Integrated stratigraphy and palaeoclimate history of the Carnian Pluvial Event in the Boreal realm; new data from the Upper Triassic Kapp Toscana Group in central Spitsbergen (Norway)

The Late Triassic climate is characterized by arid conditions interrupted by a humid phase known as the Carnian Pluvial Event (CPE). This wet phase is well documented in the Tethyan realm, but evidence from the Boreal realm is scarce. Here we present evidence from quantitative palynology for the CPE from the Kapp Toscana Group on central Spitsbergen integrated with organic carbon isotope data linked to the geomagnetic polarity time scale. Our data reveal an early to mid-Julian-1 age for the Tschermakfjellet Formation. The transition to the Julian-2 is located in the De Geerdalen Formation and the Isfjorden Member is confirmed as mostly Tuvalian-3 in age. The Aulisporites astigmosus pollen assemblage zone that marks the base of the CPE in the Tethys realm at the base of the Julian-2 is assigned to the Julian-1 in the Boreal region. Palaeoclimate proxy data inferred from principal component analysis indicate wetter conditions from the Julian-2 onwards, which is in agreement with the establishment of local swamp vegetation on top of a delta plain. The palaeotemperature curve indicates a period of cooler climate during the early Julian-1 followed by warming during the late Julian-1. Supplementary materials: A list of all identified morphotaxa of pollen, spores and aquatic palynomorphs is available at http://www.geolsoc.org.uk/SUP18879.

The CPE has not been identified at high latitudes, but Hochuli & Vigran (2010) described a climate shift from relatively dry conditions in the early Carnian to more humid conditions in the later Carnian from the Barents Sea. This change is loosely linked with more widespread coal deposits in the Boreal Carnian. Roghi et al. (2010) suggested that the interval characterized by an increase in Aulisporites and fern spores in the Boreal region (e.g. Hochuli et al. 1989;Hounslow et al. 2007a) is contemporaneous with the CPE in the Tethys realm, but confirmation of this long-distance biostratigraphic correlation by independent stratigraphic data is lacking. The Carnian was also an interval of significantly increased sedimentation rates throughout the Barents Sea region (Riis et al. 2008), a feature loosely linked with progradation of deltaic systems across the Barents Sea shelf, which form part of a longer-term process of accommodation space filling in the Barents Sea since the early Triassic (Riis et al. 2008).
Here we aim to (1) establish a quantitative palynological study of the Carnian in central Spitsbergen, (2) interpret vegetation changes and palaeoclimate trends using principal component analysis and (3) integrate stratigraphic analysis using organic carbon isotopes and magnetostratigraphy to provide an independent detailed chronology for the palynostratigraphic correlation and palaeoclimatic interpretations.

The Boreal Late Triassic
The Middle Triassic on Spitsbergen is characterized by deposition of black organic-rich shales, the Botneheia Formation of the Sassendalen Group. Near the transition to the Upper Triassic (the base of the Kapp Toscana Group), the Botneheia Formation is followed by shales and sandstones of the Tschermakfjellet and the De Geerdalen Formations,

Integrated stratigraphy and palaeoclimate history of the Carnian Pluvial Event in the Boreal realm; new data from the Upper Triassic Kapp Toscana Group in central Spitsbergen (Norway)
which were deposited in an offshore shelf, transitional to coastal and deltaic environments (Buchan et al. 1965;Mørk et al. 1982Mørk et al. , 1992Mørk et al. , 1999Mørk & Bjorøy 1984;Mørk & Worsley 2006;Riis et al. 2008;Nagy et al. 2011). Early Norian parts of the Spitsbergen Triassic record a return to marine depositional conditions. Palynology is a key tool for regional correlation of Boreal Triassic sediments as marine biostratigraphic events are rare in the absence of conodonts and ammonites in shallow marine deposition environments (Bjaerke & Dypvik 1977;Bjaerke & Manum 1977; van Veen 1985;Hochuli et al. 1989). However, independent detailed chrono-dating of these palynostratigraphic ranges is largely lacking owing to a scarcity of ash-beds. Hounslow et al. (2007aHounslow et al. ( ,b, 2008 published magnetostratigraphic data from various locations in Spitsbergen integrated with palynostratigraphic data.

Location and geological setting
This study focuses on the Kapp Toscana Group, but uses data from the late Ladinian part of the Botneheia Formation (Sassendalen Group) to the lowermost Knorringfjellet Formation. The interval covers a regressive cycle from offshore marine shelf to prodelta conditions, into overlying delta plain and coastal plain conditions, and back to shallow marine conditions. Sections at Juvdalskampen, Botneheia and Vendomdalen in central Spitsbergen have been studied (Fig. 1).

Geological setting
Deposition in the Svalbard archipelago (and the Barents Sea) during the Late Triassic took place on a stable platform that existed between the mid-Carboniferous and Mesozoic (Mørk et al. 1982;Harland 1997). During the Late Triassic delta systems discharged locally from the west, but mainly from the SE into the area of central Spitsbergen (Mørk et al. 1982;Dallmann 1999;Riis et al. 2008). The gradual progradation of sediment from the SE resulted in diachronous lithostratigraphic boundaries across the Barents Sea ( Fig. 2; Riis et al. 2008).
The overlying Kapp Toscana Group is subdivided into three formations (Mørk et al. 1982;Dallmann 1999). The lowermost Tschermakfjellet Formation in central Spitsbergen was deposited in an offshore marine to prodeltaic or delta front environment (Mørk et al. 1982). In the Isfjorden area it contains early Carnian (but not earliest Carnian) ammonoids (i.e. Stolleyites tenuis subzone of the S. tenuis zone), some 12 m above the base (Dagys et al. 1993). In the Vendomdalen area and eastern Spitsbergen the S. planus subzone of the Tenuis Zone is found some 0.5-3 m above the base of the Tschermakfjellet Formation (Dagys et al. 1993;Hounslow et al. 2007a). The overlying unit, the De Geerdalen Formation, contains no age-diagnostic macrofauna giving sufficient resolution for a Carnian subdivision (Hounslow et al. 2007a;Nagy et al. 2011). The lower part of the De Geerdalen Formation in central Spitsbergen was deposited in an interdistributary bay to delta front environment, and the upper part in a delta plain to coastal plain environment (Nagy et al. 2011). Mørk et al. (1982 described the De Geerdalen Formation as a dominantly fluvial environment deposited in a nearshore, paralic (lagoonal or delta plain) environment that was part of a NW-prograding delta system during the Late Triassic. Constant input of freshwater resulted in changing conditions from terrestrial or nearshore deltaic-dominated to shallow aquatic conditions with coastal reworking. In Svalbard, the earliest Carnian ammonoid zone with Daxatina canadensis is known only from Bjørnøya, about 300 km to the south (Tozer & Parker 1968). There it first occurs 40 m below the top of the 140 m thick Skuld Formation (Mørk et al. 1992;Dagys et al. 1993), in a unit consisting of grey shales, and siderite nodules (and some sandstone beds), similar to the lithology of the Spitsbergen Tschermakfjellet Formation. In central Spitsbergen, the Knorringfjellet Formation lies above the De Geerdalen Formation and forms the base of the Wilhelmøya Subgroup, which was deposited in a shallow marine environment (Dallmann 1999;Nagy & Berge 2008;Nagy et al. 2011). On Hopen Island the lower part of the Wilhelmøya Subgroup contains an early (not earliest) Norian ammonoid fauna (Bragin et al. 2012;Lord et al. 2014); hence part of the Wilhelmøya Subgroup on Spitsbergen may be Lower Norian.

