The Devonian-Carboniferous transition at Borkewehr near Wocklum (northern Rhenish Massif, Germany) – a potential GSSP section

The Borkewehr section near Balve at the southeastern end of the Remscheid-Altena Anticline is currently the best and most complete Devonian/Carboniferous Boundary (DCB) succession of the Rhenish Massif, Germany. Based on a multidisciplinary approach, focusing on refined conodont and ammonoid biostratigraphy, microfacies analyses, sequence stratigraphy, carbon isotopes (Corg and Ccarb), trace element geochemistry, including REE data, and cyclostratigraphy, the section is proposed as a candidate for the new basal Carboniferous GSSP. The succession represents a pelagic “seamount” setting. It spans the upper Famennian to lower Viséan, with a condensed development of most Rhenish lithostratigraphic units and sedimentary sequences. It is the type-section of the oxic and strongly cyclic, fossiliferous Wocklum Limestone and of important uppermost Famennian and basal Carboniferous index species. The First Appearance Datum (FAD) of Protognathodus kockeli s.str., in a phylogenetic succession from the ancestral Pr. semikockeli n. sp., is proposed as future GSSP level. This FAD is sandwiched between many other marker levels for global correlation. Below are, from base to top, the transgressive base of the at least partly anoxic Hangenberg Black Shale (base Lower Hangenberg Crisis Interval), the Hangenberg Regression (onset of siltstones of the basal Middle Crisis Interval), the level of maximum regression, indicated by maximum Zr/Al values, and the initial transgression of the basal Upper Crisis Interval, marked by the re-onset of carbonate deposition. Above follow the locally cryptogenic FAD of Siphonodella (Eosiphonodella) sulcata s.l. at the base of the post-Crisis Interval (Hangenberg Limestone), associated with a conodont biofacies shift, the FAD of Si. (Eo.) bransoni, and, much higher, the sharp Lower Alum Shale Event at the base of the classical middle Tournaisian. The significance of the section is greatly enlarged by its precise correlation with numerous other Rhenish DCB sections that provide important additional data on faunas, geochemistry, geochronological ages, and magnetic susceptibility. The new Borkewehr data suggest changes of weathering intensity, based on Rb/K and K/Al trends, and Milankovitch cyclicity associated with the climate changes of the Hangenberg Crisis. The local conodont extinction rate at the top of the Wocklum Limestone lies at 57 % while there are no survivors in ammonoids, trilobites, or deep-water corals. Positive carbon isotopic shifts of carbonate in the Upper Crisis Interval and of Corg in the Middle Crisis Interval are decoupled, which suggests complex underlying processes that are not yet understood.


Introduction
The Devonian/Carboniferous Boundary (DCB) was the first chronostratigraphic level to be designated at the Second Heerlen Congress in 1935 (Paeckelmann and Schindewolf 1937), using the entry of the index goniatite Gattendorfia subinvoluta in the Oberrödinghausen Railway Cut (Hönne Valley, Rhenish Massif, GPS 51°23'45.5"N, 007°50'23.8"E, ca. 7 km north-northwest of the Borkewehr section, Fig. 1). Subsequent studies (e.g. Alberti et al. 1974) proved that the selected stratotype section contains a hiatus at the boundary level, corresponding to the interval of the main Stockum Limestone. This resulted in more than 15 years of intensive search for a new DCB definition and stratotype. Finally, the entry of the conodont Siphonodella (Eosiphonodella) sulcata, in a supposed phylogenetic lineage from Si. (Eo.) praesulcata, was chosen as the Carboniferous index fossil and, following the proposal of Feist and Flajs (1988), the new GSSP was placed at the base of Bed 89 in an oolithic succession at La Serre Trench E', Montagne Noire, southern France (Paproth et al. 1991). This decision was controversially discussed and especially criticised by Ji et al. (1989) as well as by Ziegler and Sandberg (1996). Based on different views regarding their taxonomic position, intermediates within the supposed phylogenetic lineage from ancestral Si. (Eo.) praesulcata to Si. (Eo.) sulcata had been recorded from below the GSSP level . In the course of geochemical studies, and using the same morphological species definition as Feist and Flajs (1988), Kaiser (2005 discovered that the index conodont occurs at La Serre Trench E' as early as in Bed 84b in the section log in Feist et al. (2020 online). This is well below the GSSP at the base of Bed 89, in the basalmost part of the upper calcoolitic unit and just above a facies break (compare Feist et al. 2000Feist et al. , 2020Spalletta et al. 2011;Aretz et al. 2016). This confirmed the most important point of criticism by Ziegler and Sandberg (1996), who mentioned that "… the whole sequence containing the supposed Si. praesulcata-Si. sulcata lineage is an artifact created by reworked sediments derived from different source areas …". Since there are no other criteria to correlate the GSSP level with precision, another revision became necessary (e.g. Kaiser and Corradini 2008;. For this, an International Devonian/ Carboniferous Boundary Task group was founded in 2008 by Philip H. Heckel (then chairman of SCCS = Subcommission on Carboniferous Stratigraphy) and R. Thomas Becker (then chairmen of SDS = Subcommission on Devonian Stratigraphy; compare Heckel 2008). Markus Aretz (chairman) and Carlo Corradini (vice-chairman) were appointed as the task group leaders at the International Commission of Stratigraphy workshop in 2010 that was held in Prague (Richards and Task group 2010).
In the meantime and the following years, progress was made by the task group and accompanying scientists. Results and new ideas were presented and discussed at several symposia, congresses, and workshops (e.g. 2 nd International Conodont Symposium, ICOS 2009, Calgary;3 rd International Palaeontological Congress, IPC3, London 2010; International Field Symposium "The Devonian and Lower Carboniferous of northern Gondwana", Erfoud 2013; IGCP 596 -SDS Symposium, Brussels 2015). New papers/documents critically scrutinized and forwarded taxonomic problems within the supposed phylogenetic lineage of Si. (Eo.) praesulcata -Si. (Eo.) sulcata and within the genus Protognathodus (e.g. Tragelehn 2010;Corradini et al. 2011Corradini et al. , 2013Corradini et al. , 2016Corradini et al. online, 2017Kaiser and Corradini 2011;Evans et al. 2013;Kaiser et al. 2019a online, b;Becker et al. 2021). During a workshop in Montpellier (2016), the task group agreed to test a proposal combining several criteria for a DCB redefinition (for advantages and disadvantages of these criteria compare Becker et al. 2016a): (1) main extinction level and base of the Hangenberg Black Shale Event, (2) "top of the major regression (top of the Hangenberg Sandstone)", end of the mass extinction, and base of the Pr. kockeli Zone, as well as (3) base of the Si. (Eo.) sulcata/Pr. kuehni Zone and lowland plant extinction. Therefore, the search for suitable successions has become crucial. A final vote by the working group on the suitability of the Montpellier criteria in the frame of the XIX ICCP 2019 (= International Congress on the Carboniferous and Permian, Cologne, compare Aretz and Corradini 2019a), did not lead to a decision. However, a majority of more than 60 % voted for the criterion "top of the major regression (top of the Hangenberg Sandstone), end of the mass extinction, and base of the Pr. kockeli Zone" and, therefore, for a new DCB definition and GSSP level, which will be situated below the current First Appearance Datum (FAD) of Si. (Eo.) sulcata sensu Flajs and Feist (1988;= Si. (Eo.) sulcata s.l. sensu Becker et al. 2021). A precise understanding of sea level changes and of the early postextinction radiation of Protognathodus and other faunal groups will be crucial. The advantage of this new timeline will be its position near the current GSSP level, which will maintain stratigraphic stability (Aretz andCorradini 2019b, 2021).
A new terminology and subdivision of an extended Hangenberg Crisis Interval was introduced by Kaiser et al. (2015) and Becker et al. (2016aBecker et al. ( , 2021. It uses lithostratigraphic units of the northern Rhenish Massif as reference levels and includes from the base to the top the following steps (see Becker et al. 2021, fig. 1 The recent review of thirty Rhenish DCB sections by Becker et al. (2021), based on a wealth of data published by many other authors, showed that there is a limited number of regional sections that could meet the criteria outlined above. Within the pelagic realm of the northern and eastern Rhenish Massif, numerous sections are now covered or poorly accessible (e.g. Hasselbachtal, Seiler region, Apricke, Müssenberg, Hangenberg, Stockum, Eulenspiegel, Scharfenberg). Others have gaps (Oberrödinghausen Railway Cut, Reigern Quarry, easily accessible successions at Drewer) or are too condensed (Effenberg, Kattensiepen). At Oese, the basal limestones overlying the regressive, locally thick Hangenberg Sandstone are too poor in conodonts to establish firm ranges in the critical interval. This problem was also encountered when Drewer was recently re-sampled. This leaves a single section, the Borkewehr near Wocklum, which we re-sampled bed-bybed. Its revised litho-, bio-, cyclo-, and chemostratigraphy as well as a sequence stratigraphic overview are presented. We regard it as the prime Rhenish DCB stratotype candidate, especially since it is an easily accessible section that is not in danger of destruction and which can be investigated using all modern stratigraphic techniques.
during the initial geological mapping by the Prussian Geological Survey. Subsequent work by Wedekind (1914aWedekind ( , b, 1918 focused on the ammonoid fauna and installed the succession as an important type locality. Based on a more extensive geological survey in association with a renewed mapping campaign, Paeckelmann (1924) provided further descriptions, regarding the lithology of the upper/uppermost Famennian Dasberg and Wocklum limestones, Hangenberg beds, as well as overlying Tournaisian and Viséan strata. The light-coloured Viséan "lydites" are nowadays mostly covered. A first brief section log of the complete succession and few faunal records (e.g. "Posidonia venusta" from the Hangenberg Black Shale) were given in Schmidt (1924), followed by Devonian trilobite data in Richter and Richter (1926), and ostracod records in Matern (1929). Lange (1929) provided further ammonoid data, including two new species from the Wocklum Limestone. He discussed Schmidt's rejection of the term "Wocklumer Schichten" (= Wocklum Limestone), which was based on an apparent cooccurrence of ammonoids from the Wocklum Limestone (= Dasberg Limestone of Schmidt 1924) and the younger Hangenberg Limestone. Lange (1929) pointed out that Schmidt's specimens were mostly collected from the scree at the northern wall of the quarry. Based on the limited stratigraphic benefit of such material, Lange (1929) declared that the term "Wocklumer Schichten" is valid. Rabien (1954) added Upper Devonian ostracod records.
The first conodont data from the Wocklum and Gattendorfia stages (for further details on Rhenish regional chrono-and lithostratigraphy see the summary of Becker et al. 2021) of the Borkewehr section were published by Bischoff (1957), who described three new (sub)species, Tripodellus robustus, B. ultimus ultimus (as Spathognathodus spinulicostatus ultimus), and Pr. kockeli (as Gnathodus kockeli), the potential index conodont for the future base of the Carboniferous. Additional conodont records were provided by Voges (1959Voges ( , 1960, who indicated in higher parts the upper Tournaisian Sc. anchoralis Zone, amongst others. Within their geological field guide to the "Sauerland", Schmidt and Pleßmann (1961) mentioned briefly the DCB and showed a simplified geological map/figure of the outcrop area. An attempt by Streel (1969) to find spores from the Devonian-Carboniferous transition failed. Eickhoff (1970) described agglutinating foraminifera from the Wocklum Limestone, including two new species, Hyperammina aperta and ?Moravammina constricta. A new section log was illustrated in a field guide by Ziegler (1971), who provided a conodont biostratigraphy from the "Upper costatus Zone" to the Sc. anchoralis Zone.
Upper Devonian ammonoids from Borkewehr were treated in several papers, such as Korn (1981,1994), Price (1987, 2019), and Price and Korn (1989). Luppold et al. (1994) studied a few DCB conodont faunas and gave a new log of the section called "Wocklum". A sample from the uppermost 10 cm of the Wocklum Limestone yielded a typical pre-crisis fauna, including Si. (Eo.) praesulcata and the last B. muessenbergensis, B. ultimus ultimus, as well as representatives of the Pa. gracilis Group. A younger sample from the base of the assumed Hangenberg Limestone, actually from the Upper Stockum Limestone sensu Becker et al. (2021), contained Pr. kuehni, Pr. kockeli, Pr. collinsoni, as well as Neo. communis communis. Based on indifferent conodont faunas without siphonodellids, it was supposedly not possible to locate the DCB ("Si. (Eo.) sulcata Boundary") with precision. Becker (1996) illustrated a well-preserved "Mim." lentum, which originally was identified by H. Schmidt as "Aganides quadripartitus". The local record of Wocklumeriidae was revised by Becker (2000) and Ebbighausen and Korn (2007). Whilst it was possible to select lectotypes for Wo. denckmanni and Parawo. paradoxa, Wedekind's (1918) single original of Epiwo. applanata has been lost. Korn (2002) listed the Borkewehr section in his compilation of important localities for the late Middle Devonian to Late Devonian ammonoid zonation in the northern Rhenish Massif. Großhäuser (2014) studied the Tournaisian to Viséan succession but did not address the strongly condensed middle Tournaisian to lower Viséan lithological units correctly since thicknesses and facies developments deviate quite strongly from the standard of the Rhenish Kulm Basin. A new detailed bed-by-bed section log for the top Wocklum Limestone to the top of the Hangenberg Limestone, with some macrofauna records (bivalves, trilobites, but mainly ammonoids), was provided by Becker et al. (2016a). Becker and Hartenfels (2017) summarised the existing Borkewehr data in the context of the ongoing DCB revision. A new survey and re-sampling by Hartenfels et al. (2017a, b) made new geochemical and conodont data available. It was pointed out that the phylogenetic change from ancestral Pr. collinsoni to Pr. kockeli is recognisable in the first few transgressive limestones immediately after the siliciclastics of the glacioeustatic Hangenberg Regression.  proposed the Borkewehr as a DCB stratotype candidate section and gave preliminary insights into the transition from Pr. collinsoni to Pr. kockeli. Herbig et al. (2019) summarised the Borkewehr data and reported that, according to new conodont data, the interval of the Stockum Limestone is thicker than previously shown in Becker et al. (2016a). Preliminary remarks concerning sequence stratigraphy, biofacies of agglutinating foraminifera, and the middle Tournaisian to lower Viséan succession were added. Basse and Lemke (2020) began to revise all Rhenish trilobites of the Wocklum Limestone. From the Borkewehr section, they recorded nine species of phacopids and proetids, with new material collected mostly from the top 10 cm below the Hangenberg Black Shale. Due to the parallel preparations for this paper, the Borkewehr section was only briefly treated in the review of Rhenish DCB sections by Becker et al. (2021).

