Crandallite-rich beds of the Libkovice Member, Most Basin, Czech Republic: climatic extremes or paleogeographic changes at the onset of the Miocene Climatic Optimum?

K . M a c h e t a l . G e o l o g i c a A c t a , 1 9 . 1 1 , 1 2 9 , I X ( 2 0 2 1 ) D O I : 1 0 . 1 3 4 4 / G e o l o g i c a A c t a 2 0 2 1 . 1 9 . 1 1 Crandallite-rich beds of the Libkovice Member, Most Basin

We describe the occurrence and possible origin of rare beds 1-10cm thick and containing 20-70% of crandallite, a Ca-Al phosphate enriched in Sr and Ba, found within otherwise monotonous clay-rich lacustrine sediments of the Most Basin in the Central-European Neogene Ohře Rift system. The beds were formed at ca. 17.31, 17.06, and 16.88Ma, while the entire suite of monotonous clays of the Libkovice Member was deposited between 17.46 and 16.65Ma. Trace-element and organic geochemistry, Ar-Ar geochronology and C-O-Sr isotope systematics are used to infer their source and processes leading to their formation. The most enigmatic aspect of the formation of the crandallite beds is the removal of a huge amount of phosphorus from its biogenic cycle in the lacustrine system, which was otherwise stable for ca. 0.8My. Formation of detritus-poor crandallite beds could result from some exceptional environmental disruptions that hindered transport of fine clastic material to the basin floor. Silicic volcanic activity in the area of the Pannonian Basin could have triggered this disruption. Crandallite could provide evidence of long-lasting droughts and acidification of the exogenic environment, as they are roughly coeval with the onset of the Miocene Climatic Optimum at ca. 17.0Ma.

INTRODUCTION
This paper deals with phosphate-rich beds found within the Lower Miocene lacustrine fill of the continental Most Basin within the Eger Rift in Central Europe (Matys Grygar and Mach, 2013;Matys Grygar et al., 2014 Fig. 1). The main mineral component of these beds is Sr-and Barich Ca-Al phosphate of the crandallite group (CaAl 3 (PO 4 ) (PO 3 OH)(OH) 6 ) (Novák et al., 1993). Widespread beds containing crandallite intercalated in otherwise monotonous siliciclastic lacustrine sediments have not been described elsewhere, and the mechanism of their formation is thus not well constrained.
The crandallite beds (C#) in the Most Basin are found at the base of certain K-concentration minima within lacustrine clays of the Libkovice Member pointing to onsets of periods of more intensive chemical weathering in the watershed (Matys Grygar et al., 2019a). Because the K-concentration minima in the Most Basin were orbitally paced , 2019b, we hypothesize that Crandallite beds manifest some special environmental conditions related to the climatic cyclicity. The aim of characterizing environmental changes during sedimentation of the Most Formation is motivated by insufficient knowledge of what triggered the Miocene Climatic Optimum that started around 17.0Ma (Holbourn et al., 2015). The individual C# are present as outcrops in opencast mines and drill cores have been systematically sampled and documented since 2009. The geochemical research of surrounding clays was carried out concurrently and already published . This paper summarizes both new and previously published knowledge we have gathered on these beds resulting in an interpretation of their origin.

STUDY AREA AND MATERIALS Geological background
The Most Basin (Fig. 1A) is the largest sedimentary basin in the Oligocene to early Miocene tectonic structure of the Ohře Rift (Eger Graben) in the north-western part of the Bohemian Massif (Kasinski, 1991;Rajchl et al., 2009). The basement of the basin is formed by Late Proterozoic gneisses, Carboniferous felsic magmatic rocks (Altenberg-Teplice Volcanic Complex) and Late Cretaceous marine sediments (mainly sandstone, limestones and marlstones). Basin filling began with the Oligocene volcanogenic Střezov Formation, an equivalent of exposed rocks of the České Středohoří and Doupovské Hory volcanic complexes surrounding the Most basin from the SE and W. Both are represented by alkaline basalts, basanites, tephrites and various trachytic rocks. The Miocene fill is formed by the Most Formation, subdivided into the Oligocene/Miocene transitional Duchcov Member, the main coal seam-bearing Holešice Member, monotonous lacustrine clays of the Libkovice Member, variegated Lom Member containing so-called Lom Seam on its top, and the youngest lacustrine clays of the Osek Member (Fig. 1D). Clastic parts of the Most Formation have been formed by sedimentation of locally-sourced weathered material (the Krušné Hory Mts., the České Středohoří Mts., and the Doupovské Hory Mts.) as well as more distal sources in the south-west of the Bohemian Massif (Fig. 1C), as revealed by evaluation of heavy minerals associations ). According to this research of heavy minerals, older basin sediments were formed by the end of the Oligocene and beginning of the Miocene from local weathering products of volcanic rocks in České Středohoří and Doupovské Hory mountains together with Proterozoic and Paleozoic metamorphic formations and the Carboniferous felsic magmatic rocks. Sedimentation occurred in a fluvial environment, showing a gradual spreading of the river floodplains. The esediment routing system extended from catchments areas in the SW part of Bohemian Massif to the marine depositional environments of the Leipzig bay of the Miocene North Sea. Later, local sources of clastic material on the floodplain were gradually covered by expanding peat-forming swamp and more distally-transported material from Proterozoic and Paleozoic metamorphic formations in the south-western part of the Bohemian Massif with an admixture of the marine Cretaceous sediments from areas east of the Most Basin gradually became more prevalent. Within the layer of peat, this clastic alluvial input formed a widespread accumulation of sandy clay intercalations locally called the Žatec Delta (Fig. 1A). Geochemistry and mineralogical composition of associated clays correspond with the described heavy mineral associations. Peat accumulation was gradually declining as clastic input increased to the Bílina area, where a local lake was formed with fast sedimentation supported by peat compaction. This lake together with others lakes covered the entire area of the basin in stepwise fashion and started the lacustrine stage of the basin development . Heavy minerals of the Libkovice Member formed by this basin-wide lake are inaccessible, because only fine sediments have been preserved from the lacustrine phases, however the chemical composition of the Libkovice Member is comparable with the uppermost clays of the Holešice Member (the Břešťany clay). The clays within the Most Formation should be classified as silty clays, because of considerable and variable contents of quartz silt dispersed in the clay. More significant for the 200m thick Libkovice Member is the difference in the mineralogical composition, that can be attributed to the spreading and deepening of the lake during the deposition of this member (Matys Grygar and Mach, 2012). This change created distinct sedimentary conditions and led to more pronounced grain-size sorting expressed not only by variable content of silt but also by composition of clay minerals. Typical kaolinite or kaoliniteillite clay composition of the Holešice Member was replaced by sedimentation of kaolinite-illite-illite-smectite composition during the Libkovice Member formation. Hypothesis of considerable paleoenvironmental change during the Libkovice Member formation is supported by differences in the dispersed organic matter composition between the Holešice and the Lom members on one hand and the Libkovice Member on the other hand. Havelcová et al. (2015) found a higher proportion of liptinite macerals, mainly alginite (from algae) against huminite macerals (from plant detritus) in the Libkovice Member clays comparing to the Holešice Member. A climatically driven weathering regime in the basin catchment of South-western part of the Bohemian Massif led to the variability of material eroded and transported to local lacustrine reservoirs within peatforming swamp and then to a single basin-wide lake over the former peat swamp. This seemingly inconspicuous cyclic variability in geochemical composition could be used to date the sedimentary sequence by means of cyclostratigraphy in combination with magnetostratigraphy . Shallowing of the Libkovice Lake and changes in the distributary network of the alluvial system led first to the return of kaolinite-illite clay sedimentation (15-20m thick) and then to transition from lacustrine to swamp environments on its northern border . The product of swampy conditions is now represented by the so-called Lom Coal Seam, a 15m thick part of the Lom Member (Figs. 1D;2). The top of the sedimentary profile was formed in renewed lacustrine conditions. This stage is represented by silty clays of the Osek Member similar to the Libkovice one.
Authigenic minerals represent substantial components of the basin fill together with clay minerals and quartz silt or sand. They have either dispersed, concretional, or thin layer forms that occur within all sedimentary units of the Most Formation. Commonly, these minerals are represented by carbonates of the siderite-ankerite-dolomite-calcite series, and by pyrite or marcasite. Most of these minerals were formed by bacterial processes of sulphate reduction or methanogenesis (Mach et al., 1999(Mach et al., , 2001. The evolution of the Most Basin was described by Rajchl et al. (2009). The majority of the basin fill was formed during the initial S-N extension regime in four basic subdepocenters divided by E-W trending faults. Finally, there was a change to NW-SE extension, which led to interruption of sedimentary filling of the basin and formation of present system of SW-NE trending fault system. The Most Basin was a hydrologically open, overfilled basin through its time of filling, because there are no signs of dry periods leading to aerial exposure of the lake bottom such as in the nearby Sokolov Basin of the same age (Rojík, 2004). Even during peat-forming periods, input of water to the basin was in equilibrium with output. This is documented by continuing transfer of Bohemian Massif clastic material to the area of Germany .

