction of provenance based on the major elements was performed by SiO 2 /Al 2 O 3 , K 2 O/Al 2 O 3 and Na 2 O/K 2 O ratios.
The degree of chemical weathering of the source rocks was constrained by calculating the chemical index of alteration (CIA= [Al 2 O 3 /(Al 2 O 3 +CaO*+Na 2 O+K 2 O)]x100 (NESBITT & YOUNG, 1982). Oxides are expressed in molar proportions and CaO* is the amount of CaO in siliciclastic minerals only, i.e. excluding carbonates and apatite. The data for CaO* have been corrected according to the procedure described by McLENNAN (1993). Because all the analysed samples contained significant amounts of carbonate component, it was assumed that the correction of CaO is equal to that of Na 2 O.

Calcareous nannofossil analyses
The preparation method of SHAMROCK et al. (2015) was followed for TLM (transmitted light microscope) analyses of calcareous nannofossils, while for SEM (scanning electron microscope) analyses, calcareous nannofossils were coated with gold. Slides were examined using an Olympus BH-2 TLM (HGI-CGS) and JEOL JSM-6510 LV SEM (INA -Industrija nafte d.d.). The relative abundance of the calcareous nannoplankton was estimated after randomly counting more than 200 coccoliths along transects at 750x magnification (using a 60 x objective) according to SCHMIDT (1978).

Palynological analyses
Palynological analyses were carried out on four samples collected from the lower 10 m of the column. Standard palynological processing techniques were used to extract the organic matter (e.g. MOORE et al., 1991;WOOD et al., 1996). The samples were treated with sodium pyrophosphate (Na 4 P 2 O 7 ), cold HCl (15%) and HF (40%), to remove carbonates and silica. Heavy liquid (ZnCl 2 , density >2.1 kg/l) was used to separate the organic matter from the undissolved inorganic components. The organic residue was sieved through a 10 µm mesh. For palynofacies analysis slides were mounted in glycerine and for palynomorphs analysis in silicon oil. Microscopic analyses were performed using an Olympus BH-2 and Leica DM2500 microscopes (HGI-CGS). Photomicrographs were taken using an AmScopeTM camera adapter connected to the AmScope v.3.7 camera software and a Leica MC190 HD camera connected to the Leica LAS EZ software. The palynofacies analyses represent qualitative examinations of the organic matter component groups according to the classifications proposed by TYSON (1995). Samples are plotted in the AOM-phytoclast-palynomorph (APP) ternary diagram to characterize the palaeoenvironment according to TYSON (1995).

3.3.3.. Foraminiferal and ostracod analyses
Altogether 12 samples along the column were prepared for ostracod and foraminiferal analyses, both benthic and planktonic species. Approximately 200 g of sediment per sample was disaggregated by soaking in diluted hydrogen peroxide for 48 h, then washed through sieves (0.25; 0.125; 0.09 and 0.063 mm) and dried at room temperature. All fossil remains (ostracods, foraminifera, bryozoans, gastropods, pteropods, bivalves fragments, otoliths, and fish remains) were hand-picked from each dried residue and observed under a binocular microscope (WILD M3Z) and Zeiss stereomicroscope. All picked specimens were classified and counted but not statistically analyzed.
To interpret palaeoenvironmental conditions, calcareous benthic foraminifera are divided into different morphogroups based on their test morphology and mode of coiling. This widely used approach shows the relationship between the test morphology and ecological life preference of modern benthic taxa (SI-LYE, 2015 and references therein,). Ecological conditions required for each foraminiferal species are mainly accepted from MURRAY (2006) and SILYE (2015) and references therein.
Specimens considered stratigraphically important were documented using the SEM JEOL JSM-35CF system at the HGI-CGS.
All micropalaeontological samples, organic residues and paly nological slides are curated at the Department of Geology, HGI-CGS, Zagreb, Croatia.