Tectonic and palaeogeographic setting
Famennian and Mississippian strata of the northern Rhenish Massif belong to a narrow outcrop belt along the northwestern to southeastern axial ramps of the 1 st order Remscheid-Altena Anticline (Fig. 1). The latter shows the typical Variscan strike direction, extends from the Remscheid-Wuppertal region in the southwest towards the Hönne Valley region in the northeast, and disappears with the plunging fold axis to the north and northeast of Balve. Further to the northeast, the Famennian re-appears on the top of horst structures, such as the Effenberg Quarry (e.g. Hartenfels 2011;Hartenfels and Becker 2016c), which are surrounded by Mississippian strata. The Borkewehr section is situated on the southeastern flank of the Remscheid-Altena Anticline. The southeastern continuation of the anticline is the Lüdenscheid Syncline, which separates the western end of the "Ostsauerländer Hauptsattel". Amongst others, regional review papers were provided by Stoppel et al. (2006), Langenstrassen (2008), and Clausen and Korn (2008).
In the Upper Devonian, the drowned Hönne Valley Reef Complex at the eastern end of the Remscheid-Altena Anticline created a subphotic palaeohigh (e.g. Becker et al. 2016b), which descended gently from the Oberrödinghausen region to the south (e.g. Wedekind 1914b). There, Frasnian submarine volcanoes resulted in a more complex "seamount" pattern around Balve (Dornsiepen 1973;Becker et al. 2016b). The Borkewehr section is positioned on the southeastern slope of the overall palaeohigh.

Material and methods
This study is based on a new, continuous bed-by-bed logging, with new records of conodonts, ammonoids, and geochemistry. For orientation, one conodont sample was taken from the middle Famennian base of the succession. Dieter Weyer collected ammonoids and solitary rugose corals from the last 21 nodule levels of the Wocklum Limestone, which was supplemented by collections of R. T. Becker and S. Hartenfels from the last five layers and from the first limestones above the HS/ HSS equivalents. We took 12 conodont samples from the top Wocklum Limestone and sampled 26 of 29 possible micritic beds of the Stockum and Hangenberg limestones for conodonts, other microfauna, and carbonate microfacies. In the critical DCB interval, beds were re-sampled up to six times in order to establish full taxon ranges and to evaluate the initial conodont radiation with high-resolution. A medium weight of three kilograms for each sample was dissolved by a 10 % solution of formic acid. Washed residues were separated into 0.100, 0.315, and 0.630 mm fractions. If necessary, the smallest one was treated before picking by heavy liquid separation, using diluted sodium polytungstate (3Na 2 WO 4 × 9WO 3 × H 2 O). As it is common practice in high-resolution biostratigraphic conodont studies, only the Pa element taxonomy has been utilised, because many Famennian and Mississippian multi-element reconstructions are still doubtful. Ramiform elements were picked but not identified. The samples provided sparse to rich conodont faunas, which enabled the location of almost all zonal boundaries and, most importantly, of the evolutionary succession within the genus Protognathodus. Results are used to improve the regional bio-and event stratigraphy.
Using 50 × 50 and/or 100 × 75 mm thin sections, carbonate microfacies analysis and classification follow the nomenclature by Dunham (1962) as well as the modified microfacies types of Hartenfels (2011). The latter were introduced, because the standard microfacies types sensu Wilson (1975), developed for Upper Triassic reefal carbonates, do not permit a differentiated classification of (hemi)pelagic carbonates. Even the refined model by Flügel (2004) is too rough (compare discussion in Hartenfels 2011). It should be stressed that the local microfacies of all Wocklum, Stockum, and Hangenberg Limestone beds give no hints for any reworking processes Hartenfels et al. 2017a;Herbig et al. 2019).
Calcite powders were drilled from 42 freshly cut carbonate rock samples of Wocklum, Stockum, and Hangenberg limestones for carbon and oxygen isotope analyses. Powders were reacted with phosphoric acid at 70°C and the generated CO 2 gas was analysed using a MAT 253 mass spectrometer at the Nanjing Institute of Geology and Palaeontology, Chinese Academy of Sciences. For δ 13 C org studies, a total of 65 samples, including carbonate rock samples mentioned above, and 23 more samples from the HBS and HS/HSS intervals were ultrasonicated in deionized water to remove surficial organic contaminants. The dried samples were crushed to 200 mesh and acidified using 3N HCl. Insoluble residuals were repeatedly rinsed in ultrapure water, dried at 60°C, and then loaded into tin capsules. Samples were measured using a MAT 253 mass spectrometer, coupled to a high-temperature element analyser at the Nanjing Institute of Geology and Mineral Resources. All isotope ratios presented here are expressed by δ-notation as per mill (‰) relative to the Vienna Peedee Belemnite standard (VPDB). Accuracy and precision were controlled by replicate measurements of Chinese National Standard GBW-04405 (δ 13 C = +0.57 ‰, δ 18 O = -8.49 ‰) and international standard IAEA-600 (δ 13 C = -27.8 ‰), respectively, and were better than 0.04 ‰ for δ 13 C carb , 0.08 ‰ for δ 18 O carb , and 0.1 ‰ for δ 13 C org . Element composition of powdered samples was determined by inductively coupled plasma optical emission spectrometry/mass spectrometry (ICP-OES/MS; n = 35, commercial laboratory ACME, Bureau Veritas Commodities Canada Ltd laboratories, Vancouver; details of the procedures used are available at http://acmelab.com) and by energydispersive X-ray fluorescence spectrometry [ED-XRF; n = 184; MiniPal 4.0 (PANalytical, Netherlands) and a Rh lamp and Peltier cooled Si-PIN detector at the Institute of Inorganic Chemistry, Czech Academy of Sciences, Prague]. Powders for ICP-OES/MS and ED-XRF analyses were not prepared from the same parts of the samples. ICP-OES/MS provided data on concentrations of a wide spectrum of elements with high accuracy and low limits of detection (Online Resource 1), which were used for the calculation of various geochemical proxies. ED-XRF was used to get high-resolution data on selected major and trace elements, expressed in the counts per second (c. p. s.) unit (Online Resource 1).
Framboid pyrite diameters were measured by a JEOL 6490 LV scanning electron microscope at the Department of Geological Sciences, Masaryk University, Brno, Czech Republic. Only one polished thin section taken from the Hangenberg Black Shale contained altered framboid pyrites, which were analysed in backscattered electron mode. Based on Wignall and Newton (1998), more than 100 framboid pyrites were measured (n = 102). Basic statistics of min, max, mean, standard deviation, and plots were carried out using the software Statistica 12 (Statsoft; Online Resource 1).
The cyclostratigraphic interpretation in this study is based on the frequency of alternations between shales/marls and nodular limestones. To obtain insight into the frequency composition of the Borkewehr lithological rhythms, the periodogram of a lithological depth-series was calculated. This quantified litholog was constructed by assigning a code "1" to shales/marls and "2" to (nodular) limestone beds. Frequency-selective bandpass filters with rectangular windows were used to isolate and extract specific frequency-components from the litholog that we associate with either precession or obliquity. All time-series analyses were carried out using the freely-available R-library astrochron (Meyers 2014).

Upper Famennian to lower Tournaisian conodont zonation
The upper to uppermost Famennian pelagic conodont zonation was introduced by Ziegler (1962) and revised by Sandberg (1984, 1990), Kaiser et al. (2009), Hartenfels (2011), Hartenfels and Becker (2016b, Spalletta et al. (2017Spalletta et al. ( , 2020, and Corradini et al. (2020 online). Based on a phylogenetic reconstruction of the genus Siphonodella, Sandberg et al. (1978) established the lower Tournaisian siphonodellid zonation. Significant changes were introduced by Ji (1985) which were followed by Korn and Weyer (2003) and by Becker et al. (2021). A compilation and correlation of the different Tournaisian zonal schemes was given in Esteban Lopez (2021). Becker et al. (2016a) gave an overview of the DCB conodont stratigraphy, including references to shallow-water successions. These are relevant for wide regions outside Germany. Subsequently, Corradini et al. (2016 online) and Spalletta et al. (2017) installed a new zonal scheme, which is strictly based on the FADs of taxa. They decided against the Si. (Eo.) praesulcata Zone and created an expanded B. ultimus Zone, based on uncertainties stemming from the thorough analysis of the Si. (Eo.) praesulcata -Si. (Eo.) sulcata Group in Kaiser and Corradini (2011). As an upper subdivision of their B. ultimus Zone, Spalletta et al. (2020 online, following the suggestion of Corradini et al. 2020 online) introduced a Pr. meischneri Subzone in the Carnic Alps. However, in many other regions the species is very rare in the uppermost Famennian and commonly shows a delayed entry. In general, the FAD of Pr. meischneri is not a reliable biostratigraphic tool. If the index taxon is absent or enters at a higher level, they proposed to mark the base of the subzone approximately by the first occurrences of Pr. collinsoni or Si. (Eo.) praesulcata. But the FAD of Pr. collinsoni lies significantly above the FAD of Pr. meischneri. Despite the taxonomic complexity of early Siphonodella and its relativessee the concept of early Siphonodella-type conodonts by Tragelehn (2010;called siphonodelloids in Becker et al. 2013)the Si. (Eo.) praesulcata Zone has been recorded around the world. Consequently, Söte et al. (2017) proposed to keep the Si. (Eo.) praesulcata Zone. Becker et al. (2021) recognised the current nomenclatorial uncertainty by referring to a Si. (Eo.) praesulcata s.l. Zone, especially since the Montana Si. (Eo.) praesulcata holotype is not from Wocklum Limestone time equivalents but from a level well above the Hangenberg extinction (Rice et al. 2017). Corradini et al. (2016 online) as well as Spalletta et al. (2017) decided also against the costatus-kockeli Interregnum (ckI) sensu Kaiser et al. (2009), which base marks the main conodont extinction at the base of the Lower Crisis Interval. As pointed out by Kaiser et al. (2009), the main victims include important index taxa as well as dominant and "cosmopolitan" forms, such as B. bispathodus, both morphotypes of B. costatus and B. ultimus ultimus, Br. disparilis, Neo. communis collinsoni, subspecies of the Pa. gracilis Group, Po. vogesi, and Ps. marburgensis trigonicus (compare range charts in Hartenfels 2011 andSpalletta et al. 2017). Thus, it is important to mark the main extinction level and the base of the Hangenberg Black Shale Event and its equivalents (Kaiser et al. 2015 online) in the conodont scale. Söte et al. (2017) emphasised that no biostratigraphic rule discourages the use of Last Appearance Datum (LAD)-defined conodont zones. Therefore, they proposed to keep both intervals, the Si. (Eo.) praesulcata Zone as well as the ckI, but other authors may prefer to recognise these only as subzones within the expended B. ultimus Zone sensu Corradini et al. (2016). It has to be emphasised that the global conodont record from the Lower Crisis Interval is extremely poor. In the Rhenish Massif, attemps to find conodonts on bedding surfaces or by physical/chemical cracking of the Hangenberg Black Shale failed, so far. The hypoxic to anoxic environment was apparently hostile to conodonts. However, as discussed by Corradini and Spalletta (2018), the current record gap may mask the true details of extinction and short-term survival of taxa.
Through the DCB interval, the genus Protognathodus is represented by four species: Pr. meischneri, Pr. collinsoni, Pr. kockeli, and Pr. kuehni. The biostratigraphic potential of these protognathids has been recognised by Bischoff (1957), Collinson et al. (1962), Weyer (1965, Ziegler (1969), and Alberti et al. (1974). Subsequently, Ziegler and Sandberg (1984) defined their Upper praesulcata Zone by the FAD of Pr. kockeli. This led to a re-naming as Pr. kockeli Zone by Kaiser et al. (2009). Corradini et al. (2011) pointed out that the latter is the most abundant species amongst the protognathids, with a wide geographic distribution, but in many regions with an alleged variation of the First Occurrence Datum (FOD). Consequently, they concluded that "none of the four Protognathodus species has a high potential as a tool for redefining the D/C boundary interval". In the following years, detailed (re-)investigations of DCB sections throughout the world improved the knowledge and provided new insights into the global distribution and regional first occurrences of protognathids (e.g. Becker et al. 2013(e.g. Becker et al. , 2021Corradini et al. 2013Corradini et al. , 2016Corradini et al. online, 2020Mossoni et al. 2015;Aretz et al. 2016Narkiewicz et al. 2017;Hartenfels et al. 2017a, b;Kumpan et al. 2018Kumpan et al. , 2019bKumpan et al. , 2020Feist et al. 2019Feist et al. , 2020Kaiser et al. 2019a, b;Qie 2019;Spalletta et al. 2019Spalletta et al. , 2020Over 2020 online;Denayer et al. 2020 online;Matyja et al. 2020 online;Kulagina et al. 2021), especially of Pr. kockeli. Partly based on these new datasets, Corradini et al. (2016) and Spalletta et al. (2017) changed their point of view and installed an expanded Pr. kockeli Zone, which includes both the Pr. kockeli Zone sensu Kaiser et al. (2009) and the following Si. (Eo.) sulcata Zone. It is more than unfortunate that different authors will mean a different time interval hidden under a single zonal name, which has been used since many years. The Pr. kockeli Zone sensu Kaiser et al. (2009) refers strictly to the "classical interval" (Upper praesulcata Zone) between the FADs of Pr. kockeli and Si. (Eo.) sulcata s.l. The expanded Pr. kockeli Zone sensu Corradini et al. (2016 online) and Spalletta et al. (2017) also embraces the Upper Stockum Limestone and lower portions of the Hangenberg Limestone, reaching upwards until the entry of Si. (Eo.) bransoni.
The transition from Pr. collinsoni to Pr. kockeli is more complex than previously thought (compare Kaiser et al. 2019a online). Protognathodus collinsoni is characterised by one or a few irregularly distributed nodes on the upper surface of the cup. The emended (expanded) Pr. kockeli diagnosis of Corradini et al. (2011: "The coarse, nodose, ornamentation and the presence of at least one row of nodes parallel to the carina on one half of the cup …") contains also intermediates between Pr. collinsoni and Pr. kockeli sensu Bischoff (1957, whose original diagnosis translates as: "… with one or two lines of coarse nodes parallel to the carina on the inner AND outer side of the cup"). Investigations of the Borkewehr section and other localities (e.g. Puech de la Suque, Montagne Noire, France; locality description and access in Kaiser 2005 andFeist et al. 2020 online) prove that "intermediate kockeli" types with a row of nodes on only one sideherein designated as Pr. semikockeli n. sp. (see taxonomic chapter)may occur in the first limestone bed immediately above the conodont-free Hangenberg Shale/Sandstone. Feist et al. (2020) illustrated a specimen from the Middle Crisis Interval of the Montagne Noire. If it would be included in Pr. kockeli, using the Corradini et al. (2011) re-definition, the Pr. kockeli range would have to be significantly expanded downwards, into a level that has very few conodont faunas on a global scale. The holotype and originally illustrated paratypes of Pr. kockeli have two rows of nodes (second outer rows partly initially developed) on one or both sides of the cup (however, one row of nodes occurs in any case on each side). In this sense, Pr. kockeli s.str. enters in the third Stockum Limestone bed (Bed 4b) at Borkewehr .
Apart from the past inconsistencies in discriminating Si. (Eo.) sulcata from its supposed ancestor Si. (Eo.) praesulcata (compare Kaiser and Corradini 2011), we think that Si. (Eo.) sulcata is a useful zonal marker, if it is used in the sense of the lost type specimen sensu Huddle (1934). The latter came from the basal Tournaisian Henryville Bed within the Ellsworth Member of the upper New Albany Shale (Illinois Basin, southern Indiana; Evans et al. 2013). Kaiser et al. (2019a online) emphasised that it is Morphotype 5 sensu Kaiser and Corradini (2011) that should be used for the definition of a Si. (Eo.) sulcata Zone. It is long known that Siphonodella is almost lacking in the Stockum Limestone, where the FAD of Pr. kuehni provided a distinction of Lower and Upper Protognathodus faunas (Alberti et al. 1974). Later it was shown, that the FADs of Pr. kuehni and Si. (Eo.) sulcata coincide (e.g. Clausen et al. 1989a). Further support based on Austrian sections led Kaiser et al. (2019a online) to advocate a combined Si. (Eo.) sulcata/Pr. kuehni Zone. The taxonomic uncertainty of Si. (Eo.) sulcata led Becker et al. (2021) to apply a Si. (Eo.) sulcata s.l./Pr. kuehni Zone until revisions have been completed.
The subsequent Tournaisian conodont zonation of Sandberg et al. (1978) has been refined by  and discussed by Becker et al. (2016a). The latter re-named and designated accidently two specimens of Si. (Si.) hassi  using the originals of Si. duplicata sensu Hass (1959: pl. 49, figs. 17-18) as the types of Si. (Si.) jii, because Ji (1985) overlooked that the species name hassi was already occupied by Si. cooperi hassi Thompson and Fellows, 1970. Consequently, they re-named the Si. (Si.) hassi Zone sensu Ji (1985 = former Upper Si. duplicata Zone) as Si. (Si.) jii Zone. Zhuravlev and Plotitsyn (2018) re-illustrated the type material of Si. (Si.) jii and described initial second rostral ridges, at least on the inner anterior platform. Therefore, they assigned the types of Si. (Si.) jii to Si. (Si.) quadruplicata (compare taxonomic assignment by Klapper 1966). Referring to the emended quadruplicata diagnosis by Klapper (1971), the number of rostral ridges of the latter varies from three to five. Becker et al. (2021) selected the adult original of Hass (1959: pl. 49, fig. 18, USNM 115140) as the new lectotype, which is less advanced than typical Si. (Si.) quadruplicata. They noted that the specimen can be regarded as an early morphotype (named Morphotype 2) of Si. (Si.) quadruplicata or it may be an ancestral form of Si. (Si.) quadruplicata s.str. The specimen of "Si. (Si.) hassi" figured in  fig. 14; pl. 2, figs. 5-6) is lacking initial secondary rostral ridges and resembles the middle Tournaisian Si. (Si.) wilberti Bardasheva, Bardashev, Weddige, and Ziegler, 2004 (pl. 4, figs. 12-13). However, the holotype of the latter seems to be an aberrant pathological specimen displaying a peculiar, undulating, and serrate outer platform corner. In aboral view, the element shows an offset towards the inside at this point, which "interrupts" the platform. Therefore, specimens which agree with the Chinese "Si. (Si.) hassi" are listed as Si. (Si.) cf. wilberti.

