Aras Valley (northwest Iran): high-resolution stratigraphy of a continuous central Tethyan Permian–Triassic boundary section

The Permian–Triassic boundary section in the Aras Valley in NW Iran is investigated with respect to carbonate microfacies, biostratigraphy (particularly conodonts, nautiloids, and ammonoids), chemostratigraphy (carbon isotopes), and environmental setting. Correlation of the data allows the establishment of a high-resolution stratigraphy based on conodonts (with four Wuchiapingian, 10 Changhsingian, and three Griesbachian zones), ammonoids (with nine Changhsingian zones), and carbon isotopes; it forms the base for the reconstruction of the environmental changes before and after the end-Permian extinction event at the studied locality. In the Aras Valley section, there is no evidence for the development of anoxic conditions, associated with the end-Permian mass extinction.


Introduction
The end-Permian mass extinction (EPME), which is the most severe biotic crisis of the Phanerozoic (e.g. Erwin 1994Erwin , 2002Benton 2003), has been intensively studied in numerous sections all over the globe. Our knowledge about the co-evolution of life and the planet was affected by this extinction event and the causes of the extinction itself. However, our knowledge derives mainly from sections in only a few regions. High-resolution stratigraphic frameworks have been established in sections of south China, including the Global Boundary Stratotype Section and Point (GSSP) at Meishan, the western USA, and central Europe (e.g. Collinson et al., 1976;Posenato, 2008;Farabegoli and Perri 2012;Yuan et al., 2014;Jattiot et al., 2017). Geochemical proxies for environmental changes (e.g. Horacek et al., 2007a;Marenco et al., 2012;Sun et al., 2012;Yin et al., 2012) and palaeoecological changes have been intensely studied in these regions (e.g. Schubert and Bottjer 1995;Chen et al., 2010;Hofmann et al., 2014Hofmann et al., , 2015Foster et al., 2017aFoster et al., , b, 2018. However, to gain insights into global patterns of its manifestation, the EPME has to be studied in detail with a profound stratigraphic framework in numerous regions all over the world. Permian-Triassic boundary (PTB) sections in northwest Iran and Transcaucasia are well-suited for further detailed investigations concerning the EPME. Unlike many wellstudied sections in south China, the western USA, and central Europe, the Permian-Triassic transition in the Iranian-Transcaucasian sections is usually recorded as showing continuous sedimentation. These sections represent a different palaeogeographic region (i.e. the Neotethys Ocean). PTB successions at Iranian and Transcaucasian locations have previously been studied to establish a regional Permian-Triassic stratigraphic framework. The sections are known to contain large suites of macrofossils and microfossils, which allow a precise stratigraphic subdivision and correlation with the GSSP section for the Permian-Triassic boundary.
During the last decade, efforts to study PTB sections have been intensified in NW Iran, resulting in the investigation of new sections (Ghaderi, 2014;Ghaderi et al., 2014b;. One of these new sections, the Aras Valley section, was discovered in 2011 and is located only a few hundreds of metres along the strike of the west of the Dorasham section. In the following, we investigate the Aras Valley section (NW Iran) for a comprehensive and detailed bioand chemostratigraphic framework, using conodont and ammonoid zonation as well as carbon isotopes. In comparison to the other NW Iranian-Transcaucasian sections, the Aras Valley section shows some differences, concerning, for instance, the fossil content. There are, for example, extremely abundant ostracod assemblages across the PTB and a high abundance and diversity of cephalopods. Conodont data demonstrate that the section offers a complete succession without stratigraphic gaps. The EPME event can be pinpointed at the millimetre scale and the official, conodontbased Permian-Triassic boundary, with the first appearance of the index species Hindeodus parvus, can be determined.
With this stratigraphic framework, we are able to get a better understanding of the depositional setting and changing environmental conditions that occurred around the time of extinction.

History of research
Permian-Triassic boundary sections in Transcaucasia (Fig. 1a) have a long research history. Already Abich (1878) described fossils, which he considered to be of Carboniferous age, from the vicinity of the old Armenian village Dorasham in the Aras Gorge west of Dzhulfa (now belonging to the Nakhichevan Province of Azerbaijan). Von Möller (1879) discussed this stratigraphic interpretation and found more similarities to the European Zechstein Formation and, therefore, he reinterpreted the age as Early Permian. Subsequent pioneering studies were carried out by F. Frech (1900) and von Arthaber (1900), who introduced the term "Djulfa-Schichten" for fossiliferous strata with numerous brachiopods and ammonoids of Wuchiapingian age. A readable journey report of this expedition was published by V. Frech (1900), including a very precise description of the adverse field work conditions at that time. Stoyanow (1910) subdivided the lithology of the sections in the Aras Gorge into 15 clearly distinguishable units; he presented the first ammonoid occurrences from the late Changhsingian Paratirolites Limestone, a very characteristic nodular limestone unit of 4 to 5 m thickness.
The classical section near the Dorasham railway station (Dorasham II, Fig. 1b) at the northern side of the Aras River, about 17 km WNW of Dzhulfa, was studied in detail by Arakelyan et al. (1965) and . The articles of the Dorasham sections were released in a comprehensive monograph edited by Ruzhencev and Sarytcheva (1965), who also included extensive descriptions of the fossil content with respect to various groups, such as nautiloids (Shimansky, 1965), ammonoids Shevyrev, 1965), brachiopods (e.g. Sarytcheva and Sokolskaja, 1965;Sokolskaja, 1965), and ostracods (Belousova, 1965). Rostovtsev and Azaryan (1973) and Kotlyar et al. (1983), who reinvestigated these PTB successions, introduced the new name Dorashamian for the latest Permian stage, with the Dorasham II section (Fig. 1b) as type locality. The conodont succession of the Dorasham II and Sovetashen (Chanakhchi in literature; Fig. 1a) sections was studied by Kozur et al. (1978Kozur et al. ( , 1980 and Kozur in Zakharov et al. (2005); these sections were suggested as stratotypes for the PTB (Kotlyar et al., 1993). The Triassic lithological succession in the Transcaucasian sections was further outlined in great detail by Zakharov (1992). He showed that the late Permian Paratirolites Limestone is overlain by almost 2 m of the latest Permian argillites with a few marly limestone intercalations. The PTB with the first occurrence of Hindeodus parvus is marked by a first, light-green limestone bed  Stampfli and Borel, 2002). of 8 cm thickness, containing bivalves of the genus Claraia. Zakharov's study thus provided one of the most precise determinations of the PTB in the Tethyan realm.
The NW Iranian PTB sections in the Ali Bashi Mountains (Kuh-e-Ali Bashi) west of Julfa were discovered and described by Stepanov et al. (1969) and . These authors showed striking similarities to the lithological successions of the Dorasham section, which is located approximately 10 km to the north. In the latter article,  provided a detailed historical review of the Transcaucasian PTB sections.
(or Jolfa; NW Iran) in a dry tributary 200 m west of the Aras River and 2 km north-west of the Dorasham I section (Ruzhencev et al., 1965) (Figs. 1, 2). It is located about 11 km north-west of the Ali Bashi sections, which were already described in detail (Stepanov et al., 1969;Ghaderi et al., 2014b;Leda et al., 2014). The Aras Valley section belongs to a north-inclined monocline, which is composed of middle Permian to Triassic sedimentary and volcanic rocks. Half of the Wuchiapingian and a complete Changhsingian succession, as well as the lower part of the Griesbachian sedimentary rocks, are exposed at the Aras Valley section across a lateral extension of 200 m. Particularly the lower, shale-dominated part of the Changhsingian Ali Bashi Formation, the Paratirolites Limestone, and the lower Claraia Beds can be studied in great detail. The Aras Valley section belongs, together with the sections of Dorasham, Ali Bashi, and Zal, to the southernmost Permian-Triassic boundary sections of the Transcaucasian-NW Iranian region (Fig. 1). They differ from the northern sections (such as Vedi, Chanakhchi, Ogbin, Karabaglyar, and Baysal) in Armenia and Azerbaijan in the higher shale content of the early Changhsingian strata Rostovtsev and Azaryan, 1973;Kotlyar et al., 1983).
During the late Permian, NW Iran was located at the northern margin of the Sanandaj-Sirjan Terrane as part of the Cimmerian microcontinent Borel, 2002, 2004). During the Permian, the Cimmerian microcontinent drifted from Gondwana towards the north, paralleled by the opening of the Neotethys; NW Iran was located at a latitudinal position near the Equator between the Neotethys and Palaeotethys (Fig. 1c) Borel, 2002, 2004;Muttoni et al., 2009a, b). An alternative reconstruction by Ruban et al. (2007a, b) suggests that the Sanandaj-Sirjan Terrane and NW Iran were two separated units with NW Iran located east of the Sanandaj-Sirjan Terrane.
During the late Permian to Early Triassic, the Sanandaj-Sirjan Terrane was largely covered by a shallow to moderately deep sea; the NW Iranian sections represent mostly open marine deposits (Kozur, 2007b;Leda et al., 2014). The geometry of the NW Iranian carbonate platform has not been studied in detail. Some authors, however, refer to the NW Iranian and Transcaucasian PTB successions as outer-shelf deposits (Kozur, 2007a;Richoz et al., 2010;Ghaderi et al., 2014b;Leda et al., 2014).

