Composition and provenance of the Snowcap assemblage, basement to the Yukon-Tanana terrane, northern Cordillera: Implications for Cordilleran crustal growth Open Access

The North American Cordillera is regarded as a type example of an accretionary orogen, where growth occurred as the result of progressive addition of terranes to the western margin of North America (e.g., Coney et al., 1980; Monger and Nokleberg, 1996; Colpron and Nelson, 2009). Terranes were originally defined as fault-bound crustal blocks with geological records distinct from those of adjacent terranes (Coney et al., 1980; Jones et al., 1983). Differences in geological history, faunal assemblages, geochemistry (including isotopes), and paleomagnetism suggested early in terrane analysis that many of the allochthonous terranes in western North America were of uncertain paleogeographic origins and that the Cordilleran orogen represented a collage of disparate crustal fragments (Helwig, 1974; Coney et al., 1980). Ensuing studies of the North American Cordillera focused on detailed mapping of the internal framework and external relationships of the terranes, and fostered the development and application of new tools to resolve these relationships (e.g., Nd and Sr isotopes, detrital zircon geochronology; Samson and Patchett, 1991; Patchett and Gehrels, 1998; Gehrels et al., 1995; Gehrels and Ross, 1998). From these emerged a progressively better understanding of the geological history and geodynamic interpretation of many Cordilleran terranes, including genetic and/or evolutionary links between some terranes (Monger and Nokleberg, 1996; Colpron et al., 2007). It is now generally recognized that terranes of Late Jurassic and younger age originated at or near the western North American margin in the eastern Pacific Ocean. Of those terranes with Paleozoic (and older) underpinnings, the outer terranes of the Cordillera appear to have originated far from the western Laurentian margin (Paleozoic North America), either in the paleo-Arctic realm (e.g., Alexander, Farewell, Arctic Alaska, and others; Fig. 1; Colpron et al., 2007; Colpron and Nelson, 2009) or within the Panthalassic ocean, i.e., the Paleozoic World ocean (e.g., Wrangellia; Scotese, 2002; Monger and Nokleberg, 1996; Cache Creek; Monger and Ross, 1971). By contrast, terranes that occupy an inner position in the North American Cordillera (e.g., Yukon-Tanana, Quesnellia, Stikinia, and others) appear to have been generated and evolved (at least in part) along or near the western margin of Laurentia in Paleozoic time (Fig. 1). These peri-cratonic or peri-Laurentian terranes share some geological, isotopic, and provenance characteristics with source regions comparable to the western margin of Laurentia, but have a mid- to late Paleozoic evolution that differs from that of the passive continental margin (Gehrels et al., 1991; Mortensen, 1992; Creaser et al., 1997; Patchett and Gehrels, 1998). These terranes have been inferred to record the evolution of a series of superposed magmatic arcs that were developed on top of rifted continental fragments as the Slide Mountain backarc ocean opened and closed in mid- to late Paleozoic time (Nelson, 1993; Creaser et al., 1997; Nelson et al., 2006; Colpron et al., 2007). The peri-Laurentian terranes enclose oceanic rocks with Paleozoic faunal elements of Tethyan (Asian) affinity (e.g., Cache Creek terrane) that were incorporated during the early Mesozoic development of the Cordilleran orogen (Mihalynuk et al., 1994).

Recent studies of the peri-Laurentian terranes in the northern Cordillera have focused on the Yukon-Tanana terrane of Yukon and northern British Columbia. The Yukon-Tanana terrane is a stratigraphic succession of four tectonic assemblages (Colpron et al., 2006a). The basal siliciclastic assemblage, the Snowcap assemblage (Fig. 2), is overlain by as many as three unconformity-bounded volcanic and volcani-clastic successions of predominantly continental arc character (Piercey et al., 2006, and references therein). These are the Upper Devonian to lower Mississippian Finlayson assemblage, the mid-Mississippian to Lower Permian Klinkit assemblage, and the Middle to Upper Permian Klondike assemblage (Fig. 2). These arc assemblages are coeval with the oceanic assemblage of chert, argillite, and mafic volcanic rocks of the Slide Mountain terrane (Fig. 2), which forms a discontinuous belt along the eastern edge of Yukon-Tanana terrane.

