In northern Alaska, the Early Cretaceous sedimentary Yukon-Koyukuk basin documents the progressive unroofing of the adjacent Brooks Range orogen. Igneous clasts in the lower conglomerate are believed to originate from ophiolitic rocks of the two uppermost allochthons in the Brooks Range, the Brooks Range ophiolite and the Angayucham terrane. The emplacement of these oceanic terranes onto the continental margin of the Arctic Alaska terrane documents the initiation of Brookian orogenesis. While most agree that the Angayucham terrane represents a widespread distribution of Late Devonian oceanic crust and Triassic-Early Jurassic oceanic plateau(s)/island(s), the age and origin of the Brooks Range ophiolite remains controversial. We present new age, whole-rock chemistry, and isotopic data from igneous clasts as well as a few Angayucham terrane outcrop samples from the NE Yukon-Koyukuk basin. Our results show that the igneous clasts are mostly subduction-related and more likely to represent eroded material from the Brooks Range ophiolite rather than the Angayucham terrane. Our Late Triassic, and Early and Middle Jurassic zircon crystallization ages for the igneous clasts, combined with their immobile trace element compositions documenting various stages of oceanic subduction (mature arc and later rifting), suggest a long-lived subduction system that was active in the Late Triassic and throughout the Middle Jurassic. Radiogenic lead and neodymium isotopic results yield juvenile signatures for both the igneous clasts and the Angayucham terrane, pointing to their formation in an intraoceanic setting distal from the continental rocks and sediments of the Arctic Alaska terrane. These new data, combined with the published data of others, allow us to propose a revised tectonic model that integrates Late Triassic island arc formation with the tectonic development and emplacement of the Brooks Range ophiolite.

The emplacement of the suprasubduction zone Brooks Range ophiolite (BRO) onto the Arctic Alaska continental margin marks the beginning of Brookian folding and thrusting in northern Alaska [13]. This orogenic event is widely believed to have occurred in Late Jurassic to Early Cretaceous time in response to southward subduction (present-day coordinates) of the passive margin of the Arctic Alaska terrane beneath an unknown island arc terrane [46]. The style of emplacement associated with the BRO resulted in its classification as a Tethyan-type ophiolite [2, 3, 7, 8] representing either a forearc (e.g., [1, 3, 9]) or a juvenile island arc [10, 11]. The former interpretation is favored in most tectonic models, and the BRO is typically inferred to represent the obducted forearc of the Late Jurassic-Early Cretaceous Koyukuk arc terrane, located south of the Brooks Range hinterland in the sedimentary Yukon-Koyukuk basin (Figure 1).

The BRO is exposed across six klippe-like massifs in the western Brooks Range, but is considered to have been widespread across the orogen. Extensive erosion and uplift of the Brooks Range stripped away large parts of the ophiolite complex and the structurally underlying Angayucham terrane, a Mesozoic oceanic plateau which is thought to have been emplaced onto the continental margin of the Arctic Alaska terrane with the BRO [9, 12, 13]. Middle-early Late Jurassic U-Pb zircon ages from late-stage intrusive rocks provide a minimum age for the BRO (170 Ma, U-Pb zircon; [14]; 164-161 U-Pb zircon; [15]) and 40Ar/39Ar isochron ages of primary hornblende within gabbro (inferred time of crystallization) suggest that the BRO is as old as Early Jurassic (187-184 Ma; [8]). Conglomerate in the north-easternmost Yukon-Koyukuk basin, which has been eroded from the southern Brooks Range, carries basic igneous clasts with Early Jurassic ages (198 Ma, 194 Ma, and 181 Ma, U-Pb zircon; [11]) and detrital zircon from the host sediment yields Middle Triassic to Late Jurassic ages (U-Pb detrital zircon ages of 240-160 Ma with a peak at c. 200 Ma; [11]). The younger ages in the basin sediments match the published ages of the BRO, and it has been suggested that the detrital zircon and igneous clasts in the north-easternmost Yukon-Koyukuk basin have been sourced from, and represent the life span of, the BRO [11].

Mineral compositions (e.g., Cr-spinel and olivine) from ultramafic mantle rocks to lower crustal rocks in the BRO [1, 3] and the composition of detrital Cr-spinel from the northernmost Yukon-Koyukuk basin [11] indicate a suprasubduction zone (SSZ) origin, but with overlapping forearc, backarc, and midocean ridge signatures. A limited number of middle to uppercrustal samples from the BRO also have whole-rock trace element patterns indicating a SSZ origin [1]. The overlapping geochemical signatures and proposed long time-span indicate that BRO genesis is probably more complex than can be explained solely by a forearc setting. Consequently, new geochemical and geochronological data are needed to characterize the specific tectonic setting(s) associated with BRO genesis, which is of great importance for understanding the tectonics and continental growth of northern Alaska during the Mesozoic.

We present new zircon U-Pb ages and whole-rock chemistry, including lead and neodymium isotopes, for igneous clasts from the northeastern Yukon-Koyukuk basin conglomerate. A few samples from outcrop of the Angayucham terrane are also included for whole-rock and isotope geochemistry, allowing for a direct comparison to the igneous clasts. Our new U-Pb zircon crystallization ages of Late Triassic and Early and Middle Jurassic provide time constraints for the subduction-related geochemistry. These data, combined with published data of others, allow us to propose a revised tectonic model that integrates Late Triassic island arc formation with the tectonic development and emplacement of the Brooks Range ophiolite.

2.1. Brooks Range Ophiolite (BRO)

The BRO is the structurally highest allochthon (the Misheguk Mountain allochthon) of the Brooks Range and overlies the Angayucham terrane, which is an imbricate stack of Late Devonian, Triassic, and Early Jurassic mafic oceanic crustal rocks represented by the Copter Peak allochthon [16]. The contact between the BRO and the underlying Angayucham terrane is represented by an amphibolite-facies metamorphic sole consisting of mostly Angayucham-like metabasalts and sediments, indicating that the former was thrust over the latter [17]. The BRO klippe-like massifs together make up a complete SSZ ophiolite sequence that, from bottom to top, includes a metamorphic sole, a residual mantle peridotite section (dunitic and harzburgitic), transitional ultramafic cumulates, layered gabbro, massive gabbro, intermediate intrusive rocks, sheeted dikes, basaltic lavas and marine sediments, and ultramafic/mafic to felsic late-stage intrusive rocks with boninite dikes cross-cutting the main ophiolite body [1, 2, 15]. The late-stage silicic intrusive rocks have Middle Jurassic (170 Ma; U-Pb zircon; [14]) to early Late Jurassic (c. 164-161 Ma; U-Pb zircon ages; [15]) ages, similar to the age of the metamorphic sole (c. 169-163 Ma; 40Ar/39Ar biotite and hornblende ages; [2, 8]). Thus, as thrusting of the BRO over the Angayucham terrane likely occurred in the late Middle to early Late Jurassic, the late-stage silicic intrusive rocks are likely to represent emplacement-related melts [3]. It has been suggested that this thrusting event relates to the obduction of the BRO onto the Arctic Alaska continental margin [1] or to thrusting within an intraoceanic setting prior to obduction onto Arctic Alaska terrane [8, 9]. The late-stage silicic intrusive rocks of middle to early Late Jurassic represent a minimum age for the ophiolite [1, 15]. Ages of the BRO obtained from midcrustal massive gabbro and plagiogranite (187-184 Ma, 40Ar/39Ar hornblende ages) indicate that the ophiolite could be as old as Early Jurassic [8].