Lithostratigraphy
The Aras Valley section shows a sedimentary succession that is very similar to the neighbouring sections of Dorasham and the Ali Bashi Mountains. It exposes more than 37 m of sedimentary rock spanning the PTB interval without stratigraphic gaps (Fig. 3). According to the lithostratigraphic scheme developed by Stepanov et al. (1969) and Ghaderi et al. (2014b), the succession of the rock formations is outlined here in as-cending order. For the position of distinct units, beds, and horizons, we use the top of the Paratirolites Limestone as the marker position (0.00 m).

Julfa Formation
The Permian-Triassic succession in the Aras Valley section begins with the Julfa Formation, of which 15.25 m (from −29.20 to −13.95 m) was investigated in this study. The Julfa Formation can be subdivided into two members (tentatively named lower and upper Julfa Formation) that differ in rock colour and faunal assemblages (Stepanov et al., 1969).
Lower Julfa Formation. This member was separated into two subunits by Stepanov et al. (1969). The lower unit, which is characterized by abundant occurrences of the large-sized productid brachiopod genus Araxilevis, is poorly exposed in the Aras Valley section and was not investigated in this study. Only the uppermost portion of the member is well-exposed with a thickness of 3.30 m (from −29.20 to −25.90 m). This part of the lower Julfa Formation is dominantly composed of grey shales with intercalations of grey limestone beds and frequent occurrences of small marly limestone nodules. The limestone intercalations range from single beds (5 to 10 cm) to packages of thin-bedded limestones (up to 25 cm). Several of the carbonate beds are very fossiliferous and contain benthic (rugose corals, bryozoans, crinoids, ostracods) as well as nektonic (nautiloids, ammonoids, conodonts) fossil organisms. In the upper unit of this member, the brachiopod genera Araxathyris and Permophricodothyris are most common (Ghaderi et al., 2014a). Particularly the cephalopod shells are frequently encrusted by sessile organisms (e.g. tabulate corals, crinoids).
Upper Julfa Formation. The upper member is 12 m thick (ranging from −25.90 to −13.95 m) and differs from the unit below in the often pinkish to reddish or light-grey shales and limestones. Considering the colour and lithology of the strata, eight subunits, consisting of shale-limestone alternations can be separated, displaying an alteration of reddish and greyish subunits, in ascending order with the thicknesses of 2. 75, 1.90, 1.30, 0.70, 1.00, 0.40, 0.70, and 3.20 m. Specifically, these units are composed of alterations of grey to red shales in packages of 7 to 85 cm thickness and limestones (often marly). The carbonates are arranged in thin (a few centimetres thick) nodular beds that can form packages up to 1.45 m in thickness. The carbonate units are often marly and thin-bedded. At the base, the carbonate levels are pink or red and become light grey or beige at the top. The macrofossil content is much lower than in the lower Julfa Formation, recognizable by the much lower occurrences of benthic organisms such as brachiopods and rugose corals. It is noteworthy that fossils of larger nektonic organisms are also rarer; nautiloids and ammonoids have been found only in low numbers. In contrast, ostracods show mass occurrences in some of the higher carbonate beds of the unit.

Ali Bashi Formation
The Ali Bashi Formation is 13.95 m thick (from −13.95 to −0.00 m) and can be subdivided into two members: In the Aras Valley section, the Zal Member has a thickness of 9.35 m (ranging from −13.95 to −4.60 m) and is thinner than in the Ali Bashi 1 section with 12.50 m (Ghaderi et al., 2014b;Leda et al., 2014). Colour changes of the rocks seen in the underlying upper Julfa Formation continue into the Zal Member, which in turn also shows the alternation of four grey and four red subunits. Shales are the predominant lithology in the Zal member; they range in colour from deep red or purple to greenish or light to dark grey. Some of the shale portions are nearly black when non-weathered. The shales are intercalated with limestones that are usually thin-bedded (up to 5 cm thick) and marly or, more rarely, silty. Like the shales, their colour is variable and includes reddish, pink, greenish-grey, and grey tones. The upper part of the member shows a decrease in shale content and the limestone beds are amalgamated to form packages up to 55 cm in thickness. The macrofossil assemblages are dominated by cephalopods, mostly ammonoids (Korn et al., 2019a); some of the beds are very rich in ostracods.
For the Paratirolites Limestone, the member has a thickness of about 4.60 m. The base of the member is not clearly defined; here we consider the base to be marked by a compact limestone bed (20 cm thick) that occurs 25 m above the base of the section (Fig. 3). The unit is mainly composed of thin-bedded (from a few millimetres to 20 cm in thickness) pink, red, or purple nodular limestones (Ammonitico Rosso facies type). At the base of the member, shaly intercalations of up to 25 cm in thickness separate the carbonate beds, whereas higher up, the shale horizons become much thinner. Occasionally, the limestone beds are vertically amalgamated, forming packages separated by very thin clay seams. The topmost horizon of the Paratirolites Limestone and its transition to the overlaying Aras Member are marked by a sharp contact. The Paratirolites Limestone yielded a rather high number of ammonoids (Ghaderi et al., 2014b;Kiessling et al., 2018), while other macrofossils, such as brachiopods and nautiloids, are rare.
Two carbonate horizons deserve particular attention within the Paratirolites Limestone. (1) A 20 cm thick bed with lightgrey to pink colour (at −2.20 m) serves as a marker horizon in the middle part of the Paratirolites Limestone. It is, in contrast to the other beds, a coarser-grained peloidal- foraminiferal packstone to grainstone with few calcareous algae fragments (see below). This level can be recognized in similar facies in the neighbouring sections of Ali Bashi . (2) The top horizon of the Paratirolites Limestone with a 4 cm thickness displays intensive bioturbation and, in the upper 1.5 to 2 cm, a high accumulation of sponge fibres and other sponge remains.

Elikah Formation
The Elikah Formation reaches a total thickness of around 300 m in the NW Iranian sections (Stepanov et al., 1969). Only the basal 7.30 m of the Elikah Formation was analysed and sampled in this study. The Elikah Formation is divided into two members.
The Aras Member ("Boundary Clay") has a thickness of 2.45 m. It consists of purple to pink, greenish-grey, and light to dark grey shales with some marly horizons in the upper part. Some of the horizons show fine laminations caused by ostracod mass occurrences. The top of the unit is placed below the first package of thin-bedded (13 cm thick) light-grey limestone.
Nearly 5 m of the basal part of the Claraia Beds was logged and sampled (+2.50 to +7.30 m). The total thickness of the complete unit is much greater and contains, like in the Ali Bashi Mountains, thick intervals of subvolcanic sills and volcanic rocks. The basal 0.80 m of the unit consists of light-grey shales with a few marly limestone interbeds, which are overlain by platy, thin-bedded limestone beds with varying marl content and shaly interbeds. The limestone beds are commonly grouped in packages up to 25 cm but also occur as individual, single limestone beds of nearly 20 cm thickness. Their colour is usually grey, but in the lower part also brownish and greenish. The member is very poor in macrofossils; only small specimens of Claraia occur frequently at its base. Ostracods occur frequently in the carbonate beds.

Carbonate microfacies
The facies distribution of the Aras Valley succession was previously analysed by Leda et al. (2014), who outlined the depositional setting of the sections (Aras Valley, Ali Bashi, Zal) in the Julfa area. In our study, we reinvestigated the Aras Valley section in more detail, as this section differs in some aspects from the sections of Ali Bashi and Zal.
A total of 42 thin sections of limestone beds from −27.90 to +4.95 m (Figs. 4-7) were analysed regarding their textural and compositional characteristics, using transmitted-light and scanning electron microscopy and energy-dispersive Xray spectroscopy (SEM-EDX) spectroscopy. Based on the variations in lithology, texture, and constituents, a total of nine microfacies types have been identified (Fig. 8); a summary of their main characteristics and the measured carbonate content of each sample can be found in Supplement Ta-ble S1. The textural descriptions follow Dunham (1962) and Embry and Klovan (1971) classifications.

Facies 3: mudstone and wackestone with ostracods
The facies consists of (commonly bioturbated) mudstone and wackestone (Fig. 4c), in which the predominant biogenic components are ostracods (mostly disarticulated), together with undifferentiated thin shell debris, bivalves, and echinoderm fragments (Fig. 4d). Small foraminifers (mainly lagenids), brachiopods, ammonites, sponge spicules of demosponges, peloids, and poorly sorted rounded to sub-rounded micritic intraclasts (up to 3 mm) represent subordinate components. Features of bioturbation and stylolites are common. The matrix is generally recrystallized to microspar and/or dolomitized with occasional occurrences of glauconite grains and glauconitized bioclasts as well as saddle dolomite and manganese dendrites.

