Arctic sediment routing during the Triassic: sinking the Arctic Atlantis

Placement of microcontinents in the Arctic before break-up in the Early Cretaceous is a controversial issue and many different reconstructions have been proposed (e.g. Miller et al. 2013, 2018; Shephard et al. 2013; Sømme et al. 2018; Nikishin et al. 2019; Fig. 1). Understanding pre-break-up sediment transport across sedimentary basins in the Arctic could help constrain locations of microcontinents and improve plate-tectonic models, because sediment with known transport routes may serve as ‘piercing points’ in previously adjacent basins (e.g. Richardson et al. 2017). Sediment transport distance and distribution also serves as a holistic sense-check, whereby the basin configuration is considered with a source-to-sink perspective with multi-disciplinary implications for regional tectonics. Enormous sediment volumes were produced in, and prograded from, the Urals and West Siberia in the Carnian and Norian (Late Triassic) across the Greater Barents Sea Basin and Svalbard (GBSB; Klausen et al. 2019; Gilmullina et al. 2021b; Fig. 2). The mapping and budgeting of these deposits offer improved understanding of plate-tectonic process and relative positioning of terranes in the Arctic.

A microcontinent named Crockerland has previously been inferred between the GBSB and the Sverdrup Basin (Fig. 1) based on lithological and facies patterns in these two areas (Mørk et al. 1989; Embry 1993). Recent analysis shows that Triassic sediments in the GBSB, including the Late Triassic in Svalbard, are characterized by (1) a large proportion of mudstone, (2) fine- to very fine-grained sandstones (Fig. 3) and (3) a detrital zircon spectrum with a dominant Paleozoic peak and a small number of ‘young’ zircons close to depositional age (time span between c. 210 and 245 Ma; Bue and Andresen 2014; Klausen et al. 2015; Fleming et al. 2016; Flowerdew et al. 2019; Figs 4 and 5), which were supplied from sediment sources in the Urals and West Siberia, rather than Crockerland in the north (Fig. 1; Miller et al. 2013; Sømme et al. 2018; Gilmullina et al. 2021b). These sediment properties are similar to those observed in the Late Triassic in the Sverdrup Basin (Embry 1997; Omma et al. 2011; Anfinson et al. 2016). Furthermore, seismic data (Fig. 2; Gilmullina et al. 2021b) and sediment volume modelling (Fig. 6) show bypass of large amounts of sediments from the GBSB into adjacent basins (Gilmullina et al. 2021a). This raises the possibility that sediments previously believed to have originated from Crockerland in fact originated from the Urals and West Siberia and were transported a long distance across mainly subsiding basins.

An understanding of how far the sediments sourced from the Urals and West Siberia could have reached into these adjacent basins is currently lacking. Estimation of the sediment volumes bypassed off the GBSB and size of the potentially receiving basins gives necessary inputs for calculating the length of the system beyond the GBSB.

The goals of this study are fourfold: (1) to present a novel method to determine length of sediment routing systems, developed based on sediment budget calculations and investigation of provenance data; (2) to develop a model that explains Triassic sediment transport in the Arctic; (3) to evaluate whether Crockerland is a necessary concept for the Late Triassic of the Arctic; (4) to discuss how these results compare with existing plate-tectonic reconstructions for the Arctic.

During the Triassic, the Arctic comprised five main sedimentary basins: the GBSB, Sverdrup Basin, West Chukotka Basin, Arctic Alaska and East Siberian Sea Basin (Fig. 1). Our review of the stratigraphic development in these basins (based on Embry 1997; Moore et al. 2002; Tuchkova et al. 2009; Glørstad-Clark et al. 2010; Zakharov et al. 2010; Klausen et al. 2015; Rossi et al. 2019; Gilmullina et al. 2021b) shows that they share a common pattern in sedimentation rates, with large amounts of sediments supplied in the Early Triassic, small amounts in the Middle Triassic and large amounts in the Late Triassic. However, local variations are also evident.

The GBSB is filled with up to 4.5 km of sediments, supplied through a linked clinoform–mud-belt–delta–coastal plain system from the Urals and West Siberia termed the Uralo-Siberian source (Klausen et al. 2015; Gilmullina et al. 2021b; Figs 3b and 7). These sediments are represented by a large proportion of mudstone, mineralogically immature and fine-grained sandstones (Bergan and Knarud 1993), late Paleozoic to Triassic detrital zircons (Bue and Andresen 2014; Fleming et al. 2016; Klausen et al. 2022) and large sediment volumes (Gilmullina et al. 2021a). Three hundred metres of Late Triassic fluvial deposits are found in outcrops on Svalbard and Hopen Island (Riis et al. 2008; Klausen and Mørk 2014; Lord et al. 2014) and confirm a northwesterly sediment transport direction (Klausen and Mørk 2014; Haile et al. 2018), indicating that the late Carnian delta system (C3 and C4 units) reached and prograded over the most northwestern part of the GBSB. In the GBSB the early Norian (N1 unit) delta system was transgressed (Klausen et al. 2015) and prograded again over Svalbard and the western margin of the GBSB in the late Norian (N2 unit) (Fig. 3b; Klausen et al. 2015).

