The Schilfsandstein and its flora; arguments for a humid mid-Carnian episode?

Recently intensified research on the mid-Carnian episode stimulated discussions about the mid-Carnian climate and a supposed humid climate shift. This basin-scale study on the Schilfsandstein, the type-example of the mid-Carnian episode, applied sedimentological, palynological and palaeobotanical proxies of the palaeoclimate to a large dataset of cored wells and outcrops. The results demonstrate the primary control of circum-Tethyan eustatic cycles on the Central European Basin where transgressions contributed to basin-scale facies shifts. The palaeoclimate proxies point to a uniform arid to semi-arid Carnian climate with low chemical weathering and high evaporation. Consequently, transgressions into the Central European Basin led to increased evaporation forcing the hydrological cycle. The increased runoff from source areas resulted in high-groundwater stages on lowlands characterized by hydromorphic palaeosols and intrazonal vegetation with hygrophytic elements. During lowstands, reduced evaporation and runoff led to increased drainage and desiccation of lowlands characterized by formation of vertisols, calcisols and gypsisols and zonal vegetation with xerophytic elements. The proposed model of sea-level control on the hydrological cycle integrates coeval and subsequent occurrences of wet and dry lowlands, hydromorphic and well-drained palaeosols, and intrazonal and zonal vegetations. Thus, the Schilfsandstein does not provide arguments for a humid mid-Carnian episode. Supplementary material: Datasets of Palynomorph Eco Group (PEG) and Macroplant Eco Group (MEG) analyses are available at https://doi.org/10.6084/m9.figshare.c.4182593

The Schilfsandstein represents the type-example of a mid-Carnian episode of increased siliciclastic influx to Tethyan and peri-Tethyan basins and beyond (see reviews by Arche & López-Gómez 2014;Ogg 2015;Ruffell et al. 2015, and references therein). Since the proposal of a 'Carnian Pluvial Event' by Simms & Ruffell (1989) and the following rejection by Visscher et al. (1994), numerous studies have contributed to an increased knowledge (Dal Corso et al. 2018).
This paper presents results of the first integrated study employing compositional maturity, palaeosols, macroflora and palynoflora of the Schilfsandstein as climate proxies. Of special importance is the extensive dataset employed herein of the Schilfsandstein macroflora, one of the most famous and important floras of the Germanic Triassic (e.g. Schenk 1864; Schönlein & Schenk 1865;Sandberger 1882;Frentzen 1922aFrentzen , b, 1930Mader 1990;Kelber & Hansch 1995). The Schilfsandstein takes its name from the cane-or rushlike structures of the horsetail stems that are often preserved in situ in the rocks. The agglomeration of groups of stems of Equisetites arenaceus of up to 25 cm in diameter give, thus, the impression of a fossil reed bed (Frentzen 1930-31).
Methods and database associations and depositional environments followed Shukla et al. (2010) and Franz et al. (2014). Palaeosols were described and classified according to Mack et al. (1993). The results obtained were compared with published and unpublished data for 10 wells and two outcrops (Fig. 1). Wells and outcrops were sampled for granulometry, detrital mineralogy, geochemistry and palynology. For grainsize analyses, 87 samples were sieved with standard mesh sieves according to DIN 66165. Granulometric values, such as the median grain size, were calculated according to Folk & Ward (1957). For petrography and diagenesis of sandstones, 98 thin sections were investigated by means of transmitted light, scanning electron microscope-energy-dispersive spectrometry (SEM-EDX) and cathodoluminescence microscopy. The quantitative detrital mineralogy was estimated from point counting; results were classified following Pettijohn (1957) and McBride (1963).
For geochemical characterization, 335 samples of the Morsleben 52A well, 64 samples of the Neubrandenburg 2 well and 62 samples of the Apolda 1 well were analysed by means of inorganic and organic geochemical methods. Samples from the Morsleben 52A well were measured at the Federal Institute for Geosciences and Natural Resources Hannover (Bundesanstalt für Geowissenschaften und Rohstoffe, BGR). Preparation and X-ray fluorescence (XRF) measurements followed standard procedures already described by Franz et al. (2014). Samples from the Neubrandenburg 2 and Apolda 1 wells were processed at ALS Laboratories Galway (Ireland). Carbon and sulphur were determined by combustion furnace and acid digestion, major elements and base metals by inductively coupled plasma atomic emission spectrometry (ICP-AES; LiBO 2 fusion, four acid digestion), and trace elements and REE by inductively coupled plasma mass spectrometry (ICP-MS; LiBO 2 fusion).

