Global and Clumped isotope record of salinity variations in the Subboreal Province at the Middle – Late Jurassic transition

Results of clumped isotope, oxygen isotope and elemental (Mg/Ca, Sr/Ca) analyses of exceptionally well-pre- served belemnite rostra and ammonite shells from the uppermost Callovian – Upper Kimmeridgian (Lamberti – Mutabilis zones) of the Russian Platform are presented. Despite a signi ﬁ cant decrease in belemnite δ 18 O values across the Upper Oxfordian – Lower Kimmeridgian, the clumped isotope data show a constant seawater temperature (ca. 16°C) in the studied interval. The decrease in belemnite δ 18 O values and lower δ 18 O values measured from ammonite shells are interpreted as a result of the salinity decline of the Middle Russian Sea of ca. 12 ‰ , and salinity strati ﬁ cation of the water column, respectively. The postulated secular pa- laeoenvironmental changes are linked to the in ﬂ ow of subtropical, saline waters from the Tethys Ocean during a sea-level highstand at the Middle – Late Jurassic transition, and progressive isolation and freshening of the Middle Russian Sea during the Late Oxfordian – Kimmeridgian. The obtained clumped isotope data demonstrate relative stability of the Late Jurassic climate and a para- mount e ﬀ ect of local palaeoceanographic conditions on carbonate δ 18 O record of shallow epeiric seas belonging to the Subboreal Province. Variations in Mg/Ca and Sr/Ca ratios of cylindroteuthid belemnite rostra, which are regarded by some authors as temperature proxies, are, in turn, interpreted to be primarily dependent on global changes in seawater chemistry. The paleoenvironmental variations deduced from clumped and oxygen isotope records of the Russian Platform correspond well with changes in local cephalopod and microfossil faunas, which show increasing provincialism during the Late Oxfordian and the Early Kimmeridgian. Based on the review of literature data it is suggested that the observed salinity decrease and restriction of Subboreal basins during the Late Jurassic played a major role in the formation of periodic bottom water anoxia and sedimentation of organic rich facies.

The clumped isotope composition of well-preserved calcium carbonate minerals is an independent proxy for precipitation temperatures. Determination of both the clumped and the oxygen isotopic compositions of carbonates allows reconstruction of ancient δ 18 O water values (Ghosh et al., 2006;Eiler, 2011). Clumped isotope analyses have been successfully used for the reconstruction of water temperatures during the growth of carbonate fossils (cf. Finnegan et al., 2011;Dennis et al., 2013;Price and Passey, 2013;Tobin et al., 2014;Petersen et al., 2016).
The aim of the study is to determine real variations of water temperature of the Middle Russian Sea belonging to the Subboreal bioprovince during and after the Middle-Late Jurassic transition (the latest Callovian-earliest Late Kimmeridgian). For this exceptionally wellpreserved belemnite rostra and ammonite shells from the Volga Basin in the European part of Russia were analysed. The comparison between carbonate Δ 47 and δ 18 O values allow determination of ancient water δ 18 O values, which may be linked to salinity fluctuations of the restricted Middle Russian Sea. The isotope data and their interpretation are compared with the distribution of macro-and microfossils in the Russian Platform to verify palaeoenvironmental reconstructions. The present study sheds new light on the latest Middle Jurassic-Late Jurassic climate and effects of local palaeobathymetry and palaeocirculation on climate reconstructions based on δ 18 O values of marine fossils derived from epicontinental or marginal sea basins.

Material
Well-preserved and stratigraphically well-dated 19 archival and 5 newly collected belemnite rostra and ammonite shells have been analysed for their clumped isotope composition in order to obtain a continuous record of temperature variations of the epicontinental Middle Russian Sea during and after the Middle-Late Jurassic transition (the latest Callovian-mid-Kimmeridgian; see Table 1). The archival uppermost Callovian-Lower Kimmeridgian materials (cylindroteuthid and mesohibolitid belemnite rostra and ammonite shells) derived from the Dubki section near Saratov, and the Makar'ev and Mikhalenino sections in the Kostroma Region of Russia ( Fig. 1) comprise samples studied by Wierzbowski and Rogov (2011), and Wierzbowski et al. (2013). The newly collected cylindroteuthid rostra, studied for clumped isotopes, are derived from Tarkhanovskaya Pristan' section near Ul'yanovsk in the Tatarstan Republic of the Russian Federation ( Fig. 1). They are dated to the latest Early Kimmeridgian-earliest Late Kimmeridgian, i.e., the late Cymodoce Zone (=Divisum Zone) and the Mutabilis Zone (cf. Rogov et al., 2017, and Table 1). Some of these samples have been studied recently for strontium isotope composition . In addition, results of new oxygen and carbon isotope analyses of 19 Kimmeridgian cylindroteuthid belemnite rostra and 175 published values of carbonate fossils (cylindroteuthid and mesohibolitid belemnite rostra, ammonite and gastropod shell; see Wierzbowski and Rogov, 2011;Wierzbowski et al., 2013) are presented.
Collected carbonate fossils are precisely biostratigraphically dated (see Wierzbowski and Rogov, 2011;Wierzbowski et al., 2013; Table 1). The standard Boreal and the Subboreal ammonite zonal schemes are employed for dating of uppermost Callovian-Oxfordian and Kimmeridgian sediments, respectively, according to the regional biostratigraphical framework (Fig. 2). It is worth noting that occurrences of Submediterranean ammonites allow the use of Submediterranean biostratigraphical units in some parts of the studied sections (Głowniak et al., 2010;Rogov et al., 2017). Precise correlation between the (Sub) Boreal ammonite zonal scheme of the Oxfordian-lowermost Upper Kimmeridgian of the Russian Platform and the Submediterranean province of central Europe zonation is presented on Fig. 2. The employed biostratigraphical scale is matched to the assumed equal duration of Submediterranean ammonites subchrons, which are counted successively starting from the base of the studied interval (cf. Wierzbowski et al., 2013). This approach is used to enable direct comparison between the recorded isotope variations and the previously published isotope data from Western and Central Europe.

