Origin of Bentonites and Detrital Zircons of the Paleocene Basilika Formation, Svalbard

Introduction

The Paleogene was a time of profound change in the Arctic realm (Figure 1): caused by the contemporaneous spreading in the Baffin Bay/Labrador Sea and the North Atlantic, Greenland moved north and collided with Svalbard in the East and with Ellesmere Island (Canadian High Arctic) in the West (e.g., Talwani and Eldholm, 1977; Roest and Srivastava, 1989; Tessensohn and Piepjohn, 2000; Oakey and Chalmers, 2012). This led to the so-called Eurekan deformation that affected North Greenland, Ellesmere Island and Svalbard, and caused the formation of the West Spitsbergen Fold-and-Thrust Belt on Svalbard (e.g., Tessensohn and Piepjohn, 2000). Detailed timing of Eurekan deformation is still poorly constrained, but presumably commenced in the Paleocene, culminated in the Eocene, and may have lasted into the Oligocene (Tessensohn and Piepjohn, 2000; Piepjohn et al., 2013). Concurrent with Eurekan compression, extension prevailed further north, leading to the onset of spreading along the Gakkel Ridge, separation of the Lomonosov Ridge from the Barents Sea Shelf and the opening of the Eurasia Basin (e.g., Pitman and Talwani, 1972; Srivastava, 1985; Jackson and Gunnarsson, 1990; Tessensohn and Piepjohn, 2000; Døssing et al., 2013b). While the convergent movements were seemingly not associated with magmatic activity (cf. Tessensohn and Piepjohn, 2000), extensional tectonics were accompanied by magmatism with continental rift signature from the Early Cretaceous onwards until ~61 Ma (e.g., Tarduno, 1998; Estrada et al., 2001; Maher, 2001; Tegner et al., 2011; Thorarinsson et al., 2011b). During the Paleogene, magmatism in the Arctic was predominantly intrusive and basaltic effusive (e.g., Storey et al., 1998; Tegner et al., 1998; Storey et al., 2007; Thorarinsson et al., 2011b), but explosive volcanic activity also occurred, as evidenced by volcanic ash layers on Ellesmere Island and Svalbard (e.g., Dypvik and Nagy, 1978; Reinhardt et al., 2013). These ash layers are preserved in Paleogene sediments, i.e., in the Eureka Sound Group on Ellesmere Island and in the Central Tertiary Basin on Svalbard (Dypvik and Nagy, 1978; Reinhardt et al., 2013). Both depocenters are dominated by clastic deposition, and their stratigraphy and provenance are still poorly resolved (e.g., Miall, 1991; Dallmann, 1999). For the Central Tertiary Basin, a sediment source from the rising West Spitsbergen Fold-and-Thrust Belt to the west is assumed from ~mid-Eocene onwards (e.g., Helland-Hansen, 1990). Before that time, during the Paleocene and early Eocene, the basin received detritus from easterly directions, where a not further specified highland must have undergone active erosion (e.g., Helland-Hansen, 1990; Bruhn and Steel, 2003).

This study focuses on the Paleocene Basilika Formation of the Central Tertiary Basin on Svalbard. This formation comprises deltaic to deep-water clastic sediments (e.g., Steel et al., 1981), supposedly sourced from the east (Steel et al., 1981; Helland-Hansen, 1990; Müller and Spielhagen, 1990), and contains several altered volcanic ash layers that were previously characterized as bentonites (e.g., Gripp, 1927; Nagy, 1966; Dypvik and Nagy, 1978). The primary aim of this study is to geochemically characterize the Basilika bentonites in order to derive information on the origin of their volcanic sources and their plate tectonic environment. Furthermore, the ash layers will be compared to the Paleocene bentonites occurring on Ellesmere Island, to test whether both volcanic deposits may have derived from the same source (taking into account that Ellesmere Island and Svalbard were situated relatively close to each other during the Paleocene). For this, we performed main and trace element geochemical as well as Sm-Nd isotope analyses on the Basilika Formation bentonites as well as on the Ellesmere Island bentonites (complementing the data published by Reinhardt et al., 2013). No primary magmatic zircons were found in the bentonites, but we used U–Pb dating of detrital zircons contained in the bentonites to characterize the provenance of the Basilika Formation and to constrain the paleogeographic setting and sediment transport pathways during the Paleocene.

Geological Setting

Cretaceous-Paleogene Geology of Svalbard and Regional Tectono-Magmatic Framework

The Svalbard archipelago represents the uplifted north-western margin of the Barents shelf and is confined to the north by the Eurasia Basin of the Arctic Ocean and to the west by the Fram Strait (Figure 1). Svalbard has a complex tectono-magmatic history comprising multiple orogenies and magmatic events, which is to a great extent shared with that of the circum-arctic land masses, most prominently by Ellesmere Island and Greenland, but also with Baltica.

During the Cretaceous the northernmost regions of Ellesmere Island, Greenland and Svalbard were adjacent to each other at a paleolatitude of about 80°N (Lawver et al., 1990; Faleide et al., 1993; Harland, 1997). During the Early Cretaceous, fine-clastic sediments were deposited on Svalbard, followed by delta progradation from the northwest to the southeast with sediment transport possibly from Greenland (Røhr, 2009). Exhumation and erosion of northern Svalbard caused by the opening of the Amerasia Basin during the Cretaceous may have provided another source of detritus feeding the prograding delta system (Dörr et al., 2013). The Early Cretaceous sediments of Svalbard are truncated by an erosional unconformity and hence Late Cretaceous strata are completely missing (Dallmann, 1999).

Intrusive and extrusive magmatism on Svalbard in the Early Cretaceous was coeval with widespread volcanism throughout the Arctic and connected to the formation of the Cretaceous (to Paleogene) High Arctic Large Igneous Province (HALIP) (Tarduno, 1998; Maher, 2001). The HALIP is manifested in Cretaceous flood basalts, dykes and sills on eastern Svalbard and the adjacent subseafloor (Maher, 2001; Minakov et al., 2012; Corfu et al., 2013), Franz Josef Land, Ellesmere Island, and other islands in the Canadian Arctic (Embry and Osadetz, 1988; Estrada and Henjes-Kunst, 2004, 2013; Corfu et al., 2013; Jowitt et al., 2014), as well as Late Cretaceous to Paleocene dykes and extrusive magmatism on northern Greenland and the Canadian Arctic islands (Figure 1; e.g., Estrada et al., 2001; Kontak et al., 2001; Tegner et al., 2011; Thorarinsson et al., 2011a).