Sedimentology
The Hrvatska Kostajnica (KOS-I) lithostratigraphic column has a total thickness of 25.6 m, which consists of upper Badenian and Sarmatian pelitic sedimentary rocks (Figs. 1b,2). Field research and measurement of CaCO 3 has shown that the column consists of marls, and clayey limestones, while calcareous marls and silty marls are rare.
Marls contain 48-60% CaCO 3 , silty marls 29-49% CaCO 3 , calcareous marls 77% CaCO 3 and clayey limestones 81-88% CaCO 3 (Fig. 2). There is visible bedding dipping toward the north (N-NW) at a 10° angle, although the internal structure in individual layers is rarely visible. Lithological boundaries between the described intervals are mostly sharp and planar or gradational, except at 1.3 m and 6.6 m from the base of the outcrop, where the boundary is described as sharp and irregular. Horizontal lamination is present in the lower part of the outcrop and rarely in the upper part. At the top of the outcrop (from 22.6 to 25.6 m) there are bioturbation marks. Finally, there is no clearly visible discordance between the lower (Badenian) and upper (Sarmatian) parts of the lithostratigraphical column.

Mineralogy and geochemistry
The mineralogical composition and chemical composition of the marl samples are given in Tab. 1. Table 1 shows the results of the quantitative mineral composition of the bulk marl samples and semi-quantitative results of the fraction < 63 µm and < 2 µm of the insoluble residue obtained by XRPD. The main mineral components are calcite and clay minerals while quartz and aragonite are present in a lesser quantity. In the fraction < 63 µm of insoluble residue, besides clay minerals and quartz, all samples contain a small amount of plagioclase. Zeolites from the clinoptilolite/heulandite series are present in a lesser quantity in all samples except sample 15 ( Fig.  3 and Tab. 1). Sample 1 contains a small amount of pyrite. In the fraction < 2 µm of insoluble residue smectite/I-S and illite/mus- Table 1. Quantitative mineral composition of bulk samples and the semi-quantitative mineral composition of the <63 µm fraction of insoluble rock residue obtained by XRPD according to the procedure described by SCHULTZ (1964). Clay mineral content was determined from the <2 μm fraction of insoluble rock residue. Qtz -quartz, Cal -calcite, Arg -aragonite, Cli/Hul -clinoptilolite/heulandite, Pl -plagioclase, Py -pyrite, Sme/I-S -smectite/illite-smectite, Ill/Ms -illite/muscovite, Vrm -vermiculite, Kln -kaolinite, Chl -chlorite, XXX -dominant (>50%), XX -abundant (20-50%), X -subordinate (1-20%), + -traces (<1%   Rhabdosphaera sicca, Sphenolithus abies, Syracosphaera clath ratae, Umbilicosphaera rotula. Two main biostratigraphic Zones NN6 and NN7 were identified (MARTINI, 1971) (Fig. 2). Species diversity is highest (36 taxa) in Zone NN7 in sample 9, while up to 32 taxa were determined in Zone NN6 in sample 1. The most equally dominant species in the record (up to 24%) are Coc colithus pelagicus and Umbilicosphaera jafari, followed by less abundant Reticulofenestra producta (up to 17%), R. minuta (up to 16%), small Reticulofenestra sp. (up to 14%) and Reticulofen estra haqii (up to 12%). All scattered and rare taxa (≤ 9%) are omitted from the results because they did not affect the palaeoecological interpretations. On the other hand, two rare index species (Calcidiscus pataecus and Discoaster kugleri) contributed to the biostratigraphy.

Palynology
All samples contain palynomorphs in various amounts. The preservation degree of the grains is medium to good, i.e. the structure and sculpture of several grains are partially destroyed, making more precise determination impossible.
Alongside the zonal vegetation (warm-temperate mixed mesophytic forest at a very low ratio), extrazonal mountain coniferrich forests were developed.

Foraminifera
The distribution of foraminifera from the marl samples is given in Supplement Tab. 1 and Fig. 7.
Altogether, 81 benthic and 31 planktonic species of foraminifera are determined in the lithostratigraphic KOS-I column.
Large amounts of reworked Badenian microfossils within the Sarmatian sediments, together with the rare occurrence of Sarmatian index fossils, is the main characteristic of the KOS-I column.
Benthic foraminifera are predominantly represented by calcareous species, miliolids are scarce, while agglutinated species are absent. An engaging feature of the benthic foraminiferal assemblage is the slightly higher occurrence of unicameral foraminifera. Benthic foraminiferal taxa, based on their test morphology and mode of coiling, were clustered into six morphogroups. Morphogroup M1 is composed of forms with a rounded planispiral test (Elphidiella, Nonion, non-keeled elphidiids). They are present throughout the whole KOS-I column. We presume that species from this group belong to the autochthonous assemblage. Morphogroup M2 includes elphidiids (Elphidium crispum, E.   macellum) of the shallow-shelf environment. Morphogroup M3 is composed of the low trochospiral tests of Ammonia. In the KOS-I column, Ammonia species are more frequent in sample 7. Morphogroup M4 is composed of unilocular lagenids. This morphogroup characterizes the outer shelf to bathyal environments, suboxic marine environments with a low-moderate organic influx (SILYE, 2015 and references therein). In the KOS-I column, they only appear in the lower part. Morphogroup M5 is composed of two genera, Bolivina and Bulimina. Morphogroup M6 represents low-trochospiral rotalids, Cibicidoides and Lo batula species. In samples from the KOS-I column morphogroups with an infaunal mode of life prevail, while the epifaunal lifestyle is subordinate.