Stratigraphic control
According to Beutler in DSK (2005), the informal term Schilfsandstein was replaced by the formal term Stuttgart Formation. In the northern Central European Basin, the Stuttgart Formation is composed of the sandy Lower and Upper Schilfsandstein members and the predominantly shaly Neubrandenburg, Gaildorf and Beaumont members . The base of the Stuttgart Formation is drawn at the base of the Neubrandenburg Member and the top of the formation is drawn at the base of the Beaumont Member (Weser Formation). Detailed descriptions of the Stuttgart Formation in the North German Basin have been provided by Beutler & Häusser (1982), Franz (2008), Barnasch (2010) and Franz et al. (2014Franz et al. ( , 2018a. In the southern Central European Basin, the presence of the Neubrandenburg Member cannot be demonstrated so far. Therefore, the base of the formation is drawn at the base of the Lower Schilfsandstein (Fig. 2). Detailed descriptions of the Stuttgart Formation in the southern Central European Basin have been provided by Schröder (1977), DSK (2005) and Kozur & Bachmann (2010).
Biostratigraphic control on the Stuttgart Formation is provided by conchostracans, ostracods and palynomorphs (Heunisch 1999;Kozur & Bachmann 2010;Kozur & Weems 2010). In particular, the ostracod assemblage of the Neubrandenburg Member allows time-constrained correlation (Kozur & Bachmann 2010;Franz et al. 2014). Based on biostratigraphic arguments, the Stuttgart Formation was correlated with the late Julian upper Austrotrachyceras austriacum zone (Bachmann & Kozur 2004;Kozur & Bachmann 2010). Based on sequence-stratigraphic arguments, Franz et al. (2014) suggested that the Stuttgart Formation may range into the early Tuvalian. The time-spans, estimated to be about 0.8 myr for the Stuttgart Formation (Kozur & Bachmann 2010) and about 1.2 myr for the mid-Carnian episode (Zhang et al. 2015;Miller et al. 2017), are in general agreement.

135
The Schilfsandstein and its flora reconstructed a set of circum-Tethyan fourth-order transgressiveregressive (T-R) sequences.
The pre-Schilfsandstein transgression, representing the most distinct transgression of the Lower and Middle Keuper in the North German Basin, triggered the fundamental facies shift from shalyevaporitic lithologies of the Grabfeld Formation to shaly-sandy lithologies of the Stuttgart Formation. Resulting from this transgression of the first fourth-order T-R sequence, dark shales and silts of a marine-brackish inland sea and associated coastal environments covered larger parts of the Central European Basin. Following the maximum extension of the inland sea, the falling sealevel allowed the basin-wards progradation of the fluvio-deltaic Lower Schilfsandstein ( Fig. 2; Gajewska 1973). Active fluvial channel belts were flanked and delta lobes were at least partly covered by vegetated wetlands in which remnants of the Schilfsandstein macro-and palynoflora could be preserved. Avulsion of channels and abandonment of delta lobes led to drainage, desiccation, pedogenesis and decomposition of macro-and palynofloral remains. The maximum regression is reflected by the maximum basin-wards progradation of the Lower Schilfsandstein corresponding to a pedogenic maximum in some localities. The following second fourth-order T-R sequence resulted in the repetition of the depositional theme of transgressive shales and progradational clastic deposits (Gaildorf Member, Upper Schilfsandstein).