The Snowcap assemblage has been interpreted as part of a rifted fragment of the western continental margin of Laurentia (Colpron et al., 2006a, 2006b, 2007; Nelson et al., 2006). It forms the metasedimentary basement to Upper Devonian–Carboniferous strata of the Yukon-Tanana terrane, and in the absence of crystalline basement exposures in the terrane, it provides a proxy for its lower crustal section. It is inferred that the lower crustal section of the Yukon-Tanana terrane (including its Precambrian crystalline basement) was been delaminated during Mesozoic orogenesis (e.g., Cook and Erdmer, 2005; Murphy et al., 2006). Numerous lines of evidence, mainly from the mid-Paleozoic rocks, point to a northwestern Laurentian source region for the Yukon-Tanana terrane (Gehrels et al., 1991; Creaser et al., 1997, 1999; Patchett and Gehrels, 1998; Nelson et al., 2006; Nelson and Gehrels, 2007); however, detailed documentation of its oldest unit, the Snowcap assemblage, is still lacking. Many of these previous studies were of reconnaissance in nature and lacked the benefits of the stratigraphic context that now exists for the Yukon-Tanana terrane.

In this paper we describe the geological relationships, geochemical and Nd-Hf isotopic compositions, and detrital zircon U-Pb geochronology of the Snowcap assemblage in its type region, the Glenlyon map area of central Yukon (Fig. 3; Colpron et al., 2002, 2006b). These data help refine interpretations of the original location and Paleozoic evolution of the basement to Yukon-Tanana terrane, and the role of crustal recycling in the development of orogenic belts.

The Snowcap assemblage is the oldest unit in the Yukon-Tanana terrane (Fig. 2; Colpron et al., 2006a). It consists mainly of a heterogeneous assemblage of psammitic schist, quartzite, dark gray carbonaceous schist, calc-silicate rocks and marble, and locally amphibolite, greenstone, and ultramafic rocks. These rocks are commonly intruded, and locally thoroughly injected by strongly foliated and lineated Late Devonian to early Mississippian tonalite, granodiorite, and granite bodies that represent subvolcanic intrusions to overlying arc rocks of the Finlayson and Klinkit assemblages (Fig. 2). Rocks of the Snowcap assemblage are typically polydeformed and metamorphosed to amphibolite facies and commonly occupy a low structural level in the Yukon-Tanana terrane.

The Snowcap assemblage (or complex) is named after exposures on the slopes above Little Salmon Lake in the Glenlyon area of central Yukon, including those on Snowcap Mountain, where it occurs within a structural dome (Figs. 3 and 4; Colpron, 2000; Colpron et al., 2002, 2006b). In the Glenlyon area, quartzite, psammitic, pelitic and calc-silicate schists, marble, and mafic rocks are intercalated at all scales, from decimeters to hundred of meters. Massive medium gray orthoquartzite forms the prominent exposures that cap Snowcap Mountain (Fig. 4). There the quartzite overlies a sequence of coarse-grained garnet-muscovite schist (Fig. 5A) with meter-thick intercalations of quartzite, calc-silicate schist, marble, and amphibolite. Elsewhere, quartzite most commonly occurs as decimeter- to meter-thick horizons in pelitic schist (Fig. 5B). Rocks of the Snowcap assemblage are variably calcareous and commonly contain centimeter- to meter-thick marble horizons (Fig. 5C). Marble locally defines good marker horizons (50–100 m thick) that can be followed for tens of kilometers (Fig. 4; Colpron, 2000; Colpron et al., 2002, 2006b). These occur with different lithologic associations and at different structural levels, suggesting that a number of carbonate intervals are present within the Snowcap assemblage, rather than a single repeated unit. At Little Salmon Lake, marble passes laterally into a polymictic pebble to boulder conglomerate (Colpron and Reinecke, 2000). South of Little Salmon Lake, quartzite pebble to cobble conglomerate occurs locally at a lower structural level within the Snowcap assemblage.

