Dual provenance signatures of the Triassic northern Laurentian margin from detrital-zircon U-Pb and Hf-isotope analysis of Triassic–Jurassic strata in the Sverdrup Basin

Siliciclastic sedimentary successions can provide an important record of the tectonic setting and tectonic evolution of a basin through stratigraphic and detrital-zircon patterns. Previous efforts to interpret the tectonic evolution of the Sverdrup Basin have been made (e.g., Balkwill, 1978; Embry and Beauchamp, 2008), but important gaps in knowledge remain. Incipient rifting of the proto–Amerasia Basin in the Jurassic–Cretaceous (Embry, 1990, 1991; Houseknecht and Bird, 2011) was followed by opening of the Amerasian ocean basin, which separates Arctic Canada, Alaska, and northeastern Russia (e.g., Grantz et al., 1979) (Fig. 1).

Previous detrital-zircon studies in the Sverdrup Basin (Miller et al., 2006; Omma et al., 2011) have identified a detrital-zircon signature in Triassic–Jurassic strata that resembles that of the underlying Devonian clastic wedge (e.g., Anfinson et al., 2012a). Those studies also identified several different zircon assemblages in Triassic–Jurassic strata in the Sverdrup Basin and compared their age spectra to those from other Triassic strata in the circum-Arctic. Those previous detrital-zircon studies (Miller et al., 2006; Omma et al., 2011) in the Sverdrup Basin lacked data from the Late Triassic–Early Jurassic Heiberg Formation. This paper provides new detrital-zircon data from this interval, which serves to constrain the provenance of the Sverdrup Basin during the Triassic–Jurassic. Also, U-Pb detrital-zircon ages are augmented with εHf isotopic data for Permian–Triassic zircon grains. These data provide important insight into the nature of the source terrane and magmatism (cf. Vervoort and Patchett, 1996).

The Sverdrup Basin is located in the Canadian Arctic Archipelago (Fig. 2) and records near continuous sedimentation from the Carboniferous to the Paleogene (Embry and Beauchamp, 2008). The basin is underlain by an up to 10-km-thick sedimentary pile of Devonian clastic wedge strata that were deformed during the Late Devonian–Early Carboniferous Ellesmerian orogeny (Embry, 1991). Strata equivalent to the Devonian clastic wedge are widely distributed, including the northern Cordillera of North America, Arctic Alaska, and northern Russia (Amato et al., 2009; Beranek et al., 2010a; Drachev, 2011; Lemieux et al., 2011). The Ellesmerian orogeny was succeeded by initial rifting of the Sverdrup Basin that began in the Early Carboniferous and ended in the Permian (Embry and Beauchamp, 2008). Current models suggest that following the Permian, the Sverdrup Basin was tectonically quiescent and underwent thermal subsidence until rifting recommenced in the Jurassic (e.g., Embry and Beauchamp, 2008).

The Triassic stratigraphy of the Sverdrup Basin (Fig. 3) is controlled by repetitive transgressive-regressive events (e.g., Embry, 1988; Embry and Beauchamp, 2008). Early Triassic units of the Sverdrup Basin comprise the Blind Fiord Formation, which consists of shale and siltstone representing mid-outer shelf, slope, and deeper basin-floor deposits, and the Bjorne Formation, which consists mostly of sandstone confined to the basin margins interpreted to represent deltaic deposits (Embry, 1986). These two siliciclastic units mark the first major clastic influx into the basin with the accumulation of 2000 m of strata in the basin center (Embry, 1991). The Middle Triassic is marked by a major transgression that deposited bituminous source rocks of the Murray Harbour Formation (Embry and Beauchamp, 2008). These strata are overlain by clastic and/or carbonate rocks of the Roche Point Formation, which represents a short-lived regression (Embry, 1991). Earliest Carnian transgression deposited the Hoyle Bay Formation above the Roche Point Formation. Progradation of sandstone-rich, shallow marine deposits (Pat Bay Formation) extended across the basin during the late Carnian (Embry, 1993) and was terminated by a major transgression in the latest Carnian–early Norian that deposited prodelta mud and silt of the Barrow Formation (Embry, 1991). These strata progressively coarsen upward into marginal marine to nonmarine sandstones of the Heiberg Formation (Embry, 1988).

The Heiberg Formation in the central and eastern parts of the basin is subdivided into three predominantly sandstone-rich members (Embry, 1983a)—the Romulus, Fosheim, and Remus (Fig. 3). These members are stratigraphically equivalent to the Heiberg Group that comprises five formations in the western part of the basin (Fig. 3; Embry, 1983b). The formations of the Heiberg Group consist of the sandstone-rich deltaic Skybattle Formation overlain unconformably by marine mudstone of the Grosvenor Island Formation with the Triassic–Jurassic boundary occurring in its upper part (Embry and Suneby, 1994). These strata then coarsen upward into sandstone-rich strata of the deltaic Maclean Strait Formation with the upper part containing the base-Sinemurian boundary. Farther upward, these strata are overlain by marine shale of the Lougheed Island Formation capped by the sandstone-dominant King Christian Formation.

During the Triassic there were two principal sources of sediment into the basin determined by the general facies distributions shown in Figure 4 (Embry, 2009). Sediment transport was directed into the basin from the southern and eastern margins, indicating a southern and eastern sediment source area. U-Pb zircon data (Miller et al., 2006; Anfinson et al., 2012a) and Sm-Nd isotopic data (Patchett et al., 2004) are consistent with recycling from the Devonian clastic wedge and older north Laurentian strata (e.g., Hadlari et al., 2012, 2014). The facies indicate that another Triassic sediment source was derived from north of the basin (Embry, 2009), which is consistent with sandstone samples with detrital-zircon age spectra that are different from those on the south side of the basin (Miller et al., 2006; Omma et al., 2011).