Sedimentary succession
The used lithostratigraphic terminology follows the Stratigraphic Table of

Wocklum Limestone
The Wocklum Limestone at Borkewehr consists of a cyclic alternation of greenish-grey shales/marls and grey nodular limestones or nodular shales 3). Reddish nodular shales and limestones occur below, but are a part of a folded and faulted interval. Typical for the Wocklum Limestone is a pelagic fauna yielding entomozoan ostracods, ammonoids, orthoconic cephalopods, agglutinating foraminifera, thinshelled bivalves, small-sized gastropods, bellerophontid gastropods, small chonetids, blind and oculated trilobites, solitary rugose corals, as well as conodonts. Bioturbation is evident in the micritic and matrix-rich limestones, which consist of microsparitic mud-wackestones to wackestones through the entire interval ( Fig. 4a-d, f). Occasionally, reddish-brown cauliflower structures encrusted skeletal remains (e.g. in beds -9b, -7b, and -3b). They resemble Frutexites-type encrustations sensu Böhm and Brachert (1993) and were previously mentioned from the upper to uppermost Famennian (e.g. Hartenfels 2003(e.g. Hartenfels , 2011Préat et al. 2008;Hartenfels and Becker 2016a, b). Like the benthic foraminifera, they indicate very slow sedimentation rates. Thus, the Wocklum Limestone accumulated under oxic conditions, below the photic zone, and below the influence of storm waves. A single thin layer of small-sized fragmented crinoid packstone intercalated within a mud-wackestone facies indicates an episode of increased  bottom turbulence (Bed -12b, Fig. 4e). The crinoid debris must have been derived from more elevated parts of the drowned seamount. The increase in pyrite content and the simultaneous decrease in bioturbation within the last five limestone beds (-4b to -1b, Fig. 4c to 4a), immediately below the Hangenberg Black Shale, argues for a successive deterioration of oxygen availability and, therefore, of the ecological conditions for benthos on the sea-floor. Especially within Bed -1b (Fig. 4a), the content of disperse pyrite increased rapidly. Thus, the facies change between the cyclic sedimentation and the overlying Lower Crisis Interval was somewhat gradual, as at Drewer ). In this respect, the Crisis Prelude of Borkewehr shows analogies with the older, smaller-scale Annulata Events and Dasberg Crisis in the Rhenish Massif (compare Hartenfels and Hartenfels 2011).
Based on shallowing upwards, strong condensation, and non-deposition, upper to uppermost portions of the Wocklum Limestone are lacking in some neighbouring sections of the northern Rhenish Massif. At the mostly overgrown Reigern Quarry (GPS 51°23' 15.77" N, 007°59' 25.20" E) near Hachen, type locality of the conodont B. ultimus corradinii Söte, Hartenfels, and Becker, 2017 and famous for its rich Famennian ammonoid fauna, Söte et al. (2017) mentioned a major hiatus, which spans from the topmost Famennian to the upper Tournaisian. There, the middle parts of the Wocklum Limestone are disconformably overlain by laminated radiolarites and a thin metabentonite of the Mississippian Hardt Formation.

Hangenberg Black Shale and Hangenberg Shale/Sandstone equivalents
In the two lateral Borkewehr successions, the Wocklum Limestone is sharply overlain by 20 cm of very fissile, Hangenberg Black Shale (Bed 0,3). The base marks the pelagic mass-extinction level and Lower Crisis Interval I sensu Becker et al. (2021). The fine lamination suggests the lack of bioturbation due to anoxic seafloor conditions. For a review of extinction patterns and interpretation of associated palaeoclimatic and palaeoceanographic changes see Kaiser et al. (2015).
The Hangenberg Black Shale enabled a spread of pelagic facies westwards into neritic environments of the Velbert region (Becker et al. 2021) and further to the Ardennes (e.g. supposed Hangenberg Black Shale equivalent at Spontin, southeast of Namur, Denayer et al. 2020 online).
The Hangenberg Black Shale is followed by approximately 80 cm of moderately solid, greenish-grey, silty, unfossiliferous, and micaceous Hangenberg Shale/Sandstone equivalents (beds 1a-b, combined Middle Crisis Interval I-II sensu Becker et al. 2021, Figs. 2a, c, 3). In contrast to DCB localities to the northwest (e.g. Oese), north, and east, the Hangenberg Sandstone is inconspicuous and indistinguishable from the underlying Hangenberg Shale. This may reflect the increasing distance from the depocentre and northern clastic source Kołtonik et al. , 2019. This implies that the siliciclastics of Stockum, which lies to the southeast, derived from a different shedding path. The Middle Hangenberg Crisis Interval covers the major, glacioeustatic sea level fall, which is indicated by increasing Zr/Al values (see below). An absolute peak value lies approximately 45 cm below the top of this unit and is correlated by us with the base of the Hangenberg Sandstone.

Stockum and Hangenberg limestones
The overlying top 3 cm (Bed 1c) are dark-grey, marly, and represent a renewed short episode of limited oxygenation. This interval may correlate with the "Stockum Level Black Shale" (Becker et al. 2021) above the top of the Hangenberg Sandstone, described by Korn (1991) and Korn et al. (1994) from the poorly accessible western quarry wall at Drewer. It marks the gradual, initial postglacial transgression, the base of the Upper Crisis Interval.
The 11 cm thick beds 2-3 represent locally the main Upper Crisis Interval I and Basal Stockum Limestone sensu Becker et al. (2021, Figs. 2d, 3). In comparison with other Sauerland sections, it is a somewhat unusual interval of platy, thinbedded ("laminated"), dark-grey limestones with squashed goniatite impressions on bedding plains. Microsparitic mudstones are intercalated by darker layers, enriched in C org , and interrupted by thin-bedded siltstones of lighter colour ( Fig.  5e-f). Whereas the darker layers indicate an increased deposition of fine clay and organic matter, probably due to episodes of higher primary production, the siltstones probably resulted from distal turbidites and, therefore, final minor pulses of Hangenberg Sandstone sedimentation. In contrast to the conodont-free Bed 2, Bed 3 contains a low-diversity conodont fauna, a few ostracods, and thin-shelled bivalves.
The first nodular limestone (Bed 4b) represents locally the Lower Stockum Limestone (Upper Crisis Interval II sensu Becker et al. 2021, Figs. 2a, d, 3). It is a microsparitic mudstone ( Fig. 5d) with ostracods, thin-shelled bivalves, and trilobites. Clausen in Luppold et al. (1994, p. 56) mentioned silt-sized quartz grains and calcispheres. Bioturbation indicates the return to better oxygenated conditions. The microfacies of Bed 5b is similar to the preceding Bed 4b.