Palynostratigraphy
The report by Van Veen (1985) for the Norwegian Petroleum Directorate (NPD) aimed to provide a stable stratigraphic framework for the Triassic of the Barents Sea, focusing on the Troms area, and subdivided the Triassic into 16 palynozones. Hochuli et al. (1989) subdivided the Upper Triassic into six palynological assemblage zones based on a mix of outcrop samples, core data and cutting samples. Vigran et al. (2014) expanded and simplified this previous work and published palynomorph assemblage zones for Svalbard and the Barents Sea based on outcrop data, cores and wells. Paterson & Mangerud (2015) conducted a detailed palynostratigraphic study of the Carnian to Rhaetian succession on Hopen.
Whereas independent age control from ammonoids is fairly good for the Early and Middle Triassic of Svalbard, the miosporebased age assignment of the Late Triassic is mostly based on a comparison with palynomorph ranges from the Germanic and Alpine realms, owing to a rarity of other precise age diagnostic forms. However, the age ranges of some miospore taxa are different in the Boreal realm, and hence dating using long-distance relationships is of somewhat limited value.

Palynology
A total of 60 samples (31 from Juvdalskampen and 29 from Botneheia) were processed for palynological analysis. To remove the carbonates and silica, 10 g of each sample was treated with HCl (27%) and HF (40%) according to standard processing procedures described by Kuerschner et al. (2007). The residue was sieved with a 250 and a 15 µm mesh. Heavy liquid separation with ZnCl 2 was carried out to remove heavy minerals and a few samples were subsequently treated with 'Schulze Reagent' (KClO 3 and HNO 3 ) or nitric acid to eliminate some organic material such as amorphous organic matter (AOM) and increase the palynomorph density. The slides are stored in the Department of Geosciences, University of Oslo, Norway. Palynomorph preservation is better for the Botneheia section than the Juvdalskampen section. The palynomorphs have colours in the range of 3-5 on the thermal alteration scale (TAS) of Batten (2002). Palynomorph identification was mainly based on the works of Schulz (1967), Morbey (1975), Bujak & Fisher (1976) and Bjaerke & Manum (1977). Photographs of selected palynomorphs are shown in Figure 3. About 300 terrestrial palynomorphs were counted (quantitative analysis) per sample using a Leitz Diaplan microscope.
Photographs were taken with an AxioCam ERc 5s camera connected to a computer using Zen 2011 software. Sample processing for the Juvdalskampen section was carried out by APT (Applied Petroleum Technology AS, Kjeller, Norway). Relative abundances were calculated and plotted using the Tilia/TiliaGraph and TGView software (Grimm 1991(Grimm -2001. Using CONISS (Grimm 1987) palynomorph assemblage zones were established by constrained cluster analysis within Tilia (Figs 4 and 5). Two complete slides per sample were completely scanned for further taxa (qualitative analysis) to check for the presence of rare, biostratigraphically important taxa.
Principal component analysis (PCA) was calculated with the program PAST (Hammer et al. 2001) based on the quantitative terrestrial palynomorph abundance data (Fig. 6).

Organic carbon isotope analysis
Carbon isotope values (δ 13 C org ) from bulk sedimentary organic matter (δ 13 C org ) for the Juvdalskampen and Botneheia sections were published by Mueller et al. (2014). In addition, 74 samples from the Vendomdalen section were analysed for δ 13 C org . One gram of sediment was crushed and treated with 1M hydrochloric acid and left for 24 h to remove all inorganic carbon. The samples were then neutralized with water and oven dried at 60°C. The homogenized samples were analysed with an elemental analyser-isotope ratio mass spectrometer (EA-IRMS; Europa Scientific 20-20 IRMS). Isotope ratios are reported in standard delta notation relative to Vienna PDB. The analytical precision based on routine analysis of internal laboratory reference materials indicates a standard deviation of <0.08‰. IA-R001, wheat flour was used as reference material (δ 13 C V-PDB = -26.43‰). The standard deviation of the standard was 0.05. The measurements were carried out by Iso Analytical Ltd (Crewe, UK).

Terrestrial versus aquatic palynomorphs
The Juvdalskampen and Botneheia sections are generally dominated by terrestrial palynomorphs with an increase in aquatic palynomorphs at the base and tops of the sections (Fig. 7). The aquatic fraction of the upper part of the Botneheia Formation and lower interval part of the Tschermakfjellet Formation at the Botneheia section is dominated by up to 95% freshwater algae Botryococcus. Above the freshwater algae dominated interval acritarchs of Micrhystridium spp. dominate the lower Tschermakfjellet Formation, with occurrences of up to 15% at Juvdalskampen (between 0 and 8 m) and at Botneheia (below 265 m). Above this, both sections are dominated (about 95%) by terrestrial palynomorphs. In both sections the top of the Tschermakfjellet Formation shows a second increase in aquatic palynomorphs, especially acritarchs of Micrhystridium spp. (40-60 m at Juvdalskampen and 285-230 m at Botneheia). The De Geerdalen Formation is dominated by terrestrial palynomorphs. Above 180 m in the De Geerdalen Formation at the Juvdalskampen section (to the top of the formation), freshwater palynomorphs increase in abundance to 50%. The interval above this maximum consists mainly of Botryococcus sp. This maximum coincides with a generally lean palynomorph interval. In the uppermost samples from the Knorringfjellet Formation at Juvdalskampen above 330 m, aquatic palynomorphs make up about 15% and contain mostly the acritarch Veryhachium and dinoflagellate cyst Heibergella asymmetrica. Overall, the amount of aquatic palynomorphs is slightly  higher in the Botneheia section (up to 10%) than at the Juvdalskampen section.