The Uralo-Siberian source had a continental-scale drainage system, able to supply sediment volumes comparable with those at present-day continental margins, which overspilled into adjacent Arctic basins (Gilmullina et al. 2021a). Towards the basin margins to Fennoscandia and Greenland, smaller amounts of mature sediments with older detrital zircon age spectra (0.9–2.1 Ma) also occur (Bue and Andresen 2014; Eide et al. 2018; Fig. 3b). Organic-rich mudstones of the Steinkobbe and Botneheia formations were deposited in areas so distal they did not receive coarser clastic sediments from the prograding deltas, and were particularly widespread in the Middle Triassic when sediment supply to the basin was smaller (e.g. Krajewski 2008; Krajewski and Weitschat 2015; Gilmullina et al. 2021a; Fig. 3b).

The Sverdrup Basin was infilled by deltas, mainly derived from eroded Devonian strata in Arctic Canada, Greenland and local sources (Bjorne Fm) (Fig. 3a), during the Early Triassic (Omma et al. 2011; Anfinson et al. 2016; Alonso-Torres et al. 2018). The Middle Triassic was dominated by dark bituminous shales about 60 m thick (Murray Harbour Fm) (Embry 1997), similar to time-equivalent strata in Svalbard and distal parts of the GBSB (Steinkobbe and Botneheia fms). In the Late Triassic, large amounts of mudstone-rich sediments with very fine- to fine-grained sediments up to 1400 m thick (e.g. the Hoyle Bay, Pat Bay and Skybattle fms, and the Romulus Mbr) were derived from the north, and prograded as shallow-marine to deltaic environments southward across much of the basin (Johannessen and Embry 1989; Embry 1997; Fig. 3a). The traditional view is that these northerly-derived sediments were supplied from a northern landmass that has been named Crockerland (Fig. 1; Embry 1993). The detrital zircon spectra from these sediments in the Sverdrup Basin were discovered to show the typical Uralo-Siberian source-signature, as also seen in the GBSB (Figs 4a–c and 5), leading to a modification of this hypothesis by its proponents whereby these sediments were transported from the Urals and West Siberia to the Sverdrup Basin through a low-lying but emergent Crockerland (Colpron and Nelson 2011; Anfinson et al. 2016; Embry and Beauchamp 2019; Galloway et al. 2021). This modified hypothesis still requires the existence of a currently unobservable Arctic landmass. Below, we will make the case that these sediments were not supplied from Crockerland at all but are rather the result of overspill of sediments derived from the Uralo-Siberian source through Svalbard and the northern part of the GBSB.

Sediments in the West Chukotka Basin were supplied by large delta systems, but the Lower–Middle Triassic deposits were dominated by distal turbiditic, deep-marine continental slope-equivalents to these deltas (Tuchkova et al. 2009). During the Carnian, the West Chukotka Basin was dominated by shelf to base-of-slope environments and contains a thick (up to 2 km) package of turbidites, whereas the Norian interval mostly represents a shallow shelf environment, with sediments up to 1 km thick (Tuchkova et al. 2009; Fig. 3c).

During the Early Triassic, the eastern and central parts of Arctic Alaska were dominated by a fan-delta (Ivishak Fm), which was sourced locally from Laurentia and prograded basinwards to the deep shelf from the north (Houseknecht 2019). The Middle–Upper Triassic is represented by siliciclastic, carbonate and phosphatic deposits of the Shublik Fm with a clastic wedge in its upper part. This clastic wedge, the Sag River Sandstone, represents a fine-grained marine shelf sourced from Laurentia (or the NE in modern coordinates; Mozley and Hoernle 1990). Throughout the Triassic, western Alaska faced the palaeo-Pacific Ocean and was dominated by an outer shelf environment represented by phosphatic, black shale, chert, and silicified limestone of the Otuk Fm (Tye et al. 1999; Moore et al. 2002; Houseknecht 2019, Fig. 3d), characteristic of a relatively sediment-starved submarine basin. The Karen Creek Siltstone Mbr in the upper part of the Otuk Fm is, in contrast, represented by very fine- to fine-grained sandstone deposited as turbidites (Moore et al. 2002; Whidden et al. 2018). The Karen Creek Siltstone Mbr was supplied from the east, possibly from Chukotka, and it is time-equivalent to the Sag River Sandstone (Fig. 3d).