2.2. Angayucham Terrane

The Angayucham terrane consists of structurally imbricated or dismembered stack of diabase, pillow basalt, basaltic tuff, and radiolarian chert [3, 9, 12, 13]. These oceanic upper crustal rocks are faulted onto Paleozoic metasediments of the Brooks Range and dip steeply to the south, towards, and beneath the thick sedimentary fill of the Yukon-Koyukuk basin [9]. The Angayucham terrane is locally separated into three units based on interbedded radiolaria and conodont ages [13]. These units include a Devonian to Mississippian basalt (sparse data), a Middle to Late Triassic basalt, and an Early Jurassic basalt. Whole-rock geochemistry from the late 1980s indicates that the Angayucham basalts include subalkaline to alkaline Triassic basalts with E-MORB trace element compositions and Jurassic basalts from an ocean plateau/ocean island setting [12, 13]. The Angayucham basalts have been locally affected by low-grade metamorphism (prehnite-pumpellyite to lower greenschist facies). The intraoceanic geochemistry, low-grade of metamorphism, and lack of intrusive rocks (e.g., gabbro and peridotite) suggests that the Angayucham terrane represents an ocean plateau/oceanic island setting that was, due to its thickened oceanic crust, emplaced by shallow obduction [12, 13].

2.3. Brookian Orogenesis

The southern Brooks Range represents the hinterland of the Brookian fold and thrust belt and is typically divided into east-west trending belts of varying lithology and metamorphic grade. From north to south, this sequence includes the Central Belt, the Schist Belt, and the Phyllite Belt. The Central Belt is considered the geographic core of the Brooks Range and consists of inhomogeneous and ductilely deformed Neoproterozoic and Paleozoic metapelite, marble, and Devonian orthogneiss [6, 18]. The rocks of the Central Belt have been metamorphosed at lower to moderate greenschist facies, but high pressure-low temperature (HP/LT) blueschist facies metamorphism of Early Cretaceous age (120 Ma; 40Ar/39Ar white mica; [19]) is recognized in the western Brooks Range. A few mica 40Ar/39Ar plateau and isochron ages from the lower-grade metamorphic rocks in the Central Belt suggest the main greenschist event occurred in the Late Aptian-Early Albian (116.5-111-Ma; [18]). Further south, the Schist Belt is characterized by similar ages and lithologies as the Central Belt, but these rocks underwent an earlier HP/LT (blueschist facies) metamorphism believed to be of Jurassic age [6, 18, 20, 21]. Greenschist facies related to the exhumation of the Brookian metamorphic core partly overprints the HP/LT rocks of the Schist Belt and has been dated to Late Aptian (114 Ma; U-Pb zircon; [22]). Structurally, overlying the Schist Belt to the south is the Phyllite Belt which consists of Early Devonian to Triassic metapelite, metabasite, and tectonic mélange metamorphosed at low-grade pumpellyite-actinolite facies. Its contact with the Schist Belt is a south dipping normal fault [21]. 40Ar/39Ar white mica ages from this boundary yield a Late Aptian-Early Albian age of 113 Ma, which has been interpreted as the age of normal faulting in the Brooks Range [23].

2.4. Northern Yukon-Koyukuk Basin

The Yukon-Koyukuk basin in west-central Alaska is a widespread Early Cretaceous sedimentary depression with >8000 m thickness of clastic sandstone and conglomerate [24]. The basin is divided into northeastern and southwestern parts by the centrally located Koyukuk arc terrane (Figure 1). The northeastern part of the basin, the focus of this study, is wedged between the Brooks Range to the north, the Ruby terrane to the southeast, and the Koyukuk arc terrane to the southwest [18, 25, 26] (Figure 1).

Paleocurrent data [2729] and provenance investigations [11, 24] in the study area indicate that Early Cretaceous deposition of sediment (conglomerate and sandstone) along the basin margin were broadly derived and transported via local drainage systems from the north and document the progressive unroofing of the Brooks Range (Figure 2). The basin stratigraphy is commonly separated into three units: Kvg, Kmc, and Kqc [26] (Figure 2). The lowest exposed unit is Kvg and it comprises matrix-supported conglomeratic greywacke and finely laminated mudstone [26]. The conglomerate clasts are dominantly comprised of mafic to intermediate intrusive and extrusive igneous material and chert. Kvg is widespread across the entire northeast part of the basin (Figure 2) and believed to represent gravity flow deposits in a deep-water submarine fan [24]. A gabbroic clast from Kvg conglomerate yielded an Early Jurassic age (194 Ma, U-Pb zircon; [11]). The maximum age of deposition is Albian (c. 107 Ma, U-Pb detrital zircon; [11]). Kvg is overlain by Kmc (Figure 2), a poorly sorted and poorly stratified conglomeratic sand and mudstone unit. Conglomerate clasts are predominantly mafic intrusive and extrusive igneous material and chert and subordinate felsic intrusive material [26]. Fossils of Early Cretaceous marine mollusks in Kmc document a shallow marine, marginal shelf depositional environment [2426]. Gabbroic clasts from the Kmc conglomerate yielded Early Jurassic ages (c. 198 Ma and 181 Ma, U-Pb zircon; [11]).

Kmc grades upward into the third and highest unit, Kqc [26]. The Kqc consists of conglomerate, sandstone, and shale deposited in the Cretaceous [26]. The Kqc conglomerate is well-sorted and consists almost entirely of rounded metamorphic clasts. Kqc shale contains late Early Cretaceous to Late Cretaceous plant fossils which documents the transition to a nonmarine environment [24, 25].

2.5. Koyukuk Arc Terrane

The Latest Jurassic to Early Cretaceous igneous Koyukuk arc terrane crops out in the center of the Yukon-Koyukuk basin (Figure 1). The Koyukuk arc terrane is divided into four magmatic units by Box and Patton [4]: The lowest and oldest (unit 1) is regarded as the base of the arc and consists of nonsubduction-related oceanic tholeiitic basalt and mafic to ultramafic intrusive rocks [4]. The tholeiites of unit 1 are intruded by Late Jurassic (147 Ma, U-Pb zircon; [15]) subduction-related intrusive rocks of unit 2. The intrusive rocks of unit 2 are calc-alkaline and range from intermediate to felsic in composition. Unit 2 is unconformably overlain by Early Cretaceous (137-125 Ma, K-Ar hornblende and biotite; [4]) subduction-related volcanic and volcaniclastic rocks of unit 3. Unit 3 includes both tholeiitic and calc-alkaline volcanic rocks that range from mafic to intermediate compositions. Unit 3 is conformably overlain by the even younger unit 4, subduction-related mafic to intermediate shoshonitic volcanic and volcaniclastic rocks of Early Cretaceous (123-118 Ma, K-Ar hornblende and biotite; [4]) age.

3.1. Sampling

Samples were collected along the Alatna River in the northeastern Yukon-Koyukuk basin (Figure 2). Igneous clasts from the Kvg lower conglomerate (Figures 3(a)–3(c)) and metavolcanics from the Angayucham terrane (Figure 3(d)) were collected for geochemical and isotopic analyses, as well as for U-Pb geochronology. The Kvg samples include 20 igneous clasts collected from four sites (23, 24, 25, and 26; Figure 2). The Angayucham samples were collected from one site (22; Figure 2) and include a mafic intrusive sample (VP16-22a) and two pillow lavas (VP16-22b, -22c). Sampling was restricted to the least weathered material and to clasts of sufficient size, generally c. 7-15 cm across. We also present whole-rock geochemistry for the Kmc gabbro clast (BRNF15-58B1) dated by O’Brien et al. [11] to c. 180 Ma and which was collected along the North Fork of the Koyukuk River (Figure 2).