Ostracods
A total of 10 samples were used for ostracod analyses, taken successively from the KOS-I column. Twenty four ostracod taxa belonging to 15 genera were identified, of which eight species remained in open nomenclature. Distributions of ostracod species are summarised in Supplement Tab. 2, listed alphabetically. Some of the species are illustrated in Fig. 8.
Ostracods were discovered in small numbers in nine samples, and their species richness was low. Ostracods were absent in sample 6. Common species in samples from the KOS-I column are: Aurila sp. (8 records), Callistocythere cf. postvallata (6 records) and Xestoleberis glaberscens (4 records). The diversity varies throughout the column and is generally low. The maximum number of species (11) was reported in sample 11. Sample 4 contain 10 species. Other samples include relatively low numbers of species: samples 2 and 13; 7 species each, and samples 1 and 5 with 6 species each. In other samples, the number of species vari es between 3 and 4, and one sample contains only 2 species.
Pteropods, probably Limacina valvatina, are recognized in samples 4 to 9. Their occurrence is documented from the early to late Badenian and even in the lower Sarmatian (BOŠNJAK et al., 2017 and references therein). According to BOHN-HAVAS & ZORN (2002), the species was widely distributed during the Badenian.

Sedimentology
The lithostratigraphic column in Fig. 2  KOVAČIĆ, 2018). Structureless marls, that are represented in most of the lithostratigraphic profile (Fig. 2), suggest sedimentation of fine-grained siliciclastic material from suspension in an open marine environment. Similar conditions recorded on the NE part of Mt. Medvednica indicate sedimentation in the deeper part of the basin which was not significantly affected by the sea-level fall at the end of the Badenian (VRSALJKO et al., 2006). The analysed samples contain calcite and in a lesser quantity aragonite. Carbonates were predominantly deposited from seawater (discussed in section 4.2.). The small amount of carbonate content in the marls could be associated with carbonate detritus derived from fossil skeletons and bioerosion of older carbonate bedrock. The fragments of bryozoans, gastropods, pteropods, bivalves, otoliths, and fish remains, together with marine ostracods, foraminifera, calcareous nanofossils, and palynomorphs indicate a marine depositional environment. The absence of wave action indicators and the presence of palaeontological species from deeper marine environments suggests that sediments were deposited on the shelf or below the base of the wave action (offshore). In contrast, the Badenian/Sarmatian boundary is discontinuous in many cases within Central Paratethys (VRSALJKO et al., 2006;MANDIC et al., 2019b;MANDIC et al., 2019c). This discontinuity is the result of the sea-level fall during the latest Badenian, and the formation of archipelagoes which were affected by erosion resulting in the deposition of shallow-water sediments and reworked Badenian flora and fauna . Figure 9. shows the stratigraphic position of the KOS-I column correlated with the Ugljevik (MANDIC et al., 2019c) and Donje Orešje columns (VRSALJKO et al., 2006;GALOVIĆ, 2003). The depositional environment of KOS-I column is similar to the environment of the Donje Orešje. Nevertheless, during the Sarmatian deposition occured in the Donje Orešje in a lagoon, while in the studied area it continues in the deeper sea. The Badenian/Sarmatian boundary at the Ugljevik column (Bosnia and Herzegovina) as in many other parts of PBS, is marked by a weakly expressed erosive omission and hiatus (MANDIC et al., 2019c). According to AVANIĆ et al. (2003), these differences were proba bly the result of local tectonics in the earliest post-rift phase that has a significant role controlling sedimentation in these part of the PBS.