Terrestrial palynology of the Juvdalskampen section
The pollen to spore ratio for the whole section is low with less than 30% pollen for most of the section. At the base and at the top of the Tschermakfjellet Formation, two pollen maxima are recorded with up to 80% (maxima 1: 0-10 m above base and below 260 m at Botneheia; maxima 2: between 40 and 60 m at Juvdalskampen and between 290 and 300 m at Botneheia). At a stratigraphically higher level in the De Geerdalen Formation at 200 m there is another maximum of 40% pollen. Pollen abundance remains around 30% for the remainder of the Juvdalskampen section. The top of the De Geerdalen Formation is almost barren of palynomorphs. Based on cluster analysis, five assemblages were distinguished ( Fig. 4), as follows.
The Striatoabietites-Protodiploxipinus assemblage (J1) is characterized by a low diversity of pollen and spores and is dominated by striate bisaccate (S. balmei) and non-striate bisaccate pollen (Alisporites, Brachysaccus, Protodiploxypinus). Other abundant pollen taxa are Lunatisporites and Parvisaccites. Aulisporites astigmosus is also common. Spore abundance increases from base to the top from 10 to 50%. Abundant spore taxa are Concavisporites and Deltoidispora. Towards the top of this assemblage Punctatisporites becomes common. Above 5 m, the abundance of C. nathorstii increases to 4% and the abundance of Lunatisporites increases towards the top of the assemblage.
The Aulisporites-Camarozonosporites assemblage (J2) is characterized by a pollen to spore ratio of about 25%. The main characteristic for this zone is an increase in pollen A. astigmosus. Common spore taxa are Concavisporites, Deltoidispora and K. cooksonae and main pollen taxa are Parvisaccites and Protodiploxipinus. Taxa Calamospora, Duplexisporites aduncus, Semiretisporites and Baculatisporites are also abundant. Pollen Brachysaccus is abundant in the lower part of the assemblage zone.
Aulisporites astigmosus is common in the lower part and its abundance decreases to the top of this assemblage. Spore abundance varies between 20 and 50%. Abundant forms are Calamospora and Deltoidispora. At 55 m there is an increase in Convolutispora and Concavisporites. The abundance of Leptolepidites decreases to the top of this assemblage.
The Concavisporites-Semiretisporites assemblage (J4) is characterized by a steady decrease in pollen, except for an increase in abundance at 85 m height in the section. Abundant palynomorphs are D. problematicus and Deltoidispora. Abundance of Calamospora and Carnisporites decreases throughout the assemblage zone. At 125 m there is an increase of K. cooksonae to 20% as well as an increase in Convolutispora, Aratrisporites and Leptolepidites. Pollen are less abundant, with main taxa Brachysaccus and Chasmatosporites. Triadispora has its last occurrence within this assemblage zone.
The Leschikisporites-Kyrtomisporis assemblage (J5) is dominated by spore taxa. The base at around 145 m shows an acme of Leschikisporites aduncus and for the upper part of the interval above 320 m, taxa Concavisporites, Deltoidispora and Kyrtomisporites are abundant. The abundance of L. aduncus is decreasing throughout the assemblage zone. Chasmatosporites is present in moderate amounts, as well as Conbaculatisporites. The interval between 220 and 240 m is almost barren in palynomorphs.

Terrestrial palynology of the Botneheia section
At the Botneheia location the top of the Botneheia Formation was sampled with four samples, 20 samples were processed from the Knorringfjellet Formation and five from the lower third of the De Geerdalen Formation. The pollen to spore ratios from the Juvdalskampen and Botneheia sections are similar. The Botneheia Formation is almost barren in terrestrial palynomorphs. Four assemblage zones were recognized at this location ( Fig. 7), as follows.
The Striatoabietites-Alisporites assemblage (B1) is characterized by bisaccate pollen; in particular, the interval below 260 m contains abundant S. balmei. The terrestrial total count for this assemblage is low with almost barren samples having a poor preservation. Common spores are C. nathorstii and Deltoidispora sp. The assemblage corresponds to the Striatoabietites-Protodiploxipinus assemblage from the Juvdalskampen section based on the dominance of bisaccate pollen from both assemblages.
The Aulisporites-Concavisporites assemblage (B2) is dominated by various spore taxa. The spore abundance increases within this assemblage from 55 to 75%. Most common spore taxa are Deltoidispora and Concavisporites. Pollen are dominated by bisaccate taxa of Protodiploxipinus spp. A. astigmosus has a maximum abundance within this assemblage at 268 m. Taxa Lunatisporites sp., Striatoabietites balmei and Triadispora sp. occur in moderate amounts. The distinct increase in pollen A. astigmosus and decrease in bisaccate pollen correlate with the Aulisporites-Camarozonosporites assemblage from the Juvdalskampen section.
The Deltoidispora-Striatoabietites assemblage (B3) is mainly dominated by spores (up to 80%). The most common one is Deltoidispora sp. Fewer occurrences of spores Baculatisporites sp., Concavisporites spp. and D. problematicus are seen. Towards the top of this assemblage the abundance of K. cooksonae and Aratrisporites sp. increases. Bisaccate pollen dominates the top of the assemblages between 290 and 295 m with species S. balmei. Other taxa seen are Protodiploxipinus spp., Triadispora sp. and A. astigmosus. The increase in various pollen taxa and decrease in several spore taxa allows correlation with the Striatoabietites-Triadispora assemblage from the Juvdalskampen section.
A Concavisporites-Leschikisporites assemblage (B4) occurs above 300 m. The lower three samples from this assemblage contain a terrestrial assemblage consisting mainly of spores (e.g. D. problematicus, Camarozonosporites rudis, Concavisporites sp.  6. PCA plot of the pollen and spore percentage data. The plot shows a gradient from taxa we interpret as relatively cooler to warmer taxa along axis 1, whereas axis 2 represents a drier to wetter vegetation gradient. and Deltoidispora spp.). Additionally, the interval between 305 and 310 m contains abundant K. cooksonae and at 305 m pollen Triadispora sp. Although the interval above 375 m yields more pollen (e.g. Lunatisporites sp., Protodiploxipinus sp.), it is still dominated by spores (e.g. Concavisporites sp., Deltoidispora sp., L. aduncus). This section corresponds to the Concavisporites-Semiretisporites assemblage of the Juvdalskampen section because of the occurrence of several spores and pollen in both sections.