Preparation
Thin sections prepared from newly collected belemnite rostra were studied by means of an optical microscope coupled with a CCL Mk5-2 Cambridge Image Technology Ltd. cold cathode device. The rostra were cleaned, using a hand-held drill, from adherent sediment, apical-line areas, alveolar fissure infillings and, if necessary, narrow luminescent rims or veins. Fragments of the studied belemnite rostra comprising most growth rings, and derived from rostrum solidum (cf. Saelen, 1989) were powdered and homogenised in an agate mortar. Each specimen yielded 100-300 mg of carbonate powder. Aliquots of the same carbonate powders were used for chemical, oxygen, carbon and clumped isotope analyses. For clumped isotope analyses aliquots of carbonate samples studied previously by Wierzbowski and Rogov (2011), and Wierzbowski et al. (2013) were also used. The archival samples had been prepared and screened for the state of preservation in the same way (cf. Wierzbowski and Rogov, 2011;Wierzbowski et al., 2013).

Chemical analyses
Ca, Mg, Mn, Fe, Sr and Na concentrations were determined by the ICP-OES (Inductively Coupled Plasma Optical Emission Spectrometry) method using Thermo iCAP 6500 Duo system at the Polish Geological Institute-National Research Institute in Warsaw (Poland). 50-100 mg aliquots of the carbonate powders were dissolved in 5 wt% hydrochloric acid. Reproducibility of chemical analyses (2σ S.D.) was controlled by multiple analyses of measured samples and averages as follows: 1.0% for Ca, 2.4% for Mg, 2.2% for Sr, 0.8% for Mn, 2.7% for Fe, and 1.4% for Na. Repeated analyses of JLs-1 calcite and JDo-1 dolomite references (cf. Imai et al., 1996) yielded accuracies of measurements (2σ S.D.) better than 3.5% for Ca, 3.8% for Mg, 1.0% for Sr, and 4.0% for Mn. The accuracy of Fe analyses cannot be given precisely due to the employed dissolution method in weak hydrochloric acid, which is not relevant for the determination of non-carbonate iron compounds present in the both standards. In addition, Na contents of the both references are much lower than those of studied belemnite samples, therefore, the accuracy of measurements of Na concentrations cannot be precisely verified.

Clumped isotope analyses
The clumped isotope analyses were conducted between April 2015 and July 2017 at the Institute of Earth Sciences, Goethe University, Frankfurt, Germany, following the procedure described in Bajnai et al. (2018). Clumped isotope composition of all samples was measured in at least five replicates. Carbonate digestion and CO 2 purification were made on a fully automated preparation line. For each replicate analysis, ca. 6 mg of homogenised carbonate powder was reacted at 90°C, for 30 min, with > 105% phosphoric acid. The resultant CO 2 was purified cryogenically five times using a Porapak Q trap cooled down to −15°C. The sample gas was measured alternately with a reference gas of known isotopic composition using the dual-inlet system of a ThermoFisher Samples whose elemental concentrations and δ values are reported after Wierzbowski et al. (2013).
C Samples, whose elemental concentrations are reported after Wierzbowski et al. (2017).

