Paleomagnetism of the Franz Josef Land Archipelago: Application to the Mesozoic Tectonics of the Barents Sea Continental Margin

The Barents Sea marginal basin is probably one of the better studied among the shelf structures of the Russian Arctic. Special interest towards this territory was and remains not only due to the high petroleum potential of the Mesozoic deposits but also to fundamental questions of the geology of its Precambrian–Paleozoic basement and of Arctic tectonics in general (Pusharovsky, 1976; Zonenshain and Natapov, 1987; Bogdanov and Khain, 1996, 1998; Laverov et al., 2013). Nonetheless, many issues of the geologic structure remain highly debatable today, including several opinions on the tectonic nature and development of the Barents platform cover (Gramberg, 1988; Aplonov et al., 1996; Bogdanov and Khain, 1996, 1998). Published models have much in common and come down to an attempt to find the continuation of tectonic structures from the Eurasian continental crust on the Barents and adjacent Kara shelf, as well as traces of its alteration due to the opening of the Arctic Ocean basins.

The Franz Josef Land archipelago (FJL), which includes almost two hundred small islands in the northeast of the Barents Sea, is one of the key objects of study for these questions. On the one hand, FJL is among the rare uplifts of the basement of the Barents Sea platform. On the other hand, it is a center of late Mesozoic mantle plume magmatism at 130–80 Ma that was coeval to and accompanied the opening of the Amerasia oceanic basin and the formation of widespread magmatic formations on the territory of the High Arctic (High Arctic Large Igneous Province, HALIP) (Fig. 1) (Grantz et al., 1998; Filatova and Khain, 2009; Shipilov et al., 2009; Korago et al., 2010; Gaina et al., 2014). Other HALIP mantle plume formations of similar age occur in the Canadian Arctic Archipelago, on the territory of the Central and Eastern Arctic, including the submerged area of the Central Arctic uplifts, the Alpha and Mendeleev ridges, and the De Long Islands archipelago (Fig. 1a) (Dobretsov et al., 2013; Vernikovsky et al., 2013a; Ernst, 2014; Pease et al., 2014; Shipilov, 2016). The formation of HALIP and the younger North Atlantic province (NAIP) is linked to the activity of the mantle plume that manifests on the surface as the Iceland hot spot (Lawver and Müller, 1994; Lundin and Doré, 2005; Gaina et al., 2014). Some studies propose that the Iceland plume also caused the formation of the Siberian traps (SLIP) ca. 250 Ma, as well as older provinces in North-East Asia (Lawver et al., 2002; Kuzmin et al., 2010; Kuz’min et al., 2011; Dobretsov et al., 2013). All this leads to the conclusions that the deep source of the Iceland plume was stationary within the African (Tuzo) Large Low Shear Velocity Province (LLSVP) (Metelkin et al., 2021) and that individual upwelling plumes of the LLSVP were stable for a long time (Zhao, 2001; Torsvik et al., 2008). However, several new Ar-Ar age determinations indicating a possible Jurassic episode of mantle plume magmatism on FJL (Koryakin and Shipilov, 2009; Shipilov and Karyakin, 2011) led to a renewed heated debate on the duration of formation of the Barents Sea province (Tarakhovsky et al., 1982; Grachev, 2001; Corfu et al., 2013; Dobretsov et al., 2013; Simonov et al., 2019; Abashev et al., 2020). Consequently, solving this question would affect the validation of kinematic parameters of plates drift in the Arctic and the arguments on the tectonic mode of development of the Arctic Ocean basins.

The common features of epicontinental and epioceanic structural associations in the Barents Sea and adjacent regions are strike-slip systems (Bogdanov et al., 1997; Dobretsov and Vernikovsky, 2001; Puchkov, 2003; Shipilov, 2016; Abashev et al., 2017) that usually inherit deformations from ancient sutures and subsequently form the present-day morphological and structural framework of the continental margin. In addition to direct geophysical (mainly seismologic) observations (Shipilov and Senin, 1988; Bogdanov and Khain, 1996, 1998; Minakov et al., 2012; Startseva et al., 2017), the degree of continuity in the youngest structure of the shelf can also be determined using paleomagnetic data. However, the paleomagnetic database available today for the Russian Arctic is limited to determinations mainly for the early Paleozoic for the Central and Eastern regions (Metelkin et al., 2000, 2005, 2014, 2016; Vernikovsky et al., 2011, 2013b; Chernova et al., 2017). Some of the first data for the FJL traps was published in (Mikhaltsov et al., 2016; Abashev et al., 2018, 2019). Here we report additional paleomagnetic and geochronological data for the FJL basalts and results of a special study of the host and underlying sedimentary rocks. We also present a final comparison of our data with the rare available paleomagnetic determinations for the late Paleozoic and Mesozoic rocks of the Barents Sea region and with detailed series of paleomagnetic poles for Eurasia. This enables us to estimate the role of late Mesozoic strike-slip tectonics for the Barents Sea continental margin.

The FJL archipelago is one of the largest uplifts on the Svalbard microplate, the latter forming the continental basement of the present-day North Barents Sea basin. The crust reaches 35–38 km in thickness in this region (Seredkina and Filippov, 2021) (Fig. 1b). The age of the plate is presumed Grenvillian, and the oldest Meso–Neoproterozoic rocks of the folded basement are known on adjacent islands including in the east of the Svalbard archipelago and individual locations on Novaya Zemlya (Korago et al., 1992; Bogdanov and Khain, 1996, 1998; Korago and Timofeeva, 2005; Shipilov and Vernikovsky, 2010; Piskarev et al., 2018; Petrov and Pubel’e, 2019).

In the south, in the coastal part of the Barents Sea, the Svalbard plate is sutured to the East European craton (Baltica). This suture manifests as the Timanide orogen formed during the accretionary-collisional transformation of the Proto–Ural Ocean at the Neoproterozoic–Cambrian boundary ca. 650–520 Ma (Gee et al., 2006; Kuznetsov et al., 2007; Miller et al., 2018). The resulting continental lithosphere underlies the marginal southern part of the shelf corresponding to the South Barents basin. It also forms the basement for the Bolshezemelsky block of the Timan–Pechora continental crust, and participates in the structure of the western slope of the subpolar and polar Urals and the south of the Paikhoi–Novaya Zemlya fold-and-thrust system (Puchkov, 2003; Gee et al., 2006; Kuznetsov et al., 2007; Pease and Scott, 2009; Drachev, 2016). These orogenic events are reflected in the inner part of the Svalbard plate as reconstructed transformations of sedimentation conditions and basins configurations (Nikishin et al., 1996). The boundary of the Svalbard plate is traced along the zone of intense deformations of mainly strike-slip kinematics and up to Mesozoic age in the basement of the Barents Sea basin located on the continuation of the Baidaratsky fault (Fig. 1b). (Aplonov et al., 1996; Bogdanov and Khain, 1996, 1998; Bogdanov et al., 1997; Abashev et al., 2017).