The HALIP comprises a tholeiitic suite (ca. 130–90 Ma) and an alkaline suite (ca. 90–60 Ma; Estrada and Henjes-Kunst, 2004; Buchan and Ernst, 2006; Tegner et al., 2011; Thorarinsson et al., 2011b; Estrada, 2015). The tholeiitic suite is associated with extensional tectonics in the High Arctic related to the breakup of the Arctic continental land masses and the subsequent opening of the Amerasia Basin (Tegner et al., 2011; Døssing et al., 2013b). The extrusion of the alkaline suite of the HALIP is typical for continental rifts and does only partly overlap geographically with the tholeiitic suite (Tegner et al., 2011). Alkaline mafic to felsic volcanism is known from northwest Ellesmere Island (Hansen Point Volcanic Complex at around 80 Ma; Embry and Osadetz, 1988; Estrada and Henjes-Kunst, 2004, 2013), the Kap Washington Group in North Greenland (71–61 Ma; Tegner et al., 2011; Thorarinsson et al., 2011a,b), from the Cape Back Basin on Ellesmere Island, evidenced by deposition of volcanogenic detritus (61–58 Ma; Estrada et al., 2010) and from Freeman Cove, Bathurst Island (56 Ma; Day et al., 2005; Figure 1).

In the High Arctic volcanism ceased during the Paleogene probably due to compressional tectonics induced by the northwards movement of Greenland (Tegner et al., 2011). Coevally with the termination of volcanism in the High Arctic, a Large Igneous Province formed in the North Atlantic region (NALIP). The NALIP is expressed in flood basalts on eastern Greenlandas well as flood basalts and alkaline tuffs in western central Greenland (Storey et al., 1998, 2007; Tegner et al., 1998; Larsen and Pedersen, 2009; Larsen et al., 2016). Moreover, Paleocene bentonites were observed on Svalbard as well as Ellesmere Island and are particularly widespread in the North Atlantic region (Figure 1; Major and Nagy, 1972; Dypvik and Nagy, 1978; Larsen et al., 2003a; Grist and Zentilli, 2004; Reinhardt et al., 2013).

Until ~55 Ma, Greenland was part of the Eurasian plate. The plate boundaries between the Eurasian and North American plates were located in the Labrador Sea/Baffin Bay and within the Eurasia Basin. Both plate boundaries were connected by a putative sinistral transform fault, the existence and exact positioning of this fault currently being under debate (e.g., Tessensohn and Piepjohn, 2000; Oakey and Chalmers, 2012; Greiner and Neugebauer, 2013; Frisch and Dawes, 2014; Neugebauer and Greiner, 2014). From 55 Ma onwards, Greenland was separated from both Eurasia and North America by coeval seafloor spreading in the Labrador Sea/Baffin Bay and in the northern North Atlantic Ocean (Pitman and Talwani, 1972; Roest and Srivastava, 1989; Tessensohn and Piepjohn, 2000), as well as in the Eurasia Basin (Tessensohn and Piepjohn, 2000). Seafloor spreading in the Labrador Sea/Baffin Bay ceased at ~33 Ma (Kristoffersen and Talwani, 1977), while the North Atlantic spreading system continued to propagate northwards, separating Greenland/North America from Eurasia/Svalbard from the Oligocene onwards (e.g., Talwani and Eldholm, 1977; Gaina et al., 2009, 2015; Seton et al., 2012; Døssing et al., 2013a, 2016).

The northwards movement of Greenland as an independent micro-plate between 56 and 33 Ma led to intracontinental compression along the margins of Eurasia and North America, which is expressed in the Eurekan deformation of Svalbard, northeast Greenland, and Ellesmere Island (Tessensohn and Piepjohn, 2000; CASE Team, 2001). These deformations were initiated during the Early Paleocene and culminated during the Eocene (Oakey and Chalmers, 2012; e.g., Piepjohn et al., 2013), forming the West Spitsbergen Fold-and-Thrust Belt (Bergh et al., 1997; Piepjohn et al., 2001; von Gosen and Piepjohn, 2001). The NNW-SSE striking Central Tertiary Basin (CTB, Figure 2) was formed as a foreland basin of the fold-and-thrust belt and filled with the eroded material of the bulge. Compression in the West Spitsbergen Fold-and-Thrust Belt ceased during the Late Eocene followed by a short extensional episode during the early Oligocene (Braathen et al., 1999; CASE Team, 2001).

Stratigraphy of the Central Tertiary Basin (CTB)

The main occurrence of Paleogene strata on Svalbard is in the CTB of Spitsbergen (Figure 2), the main island of the Svalbard archipelago (reviewed in Dallmann, 1999). The CTB comprises a Paleocene to Eocene clastic sedimentary sequence, the Van Mijenfjorden Group, with a preserved thickness of more than 1900 m, which lies unconformably upon Lower Cretaceous sandstones of the Carolinefjellet Formation (Figures 3A, 4; Major and Nagy, 1964; Livšic, 1967; Harland, 1969; Major and Nagy, 1972; Livšic, 1974; Dallmann, 1999).

The base of the Van Mijenfjorden Group is formed by the Paleocene Firkanten and Basilika formations. These represent an overall transgressive succession, evolving from continental to marginal marine sandstones of the Firkanten Formation to the marine shales of the Basilika Formation (Steel et al., 1981; Figure 4). The Basilika Formation consists of black mudstones, shales and siltstones with intercalated bentonite horizons that were deposited in a prodelta environment (Steel et al., 1985; Müller and Spielhagen, 1990). The lower part of the formation features a fining upwards sequence, which reverses into coarsening upwards in the upper part of the formation (Dallmann, 1999). This upper part of the formation marks the onset of a regression, leading to a gradual transition from marine shales into the marine glauconitic, highly bioturbated sandstones of the Grumantbyen Formation (Steel et al., 1981). The overlying strata comprise a transgression-regression cycle recorded in the Paleocene-Eocene Frysjaodden Formation and the Eocene Battfjellet and Aspelintoppen formations (Charles et al., 2011; Figure 4). Sedimentation in the CTB ceased during the Middle Oligocene (Harland, 1997). The thickness of the Paleogene formations varies greatly throughout the CTB. The Basilika Formation thins out from more than 350 m in the southwest to ~10 m in the northeast (Dallmann, 1999).

Bentonites are very common in the Basilika Formation. They were first described by Gripp (1927) as thin, laterally consistent beds of plastic clay and have been observed in several locations within the CTB (Nagy, 1966; Vonderbank, 1970; Major and Nagy, 1972; Dypvik and Nagy, 1978; Major et al., 2001). Dypvik and Nagy (1978) described the bentonites as very fine-grained but poorly sorted, with maximum grain sizes between 15 and 40 μm and consisting of ~80% clay minerals—mostly illite/smectite—with minor fractions of quartz and feldspar, indicative of altered, aeolian-transported volcanic material. Based on the grain size they concluded that the source area was located at a distance of 100−200 km from the area of deposition but more distant sources were not excluded. Except for the bentonites of the Basilika Formation, only few indications of Paleogene volcanism have been found on Svalbard. Major and Nagy (1972) reported that bentonites of 10−20 cm thickness had been observed adjacent to the coal seams of the Firkanten Formation. Further bentonite layers of Paleocene age are known from numerous coal exploration wells (Jones et al., 2015). Additionally, bentonites were observed in the Frysjaodden Formation, situated in the interval of the carbon isotope excursion of the Paleocene-Eocene Thermal Maximum, and thus provide constraints on the numerical age of the Paleocene-Eocene boundary on Spitsbergen (Charles et al., 2011).

