Sediment petrography, mineralogy and geochemistry of the Miocene Islam Dağ Section (Eastern Azerbaijan): Implications for the evolution of sediment provenance, palaeo‐environment and (post‐)depositional alteration patterns

The reconstruction of regional long‐term patterns recorded in marine sedimentary successions of the Eastern Paratethys is important in understanding the role of Cenozoic climate change and orogenic activity on the depositional environment and sedimentation dynamics in Western Asia. In this study, the environmental conditions in the early to middle Miocene (Islam Dağ section) in eastern Azerbaijan are elucidated using petrographic–mineralogical relations, detrital indicators, weathering indices and δ13C and δ18O signatures of organic‐rich (total organic carbon: ca 3 to 6 wt. %) argillites. Sedimentary facies and chemical proxies (Na/K, K/Al, Si/Al, Ti/Al ratios, chemical index of alteration values) indicate arid conditions, reduced weathering rates in the hinterland and sediment deposition in an euhaline and poorly oxygenated deep‐water basin during the early Miocene, followed by a shift to humid conditions, higher weathering rates and an oxygenated water column in the mid‐early Miocene. Long‐term aridification and deposition of gypsiferous and calcareous argillites under generally more oxygenated, euhaline to polyhaline conditions in a lacustrine or restricted shelf setting until the middle Miocene is evidenced by gradual changes in element ratios and the chemical index of alteration. Discriminant function analysis suggests the Russian Platform, drained by the Palaeo‐Volga and Palaeo‐Don river systems, to be the source area for the siliciclastic input throughout the Miocene, although a minor contribution of volcanogenic detritus and mafic components from the Greater Caucasus is possible. The C–S–Fe associations and increasing Fe/Al ratios towards the middle Miocene support the concept of continuous influx of detrital Fe and total organic carbon. The formation of ferruginous smectite from alteration of volcanic ash layers could have affected the preservation of total organic carbon and therefore the sedimentary C and Fe budget in the Eastern Paratethys basins. Palaeo‐climatic reconstructions based on δ13C (−34·5 to +1·7‰ Vienna Pee Dee Belemnite) and δ18O (−34·7 to −4·8‰ Vienna Pee Dee Belemnite) records of authigenic carbonates should be made with great caution, as the pristine marine signatures may be affected by the oxidation of organic matter and meteoric diagenesis.


INTRODUCTION
In the terminal Eocene and early Oligocene the collision of the African-Arabian-Eurasian plates, a eustatic sea-level fall and the uplift of the Alpine fold belt resulted in the progressive isolation of the Alpine-Carpathian and Euxinic-Caspian basins from the Tethys Ocean, causing variations in biodiversity, hydrological regime and sedimentation dynamics from central Europe to western Asia (e.g. Baldi, 1980;Jones & Simmons, 1996;Feyzullayev et al., 2001;Popov et al., 2004Popov et al., , 2010Sachsenhofer et al., 2015). Subsequent palaeo-oceanographic processes in the Crimea-Caucasian-Kopetdagh deep-water environments of the Eastern Paratethys Sea provided conditions suitable for the establishment of a thermohaline water stratification and estuarine water circulation pattern with recurrent periods of anoxia (Baldi, 1980;Golonka, 2004;Johnson et al., 2010). This is expressed by the deposition of ca 1000 to 1500 m thick, organicrich clayey sequences of Oligo-Miocene age (the so-called 'Maikop' facies), which are source rocks for hydrocarbons of economic significance, for example, in eastern Azerbaijan (e.g. Feyzullayev et al., 2001;Guliyev et al., 2001;Johnson et al., 2010;Bechtel et al., 2013;Sachsenhofer et al., 2015). During the early to middle Miocene, a prominent change from deep-marine settings with fluctuating salinities into shallowmarine and evaporative regimes is traceable in the Euxinic-Caspian basins, resulting in the deposition of marly clays with intercalated limestone beds (Jones & Simmons, 1996;Popov et al., 2004Popov et al., , 2010Abdullayev & Leroy, 2016).
Several studies have been published in the past describing sedimentary environments and facies, palaeo-geography and source rock potential of the Oligo-Miocene rocks from the Euxinic-Caspian basins (e.g. Jones & Simmons, 1996;Popov et al., 2008;Sachsenhofer et al., 2018). However, sedimentological, petrographic, mineralogical and (isotope) geochemical studies focusing on the characterization and reconstruction of the spatiotemporal evolution of depositional environments, palaeoclimate and diagenesis are still scarce. Recently, a chemostratigraphic correlation between different outcrops (Islam Da g, Boyanata, G€ oytepe, Peri-k€ us ßk€ ul, etc.) has been developed (Hudson et al., 2008(Hudson et al., , 2016Johnson et al., 2010) utilizing redoxsensitive trace metals (V, Ni, Mo, U, etc.), biomarkers and chemical parameters (total organic carbon, d 15 N tot , d 13 C org , field c-ray values, C-S-Fe associations, ratios of Fe/Al and Ti/Al, etc.). However, problems still persist with the age and lateral correlation of the Maikopian strata in eastern Azerbaijan, because of major lateral variability in the source rocks, lack of diagnostic microfaunal elements and inconsistencies in age dating of clayey deposits (Bechtel et al., 2014;Sachsenhofer et al., 2018). Until recently, the stratigraphy of the Maikopian strata is still poorly constrained (Washburn et al., 2018).
The Islam Da g section (eastern Azerbaijan) is a key area for investigating the impact of local tectonic activity, basin evolution and regionalscale response of the Eastern Paratethys to longterm Cenozoic global cooling, because this location exhibits an almost complete succession of offshore marine argillaceous rocks of early to middle Miocene age (Washburn et al., 2018). This study presents a novel mineralogical, major/minor elemental and d 13 C and d 18 O dataset of the argillaceous rocks from this famous study site, and discusses implications for the provenance of sediments, evolution of climate, salinity, redox conditions and post-depositional processes. The obtained results are of great significance for reconstructing climate change in eastern Azerbaijan on a regional scale and for correlating the marine, organic-rich succession with other areas (for example, the Vienna Basin, Black Sea and Mediterranean Sea) of the Paratethys in the Miocene.