Facies 4: burrowed bioclastic mudstone and wackestone
The facies consists of bioclastic mudstone and wackestone (Figs. 4e, 5a-c). Intense bioturbation is indicated by sub-vertical and horizontal burrows (Fig. 5a, b). Main components are ammonites, fine shell debris (mainly ostracods and thin-shelled bivalves), echinoderm fragments (mainly crinoids), ostracods (articulated and disarticulated), and foraminifera (lagenids). Subordinate components are sponge spicules and other sponge skeletal fragments of demosponges, gastropods, brachiopods, putative radiolarians, intraclasts, and glauconite grains. The intraclasts are micritic and show the same composition as the rock matrix. They are poorly sorted (from less than 1 mm up to 5 mm across), rounded to sub-rounded, and occasionally show features of ferruginization. The matrix shows small, irregular cavities filled with calcite cements; the matrix is often patchily recrystallized and/or dolomitized. Fractures, manganese dendrites, stylolites, and seams are abundant and give the rock a breccia-like appearance.

Facies 5: burrowed bioclastic-intraclastic wackestone
Facies 5 has a breccia-like appearance and is compositionally similar to the preceding bioclastic mudstone and wackestone beds of facies 4. The most significant differences are the higher number of sponge fibres of possible keratose sponges and the higher concentration of intraclasts in the upper half of the unit. The number of intraclasts increases towards the topmost part of the interval; the clasts show the same composition as the rock matrix (bioclastic and spicule-rich wackestone). They are poorly sorted (from less than 1 to 5 mm across) and are sub-rounded to subangular. Some of the clasts are coated with dark ferruginous crusts and/or encrusted by foraminifera. The matrix shows small, irregular cavities filled with calcite cements. Most common biogens are ostracods, ammonoids, bivalves, sponge spicules and echinoderms, but also foraminifera, and putative radiolarians (Fig. 5d).

Facies 6: sponge packstone
Facies 6 corresponds to the upper 2 cm of the topmost 5 cm thick limestone bed of the Paratirolites Limestone. It consists of coarse-grained packstone with a micritic (occasionally clotted) matrix and conspicuous accumulations of sponge fibres ("sponge spike" following Leda et al., 2014). The sponges are possibly keratose sponges (sensu Luo and Reitner, 2014), which are characterized by a fibre network and lack of spicules; the skeletal parts of the sponges are replaced by calcite. This sponge packstone overlies the preceding facies with a sharp and irregular contact, recognizable in thin sections (Fig. 6c). The matrix is micritic and occasionally clotted. Articulated skeletons and fibres of sponges are the most common bioclasts in this facies, whereas ammonites, gastropods, ostracods, shell fragments, echinoderms, and glauconite are subordinate. The sponge packstone is burrowed and shows internal clotted peloidal structure and small irregular cavities with geopetal infill within the sponge remains. This characteristic sponge packstone facies marks the extinction horizon in all sections between the Aras Valley and the Ali Bashi Mountains   (Fig. 6c).

Facies 7: burrowed mudstones with sponge remains and calcite crystal fans
The calcareous intercalations (+1.70 to +2.00 m) of the Aras Member consist of burrowed limestones with sponge remains (fibres) and calcite fan cements. The microfacies consists of mudstone with fragmented sponge networks and fibres (possibly of the keratose sponges). The cements consist of millimetre-scale fans (occasionally crusts) formed by large calcite crystals and small cavities filled with calcite cements. Other minor components are disarticulated and complete ostracods and muddy intraclasts. The matrix is often recrystallized to microspar. In comparison to the Baghuk Mountain section , the "calcitic fans" of the Aras Valley section do not appear as branching digitate structures. Often, the fans appear to be slumped and/ or reworked, often with characteristic flat and square terminations, radiat-ing crystal fibres, spherulites, and mineral splays suggesting that they possibly correspond to aragonite pseudomorphs ( Fig. 7a, b). Several types of cement are recognized: (1) microcrystalline with rhombic crystals, (2) bladed crystals with broad, flattened terminations, and (3) heterogranular blocky calcite, characterized by inequigranular, xenotopic, and hypidiotopic fabrics .

Facies 8: bioclastic (gastropod) mudstone and wackestone
Facies 8 consists of mudstones and wackestones with diverse bioclasts. The matrix and grains are often recrystallized and/or dolomitized, often hindering the identification of the components. The most abundant components are small gastropods, sponge fibres and other sponge fragments (possibly of keratose sponges) and ostracods. The sponges show internal clotted fabrics and are often burrowed. Small (≤ 1 mm across) and rounded micritic intraclasts and occasionally millimetre-scale levels with packstone texture formed by a densely packed concentration of small gastropods and ostracods also occur ( Fig. 7c). Coarse-grained calcite spar cements filling large (few centimetres) cavities are visible in the thin section. Bioturbation, pressure-solution seams, and manganese dendrites are also common.

Facies 9: laminated? mudstone with sparry calcite crystals
The facies consists of fine-grained and crudely laminated mudstone and bindstone (Fig. 7d, e), partially recrystallized, with a few microfossils (ostracods and crinoids). The bindstone is composed of very thin, planar to wavy discontinuous layers of dense or clotted micritic layers, alternated with sparitic layers. Closely spaced equant sparry calcite (occasionally celestine) crystals are incorporated within the laminae. Most of the crystals are concentrated in millimetre-scale layers that are occasionally inversely graded (Fig. 7e). The sparry crystals are 0.5-2 mm across and show diverse morphologies varying from more or less geometrical to irregular to spherical. Some of the crystals show uniform extinction or contain small euhedral dolomite rhombi. The matrix surrounding the crystals is often recrystallized and indicates clotted micritic fabrics resembling automicrite. The identification of laminae is not straightforward and is often hindered by the pseudo-laminae created by abundant horizontal pressure-solution seams and stylolites.