Amphibolite and greenstone occur as 1–10-m-thick horizons locally throughout the Snowcap assemblage in the Glenlyon area, but are more common near Little Salmon Lake. On Snowcap Mountain, coarse-grained garnet amphibolite forms a prominent horizon as much as 300 m thick and ~7.5 km long (Figs. 4 and 5D). Marble intercalations (from a few centimeters to 50 m) within the amphibolite suggest a volcanic protolith. Coeval deposition of carbonate and mafic igneous rocks is also indicated by the geochemistry of the marble, as described below. The garnet amphibolite on Snowcap Mountain occurs within a metasedimentary sequence dominated by coarse-grained garnet-muscovite schist (Fig. 5A) and quartzite.

In the Glenlyon area, the Snowcap assemblage is unconformably overlain by arkosic metawacke and conglomerate of the Drury formation and orthoquartzite of the Pelmac formation, both part of the regional Finlayson assemblage (Figs. 3 and 4; Colpron et al., 2006b). The Drury formation locally contains felsic metavolcanic rocks that were dated by the U-Pb method on zircons as ca. 350 Ma (Colpron et al., 2006b). Detrital zircons from the Drury formation yielded almost exclusively Late Devonian ages clustered at ca. 365 Ma and ca. 378 Ma (Colpron et al., 2006b). These relationships constrain the age of the Snowcap assemblage in central Yukon as pre–Late Devonian. In the southern Yukon-Tanana terrane, near the 60th parallel, lower Mississippian felsic meta volcanic rocks in the upper Dorsey complex (part of the Snowcap assemblage) suggest local transition into the overlying Finlayson assemblage (Roots et al., 2006).

Rocks of the Snowcap assemblage have a more complex deformational and metamorphic history than overlying Carboniferous strata. Psammitic and pelitic rocks of the Snowcap complex typically have garnet-grade metamorphic mineral assemblages (Fig. 5) and record multiple metamorphic and deformational events. Syntectonic to posttectonic garnet porphyroblasts are commonly partially to completely retrograde metamorphosed to chlorite. Near Little Salmon Lake, garnet appears to be more extensively retrograde metamorphosed near its upper contact with the Pelmac formation (Fig. 4). At deeper structural levels, such as on Snowcap Mountain, garnet tends to be coarser grained and only locally retrograde metamorphosed (Figs. 5A, 5D).

Rocks of the Yukon-Tanana terrane in the Glenlyon area are typically deformed by tight to isoclinal folds (F₂) that locally deform an earlier foliation (S₁) (Colpron et al., 2006b). The tight to isoclinal F₂ folds are commonly affected by younger open folds (F₃) that locally produce basin and dome interference patterns (Colpron et al., 2006b). The local occurrence of foliated quartzite clasts in basal conglomerate of the Mississippian Little Salmon formation (Klinkit assemblage; Colpron et al., 2005) and intrusion of deformed Snowcap assemblage rocks by undeformed granitoid rocks of the ca. 340 Ma Tatlmain batholith and ca. 357 Ma Ragged pluton (Fig. 3; Colpron et al., 2006b) indicate that at least part of the deformation in the Snowcap assemblage is pre-Mississippian in age. Geochronological constraints from the western Yukon-Tanana terrane document tectonometamorphic events ca. 365–350 Ma, ca. 260–240 Ma, and ca. 195–187 Ma (Berman et al., 2007).

A representative suite of rock types from the Snowcap assemblage of Glenlyon map area was sampled for major, trace, and rare earth element (REE) analysis (Table 1; Supplemental File 1¹). The samples were analyzed at Activation Laboratories in Ancaster, Ontario, Canada. Major elements were determined using fused bead X-ray fluorescence (XRF). Trace elements and REE were determined via an inductively coupled plasma–mass spectrometer (ICP-MS) using research grade analyses (4LITHORESEARCH package; www.actlabs.com). Samples for ICP-MS analysis were prepared using a lithium metaborate/tetraborate fusion and the fused sample was subsequently dissolved in acid and analyzed. During the course of the study numerous reference materials were run to test precision and accuracy. Reference material data are presented in Supplemental File 2².