The subsequent orogeny related to the formation of the Scandinavian Caledonides belt caused by the collision with the North America paleocontinent (Laurentia craton) during the closing of the Iapetus Ocean is poorly defined on FJL. During the final stages of this tectonic event in the Silurian–Devonian period, the deformation affected the western part of the Svalbard plate including Spitsbergen Is. (Fig. 1b) (Gramberg, 1988; Vernikovsky et al., 2013a, Metelkin et al., 2015, Domeier, 2016).

Lithological-paleogeographic reconstructions show that the FJL territory could have been uplifted during all this time up to the end of the Devonian period (Gramberg, 1988; Nikishin et al., 1996). Flat lying Carboniferous deposits overlie the intensely reworked quartzite-schistose sequence (Nagur Group) of the folded basement with a sharp angular unconformity, whereas most of the Svalbard plate has a weakly deformed Ediacaran/Vendian(?)–early Paleozoic sedimentary complex (Gramberg et al., 1985, 2004; Gramberg, 1988; Makar’ev, 2006; Stolbov et al., 2006). According to seismic profile shooting and modeling, the FJL territory subsided due to intracontinental rifting during the Frasnian age. During this time, graben-rift depressions began forming adjacent to the uplift and had active sedimentation during the Carboniferous–Permian, which marks the start of formation of the current structure of the North Barents Sea basin (Basov et al., 2009b; Startseva et al., 2017).

The late Paleozoic–early Mesozoic orogeny was due to the closing of the Ural Ocean, the formation of fold-and-thrust systems of the Ural and Paikhoi–Novaya Zemlya, as well as to collisional events in the adjacent structures of the Kara continental margin (Vernikovskii et al., 1995; Vernikovsky et al., 1999, 2018; Metelkin et al., 2000, 2005). It also manifested along the entire eastern periphery of the Barents Sea region (Puchkov, 2003; Shipilov and Vernikovsky, 2010). During this time, the general shape of the present-day boundary between the Svalbard and Kara microplates was formed among other structures. Geographically, this boundary corresponds to the St. Anna trough (Fig. 1b). Judging from the entire set of geologic-geophysical data, this tectonic boundary had mostly strike-slip kinematics both on the initial late Paleozoic–Mesozoic and on the subsequent Mesozoic–Cenozoic development stages of the Arctic shelf (Bogdanov and Khain, 1998; Shipilov and Vernikovsky, 2010).

Another manifestation of early Mesozoic strike-slip tectonics is the formation of the present-day arcuate shape of the structure of the Paikhoi–Novaya Zemlya orogen (Bogdanov and Khain, 1996; Bogdanov et al., 1997; Puchkov, 2003). For instance, paleomagnetic determinations confirm a left-lateral strike-slip of the northern Novaya Zemlya segment, possibly along the Baidaratsky suture (Abashev et al., 2017). The orogeny and related tectonic processes did not just disrupt the general shape of sedimentary systems in the eastern part of the North Barents Sea basin, but also led to an abrupt change in sedimentation mode (Basov et al., 2009b; Startseva et al., 2017). During the Late Triassic–Jurassic–Early Cretaceous, this region was dominated by continental and shallow marine sedimentation, including the accumulation of a clinoform complex typical for northern Siberia; the similar sedimentation mode, common paleogeography, and tectonic history are confirmed by taxonomic composition of the biota of both territories (Basov et al., 2009a).

Tectonic models show that the mostly strike-slip deformation mode including the emplacement of “hot” strike-slips (Dobretsov and Vernikovsky, 2001) and extension zones associating with LIPs persisted in the region up to the late Cenozoic (Bogdanov et al., 1997; Shipilov and Vernikovsky, 2010; Shipilov, 2016; Startseva et al., 2017). This was significantly influenced by the formation of oceanic basins of the Arctic Ocean (Lavwer et al., 2002; Golonka et al., 2003; Laverov et al., 2013; Jowitt et al., 2014). However, a regional analysis of available data lets us infer that the causes of these deformations are not related only to upper mantle processes dynamics during the opening of the Arctic Ocean, but also affected extensive areas beneath the Eurasian continent and could have been related to reconstructed intraplate drift between its European (Baltica) and Asiatic (Siberia) tectonic units (Metelkin et al., 2010; Timofeev et al., 2011).

Such “hot” (LIP-related) tectonics is vividly present in the Barents Sea region in trap magmatism, whose formations are the main constituents of the FJL islands (Fig. 1c). Geologic and geophysical data including anomalies in the magnetic and gravitational fields, and seismic data analysis (Shipilov and Karyakin, 2011; Minakov et al., 2012; Shipilov, 2016; Abashev et al., 2018), and well data (Gramberg et al., 1985; Komarnitskii and Shipilov, 1991; Koryakin and Shipilov, 2009) show that the main volume of traps including both the effusive and intrusive facies (a total close to 200 thous km³ (Polteau et al., 2016)) is concentrated in the offshore area around FJL and the adjacent Novaya Zemlya and Svalbard archipelagos. According to estimates, it occupies no less than a third of the Barents Sea floor and covers its entire northeastern part as well as a significant area to the south of FJL along the North Island of Novaya Zemlya towards Cape Kanin Nos (Fig. 1). However, the FJL islands are the most accessible objects for direct observation including paleomagnetic analysis.

Magnetotectonic reconstructions based on paleomagnetic data begin with determination of geologic age. Regarding FJL, this issue has been researched systematically for a long time; however, there still are some contradictions (Tarakhovsky et al., 1982; Grachev, 2001; Shipilov and Karyakin, 2011; Corfu et al., 2013; Dobretsov et al., 2013; Simonov et al., 2019; Abashev et al., 2020).

Individual high-precision U–Pb dates show that the magmatic event on FJL was brief and not older than the Barremian–Aptian interval ca. 130–110 Ma (Corfu et al., 2013). Much more abundant results of Ar-Ar dating, paleontological data, geologic mapping results and interpretations of regional geophysical lines, as well as other geologic and geophysical observations mostly confirm that and provide evidence that the FJL flood basalts lay on sedimentary rocks not older than the Oxfordian stage (Repin, 1999; Stolbov and Suvorova, 2010; Polteau et al., 2016). The Early Cretaceous age is also supported by our own paleomagnetic data (Abashev et al., 2018, 2019). Specifically, for an extensive collection (>500 samples) from the most complete trap sections on eight islands of FJL, only one case (Ametistovaya dike, Heiss Is.) recorded a reversed polarity magnetization. Keeping in mind the Early Cretaceous geochronological estimates, prevalent normal polarity is the basis for comparing the FJL trap event with the beginning of the Jalal superchron (C34n) – one of the longest “no-reversals” periods in the history of the geomagnetic field that began ca. 124 Ma and continued for almost 40 Myr. The question of mutual relationship and correlation between the frequency of magnetic reversals and the periodicity of mantle plume magmatism was raised many times, and in this case, the formation of the FJL province is in good agreement with global geodynamic models (Didenko, 2011; Gallet and Pavlov, 2016; Dobretsov, 2020; Dobretsov et al., 2021).