Granulometry, detrital and authigenic mineralogy of sandstones
The median grain sizes (Folk & Ward 1957) of 87 samples of the Lower and Upper Schilfsandstein range from 2.1 to 4.2 Phi. According to Wentworth (1922), 24 samples are classified as fine sand, 53 samples as very fine sand and 10 samples as coarse silt. Samples from channel fills, proximal crevasse splays and sheetsands are characterized by coarser grain sizes whereas samples from levees, distal crevasse splays and sheetsands as well as wetlands are characterized by finer grain sizes. Thus, lateral facies shifts and associated changes from bedload to suspensionload processes are considered responsible for these grain-size variations.
Detrital grains of proximal and distal samples are of low to moderate sphericity and subangular to subrounded grain shape (Wentworth 1922). Resulting from this uniform low textural maturity, a downstream trend towards a more accentuated roundness in the basin centre is not indicated.
The compositional maturity of the Schilfsandstein is remarkably low. According to Pettijohn (1957), the mineralogical maturity indices of 98 samples range from 0.2 to 1.4. Following McBride (1963), six samples are classified as arkose, 86 as lithic arkose and five as feldspathic litharenites (Fig. 3). In individual samples, feldspar has a range of 24.4-68.5%, quartz has a range of 17.4-58.8% and lithic fragments have a range of 3.9-42.3% of all detrital grains identified to feldspar, quartz and lithic fragments. The feldspar assemblage is dominated by untwinned and twinned plagioclase with lesser K-feldspar. The quartz assemblage is dominated by monocrystalline quartz followed by polycrystalline quartz followed by chert. The assemblage of lithics is composed of metamorphic rock fragments, mainly schists and gneisses, and igneous rock fragments outweighing rock fragments of sedimentary sources.
The detrital mineralogy, in particular the abundance of lithic fragments, seems to be grain-size dependent. An increase of grain size corresponds to an increased abundance of lithic fragments. Within fine-grained samples, the lowered abundance of lithic fragments is balanced by increased abundances of quartz and feldspar. Interestingly, the detrital mineralogy is obviously not related to transport distances, as a downstream increase of compositional maturity could not be observed. Likewise, in situ feldspar weathering is not indicated, as feldspar ratios of pedogenic sandstones are not systematically lowered (Fig. 3).
Intense burial diagenesis is witnessed by corrosion of and authigenic overgrowth on detrital grains and authigenic cementation of open intergranular pore space. Detrital feldspars range from apparently unaltered to almost completely corroded but, important to note, unaltered to slightly altered feldspar grains predominate. Lithic fragments, such as quartz, are only subordinately affected by alteration. Authigenic quartz and feldspars occur as overgrowths on detrital quartz and feldspar grains. Pedogenic sandstones show hematite and clay mineral overgrowth on practically all detrital grains. Open pore space is partly to completely filled with nests of small euhedral analcime crystals or larger poikilotopic patches as well as larger patches of calcite, dolomite, anhydrite and gypsum. Authigenic clay minerals are mainly represented by illite/smectite and chlorite, and less abundantly by kaolinite (Fig. 3).

Palaeosols
The distribution of palaeosol orders is clearly related to depositional environments. Protosols and spodosols sensu Mack et al. (1993) were frequently observed throughout the Stuttgart Formation (Table 1). Histosols sensu Mack et al. (1993) occur mainly in the Neubrandenburg Member but are sparse above. Vertisols, calcisols and gypsisols sensu Mack et al. (1993) occur from the Lower Schilfsandstein upwards.
A typical feature is the formation of a hydromorphic palaeosol ( protosol or gleysol) during early pedogenesis followed by overprint to a vertisol, calcisol or gypsisol owing to increased drainage, desiccation and evaporation (Fig. 4;Nitsch 2005; present study). According to basin topography, the earliest calcisols and gypsisols were formed in well-drained floodplains of the Lower Schilfsandstein in South and Central Germany. For the less welldrained basin centre, these palaeosol types are proven for the Upper Schilfsandstein (Fig. 2).
Histosols are associated with floodplain wetlands flanking fluvial channels and delta plain wetlands of the Neubrandenburg Member and Lower Schilfsandstein (Fig. 2). All observed cases of histosols were characterized by remarkably thin coaly horizons not exceeding 20 cm thickness. At many localities, only one histic horizon of a few centimetres thickness was evolved. At only a few localities, for example the Neubrandenburg 2 well, several thin histic horizons were evolved but occurred in a narrow interval of less than 4 m thickness . Moreover, the outcrop example Am Hohnert showed that Schilfsandstein histosols are laterally restricted.
Spodosols sensu Mack et al. (1993) are recognized by reddish to violet spodic horizons, which are a typical feature of sand-prone successions of crevasse splays and sheetsands. These horizons formed from subsurface illuviation of iron oxides and/or iron hydroxides resulting in impregnation of sandstones (Mack et al. 1993). Because of this, iron oxides and/or iron hydroxides form ferritic cutans, spots, irregular nests or nodules, or may accentuate bedding structures. The spodosols observed herein lack clear evidence of illuviation of organic matter. This may result from alteration of illuvial organic matter during burial diagenesis and/or erosion of Bh horizons owing to truncation of palaeosols as already noted by Nitsch (2005)