Crandallite in the Most Formation
In the Most Formation phosphates were accumulated in form of layers with thicknesses of a few cm (crandallite layers or beds) or in the form of phosphate-rich concretions. The latter are known from the Holešice Member, in particular in the clay partings of the Main Coal Seam in the Chomutov area (Novotný and Mach, 2017) but they are not described in this work. Crandallite beds are a volumetrically minor, but very specific part of the Libkovice Member (Fig. 2). The Libkovice Member includes monotonous lacustrine silty clays with a significant portion of clay minerals with smectite structures. Several laterally-continuous crandallite-rich beds, each with thicknesses of 0.5 to several cm, have a specific stratigraphic position within sediment chemostratigraphy, developed by paleomagnetic dating and cyclostratigraphy , i.e. they were formed coevally on the paleolake bottom. The crandallite-rich layers are overlaid by sediments with clear Sr concentration minima, these both are located in the bottom part of the K-concentration minima identified in the entire basin floor (Matys Grygar and Mach, 2013) (Figs. 2;3). The composition of the crandallite-rich beds was first studied in the Merkur and Libouš coal mines in the western part of the basin (Coufal and Mejstříková, 1996;Novák et al., 1993). According to these previous investigations, it consists of Ca-Al phosphate with some other mineral admixtures. Usually, the phosphate is defined as a Sr-, Ba-rich crandallite, a member of isomorphous series of crandallite-goyazite-gorceixite minerals. The crandallite bed contains up to 70% of phosphate, a significant portion of which can consist of metastable SiO 2 polymorphs, mainly cristobalite and tridymite (Novák et al., 1993) and clay minerals (kaolinite, illite). There are three prominent crandallite-rich beds within the Libkovice Member; they are numbered consecutively C# (C1-C3) in an upward direction (Matys Grygar and Mach, 2013;. Several other not-so-striking occurrences have been found later. These were labelled C1a and C2a, the latter divided to C2' and C2''. During detailed geochemical research it has been found that not only centimeter thick C# beds contain crandallite. Dispersed crandallite particles result in anomalously high contents of Sr, Ba and P in layers of clays with several meter thicknesses, typically with Sr concentrations of 100-350ppm Sr in comparison to ca. 50-70ppm base concentrations, i.e. ca. 2-5 times above the background (Fig.  3). These anomalies are correlated over 20-80 kilometers between drill cores (Fig. 2) within the topmost parts of the Holešice Member and the Libkovice Member clays (Matys Grygar et al., 2017). Even so, this anomalous Sr/Ba/P content is two orders of magnitude lower than in C# beds. In the Libkovice Member clays, elevated Sr concentrations correlate with elevated Al/Si (Fig. 3), corresponding with stages of higher supply of kaolinite clay minerals in relation to silt-sized quartz grains. As a result of this correlation, higher Sr/Ba/P content is well displayed by cation changing capacity curve (minima) and a geophysical log of electric induction (minima). All these features and methods show a specific composition of the clay content and mineralogy associated with crandallite (Fig. 3).

Studied sediments
A complete correlation scheme of drill hole profiles down the Most Basin longitudinal axis (Figs. 1; 2) shows relations of lithostratigraphic units above than Crandallite-rich beds of the Libkovice Member, Most Basin 6 main coal seam together with magnetic polarity and chemostratigraphic horizons. As a representative profile, the LB433 drill hole from the central part of the basin was chosen (Fig. 3). The depth-age model for the LB433 core was obtained by matching K minima to the orbital eccentricity maxima as described earlier by Matys Grygar et al. (2019b). The depth-age model is based on integrated stratigraphy based on biostratigraphy, magnetic polarity, and cyclostratigraphy, corroborated and validated by extensive lateral correlation of sediment cores , 2019b. The primary age model for LB433 was obtained by alignment to the stratigraphic scheme published in Matys Grygar et al. (2019b). The profile of LB433 borehole consists of all known stratigraphic units of the Most Basin younger than the main coal seam of the Holešice Member. Profile starts from the 18m thick upper half of main coal seam. Then starts clayey part of the profile by 65m of clays of Holešice Member local Bílina lake. Younger monotonous clays of the Libkovice Member reach 204m. Libkovice Member clays are usually gray. As it is usual on the central part of the basin, clays of the Libkovice Member are fossil barren, but we can distinguish ichnofossils representing some token of lacustrine life. Two sub-units can be distinguished in the Libkovice member profile, the lower half frequently bioturbated by Planolites montanus borings (Mikuláš et al., 2006) and the upper half with abundant occurrence of sulphidic hairy structures related to "Trichichnus" ichnofossils (Kedzierski et al., 2015). Dense, 0.5-2mm in diameter tube-like formations, Planolites montanus (Mikuláš et al., 2006) are interpreted as domichnia of unknown small unvertebrates. Their occurrence points at more oxic conditions in top 1-2cm of the lake bottom mud. On the contrary thin pyritic hairy fibers, "fossilized bioelectric wires" of Trichichnus (Kedzierski et al., 2015) are interpreted as a product of bacterial activities in the oxygen-depleted part of sediments and so they are related nearly to at least temporary anoxic conditions on the surface of bottom sediment. Libkovice Member hosts C# beds. A typical feature of C# beds occurrences within outcrops in open cast mines is a contrast of the color of clays on either side of thin C# beds (Fig. 4). Mineralogical and geochemical differences lead to the darker gray color of underlying clay and light gray color of the overburden. The overburden layers commonly display better visible layering and seem lighter within several meters thickness (Fig. 4). In the case of C1 in the Libouš Mine this lighter overburden clay is even fossiliferous (unpublished association of macroflora with rare insect imprints) unlike the uniform fossil-barren Libkovice Member.
The subsequent 18m of the Most Formation in the LB433 profile are represented by clays of lacustrine Lom Member clays and 17m of the Lom Member Coal Seam. The top of the profile is formed by 15m of the Osek Member

METHODS
We used a variety of methods to describe the extraordinary phosphate occurrences, their specifics within the Holešice and Libkovice members and thereby searched for evidence of their origin.

Microscopic description and mineral chemistry
Twenty 2×3cm petrographic thin sections were prepared at the Czech Geological Survey in Prague (CGS). The sections represented the majority of C# beds, in both horizontal and vertical directions. The textures, structures, granularity, and mineral composition of the beds were studied. Five polished sections were also prepared from the C1 bed in the Libouš Mine. The sections were used to study the chemical composition of individual clasts, crystals or amorphous phases using a JEOL JxA-8600 electron microprobe equipped with SE (Secondary Electron), BSE (BackScattered Electron), EDS (Energy-Dispersive X-ray Spectroscopy) and WDS (Wavelength-Dispersive X-Ray Spectroscopy) detectors housed at the Dept. of Geology, Faculty of Science, Palacký University in Olomouc. An accelerating voltage of 15kV, beam current of 10nA and counting time of 100s were used for analysis of silicates and oxides. The following standards were used: jadeite (Na), barite (Ba), apatite (P), diopside (Mg, Ca), microcline (Si, K, Al), elemental manganese (Mn), magnetite (Fe), sphalerite (S, Zn), albite (Na), ilmenite (Ti). Chemical composition of biotite crystals from the sandy basis of the C2 bed sampled at Bílina and Tušimice mines, together with biotite flakes from the Sokolov Basin (Rojík, 2004) were studied separately using the Tescan MIRA 3GMU Field Emission Gun-equipped Scanning Electron Microscope (FEG-SEM), housed at the CGS, equipped with the X-Max 20 (Oxford Instruments) EDS microanalysis system to avoid preferential evaporation of alkali elements from the matrix (cf. WDS). EDS analyses were conducted with an accelerating voltage of 15kV, beam current of 3nA and the working distance of 15mm. For quantitative EDS analyses, the SPI set of mineral standards was used for standardization, pure Co for quant optimization.

Chemical and mineralogical composition
Selected samples of the parts of the C# beds richest in crandallite and also sandy bases of the beds were collected for bulk-rock chemical analyses. Conventional wet chemical analyses (major elements) were provided by laboratories of the CGS. Trace element concentrations (including Rare-Earth Elements (REEs) were determined using the ThermoFisher Scientific X Series II ICP-MS Crandallite-rich beds of the Libkovice Member, Most Basin 8 (mass spectrometry with inductively coupled plasma; Ba, Cr, Ga, Hf, Nb, Ni, Pb, Rb, Sr, Th, U, Zr, Y, REEs) housed at the CGS. For the ICP-MS determinations, the samples were decomposed fusion with LiBO 2 -Na 2 CO 3 mixture in a Pt-beaker.
Mineral composition of 18 representative samples was determined semi-quantitatively by conventional powder X-Ray Diffraction (XRD). The majority of these samples (14) were analyzed in the Brown Coal Research Institute accredited according to ČSN EN ISO/IEC 17025. Analyses of powder samples were realized using a D 5000 Siemens diffractometer and evaluated with the DIFFRAC plus EVA application. Part of the software is database "Powder Diffraction File" (PDF) updated by the International Centre for Diffraction Data (ICDD). The main analytical parameters were: acquisition time 1 hour, angular spectrum 3 ÷ 80° 2θ detector slit 0.2mm, primary X-rays slit 2mm, secondary X-rays slit 2mm, angular step 0.02°.
Composition of additional four samples was analyzed semi-quantitatively in laboratories of the CGS. Powder X-ray diffraction data were here collected in Bragg-Brentano geometry on a Bruker D8 Advance diffractometer equipped with a LynxEye XE detector and Soller slits (2.5°) in primary and secondary beams. CuKα radiation was used. The samples mixed with acetone were gently pulverized in an agate mortar. The data were collected in an angular range from 4 to 80° of 2θ, with a 0.015° step and counting time of 0.8 sec. per step. Qualitative phase analysis was performed using the DIFFRAC. Eva (Bruker AXS, 2015) program and the PDF-2 database (ICDD, 2002). Subsequent quantitative phase analysis was carried out by the Rietveld method using the Topas 5 (Bruker AXS, 2014) program. Crystal structure models for the detected phases were adapted from the Inorganic Crystal Structure Database (Zagorac et al., 2019).

Analysis of C and O stable isotopes
Compositions of stable C and O isotopes in carbonates (siderite) in the sediment and the C# beds (C1 from the Libouš Mine) were examined at the CGS. Stable isotope composition of carbon (δ 13 C‰) and oxygen (δ 18 O‰) in samples of siderite was determined in CO 2 gas produced by the decomposition of the sample in 100% phosphoric acid under vacuum condition, at the temperature of 100°C for 24 hours (Rosenbaum and Sheppard, 1986). The isotopic composition of carbon and oxygen in emitted CO 2 gas was measured on Delta V Advantage (ThermoFisher, Bremen, Germany) mass spectrometer. The total error yields ±0.1‰. Isotopic composition of carbon is relative to V-PDB international standard. In the case of siderite, the δ 18 O value is corrected by the value +1.44.

Strontium isotope systematics
Strontium isotopic composition of the studied sediments as well as of potential source rocks were obtained at the CGS. Silicate samples were dissolved using a combined HF-HNO 3 -HCl digestion. Carbonate samples were first digested in HCl and the remaining undissolved silicate residue was dissolved by HF-HNO 3 -HCl; the extracts were combined prior to chromatographic separation. Strontium was isolated from the bulk matrix using Sr. spec resin (Triskem Intl.) (Pin et al., 2014). Isotopic analyses were performed with a Triton Plus (Thermo Fisher Scientific Inc.) thermal ionization mass spectrometer in static mode using a single Ta filament assembly. The measured 87 Sr/ 86 Sr ratios were corrected for mass fractionation assuming 86 Sr/ 88 Sr= 0.1194. External reproducibility and measurement accuracy were demonstrated by repeated analyses of the NBS 987 ( 87 Sr/ 86 Sr= 0.710253±0.000018 (2σ, n= 54) reference material (see Jochum et al., 2005). Analyzed samples represent Miocene phosphates and carbonates, clays with normal and anomalous Sr content, and also by rocks representing potential sediment sources gneisses and acid magmatic rocks of the Krušné Hory Mountains and Cretaceous marls from the Louny area south of the Most Basin. All data, newly acquired and found in the literature, were recalculated to the time of deposition, i.e. 17Ma, based on constraints given by Steiger and Jäger (1977). Sr and Rb concentrations for the samples not analyzed by ICP-MS were obtained using portable XRFanalyzer Niton 3 Gold+ Thermo Fischer Scientific as an average of 5 measurements.