GEOLOGICAL SETTING AND LITHOSTRATIGRAPHY
The sedimentary rocks from the Islam Da g section are exposed in several outcrops alongside the Islam Da g ('da g' means mountains) range, which is located about 50 km north-west of Baku (Azerbaijan) within the Gobustan-Absheron depression of the South Caspian Basin Province (Fig. 1). This location exhibits a sedimentary succession of argillaceous rocks of late Oligocene to middle Miocene age (Fig. 2). The lowermost part of this section, comprising late Oligocene argillites of greyish colour (Johnson et al., 2010), is currently not exposed due to overburden by local colluvium. The bottom part of the profile comprises massive (metre-scale) and structureless argillites of greyish-black colour that are intercalated with thin (few centimetres) siltstone lenses. These organic-rich sequences have been assigned to the Oligo-Miocene Maikop facies and were interpreted as deposits of a 'deep-water' depression under dysoxic to anoxic conditions (Popov et al., 2004(Popov et al., , 2008. This unit is overlain by greyish-black to yellowish-green (when oxidized) argillites, which are thought to have been deposited in a restricted shelf setting during the Miocene (R€ ogl, 1999). Due to sea-level fluctuations and uplifting of the Russian Platform, the Ural and the Kazakhstan Highs in the north and the Lesser Caucasus, Elburz and Kopetdagh mountain system in the south, a strong differentiation of the depositional environment of the South Caspian Basin occurred in the early to middle Miocene (Krhovsky et al., 1993;Jones & Simmons, 1996). This is expressed by occurrences of chocolate brown argillites and brownish carbonate argillites that are intercalated with sandstone and limestone beds of yellowish-brown colour, deposited in a lacustrine or restricted shelf setting (Popov et al., 2008).
Studies of Maikopian strata in eastern Azerbaijan did not yield a consistent timing of deposition, because index fossils are absent and other faunal elements are either scarce or poorly preserved in most beds (Hudson et al., 2008;Bechtel et al., 2014). However, based on biostratigraphic indications and chemostratigraphic correlations between different outcrops (Boyanata, G€ oytepe, Perik€ us ßk€ ul, etc.), Hudson et al. (2008) and Popov et al. (2008) have assigned the lowermost part of the Islam Da g section to the late Oligocene, while the middle and upper parts have been dated to the early Miocene. Very recently, Washburn et al. (2018) have revisited the stratigraphy of the Islam Da g section and have re-assigned the recently exposed middle part to the early Miocene and the upper part to the middle Miocene, respectively, based on Re-Os geochronology.
Little is known about the burial history of this offshore sedimentary succession, but it is likely that rapid subsidence and burial have occurred since the late Miocene (Devlin et al., 1999), followed by fast uplifting of the Cenozoic strata during the development of the Greater Caucasus (Forte et al., 2015). From Rock-Eval pyrolysis data (i.e. <435°C T max ; a maturity indicator recording the temperature at which the maximum S2 peak is achieved during pyrolysis) it is evident that the Miocene rocks are thermally immature and thus a maximum burial temperature slightly below or close to the early oil window (ca <80°C) can be inferred (Hudson et al., 2016).

Field work, samples and sample preparation
Geological mapping of the Islam Da g section was carried out during several field campaigns from April to May 2018, where the outcrops were logged and studied bed by bed at a resolution of ca 5 cm. The mostly fine laminated argillites were categorized using the classification of F€ uchtbauer (1988). The main types of limestone microfacies were classified using the scheme of Tucker & Wright (1990). Hand-size specimens (37 in total) were collected from the Islam Da g section (Fig. 2) for petrographic, mineralogical and chemical characterization. A 20 to 30 cm thick surface layer was removed prior to sampling to avoid oxidized samples. Moreover, the collection of samples containing siltstone lenses, limestone intercalations and gypsum nodules has been avoided to ensure comparison between sedimentary facies and geochemical proxies recorded in the argillites. For correlation of the sedimentary units the stratigraphy of Washburn et al. (2018) has been applied.
For the separation of the clay mineral fraction, six bulk samples (taken at 5Á6 m, 24Á5 m, 47Á3 m, 56Á3 m, 61Á6 m and 81Á4 m in the Islam Da g section) were treated with 10% acetic acid for 1 h to dissolve carbonates, if present. Then, the acid-insoluble fraction was dispersed in an ultrasonic bath for 10 min. Subsequent to Atterberg sedimentation the <2 lm size fraction was collected, following separation by 0Á45 lm cellulose acetate membrane (Sartorius) filters using a suction filtration unit and drying at 40°C.

X-ray diffraction
X-ray diffraction (XRD) patterns were collected for mineral identification and quantification on a PANalytical X'Pert PRO diffractometer operated at 40 kV and 40 mA (Co-Ka radiation) and equipped with a high-speed Scientific X'Celerator detector (Malvern PANalytical, Malvern, UK), 0Á5°antiscattering and divergence slits, primary and secondary soller and spinner stage. Bulk samples were mixed with 20 wt. % zincite (ZnO) as an internal standard and ground in a McCrone micronizing mill (The McCrone Group, Westmount, IL, USA) for 8 min. Random preparations were then made using the front loading technique and examined from 4 to 85°2h with a step size of 0Á008°2h and a count time of 40 secÁstep À1 . Mineral quantification was carried out by Rietveld analysis of the XRD patterns using the PANalytical X'Pert HighScore Plus Software and a pdf-4 database. The accuracy of the XRD results was verified by comparison with mass-balance calculations based on chemical compositions of bulk samples (see section on X-ray fluorescence). Assuming idealized compositions for quartz, feldspar (K-feldspar and albite/plagioclase), illite, kaolinite, clinochlore (Mg-chlorite), nontronite (Fe-smectite), zeolite (K-clinoptilolite), gypsum, jarosite, pyrite, calcite, anatase and the amorphous phase (organic matter), the deviation of XRD and chemical data was ≤3 wt. %. For further XRD analysis of the clay mineral (<2 lm) fraction, oriented preparations were made using the procedure described in Baldermann et al. (2014). Briefly, ca 50 mg of sample was mixed with 5 ml of ultrapure water (Millipore Integral 3: 18Á2 MOÁcm À1 ; MilliporeSigma, Burlington, MA, USA) and dispersed ultrasonically for 10 min, following suction of the clay in suspension through a ceramic plate of ca 4 cm 2 . The clay films were analyzed from 3 to 30°2h with a step size of 0Á02°2h and a count time of 2 secÁstep À1 each at air-dried states, after solvation with ethylene glycol (EG) and after heating to 550°C for 1 h. The XRD patterns were recorded on a Philips PW 1830 diffractometer (Cu-Ka: 40 kV, 30 mA; Philips, Amsterdam, The Netherlands) and outfitted with automatic slits, a graphite monochromator and a scintillation counter. The illite crystallinity and the Esquevin index were derived from these XRD patterns in order to track changes in the evolution of the detrital flux within the studied interval (Bout-  Washburn et al. (2018). The width of the lithological columns indicates the structural integrity (i.e. weathering resistance) of the beds. Thin laminae (ca 1 cm) of volcanic ash layers are exposed at 8Á7 m, 13Á3 m, 21Á7 m and 47Á1 m in the profile (marked by purple circles). Persons for scale are 1Á8 m tall. Photographs of outcropping Miocene rocks (A) and close-ups of argillites from units 1 to 3 are shown (B) to (D). Greyish-black argillites exposed in the lowermost part of the profile (B) comprise siderite concretions and show no signs of oxidation. Light grey argillites (C) contain patches of yellowish jarosite visible on the outcrop. Chocolate brown argillites (D) are gypsiferous and frequently show polygonal desiccation cracks. Carbonate argillites from unit 4 are exposed above ca 108 m in the profile and are intercalated with massive limestone beds. Hammer for scale is ca 35 cm long. Roumazeilles et al., 2013). The illite crystallinity (in°2h) refers to the full width at half maximum (FWHM) measured on the d 001 -reflection (ca 10 A) of illite at EG-solvated states (Chamley, 1989). This parameter is a measure for the metamorphic degree and temperature in pelitic rocks, i.e. a low illite crystallinity reflects a high metamorphic facies. The Esquevin index (which is referred to as illite chemistry index) is defined by the intensity ratio of the illite peaks measured at 5 A and at 10 A (Esquevin, 1969). This ratio and can be used as a chemical weathering index.