Mineralogical composition of marls
The bulk analyses of marls show that the main mineral components are calcite and clay minerals, while quartz and aragonite are present in lesser quantities. The amount of aragonite is probably of biogenic and/or chemical origin. It is well known that the type of carbonate and its morphology depends on the temperature of the water in which it is generated, together with the concentration of Ca 2+ and Mg 2+ ions in the solution, salinity and pressure of CO 2 (RAO, 1996). Calcite may form when the Mg/Ca ratio is < 2, while aragonite occurs when the Mg/Ca ratio is between 2 and 12 (MÜLLER et al., 1972). Moreover, calcite may occur in different environments, while aragonite is usually formed in warm (20-30 °C), shallow marine environments by direct precipitation from seawater, or it forms skeletons of various organisms (LIPPMANN, 1973;RAO, 1996;CHANG et al., 1998, TIŠLJAR, 2001. Such marine and marine environments with reduced salinity existed during both the Badennian and the Sarmatian. Also, part of the aragonite, that is unstable in the subsurface and at standard temperature and pressure, could be altered to the more stable isomorph calcite. In the fraction < 63 µm of insoluble residue, besides clay minerals and quartz, all samples contain a small amount of plagioclase and zeolites from the clinoptilolite/heulandite series. Zeolites are present in almost all samples, indicating that they were probably transported from some local source of origin. Such a local source could be the upper Cretaceous deposits in the locality of Volinja near KOS-I column, in which MARKOVIĆ (2002) mentions the occurrence of zeolite minerals. The results of previous provenance studies suggest also that the neighbouring Inner Dinarides produced a significant amount of clastic detritus during the Middle Miocene (KOVAČIĆ   al., , GRIZELJ et al., 2017. In the < 2 μm fraction of insoluble residue, the main components are smectite/I-S and illite/ muscovite, while kaolinite is present in a lesser amount. In some samples, chlorite and vermiculite are also present. While illite/ muscovite and chlorite are considered typical terrigenous mineral species, formed directly from disintegrating intrusive and metamorphic rocks, kaolinite and vermiculite are characteristic products of chemical weathering (CHAMLEY, 1989). It should also be noted that chlorite is poorly resistant to chemical weathering, which probably explains its presence in small amounts in the analysed marls. According to WEAVER (1989), most smectite occurs as interlayer I-S. Smectites may form from the alteration of volcanic rocks in subaerial and submarine environments and do not suffer from marine and river transportation, and are often carried in the marine environment farther than other minerals because of their high buoyancy (CHAMLEY, 1989). In the investigated profile there were no traces of volcanic activity in the sediment. The occurrence of smectite could be explained by re-deposi tion from volcanic material from older formations such as tuffs that were determined in the area of Banovina (MANDIC et al., 2012 andMARKOVIĆ, 2017) or alteration of volcanic material which was transported over long distances. Intensive volcanic activity was reported for the Middle Miocene in the wider area of the PBS (PAMIĆ et al., 1995;KOVÁČ et al., 2007;MANDIC et al., 2012;MARKOVIĆ, 2017;BRLEK et al., 2020).