Aquatic palynology
The sampled interval of the Botneheia Formation contains dominant freshwater algae Botryococcus and occasional Tasmanites sp. The base and top of the Tschermakfjellet Formation are characterized by an increase in aquatic palynomorphs (Fig. 7). It also contains moderate amounts of acritarchs Micrhystridium and Veryhachium as well as occasional Pediastrum and Pterospermella and rare foraminifera test linings. In the De Geerdalen Formation, acritarchs Micrhystridium and Veryhachium occur in the section at Botneheia and the freshwater algae Botryococcus is abundant in the Juvdalskampen section with up to 40% occurrence (interval between 200 and 240 m). Towards the Knorringfjellet Formation a shift to more marine palynomorphs such as dinoflagellate cyst H. asymmetrica and moderate amounts of acritarch Veryhachium takes place. Overall the Botneheia section contains a higher total count of aquatic palynomorphs compared with the Juvdalskampen section.

Principal component analysis
The terrestrial palynomorph counts are shown as species scores on the first and second axes of a PCA ordination diagram (Fig. 6). The two axes are the dimensions through the dataset that explain the largest variance in species composition, interpreted to result from climatic or environmental factors that control vegetation type. Fig. 7. Stratigraphy, lithology, sample positions, δ 13 C org , PCA axes, distribution of the Sporomorph Ecogroups (SEGs), ratio of hygrophytic to xerophytic floral elements, terrestrial to marine palynomorph ratio, pollen to spore ratio, aquatic palynomorph and total counts of aquatic palynomorphs for the (a) Juvdalskampen and (b) Botneheia section.
PCA axis-1 explains 40% of the total variance of the data and PCA axis-2 explains 28%. Forms such as S. balmei, Cycadopites spp., A. astigmosus, Protodiploxipinus spp. and Lunatisporites spp. have a positive score on the first axis and are characteristic of warm temperatures. Therefore, the first axis can be interpreted as the ratio between taxa indicating relatively warm versus cold conditions. Taxa such as A. astigmosus, Cycadopites spp. and Kraeuselisporites cooksonae have a high positive value on the second axis (wetter) and S. balmei has a high negative value on the second axis (drier). The second axis is interpreted to represent taxa indicative of relatively drier versus wetter conditions.

Organic carbon isotope stratigraphy
At Vendomdalen, samples from two sections located c. 5 km apart were sampled. The bulk organic carbon isotope values (δ 13 C org ) for the older, east Milne Edwardsfjellet (MEE) section shows a trend from -30‰ at 0 m to -28‰ at 20 m (Figs 8 and 9). For the younger Dalsnuten (DA) section, the δ 13 C org has a larger variability and ranges between -24 and -30‰. The δ 13 C org values have two negative excursions in the Tschermakfjellet Formation whereas the De Geerdalen Formation contains four possible negative excursions to -30‰. The spore colour is in the range 3-5 at Juvdalskampen and Botneheia, indicating that modification of δ 13 C org owing to heating from Cretaceous intrusions (e.g. on Edgeøya; see Brekke et al. 2014) is unlikely. No dolerite intrusions occur in the Triassic in Vendomdalen.

Juvdalskampen-Botneheia section correlation
Correlation between the Juvdalskampen and Botneheia sections is based on their palynomorph assemblages, lithostratigraphy and δ 13 C org stratigraphy (Fig. 7). Several bio-events can also be matched in the two sections. The correlation between the Botneheia and Juvdalskampen sections was quantified using sequence slotting (Clark 1985;Thompson & Clark 1989;Thompson et al. 2012), as implemented in the CPLSLOT Windows program (http:// www.geography.lancs.ac.uk/cemp/resources/software/cplslot. htm); an objective method of quantitative correlation that can deal with multivariate datasets, including palynomorph counts. We used the palynology data to produce two likely correlation models, and independently the bulk δ 13 C org data to produce a third correlation model. The final height composite of δ 13 C org used the section heights at Juvdalskampen and correlated the heights at the Botneheia section to these, using an average of these three correlation models (Fig. 10). The negative carbon excursion from height 60 m in the Juvdalskampen section was assumed to be the same feature as seen at 296 m at Botneheia (Fig. 8). This provided the single correlation constraint for the sequence slotting. Rock-Eval pyrolysis (Mueller et al. 2014) does not reveal significant changes in the organic matter composition across these δ 13 C org fluctuations in the upper part of the Tschermakfjellet Formation, which we interpret to indicate that there is little organic matter compositional control on the δ 13 C org changes.
The success of the sequence slotting was evaluated by the Δ parameter, which takes values from greater than zero to about unity (Δ < c. 0.5 indicates very good similarity; Thompson & Clark 1989), and the random variable (RV) and modified RV (here called RV2) coefficients, which are somewhat similar to conventional regression coefficients (Smilde et al. 2009). The good degree of similarity in the correlation models ( Fig. 10) using δ 13 C org or the palynological data independently indicates the robust nature of these correlations through the Tschermakfjellet Formation. For the palynological data the Euclidean distance metric (an equal weight metric; Gavin et al. 2003) produces the smaller Δ, whereas the squared chord distance metric (a 'signal-to-noise' metric of Overpeck et al. 1985;Gavin et al. 2003) produces larger (i.e. better) association (RV, RV2) coefficients (Figs 9 and 10). Hence, these indicate better short-range similarity for the Euclidean model, and slightly better longer-range similarity for the squared chord model. The correlation relationships are much poorer (larger sensitivity statistics; Fig. 10) through the lower parts of the De Geerdalen Formation, which may be a reflection of the sparse data in both sections through this interval.

Palynological correlation between sections
It is feasible to compare results from the Juvdalskampen-Botneheia sections with the semi-quantitative palynomorph counts at Vendomdalen (MEE and DA sections) of Hounslow et al. (2007a). The palynomorph assemblages from the MEE section and from the Striatoabietites-Alisporites assemblage from Botneheia are characterized by an acme of various bisaccate pollen; mainly S. balmei and aquatic palynomorphs such as Tasmanites sp. and Micrhystridium. The base of the Tschermakfjellet Formation at Juvdalskampen is also characterized by an acme of various bisaccate pollen in the  Nagy et al. (2011) described potentially reworked foraminifera from the base of the Knorringfjellet Formation; this could also be the case for the palynomorphs at this level at Juvdalskampen. Comparison with Hounslow et al. (2007a) suggests that some of the samples from the upper Isfjorden Member may contain reworked taxa that would influence their palynological age assignment.