Depositional environments
The carbonate litho-and microfacies data (lithology, sedimentary structures, texture, and skeletal and non-skeletal components) collected in the Aras Valley section indicate deposition in open marine, mostly deep, outer-shelf environments dominated by quiet waters.
The upper Julfa Formation and the lower part of the Ali Bashi Formation (facies 1, 3) are composed of shales with numerous carbonate beds, which contain fossils of marine organisms (ammonoids, nautiloids, conodonts, putative radiolarians, echinoderms, brachiopods, thin-shelled bivalves, lagenid foraminifera). These indicate a deposition in distal, relatively deep (below storm wave base) and open marine settings. Moreover, the occurrence of a mud-dominated carbonate facies (mudstone and wackestone) and glauconite, as well as the common to intense bioturbation, suggests deposition under low-energy and normal oxic conditions with low sedimentation rates. The occurrence of coarser grainy facies (peloidal-foraminiferal packstone) in the middle part of the upper Julfa Formation may indicate episodic conditions of higher water energy, possibly in open marine, middle-shelf settings.
The facies of the Paratirolites Limestone shows a marked change into nodular fabrics (Ammonitico Rosso facies type), a decrease in shale content, and an increase in pelagic fauna (e.g. ammonoids, putative radiolarians) in comparison to the underlying strata. These faunal and depositional changes indicate a successive deepening upward trend from the Julfa Formation to the Paratirolites Limestone. Therefore, the nodular, bioturbated bioclastic and bioclastic-intraclastic muddy facies of the Paratirolites Limestone (facies 4) was deposited in an open marine and deeper (below wave base; dysphotic zone) shelf setting, mainly dominated by low water energy and low terrigenous input. The commonly intense bioturbation and the relatively diversified benthic macrofauna and microfauna (e.g. crinoids, brachiopods, bivalves, gastropods, and sponges) indicate oxic conditions. A similar deepening trend has been suggested by Ghaderi et al. (2014b) based on the decrease in abundance and diversity of brachiopods. Episodically, the deep and quiet water environment may have changed into a shallower environment with higher water energy (middle shelf) as supported by the grainy facies with peloids, diverse foraminifers, and calcareous algae (facies 2), which characterize the deposits of the middle part of the Paratirolites Limestone.
The facies of the topmost part of the Paratirolites Limestone (facies 5) does not show sedimentological or palaeontological evidence for large-scale sea-level changes but indicates marked variations in the rate of sedimentation. In agreement with previous interpretations for other central and northern Iranian successions (Heydari et al., 2003;Richoz et al., 2010;Leda et al., 2014), the topmost part of the Paratirolites Limestone (bioclastic-intraclastic limestone) is interpreted here to be deposited in an outer-shelf setting. The occurrence of omission or condensation surfaces, breccialike appearance, high concentration of intraclasts (sometimes coated with Fe crusts or bioencrusted), and the cavities filled with cements are not indicative of subaerial exposure but of early lithification and reworking during conditions of very low sedimentation rates and condensation. Similar interpretations, suggesting a continuous sedimentation without sub- (c) Microfacies sample from the topmost 4 cm of the Paratirolites Limestone (sample AJ200; = 0.00 to −0.04 m). Lower part: burrowed bioclastic-intraclastic wackestone with ammonoids, bivalves and ostracods, lithoclasts, and micrite clasts. Upper part: sponge packstone with ammonoids, bellerophontids, and ostracods; uppermost 10 mm with a densely packed sponge meshwork of possible keratose sponges. Scale bar units = 1 mm. aerial exposure, have been previously made for the NW Iranian Paratirolites Limestone (Altiner et al., 1980;Richoz et al., 2010;Leda et al., 2014). The conditions of starvation and early lithification of the sea bottom, which are observed for the top Paratirolites Limestone, may have favoured the development and colonization of the substrate by sessile sponges (sponge packstone facies), which characterize the top 1-2 cm of this unit (facies 6). This dense accumulation of keratose sponge remains (sponge spike) may mark the EPME event, as keratose sponges have been previously interpreted as post-extinction facies (Friesenbichler et al., 2018;Heindel et al., 2018;Foster et al., 2020). Similar occurrences, but with a stratigraphically younger position, have been reported from other Tethyan PTB sections in Armenia and central Iran (Richoz et al., 2010;Leda et al., 2014;Friesenbichler et al., 2018;Heindel et al., 2018). In these regions, the sponge network remains are usually associated with microbial, crystal fan-like structures, representing post-extinction spongemicrobial build-ups (Friesenbichler et al., 2018;Foster et al., 2020;Martindale et al., 2019), but in this study no clear indication of microbial fabrics is observed.
The facies of the post-extinction Aras Member (Boundary Clay) does not indicate bathymetric or hydrodynamic changes when compared with the Paratirolites Limestone. However, a strong decline and/or cessation of the skeletal carbonate production may have caused the abrupt lithological change from the carbonate-dominated Paratirolites Limestone to the shale-dominated Aras Member. The restart of the carbonate factory in the upper part of the Aras Member is marked by the deposition of burrowed mudstone-wackestone facies with sponges and conspicuous calcite crystal fans and crusts (facies 7). These structures are similar to the calcite crystal fans already known from PTB sections in various regions (Taraz et al., 1981;Baud et al., 1997;Kershaw et al., 1999;Woods et al., 1999;Heydari et al., 2003;Baud et al., 2005a, b;Pruss et al., 2005;Richoz et al., 2010;Friesenbichler et al., 2018). Interpretations of the origin of these structures differ mainly in the hypothesis of whether they were  produced abiotically or biotically (Woods et al., 1999;Wignall and Twitchett, 2002;Heydari et al., 2003;Wignall et al., 2005;Leda et al., 2014;Friesenbichler et al., 2018). In contrast to the "digitate stromatolites" described by Friesenbichler et al. (2018), which are suggested to be of biotic origin, the crystal fans in the Aras Valley section do not show internal lamination. Therefore, we interpret these structures as abiotically formed aragonite crystals, induced by unusual water chemistry (i.e. supersaturated with respect to calcium carbonate), which were later recrystallized to calcite. An increase in total alkalinity, induced by intensified silicate weathering, a "carbonate overshoot" following an ocean acidification event or more passively through the extinction of many calcifying organisms, could have led to a supersaturation of ocean water with respect to calcium carbonate and consequently to the precipitation of the crystal fan structures (Zeebe and Westbroek, 2003;Knoll et al., 2011;Leda et al., 2014;Lehrmann et al., 2015;Foster et al., 2019Foster et al., , 2020. No sedimentary or palaeontological features indicative of shallow-water or high-energy conditions have been recognized in the deposits of the Aras Member in the Aras Valley. Therefore, we interpret that the Aras Member was deposited in an outer-shelf setting with low water energy, comparable to the depositional environment of the Paratirolites Limestone. Similar deep settings have been suggested for neighbouring sections such as Ali Bashi and Zal (e.g. Richoz et al., 2010;Leda et al., 2014). This contrasts with the interpretation of Mohtat Aghai et al. (2009), who postulated that the Boundary Clay was deposited in relatively shallow environments with quiet and stagnant waters, or with the interpretation of Heydari et al. (2003), who suggested a shallow environment based on geochemical data and on the occurrence of reefal and grainy facies coeval with synsedimentary carbonate cements studied in sections near Abadeh (central Iran).
The platy limestone beds of the basal Claraia Beds are mostly composed of mudstone and bindstone, which are poor in fossils and often partially recrystallized. The lowermost part consists of mud-supported facies (facies 8) with scarce fossils (i.e. high-spired microgastropods and ostracods), which are overlain by recrystallized mudstone and crudely laminated bindstone with sub-rounded sparry calcite crystals (facies 9). The occurrence of irregular lamination and clotted texture possibly indicates a microbial origin, but the poorly preserved fabric and the abundant chemical compaction features make any interpretation equivocal. No remains of microbial or algal structures have been recognized. Thus, the origin of these crystals remains unclear. Similar structures have also been observed in association with microbial mats or oncoids in other sections from the central and western Tethys (Hips and Haas, 2006;Richoz et al., 2010;Leda et al., 2014;Foster et al., 2020). Based on the similar morphology and size of the crystals and the skeletal grains, these sparry calcite crystals may be interpreted as recrystallized bioclasts (i.e. crinoids and/or microgastropods) trapped during the halt of the microbial mat growth. Alternatively, the crystals may represent calcified coccoidal remains as suggested by Hips and Haas (2006), but the size of the crystals from the Aras Valley section is much larger. The presence of this microbial facies may indicate an equivalent formation to basal Triassic microbialites. These microbial buildups, which are a typical feature in other Tethyan sections, e.g. in south China (Kershaw et al., 2007(Kershaw et al., , 2012Lehrmann et al., 2015;Bagherpour et al., 2017;Foster et al., 2020), Slovenia (Kolar-Jurkovšek et al., 2018;Foster et al., 2020), Turkey (Baud et al., 1997(Baud et al., , 2005Pruss et al., 2006;Richoz, 2006;Kershaw et al., 2007;Heindel et al., 2018;Foster et al., 2020), Armenia (Friesenbichler et al., 2018;Foster et al., 2020), and central Iran (Taraz et al., 1981;Heydari et al., 2003Heydari et al., , 2008Leda et al., 2014;Foster et al., 2020), are absent in the NW Iranian sections. The lack of clear sedimentary and palaeontological indicators for water depth and water energy in the deposits of the Claraia Beds hinders their palaeoenvironmental reconstruction. However, there are indications of a temporal shallowing-upward trend during this time in the neighbouring Ali Bashi and Zal sections , as suggested by the occurrence of oncoidal facies.