Organic Geochemistry
Powdered samples (<0.2mm) of C1, C2 and C3 beds and clays surrounding the C# beds sampled from the outcrops in the Bílina Mine were extracted using ASE extractor (Dionex) by dichloromethane. Gas chromatography/mass spectrometry (GC/MS) analysis were performed using a Trace 1310 1310 Gas Chromatography (GC) equipped with an ISQ (In Situ Quantitative) single quadrupole MS (Thermo Scientific) instrument equipped with a TR-5MS column (60m × 0.25mm × 0.25μm). The oven temperature program was set to 1 min step at 40°C followed by increase to 120°C at rate of 15°C/min, than increase to 220°C at rate of 6°C/min followed by increase to 300°C at 12°C/min and finally 5 min delay at 300°C. Samples were injected in splitless mode, the injector temperature was set at 250°C and helium was the carrier gas. Mass spectra were obtained by scanning from m/z 45 to 650 in full scan mode. Aliquots of the total extracts were converted to ester derivatives by reaction with methanol and 14% boron trifluoride (BF3) for 1 h at 90°C and analyzed under the same conditions. The mass spectrometer detector was operated in the Total Ion Current acquisition mode and was combined with the Selected Ion Monitoring acquisition mode for alkanes, and methyl ester fatty acids. For data processing, the Chromeleon software (Thermo Scientific) was used.

40
Ar-39 Ar geochronology of biotite Biotite samples were separated from the bottom part of C2 from the Bílina and Libouš mines. For comparison, a biotite sample from the so-called greigite horizon of probably aeolian origin (Rojík, 2004) of the lower Miocene in the Sokolov Basin (the Jiří open cast mine) was prepared.
The biotites were prepared after coarse mechanic separation in the field followed by physical-chemical separation in the Czech Geological Survey laboratory. The separates were irradiated for 7 hours in the NM-304 package at the Oregon State University TRIGA reactor (CLOCIT position) along with the standard Fish Canyon tuff sanidine as a neutron flux monitor. The samples were analyzed in the New Mexico Geochronology Research Laboratory (NMGRL) in Socorro by a combination of bulk aliquot step-heating and single crystal laser fusion methods using a defocused diode laser to heat the samples.

Specific weight
5 whole-rock samples of C# beds (weight 25-40g) were tested to determine approximate specific weight (density) in nature and dry stage. A simple method of weighing samples and measuring their displacement volume in a 200ml water cylinder were applied with approximately 5% precision in laboratory of the Severočeské doly a.s.

Macroscopic description of individual C# beds
C1 is stratigraphically the lowest, and hence the oldest C# bed. Because of its position in the oldest part of the lacustrine sequence, it is present in major part of the basin (Fig. 4A, B, D, F). C1 is less than 1cm thick in the Bílina area ( Fig. 4A), whereas in the Chomutov-Kadaň area (the Libouš Mine and its vicinity) its thickness is 6-12cm (Fig. 4B, F). The C1 bed always contains laminae of fine biotite-quartz sand at its base. In contrast to the later C2 bed (described below), the biotite at the base of C1 is commonly strongly altered (sericitized) and could not hence be separated and further analyzed. The upper half of C1 in the Libouš Mine area is hard, while the lower parts of the bed in the Libouš Mine and the entire bed in other areas is rather soft. Convolute structures approximately in the middle part of the C1 bed (Figs. 4F; 5A) may be evidence of Kelvin-Helmholz and/or Rayleigh-Taylor instabilities (Fernando, 1991;Gladstone et al., 2018), whereas the top third of the bed shows clear planar lamination (Fig. 4F).
C1A bed was discovered in the Libouš Mine as a discontinuous horizon with a maximum thickness of 0.5cm. It is located between the C1 and C2 beds near the bottom of K4 orbital-triggered minimum of potassium (or K/Al) concentrations. A similar discontinuous, up to 1mm thick layer of anomalous P, Sr, and Ba contents occurs near the K4 minimum (according to the stratigraphic scheme in Matys Grygar et al., 2017) in the Bílina Mine. It could be a lateral equivalent of the C1A bed in the Libouš Mine. C1A has not been found in any heretofore studied drill core.
C2 bed (Fig. 3C) has the typical appearance of the C# beds mainly in the central part of the basin. It also occurs in the Chomutov-Kadaň part of the basin, while further to the south and west the strata with C2 have not been preserved. The strata with C2 are also absent in the area near Krušné Hory Mountains (the ČSA open cast mine). The bed is 1.8-2.5cm thick in most localities. At the base of C2, there is a 2mm thick lamina of fine quartz-biotite sand (Fig. 4G).
In the vicinity of Teplice in the eastern part of the basin, this bed is up to 1cm thick, and differs from typical C2 by a lack of biotite and quartz sand on the bottom. Biotite is macroscopically dark brown, reflective on cleavage planes, without any signs of alteration (Fig. 4G). The base of C2 bed in DU7 drill core is formed by fine rounded grains (clasts) of kaolinized rocks cemented by crandallite.
The bed C32' was first found in outcrop in the Libouš Mine but later was also identified during documentation of drill cores in the Bílina Mine area (LB433, LOM30, HK930). It is represented by a 0.5cm thick, continuous crandallite-rich bed with less than 1mm thick laminae of quartz-biotite sand at its base.
Similar bed C2'' has until now only been found in two drill cores --LOM30 and LB433--in the Bílina Mine area. The maximal thickness of this discontinuous orangecolored layer is 3mm. Biotite flakes at its base were not identified.
C3 is the youngest identified C# bed. Its similarity to C2 (the same thickness and shape) prevented the earlier distinguishing of C2 and C3 in the outcrops of the Bílina Mine. The C3 thickness is 1.8-2.5cm. It differs from C2 only by a very thin or indistinct sandy lamina on its bottom, usually with biotite flakes not observed macroscopically due to their intensive weathering. As the C3 bed occurs in the topmost part of the Libkovice Member, its presence (preservation) is limited to the deepest part of the Most Basin between cities of Bílina, Litvínov, and Osek (Fig. 4C). Only a limited erosional remnant of C3 occurred several m from the surface within the deepest overburden profile of the Libouš Mine. In all C# beds, crandallite shows the microcrystalline character and gray or yellow to orange colors on their fresh cuts. These colors quickly turn white by weathering in outcrops. The C# beds typically disintegrate to prisms via two systems of perpendicular cracks. The bottoms of the beds show irregular but very contrasting surfaces dividing the underlying clay and C# bottom sand with many dents and mounds reflecting the original shape of the lake bottom (Fig. 4D). In some places it resembles circular bioturbation holes of Selenichnites, distinguished in sediments of the Holešice Member (Mikuláš et al., 2006). On the other hand, the upper contacts of the beds with clays are sharp and flat in all cases ( Fig. 4A, C, E, F). The internal structure of the majority of C# beds is similar and does not depend on the bed thickness. Substantial difference of C1 structure in the Libouš Mine is described above. In contrast to the situation in hosting Libkovice Member clays described above (LB433 borehole Fig. 3) no signs of bioturbation were found in any occurrence of C# beds or its upper surface.