Middle Tournaisian to Viséan
The uppermost carbonate layers of the Hangenberg Limestone are overlain by the middle Tournaisian Kahlenberg Formation ("Liegende Alaunschiefer", "Lower Alum Shales"). The fissile, laminated, unfossiliferous, dark-grey to blackish shales, exposed at the right (southeastern) side of the quarry, show a very sharp base (Figs. 2e, 7a, b). The strongly condensed, only 56 cm thick formation (Bed 31, Fig. 8), which in most sections along the northern margin of the Rhenish Massif measures about ten metres, is also somewhat lighter coloured than in standard facies; the intensively black-coloured alum shales are missing.
Most interestingly, the conspicuous black bedded cherts and siliceous shales of the upper Tournaisian Hardt Formation ("Schwarze Kieselschiefer") are missing due to a hiatus. This is also observed in sections at the western basin margin and on intrabasinal swells, for example in the Drewer Quarry in the northeastern Rhenish Massif (Herbig 2016;Herbig et al. 2019).
Above, a 51.5 cm thick, unnamed unit crops out (beds 32a-35), which is a stratigraphic equivalent of the Erdbach Limestone II of Krebs (1966Krebs ( , 1968) and of two further formations in the Rhenish Massif (Kohleiche Formation, Korn 2005a; Kattensiepen Formation, Korn 2005b) as mentioned by Herbig et al. (2019, Fig. 7a). At the base, a 15 cm thick dark shale horizon (Bed 32a) enriched in phosphorite nodules and up to 8 cm thick and up to few decimetres long limestone lenses contain phosphorite pebbles, as already indicated by Voges (1959: fig.1; 1960: fig. 2; Fig. 7c). The phosphorite nodule horizon indicates reworking. Therefore, despite of the lithological similarity, it cannot be included in the dark shales of the underlying Kahlenberg Formation. This is also shown by its younger, uppermost Tournaisian conodont fauna (see below). The upper 36.5 cm of the unit are a true lithological equivalent of the Erdbach Limestone II. Already Schmidt (1924, p. 105, horizon f of the described section) and later Voges (1959Voges ( , 1960 attributed this unit to the "Erdbacher Schichten", respectively "Erdbach Horizon". It consists of two decimetric limestone beds (beds 32b and 34b = beds K1 and K2 in Großhäuser 2014) separated by a 5 cm thick shale interval (beds 33a to 34a) that includes an intervening 2 cm thick limestone veneer (Bed 33b, Fig. 8). Beds 32b and 34b are moderately bioturbated microbioclastic-microintraclastic wackestone/packstone (Fig. 9a). Only the intervening limestone veneer (Bed 33b) shows lamination. Besides differing packing density of the components and bioturbation, differences are faint. Fossils are restricted to some conodonts, ostracods, diminuitive echinoderm debris, and spheres, which at least in part, are radiolarian ghost structures. Extremely rare are the calcareous microbiota Archaesphaera, Radiosphaera, Diplosphaerina, and Eotuberitina. Multichambered foraminifera and calcareous algae are missing. Small calcite crystals occur, which in part are dissoluted bioclasts, in part recrystallized matrix. The 17 cm thick upper bed (Bed 34b) is bipartite. A siltitic mudstone at its base is bored (Fig. 9b). Above, renewed sedimentation and bottom currents caused erosion of some well-rounded clasts that float in identical siltitic mudstone as at the base of the bed (Fig. 9b). In general, microfacies indicates deposition in deeper water below storm wave base. Sedimentation processes, however, remain unclear. Involvement of gravitative (turbiditic) redeposition, indicated by dense packing, strong grain diminution and sorting, as well as by relictic lamination seems reasonable. Components indicate a deeper water source. The 2.5 cm thick top (Bed 35) of the upper limestone presumably is a nodular, light-weathering hardground (Herbig et al. 2019).
The microfacies of these calciturbidite beds is a predominantly fine-grained, diagenetically flasered peloidal-bioclastic packstone as a result of intense pressure solution. Therefore, in many cases only small stringers and lenses of the primary facies were preserved, which on first sight are easily confused with intraclasts (Fig. 9e). Calcispheres predominate. Wellpreserved primary facies within an early diagenetic nodule shows peloidal-bioclastic grainstone with some foraminifera, echinoderm debris, and brachiopod shells. Algae are missing except for rare kamaenids. Peloids have been derived from small intraclasts, micritized bioclasts, and some ooids. Thus, in contrast to the lower calciturbidite set that was derived from a pelagic deeper water source, these calciturbidites reworked shallow-water platform material. Besides Radiosphaera sp. (Fig. 9e), Calcisphaera pachysphaerica (Pronina, 1963), Priscella cf. prisca (Rauser-Chernoussova and Reitlinger in Rauser-Chernoussova et al. 1936;Fig. 9f), and Pseudotaxis sp., only Uralodiscus rotundus (Chernysheva, 1948, Fig. 9g) were determined with confidence. An oblique section of an archaediscid might belong to either Glomodiscus sp. or 'Archaediscus' at involutus stage (Fig. 9f). In general, the assemblage suggests the middle Moliniacian MFZ 11α or MFZ 11β Zone of Cózar et al. (2020), respectively the top of Cf4β or Cf4γ (Conil et al. 1991). In the review of Herbig (2006), oldest foraminifera from the Hillershausen Formation ("Kieselkalk and equivalents") were mentioned from the Cf4δ. Thus, this might be the oldest foraminifer faunula hitherto recorded from the Hillershausen Formation, which correlates well with the sample position from the basal part of the formation.

Sequence stratigraphy
The sequence stratigraphic interpretation of the section follows Herbig (2016), who first established sequences for the uppermost Devonian and Mississippian of the Rhenish Kulm Basin and correlated them with the sequences of the Northwest European shallow water platform (Hance et al. 2001(Hance et al. , 2002Poty et al. 2014; see also Poty 2016).
The lowermost sequence boundary in the Rhenish Kulm Basins discussed by Herbig (2016) coincides with the top of the Wocklum Limestone. The uppermost part is considered to record a falling systems stage tract (FSST) due to condensation and concomitant thinning of calcareous beds, increased faunal content as described in many sections by Korn (1995), Becker (1996), Korn and Weyer (2003), and Becker et al. (2016aBecker et al. ( , 2021 and by increased Zr/Al values (Kumpan et al. 2015, and herein). In the Borkewehr section, the inconspicuous FSST could be tentatively related to the interval above Bed -9b (Figs. 2b, 3). It has to be stressed that the gradual deterioration of ecological conditions in the uppermost beds of the Wocklum Limestone that are interpreted to be a prelude of the overlying Hangenberg Black Shale, point to an extraordinary complete sedimentary succession.
A relictic sequence is recorded by the Hangenberg Black Shale (Bed 0, transgressive systems tract/maximum flooding surface, TST/mfs) and the lower part of the overlying Hangenberg Shale/Sandstone equivalents that record the highstand systems tract (HST, Fig. 2c) or initial regression (Kaiser et al. 2015 online;Becker et al. 2016a; references therein). Considerable non-deposition or even erosion, as seen elsewhere by incised valley fills, e.g. some 20 km further northwest by the famous Seiler conglomerate, have to be assumed for this sequence. This is stressed by the fact that Kaiser et al. (2015) estimated in southern Morocco a relative sea level fall in the scale of more than 100 m before deposition of Hangenberg Sandstone equivalents. The continuous finegrained siliciclastic facies development of the Borkewehr section does not show any macroscopic impact of such a change.
However, Zr/Al values indicate an absolute peak some centimetres above the base of Bed 1b (see below). In comparison with other sections, this level is correlated with the base of the Hangenberg Sandstone, which was addressed as the lowstand systems tract (LST) of Sequence 1 in the Rhenish Kulm Basin by Herbig (2016). In the Borkewehr section, the onset of the TST is seen in the dark grey marly Bed 1c at the top of the Hangenberg Shale/Sandstone equivalents (Fig. 3), which might be an equivalent of the "Stockum Level Black Shale" of Becker et al. (2021; see above). The following Stockum and Hangenberg limestones are regarded as the undifferentiated TST/HST of Sequence 1 (Fig. 2d).
In regions not reached by sandstone shedding, Herbig (2016) placed the base of Sequence 1 at the base of the Hangenberg Shale. Though this is a pragmatic approach, geochemical results from the Borkewehr section show that the sequence boundary has to be seeked inside the Hangenberg Shale succession. However, in general Herbig (2016) followed the earlier model of van Steenwinkel (1993b) that was developed by comparison with the Belgian Namur-Dinant Basin (van Steenwinkel 1990Steenwinkel , 1993a and similar models for the latter by Kumpan et al. (2014) and Bábek et al. (2016a). On the contrary, Hance et al. (2001Hance et al. ( , 2002 and later publications of the Liège working group placed the base of the Belgian Sequence 1 below, at the base of the mixed siliciclastic-carbonate succession of the Etrouengt Formation and its lateral equivalents (Comblain-au-Pont and Dolhain formations). Poty (2016), who also gave a thorough review of the development of Dinantian sequence stratigraphy in Northwest Europe, and particularily in the Namur-Dinant Basin, interpreted the Hangenberg Sandstone interval and the concomitant sea level fall not as a third-order sequence boundary close to the DCB, but as an "out-of-sequence event". This was based on the identical pattern of precession cycles below and above the Hangenberg Sandstone interval, and further arguments forwarded by Hance et al. (2001) and Denayer et al. (2015Denayer et al. ( , 2019Denayer et al. ( , 2020, i.e. very similar facies across the DCB, which hinders separation into different system tracts, and homogeneous facies and wide extension of the middle Hastière Formation, which typify a HST. The dark-grey to blackish shales of the Kahlenberg Formation correspond to the TST of Sequence 2 (Figs. 7a, 8). By means of facies and biostratigraphy, the formation is well correlated with the Belgian Pont d'Arcole Formation, which corresponds to the same sequence stratigraphic tract. For the Rhenish Massif, Herbig (2016) showed the unconformable contact at the base and disproved earlier interpretations of a single sequence consisting of the Hangenberg beds below (LST) and the Kahlenberg Formation above (TST; van Steenwinkel 1993b) or of a sequence even continuing into the uppermost Tournaisian, including the Richrath Limestone (HST; Siegmund et al. 2002). In the Borkewehr section, a hiatus at the base of the Kahlenberg Formation is clearly shown by the sharp contact with the underlying Hangenberg Limestone (Figs. 2e, 7a). Like elsewhere in the Rhenish Massif, the LST of Sequence 2 is not recognised.
The upper Hastarian-lower Ivorian Sequence 3, elsewhere in the Rhenish Massif represented by the black siliceous shales and bedded cherts of the Hardt Formation, is missing. This corresponds to the development in the Velbert Anticline at the western margin of the Kulm Basin (Herbig 2016). Also, on the intrabasinal deep Drewer Swell the Hardt Formation is missing, though a very thin Sequence 3 was postulated by Herbig (2016).
The LST/TST of Sequence 4 is well documented by a thin, dark horizon with enriched phosphorite nodules and phosphorite pebbles (Bed 32a) at the base of the thin limestone package Figs. 7a,c,8), which are equivalents of the Erdbach Limestone II. The limestones represent the HST of Sequence 4, which ends with a questionable hardground. However, condensation and omission are already indicated by a bored horizon in the limestone bed below (Fig. 9b). Again, the succession is closely comparable to the western reaches of the Rhenish Kulm Basin. In the Herzkamp Syncline sections at Kohleiche and Riescheid, the Kohleiche Formation starts with a thin horizon bearing detrital quartz, phosphorite nodules, and phosphorite clasts that is overlain by a limestone horizon. Similar limestones are also known from other intrabasinal swells in the Kulm Basin, e.g. from the Warstein and Belecke anticlines (Kattensiepen Formation) or from the southernmost Dill Syncline (Erdbach Limestone II on top of the drowned Givetian-Frasnian Langenaubach-Breitscheid Reef).
reworking of deeper water, slope-derived material and, thus, still relatively low sea levels ( Fig. 9c-d). This is a corroboration of preliminary observations by Herbig et al. ( , 2014 from the Zippenhaus Member of the Heiligenhaus Formation, which is a lateral equivalent of the Hillershausen Formation, and also observations from other depositional systems (Reijmer and Everaars 1991;. Reworking of platform materialthis means further rising sea level and initial flooding of the platformis first documented in the upper calciturbidite set close to the top of the Borkewehr section ( Fig. 9e-g).

Depositional setting
The regional geological context, lithofacies, lithostratigraphy, and sequence stratigraphic interpretation enable a characterisation of the depositional setting of the extraordinary Borkewehr section.
In general, the section is situated on the southeastern slope of an overall palaeohigh related to the drowned lower Givetianlower/middle Frasnian Hönne Valley Reef Complex. Volcanic activity ("Schalstein volcanism") occurred during the uppermost Givetian in the surroundings of Balve, resulting in a 250 m thick, mostly tuffitic succession at the Husenberg, directly north of the Borkewehr section (Denckmann 1903;Paeckelmann 1938;Paeckelmann and Kühne 1938). Thus, an even more complex submarine relief ("seamount topography") was created that influenced later sedimentation (Dornsiepen 1973;Becker et al. 2016b). The complex palaeotopography of the region caused variously extended sedimentary breaks (e.g. Oberrödinghausen Railway Cut, Reigern Quarry, Becker et al. 2021) and special facies developments of the Famennian to Viséan successions (Schäfer 1975(Schäfer , 1978. For the Borkewehr section, sedimentation on a submarine high is evident by the well-developed Wocklum, Stockum, and Hangenberg limestones (Fig. 3). Continous pelagic carbonate input ("pelagic rain") resulted in a remarkably complete carbonate sequence, as demonstrated by the inconspicuous FSST and missing hiatus at the top of the Wocklum Limestone andmost importantby the complete conodont succession. Vice versa, the topographic high almost completely prevented coarse-grained clastic influx and suppressed the development of the Hangenberg Sandstone. Since the source of the sand was in the north (Kołtonik et al. , 2019, the Borkewehr section was in a more distal position than the typical Hangenberg Sandstone localities, such as Oese (compare Becker et al. 2021). Another phase of condensation started in the middle Tournaisian (Herbig et al. 2019). Already Paeckelmann (1938) described the presence of a lower Carboniferous swell facies. The Kahlenberg Formation is locally extremely thin (Fig. 8) and deviates from the typical black alum shale facies observed elsewhere in the Rhenish Massif. The upper Tournaisian siliceous shales and bedded cherts of the Hardt Formation are missing and typical deep swell limestones of the Erdbach type overlie the hiatus, but only as a thin veneer. Above, the exposed part of the Viséan Hillershausen Formation is remarkably poor in calciturbidites. This indicates either a missing well-developed source area on top of the Hönne Valley Reef Complex or, alternatively, turbidite currents that by-passed the swell according to the model first sketeched by Bender et al. (1977) for Frasnian calciturbidites in the northeastern Rhenish Massif.
The top part of the type Wocklum Limestone (beds -3b to -1b, Crisis Prelude sensu Becker et al. 2021) is especially rich in conodonts. The last nodular layers yielded a typical preextinction fauna. Whereas Caenodontus sp., Pa. gracilis semisigmoidalis (Fig. 10g), Pa. gracilis gracilis, Pa. gracilis sigmoidalis, Pa. gracilis expansa M1 (Fig. 10e), both morphotypes of B. costatus (Fig. 10a) and B. ultimus ultimus (Fig. 10d), B. bispathodus (Fig. 10b), B. muessenbergensis, Br. suprema (Fig. 10c), Ps. marburgensis trigonicus, and N. Gen. 2 n. sp. sensu Becker et al. (2013) Luppold et al. (1994) recorded, in addition, Po. inornatus (not figured) from the top Wocklum Limestone, but it could not be found in our samples. The extinction of B. costatus at the top of Bed -1b, the base of the Hangenberg Black Shale, defines the base of the ckI. It is a slightly higher level than the base of the former Middle praesulcata Zone, which was defined by the extinction of Pa. gracilis gonioclymeniae. Since the subspecies is lacking in our Wocklum Limestone samples at Borkewehr, the level of the former Middle preasulcata Zone is not recognisable locally. The Hangenberg Black Shale at Borkewehr is too fissile to search for bedding plane conodonts. We did not observe any specimens on partings of the Hangenberg Shale/Sandstone equivalents (beds 1a-b, Fig. 3).
As typical for pelagic settings, a Crisis Prelude Palmatolepid-Bispathodid Biofacies shifted locally into an Upper Crisis Interval Neopolygnathid-Protognathid Biofacies. Conodont samples from Bed 2, the basal Upper Crisis Interval I, were all barren. Bed 3 yielded Pr. meischneri, Pr. collinsoni, as well as Pr. semikockeli n. sp. (Fig. 11p). It documents the conodont radiation with continuing transgression in higher parts of Upper Crisis Interval I. Associated is Neo. communis communis (Fig. 11q), an opportunistic taxon (Söte et al. 2017). This assemblage marks the top of the costatus-kockeli Interregnum.
Based on the entry of Si. (Eo.) sulcata s.l. M5 and of a transitional form between morphotypes 4 and 5 (Fig. 12m, lacking the Morphotype 4 typical constriction in the anterior part of the platform, compare Kaiser and Corradini 2011), Bed 6b marks the base of the Si. (Eo.) sulcata s.l./Pr. kuehni Zone and, thus, the currently valid base of the Devonian/ Carboniferous Boundary. Associated are Neo. communis aff. dentatus sensu Luppold et al. (1984, Fig. 12s-t), Po. purus purus (Fig. 12q-r), and Ps. primus primus M2 (Fig. 11n-p). Bed 6b coincides locally with the base of the post-Crisis Interval and the base of the strongly cyclic Hangenberg Limestone. In terms of conodont biofacies, the local Upper Crisis Interval Neopolygnathid-Protognathid Biofacies shifted to a post-Crisis Siphonodella Biofacies. Two carbonate-marl cycles above, Po. purus subplanus first occurs, with a local delay, in Bed 8b, together with Ps. primus primus M3. The latter was recorded by Hartenfels and Becker (2016b online) from the Famennian of the Tafilalt, SE Morocco, where it links Ps. primus primus with the more weakly ornamented Ps. controversus.