Neodymium and hafnium isotopic data were collected on whole-rock powders at the Pacific Centre for Isotopic and Geochemical Research (PCIGR) at the University of British Columbia following the methods of Weis et al. (2005). Samples were dissolved in bombs prior to chemical separation and analysis. Neodymium isotopic ratios were measured on a Finnigan Triton thermal ionization mass spectrometer (TIMS) in static mode multicollection. Samples were fractionation corrected by normalization to ¹⁴⁶Nd/¹⁴⁴Nd = 0.7219, and during this study the La Jolla Nd standard yielded values of ¹⁴³Nd/¹⁴⁴Nd = 0.511854 ± 11 (2 standard deviation, SD; n = 8). Samarium and Nd concentrations were determined using ICP-MS as outlined above. Initial ¹⁴³Nd/¹⁴⁴Nd ratios and ϵNd were calculated at 360 Ma, the youngest age the samples could be, and in order to facilitate comparison with existing data within the Yukon-Tanana terrane (Creaser et al., 1997, 1999; Grant, 1997; Piercey et al., 2003, 2004; Ruks et al., 2006). The chondritic uniform reservoir (CHUR) values used for ϵNd calculations are ¹⁴³Nd/¹⁴⁴Nd = 0.512638 and ¹⁴⁷Sm/¹⁴⁴Nd = 0.1967 (Hamilton et al., 1983). Depleted mantle model ages (T_DM) are calculated using the values of ¹⁴³Nd/¹⁴⁴Nd = 0.513163 and ¹⁴⁷Sm/¹⁴⁴Nd = 0.2137 (Goldstein et al., 1984). Hafnium isotope ratios were determined on a Nu Instruments (Nu 201) multicollector ICP-MS (MC-ICP-MS) run in static mode. Samples were fractionation corrected by normalization to¹⁷⁹Hf/¹⁷⁷Hf = 0.7325. The JMC-475 Hf standard has yielded ¹⁷⁶Hf/¹⁷⁷Hf = 0.282152 ±17 (2SD) for the PCIGR facility (Weis et al., 2005). Lu and Hf concentrations were determined using ICP-MS as outlined above. Initial ¹⁷⁶Hf/¹⁷⁷Hf ratios and ϵHf were calculated at 360 Ma similar to the Nd isotopic data. The chondritic uniform reservoir (CHUR) values used for ϵHf calculations are ¹⁷⁶Hf/¹⁷⁷Hf = 0.282772 and ¹⁷⁶Lu/¹⁷⁷Hf = 0.0332 (Blichert-Toft and Albaréde, 1997). T_DM ages are calculated using the values of ¹⁷⁶Hf/¹⁷⁷Hf = 0.28325 and ¹⁷⁶Lu/¹⁷⁷Hf = 0.0334 (Vervoort and Blichert-Toft, 1999). The results of Nd and Hf isotopic analyses are presented in Table 2.

The mafic metaigneous rocks of the Snowcap assemblage have elevated TiO₂ and low Al₂O₃/TiO₂ (Table 1) relative to normal mid-ocean ridge basalts (N-MORB; TiO₂ = 1.36; Al₂O₃/TiO₂ ~11). The samples straddle the alkalic-subalkalic boundary (Nb/Y ~0.7) on the Zr/TiO₂ versus Nb/Y diagram (Fig. 6A) and have elevated Ti/V ratios similar to rocks from MORB or alkalic sources (Fig. 6B). Their high Zr/Yb, Nb/Yb, and Nb/Th ratios plot near enriched (E) MORB to ocean island basalts (OIB; Figs. 6C, 6D), indicative of derivation from incompatible element-enriched mantle source regions (Pearce and Peate, 1995). The mafic rocks have light (L) REE-enriched primitive mantle-normalized signatures with flat to positive Nb anomalies relative to Th and La, typical of rocks from E-MORB to OIB sources (Fig. 7). Tracer isotopic data for the mafic rocks include: ϵNd_t = +5.5 to +7.8 and ϵHf_t = +6.7 to +10.5 (Fig. 8; Table 2). Depleted mantle model ages for the rocks range from T_DM(Hf) = 0.88–1.27 Ga for the Hf depleted mantle model (model of Vervoort and Blichert-Toft, 1999). The mafic rocks yield a single Nd depleted mantle model age (model of Hamilton et al., 1983) T_DM(Nd) of 1.43 Ga (Fig. 8C; Table 2).