Triassic deposits on the New Siberian Islands are characterized by thin (up to 600 m) shale-dominated deep-water deposits with carbonates, phosphorite and siderite concretions (Egorov et al. 1987; Zakharov et al. 2010). Detrital zircons of New Siberian Island show great similarity to those from Chukotka Basin samples and suggest that these regions share sediment sources (Miller et al. 2013). A clear understanding of the stratigraphic variation through Triassic time is currently lacking owing to complicated tectonic deformation and relatively few studies.

Thus, as shown above, the Arctic basins show two general patterns: (1) the sediment supply is high in the Early Triassic, low and dominated in distal areas by marine productivity during the Middle Triassic, and high again during the Late Triassic; (2) sediments shed from local sources become gradually replaced by mudstone-rich sediment with a Late Paleozoic and Triassic detrital zircon age peak. This would indicate that these now separated basins were linked prior to break-up, and that sediments were supplied to these basins across significant distances. Whether the sediment budget and catchment characteristics were sufficient to provide enough material to prograde these distances is a key question. Sediment budgets for individual time intervals and their provenance character can tell us whether the progradation length is reasonable and if the sediment source is similar in these areas, and this will be addressed below.

Here we used a database of sediment volumes, stratigraphic seismic interpretations and sediment transport directions based on analysis of 3238 seismic 2D lines, 20 3D seismic datasets, 257 wells and 39 biostratigraphic datings (presented first by Gilmullina et al. 2021b); a detrital zircon database consisting of two new (DR1) and 16 published samples (Fig. 4); and sediment volumes and sediment supply rates in the GBSB of all stratigraphic units shown on Figure 3b (based on Gilmullina et al. 2021a).

Gilmullina et al. (2021a) estimated sediment volumes supplied to the GBSB per year using two different methods based on (1) observed volumes calculated from the seismic dataset and (2) modelled volumes from the BQART approach involving Monte-Carlo Simulation (MCS). Observed volumes calculations are based on (1) estimation of the time–thickness of each stratigraphic time unit, determined by interpreting the available dataset described above, (2) depth-conversion of top and bottom surfaces of each time unit, (3) calculation of the mass of each unit by multiplying thickness maps by density maps, created based on density logs from available wells, and (4) division of mass of each time unit by duration determined by biostratigraphic data.

In this paper we use the sediment load calculated for the Late Triassic, and input parameters for the BQART MCS are described below.

The lithology parameter (L) was assigned a normal distribution with a range between 0.75 and 1.5 and an overall average value of unity, estimated for a catchment area with a wide variety of clastic, igneous and metamorphic lithologies (Petrov et al. 2012, 2016; https://vsegei.ru/ru/info/webmapget/).

Temperature values (T) were assigned to a normal distribution between 0 and 12°C, associated with cooler and more humid climate after greenhouse dominated in the Early Triassic (Scotese and Moore 2014).

The catchment area (A) for the Late Triassic includes a vast area of the Uralo-Siberian source, including the West Siberia Basin with the western part of the Siberian Platform, northern Central Asian Orogenic Belt in the south, Taimyr in the north and the Novaya Zemlya Fold and Thrust Belt, supported by detrital zircon studies (Omma 2009; Tevelev 2013; Bue and Andresen 2014; Soloviev et al. 2015; Fleming et al. 2016; Flowerdew et al. 2019; Khudoley et al. 2019; Klausen et al. 2017, 2019), and quartz fluid inclusions (Haile et al. 2021). Separate normal distributions were assigned for each sub-catchment within the overall catchment area. The maximum catchment relief parameter (R) was assigned to a normal distribution according to sub-catchments and was based on a general understanding of tectonic evolution and modern relief analogues: the Urals 2–6 km, the Central Asian Orogenic Belt 3.5–7 km, the West Siberia Basin 0.5–1.5 km and the Taimyr 0.5–9 km.

Here, we present new detrital zircon U/Pb ages from two samples: one outcrop sample from the Induan Vardebukta Fm in the Festningen section on Svalbard (Fig. 4j), and one sample from the Induan Havert Fm from a cored stratigraphic borehole on the Finnmark Platform (well 7128/9-U-1 at 83.40 m depth, Fig. 4d). The drill core sample was made available by the Norwegian Petroleum Directorate (NPD).