3.2. Petrography

The Kvg igneous clasts consist of medium grained mafic and intermediate and felsic intrusive rocks, as well as minor mafic to intermediate porphyritic volcanic rocks. Mafic intrusive clasts dominate. Primary mineralogy of the mafic samples includes plagioclase, clinopyroxene, ± hornblende, and oxide(s). More evolved samples have a primary mineralogy represented by plagioclase±alkali feldspar, ± hornblende, quartz, biotite, and oxide(s). One mafic volcanic sample has clinopyroxene phenocrysts, and the intermediate volcanic samples generally have plagioclase and/or alkali-feldspar phenocrysts. All samples show variable hydrothermal alteration to greenschist facies as indicated by a secondary mineral assemblage of actinolite, chlorite, sericite, epidote, and calcite. Common reaction textures include uralitization of pyroxene, chloritization of biotite, seritization, and/or saussuritization of feldspar (Figures 4(a) and 4(b)). Hornblende is occasionally altered to chlorite and/or actinolite. The intermediate volcanic samples and a felsic intrusive sample have mafic enclaves with mineralogy distinct from the groundmass indicating that magma mixing and/or wall-rock assimilation was likely a contributing process to magma differentiation (Figure 4(a)). The fine-grained mafic intrusive samples typically display intergranular texture indicating fast cooling/shallow intrusion (Figure 4(c)). Mafic and intermediate intrusive samples with primary hornblende indicate crystallization of hydrous magmas [30] (Figure 4(d)). The Kmc gabbro clast (BRNF15-58B1) is melanocratic in hand specimen, and its thin section shows a cumulate texture of layered plagioclase crystals.

The Angayucham samples (VP16-22a, -22b, and -22c) consists of a medium grained, mafic intrusive rock and aphanitic pillow lavas. Their primary mineralogy includes plagioclase, clinopyroxene, and iron oxide(s). Secondary mineralogy consists of calcite, sericite, chlorite, and epidote, indicating greenschist facies metamorphism. The mafic intrusive sample has a subophitic texture of plagioclase laths partly enclosed by fractured clinopyroxene indicating a crystallization sequence of plagioclase before pyroxene, which is typical of dry magmas [30] (Figure 4(e)). The pillow lava samples have flow-oriented amygdales filled with secondary calcite and chlorite (Figure 4(f)).

4.1. Whole-Rock Geochemistry

Loss on ignition (LOI) and whole-rock geochemical analyses were obtained for 24 samples including one gabbro clast (sample BRNF15-58B1) from O’Brien et al. [11]. Weathered surfaces were removed from the samples prior to making powders for geochemical analyses. Whole-rock powders were made using an agate vibratory disc mill. X-ray fluorescence (XRF) and laser ablation inductively plasma mass spectrometry (LA-ICPMS) analyses were performed using instruments at the PetroTectonics analytical facility, Stockholm University, Sweden. Major element analyses were performed using a Rigaku ZSX Primus II XRF on homogenized-fused glass disks. The instrument was calibrated using 24 international standards. Matrix matched USGS reference materials AGV-2 and BCR-2 were run every 10th sample to confirm analytical performance. Information regarding accuracy, precision, and detection limits is reported in supplementary data table (S1).

Trace element analyses were performed on the XRF disks using a New Wave 193 nm ArF excimer laser coupled to a Thermo XSeries II quadrupole ICPMS. NIST 612 glass was used as a reference material for instrument calibration and Si as an internal standard. BCR-2 and SARM-1 were used as secondary reference materials to confirm instrument performance. The trace element data were reduced using Iolite 4 and the software’s trace element data reduction scheme (DRS). Accuracy, precision, and detection limits are reported in supplementary data table (S1).

4.2. U-Pb Geochronology

All samples were processed for heavy minerals. No baddeleyite was found but five samples yielded zircon: five Kvg clasts from the Alatna River. Zircon separation included crushing, milling, Wilfley table, and removal of magnetic minerals with a hand magnet. The zircon yield was cleaned in analytical-grade ethanol and rinsed with Milli-Q ultrapure water before being mounted in epoxy resin. Mounting, polishing, carbon coating (gold coating for the sample analyzed by secondary ionization mass spectrometry, SIMS), and cathodoluminescence imaging (CL) using a Hitachi S-4300 scanning electron microscope were done at the Swedish Museum of Natural History, Stockholm. U-Pb analyses were performed using LA-ICPMS and SIMS. Cores of magmatic zircon were targeted to determine the crystallization age(s) of the dated samples.

U-Pb geochronology was performed by LA-ICPMS using a New Wave 193 nm ArF excimer laser coupled to a Thermo XSeries II quadrupole at the PetroTectonics analytical facility, Stockholm University. Samples analyzed by LA-ICPMS include VP16-23e, VP16-25a, VP16-25d, and VP16-26f. A spot size of 20 μm was used for all analyses (samples and reference materials). The Plešovice natural zircon [31] was used as reference material for instrument calibration, and the FC-5z [32] was used as secondary reference material for quality assessment. Data reduction was performed using the U-Pb geochronology data reduction scheme in Iolite 4 which corrects for downhole fractionation and instrumental drift, including the uncertainty propagation routine, according to Paton et al. [33]. Common lead was monitored by measuring 202Hg and assuming a 202Hg/204Hg natural ratio of 4.35. Due to low common lead concentrations in the samples, no correction was needed. Analyses with (>10%) discordance were excluded from the final analysis. For accuracy and precision, see supplementary data table (S2).

In addition, a single sample (VP16-23g) was analyzed by SIMS due to having zircon grains too small for the LA-ICPMS. SIMS analyses were performed on a CAMECA IMS 1280 at the NordSIMS facility at the Swedish Museum of Natural History. The analytical procedure follows Whitehouse et al. [34] and Whitehouse and Kamber [35]. Sputtering with a rastered O2-, primary beam from an Oregon Physics Hyperion H201 RF+plasma source created slightly elliptical craters of c. 15 μm in diameter.

The Pb/U and UO2/U calibration scheme and measurements of the 91500 natural zircon reference material [36] were used to calculate the actual U/Pb ratios. For details and analytical statistics related to the calibration scheme, see Jeon and Whitehouse [37]. Data reduction was performed using a NordSIMS in-house developed software. Measured ratios were corrected for common lead according to the model of Stacey and Kramers [38]. Analyses with discordance (>2%) were excluded from the final analysis.

4.3. Neodymium Isotopes

Neodymium isotope ratios of a subset of 10 samples were analyzed with a Thermo Scientific TRITON thermal ionization mass spectrometer (TIMS) at the Laboratory of Isotope Geology, Swedish Museum of Natural History. Selection of samples chosen for analysis was restricted to the least altered samples based on thin section mineralogy and LOI contents.

Whole-rock powders of c. 100-200 mg from each sample was kept in Teflon capsules and spiked with a mixed 149Sm/150Nd tracer. Each spiked sample was dissolved in a 10 : 1 mixture of concentrated HF and concentrated HNO3 in autoclaves at 205°C for 5 days. Residues were eliminated with further acid (HF and HNO3) treatment and subsequent centrifugation. A two-step ion exchange procedure was performed to extract samarium and neodymium. The initial ion exchange was performed in columns charged with AG50W×8 hydrogen form resin and REE eluted with 6N HNO3. Samarium and neodymium were further separated from the remaining REE during a second ion exchange in Ln-spec columns [39].

Samarium concentrations were determined in multicollector static mode on double rhenium filaments, and neodymium was run in static mode on double rhenium filaments using rotating gain compensation. Concentrations and ratios were reduced assuming exponential fractionation. Samarium and neodymium ratios were corrected for mass fractionation by the normalizing factors 147Sm/148Sm=1.33386 and 146Nd/144Nd=0.7219, respectively. The external precision for 143Nd/144Nd was judged from values for La Jolla standard at 18 ppm. Accuracy correction was applied since replicate analyses of the La Jolla Nd reference material yielded a mean 143Nd/144Nd value of 0.511867±09 (n=12) (reference 143Nd/144Nd value=0.511858, [40]). Consequently, measured 143Nd/144Nd ratios were brought to accordance with the reference material by subtraction of 0.3 epsilon units. The USGS reference material BCR-2 was used for quality control, and measured values were judged against the preferred values of Jochum et al. [41]. The column blank was 0.15 ng.