The biostratigraphy of the Van Mijenfjorden Group is poorly resolved (reviewed in Manum and Throndsen, 1986; Čepek and Krutzsch, 2001). While earlier publications proposed just a Paleocene age for the Firkanten Formation (Manum, 1962), more recent studies place it in the Selandian to Thanetian (Middle to Upper Paleocene) based on the occurrence of a coccolithophore species (Čepek, 2001; Čepek and Krutzsch, 2001). Normapolles flora found in the Basilika Formation was assigned to a wide age range from uppermost Cretaceous to middle Paleocene (Čepek and Krutzsch, 2001). Manum and Throndsen (1986) suggested that the Basilika Formation was deposited in the Lower to Upper Paleocene transition based on dynocysts. However, a Danian (Early Paleocene) age for the lower part of the Basilika Formation is not very likely due to the Middle to Upper Paleocene deposition age of the underlying Firkanten Formation (Čepek, 2001; Čepek and Krutzsch, 2001). The Late Paleocene age of the overlying Grumantbyen Formation is again poorly constrained. The Grumantbyen Formation is overlain by the Frysjaodden Formation, which contains a bentonite layer dated as 55.9 ± 0.1 Ma (Paleocene/Eocene boundary after Gradstein et al., 2012) by U–Pb geochronology (Charles et al., 2011). Summarizing the biostratigraphic and radiometric age constraints as well as the lithostratigraphic correlations for the CTB formations and considering recent changes in the Geologic Time Scale, a latest Danian to early Thanetian time interval (ca. 62–58 Ma) for the deposition of the Basilika Formation can be estimated.

Sampling and Field Observations

The Bjørndalen section of the Basilika Formation consists of two outcrops along the river Bjørndalselva and contains a total of four bentonite layers (Table 1, Figures 3, 5). The sandstone cliffs of the Firkanten Formation are well above the river at the seaward valley entrance (Figure 3A). Further up the valley, the river cuts first into the Firkanten Formation and subsequently into the Basilika and Grumantbyen Formations, forming well exposed outcrops. The exposures of the Firkanten and Basilika Formations along the river were logged stratigraphically (Figure 5).

The lower, middle, and uppermost parts of the Firkanten Formation are scree-covered, so that just a section of 11.5 m of the upper part could be investigated. The formation comprises mostly fine-grained sandstones with intercalated siltstones and silty shales with an overall fining upwards trend. Mudclasts, clay-ironstone nodules as well as coal or plant fragments are frequent in the upper part. Bioturbation is common throughout the investigated section. The Basilika Formation is 73.5 m thick, dips 5°SE and is partly debris-covered. Therefore, the boundary to the underlying Firkanten Formation could not be observed and the outcrop of the Basilika Formation is split into several parts. The Basilika Formation consists of dark gray to black, silty shales, and siltstones with few intercalated sandstone horizons. Bioturbation is common in the lower part of the formation but absent in the middle and upper part. The formation comprises several coarsening upwards sequences from shale to siltstone. Dropstones, mostly in the range of 1–5 cm in diameter but rarely up to 20 cm can be observed throughout the formation. These consisted in all cases either of chert or quartz/quartzite. The boundary between the Basilika Formation and the overlying Grumantbyen Formation is well exposed and marked by the occurrence of a fine-grained glauconitic sandstone bed.

Two pairs of laterally continuous horizons of plastic, gray-weathered brownish clay were observed in the Basilika Formation and interpreted as bentonites. Horizons H1 and H2 are located in the lower part of the formation about 50 cm apart and are 5 and 3 cm thick, respectively. Horizons H3 and H4 are located in the middle part of the formation and are 2 m apart, with thicknesses of 3 and 5 cm, respectively. Horizon H3 was subdivided into a main horizon and sub-horizon H3b. The latter does not form a coherent layer but occurs sporadically several cm above the H3 main horizon.

The bentonite layers comprise plastic, gray-weathered brownish clay intercalated into shales of the lower Basilika Formation (Horizons H1 and H2) and siltstones of the upper Basilika Formation (H3 and H4; Table 1, Figure 5). The horizons were laterally traceable as continuous lines of vegetation on the scree cover. Two to three samples were taken from each horizon (about 1.5 kg per horizon) in 2010 for petrographic, geochemical, and geochronological analyses (see Table 1 for sample numbers).

For geochemical comparisons, additional bentonite samples from the Basilika Formation collected during the CASE 1 expedition of the German Federal Institute for Geosciences and Natural Resources (BGR) in 1992 (Tessensohn et al., 2001) and two bentonite samples from the lower Paleogene Mount Lawson Formation (Eureka Sound Group) of Ellesmere Island, Canada (Reinhardt et al., 2013) were included in this study. Whole-rock geochemical data of bentonites of the Basilika Formation sampled during the CASE 1 expedition in 1992 in the northern part of the CTB complements the dataset obtained in this study (Supplementary Material). The sampling sites of the CASE 1 bentonites are located on the Erdmannflya peninsula on the northern shore of Isfjorden, as well as in Fossildalen, Russekollen, and Battfjellet to the southwest of Longyearbyen (Table 1, Figure 2). Their stratigraphic positions relative to the bentonites recovered in Bjørndalen is unknown.

Analytical Methods

Compositional Analyses

Sample preparation and compositional analyses were conducted at the Federal Institute for Geosciences and Natural Resources (BGR) in Hannover, Germany. Identification of mineral phases and geochemical analyses were initially performed on homogenized whole-rock samples. Subsequently, clay fractions (<2 μm) were separated from the whole-rock samples using ultrasonication and analyzed separately in order to minimize the influence of detrital components. Identification of mineral phases in whole-rock samples and <2 μm fractions was achieved by infrared spectroscopy and X-ray diffraction (XRD; Cu radiation) measurements of air dried and ethylene glycol vapor-solvated samples. Cation-exchange capacity (CEC) was measured as a qualitative indicator for bentonite using Cu-trien (Dohrmann et al., 2012). Additionally, scanning electron microscopy was performed on samples H2 and H3 for identification of volcanic glass particles. The chemical composition of the whole-rock samples and <2 μm fractions was analyzed by X-ray fluorescence (XRF) spectrometry and inductively-coupled plasma mass spectrometry (ICP-MS). Sm–Nd isotope data were obtained on the acid-digested whole-rock samples and <2 μm fractions. U-Pb ages of density-separated, hand-picked zircons were determined using laser ablation ICP-MS (LA-ICP-MS). Concordia diagrams were plotted using the Isoplot program of Ludwig (1991). Probability density plots were created using a program written by Sircombe (2004). ²⁰⁶Pb/²³⁸U ages are plotted for zircons < 1000 Ma and ²⁰⁷Pb/²⁰⁶Pb ages for zircons >1000 Ma. A detailed description of the methods can be found in the Supplementary Material (Section Analytical Methods). Results are given in Tables 2, 3 and the Supplementary Material. All data are stored in the Pangaea database and are accessible via http://doi.pangaea.de/10.1594/PANGAEA.860807.