Oxygen and carbon isotope analyses
Oxygen and carbon isotope analyses of newly collected mid-Kimmeridgian belemnite rostra have been performed at the GeoZentrum Nordbayern, University of Erlangen-Nuremberg, Erlangen, Germany. Samples were reacted with 100% phosphoric acid at 70°C using a Gasbench II connected to a ThermoFisher Delta V Plus mass spectrometer. All values are reported in per mil relative to VPDB scale by assigning a δ 13 C and δ 18 O value of +1.95‰ and − 2.20‰ to NBS19 and − 46.6‰ and − 26.7‰ to LSVEC, respectively. The reproducibility and accuracy of the measurements was monitored, over the course of analyses, by replicate analysis of laboratory standards Sol 2 (n = 10) and Erl 5 (n = 8). Reproducibility for δ 13 C and δ 18 O values was 0.04‰ and 0.03‰ ( ± 1σ S.D.) for Sol 2, and 0.06‰ and 0.05‰ ( ± 1σ S.D.) for Erl 5, respectively.
To calculate δ 18 O-derived temperatures and water δ 18 O values for calcite a relationship of O'Neil et al. (1969) modified by Friedman and O'Neil (1977), Eq. (2) along with the SMOW to PDB scales conversion given by Friedman and O'Neil (1977), was used. (2) where α calcite-water is the oxygen isotope fractionation factor between calcite and water, and T is the temperature in Kelvin.
To calculate δ 18 O-derived temperatures and water δ 18 O values for aragonite a relationship established by Böhm et al. (2000), Eq. (3), along with the VSMOW to VPDB scales conversion given by Coplen et al. (1983), was used.
18. 45 10 /T-32.54 aragonite-water 3 (3) where α aragonite-water is the oxygen isotope fractionation factor between aragonite and water and T is the temperature in Kelvin. Since the 1000lnα aragonite-water temperature relationship of Böhm et al. (2000) is based on δ 18 O values of aragonite samples assigned directly to a calcite reference NBS 19, and measured at similar laboratory conditions as our samples, in the present study we report δ 18 O values of aragonite ammonite and gastropod samples referenced to calcite standards (Table 1).
The reported values differ slightly from those published previously by Wierzbowski and Rogov (2011), and Wierzbowski et al. (2013), which were re-calculated using a temperature-specific CO 2 -acid fractionation factor for aragonite.
Newly collected mid-Kimmeridgian belemnite samples are characterized by the lack of luminescence and very low Mn (≤20 ppm) and Fe (≤33 ppm) as well as high Sr (≥1040 ppm; Table 1) concentrations. Their minor and trace element contents is similar to those of older belemnite material from the Russian Platform (cf. Wierzbowski and Rogov, 2011;Wierzbowski et al., 2013). Archival and new belemnite materials selected for the clumped isotope analyses show exceptionally low concentrations of Mn (≤ 9 ppm) and Fe (≤20 ppm) as well as high contents of Sr (≥824 ppm). Low contents of Mn and Fe in studied samples, which are buried in dark, often organic rich sediments, contradict suggestions that belemnite rostra might have been originally porous and cemented during early marine diagenesis (cf. Benito et al., 2016;Hoffmann et al., 2016; see also Supplementary Data 2). In addition, pervasive calcite cementation of fossils could not have occurred in fine-grained siliciclastic sediments of low permeability and carbonate content, which predominate in the Jurassic of the Russian Platform.
Although the physical and chemical screening methods of the preservation state of calcium carbonate are well-established in a case of oxygen and carbon isotope studies the solid-state reordering of CeO bonds in the carbonate lattice can affect its Δ 47 values without discernible microstructure and chemical changes. Recent studies of Passey and Henkes (2012), and Henkes et al. (2014) imply that Δ 47 values of calcite preserve at burial temperatures below 100-120°C. Models of Stolper and Eiler (2015) suggest, however, that calcites can undergo small decreases in their Δ 47 values at temperatures of 80-100°C. The clumped isotope composition of metastable aragonite is also very susceptible to the bond re-ordering and aragonite Δ 47 values have been experimentally modified at 100°C in aqueous fluids (Staudigel and Swart, 2016;Ritter et al., 2017).
The geothermal gradient of the Russian Platform is low (0.7 to 2.5°C/100 m), and is estimated to have been ca. 3°C/100 m during the Jurassic and the Cretaceous (Bazhenova, 2008). As a consequence, catagenic processes of hydrocarbon generation (above 75°C) are expected to have occurred in sediments buried below 2.5 km. They affected Precambrian deposits of the Russian Platform (Bazhenova, 2008). Jurassic sediments of the Russian Platform occur close to the Earth surface and there is no evidence that they were deeply buried. Organic matter present in the Upper Jurassic in the study area is immature as it is indicated by low T max in Rock Eval analyses Shchepetova and Rogov, 2013; and Supplementary Data 2). Very low thermal maturity of the organic matter from Oxfordian strata near Makar'ev is additionally proved by high concentrations of original biomolecules (Bushnev et al., 2006). The presence of metastable aragonite in the sediments studied and similar Δ 47 values of calcite and metastable aragonite may also point to the low thermal alteration. Although organic aragonites transform into calcite on heating above 250°C (Yoshioka and Kitano, 1985), the aragonitecalcite transformation in aqueous solutions, which are common in diagenetic environments, were observed in the temperature range of 50-175°C (Bischoff, 1969;Ritter et al., 2017).

Oxygen isotopes
Various fossil groups, which co-occur in the uppermost Callovian and the Oxfordian of the Russian Platform, are characterized by different δ 18 O values (Fig. 3). Mesohibolitid belemnite rostra are reported to have lower δ 18 O values than those of coeval cylindroteuthids (cf. Wierzbowski and Rogov, 2011;Wierzbowski et al., 2013). Ammonites are divided into two groups: one from the Callovian-Oxfordian boundary and the lower part of the Tenuiserratum Zone of the Middle Oxfordian with δ 18 O values between 0.0 and 1.5‰ (mean of 0.8‰) and another one from the Densiplicatum Zone of the Middle Oxfordian and the Middle-Late Oxfordian boundary with much lower δ 18 O values between −0.5 and −2.5‰ (mean of −1.3‰; Fig. 3). δ 18 O values of aragonitic gastropod shells vary between 1.0 and 1.4‰ (Fig. 3).

Correlation between isotope and chemical ratios of cylindroteuthids
Temporal trends and relationships between various isotope and chemical data can be studied within the cylindroteuthid dataset, which covers the whole interval from the uppermost Callovian to the lowermost Upper Kimmeridgian (Fig. 3). There is a weak (R = 0.41) but statistically significant (at 1% level) linear correlation between δ 18 O and δ 13 C values of cylindroteuthid rostra (Fig. 4A) and a strong correlation (R = 0.83) between their δ 18 O values and Sr/Ca ratios (Fig. 4C). A very weak (R = 0.26) linear correlation, significant at 5% level, is observed between cylindroteuthid δ 18 O values and Mg/Ca ratios (Fig. 4B). No correlation is, in turn, noticed between cylindroteuthid Δ 47 (CDES 25) and both δ 18 O values and elemental proxies (R ≤ 0.11; Fig. 4D-F). There is also no discernible differentiation in the isotope and chemical composition of various genera of studied cylindroteuthid belemnites.