Microscopic examination of C# beds
Both optical and electron microscopy confirmed the occurrence of distinct thin (1-3mm) sandy to silty laminae at the bottom of C# beds (Fig. 6A, B, C). In C3, the basal coarser lamina was not apparent macroscopically (Fig. 6C). Basal sandy to silty laminae consist of very fine-grained phosphate matrix with irregular to amoeboid crandallite grains. Individual idiomorphic biotite flakes (up to 0.5mm in their largest dimension, Fig. 6D) and sharply limited irregular fragments of quartz (0.1-0.5mm in diameter, Fig. 6E, H, I) are embedded within this matrix. Flakes of green-brown biotite have commonly a perfectly euhedral shape of hexagonal plates (Fig. 6F) in some samples, with pronounced pleochroism (Fig. 6H, I). In some vertical cross-sections, the weathering of biotite crystals is obvious as the brush-shaped cleavage of their rims (Fig. 6D). In more weathered samples, e.g. C2 from the Libouš mine, the biotite crystals have lost their color and pleochroism. Present in every sample within the basal sandy laminae are apatite grains or crystals (Figs. 5; 6F) and irregular grains of some mineral completely transformed to a very fine mix of probably amorphous SiO 2 and clay minerals. These are oval to angular grains with medium-high relief, often with inclusions of opaque minerals (Fig. 6F in the middle or 5D in the upper right part of the photo). Grain size of apatite and transformed minerals is similar to grain size of quartz. Apatite is also present as inclusions in biotite (Figs. 1; 6F). Several grains of a grass-green mineral with low relief without any cleavage were found; we interpret them as glauconite ( Fig. 6E) but the small amount of material did not allow more accurate identification. Other common constituents of basal laminae are opaque (ore) minerals in globular concretions (up to 10μm in diameter Fig. 6F) or in the form of grains of various forms with similar dimensions (Fig. 6D, E) or they highlight the lower contact of the C# bed with underlaying clay (Fig. 6A, B, C). High relief enhances visibility of a common minor admixture of irregularly dispersed fine idiomorphic crystals of 5-10µm large siderite (Fig. 6H, F, G). Biotite and quartz represent up to 50vol.% of the rock near the bottom of the bed, but its occurrence significantly decrease upwards (Fig. 6A, B). Their decrease correlates with the decreasing grain size. Although the admixture of quartz and biotite within the upper half of the C# beds is not visible macroscopically, microscopy in many cases confirms their sparse presence. Their proportion in the upward direction decreases below the XRD detection limits. In all cases, the uneven, sharp contact with the underlying clays was confirmed (Fig. 6A, B, C, D). Biotite flakes near the contact are commonly oriented sub parallel to this contact or they lay just on it ( Fig. 6D). Upward in the profile, this subparallel orientation of biotite flakes prevails but is not universal (Fig. 6I).
Clusters of crandallite precipitate highly variable in form (Figs. 1-9; 6F, G) or homogeneous orange mass ( Fig.  6D) are the main constituent of every C# bed. Individual grains can only be distinguished by SEM (Figs. 1-9; 6F), as their dimensions vary from 3 to 15μm. A similar grain size is typical for idiomorphic crystals of carbonates, mostly siderite (Figs. 1-3; 6G; 7-9). Rare larger idiomorphic grains of carbonate were identified as dolomite by electron microprobe (Figs. 2; 6G; 7). Sparsely distributed flakes of biotite (partly sericitized) and irregular fragments of quartz and altered feldspar with a size of 20µm were identified. In the majority of samples, carbonate reaches up to 5vol.%. Only in the harder upper part of C1 from the Libouš Mine does its proportion exceed 50 vol.% (Fig. 6G), i.e. the bed contains more carbonate than phosphate. In this part of C1 bed, the dimensions of all mineral grains are approximately 1/3 or 1/2 compared to the underlying phosphate-rich part of C#. Syndepositional deformation convolute structures 1-2cm thick in the lower half of siderite rich part of this bed ( Fig. 6A) commented above represent a special feature of C1 of the Libouš Mine.
The upper boundary of C# beds is in all cases flat with a very sharp transition to overlying clay. Mineral composition varies at the scale of several mineral grains (Fig. 6J). Phosphate --or mixed phosphate--carbonate-bearing mass is upward replaced by clay with an admixture of carbonate and organic matter. Carbonate grains dispersed within clays are considerably coarser than the grains dispersed in the C# bed (Fig. 6J). Clay minerals in the lacustrine clays are not distinguishable optically. Irregular dark fragments of very fine organic matter but often subangular silt-sized grains of quartz were usually found together with clay minerals and carbonates. Clays in the overburden and bellow C# beds are intensively bioturbated (Planolites montanus according to is changed predominantly to mix of clay minerals, E) C1, CSA Mine, horizontal section through the sandy basal laminae, in the middle notice green grain of probably glauconite, white grains are represented by quartz, grainy gray, unidentified mineral grains (probably altered feldspar) completely transformed to the mix of clay minerals. F) C1, Libouš Mine, SEM detail of horizontal section of bottom sandy laminae. Bt: biotite, Sid: siderite (minute light gray grains), Q: quartz (angular gray grains), Cr: crandalite, grainy gray matrix, Py: pyrite framboids (white, circular shape), Ap: apatite, thin crystal penetrating plate of biotite, G) C1, Libouš Mine, SEM detail of the lower part of the bed, prevalence of gray matrix of crandallite (Cr) and isometric grains of siderite (Sid), rare rhomboedric crystal of dolomite (Dol). H) C1, CSA Mine, vertical section of bottom sandy laminae with orange largely crandallite matrix, white angular quartz grains (Q) and green-brown flat idiomorfic grains of biotite (Bt), Fsp: altered grains, probably former feldspar, Gl: glauconite, I) the same as H with analyzer (x Nicols), J) SEM detail of sharp upper contact (yellow line) of top sideritic C1 bed, Libouš Mine, with overlying silty clay. We can distinguish difference of both layers in granularity and content of light gray siderite. Top of crandallite bed has finer granularity and more siderite. Both features are changing in the profile during several grains. Mikuláš et al., 2006). It is very important that no traces of bioturbation were found in any of the C# beds. Commonly, in the case of Bílina C2 and C3, wedge-like structures transect the beds from the top to the bottom (Figs. 5C; 6B); however, no differences of mineral composition or structures are visible in these wedge areas. We interpret wedge-formed coloring as a product of weathering along diagenetic cracks, not an original sedimentary structure.
Microfossil-like structures were found only in two cases in the connection with C# beds. In both cases, they are represented by flat circular objects probably composed of amorphous SiO 2 (Fig. 6A, B), with their shapes resembling some circular diatoms. They were located in the 2mm thick zone of an overburden of C3 bed in Bílina Mine and the vicinity of C1A bed in its Libouš Mine occurrence.

Mineralogy by XRD
XRD mineralogy of bulk samples is summarized in Table 1. Mineralogically the C1 crandallite bed occurs in two distinct forms. In the central part of the Most Basin, it is present as a 1cm thick bed with prominent basal 1mm thick fine-sand lamina with prevailing detrital quartz, biotite, kaolinite and a crandallite admixtures. Crandallite is dominant within the main part of C1; the admixtures of quartz, biotite and clay minerals (kaolinite, illite) were also found. The C1 bed from the western part of the basin near the Libouš Mine is thicker than in the central part of the basin; with siderite more abundant than crandallite in the upper half of C# and crandallite more abundant than siderite in the soft, lower part of the bed. The main detectable admixture in siderite an crandallite is quartz. In the lower    Crandallite-rich beds of the Libkovice Member, Most Basin 14 third of C1, the diffraction lines of clay minerals (kaolinite, illite, especially illite-smectite) are more intensive than those of siderite and crandallite. The composition of the 5mm thick bottom sandy lamina is similar as in the central part of the basin (Bílina Mine, ČSA Mine areas). Biotite, quartz, and crandallite are dominant minerals, siderite and clay minerals are detectable, and trace amounts of pyrite were also found. The content of illite-smectite is higher and could indicate weathering of biotite or feldspar.
C2 bed is laterally very homogeneous throughout the whole area of its occurrence. Basal sandy laminae are distinct mainly in non-weathered outcrops of the Bílina Mine (Fig. 4G); their composition is dominated by biotite with significant admixtures of crandallite and quartz and a minor amount of kaolinite. A major component of the soft, main part of C2 bed is crandallite, minor admixtures are kaolinite and some amorphous matter, probably opal, minor amounts of biotite were also detected in the Bílina Mine. Because of intensive weathering of C2 on the Libouš Mine outcrops, the sandy lamina on the bottom was not macroscopically distinguishable and thus were not subjected to XRD analysis.
The youngest bed C3, preserved mainly in the deepest central part of the basin and little erosional remnant in western part, shows mineralogy very similar to C2. C3 consists mainly of crandallite with an admixture of biotite and traces of clay minerals.
The lenticular C1A bed in the Libouš Mine resembles C2 in composition, exclusively crandallite with some admixture of amorphous matter (probably opal). Other C# beds were not analyzed because of the insufficient amount of representative material.
Generally, XRD analyses confirm the above-described strong mineralogical difference between bottom sandy laminae with prevalence of detrital biotite and quartz and main bed body formed predominantly by phosphate or siderite only with some admixture of other minerals.

Chemical composition
The presence of crandallite-group phosphates is indicated by high contents of P 2 O 5 , CaO, Sr and Ba. Minimal P 2 O 5 content in C# samples slightly exceeds 6wt.%, maximum reaches 22wt.% ( Table 2). The content of phosphate mineral calculated from P 2 O 5 content is equivalent to entire content of SrO, BaO, CaO and the stoichiometric part of Al 2 O 3 . In such case, the content of crandallite in samples gives 18-69wt.% in samples within the C# bed. The content of crandallite is generally lowest in C1 in the Libouš Mine and basal parts of other C# occurrences. In the first case, it is due to the high content of siderite. In the other cases crandallite is diluted by clastic biotite, quartz and other minerals. On the contrary, the highest phosphate contents were detected in the soft parts of the C# beds above the basal sandy laminae. Samples of the harder, upper half of C1 from the Libouš Mine show a significant excess of FeO and MgO over CaO in the residue after crandallite subtraction from the entire composition. The high content of CO 2 in these samples corresponds with that fact, so it can be concluded that MgO and FeO are bound in carbonates, mainly siderite identified according to XRD. In the soft lower part, the P 2 O 5 content is lower than in other samples of C# beds what is compensated by higher SiO 2 and Al 2 O 3 content. This observation agrees with XRD results giving a prevalence of clay minerals over phosphates or carbonates. In C2 and C3 a slight excess of Ba over Sr was found. On the contrary, C1 shows more Sr than Ba. Excess of Al 2 O 3 over the crandallite content is FIGURE 8. Cr REE-REE patterns of crandallite beds normalized to: A) chondrite (Boynton, 1984) and B) North American Shale Composite (NASC: Haskin and Haskin, 1966). Samples: C3, represent two samples of phosphate bed divided to two ones, locality Bílina, A) upper half of profile, B) lower half of profile; C2, sample of bottom sandy laminae + sample of the whole phosphate precipitate part, both Bílina locality; C1, sample of bottom sandy laminae of Bílina locality + sample of Bílina (B) and Libouš (L) phosphate precipitate part of bed.
probably bound in micas, kaolinite, or other clay minerals. Higher content of SiO 2 in the main crandallite parts of some samples confirms probable content of opal, which could be the XRD amorphous phase. SiO 2 in the form of opal were often identified together with other minerals by electron microprobe (Table I; Figs I-IX) as inseparable ingrowths with crandallite clusters.
In some samples ca. 1wt.% S was found, however neither sulfate nor sulfidic phase was detected by XRD. This could mean the presence of SO 4 2+ in the crandallite (partial PO 4 3replacement by SO 4 2-; Dill, 2001). Alternatively, the sulfide micro-concretions visible by optic and SEM microscopy could have been present in abundances below the XRD detection limit.
Carbon not bound in carbonates (mainly organic) does not exceed 4wt.%. Fewer concentrations of non-carbonate C were detected in C1 bed of Libouš Mine where the content of carbonates is maximal.
REE content was determined in 7 samples of C# beds prevalently from the Bílina Mine area. One sample represents part of C1 of Libouš locality. Results are summarized in Table 3 and presented also in graphic form (Fig. 8). Data normalized to chondrite ( Fig. 8A; Boynton, 1984) and North American Shale Composite ( Fig. 8B; NASC: Haskin and Haskin, 1966) show a significant positive Eu-anomaly in C# samples from the Bílina Mine except for quartz-biotite laminae. Only the sample of C1 from the Libouš Mine shows a slight negative Eu-anomaly.