Major elements
Variation in the chemical composition of major elements of the samples may be explained by the observed variation in the mine ralogy of the samples. Al 2 O 3 values range between 3.00-12.29 wt %, an amount related to clay minerals, zeolites and plagioclase. The amount of Al 2 O 3 and other major elements is dictated to by the amount of CaO, which is mainly related to the amount of calcite and/or aragonite in the samples. This is illustrated in Fig. 4, which shows good positive correlations of SiO 2 , Fe 2 O 3 , MgO, Na 2 O, K 2 O and TiO 2 with Al 2 O 3 , while CaO is strongly negatively cor- related. Except for clay minerals, zeolites and plagioclase, a certain amount of SiO 2 is also associated with the amount of quartz (Tab. 1).
The SiO 2 /Al 2 O 3 ratio proposed by CULLERS (2000) was used as an indicator of the maturity of the clastic sedimentary rocks, as well as for the presence of quartz in relation to the clay minerals and feldspar. This ratio in all samples ranges from 3-3.5 (Tab. 2), and is lower than the values characteristic for the Upper Continental Crust (TAYLOR & McLENNAN, 1985).
The Al 2 O 3 /TiO 2 ratio in clastic sedimentary rocks is used to distinguish the types of source rocks (GARCIA et al., 1994;AN-DERSSON et al., 2004). This ratio < 14 is indicative of mafic source rocks, whereas a ratio ranging from 19-28 is characteri stic for felsic source rocks. The Al 2 O 3 /TiO 2 ratio in the Badennian--Sarmatian marls of KOS-I column ranges from 17.56 to 20.39 (Tab. 2), suggesting that these sediments are derived from felsic source rocks.
The K 2 O/Al 2 O 3 ratio is used as an indicator of the source composition of pelitic sedimentary rocks and is significantly different for clay minerals and feldspars. The K 2 O/Al 2 O 3 ratio ranging from 0-0.3 is indicative of clay minerals, and for feldspars, it ranges from 0.3-0.9 (COX et al., 1995). A K 2 O/Al 2 O 3 ratio of sediments > 0.5 indicates a significant content of alkali feldspar relative to other minerals in the source rocks, while a K 2 O/Al 2 O 3 ratio < 0.4 suggests recycling of pelitic sedimentary rocks (COX et al., 1995). In the analysed samples, this ratio is less than 0.2 (Tab. 2) suggesting, therefore, they could have been derived from older pelitic sedimentary rocks. From the mineral composition of the analysed marls, it could be presumed that the potassium content in the samples originates from the presence of illite/muscovite. Such a conclusion is supported by the perfect correlation of K 2 O and Al 2 O 3 (Figure 10f) and the mineral composition of the samples (Tab. 1).
The Na 2 O/K 2 O ratio ranges from 0.20-0.33 in the analysed samples. Sodium in marls has been associated mainly with plagioclase, and in lesser amounts it could be related to the exchangeable interlayer cations of clay minerals. Na 2 O has a perfect correlation with Al 2 O 3 in the analysed samples (Fig. 10e).
The degree of chemical weathering of the source rocks can be constrained by calculating the chemical index of alteration (CIA) proposed by NESBITT & YOUNG (1982). According to NESBITT & YOUNG (1982), the CIA values for average shale range from 70-75, and for clay minerals, illite and montmorillonites from 75.00-85.00. For the analysed marls, the CIA values range from 71.81-75.01 (Tab. 2), and are higher than the UCC values (< 50;TAYLOR & McLENNAN, 1985). The intensity of weathering of the source rock was also deduced by the triangular Al 2 O 3 -(CaO+Na 2 O)-K 2 O plot (after NESBITT & YOUNG 1982;1984) with the values of the Upper Continental Crust given by TAYLOR & McLENNAN (1985), and the idealized compositions of plagioclase, K-feldspar, illite/muscovite, smectite and kaolinite/chlorite (NESBITT & YOUNG, 1984) (Fig. 11).  (after NESBITT & YOUNG 1982;1984) in comparison to the data for Upper Continental Crust (UCC) given by TAYLOR & McLENNAN (1985) and the idealized mineral compositions of plagioclase, K-feldspar, kaolinite, chlorite, muscovite, illite and smectite (from NESBITT & YOUNG, 1984). The marl samples were plotted between the idealized compositions of smectite and illite/muscovite, implying that the source rocks underwent an intermediate degree of weathering.

Trace elements
The chemical composition and element ratios of samples critical for understanding provenance, such as La/Co, Th/Co, Th/Sc, La/Sc, Th/Cr, Eu/Eu * , LREE/HREE, ΣREE are given in Tab. 2. Values of the La/Co, Th/Co, Th/Sc, La/Sc, Th/Cr, and Eu/Eu * ratios correspond to the values for acid (Si-rich) rocks, according to CULLERS (1994CULLERS ( , 2000. Figure 12. represents the REE plots of samples normalized to chondrite (TAYLOR & McLENNAN, 1985).
All samples show LREE > HREE and a negative Eu-anoma ly, which is in accordance with the Eu/Eu * ratio as indicators of the size of Eu-anomalies (COX et al., 1995). The sum of REE, depending on the carbonate content, varies in the analysed samples from 36-123 ppm, while the LREE/HREE ratios are more uniform (LREE/HREE = 5.8-7.3 ppm). High LREE/HREE ratio indicate predominantly acid (Si-rich) source rocks. Figure 13.
shows Nb/Y -Zr/TiO 2 diagrams proposed by WINCHESTER & FLOYD (1977) and illustrates that all the analysed samples plot to the rhyolite field, further suggesting acid (Si-rich) source rocks.
According to these diagrams, the Badennian and Sarmatian marls plot close to the average composition of the Upper Continental Crust (TAYLOR & McLENNAN, 1985), and within the field of the Continental Island Arc, representing inter-arc, fore-arc, or back-arc basin, adjacent to a volcanic-arc developed on a thick continental crust or thin continental margins (BHA-TIA & CROOK, 1986). The inferred tectonic position corresponds to the fact that the PBS was one of the Mediterranean basins that existed from the Miocene to the Pliocene (HORVATH & ROYDEN, 1981;ROYDEN, 1988).