Conodont stratigraphy
For the study of the conodont succession, a total of 97 samples, each of about 2 kg weight, were processed following the standard protocol. The samples were dissolved using diluted acetic or formic acid, sieved with a 0.063 mm mesh, and the dry residue was then processed by magnetic separation.
The conodont biostratigraphy of the Late Permian sedimentary succession of Transcaucasia focused on the Ali Bashi sections (Sweet in  and the Dorasham sections (Kozur et al., 1978(Kozur et al., , 1980. Subsequently, Kozur (2004aKozur ( , 2005 proposed a more precise resolution of these sections including 12 conodont biozones. A revision of this subdivision was suggested by Shen and Mei (2010); they proposed a zonal scheme with eight zones spanning the interval from the late Wuchiapingian C. orientalis zone to the latest Changhsingian C. hauschkei zone.
Ghaderi (2014), Ghaderi et al. (2014b), and Isaa et al. (2016) revised the conodont succession of the sections in the Julfa region. In these studies, the late Permian interval was subdivided into 16 conodont zones, of which six characterize the Wuchiapingian and 10 the Changhsingian (Fig. 9). In these studies, it was also shown as remarkable that the stratigraphic resolution increases towards the end of the Late Permian, as the rock intervals representing distinct conodont zones become thinner because of increased evolutionary rates of the genus Clarkina. The lower Julfa Formation, for instance, belongs to only one conodont zone (C. leveni zone), while the Paratirolites Limestone alone has six conodont zones. Only a part of this pattern may be caused by condensation of the uppermost beds of the Paratirolites Limestone.
The base of the Changhsingian stage in the NW Iranian sections was defined by Ghaderi et al. (2014b) with the C. orientalis-C. subcarinata interval zone, which was taken as equivalent to the absent basal Changhsingian conodont marker species C. wangi (Jin et al., 2001). The stage boundary is located close to the base of the Zal Member. The end-Permian extinction horizon marks the boundary between the C. hauschkei and Hindeodus praeparvus-H. changxingensis zones. The upper limit of the Changhsingian is characterized by the first occurrence of H. parvus, which indicates the official Permian-Triassic boundary (Kozur, 2005(Kozur, , 2007b. The conodont succession of the Aras Valley section in ascending order is as follows ( Fig. 9; for representative species, see Figs. 10, 11).
1. Clarkina leveni interval zone. The lowermost logged and sampled part of the Aras Valley section with 3.30 m thickness belongs to the lower Julfa Formation and the C. leveni zone, which was recorded for the first time by Kozur (1975) in the Achura section of Transcaucasia. C. asymmetrica, H. minutus, and H. typicalis are the most common associated species.
2. Clarkina guangyuanensis interval zone. This interval zone was established by Mei et al. (1994) in the Dukou and Nanjing sections of Sichuan (south China). The first occurrence of the zonal species at the base and the first occurrence of C. transcaucasica at the top define the interval. The C. guangyuanensis interval zone, with 3.30 m thickness in the Aras Valley section, is the first conodont zone in the upper Julfa Formation. C. liangshanensis arises in the upper part of this zone; however, it ranges to the middle part of the C. orientalis zone; H. typicalis is another associated species.
3. Clarkina transcaucasica interval zone. As in the case of the previous zone, the C. transcaucasica zone was first defined by Mei et al. (1994) at Dukou and Nanjing in Sichuan. It comprises the interval from the first occurrence of C. transcaucasica up to the first occurrence of C. orientalis and comprises a thickness of 4.55 m in the Aras Valley section. C. liangshanensis and H. typicalis are co-occurring in the entire interval, while C. guangyuanensis is present in some of the lowest beds.
4. Clarkina orientalis range zone. This is the youngest conodont zone within the upper Julfa Formation. The zone was introduced by Kozur (1975) and is defined by the first occurrence of the zonal species; however, the position of its top is variable in different localities (Shen and Mei, 2010). For instance, the zone is overlain by the C. inflecta zone in the Dukou and Nanjing sections (Mei et al., 1994), but C. orientalis is still present in the C. inflecta zone. In the Meishan section (Zhejiang, south China), it is followed by the C. longicuspidata zone, but C. orientalis ranges into the basal Changhsingian C. wangi zone (Mei et al., 1994;Jin et al., 1997). In the Aras Valley section, the C. orientalis zone has a thickness of 4.15 m and is completely restricted to the grey to beige nodular limestone of the uppermost Julfa Formation. Associated species in the C. orientalis zone are C. transcaucasica and C. liangshanensis, which are mostly common in the lower parts of this zone.

5.
Clarkina orientalis-Clarkina subcarinata interval zone (Clarkina wangi zone). The zone is used here for the base of the Changhsingian stage in the NW Iranian sections. In the Aras Valley section, it is limited to the 1.65 m fossiliferous interval between the last occurrence of C. orientalis and the first occurrence of C. subcarinata (Ghaderi et al., 2014b). The zone is accompanied by the species of H. typicalis and the first occurrence of H. julfensis, with the latter mostly considered to be a Changhsingian hindeodid species (e.g. Metcalfe, 2012).
According to Kozur (2005), the C. orientalis zone is overlain by the "Dorashamian" (Changhsingian) C. hambastensis zone in Iran. However, C. hambastensis is limited to the sections in the Abadeh region and has never been recovered from the NW Iranian sections (Ghaderi, 2014) including the Aras Valley section. In the Meishan section of south China, the C. wangi zone was considered the first Changhsingian conodont zone by Mei et al. (2004). Neither Shen (2007) nor Shen and Mei (2010) strictly regarded the C. wangi zone as the overlaying biozone of the C. orientalis zone in Iran, but they did not present any evidence to confirm this claim. In our investigations, no trace of C. wangi and C. hambastensis could be found in the Aras Valley section or any other section of the Julfa region.
6. Clarkina subcarinata interval zone. The nominal species was coined by Sweet in  as the subspecies Neogondolella carinata subcarinata and amended by Mei et al. (1998b) as C. subcarinata. This biozone has a thickness of 4.58 m in the Aras Valley section and is here the thickest Changhsingian biozone. Its upper boundary is limited to the first occurrence of C. changxingensis. H. typicalis and H. julfensis are associated species.
7. Clarkina changxingensis interval zone. The first occurrence of C. changxingensis indicates the zone base and the first occurrence of C. bachmanni validates its top. Kozur (2005) stated that C. bachmanni follows the C. subcarinata zone and C. changxingensis occurs in association with C. deflecta in an assemblage zone above the C. nodosa zone. This cannot be confirmed and the actual position of the C. changxingensis zone, as Shen and Mei (2010) and Ghaderi et al. (2014b) demonstrated, is   8. Clarkina bachmanni range zone. This zone was firstly described by Kozur (2005) for the Changhsingian sections of Iran. The zone is defined by the total range of C. bachmanni which overlays the significant presence of hindeodids on top of the previous C. changxingensis zone. According to Chen et al. (2008), C. bachmanni represents transitional morphotypes between the round and narrow ones. C. changxingensis and C. de-flecta are associated species and the zonal thickness is only 0.38 m.
9. Clarkina nodosa range zone. This zone was established by Kozur (2005) as a total range zone of the naming species C. nodosa. It has a thickness of 0.50 m in the Aras Valley section; associated species are C. postwangi and C. deflecta.
10. Clarkina yini interval zone. In the Changhsingian stratotype section in Meishan (south China), this biozone comfortably overlays the C. changxingensis zone and is overlain by the C. meishanensis zone (Mei et al., 1998a;Yuan et al., 2014). In contrast, the sections in NW Iran show biozones between these two zones; they are defined by C. bachmanni and C. nodosa. The C. yini 11. Clarkina abadehensis range zone. An equivalent was introduced as the Clarkina iranica zone by Kozur (2005), but Shen and Mei amended the species to C. abadehensis. According to Ghaderi et al. (2014b) the biozone comprises the whole range of C. abadehensis (with both the supposed subspecies C. abadehensis abadehensis and C. abadehensis iranica). Associated species of this biozone are C. meishanensis, C. tulongensis, C. taylorae, and H. bicuspidata. It is only 0.18 m thick in the Aras Valley section.
12. Clarkina hauschkei range zone. This is the topmost conodont zone in the Ali Bashi Formation; it comprises only the top 0.13 m of the Paratirolites Limestone up to the extinction horizon. The C. hauschkei range zone was firstly defined by Kozur (2005) based on the full range of the nominal species and shows the occurrence of the two subspecies of C. meishanensis meishanensis and C. meishanensis zhangi. In addition, C. tulongensis and C. taylorae are present in this biozone. It is the thinnest of the conodont zones in the section.
13. Hindeodus praeparvus-Hindeodus changxingensis assemblage zone. Kozur (2005) defined, at the base of the Aras Member, an assemblage zone equivalent to the C. meishanensis-H. praeparvus zone. This biozone thus immediately overlays the mass extinction horizon. Notwithstanding, C. meishanensis, which is already present at the base of the C. hauschkei zone, is very rare at the top of the Paratirolites Limestone in the NW Iranian sections. In the Aras Valley section, this species is completely absent in the Aras Member. In contrast, H. changxingensis, which first occurs in the lowest samples of the Aras Member in all of the sections, is very abundant. Therefore, Ghaderi et al. (2014b) suggested the introduction of the H. praeparvus-H. changxingensis assemblage zone as a replacement of the C. meishanensis-H. praeparvus zone of Kozur (2005). The thickness of this biozone is 2.30 m in the Aras Valley section.
14. Merrillina ultima-Stepanovites mostleri assemblage zone. In the Aras Valley section, this biozone contains ramiform elements of Merrilina ultima and Stepanovites mostleri in association with H. praeparvus and rare Merrilina postdivergens. The biozone was firstly established by Kozur (2005); it is considered to be correlated with the C. zhejiangensis zone of Mei et al. (1998a) in the Meishan section (Kozur, 2005). In agreement with the previous studies by Ghaderi et al. (2014b), the presence of this biozone with 0.25 m thickness is confirmed here in the Aras Valley section.
15. Hindeodus parvus interval zone. The base of this biozone is defined by the first occurrence of H. parvus and marks the beginning of the Triassic system (Yin et al., 2001). This key level is situated at a position of +2.55 m in the Aras Valley section; the biozone has a thickness of 1.55 m. Accompanying species within this biozone are H. anterodentatus and H. eurypyge.
16. Isarcicella staeschei interval zone. This biozone is defined by the first occurrence of I. staeschei (Wang, 1996;Wang and Wang, 1997 17. Isarcicella isarcica zone. The first occurrence of I. isarcica is the marker for definition of another conodont zone on top of the I. staeschei interval zone (Wang and Wang, 1997;Perri and Farabegoli, 2003). The I. isarcica zone in the Aras Valley section continues till the end of the logged section and thus has at least 2 m thickness.