Strontium 87 Sr/ 86 Sr isotope systematics
All results of strontium isotopic analyses are presented in Table II and Figure 9. Strontium isotope ratios were analyzed in C1, C2 and C3 from the Bílina Mine. Basal sandy laminae and main crandallite parts of beds were separated and analyzed in the case of C1 and C2, whereas C3 was only split to lower and upper halves. The initial 87 Sr/ 86 Sr, recalculated to 17Ma, vary from 0.70974 to 0.71022ppm and display only subtle variations (Table II; Fig. 9). The same Sr isotope ratios were found in basal sandy layers and upper massive parts of C3. Negligible variations in 87 Sr/ 86 Sr isotope ratios, in addition, do not reflect the relatively wide variations in Sr contents (2300-15700ppm). Similar isotope ratios ( 87 Sr/ 86 Sr 17 = 0.70859-0.70992) were also obtained for phosphate concretions in the Holešice Member (the main coal seam) from the Libouš Mine.
On the other hand, strontium isotope ratios in the Most Basin clays display unusual variability ranging from 0.7122    o l o g i c a A c t a , 1 9 . 1 1 , 1 -2  For comparison, we also present isotopic and concentration data for Sr in the Oligocene to lower Miocene alkaline volcanic complexes of the České Středohoří Volcanic Complex (CSVC) Mts. and the Doupovské Hory Volcanic Complex Mts., were considered possible resources for crandallite beds and the Most Basin clays. A relatively narrow range in 87 Sr/ 86 Sr 17 ratios (0.70359-0.70495) of the Doupovské Hory volcanic complex encompasses primitive as well as FIGURE 9. Strontium isotope diagram of Miocene sediments and potential sources (data see Table I). Text abbreviations, CSVC: České Středohoří Volcanic Complex, DHVC: Doupovské.
highly fractionated rocks with highly variable Sr contents (520-3350ppm, Holub et al., 2010;Rapprich and Holub, 2008). Most of th<<e volcanic rocks within the CSVC isotopically resemble those of DHVC. ( 87 Sr/ 86 Sr 17 = 0.703155-0.70513; Sr= 70-1700ppm: Ackerman et al., 2015;Ulrych et al., 2002), but several strongly fractionated phonolites with Sr contents decreased to 11-133ppm revealed significantly modified ratios (up to 0.72581, Ackerman et al., 2015) reaching the values of the metamorphic basement. The phonolites with the most radiogenic signatures differ from gneisses in even lower Sr contents (11-13ppm). The phonolite domes and exposed laccoliths also represent relatively small bodies, not providing enough sedimentary material to significantly modify the geochemical signatures of sediments in the Most Basin.

O and C stable isotope composition of siderite
Because mineralogical analysis and optical and SEM microscopy detected a higher content of siderite mainly in C1 in the Libouš Mine, stable isotope C and O analyses were performed to estimate under what conditions carbonate precipitated in this bed. The C and O stable isotope ratios plot to a single line in the scatterplot for all analyzed samples (Fig. 10), with δ 13 C ranging from 1.1 to -3.2 and δ 18 O‰ from -4.5 to -6.0‰ vs. Pee Dee Belemnite (PDB) ( Table III).

Organic geochemistry
No substantial differences in the micropetrography and organic geochemistry in the clay sediment below and above C# beds was found by Havelcová et al. (2015). However, that previous study (Havelcová et al., 2015) did not include directly the C# beds. The organic compounds in C# beds (B samples) and in overlying (A samples) and underlying clays (C samples) are characterized in Table IV. The organic analysis showed features indicating the biogenic origin of organic compounds in sediment, but it also revealed significant differences in parameters documenting specific conditions during the C# bed formation. Parameters that are similar for all sample extracts include the Carbon Preference Index (CPI) or the distribution of n-alkanes. The CPI ranges from 2.4 to 4.0, which is typical for land-plant input and in the distribution of n-alkanes only a very slight predominance of shortchain n-alkanes (up to 20) in C # beds are documented. The parameter TAR (Terrigenous/Aquatic Ratio) is one of the results indicating a difference in the conditions of C# beds formation in comparison with overlying and underlying clays (Fig. 11A). The lower TAR values for extracts from C# beds suggests a reduced supply in the proportion of substances derived from terrestrial plants in relation to materials of aquatic ecosystems (Bourbonniere and Meyers, 1996). The Pr/Ph ratio of isoprenoid hydrocarbons of pristane (Pr) and phytane (Ph), which are  o l o g i c a A c t a , 1 9 . 1 1 , 1 -2  Crandallite-rich beds of the Libkovice Member, Most Basin 18 diagenetic products of phytol (chlorophyll) is reduced in the case of C# beds compared to the surrounding clays (Fig. 11B). This ratio characterizes the redox conditions in situ in processes of microbial degradation of organic matter (Didyk et al., 1978). Other ratios that differ for C# beds are Pr/n-C 17 (Fig. 11C) and Ph/n-C 18 (Fig.  11D). These ratios are based on the greater resistance of isoprenoid hydrocarbons to microbial degradation compared to n-alkanes and are mainly used to determine the maturity and biodegradation of the material (ten Haven et al., 1987). Using derivatized extracts, significant differences were found between C# beds and surrounding clays for fatty acids (Fig. 11E). Extracts from C# beds extracts contain a higher proportion of Fatty Acids (FA) with shorter carbon chains and used FA ratio (FA 20-28 / FA 12-18 ) documents the proportion of fatty acids derived from vegetable waxes (longer chains) and bacterial -algal origin (shorter chains) (Volkman et al., 1980).

Ar/ 39 Ar age and chemical composition of biotite
Biotite from C2 bed in the Bílina and Libouš mines produced ages of 17.37±0.04Ma and 17.63±0.02Ma, respectively. Both these ages are older than 17.07Ma, which was inferred for the C2 age obtained by the combination of paleomagnetic analysis and geochemical cyclostratigraphy for drill core MR93 and other long cores studied in detail (Matys Grygar et al., 2017, 2019b, 2021. Both C2 biotite samples have very similar compositions (Table 4). This supports the idea of their unique magmatic source.
The chemical composition of biotite from both localities in the Most Basin differs from the composition of the biotite sample collected from the so-called "greigite horizon" (Rojík, 2004) in the Sokolov Basin (western part of the Ohře Rift). C2 bed biotite in the Most Basin has higher contents of Ti, Mg and Fe, whereas the Sokolov Basin biotite is enriched in Al and K. The biotite from the Sokolov Basin provided Carboniferous age (322.7±0.2Ma) corresponding well to the magmatic age of the Nejdek-Eibenstock granite pluton ( 207 Pb/ 206 Pb age of 320±8Ma: Kempe et al., 2004).

DISCUSSION
We assign the C# beds within the lacustrine clays of the Libkovice Member to combination of the deposition of clastic components with organic matter and extensive formation of authigenic components. The clastic components were transported to the basin in the same manner as "normal" lacustrine clays below and above the C# beds. The authigenic compounds were precipitated near the bottom or under the surface of bottom of the lake from solutions, or by transformation of some extraordinary material. Our observations show that the clastic portion of the C# beds comprises biotite and quartz grains of sand-to silt grain size. Clay minerals identified by XRD can be detritic as in the entire Libkovice Member, or they could result from alteration of mineral grains like feldspars and micas. Carbonates, sulfides, amorphous silica and phosphates found in the C# beds were likely precipitated near the bottom of the lake.

Origin of biotite-quartz horizons in C# beds
Clastic material is found at the base of each C# bed. Novák et al. (1993) interpreted the biotite flakes abundant in the C# bases as a sign of their volcanic origin. Dating, however, showed that this biotite was probably formed before the C# deposition in the basin. The wide and flat plateaus in the Ar/Ar geochronology data of C2 biotite point towards a simple crystallization history without any recrystallization events. The obtained ages (17.37±0.04 and 17.63±0.02Ma) therefore roughly correspond to their crystallization. Accordingly, any older formations containing biotite and/ or quartz (all Proterozoic and Paleozoic metamorphic and magmatic formations of Bohemian Massif) must be excluded from further consideration. The studied time interval, however, post-dates the extinction of the two main volcanic complexes in the Ohře Rift (e.g. Ulrych et al., 2011) and includes the period when only minor scattered basaltic eruptions occurred more than 100 km east of the Most Basin (the Jičín Volcanic Field, Rapprich et al., 2007). Small volumes of dry basaltic magmas with low explosivity could not have produced so much biotite and distributed it over such a large area. Similarly, volcanic eruptions occurring in that time along the Rhine Graben in Germany were mostly effusive associated with rather small explosive eruptions (e.g. Lippolt, 1983). The large eruptions of alkaline magmas in French Massif Central, which could distribute pyroclastic material over the vast area are on the other hand at least by some 6My younger (e.g. Baubron and Demange, 1982). The lower Miocene was a period when volcanic activity in Carpatho-Pannonian region started (Pécskay et al., 2006). After the earliest intrusions around 23Ma, volcanic activity started there at 21Ma and lasted until sub-recent times. Besides numerous volcanoes with remarkable morphology in the Carpathian arc, there were other volcanic centers in the Pannonian Basin, now buried by sediments (Fig. 1C). Volcanic products and, to a limited extent, some volcanic edifices were partly revealed by drilling and geophysical surveys (e.g. Zelenka et al., 2004). Some of thevolcanoes produced widespread ignimbrites, outcropping on the margins of the Pannonian Basin. This is also the case of the so-called Middle Tuff Complex (MTC) exposed on the Bükk Foreland (Fig. 1C). This volcanic sequence consists of at least two, or maybe even three ignimbrite units, which erupted in the time interval 17.5-16.0Ma (Szakács et al., 1998). The source vent of these ignimbrites was inferred to be located east of Bogács and south of Harsány towns (580km ESE from the Most Basin) with the main direction of pyroclastic transport towards the west. In addition, a pyroclastic fall deposit correlated with the MTC was discovered on the boundary between Bohemian Massif and Carpathian Foredeep in northern Austria. This tuff ("Stranning ash", Fig. 1C) was dated using the Ar-Ar method on K-feldspar to 17.23±0.18Ma (Roetzel et al., 2014). The Austrian occurrence also contains abundant biotite, but K-feldspar was used for geochronology as a more suitable mineral phase. In the C# beds in Most Basin, K-feldspar was not found. As the biotite and K-feldspar differ in atmospheric transport, biotite from volcanic eruptions can be distributed over larger distances, while K-feldspar could not reach the Most Basin. Alternatively, the K-feldspar from MTC eruption could have also been deposited in the Most Basin but might have been completely altered (Fig. 6E).   o l o g i c a A c t a , 1 9 . 1 1 , 1 -2 , 2014, 2017), and there is also a mismatch of 40 Ar/ 39 Ar age of biotite in C2 with K-feldspar Ar-Ar geochronology of Straning tuff (17.23±0.18Ma, Roetzel et al., 2014). On the other hand, within the error given, the age of the Straning tuff matches well the cyclo-/magnetostratigraphy-based age of the C1 bed (17.31Ma, Matys Grygar et al., , 2017. The discrepancy between K-feldspar and biotite Ar-Ar ages was already noticed from many volcanic systems (e.g. Bachmann et al., 2007;Spell and Harrison, 1993). Systematic research comparing biotite and sanidine Ar-Ar geochronology under control of independent method ( 238 U-230 Th disequilibria) revealed a systematic shift of 200-500ky in biotite Ar-Ar ages even for very young (<50ka) volcanic products due to Ar partitioning and preeruption closure in biotite (Hora et al., 2010). This effect is highly pronounced namely in cases of young (Neogene) geological events, while in older samples this discrepancy is comparable to analytical uncertainty. Considering this systematic bias of the biotite Ar-Ar geochronology, the analytical ages of C2 beds are in broad agreement with other methods and confirm the possible volcanic origin of the biotite settled at the base of C# beds. Association of biotite with quartz and transformed feldspar is not in contradiction with silicic volcanic origin.