X-ray fluorescence
The chemical composition of bulk rock samples was analyzed using a PANalytical PW2404 wavelength dispersive x-ray fluorescence (XRF) spectrometer (Malvern PANalytical). Powdered samples (ca 2Á0 g) were heated to 950°C for 1 h to remove volatiles (for example, H 2 O and CO 2 ) and then the loss on ignition (LOI) was determined by gravimetric analysis. Subsequently, 1Á0 g of material (dried at 105°C) was fused at ca 1200°C in a fully-automatic PANalytical Perl'X 3 bead preparation system using 6Á0 g of Li 2 B 4 O 7 as fluent agent (Malvern PANalytical). The fusion time was reduced to <3 min in order to avoid volatilization and loss of S. The tablets were analyzed together with a range of United States Geological Survey (USGS) standards. The analytical error is AE0Á5 wt. % for the major elements.
For the determination of distinctive provenance signals discriminant function analysis of major elements data was carried out with the SPSS package, following the manual of Roser & Korsch (1988). The MnO and P 2 O 5 were not considered in the analysis, because their concentrations are low and the analytical precision worse than for the major elements. For palaeo-climate reconstructions the chemical index of alteration (CIA) can be used (Nesbitt & Young, 1982), which expresses the molar volumes of [Al 2 O 3 / (Al 2 O 3 + Na 2 O + K 2 O + CaO*)]Á100. This weathering proxy is based on the progressive transformation of unstable minerals, like alkali feldspar, to more stable clay minerals under ambient environmental conditions, and can be quantitatively traced by following changes in the ratio of immobile Al 2 O 3 to the mobile cations (K + , Na + and Ca 2+ ), expressed as oxides (Richoz et al., 2017). Calcium oxide present as carbonate and/ or gypsum was subtracted from the total CaO content (on the basis of XRD data) to obtain CaO* of the silicate fraction. The CaO present as phosphate was not considered in the calculation, because it increases the CIA value by less than two units if all P 2 O 5 is assigned to apatite. Detrital indicator ratios (K/Al, Ti/Al, etc.), C-S-Fe distributions and Fe/Al ratios were calculated to constrain the detrital input flux, to track changes in the intensity and pathways of weathering in the source rock areas and to trace early diagenetic processes (e.g. Johnson et al., 2010).

Total organic carbon
The total organic carbon (TOC) content was determined on powdered samples (37 in total) following catalytic combustion and analyses of the originated CO 2 by non-dispersive infrared detection using a Shimadzu TOC-VcPH+ASI-V analyzer (Shimadzu Corporation, Kyoto, Japan). The analytical error of TOC analyses is AE5%, as determined by replicate measurements of sodium salicylate (C 7 H 5 NaO 3 ), sodium carbonate (Na 2 CO 3 ) and D-(+)-glucose (C 6 H 12 O 6 ) standards (all p.a. quality, from Roth; Carl Roth GmbH + Co. KG, Karlsruhe, Germany).

Scanning electron microscopy
The mineralogical composition, the particle form and the particle shape of authigenic and detrital (clay) mineral phases were studied by scanning electron microscopy (SEM) using a JEOL JSM-6610LV microscope (JEOL Limited, Tokyo, Japan) operated at an accelerating voltage of 20 kV (Institute of Geology and Geophysics, Azerbaijan National Academy of Science). The microscope is equipped with a low vacuum secondary electron (SE) detector and an Oxford Instruments silicon drift detector for energy-dispersive X-ray spectrometry (EDS) analysis (Oxford Instruments, Abingdon, UK). Rock chips were therefore prepared on standard SEM stubs and sputter coated with carbon to reduce charging.

Stable isotopes
The oxygen and carbon isotopic composition of carbonates was measured using a ThermoFisher Scientific Gasbench II connected to a ThermoFinnigan DELTA plus XP mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) at the stable isotope laboratory of the JR-AquaConSol GmbH (Graz, Austria). Samples were prepared following the procedure described in Dietzel et al. (2009). Briefly, concentrated phosphoric acid was injected into the individual sample vials containing 300 to 600 lg of sample and reacted with the carbonates for 2 h at 70°C prior to the continuous-flow isotopic ratio mass spectrometric analyses. The measured oxygen and carbon stable isotope values are reported in terms of delta notation (for example, d 18 O and d 13 C) relative to the Vienna Pee Dee Belemnite (VPDB) standard. The analytical precision is AE0Á1& for d 13 C and AE0Á08& for d 18 O, respectively.