The Schilfsandstein macroflora
Most plant remains available in collections were recovered in Schilfsandstein quarries of South Germany (see Kelber & Hansch 1995). Accordingly, 80% of all specimens investigated originate from localities in Baden-Württemberg (61 localities out of 82 in total) and 14% from Bavarian localities (14 localities). Thuringia, Hesse, Lower Saxony and Rhineland-Palatinate are represented by fewer than five localities. The localities of about 3% of all specimens are questionable. The localities Stuttgart (177 specimens), Feuerbacher Heide (105 specimens; today part of Stuttgart), Eyershausen (45 specimens) and the surroundings of Eberstadt-Lennach-Buchhorn (33 specimens) were the most productive fossil lagerstätten, whereas 31 localities were represented by only one fossil.

137
The Schilfsandstein and its flora banks, with fern-and bennettitalean-dominated wetlands (wet lowland) flanking fluvial channels, seed fern-dominated dry floodplain areas (dry lowland) and a conifer-dominated hinterland.
Comparing the neighbouring localities Stuttgart and Feuerbacher Heide, the composition of the Schilfsandstein macroflora shows only minor changes (Fig. 9). Both macrofloras are dominated by the River MEG (Stuttgart 58%; Feuerbacher Heide 61%), followed by the wet Lowland MEG (22% and 16%) and the dry Lowland MEG (10% and 13%). The occurrence of the Coastal MEG should be noted, with 1% at Stuttgart and 3% at Feuerbacher Heide.
The comparison of the distant localities Eberstadt and Eyershausen (about 190 km distance) suggests more pronounced changes. The Eberstadt macroflora is dominated by conifers (Hinterland MEG) with several not better defined plant remains and a reduced amount of horsetails, supporting a flora that may have been subjected to a higher transport and/or drier conditions. Interestingly, the Hinterland MEG is completely missing in the Eyershausen macroflora. Instead, the flora is composed of 44% wet Lowland MEG, followed by 33% River MEG and 22% dry Lowland MEG (Fig. 9). However, it is important to note that both macrofloras are represented by a limited amount of specimens.

The Schilfsandstein palynoflora and its associations
Samples investigated in this study could be assigned to palynomorph zones GTr 13-15 of Heunisch (1999); 35 productive samples are assigned to the GTr 14 zone covering the Stuttgart Formation. As these samples mainly originate from the lower part of the formation, whereas most samples of the upper part were unproductive, the preservation potential was obviously higher for marine-brackish and fluvio-deltaic environments associated with the first fourth-order T-R sequence (Fig. 10).

139
The Schilfsandstein and its flora PEG (Aratrisporites spp., Leschikisporis aduncus, Duplicisporites granulatus) may occur with higher abundances. Deltoidospora spp. are of accessory occurrence. The Ovalipollis association occurs in dark grey but also variegated shales and siltstones of the topmost Grabfeld and basal Weser Formations. In the North German Basin, the association ranges into the basal Neubrandenburg Member. As the Ovalipollis association appears to be limited to these intervals, it most probably represents the pre-and post-Schilfsandstein palynoflora dominated by dry Lowland and Hinterland PEGs.

141
The Schilfsandstein and its flora

Leschikisporis association
Leschikisporis aduncus constitutes 25-75% of the assemblage and Aulisporites astigmosus is present at less than 10% (Fig. 11). For individual localities, A. astigmosus may be missing in the palynomorph association (Kahlert et al. 1970). Another important taxon is Aratrisporites spp., which is present at up to 7%. The taxa Deltoidospora spp., Calamospora spp., Osmundacidites wellmanii, Baculatisporites sp., Punctatisporites spp., Cycadopites spp., undifferentiated bisaccate pollen and acritarchs may occur as accessories. The Leschikisporis association is dominated by the Coastal PEG, but also the wet Lowland, dry Lowland, River and Highland PEGs (up to 15%) may be common. As the association could be observed only in dark shales of lower delta plain wetlands in the Neubrandenburg 2 well, its occurrence seems to be limited to central parts of the North German Basin (Fig. 10).