Results and Interpretation

Mineralogical Composition of the Basilika Formation Bentonites

Multiple lines of evidence indicate that the plastic clay horizons in the Basilika Formation are indeed bentonites. XRD analyses on the whole-rock powder samples revealed a composition dominated by quartz, feldspar, and smectite-rich interstratified (mixed-layered) minerals for all samples. While all samples contain traces of kaolinite and muscovite-illite, only the lower horizons H1 and H2 additionally comprise traces of anatase (TiO₂) and barite (observed during heavy mineral separation). High loss on ignition (6.5–10.0 wt%; see Supplementary Material) in all samples and a lack of relict volcanic glass indicate strong alteration of the primary volcanic ashes. The <2 μm fractions were analyzed separately to obtain a more comprehensive characterization of the clay minerals. The clay fractions showed CEC values of 27–35 meq/100 g (H1 and H2) and 24–32 meq/100 g (H3 and H4), consistent with a high abundance of clay minerals, but lower than the CECs typical for smectite-rich bentonites (Rollins and Pool, 1968; Reinhardt et al., 2013). XRD analysis of air-dried and ethylene glycol solvated samples revealed that the <2 μm fractions were dominated by illite-smectite type expandable clay minerals, with R0 illite (0.45)-smectite in samples H1 and H2 and R1 illite (0.6)-smectite in samples H3 and H4 (Figure 6). This is consistent with previous observations (Dypvik and Nagy, 1978). Furthermore, quartz, kaolinite, and chlorite (Figure 6) were present in all <2 μm fractions, while anatase was abundant in samples H1 and H2 (Figure 6). The dominance of illite-smectite in contrast to smectite in other bentonites might be indicative of diagenetic transformation of smectite to illite (Šucha et al., 1993), consistent with the burial history of the CTB (Blythe and Kleinspehn, 1998) and is therefore a further indication of the strong diagenetic alteration of the bentonites.

Geochemical and Isotopic Characterization of the Basilika Formation Bentonites

To largely exclude detrital contaminants, major and trace element geochemical analyses and Sm-Nd isotope ratio measurements were performed using the <2 μm fraction assuming that the clay minerals represent the alteration products of the original volcanic glass.

The bentonite samples exhibited similar SiO₂ concentrations (45–47 wt%; Table 2). They can be divided into two groups, a stratigraphically older suite comprising H1 and H2 with elevated TiO₂ (~5 wt%) and lower Al₂O₃ (~20 wt%) abundances and a younger group comprising H3 to H4 with lower TiO₂ (~1 wt%) and higher Al₂O₃ (~23 wt%) contents. Furthermore, the K₂O content is slightly higher in H3 and H4 (~3.5 wt%) compared to H1 and H2 (~2.5 wt%). Additionally, V, Cu, and Sc are higher in H1 and H2, whereas Nb, Y, Th, Zr, and the rare earth elements (REE; except Eu) are higher in H3 and H4 (Supplementary Material).

The geochemical fingerprint of the analyzed bentonites was in part inherited by the parental volcanic ashes but likely also influenced by post-depositional alteration. Alteration may have occurred at several stages. Firstly, alteration by seawater via ion exchange after deposition of the ashes in the pelagic zone of the marine basin, which occupied the CTB during the time of deposition. Subsequently, the volcanic glass and the less stable minerals were transformed to clay and other minerals. Finally, the volcanic ashes may have been altered by diagenesis at a burial depth of up to 2 km, which was followed by Neogene uplift and exhumation as well as recent weathering (Blythe and Kleinspehn, 1998; Dörr et al., 2013; Schlegel et al., 2013). During all of these phases, ions may have been exchanged with the surrounding rocks and water, thus altering the original geochemical fingerprint. Specifically, seawater-mediated alteration of felsic and intermediate volcanic rocks to bentonites may result in leaching of Na, K, Sr, Rb, as well as uptake of Mg and Fe²⁺, while Al, Ti, Zr, Nb, V, and Ni are residually enriched and the behavior of Si and Ca depends on the chemistry of the parent rock (Christidis, 1998). Similarly, seafloor alteration of basaltic ashes may result in enrichment in Si and Al as well as depletion in Fe, Mg, and Ca (Larsen et al., 2003a). Furthermore, the diagenetic transformation of smectite to illite should theoretically yield enrichment in K and Al and decrease in Mg and Na content, based on the stoichiometry of the montmorillonite and illite end members. Therefore, conventional discrimination diagrams for the characterization of the volcanic source rocks may not be applicable to the classification of bentonites. Still, the strong systematic differences in TiO₂ and Al₂O₃ content among the Basilika Formation bentonites likely originate from differences in the chemical composition of the volcanic ash precursors and may therefore be characteristic for the magmatic source.

The behavior of the REE and Y during alteration is controlled by the presence of phosphate minerals (Christidis, 1998; Reinhardt et al., 2013). There is no indication for the presence of such minerals in the Basilika bentonite samples. Furthermore, the concentrations of REE and Y are generally higher in the <2 μm fraction than in the whole-rock fraction (see Supplementary Material) indicating that these elements are hosted by the clay minerals (as most other trace elements, except Ba and Sr). Thus, we infer that the REE and some other relatively immobile trace elements (e.g., Zr, Ti, Y, Nb) as well as their ratios can provide a geochemical fingerprint of the parent volcanic ash. The chondrite-normalized REE patterns of the bentonite samples generally show enrichment of light REE (LREE) vs. heavy REE (HREE) with H1 and H2 (La_n/Lu_n = 11−12) being less enriched than H3 and H4 (La_n/Lu_n = 15−19) (Figure 7). In the range of the HREE, the patterns are subparallel to each other (Tb_n/Lu_n = 1.9−2.2 for H1/H2 and 2.1−2.3 for H3/H4), indicating magma generation under similar conditions with the H3/H4 magma being more evolved compared to H1/H2. This is supported by a negative Eu anomaly increasing from H1 (Eu/Eu* = 0.87) to H4 (Eu/Eu* = 0.59) indicating fractional crystallization of feldspar (Philpotts and Schnetzler, 1968). An alkali-basaltic original ash composition is suggested for the H1 and H2 bentonites, while a trachytic to alkali-rhyolitic original ash composition is suggested for the H3, H3b, and H4 bentonites, assuming that the Zr/Ti and Nb/Y ratios are only slightly affected by alteration processes (Figure 8). Thus, the younger bentonites on Svalbard (H3–H4) originated from a more evolved magma than the older bentonites (H1–H2).

