The crustally derived Tungsten Plutonic Suite (TPS) in the northern Canadian Cordillera is responsible for several of the world’s most important tungsten (W) resources, but the actual source rocks to these exceptionally metallogenic magmas have never been identified. Detrital zircon studies have improved our knowledge of exposed supracrustal packages in the region, such that a U-Pb study of inherited zircon cores in the TPS provides an opportunity to determine the age and provenance of the melt source to these magmas. The TPS is dominated by Neoarchean to Paleoproterozoic inherited zircons that are typical of a northwest Laurentian “cratonic” detrital zircon signature (e.g., 2.8–1.8 Ga). Cross-correlation statistical analysis of detrital zircon populations combined with whole-rock neodymium isotopic compositions indicate that mid- to late Cambrian sedimentary rocks in the Selwyn Basin are the most likely melt source for the TPS magmas. Despite the presence of prospective intrusions across the Selwyn Basin, world-class W metallogeny is limited to the narrow belt of TPS plutons in the eastern Selwyn Basin near the ancient continental margin, possibly due to depositional or structural controls on the source strata. Today, the early Phanerozoic melt source to the TPS must be present at depths of at least 15–17 km and is notably younger than suggested by current geophysical interpretations for the middle crust, requiring a reconsideration of the crustal architecture in the northern Canadian Cordillera.

Tungsten (W) is a critical raw material due to its importance in the manufacturing of essential heat- and wear-resistant metals needed for future technologies (Burton, 2022). The northern Canadian Cordillera hosts two of the world’s most important W skarn deposits at Cantung (Northwest Territories) and Mactung (Yukon Territory) (Green et al., 2020) in association with the Tungsten Plutonic Suite (TPS). Previous work on the TPS and W mineralization demonstrated that melt and metal source(s) were extracted from the mantle during the Mesoarchean to Paleoproterozoic (Elongo et al., 2022). However, specific protoliths for the associated magmas could not be identified on the basis of isotopic composition alone, and the lack of W deposits associated with other compositionally similar plutons in the same paleogeographic province was not addressed. Furthermore, our understanding of the age and composition of potential source rocks in the basement is speculative and inferred mainly from geophysical data. The ongoing characterization of detrital zircon signatures for much of the exposed stratigraphy in the northern Canadian Cordillera, combined with our study of inherited zircon cores from crustally derived magmas of the TPS, has the dual purposes of identifying the source of melt to the W-associated plutons and testing geophysical interpretations of crustal architecture in the northern Canadian Cordillera.

The TPS comprises small crustally derived 98–95 Ma plutons of reduced biotite monzogranite and leucogranite associated with abundant W occurrences in the eastern Selwyn Basin (Figs. 1A and 1B; Gordey and Anderson, 1993; Hart et al., 2004; Rasmussen et al., 2011; Rasmussen, 2013), which was deposited on the northwestern margin of Laurentia in the early Phanerozoic. The TPS magmas are generally inferred to have been derived from a metasedimentary protolith despite their low peraluminosity and a paucity of highly aluminous minerals (e.g., cordierite, sillimanite, garnet; Hart et al., 2004; Rasmussen, 2013). This post-collisional magmatism was emplaced as a narrow belt (Fig. 1A) at the end stages of Mesozoic orogeny related to terrane accretion in the northern Canadian Cordillera (Nelson and Colpron, 2007).

Paleozoic and Neoproterozoic rocks in the upper 5–7 km of the crust in the Selwyn Basin have been well constrained by bedrock mapping (Gordey and Makepeace, 2003). In contrast, the middle crust is characterized by seismically slow and electrically resistive layered rocks that are thought to overlie attenuated crystalline lower crust (Fig. 1C; Clowes et al., 2005). The layered rocks are interpreted to be Meso- to Neoproterozoic siliciclastic and calcareous packages deposited during extensional and/or rifting events along the northwestern margin of Laurentia (Fig. 1C; Cook et al., 2004). Analogous Proterozoic supracrustal units occur north and east of the Selwyn Basin in several Proterozoic inliers and the Mackenzie Mountains, and in the sub-surface to the southeast (Fig. 1A). In the northern Canadian Cordillera, rifting and breakup of Rodinia in the late Neoproterozoic through to the middle Cambrian was followed by the onset of passive-margin sedimentation in the Selwyn Basin (Nelson and Colpron, 2007; Busch et al., 2021). The Selwyn Basin has undergone 30%–50% northeast-verging shortening since the Mesozoic (i.e., 150–350 km; Gordey and Anderson, 1993; Gordey, 2013), although there are no constraints on how far post-magmatic shortening may have transported the TPS plutons.