Sediment petrography and bulk mineralogy of the Islam Da g section
The combination of field work and sedimentological-mineralogical analyses of the bulk samples is given in Table S1. This provides a ca 145 m thick lithostratigraphic profile of the Islam Da g section (Fig. 2), which is described below.
Unit 1: Greyish-black argillites Argillites of greyish-black colour occur at the base of the Islam Da g section (ca 0 to 31Á5 m in the profile; Fig. 2A and B). These rocks are intercalated with reddish-brown siltstone lenses (1 to 20 cm thick) and yellowish-brown siderite concretions (10 to 20 cm across; very seldom). Very thin (ca 1 cm) and whitish to light grey laminae of volcanic ash crop out at ca 8Á7 m, ca 13Á3 m and ca 21Á7 m in this unit. The argillites consist mainly of subrounded quartz (18 to 24 wt. %), partly dissolved feldspar (13 to 22 wt. %) and platy illite (35 to 39 wt. %), in addition to minor amounts of kaolinite (3 to 8 wt. %), veil-like smectite (3 to 9 wt. %), organic matter (expressed as amorphous phase content; 8 to 12 wt. %), zeolite (0 to 4 wt. %) and framboidal pyrite (0 to 2 wt. %) (Figs 3 and 4A). Jarosite, chlorite and anatase occur only in low amounts (<2 wt. %). The amount of carbonates in these rocks always stays below the detection limit of XRD analyses (ca <0Á5 wt. %), corroborating the classification of these argillites as 'carbonate-free', according to Popov et al. (2008). The current microscopic study, however, reveals trace amounts of rhombohedral low-Mg calcite spar (LMC), about 5 lm in largest dimension, infilling former pore space of these rocks, which is typical for neomorphic calcite spar. With the naked eye, the argillites do not show distinctive signs of oxidation. These rocks have been deposited in the early Miocene, applying the stratigraphy of Washburn et al. (2018).
Unit 2: Light grey argillites Massive and 'carbonate-free' argillites of light grey to greyish-black colour are exposed between ca 31Á5 m and 73Á5 m in the profile ( Fig. 2A and C). Macroscopically oxidized samples from this unit are characterized by large occurrences of patches of rhombohedral jarosite crystals of yellowish-green colour on the sample surfaces (Fig. 4B). Unaltered samples consist dominantly of quartz (16 to 35 wt. %), feldspar (12 to 22 wt. %), illite (31 to 40 wt. %) and organic matter (8 to 13 wt. %). Minor amounts belong to smectite (3 to 9 wt. %), kaolinite (1 to 8 wt. %), chlorite, zeolite and jarosite (each 0 to 3 wt. %) as well as anatase and pyrite (<1 wt. %) (Fig. 3). Rhombohedral to rounded LMC spar, ca 2 to 5 lm in size, is barely present in these rocks. In contrast to the argillites from unit 1, these rocks do not contain siderite concretions. A single volcanic ash layer crops out at ca 47Á1 m in the profile. The transition between the early Miocene and the middle Miocene is located near the top of this unit (Washburn et al., 2018), although the precise stratigraphic boundary could not be resolved.
Unit 3: Chocolate brown argillites with mudstone beds Massive argillites of light chocolate brown to greyish-brown colour are exposed between ca 73Á5 m and 108 m in the profile ( Fig. 2A and D). The argillites are traversed by thin lamellar beds (ca 1 to 2 cm) and nodules of gypsum. Polygonal desiccation cracks, dissolution breccia and tepee structures can be found in this interval. The argillites consist of quartz (17 to 22 wt. %), feldspar (17 to 21 wt. %), illite (34 to 39 wt. %) and organic matter (7 to 12 wt. %). Kaolinite, smectite (each 4 to 6 wt. %), platelet-prismatic gypsum (0 to 5 wt. %), rhombohedral to rounded LMC (0 to 4 wt. %), zeolite and anatase (1 to 3 wt. % in total) represent minor constituents (Figs 3 and 4C). In the upper part of this unit marly mudstone beds, about 0Á1 to 0Á5 m thick, of yellowish-beige colour crop out. These rocks comprise high amounts of LMC (>90 wt. %), besides quartz and clay minerals, and show a sharp erosional contact to the surrounding 'low-carbonate' argillites. However, these rocks still contain trace amounts of rhombohedral LMC spar, ca 5 lm in size. Sandstone beds were not found in this unit, contrasting prior results of Popov et al. (2008). This section has been deposited in the early middle Miocene (Washburn et al., 2018).

Clay mineralogy of the Islam Da g section
Four types of clay minerals were identified based on the XRD study of the <2 lm size fraction (Fig. 5) and bulk fraction, namely illite, chlorite, kaolinite and ferruginous smectite.
Illite displays sharp reflections at 9Á94 A (d 001 ), 4Á99 A (d 002 ) and 3Á34 A (d 003 ) and contains ca 98% illite layers and ca 2% smectite layers (Fig. 5), as calculated from the relationship between the position of d 002 peak after EGsolvation and the percentage of illite layers in mixed-layered illite-smectite (Baldermann et al., 2013. Polytype diagnostic reflections reveal a 2M 1 polytype, which is common for detrital illite in marine sediments. Measurements of the illite crystallinity and of the Esquevin index yielded values between 0Á4 to 0Á5°2h and from 0Á3 to 0Á4, respectively, independent of the stratigraphic position of the samples. In the bulk samples, chlorite (i.e. clinochlore) was identified based on sharp reflections at 14Á23 A (d 001 ), 7Á08 A (d 002 ), 4Á74 A (d 003 ) and 3Á52 A (d 004 ). The absence of chlorite in the <2 lm size fraction (Fig. 5) and the relatively low burial temperature (for example, <80°C) point to a detrital nature of this mineral phase.
Poorly crystallized kaolinite shows broad reflections at 7Á15 A (d 001 ) and 3Á57 A (d 002 ) (Fig. 5), which disappeared after heat treatment due to dehydroxylation. The SEM results reveal neither an association of kaolinite with altered feldspar nor a vermiform structure of kaolinite aggregates (Fig. 4), implying a detrital origin of this phase. However, kaolinization of alkali feldspar cannot be fully ruled out, meaning that a fraction of kaolinite could be diagenetic in origin.
Smectite displays a broad d 001 -reflection at ca 12Á6 A at air dried states, which indicates that Na + ions mainly occupy the interlayer space. After EG-solvation and subsequent to heating the basal spacing changed to ca 17Á1 A and ca 10Á0 A, respectively, which is typical for smectite (Fig. 5). The diagnostic d 060 -reflection was at

1Á51
A, reflecting the dioctahedral nature of this mineral phase and the presence of sufficient amounts of structural Fe(III) Voigt et al., 2018). The flaky to veil-like particle form ( Fig. 4A and C) of ferruginous smectite and its association with zeolite minerals and volcanic ash layers imply a diagenetic origin.
Changes in the clay mineral assemblages across the entire Islam Da g section are shown by the illite+chloritekaolinitesmectite ternary diagrams (Fig. 6). Illite is the dominant component (64 to 87%) of the clay mineral fraction with an average of 78%. Smectite (7 to 20%) and kaolinite (3 to 15%) are present in lesser abundance with a similar average content of 10 to 11% throughout in the profile. Chlorite (0 to 7%) is scarce in units 1 and 2 with an average content of 1%, and it disappears in units 3 and 4. Apart from the observed relatively homogenous distribution of the clay mineral assemblage across the Islam Da g section (Fig. 6), some distinctive trends can be recognized when considering the individual proportion of each clay mineral among the total clay mineral fraction over the stratigraphic profile (Table S2). Smectite and kaolinite generally show a similar trend, which is opposed to that of illite, i.e. they progressively increase in unit 1 reaching the highest values in unit 2, subsequently show a sudden drop at the base of unit 3 and then stay comparably low in unit 4.