Biostratigraphy, palaeobiogeography, palaeoecology
Lithologically and according to microfossil content, three different depositional environments can be distinguished (Fig. 2). 5.4.1. Late Badenian; lower part of the column (0.0 -5.2 m; samples 1 -5) Silty marls dominate, followed by marls with intercalations of sand. The depositional environment varies from the distal dysoxic-anoxic shelf through a mud-dominated oxic shelf without terrigenous input to the proximal suboxic-anoxic shelf. Such conditions, affected by a humid climate, caused the development of a stratified water column and low oxic conditions at the sea bed. Low oxic conditions and a stratified environment were also confirmed by the presence of pyrite inclusions. These conditions are confirmed by the domination of buliminids, bolivinids and uvigerinids and the mass occurrence of Globigerina bulloides and Turborotalita quinqueloba (KOVÁČ et al., 2017 and references therein). Based on the different genera identified, the ostracod fauna from the KOS-I column can be defined as a shallow-water marine fauna. Some species of the genera Loxoconcha, Amino cythere and Xestoleberis tolerate lower water salinity, ranging from meso to oligohaline values. A Late Badenian age is determined based on the calcareous nannofossil assemblages (samples 1 and 5, Fig. 2) that identifies Subzone NN6c (MĂRUNŢEANU, 1999), as well as palynomorphs (samples 1 and 4, Fig. 2), foraminifera (samples 1-5, Fig. 2), and ostracods (sample 4, Supplement Tab. 2).
During the late Badenian, vegetation in the southernmost parts of the Central Paratethys was thermophilous, partly sub-humid under subtropical climatic conditions (KVAČEK et al., 2006;UTESCHER et al., 2007). Nevertheless, coccolith Umbilico sphaera is typical for the warm water, oligotrophic marine environments of subtropical to temperate climate (KINKEL et al., 2000;KRAMMER et al., 2006), while small Reticulofenestra are typical for oligotrophic, warm waters of lower latitudes (HAQ, 1980;RAHMAN & ROTH, 1990). Umbilicosphaera jafari in assemblage with small Reticulofenestra species, noticed in sample 1, implies warmer and saltier shelf waters. This interval ( Fig. 2; samples 1-2) probably belongs to the beginning of the late Badenian period of climate change from subtropical to warm, temperate humid climate recorded from 13.5 Ma according to BÖHME (2003), known as the Middle Miocene Climatic Transition (MMCT). Occurrence of the dinocyst Achomosphaera andalous iensis in sample 4 points to an age younger than 13.3 Ma -Serravallian (LOURENS et al., 2004). Cleistosphaeridium placacan thum and Cerebrocysta poulsenii together indicate an age no younger than the latest Serravallian or earliest Tortonian (JIMENEZ-MORENO et al., 2006). The HO of Cerebrocysta poulsenii is calibrated against basal C5r (~11.7 Ma) in Italy (ZEVENBOOM, 1995). An assemblage from the lower part (sample 1) is very similar to an assemblage from the Cerebrocysta poulsenii Assemblage Biozone of JIMENEZ-MORENO et al. (2006). They stated that the Cpo Zone indicates a late early Serravallian age and lies within nannofossil Zone NN6, as we detected here, and the Bolivina-Bulimina Zone of the regional benthic eco-zonation of GRILL (1941). This zone is also a correlative of the Achomosphaera andalousiensis (Aan) Interval Zone in the Mazzapiedi and Cassinasco sections of northwest Italy of Late Serravalian age (ZEVENBOOM, 1995). However, JIMENEZ-MORENO et al (2006) assigned this zone to the latest Badenian (late early Serravallian) in the Central Paratethys based on other fossils. The absence of agglutinated species with a dominance of Bulimina spp. and Bolivina spp. in the foraminiferal assemblage suggests a late Badenian age as well (PAULISSEN et al., 2011). The oceanic to outer neritic Nematosphaeropsis labyrinthus suggests that relatively open-marine conditions existed, which is in accordance with palynofacies that indicate a dysoxic-anoxic shelf environment (Fig. 6). Among the foraminifera (sample 1-6), morphogroups M4 and M5 dominate. The presence of the aforementioned unicameral foraminifera also indicates a late Badenian age (POPESCU & CRIHAN, 2004). The main characteristic of this assemblages is its infaunal lifestyle in the muddy sediments mostly restricted to a shelf and slope environment (LI & McGOW-RAN, 1994). They prefer high productivity or a cold water environment (LI & McGOWRAN, 1994), which is probably a consequence of a periodic influx of the cold Mediterranean current that affected the climate as well as humid periods under the subtropical climate detected here. The proportions of juveniles/adults and valves/carapaces of ostracods in the sedimentary record provide an estimate of wave energy during deposition (WHATLEY, 1988 andBOOMER et al., 2003). Higher energy conditions lead to the removal of smaller juvenile valves and cause the disarticulation of carapaces. Previous research showed that this relationship is autochthonous and characterizes the environment in which it is found when most ontogenetic stages are present in samples (WHATLEY, 1988;BOOMER & EISENHAUER, 2002). In all samples of KOS-I column valves and a few fragments were found. In three samples (KOS-I 2, 4 and 5) juvenile valves are dominant. The population structure suggested that ostracods from the KOS-I column are allochthonous. Most of them are transported from a high-energy environment, and deposited in a low-energy environment. Resediment ostracods can be readily distinguished by examining adult to juvenile ratios. The ratio of adults and juveniles, as well as their general low frequency, suggests unfavourable conditions for the development of ostracods. Finer grained particles, such as juvenile valves, are more readily transported over greater distances such as transported assemblages have cleary skewed adult to the juvenile ratio (LORD et al., 2012). Besides, a small portion of foraminifera morphogroups M1 and M2 also suggests reworking from the shallower, well-oxygenated environment. 5.4.2. Thin layer of clayey limestone (5.70 -6.05 m; sample 6).