The samples were crushed with a disc-mill, before the zircons were concentrated, using panning and density separation techniques. Instead of hand-picking, the zircons were extracted for mounting by pipetting of ethanol to limit bias during picking. The zircons were then embedded in epoxy, ground to about half the grain thickness and polished to expose the grain cores. The grain mounts were further photographed with backscatter and cathodoluminescence detectors, using a Zeiss Supra 55VP scanning electron microscope, prior to laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) analyses at Bergen Geoanalytical Facility, University of Bergen.

For each sample, 331–349 zircons were analysed by a Nu AttoM high-resolution ICP-MS system, coupled to a 193 nm ArF excimer laser (Resonetics RESOlution M-50 LR). The laser was fired at a repetition rate of 5 Hz and with an energy of 90 mJ, using a spot size of 26 µm. Typical acquisitions consisted of 15 s measurement of blank, followed by 30 s of measurement of U, Th and Pb signals from the ablated zircon. The data were acquired in time resolved–peak jumping–pulse counting mode with one point measured per peak for masses ²⁰⁴Pb + Hg, ²⁰⁶Pb, ²⁰⁷Pb, ²⁰⁸Pb, ²³²Th, ²³⁵U and ²³⁸U. The raw data were preprocessed using a purpose-made Excel macro because of a nonlinear transition between the counting and attenuated (= analogue) acquisition modes of the ICP instruments. As a result, the intensities of ²³⁸U were left unchanged if measured in a counting mode and recalculated from ²³⁵U intensities if the ²³⁸U was acquired in an attenuated mode. The data reduction (correction for gas blank, laser-induced elemental fractionation of Pb and U, and instrument mass bias) was carried out off-line using the Iolite data reduction package (v. 3.0), with VizualAge utility (Petrus and Kamber 2012). Details of the data reduction method have been given by Paton et al. (2010). For the data presented here, blank intensities and instrumental bias were interpolated using an automatic spline function, and down-hole interelement fractionation was corrected using an exponential function. No common Pb correction was applied to the data, but the low concentrations of common Pb were controlled by observing the ²⁰⁶Pb/²⁰⁴Pb ratio during measurements. Residual elemental fractionation and instrumental mass bias were corrected by normalization to the natural zircon reference material 91500 (1065 Ma: Wiedenbeck et al. 1995). Zircon reference materials GJ-1 (609 Ma: Jackson et al. 2004) and Plešovice (337 Ma: Sláma et al. 2008) were periodically analysed during the measurement for quality control. The GJ-1 and Plešovice standards provided ages of 599.2 ± 0.4 Ma and 345.2 ± 0.3 Ma, respectively, when calibrated against the 91500 standards.

To compare previously published datasets with the new data, all analyses have been filtered in a similar way. The data have been filtered for discordance >10% or <−10% and relative error on age <20% (2$σ$⁠). For the new data, 66 out of 680 analyses were rejected. The detrital zircon data are visualized and analysed by the Python-based detrital Py-package (Sharman et al. 2018). For grains <1000 Ma, the ²³⁸U/²⁰⁶Pb age was used, whereas the ²⁰⁷Pb/²⁰⁶Pb age was used for the older grains. Uncertainties are given at 2$σ$ confidence level.

As indicated above, we have used a novel approach to reconstruct the distribution and the length of the easterly derived Triassic sediment beyond the GBSB, which was developed based on sediment budget calculations.

The well-established BQART approach (Syvitski and Milliman 2007; Sømme et al. 2013; Zhang et al. 2018; Brewer et al. 2020; Nyberg et al. 2021), can provide an estimate of sediment supply (in mass per time) to sedimentary basins when a series of key parameters about the catchment are provided (lithology, relief, area, temperature, degree of glacial coverage and water discharge). Gilmullina et al. (2021a) compared the sediment load to the GBSB measured from the seismic data with what could be expected to have been delivered from the Uralo-Siberian source throughout the Triassic using a BQART–MCS approach to quantify and represent the uncertainty for the unknown input values. Their results showed that there was generally an excellent fit with the estimated sediment load of the sedimentary units that were fully constrained within the seismic data (Induan, Olenekian and Carnian C1). Sediment loads in late Carnian (Carnian C2, Carnian C3 + 4), and Norian (Norian N2) units, as determined from seismic data, are all towards the lower end of the modelled sediment loads, constituting 40, 30 and 25% of the mode of the modelled sediment loads, respectively (Fig. 6). This indicates loss of a significant amount of sediment from the GBSB, and large-scale sediment bypass beyond the present-day limits of the GBSB can explain the documented similarities of the Upper Triassic sediments in other Arctic basins.