Major and minor element composition of the Islam Da g section
The major and minor elemental composition of the argillaceous rocks from the Islam Da g section is highly variable (Table S3) due to changes in lithology and grain-size variations. Accordingly, SiO 2 (39Á2 to 62Á4 wt. %), Al 2 O 3 (11Á8 to 23Á4 wt. %), Na 2 O (0Á6 to 2Á2 wt. %), K 2 O (1Á7 to 2Á9 wt. %) and CaO (0Á1 to 14Á3 wt. %) contents are directly attributable to varying abundances of quartz, feldspar, illite, calcite and minor kaolinite and zeolite, complying with the mineralogical results (Fig. 3). The MgO (1Á4 to 3Á8 wt. %), Fe 2 O 3 (2Á4 to 9Á8 wt. %), SO 3 (0Á01 to 2Á4 wt. %) and TiO 2 contents (0Á4 to 0Á9 wt. %) mostly reflect variations in minor amounts of smectite, chlorite, pyrite, jarosite, gypsum and anatase, while the LOI (11Á1 to 23Á0 wt. %) contents belong mainly to organic matter, phyllosilicates and carbonates. The MnO, P 2 O 5 , ZrO 2 and SrO (0Á01 to 0Á2 wt. %) occur in low amounts and reflect trace mineral impurities. The concentrations of redox-sensitive trace metals (for example, Cr, Co, V, Ni, Mo, Th and U) were always below the detection limit of the XRF analyses (ca 0Á01 wt. %) and hence cannot be used for the interpretation of palaeo-redox conditions.
The CIA values (left part of Fig. 7A) range between 75Á5 and 89Á1 (CIA; 81Á9 AE 4Á2, on average) and plot well around the fields of illite and smectite (75 to 90), which reflects intense weathering rates, according to McLennan et al. (1993), and/or recycling of clayey sediments in the basin. These high CIA values correspond to high proportions of clay minerals (especially illite) in the argillaceous rocks and suggest that the fine-grained material is derived from chemical weathering of endogenous or metamorphosed host rocks, considering that fresh basalt, granite/granodiorite and feldspar have CIA values of 30 to 45, 45 to 55 and 50, respectively (Nesbitt & Young, 1984). Plotting the XRF data in the A-CN-K (Al 2 O 3 -CaO* + Na 2 O -K 2 O) diagram (Fig. 7A) reveals that most samples plot slightly above the compositional ranges of the Post-Archean Australian Shale (PAAS) and Average Proterozoic Shale (APS) or follow predicted weathering trends for Upper Continental Crust (UCC) rocks. Given that kaolinization of alkali feldspar was not observed, it can be inferred that the argillites were affected by sorting, as seen in the slight shift towards the A pole of the diagram. Variations in the tectonic setting and/or burial diagenetic illitization of smectite have played only a minor role. Accordingly, the plot of the chemical data in the A-CNK-FM (Al 2 O 3 -CaO* + Na 2 O + K 2 O -Fe 2 O 3 + MgO) diagram (Fig. 7B) shows a mixed composition, which is typical for heterogeneous, clayey sediments that are subjected to a certain degree of intra-basin recycling (McLennan et al., 1993).

Total organic carbon and amorphous phase contents
The TOC content varied from 1Á8 to 6Á4 wt. %. No systematic trend in TOC was recognized in the profile, although two different types of samples can be distinguished: (i) samples with a low TOC content (1Á8 to 2Á3 wt. %; n = 5); and (ii) samples with a moderate to high TOC content (3Á2 to 6Á4 wt. %; n = 32). The low TOC values coincide with samples having low pyrite contents (<1 wt. %), but high jarosite contents (up to 3 wt. %), which suggests that these samples could have suffered from post-depositional oxidation. The high amorphous phase content of these oxidized samples may be due to 'contamination' by glassy components from adjacent volcanic ash layers.
The TOC values obtained for the non-oxidized samples show a positive linear correlation with the amorphous phase content determined by XRD (R² = 0Á69, n = 32). The slope of 2Á03 of the best-fitting function is close to the empirical (1Á9) and theoretical (2Á0) factors frequently used for converting soil organic carbon to soil organic matter (Pribyl, 2010). This relationship implies that other amorphous phases (volcanic ash, silicifying organisms, etc.) are negligible and that the amorphous phase content is about equivalent to the organic matter content.

Oxygen and carbon isotopic composition of carbonates from the Islam Da g section
The rhombohedral LMC spar from the lower part of the Islam Da g section (units 1 and 2) displays a high variation in the d 13 C and d 18 O values in the range from À34Á5 to À17Á3& VPDB for d 13 C and from À34Á7 to À12Á5& VPDB for d 18 O ( Fig. 8; Table S3). A slightly lower scatter in the values of d 13 C (À17Á3 to +1Á7& VPDB) and d 18 O (À17Á6 to À4Á8& VPDB) is evident for the neomorphic LMC and HMC spar, and their admixtures, exposed in the upper part of the profile (units 3 and 4; Fig. 8), although petrographically these authigenic carbonates do not differ much from the lower units.

Post-depositional alteration
The argillites from the Islam Da g section show unequivocal evidence for post-depositional alteration, such as oxidation of pyrite and formation of jarosite as well as alteration of volcanic ash and precipitation of ferruginous smectite and zeolite. An influence of meteoric diagenesis is indicated by the occurrences of neomorphic LMC (and HMC) spar throughout in the profile. Some samples showing low TOC contents have suffered from post-depositional oxidation. These samples were not considered in the further discussion. In contrast, neither signs of pedogenesis, such as formation of calcrete, caliche and soil clays, nor smectite-illitization were found, which suggests that long periods of subaerial exposure and burial diagenetic processes did not modify the original petrographic, mineralogical and chemical signatures of the argillites (Mackenzie, 2005;Richoz et al., 2017;Hellwig et al., 2018). Cementation and compaction by late diagenesis did not play a role taking the low burial temperatures (ca 50 to 80°C) these offshore Miocene deposits have seen (Hudson et al., 2016). The latter features are an important pre-requisite for the interpretation of depositional environments and for the reconstruction of palaeo-climate based on chemical proxies and weathering indices.
Early diagenetic processes and sub-recent oxidation Microbial reduction of labile Fe-oxyhydroxides coupled with microbial heterotrophic and abiotic sulphate reduction soon after sediment deposition (Raiswell & Canfield, 2012) could have provided a secondary pool of dissolved Fe 2+ and S À ions that are required for the formation of pyrite (Figs 3 and 4A). Under such reducing conditions, degradation of organic matter is less efficient compared to oxidizing conditions, explaining the high TOC contents of the argillites from unit 1 (4Á9 AE 0Á6 wt. %). This coincides with the observed low degree of alteration of sediments from unit 1, judging from visual inspection of the outcrop. In contrast, the argillites from unit 2 show distinct signs of oxidation, such as occurrences of patches of jarosite on the rock's surface (Figs 2 and 3) and partly reduced TOC contents (ca ≤2 wt. % in Table S3). It is possible that jarosite has been formed by the microbial mediated oxidation of pyrite under low pH conditions (for example, pH < 4) and at ambient temperature (e.g. Parafiniuk et al., 2016;Lewis et al., 2018). The irregular to patchy appearance of the rhombohedral jarosite crystals may support this hypothesis (Figs 3 and 4B). The in situ conversion of pyrite into jarosite during marine diagenesis is unlikely, given the high solubility of Na/K-or hydronium-jarosite in aqueous environments (K sp = 10 À14Á8 to 10 À7Á1 at 25°C). The absence of oxidation products of pyrite, such as goethite and hematite, and the overall high TOC contents (4Á9 AE 0Á8 wt. %) of argillites from unit 2 support this assertion (Baron & Palmer, 1996;Kotler et al., 2008). For the same reasons, the formation of gypsum through alteration of pyrite seems implausible. Indeed, desiccation cracks, dissolution breccia and tepee structures associated with the gypsiferous argillites from unit 3 suggest that gypsum is primary. Considering the sedimentary structures mentioned before and occurrences of subhedral to rhombohedral HMC in the subsequently deposited carbonate-bearing argillites from unit 4, it seems likely that the gypsum nodules and the neomorphic HMC spar have been precipitated from seawater-derived pore solutions under evaporitic conditions (Mackenzie, 2005;Swart, 2015;Mavromatis et al., 2017). The high TOC contents of units 2 and 4 (5Á1 AE 1Á2 wt. % and 4Á6 AE 1Á1 wt. %, respectively) indicate that post-depositional alteration due to oxidation is negligible.
Occurrences of ferruginous smectite and zeolite minerals throughout the Islam Da g section (Fig. 3) are likely related to alteration of volcanic ash layers, although a minor contribution from detrital sources cannot be fully ruled out. Authigenic smectite can be differentiated from the detrital clay minerals by the deviant particle morphology and the smaller grain size. Ferruginous smectite particles have a flaky to veil-like form and are <1 lm in largest dimension ( Fig. 4A and C). Detrital clay minerals are generally larger in size (ca >1 to 5 lm) and have a platelet, platy or rounded particle form (Fig. 4B). The close association of volcanic ash layers, ferruginous smectite having Na + ions in the interlayer sites and K/Na-clinoptilolite, especially in units 1 and 2 (Figs 3 and 5), suggests their formation during early marine diagenesis at low temperature (Lawrence et al., 1979;Abdullayev & Leroy, 2016).