Correlation to the Vendomdalen sections
Correlation between the Juvdalskampen-Botneheia δ 13 C org composite and δ 13 C org at east Milne Edwardsfjellet was visually evaluated because of limited stratigraphic overlap. In the MEE section (which crosses the Botneheia-Tschermakfjellet Formation boundary), δ 13 C org is lower compared with the other sections, which is interpreted to indicate a slightly older part of the Botneheia Formation than that seen in the Botneheia section. We matched the δ 13 C org values at 261 m from Botneheia with the similar δ 13 C org values at 16 m from the MEE section, to produce the composite through the Ladinian-Carnian transition (Fig. 9). The Botneheia Formation-Tschermakfjellet Formation boundary in central Spitsbergen is typically characterized by a hiatus (or highly condensed interval; Hounslow et al. 2007a), which explains the rapid reduction in δ 13 C org values across this boundary. and c. 90 m, respectively. Attempting sequence slotting for matching to Dalsnuten was not feasible owing to widely varying spacing and sparse data at Juvdalskampen. The δ 13 C org height composite (Fig. 10) uses the relative heights of remaining samples at Dalsnuten, with respect to these correlation points. The correlation defined by the δ 13 C org stratigraphy indicates that the base of the De Geerdalen Formation is diachronous between these sections. This is not unexpected, as the definition of the base of the De Geerdalen Formation is defined by lithostratigraphy at the first significant sandstone bed. Diachronous Late Triassic lithostratigraphic boundaries have also been widely inferred in the Barents Sea successions (Riis et al. 2008;Høy & Lundschien 2011;Lundschien et al. 2014;Rød et al. 2014). Higher in the De Geerdalen Formation (above c. 390-480 m at Dalsnuten), channel sandstones predominate in the section, and probably complicate an already data-sparse interval, owing to complex lateral relationships. The variations could also reflect carbon storage on the De Geerdalen floodplain, or a climatic imprint on the δ 13 C org . Rock-Eval pyrolysis (Mueller et al. 2014) does not reveal significant changes in the organic matter composition across these δ 13 C org changes at Juvdalskampen.

Carnian carbon isotope stratigraphy
The carbon isotope stratigraphy through the Carnian is not completely known; the most complete stratigraphic coverage comes from the δ 13 C carb of Korte et al. (2005), with more limited, but detailed, stratigraphic coverage of δ 13 C org data from Dal Corso et al. (2012,2015). These datasets show increasingly more positive values from the late Ladinian into the Julian-1 (Figs 9b and 11), a feature that is clearly displayed in our data. A negative carbon isotopic excursion (CIE; -1) in δ 13 C org may mark the Ladinian-Carnian boundary, or perhaps more complex relationships in the δ 13 C carb data (Fig. 11). A broad positive CIE in δ 13 C org in the mid and upper parts of Julian-1 is matched with flat δ 13 C carb values, with some evidence of smaller positive excursions in δ 13 C carb (Fig. 9b). Our composite data through the Tschermakfjellet and lowest De Geerdalen Formation are inferred to match this broad positive CIE, in the age interval mid-to late Julian-1 (Fig. 9). This is compatible with the Fig. 9. Chronology inferred in Spitsbergen sections using the carbon isotope stratigraphy and magnetostratigraphy, correlated to Tethyan sections. (a) Composite organic carbon isotope stratigraphy through the Tschermakfjellet and De Geerdalen formations, using the isotope and palynological data. Composite heights are those at Juvdalskampen. Grey curve is a slightly smoothed line between the data points at their composite depths. The Dalsnuten section magnetostratigraphy is from Hounslow et al. (2007a). (b) Carbonate carbon isotopic data from Korte et al. (2005), with interpreted correlative intervals indicated with symbols and arrows. For all sections other than Silicka Brezova, the height scale is their relative sample order. The isotope stratigraphy and magnetostratigraphy at Silicka Brezova is plotted with respect to stratigraphic heights of Korte (1999), and biochronology of Korte et al. (2005). Silicka Brezova magnetostratigraphy from Channell et al. (2003). Section data sources are indicated: b, brachiopod data; wr, whole-rock δ 13 C carb . Correlative positive and negative Carbon Isotopic Excursions (CIE) numbered from the base (-1, +1), partly derived from the data in Figure  11. Shading of similar isotopic intervals matches that of Figure 11. inferred biochronology, derived from the magnetostratigraphy of Hounslow et al. (2007a). Our data also appear to display the positive CIE (+1) seen in the δ 13 C carb data, which is also shown in δ 13 C org in the Balatonfüred core (Fig. 11). The single sample spike in the δ 13 C carb data from the St Cassian section may be anomalous, but the low Mn and high Sr content of the calcite suggests that it is diagenetically unaltered (Korte et al. 2005). This isotopic excursion may be amplified by the coeval changes in palynological composition in the upper Tschermakfjellet Formation.
A strong negative CIE (-2) during the early parts of Julian-2 is seen in δ 13 C carb , δ 13 C org (Figs 8, 9b and 11) and marine algal biomarker material (Dal Corso et al. 2015). The CIE (-2) appears to be displayed in our data from the lower parts of the De Geerdalen Formation (Fig. 9a). The recovery of δ 13 C org from the CIE (-2) is less clear. The Milieres-Dibona section appears to show complete recovery to more positive values by the late Julian-2 (Fig. 11), but the δ 13 C carb shows continuation of the CIE until early Tuvalian-1 (Fig. 9b). This may reflect the disconnection between marine carbonate and atmospheric carbon sources (Dal Corso et al. 2015).
Following CIE (-2) the bulk δ 13 C org shows rather more complex changes (Fig. 11), which show only partial recovery, involving much variability, but involving initial recovery after CIE (-2) to a weak positive CIE (+2). This behaviour is tentatively related to what we see in the mid parts of the De Geerdalen Formation, but is incompletely defined by our sparse data through this interval (Figs  9a and 11). The channel sandstones at Dalsnuten in the lower part of this interval may have given rather complex interrelationships to the Juvalsdalen section, so it is not clear if the simple height matching we used is appropriate in this part of the sections. The Julian-1 to Julian-2 boundary previously inferred in the De Geerdalen Formation was based on application of conodont biostratigraphy of the magnetostratigraphically matched sections of Hounslow & Muttoni (2010), and was apparently placed too high, based on the isotope correlations proposed here. The reverse polarity chrons UT2r and UT3r accordingly may be present in the unsampled intervals in the Dalsnuten sections (Fig. 9a). However, the Tethyan sections of Dal Corso et al. (2015) also have considerable uncertainty in the placement of the Julian-1 to Julian-2 boundary (Fig. 11), which emphasizes the uncertainty in regional biochronological correlations, which also underpin the magnetic polarity to substage relationships in the time scale of Hounslow & Muttoni (2010). The Lunz section of Dal Corso et al. (2015) has probably the best defined (i.e. by conodonts and ammonoids) Julian-1 to Julian-2 boundary interval, and therefore this is chosen as the best compromise in our final age model.
Following Lord et al. (2014), primarily using the magnetostratigraphy, we relate the Isfjorden Member at Dalsnuten to the latest Tuvalian-2 to Tuvalian-3 interval. Using this appears to show a correspondence between the δ 13 C org changes at Dalsnuten and the δ 13 C carb changes of Korte et al. (2005) through this interval (Fig. 9), relating to the CIE numbered +3 and -3 in the Tuvalian. The δ 13 C carb data from Silická Brezová showing the CIE (-3) excursion (with δ 13 C carb < 3‰) are mostly from brachiopod data showing some diagenetic alteration (Korte et al. 2005), so there is less confidence about the reliability of this excursion. However, as Korte et al. (2005) noted, their apparently non-diagenetically altered data brachiopod data follow similar trends to diagenetically altered data through the Triassic, so the impact of the alteration may be small. Hence, we infer from these relationships that the magnetochron UT10 represents most of the Isfjorden Member at Dalsnuten, with the interval represented by UT6r to UT9n (late Julian-2 to late Tuvalian-1) missing at the disconformable base of the Isfjorden Member. The tentative δ 13 C org correlations proposed here allow for a refined age model for the Isfjorden Member-De Geerdalen Formation from that inferred by Lord et al. (2014).