In the Jurassic–Cretaceous, a narrow paleohigh, the Sverdrup Rim (Fig. 2), separated the Sverdrup Basin from the rift grabens of the proto–Amerasia Basin (Meneley et al., 1975; Embry, 1993). The northern source region was fully separated from northern Laurentia by the opening of the Amerasia Basin in the Cretaceous (Embry, 2009).

The tectonic interpretation of the Arctic region prior to the opening of the Amerasia Basin is complex and is the subject of much debate (e.g., Grantz et al., 1979; Embry, 1990; Lawver and Scotese, 1990; Lane, 1997; Lawver et al., 2002; Grantz et al., 2011; Pease, 2011; Pease et al., 2014). The region that is central to reconstruction is typically referred to as the Arctic Alaska–Chukotka microcontinent (AACM; Fig. 1), a lithospheric block occupying northeastern Russia, Arctic Alaska, and their respective offshore shelves (Pease et al., 2014). The AACM was separated from the Siberian craton by the Anyui Ocean, which closed when the Amerasia Basin and Arctic Ocean opened. The timing and mechanism of this opening is the subject of several different models (Grantz et al., 1979; Embry, 1990; Lane, 1997; Nokleberg et al., 2000; Miller et al., 2006; Kuzmichev, 2009; Grantz et al., 2011; Lawver et al., 2011).

This paper tests the rotational opening model of Grantz et al. (1979, 2011) and Embry (1990) that places the AACM against the Canadian Arctic Islands margin prior to the Cretaceous. Restoration of the AACM to its location prior to separation and rotation from the northern Laurentian margin places the North Slope of Alaska adjacent to Banks and Prince Patrick islands. As a result, the provenance of the Sverdrup Basin during the Triassic should resemble that of the North Slope of Arctic Alaska and Chukotka, since the Sverdrup Basin, Hanna Trough, and Arctic Alaska Basin would have formed a continuous sedimentary basin from the Carboniferous to Jurassic (e.g., Gottlieb et al., 2014). Seismic interpretation of an extinct spreading ridge has a trend that is parallel to the shelves offshore of the western Canadian Arctic Islands and Chukotka (Pease et al., 2014, and references therein).

In a restored position, the outboard margin of the AACM is typically interpreted to have been a passive margin to the Anyui and Angayucham oceans in the Triassic (Nokleberg et al., 2000; Miller et al., 2006; Sokolov et al., 2009; Tuchkova et al., 2009; Miller et al., 2010; Tuchkova et al., 2011; Miller et al., 2013; Amato et al., 2015). This was followed in the Cretaceous by collision with the Siberian craton–Verkhoyansk margin, known as the South Anyui suture (SAS; Fig. 1), after closure of Anyui Ocean and concomitant opening of the Amerasia Basin (e.g., Drachev, 2011; Houseknecht and Bird, 2011; Laverov et al., 2013; Amato et al., 2015). The SAS contains remnants of a Jurassic–Early Cretaceous convergent system, made up of island arc, continental terrane and oceanic basin rocks (Kuzmichev, 2009; Drachev, 2011; Amato et al., 2015).

It is not clear whether the SAS extends south of the New Siberian Islands (NSI), and therefore is part of the AACM, or if it extends north of the NSI, and therefore is unrelated to the AACM (see discussion in Kuzmichev, 2009; Pease, 2011; Pease et al., 2015). Detrital zircons from the Triassic Burustas Formation, exposed on the NSI, have a strong fraction of ca. 252 Ma zircon that, in addition to geochemical and petrographical characteristics of the zircon, are consistent with zircon ages from Siberian Trap magmatism (Miller et al., 2006), possibly indicating that the NSI were connected to Southern Taimyr and Siberia in the Early Mesozoic (e.g., Kuzmichev and Pease, 2007). Recent studies (Ershova et al., 2015; Pease et al., 2015) show a Baltican detrital-zircon character in the Carboniferous that changes to Uralian in the Permian.

The overall detrital-zircon spectrum for the Devonian clastic wedge in the Canadian Arctic has a characteristic signature of 700–360 Ma ages and a broad spectrum of Proterozoic ages (Fig. 5; Anfinson et al., 2012a, 2012b). Studies of strata tectonostratigraphically equivalent to the Devonian clastic wedge confirm a consistent and diagnostic detrital-zircon signature of this clastic package across northwestern Laurentia and the AACM (Beranek et al., 2010a, 2010b; Drachev, 2011; Lemieux et al., 2011; Anfinson et al., 2012a, 2012b), which could have been ultimately derived from Paleozoic arc rocks of the AACM (e.g., Amato et al., 2009, 2015; Lemieux et al., 2011; Hadlari et al., 2014). Analyses of detrital zircon from Triassic strata on the North Slope of Alaska (Gottlieb et al., 2014) and Triassic and Jurassic strata (Bjorne and Pat Bay formations: Miller et al., 2006; Sandy Point Formation: Omma et al., 2011) from the Sverdrup Basin have a similar signature as the Devonian clastic wedge (Fig. 5). Sedimentary recycling of detrital zircon is a common way that younger strata can mirror the detrital-zircon spectra of older strata (Hadlari et al., 2015), and so the signature of the Devonian clastic wedge provides a useful reference curve for circum-Arctic Mesozoic spectra (Fig. 5).