We assume that the difference between averages of modelled sediment loads and observed sediment loads approximate the amount of sediment that prograded over from the GBSB into adjacent sedimentary basins. The minimum width of the Arctic Basin was estimated as the distance between Svalbard and the Severnaya Zemlya Archipelago, essentially the area with confirmed distribution of the Triassic sediments (Shneyder et al. 1989). The maximum width of the circum-Arctic basinal areas was based on the 200 Ma reconstruction of Shephard et al. (2013). We used an average basin depth of 500 m as many of the backstripped second-order clinoform surfaces in the GBSB scale to such depths (Klausen and Helland-Hansen 2018), and the thickness of late Triassic formations seems to have scaled to such thicknesses before post-depositional erosion (Klausen et al. 2017). Thicknesses of the second-order Carnian and Norian sequences in the Sverdrup Basin are accordingly c. 300 and 400 m (Embry 2011).

Estimation of the sedimentary system's progradation length was made by the following workflow: (1) the volume of missing sediments per unit was calculated as a difference between modelled values that lay within the centre of the normal distribution and observed sediment load and (2) divided by the mean basin depth and (3) basin width (Table 1). This leads to a depositional model for the Arctic, which is independently verified using published and new (DR1) detrital zircon age data (Fig. 4).

The BQART model shows that the Uralo-Siberian source had potential to generate 670 (±190) MT of sediments per million years in the Carnian (Gilmullina et al. 2021a). During the Triassic until the early Carnian, sediments from the Uralo-Siberian source were largely contained within the GBSB, but after this sedimentary geometries show that progressively greater amounts of sediments prograded from the GBSB to basins to the north (Figs 1–3). This is also seen as a progressively increasing mismatch between sediment load observed in seismic data and modelled sediment load (Fig. 5, Table 1). Assuming constant sediment production in the catchment through the Carnian, which seems reasonable based on the uniform clinoform thickness and relief, 6.7 × 10⁹ MT of sediments were produced in the Uralo-Siberian source, and c. 60% of these sediments bypassed to basins to the north and NW of the GBSB (DR2).

Norian strata in the GBSB are strongly eroded, especially towards the Finnmark Platform, Loppa High and Svalbard, but also locally around salt domes reactivated at the Triassic–Jurassic transition (Müller et al. 2019). Estimates of Norian sediment supply are, therefore, more uncertain than those for the Carnian. If the Uralo-Siberian source continued to generate the same amounts of sediments, 12.4 × 10⁹ MT were generated. c. 25% of these sediments in the GBSB were later eroded and 65% probably bypassed to basins beyond.

Using the sediment volumes calculated above, it is possible to estimate how far the sediments that bypassed the GBSB prograded into the adjacent basins. Basin geometry is approximated using a simple rectangular prism, where the width is assigned to normal distribution between a minimum and maximum distance equal to the distance between Svalbard and Severnaya Zemlya of 1400 km and circum-Arctic reconstruction by Shephard et al. (2013) of 2600 km. Prism height equals average basin depth, approximated by the decompacted sediment thicknesses in the Sverdrup Basin. Average thicknesses of Late Triassic deposits in the Sverdrup Basin are up to c. 310 m, which translates to thicknesses of 560 m (DR2) when applying similar decompaction parameters and method, as used in a study on time-equivalent strata in the Barents Sea by Klausen and Helland-Hansen (2018).

Using this simple model, the mean progradation lengths of bypassed sediments beyond the GBSB become 1000 (±500) km for the Carnian and 3300 (±1300) km for the Norian (DR2) (Figs 1a and 7). This implies that sediments from the Uralo-Siberian source, bypassing the GBSB, could have supplied sediment through nearly the entire Sverdrup Basin in the Carnian, and all the way to Arctic Alaska in the Norian.

The calculated progradation lengths are supported by a compilation of new and previously published detrital zircon age data in the Arctic basins throughout the Triassic (Figs 4 and 5). These data show that most areas were initially dominated by locally derived sediments. In the Southern Barents Sea (Lower Triassic), the local source is dominated by Paleoproterozoic–Early Neoproterozoic detrital zircon ages, typical for a Fennoscandian sediment source (Figs 4d and 8–12; Eide et al. 2018). On Svalbard, the local source is documented in the Lower–Middle Triassic sections (Fig. 4g–j; Bue and Andresen 2014). Although there are similarities to the typical Fennoscandian age signature, these age distributions additionally include a more prominent early–middle Paleozoic age component, which indicates a Greenland sediment source. Lower Triassic samples from the Sverdrup Basin (Anfinson et al. 2016; Alonso-Torres et al. 2018) are mostly characterized by Proterozoic and early–middle Paleozoic detrital zircon ages (Fig. 4m and n). Also, a small c. 290–250 Ma age peak indicates a short-lived, local magmatic event that occurred in the Sverdrup Basin during the Permian (Omma 2009; Omma et al. 2011; Gottlieb et al. 2014; Anfinson et al. 2016; Hadlari et al. 2016; Midwinter et al. 2016; Alonso-Torres et al. 2018). These age distributions are typical for a Laurentian sediment source in the Sverdrup Basin. The local sources recorded in the Southern Barents Sea, Svalbard and Sverdrup Basin remained dominant until the prograding Uralo-Siberian sedimentary system arrived at different times in different locations.