Palynological-based age assignment in the Boreal Carnian
Several palynomorph range charts for the Boreal Triassic exist (e.g. van Veen 1985;Hochuli et al. 1989;Vigran et al. 2014), varying in their details through the Carnian.
The assemblages from the lower Tschermakfjellet Formation correlate with assemblage F of Hochuli et al. (1989). Key forms present are Angusticulcites klausii, Podosporites sp., Protodiploxipinus spp., S. balmei and Triadispora sp. These forms are typical for assemblages younger than Anisian according to Hochuli et al. (1989). However, Hochuli et al. (1989) based their age assignment on extrapolated ranges from the Tethys although they speculated that the age ranges in the Boreal region could be different, owing to the different latitudinal affinity having an impact on the true age ranges. The Tschermakfjellet Formation in Fig. 10. Sequence slotting correlations between the Juvdalskampen and Botneheia (a) Correlation models using the palynological data (euclidean and squared chord distance metric models) and the δ 13 C org data (Euclidean distance metric). A blocking length of 3 was used in all models, and the low in the isotope curve in the upper part of the Tschemakfjellet Formation was used as an approximate initial correlation constraint. Δ= slotting statistic of Clark (1985), RV/RV2= association coefficients (Smilde et al. 2009). Rp, Rs are conventional Pearson and Spearman correlation coefficients respectively for the isotope data. (b) Sensitivity statistic (reduction in combined path length, CPL), showing how sensitive the correlations are, if data from each level at Juvdalskampen are successively removed, and then the analysis re-run. Slotting most robust for palynological data below 60 m (Juvdalskampen) and less robust above. the MEE section contains both ammonoids and conodonts indicating the early Carnian, and when linked with the magnetostratigraphy in the Vendomdalen sections (Hounslow et al. 2007a) indicates a Julian-1 (I) age for the sampled base of the Tschermakfjellet Formation from the Botneheia section.
The assemblages from the Striatoabietites-Protodiploxipinus (Juvdalskampen) and Striatoabietites-Alisporites assemblages (Botneheia section) belong to zone VIII of van Veen (1985) and assemblage F of Hochuli et al. (1989) based on the previously mentioned forms and the first occurrence of Chasmatosporites (Botneheia at 264 m). Taeniate bisaccate pollen (e.g. Lunatisporites, Striatoabieites) are common in palynozone VIII and were used for correlation (Fig. 2). The younger Aulisporites-Camarozonosporites (Juvdalskampen) and Aulisporites-Concavisporites (Botneheia section) assemblages are characterized by an increase in A. astigmosus. Hounslow et al. (2007a) reported this form to be dominant at 280.7 m at Dalsnuten. It correlates well with assemblage VII of van Veen (1985) to which an early Carnian (Julian) age was assigned. In addition, various forms of Kyrtomisporites have their first occurrence in this assemblage.
The Striatoabietites-Triadispora assemblage from the Juvdalskampen section records an increase in Brachysaccus and has a resemblance to assemblage D of Hochuli et al. (1989). Assemblage E is distinguished on quantitative criteria which were not recognized in the Juvdalskampen section.
The Concavisporites-Semiretisporites assemblage corresponds to zone VI of van Veen (1985), characterized by the presence of Striatoabieites spp., Schizaeosporites worsleyi, Triadispora sp. and Aratrisporites spp. The lower part of the De Geerdalen Formation (lower part of the Concavisporites-Semiretisporites assemblage) with the last occurrence of K. cooksonae and an increase in Triadispora corresponds to the top of assemblage D of 'mid-Carnian' age.
The bottom of the Leschikisporites-Kyrtomisporis assemblage at Juvdalskampen is characterized by an acme of L. aduncus (145 m). This acme was not recognized at the Botneheia section. Hounslow et al. (2007a) described L. aduncus to be common in the higher De Geerdalen Formation and dominant in the Isfjorden Member. The findings are in accordance with Van Veen's zonation scheme of being typical for assemblage Va, to which he assigns a late Carnian (Tuvalian) age. A similar assemblage was described by Hochuli & Vigran (2010) in their floral phase 12, which is dominated by monolete spores based on a well from the southern Barents Sea. Higher in the section the abundance of L. aduncus decreases and the top of the De Geerdalen Formation is characterized by almost sporomorph barren samples (180-240 m Juvdalskampen). Van Veen (1985) described a similar sequence for assemblage Va. Lord et al. (2014) and Paterson & Mangerud (2015) also described a L. aduncus acme palynozone for the De Geerdalen Formation from Hopen. The last occurrence of species Thomsonisporites toralis at 180 m at the Juvdalskampen section suggests an age older than that of assemblage C of Hochuli and of assemblage IV of van Veen. The occurrence of S. balmei at 380 m at the Botneheia section, together with S. worsleyi at 200 m at the Juvdalskampen section and T. toralis at 180 m at Juvdalskampen, suggests a Carnian age for the top of the sampled interval from the De Geerdalen Formation. Vigran et al. (2014) introduced 15 composite assemblage zones for the Triassic of the Barents Sea and Spitsbergen. Most of the sections from this study correlate with their Aulisporites astigmosus Composite Assemblage Zone. This zone is, among others, characterized by an abundance of A. astigmosus and A. klausii. Also, L. aduncus is common to abundant for some intervals. From the Barents shelf this assemblage was recognized in the middle part of the Snadd Formation (Hochuli & Vigran 2010). Vigran et al. (2014) correlated this zone to the assemblages G to D of Hochuli et al. (1989) and assigned an early to mid-Carnian age. Vigran et al. (2014) also recognized this assemblage from other sections of the Svalbard Archipelago (e.g. Isfjorden in Spitsbergen and Blanknuten in Edgøya).
The Knorringfjellet Formation in the upper part of the Leschikisporites-Kyrtomisporis assemblage is characterized by an increase in Kyrtomisporites spp., presence of K. cooksonae and the first occurrence of Quadreaculina anellaeformis. This suggests an affiliation to zone IVb of van Veen and B2 of Hochuli of Norian age. Both dinoflagellate cysts (H. asymmetrica) and acritarchs (Veryhachium, Cymatiosphaera) were recognized in this interval. Nagy et al. (2011) described reworking for the lower Knorringfjellet Formation. Hounslow et al. (2007a) described the presence of typical sporomorphs in the Isfjorden Member of older assemblages according to Hochuli et al. (1989). This suggests a high potential for reworking in combination with hiatuses that makes palynostratigraphy difficult. The Knorringfjellet Formation from the Juvdalskampen section correlates with the Limbosporites lundbladii Composite Assemblage Zone of Vigran et al. (2014). The lower Rhaetogonyaulax spp. Composite Assemblage Zone below was not recognized, which is probably due to the hiatus between the top of the De Geerdalen Formation and base of the Knorringfjellet Formation at Juvdalskampen. The Limbosporites lundbladii Composite Assemblage Zone is characterized by spores and regular occurrences of dinoflagellates such as H. asymmetrica. Vigran et al. (2014) also assigned a Norian age to this assemblage.