Water temperatures
A remarkable decrease (ca. 3‰) in δ 18 O values of cylindroteuthid belemnite rostra (Fig. 3; see also Dromart et al., 2003;Wierzbowski and Rogov, 2011;Wierzbowski et al., 2013), if treated solely as a result of a temperature increase, may point to a warming of bottom waters of the Middle Russian Sea of ca. 12°C during the Oxfordian and the Kimmeridgian. Differentiation of oxygen isotope compositions of various fossils (nektobenthic cylindroteuthid and mesohibolitid belemnites, and nektonic ammonites) from the Russian Platform is, in turn, regarded as a result of the presence of a significant thermal gradient in the water column of the Middle Russian Sea (Wierzbowski and Rogov, 2011;Wierzbowski et al., 2013).
The Middle Russian Sea was a huge and shallow epicontinental basin, whose salinities may have varied during the Jurassic, along with changes in the local palaeobathymetry, circulation patterns and connections with the open Tethys Ocean and the land-locked Boreal Sea (Fig. 5). Therefore, palaeotemperature reconstructions for this area based on the assumed constant water salinity and seawater δ 18 O value should be treated with caution (Wierzbowski et al., 2013). Nevertheless, a Late Oxfordian-Early Kimmeridgian bottom water warming of 6.5-9.0°C was suggested for the Middle Russian Sea as based on the changes in δ 18 O values and Sr/Ca ratios of cylindroteuthid belemnite rostra (Wierzbowski et al., 2013).
Recent clumped isotope studies based on well-preserved carbonate fossils have proved the applicability of this method for the reconstruction of conditions of Mesozoic marine and brackish environments (Dennis et al., 2013;Price and Passey, 2013;Tobin et al., 2014;Petersen et al., 2016). Obtained clumped isotope data are derived from exceptionally well-preserved fossils of the Russian Platform and are fully reliable. Similar clumped isotope temperatures of ca. 16°C are recorded for the whole studied interval encompassing the uppermost Callovian-mid Kimmeridgian (Fig. 3). Observed temperature variations are broadly within the ranges of measurement errors (ca. ± 2.3°C) and likely arise from a general scatter of data points, with a few outliers. Since studied fossils (belemnites and ammonites) are considered to have been nektobenthic and nektonic, respectively (cf. Anderson et al., 1994;Wierzbowski and Joachimski, 2007;Wierzbowski and Rogov, 2011;Wierzbowski et al., 2013), the calculated clumped temperatures are expected to be relevant to bottom and intermediate waters of the Middle Russian Sea.
The studied basin was sensitive to a climate change due to its shallow, epicontinental character and location in mid-latitudes (40-45°N; Thierry et al., 2000) The constant clumped isotope temperatures evidence, therefore, long-term stability of the latest Middle Jurassic-Late Jurassic climate. The measured temperature of ca. 16°C is a few degrees centigrade higher than modern average seawater temperatures (8.0 and 11.4°C) at 100 m depth in the 40°N and 45°N, respectively (Locarnini et al., 2013). This may be a result of a warmer and more equable Jurassic climate and the absence of cold bottom water masses. It is worth noting that the calculated clumped isotope temperatures are within the range of previously reported belemnite clumped isotope temperatures (13-19°C) for the lowermost Cretaceous of the sub-polar Urals (Price and Passey, 2013 (Fig. 3) are likely due to δ 18 O water and salinity variations. In addition, the clumped isotope data suggest that the previously postulated thermal stratification of the Middle Russian Sea (cf. Wierzbowski and Rogov, 2011;Wierzbowski et al., 2013) was apparent and arising from salinity stratification of the water column.
The positive correlation observed between δ 18 O and δ 13 C values of the studied fossils could either result from significant salinity fluctuations in marginal marine environments (Surge et al., 2001(Surge et al., , 2003Swart et al., 2001) or be related to kinetic effects (McConnaughey, 1989a, 1989b: McConnaughey et al., 1997Auclair et al., 2003). It is, however, possible that a moderate positive correlation between δ 18 O and δ 13 C values of Russian cylindroteuthid belemnites (Fig. 4A) is a result of a more or less casual coincidence between a Kimmeridgian period of decreased belemnite δ 18 O values in the Russian Platform and a global fall in δ 13 C values of marine carbonates after the Middle Oxfordian positive carbon isotope excursion (cf. Wierzbowski et al., 2013;Wierzbowski, 2015).
Mg/Ca and Sr/Ca ratios of brackish waters remain relatively The weak correlation between cylindroteuthid Mg/Ca ratios and δ 18 O values, along with a little change in the Mg/Ca ratios, suggests that other factors controlled magnesium uptake of belemnite rostra. They may be related to temporal variations in water chemistry. Such explanation is plausible as the Mg/Ca ratio of seawater is considered to increase slightly in the Late Jurassic (Hardie, 1996;Dickson, 2004). The studied strata were deposited during a geological time of ca. 10 Ma (cf. Ogg et al., 2012Ogg et al., , 2016Wierzbowski et al., 2017), which is long enough to record a temporal change of global seawater Mg/Ca ratio.
Investigations of modern planktonic foraminifers, brachiopods and bivalves (Lea et al., 1999;Lorrain et al., 2005;Kısakürek et al., 2008;Freitas et al., 2012;Butler et al., 2015; indicate that the temperature sensitivity of Sr/Ca ratio of biogenic calcites is much weaker than the temperature sensitivity of the Mg/Ca ratio. In addition, prominent changes in the seawater Sr/Ca ratio, concomitant with 87 Sr/ 86 Sr variations, occurred during the Early and the Middle Jurassic (Ullmann et al., 2013). As the lowest seawater Sr/Ca ratio, which coincides with Phanerozoic minimum of the 87 Sr/ 86 Sr ratio, is postulated for the Middle-Late Jurassic transition (Ullmann et al., 2013) it is probable that the ratio increased after this turning point. The temporal change of the cylindroteuthid Sr/Ca ratio and its correlation with local record of belemnite δ 18 O values ( Fig. 4C; see also Wierzbowski et al., 2013) may, thus, be interpreted as arising from global variations in strontium and calcium concentrations in seawater. This explanation of cylindroteuthid Sr/Ca variations differs from that of Wierzbowski et al. (2013) but is in agreement with the lack of significant fluctuations of bottom water temperatures of the Middle Russian Sea during the latest Callovian-Kimmeridgian deduced from clumped isotopes.
Little variability in measured Δ 47 (CDES 25) values along with significant standard errors of the measurements (Fig. 4E, F) precludes assessment if Mg/Ca and Sr/Ca ratios of cylindroteuthid calcite were controlled, to some degree, by the ambient water temperature during biomineralization processes.