Ostracod succession
The Aras Valley section was sampled for ostracods from −3.80 to +7.00 m with respect to the extinction horizon; this covers almost the entire Paratirolites Limestone, the Aras Member, and the lower part of the Claraia Beds. In total 59 samples were taken for subsequent processing. The limestone samples (around 400 g each) were processed by hot acetolysis as described in Crasquin et al. (2005). The clay material (1 kg each) from the Aras Member was treated with water and, in some cases where it was necessary, with tensides and was subsequently wet sieved. For the sieving processes we used two sieves with mesh widths of 0.063 and 1.5 mm.
The ostracod taxonomy of the nearby Dorasham II section was already investigated by Belousova (1965); her study comprised 500 late Wuchiapingian to late Changhsingian specimens of 26 species. Belousova showed that there was a gradual taxonomic change between the assemblages with a relatively constant diversity. However, Belousova (1965) and the subsequent investigation by Kotlyar et al. (1983), who described a less diverse assemblage from the Paratirolites Limestone, only described the succession below the extinction horizon, not allowing any investigation of the extinction event itself.
Ostracods were present in all 59 samples. The abundances of ostracods through the studied interval range between four and approximately 31 500 specimens per 500 g of sample material (Fig. 12). In total, 3428 specimens were identified at the species level. A total of 62 species are present, belonging to 23 genera and 12 families (Fig. 13). The three sampled lithological units show the following characteristics in terms of abundance, diversity, and taxonomic patterns.
1. Paratirolites Limestone. Most of the Paratirolites Limestone samples show specimen numbers between 200 and 300, but higher occurrences reach up to 700 specimens. The diversity is generally low with only four species in most of the sample horizons; however, it reaches eight species at −0.40 m in the section. The highest Paratirolites limestone sample at −0.01 m, corresponding to the sponge-rich layer, yielded a monospecific assemblage, consisting exclusively of Fabalicypris obunca Belousova, 1965.
The assemblages of the Paratirolites Limestone are mainly composed of specimens of the order Podocopida, generally dominated by the genus Fabalicypris ( Fig. 12l-m). The most common species in this preextinction interval is Fabalicypris obunca. Hungaroleberis sp. nov. (Fig. 12o) as the only myodocopid species in the section is the only exception among the predominant podocopid species and occurs in the interval from −3.30 to −1.40 m.

Aras Member. The abundance distribution in the Aras
Member is very heterogeneous, ranging from 6 to 31 500 specimens per 500 g of sample material. The samples between +0.45 and +0.75 m and at +1.25 m are marked by very low values of 6 to 46 specimens. Higher up in the section, at +1.71 m, there is a characteristic shale layer with ostracod mass occurrences of around 31 500 specimens per 500 g (Fig. 14). Species richness is increasing in the Aras Member, starting just above the extinction horizon and reaching its maximum of 16 at +1.16 m in the section. A subsequent diversity drop to four species occurs at +1.71 m and is followed by an increase that extends up to the overlying Claraia Beds.
The first diversity increase, which begins immediately above the extinction level, is paralleled by a complete faunal turnover (Fig. 13). The poorly diverse Fabalicypris-dominated community of the Paratirolites Limestone was replaced by a relatively diverse Bairdiacypris-dominated community. This faunal turnover takes place at the transition from the Paratirolites Limestone to the Aras Member. The nearby Zal section, which was studied by Mette (2008Mette ( , 2010, records a similar pattern of complete faunal turnover with the first deposition of the Boundary Clay, with the crucial difference that the assemblage of the Aras Member in the succession of Zal is of low diversity and dominated by a single species of the genus Indivisia. In contrast, in the Aras Valley section, the lower to middle part of the Aras Member yielded a diverse assemblage, consisting of mainly Bairdiacypris ottomanensis Crasquin-Soleau, 2004, but also of species of the orders Platycopida and Palaeocopida, such as species of Cavellina and Carinaknightina (Fig. 12g, j, k).
3. Claraia Beds. The lower 70 cm of the Claraia Beds shows relatively high abundances of about 1000 to 2500 specimens per 500 g sampled material, but higher samples are characterized by very low abundances that do not exceed 40 specimens. The accompanying diversity records a peak of 15 species at +2.65 m, followed by decreases in higher beds (from +3.10 m) to values between two and eight species per sample horizon. Up to around +6.05 m in the section, the assemblages are dominated by B. ottomanensis. In the uppermost sample (+7.00 m), the taxonomic composition changes completely. In this sample, the specimens are mainly represented by a species of Bairdia together with Liuzhinia (Fig. 12c) and only one specimen of B. ottomanensis.
Most of the investigated ostracods from the Aras Valley are smooth-shelled species of the family Bairdiidae. The same pattern was found at the nearby Dorasham II section, where the majority of the 26 identified species also belong to the Bairdiidae (Belousova, 1965). In general, the occurrences of Bairdiidae are considered to indicate normal, oxygenated, open marine conditions (Melnyk and Maddocks, 1988;Crasquin-Soleau et al., 1999).