4.4. Lead Isotopes

Lead isotopic compositions of whole-rock powders from a subset of 14 samples were analyzed with a VG Sector 54 IT thermal ionization mass spectrometer (TIMS) at the Danish Center for Isotope Geology, University of Copenhagen. Selection of samples chosen for analysis was restricted to the least altered samples based on thin section mineralogy and LOI contents. All steps of sample preparation were carried out in Class 1000 and 10,000 clean labs. Whole-rock powders of 200 mg or 600 mg, depending on the Pb bulk concentration in the sample, were gently leached in 2 N HCl for 15 minutes prior to dissolution. Leachates were pipetted off and attacked with aqua regia (1 : 3 mixture of concentrated HNO3:HCl) in Savillex™ vials on a c. 125°C hotplate for 24 hours. Residues were attacked with aqua regia and concentrated HF in Savillex™ vials and dissolved on a c. 125°C hotplate for 48 hours. Lead was subsequently extracted from the leachates and residues following a conventional HCl-HBr elution method using chromatographic columns charged with Bio-Rad AG1×8 anion exchange resin (100-200 mesh). The samples were loaded with phosphoric acid and silica gel onto outgassed Re-filaments and measured in static multicollection mode at a preferred temperature of c. 1150°C. Total procedural blank for Pb was <200 pg, which is considered to have an insignificant influence on the measured ratios. Isotope fractionation was monitored by repeated measurements on the NBS 981 reference material and correction was made after the values of Todt et al. [42]. The USGS reference material BCR-2 was used for quality control, and measured values were judged against the preferred values of Jochum et al. [41].

5.1. Geochemistry and Sample Classification

The major, minor, and trace element concentrations of the 24 samples are reported in supplementary data table (S1). The samples are altered to variable degrees as seen in thin section and as indicated by LOI values which range between 2-8 wt%, with about half of <4 wt%. While high-water contents may be associated with arc environments (see [43]), samples with higher LOIs presented here are also associated with greater secondary alteration as visible in thin section. It is well known that some major elements (e.g., the alkalis) and some trace elements (e.g., large ion lithophile elements) can be mobilized by weathering and alteration (hydrothermal and metamorphic) [4450], and due to the variable degrees of alteration shown by the samples in this study, we avoid using these mobile elements. Instead, we use the more robust classification diagram of Winchester and Floyd [44] that is based on an immobile element ratio (Nb/Y) as a proxy for alkalinity (total alkalis) (Figure 5). Immobile trace element abundances are shown on multielement and rare earth element (REE) diagrams normalized to normal midocean-ridge basalt (NMORB) values of Sun and McDonough [51] (Figure 6).

5.1.1. Kvg Clasts

(1) Major Elements. Silica contents for the conglomerate clasts vary between 45-70 wt%, but samples with 49-53 wt% SiO2 dominate. Alumina contents are typically between 15-20 wt% and show no covariation with SiO2. Magnesia, Fe2O3, and CaO contents decrease with increasing SiO2 and vary between 1-10 wt%, 3-13 wt%, and 3-11 wt%, respectively. Na2O shows a positive trend with increasing SiO2 and varies between 2-6 wt%. K2O is generally low (0.3-3.0 wt%), scattered, and shows no covariation with SiO2 (see supplementary data file S3). Both TiO2 and MnO define negative trends with increasing SiO2 and vary between 0.3-2.0 wt% and 0.05-0.25 wt%, respectively. P2O5 detection limits are generally <0.1 wt% but interday repeatability varies by 0.1 wt%; many P2O5 analyses are below this and therefore not considered further. Bivariate diagrams are available in the supplementary material (S2).

(2) Trace Elements. The conglomerate clasts are subalkaline and range from basic to silicic in composition, with the exception of one sample which plots in the alkaline part of the diagram, (Figure 5). All Kvg samples have negative Nb-anomalies as displayed by the depletion in Nb relative to Th and La (Nb/ThN=0.10.4; Nb/LaN=0.10.7) (Figure 6). All of the silicic samples and most of the intermediate samples show negative Ti-anomalies (Ti/TiN=0.30.4 and 0.5-0.9, respectively), whereas the basic samples lack pronounced Ti-anomalies (Ti/TiN=0.91.2). The basic samples show weak to moderately enriched light (L) REE (La/YbN=25 and La/SmN=23) and flat heavy (H) REE patterns (Dy/YbN=1.01.1) with no pronounced Eu-anomalies (Eu/EuN=0.91.2). The intermediate samples show weak to moderately enriched LREE (La/YbN=27 and La/SmN=23) and flat HREE patterns (Dy/YbN=0.91.2) with no pronounced Eu-anomalies (Eu/EuN=0.81.1). The silicic samples show moderately to strongly enriched LREE (subalkaline La/YbN=49 and La/SmN=34; alkaline La/YbN=21 and La/SmN=9) with flat HREE patterns (Dy/YbN=0.91.0) and lacking pronounced Eu-anomalies (Eu/EuN=0.91.1).

5.1.2. Kmc Clast

The major element composition of the Kmc clast (BRNF15-58B) is given in supplementary data table (S1).

(1) Trace Elements. The Kmc clast is subalkaline and basic in composition (Figure 5). The sample has a negative Nb-anomaly (Nb/ThN=0.2; Nb/LaN=0.1) and a positive Ti-anomaly (Ti/TiN=2), displayed on the NMORB normalized multielement diagram (Figure 6). The Kmc clast is weakly enriched in LREE (La/YbN=2 and La/SmN=1) and has fairly flat HREE pattern (Dy/YbN=1.4) with a positive Eu-anomaly (Eu/EuN=1.7).

5.1.3. Angayucham Terrane

(1) Major Elements. The Angayucham terrane samples include VP16-22a, VP16-22b, and VP16-22c. Silica and alumina contents vary between 46-49 wt% and 15-16 wt%, respectively. The alkali content varies between 0.04-0.06 wt% for K2O and 3-4 wt% for Na2O. Concentrations of MgO, Fe2O3, and CaO vary between 7-9, 10-13, and 10-12 wt%, respectively. Titanium oxide varies between 0.7-2.5 wt%, and MnO concentrations are typically about 0.2 wt%. P2O5 is at or below detection and is not discussed further. Given the small number of samples, there are no correlations on bivariate diagrams (see supplementary data file S3).

(2) Trace Elements. The samples are basic in composition (Figure 5). Two of the samples (VP16-22a, -22c) are subalkaline and the other sample (VP16-22b) plot in the alkaline part of the diagram (Figure 4). The Angayucham samples lack negative Nb-anomalies (Nb/ThN=0.70.9; Nb/LaN=12; Figure 6). These samples are superimposed on the field from Pallister et al. [13] for the inferred Triassic and Jurassic units of the Angayucham terrane. Two of the samples (VP16-22b, -22c) have enriched LREE (La/YbN=2.13.2 and La/SmN=1.21.3) and slightly depleted HREE (Dy/YbN=1.21.4) (Figure 6). The third sample (VP16-22a) has depleted LREE (La/YbN=0.4 and La/SmN=0.5) and flat HREE (Dy/YbN=1.0) (Figure 6). The LREE-enriched samples show weakly positive Ti-anomalies (Ti/TiN=1.51.6), whereas the LREE-depleted sample lacks a distinct Ti-anomaly (Ti/TiN=0.9). None of the samples has a distinct Eu-anomaly (Eu/EuN=1.11.2).

5.2. U-Pb Geochronology

LA-ICPMS and SIMS analyses are provided in supplementary data tables (S3). Representative CL images of each sample and their respective 206Pb/238U ages are provided in supplementary data table (S4). All the zircons display typical magmatic characteristics such as oscillatory and sector zoning (e.g., [52]) and Th/U ratios ≥0.1 (metamorphic zircon commonly have Th/U<0.1; [53]). Therefore, the concordia age calculated for each sample, plotted on conventional concordia diagrams with 2 s absolute uncertainty limits (Figure 7), are interpreted to represent the crystallization age of each sample.