The mafic metaigneous rocks can be subdivided into two groups. Group 1 mafic rocks contain higher TiO₂, P₂O₅, high field strength element (HFSE: Zr, Nb, Th), and REE contents, and lower Al₂O₃/TiO₂, ϵHf_t, and ϵNd_t relative to Group 2 rocks. These variations are attributed to a greater incompatible-element–enriched mantle (lithospheric?) component in the genesis of Group 1 mafic rocks relative to those of Group 2.

The clastic metasedimentary rocks of the Snowcap assemblage have been divided for presentation purposes into quartzites, psammites and pelites, but their attributes are discussed both individually and collectively. The meta-clastic rocks have chemical index of alteration values [CIA = Al₂O₃/(Al₂O₃ + CaO* + K₂O + Na₂O), where all values are molar and CaO* is the CaO derived from the silicate fraction in the rock; Nesbitt and Young, 1984] that range from ~55 to 75 with SiO₂/Al₂O₃ ratios ranging from 3 to 36 with the higher SiO₂/Al₂O₃ values associated with quartzitic rocks (Fig. 9A). On molar Al-(Ca + Na + K)-(Fe-Mg) and Al-(Ca + Na)-K diagrams, the Snowcap meta-clastic rocks form linear arrays between the composition of typical upper crustal material (e.g., granodiorite in Fig. 9B and the feldspar tie line in Fig. 9C) and the more potassic clay minerals, suggesting variations in clay mineral abundances (Figs. 9B, 9C). With increasing SiO₂, the metaclastic rocks show a decrease in TiO₂, Al₂O₃, and K₂O (Supplemental File 3³), consistent with increasing quartz, and decreasing clay, abundances.

The trace element and isotopic attributes of the metaclastic rocks are presented in Figures 8, 10, 11, and 12. Figures 10A and 10B show elemental ratio plots of elements enriched in upper crustal materials (e.g., Th, Zr, and La) relative to elements enriched in mafic and mantle-derived materials (e.g., Sc). These ratios were chosen to test potential mantle versus upper crustal sources for metasedimentary rocks of the Snowcap assemblage (e.g., Taylor and McLennan, 1985; Creaser et al., 1997; McLennan et al., 2003, and references therein). The compositions of several potential source materials for metasedimentary rocks of the Snowcap assemblage are also plotted on the diagrams in Figure 10, including: (1) mafic rocks of the Snowcap assemblage (this study); (2) values for N-MORB (Sun and McDonough, 1989); (3) values for modern upper continental crust (McLennan, 2001); and (4) the average value of pre–365 Ma Cordilleran miogeo clinal sedimentary rocks (Boghossian et al., 1996; Garzione et al., 1997; Patchett and Gehrels, 1998).

The Snowcap metaclastic rocks have element ratios that cluster near those of the upper continental crust and rocks of the Cordilleran miogeocline with high Th, La, and Zr relative to the compatible element Sc (Fig. 10). For most samples, La/Sm_UCN (UCN = upper crust normalized) ratios of ~1 (Fig. 10C) and primitive mantle normalized trace element patterns are consistent with derivation from upper crustal sources (Fig. 11). Three quartzite samples (02DM144, 02GGA734–1, 02JN063) exhibit different trace element patterns with LREE depletions, elevated Zr and Hf, flat to positive Ce anomalies, and flat heavy (H) REE (Fig. 11C). The enrichment of HREE relative to the LREE and the positive Ce anomalies are also clearly illustrated when the samples are normalized to post-Archean Australian Shale (PAAS), a proxy for the upper crust (Fig. 11E). These features are consistent with the relative pure composition of the quartzite (~88%–94% SiO₂; Table 1), and REE patterns similar to that of zircon (Hoskin and Schaltegger, 2003; Whitehouse and Kamber, 2003) suggest minor zircon enrichment.

The clastic metasedimentary rocks have ϵNd_t = −4.3 to −22.9 (most have ϵNd_t <−14) and ϵHf_t = −0.7 to −40.1 (most have ϵHf_t <−15; Fig. 8; Table 2). Depleted mantle model ages for the rocks range from T_DM(Hf) = 1.15 to 3.23 Ga (Hf depleted mantle model of Vervoort and Blichert-Toft, 1999) and T_DM(Nd) = 1.98 to 3.37 Ga (Nd depleted mantle model of Hamilton et al., 1983; Fig. 8; Table 2). In addition, when ϵNd_t and ϵHf_t are plotted against Nb/Th and Th/Sc, the Snowcap samples show hyperbolic arrays between the mafic rocks and the more unradiogenic clastic rocks, with the LREE-depleted quartzite sample 02DM144 as the other end member (Fig. 12; Table 2).