Despite this, several Ar-Ar determinations (Koryakin and Shipilov, 2009; Shipilov and Karyakin, 2011) and results of petrological-geochemical modeling and studies of melt inclusions (Koryakin and Shipilov, 2009; Simonov et al., 2019) imply a significantly longer, and probably multistage history of mantle plume magmatism of FJL. Specifically, Ar-Ar dates suppose three relatively brief pulses with an approximate 30 Myr periodicity of the main peaks of magmatic activity at ca. 190, 160–155 and 130–125 Ma (Dobretsov et al., 2013). Basalts of supposed early episodes are mainly identified in the southwestern part of the FJL archipelago, and the most representative sections are believed to be on Hooker, Alexandra Land, George Land, and Alger islands, and some others (Fig. 1c) (Koryakin and Shipilov, 2009; Piskarev et al., 2009; Shipilov and Karyakin, 2011; Dobretsov et al., 2013).

Accepting the stationarity of the upwelling currents of the LLSVP (Zhao et al., 2001; Torsvik et al., 2008, 2012), variously aged formations (in this case responsible for the generation of HALIP) from the same mantle plume overlapping on the limited territory of the Barents Sea province would imply that the Svalbard plate and, probably, the adjacent Arctic shelf plates were quasi-stationary during 60–70 Myr. To check this hypothesis and for paleomagnetic verification of the described tectonic and geodynamic models, we sampled some key units on Hooker and Alexandra Land islands that correspond, according to (Koryakin and Shipilov, 2009; Shipilov and Karyakin, 2011), to two older, Jurassic magmatic pulses. We also sampled several sites with supposedly Early Cretaceous basalts included in the terrigenous Triassic–Jurassic section of Heiss and Fersman islands (Fig. 1c). We determined the absolute age of basalts in the Analytical Center of IGM SB RAS (Novosibirsk) using the ⁴⁰Ar/³⁹Ar method following standard procedure (Yudin et al., 2021). We assume that the dated pyroxene and plagioclase monomineral fractions were not affected by recrystallization processes, so that the isotopic system remained closed from the crystallization of the minerals to rock sampling and reflects the true age of the rock. Preliminary data was published in brief in (Abashev et al., 2020).

Heiss and Fersman islands. The trap formation of Heiss Is. is distinguished by a series of subparallel northwest striking dikes cutting Triassic–Jurassic deposits on the entire extent of the island. The flood basalt flow capping the sedimentary section together with several sills are located along the eastern periphery of the island and on the small Fersman Is. located near its northeastern point (Fig. 2a). The flow’s Ar-Ar age for a pyroxene monofraction is 128.8 ± 12.1 Ma (Shipilov and Karyakin, 2014). Dikes and sills also have similar ages that are in general not older than 140 Ma (Grachev et al., 2001; Shipilov and Karyakin, 2014; Shipilov, 2016). For instance, we previously studied the Ametistovaya dike near the E. Krenkel polar station in the northeast of Heiss Is. According to Ar-Ar data (Shipilov and Karyakin, 2014), it is the youngest body of basaltoid magmatism (125.2 ± 2.0 Ma) along with two sills: the lower (126.2 ± 2.8 Ma) and the upper (132.0 ± 2.0 Ma) one (Mikhaltsov et al., 2016; Abashev et al., 2018).

In the course of this new study, we sampled the Skvoznaya dike on the northern coast of Heiss Is. close to Cape Zenit (Fig. 2b). Its age according to Ar-Ar data (Shipilov and Karyakin, 2014) is 138.1 ± 2.6 Ma. In the same area, a few tens of meters away we found and sampled two outcrops of yellowish-grey medium-grained feldspar-quartz sandstone (17z06 and 17z07) that correspond to the Upper Triassic Kheis Formation (Makar’ev, 2006) (Fig. 2c). The sandstone has small (up to 3 cm) casts of leaf flora (Fig. 2d) and its upper part – on the bedding plane – has small angular basalt fragments (Fig. 2e). Considering available Ar-Ar data for the Heiss Is. basalts, we can assume a younger age of this sequence or interpret this surface as an erosional one that formed during weak lithification of the sediment coeval with the active phase of magmatism at the beginning of the Cretaceous period. The sequence has a monoclinal attitude with a shallow (5–20°) southeastern dip of individual layers; therefore, it did not undergo significant deformation.

On the northeastern coast of Fersman Is., we studied three sub-conformable magmatic bodies in the terrigenous, mainly argillaceous rocks of the Heiss(?) Formation (Fig. 3a). They could probably be correlated with the dated sills of Heiss Is. Unfortunately, we do not have direct observations; however, from general topography the basalt in the uppermost body in the section (17z03) can correspond to the base part of the Heiss Is. flow. The visible thickness in the outcrop is approximately 10 m (Fig. 3b). There is a clearly visible “hot” bottom contact with the underlying sandstone, although the baked contact zone is no more than 10 cm (Fig. 3b). The boundary is even, without indicators of fluid motion, which corresponds better to the intrusive nature of this body. The total sequence is undisturbed, subhorizontal, with measured dip angles not exceeding 10°.

Approximately 5–10 m downsection is another sill (17z04) about 5–7 m thick; and further down – close to the water’s edge is the lower sill (17z05) with a visible thickness of about 10 m. The lower sill also has a distinct contact with the host rocks; it can be observed 30 m northerly from the sampling site (Fig. 3c). The interval between the basalt bodies is covered in talus deposits throughout with pieces of sandstone and mudstone that are also visible in individual small outcrops including mainly argillaceous rocks.

To validate the age of the Fersman Is. basalts we sampled the lower sill. Pyroxene from sample 17z05 yielded an upwards staircase spectrum (Table 1, Fig. 3d). For the high-temperature step corresponding to 60% of released ³⁹Ar the calculated age is 142.8 ± 1.6 Ma, Ca/K = 17. The high-temperature part consisting of two steps and corresponding to 89% of released ³⁹Ar has an age of 133.6 ± 1.6 Ma. The total age is 131.2 ± 1.4 Ma. Plagioclase from the same sample yielded a spectrum with a distinct 2-steps plateau characterized by 98% of released ³⁹Ar and an age of 130.6 ± 3.2 Ma, Ca/K = 73–80. Thus, there is grounds for considering that the obtained ages of approximately 130 Ma correspond to the closing time of the isotopic system of the minerals, i.e., they are close to the formation time of the rock. This data is in good agreement with Ar-Ar age determinations for the upper sill of Heiss Is. from (Shipilov and Karyakin, 2014).

Hooker Island. The trap section of this island is considered one of the best studied, including by paleomagnetic method (Mikhaltsov et al., 2016; Abashev et al., 2018). Various manifestations of basalt magmatism, mostly large flows, are known throughout the island. Specifically, in the region of Tikhaya Bay one of the first Early Jurassic Ar-Ar dates was obtained for a basalt at Cape Sedov (Fig. 4a, b) (Koryakin and Shipilov, 2009; Shipilov and Karyakin, 2011). It is assumed that similar Early Jurassic basaltoid magmatic formations could exist in the base of the cliff section on Cape Medvezhii and under the Voronin Glacier.