The Uralo-Siberian sediment source is typically characterized by a group of c. 200–250 Ma zircons (c. 20–30%), a dominant population of c. 250–370 Ma zircons (c. 50%) and smaller groups of zircons with ages of c. 390–500 Ma and c. 540–620 Ma. Scattered older ages can also occur in minor amounts (Figs 4a–c and 8; Fleming et al. 2016; Klausen et al. 2017; Flowerdew et al. 2019). However, there is a change from the early to the late Carnian, where the abundance of zircons younger than c. 250 Ma strongly increases (Omma 2009; Bue and Andresen 2014; Fleming et al. 2016; Klausen et al. 2022). Unfortunately, too few detrital zircons age spectra have been published from the relevant areas, especially the Northern Urals Foreland Basin and South Kara. It is, therefore, difficult to say whether this change is caused by (1) increased magmatic activity in the West Siberia and Central Asian Orogenic Belt, (2) onset of magmatic activity in the Novaya Zemlya Fold and Thrust Belt or (3) a combination of these. The exhumation of Novaya Zemlya probably occurred during the Carnian (Klausen et al. 2014; Gilmullina et al. 2021a), and the possibility that it acted as a barrier for sediments coming from West Siberia could not be excluded. However, detrital zircon age signatures for Triassic samples from the Moscow Basin (Western Urals Catchment) do not include any Triassic zircons (Chistyakova et al. 2020), and it is therefore likely that the young zircons were derived from West Siberia or Novaya Zemlya, or a combination of these.

In areas close to the Uralo-Siberian source, such as the Finnmark Platform (Figs 1, 4d and 8), locally derived sediments were already replaced by sediments from the Uralo-Siberian source in the Induan (Early Triassic). The Uralo-Siberian source-signature is characteristic of the succession in Chukotka throughout the Triassic, indicating that it was located close to this provenance throughout the Triassic. In medial areas, such as Svalbard, locally derived sediments persist until sediments from the Uralo-Siberian source arrive in the earliest Carnian (C1, Figs 3 and 4e, f). At the GBSB, Svalbard, Chukotka and Sverdrup Basin (Fig. 1), an incursion of a mudstone-rich sedimentary system with sparse fine-grained sandstones with a typical Uralo-Siberian source detrital zircon signature occurs in the Late Carnian (Figs 4j, k and 9). In Arctic Alaska, locally derived zircon age spectra are observed in the Norian (Figs 4s, 8 and 9), but the characteristic young Uralo-Siberian source-signature becomes mixed in with the local signal in the late Norian (Figs 4r and 10) suggesting that the system reached all the way to Arctic Alaska. This distribution of detrital zircon ages fits well with calculated progradation lengths of bypassed Uralo-Siberian source sediment for each unit prograding sequentially from the GBSB (Fig. 1).

Looking at the current basin structure in the Arctic, the youngest ocean basin is the early Cenozoic to Recent Eurasia Basin (Fig. 13). Before this basin opened, the Lomonosov Ridge is by all researchers reconstructed at the edge of the Barents Shelf. The earlier phase of opening formed the Amerasia Basin in the Cretaceous, but the lack of magnetic anomalies (Gaina et al. 2011; Zhang et al. 2019) and lack of good understanding of the kinematics of the opening make it difficult to choose one unique palaeogeographical model for the structure of the Arctic prior to rifting. This is where the model for sediment dispersal presented above has important implications for plate-tectonic reconstructions in the Arctic.

The sedimentary record of the Chukotka Basin follows the same sediment supply trend and contains late Paleozoic and Triassic zircons best explained by bypass from a Uralo-Siberian source throughout the Triassic (Figs 8–12). This implies a close docking of the Lomonosov Ridge against the northern GBSB, and Chukotka docked close to the Lomonosov Ridge, as suggested by Miller et al. (2013, 2018; Fig. 1a). Chukotka is then located closer to the GBSB (different from Nikishin et al. 2019) and rotated more than in the reconstruction of Shephard et al. (2013) and Sømme et al. (2018) (Fig. 13). The GBSB and Greenland blocks were in that case located very close to the Sverdrup Basin (Fig. 13).