The carbonate-bearing metasedimentary rocks are divided into calc-silicate rocks and marble. They have geochemical and isotopic signatures that are distinct from meta-clastic rocks of the Snowcap assemblage. The carbonate-bearing rocks are characterized by lower TiO₂, Al₂O₃, and K₂O at a given SiO₂ content and higher CaO and loss on ignition (LOI) relative to the clastic metasedimentary rocks (Table 1; Supplemental File 3 [see footnote 3]). The sample of marble has low absolute trace element concentrations, in particular the REEs, and exhibits a LREE-depleted trace element pattern with a negative Ce anomaly (Ce/Ce* <1; Fig. 13A; Table 1), and an elevated Y/Ho ratio (Figs. 13B, 13C). These trace element patterns are common to seawater-derived carbonates and surface seawater (Elderfield et al., 1988; Shimizu et al., 1994; Nozaki et al., 1997; Kamber and Webb, 2001). The calc-silicate rocks have similar trace element patterns but with less depleted LREE, higher total REE, less pronounced negative Ce anomalies, and lower Y/Ho ratios (Fig. 13). This is consistent with their mixed carbonate-siliciclastic composition. The Snowcap calc-silicate rocks have trace element compositions intermediate between the clastic metasedimentary rocks and marble (Fig. 13).

The carbonate-bearing metasedimentary rocks of the Snowcap assemblage have ϵNd_t = −1.8 to +0.7 and ϵHf_t = +2.0 to +13.4 (Fig. 8; Table 2). Depleted mantle model ages for these rocks range from T_DM(Hf) = 1.55 to 1.64 Ga (Hf depleted mantle model of Vervoort and Blichert-Toft, 1999) and T_DM(Nd) = 1.91 to 2.22 Ga (Nd depleted mantle model of Hamilton et al., 1983; Fig. 8; Table 2). These values are younger than all metaclastic rocks and overlap with isotopic values for mafic rocks of the Snowcap assemblage (Fig. 8; Table 2).

A quartzite sample from roadside exposures of the Snowcap complex along the north shore of Little Salmon Lake (Fig. 4; sample 04MC155, Supplemental File 1 [see footnote 1]) was collected for U-Pb analyses of detrital zircon. These analyses were completed at the University of Arizona following methods outlined in Gehrels et al. (2008). The U-Pb analytical results are reported in Table 3 and plotted on a concordia diagram in Figure 14. The uncertainties reported in Table 3 and Figure 14 include both random errors (mainly from measurement of ²⁰⁶Pb/²³⁸U, ²⁰⁶Pb/²⁰⁷Pb, and ²⁰⁶Pb/²⁰⁴Pb) and systematic errors (mainly from common Pb composition, calibration correction, decay constants, and age of the calibration standard). Of the 95 zircon grains analyzed, 92 yielded concordant or near-concordant results (<10% discordance; Fig. 14; Table 3). These results are plotted on a probability density diagram in Figure 15A. As all ages reported in Table 3 are older than 1.0 Ga, the ²⁰⁷Pb/²⁰⁶Pb ages are used in Figure 15A.

The majority of detrital zircons from the Snowcap assemblage yielded Paleo proterozoic and Archean ages with prominent peaks in probability density ca. 1870 Ma and ca. 2720 Ma, and secondary peaks ca. 2080 Ma and ca. 2380 Ma (Fig. 15A). Single concordant analyses at 1135 ± 53 Ma and 1359 ± 79 Ma (Fig. 14; Table 3) suggest a possible minor Grenville source and provide the best estimate of maximum age for this sample. The pattern of detrital zircon ages for the Snowcap assemblage in Glenlyon area resembles published data for Mississippian and older units of the southern Yukon-Tanana terrane in Yukon and British Columbia (Figs. 15B, 15C; Devine et al., 2006; Nelson and Gehrels, 2007) and the Coast Mountains of southeastern Alaska (Fig. 15D; Gehrels et al., 1991; Gehrels and Kapp, 1998). Collectively these patterns match well the distribution of Paleoproterozoic–Archean ages in the Laurentian miogeoclinal detrital zircon reference for northern British Columbia (Fig. 15F; Gehrels, 2000), including a distinct subpopulation in the 2.2–2.4 Ga range.