Origin of authigenic compounds
Because phosphorus has no specific indicator of its origin, we initially focused on possible fingerprints of other element constituents of the crandallite. In the Most Basin sediments, Sr 2+ and Ba 2+ ions for C# beds were incorporated from the pore-and lake water. In the case of Sr 2+ , this mechanism would explain the prominent decrease of Sr and P concentrations in the monotonous lacustrine clay layer covering the C# beds (Matys Grygar and Mach, 2013; this paper, Fig. 3). The Sr depletion above the C# beds was proposed to identify them, in particular when the beds themselves were destroyed mechanically during drilling or overlooked in sampling (Matys Grygar and Mach, 2013). The Sr depletion is not observed below the C# beds, which demonstrates the Sr was not brought to the C# beds by postdepositional diffusion from surrounding sediment.

Sr/ 86 Sr isotope ratio and C# provenance
The C# beds show minimal differences in Sr isotope ratio relative to a wide scatter of Sr isotope signatures in rocks relevant for the Most Basin ( Fig. 9; Table I). Samples of various clays of the Osek, Lom, Holešice, and Libkovice members were also analyzed to find the extent of variability in 87 Sr/ 86 Sr isotope ratios of the Most Formation sediments.
We used 87 Sr/ 86 Sr to find sources of other soluble matter inputs to the basin. The Sr isotope datasets used for comparison were obtained in the literature on the areas of the Most Basin catchment that were the main source of water and clastic material for the Most Basin fill during its lacustrine evolution stage . The 87 Sr/ 86 Sr 17 in crandallite differs significantly from common volcanic rocks (mainly alkali basaltic rocks) of the Ohře Rift (Ackerman et al., 2015;Holub et al., 2010;Rapprich and Holub, 2008;Ulrych et al., 2002). Only some fractionated phonolites show higher 87 Sr/ 86 Sr 17 . These rocks are very poor in Sr and form small rock bodies that could not have provided enough material to change the Sr isotopic ratios of the Most Basin sediments. All these facts make it possible to exclude the Ohře Rift volcanic rocks from possible sources of authigenic components of crandallite beds. This comports with the lack of high contents of typical mafic "volcanogenic" elements such as Ti, Ni, Cr, and Cu in any parts of the C# beds when compared with the surrounding clays. On the other hand, the isotopic characteristics of the crandallite beds are very close to that of carbonates, and also silicic residua after leaching carbonate from the marine sediments of the Bohemian Cretaceous Basin (0.707353-0.708271, Nádaskay et al., 2019 and newly obtained data). The Sr isotopic ratios of the Cretaceous deposits partly overlap with the presentday waters of European rivers (the Donau River= 0.70886, Veizer, 1993; the Odra River= 0.7010-0.7108, Zieliński et al.,, 2017). The isotopic signature of C# beds is also close to that of Miocene (17Ma) seawater 87 Sr/ 86 Sr (McArthur et al., 2001). The isotopic signature and namely amounts of Sr available from the Cretaceous sediments could suggest that most of the Sr in C# originated from chemical weathering of the Cretaceous rock outcrops. We expect that carbonate shells (with high amounts of Sr, Nádaskay et al., 2019) in outcrops of marlstones and claystones were weathered and leached preferentially. The shift in Sr isotopic signatures of the crandallite beds from that typical of Cretaceous rocks towards more radiogenic values in the clays can be explained in terms of admixture of more radiogenic strontium. As all the Rb-rich rocks (Rb produces radiogenic Sr signature) are poor in Sr, which could be mobilized during weathering, rather very small amounts of highly radiogenic strontium signature seem to be a good candidate to represent additional strontium derived from the Cretaceous rocks. In this aspect, granites and rhyolites of the Altenberg-Teplice Volcanic Complex with strongly elevated 87 Sr/ 86 Sr values (except for their early stages with less radiogenic signatures, Walther et al., 2016) would have potential to modify the 87 Sr/ 86 Sr ratio in the Most Basin paleolake.
In contrast to the C# beds, the trend of clastic sediments of the Main Coal Seam in 87 Sr/ 86 Sr 17 vs Sr (Fig. 9) follows that of metamorphic basement and granitic rocks of carboniferous age occurring in potential watershed. It can be explained by a difference in conditions in the lake during the formation of C# beds and deposition of clastic sediments. The crandallite beds exclusively incorporated Sr dissolved in lacustrine water (mostly originating from the Cretaceous rocks), while the monotonous lacustrine clays incorporated more detritic grains from non-carbonate rocks. The negative correlation between the Sr contents and 87 Sr/ 86 Sr 17 ratios (Fig. 9) thus suggests the mixing of two contrasting sources. A source poor in Sr, but with highly radiogenic signature (gneisses and namely granites or rhyolites), dominates over the source of less radiogenic strontium (the Cretaceous sediments and possibly also Ohře Rift volcanic rocks). The higher Sr contents in monotonous lacustrine clays is accompanied by a higher P, Ca, and Ba contents and so we can deduce that this extra Sr has the same form as Sr in C# beds, crandallite, with Sr originating from the lake water. Within clays containing less than 70ppm Sr, the clastic signal prevails; this comes from weathered Paleozoic acid magmatic rocks and metamorphic rocks of SW part of the Bohemian Massif, the main source of clastic sediment components in the Most Basin fill. The isotopic composition of this Sr is also similar to that of Proterozoic gneisses of the Krušné Hory Mts. (Table I).
Sr isotopes help explain mixing of inorganic sources of basin inputs. The other question is how and in which conditions authigenic minerals precipitated; methods including REE spectra, stable C and O isotopes in carbonate, and organic geochemistry analyzing provide insight.

Conditions of crandallite precipitation
Generally, crandallite is a mineral, which requires quite specific conditions to be formed (Dill, 2001). Relatively minor differences in solution acidity around a pH value of 7 controls whether apatite (in alkaline medium) or crandallite (in acid medium) is precipitated. With higher HPO 4 2activities, acidic environment (to pH= 5) are needed for the crandallite precipitation. The crandallite physical-chemical stability was examined especially in weathering crusts including phosphate deposits. Unlike more ubiquitous Caphosphates of the apatite group, which are stable in neutral or slightly acidic conditions with a sufficiently high PO 4 3anion activity, crandallite is stable in more acidic conditions (pH between 6.5 and 5.5). Besides sufficient availability of phosphate ions, the presence of Al 3+ or Al(OH 2 )+ is necessary for crandallite crystallization (Huang and Keller, 1972;Vieillard et al., 1979). Crandallite crystals scavenge Sr 2+ and Ba 2+ and other larger cations from surrounding solutions, substituting them for Ca 2+ cations in its crystal lattice (May et al., 1963). Thus, crandallite commonly occurs as a solid solution with Sr-or Ba-rich isostructural minerals goyazite and gorceixite, respectively. REE concentrations normalized to chondrite ( Fig. 8A; Boynton, 1984) and North American Shale Composite ( Fig.  8B; NASC, Haskin and Haskin, 1966) show significant positive Eu anomalies, attributable to special chemical properties of Eu within the lanthanides group. Europium usually existing in poorly water-soluble Eu 3+ form can be under certain conditions reduced to more soluble Eu 3+ . The specific conditions of crandallite precipitation (reduction by organic matter, a narrow interval of acidity, high phosphate activity) enabled its extraction from detritic minerals like monazite. Subsequently, its reduced form (Eu 2+ ) became available for incorporation into crandallite, analogous to Sr 2+ or Ba 2+ . This presumption is in good agreement with observations from organic geochemistry, which provided evidence of high bacteria/algae activity at the lake bottom during crandallite precipitation. Disparate conditions for C1 formation in the Libouš Mine area are reflected in differences in mineralogical composition (high content of detritic clay minerals and authochthonous siderite) and bed thickness. Europium is thus a sensitive indicator of crandallite precipitation conditions across the basin. The Ce-anomaly (defined as log [Ce/Ce*], where Ce/Ce*= 3×Ce N /(2×La N +Nd N ); normalized to the North American Shale Composite (NASC) of Haskin and Haskin (1966) was used by Elderfield and Pagett (1986) to distinguish between oxic and anoxic conditions during phosphorite sedimentation, with boundary set at zero. Wright et al. (1987) then modified the redox boundaries to anoxic/ transitional for Ce anomaly of 0 and transitional/oxic for Ce anomaly of -0.1. Ce-anomaly values for C1 precipitates for Bílina and Libouš Mine at 0.06 and 0.09, respectively suggest anoxic conditions at the lake bottom during the C1 bed sedimentation, whereas values of this anomaly in the range -0.01--0.03 indicate transitional oxic/anoxic conditions for sedimentation of remaining C# beds.
C and O stable isotopic ratios in the C1 contained in siderite can be compared with data obtained in earlier studies (Mach et al., 2001, R. Lojka, unpublished data of carbonates dispersed in clays of the Most Formation above the main coal seam, data overview in Figure 10). In the δ 13 C/δ 18 O graph, the newly-studied C# plot along the border of the main field of diagenetic carbonates. Graph subarea of C1 samples is extreme by lower values of both δ 13 C and δ 18 O (Fig. 10). It shows slightly different conditions of its formation, than in the case of dispersed carbonates. Interestingly the siderite from C1 in the Libouš Mine differs from carbonates of the Libkovice Member clays from both central and Bílina parts of the basin and especially from the samples from the drill core SP257 (black points in Fig.  10) situated near the C1 sampling site (Fig. 1). Isotopic compositions of clays from the Holešice Member in the central part of the basin (yellow rectangles in Fig. 10) are nearer to siderite in C1. Nevertheless, C1 siderites were formed by the same manner as siderites in lacustrine silty clays of the Libkovice or Holešice members, i.e. early after sedimentation. As preceding studies assumed (Mach et al., 2001), the isotopic, mineralogical, and chemical signatures G e o l o g i c a A c t a , 1 9 . 1 1 , 1 -2 9 , I -X ( 2 0 2 1 )  D O I : 1 0 . 1 3 4 4 / G e o l o g i c a A c t a 2 0 2 1 . 1 9 . 1 1 Crandallite-rich beds of the Libkovice Member, Most Basin 22 of carbonates formed in more extreme conditions are more different. Siderite of the C1 bed thus precipitated during early diagenesis in slightly higher temperature, different acidity, and more reduction conditions than carbonates dispersed in the surrounding monotonous lacustrine clays, any case not in the zone of prevailing methanogenesis or sulfate reduction. The high content of siderite in the upper part of C1 (up to 50 %) shows, that phosphate-containing sediment was very porous in the time of carbonate crystalization, also in the agreement with high porosity calculated from C# bulk density of other samples. The very high porosity of the C1, C2 and C3 crandallite rock reaching 66% of sample volume documents presence not only of water in pores but also some other material existing in pores before and during crandallite precipitation; the material subsequently diluted or consumed diagenetically. The most probable explanation of this material would be organic matter later consumed by bacteria, finally removed in the form of H 2 O, CO 2 and C4 as products of their metabolism.
GC/MS analyses of the extractable organic compounds indicate significant differences in C# beds in comparison with the surrounding sediments and demonstrate different conditions of their formation. The reduced TAR values (Fig.  11A) indicate a higher algal presence and bacterial activity compared to the surrounding clays. Decreased Pr/Ph values in C# beds (Fig. 11B) can be interpreted as evidence of a significantly reducing environment during phosphate bed formation. Similarly, Pr/n-C 17 and Ph/n-C 18 ratios based on higher resistance of Pr and Ph compared to n-alkanes (Fig.  10C, D) show that C# beds, having higher ratios, endured more significant bacterial degradation of organic matter than in the surrounding clays. Finally, the higher proportion of shorter-chain fatty acids (Fig. 11E) documents, similarly to TAR, increased presence of bacterial-algal mass entering the set of organic compounds in C# beds.
Stable C and O isotopes together with organic geochemistry and REE patterns define the conditions of crandallite and carbonate precipitation. These minerals precipitated in anoxic to transitional anoxic/oxic conditions surrounded by organic matter decomposing by vigorous bacterial activity. Organic matter was prevalently of bacterial or algal origin.