The Kvg clasts yield Late Triassic and Early and Middle Jurassic ages. The silicic intrusive clast VP16-23g yields a Late Triassic concordia age of 231±2 Ma (Figure 7(a)). A single grain is somewhat rounded, but this silicic clast contains an aphanitic mafic enclave that may explain this minor resorption; in addition, this zircon is large in size (>100 μm), oscillatory zoned, and the combined single-age population argues for crystallization from the silicic magma. One basic intrusive clast (VP16-23e) yields a Late Triassic concordia age of 214±2 Ma (Figure 7(b)), while another (VP16-26f) yields an Early Jurassic concordia age of 196±2 Ma (Figure 7(c)). The silicic intrusive clasts VP16-25a and VP16-25d yield Early and Middle Jurassic concordia ages of 192±2 Ma (Figure 7(d)) 164±2 (Figure 7(e)), respectively.

5.3. Neodymium Isotopes

The neodymium analytical results from 10 samples are provided in supplementary data table (S5) and show that all samples have enriched 143Nd/144Nd and positive epsilon neodymium (ɛNd) values. Initial ratios are calculated using the U-Pb zircon crystallization ages for the dated samples, and an assumed initial age of 200 Ma for the undated samples (these “young” ages do not permit significant radiogenic in-growth). The Kvg clasts define juvenile initial neodymium values with ɛNdi ranging between +9.4 and +7.4. For the Angayucham terrane, the LREE-depleted intrusive sample (VP16-22a) has the highest ɛNdi (+7.6), whereas the LREE-enriched volcanic sample (VP16-22c) has a lower ɛNdi (+5.8).

5.4. Lead Isotopes

The results from 14 lead analyses are provided in supplementary data table (S5). The clast samples (Kvg+Kmc) have initial lead compositions that vary between 17.84 and 18.97 for 206Pb/204Pb, 15.37 and 15.55 and 36.84 and 38.08 for 207Pb/204Pb and 208Pb/204Pb, respectively. The Angayucham samples have initial lead compositions of 19.04 and 19.42 for 206Pb/204Pb, 15.54 and 15.56 for 207Pb/204Pb, and 38.55 and 39.21 for 208Pb/204Pb.

6.1. Ages and Geochemistry

Conglomerate clasts from the northeastern Yukon-Koyukuk Basin, together with samples from the Angayucham terrane, have been analyzed for whole-rock geochemistry including radiogenic isotopes and U-Pb zircon geochronology. Geochemical “types” can be recognized, despite the limited number of samples. Our results indicate that Yukon-Koyukuk basin clasts can have different ages despite having similar geochemical signatures, and we are unable to fully integrate all samples into an evolutionary framework. Nonetheless, linking geochemical signatures when U-Pb zircon ages are known provides some constraints on arc evolution through time (Table 1). Note that the Kmc clast (BRNF15-58B1) is not used in the chemical discrimination as it displays plagioclase accumulation in both thin section and chemistry (e.g., low MgO relative to high Al2O3 and CaO contents, depleted NMORB normalized REE pattern with a strong positive Eu-anomaly) and therefore does not represent a melt composition.

6.2. Magma Genesis

The mantle wedge of subduction zones is enriched in fluid-mobile incompatible elements (“subduction-mobile”, e.g., Th and LREEs) as the downgoing slab is dehydrated and at the same time is relatively depleted in the less-mobile incompatible elements (“subduction-immobile”, e.g., Nb, Ta, and HREEs) [30, 45]. The input of slab-related fluids to the mantle wedge results in metasomatism and oxidizing conditions that lead to early saturation and fractionation of Fe-Ti oxide, producing calc-alkaline magmas [54]. The decoupling of incompatible element behavior, combined with oxidizing mantle conditions, is specific to subduction zone settings and commonly used to identify subduction-related magmas [30, 5561].

The discrimination of subduction-unrelated and subduction-related samples can be approximated using Th/Yb as a proxy for the subduction component input and Nb/Yb as a proxy for mantle fertility (or to a lesser extent the degree of melting) [47]. Oceanic samples without the subduction component (e.g., MORB or OIB), define a MORB-array, while samples with a subduction component have increased Th/Yb and lie above the MORB-array. Because thorium is subduction-mobile and ytterbium is not, subduction-related magmas typically have enriched Th/Yb [45, 50]. The Nb/Yb proxy consists of two subduction-immobile elements, but niobium is slightly more incompatible than ytterbium, hence depleted sources (or high % melting) such as NMORB have low Nb/Yb, whereas enriched sources (or low % melting) such as EMORB or OIB have comparatively higher Nb/Yb [47, 50, 62]. In addition, subalkaline magmas generated in subduction zone settings can be recognized due to the increasing mobility of Th with increasing temperatures and slab depth [45, 63]. Thus, shallow tholeiitic magmas typically have lower Th/Yb values than deeper calc-alkaline and shoshonitic magmas. Although originally developed using basic rocks, combining immobile elements with similar partition coefficients reduces the effects of differentiation (except for very low degrees of partial melting), thus the Th/Yb and Nb/Yb proxies are relatively robust and can also be applied to more evolved rocks [48].

Our conglomerate clast samples have subduction-related chemistry as shown by their normalized multielement patterns in Figure 6 (negative Nb-anomalies, decoupled LREE, and HFSE; Kvg+Kmc) and lie above the MORB-array in Figure 8(a) (Kvg only). Samples with lower Th/Yb values represent tholeiites, whereas samples with higher Th/Yb values are regarded as calc-alkaline. The calc-alkaline clasts also have pronounced negative Ti-anomalies, consistent with Fe-Ti oxide fractionation and typical of calc-alkaline magmas [54].

In contrast, the Angayucham terrane samples clearly lack subduction-related geochemical signatures (lack negative Nb- and Ti-anomalies, no LREE and HFSE decoupling; Figure 6) and lie along the MORB-array, ranging from NMORB to more enriched values transitional between EMORB and OIB (Figure 8(a)).

The geochemical features of the Kvg clast samples are consistent with subduction-related magmas formed under different redox conditions and with varying input of the subduction component (Figure 8(a)). Consistent with previously published work [12, 13], the Angayucham terrane samples represent a nonsubduction-related oceanic setting (Figure 8(a)).

6.2.1. Magma Type(s)

We attempt to refine the subduction-related magmatic environment(s) of our samples (e.g., backarc and forearc) using vanadium and titanium following Shervais [64] and Pearce [48]. Titanium is incompatible, and due to the increased degree of partial (flux) melting during release of subduction-related fluids, it results in Ti-depletion of the mantle wedge-derived melts. Vanadium is also incompatible but becomes increasingly incompatible under oxidizing mantle conditions, and as subduction progresses, V-enrichment occurs due to the increasing oxidation of the wedge. It is possible to discriminate between (i) high V/Ti boninites (highly Ti-depleted and V-enriched magmas), such as magmas generated by high-degree melting of an already depleted source; (ii) intermediate V/Ti island arc tholeiites (IAT), including intra-arc tholeiitic magmas and the less depleted forearc and trench-proximal backarc magmas; (iii) low V/Ti MORB-type tholeiites with higher titanium and lower vanadium concentrations due to the minimal input of subduction-related fluids in their melt source (more reducing environment) and the relatively lower degree of melting, such as trench-distal backarc magmas or forearc basalts formed by spreading in the earliest stage of subduction initiation. Vanadium and titanium are less robust indicators for calc-alkaline and evolved rocks which have undergone magnetite or ilmenite fractionation which results in Ti-depletion and misclassification. Therefore, calc-alkaline samples and/or samples with negative Ti-anomalies (Ti/Ti<0.85 after [47]) are excluded from the V-Ti discrimination diagram (Figure 8(b)) and other Ti-based discrimination diagrams presented herein.