Meteoric diagenesis and carbonate formation
Two types of authigenic carbonate minerals have been identified in the Islam Da g section: LMC (ca 4 mol % MgCO 3 ) and HMC (ca 9 mol % MgCO 3 ). The LMC occurs in trace amounts (ca <0Á5 wt. %) between 0 m and 87 m in the profile, becoming more enriched towards the top of the section, while HMC only occurs in the uppermost samples of unit 4 (Fig. 3). The characteristic d 13 C and d 18 O values (Fig. 8) and the spatial distribution of LMC and HMC in the profile suggest a different origin of these carbonates.
Low magnesium calcite is strongly depleted in 13 C and in 18 O with respect to common marine carbonates (Veizer, 1983). Such extreme 12 C enrichments are frequently related to oxidation of methane, liberated from decaying organic matter, followed by the conversion of methane into dissolved carbonate species (DIC) and subsequent precipitation of isotopically light carbonates. Such negative d 13 C values have been reported for the methane-derived carbonates (À46Á8 to À41Á5& VPDB) from the native sulphur deposit at Mach ow (south-east Poland) and for the cold seep carbonate deposits (À40Á2 to À0Á1& VPDB) from Marmorito (Italy) (B€ ottcher & Parafiniuk, 1997;Peckmann et al., 1999). In this line, Johnson et al. (2010) have shown the bulk organic matter from the Islam Da g section to be isotopically light (À18 to À30& VPDB for d 13 C org ), suggesting that its oxidation could have provided the source of carbon for LMC precipitation. However, the d 18 O isotope signatures of such methane-derived carbonates are completely different (À7 to +5& VPDB), compared to LMC from the Islam Da g section (up to À35& VPDB), implying more complex formation processes. Considering the d 18 O isotopic composition of the present-day Kura and Samur rivers and of local groundwater from Baku (ca À10Á2 to À9Á6& SMOW -standard mean ocean water) (Lavrushin et al., 2015;Nadiri et al., 2018), it is clear that LMC should have been precipitated under the influence of meteoric waters depleted in 18 O (Beck et al., 2005;Dietzel et al., 2009;Peryt et al., 2012;Boch et al., 2018). It is proposed that recent sub-surface precipitation of jarosite may have acted as a mineral sink for 18 O, i.e. isotopic equilibrium between the meteoric solution and dissolved SO 4 2À (from weathering of pyrite) could have been approached, producing meteoric solutions enriched in 16 O. The interaction of fluids heavily depleted in 18 O with isotopically light methane at ambient temperature could produce DIC signatures that are suitable for the formation of LMC enriched in 16 O and 12 C.
High magnesium calcite is thought to have been formed during early marine diagenesis at elevated Mg/Ca ratios of the pore water (Tucker & Wright, 1990;Swart, 2015;Mavromatis et al., 2017;Purgstaller et al., 2017); thus recording isotopic signatures (ca AE 5& VPDB for d 13 C and d 18 O; Fig. 8) similar to other marine Miocene carbonate deposits, like the middle Miocene lacustrine carbonates from the Bannockburn Formation at Vinegar Hill in New Zealand (Horton et al., 2016) and the bryozoan-rich sediments from the early to middle Miocene Central Paratethys (Key et al., 2013). Assuming a d 18 O isotopic composition of Miocene seawater of À1& SMOW, HMC has been formed at ca 35 AE 2°C, using the equations of Epstein et al. (1953) and Craig (1965) modified by Anderson & Arthur (1983). This temperature estimate is higher than reported for the time-equivalent deposits of the Central Paratethys (for example, 12 to 21°C), but similar to modern and ancient carbonates formed in evaporative, lacustrine and restricted shelf environments (Veizer, 1983;Tucker & Wright, 1990;Baldermann et al., 2012Baldermann et al., , 2015bSwart, 2015). In essence, the d 13 C and d 18 O isotopic signals of authigenic carbonates from the Islam Da g section are not particularly useful for stratigraphic correlations and for palaeo-environmental studies, in contrast to chemical proxies and weathering indices based on the siliciclastic fraction.

Provenance of sediments
Discriminant function analysis of major elements indicates generally a mixed provenance of the Miocene argillaceous rocks from Islam Da g section (Fig. 9). The majority of the samples plot in the 'quartz sedimentary provenance' field and few fall into the 'mafic igneous provenance' group (Roser & Korsch, 1988). No distinct difference in the provenance was observed between the four sedimentary units, which suggests a rather constant detrital input of siliciclastic material during the Miocene and no significant changes in the source area(s).
The current study infers that the Russian Platform was the main source area for the argillaceous rocks from the Islam Da g section at this time, because this area is represented mainly by Precambrian gneiss, amphibolite, conglomerate and sandstone, Cambrian sandstones with clay lenses, Palaeozoic schist, and carbonate and Mesozoic sandstone and shale deposits (Mammadov & Karimov, 1987;Buryakovsky et al., 2001). Physical weathering of these potential host rocks coupled with long transportation distances and intra-basin reworking action will produce mixed, fine-grained sediments rich in detrital components, similar to those of the Islam Da g section (Figs 4 and 7B). If considering that there was no important illite diagenesis, the measured illite chemistry index (0Á3 to 0Á4) and illite crystallinity (0Á4 to 0Á5°2h) may support this assertion: an illite chemistry index lower than 0Á5 represents Fe-rich and Mg-rich illites, which are formed during physical erosion and weak chemical weathering (Esquevin, 1969), whereas a high illite crystallinity index and corresponding low K€ ubler indices (see Table S2) are indicative of a provenance from low-grade metamorphosed domains (Bout-Roumazeilles et al., 2013), in agreement with regional geology and observed trends of terrestrial input fluxes, based on d 13 C org isotopes and hydrogen indices (Hudson et al., 2008). In this line, Popov et al. (2004) have argued that the major positive topographic features, including the Russian and Pre-Ural Highlands, the Ural Mountains and the Ukrainian and Donets lands, drained by the Palaeo-Volga, Palaeo-Don and Palaeo-Donets, could have acted as provenance areas for the sediments of the Eastern Paratethys basins during the early to middle Miocene. Apart from that, a minor contribution of volcanogenic detritus and sporadic input of basaltic-intermediate material from the Greater Caucasus Mountains, for example, from the early Jurassic diabase and gabbro-diabase complex, is indicated by this study's data (Fig. 9). The latter supports the hypothesis that the Greater Caucasus Mountains were emerging during deposition of the Maikop facies and the Spiralisian Formation in the late Oligocene and early Miocene, although major uplifting probably did not occur before the middle Miocene (e.g. Ershov et al., 2003;Vincent et al., 2007;Johnson et al., 2010).