The East Siberian Sea shelf, including New Siberian Islands (NSI), is one of the most complex and poorly understood areas in the Arctic (Piepjohn et al. 2018; Prokopiev et al. 2018). The pre-break-up location of the NSI and its affiliation to Arctic or Siberia is disputed (Kuzmichev 2009; Ershova et al. 2015). The NSI deposits represented the deepest and most distal facies of the Uralo-Siberian system throughout the Triassic (Figs 8–12, Egorov et al. 1987; Zakharov et al. 2010). Thus, a position of the NSI adjacent to the Sverdrup Basin or Severnaya Zemlya and the GBSB is unlikely because these areas are dominated by fluvial deposits in the Carnian and late Norian (Figs 9–11). Only very distal facies, mainly thinly bedded shales with carbonate interbeds, are present during these times in the NSI. To deposit such distal deposits and still contain zircons with a Uralian signature, the location of the NSI must have been far offset from the main sediment transportation route, in more distal locations in line with suggestions made by Nikishin et al. (2019; Fig. 1). The precise Triassic location of the NSI remains to be resolved and is an interesting topic for future study.

A location of Arctic Alaska near Laurentia (Miller et al. 2013; Shephard et al. 2013; Nikishin et al. 2019; Døssing et al. 2020; McClelland et al. 2021) is the least controversial among reconstructed terranes; however, the angle of rotation of the continent, associated with Amerasia Basin opening, is very different depending on the researcher (Fig. 1). Distribution of sedimentary environments and published detrital zircon data support a rotation of Arctic Alaska as suggested by Shephard et al. (2013) and Gottlieb et al. (2014). Such a rotation is in accordance with the fact that sediments with an Uralo-Siberian source-signature are found only in the Lisburne Hills in the SW part of Arctic Alaska (Figs 4t and 10).

During the Carnian–Norian, the Arctic basins (GBSB, Sverdrup, Chukotka, New Siberian Island, Wrangel Island, Alaska) received clastic sediments with a significant group of zircons with ages close to the depositional age (Fig. 4; Miller et al. 2013; Flowerdew et al. 2019). Many studies have discussed the origin of these zircons and suggested different potential sources such as Taimyr (Omma et al. 2011; Fleming et al. 2016), the ‘Pangean Rim of Fire’ or a subduction zone along the western margin of Laurentia (Hadlari et al. 2017). Our results also imply that the presence of these ‘young’ zircons in the Upper Triassic deposits do not argue in favour of a magmatic arc system (e.g. Midwinter et al. 2016; Hadlari et al. 2017) extending all the way into the Arctic region. This is because similar zircon age populations were produced by the Uralo-Siberian source (Figs 8–12; Tevelev 2013; Klausen et al. 2017; Gilmullina et al. 2021a), and because sedimentary systems sourced from the Urals and Siberia, and prograding northwestwards across the GBSB into the wider Arctic, are the most likely prime cause of the Triassic zircon distribution.

The Crockerland terrane is a hypothetical landmass proposed to explain the facies distribution in Svalbard and the Sverdrup Basin (Fig. 1; Embry 1993). There are, however, numerous problems with this suggestion. First, clinoforms in the GBSB show sediment transport towards the NW (Riis et al. 2008; Glørstad-Clark et al. 2010; Gilmullina et al. 2021b) and Late Triassic channels in the GBSB and Svalbard also show sediment transport towards the NW (Klausen and Mørk 2014; Haile et al. 2018), which implies that sediment was transported from the Uralo-Siberian source across the Barents Sea over Svalbard throughout the Late Triassic in a direction trending directly towards where the Sverdrup Basin was located (Miller et al. 2013; Gilmullina et al. 2021b). The deep basin that was situated between the Uralo-Siberian source and Laurentia–Greenland accommodated thick, organic-rich marine shales of the Middle Triassic Steinkobbe, Botneheia and Murray Harbor formations, until the basin was finally filled in the Late Triassic. Previous studies have inferred a land bridge across the Arctic, merging with the proposed Crockerland landmass to explain Uralo-Siberian detrital zircons in the Sverdrup Basin (Anfinson et al. 2016; Sømme et al. 2018). This explanation implies both delta progradation without pronounced avulsion across a deep basin and that the delta circumvented the emergent landmass to reach the Sverdrup Basin. The direct route seems more credible than transport of sediments across an emergent, low-lying landmass. Second, large amounts of sediments prograded over to basins to the NW (Fig. 6) (Klausen et al. 2019), and results of the modelling presented here show the potential for the Uralo-Siberian sediment source to supply clastic material across many hundreds of kilometres. In addition, the relatively short distance between the GBSB, Svalbard and the Sverdrup Basin throughout the Triassic (Fig. 1) (Shephard et al. 2013) and the late Carnian and late Norian timing of bypass in the GBSB coincide with the timing of the Pat Bay and Hoyle Bay formations (Fig. 3) in the Sverdrup Basin. Third, it is unlikely that a very proximal landmass would supply the fine-grained and well-sorted sandstones observed in the Sverdrup Basin, and no evidence for a northern source or southerly transport directions is observed in time-equivalent strata on Svalbard (e.g. Riis et al. 2008; Gilmullina et al. 2021b). Finally, the great similarity between the detrital zircon age spectra in the late Carnian and Late Norian of the Sverdrup Basin (Fig. 4k and l) and the GBSB, including Svalbard, shows that the two basins had a common Uralo-Siberian source.