Our clast samples show some overlap with BRO crustal samples from earlier studies ([1] and references therein); the clast samples plot in the MORB- and transitional to the IAT-fields, whereas the BRO crustal samples show a wider range plotting in all three fields (MORB, IAT, and Boninite) (Figure 8(b)). Our clast samples are similar to backarc basalts from Izu-Bonin and the Mariana trough, but different from forearc basalts and boninites from the Izu-Bonin-Mariana system (Figure 8(b)). There is, however, a problematic overlap between backarc and forearc basalts in the MORB-field of Figure 8(b) making it difficult to distinguish between these two settings based solely on this diagram. It may be possible to distinguish between these two settings by evaluating the involvement of a deep subduction component in the magma genesis.

During shallow subduction, the elements released from the slab into the mantle wedge consists of low temperature, fluid-mobile large ion lithophile elements (LILE, e.g., Ba, Rb, Sr, and Pb); a deeper subduction signature includes elements, in addition to LILE, mobilized at higher temperatures (e.g., Th and LREE) [62, 63]. Pearce et al. [63] used the temperature dependent mobility differences of incompatible trace elements to define proxies for identifying input of shallow (Ba/Th), deep (Th/Nb), and total subduction (Ba/Nb) components in a magmatic system. As mentioned in Section 5.1, elements sensitive to weathering and alteration (such as LILE) are not used in this study. Thus, only the deep subduction component proxy is applied since it is based on immobile elements.

Forearc basalts are believed to form by spreading and decompression melting in the earliest (shallow) stage of subduction initiation [65, 66] and should theoretically lack deep subduction components; in contrast, backarc magmas form at greater distance from and above the slab and are commonly enriched in deep subduction components [62]. It should be noted that the degree of enrichment of subduction components in backarc magmas may vary depending on proximity to the trench (see [62]).

The tholeiitic clast samples and one recently published BRO gabbro [1] shown in Figure 8(b) are enriched in deep subduction components (high Th/Nb and La/Nb) and differ from the Izu-Bonin-Mariana forearc basalts which lack a distinct deep subduction component enrichment (low Th/Nb and La/Nb) (Figure 8(c)). This is consistent with Figure 8(a) in which our clast samples and the BRO gabbro sample from Biasi et al. [1] occur along the volcanic arc-array, and the Izu-Bonin-Mariana forearc basalts lie along the MORB-array—further highlighting the lack of a subduction component in forearc basalt magmas. We therefore suggest that our clast samples and the recently published BRO gabbro sample which plot in the MORB- and transitional to IAT-fields (Figure 8(a)) are not forearc basalt-type magmas. We therefore suggest that our clast samples and the recently published BRO gabbro [1] which plot in the MORB- and transitional to IAT-fields (Figure 8(a)) are not forearc basalt-type magmas. Instead, the transitional geochemistry of these clasts and the BRO gabbro is similar to magmas generated by extension in trench-proximal settings. Magmas produced in backarc settings typically include both MORB and IAT but characteristically fall between these two end-members [62]. Another alternative is that the magmas to our MORB-IAT transitional clast samples, possibly also to the BRO gabbro, was formed by extensional magmatism in an existing forearc setting. The latter is more consistent with the recently published study of the BRO by Biasi et al. [1] whom concluded that the BRO most likely formed in a forearc setting.

One intermediate clast sample (VP16-26f) has chemistry consistent with the boninite magma series (high in SiO2, MgO, and Cr and low in TiO2; [67, 68]). Because this sample shows weak Ti-depletion (Ti/Ti<0.85), the assumed boninitic composition is not evaluated using the Ti-V diagram. Instead, we apply the discrimination of Saccani [69] which is based on the absolute values of Dy and Yb normalized to chondrite (values of Sun and McDonough [51]) (Figure 8(d)). These elements were chosen by Saccani [69] to account for different degrees of partial melting and/or different degrees of mantle source depletion, which are the two processes believed to be responsible for generating boninite magmas [70, 71]. Since this diagram is used to discriminate between boninite and tholeiite magmas, calc-alkaline samples are not included. The intermediate clast (VP16-26f) plots in the boninite field together with one of the basic clast samples (VP16-24d) (Figure 8(d)). Only the intermediate clast is considered to reflect a boninitic composition because the basic clast does not fulfill the compositional criteria for boninite series classification; its Cr is too low and MgO and TiO2 are too high for its SiO2 content (see [67, 68]). Additional discrimination diagrams showing the boninitic composition of this intermediate clast are provided in the supplementary material (S6).

6.2.2. Depth of Melting

The flat nonfractionated normalized HREE patterns of the clast samples indicate no residual garnet in the melt source(s) and instead suggest that melting occurred in the shallow, spinel-bearing mantle. The Angayucham terrane samples, on the other hand, have HREE normalized patterns varying from nonfractionated to fractionated, suggesting either (i) two sources (a shallow spinel-bearing mantle source and a deep garnet-bearing mantle source), (ii) a changing source (deepening or shallowing), or (iii) repeated melting of a common source that is not continuously rejuvenated.

Pearce [47] suggested that basic rocks formed by melting deep mantle in equilibrium with residual garnet (e.g., ocean island basalts) will have higher TiO2/Yb values due to the compatibility of Yb in garnet and defined an OIB-array, whereas those formed from shallow melting with no residual garnet (e.g., midocean-ridge basalts) will have lower TiO2/Yb values and define a MORB-array. Plume-ridge interactions and oceanic plateaus typically plot along a diagonal trend that can go from OIB-EMORB and all the way to NMORB [47]. On the Nb/Yb versus TiO2/Yb discrimination diagram of Pearce [47], the clast samples plot in the NMORB part of the MORB-array consistent with their flat normalized HREE patterns and a shallow spinel-bearing mantle source (Figure 9). Two out of three Angayucham terrane samples (one is alkaline on the Nb/Yb versus Zr/Ti classification diagram in Figure 5) represent the tholeiitic part of the OIB-array, and one sample plots in the MORB-array (Figure 9). Together with superimposed Angayucham terrane data from Pallister et al. [13], the samples define a plume-ridge interaction trend going from shallow melting associated with the Triassic basalts and possibly this study’s LREE-depleted sample (VP16-22a) to deeper melting associated with the Jurassic basalts and this study’s LREE-enriched samples (VP16-22b, -22c) (Figure 9).

6.2.3. Mantle Source(s)

The samples are derived from a juvenile mantle source as defined by their radiogenic lead isotopic signature which follows the northern hemisphere reference line (Figures 10(a) and 10(b)) and their highly positive epsilon neodymium values (Figure 10(c)). Isotopically, the clast samples document a primary MORB-type mantle source and lack mixing trends with enriched mantle sources (e.g., -EMI or EMI) (Figures 10(a)–10(c)). The juvenile isotopic composition of the intermediate to silicic intrusive clasts is inferred to represent the isotopic composition of the mantle source from which the parental melt(s) was derived. Thus, the clasts represent magmas derived from a shallow depleted MORB-like mantle which have had little to no contamination by old sialic crust (or its sediment) in their genesis. This is consistent with their formation in an intraoceanic setting.

Similar to the clasts, the Angayucham samples lack mixing trends with EMI or EMII enriched mantle sources and define a mantle source that is intermediate between the MORB and OIB reference fields (Figures 10(a)–10(c)). The Angayucham volcanic sample (VP16-22c) has a more enriched radiogenic lead composition, a less juvenile εNd, and plots above the MORB array similar to an OIB source, indicating a more radiogenic and distinct mantle source from the clasts. The Angayucham intrusive sample (VP16-22a) has a less radiogenic lead composition and more juvenile εNd, similar to a MORB source and not so different from the clast samples (Figures 10(a)–10(c)). The variation in the Angayucham isotopic data is supported by the immobile trace element chemistry. The LREE-enriched volcanic sample has chemistry transitional between EMORB/OIB, consistent with its isotopically enriched OIB-like mantle source. The LREE-depleted intrusive sample has chemistry similar to NMORB, consistent with its isotopically less enriched and MORB-like mantle source.

6.2.4. Tectonic Setting(s)

The tholeiitic basic clast samples resemble backarc magmatism with their MORB-IAT transitional chemistry and most likely represent shallow melting in a trench proximal extensional setting. Whether this extensional setting was located in a backarc or forearc position is difficult to say, and more chemistry including additional age data is needed to investigate this. Despite the lack of age data, these samples have consistent and similar immobile trace element and isotopic compositions.