The Hooker Is. flood basalts overlie argillaceous-sandy beds attributed to the Fium and Tegetkhof formations, which have very few outcrops. Paleontological data place the Tegetkhof Formation in the Upper Triassic system (?)–Toarcian stage (Basov et al., 2009a), while the Fium Formation is interpreted as upper Bajocian–lower Oxfordian based on ammonite and other fossil finds (Repin et al., 1999; Suvorova et al., 2008). According to Stolbov and Suvorova (2010), the near ubiquitous occurrence of the Fium Formation, including in the areas described, is a natural geologic limit that precludes the possibility of Early Jurassic magmatism on Hooker Island. However, supporters of the multistage hypothesis of formation for the FJL province argue that most known outcrops of argillaceous-sandy deposits with abundant Middle–Upper Jurassic fauna are results of periglacial resedimentation. This would imply that they do not underlie the flood basalts, but were rather propped against them during transport of parts of the section directly upon the glacier or are the matrix of its flank moraines.

The basalt flows we studied have a massive structure, a typical porphyraceous or poikilitic texture and are composed of plagioclase, clinopyroxene, and a small amount of volcanic glass. Phenocrysts are rarely composed of olivine. All samples also contain a diffused ore impregnation (mostly titanomagnetite) not exceeding 5% of the rock volume (Fig. 4c).

Our latest Ar-Ar determinations for key objects on Hooker Is. including the basalt flow near Cape Sedov and the Cape Medvezhii basalt did not confirm an Early Jurassic age (Abashev et al., 2020). Plagioclase from sample 17z01 taken at the base of the outcrop near Cape Sedov yielded a 3-steps plateau with 100% of released ³⁹Ar, an age of 145.8 ± 5.3 Ma and a high Ca/K ratio between 108 and 227. The pyroxene spectrum for the same sample displays a distinct 5-steps plateau for 82% of released ³⁹Ar, an age of 138.8 ± 3.9 Ma with a Ca/K ratio from 61 to 97 (Table 1, Fig. 4d).

In outcrop 17z10 located at most 5 m above sea level near Cape Medvezhii we sampled a thick (about 30 m) body composed of well crystallized basalt (Fig. 5a). Its effusive or intrusive nature cannot be determined reliably. Plagioclase from this rock displays an upward staircase type of spectrum with Ca/K varying in the range 63–192. For the high-temperature step that yielded 83% of released ³⁹Ar the calculated age is 136.8 ± 7.7 Ma (Table 1, Fig. 5b).

Pyroxene from the same sample produced a 3-steps age spectrum. The low-temperature part includes a two-step plateau corresponding to 56% of released ³⁹Ar and an age of 127.3 ± 3.8 Ma. The remaining high-temperature step (44% of released ³⁹Ar) corresponds to an age of 182.4 ± 4.3 Ma, which is close to a date published previously (Koryakin and Shipilov, 2009). The total age is 151.9 ± 2.8 Ma. An even older age of 234.3 ± 8.5 Ma can be obtained from the high-temperature step (12% of released ³⁹Ar) for pyroxene from sample 17z01 from the Cape Sedov flood basalt (Table 1, Fig. 5b). The Ca/K ratios are high in both cases – 83 and 159 respectively. However, according to accepted standards, the age spectra with plateaus corresponding to less than 50% of released ³⁹Ar are not considered valid. Based on this, we consider that the Early Jurassic and older ages are explainable by the presence of excessive radiogenic argon, and that the rocks actually formed closer to the Jurassic–Cretaceous boundary. For pyroxene from sample 17z10 it is closest to the age of an intermediate low-temperature plateau, specifically, about 130 Ma.

A close Early Cretaceous Ar-Ar age has been obtained for the Rubini Rock basalt. This stock-like body has already been dated by K-Ar method and is not older than 145 ± 7 Ma (Stolbov, 2005). To check the quality and coherence of new Ar-Ar data and to verify available K-Ar data we studied sample 17z11 from an outcrop located near Tikhaya Bay (Figs. 4a, 6a). The pyroxene monofraction displays an upward staircase age spectrum, with no reliable plateau however (Table 1, Fig. 6b). Two high-temperature steps forming a narrow intermediate plateau with 25% of released ³⁹Ar yield a calculated age of 120.7 ± 1.3 Ma. The Ca/K spectrum for this interval increases from 13 to 25; the total age is 108.4 ± 1.1 Ma.

Plagioclase from the same sample yielded a spectrum with a distinct 8-steps plateau corresponding to 100% of released ³⁹Ar and an age of 134.0 ± 2.9 Ma. The Ca/K ratio varies between 42 and 94. Thus, the closest age estimate for the Rubini Rock basalt can also be considered close to 130 Ma.

All outcrops from which samples for Ar-Ar dating were gathered have been studied by paleomagnetic method. For instance, outcrop 17z01 on Cape Sedov corresponds to paleomagnetic data for sites 11z01 and 11z02 published together with other data for Hooker Is. in (Abashev et al., 2018). This publication also reports data for the Rubini Rock basalt (site 11z14), that correspond to geochronological sample 17z11.

The new paleomagnetic data presented in this study characterizes the Early Cretaceous flood basalts of outcrop 17z10 near Cape Medvezhiy. Additionally, we have supplemented the data for the northwest of Hooker Is. At the base of the cliff section beneath Voronin Glacier, where according to (Koryakin and Shipilov, 2009) Early Jurassic magmatic formations might also occur, we sampled a basalt flow with multidirectional columnar jointing. The outcrop is about 50 m above sea level, has a thickness close to 15 m and can conventionally be correlated to the studied flow on Cape Sedov. There were no outcrops suitable for paleomagnetic sampling below this flow.

Alexandra Land Island. The flood basalts of this island are considered in (Koryakin and Shipilov, 2009; Shipilov and Karyakin, 2011; Shipilov, 2016; Simonov et al., 2019) as the reference object for the multistage hypothesis of formation of the FJL basaltoid province based on geochemical and Ar-Ar data. It is assumed that flood basalt flows of all three potential magmatism episodes occur here in the same section along coastal cliffs of Severnaya Bay (Fig. 7a). We reported paleomagnetic determinations for these outcrops (sites 10z04, 10z05 and 10z06) in (Mikhaltsov et al., 2016; Abashev et al., 2018).