The Urals, Taimyr and Siberia have been suggested as a source for the Triassic sediments in the Sverdrup Basin in previous studies (Omma 2009; Omma et al. 2011; Miller et al. 2013; Anfinson et al. 2016), and the GBSB has even been proposed as an alternative pathway for sediment transport (Anfinson et al. 2016). Our data add weight to the idea that the Uralo-Siberian source is the primary source for the Late Triassic sediments in Arctic basins, and that the GBSB is the main sediment route through which the bypass took place (Figs 9–11). Based on the evidence presented above, we suggest that there is no need for an extra sediment source in the middle of the Triassic Arctic, as shown in many reconstructions as a Chukotka–Alaska microcontinent (Sømme et al. 2018) or as local highs (Miller et al. 2018). In fact, inferring such a terrane sets up an artificial constraint on sediment dispersal patterns and plate reconstructions because models need to account for an ‘Arctic Atlantis’. We propose that the Crockerland concept should be abandoned, that a more useful view is that the Arctic basins were connected in the Triassic, and that the Polar Urals together with source areas in West Siberia supplied the majority of the basin-filling sediment, consisting of vast amounts of mudstone-rich sediments with mineralogically immature sandstones and a characteristic detrital zircon age spectrum.

In this study, we present a novel approach, based on sediment budget modelling and support from provenance data, that helps to constrain Arctic sediment transport pathways and improve plate-tectonic and palaeogeographical reconstruction. The source-to-sink approach shows the importance of evaluating sediment bypass and the connectedness of adjacent sedimentary basins, and of using a mass-balance approach. Based on this work, we suggest revisions to the Triassic plate-tectonic reconstruction of the Arctic, although our interpretation should be tempered by the fact that this is essentially a sedimentological, not geodynamic approach. Future work would ideally include a rigorous geodynamic testing of the ideas presented.

The present study presents a revised, uniform Triassic lithostratigraphy for the Arctic, explaining the sediment supply patterns that created the characteristic detrital zircon spectra found throughout the Triassic within the Arctic sedimentary basins. Results show that the Uralo-Siberian signature was found in detrital zircons across all basins in the Carnian and the Norian, which implies that the Arctic basins were closely connected.

The results imply that the Uralo-Siberian source dominated the Arctic basins in the Late Triassic, and that enigmatic local terranes such as ‘Crockerland’ or the Pangean ‘Rim of Fire’ are not needed to explain Arctic sediment supply. Finally, we show how the reconstructed Arctic sediment routing system constrains plate-tectonic models and we offer new plate-tectonic and palaeogeographical reconstructions based on this concept.

We thank G. Hampson and one anonymous reviewer for constructive comments, which considerably improved this paper. We also thank E. Miller and A. Owen for important discussions and constructive comments. PGS, TGS and the Norwegian Petroleum Directorate are acknowledged for providing seismic data and for permission to publish seismic lines. We also thank Schlumberger for access to Petrel under an educational licence to the University of Bergen.

AG: conceptualization (equal), visualization (lead), writing – original draft (lead); TGK: conceptualization (supporting), methodology (supporting), supervision (supporting), writing – original draft (supporting), writing – review & editing (equal); AGD: conceptualization (supporting), supervision (supporting), writing – review & editing (equal); HS: formal analysis (lead), writing – review & editing (supporting); AS: data curation (supporting), writing – review & editing (supporting); CHE: funding acquisition (lead), methodology (equal), supervision (lead), validation (supporting), writing – original draft (supporting), writing – review & editing (lead)

This work was funded by the Norges Forskningsråd through the Petromaks2-program to the ISBAR (267689) and FueBAR (308799) projects, and the Russian Foundation for Basic Research through project 20-55-20007 (to A.S.).

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

All data generated or analysed during this study are included in this published article (and its supplementary information files)