Triassic strata within the Sverdrup Basin (Blind Fiord and Pat Bay formations: Omma et al., 2011) have a detrital-zircon signature that has been attributed to a source in western Siberia (e.g., Taimyr, Urals, and Siberian Traps, Fig. 1; Miller et al., 2006, 2013). The diagnostic age fraction for the northern Sverdrup Basin source is a nearly continuous spectrum of near-syndepositional Permian–Triassic ages (Fig. 5). Embry (1993, 2009) postulated a northwestern provenance for Lower and Upper Triassic strata of the Sverdrup Basin based on lithofacies distribution (Fig. 4) and that this sediment source remained active until the lower Middle Jurassic. The northern source of Triassic sediment in the Sverdrup Basin, with its characteristic Permian–Triassic zircon, is observed elsewhere in the Arctic (Fig. 5), specifically in Chukotka (Miller et al., 2006; Tuchkova et al., 2011; Amato et al., 2015), Wrangel Island (Miller et al., 2006, 2010), and Lisburne Hills, Alaska (Miller et al., 2006). The Western Interior Platform adjacent to the Yukon Tanana terrane has a similar dual detrital-zircon signature with one source displaying strong similarities to the Devonian clastic wedge, whereas the other source has abundant near-syndepositional Permian–Triassic ages derived from arc rocks of the Yukon Tanana terrane (Beranek et al., 2010b; Beranek and Mortensen, 2011).

Sm-Nd isotopic data from Carboniferous to Cretaceous sedimentary rock samples from the Sverdrup Basin have a nearly uniform εNd signature throughout the basin’s history (Fig, 3), which is interpreted to represent the progressive recycling of Devonian clastic wedge strata during the Mesozoic (Patchett et al., 2004). Two excursions to more positive εNd isotopic values occurred during the Late Triassic–earliest Jurassic and Late Cretaceous (Fig. 3). Patchett et al. (2004) hypothesized that the Nd isotopic shift in the Late Triassic–earliest Jurassic resulted from minor volcanic contribution to the basin, although no evidence of volcanism had been identified at that time. The shift in the Late Cretaceous is consistent with sedimentary input from juvenile volcanic rocks of the High Arctic Large Igneous Province (e.g., Buchan and Ernst, 2006; Estrada and Henjes-Kunst, 2013).

The earliest record of volcanism within the Sverdrup Basin is the Lower Carboniferous Audhild volcanics (Trettin, 1988; Embry and Beauchamp, 2008), which coincided with early rifting. Mafic volcanic rocks within Early Permian carbonates of the Nansen Formation have been interpreted to mark the end of the Asselian stage (Mayr et al., 2002; Embry and Beauchamp, 2008). Early Permian volcanic rocks are termed the Unnamed Lower volcanics (ULV) at the base of the Great Bear Cape Formation, and the Esayoo volcanics in the Sabine Bay Formation (Morris, 2013). The Esayoo volcanics have been interpreted as intra-plate basalts with alkaline to transitional affinities (Cameron and Muecke, 1996). Although there are no previously reported records of Triassic to Jurassic volcanic rocks in the Sverdrup Basin, magmatism was active during the Cretaceous (e.g., Evenchick et al., 2015).

Samples of the Bjorne Formation and Romulus Member were collected from Ellesmere Island by the authors in 2011, and the King Christian Formation sample was collected by Carol Evenchick in 2010. U-Pb ages of detrital zircon were analyzed by secondary ion microprobe, and performed using the sensitive high-resolution ion microprobe (SHRIMP) at the Geological Survey of Canada (GSC), Ottawa. SHRIMP analytical procedures followed those described by Stern (1997), with standards and U-Pb calibration methods following Stern and Amelin (2003). Briefly, zircon were cast in 2.5-cm-diameter epoxy mounts along with fragments of the GSC laboratory standard zircon (z6266, with ²⁰⁶Pb/²³⁸U age = 559 Ma). The midsections of the zircon were exposed using 9, 6, and 1 µm diamond compound, and the internal features of the zircon (such as zoning, structures, alteration, etc.) were characterized in backscattered electron mode utilizing a Zeiss Evo 50 scanning electron microscope. Mount surfaces were evaporatively coated with 10 nm of high-purity Au. Analyses were conducted using a ¹⁶O− primary beam, projected onto the zircon at 10 kV. The sputtered area used for analysis was ∼18 µm in diameter with a beam current of ∼8–9 nA. The count rates at ten masses including background were measured over five scans with a single electron multiplier and a pulse-counting system with dead time of 23 ns. Offline data processing used SQUID 2.5 software written by Ludwig (2003). The 1σ external errors of ²⁰⁶Pb/²³⁸U ratios reported in the GSA data repository item¹ (Table DR2) incorporate a ±1.0% error in calibrating the standard zircon (see Stern and Amelin, 2003). No fractionation correction was applied to the Pb-isotope data; common Pb correction utilized the Pb composition of the surface blank (Stern, 1997). Isoplot v. 3.00 (Ludwig, 2003) was used to generate concordia plots and calculate weighted means.

Probability density function plots use ages based on the concordia age calculation method (Ludwig, 1998) as outlined in Nemchin and Cawood (2005) to avoid using different isotopic ratios for age interpretations based on an arbitrary age cutoff and to be able to evaluate probability of concordance for Phanerozoic zircons for which the traditional discordance measurement is subject to very large errors. Rather than picking a subjective cutoff between using ²⁰⁶Pb /²³⁸U and ²⁰⁷Pb/²⁰⁶Pb age, which is common practice in the literature (e.g., 1200 Ma for Omma et al., 2011; 1000 Ma for Anfinson et al., 2012a), the concordia function calculates a single age based on the relative errors of the measured ²⁰⁶Pb /²³⁸U and ²⁰⁷Pb/²⁰⁶Pb ratios (Ludwig, 1998). The approach outlined by Nemchin and Cawood (2005) involves a minimum degree of decision making and the use of probability of concordance as a screening parameter to reduce bias. In this study, a sample is not included if either the probability of concordance is <0.01 or if discordance measurements are <−5 or >5 (Table DR2).