The high Zr concentrations of up to ~2400 ppm in the <2 μm fractions potentially represent a combination of inheritance from the precursor rock and detrital contamination. Since no major Zr-bearing mineral phase could be identified, the Zr was likely associated with the glass matrix of the precursor rock resulting from rapid cooling before any zircon could crystallize. Likewise, high Zr concentrations have been observed in melt inclusions in Paleogene alkaline tuffs from Greenland (Heister et al., 2001). In contrast, contamination by detrital zircons during deposition or diagenesis could also lead to elevated Zr concentrations of the bentonites. However, the interlayered sandstones of the Basilika Formation have up to tenfold lower Zr concentrations (Schlegel et al., 2013), which are well within the range commonly observed for sedimentary rocks and upper continental crust (Taylor and McLennan, 1985). A detrital origin of the high Zr concentrations in the <2 μm fractions would thus require extraordinarily strong, selective enrichment of zircon in this size class, which is not commonly observed during diagenesis of sedimentary and magmatic rocks (González López et al., 2005; Taboada et al., 2006; Cavalcante et al., 2014). Contrasting evidence for contamination by detrital zircons comes from the distinct Zr/Nb ratios of the bentonites. The high Nb contents of H1-H4 (75–142) suggest that the bentonites originate from alkaline precursor rocks. Indeed, the Zr/Nb ratios of around 7 for H1 and H2 are well within the range suggested for alkaline rocks (<10; Larsen et al., 2003a), while those of H3 and H4 (~12−21) are far higher. Assuming that Zr and Nb are mobile to a similar degree during diagenesis, this would suggest that contamination by detrital zircons is negligible for H1 and H2, while up to half or more of the Zr content in H3 and H4 could be derived from detrital material. Accounting for this detrital Zr would yield lower Zr/TiO₂ ratios for H3/H4 and would thus suggest a lower degree of magmatic differentiation. However, the Zr/TiO₂ vs. Nb/Y discrimination diagram would still indicate that H3/H4 originate from a more evolved magma than H1/H2 even when accounting for a large degree of contamination by detrital Zr of 50% or more.

The different enrichment of LREE vs. HREE in H1/H2 and H3/H4 likely reflects different degrees of partial melting of the mantle source. A similar mantle source is supported by the Sm–Nd isotope data (Table 3). The εNd(t) values of H1/H2 (1.2–1.3) and H3/H4 (0.3–1.6) overlap and the Nd model ages (t_DM) vary in a narrow range for all horizons (0.55–0.67 Ga). This mantle source is very similar to those of alkaline basalts from ocean island and continental rifting settings (Wilson, 1989), pointing to either a fertile mantle source as in hot-spot settings, melting in the subcontinental lithospheric mantle and/or the contamination of the magma by continental crust. Overall, the REE data of the H1–H4 bentonites suggest that all analyzed ash layers were sourced from the same tectono-magmatic setting and possibly also from the same volcanic center within a continental rift zone, exhibiting minor differences in the parental magma composition and different degrees of differentiation.

Based on the Zr/Ti-Nb/Y discrimination diagram, the CASE 1 bentonites originate from an alkaline magmatic suite (Figure 8). Their whole-rock REE can be subdivided into two groups: A group with highly enriched LREE vs. HREE and a group with less enriched LREE vs. HREE and a trough in the HREE from Tb to Lu that is probably the result of stronger alteration (Figure 7B). The REE patterns of the second group actually look similar to pelitic sediments, but the positive ε_Nd(t) value of sample TH 145 and the small positive Eu anomalies of the FTS 78 samples point to a volcanic origin. The small positive Eu anomalies point to a source in North Greenland or northeastern Ellesmere Island (Thorarinsson et al., 2011a; Estrada, 2015). The REE patterns of the FTS 35−36 samples (Erdmannflya) are similar to those of the geochemically evolved Bjørndalen bentonites H1 and H4, but are steeper. The REE pattern of sample TH 144 is comparable to those of H3 and H4, but the Sm-Nd data of this sample [ε_Nd(t) = −0.9; t_DM = 0.9 Ga] indicate stronger contamination with crustal materials or potentially a different mantle source. Thus, the bentonite layers sampled during CASE 1 cannot be considered to be equivalents of the Bjørndalen bentonites, although they originate most likely also from continental-rift related magmatism. The geochemical fingerprints of the CASE 1 samples indicate that different volcanic centers may have sourced the various bentonite layers within the Basilika Formation.

Results of Zircon U–Pb Dating

All four samples from the Basilika Formation yielded similar zircon populations consisting mostly of subrounded and rounded zircon that varied from colorless to pale violet. Euhedral grains were rare and tended to be small. U–Pb LA-ICP-MS analyses were performed on a total of 430 zircons from the four bentonite horizons (Supplementary Material). The 261 zircon age analyses with ≤10% central discordance were used for further interpretation (Figure 9). Age distributions were highly similar between the four samples from the Basilika Formation, indicating constant sediment sources during deposition (Figure 10). The zircon ages with < 10% discordance span a range from 88 ± 1 Ma to 2950 ± 10 Ma (Supplementary Material) that can be subdivided into four distinct populations (Figure 10):

(i) Neoarchean (2500–2800 Ma) with a peak at 2700–2800 Ma;

(ii) Paleoproterozoic (1700–2100 Ma) with a peak around 1800−1900 Ma;

(iii) a small population at ca. 900–1000 Ma (8 of 11 ages between 974 Ma and 995 Ma);

(iv) a broad cluster of ages scattered fairly uniformly between 200 and 650 Ma.

Only a few scattered ages <200 Ma are present (88 ± 1, 152 ± 1, 154 ± 2, 162 ± 1, 188 ± 3 Ma), all older than the Paleocene depositional age of the Basilika Formation. The pale violet grains are mostly Precambrian in age. The euhedral grains show a similarly wide age range as the subrounded ones, although the youngest grains are mostly euhedral. Thus, morphology cannot be used as a discriminant for primary volcanic zircons. Th/U profiles for many of the younger grains show a characteristic, step change from low to high ratios that might indicate the presence of core-overgrowth relationships. Therefore, LA-ICP-MS profiles were measured over a single component where possible, but in some cases ages may represent mixtures of Neoproterozic and Phanerozoic components.

Only 18 concordant zircon ages were detected in the Ellesmere Island samples, which cluster around 1800−2000, 2500−2900 Ma, with minor age groups around 1500−1700 and 1000−1200 Ma (Figure 11). Mesozoic/Paleozoic ages are almost completely absent.

Discussion

Potential Volcanic Sources of Basilika Formation Bentonites

Due to stratigraphic constraints, the interval of ~62−58 Ma is a realistic estimate for the depositional age of the Basilika Formation and the volcanic ashes. This volcanic activity predates the main phase of the NALIP around 55−50 Ma, which produced the ash layers widespread in the North Atlantic region (Figure 1; Tsikalas et al., 2002; Larsen et al., 2003a; Storey et al., 2007; Tegner et al., 2011), but it is temporally related to the initial phase of the NALIP at around 61 Ma and several alkaline volcanic events in the high Arctic (Figure 1).