Development of the depositional environment
The Kapp Toscana Group in Spitsbergen consists mainly of deltaic deposits forming an overall regressive setting (e.g. Mørk et al. 1982;Riis et al. 2008;Nagy et al. 2011). The top of the Botneheia Formation contains terrestrial palynomorphs and abundant aquatic palynomorphs such as Botryococcus and Tasmanites. This suggests deposition in a restricted shallow marine environment with a reduced salinity (Guy-Ohlson 1992;Krajewski 2008Krajewski , 2013. A palynofacies analysis of this section (Mueller et al. 2014) shows abundant AOM typical for a marine environment with a restricted circulation (Tyson 1993(Tyson , 1995. Marine palynomorphs include Micrhystridium and Veryhachium. The terrestrial palynomorph component records an increase in typical upland vegetation following the Sporomorph Ecogroup model (SEG; Abbink 1998; Abbink et al. 2004) in the Striatoabietites-Protodiploxipinus and Striatoabietites-Alisporites assemblage zones (Fig. 7). Because upland vegetation consists mostly of bisaccate pollen grains it tends to be transported easily by wind ('Neves effect' described by Chaloner & Muir 1968;Abbink et al. 2001). This leads to a selective enrichment of upland vegetation forms in a marine setting and explains the apparent contradiction. The marine influx gradually decreases higher in the succession, with an exception for the top of the Tschermakfjellet Formation where marine palynomorphs such as Micrhystridium and Veryhachium increase in abundance. At this level upland vegetation becomes more abundant again (Deltoidispora-Striatoabietites and Striatoabietites-Triadispora assemblage zones). This interval could represent a short marine ingression within the main regressive depositional trend as already suggested by Nagy et al. (2011) for their biofacies MB2 representing a prodelta setting. However, we note that mainly only one bisaccate pollen type, Striatoabietites, is affected in this interval whereas other bisaccates remain rather stable. If the abundance of bisaccate pollen were influenced by sea-level changes alone, we would expected a more even increase in all bisaccate pollen types rather than an increase in only Striatoabietes. Therefore, we suggest that although there is some taphonomical transport effect in the pollen record, the pattern in Striatoabieties abundance is related to climate changes as indicated by the PCA 1 curve (see also discussion below).
Marine palynomorphs in the De Geerdalen Formation above are rare at the base and absent for the mid and upper part. The palynomorph records show a mostly lowland or river SEG. This combined with sedimentological observations from sandstone bodies suggests a delta plain setting with river deposits. This setting is also the reason why a direct correlation between the channel sandstone units within the De Geerdalen Formation is challenging.
The youngest unit, the Knorringfjellet Formation, yields more marine palynomorphs of types Veryhachium and Cymatiosphaera and dinoflagellate cysts of H. asymmetrica (around 330 m at Juvdalskampen). The terrestrial component consists mostly of a lowland SEG but does also record a slight increase in upland vegetation. This suggests that the Knorringfjellet Formation is separated from the De Geerdalen Formation by a transgression, and forms the base of a new depositional sequence in a shallow marine setting.