Conodont diversity trends
Within the Wocklum Limestone, Bed -4b yields the most diverse fauna, with a total of 18 different taxa, including four morphotypes and the transitional form between Br. inornata and Br. suprema. The main conodont extinction took place at the base of the Hangenberg Black Shale, followed by the clastic interval without faunas. This local extinction rate of 57.1 % for the main Hangenberg Event lies in the lower range (55-72 %) of rates known from sections around the globe ). However, it lies above the total pelagic extinction rate, which is at 38 % (compare discussion in Hartenfels 2020).
The onset of the new conodont radiation is locally recorded in the second limestone bed (= upper part of the "laminated" limestones) above the Hangenberg Shale/Sandstone equivalents, at the top of the ckI (Fig. 3). Beds 2 and 3 are lithologically similar. Although sampling was repeated, Bed 2 remained barren. The first conodont-bearing Bed 3 yielded four taxa, including the entry of Pr. semikockeli n. sp. (Fig.  11p). Intensive re-sampling, six successive attempts based on comparable sample sizes, did not change the result. The temporary disappearance of eight taxa known from the Wocklum Limestone marks an episodic regional diversity reduction (Lazarus phase), triggered by the ecological changes of the Hangenberg Crisis. With continuing transgression in the basal Pr. kockeli Zone, the overlying Bed 4b (which was also re-sampled several times) provided 12 different taxa, including four FODs of species and subspecies, and four local Lazarus taxa ranged through. This tendency culminates in Bed 10 at the base of the Si. (Eo.) bransoni Zone. The contrast between 16 proven (sub)species versus three Lazarus taxa shows that the ecological conditions were now favourable for many conodont taxa. Nevertheless, it represents a turning point in the local conodont diversity. Starting from the immediately overlying Bed 11b at the top of the Si. (Eo.) bransoni Zone, the diversity decreases significantly. In contrast to the locally delayed entry of Pr. kuehni in our samples, there is a (local) loss of seven taxa. A total of six (sub)species is recorded, excluding seven local Lazarus taxa. Apart from some FODs and LODs and changes in the number of Lazarus taxa, the local palaeodiversity (between 14 and ten different taxa, including Lazarus taxa) remains fairly consistent until Bed 26b of the Si. (Si.) sandbergi Zone. At that level, there are seven (sub)species, excluding five Lazarus taxa. The conodont record in the subsequent carbonate layers decreases strongly. Whereas Bed 28b at the base of the Si. (Si.) lobata M1 Subzone yielded four conodont (sub)species and six Lazarus taxa, seven (sub)species are known from the topmost Hangenberg Limestone, Bed 30 (Fig. 3). This suggests a sudden local degradation of palaeoenvironmental conditions for conodonts, as a prelude to the subsequent Lower Alum Shale Event.
The lower part of the Wocklum Limestone is locally strongly affected by folding and faulting. Underlying variably reddish and greenish nodular limestones ("Foßley") yielded Wedekind (1914aWedekind ( , 1914b Clymenia laevigata, an index clymeniid of the Dasbergian (UD V). We found a large, poorly preserved Kalloclymenia in the last reddish interval, which indicates the lower Wocklumian (UD VI-A/B). It confirms a former record of Kallo. subarmata by Wedekind (1914a). The collection of H. Kaufmann includes a well-preserved specimen of Kallo. pessoides, with whorls that are slightly wider than high, but its precise level is not known.
The Hangenberg Black Shale at Borkewehr is very fissile and disintegrates in the field. This is probably the reason why the post-extinction Postclymenia evoluta Zone (UD VI-E) has not yet been recognised. Schmidt (1924) recorded three prionoceratid species that indicate the Ac. (St.) prorsum Zone (UD VI-F to LC A 1 ) -"Aganides infracarbonicus", "Ag. guerichi", and Ac. (Streeliceras) carinatum. Lange (1929) added Imitoceras subbilobatum, Im. cf. intermedium, and Im. substriatum. Without a re-examination of the types, it is x difficult to translate most of the old identifications into modern taxonomy. In addition, there are unsolved uncertainties whether Rhenish specimens are really conspecific with acutimitoceratid taxa described from Franconia (Kirchgattendorf), as it has been assumed in the past (e.g. Schmidt 1924;Vöhringer 1960; Korn 1984Korn , 1994, but without sufficient knowledge of the ontogenetic morphometry of populations. New poorly preserved specimens of Acutimitoceras (Stockumites) sp. were recovered from beds 2, 3, and 4b, which fall in the upper part of the Ac.  (Fig. 16), with typical, markedly biconvex growth lirae, constrictions, and relatively wide umbilicus, and a form close to Ac. (St.) intermedium were collected by H. Kaufmann from Bed 5b. The combination of old and new records shows that the Borkewehr section includes index goniatites around the revised DCB. Schmidt (1924) mentioned a juvenile G. subinvoluta. This is sparse evidence for the post-Crisis G. subinvoluta Zone and for the occurrence of goniatites in the Hangenberg Limestone. We found one poorly preserved Acutimitoceras (Stockumites) sp. in Bed 7b (compare Becker et al. 2016a). The thin section of Bed 21b cuts through a widely evolute juvenile goniatite, probably an Eocanites of the nodosus Group (Fig. 6e), and a ribbed goniatite, which points to a member of the Pseudarietitinae, according to the conodont level (basal Si. (Si.) mehli Zone; Fig. 3), possibly Paprothites. In the Rhenish Massif, there is practically no ammonoid record from the middle Tournaisian and only very few, poorly preserved specimens are known from the upper Kahlenberg Formation. Therefore, the discovery of a fragmentary, involute goniatite impression from dark-grey to blackish, slightly calcareous shales at the base of the Lower Alum Shale (Bed 31, Fig. 8) is exceptional. Seven, widely-spaced, convexly arches "ribs" may be the remnants of septa (wide dorsolateral saddles). A generic identification is not possible.

Coral faunas
Corals are extremely rare in the DCB beds of the Borkewehr section, as usual in sediments of the dysphotic/aphotic cephalopod facies, where rich anthozoan communities occur only exceptionally. Besides, they are usually ignored and remain uncollected. A special one-day-search (collection K. Bartzsch and D. Weyer 1992) resulted in 13 determinable Rugosa and two Heterocoralliamostly of dwarfish growthfrom the upper 0.85 m of the Wocklum Limestone (Fig. 17).
This uppermost Famennian fauna is characterised by Neaxon regulus (Richter, 1848, Bed -19, Fig. 18b-c), the best index fossil among corals, ranging from the Clymenia laevigata Zone to the Wocklumeria sphaeroides Zone, up to the Hangenberg extinction Event, in Thuringia (Korn and Weyer 2003: fig. 19). It is known in Europe from the Holy Cross Mountains (central Poland) through Germany to the Montagne Noire (southern France) and has also been discovered at five localities in the Anti-Atlas (southern Morocco).
Laccophyllum rozkowskae (Fedorowski, 2003) was found in the upper Famennian of the Holy Cross Mountains and at Dasberg near Hövel in the Rhenish Massif. Better preserved and analysed material is needed for a revision of this species, which might include Laccophyllum aff. rozkowskae (Bed -5b, Fig. 18a) from the Wocklum Stufe as the hitherto unknown juvenile stages. Unpublished upper/uppermost Famennian    A questionable Famennelasma sp. indet is represented by a crushed specimen of normal, large size (calice diameter ca. 20 mm, Bed -19, Fig. 17). This upper/uppermost Famennian genus was recorded from the Rhenish Massif and Thuringia and occurs also in Morocco. In Poland, it was described as Amplexizaphrentis conus Różkowska, 1969. Its relationship to the problematical genus Friedbergia Różkowska, 1969remains unclear (Weyer 1973bBerkowski 2002;Korn and Weyer 2003).
The two small fragments of the heterocoral Oligophylloides sp. indet (diameter 0.6-0.9 mm) are now usually interpreted as parts of destroyed colonies, indicated by budding in one specimen, determinable only at the generic level (Bed -4b, Fig. 18e). Middle to uppermost Famennian Oligophylloides are widely distributed in Poland, Germany, France, and especially in Morocco, with only one fragmentary specimen as questionable survivor of the global Hangenberg Crisis in the higher part of the Si. (Eo.) sulcata s.l./Pr. kuehni Zone at La Serre (Bed 93, Montagne Noire, Vachard 1988: pl. 2, fig. 1;Weyer 1995: fig. 5).
The genera Drewerelasma and Cyathaxonia are typical and dominant elements in the richer German coral faunas of the Gattendorfia Stufe (Saalfeld in Thuringia, Oberrödinghausen Railway Cut and Drewer in the Rhenish Massif).

Carbon isotopes
The Devonian-Carboniferous transition was associated with a significant perturbation to the global marine carbon cycle, as indicated by distinct positive excursions of both δ 13 C carb and δ 13 C org in the Hangenberg Crisis Interval worldwide (Brand et al. 2004;Kaiser et al. 2006Kaiser et al. , 2015Cramer et al. 2008;Kumpan et al. 2015Kumpan et al. , 2020Qie et al. 2015Qie et al. , 2021. The Hangenberg carbon isotope excursion was generally attributed to either abundant organic carbon burial during the Lower Crisis Interval (HBS and equivalents) or enhanced primary productivity in the Upper Crisis Interval (initial post-glacial transgressive interval), although the precise timing, pattern, and controlling mechanism remain debatable (see discussions in Kaiser et al. 2015 online andQie et al. 2015).
At Borkewehr, high-resolution δ 13 C carb records (n = 42) were obtained from the Wocklum, Stockum, and Hangenberg limestones at intervals between 1 to 12.5 cm, whereas the 103 cmthick HBS (Bed 0), HS/HSS (beds 1a-1b), and possible basal Stockum Level Black Shale (Bed 1c) equivalents yielded no δ 13 C carb data (Fig. 19). Meanwhile, two samples from beds -6a and 27 show low δ 13 C carb values of -0.3 ‰ and -2.5 ‰, respectively, which are best explained by diagenetic alteration, evidenced by abnormal δ 18 O carb values (Fig. 19). In the pre-Crisis Interval, the Wocklum Limestone records baseline δ 13 C carb values of 1.7 ‰ that decrease slightly to 1.4 ‰ at the top of the Si. (Eo.) praesulcata s.l. Zone (Fig. 19). Above the HBS and fine-grained siliciclastic successions, δ 13 C carb drifted toward higher values in the Upper Crisis Interval I and II (beds 2 to 6a), reaching a peak value of 2.2 ‰ in the Pr. kockeli Zone (Bed 4b), and representing the Hangenberg Crisis carbon isotope excursion in the Rhenish Massif. Upsection, a long-term gradual decrease to lower value (~0.2 ‰) in the lower to middle Tournaisian is observed, except for two minor positive shifts in the Si. (Eo.) bransoni Zone and basal Si. (Si.) sandbergi Zone (Fig. 19). A similar pattern of δ 13 C carb records was observed from other pelagic DCB successions in the Rhenish Massif (Hasselbachtal), Carnic Alps (Grüne Schneid, Kronhofgraben), and South China (Muhua II, Gedongguan, Daposhang), characterised by relatively constant δ 13 C carb values in the pre-Crisis Interval, a minor positive shift (generally < 1 ‰) in the Pr. kockeli Zone, and gradual decrease to~0 ‰ in the aftermath of the Hangenberg Crisis (Kaiser 2005;Qie et al. 2021). Global time-varying signals of δ 13 C carb allow for high-resolution regional and intercontinental correlation of the DCB intervals.
Unlike the widely studied δ 13 C carb excursion in the DCB successions, δ 13 C org records were merely reported in several localities from Canada, the Rhenish Massif, Carnic Alps, Poland, and Vietnam (Caplan et al. 1996;Kaiser 2005;De Vleeschouwer et al. 2013;Paschall et al. 2019). A global δ 13 C org trend has not been well established due to complex isotopic fractionation between dissolved CO 2 and primary biomass, organic matter source variation, secondary biological fractionation (e.g. heterotrophy), widespread sedimentary gaps, and lack of precise biostratigraphic control. In the Rhenish Massif and Carnic Alps, Kaiser et al. (2006Kaiser et al. ( , 2015 reported peak δ 13 C org values within the Hangenberg Crisis Interval, mainly in the Lower and Upper Crisis intervals and their equivalents, and suggested that the shifts of δ 13 C org values reflect enhanced burial of organic carbon during the major climate and environment perturbations around the Devonian-Carboniferous transition. At Borkewehr, δ 13 C org values range from -22.7 to -28.3 ‰, and the profile exhibits a major positive excursion from -25.4 ‰ to -22.8 ‰ in the HS/HSS equivalents (beds 1a and 1b) rather than in the Lower and Upper Crisis intervals (Fig. 19). δ 13 C org values decrease to -26.6 ‰ near the basal Upper Crisis Interval I (Bed 2), stay stable around -26.9 ‰ in the Upper Crisis Interval (top ckI to lower Si. (Eo.) sulcata s.l./Pr. kuehni Zone), then return to -25.1 ‰ in the Si. (Eo.) bransoni Zone, before recording a gradual decrease to -28.2 ‰ at the top of the Hangenberg Limestone. At Borkewehr, the δ 13 C org profile of the Hangenberg Crisis Interval is markedly different from trends in δ 13 C carb and the previous description of Kaiser et al. (2015) in this region, an indication that other factors (e.g. changes of organic matter compositions and atmospheric pCO 2 ) rather than changes of δ 13 C of oceanic dissolved inorganic carbon (DIC) are responsible for the decoupling of δ 13 C org and δ 13 C carb . High-resolution studies of paired δ 13 C carb and δ 13 C org records in widely separated regions are needed to Fig. 19 Stable isotope chemostratigraphy of the Borkewehr DCB section, including profiles of δ 18 O carb , δ 13 C carb , and δ 13 C org as well as TOC [wt. %] contents. reveal controlling factors of major carbon cycling perturbations during the Devonian-Carboniferous transition.