When comparing multiple samples, the probability density function (PDF) is assumed to represent all the possible ages in the samples, and all the zircon ages must fall within the PDF curve. Accordingly, the area below the curve for each sample is comparable. Another comparative technique uses the cumulative distribution function (CDF). Although similar to the PDF, the CDF sums the probabilities with increasing age but requires equivalence in the population size. This step function does not account for uncertainties in the measured values. In spite of the differences, a PDF and CDF display the same information, but each has its own strength: PDFs are easier to use when evaluating the presence or absence of specific ages in age distributions, whereas CDFs are more useful when assessing similarities or differences within a set of age distributions (Guynn and Gehrels, 2010). Kernel density estimation (KDE) is displayed with PDF in Figure 6. An advantage of KDE is that the bandwidth is adaptive; therefore, with sparse data density, the density estimate becomes increasingly smooth. Kernel plots were produced using Density Plotter software (Vermeesch, 2012). Results are presented in Figure 6 with full U-Pb analytical data compiled in Table DR2.

Two samples were analyzed for Hf isotopes with grains selected specifically for their young U-Pb zircon ages with a total of 24 analyses conducted. Hf analyses were conducted with a Photon excimer laser and a Nu Plasma multi-collector inductively coupled plasma mass spectrometer in time-resolved analyses mode at the GSC, Ottawa. Data were acquired using either a 30, 40, or 50 μm beam size selected based on grain size of the target.

Chondritic uniform reservoir (CHUR) values of ¹⁷⁶Lu/¹⁷⁷Hf = 0.0336 and ¹⁷⁶Hf/¹⁷⁷Hf = 0.282785 are from Bouvier et al. (2008). Depleted mantle model based on ¹⁷⁶Lu/¹⁷⁷Hf = 0.03902 and ¹⁷⁶Hf/¹⁷⁷Hf = 0.28327 and new crust model age calculations utilized values of Dhuime et al. (2011) (¹⁷⁶Lu/¹⁷⁷Hf = 0.03781 and ¹⁷⁶Hf/¹⁷⁷Hf = 0.28316). ¹⁷⁶Lu decay constant of 1.867 × 10−¹¹ yr−¹ was used (Söderlund et al., 2004).

Elemental fractionation of Lu and Hf was monitored and corrected based on the 91500 standard and an accepted ¹⁷⁶Lu/¹⁷⁶Hf value of 0.000309 (Blichert-Toft, 2008). Measured values were ∼15% higher than accepted values with a reproducibility of ∼23% at 95% confidence. Reproducibility of the ¹⁷⁶Lu/¹⁷⁶Hf value was better than 5% for the 6266 standard.

Accuracy and reproducibility were monitored by replicate analyses of four zircon standards (91500, Temora 2, 6266, and Mud Tank), each showing excellent agreement with published data (Woodhead and Hergt, 2005; Wu et al., 2006; Blichert-Toft, 2008). Based on internal precision, each of the standards exhibits excess scatter as indicated by high mean square of weighted deviates (MSWD), and external errors in the range from 0.9 to 1.2 epsilon units at 2σ are required. A minimum external error of 1.2ε (2σ) is assumed for the data.

This conversion allows comparison of more geographically widespread εNd data on the Hf-evolution plot.

For the calculation of hafnium-depleted model ages (T_DM), crustal evolution trajectories assume present-day ¹⁷⁶Lu/¹⁷⁷Hf ratio of 0.0093 (Vervoort and Patchett, 1996; Amelin et al., 2000; Gehrels and Pecha, 2014). These model ages or crustal residence ages for zircon provide a qualitative estimate of the time of separation of the source rocks, or their precursors, from a hypothetical depleted mantle reservoir (Bahlburg et al., 2011). While these model ages do not necessarily provide real age information, it is useful for comparative analyses. When comparing the model ages (T_DM) of this study with other published isotopic studies, no conversion was applied to the T_DM if it was originally calculated from εNd values. Full Hf analytical results can be found in Table DR3.

The Bjorne Formation is more than 1000 m thick along the southern and eastern margins of the Sverdrup Basin and, along with the Blind Fiord Formation, represents the first significant episode of siliciclastic deposition in the basin. It has been subdivided into the predominantly sandstone Cape Butler, Pell Point, and Cape O’Brien members (Embry, 1986). A single sample was analyzed for detrital-zircon geochronology from the Cape O’Brien Member (Olenekian) on central Ellesmere Island. Zircon ages fall into three modes—a tight late Silurian peak (ca. 460–420 Ma); a broad range of Paleo- to Mesoproterozoic (ca. 2100–950 Ma); and Archean (ca. 3000–2500 Ma) ages (Fig. 6). The robust, Caledonian-aged peak has 10 grains (∼10% of the sample population), with the major peak at 430 Ma. The Mesoproterozoic range, the primary age fraction, has 74 grains (∼76% of the sample) with several peaks at 1650, 1450, and 1100 Ma. The Archean range has 12 grains (∼13% of the sample) with two minor peaks at 2760 and 2530 Ma.