Ammonoid stratigraphy
The Aras Valley section is the NW Iranian section that has yielded the most diverse ammonoid assemblages to date (Fig. 15). However, the abundance of specimens varies strikingly between the section intervals. While the Julfa Formation is rather poor in specimens, their number increases in the Zal Member, and finally the Paratirolites Limestone is comparatively rich in ammonoids. A scheme for the biostratigraphic subdivision based on ammonoids of the Permian-Triassic boundary beds in the area of Julfa was proposed by . They separated seven units (of which the upper four were, by Ruzhencev and Shevyrev, placed into the Triassic) now regarded as representing the Wuchiapingian and Changhsingian, in ascending order by the genera Araxoceras, Vedioceras, Phisonites, Tompophiceras, Dzhulfites, Bernhardites, and Paratirolites. Kotlyar et al. (1983) used these genera, after a revision by  and named biozones, from bottom to top with the species Arax-  oceras latissimum, Vedioceras ventrosulcatum, Phisonites triangulus, Iranites transcaucasinus, Dzhulfites spinosus, Shevyrevites shevyrevi, and Paratirolites kittli. Kotlyar et al. (1989) and Zakharov (1992) added another ammonoid zone at the top, the Pleuronodoceras occidentale zone; this zone should be represented in the lowest 2 m of the "lower Karabaglyar Formation", corresponding to the Aras Member of the Elikah Formation.
This general scheme of ammonoid stratigraphy can also be applied in the Aras Valley section, but as Ghaderi et al. (2014b) and Korn et al. (2016) have shown, the ammonoids of the Paratirolites Limestone do not represent only one single zone (Paratirolites kittli zone) but a succession of clearly separable units. By contrast, Leonova (2016), in her review on the Permian ammonoid stratigraphy, proposed a much less detailed zonation for the late Permian ammonoid succession. She only accepted three global ammonoid zones for the Changhsingian, the Phisonites triangulus zone, Paratirolites kittli zone, and Pleuronodoceras multinodosum-Rotodiscoceras asiaticum zone. This is remarkable in two aspects. (1) It neglects the rapid turnover rates of the ammonoid genera and species, as demonstrated by empirical data from the Julfa region (e.g. Kotlyar et al., 1983;Ghaderi et al., 2014b;Korn et al., 2016Korn et al., , 2019a. (2) It accepts the previous assumption by Zhao et al. (1978) that the Transcaucasian genus Paratirolites is stratigraphically older than the south Chinese genera Pleuronodoceras and Rotodiscoceras. As Korn et al. (2016) pointed out, such an assumption was probably based on misidentification of specimens from south China as belonging to Paratirolites by Zhao et al. (1978).
From Aras Valley, more than 200 ammonoid specimens are available for study, of which the majority are from the Paratirolites Limestone. Because of the scarcity of ammonoids in the Wuchiapingian part of the section, most of the material from this interval was not collected in situ but from float and can thus only roughly be attributed to distinct rock units. The following ammonoid units can be separated in the Aras Valley section (Fig. 16).
Lower Julfa Formation (Araxoceras beds). Apart from the goniatite Pseudogastrioceras, the araxoceratid genera Araxoceras, Prototoceras and Vescotoceras are most common. With this composition, the Aras Valley section complements the neighbouring Dorasham II section , but the number of specimens appears to be lower.
Zal Member (Iranites transcaucasinus-Phisonites triangulus zone to Shevyrevites shevyrevi zone). The ammonoids of this member from NW Iran have been monographically described by Korn et al. (2019a), including specimens from the Aras Valley section. Phisonites triangulus occurs in a single platy and marly limestone bed near the base of the Zal Member at −12.90 m. It is accompanied by poorly preserved specimens of Xenodiscus dorashamensis. As Ghaderi et al. (2014b) figured out, Iranites transcaucasinus occurs already below at −13.50 m, and hence the two zones separated by  need to be merged.
The Dzhulfites nodosus zone is recorded, in the Aras Valley section, with the two species Dzhulfites nodosus and D. spinosus, which first occur at −9.50 m. A second horizon in the Aras section lies at −8.20 m, where Pseudogastrioceras relicuum occurs with specimens of Dzhulfites. Higher horizons with Dzhulfites occur at −7.60 and −6.15 m. The zone is thus rather well-represented in the Aras section, although the diversity is, with a total of five species, still rather low.
In the Aras section, the Shevyrevites shevyrevi zone is best seen in a rather fossiliferous purple shale package about 0.80 m below the base of the Paratirolites Limestone at −5.40 m. Unfortunately, the ammonoid specimens are mostly fragmented in this bed and only rarely allow the study of more than one volution. Araxoceltites cristatus is by far the most common taxon; it is accompanied by specimens of Shevyrevites and rare Dzhulfites (Korn et al., 2019a). Shevyrevites has obviously only a very limited stratigraphic range and characterizes only this thin interval at the top of the Zal Member below the Paratirolites Limestone.
Paratirolites Limestone. The ammonoid succession of the Paratirolites Limestone was outlined in detail by . A total of 140 specimens were then available from the Aras Valley section; 90 of these were collected in situ with precise distance from the extinction horizon. Eight ammonoid zones can be separated, from the Dzhulfites zalensis zone to the Arasella minuta zone, of which all index ammonoids were recorded from Aras Valley. These zones show a succession of the genera Dzhulfites, Paratirolites, Alibashites, Abichites, and Arasella, which occur with some overlap. The diversity dynamics, size decrease, and morphological simplification were illustrated and discussed by Kiessling et al. (2018).
Neither ammonites nor nautiloids were recorded from the post-extinction Aras Member; only rare and poorly preserved small ammonoids occur in the basal Claraia Beds. We interpreted this disappearance of cephalopods at the extinction horizon as true extinctions of the taxa. A decrease in the preservation potential for calcitic fossils from the Paratirolites Limestone to the Aras Member, which may also mimic an extinction, is rather unlikely in this case. Ammonoids occur, for instance, in a similar lithofacies in the Zal member as impressions in shale or 3-D specimens in marly nodules.
It was shown in previous articles that the late Permian nautiloid assemblages from Dorasham (Shimansky, 1965) and Ali Bashi  are remarkable for their high diversity when compared with other Permian occurrences. Already Shimansky (1965) showed that particularly the lower Julfa Formation contains diverse nautiloids with the main nautiloid groups represented.
A total of 62 coiled nautiloid specimens are available from the Aras Valley section, of which some are figured here (Fig. 17). There is a steep decrease in the frequency of specimens within the section: 53 come from the lower Julfa Formation, six from the upper Julfa Formation, two from the Zal Member, and only one from the Paratirolites Limestone. Parallel to the frequency of nautiloid specimens, there is a decrease in genus and species richness within the section (Table 1). In the lower Julfa Formation occur at least six genera and 15 species of coiled nautiloids, and each of the four major late Permian nautiloid superfamilies is represented: Tainoceratoidea (Metacoceras, Pleuronautilus), Trigonoceratoidea (Endolobus), Aipoceratoidea (Domatoceras), and Liroceratoidea (Liroceras, Permonautilus). These taxa cover almost the entire morphological spectrum of late Permian nautiloid conch geometry, from thinly discoidal to globular and from widely umbilicate to involute (Korn et al., 2019b). It is evident that this early Wuchiapingian nautiloid assemblage differs, in its generic composition, from the other two major regions with time-equivalent nautiloid assemblages, the Salt Range (Waagen, 1879;Reed, 1931Reed, , 1944, and south China (Zhao et al., 1978).
The six nautiloid specimens from the upper Julfa Formation belong to five genera and only two superfamilies are present: Tainoceratoidea (Tainoceras) and the dominant Liroceratoidea (Liroceras, Permonautilus, Peripetoceras, Coelogasteroceras). It is interesting that this assemblage resembles, in the morphological spectrum, the Changhsingian nautiloid assemblage from the Bellerophon Formation of the Dolomites (Prinoth and Posenato, 2007).
The impoverished nautiloid assemblages from the Zal Member (with only the two tainoceratoid genera Pleuronautilus and Tirolonautilus) and Paratirolites Limestone (with only one single poorly preserved specimen of Paranautilus) show that the nautiloid-dominated cephalopod communities of the lower Julfa Formation have been replaced by ammonoid-dominated cephalopod communities towards the end of the Permian sedimentary succession.