Weathering trends and palaeo-climate
Geochemical proxies and clay mineral assemblages are commonly used for palaeo-environmental analyses of argillaceous rocks (McLennan et al., 1993;Pearce et al., 2005;Chen et al., 2013;Hofer et al., 2013;Sachsenhofer et al., 2015;Richoz et al., 2017;Hellwig et al., 2018). Specifically, major and minor elemental compositions of argillites and clay mineral distributions often record the dominant weathering processes and paths in the source areas driven by palaeoclimate. Physical weathering dominates under arid (or cold) climate conditions and results in the mechanical breakdown of parent rocks and minerals without significant mineralogical and chemical alteration. Chemical weathering is enhanced under humid climatic conditions, leading to distinct changes in the chemical and mineralogical composition of the sediments, as mobile cations (K + , Na + , Ca 2+ , etc.) are preferentially leached away during this process, leaving immobile elements (Al, Si, etc.) in the residue (Nesbitt & Young, 1982). Considering the major input of sediments from the Russian Platform in the early to middle Miocene, as shown before, changes in weathering indices, detrital indicators and clay mineral distributions throughout the Islam Da g section should reflect variations in  (Roser & Korsch, 1988). palaeo-climate in the hinterland. In the present study, the detrital indicators show a co-variation with one another (for example, Na/Al, K/Al, Si/ Al and Ti/Al ratios) and with the percentage of illite (i.e. the most abundant mineral in the clay fraction, see Table S2), and accordingly a clear anti-correlation with the CIA values (Fig. 10). Thus, shifts in these parameters across the Islam Da g section can be used to track changes in palaeo-climate and in the intensity of chemical weathering in the source areas during the early to middle Miocene.
In detail, comparatively low CIA values (82 AE 2), high ratios of Na/Al (0Á10 AE 0Á03), K/ Al (0Á23 AE 0Á02), Si/Al (2Á67 AE 0Á17) and Ti/Al (0Á04 AE 0Á01), and high amounts of illite in the clay mineral fraction are seen for the argillites from unit 1, which could be attributed to less intensified chemical weathering under arid conditions during the early Miocene (Hofer et al., 2013). At the boundary between units 1 and 2, an abrupt shift to comparatively higher CIA values and lower ratios of Na/Al, K/Al, Si/Al and Ti/Al can be seen, probably reflecting an increase in humidity in the mid-early Miocene and intensified chemical weathering (Fig. 10). The gradual increase of the contents of kaolinite a typical weathering product of feldspar (Singer, 1980(Singer, , 1984 and smectite, if detrital, in this interval ( Fig. 3; Table S2) may corroborate this assertion. A 'long-term' transition towards comparatively lower CIA values, higher element/Al ratios and increasing proportions of illite in the clay mineral fraction is observable between units 2 and 3, which may be related to an aridification trend and related reduced rates of chemical weathering during the mid-early Miocene to early middle Miocene (Fig. 10). Physical erosion and reduced chemical weathering under semi-arid to arid conditions could have prevailed during the middle Miocene, as indicated by comparatively low CIA values (76 AE 1), high Na/Al (0Á18 AE 0Á02), K/Al (0Á26 AE 0Á02), Si/Al (3Á07 AE 0Á12) and Ti/Al (0Á04 AE 0Á01) ratios as well as the highest illite contents (see Fig. 10; Table S2). These spatiotemporal variations in palaeo-climate of eastern Azerbaijan in the Miocene generally match with prior climatic reconstructions (for example, seasonal warm to moderately humid) derived from the study of onshore marine deposits from northern Azerbaijan and adjacent areas (Jones & Simmons, 1996;R€ ogl, 1999;Hudson et al., 2008Hudson et al., , 2016Popov et al., 2008;Johnson et al., 2010).

Palaeo-salinity
The TOC/S (=C/S) ratio is often applied to determine the palaeo-salinity (and the oxygenation level) of environments that are represented by organic rich, fine-clastic sediments (Raiswell & Canfield, 2012). Average C/S ratios in non-marine samples (12Á2 AE 13Á9) are generally higher compared to marine samples (4Á3 AE 6Á4), because Fe-sulphides (for example, pyrite) are more abundant in pristine marine deposits than in the freshwater equivalents (Berner & Raiswell, 1984). For the Islam Da g section, the C/S ratios obtained for non-oxidized samples vary in the range from 7 up to 998; thus not recording a clear stratigraphic trend. Nevertheless, the disappearance of pyrite in units 3 and 4 might be an indicator for increased freshwater conditions and related decrease of the sulphate concentration in the basin towards the middle Miocene, taking that the nature and amount of organic matter are similar in all sedimentary units (see TOC contents in Table S3 and the discussion in Hudson et al., 2016).
However, variations in microfossil and palynomorph assemblages as well as ichthyofaunal elements over the Islam Da g section seem to be a more robust tracer for palaeo-salinity, although not specifically investigated in this study. However, findings of Clupeidae, Merlucciidae and dinocysts in the lowermost part of the profile (unit 1) indicate sediment deposition in a 'deepwater' depression at ca 300 to 600 m water depth and under normal-marine conditions during the early Miocene. Changes in ichthyofaunal elements (Argentinidae, Syngnathidae, Stromateidae, etc.) towards unit 2 suggest sediment accumulation near to the coast line under normal-salinity conditions through the mid-early to late-early Miocene (Popov et al., 2008). Remnants of fish and insects, and plant debris, as well as ichthyofossils (phytoplankton is scarce) can be found in early-middle Miocene strata and suggest polyhaline depositional environments (Popov et al., 2008), which matches with the deposition of gypsiferous argillites (unit 3) under evaporitic conditions. Occurrences of Spiratella and of different endemic species (Popov et al., 2008) in the uppermost part of the Islam Da g section indicate restricted shelf basins and seasonal brackish to polyhaline conditions throughout the middle Miocene. These changes in palaeo-salinity coincide well with the proposed aridification and freshening events established on the basis of chemical proxies (Fig. 10), facies changes observed across the Islam Da g section (i.e. shift from the deposition of carbonate-free argillites in a deep-basin setting to sedimentation of calcareous and gypsiferous argillites in a partly evaporitic, shallow water environment; Fig. 2) and with sedimentological and palaeontological analyses of carbonate argillites from the Spiralisian Formation and Upper Miocene Diatom Suite (Ershov et al., 2003). This study suggests that seasonal freshening may be attributed to humid climate and freshwater input by the Palaeo-Volga and Palaeo-Don rivers.