Aulisporites association
Aulisporites astigmosus constitutes >30% of the assemblage whereas Leschikisporis aduncus is present at less than 10% (Fig. 11). For individual samples, the maximum abundances of A. astigmosus (wet Lowland PEG) may even exceed 80% (Supplementary material). Other important forms of the Aulisporites association are elements of the Coastal PEG (Araucariacites australis, Aratrisporites spp.), River PEG (Punctatisporites spp.) and Hinterland PEG (bisaccate pollen grains). Prasinophycean algae, such as Leiosphaeridia spp., may occur as accessories. The Aulisporites association is recorded in dark grey to blackish and often carbonaceous shales and coals of delta plain and floodplain wetlands (Fig. 10).

Trilete laevigate spores association
This association is characterized by a mixture of palynomorphs of the dry Lowland, River and Hinterland PEGs. Trilete laevigate spores, such as Deltoidospora spp., Calamospora spp. and Punctatisporites spp., are present at percentages of up to 30%. A. astigmosus and L. aduncus are also present but at percentages below 25%. Of further importance are Triadispora spp. (<15%), bisaccate pollen grains (<25%) and ornamented spores (<15%). The trilete laevigate spores association is recorded in dark shales but also in laminated silts of floodplain and delta plain wetlands in North Germany (Fig. 10).

Low chemical weathering and high evaporation
For this study, only palaeosols and compositional maturity of sandstones were evaluated as palaeoclimate proxies. As authigenic feldspar overgrowths and formation of clay minerals, as a result of burial diagenesis, are common features (Fig. 3;Wurster 1964;Heling 1965;Förster et al. 2010;present study), clay mineral analysis and geochemical weathering indices (e.g. chemical index of alteration; CIA) were ruled out.

The Schilfsandstein and its flora
Together with other arguments, these examples were employed to reconstruct a long-term change from arid Early Triassic climates to humid Jurassic climates (e.g. Ahlberg et al. 2002Ahlberg et al. , 2003Decou et al. 2017). The low compositional maturity of the Schilfsandstein is apparently not in agreement with this long-term trend. From the Buntsandstein to Middle Jurassic, the North German Basin was mainly fed from Scandinavian sources to the north Lott et al. 2010;Zimmermann et al. 2018). Considering the stable basin centre and, at least for the Keuper, comparable fluviodeltaic routing systems (see Franz et al. 2018b), the climatic significance of the low compositional maturity of the Schilfsandstein is underlined by comparisons with the Lower Keuper and Rhaetian (Fig. 3). For the Danish, North German and Polish Basins, the Ladinian climate was reconstructed as semihumid to humid and the Rhaetian climate was reconstructed as humid (Ahlberg et al. 2002(Ahlberg et al. , 2003Fijałkowska-Mader 2015;Franz et al. 2015;Szulc 2000). Consequently, the low compositional maturity of the Schilfsandstein is ascribed to an arid to semi-arid mid-Carnian climate herein.
High abundances of chemically unstable detrital feldspars and also rock fragments are related to low degrees of chemical weathering and short residence time of the sediment (e.g. Johnsson & Meade 1990;Ibbeken & Schleyer 1991). According to Van de Kamp (2010), arkoses in which plagioclase predominates over K-feldspar are the product of feldspathic crystalline sources and glacial or arid to semi-arid climates characterized by a predominance of physical weathering. Chemical weathering in source areas and the basin (e.g. during pedogenesis) will result in hydration of feldspars and formation of clay minerals (Nesbitt & Young 1982). Longer times of residence (e.g. in intermediate deposits) and subsequent reworking will also result in an increase of alteration in the downstream direction (Johnsson & Meade 1990;Van de Kamp 2010). Consequently, the low maturities of mid-Carnian sandstones of the NW Tethys and peri-Tethyan realm indicate low degrees of chemical weathering in source areas and basins.
The record of Schilfsandstein palaeosols described herein corresponds to that of Nitsch (2005). The occurrence of histosols and hydromorphic soils in the Neubrandenburg Member and Lower Schilfsandstein is clearly related to facies shifts triggered by the pre-Schilfsandstein transgression (Fig. 2). During the following regressive phase, the falling sea-level resulted in increased drainage and desiccation of floodplains and delta plains, overprint of hydromorphic soils, and formation of vertisols, calcisols and gypsisols ( Fig. 4; Nitsch 2005; present study). Owing to shortterm flooding of the intra-Schilfsandstein transgression, hydromorphic soils evolved only locally in the Upper Schilfsandstein whereas stacked successions of gypsisols became abundant (Fig. 6). The widespread occurrence of vertisols, calcisols and gypsisols in the Schilfsandstein provides evidence for a mid-Carnian hydrological regime with high evaporation rates exceeding precipitation rates (e.g. Dan & Yaalon 1982;Retallack 1994;Sheldon & Retallack 2004;Retallack & Huang 2010) in which hydromorphic soils and histosols could develop locally on delta plains and floodplains during phases characterized by high-groundwater stages. A comparable scenario was described by Palain (1966) from the Gres à Roseaux, the equivalent of the Schilfsandstein in the Paris Basin.