Carbon isotopes
A total of 127 carbonate samples from −19.95 to +5.40 m, spanning from the upper part of the Julfa Formation, to the Ali Bashi Formation including the Paratirolites Limestone, the Aras Member, and the lower part of the Claraia Beds, were taken for the study of δ 13 C carb (Fig. 18, Table S3). The carbon isotopic ratios were measured from resulting CO 2 using an isotope-ratio mass spectrometer (IRMS) of the type Thermo-Finnigan Delta V Advantage coupled with a Gas-Bench II at the Museum für Naturkunde Berlin (Schobben et al., 2017). The used standard "shattercone" reference material was created by the Museum für Naturkunde Berlin by calibrating the homogenized carbonate against Vienna Pee Dee Belemnite (VPDB) using the IAEA reference materials NBS-18 and NBS-19. The resulting isotope ratios are expressed in the conventional delta notation (δ 18 O and δ 13 C) in relation to VPDB.
Following stable δ 13 C carb values of around +4 ‰ in the Wuchiapingian to early Changhsingian part of the section, a first-order δ 13 C carb trend to 4 ‰-5 ‰ lower values starts at the base of the C. yini zone above the marker bed at −2.20 m (AV190; peloidal-foraminiferal packstone). It reaches a value of +1.2 ‰ immediately below the extinction horizon in the uppermost centimetres of the Paratirolites Limestone. The negative trend continues throughout the Aras Member; it reaches the zenith with the most negative δ 13 C carb value (−0.7 ‰) in the lowermost sample of the Hindeodus parvus zone at the base of the Claraia Beds.
The carbon isotope composition of bulk micritic rock has become an important correlative tool for the PTB strata (Scholle and Arthur, 1980;Jenkyns and Clayton, 1986;Baud et al., 1989;Korte and Kozur, 2010). Global correlative schemes based on these chemostratigraphic records hinge on the observation that the isotope signal traces the ancient oceanic dissolved carbonate (DIC)-carbon isotope composition (Kump, 1991;Dickens et al., 1995). Due to the long residence time of DIC in the marine reservoir and changes in the cycling of carbon between the Earth's surface reservoirs, the so-called biogeochemical carbon cycle should result in C isotopic excursions of the same sign across the world's oceans (but see Swart, 2008, for a different view). These trends in DIC carbon isotope composition can then be recorded in carbonate minerals that are precipitated in close equilibrium to seawater (such as shells, tests) (Marshall, 1992). This primary observation underlies the assertion that broad-scale stratigraphic trends to both negative and positive δ 13 C carb excursions can be correlated between sedimentary sequences around the globe. For instance, Early Triassic carbon isotope trends recorded in the Zal section have been correlated globally (Clarkson et al., 2013;Grasby et al., 2013). However, the degree to which stratigraphically relevant information is captured in δ 13 C carb records can vary and depends on many factors, such as carbonate mineralogy changes (Brand et al., 2012), diagenetic resetting (Marshall, 1992;Schobben et al., 2016), sedimentary continuity in conjunction with the sampling resolution, and bioturbation intensity (Schobben et al., 2017). As such, we refrained from an overly detailed description of the here-produced δ 13 C carb record, and we only describe first-order trends over broad stratigraphic intervals (Schobben et al., 2019).   The first-order stratigraphic pattern in the Aras Valley sequence is compatible with δ 13 C records from various regions (e.g. Dolomites, Bűkk Mountains, Oman, central and NW Iran, Tibet, and Kashmir) (Korte and Kozur, 2010;Schobben et al., 2017), as well as organic carbon trends in western and Arctic Canada and Spitsbergen (Grasby and Beauchamp, 2009;Grasby et al., 2013Grasby et al., , 2015Wang et al., 1994). The lowest δ 13 C values were documented near the PTB in the Aras Valley section in the uppermost sample of the Aras Member in the M. ultima-S. mostleri Assemblage zone. This zone can be correlated with the upper part of the H. changxingensis-C. zhejiangensis zone at Meishan, which shows a similar δ 13 C carb excursion (Yuan et al., 2014).
This trend occurs over an estimated time span of 0.5 to 1 Myr when considering the independent biostratigraphic information gained in our work and updated age models (e.g. Burgess et al., 2014Burgess et al., , 2017Black et al., 2018). Several scenarios have been put forward over the last decades that tried to identify the external force behind this chemostratigraphic marker. Most scenarios indicate a global-scale carbon cycle perturbation that simultaneously relates to the dramatic ecological and biological disaster unfolding at this system boundary (e.g. Berner, 2002;Xie et al., 2007;Rothman et al., 2014), although some view the δ 13 C carb excursion as a result of the extinction (Rampino and Caldeira, 2005). A sudden ocean overturn of a previously stagnant ocean resulting in the upwelling of 13 C-depleted DIC and erosion of organic-rich sediments stored on the continental shelf forced by a global sea-level drop have been implicated in the formation of limestone with a low δ 13 C value (Knoll et al., 1996;Berner, 2002). Another frequently proposed scenario is the catastrophic destabilization of methane clathrates, injecting 13 C-depleted carbon in the exogenic carbon reservoirs (Benton and Twitchett, 2003). However, in view of recent radiometric dating developments, these scenarios are generally assumed to have acted on short timescales (< 100 kyr); they are incompatible with the long duration of the negative inflection of the PTB δ 13 C record (e.g. Burgess et al., 2014Burgess et al., , 2017Majorowicz et al., 2014;Black et al., 2018). In addition, a sud-den ocean overturn is incompatible with currently favoured scenarios involving a well-ventilated ocean superseded by an anoxic and more sluggish ocean circulation (Hotinski et al., 2001;Winguth and Winguth, 2011), and, although a shortterm regression does mark the PTB beds, a long-term major transgression follows this more subtle sea-level drop (Hallam and Wignall, 1999;Grasby and Beauchamp, 2009). It thus follows that these scenarios are likely only of moderate applicability for driving the negative δ 13 C trend. Moreover, the resistance of the DIC pool to a short-term C injection by methane clathrate dissociation might have been much larger due to the much larger size of this marine chemical reservoir comparative to the modern ocean (Ridgwell, 2005). In spite of this, sub-100 kyr carbon isotope excursions (CIEs) superimposed on the first-order trend, as recognized by some authors, might be compatible with previously cited scenarios. The global importance of sub-100 kyr CIEs has yet to be proven, and we therefore refrain from using them for correlative purposes in this study.
On a more positive note, updated age models can also help to better deconvolve the nature and drivers of the negative inflections in the PTB δ 13 C record (e.g. Burgess et al., 2014Burgess et al., , 2017Baresel et al., 2017;Black et al., 2018). Grasby et al. (2013) showed that both negative and positive CIEs through the late Permian to Early Triassic are consistent in both carbonate carbon and organic carbon records; this rules out something like changes in rates of marine organic carbon burial and supports changes to the external C pool as the driver for the CIEs.
The state-of-the-art radiometric ages do provide robust support for coeval massive magmatic activity associated with the large igneous province (known as the Siberian Traps basalts) and the first-order negative CIE (e.g. Burgess et al., 2014). More specifically, the lowest δ 13 C values seem to coincide with an episode of extensive intrusive basalt emplacement of this LIP (Burgess et al., 2017). This intrusive phase of Siberian Traps basalts would encompass sill emplacement in the sedimentary basins beneath the flood basalts. Such intrusions baked the rock units in which the magma penetrated, which included both evaporites and organic-rich strata, causing the release of large amounts of halocarbons and 13 Cdepleted carbon gas. Supporting evidence comes from abundant breccia pipes in Siberia, which formed as a result of explosive, gas-release events (Svensen et al., 2009;Burgess et al., 2017). Besides being the harbinger of environmental destruction, venting of greenhouse gas would have introduced 13 C-depleted carbon in the exogenic carbon reservoirs. This observation is important as it is normally assumed that volcanogenic carbon dioxide has a δ 13 C value of −5 ‰, which is only moderately lighter than the value in the oceans and atmosphere to which it is being added. Thermogenic gas release with δ 13 C values that could have approached as much as −25 ‰, or even lower (∼ −50 ‰) when concerning methane gas, therefore has a higher potential to have caused the negative carbon isotope excursion as recorded in limestone, without having to invoke unrealistically high carbonrelease scenarios (Berner, 2002). Identifying the cause for observed carbon isotope fluctuations remains challenging, an effect, above all, related to limitations of the isotope recording potential of bulk carbonate rock and uncertainties in flux estimates. The release of greenhouse gas by intrusive volcanism provides, however, a promising prospect in the search for a common driver of the negative C isotope inflection and the causes behind this mass extinction event. In addition, this volcanically induced carbon gas release might have generated a geochemical anomaly that can serve as a first-order feature in a chemostratigraphic scheme to correlate PTB beds from across the globe (Korte and Kozur, 2010;Schobben et al., 2019). Hence, the negative carbonate-carbon isotope excursion as recorded in the Aras Valley succession is supportive of a relatively complete sequence of the PTB interval.

Conclusions
The multidisciplinary investigation of the Aras Valley section shows that this complete sedimentary succession spanning the Permian-Triassic boundary has great potential for studying the change of environmental conditions that paralleled the end-Permian mass extinction. The lithological succession; carbonate microfacies characteristics; stable isotope dynamics; and conodont, ostracod, and ammonoid stratigraphy allow for a detailed study of the chronological succession of events and correlation with other PTB sections. In summary, the investigation of the fossil inventory and the geochemical patterns contains the following information.
1. The lithological succession allows the clear separation of three formations (Julfa, Ali Bashi, and Elikah formations), all of which can be subdivided in two members.
2. The end-Permian extinction horizon is marked by the occurrence of a sponge packstone in the uppermost Paratirolites Limestone, which is followed by the shaledominated Aras Member (Boundary Clay).
3. The carbonate microfacies analysis indicates an open marine setting with a general deepening trend from the Julfa Formation (Wuchiapingian), continuing through the Zal Member and the Paratirolites Limestone (end-Changhsingian). Small-scale changes appear to indicate shallower settings, which are indicated in two horizons in the upper Julfa Formation and the Paratirolites Limestone. The extinction horizon in the uppermost part of the Paratirolites Limestone is marked by the occurrence of a sponge packstone, consisting of keratose demosponges (sensu Luo and Reitner, 2014). This characteristic sponge spike is a typical marker bed for the end-Permian mass extinction in all sections between the Aras Valley and Ali Bashi . The subsequent transition to the Aras Member does not show significant bathymetric changes. However, cessation of the carbonate production caused a lithological change to shale-dominated deposits. Calcitic fan-like structures, typical sediment structures of post-extinction deposits, suggest an unusual seawater chemistry with possibly increased total alkalinity. Possible microbial structures were identified in the Claraia Beds, which may represent a formation equivalent to Early Triassic microbialites of other Tethyan sections. The facies analysis does not reveal any indications about a bathymetric change within the Claraia Beds; however, nearby sections (Ali Bashi and Zal) indicate a temporally shallower environment .
4. The fossil content of the Aras Valley section is exceptionally high compared to time-equivalent sections in the Transcaucasian-NW Iranian region.
a. The conodonts show a complete succession with four Wuchiapingian, 10 Changhsingian, and three Griesbachian zones; all carbonate beds yielded productive samples. The Permian-Triassic boundary can precisely be determined at 2.55 m above the extinction horizon.
b. The ostracod analysis reveals a local extinction and complete faunal turnover, immediately at the base of the Aras Member (Boundary Clay). The speciespoor Fabalicypris-dominated pre-extinction assemblage changes to a diverse Bairdiacypris-dominated post-extinction assemblage in the Aras Member. The middle part of the Aras Member is marked by mass occurrences mainly of Bairdiacypris ottomanensis, which may be regarded as a disaster taxon, appearing in several Tethyan sections at the end-Permian mass extinction event. The high abundance of Bairdiidae in the whole succession suggests an open marine environment with normal marine salinity and oxygen concentrations.
c. The nautiloid and ammonoid successions indicate a change from a shallow-water nautiloid-dominated cephalopod community in the lower Julfa Formation to a deep-water ammonoid-dominated cephalopod community towards the extinction horizon. Neither of the groups were recorded from the post-extinction Aras Member; only rare and poorly preserved small ammonoids occur in the basal Claraia Beds.
5. The carbon isotope results document a negative δ 13 C carb trend starting 2 m below the extinction horizon, above the marker bed of the peloidal-foraminiferal packstone.
The δ 13 C carb values range from 4 ‰ in the Wuchiapingian and early Changhsingian to −0.7 ‰ in the lowermost sample of the Hindeodus parvus zone. This firstorder δ 13 C signal can serve as a chemostratigraphic marker for the correlation with other PTB successions.
Data availability. Data (CaCO 3 content, δ 13 C values) related to this article are available in the Supplement.
Author contributions. JG, NG, AG, LL, MS, and DK collected the investigated material in the field. US performed the lab analysis of stable isotopes. The various aspects were studied by AG, DK, and LL (lithological succession); LL and ST (facies); MS and US (stable isotopes); AG (conodonts); JG and MBF (ostracods); DK and AG (ammonoids); and DK and NGT (nautiloids). All authors contributed to the concept of the study and writing of the manuscript.