Like on Hooker Is. mainly flood basalts are exposed on Alexandra Land Is. However, the general geology is known from well data from its northern point near Cape Nagursky (Gramberg et al., 1985). According to this description, the part of the island from its surface to depth 283 m is composed of a sub-horizontally occurring sequence of flood basalts, layers of clay, siltstone and carbonaceous mudstone of Barremian–Aptian (!). The wells then penetrated the Triassic–Carboniferous sedimentary rock complex cut by multiple dolerite intrusions, and Neoproterozoic metamorphic rocks corresponding to the basement of the Svalbard plate. Early isotope dates for basalt and dolerite of the trap complex were quite diverse and the magmatism was believed to have started in the Late Triassic and ended in the Early Cretaceous (Gramberg et al., 1985). Verification of these data by K-Ar method showed an exclusively Early Cretaceous for the intrusions penetrated by the well with all ages corresponding to the 136–108 Ma interval within error margins (Grachev et al., 2001). This result is also supported by the single precise U–Pb date (122.2 ± 1.1 Ma) for this section obtained for zircon from a sill at depth 2944 m (Corfu et al., 2016). Therefore, the Jurassic Ar-Ar dates reported in (Koryakin and Shipilov, 2009; Shipilov and Karyakin, 2011) for the Alexandra Land Is. flood basalts raise some questions. Not doubting the dates themselves, we can infer that they were obtained for Jurassic-aged xenoliths captured by the melt directly from the long-lived magma chamber during its activation in the Early Cretaceous.

We were unable to visit the reference cliff section of Severnaya Bay during fieldwork; instead, we studied the rocks close to Cape Melekhov (Fig. 7a, b). We confirmed the subhorizontal attitude of the traps and host sedimentary rocks; therefore, it is highly probable that our data was taken from the flows of the lower or middle part of the Severnaya Bay section.

Directly at the water’s edge, we sampled the cliff outcrop 17z12 which consists mainly of massive black cryptocrystalline basalt. The visible thickness of the body is about 30 m (Fig. 7c). Ar-Ar dating of pyroxene yielded an intermediate 2-steps plateau corresponding to 95% of released ³⁹Ar and an age of 134.3 ± 3.6 Ma. The Ca/K ratio increases from 64 to 77 in this interval (Table 1, Fig. 7d). The plagioclase spectrum from the same sample displays a 2-steps plateau with 100% of released ³⁹Ar, an age of 130.2 ± 4.4 Ma and Ca/K varying between 100 and 139.

The second sampled flow (outcrop 17z13) corresponds hypsometrically to the uppermost point in the area of Cape Melekhov and, probably, can be correlated with the capping flood basalt of this island. The cliff outcrops of the lower and upper flows are separated by scree debris exclusively made up of basalt, including large blocks (Fig. 7b). The upper flow is represented by massive basalt with singular calcite amygdales and has a visible thickness of 15 m at most. The new Ar-Ar data for this flow is the most contradictory. Pyroxene yielded an upwards staircase age spectrum without a distinct plateau. The Ca/K ratio varies from 26 to 70. The high-temperature step corresponding to 38% of released ³⁹Ar has a calculated age of 125.3 ± 3.5 Ma. The total age is 118.4 ± 2.4 Ma (Table 1, Fig. 7e). Plagioclase from the same sample displays a spectrum with an intermediate 2-steps plateau corresponding to 95% of released ³⁹Ar and, age of 159.0 ± 15.5 Ma and very high Ca/K ratios: 405–409. The plateau age and the total age for plagioclase 157.0 ± 10.7 Ma are not technically different from age estimates of the intermediate flow in the middle part of the cliff section of Severnaya Bay from (Koryakin and Shipilov, 2009). However, considering the totality of the isotope and geologic data, we consider this date to be unreasonably high. Taking the error parameter into account, the lower boundary for the formation time of the Cape Melekhov flood basalts should be no older than approximately 145 Ma, although 130 Ma is more likely.

In this study, we investigated the flood basalts for which there are Ar-Ar dates described above, as well as the host Triassic–Jurassic sandstones. The samples (115 oriented cores) were collected using a portable gasoline drilling machine. The local magnetic declination was accounted for according to the IGRF-13 model corrections that were verified by measuring the solar azimuth at sampling sites. The local magnetic declination is 35 degrees on average. The laboratory study procedure is standard (Butler, 1992). The measurements were conducted using the instruments of the Novosibirsk Paleomagnetic Center (NSU and IPGG SB RAS). To validate the component analysis of the natural remanent magnetization (NRM), all samples underwent incremental demagnetization by alternating field (AF). Orientations and values of the NRM directions were measured on a 755 SRM cryogenic magnetometer (2G Enterprises, USA). The composition of the dominant magnetic fraction was determined by studying the temperature relation to magnetic susceptibility k(T) using a MFK1–FA Kappabridge multifunction instrument (AGICO, Czech Republic). Magnetic hysteresis parameters were measured on a J-Meter magnetometer-coercimeter (Russia).

Scalar magnetic characteristics. The studied flows and intrusions of the FJL trap complex do not differ strongly in their magnetic characteristics (Abashev et al., 2018) and our new data supports this conclusion (Fig. 8a). Magnetic susceptibility in the basalts and dolerites varies in the range 10^–2–10^–1 SI units and the NRM varies from 1 to 20 A/m. The ratio of these parameters – the Koenigsberger ratio (Qn) – is on average higher than 1. Such a distribution of the figurative points is typical for igneous rocks that have not undergone significant alteration subsequent to crystallization, that is to say, it is an indirect indicator that the primary thermoremanent magnetization was preserved (Nagata, 1961).

The NRM and k values of the Heiss Is. sandstone and contact rocks of Fersman Is. is predictably lower compared to those of the flood basalts (Fig. 8a). The Qn figurative points on the corresponding diagram have a noticeable vertical trend, meaning their NRM varies significantly from 10^–3 to 10^–2 A/m, at the same k value of approximately 10^–4 SI units. Additionally, the baked sandstone on Fersman Is. has the highest Qn > 1. Such a pattern of changes in magnetic properties of sedimentary rocks can be explained by differing degrees of secondary alterations caused by heating during the formation of the traps. Thus, the potential for preservation of the primary detrital remanent magnetization in the sandstone is low.

Domain state. The Day plot diagram (Day et al., 1977; Dunlop, 2002) representing the ratio of hysteresis parameters (Fig. 8b) primarily reflects the size of magnetic particles present in the rock. The larger they are, the closer their figurative points are to the area corresponding to the multidomain state. We observe that mainly relatively small grains are present in the studied rocks. At the same time, the figurative points of the basalts are evenly distributed along lines of experimental values for magnetite. This, again, indirectly points to the magmatic origin of the magnetic minerals in the basalts and to a lack of increase in their grain size that usually accompanies any chemical alteration.

Magnetic mineralogy. Experiments conducted for studying k(T) confidently indicate that titanomagnetite is the main magnetic mineral in the basalts (Fig. 8c). At the same time, the relative content of Ti and Fe in the primary titanomagnetite can be different, and so the corresponding Curie temperature can vary from 200 to 550 °C. During heating in the laboratory, titanomagnetite homogenizes resulting in the Curie temperature moving on the cooling graph towards higher temperatures, which indicates insignificant secondary alteration (Fig. 8c). Thus, we can confirm that the breakup of the primary titanomagnetite took place at once, during cooling of the magma melt. The irreversible character of the alteration is confirmed by subsequent experiments and was identified for all studied flood basalts, which shows that the products of an earlier magmatic pulse could not have been remagnetized by a later one. This is another indirect confirmation that the primary magnetization was preserved.