Restricted marine conditions are inferred from the low species richness of the dinocyst assemblages, but some oceanic influence was present as indicated by the rare occurrence of Impagidinium species in sample 6. According to JIMENEZ-MORENO et al. (2006), the Cerebrocysta poulsenii Assemblage Biozone (Cpo) reflects shallowing of the sea, before the emergence and erosion that led to the unconformity between the Cpo and Cpl zones at the Badenian-Sarmatian boundary. A general cooling trend appears as reflected by an increasing role of deciduous elements. Sometimes it overlaps with the lowermost Sarmatian, which is floristically barely distinguishable (KVAČEK et al., 2006). Based on calcareous nannofossils, a cooling event defines the Badenian/ Sarmatian transition as recorded in Subzone NN6c, while max. regression at the base of NN6d, close to the base of the Sarmatian (GALOVIĆ, 2017; GALOVIĆ, 2020), which is in accordance with HOHENEGGER et al. (2014). Sample 6 has not been adequate for calcareous nannofossils, but this interval could apply to Subzone NN6d, and stress developments recognised near the Badenian/Sarmatian boundary. A characteri stic feature of the period is the predominance of Coniferae pointing to extrazonal mountain forests.
The stressed marine environment is confirmed by prasinophytes. Some prasinophytes are "disaster species" because of their ability to survive in a stressed environment. They have a short-lived population increase, and they are most often replaced by opportunists (HARRIES et al., 1996, VAN DE SCHOOT-BRUGGE et al., 2007. The environment was probably stressful, and the assemblage was dominated by disaster species which explains the co-occurrence of prasinophyte phycomata and redeposited dinocysts. Lingulodinium machaerophorum is restricted to coastal regions and regions in the vicinity of continental margins. High relative abundances can be observed in sediments near upwelling cells or below river discharge plumes or in highly stratified waters (ZONNEVELD et al., 2013).
Domination of foraminifera morphogroups M3 and M6 also suggests stressful and unfavourable environmental conditions. Small-sized and fragile tests of Cibicidoides species and prevailing Ammonia species are considered to be pioneer specimens. Based on all the investigated palaeontological data the Badenian/ Sarmatian boundary in the KOS-I column is in line with the proposed position of the boundary (Fig. 2). Marls dominate, followed by silty marls with intercalations of clayey limestone in the lower part to calcitic marls in the upper part.
Following the lowstand in the early Sarmatian, sea level rose in Paratethys at ca. 12.5 Ma (KOVÁČ et al., 2001) which influenced connections with the Mediterranean Sea and Indo-Pacific Ocean (GALOVIĆ, 2017 and references therein). Occurrence of the Sarmatian index species Nonion bogdanowiczi in sample 7 characterised the Sarmatian age. An early Sarmatian age is confirmed in sample 9 based on assemblages of calcareous nannofossils that identify Subzone NN7a (Fig. 2). The appearance of Discoaster kugleri in the assemblage with rare Calcidiscus pa taecus, Helicosphaera walbersdorfensis, Rhabdosphaera spp. and Pontosphaera spp. characterises Subzone NN7a (GALOVIĆ, 2019). The base of Zone NN7 for Paratethys is close to the FO (global) D. hexapleuros at 12.186 Ma, De KAENEL et al. (2017) and possibly sets the sample 9 close to 12.18 Ma for Paratethys (GALOVIĆ, 2020 and references therein) (Fig. 2).
According to JIMENEZ-MORENO et al. (2006), the Cleis tosphaeridium placacanthum Assemblage Biozone (Cpl), contains a low diversity assemblage indicating environmentally unfavourable conditions for most marine dinoflagellates. They correlate this event to the shallow depths and oscillating environmental conditions of the early Sarmatian. A similar assemblage is also found in sample 9 from the KOS-I column. Impagidinium spp. and Nematosphaeropsis labyrinthus disappeared there, indicating marginal marine environments also documented by the common occurrence of the genus Batiacasphaera (GEDL, 2005). High relative abundances of Lingulodinium machaerophorum indicate enhanced nutrient levels and hyposaline conditions with seasonal stratifications (JIMENEZ-MORENO et al., 2006)., while increments of the coccolith R. producta, as in high latitudes (WEI & THIERSTEIN, 1991), indicates periods of weaker upwelling and warmer surface water. Enhanced species diver sity in Zone NN7 implies a deeper and more stable environment.
The increase of Coccolithus pelagicus and small Reticulofe nestra species in the upper part (Fig. 2) may be interpreted as the continues influence of weaker upwelling waters that slightly induced productivity extended from the margin, where a warmer surface current along the outer shelf and upper slope appears during the winter period, similar to the present west coast of Portugal (CACHAO & MOITA, 2000).