The Schilfsandstein macroflora
The Schilfsandstein macroflora database may be influenced by collecting and taphonomic biases, as no material could be collected in situ and/or by the authors themselves. However, the presence of abundant badly preserved plant fossils (especially sphenophyte stem remains) suggests that perhaps this bias is not too high. As most of the Schilfsandstein macroflora was recovered from sandstones the preservation is often poor. Nonetheless, is it possible to observe in situ stems as well as several generations of stems growing on top of each other (e.g. Kelber & Hansch 1995). Frentzen (1930-31) found that the Schilfsandstein flora does not show xerophytic characters. The large leaves of Chiropteris and Danaeopsis demonstrate that there was enough water available and plants did not need to reduce the lamina or use other adaptations to reduce transpiration. Frentzen reconstructed the vegetation as belonging to swamped lowlands laterally associated with fluvial channels.
According to our database, the swamped lowlands (e.g. backswamps) were mainly colonized by horsetails (Equisetites, Neocalamites, Schizoneura) and Pterophyllum. Although Bennettitales are generally considered xerophytes and are indicators of arid environments, Pott et al. (2008) have shown that the macromorphological and epidermal features of Pterophyllum leaves from the mid-Carnian Lunz ecosystem supported peat-forming environments. In the Schilfsandstein ecosystem, the osmundaceous ferns Cladophlebis and Neuropteridium were growing close to and on the riverbanks. The community of the wet Lowland MEG was probably richest in species. The understorey and clearings were filled with a high diversity of ferns (Danaeopsis, Chiropteris, Osmundites, Dictyophyllum; Kustatscher et al. 2012) and shrubby conifers (Pelourdea; Kustatscher et al. 2014). Sphenopteris schoenleiniana would be mostly climbing on the tree-like plants .
The cuticles in the Ladinian specimens of Scytophyllum show adaptations to drier or more stressful environments Kustatscher et al. 2012). Also, some other seed ferns (Sagenopteris, Lepidopteris) and Bennettitales (Anomozamites, Zamites) may have grown in dry lowlands or in the hinterland. Some ferns (Asterotheca, Clathropteris) are characterized by a thick, leathery lamina of the pinnules suggesting drier (low groundwater table) or stress-related habitats (e.g. Kustatscher et al. 2012). Lepacyclotes, the only lycophyte present in the flora, is considered to grow in coastal areas (Kustatscher et al. , 2012, as is the fern Symopteris (Kustatscher et al. , 2012 and the seed fern Ctenozamites (comparable with Ptilozamites; see Kustatscher & Van Konijnenburg-van Cittert 2005;Popa & McElwain 2009). The conifers grew mainly in drier and more distant floodplains or the hinterland. The longer transport could explain the stage of fragmentation and preservation and the relatively low amount of material, in comparison with the palynoflora.
In accordance with the Leschikisporis association proposed herein, the abundances of A. astigmosus are below 25%. Associations characterized by a mixture of trilete laevigate spores, ornamented spores, bisaccate pollen grains and other palynomorphs were also reported by Orłowska-Zwolinśka (1983).