The irreversible nature of alterations has also been determined for the sandstone, however, there are differences in the mineral composition of the initial magnetic phase, the changes during laboratory experiments and, correspondingly, the shape of the k(T) graphs (Fig. 8d). Firstly, the sandstones are characterized by low values of the initial magnetic susceptibility that deviates from zero at heating to about 380 °C. After this, k(T) rises abruptly, an order of magnitude higher that the initial one, and then in the approximate interval 540–580 °C it falls just as sharply to zero. During cooling, the heating curve is not repeated, and the value of magnetic susceptibility becomes several orders of magnitude higher than the initial one. Such particularities of k(T) are easiest to explain by newly formed magnetite during the experiment.

Component analysis of magnetization. Because during heating there are changes in the composition and properties of the mineral magnetic phase, AF-demagnetization was used to identify the component of remanence.

The median destructive field (MDF) value during demagnetization of basalts varied from 5 to 50 mT depending on the composition, and correspondingly, on the magnetic hardness of the primary titanomagnetite, but more often it was about 20 mT (Fig. 9a). Usually, during the first increments and application of an alternating field of 4–6 mT (less often 10–14 mT), the viscous and/or laboratory viscous component was destroyed. The true input of the viscous component that formed in situ is hard to evaluate due to the high latitude position of the studied rocks (close to the true geomagnetic pole). Nonetheless, during AF-demagnetization in a >20 mT field, the destruction of the regular characteristic remanent magnetization (ChRM) is clearly recorded. The latter was probably not distorted by the viscous component and, considering the geologic and mineralogical facts and rock magnetic particularities reported above, it could be considered as primary thermoremanent magnetization. This component is highly stable, full demagnetization was usually achieved in fields of 120 mT and higher. The results of component analysis of flood basalts are summarized in Table 2.

The AF-demagnetization curves of the Heiss Is. sandstone and its analogs mainly on the hot contact with the upper body basalt of Fersman Is. are generally similar (Fig. 9b, c). The MDF is somewhat higher; usually its value was about 40 mT. The diagrams for the Heiss Is. sandstone sometimes have a more “noisy” signal compared to the exocontact. Full demagnetization was not always achieved; in some samples even at the highest fields of 160 mT, no more than 80–90% of the magnetization was destroyed. Nonetheless, in all samples, a ChRM is clearly identified in fields higher than 20–40 mT and its direction in geographic coordinates coincides with the basalt ChRM within the margins of error (Fig. 9d).

The contact sandstone of the upper basalt body on Fersman Is. (site 17z03) has a subhorizontal attitude, therefore, the selection of a coordinate system in unimportant when comparing the ChRM directions. The mean directions are not different from each other, the angular distance between them is γ = 4.70, the critical value being γ_c = 10.83. This confirms our supposition that the rocks directly in the exocontact of the basalt body were completely remagnetized during its emplacement.

The Heiss Is. sandstone is in a field where large trap formations are not exposed on the present-day surface. The rocks do not have visible alterations that could have resulted from thermal impact. The Skvoznaya dike is closest to them. The baking of the host rocks caused by its intrusion should not have affected the rocks occurring several tens of meters away (Metelkin et al., 2019). Moreover, despite rock magnetic indicators of remagnetization (see above), theoretically, they could have recorded the detrital remanent magnetization corresponding to the time of sedimentation. The degree and possible time of deformation of the sandstone sequence also raise questions. The rocks have a general subhorizontal attitude (Fig. 2). Nonetheless, in individual fragments of sampled outcrops some layers dip at 15–20°. Our regional observations show that the trap complex was generally not disturbed tectonically after its emplacement. Therefore, the upper age limit for the deformation of the sequence, if it happened, is the Early Cretaceous.

Using the standard fold test algorithm (Watson and Enkin, 1993), we attempted to estimate the time of magnetization fixation relative to the potential event related to deformation. Because the amount of data is limited, the calculations involved distributions of ChRM directions in samples (n = 21). The precision in the stratigraphic coordinates is noticeably lower. The ratio of the precision in stratigraphic and geographic coordinates is ks/kg = 0.5. Maximum precision is achieved at 4.6 ± 2.8 per cent untilting, i.e. in geographic coordinates. Therefore, even if deformation took place, the magnetization identified in the Heiss Is. sandstone formed after this event. If we compare its mean direction (D = 47.0, I = 76.5, α95 = 4.2) with the mean direction in the flood basalts and the upper body exocontact (17z03) of Fersman Is. (D = 32.2, I = 74.4, α95 = 5.3, n = 19), they are also not different from each other: γ = 4.50, less than γ_c = 6.57.

It is interesting to note that the mean ChRM direction of the Skvoznaya dike is different (if only slightly) from the mean direction for the Heiss Is. sandstone: γ = 8.18, higher than γ_c = 6.56 (Fig. 9d). This confirms the supposition that there was no thermal alteration due to the intrusion of the dike and points to the regional character of the remagnetization. Differences in paleomagnetic directions of the Heiss Is. dike and the Fersman Is. flood basalts reflect variations in the geomagnetic field in the Early Cretaceous (Abashev et al., 2018). At the same time, the age of formation of the basalt body on Fersman Is. is probably closer to the main pulse of magmatism that caused the regional heating.

In summary, comparing the paleomagnetic directions in the flood basalts and host sedimentary rocks of Fersman and Heiss islands indicates that the detrital remanent magnetization was destroyed and that there was a total regional remagnetization during the basaltic magmatism on FJL. Differences in paleomagnetic directions in the flood basalts of the archipelago are due to the thermal nature of the remanent magnetization, which records the instant state of the geomagnetic field. Its configuration differs from the field of the Geocentric Axial Dipole (GAD). The quality of the paleomagnetic result for FJL, the possibility of its use for solving tectonic issues depend on the degree of averaging of short-period (secular) variations and on the approximation to the GAD model (Dobretsov et al., 2021).

The Ar-Ar dates obtained in this study are in the interval 150–120 Ma. One date for pyroxene for the upper part of the Cape Melekhov section on Alexandra Land Is. is the only exception – this date is older, its reliability and correspondence to the crystallization age of the basalt melt are doubtful. Moreover, the plagioclase date is noticeably younger and corresponds to the upper part of the identified interval. The distribution of the new ages on a histogram has only one clearly defined mode (Fig. 10a). The main peak is at about 133 Ma with an additional peak at 122 Ma. Without the controversial date for Cape Melekhov, the mean age for our determinations is 132.5 ± 3.7 Ma.

In general, a similar distribution of the events probability density parameter can be also noted for the entire batch of Ar-Ar dates available for analysis (Makar’ev, 2006; Koryakin and Shipilov, 2009; Shipilov and Karyakin, 2014). Three peaks at 117.8, 123.8 and 132.6 Ma are clearly visible in the distinct Early Cretaceous maximum. The youngest peak corresponds to the mean value of 132.9 ± 5.3 Ma for the accepted batch (Fig. 10b). Moreover, there is a noticeable probability density for the Late Jurassic – a peak near 151 Ma – but its value is less than half of the main peak (Fig. 10b).