On central Axel Heiberg Island (Depot Point), a sample was collected from the Romulus Member in the lower part of the Heiberg Formation. The Romulus Member is interpreted to be part of a delta front and is given a Norian age (Embry, 1983a, and references therein). Detrital-zircon ages can be divided into three distinct fractions—ca. 2580–1690 Ma, 920–750 Ma, and 560–215 Ma—with a notable absence of Mesoproterozoic ages (Fig. 6). The Cambrian to Triassic zircon fraction has 30 grains (∼63% of the sample population) with peaks at 425 and 270 Ma. The Neoproterozoic ages, of which there are five grains (∼10% of the sample), has one peak at ca. 850 Ma. The Paleoproterozoic ages have a broad spectrum with three peaks, 2500, 2150, and 1750 Ma (∼27% of the sample).

The Sinemurian to Pliensbachian King Christian Formation sample is from western Ellef Ringnes Island, where the formation is the thickest (180 m) in the basin, and here the unit is interpreted to be a deltaic deposit (Embry, 1983b). Zircon ages have two distinct age ranges—specifically Paleozoic and Precambrian age fractions. The Paleozoic ages range from ca. 441 to 262 Ma with peaks at 425 and 275 Ma (∼18% of the sample). The Precambrian can be subdivided into a Tonian-age range and a Paleo- to Mesoproterozoic range. The Tonian zircon, specifically ca. 1020–910 Ma, makes up 8% of the sample. The predominant age range is ca. 2100–880 Ma with peaks at 2000, 1650, 1450, and 1200 Ma (73% of the sample), which is similar to zircon age distribution in the Bjorne Formation (Fig. 6).

Twenty-four zircon grains were analyzed to provide further information on the source of Permian–Triassic zircon from the Lower Heiberg and King Christian formations. The εHf values range from +16 to −17 with no discernible εHf groupings within the data (Fig. 7). There is no trend (R² = 0.06) for εHf values becoming more depleted as the U-Pb age of the zircon decreases. T_DM values show a broad spectrum, but the majority of ages are Meso- to Neoproterozoic.

The study location along northern Axel Heiberg Island (Fig. 2) represents a geographically proximal location within the Sverdrup Basin to the proposed northern sediment source. At Bunde Fiord, ochreous horizons were commonly observed in marine mudstones of the Blind Fiord, Murray Harbour, Hoyle Bay, and Barrow formations (Fig. 8). The ochre layers are typically less than 10 cm thick and laterally continuous, and they are interpreted to be volcanic ash beds due to their field properties (color, moisture, and plasticity). X-ray powder diffraction (XRD) using CuKα radiation with a scanning speed of 1°2θ/min show that these ash beds consist of quartz, hydrotalcite, illite, jarosite, with lesser zircon and halloysite (Fig. 8A). Halloysite is derived from the dissolution of volcanic glass or weathered volcanic ash with a crystalline structure similar to kaolinite (Joussein et al., 2005). Similarly, hydrotalcite is formed from volcanic glass, and it has been produced experimentally by a reaction between basaltic glass and seawater (Crovisier et al., 1982). Hall and Stamatakis (2000) observed hydrotalcite infilling the molds left by the dissolution of volcanic glass shards. Collectively, the presence of volcanically derived minerals and macroscopic textural characteristics confirm that the ochreous layers are volcanic ash beds.

The detrital-zircon signature from the Bjorne Formation is remarkably similar to samples from the Devonian clastic wedge (Anfinson et al., 2012a) and is consistent with sample AE1 reported by Miller et al. (2006). The detrital-zircon signature of the Bjorne Formation is similar to Triassic strata from the North Slope of Alaska (Gottlieb et al., 2014) and similar aged strata from the northwestern Cordillera (Fig. 9A), the latter being recycled from the strata of northwestern Laurentia (Beranek et al., 2010b). Collectively the geographically dispersed areas suggest a common, areally expansive source, which most probably was the Devonian clastic wedge and equivalents (Fig. 9B), and so provenance is from a recycled source.

A sample from the Blind Fiord Formation has a wholly different signature interpreted to represent the northern source to the basin (Omma et al., 2011). Notable is the occurrence of a suite of Permian ages (ca. 290–265 Ma) and the absence of Caledonian and Ellesmerian orogen ages that typify the Devonian clastic wedge (ca. 700–360 Ma). Omma et al. (2011) suggested the source of the young ages was either Early Permian basaltic magmatic activity within the basin (e.g., Thorsteinsson, 1974), or mid-Permian syenites associated with the Uralian orogeny in the Taimyr region in central Russia (e.g., Vernikovsky et al., 1995; Zhang et al., 2013). The majority of Permian volcanic rocks are observed within the northern part of the Sverdrup Basin, which suggests there were probably volcanic equivalents to the north of the basin. The Lower Permian volcanic rocks in the Sverdrup Basin (the Esayoo, the Unnamed Lower volcanics [ULV], or equivalents) could provide the limited age range of Permian zircon within the Blind Fiord Formation; basalts may be able to supply zircon even though they typically have poor zircon potential (Rioux et al., 2012; Candan et al., 2015; Iles et al., 2015). The converted εNd values and the T_DM from the Esayoo volcanics (ca. 276 Ma) are comparable to Hf isotope data from similar aged zircon from this study (Fig. 10). Long distance transport from the Urals is unlikely to produce the tight range of Permian ages within the Blind Fiord Formation, particularly because Uralide granitoids formed at an almost constant rate from 370 Ma to 250 Ma, older in the south and younger in the north (Vernikovsky et al., 1995; Bea et al., 2002). Granitoids in the northern Urals have ages from ca. 300 to 280 Ma, and post-tectonic granitoids dated at ca. 260 Ma (Pease et al., 2015). Transport from the Urals is possible but would likely provide a broad spectrum of U-Pb ages rather than narrow peaks as seen in the Blind Fiord Formation.