The initial phase of the NALIP is expressed mainly as tholeiitic intrusives and flood basalts on the British Isles, East Greenland, Baffin Island and West Greenland including Disko Island, and is related to continental rifting to the east and west of Greenland (Storey et al., 2007). The REE patterns of the Basilika Formation bentonites show no resemblance to those of the tholeiites, but they fit well to the REE signatures of alkaline basalts from Disko Island (Figure 7; Larsen et al., 2003b; Larsen and Pedersen, 2009). However, the NALIP is unlikely to have sourced the Basilika Formation bentonites as the initial phase of the NALIP was not commonly associated with explosive activity (O'Nions and Clarke, 1972; Thompson, 1982; Storey et al., 1998; Larsen and Pedersen, 2009). The only evidence for early Paleocene explosive volcanism related to the NALIP are ~58 Ma old basalt tuffs from the Svartenhuk Formation in the Nuussuaq basin of western Greenland. However, the REE patterns of the Svartenhuk Formation basalts (Larsen et al., 2016) do not resemble those of the bentonites from the Basilika and Mount Lawson formations.

Instead, some of the REE patterns of the Basilika Formation bentonites correlate relatively well with those of Lawson-ash 1, which was deposited on southern Ellesmere Island at ~60 Ma (Figure 7; Reinhardt et al., 2013), both showing steep REE patterns with strong enrichment in LREE over HREE and small negative Eu anomalies. Additionally, Lawson-ash 1 and the Basilika Formation bentonites show broadly similar major element composition but some differences in MgO, Na₂O, and SiO₂ contents (Table 2), which may have been caused by different degrees of alteration.

The REE patterns of some of the Basilika Formation bentonites as well as Lawson-ash 1 broadly resemble those of the evolved members of the Kap Washington Group on North Greenland as well as volcanic detritus in the Cape Back Basin near Nares Strait on Ellesmere Island (Estrada et al., 2010; Thorarinsson et al., 2011a). The Kap Washington Group comprises a 5000 m thick suite of bimodal, basaltic to rhyolitic rocks, which was emplaced between 71 and 61 Ma (Brown et al., 1987; Estrada et al., 2001; Tegner et al., 2011; Thorarinsson et al., 2011a,b). The evolved phase of the Kap Washington Group was associated with explosive volcanism producing tuffs, ignimbrites, and pyroclastic sandstones (Brown et al., 1987; Thorarinsson et al., 2011a) and may potentially have sourced part of the bentonites on Svalbard and Ellesmere Island. In this case, the time of deposition of Lawson-ash 1 would need to coincide with the last volcanic activity on northern Greenland, which can be reconciled only if the maximum errors of age determinations are considered (Reinhardt et al., 2013). Moreover, the initial εNd(t) values of Lawson-ash 1 [εNd(60) = 2.9] exclude a provenance from the evolved phase of the Kap Washington Group [εNd(70) < 1]. Similarly, the εNd(60) values of the Basilika Formation bentonites (0.3−1.6) are higher than or within the upper range of the most positive εNd(t) values observed in evolved rocks of the Kap Washington Group (Estrada and Henjes-Kunst, 2004; Thorarinsson et al., 2011a).

The alkaline volcanic detritus from the Cape Back basin, originating from ignimbrites and lava flows produced between 61 and 58 Ma, exhibits REE patterns that resemble those of Lawson-ash 1 and parts of the Basilika Formation bentonites (CASE 1 samples FTS35-36) including the lack of a negative Eu anomaly (Figure 7; Estrada et al., 2010). Additionally, the εNd(t) values of both Lawson ash 1 and the Basilika Formation bentonites fall into the wide range of εNd(t) values of the Cape Back basin detritus. Given the close temporal and paleogeographic proximity, it is plausible that the volcanic detritus of the Cape Back basin and parts of the bentonite layers from Svalbard and Ellesmere Island share a common, alkaline, continental rift-related source and potentially originate from the same volcanic center (Estrada et al., 2010).

The dating of Lawson-ash 2 from Ellesmere Island was associated with relatively high uncertainties, but if the preliminary age of 53.90 ± 3.1 Ma (Reinhardt et al., 2013) is correct, it would postdate the Basilika Formation bentonites and the youngest known alkaline volcanic suite in the High Arctic at Bathurst Island (~56 Ma; Day et al., 2005). The Lawson-ash 2 bentonite layer shows significantly less enrichment in LREE (Figure 7) and seems geochemically and temporally related to the later, major phase of the NALIP. Similar to coeval ash deposits in the North Atlantic Ocean (Figure 1; cf. Larsen et al., 2003a), this ash may have emanated from volcanic centers in eastern Greenland (Heister et al., 2001) or western Greenland (Thompson, 1975; Saunders et al., 1997; Larsen et al., 2016), and might be causally linked to the occurrence of early Eocene bentonites in the Frysjaodden Formation in the CTB of Svalbard (cf. Charles et al., 2011).

In conclusion, the Paleocene bentonites on Svalbard and Ellesmere Island fit well into the emerging view of widespread continental rift-related alkaline volcanism in the Canadian High Arctic and North Greenland that predates the almost simultaneous start of spreading in the Eurasia Basin, Baffin Bay, and the North Atlantic Ocean. This volcanism was possibly originally of a larger extent than it can be assumed from the preserved volcanic centers, i.e., the outcrops of the ca. 80 Ma old Hansen Point Volcanic Complex on northwestern Ellesmere Island (Estrada and Henjes-Kunst, 2004, 2013) and the ~71−61 Ma old Kap Washington Group of North Greenland (Tegner et al., 2011; Thorarinsson et al., 2011b). This volcanic phase is furthermore recorded by the ~61−58 Ma old volcanic detritus in Paleocene basins (Cape Back Basin) on northeastern Ellesmere Island (Figure 1; Estrada et al., 2010) and ~58 Ma old tuffs in the Nuussuaq basin in western Greenland (Larsen et al., 2016).

Provenance of Detrital Zircons in the Basilika Formation

Potential Sources of Detrital Zircons in the Basilika Formation

The dominant zircon age populations of the Basilika Formation bentonites imply contributions of cratonic sources to the Paleoproterozoic (1800−1900 Ma) and the smaller Neoarchean (2700−2800 Ma) populations, as well as from Uralian, Caledonian and Timanian sources and probably the Siberian trap volcanism to the 200−650 Ma population. A characteristic feature of the Basilika Formation is the almost complete lack of Grenville-aged and older Mesoproterozoic-aged grains (apart from a small early Neoproterozoic population of 900−1000 Ma).