Inorganic geochemistry
Concentrations of the elements used in the present study are shown in the supplementary material (Online Resource 1). The principal component analysis of ICP-OES/MS dataset revealed common relationships among measured elements and discriminated groups of elements with similar geochemical behaviour. The principal components PC1 and PC2 explain 69.52 % and 16.39 % of dataset variance, respectively. The PC3 explains 4.16 % of the total variability (Online Resource 3). Group 1 comprises Ca, Sr, and Mn, elements with affinity to carbonates, which have a strong positive score on PC1 and low negative score on PC2. Group 2 involves mainly lithophile elements (e.g. Al, Si, Fe, Zr, Ti, K, Rb) and Rare Earth Elements (REEs), with strong negative score on PC1 and low positive score on PC2. Group 3 is represented by Zn, Cu, Pb, Co, U, as well as Ni and is characterised by positive scores on PC1 and negative scores on PC2 (Online Resource 3.1). The main variability of the groups 1 and 2 can be explained by carbonate dilution of siliciclastic detritus (PC1). Group 3 consists of elements, typically related to autigenic phases or organic matter, deposited in reducing or high-productivity conditions (Craigie 2018). The main lithologic types can be categorised by CaCO 3 content to carbonate facies, mixed facies, and siliciclastic facies (Fig. 20a). The carbonate facies are rigid limestone beds, which contain 89 to 60 % CaCO 3 and correspond to limestones, marly limestones, calcareous marlstones, and marls based on the scheme of Correns (1968). The mixed facies, represented by soft and/ or fissile beds, are marlstones to calcareous shales with 59 to 15 % CaCO 3 . The siliciclastic facies are shales and siltstones (14 to < 1 % CaCO 3 ). To eliminate the effects of biogenic carbonate dilution, we calculated enrichment factors (EFs) based on Al normalized element concentrations, relative to the average Upper Continental Crust (UCC) standard of McLennan (2001), following the common formula: Aluminium normalisation was used only for elements with higher or equal coefficients of variation compared to the coefficients of variation of Al (van der Weijden 2002, Online Resource 1). Statistical evaluation of the correlation among the reported elements data was performed by non-parametric statistical tests (Spearman's correlation coefficientrs, Online Resource 2) as the Shapiro-Wilk and the Anderson-Darling statistical tests rejected the assumption of normality for all variables.
REEs concentrations have been normalized to the REEs composition of average upper continental crust (UCC, McLennan 2001), where N stands for REEs normalized to UCC:

Detrital input
The terrigenous elements, namely K, Rb, and Ti, have high statistic correlation with Al (rs = > 0.9, n = 35, p < 0.05), indicating their affinity to clay minerals. Faintly lower correlation between Al and Si (rs = 0.83, n = 35, p < 0.05) is probably influenced by presence of biogenic silica. Average content of Al is 1.2 % in carbonate facies, 4.1 % in mixed facies, and 7.9 % in siliciclastic facies. The Al/Ti ratio is between 14 and 22 (mean 18.8), which is close to the UCC values (Al UCC /Ti UCC = 19.6; McLennan 2001) or Post Archaean Australian Shale (Al PAAS /Ti PAAS = 17; Taylor and McLennan 1985), and thus indicates detrital Al and Ti source (Murray and Leinen 1996). Also, an average Zr/Al ratio (0.0029) is similar to the Zr/Al of UCC (0.0024; McLennan 2001), revealing a detrital origin of Zr. Zr/Al is used as a proxy for tracing the variations in grain size of detrital particles in fine grained sediments, as Zr is enriched in the silty and sandy fraction, whereas Al is carried mainly by clay minerals (e.g. Ganeshram et al. 1999;Calvert and Pederson 2007). Correlation between Al and Zr is strong (rs = 0.83, n = 35, p < 0.05) but slightly lower compared to other terrigenous elements. This can be explained by the combination of carbonate dilution and the effect of sorting in the siliciclastic facies (Online Resource 3.2). The Zr/Al curve displays a wellmarked cyclic pattern through the section, especially in the Wocklum Limestone. These cycles correlate well among northern Rhenish Massif DCB sections (e.g. Oberrödinghausen Railway Cut, Oese, Drewer; compare Kumpan et al. 2015). Zr/Al values drop in the lower portion of the studied Wocklum Limestone (Fig. 20a) as a record of fining of detrital grain size, probably related to the deepening of the depositional environment. Up section, Zr/Al values increase in the cyclic pattern indicating coarsening, which matches with the shallowing trend inferred from facies analysis (see above). The decreasing Zr/Al pattern in the Hangenberg Black Shale (Fig. 20a) supports a transgressive nature of the unit. The overlying Hangenberg Shale/Sandstone equivalents start with a sharp Zr/Al increase reaching the highest values of the dataset approximately 45 cm below the top of this unit (Fig.  20a) and is correlated by us with the base of the Hangenberg Sandstone. It reflects the presence of the coarsest clastic material and records the maximal regression. Zr/Al values decrease from the upper part of the Hangenberg Sandstone equivalent to the dark-grey platy limestones of the Upper Crisis Interval (beds 2 and 3). In general, Zr/Al values increase in cyclic manner from the Wocklum Limestone up to the top of the Hangenberg Limestones. Higher Zr/Al ratio and Al content distinguish the Hangenberg Limestone from the Wocklum Limestone and reveal more proximal depositional setting of the former. Rb/K and K/Al ratios are proxies for weathering intensity related to mobilisation of K during chemical weathering, although K/Al is partly controlled by grain size (Bábek et al. 2016a). Rb/K values are lower and K/Al values are higher in the studied limestones than in siltstones, revealing the more dominant role of a mechanical weathering during deposition of limestones. The highest K/ Al values are below the top of the Wocklum Limestone (Bed -4b), followed by K/Al decrease toward the top of the Hangenberg Crisis interval (Fig. 20a). A change in weathering character took place at the base of the Lower Crisis Interval, where low K/Al and high Rb/K ratios indicate the deposition of products of more weathered clastic material. Higher Rb/K ratios indicate deposition of chemical weathering products, both in the Hangenberg Shale/ Sandstone equivalents and in the overlying dark-grey platy limestones. A decreasing K/Al trend demonstrates the combined influence of increased weathering and grain size fining upward. K/Al increases and Rb/K decreases again in the Stockum and Hangenberg limestones (Fig. 20a).

Redox and palaeoproductivity
The redox-sensitive trace elements V, Cr, Ni, and U revealed strong affinity to the detrital component, as they show good statistical correlations with Al (rs = > 0.8, p < 0.05) and low enrichment factors, which complicates their application as redox proxies. Moderate correlation with Al is shown by Mo and P (rs = 0.7, p < 0.05). Weak correlation with Al has As (rs = 0.3, n = 35, p < 0.05) and Pb, Cu, and Sb have no statistically significant correlation with Al (Online Resource 2). These elements have higher enrichment factors in several samples. Their enrichments can be considered as a product of authigenic accumulation in oxygen deficient conditions (Tribovillard et al. 2006).
Wocklum Limestone: The enrichments of Cu (EF up to 13) and Sb (EF = 5) first occur in the upper part of the Wocklum Limestone (4 to 5 m of the section thickness, Fig. 20a) and are followed by enrichments of Cu (EF = 5.7), Sb (EF = 21.8), Pb (EF = 6.7), As (EF = 134.0) (Fig. 20a), and P (EF = 4.6, Online Resource 3.3) in the uppermost part of the limestone unit. The enrichment of redox-sensitive trace elements is interpreted as a record of the iron shuttle process and episodic oxygen-deficient conditions (Tribovillard et al. 2013(Tribovillard et al. , 2015. The oxic-anoxic fluctuations are also supported by the REE Fig. 20 Inorganic geochemistry of the Borkewehr section. a Vertical distribution of selected element concentrations, their ratios, and enrichment factors. CaCO 3 , Zr/Al, K/Al, Rb/K, and Ti/Al (white circles carbonate facies, grey circles mixed facies, black circles siliciclastic facies) plotted from ED-XRF data, Cu EF , Pb EF , As EF , Sb EF , U/Th, Mo/ TOC, and REEs pattern (blue circles terrigenous pattern, orange circles MREE bulge pattern) plotted from the ICP-OES/MS dataset. Stratigraphic column key: 1 height above the section base in metres, (see section REE geochemistry). Oxygen-deficient conditions are also evidenced by the presence of pyrite and decreased bioturbation in the uppermost part of the Wocklum Limestone (see above). However, Cu, Sb, Pb, and As enrichments do not correspond with low Mo EF and U EF (Online Resource 3.3) and thus exhibit differential responses to weathering or high-frequency redox fluctuations (Jin et al. 2018). Low Mo EF and U EF may indicate the loss of Mo and U during weathering. This may be supported by a high Co/Mo ratio (Online Resource 1), which may reflect more intense dissolution of the original metal enrichment as Co is preferentially retained in the original sediment layer because of the high insolubility of Co(II)oxyhydroxides and the slow reaction kinetics of Co(III) (Märtz et al. 2011). Another factor influencing low Mo is caused by the distinct hydrological restriction of the basin indicated by low Mo/TOC ratios ranging between 4.3 and 21.3 (Fig. 20a, Online Resource 1; Algeo 2004; Algeo and Rowe 2012). A slightly lesser basin restriction is apparent at the top of the Famennian with no restriction at the Hangenberg Black Shale Event (Mo/TOC = 121, Fig.  20a), regarded as transgressive by many authors (e.g. Kaiser et al. 2015 online). These data are similar to those obtained in the Namur-Dinant Basin (Kumpan et al. 2019a, b).
Hangenberg Black Shale and Hangenberg Shale/Sandstone equivalents: The lower part of the Hangenberg Black Shale has higher Cu EF (7.5) and As EF (16.5), which increase together with EFs of other trace elements (Cu EF = 4.1, Zn EF = 4.7, As EF = 91.9, U EF = 4.8, Pb EF = 20.2, Sb EF = 14.0, Fig. 20a) and contents of S, Hg, and Mo (Online Resource 3.3) in the uppermost part of the black shale, revealing deposition in anoxic conditions. The calculated redox proxies suggest a conflicting redox interpretation of the lower part of the Hangenberg Black Shale with redox conditions between ferruginous and oxic conditions. The U/Al, V/Al, and Mo/Al proxies of Bennett and Canfield (2020) show an oxic or seasonal oxygen minimum zone (OMZ) in the lower part of the Hangenberg Black Shale and a seasonal OMZ in the upper part (Online Resource 3.4). High As EF and medium Sb EF may indicate the iron shuttle process and the lack of significant Mo EF may suggest the release of Mo during weathering, supported again by the high Co/Mo ratio (Märtz et al. 2011). The U/Th ratio reaches the value 1.25 in the uppermost part of the Hangenberg Black Shale (Fig. 20a), indicating the transition from suboxic to anoxic conditions (Jones and Manning 1994). The low content of all redox-sensitive trace elements may indicate oxic conditions for the overlying Hangenberg Shale/ Sandstone equivalents (Fig. 20a).
Stockum and Hangenberg limestones: The low content of redox-sensitive trace elements may indicate prevailing oxic conditions in the Stockum and Hangenberg limestones. The only exceptions are the enrichments of As (EF = 37) and Sb (EF = 9), accompanied by higher Mo/ Al at 7.9 m of the section thickness (Fig. 20a, Online Resource 3.4). The enrichment indicates the role of the Fe and Mn oxyhydroxide shuttle, which may have been connected with suboxic conditions. Increased U/Al and V/ AL ratios at 8.5 m of the section thickness (Fig. 20a) correspond to oxic to seasonal OMZ in the model of Bennett and Canfield (2020;Online Resolurce 3.4). Pyrite occurrences restricted to beds 15b (~7.9 m) and 29 (~8.5 m, see above) may support this interpretation.