The overall character and composition of the Snowcap assemblage suggest mixed siliciclastic-carbonate deposition punctuated by localized mafic volcanism in a continental margin setting. Most clastic metasedimentary rocks of the Snowcap assemblage have trace element and radiogenic isotope signatures (La/Sm_UCN ~ 1; high Th/Sc, Th/Co, La/Sc, and La/Co; low Nb/Th; and negative ϵNd_t and ϵHf_t; Figs. 10 and 12; Table 2) that are consistent with derivation from upper crustal sources similar to sedimentary rocks of the Cordilleran miogeocline (Figs. 10 and 12). The evolved Nd-Hf isotopic signature and Paleoproterozoic–Archean depleted mantle model ages of most Snowcap metasedimentary rocks are consistent with, but not unique to, a western Laurentian origin for the basement to the Yukon-Tanana terrane. These results are consistent with previous reconnaissance studies of the Yukon-Tanana terrane (e.g., Creaser et al., 1997, 1999; Patchett and Gehrels, 1998). Xenocrystic zircon cores from Late Devonian–Mississippian plutonic and volcanic rocks of the Yukon-Tanana terrane also generally yield Paleoproterozoic–Archean average inheritance ages broadly consistent with a western Laurentian basement (e.g., Mortensen, 1992; Colpron et al., 2006b).

The detrital zircon signature of the Snowcap assemblage (and other Yukon-Tanana units) provides more specific ties to sources in northwestern Laurentia (Figs. 15 and 16). The Proterozoic makeup of the Laurentian craton is well established, with its Archean cores (e.g., Slave, Superior, Wyoming), predominance of Paleoproterozoic magmatic arcs in the northern and western parts of the craton (e.g., Hottah, Great Bear, Fort Simpson), and Mesoproterozoic crust, including the Yavapai-Mazatzal and Grenville orogens, in its southern and eastern parts (Fig. 16; Hoffman, 1988). Along western Laurentia, these variations in basement ages are reflected in the series of detrital zircon reference spectra that were developed for autochthonous to parautochthonous strata of the Cordilleran miogeocline (Figs. 15E–15I; Gehrels et al., 1995; Gehrels and Ross, 1998; Gehrels, 2000). These reference age spectra closely reflect the Precambrian source regions present to the east of a specific segment of the Cordilleran miogeocline (Figs. 15 and 16). The Yukon-Tanana detrital zircon data (Figs. 15A–15D) are dominated by ages in the 1.7–2.0 Ga range and subordinate populations in the 2.5–2.75 Ga range, a pattern that matches well the abundance of Paleoproterozoic magmatic arc rocks and adjacent Archean rocks in northwestern Laurentia (e.g., Great Bear and Fort Simpson arc terranes; Slave province; Fig. 16), and the detrital zircon references for either northern British Columbia or Alaska (Figs. 15E, 15F). The Yukon-Tanana data differ from the Alaska detrital zircon reference in the paucity of Mesoproterozoic ages (1.0–1.4 Ga; Fig. 15E) and in the low, but consistent occurrence of zircons in the 2.2–2.4 Ga range (hachured area in Fig. 15), a feature more characteristic of the northern British Columbia reference spectra (Fig. 15F). Occurrence of juvenile magmatic crust of 2.1–2.3 Ga age is unique to the Hottah and Buffalo Head terranes of northwestern Laurentia (Fig. 16), a characteristic consistent with derivation of the Yukon-Tanana terrane from a source in northwestern Laurentia, not far from its current position in the northern Cordillera.