The intermediate volcanic clasts show physical signs of magma mixing/mingling, and the lack of ages plus contrasting isotope data makes it difficult to fingerprint the specific tectonic association of these lavas. The three intermediate intrusive clasts have chemistry and isotopic signatures different from one another. The Early Jurassic intermediate clast (VP16-26f) belongs to the boninite series implying high-degree melting of a residual mantle source (two-stage melting, see [68]). Another intermediate clast (VP16-24b) has MORB-IAT transitional chemistry similar to the basic tholeiitic clasts suggesting formation in a trench proximal extensional setting. The last intermediate intrusive clast (VP16-25f) has similar chemistry as the previous sample (VP16-24b) but differ from it by being highly depleted in radiogenic lead.

The silicic clasts have arc-related trace element chemistry (e.g., negative Nb- and Ti-anomalies, decoupled incompatible element behavior, and calc-alkaline character). The Late Triassic silicic clast (VP16-23g) is calc-alkaline and has isotopic lead compositions similar to the Late Triassic basic clast sample (VP16-23e), which is also calc-alkaline. We regard these two clast samples to represent Late Triassic island arc magmatism. The Early Jurassic silicic sample (VP16-25a; 192 Ma) has a lead isotopic composition nearly identical to the Early Jurassic boninitic sample (VP16-26f; 196 Ma), and they are close in age, thus the silicic sample might represent a differentiated member of the boninite magma (Figure 10). The Early Jurassic cumulate gabbro clast (BRNF15-58B1; 181 Ma; [11]), though not plotted on discrimination diagrams, has subduction-related chemistry (negative Nb-anomalies, decoupled LREE and HFSE; Figure 6) and has the most depleted radiogenic lead composition together with one of the intermediate clast samples (VP16-25f). The depleted mantle-like isotope values of these two samples suggest they originate from a mantle wedge that had been depleted by a previous melt extraction event.

The Middle Jurassic silicic clast (VP16-25d) differs from the other clasts in having a less pronounced subduction-signature (less negative Nb-anomaly), and its high Nb/Y puts it into the alkaline part of the classification diagram (Figure 5). Although this could be explained by very low-degrees of partial melting, the Middle Jurassic clast also differs isotopically from the others in having a more enriched radiogenic lead composition at a less evolved intial εNd value, making it an outlier in this data set.

6.3. Clast Correlation and the Brooks Range

Earlier provenance and paleocurrent studies in the northeastern Yukon-Koyukuk basin have shown that the sedimentary units document the progressive unroofing of the southern Brooks Range [11, 24, 2729]. The large clast size of c. 10 cm and the southerly directed paleocurrents suggest their derivation from a relatively proximal source to the north. The igneous clasts from the lower conglomerate presented here record Late Triassic, Early-Middle Jurassic intraoceanic subduction-related magmatism, consistent with the results of O’Brien et al. [11]. Mesozoic oceanic rocks are found in the structurally highest allochthons of the Brooks Range and include the Misheguk allochthon (BRO) with a suprasubduction zone origin [13] and the Copter Peak allochthon (Angayucham terrane) with an origin unrelated to subduction [12, 13]. The currently known exposures of the former (the BRO) are located c. 350-400 km away in the western Brooks Range, which seems too far away to be the source of our clasts given their large size. Exposures of the latter (Angayucham terrane) are widespread along the basin margin and close to our sampling area, but our whole-rock geochemistry and isotopic signatures show that the igneous clasts and the Angayucham terrane are not genetically related. Our Early Jurassic ages from one boninite and one possibly related silicic clast (196 and 192 Ma, U-Pb zircon), including clast ages reported in the O’Brien et al. [11] study (198 Ma, 194 Ma, and 181 Ma, U-Pb zircon), are similar to the published Early Jurassic ages of midcrustal gabbro in the BRO (187-184 Ma, 40Ar/39Ar hornblende ages; [8]). Our Middle Jurassic age (164 Ma, U-Pb zircon) of one silicic intrusive clast matches the age(s) of late-stage emplacement-related silicic intrusions in the BRO (164-161 Ma, U-Pb zircon; [15]). The similarity in ages and subduction-related chemistry together with paleocurrent data and provenance studies supports a possible BRO origin for the Early-Middle Jurassic Kvg and Kmc igneous clasts, whereas the provenance of the Late Triassic igneous clasts is more difficult to constrain.

Although there are to date no reported findings of Late Triassic island arc-related magmatism in the Brooks Range, the Middle-early Late Jurassic U-Pb zircon ages obtained from late-stage intrusions of the BRO represent a minimum age for the ophiolite. It is therefore possible that the Late Triassic igneous clasts represent erosional products from older parts of the BRO or an arc that predates the BRO that has not yet been recognized/dated. Other scenarios that may explain the lack of Late Triassic island arc-related rocks include: (1) the igneous clasts represent parts of an arc that predates the BRO that has been completely denuded and is only preserved in the sedimentary fill of the Yukon-Koyukuk basin, (2) the clasts represent parts of an arc that predate the BRO that was downdropped by late normal faulting of the southern Brooks Range and is now buried beneath the basin [10, 72], or (3) some combination of these. We prefer the last scenario (3) because at present, there is not enough data to dismiss any of these possibilities. Consequently, we propose a tectonic model in the next section that includes aspects of all of these scenarios to explain the absence of Late Triassic subduction-related rocks in the present Brooks Range. It must be stressed that there is as yet no known preserved source for these clasts.

6.4. Tectonic Model

By integrating our geochemical and geochronologic results with the existing data of others, we develop a tectonic model to explain the genesis of the NE Yukon-Koyukuk conglomerate clasts and how they might be related to the BRO genesis and emplacement (Figure 11). In the Late Triassic, our model suggests that an intraoceanic arc is active to the south of the Arctic Alaska continental margin, as indicated by the Late Triassic calc-alkaline magmatism of our intermediate and silicic intrusive clast samples (231 Ma and 214 Ma, plus U-Pb detrital zircon ages in [11]). The Late Triassic arc is situated on the upper oceanic plate above a south-dipping subduction zone (present-day coordinates) which is separated from the Arctic Alaska continental margin by the Angayucham ocean of unknown size (break in section, Figure 11).

In the Early Jurassic, slab rollback-induced extension and rifting in the forearc region of the Late Triassic island arc leads to shallow, NMORB-like, relatively low-degree melting of the metasomatized mantle wedge producing magmatism that is transitional between MORB-IAT, as indicated by our backarc-like clasts. Slab rollback may be due to progressively older and denser oceanic lithosphere getting subducted as the Arctic Alaska continental margin approaches the trench. Production of MORB-IAT magmatism in the initial stage of rifting depletes the mantle wedge. Further melting of the residual mantle wedge due to input of slab fluids leads to boninitic magmatism, as documented by our Early Jurassic clast (196 Ma), possibly related silicic clast (192 Ma), and by the existence of boninite dikes intruding the BRO [1, 15].

Southward subduction likely continues throughout the Middle Jurassic as indicated by subduction-related magmatism of the Kmc cumulate clast (181 Ma; plus 187-184 Ma 40Ar/39Ar hornblende ages from gabbro within the BRO; [8]; and U-Pb detrital zircon ages in [11]) and the Kvg silicic clast (164 Ma; plus 164-161 U-Pb zircon ages from late-stage intermediate and silicic intrusions within the BRO, [15]). The late-stage melting in the Middle Jurassic (Figure 11) is related to either alternative (1) emplacement of the BRO (with or without island arc crust) by thrusting over the Angayucham oceanic crust with plateaus/islands in an intraoceanic setting before obduction onto the Arctic Alaska continual margin, or alternative (2) obduction of BRO (with or without island arc crust) together with upper crustal parts of the Angayucham oceanic crust with plateaus/seamounts onto the Arctic Alaska continental margin. We prefer alternative (1) due to the juvenile radiogenic isotope compositions of the igneous clasts and the Angayucham terrane which indicate little or no involvement of old continental crust in their genesis.