Prerequisites for C# bed formation: phosphate accumulation and environmental disruptions
We principally envisage the origin of crandallite in three ways: 1. crandallite grains were detritic, transported to the lake by rivers, 2. crandallite precipitated in the water column and then deposited on the bottom of the lake, 3. crandallite is a product of transformation of some other source of phosphate within the sediment.
The first hypothesis is unsupported, no such significant source of crandallite is known within potential source areas. It is also questionable whether crandallite could be transported for such long distances without dissolution in water and dilution by detritus. If there were to be some source, e.g. older kaolinic weathering crust, the crandallite would be diluted by a larger amount of clay minerals, which was not the case in C# beds.
The second possibility would require an abundant source of diluted phosphate in lacustrine water together with high concentrations of Al 3+ and Ca 2+ cations and special acidic conditions (Vieillard et al., 1979) in the water column. Under usual lacustrine conditions phosphate is primarily a part of its biogenic cycle bound in living organisms, which tend to recycle this nutrient as soon as possible after death of organism and their decomposition. Persistent combination of high activities of all needed chemical components and acidic conditions in the column lacustrine water is not probable.
The third idea is most probable for the lake in the Most Basin (Fig. 12). We propose that dead organisms (especially algae and bacteria according to our results) accumulate on the lake bottom, where anoxic bacterial decomposition under acidic conditions make crandallite precipitation possible. Low pH produces Al 3+ from clay minerals from suspension fluvially transported to the lake. Acidic decomposition of clay minerals would also produce amorphous SiO 2 , indeed found in C# beds by XRD and chemical analysis. Apatite containing skeletons from higher organisms could contribute phosphate beside phosphate from soft tissues of all organisms. The lack of the original bone structure in the crandallite bed could be explained by recrystallization of the original biogenic apatite similarly as reported by Dericquebourg et al. (2019). Geochemical data obtained from above described analysis like organic geochemistry, REE geochemistry and C, O stable isotope geochemistry document intensive decomposition of organic matter under acidic anoxic conditions. An enigmatic aspect is the fixation of such a large amount of P from biogenic cycle in the Most Basin. How could such a large amount of a "phosphate concentrate" be so regularly deposited on the bottom of the basin-wide lake? The limiting factor for any of the hypothetic processes responsible for crandallite bed formation would be the quantity of the biogenic material needed. To create a 2cm thick C2 bed containing 9% of phosphorus in the Most Basin, ca. 2kg per 1m 2 of phosphorus must have been deposited on the lake bottom (using dry density around 1g/cm 3 obtained for non-sideritized beds). According to Whittaker and Likens in Lieth and Whittaker (1975), normal lacustrine primary production of organic matter is 0.1-1.5kg of dry mass (or 0.02-0.95kg of C) per m 2 and year (mean 0.4kg of dry mass per m 2 and year), depending on the distance from the tributary river mouths. Depending on the form of biomass, the P content can vary from 0.5 to 5%. The measured state of biomass content was 0.0 to 0.1kg per m 2 after Whittaker and Likens in Lieth and Whittaker (1975). The stoichiometric content of P in freshwater fish bodies is usually in the range of 1.3-5.7%, averaging 2.9% in dry mass (McIntyre and Flecker, 2010). Similar data for invertebrates (Evans-White et al., 2005) gives phosphorous content in the interval of 0.3 to 1% in dry mass. Data for 5 species of freshwater zooplankton from the Kjelsiisputten Lake, Norway given by Hessen and Andersen (1990) are in the interval of 0.6-1.4% of P in dry biomass. Taking a very rough average primary productivity of dry biomass as 0.4kg per year (Whittaker and Likens in Lieth and Whittaker, 1975) and content of P in this mass to 1.5%, we obtain 0.4 × 0.015= 0.0060kg/ m 2 . Similarly, the actually measured state of biomass around 0.05kg/m 2 gives 0.05 × 0.015= 0.00075kg/m 2 of P. Another direct view to the potential P sources in lacustrine water is provided by an aggregate of P content in particles floating in lacustrine water. Hecky et al. (1993) published dataset on P content in seston for 51 recent lakes of various climatic regimes. According to this dataset, the P content in particles (including bioseston phyto and zooplankton, nekton etc.) is in the range 0.09 to 1.81μmol·l-1 with a median value around 0.2μmol/l= 0.0002mol/m 3 . Accordingly, in a 10m column of lacustrine water 0.0002*30.97*10= 0.062g/m 2 of P could potentially be in particle source, which can at any one moment be deposited onto the lake bottom.
Roughly estimated values of potential P sources in lacustrine water calculated by various methods (above) scatter across a few orders of magnitude and reach a maximum of a few grams of P per m 2 and year. In any case, they show that it was not possible to form C2 bed by a single-event extinction of all lacustrine organisms. A longer deposition of P would thus be needed under normal lacustrine productivity, in particular lasting for a thousand years in the Most Basin C# beds. The usual clastic sediment deposition rate of the Libkovice Member clay was in average 17-20cm/ky ) thus a several-ky-long interval for the formation of C# would represent a disruption on the order of tens cm to several meters of the sediment condensed to a few cm, something that could not be revealed by the integrated magneto-/cyclostratigraphy age model for the entire 200m of clays of the Libkovice Member (Figs. 2; 3).
Considering all facts and assumptions above, the C# beds can be interpreted as the sediment-starved intervals, with the usual clastic input temporarily suppressed and lacustrine deposits formed mainly by biogenic/ autochthonous chemogenic components (Fig. 12). One possible condition enabling a radical decrease of clastic input to the basin would be unusual seasonal stability of climatic conditions in the catchment area. If the seasons have differed minimally, flooding conditions crucial for clay suspension forming in watershed and transport to the lake could have been absent. Such a climatic regime could be imagined in periods of the weakest seasonal contrasts, i.e. either during minimal eccentricity of the Earth orbit and minimal obliquity of the Earth rotational axis, or maximal eccentricity with the Earth in perihelion in winter and aphelion in summer according to the actual precession (cycle ca. 20ky). Especially occurrences of C1 and C2 together with C' and C'' fall very well (Fig. 3) to the latter pattern periods (Matys Grygar et al., 2017). The coincidence of eccentricity maximum and C# bed is apparent in the case of C1, C1a and C3. The precise timing of C2 is within one precession cycle near an eccentricity maximum. It is necessary to mention here crandallite dispersed in the lacustrine sediments also in Sr maxima outside C# beds documents the chemical conditions were favorable for crandallite formation in the entire lacustrine deposition.
At first glance, it is hard to judge whether these stable climate periods were humid or dry. Stable humid climate could result in very dense plant cover of the catchment with minimal physical erosion and intense pedogenetic hydrolysis of minerals rather than their transformation to clay minerals, lack of slope erosion, vegetation covering rock outcrops, and stable vegetated riverbanks limiting floodplain reworking. Eutrophication of the lake in pluvial with development of sub-oxic or anoxic conditions in the lake bottom would prevent P recycling by aquatic organisms and favor bacterial metabolism in the lake bottom as it was proposed by Dericquebourg et al. (2015) for similar phosphate accumulation in African lacustrine sediments.
On the other hand, stable dry periods could produce very limited detritic input, transport by aeolian rather than fluvial processes. Both such conditions would considerably decrease clastic input. Either huge explosive volcanic events, or dry climatic events with an associated aeolian process could produce the regular, continuous, thin sandy laminae without any ripples or deformational balls and pillows structures such as those present at the base of C#. Results of organic geochemistry analyzes support the "dry periods" hypothesis by higher content of organic matter having its origin in autochthonous (bacterial and algal) input in relation to allochthonous sources in the basin watershed. Lower content of terrigenous organic matter is associated with lower input not only of clastic sediment components, but also of water generally.
We cannot exclude other possibilities explaining the close relations of C# beds occurrences and general G e o l o g i c a A c t a , 1 9 . 1 1 , 1 -2  Crandallite-rich beds of the Libkovice Member, Most Basin 24 geochemical characteristics of K/Al, Al/Si, Ti/Al, Mg/Al or cation exchange capacity. Changes of some geochemical parameters --proxies of mineralogical composition of clay sediment during that time--can be explained in other ways. One such idea involves autocyclic processes working in the lake especially the changing of deltas and depocenters position within basin. Evidence of this can be found in local lakes and of Holešice Member deltas, especially case of Bílina Delta (Dvořák and Mach, 1999;Rajchl and Uličný, 1997). Conditions in local lakes surrounded by peat swamp is rather different from the whole basinwide lake represented by sediments of Libkovice Member excluding the topmost part of Holešice Member, 10-20m thick Břešťany Clays. Břešťany clays as a widespread layer represent the stage of expansion of the local lakes, their interconnection and creation of one basin-wide lake covering the entire Most Basin. The two topmost delta fans connected to these basin-wide clay layers were documented in southern part of Bílina Mine (Dvořák and Mach, 1999). Bottom part of Břešťany Clays is split by wedges of these delta bodies. Břešťany clays differ from Libkovice Member clay only by absence of illite-smectite in its composition on major part of the basin. Some portion of smectite structures occur only in the central part of the Most Basin between ČSA Mine and Most, where it is present in lower half profile of this layer. It was detected by higher values of cation changing capacity (Fig. 2). This Illite-smectite area is located approximately 15km W of the abovementioned delta bodies in the area of Bílina Mine. This situation gives rise to the possibility that at the scale of the Most Basin some lateral changes of lacustrine sediment composition caused by different rates of various clay minerals sedimentation depending on the distance from the deltas. Consequently, changing clay composition within the profile could be the result of changing delta position. Delta position changes were identified as being caused by changes in water level in the lake (Rajchl and Uličný, 1997). On the other hand, we found that content of illite-smectite in the Libkovice Member clays well documented by cation changing capacity is not dependent on the Illite, kaolinite and silt of quartz content in the clay displayed by K/Al (Mg/Al) or Al/Si relations (Matys Grygar et al., 2019a). This fact was then confirmed by methods of lithological correction of previous results (Matys Grygar et al., 2020). Stated simply, the suspended fractionation within the lake is only one but not the main reason of clay mineral composition variations in the profile. Because occurrences of C# beds are specifically related to K/Al stratigraphy led by Milankovich cycles of orbital Earth parameters (Matys Grygar et al., 2017) and not to other parametres closed to physical properties of sediment we can exclude some autocyclic reason of its formation as the main reason.
Close relation of C# beds cooccurrence with K, minima is not sole fact supporting the idea of climatically driven phosphate accumulation. Likewise, it can be supported by the co-occurrence of several meter thick layers of clay with several times higher content of phosphates (indicated as Sr anomalies) and layers with increasing of Al/Si ratio of sediment (Fig. 3). The Al/Si variation in the class of the Libkovice Member represents another form of climatically driven cyclicity similar to the K/Al ratio. Higher Al/Si in K/ Al minima indicates higher contents of kaolinite that can be interpreted as higher intensity of chemical weathering (deeper hydrolysis of clay minerals) in the watershed.
Noteworthy from an environmental point of view are also the chemical pre-requisites for the crandallite crystallization. It needs a pH of 5.5 to 6.6, lower than apatite depending on the phosphate ion activity (Dill, 2001). Such a slightly acid solution would prevent the precipitation of calcite and common Ca-phosphates. Global acidification was apparently among the general geochemical aspects of global environmental crises related to volcanism (Borruel-Abadía et al., 2019). If such a crisis was behind C# formation, it could also explain the basin-wide biotic crisis needed to remove a sufficient amount of P from biogenic recycling. This crisis alone cannot explain the persistent deficit of clastic input to the lake. Occurrence of invertebrate bioturbation in the overlying and underlying clays together with the lack of bioturbation in the C# beds could be the result of reducing/acidic conditions during the formation of the C# beds. Reducing conditions are documented also by Pr/Ph ratios in C# beds and surrounding clays.
A very interesting aspect of all C# beds is the sudden re-establishment of the clay sedimentation after the crandallite-formation interval (Fig. 6B, C, J). This can be connected only with a rapid change in the watershed conditions, maybe in combination with rapid progradation of the river deltas to the lake, with hyperpycnal suspension of clay brought again to the bottom of the lake, all under renewed oxidizing conditions documented by Planolites montanus occurrence. Such scenario would evidence extraordinarily rapid climate changes in the lower Miocene history.
Convolute bedding of Kelvin-Helmholtz (Fernando, 1991) and/or Rayleigh-Taylor type instabilities (Gladstone et al., 2018) described in the middle of C1 (Fig. 4F) could be interpreted as evidence of seismic activity during formation of the bed. The density contrast between the upper part of C1 bed (denser due to containing siderite filling pores) and the lower half of it (very porous, containing prevalent phosphate + clay minerals) led to a probable reaction to shear stress produced by seismic waves (Heifetz et al., 2005) generating typical billows (Fig. 4F).
Another subject of discussion can be formation of bottom sandy lamina of C# beds. According to above-offered idea of formation of this lamina by falling its particles from the air to the water surface (by volcanic or aeolian event) and after that to the bottom of the lake. This process covers a wide area by very regular layers of sandy to silty material in a short time interval. We excluded the competing hypothesis of water because, in our opinion, the Holešice Member study (Dvořák and Mach, 1999;Rajchl and Uličný, 1997) showed that thin laminae of fine sand could be formed in lacustrine conditions only after transport of the sand by dense clay suspension. Such suspension sediments within the prodelta area in the form similar to turbidites, prodeltaic heteroliths. If the sand is not a part of suspension, it can be usually transported only by bottom transport (moving ripples) and in that case, it finishes its course just near the mouth of tributary on the delta front where the rate of water stream decreases suddenly. According to our observations, prodeltaic heteroliths containing thin sandy laminae produced by sedimentation from hyperpycnal suspension did not reach more than several hundred meters from delta front covering areas maximum of couple square kilometers often characterized by syndepositional deformations near the delta (balls and pillows or micro-ripples). Biotite containing laminae on the C# beds cover hundreds square kilometers and have no signs of horizontal bottom transport like ripples. A typical sign of sandy laminae which are a part of prodeltaic heteroliths (turbidites) is their coupling with overburden clay bed formed in the second part of the process of turbidite flow sedimentation. Sandy lamina can´t occur without normally graded clay bed in the overburden and usually, the clay portion is much thicker than the sand one. The biotite-containing sandy laminae have no clay cover in the case of C# beds, it is covered by a phosphate G e o l o g i c a A c t a , 1 9 . 1 1 , 1 -2  layer. Another argument supporting the hypothesis of ash fall or aeolian origin of bottom sandy laminae is the relatively low degree of biotite weathering and usual form of its individua. They are very frequently represented by well-preserved crystals without any signs of destruction by transport or pre-sedimentary weathering. This can be hardly imagined on biotite that experienced protracted weathering and fluvial transport. Biotite was exclusively identified on the bottom of C# beds within the whole profile of the Most Formation although it is common in rocks forming the Most Basin watershed. This means that any other biotite of the Most Basin watershed was completely transformed by weathering and destroyed by transport to the basin. We can therefore exclude water transport of biotite, quartz sandy material to the whole basin area.