Palaeoclimate and the Carnian Pluvial Event
The principal component analysis (Fig. 7) suggests a decrease in temperature (PCA 1 curve) in the lower part of the Tschermakfjellet Formation and an increase at the top, coinciding with the negative CIE preceding the positive CIE (+1). These temperature fluctuations have not been reported elsewhere in the Boreal region. However, the Tethyian Early Carnian oxygen isotope curve shows a transient rise by about 3‰ within the late aon and aonides ammonite Zone indicating a cooler period during the early Julian-1 (Cordevolian) followed by warming during the late Julian-1 (Korte et al. 2005). These oxygen isotope fluctuations suggest a rapid temperature decline and increase that could well correlate with those indicated in the PCA 1 curve of the present study. The PCA 1 curve is mainly influenced by the abundance of the bisaccate Striatoabietites. The temperature maxima correlate with peak abundance of Striatoabietes of about 50-60% above the normal background abundance in the miospore assemblages extracted from the Tschermakfjellet Formation. Potentially, bisaccate pollen abundance can be also biased by sea-level changes (Neves effect as discussed above). However, the pollen record shows mainly changes in the proportion of Striatoabietites whereas other bisaccate pollen types show only minor variations. Therefore, we suggest that most of the PCA 1 variation is related to temperature changes rather than short-term sea-level fluctuations.
Concerning changes in humidity as indicated in PCA 2 curve, the climate was relatively wetter during the deposition of the Botneheia Formation (Longobardian) and changed to drier conditions during the deposition of the Tschermakfjellet Formation (Julian-1). From the lowest part of the De Geerdalen Formation the climate became increasingly humid until the top of the formation. The interval from the late Ladinian to early Carnian could correspond to floral phases 10 and 11 of Hochuli & Vigran (2010). This interval is followed by a change at the base of the De Geerdalen Formation to dominant hygrophytic floral elements and an increase in monolete spores of type L. aduncus (floral phase 12 of Hochuli & Vigran 2010). The superabundance of L. aduncus, typical for swamp environments, along with the occurrence of coal beds and freshwater palynomorphs, such as Botryococcus, at the top of the De Geerdalen Formation, suggest a humid delta plain environment with rivers or lakes (Fig. 7).
The CPE in the Tethys has been described from the Julian-2-II A. austriacum ammonoid zone, there coincident with the A. astigmosus acme zone (e.g. Roghi et al. 2010;Dal Corso et al. 2012). However, in the Boreal region the A. astigmosus zone is located in the lower parts of the Tschermakfjellet Formation (Vigran et al. 2014; this paper), within the Tethyan T. aonoides ammonoid zone. Therefore, it appears that the A. astigmosus zone is older in the Boreal region (Fig. 8). It is likely that the Aulisporitesproducing mother-plant, which was a hygrophytic gymnosperm (Bennetitales, Kräusel & Saarschmidt 1966;Balme 1995) migrated southwards over the time interval of the Carnian Pluvial Event when palaeoenvironmental conditions became favourable for its proliferation in the Tethys realm. In Spitsbergen, the mid to upper interval of the De Geerdalen Formation is contemporaneous with the CPE in the Tethys (Julian-2; e.g. Roghi et al. 2010;Mueller et al. 2015). This interval of the De Geerdalen Formation is characterized by terrestrially dominated lowland or river vegetation with the L. aduncus acme and an increase in humid conditions.
There is evidence in the Tethys realm for a vegetation shift from xerophytic floral elements in the early Julian to hygrophytic elements in the late Julian, with a return to xerophytic elements in the early Tuvalian (e.g. Roghi 2004;Hornung et al. 2007a,b;Rigo et al. 2007;Kolar-Jurkovsek & Jurkovsek 2010;Breda et al. 2009;Kozur & Bachmann 2010;Roghi et al. 2010). This study shows that time-equivalent events are reflected in the lower De Geerdalen Formation on Spitsbergen, an interval that consists mainly of spore-bearing plants that indicate a humid climate and has a Julian-1 to Julian-2 transitional age (Fig. 8). The carbon isotope perturbations documented in our Boreal sections correlate to those seen on the Palaeothetys margins (Fig. 9).
A possible cause for the CPE climate shift could be the eruption of the Wrangellia oceanic plateau (Jones et al. 1977;Greene et al. 2010;Dal Corso et al. 2012 that started in the latest Ladinian (Xu et al. 2014). This eruption resulted in the release of large amounts of carbon dioxide into the atmosphere, which probably caused the global climatic change to more humid conditions, with later perturbations of the carbon cycle represented primarily by CIE (-2). The sedimentological expression of the CPE in the Barents Shelf sediments is not entirely clear, as the prograding deltaic systems, sourced from the Urals, seem to have been stable and progressive systems since the early Triassic (Riis et al. 2008;Glørstad-Clark et al. 2010). However, the more humid event represented by the CPE may be expressed in the overall increased sedimentation rates, increased channel-sandstone development and incision rates. To some extent this is what we see at Dalsnuten, with the development of major channel-sandstone bodies immediately after CIE (-2) event, in the earliest Julian-2. However, the synchronicity of such major sandstone bodies remains to be discovered in other areas of Svalbard and the Barents shelf. There may also be coeval changes in palaeosol style and development associated with the humid CPE. On Edgeøya, major units of listric fault development characterize the lower part of the De Geerdalen Formation (Osmundsen et al. 2014), which may also be a response to delta-front overloading by high sedimentation rates stimulated by the CPE.

Conclusions
Rarity of marine index fossils prevents a detailed biostratigraphy correlation of the Late Triassic sediments from central Spitsbergen with Tethyan biozonations (Fig. 2). However, this study successfully integrates palynology, δ 13 C org stratigraphy and magnetostratigraphy (Hounslow et al. 2007a) to produce a composite stratigraphy that provides a detailed understanding of the stratigraphic changes through the Kapp Toscana Group at Juvdalskampen, Botneheia and Vendomdalen (Fig. 8). The composite organic carbon isotope stratigraphy provides sufficient detail for a correlation to Carnian mid-latitude sections, in that it displays the major isotopic excursions seen in other mid-latitude sections, but in a lower Carnian interval with high sedimentation rates (Fig. 9). Palynology reveals nine assemblage zones (Figs 4 and 5), from two sections, that correlate with the existing stratigraphic schemes from van Veen (1985), Hochuli et al. (1989) and Vigran et al. (2014). The A. astigmosus acme zone from the Tethys of early Julian-2 age, which characterizes the onset of the CPE in the Tethys, has a Julian-1-I age in the Boreal realm (Fig. 8). The integrated stratigraphy reveals a diachronous age for the beginning of the De Geerdalen Formation, the base of which is around the upper boundary of the Tethyan ammonoid Aon biozone. The remainder of the De Geerdalen Formation (below the base of the Isfjorden Member) is located in the substage interval upper Julian-1 to mid-Julian-2. Organic carbon isotopic excursions in the Isfjorden Member confirm the latest Tuvalian-2 to Tuvalian-3 age of this member, previously inferred using magnetostratigraphy (Fig. 9). Marine palynomorph taxa in the Knorringfjellet Formation suggest a latest Carnian or Norian age for the top of the sections.
Application of the Sporomorph Ecogroup (SEG) model and the ratio of marine to aquatic palynomorphs suggest a restricted marine with freshwater influx setting for the base of the sections from Botneheia (Fig. 7). The Kapp Toscana Group is characterized by an overall regressive trend with marine conditions for the base of the Tschermakfjellet Formation and change from an upland SEG to a lowland and river SEG. An interval close to the top of the Tschermakfjellet Formation records a short return to more marine influx. The De Geerdalen Formation is terrestrially dominated with mostly lowland or river SEGs. Marine palynomorphs (dinoflagellate cysts and acritarchs) in the Knorringfjellet Formation above suggest a return to shallow marine conditions at the top of the section.
Multivariate statistical analysis reveals a relatively drier and warmer climate for the Botneheia Formation and lower Tschermakfjellet Formation of Julian-1 age (Figs 7-9). This is followed by a cooler phase and a short increase in temperature during the upper negative carbon isotope excursion in the Tschermakfjellet Formation. Humidity increased around the Julian-1-Julian-2 boundary in the De Geerdalen Formation. Botanical affinities show a general dominance of hygrophytic floral elements. In Spitsbergen a dominance of spores and occurrence of local coal beds suggest a humid swamp or marsh setting during the late Julian, which is the Boreal equivalent of the Carnian Pluvial Event known from the Tethys in the De Geerdalen Formation. A clearer sedimentological expression of the CPE in the Carnian sediments of Svalbard awaits further exploration.