Detrital-zircon U-Pb age data of the Pat Bay Formation (Miller et al., 2006; Omma et al., 2011) vary considerably between samples indicative of two different sources during deposition (e.g., Embry, 2009). Sample AE2 from Miller et al. (2006) has an assemblage of ages that overlap within the Devonian clastic wedge spectrum, including a single 376 Ma age, an age fraction of ca. 620–505 Ma, and a broad range of Paleo- to Mesoproterozoic ages. Miller et al. (2006) hypothesized the sample represented sediment derived from north of the Sverdrup Basin because those ages were unknown in northern Canada at the time. Subsequent studies (e.g., Lemieux et al., 2011; Anfinson et al., 2012a) report abundant ca. 700–500 Ma zircon ages from rocks of the Late Devonian clastic wedge as well as the Silurian flysch (Beranek et al., 2015). A prominent ca. 700–500 Ma age-fraction coupled with an absence of Permian–Triassic zircon ages would suggest a source much like Silurian and Devonian strata in the Franklinian Basin; these strata were probably ultimately derived from rocks in Arctic Alaska–Chukotka of Timanide age (e.g., Cecile et al., 1991; Amato et al., 2009, 2014). In contrast, the later work of Omma et al. (2011) reported a prominent range of near-syndepositional ages (ca. 255–217 Ma), in addition to a broad Paleozoic spectrum (ca. 490–295 Ma), which is indicative of derivation from the active margin source region.

The Lower Heiberg Formation (Romulus Member) sample has a similar detrital-zircon signature to the Pat Bay sample analyzed by Omma et al. (2011). As with the two previous northerly-derived samples from the Pat Bay and Blind Fiord formations, there is a prominent near-syndepositional–age fraction in the Lower Heiberg Formation with a continuous spectrum of ca. 300–215 Ma ages. In contrast to the Pat Bay and Blind Fiord formations, there is a notable absence of Mesoproterozoic ages (Fig. 9A) that are typically present in Laurentian strata (e.g., Hadlari et al., 2012).

The consistent zircon signatures from the Blind Fiord, Pat Bay, and Lower Heiberg formations are similar to those reported from Triassic strata of the AACM from Lisburne Hills, North Slope of Alaska, Wrangel Island, and Chukotka (Fig. 4). Samples from the Upper Triassic Otuk Formation from the Lisburne Hills have an age fraction from ca. 275–220 Ma, in addition to Carboniferous, Lower Paleozoic ages, and minor Mesoproterozoic ages (Phanerozoic peaks at 420, 355, 315, 255, and 220 Ma) (Miller et al., 2006). The detrital-zircon signature from Triassic strata along the North Slope is remarkably similar to the Devonian clastic wedge with a strong age fraction of ca. 700–360 Ma and a broad spectrum of Proterozoic ages (Gottlieb et al., 2014). When comparing detrital-zircon data from the Lisburne Hills and the North Slope, the two regions have notably different age distributions, which strongly resemble the pattern of the two different provenance signatures within the Sverdrup Basin (Fig. 9).

Similar to Lisburne Hills, Triassic samples from Wrangel Island and Chukotka document a general assemblage of near-syndepositional zircon ages and a minor representation of Mesoproterozoic ages (Miller et al., 2006, 2010; Tuchkova et al., 2011; Amato et al., 2015). More specifically, Wrangel Island samples have nearly continuous ages from ca. 480 to 205 Ma (Phanerozoic peaks at 440, 350, 305, 250, and 230 Ma) (Miller et al., 2010), and samples from Chukotka (Miller et al., 2006; Tuchkova et al., 2011; Amato et al., 2015) have a range of ca. 405–216 Ma (robust Phanerozoic peaks at 500, 300, and 250 Ma). Both regions have minor contributions from the Paleoproterozoic to Silurian. In summary, both the Sverdrup Basin and the broader AACM share the active margin source, which is consistent with the rotational model for the opening of the Arctic Ocean. Alternatively, Zhang et al. (2015) presented a reconstruction in which littoral currents allow for the redistribution of detrital material from the Polar Urals and Taimyr during the Triassic due to the sublithospheric spreading associated with the Siberian mantle plume at ca. 250 Ma, which does not explain the post–Siberian Trap range of detrital-zircon ages of 240–210 Ma.

A similar pattern of dual detrital-zircon spectra is observed along the northwestern margin of Laurentia in the Yukon. Sediment recycled from western interior basins of Laurentia has a detrital-zircon signature consistent with derivation from a Devonian clastic wedge-type source (Beranek et al., 2010b)—namely broad Phanerozoic peaks at 450, 420, and 365 Ma—and a wide spectrum of Proterozoic and Lower Paleozoic ages (YTT platform, Fig. 9). This signature remains relatively stable through Early to Late Triassic strata.

The other provenance signature is from sediment derived from the west, along the active margin of the Yukon-Tanana terrane (YTT). The YTT is a pericratonic terrane that comprises a series of deformed arc, backarc, and continental margin assemblages (Colpron et al., 2006; Nelson et al., 2006). During the Middle Permian it was separated from ancestral North America by a backarc basin, the Slide Mountain Ocean (Plafker and Berg, 1994); however, the closure of the backarc basin in the Late Permian and shortening across the YTT and the western margin of Laurentia (Klondike orogeny) formed a Triassic foreland to backarc basin to the east of the YTT (Beranek and Mortensen, 2011). Detrital-zircon signatures from Triassic strata sourced from the convergent, western margin of the YTT exhibit a suite of near-syndepositional Permian–Triassic zircon ages (Beranek and Mortensen, 2011).