Rocks with Caledonian, late Grenvillian and Paleoproterozoic to Archean ages are exposed today on northern and westernmost Svalbard (Figure 2). Northern Svalbard has experienced continuous erosion during the Jurassic to Cretaceous, followed by fault-related Late Cretaceous to Early Paleocene exhumation and erosion mainly in NE Spitsbergen. Thus, northern Svalbard potentially shed detritus toward the south during the Paleocene (Dörr et al., 2012). Potential cratonic sources in western Ny Friesland comprise ca. 1750 Ma old granitic orthogneisses, 2709 ± 28 Ma old quartz-monzonite and metasediments with Paleoproterozoic (1870−2040 Ma) and Neoarchean detrital zircon ages (Hellman et al., 1997, 2001). Detrital zircons from another related metasedimentary unit, the Smutsbreen Formation of southwestern Ny Friesland, yielded single zircon Pb-evaporation plateau ages of 2560−2680 Ma, but also of ca. 1200−1300 and 1560−1710 Ma (Gee and Hellman, 1996), i.e., in a range that is not typical for the Basilika age spectrum. Caledonian granitoids, known from Nordaustlandet and the Northwestern Basement Province, can contribute 410−450 Ma old zircons. Further, Grenvillian granites, known from Nordaustlandet and the Northwestern and Southwestern Basement provinces, were intruded at ca. 930−960 Ma (e.g., Johansson et al., 2004, 2005; Gasser, 2014, references therein). However, both the Grenvillian and Caledonian granites have intruded metasedimentary rocks with abundant zircon ages in the range of 1000−1800 Ma (e.g., Pettersson et al., 2009, 2010; Lorenz et al., 2012; Gasser and Andresen, 2013). A selective erosion of the granitoids without the country rocks is very unlikely, unless such an intrusive body has experienced a stronger exhumation as part of a fault-bounded block.

Reworking of Mesozoic sediments surrounding the CTB may represent an alternative source for the detrital zircons in the Basilika Formation. Mesozoic sediments have covered the whole of Svalbard up to the onset of uplift of northern Svalbard and the formation of the West Spitsbergen Fold-and-Thrust Belt (e.g., Smelror et al., 2009). These rocks also covered large areas of the northern and north-eastern Barents Shelf, where uplift and erosion took place probably already since the end of the Early Cretaceous, expressed by a regional unconformity between Aptian-Albian and Paleocene strata and missing Upper Cretaceous sediments on Svalbard (e.g., Smelror et al., 2009; Dallmann, 2015). Mesozoic sediments exposed in central and southern Svalbard comprise Triassic to Lower Cretaceous strata. Detrital zircon age spectra of these sediments, reported by Pózer Bue and Andresen (2014), differ strongly from those of the Basilika Formation. The characteristic 1800−1900 and 2700−2800 Ma populations of the Basilika Formation are absent or only scarcely present in Late Triassic and Early Jurassic formations. Instead these formations are dominated by zircon ages in the range of 200−500 Ma. In contrast, the other Mesozoic formations contain the 1800−1900 and 2700−2800 Ma age populations, but additionally abundant 1100−1600 Ma zircon ages, which are not typical for the Basilika Formation. In summary, reworking of and mixing between Mesozoic formations of Svalbard combined with sediment input from uplifting Caledonian and pre-Caledonian sources in northern Svalbard cannot produce the zircon age spectrum of the Basilika Formation.

Another possibility is a western provenance as indicated by similarities between the detrital zircon age spectra of the Basilika Formation (Figure 10) and the Mount Lawson Formation from southern Ellesmere Island (Figure 11). The source area for the Mesoproterozoic to Neoarchean zircons in the Mount Lawson Formation is presumably the Canadian-Greenland Shield. However, similar to the cratonic rocks of North Svalbard, the Canadian-Greenland Shield provides not only Neoarchean to Paleoproterozoic zircon ages, but also abundant zircon ages in the range of ca. 1000−1500 Ma (Røhr et al., 2008; Kirkland et al., 2009), which in turn rarely occur in the Basilika Formation. This makes a westerly source of the Basilika sediments unlikely, and points toward an easterly provenance (Figure 12).

Surprisingly, the best fit is with detrital zircon age spectra of sediments from northern Siberia, which are exposed in the Verkhoyansk Fold-and-Thrust Belt, the South Anyui Zone, Chukotka, and the New Siberian Stolbovoi Island (Miller et al., 2006, 2008; Prokopiev et al., 2008; Soloviev and Miller, 2014). The Verkhoyansk Fold-and-Thrust Belt contains a 15-km-thick Carboniferous to Jurassic sequence that was deposited along the eastern passive margin of the Siberian craton and deformed during the collision with the Kolyma-Omolon superterrane in the Late Jurassic (Prokopiev et al., 2008). Late Carboniferous to Middle Jurassic sediments from Verkhoyansk show detrital zircon age spectra that are characterized by populations from 250 to 700 Ma (with peaks at around 300, 400, and 480−500 Ma), from 1800 to 2100 Ma, a small population from ca. 2400 to 2800 Ma, and a hiatus between 900 and 1600 Ma (Miller et al., 2006; Prokopiev et al., 2008). The Middle Jurassic sediments bear additional small populations with peaks at 176 and 229 Ma (Prokopiev et al., 2008). Late Jurassic to Early Cretaceous sediments from the South Anyui Zone, Stolbovoi Island and Chukotka have very similar zircon age spectra (Miller et al., 2008). The 250−330 Ma ages, the early Paleozoic ages and the Paleoproterozoic ages in the Verkhoyansk sediments can be linked to intrusive rocks in mountain belts in the Baikal region, from where the detritus was transported by large river systems toward the Verkhoyansk passive margin (Miller et al., 2006; Prokopiev et al., 2008). The youngest ages can be correlated with the Main Granitoid Belt of the Verkhoyansk Fold-and-Thrust Belt formed during collision-related arc magmatism between 143 and 160 Ma (U-Pb SHRIMP zircon data; Akinin et al., 2009). These ages fit also well with the youngest ages of the Basilika Formation age spectrum. However, how likely is sediment transport from such remote areas to Svalbard over a distance of about 3000 km during the Paleocene?

Sediment Transport Pathways in the Arctic during the Early Paleocene

For the present-day Arctic configuration, coarse-clastic sediment transport from the river mouths along the Siberian coast toward the Fram Strait is possible by sea ice, following the Transpolar Drift (Krylov et al., 2008; Thompson et al., 2012). However, perennial sea ice cover and current conditions including the Beaufort Gyre and the Transpolar Drift were likely established during the Miocene (Krylov et al., 2008; Thompson et al., 2012). During the Paleogene, the Arctic Ocean was a much more enclosed basin with only restricted exchange toward the North Atlantic (Thompson et al., 2012). There is, however, increasing evidence for earlier Arctic glaciation periods (e.g., Moran et al., 2006; Tripati et al., 2008; Immonen, 2013), and the Basilika Formation contains erratics, glendonite (a cold climate indicator), as well as dropstones (Spielhagen and Tripati, 2009, and own field observations), which suggests episodic cold periods with at least seasonal sea ice, intermittent with warmer periods (Spielhagen and Tripati, 2009). Thus, transport of the Basilika detritus from Siberia toward the west by sea ice following similar pathways as today cannot be completely ruled out, but the observed phenomena may hint also to a seasonal, more local ice transport.