A tentative reconstruction of the Schilfsandstein flora
Any integration of macro-and palynoflora datasets is biased by statistical, palaeogeographical, stratigraphic and palaeoecological limitations. Statistical limitations may exist because for Palynomorph Eco Group analysis (PEG) more than 200 specimens have been counted per sample, whereas only for two localities, Stuttgart and Feuerbacher Heide, were more than 100 macroplant specimens available in collections. As the macrofloras of both localities show comparable compositions, the Stuttgart macroflora is considered statistically relevant. Palaeogeographical limitations result from the lack of macroflora data from Northern Germany. A palaeogeographical overlap may be provided by the Obernsees palynoflora and the macrofloras of Eyershausen, Gochsheim, Iphofen and Zeil am Main, but as only 14-45 specimens were available from these localities, none of the macrofloras is statistically significant. However, as the macroflora of Iphofen resembles the macroflora of Stuttgart it may be considered statistically relevant (Fig. 9).
The Obernsees palynoflora is well constrained to the Lower Schilfsandstein but the Iphofen macroflora can only be tentatively assigned to this unit. Because of the grey to yellowish rock slabs lacking features of intense pedogenesis, the assignment to the Lower Schilfsandstein appears feasible. The macroflora was recovered from a sandy near-channel succession and the palynoflora from floodplain wetlands, but as both originate from the same channel-floodplain system a comparison is attempted.
The Iphofen macroflora is dominated by the River MEG (63%) followed by the wet Lowland MEG (26%), dry Lowland MEG (5%) and Hinterland MEG (5%). The Obernsees palynoflora is dominated by the wet Lowland PEG (up to 49%), followed by the Hinterland PEG (up to 23%), coastal PEG (up to 21%), River PEG (up to 11%) and dry Lowland PEG (up to 5%). The different abundances of the River MEG and PEG are ascribed to the different environments. Accordingly, the macroflora is dominated by a ( para) autochthonous riparian vegetation of sphenophytes, mainly large articulated plant remains and in situ stems, growing on river banks. In the palynoflora the River PEG is less abundant owing to episodic introduction of spores to floodplain wetlands; for example, by crevassing. In accordance with lithofacies, the palynoflora shows high abundances of the wet Lowland PEG and subordinate abundances of the coastal PEG. This is due to high abundances of A. astigmosus, a Bennettitean pollen grain assigned to Williamsonianthus keuperianus (Kräusel & Schaarschmidt 1966). Bennettitales are allocated to the wet Lowland MEG (see above), and therefore A. astigmosus was assigned to the wet Lowland PEG herein. Considering these results, a vegetation of ferns (mostly Marattiales, Osmundales and Dipteridaceae), cycadophytes (Pterophyllum), and shrubby conifers (Pelourdea) is reconstructed for the floodplain lowland and a vegetation of sphenophytes (Equisetites mostly) and osmundaceous ferns (Cladophlebis, Neuropteridium) for the banks of backswamps. The dry Lowland and Hinterland elements represent distant localities. The 26% abundance of the wet Lowland MEG may result either from mixed riparian vegetation of sphenophytes, bennettiteans and ferns, or from mixing during transport processes.
Comparisons of the Schilfsandstein flora with pre-or postcursor Keuper floras are limited, as the Grabfeld and Weser Formations yielded only palynomorphs but not macroplant remains. This and the high abundances of the dry Lowland PEG in these palynofloras emphasize two aspects: (1) wet floodplains represent a taphonomic window allowing the preservation of the Schilfsandstein flora; (2) the Schilfsandstein flora represents an intrazonal flora related to a sea-level highstand in the Central European Basin. Recently, examples of taphonomic windows were reported by Kustatscher et al. (2014Kustatscher et al. ( , 2017. In particular, the intrazonal riparian Bletterbach flora, which occurred in an arid environment owing to sea-level highstand (Kustatscher et al. 2017), seems to be an analogue for the Schilfsandstein flora.