Our statistical analysis shows that published K-Ar dates (Grachev, 2001; Stolbov, 2005) imply a somewhat younger age of the trap magmatism on FJL. The main peak corresponding to the mean value of the whole batch corresponds to the age 114 ± 6.0 Ma, and the weakly defined additional secondary peak – to about 95 Ma (Fig. 10c). The displacement of K-Ar ages towards younger values is probably due to the particularities of the method.

Thus, the analysis of the totality of K-Ar and Ar-Ar geochronological data indicates that the interval of active trap magmatism on FJL corresponds to the Hauterivian–Aptian period. The Jurassic age estimates, probably, are an artefact and related to excess argon that can be caused by the occurrence of xenocrysts, seawater inclusions in the grains captured during rock formation and other factors (Konrad et al., 2019). They could also reflect processes taking place in the long-lived magma chamber that, however, did not have anything to do with the effusion of the flood basalts forming FJL (Abashev et al., 2020).

We also consider our new paleomagnetic data as evidence for the Early Cretaceous age of the basalts of Hooker and Alexandra Land islands with Jurassic Ar-Ar ages. There is virtually no difference in the angular distances between paleomagnetic directions in the outcrops dated by us and the directions of their inferred analogs including rocks with Jurassic Ar-Ar dates (Table 2). For example, taking error parameters into account, for Hooker Is. the differences between ChRM component directions in Cape Sedov flood basalts (Mikhalstov et al., 2016; Abashev et al., 2018) and the inferred analogs near Cape Medvezhii and under Voronin Glacier reported in this study are not over 10–15°, respectively. For Alexandra Land Is., the differences in paleomagnetic directions in the upper flow on Cape Melekhov and the middle flow in the Severnaya Bay cliff section are statistically insignificant, and the difference in directions in the lower flows is 10° at most. Because the rock magnetic data clearly confirms the thermal nature of the remanent magnetization in the flood basalts as well as the lack of any indicators of remagnetization on the latest magmatism stage, there is no way such a coincidence can be explained by the multistage hypothesis of formation of the FJL province.

The identified differences in paleomagnetic directions (and respectively, in coordinates of virtual geomagnetic poles) are explained by insignificant differences in formation age and secular variations of the geomagnetic field (Abashev et al., 2018, 2020). The amplitude of secular variations is determined by the value of angular dispersion (S) – this parameter is also used for standard quality estimation of the averaging of secular variations (McFadden, 1980; Merrill and McElhinny, 1983). To calculate the mean coordinates of the paleomagnetic pole and the value of the S parameter, we used the entire batch of data available for FJL – 55 determinations (for over 500 samples) including new data from this study (Table 2), data from (Abashev et al., 2018) as well as one determination for George Land Is. flood basalts (Gusev, 1970). The resulting paleomagnetic pole with parameters: Plat = 69.1°, Plon = 180.5°, A95 = 3.6°, K = 29.4 places FJL at a paleolatitude of 62.8° (Table 3) during the formation of the trap province. The S parameter equal to 14.9° completely corresponds to the modeled condition of a reliable averaging of secular variations – 10°< S <20° (Merrill and McElhinny, 1983).

Because not all determinations used in the statistics have a precise geochronological estimate, the age of the new paleomagnetic pole is accepted as the average in the Hauterivian–Aptian interval – 125 Ma. This age is also supported by the exclusive dominance of normal polarity in the FJL flood basalts. Except for the Ametistovaya dike on Heiss Is. (Mikhaltsov et al., 2016), all currently studied magmatic bodies including those without absolute age estimates were magnetized during normal polarity of the geomagnetic field. For the entire Jurassic interval up to the middle Barremian, the geomagnetic field was characterized by very frequent reversals – at least one per million year (Walker et al., 2013). Such a state of the field would certainly be reflected in the paleomagnetic record. In the Geomagnetic Polarity Time Scale (GPTS), the period of dominating normal geomagnetic field corresponds to the Cretaceous Jalal superchron C34n, begins at the end of the Barremian age (chron M1n) and continues until and including the Santonian age, which is more than 40 Myr. Therefore, the exclusive dominance of normal polarity means that the vast majority of the studied rocks probably acquired their magnetization during the Jalal superchron. Taking into account the geochronological data, this was probably either during the M1n chron or at the very beginning of C34n. These two units of the GPTS are separated by the narrow (125.9–126.3 Ma) period of reversed polarity M0r (Gradstein et al., 2012). This latter time corresponds to the formation of the reverse-magnetized dolerite of the Ametistovaya dike ca. 125.2 ± 5.5 Ma according to Ar-Ar data (Shipilov and Karyakin, 2014).

Moreover, the 125 Ma date has the best agreement with published U–Pb data for both FJL and Svalbard (Corfu et al., 2016). Although paleomagnetic determinations for Svalbard traps are limited (Krumsiek et al., 1968; Vincenz and Jelenska, 1985; Halvorsen, 1989), the coordinates of paleomagnetic poles are almost the same as for FJL (Table 3). We suppose that the insignificant angular distance between paleomagnetic directions expected for Svalbard and those observed for FJL is due to insignificant statistical paleomagnetic data for Svalbard and incomplete accounting for secular variations. Therefore, the calculated kinematic parameters describing relative drift between Svalbard and FJL cannot be considered reliable.

A much more dependable comparison can be done with synthetic data calculated from apparent polar wander paths (APWP) because their plotting methodology takes into account all relatively short-period variations.

Analysis of Mesozoic paleomagnetic poles for northern Eurasia and paleotectonic schemes created from it imply a significant intraplate strike-slip drift between the main units of the Eurasian continental plate – Europe and Siberia – with the latter undergoing a relative clockwise rotation (Metelkin et al., 2008, 2010, 2012). Various approaches and criteria are used for analyzing the available complex of Mesozoic–Cenozoic paleomagnetic poles and for debating the time following which the APW paths of Siberia and Europe become identical (Didenko, 2015). According to (Didenko, 2015), these units of the Eurasian plate were tectonically unified from the Late Jurassic (150 Ma); according to (Metelkin et al., 2010), its inner structure was not stable up to the Late Cretaceous. Timofeev et al. (2011) report that the reorganization of the inner structure of the Eurasian plate accompanied by strike-slip tectonics between Siberia and Europe recorded in paleomagnetic data for the pre-Cretaceous time could have continued into the Cenozoic era (although the scales of such displacement are below the detection limit of paleomagnetic reconstructions). Regardless of the model selected, according to general tectonic concepts, the structures of the Barents Sea shelf including FJL should seemingly have been completely united with the European unit at ca. 125 Ma. Nonetheless, the paleomagnetic directions we are observing have a better correlation with data for Siberia (Table 3). The expected paleomagnetic directions calculated for FJL coordinates are not different from the observed ones. Whereas the directions calculated from the APWP for Europe deviate at an angle 4.8 ± 2.1° and a comparison of paleomagnetic poles implies a statistically relevant relative drift to a distance of 8.4 ± 3.6°. This would mean the possibility of left-lateral strike-slip drift between FJL and Europe to an amplitude of 400 km and more. We suppose that the scale of these strike-slips is exaggerated, however the kinematics of this process is indeed in good agreement with published tectonic models (Puchkov, 2003; Metelkin et al., 2010).