REE geochemistry
There is no systematic change of REE patterns through the Crisis interval (Fig. 20) but the REE distribution develops two characteristics through the section: (1) a 'flat distribution' signifying predominantly terrigenous siliciclastic influence supported by a very good correlation of REEs and Al (rs = 0.82, n = 35, p < 0.05, Fig. 20b, Online Resource 3.5), (2) a 'middle-REE bulge' (Fig. 20b, Online Resource 3.5) due to adsorption of light and heavy REEs to Mn and Fe oxyhydroxides, respectively, documented by a positive MREE/MREE* anomaly and supported by a relatively high content of Fe and Mn, high Al/(Al+Fe+Mn) ratios, and a low Y/Ho ratio (Online Resource 4). The MREE bulge represents an early diagenetic signature associated with suboxic to anoxic conditions, possibly related to diagenetic dissolution of Mn and Fe oxyhydroxides (Haley et al. 2004;Himmler et al. 2010;Chen et al. 2015;Fig. 20b). The early diagenetic REE enrichment appears to vary as a function of the influence of the Fe-Mn oxyhydroxide shuttle and redox cycling and the content of clay minerals and phosphates (Abbott et al. 2015;Chen et al. 2015;Yang et al. 2017).
The (La/Yb) N and (La/Sm) N proxy (Reynard et al. 1999;Lécuyer et al. 2004;Fazio et al. 2007) shows the influence of early diagenesis and adsorption (Online Resource 3.6). The role of diagenesis is supported also by (La/Nd) N and Y anom crossplots, by a weak positive correlation between (La/Sm) N and Ce anomaly (Ce/Ce*= 0.5 x (La N +Pr N )), weak negative correlation between Eu (Eu/Eu* = Eu N / (Sm N 2 × Tb N ) 1/3 ) and Ce anomalies, and weak negative correlation between the Dy/ Sm N ratio and Ce anomalies, which also implies that Ce/Ce* values have been affected by MREE arching (Shields and Stille 2001;Fazio et al. 2007;Online Resource 4).
Positive Eu anomalies seem to indicate a hydrothermal source (Bau 1991;Fitzgerald and Gillis 2006;Jiang et al. 2007). Relatively low Ba/Sm and Ba/Nd ratios and a negative correlation between Ba/Sm and Eu/Eu* indicate that the anomaly is not influenced by Ba interference (Jiang et al. 2007;Online Resource 3.6). Positive Eu anomalies have been distinguished in the uppermost Devonian of different parts of the Rhenohercynian Zone (Kalvoda et al. 2018;Kumpan et al. 2019a).
The oxic-anoxic fluctuations, inferred from the redox sensitive trace element enrichments for the upper part of the Wocklum Limestone, are supported by the MREE bulge ( Fig. 20b; Reynard et al. 1999;Shields and Webb 2004;Lécuyer et al. 2004;Gu et al. 2019) and a changing Ce anomaly ranging between 0.46 and 1.02 (Online Resource 4). The role of REE in redox interpretation is obscured in shale samples from the Lower and Middle Crisis intervals. The samples with a strong correlation between Al/ΣREE indicate a modification of the original 'seawater' REE signal and Ce anomaly by detrital input (Fig. 20b; e.g. Frimmel 2009;Tostevin et al. 2016). The MREE bulge supports episodic suboxic to anoxic conditions interpreted in the Hangenberg Limestone (7.9 m, Fig. 20b, Online Resource 4).

Pyrite framboids
Pyrite framboids, altered by weathering to Fe-oxyhydroxides, were only observed in the uppermost part of the Hangenberg Black Shale, where diameters of 102 pyrite framboids were measured. The minimum diameter was 1.3 μm, the maximum diameter was 16.2 μm, the mean 3.9 μm, and the standard deviation 2.1 μm (Online Resource 1). In the sample, small framboids predominate (< 5 μm, n = 88), whereas large framboids (> 10 μm) were rare (n = 3). The size distribution of pyrite framboids (Fig. 20c) indicates anoxic conditions (Wignall and Newton 1998) during deposition of the uppermost part of the Hangenberg Black Shale and supports the geochemical interpretation based on the high trace metals enrichment.

Cyclostratigraphy
The Borkewehr section is particularly well-suited for cyclostratigraphy across the DCB for two reasons. First, based on the microfacies survey, there is no evidence for sedimentary hiatuses or major reworking events. Second, the section exhibits a steady lithological rhythm between marls/shales and limestones at the centimetre scale, prior to, as well as after, the Hangenberg Crisis. A typical cycle starts with a thinbedded marly or shaly layer, followed by a slightly thicker (nodular) limestone bed. However, prior to a cyclostratigraphic assessment of the lithological rhythms, the primary nature of the limestone/marl alternations needs to be assessed. This is because differential diagenesis may produce secondary limestone/marl alternations that mimic astronomical signals. Here, we adopt a technique first suggested by Westphal et al. (2010) and later refined by Nohl et al. (2021), which considers the ratio of diagenetically inert elements such as Al and Ti. The marls and limestones of the Wocklum Formation at the Borkewehr section exhibit two separate trend lines in the Al vs. Ti cross-plot, indicating a bimodal chemical composition of the non-carbonate fraction. This is a clear indication of varying environmental conditions, notably of changes in the hinterland weathering regime, and is a strong indication that the observed marl/limestone rhythms are of primary origin (Nohl et al. 2021). Titanium in soils is less soluable than aluminium. Hence, the relatively low Al/Ti ratios in the Borkewehr marls and shales (~0.36 vs.~0.54 for limestone layers) indicate that these lithologies were formed under a more intense weathering regime compared to the interbedded limestone layers (Fig. 21).
The thicknesses of lithologic couplets range between 3 and 10 cm. This relatively large range can be explained with three scenarios, of which the third is advocated in this study.
& In a first scenario, all lithological cycles are the result of the same forcing factor with a quasi-constant periodicity. The changing thickness of lithological couplets can be ascribed to changes in accumulation rate. We consider this scenario unlikely, as the section was deposited below the photic zone and storm wave base. Sedimentation rate variability, with two-fold or three-fold changes over relatively short periods of time, are not in line with expectations for such a calm sedimentary environment, without major reworking.
& In a second scenario, a relatively steady sedimentation rate is assumed and all lithological cycles are the result of the same forcing factor with a quasi-constant periodicity. In this scenario, the variations in couplet thickness are explained by the sedimentary system, occasionally failing to develop a thin-bedded marl/shale during a forcing extreme. In other words, the sedimentary system sometimes misses a beat, but sedimentation rates are more or less constant. We also deem this scenario unlikely because a random distribution of thicker cycles throughout the lithological column would be expected. This is not the case in Borkewehr.

Fig. 22
Cyclostratigraphic interpretation of the lithologic couplets at the Borkewehr sections. Thinner cycles (3.5-5.5 cm) are interpreted to reflect precession cycles (green), whereas somewhat thicker cycles (7-12.5 cm) are associated with obliquity (blue). The lithological cyclicity is used as a geologic metronome by applying the theoretical periodicities of precession (18.5 kyr) and obliquity (33 kyr), as calculated by Berger et al. (1992). Whereas the lower power spectrum refers to the cyclic succession (Wocklum Limestone) below the siliciclastics of the Hangenberg Black Shale and Hangenberg Shale/Sandstone equivalents, the upper power spectrum refers to the cyclic succession (Stockum and Hangenberg limestones) above.
& In a third scenario, relatively steady sedimentation rates and lithological cycles with different thicknesses reflecting the sedimentary response to different astronomical forcing factors are assumed. In this scenario, the thinner cycles, spectral peaks with a periodicity around 4 cm (Fig. 22) are attributed to precession, as previously suggested by Becker et al. (2012). The thicker cycles that have spectral peaks with a periodicity of~8.3 cm in the Hangenberg Limestone (Fig.  22) are ascribed to obliquity. Considering an 18.5 kyr and 33 kyr periodicity for precession and obliquity, respectively (Berger et al. 1992), the suggested relationship between astronomical parameters and lithological couplets implies a sedimentation rate of~0.23 cm/kyr (after compaction). This value chiefly complies with the first-order chronologic constraints from biostratigraphy.
Here, we advocate the third scenario because the Borkewehr section exhibits interesting stacking and amplitude modulation patterns. In the Wocklum Limestone as well as in the Hangenberg Limestone series of four to six thinner cycles are interrupted on both sides by a thicker cycle. As the thinner cycles are thought to be of precessional origin, it is expected to see indications of amplitude-modulation of those cycles by orbital eccentricity with 100-kyr and 405-kyr periods. This is because precession cannot exert any influence on insolation patterns when Earth's orbit is perfectly circular and has zero eccentricity. However, with increasing eccentricity, the imprint of precession on insolation patterns worldwide rises. In other words, when eccentricity decreases, obliquity gains in relative importance as eccentricity-modulation precession lessens its control on insolation patterns. In the studied section, we interpret the disruption of a series of thinner cycles by a thicker cycle as such a shift from a precession-dominated insolation regime (green intervals on Fig. 22) to an obliquitydominated regime (blue intervals on Fig. 22).
The five successive obliquity cycles at the top of the studied interval, however, no longer fit into this eccentricity-paced pattern. There seems to be a permanent transition from a mainly precession-sensitive climate and sedimentary environment to a more obliquity-sensitive system in the lower Tournaisian. However, this palaeoclimate interpretation needs scrutinization by future work, e.g. by means of a cm-resolution, geochemistry-based cyclostratigraphy throughout the section. Future work will also include cyclostratigraphic studies of other lower Tournaisian sections to lay out the global or regional nature of the proposed shift in Earth System response to Milankovitch forcing. A multi-section approach is also a prerequisite for adopting cyclostratigraphy as a tool to estimate absolute durations of successive conodont and/or ammonoid zones by means of cycle-counting. In addition, large sedimentation rate changes within the Crisis Interval are suspected, but these have yet to be determined. Protognathodus Ziegler, 1969 Representatives of the genus Protognathodus are well known in uppermost Famennian and Tournaisian strata. They are widely distributed and were studied by various authors in the last decades (compare synonymy lists given below). As suggested by Ziegler (1973) and Corradini et al. (2011), the ornamentation on the oral surface and the shape of the cup indicate a phylogenetic lineage within early protognathids. Herein we concentrate on the evolution from ancestral Pr. collinsoni to Pr. kockeli, because this transition is rather complex (e.g. Luppold et al. 1994;Kaiser et al. 2019a online). As outlined by Kaiser et al. (2019b), "almost all diagnostic features regarding the typical shape of cup of one taxon can be applied to all other ornamentation-types" in early protognathids. Nevertheless, general trends exist. Therefore, we focused mainly on the arrangement of cup ornamentation, which allows a clear distinction of species within the genus (Fig. 23). However, atypical characteristics, especially in transitional forms, result in a significant morphological complexity.

Conodonts
With the introduction of Pr. semikockeli n. sp. (Figs. 12g, 23c) we ensure the use of Pr. kockeli s.str. as a biostratigraphic tool for defining a future Devonian/Carboniferous Boundary, within the frame of successive other stratigraphic markers. It avoids the problem that the siliciclastic intercalation of Hangenberg Black Shale, Hangenberg Shale, and/or Hangenberg Sandstone normally lacks a conodont record. It would be difficult to place a future GSSP directly above such a facies break, since any defining conodont could have existed in the record gap. Furthermore, within Pr. kockeli, three new morphotypes ( Fig. 23d-f) are defined.
The three considered species are discussed in an evolutionary order -Pr. collinsoni, Pr. semikockeli n. sp., and Pr. kockeli. Our review of their geographic distributions is grouped in figured and not figured records (Tabs. 2-4). Not figured quotations of Pr. collinsoni as well as Pr. kockeli may possibly include the transitional Pr. semikockeli n. sp. The evolution from Pr. kockeli to Pr. kuehni was investigated conscientiously by Kaiser et al. (2019a online) but it will be helpful to designate in future clearly defined morphotypes along the transition.
As pointed out by Sanz-López et al. (2018), the genus Protognathodus is homeomorphic with the younger Pseudognathodus, which ranges from the upper Tournaisian to the upper Viséan.
Synonymy lists, abbreviations: * = introduction of a new taxon, ? = the allocation to a taxon is questionable, e.p. = ex parte, only a specified part of the material of a publication belongs to the discussed taxon, non = the specimen does not belong to the taxon. Ziegler, 1969 (Figs. 11m, 12h-i, 23b, Tab. 2) Fig. 23 Phylogenetic lineage of early protognathids, Puech de la Suque, Montagne Noire, southern France. a Protognathus meischneri Ziegler, 1969, GMM B9A.11-65, Bed 10A1, Pr. kockeli Zone, oral view. b Protognathus collinsoni Ziegler, 1969 kockeli Zone, oral view, two isolated nodes on both sides of the cup surface (red circles). c Protognathus semikockeli n. sp., GMM B9A.11-67, Bed 10A1, Pr. kockeli Zone, oral view, one node on the left cup surface (red circle) and a row of three nodes parallel to the carina on the right side of the cup (red line). d Protognathus kockeli (Bischoff, 1957) Morphotype 1:1, GMM B9A.11-68, Bed 10A2, Pr. kockeli Zone, oral view, one row of nodes (consisting of at least three nodes) parallel to the carina on each side of the cup (red lines, Morphotype 1:1 already occurs in Bed 10A1). e Protognathus kockeli (Bischoff, 1957) Morphotype 1:2, GMM B9A.11-69, Bed 10A1, Pr. kockeli Zone, oral view, one row of nodes parallel to the carina on the right side and two rows of nodes on the left side of the cup (red lines). The row of nodes on the right side is accompanied by an additional node (red circle). f Protognathus kockeli (Bischoff, 1957) Morphotype 2:2, GMM B9A.11-70, Bed 10A2, Pr. kockeli Zone, oral view, two rows of nodes run on both cup sides parallel to the carina (red lines).