Water δ 18 O values
Clumped and oxygen isotope data allow calculation of δ 18 O values of water from which calcium carbonate was precipitated. The δ 18 O water values of the Middle Russian Sea calculated from the cylindroteuthid data gradually decrease from ca. 2.0‰ at the Callovian-Oxfordian transition to ca. -1.4‰ VSMOW at the Early-Late Kimmeridgian transition (Fig. 6). Water δ 18 O values calculated from Lower Oxfordian mesohibolitid data are ca. 0.5‰ lower than those of coeval cylindroteuthids. Ammonite data from the uppermost Callovian-Middle Oxfordian translate, in turn, into distinctly lower δ 18 O water values (from −4.0 to −0.1‰ VSMOW).
The δ 18 O water values for the Callovian-Oxfordian transition calculated from cylindroteuthid data are very heavy assuming the average seawater isotope composition of ca. -1‰ VSMOW (Shackleton and Kennett, 1975) as distinctive of the Jurassic world, which was devoid of  Sazonova and Sazonov, 1967;Matyja and Wierzbowski, 1995;Lyyurov, 1996;Wierzbowski et al., 2015). continental glaciation. Seawater δ 18 O values above +2‰ VSMOW are not observed in modern oceans (Craig and Gordon, 1965;LeGrande and Schmidt, 2006;Rohling, 2013). Increases in δ 18 O water values, higher than 2‰ above average composition of seawater, are also not predicted in models for the ice-free mid-Cretaceous (Zhou et al., 2008). Recent studies have, however, shown that palaeotemperatures calculated from δ 18 O values of cylindroteuthid rostra are 1-3°C lower than those derived from co-occurring bivalve shells (Anderson et al., 1994;Mettam et al., 2014;Price et al., 2015). There is also a discrepancy between oxygen isotope palaeotemperatures derived from calcitic rostra and aragonitic phragmocones of the same cylindroteuthid belemnites (Price et al., 2015). Phragmocones are depleted in 18 O, which may apparently suggest higher temperature of precipitation. Although a reason for this discrepancy is not clear, it seems that temperatures calculated from δ 18 O values of cylindroteuthid rostra, using commonly applied isotope equations, are slightly underestimated. On the other hand, palaeotemperatures derived from δ 18 O values of Jurassic mesohibolitid rostra are in better agreement with those from coeval bivalves and brachiopods (Wierzbowski and Joachimski, 2007;Price and Page, 2008;Price and Teece, 2010;Wierzbowski, 2015). If presently reported mesohibolitid δ 18 O values from the uppermost Callovian-Middle Oxfordian (Fig. 3) of the Russian Platform reflect oxygen isotope fractionation predicted by Friedman and O'Neil (1977), the offset between coeval mesohibolitid and cylindroteuthid oxygen isotope data (ca. 0.75‰) may suggest a necessity of the correction of the cylindroteuthid dataset.
δ 18 O water values calculated using the cylindroteuthid data, corrected for the 0.75‰ offset from mesohibolitid data, are distinctly lower. They average 1.2‰ for the Callovian-Oxfordian transition, and ca. -2.1‰ VSMOW for the Early-Late Kimmeridgian transition (Fig. 6). The corrected cylindroteuthid data probably better reflect true variations in ancient seawater chemistry of bottom waters of the Middle Russian Sea showing a gradual decrease in δ 18 O water values and salinity during the Oxfordian and the Early Kimmeridgian.