A similar result was obtained from paleomagnetic data for the upper Paleozoic rocks of the Novaya Zemlya archipelago (Abashev et al., 2017). The proposed model implies that the main structure controlling the strike-slip kinematics could be the Baidaratsky suture that connects the basement of the North Barents Sea and South Barents Sea tectonic units and can be considered as the southern boundary of the Svalbard plate (Fig. 1b). The left-lateral strike-slips during the early Mesozoic period accompanied the opening of the West Siberian rift system, which reached its peak in the South Kara basin. Therefore, they led to the northwestern (in geographic coordinates) drift of the Svalbard unit relative to the arctic margin of Europe. As a result, at the Permian–Triassic boundary, the arcuate structure of the Paikhoi-Novaya Zemlya orogen was formed. In the North Novaya Zemlya segment of this orogen, the deformation is mostly thrust-sheet in nature with structural axes perpendicular to the strike-slip front; in the South Novaya Zemlya segment, which associates with the strike-slip zone, the axes of structural elements including thrusts are oriented along the strike-slip (Abashev et al., 2017).

The presented paleomagnetic comparisons indicate that such a strike-slip style persisted until the Early Cretaceous. The Late Jurassic–Early Cretaceous intensification of intraplate activity in the North Barents Sea region against the background of continued strike-slip drift and clockwise rotation of the Siberian unit probably accompanied rifting in the Eastern Arctic during the opening of the Amerasia basin and accommodated the formation of “hot” strike-slip zones and extension structures adjacent to them (Dobretsov and Vernikovsky, 2001) in the Central Arctic including the Barents Sea region.

The latitude for FJL ca. 125 Ma we reconstructed based on our paleomagnetic data (62.8° ± 2.0° N) coincides with the latitude of present-day Iceland (65° N). This, together with petrological-geochemical evidence (Ntaflos and Richter, 2003; Jowitt et al., 2014), is the basis for a model that presents the igneous provinces of Iceland and the High Arctic as the trace of the same long-lived mantle plume (Lawver and Müller, 1994; Kuzmin et al., 2010, 2011; Dobretsov et al., 2013; Metelkin et al., 2021).

Using the principles of this model, as well as information on linear magnetic anomalies that helps to reconstruct the oceanic space of the Northern Atlantic and Arctic (Gaina et al., 2014; Seton et al., 2012), we propose the following conceptual paleotectonic scheme for the Eurasian–Arctic continental margin for 125 Ma (Fig. 11). According to this model, directly after the Canada basin started opening, the FJL archipelago, the submerged Alpha and Mendeleev ridges, the adjacent offshore part of the Canadian Arctic Archipelago and the De Long Islands, while part of the same unified territory, drifted in the area of effect of the Iceland plume, which resulted in the formation of the extensive HALIP. The Amerasia basin seafloor spreading ridge manifested on the continent as areas of diffuse intracontinental rifting, which probably were the main transport system for the melts and controlled the shape of HALIP. We consider its center to be in the region of the Franz-Victoria trough (Fig. 11). The orientation of the transforms of the spreading ridge was inherited and coincided with the spatial configuration of the strike-slip zones on the continental margin. In the Barents Sea region, the largest of these zones associated with the extension area at the base of the North Barents Sea epicontinental basin and now corresponds to the rift valley between the FJL and Novaya Zemlya uplifts (Aplonov et al., 199; Bogdanov et al., 1997; Bogdanov and Khain, 1998).

The dominating extension setting is one of the important regional features of the late Mesozoic tectonics of the Barents Sea region marking its direct connection to intraplate strike-slip systems of Eurasia that continued in the Early Cretaceous according to our paleomagnetic data (Metelkin et al., 2010). Because the Siberian tectonic unit rotated clockwise faster than the European one, an area of stable thinning of the continental basement formed between them in the north of the West Siberian platform and in the South Kara basin, leading to a maximum subsidence of this territory. The North Barents Sea basin was on the continuation of this zone (Fig. 11). The strike-slip systems framing it from the European side including the Baidaratsky fault zone are of left-lateral kinematics (Bogdanov et al., 1997; Abashev et al., 2017). In contrast, the large continental fragments of the Eastern Arctic including the adjacent Kara unit (Lavwer et al., 2002; Golonka et al., 2003; Metelkin et al., 2005; Chernova et al., 2018) have the opposite dextral kinematics (Fig. 11). Thus, the Barents Sea basin can be represented as a “gigantic Mesozoic pull-apart” basin. The strike-slips contributed to the destruction of the continental crust and opening of the Amerasia basin on the one hand, and to the emplacement and development of Mesozoic deformation structures of the Barents Sea region on the other. By drifting above the hot upwelling current of the Iceland mantle plume in the Early Cretaceous, its “prepared” continental lithosphere was intensively intruded leading to the formation of HALIP, which is similar in size and volume to most known mantle plume provinces.

In summary, the presented new paleomagnetic and geochronological data help to improve several aspects of the current views on mantle plume magmatism in the Arctic (Filatova and Khain, 2009; Shipilov et al., 2009; Dobretsov et al., 2013; Ernst, 2014). Our results also expand upon the role of strike-slips during the formation of the Barents Sea region and the Eurasian-Arctic continental margin in general (Zonenshain and Natapov, 1987; Aplonov et al., 1996; Bogdanov et al., 1998; Shipilov and Vernikovsky, 2010; Laverov et al., 2013; Shipilov, 2016).

First, the obtained age estimates for the FJL archipelago basalts validate the formation of the Barents Sea sub-province of HALIP as one single mantle plume event in the Early Cretaceous geologic history of the Arctic.

The reconstructed paleolatitude for the FJL traps connects the formation of HALIP with the activity of the present-day Iceland hot spot, and therefore helps to specify the location of its Early Cretaceous trace, as well as to validate the stationarity of this hot field in the mantle numerically at least during the last 125 Myr.

Our new paleotectonic reconstructions show the leading role of strike-slip/transform tectonics during the late Mesozoic for the Barents Sea continental margin and confirms the direct link of the intraplate strike-slip systems in the Eurasian continent with the configuration and development mode of intracontinental rifts and spreading zones during the initial opening stage of the Arctic Ocean.

We are grateful to M.V. Lomonosov Northern (Arctic) Federal University, the Russian Arctic National Park, the crew of the ship Professor Molchanov, and the participants of the Arctic Floating University educational project for assistance in fieldwork on the Franz Josef Land archipelago in 2017. Corresponding member of the RAS A.N. Didenko and Dr. V.F. Proskurnin are thanked for constructive remarks and review of the manuscript.

This work was supported by the Russian Science Foundation (projects 19-17-00091 and 21-17-00052).