CONCLUSIONS
1. The lithostratigraphic Hrvatska Kostajnica column (KOS-I) represents continuous sedimentation from the late Badenian to the early Sarmatian. An interpretation of the depositional history was performed based on the integration of data from the sedimentological fieldwork, mineralogical, geochemical and palaeobiological (calcareous nannofossil, foraminiferas, ostracods, palynomorphs) analyses that suggest marine, offshore sedimentation.
2. The mineral composition of pelitic sedimentary rocks is common for Middle Miocene deposits. However, the appearance of smectite indicates possible local volcanic activity in this period, while the presence of minerals from the clinoptilolite/heulandite series indicates a local source of material. Provenance analyses based on the chemical composition show that the Badenian-Sarmatian marls were predominantly formed by the weathering of acidic (Si-rich) source rocks in the Continental Island Arc area.
3. The investigated column was recorded from the subtropical late Badenian to a warmtemperate climate of early Sarmatian. A high rate of reworked microfossil species, especially foraminife ra, together with small changes in the sedimentological composition and lack of significant tectonic features make it difficult to precisely define the Badenian/Sarmatian boundary. The biostratigraphy based on the fossil assemblages shows that the Badenian/ Sarmatian boundary could be located within the column at 6.05 m from the base.
4. The position of the Hrvatska Kostajnica column (KOS-I) at the edge of the NCB shows different attributes in depositional settings and ecological conditions than the rest of the area studied to date (the Slavonian Mts. and Mt. Medvednica). The main feature of the KOS-I column, concerning fossil assemblage, reworking, can also be explained due to its position mostly detected at the edge of the basin.

ACKNOWLEDGEMENT
This study was co-financed by the project INTERREG -IPA CBC Croatia-Bosnia and Herzegovina-Montenegro 2014 -2020 "safEarth" (Transnational advanced management of land use risk through landslide susceptibility maps design). We are very grateful to Stjepan ĆORIĆ (Geological Survey of Austria) and Anonymous Reviewer for their valuable and helpful reviews.