Water salinity
Bottom water salinity of the Middle Russian Sea can be estimated using the δ 18 O water values based on corrected cylindroteuthid data and a universal salinity-δ 18 O relation models for marine basins given by Railsback et al. (1989), which take into account effects of freshwater runoff (Eq. (5)) and evaporation (Eq. (6)). The two equations are used for salinity calculations below and above the normal marine, respectively.
water 18 0 f w 0 w a t e r 0 where δ 18 O 0 is the average oxygen isotope composition of Jurassic seawater assumed as −1‰ VSMOW (after Shackleton and Kennett, 1975), Δ fw is a difference between δ 18 O values of average seawater and local meteoric water, and S 0 is average seawater salinity.
where δ 18 O 0 is the average oxygen isotope composition of Jurassic seawater assumed as −1‰ VSMOW (after Shackleton and Kennett, 1975), m 0 is an evaporative enrichment parameter being a relation between δ 18 O and salinity observed in evaporative basins, and S 0 is average seawater salinity. Average Late Jurassic seawater salinity of 34‰, δ 18 O precipitation at 45°N equal to −9‰ VSMOW, and an evaporative enrichment parameter of 0.35‰VSMOW per 1 salinity per mil may be deduced from their modern equivalents, according to Railsback et al. (1989). Alternatively, another average salinity of 39.4‰ and δ 18 O precipitation at 45°N of −7.5‰ VSMOW may be assumed for the Late Jurassic according to data given by Hay et al. (2006) and Zhou et al. (2008), respectively. Assumption of the higher average salinity in the second model results in a need of adjustment of evaporative enrichment parameter (m 0 ) to a wider salinity range. Its calculated value is 0.31‰VSMOW per 1 salinity per mil.
The constructed models suggest higher, than average, salinity of bottom waters of the Middle Russian Sea during the latest Smoothed curves and corresponding 95% confidence intervals, calculated using locally weighted polynomial regression, represent cylindroteuthid (dashed) and corrected cylindroteuthid (shaded) δ 18 O water salinity trends.
Callovian-Middle Oxfordian and freshening of the basin during the Late Oxfordian-earliest Kimmeridgian (Fig. 7). Although the modelled data are tentative because of i) uncertainty of published clumped isotope temperature scales (cf. Wacker et al., 2014) and considerable standard errors of measurements of Δ 47 (CDES 25) values, ii) problems with the estimation of real oxygen isotope fractionation for cylindroteuthid belemnites rostra (cf. Price et al., 2015), and iii) assumption of non-actualistic values of seawater salinity and oxygen isotope composition of precipitation for the Jurassic (cf. Hay et al., 2006;Zhou et al., 2008), the data provide strong evidence for salinity variations in the Subboreal Realm. Waters of the Middle Russian Sea may have been characterized by increased salinities during the latest Callovian-Middle Oxfordian because of diminished freshwater influx and the presence of saline, subtropical, Tethyan waters. A general decrease in sea level after the worldwide Middle-Late Jurassic transition highstand (cf. Sazonova and Sazonov, 1967;Norris and Hallam, 1995;Hallam, 2001;Wierzbowski et al., , 2013 may have contributed to the enhanced freshwater runoff, and the limitation of oceanic water exchange with the restricted Middle Russian Sea. This probably led to a decrease in salinity and water δ 18 O values of this basin. The decrease is likely recorded in δ 18 O values of carbonate fossils from the Russian Platform, whose fall was previously interpreted as an evidence for the prolonged period of warm Kimmeridgian-Volgian climate (Riboulleau et al., 1998;Price and Rogov, 2009). Coeval oxygen isotope data from other (Sub)boreal basins i.e. Scotland, Northern Siberia (Nunn et al., 2009;Nunn and Price, 2010;Žak et al., 2011) suggest that the Kimmeridgian-Volgian decrease in carbonate δ 18 O values is a common feature in this palaeobiogeographical province. In addition, clumped isotope data imply that previously postulated thermal stratification of the Middle Russian Sea (cf. Wierzbowski and Rogov, 2011;Wierzbowski et al., 2013) was apparent and resulting from salinity and δ 18 O value stratification of the water column. Unfortunately, relatively scarce and scattered δ 18 O water values (Fig. 6) calculated from ammonite data do not allow precise estimation of the salinity of subsurface waters, where ammonites lived (cf. Wierzbowski et al., 2013).
The increasing restriction of the Middle Russian Sea during the Late  Friedman andO'Neil, 1977, andWacker et al., 2014). Average seawater salinity of 34.0‰ (model A) and 39.4‰ (model B) is accepted for Late Jurassic oceans after Railsback et al. (1989) and Hay et al. (2006), respectively. Smoothed curves of salinity trends and corresponding 95% confidence intervals, are calculated using locally weighted polynomial regression. Oxfordian-earliest Kimmeridgian is substantiated by a fall in ε Nd (t) values of local sediments, which is linked to an increasing rate of neodymium derived from adjacent Precambrian cratons (Dera et al., 2015). Limitation of water circulation and salinity stratification of the Middle Russian Sea may have contributed to the formation of oxygen depleted bottom layer and the deposition of black shales rich in organic matter, which are common in the Russian Platform starting from the Middle-Upper Oxfordian boundary Zakharov et al., 2017). Black shale layers occur in the studied sections of the Russian Platform in the Maltonense Subzone of the Densiplicatum Zone of the uppermost Middle Oxfordian, in the Ilovaiskii Subzone of the Glosense Zone of the lowermost Upper Oxfordian, and in the Lalieranum Subzone of the Mutabilis Zone of the lowermost Upper Kimmeridgian (Głowniak et al., 2010;Wierzbowski et al., 2013;Rogov et al., 2017;Zakharov et al., 2017; see also Supplementary Data 2). A local organic rich layer is additionally found in the lowermost Kimeridgian (the Densicostata Subzone of the Baylei Zone) of the Kostroma Region of Russia (Zakharov et al., 2017). Interestingly, black shales are even more widespread and thick in the uppermost Kimmeridgian-Volgian of the Russian Platform Shchepetova et al., 2011;Zakharov et al., 2017). Organic rich strata are also common in the uppermost Oxfordian-Lower Tithonian of NW Europe Morgans-Bell et al., 2001;Smelror et al., 2001) and attributed to restricted water circulation during sealevel lowstand (Mutterlose et al., 2003). Weakening of water movement and inflow of freshwaters into Subboreal basins in the Late Jurassic is consistent with postulated models of deposition of the organic rich beds under permanent or seasonal stratification of the water column (cf. Tyson et al., 1979;Oschmann, 1991) or during a spread of stagnant deep waters in submarine depressions (cf. Wignall and Hallam, 1991).