The two provenance signatures from Triassic strata of the YTT, specifically an active margin source and a recycled source, share a close resemblance to the dual provenance signature of Triassic strata of the Sverdrup Basin and Arctic Alaska (Fig. 9B). This suggests that the tectonic setting for YTT basin could have been similar to the tectonic setting of the Sverdrup Basin during the Triassic.

Identifying source(s) for the consistently near-syndepositional–aged zircon in Triassic strata throughout the Sverdrup Basin and AACM is essential to understanding the tectonic setting of the Sverdrup Basin during the early Mesozoic. The broad range of εHf values from this study of U-Pb zircon ages between 330 and 200 Ma is interpreted to record a mixed source with a combination of juvenile, intermediate, and evolved crust (Fig. 7). Other εHf or εNd values from the circum-Arctic display a variety of isotopic compositions and T_DM. The ca. 710–380 Ma zircon population from Devonian strata of the Franklinian Basin provides εHf values (Anfinson et al., 2012b) that are congruent with a juvenile to intermediate source (Fig. 10B) in the AACM. Neoproterozoic to Devonian igneous rocks from Arctic Alaska have converted εNd values that are slightly less juvenile than the Devonian strata (Amato et al., 2009). The T_DM of both studies overlap with the T_DM of Permian–Triassic zircon of the Lower Heiberg Formation with predominantly Meso- to Neoproterozoic model ages (Fig. 10A). It is plausible that the lithospheric source involved during the formation of the older zircon was reincorporated during the Permian–Triassic magmatism in the AACM.

The other proposed source for Permian–Triassic zircon in the Sverdrup Basin is from the Siberian region (e.g., Urals, Taimyr, and NSI: Fig. 1; Omma et al., 2011; Miller et al., 2013, and references therein). Miller et al. (2013) and Zhang et al. (2015) recently outlined possible sources of the younger zircon ages, which included the Carboniferous to Early Permian plutonic belts from the northernmost Urals; granites as young as ca. 250 Ma from the southern Urals; mafic magmatism (ca. 252 Ma) from the initiation of the Siberian Traps; earliest Triassic felsic and mafic magmatism in the Taimyr region; Permian–Triassic to Triassic rift-related magmatism in the Kara Sea; and granites and syenites (ca. 249–241 Ma) and basalts from southern and central Taimyr. The εHf values from Carboniferous to Triassic magmatic rocks along the northwestern part of the Siberian Craton are substantially more juvenile (Malitch et al., 2010) than similar-aged zircon within the Sverdrup Basin. The T_DM of the Siberian zircon are younger than the T_DM from this study, which suggests that the Siberian region was not a source region for the Sverdrup Basin during the Triassic. The εHf data from basement rocks of the NSI are within the spectrum of εHf and T_DM values of Arctic Alaska and Devonian strata from northern Canada, and these data interpreted to represent juvenile magmatic addition during the Neoproterozoic (Akinin et al., 2015).

The new observations of volcanic ash beds throughout the Triassic section on Axel Heiberg Island are noteworthy because they are a record of ash air fall on the northern side of the basin. To date there have been no observations of volcanic ash beds in Triassic strata from the southern Sverdrup Basin, implying that ash transport was confined to the northern part of the basin and not carried a great distance; otherwise they would be widely distributed throughout the basin. A stratigraphic section of Triassic strata from ∼200 km south of Bunde Fiord had no observable volcanic ash beds (Embry, 1983b); therefore, Triassic zircon being sourced from the north of the basin supports the sediment provenance direction suggested by Embry (2009), rather than the near-syndepositional zircon originating from a great distance east of the basin (e.g., Urals, Taimyr; Omma et al., 2011; Miller et al., 2013). If the active margin outboard of the YTT along the western margin of Laurentia was generating the Permian–Triassic zircon observed within Triassic strata, then the same tectonic process could generate similar-aged zircon along the outboard margin of the AACM. This is consistent within the framework of the rotational model, with the Devonian clastic wedge (e.g., Lemieux et al., 2011; Anfinson et al., 2012b) derived from an Arctic Alaska–Chukotka arc such as at Seward Peninsula (Amato et al., 2009); it is logical that a similar arc in the Triassic (Fig. 11) would incorporate a similar evolved εHf signature and T_DM.

The Lower Jurassic King Christian Formation sample has a similar cumulative trend to sediment that would be sourced from erosion of the Devonian clastic wedge. There is a peak at 420 Ma and a broad range of Proterozoic ages that are similar to the detrital-zircon signature of the Bjorne Formation (Fig. 9A). The difference between the King Christian Formation from the southerly-sourced Bjorne Formation is the presence of three young zircon grains (315, 279, and 262 Ma), which could have been cannibalized from Triassic strata. Zircon assemblages in the Middle Jurassic to Early Cretaceous (Sandy Point, Deer Bay, and Isachsen formations) are consistent with the provenance signature of the Devonian clastic wedge with minor recycled Permian–Triassic zircon (Røhr et al., 2010; Omma et al., 2011; Fig. 9A).