Inherited zircon cores commonly suffer from substantial resorption during anatexis that results in small, frequently altered crystals that may have been further compromised by complex igneous, metamorphic, and/or hydrothermal histories (Keay et al., 1999; Huang et al., 2020). This study employs a method modified from Reimink et al. (2016) to deal with discordance by identifying the most likely lower intercept age at which a common discordance event could have occurred (e.g., anatexis), then anchoring the lower intercept at that age to remove the effect of discordance on a recalculated upper intercept age (Appendix 1 of the Supplemental Material1). Another complicating factor is distinguishing inherited cores originating in the magma source rocks from xenocrystic cores that were entrained during ascent and emplacement. Detailed cathodoluminescence images were used to rigorously filter out potential xenocrystic cores based on several criteria, although the possibility of some xenocrystic cores being included in the data set cannot be entirely ruled out (Appendix 1). We report U-Pb laser ablation–inductively coupled plasma–mass spectrometry results for inherited zircon cores from ten TPS plutons, which yielded 248 dates; of these, 190 were accepted for consideration after processing and filtering (Appendix 1; Data Set S1).

Concordant analyses for inherited zircons have two dominant populations at 2800–2500 Ma and 2100–1750 Ma as well as a number of individual dates at ca. 3000 Ma, 2400 Ma, 2150 Ma, 1550–1400 Ma, and 1100–1000 Ma (Fig. 2). Age spectra for the processed and filtered discordant data essentially mirror the concordant results but do outline one subsidiary population from 2400 to 2150 Ma that would have been lost if a conventional discordance filter had been used (Fig. 2B). Most of the discordant analyses, therefore, retain useful information for the characterization of magma source rocks. The bimodal Neoarchean and Paleoproterozoic populations that largely characterize the TPS are typical of a northwest Laurentian or “cratonic” signature (e.g., Furlanetto et al., 2016). Previous work has demonstrated that the Mesoproterozoic packages and some of the Ediacaran to early Phanerozoic strata also have detrital zircon distributions indicative of a similar “cratonic” source, although in the younger rocks, this signature is likely due to recycling of the older strata (e.g., Hadlari et al., 2012, 2015; McMechan et al., 2017). In contrast, a Mesoproterozoic-dominated or “Grenvillian” detrital zircon signature is exhibited by the Neoproterozoic Mackenzie Mountains Supergroup, Cryogenian units in the Windermere Supergroup thought to have been derived by recycling of the former, and in much of the younger Phanerozoic strata via local recycling of those older rocks (e.g., Hadlari et al., 2012, 2015; Lane and Gehrels, 2014; Rainbird et al., 2017).

To identify possible TPS melt source rocks, the inherited zircon core dates were compared to a compilation of detrital zircon data for Paleoproterozoic to Ordovician strata exposed in the Proterozoic inliers, the Richardson Mountains, the Mackenzie Mountains, and the Coal River and Liard areas (Fig. 1A). The cross-correlation coefficients (R2) for probability density plots (PDPs) and kernel density estimates (KDEs) were selected as the best statistical tools to determine the probability that two or more samples were derived from the same parent population of zircon (Appendix 1 [see footnote 1]; Saylor and Sundell, 2016). While the PDP and KDE tests identified the same units as being very likely to have the same parental source of zircon (Fig. 3), due to the sample size limits specified for using the KDE metric (Saylor and Sundell, 2016), only the PDP results are discussed here. PDP R2 values are very high between the TPS and the Muskwa assemblage (0.78), the Wernecke Supergroup (0.83), unit “Ps” (Pigage, 2009) (0.74), the Ediacaran to lower Cambrian group (0.79), the mid- to late Cambrian “cratonic” sub-group (0.86), and the Ordovician “cratonic” sub-group (0.74; Fig. 3). These results are compatible with one or more of these stratigraphic groups being the source of inherited zircons in the TPS.

Whole-rock neodymium isotopic compositions (εNd[100]) for most of the Mesoproterozoic and Ediacaran to lower Cambrian units are typically more evolved than those of the TPS, demonstrating that these units cannot be the sole melt source to the magmas (Fig. 4). Neodymium isotopic compositions for the Pinguicula Group, however, do overlap with the range exhibited by the TPS (Fig. 4), but this unit did not have a particularly strong PDP R2 with the TPS (0.64; Fig. 3). Furthermore, the TPS contains sub-populations of inherited cores that are younger than the youngest detrital zircons found in all of the Mesoproterozoic strata (Fig. 3). While εNd[100] data for the mid- to late Cambrian units do span the entire range exhibited by the TPS, the Ordovician and younger strata tend to be more radiogenic (Fig. 4). The trend toward more radiogenic εNd[100] in the early Phanerozoic is likely due to an influx of detritus eroded from relatively isotopically juvenile igneous rocks (e.g., Proterozoic to Cambrian volcanic rocks, dikes, and plutons in the region; Pigage, 2009; McMechan et al., 2017). Together, the available U-Pb and εNd[100] data sets suggest that mid- to late Cambrian rocks exhibiting a “cratonic” detrital zircon signature are the likeliest melt source to the TPS magmas. The strata composing the “cratonic” sub-group are not contiguous across the Selwyn Basin, so these particular units together cannot compose the actual melt source. However, the source rocks to the TPS have εNd[100] and a detrital zircon signature that are the most similar to those of the mid- to late Cambrian “cratonic” sub-group and are, therefore, most likely composed of sediment recycled from older rocks in the region that may have mixed with juvenile mafic or otherwise zircon-poor detritus. The actual mid- to late Cambrian protolith(s) for the TPS magmas may have been strongly influenced by detritus from the Wernecke Supergroup, which also has a very strong correlation coefficient with the TPS data set (0.83; Fig. 3) but lacks the younger zircons and εNd[100] necessary to be the direct source of the magmas.