Changes in cephalopod assemblages
Significant decrease in salinity of the Middle Russian Sea, indicated by clumped and oxygen isotope data, should have affected local marine faunas. Although the suggested drop of salinity lies within a common range of marine salinities, its impact on biota could be noticeable constituting a major reason for periods of occurrences of a mixed Boreal-Mediterranean and endemic Boreal cephalopod faunas in the study area (cf. Wierzbowski and Rogov, 2011;Głowniak et al., 2010).
Cephalopods do not tolerate large decreases in salinity because of the salt sensitivity of the hemocyanin in their blood (Mangum, 1991). Mangold and Boletzky (1988) show that no modern cephalopod survives at salinities lower than 15‰ or higher than 45‰ although some species are moderately tolerant to salinity variations. For example, Lolliguncula brevis occurred at salinities between 17 and 38‰ (Zuev and Nesis, 1971) while Sepia officinalis is caught in regions characterized by fluctuations in salinity from 20 to 35.5‰ (Guerra and Castro, 1988). It should be noted that areas characterized either by low or high salinities are poor in cephalopods (Zuev and Nesis, 1971). Response of extinct ammonites to low or high salinity remains unclear, although a general relationship between their diversity and salinities is suggested. Some ammonites, e.g. some Triassic ceratitids, are interpreted to be tolerant to high salinities (Westermann, 1996). Ammonites from the Late Cretaceous Western Interior Seaway of North America are, in turn, interpreted to have occurred at a wide range of salinities between 20 and 35‰ (Cochran et al., 2003). Hallam (1969) argue that reduced salinity in Boreal seas was a key factor responsible for the diversity contrast between Tethyan and Boreal faunas. This view is criticized by Stevens (1971) who suggests climatic zonation as a dominant factor responsible for differentiation of marine faunas.
Callovian-Kimmeridgian ammonite faunas of the study areas are characterized by the mixed origin and consist of Boreal, Subboreal, Submediterranean and Mediterranean taxa (Zakharov and Rogov, 2003;Głowniak et al., 2010;Wierzbowski and Rogov, 2011;Kiselev et al., 2013). Ammonite assemblages have been studied in sections sampled for clumped isotopes and two additional sections (Mal'gino and Yakimikha) located in the Kostroma Region of Russia (Fig. 8).
Nearly all studied uppermost Callovian-Oxfordian ammonite faunas are dominated by Boreal cardioceratids except for an assemblage of the Bukowskii Subzone of the Cordatum Zone (Lower Oxfordian, the Dubki section), which is crowded by Mediterranean oppeliids (Wierzbowski and Rogov, 2011;Kiselev et al., 2013;Fig. 8). Ammonites of the Submediterrranean origin occur in the Lower and the Middle Oxfordian. Starting from the Middle-Upper Oxfordian boundary Submediterrranean ammonites decline in frequency, and they are missing from the uppermost Oxfordian (Głowniak et al., 2010;Fig. 8). Lowermost Kimmeridgian assemblages are dominated by dwarf Boreal Plasmatites with the admixture of Subboreal aulacostephanids; only in a short interval near the Baylei-Cymodoce boundary an assemblage with relatively common Submediterranean faunal elements (Aspidoceras, Lingulaticeras) is recognized (Głowniak et al., 2010;Rogov, 2017;Fig. 8). The Lower-Upper Kimmeridgian boundary studied recently in the southern Tatarstan is also characterized by the presence of Submediterranean taxa. They are represented by Crussoliceras spp. in the uppermost Cymodoce Zone and by Aspidoceras spp. in the Mutabilis Zone Fig. 8). Taking into account the absence of aspidoceratids in the Mutabilis Subzone of the Subboreal succession of Western Europe, it may be concluded that these ammonites penetrated in the Middle Russian Sea via a southern strait which connected it with Caucasian margin of the Tethys.
Ammonite assemblages of the Russian Platform are characterized by a drop of diversity near the Callovian-Oxfordian boundary. Appearance of Submediterranean ammonites in some intervals possibly reflect changes in water circulation (cf. Rogov et al., 2017). Oscillations in ammonoid assemblages of the Russian Platform are not consistent with palaeotemperature data based on oxygen isotopes (cf. Riboulleau et al., 1998;Dromart et al., 2003;Price and Rogov, 2009;Wierzbowski et al., 2013). Late Oxfordian-Early Kimmeridgian ammonite assemblages of the Russian Platform are strongly dominated by Boreal and Subboreal taxa, which is in conflict with a temperature rise postulated previously based on the oxygen isotope composition of carbonate fossils. Observed changes in the ammonite assemblages are, however, in general agreement with a gradual decrease of salinity detected by present clumped and oxygen isotope analyses.
The Mediterranean mesohibolitid belemnites of the Russian Platform are represented by the genus Hibolithes (Gustomesov, 1976). They are characterized by small size (maximal diameter of ca. 8 mm) compared to the much bigger Hibolithes rostra found in Central or Southern Europe (Gustomesov, 1976;Wierzbowski et al., 2013). Occurrences of Hibolithes belemnites in the Russian Platform become discontinuing at the Middle-Upper Oxfordian boundary; they disappear totally in the Upper Oxfordian (Rogov, 2003;Wierzbowski et al., 2013). This is consistent with postulated restriction and a decrease in salinity of the Middle Russian Sea.