Another option would be a transport by (possibly alongshore) currents, which redistributed sediments from Siberia toward the Paleocene position of Svalbard, then still located somewhere north of Greenland (Figure 12). Little is known about the current system at that time. Sediment pathways may have followed the area between the rising northern edge of the Barents Shelf and the Lomonosov Ridge. The latter was presumably exposed above sea level during the Paleocene and Late Cretaceous, as indicated by a pronounced unconformity between ca. 80 and 56 Ma (O'Regan et al., 2008 and references therein), and thus formed a barrier for sediment currents from the Siberian coastal areas against the Amerasia Basin (Figure 12B). However, this scenario appears unlikely as the uplift of the northern edge of the Barents Shelf would also prohibit sediment transport toward the CTB in the South-West.

A third possibility is that the Basilika Formation clastics of “Siberian origin” were transported to the CTB in several steps including intermediate storage in Mesozoic sediments of the northeastern Barents Shelf. Indeed, an easterly (Uralian and Timanian) provenance has been suggested earlier for Paleozoic detrital zircons in Late Triassic strata on Svalbard and in the Sverdrup Basin (Pózer Bue and Andresen, 2014; Anfinson et al., in press). In this context it should be noted that (i) Mesozoic sediments exposed east of Svalbard (Franz Josef Land) display different detrital age patterns than those exposed on Svalbard, and do NOT contain the Grenvillian ages characteristic for the “westerly” provenance of most of the Svalbardian Mesozoic sediments (Soloviev et al., 2015); and (ii) that Mesozoic sediments of the Barents Sea underwent extensive erosion during the Paleocene, as evidenced, amongst others, by Triassic, Jurassic and Early Cretaceous sediments subcropping over wide areas of the northeastern Barents Shelf beneath the actual seafloor (e.g., Asch, 2005).

Thus, Mesozoic sediments originally sourced from the Verkhoyansk Fold-and-Thrust Belt and mixed with clastic detritus derived from the Siberian traps, the Uralides and the Timanides may have been reworked and redeposited during the Paleocene (Figure 12B). However, a convincing answer regarding transport mechanisms across the Arctic including larger scale fluvial drainage patterns will only be possible when more data become available, both on the provenance of clastic sediments and the Paleogene climate evolution of the Arctic.

In any case, it is striking that the Basilika Formation apparently received only little detritus from the immediate surroundings, including the Mesozoic sediments presently exposed directly east of the CTB, the area of the present West Spitsbergen Fold-and-Thrust Belt toward the west, and northern Svalbard, which was emergent and underwent erosion during the Paleocene (Dörr et al., 2012). Only the small late Grenvillian age group contained in the Basilika sediments seems to be derived from local sources, most likely from the Newtontoppen area toward the north, which was relatively rapidly exhumed during the Paleocene (Dörr et al., 2012; the Newtontoppen is the highest peak of present-day Svalbard, consisting of a Caledonian granitic body intruded into Proterozoic metamorphic rocks). The paucity of local detritus implies that the West-Spitsbergen Fold-and-Thrust Belt was still not active, in agreement with previous assumptions (Tessensohn and Piepjohn, 2000), and that the Mesozoic cover, today exposed east of the CTB, was still an area of deposition rather than of erosion. Higher subsidence rates occurred in the Eocene, when the West Spitsbergen Fold-and-Thrust Belt became active, and sediment transport changed from an easterly to a westerly source area (Helland-Hansen, 1990; Bruhn and Steel, 2003). The lack of local detritus also implies that the contribution from North Svalbard was either minor and thus strongly diluted by the dominant eastern source (in agreement with relatively low erosion rates reported for North Svalbard; Dörr et al., 2012), or that North Svalbard was drained toward the North into the assumed lowland between Svalbard and the Lomonosov Ridge, with the local drainage divide presumably located in the Newtontoppen area.

Conclusions

In this study we investigated the characteristics of bentonite layers contained in the Paleocene Basilika Formation, as well as the provenance of their clastic host rocks. Bentonite layers were sampled from the lower and upper parts of the Basilika Formation in the Bjørndalen area on central Spitsbergen. For geochemical comparisons, additional bentonite samples from the Basilika Formation taken during the CASE 1 expedition of BGR in 1992 and two samples from the Lower Paleogene Mount Lawson Formation of Ellesmere Island (Canadian Arctic) were included.

From our data we draw the following conclusions:

(i) The four analyzed bentonite layers from Bjørndalen represent altered volcanic ashes that originate from a common mantle source, but with different degrees of partial melting and having experienced different fractionation. The older layers are less evolved than the younger layers and were derived from an alkali-basaltic parental magma, whereas the younger layers were derived from a trachytic to alkali-rhyolitic precursor.

(ii) The geochemical fingerprints of the CASE 1 bentonite samples show similarities with, but also differences to, the Bjørndalen bentonites, which renders an origin of all Basilika bentonites from a common volcanic center unlikely. However, some of the CASE 1 samples may share a common source with the volcanic detritus from the Nares Strait area (Estrada et al., 2010).

(iii) The Basilika Formation bentonites, the ~60 Ma bentonite layer on Ellesmere Island (Lawson-ash 1; Reinhardt et al., 2013) and the 61–58 Ma volcanic detritus in the Nares Strait area (Estrada et al., 2010) all originate from alkaline, continental-rift magmatism related to the last, explosive stages of the HALIP. This magmatism was probably more widespread in the Arctic during the Paleocene than it can be assumed from the present-day extent of volcanites (e.g., the 71−61 Ma Kap Washington Group of North Greenland; Tegner et al., 2011; Thorarinsson et al., 2011b). Other volcanic centers may have been eroded or remain hidden beneath ice caps.

(iv) All Bjørndalen bentonite layers contain detrital zircons with four distinct U–Pb age groups of Neoarchean (2800−2700 Ma), Paleoproterozoic (1900−1800 Ma), late Grenvillian (1000−900 Ma), and Neoproterozoic/Phanerozoic age (650−200 Ma). The constant zircon age patterns indicate that the clastic sediments hosting the bentonite layers originated from the same source over the whole time of deposition of the Basilika Formation.

(v) The Grenvillian age group is very small, thus excluding a significant input from local sources or from westerly sources and pointing to a predominantly eastern provenance. The most important source seems to have been the Cretaceous Verkhoyansk Fold-and-Thrust Belt exposed in Siberia more than 3000 km away from Svalbard.

(vi) Several sediment transport mechanism are conceivable, such as ice rafting or alongshore currents following the northern edge of the Eurasian continental margin. Also, multi-stage sediment transport including intermediate storage and later reworking of the Siberian detritus in Mesozoic sediments may explain the zircon age signature of the Basilika Formation. Further provenance studies on different stratigraphic layers and locations across the Arctic are required to gain a better understanding of Paleocene sediment pathways, current systems, fluvial drainage systems, and transport mechanisms.

Author Contributions

FE, CS, SE, FL designed research; FE, SE, DD, LR, FH, NA, RD performed research; FE, CS, SE, DD, LR, FH, RD, KP analyzed data; and FE, CS, SE wrote the paper with input from all co-authors.

Conflict of Interest Statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

We thank Franz Tessensohn and Hans Paech for providing information on CASE 1 bentonite samples. Lotte M. Larsen and Fernando Corfu are thanked for providing comments that improved this manuscript.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/article/10.3389/feart.2016.00073

References