The current depth of the Moho below the TPS belt (35–40 km; Audet et al., 2019) combined with the emplacement of the TPS 8–10 km below the paleosurface (e.g., Gordey and Anderson, 1993) imply that the regional crust was as much as 50 km thick in the Cretaceous (Fig. 1C). Given the lack of magmatic muscovite in most TPS intrusive phases, anatexis was likely dominated by biotite-dehydration melting requiring relatively high temperatures (>850 °C) and mid-crustal depths (>25–30 km; e.g., Patiño Douce et al., 1990; Sawyer et al., 2011). This places the source rocks to the TPS magmas at depths of at least 15–17 km today, well within the mid-crust and near a proposed regional décollement that broadly aligns with the upper boundary of the interpreted Mesoproterozoic layered rocks (Fig. 1C; Cook et al., 2004; Hayward, 2019). This structural and/or stratigraphic transition occurs at a consistent depth of 25 ± 5 km across the Selwyn Basin; therefore, regardless of how far northeast the TPS intrusions have been transported from their basement source rocks by post-magmatic shortening, our data suggest that mid- to late Cambrian strata are present near this transition in the middle crust. While our depth estimate appears to constrain anatexis to below the décollement structure (Fig. 1C), given uncertainties with geophysical modeling and pluton depths of anatexis and emplacement, the magma source rocks could also be above the inferred structure. Regardless, the possibility of early Phanerozoic rocks at these depths has not yet been considered for this region.

Magma source rocks are a critical factor in W metallogeny (e.g., Romer and Kroner, 2015). In the northern Canadian Cordillera, the TPS magmas generated two globally important W deposits, and the likeliest source rocks to these magmas are mid- to late Cambrian strata. These rocks comprise detritus from Neoarchean to Paleoproterozoic basement that was originally eroded from the northwestern Laurentian craton and deposited in large basins along the continental margin in the Mesoproterozoic. This material may have been reworked, redistributed, and mixed with younger detritus more than once prior to—and during—the breakup of Rodinia. At the onset of the passive-margin sedimentation that characterizes the Selwyn Basin, these sediments were then recycled once more in the mid- to late Cambrian and likely mixed with detritus derived from more isotopically juvenile mafic igneous rocks. The emplacement of the TPS magmas as a narrow belt in the Selwyn Basin near the ancestral margin of northwestern Laurentia may reflect a depositional or structural control on the basinward extent of mid- to late Cambrian rocks in the middle crust. This could explain the lack of world-class W mineralization in other areas of the Selwyn Basin, despite the presence of compositionally similar granitic magmatism. In order to attain conditions that allowed biotite-dehydration-driven anatexis, the mid- to late Cambrian strata must have resided at depths of at least 25 km in the Cretaceous (Fig. 1C). Our new evidence for early Phanerozoic rocks in the middle crust of the northern Canadian Cordillera contrasts with earlier geophysical models that invoked considerably older crustal components at similar depths (e.g., Cook et al., 2004; Clowes et al., 2005).

1Supplemental Material. Appendix S1 (methods); Data Set S1 (sample summary, analytical conditions, new U-Pb data, compiled U-Pb data, new Sm-Nd data, compiled Sm-Nd data); Data Set S2 (results of cross-correlation statistical analysis). Please visit https://doi.org/10.1130/GEOL.S.22661677 to access the supplemental material, and contact editing@geosociety.org with any questions.

The authors are grateful for A. Gebru’s contribution of “MAC” samples from the Mactung area. Discussions with L. Beranek, J. Saylor, and D. Moynihan greatly assisted with data processing and interpretation, and feedback received from reviewers T. Hadlari and M. McMechan substantially benefited the manuscript. Funding was provided by Natural Resources Canada (Targeted Geoscience Initiative), the Northwest Territories Geological Survey, and the Polar Continental Shelf Program.

Gold Open Access: This paper is published under the terms of the CC-BY license.