Protognathodus collinsoni
Material: Twelve Borkewehr and two Puech de la Suque (Montagne Noire, southern France) specimens; additional elements from other Rhenish sections, Franconia, and Montagne Noire. Diagnosis (emended): A species of Protognathodus with a single or several nodes on one or both sides of the carina that do not form rows that consist of at least three nodes, parallel to the carina. Discussion: The longitudinal axis of Pr. collinsoni is straight to slightly curved laterally. The latter leads to a slightly asymmetrical shape of the cup and a more expanded outer side. In many specimens, the anterior part of the inner portion and the posterior part of the outer side show convexities of the cup margins (e.g. Fig. 12i; Bartzsch and Weyer 1982: pl fig. 4.4). However, more or less symmetrical forms occur, too (e.g. Fig. 23b; Hou et al. 1984: pl. 4, fig. 12;Ji et al. 1989: pl. 18, fig. 9;Corradini et al. 2011: pl . 1, fig. 4). The cup, which extends to the posterior tip, bears one node or more scattered nodes on one or both sides of the surface. The nodes are situated at variable distances to the carina (e.g. Perri and Spalletta 2000b: pl. 1, fig. 4;Feist et al. 2020 online: fig. 6g) or further away and closer to the cup margins (e.g. Bartzsch and Weyer 1982: pl. 3, fig. 1). In the majority, the anterior cup margins begin directly opposite at the blade (e.g. Figs. 12h,23b), but there are also specimens in which the anterior terminations show an offset (e.g. Kaiser et al. 2019a: pl. 1, fig. 6;Plotitsyn and Grudzev 2019: fig. 4.4).
The free blade is about one third to one half of the element length. It consists of laterally compressed, fused denticles with free tips, slightly decreasing in height towards the cup. The blade extends to the posterior end as a carina. In some specimens, the denticles of the carina get broader, thickened laterally (e.g. Szulczewski 1973: pl. 2, fig. 10;Corradini et al. 2011: pl. 1, fig. 5).
In aboral view, a moderately deep cup extends to the posterior end. The edge of the free blade is marked by a narrow, sharp median groove. Rare juveniles, as well as abundant medium sized and adult specimens, are available from the literature and our collections. No variation in cup shape and ornamentation can be recognised depending on growth stages. Discussion: Protognathodus semikockeli n. sp. is introduced to ensure the biostratigraphical use of Pr. kockeli s.str. as a potential marker for defining a future Devonian/Carboniferous Boundary. It falls outside the original definitions of both Pr. collinsoni and Pr. kockeli. Intermediates between Pr. collinsoni and Pr. kockeli are well known but they were assigned to Pr. collinsoni, expanding the original diagnosis sensu Ziegler (1969), or to Pr. kockeli, in the sense of the emended diagnosis sensu Corradini et al. (2011). At Borkewehr and in other sections (e.g. Puech de la Suque, Montagne Noire, France), Pr. semikockeli n. sp. occurs in the first limestone bed immediately above the conodont-free Hangenberg Shale/Sandstone. Therefore, its entry is not useful for a future boundary definition. The oldest Borkewehr specimens are not preserved well enough to select a holotype.
Protognathodus semikockeli n. sp. differs from its ancestor Pr. collinsoni in the row arrangement of cup nodes on one side. A row is defined by at least three nodes (Fig. 12g). In some specimens the nodes are fused to a ridge (Fig. 11l). The total number of cup nodes is not decisive. In Pr. collinsoni, one or two nodes occur or, in advanded forms, more nodes can be scattered on the cup. Its descendant, Pr. kockeli s.str., bears at least one row of nodes/ridge parallel to the carina on each side of the cup. In Kaiser et al. (2019a online), representatives of our new species were named "Pr. kockeli "one-row" morphotype". However, other specimens were hidden under the designations "Pr. kockeli cf. "one-row" morphotype", "Protognathodus aff. Pr. kockeli "one-row" morphotype", and "Pr. kockeli "two-row" morphotype". This reflects the intraspecific variation of Pr. semikockeli n. sp.  recognised it as "collinsoni-kockeli intermediate", Herbig et al. (2019) as "transitional form to Pr. kockeli", and Feist et al. (2020 online) as "Pr. collinsoni, transitional form towards Pr. kockeli".
An uncommon narrow morphotype with an unornamented and strongly reduced inner cup side was illustrated by Ji et al. (1989: pl. 18, fig. 7) from the Upper praesulcata Zone at Dapoushang, South China. Somewhat atypical specimens were recorded by Corradini et al. (2003: pl. 4, fig. 1, re-figured in Corradini et al. 2011: pl. 1, fig. 15) from the Monte Taccu North A section, Sardinia, Italy and by Habibi et al. (2008: fig. 4.6) from the Shahmirzad section, central Elburz Mountains, Iran. Both specimens show nodes arranged in a line on one side of the cup and a cluster of irregularly distributed nodes on the upper cup surface of the other side. Especially the Italian specimen shows additional scattered nodes between the carina and the parallel running row of nodes. On both elements further nodes occur closer to the cup margin.
In addition, it is possible that a record of "Pr. kockeli early form" in Conil et al. (1986) from the French side of the Ardennes refers to Pr. semikockeli n. sp.
Since our new species overlaps only partially in time with its ancestor (Pr. collinsoni) and descendent (Pr. kockeli s.str.), it does not represent a simple intermediate stage ("chronomorphotype") of a single taxon that changes gradually. Due to its shorter time range (see below) it cannot be simply regarded as an intraspecific variant. However, in a more conservative approach, authors may prefer to recognise it at subspecies level only, which has no affect on its biostratigraphic value. Stratigraphic range: Unlikely Pr. collinsoni, the new species is not known from pre-crisis levels, such as the Wocklum Limestone. It enters at Borkewehr at the top of the costatuskockeli Interregnum, in the upper part of the "laminated" limestones, just above the Hangenberg Shale/Sandstone equivalents, which were divided into the lithologically similar beds 2 and 3. An equal FAD was indirectly reported by Kaiser et al. (2019a online) from the Graz Palaeozoic, Austria. At Trolp, Pr. semikockeli n. sp. enters in the first limestone bed immediately below the FAD of Pr. kockeli s.str. Sanz-López et al. (1999) described the first specimens assigned by us to Pr. semikockeli n. sp. from the second nodular limestone bed above the extinction of B. ultimus, within an interval of darker interbeds, interpreted as Hangenberg Crisis Interval. There is a lack of Pr. kockeli in the subsequent layers and Si. (Eo.) sulcata s.l. enters in the twelfth (nodular) limestone bed above. In the absence of illustrations, it cannot be ruled out that the oldest Pr. kockeli reported from Stockum Trench II (Clausen et al. 1994, Bed 108) and Drewer , Bed 100) belong to Pr. semikockeli n. sp. The oldest known Pr. semikockeli n. sp. record is from Bed 21 at La Serre, Trench F (Feist et al. 2020 online, their fig. 6h), the top of shallow-water limestones (lower carbonate unit) that fall in the top of Middle Crisis Interval I.
At Borkewehr, the last occurrence of Pr. semikockeli n. sp. is in the basal part of the Si. (Eo.) bransoni Zone. However, the atypical specimen of Corradini et al. (2003) occurs in the Si. (Si.) mehli (= Upper duplicata) Zone, whereas the one of Habibi et al. (2008) was extracted from a limestone bed at the top of an interval, which was dated as sandbergi to Lower crenulata Zone.
With respect to ontogeny and variability, we propose an emended diagnosis that enables an easy identification independent of size and an easy distinction from Pr. semikockeli n. sp. The holotype is clearly an adult and morphologically advanced form. Based on past taxonomic uncertainties and discrepancies, a new synonymy list is supplied. As pointed out by Ziegler (1973), the cup ornamentation is closer to or at the level of carina denticles. The longitudinal axis can be straight (Fig. 11e, j), but in the majority of specimens it is (slightly) curved laterally (Figs. 11d,. This results in an asymmetrical cup, with an outer half more strongly expanded than the inner one. As in its ancestors, two convexities of the cup margins are developed, one at the inner anterior and one at the outer posterior sector (Fig. 11d). However, specimens with a more or less symmetrical cup shape also exist.
In aboral view, Pr. kockeli equals Pr. semikockeli n. sp. A moderately deep cup reaches the posterior tip and the edge of the free blade shows a narrow, sharp median groove.
We recognise three Pr. kockeli morphotypes based on the number of rows of nodes or ridges of fused nodes on each side of the cup. They can be distinguished easily and strictly. At least three nodes arranged in a line parallel to the carina are necessary to build a row. Morphotype 1:1 (Figs. 11e, i-k, 12d, 23d), the most common form, bears one row of nodes/ridge on both cup sides and equals the "Pr. kockeli "two-rows" morphotype" sensu Kaiser et al. (2019a online). Morphotype 1:2 (Figs. 11d, h, 12b-?c, 23e) has one row of nodes/ridge on one side and two rows of nodes/ridges on the other side. It does not matter if the two rows/ridges are developed on the inner or the outer side of the cup. In Morphotype 2:2 (Fig. 23f), which includes the holotype, two rows of nodes/ridges run on both cup sides parallel to the carina. However, these rows of nodes/ridges can be accompanied by additional node(s) (e.g. Fig. 23e) in each of the three morphotypes.
Protognathodus kockeli s.str. differs from Pr. semikockeli n. sp. in the development of at least one row of nodes/ridge parallel to the carina on both sides of the cup. The descendant Pr. kuehni Ziegler and Leuteritz in Koch et al. (1970) can be distinguished by the fusing of nodes to transverse ridges. However, intermediates between Pr. kockeli s.str. and Pr. kuehni are well known 12a) and bear rounded nodes, transversally elongated nodes, as well as incipient transverse ridges. These transitional forms were discussed conscientiously in Corradini et al. (2011) and Kaiser et al. (2019a online) and may be recognised as additional, distinctive morphotype(s). The distinction of Pr. kockeli s.str. from the younger Pr. praedelicatus, characterised by indistinct longitudinal rows (e.g. type material of Lane et al. 1980: pl. 3, figs. 5-6), requires further revision. Kalvoda et al. (2013 online: fig. 5.16, re-illustrated in Kumpan et al. 2020 online: fig. 5.9) as well as Kaiser et al. (2019a: pl . 1, fig. 8) figured Pr. kockeli specimens, which have not directly opposed anterior cup margins. Stratigraphic range: After the current DCB definition, the species enters in the uppermost Famennian and defines the base of the eponymous Pr. kockeli (= Upper praesulcata) Zone (compare Ziegler and Sandberg 1984). We propose to use the FAD of Pr. kockeli s.str. to define in the future the Carboniferous base. It ranges at least into the Lower crenulata Zone (= crenulata/isostichia zones; Sandberg et al. 1978, North America;compare Corradini et al. 2011 andSpalletta et al. 2017). Kaiser et al. (2009, Puech de la Suque, Montagne Noire) described the youngest specimens from the Si. (Si.) quadruplicata Zone (= supposed upper subdivision of the Si. (Si.) sandbergi Zone sensu .

Conclusions
Becker et al. (2021) reviewed thirty DCB sections of the Rhenish Massif. In order to appreciate the quality of the Borkewehr section, comparisons of the immediate DCB interval of the best-known Rhenish sections is helpful. These are Hasselbachtal, Oese, Oberrödinghausen Railway Cut, Stockum Trench II, and Drewer (especially the poorly accessible sections WI and WJ of Korn et al. 1994;Fig. 24). For the extensive literature on the named sections see Becker et al. (2021). The lateral DCB sections provide important additional data on faunas, especially on lower Tournaisian ammonoids and trilobites, pelagic-neritic correlation, cyclostratigraphy, element and isotope geochemistry, magnetic susceptibility, the provenance of the siliciclastics, and the geochronological dating of bentonites.
Our proposal of the Borkewehr section as a DCB GSSP section should be viewed in the context of the enormous regional package of information. The Borkewehr is currently the only section with a record of the Pr. semikockeli n. sp.-Pr. kockeli s.str. sequence, which we propose to use for a new DCB definition. In many other sections, with a slightly later onset of carbonate deposition after the Hangenberg Regression, the FOD of Pr. kockeli, in its original sense, enables an unequivocal recognition of the Carboniferous base. At Borkewehr, conodont faunas of the Upper Hangenberg Crisis Interval are not very rich but ranges were confirmed by several rounds of re-sampling. The base of the Si. (Eo.) sulcata s.l./Pr. kuehni Zone can be recognised in the sixth limestone bed above the Hangenberg Shale/Sandstone equivalents, which coincides with a marked shift of conodont biofacies, as it is typical for the base of the Hangenberg Limestone. Borkewehr is the type section of the Wocklum Limestone, of several important topmost Devonian ammonoids, and both of Pr. semikockeli n. sp. and Pr. kockeli s.str. Other important faunal groups with stratigraphical significance, such as ostracods, trilobites, and deep-water corals, are present. In the critical interval, there is no evidence for sedimentary gaps or reworking events. As further important tie points for global correlation, the Hangenberg Black Shale is well-developed as a transgressive and at least partly anoxic interval and Zr/Al ratios show the precise level of maximum regression within the locally undivided Hangenberg Shale/ Hangenberg Sandstone unit. There is evidence for a change of Milankovitch cyclicity by climatic developments associated with the Hangenberg Crisis. Above the DCB interval, the middle Tournaisian Lower Alum Shale Event is sharp, but remarkably thin, followed by a condensed Tournaisian-Viséan succession that is exceptional for the missing upper Tournaisian bedded cherts (Hardt Formation) and development of equivalents of the Erdbach Limestone II. The Borkewehr is very easily accessible, not in danger of destruction, and suitable for the application of numerous modern stratigraphic techniques, as demonstrated by us for element chemostratigraphy, carbon isotope stratigraphy, cyclostratigraphy, and sequence stratigraphy. The proposal of the Borkewehr section as a new GSSP candidate ensures its return to the type region of the first Devonian/ Carboniferous Boundary GSSP. his team (Münster) as well as Kathrin Jung (Cologne) prepared thin sections; additional Carboniferous thin sections were provided by Markus Großhäuser (formerly Cologne). Pedro Cózar (Madrid) is thanked for identification of foraminifera and biostratigraphic advice. Hartmut Kaufmann (Burgwald) supplied ammonoid data based on his collection and the original of Figure 16. Hanna Cieszynski (Cologne) took some of the conodont photographs and Traudel Fährenkemper (Münster) produced figures 1, 15, and 24. We appreciate the helpful reviews of Markus Aretz (Toulouse) and D. Jeffrey Over (Geneseo). The element geochemistry research was funded from the project 19-17535S of the Czech Science Foundation (GAČR). This paper is a contribution to the International Geoscience Programme (IGCP) Project 652 "Reading geological time in Paleozoic sedimentary rocks".
Funding Open Access funding enabled and organized by Projekt DEAL.
Data availability All data generated or analysed during this study are included in this published article [and its supplementary information files]. All specimens (if not stated otherwise) are stored in the Geomuseum of the Westfälische Wilhelms University Münster (GMM).

Declarations
Conflict of interest The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.