Previous authors (Embry, 1993; Embry and Beauchamp, 2008; Embry, 2009) proposed that Jurassic extension dismembered the northern sediment source region and prevented communication with the Sverdrup Basin by trapping sediment in extensional basins such as the proto–Amerasia Basin. The new detrital-zircon data from the Heiberg Group suggest that the provenance change occurred below the King Christian Formation and above the Romulus Member, which is equivalent to the Skybattle Formation (Fig. 3; ca. 210–190 Ma). After deposition of the King Christian Formation in the Early Jurassic, there was no longer a supply of near-syndepositional zircon into the basin. The Early Jurassic onset of rifting is supported by the presence of upper Heiberg Group strata at the base of half-grabens on Prince Patrick Island (Harrison and Brent, 2005). Deposition of the Isachsen Formation occurred after the breakup unconformity at ca. 130 Ma (Embry and Beauchamp, 2008). Detrital-zircon ages from the Lower Cretaceous Isachsen sandstone exhibit only a single grain (ca. 244 Ma) younger than 400 Ma (Røhr et al., 2010).

The abundance of Permian–Triassic detrital zircon and confirmation of Triassic volcanic ash beds in the Sverdrup Basin suggests that the region to the north was tectonically active during the Triassic, and likely part of the Permian as well. We interpret that the combination of volcanic ash beds and identification of remarkably similar dual detrital-zircon provenance patterns from the Sverdrup Basin, YTT, and AACM allows for an interpretation of the outboard margin of the AACM as convergent, as shown in Figure 11 (adapted from Plafker and Berg, 1994; Nokleberg et al., 2000), thereby providing a geodynamic model for rifting in the Jurassic concurrent with closure of the South Anyui ocean basin. The Nokleberg et al. (2000) reconstruction is modified by indicating that rifting of the proto–Amerasia Basin developed as a retroarc system driven by a subducting slab, which in turn caused AACM to rift from northern Laurentia (as proposed by Kuzmichev, 2009). During slab rollback, retroarc extension can occur far inboard from the convergent plate boundary (Lawton and McMillan, 1999; Hadlari and Rainbird, 2011), which is how we interpret the Jurassic to Early Cretaceous rifting of the Sverdrup Basin. Our detrital-zircon data support the hypothesis of Embry (2009) that crustal extension dismembered the northern sediment source from the Sverdrup Basin and show that this transition occurred in the Early Jurassic (Fig. 11), in contrast to Arctic Alaska, which continued to receive sediment from Chukotka in the Jurassic and into the Albian–Aptian (Moore et al., 2015).

The lack of near-syndepositional zircon in Jurassic–Cretaceous strata is important to understanding the nature of the basin because during the synrift phase (ca. 190–130 Ma), there is a limited record of magmatism other than the single 158 Ma zircon grain within the Deer Bay Formation. The breakup unconformity (130 Ma; see Embry and Beauchamp, 2008) coincides with the onset of extensive magmatism in the Sverdrup Basin (e.g., Evenchick et al., 2015). Postrift-phase magmatism persisted for more than 30 m.y.

The South China Sea provides an analog for the Sverdrup and Amerasia basins. The South China Sea formed in a retroarc setting and had limited synrift magmatism and persistent magmatism in the postrift phase (Franke, 2013). Slab roll back is thought to have been the driving mechanism in South China Sea causing backarc extension (Zhou and Li, 2000; Doust and Sumner, 2007). Rifting began at 110 Ma (Li et al., 2008), and generation of oceanic crust likely began ca. 40–30 Ma (Franke, 2013, and references therein), which indicates a protracted period of extension with no associated magmatic activity followed by extensive postrift magmatism, similar to the Cretaceous record of the Sverdrup Basin and possibly also Amerasia Basin.

The Sverdrup Basin has two identifiable detrital-zircon provenance signatures during the Triassic—an active margin source to the north and a recycled source to the south and east of the basin. These two signatures are similar to the two distinct sources observed in Arctic Alaska and the YTT. A hypothetical northwestern convergent margin of Laurentia would explain the remarkable similarity in the source age profiles of the Sverdrup Basin and the overlap assemblage between YTT and the North American craton. The presence of volcanic ash beds throughout Triassic strata, apparently restricted to the northern margin of the basin, indicates that volcanic activity that lasted more than ∼50 m.y was relatively proximal to this northern margin, which is consistent with a convergent outboard margin of the AACM during the Triassic. Previous interpretations attributed the source of the Permian–Triassic zircon to be in Russia, specifically the Siberian Traps, Uralian granites, and Taimyr magmatism, none of which make for an ideal candidate for the spectrum of ages, particularly for the Middle to Late Triassic zircon within sandstones of the Sverdrup Basin. The more negative εHf values of Permian–Triassic zircon compared to juvenile values of Permian–Triassic igneous rocks in Siberia support that the provenance was from the AACM instead of western Siberia. The contribution of U-Pb detrital-zircon geochronology, εHf isotope values, and newly described volcanic ash beds from Triassic strata of the Sverdrup Basin help to develop an argument that active magmatism north of the basin supplied Permian–Triassic zircon. Furthermore, the provenance signature from Jurassic strata suggests initial rifting isolated the Sverdrup Basin from the active margin source by the Early Jurassic.

This work was supported by the Geological Survey of Canada and the University of Ottawa. Fieldwork was assisted by the Polar Continental Shelf Program. Dr. Carol Evenchick provided a sample of the King Christian Formation, and the authors are grateful to Dr. Ashton Embry for guidance in the field in 2011. We are grateful for Tara Kell’s assistance with the XRD analysis. Ray Chung, Ron Christie, and Tom Pestaj are thanked for technical assistance preparing mineral separates and maintaining the SHRIMP ion probe at the Geochronology Laboratory in Ottawa. We thank Victoria Pease, Elizabeth L. Miller, Jeff Amato, and associate editor Damian Nance for their thoughtful comments that helped improve the manuscript. This is a Circum-Arctic Lithosphere Evolution (CALE) contribution. This is GSC Contribution #20160017.