The present-day position of the displaced parts of the collided oceanic complex (Angayucham terrane and BRO with or without the Late Triassic island arc; Figure 11) has been attributed to postcollisional normal faulting of the southern Brooks Range [10]. This postcollisional extensional event is proposed to have occurred in the Late Aptian [22, 23], resulted in exhumation of underlying HP/LT metamorphic rocks in the Brooks Range hinterland, and contributed to the formation of the Yukon-Koyukuk basin to the south [10, 22, 73].

This model satisfies our results of subduction-related magmatism with diverse ages and chemical signatures from diverse stages of oceanic subduction (mature arc and slab rollback-induced forearc extension). This is consistent with the BRO chemistry recently published by Biasi et al. [1], who also suggested slab-rollback as a possible genesis for the BRO but favored subduction initiation as the most likely mechanism for the genesis for the BRO. Boninites alone are not characteristic of subduction initiation as boninitic magmas can also be produced in several other settings (see [68]). Boninites formed during subduction initiation typically occur in a diagnostic sequence following forearc basalts, as has been documented in the Izu-Bonin-Mariana subduction zone [65]. Due to the lack of forearc basalt-chemistries among our samples and samples from the BRO, we favor slab rollback-induced magmatism over subduction initiation for the genesis of the BRO and our possibly related igneous clasts. More sampling of the BRO is needed to investigate this further.

The involvement of a Late Triassic island arc has also been proposed in other studies (e.g., [10, 11]). There are, however, uncertainties related to this model:

  • (1)

    The model presumes that the Yukon-Koyukuk basin Late Triassic igneous clasts have been sourced from the BRO. Although paleocurrent data and provenance studies show that the Yukon-Koyukuk lower conglomerate records the progressive erosion of the southern Brooks Range, there is still debate regarding whether these igneous clasts have been sourced from the BRO or if they have been sourced from other terranes in the Alaskan interior (e.g., [1])

  • (2)

    There are as yet no age data to confirm that our tholeiitic clasts represent Jurassic magmatism of the BRO, as proposed here. Instead the timing of this suggested extensional magmatism in a forearc setting is based on the documentation of boninite dikes cutting massive and layered gabbro in the BRO (the likely source of our tholeiitic clasts). If our tholeiitic clasts are truly representative of the BRO gabbro, it would mean that they are older than the Early Jurassic boninite clast (196 Ma)

  • (3)

    There are known examples of boninite dikes in the BRO, but as yet no published ages to correlate them with our boninite clast

6.5. Tectonic Implications for Long-Lived Arcs and Arc-Accretion in the Region

Other accreted Late Triassic and Early Jurassic island arcs in the North American Cordillera include the Talkeetna arc in south-central Alaska, and the Stikine and Quesnel terranes in the British Columbia. As in our study, these subduction-related rocks also have juvenile initial isotope compositions (e.g., highly positive εNdi and low initial Sr isotope ratios) indicating little or no involvement of continental crust [7476]. O’Brien et al. [11] suggested the remnants of Middle Triassic to early Late Jurassic island arc material preserved in the Yukon-Koyukuk basin sediments was a northward continuation of an oceanic arc system that included all the arc terranes mentioned above. They proposed that this regional oceanic arc system developed offshore of North America and was later emplaced onto the North America continental margin due to craton-ward subduction.

An oceanic plateau-island arc collision, similar to what has been proposed for the BRO and Angayucham terrane, has also been implied for the Quesnel terrane in the North American Cordillera. Lapierre et al. [77] suggested that the Cache Creek terrane, a Late Triassic oceanic plateau in the south-central British, collided with the Quesnel Arc before accretion onto the North American margin. The radiogenic lead and positive εNd values of the Angayucham samples are shown with the basalts of the Cache Creek terrane (Figure 10). The matching age and isotopic composition of the Angayucham terrane with the Cache Creek terrane is consistent with their derivation from a similar mantle source. Lapierre et al. [77] concluded that Late Triassic plume-related Cache Creek Terrane documented a Permian to Triassic event of major plume activity. It is possible that the Angayucham terrane also documents this event and might represent a part of the Cache Creek oceanic plateau; however, the timing of the Cache Creek collision with the Quesnel arc has been suggested to be Early Jurassic, which is earlier than the suggested Middle-Late Jurassic collision of the Angayucham terrane with the Jurassic BRO accompanied by our suggested Late Triassic island arc. The occurrence of other Late Triassic and Early Jurassic island arc terranes and oceanic plateaus in the North American Cordillera suggests that our Late Triassic island arc, the BRO, and Angayucham terrane were once part of a complex system involving oceanic plateaus/islands and intraoceanic arcs developing outboard of the North American paleomargin, similar to today’s southwestern Pacific margin.

Our new whole-rock major, trace, and isotopic geochemistry, combined with U-Pb zircon crystallization ages, for igneous conglomerate clasts from the northeastern Yukon-Koyukuk basin, and outcrop samples of the Angayucham terrane have the following implications.

  • (i)

    The igneous clasts document Late Triassic (231 Ma and 214 Ma), Early Jurassic (196 Ma-181 Ma), and late Middle Jurassic (164 Ma) subduction-related magmatism, while the Angayucham terrane samples are unrelated to subduction. The subduction-related clasts have chemistry that suggest they are more likely related to the Brooks Range ophiolite than the Angayucham terrane

  • (ii)

    The geochemistry and age data imply that the clasts may have been part of a long-lived subduction zone that was active in the Late Triassic and throughout the Middle Jurassic. The clasts record several magma types such as mature island arc calc-alkalic, extensional MORB-IAT transitional, and boninitic magmatism. All possibly generated at different times but associated with a single subduction zone that was affected by slab rollback

  • (iii)

    The Brooks Range ophiolite formed by slab rollback-induced rifting in the forearc region of a Late Triassic island arc before it was emplaced together with the Angayucham terrane onto the continental margin of the Arctic Alaska terrane

  • (iv)

    Radiogenic Nd and Pb isotope compositions of the igneous clasts and the Angayucham terrane indicate little or no involvement of old continental crust in their genesis; instead thrusting of the Brooks Range ophiolite over the Angayucham oceanic plateau likely occurred in an intraoceanic setting distal to any potential continental input

  • (v)

    The occurrence of other Late Triassic and Early Jurassic island arc terranes (the Stikine and Quesnel terranes) and oceanic plateaus (e.g., Cache Creek) in the North American Cordillera suggests that our Late Triassic island arc, the Brooks Range Ophiolite, and Angayucham terrane were once part of a complex system of intraoceanic arcs with oceanic plateau(s) formed on the distal side to the North America paleomargin, similar to today’s southwestern Pacific margin

Data tables are available in manuscript and in supplementary material S1-S6.

Preliminary results of this work have only been published as abstracts.

The authors declare that they have no conflicts of interest.

The authors of this study wish to thank Robert Frei for providing access and Toni Larsen and Toby Leeper for help and assistance in the isotope laboratory facilities at the Danish Center for Isotope Geology, University of Copenhagen. Staff of the NordSIMS facility, Martin Whitehouse, Heejin Jeon, and Kerstin Lindén are thanked for sample preparation and technical assistance. The NordSIMS facility is funded by Swedish Research Council infrastructure grant 2017-00671; this is NordSIMS contribution no. 722. Elizabeth Miller is thanked for sharing the Kmc clast (BRNF15-58B1) for whole-rock geochemical and isotope analyses. This work is funded by a grant to V. Pease by the Swedish Research Council. We thank Steve Box and Frances Deegan for their feedback on an earlier version of this paper.

Exclusive Licensee GeoScienceWorld. Distributed under a Creative Commons Attribution License (CC BY 4.0).