Sea-level-controlled evaporation forcing the hydrological cycle of the Central European Basin
The data presented herein argue for a rather uniform arid to semiarid Carnian climate. Considering this and the primary control of circum-Tethyan eustatic third-order and fourth-order cycles (Köppen 1997;Franz et al. 2014), the Schilfsandstein is herein related to changes of the hydrological cycle of the Central European Basin. This was already concluded by Kozur & Bachmann (2010), but those researchers considered the short-term redirection of the NW Tethyan monsoon across the Central European Basin responsible for a pluvial climate in the Scandinavian source area. However, such an unusual circulation pattern remains speculative and a pluvial climate in the Scandinavian source area is not in agreement with the immature Schilfsandstein. Moreover, the monsoonal moisture transport would have also caused seasonal precipitation over the Variscan source areas, the main orographic barrier between the Tethys and Central European Basin. This was shown by Reinhardt & Ricken (2000) on the example of the Norian Löwenstein and Arnstadt Formations, for which seasonal precipitation resulted in intensified chemical weathering (Reinhardt & Ricken 2000;Vollmer et al. 2008) and higher compositional maturity of sandstones (Kulke 1969). Considering this and data presented herein, the Schilfsandstein may not be explained in terms of monsoonal circulation or increased seasonality.
Instead, the accelerated hydrological cycle seems to be related to mid-Carnian transgressions into the Central European Basin. Under arid to semi-arid Carnian climate, the flooded basin contributed to increased evaporation and formation of moisture-laden air masses. Probably transported by northwesterly flowing trade winds, these air masses led to increased rates of precipitation over source areas. During regressive phases, fluvio-deltaic environments prograded basin-wards and supplied detritus predominantly formed of firstcycle sands (Wurster 1964;Van de Kamp 2010). High-groundwater stages favoured the establishment of an intrazonal flora and its preservation in backswamps and floodplain lakes (taphonomic window). From late to maximum regression, the hydrological cycle ceased owing to the retreat of the inland sea and resulting lowered evaporation. Low-groundwater stages were responsible for increased drainage, desiccation and pedogenesis.
The pre-Schilfsandstein transgression represents the most pronounced transgression of the Lower and Middle Keuper. The

145
The Schilfsandstein and its flora control of third-and fourth-order mid-Carnian T-R sequences on the hydrological cycle is comparable with conditions for the Upper Muschelkalk and Rhaetian, where pronounced transgressions were followed by progradations of fluvio-deltaic environments from northern sources (Franz et al. 2015;Barth et al. 2018) but, as discussed above, under semi-humid to humid climates.

Conclusions
This study of sedimentary, palynological and palaeobotanical climate proxies of the Schilfsandstein points to the following conclusions.
(1) The Stuttgart Formation was primarily controlled by thirdand fourth-order mid-Carnian T-R cycles. Transgressions into the Central European Basin resulted in formation of large inland seas, and regressions allowed the progradation of fluvio-deltaic environments.
(2) For this epicontinental setting, the detrital mineralogy and palaeosols represent reliable proxies of the palaeoclimate. Owing to burial diagenesis, clay minerals and geochemical weathering indices reflect detrital and authigenic signals, and thus do not represent reliable proxies of the palaeoclimate. (3) The palaeoclimate proxies employed point to a rather uniform arid to semi-arid Carnian climate with low chemical weathering and high evaporation in which flooding of the Central European Basin contributed to increased evaporation rates and formation of moistureladen air masses. The forced hydrological cycle led to increased precipitation over source areas and increased runoff led to high-groundwater stages on delta plains and floodplains. (4) The application of MEG and PEG analyses to macro-and palynoflora datasets allows the preliminary reconstruction of the Schilfsandstein flora. Local high abundances of the wet Lowland PEG and MEG are considered intrazonal vegetation of wet lowland habitats. These habitats represent a taphonomic window that was controlled by high-groundwater stages on delta plains and floodplains. Laterally, the (more xerophytic) zonal Carnian vegetation prevailed in habitats of dry floodplains. (5) The proposed model of sea-level control on the hydrological cycle integrates the low compositional maturity and the coeval and subsequent occurrences of wet backswamps and dry floodplains, hydromorphic and well-drained palaeosols, and intrazonal and zonal vegetations. On the basis of this model, the Schilfsandstein does not provide arguments for a humid mid-Carnian episode. (6) The model proposed herein may be of significance for other epicontinental basins of the NW peri-Tethyan realm as well as for basins of the NW Tethys controlled by circum-Tethyan eustatic cycles. In these basins, the mid-Carnian clastic equivalents are of low compositional maturity and the occurrence of hydromorphic soils is related to high-groundwater stages on coastal plains.