Changes in ostracod and other microfossil assemblages
Uppermost Callovian-Lower Kimmeridgian ostracod fauna of the Russian Platform was studied in detail by Tesakova (2008) and Tesakova et al. (2012). Ostracods occur at wide ranges of salinity and cannot be used as an indicator of moderate salinity oscillations. Despite attempts of the reconstruction of variations of seawater temperatures based on the distribution of ostracod faunas (see Tesakova, 2014b) there are still problems with interpretation of thermal preferences of cosmopolitan taxa; observed changes in distribution of ostracods may also be affected by sea-level fluctuations.
An approach to the reconstruction of relative depth of the basin based on analysis of ostracod communities has been done by Tesakova (2014a). The lower boundary of the upper sublittoral zone (below 50 m), which is penetrated by visible light and grown by seaweeds, is usually inhabited by large-sized benthic ostracods (carapace length of 0.48-1.2 mm). Another small-shelled ostracod assemblage (carapace length 0.25-0.32 mm) is typical of deposit feeders, which occur at all depths but predominate in the middle part of the shelf and in deeper environments (Yasuhara et al., 2009;Tesakova, 2014a). The shallowwater environments can, thus, be recognized based on high abundance (70-90%) of large-shelled ostracods.
Variations in relative diversity of small-shelled and large-shelled assemblages studied in the Dubki and the Mikhalenino sections are shown (Fig. 9). A predominance of the shallow-water environment is inferred for the latest Callovian due to the high abundance of largesized ostracods in the Lamberti Zone. Since the beginning of the Oxfordian relative diversity of small-sized genera grows drastically. The small-sized ostracods prevail in the Oxfordian and the lowermost Kimmeridgian. It reveals deep-water settings of the studied part of the Middle Russian Sea in this time period. Shallowing, marked by increasing diversity of large-sized ostracods, began in the latest Baylei Chron and is evident in the Cymodoce Chron of the Early Kimmeridgian. The large-sized ostracods also predominate in the Volgian of the Russian Platform Ustinova and Tesakova, 2017). The sea-level fall and the restriction of the Middle Russian Sea during the Early Kimmeridgian are, thus, suggested based on the ostracod fauna.
Worth noting is also the distribution of other microfossils in the Russian Platform. Planktonic foraminifera are abundant in the Middle and the lowermost Upper Oxfordian of the Makar'ev and Mikhalenino section. They become rare in younger sediments and gradually disappear around the Oxfordian-Kimmeridgian boundary (Ustinova, 2012;Colpaert et al., 2016). Studied assemblages of calcareous nannoplankton from the Moscow region show a general decline in abundance and a noticeable decrease in number of cosmopolitic Watznaueria spp. in the uppermost Oxfordian (Ustinova, 2009).
Microfossil data substantiate the validity of present interpretations of temporal changes in salinity and circulation of the Middle Russian Sea based on isotope proxies. They are not consistent with previous views about the existence of a few short-term falls in sea-level of the basin during the Oxfordian, which might have overlapped the longterm period of high sea-level (cf. Sahagian et al., 1996).

Conclusions
Clumped isotope analyses of chemically and thermally well-preserved belemnite rostra and ammonite shells from the Russian Platform have allowed calculation of absolute temperatures of seawater of the Middle Russian Sea and verification of previously presented opinions on latest Middle-Late Jurassic palaeoenvironmental and palaeoclimatic variations.
The clumped isotope data indicate constant temperatures of water of the Middle Russian Sea of ca. 16°C in the course of the latest Callovian-earliest Late Kimmeridgian. This documents long-term stability of the Late Jurassic climate and challenges previous opinions on a short-term cooling at the Callovian-Oxfordian transition and a pronounced warming during the Late Oxfordian-Early Kimmeridgian. Minor variations in belemnite Mg/Ca ratios and major variations in their Sr/Ca ratios are presently interpreted as reflecting secular variations in global seawater chemistry rather, than temperatures of belemnite growth.
The clumped and oxygen isotope data derived from cylindroteuthid belemnites, which occur continuously in the studied sediments, enable calculation of variations in δ 18 O water values and estimation of fluctuations in salinity of the Middle Russian Sea around the Middle-Late Jurassic transition. The data show the presence of waters of possibly slightly increased salinities during the latest Callovian-Middle Oxfordian and a gradual decrease in salinity (of ca. 12‰) during the Late Oxfordian-earliest Kimmeridgian. The observed change may be connected with the inflow of saline, subtropical waters from the Tethys Ocean during the sealevel highstand at the Middle-Late Jurassic transition and a subsequent period of progressive isolation of the Middle Russian Sea resulting in freshening of its waters. This is consistent with increasing provincialism of local macro-and microfauna in the Late Oxfordian-Early Kimmeridgian.
Similar circulation changes probably occurred in all Subboreal basins of Europe. They are manifested by decreases in δ 18 O values of carbonate fossils, decreases in ε Nd (t) sediment values, and increasing faunal provincialism. The restriction of epeiric Subboreal basins, diminishing water circulation and freshwater input may have contributed to temporal bottom water anoxia and widespread sedimentation of Upper Jurassic organic-rich facies.