Palaeo-redox conditions
Carbon-sulphur-iron associations and Fe/Al ratios can be used to constrain the oxygenation level in marine sediments, as these proxies typically reflect fluxes in the detrital or hydrothermal input, primary bioproductivity and benthic shuttling of Fe during and after deposition. Based on the study of 13 outcrop localities in eastern Azerbaijan, Johnson et al. (2010) have shown that the C-S-Fe distributions are predominated by the contribution of detrital Fe, especially in the Palaeo-Eocene and middle to late Miocene strata, where the TOC and pyrite-S contents are low (<1 wt. %). All Oligo-Miocene samples display higher amounts of TOC (up to 6Á3 wt. %; Hudson et al., 2008) and pyrite-S (ca 1 to 2 wt. %; Johnson et al., 2010).
The argillites from the Islam Da g section exhibit generally high TOC contents (4Á9 AE 0Á8 wt. %), low pyrite-S contents (0Á17 AE 0Á22) and S/Fe ratios significantly smaller than that of pyrite (for example, 0 to 0Á2 versus 1Á15). This indicates that most of the Fe is present in Fe-bearing silicates and labile Fe-oxyhydroxides, as indicated before. Taking that hydrothermal sources for Fe are not available and that the contribution of Fe by the volcanic ash deposits is minor (i.e. only four and thin laminae of volcanic ash were recognized in the studied profile; Fig. 2), it can be inferred that the major input of Fe is of detrital origin (i.e. bound to illite and minor chlorite and Fe-oxyhydroxides). Regarding the nature of organic matter in the argillites, the presence of ubiquitous amounts of terrestrial higher plant material as well as evidence from d 13 C org isotopes and biomarkers (i.e. C 30 sterane index and the terrigenous/aquatic ratio: TAR) suggest that a high amount of terrestrial TOC input to the basin was present throughout (Hudson et al., 2016). Interestingly, offshore deposits of the South Caspian Basin with high terrestrial TOC contents are more prolific oil-prone source rocks than timeequivalent onshore deposits having higher contents of algae-TOC, which do not comprise excellent oil source rocks (Devlin et al., 1999;Feyzullayev et al., 2001;Hudson et al., 2016). The reasons for this are not completely understood yet, but it seems likely that stratified (partly lacustrine) basins with a restricted oxygen circulation in bottom waters have been developed in the Miocene, as the Arabian Peninsula moved northward, with apparent implications for the preservation of TOC (Vincent et al., 2007).
The lack of co-variance between the detrital indicators and the Fe/Al ratio as well as the low-S, moderate-Fe and high-TOC compositions ( Fig. 10) suggests further that syndepositional and early diagenetic processes have affected the sedimentary inventory of Fe and TOC. Specifically, the increasing trend in the Fe/Al ratio towards the top of the studied profile indicates increased accumulation of Fe-oxyhydrates and Fe-illite due to the establishment of a well-oxygenated water column towards the middle Miocene. Abundant Fe(III)-bearing smectite throughout in the Islam Da g section further suggests a certain degree of benthic Fe redox recycling and scavenging of dissolved Fe immediately at the interface between reducing pore waters and oxygenated bottom waters, transforming volcanic ash into Fe-smectite and zeolite minerals (see Fig. 2). The formation of smectite could have contributed further to the preservation of sedimentary TOC by lowering the permeability of the bulk sediment and acting as a reactive template for the sorption of organic molecules onto the charged Fe-smectite surfaces (Canfield et al., 1993;Poulton et al., 2010;Taylor & Macquaker, 2011;Raiswell & Canfield, 2012;Baldermann et al., 2015aBaldermann et al., , 2019. This study agrees with Johnson et al. (2010) and Hudson et al. (2016) that the majority of Fe and TOC may be of detrital origin, but questions whether the Fe/Al ratio can be used as a proxy for the Fe sources. In essence, it is plausible that the Oligo-Miocene Maikop facies has been deposited under anoxic conditions, as previously suggested by Hudson et al. (2008) and Johnson et al. (2010). However, the early to middle Miocene strata of the Islam Da g section have been accumulated under oxygenated conditions.

SUMMARY AND CONCLUSIONS
The early to middle Miocene sedimentary succession of the Islam Da g section in the Gobustan-Absheron depression (eastern Azerbaijan) was deposited in a restricted marine to lacustrine basin with pronounced recurrent periods of poorly oxygenated conditions, intensified evaporation and freshening events. The succession consists of organic-rich argillaceous rocks, accumulated during phases of high influx of siliciclastic and volcanogenic components from Russian Platform and Greater Caucasus source areas, respectively, with intercalated thin siltstone lenses, siderite concretions, volcanic ash layers and mudstone beds, deposited during phases of low detrital input. Detrital indicators, weathering indices and palynomorph assemblages reveal varying arid or semi-arid to moderately humid climates and related alternating polyhaline and euhaline to evaporative conditions in the depositional basin, concurrent with large-scale fluctuations in palaeogeography, orogenic activity and basin configuration in the Cenozoic. Trends in C-S-Fe distributions and Fe/Al ratios suggest high detrital inputs of Feillite, Fe-oxyhydrates and total organic carbon in a mostly well-oxygenated water column as well as significant syngenetic Fe scavenging through Fe(III)-smectite formation. Microbial heterotrophic and abiotic reduction of sulphate and Fe-oxyhydrates immediately at the sediment-seawater interface caused pyrite precipitation under reducing conditions, which further suppressed the oxidation of organic matter, making these offshore deposits excellent oil source rocks, compared to time-equivalent onshore deposits. In conclusion, this study illustrates the potential of the deposits of the Islam Da g section to be used for regional-scale palaeo-environmental reconstruction of the Miocene Eastern Paratethys and for lithostratigraphic correlation with other onshore deposits, taken that this succession is largely unaffected by pedogenesis, cementation and compaction. Significant post-depositional alteration, such as oxidation of pyrite and formation of jarosite, the conversion of volcanic ash into smectite and zeolite and the formation of carbonates associated with meteoric diagenesis, can smear the sediment's pristine signatures, thus hampering the interpretation of depositional and palaeo-environmental trends recorded in proxies based on isotopes. Future work focusing on the in-depth characterization of the petrographic, mineralogical and chemical signatures of Miocene onshore and offshore deposits of the Euxinic-Caspian basins is needed to constrain the regional-scale response of the Eastern Paratethys to plate tectonics, weathering paths and climate change.

Supporting Information
Additional information may be found in the online version of this article: Table S1. Bulk mineralogy of argillaceous rocks from the Islam Da g section (Shamakhy-Gobustan area, eastern Azerbaijan). Table S2. Distribution of illite (Ilt), smectite (Smc), chlorite (Chl) and kaolinite (Kao) (in per cent) among the total clay mineral fraction (∑clays) across the Islam Da g section. Table S3. Bulk geochemistry and d 18 O and d 13 C data of the Islam Da g section (Shamakhy-Gobustan area, eastern Azerbaijan).