Role of volcanism in C# formation
The biotite in the basal strata of crandallite beds was of volcanic origin, suggesting some large volcanic eruptions coincided with the onset of crandallite precipitation (or accumulation of P-rich crandallite precursor). On the other hand, the climatic extremes lasting over several thousands of years could hardly be driven by a volcanic eruption. Similarly, the largest ever known volcanic eruption of Toba (74ka) preceded glaciation, but the volcanic winter resulting from the eruption only accelerated the advent of the glacial stage (Rampino and Self, 1992). Volcanism is a good hypothetic trigger of a sudden change of conditions in the catchment and the lake, in particular in the period of the Earth evolution approaching the onset of the Middle Miocene Climatic Optimum (MMCO), as this period was close to a global climate change. On the other hand, we cannot exclude the possibility of post-volcanic aeolian transport of volcanic products to the Most Basin by extraordinarily strong aeolian processes. This possibility can be supported by the content of probable glauconite grains within the basal sand mineral association.

CONCLUSIONS
Contrary to preceding works, we conclude that Sr-and Ba-bearing Ca-Al phosphate beds in lacustrine clays of the Libkovice Member of the Most Basin resulted from climatic disruptions in periods of large orbital eccentricity and coincidence with other orbital factors, in particular a specific phase of precession cycles, by an unusually stable precipitation regime, i.e. a very low seasonality. C# beds were formed in a sediment-starved interval under persistent dry or persistent humid climate. The climatic trigger for formation of the C# beds in the Most Basin is supported by their association with K concentration minima driven by orbitally-controlled sedimentary cyclicity. The precipitation of the phosphate from biogenic precursors in contact with lacustrine water is proven by Sr isotopic signatures and depletion of the lacustrine clays above the C# beds in Sr, Ba, and P. The presence of biotite crystals of volcanic origin, mostly concentrated in the bed base remains enigmatic, but can be explained by two hypotheses: aeolian transport from rock outcrops (earlier volcanic ash fall) of the basin vicinity to the basin floor (under dry climate), or direct volcanic influence (under pluvial climate). Volcanic events could trigger several kiloyears-lasting environmental changes in the Libkovice paleolake body or watershed. The available data do not provide unequivocal evidence for preferring either of the two above proposed hypotheses for crandallite bed conditions: pluvial climate, or drought. The position of the C# beds at the base of K minima would point to generally humid climate and bottom anoxia caused by lake eutrophication, and in this case the biotite-quartz sand on the bottom of the bed would require a coincidence with volcanogenic fallout. Both cases, however, represent extremes of the precipitation regime in the Most Basin catchment and could point to yet unidentified environmental disruptions around the time of the MMCO onset.  l o g i c a A c t a , 1 9 . 1 1 , 1 -2 l o g i c a A c t a , 1 9 . 1 1 , 1 -2 o l o g i c a A c t a , 1 9 . 1 1 , 1 -2 l o g i c a A c t a , 1 9 . 1 1 , 1 -2 l o g i c a A c t a , 1 9 . 1 1 , 1 -2