This conclusion is similar to that of Nelson and Gehrels (2007), who reported the results of detrital zircons from younger units in the Yukon-Tanana terrane, and the Swift River and Klinkit Groups (the Swift River Group is part of the regional Finlayson assemblage; Colpron et al., 2006a; Fig. 2). Our results from the older metasedimentary rocks of the Snowcap assemblage, the basement to the Yukon-Tanana terrane, are consistent with the Snowcap being the local source for these younger sedimentary rocks. The prevalent tectonic model for the peri-Laurentian realm in mid-Paleozoic time suggests that by the time the Finlayson and Klinkit assemblages were being deposited, the Yukon-Tanana terrane was isolated from the western Laurentian craton by a marginal ocean basin, the Slide Mountain ocean (Nelson, 1993; Creaser et al., 1999; Nelson et al., 2006; Colpron et al., 2007), thus preventing direct clastic input from the craton.

Although the detrital zircon and isotopic signatures of the Snowcap assemblage are consistent with a northwestern Laurentian origin for the Yukon-Tanana terrane, it could be argued that similar signatures could be derived from other cratons, for example, such as the North China block (Darby and Gehrels, 2006). However, the similarities in the Late Devonian development and metallogeny of both the western Laurentian margin (the miogeocline) and the Yukon-Tanana terrane (Finlayson assemblage) provide strong support for a northwestern Laurentian origin (Nelson et al., 2002, 2006; Peter et al., 2007). In particular, the Pb isotopic signatures of Late Devonian–early Mississippian syngenetic sulfide occurrences in the Yukon-Tanana terrane are along the Pb evolution curve defined by syngenetic sulfide occurrences in miogeoclinal rocks of northwestern Laurentia (the shale curve of Godwin and Sinclair, 1982). This Pb evolution model is more highly radiogenic than other Devonian–Mississippian massive sulfide districts in the world and unique to the northwestern Laurentian margin (Godwin and Sinclair, 1982; Mortensen et al., 2006; Nelson et al., 2002, 2006).

Although the bulk of the metasedimentary rocks in the Snowcap assemblage have evolved trace element and isotopic signatures, some samples show a mixed signature between evolved and juvenile (mafic) end members (Fig. 12). This mixed signature was previously reported from metasedimentary rocks of the Yukon-Tanana terrane in southern Yukon and Alaska (Creaser et al., 1997; Grant, 1997; Patchett and Gehrels, 1998; Patchett et al., 1998). In these previous studies, the evolved signature was interpreted to be derived from upper crustal sources in the western Laurentian miogeocline, whereas the source of juvenile detritus was inferred to be in the coeval early Mississippian arc of the Yukon-Tanana terrane (Creaser et al., 1997; Grant, 1997; Finlayson assemblage in Fig. 2). A significant difference between our new data from the Snowcap assemblage and these previous studies is that these older studies predate the establishment of a regional stratigraphic framework for the Yukon-Tanana terrane. Deposition of the Snowcap assemblage metasedimentary rocks predated the Late Devonian–early Mississippian arc magmatism of the Finlayson assemblage (Fig. 2). Therefore the mixed evolved and juvenile isotopic signatures in the Snowcap assemblage cannot be explained by mixing of this juvenile arc material. More localized sources, such as the greenstone and amphibolite of the Snowcap assemblage, may be more plausible sources of juvenile material.

The mafic rocks of the Snowcap assemblage have the characteristics of alkalic rocks found within continental rifts: high TiO₂ and P₂O₅, Nb/Y ~0.7, elevated HFSE and REE contents, and OIB- and E-MORB–like primitive mantle normalized signatures (Figs. 6 and 7; McDonough, 1990; Ellam, 1992; Goodfellow et al., 1995; Piercey et al., 2002). Their incompatible element ratios (Zr/Yb versus Nb/Yb; Fig. 6C) are between the average values for E-MORB and OIB, suggesting derivation from incompatible element–enriched mantle sources. This is also reflected in ϵNd_t and ϵHf_t values that are lower than the depleted mantle at 360 Ma (Fig. 8; Table 2; see following).

Mafic rocks in the Snowcap assemblage form two distinct groups based on variations in HFSE and REE contents. Both groups show similar trace element patterns, but Group 1 shows a greater HFSE and REE enrichment and steeper primitive mantle normalized patterns than Group 2 (Fig. 7). Variations in the incompatible HFSE and REE contents can be attributed to: (1) variation in the HFSE and REE contents between two distinct sources; or (2) variation in degree of partial melting of a common source. Isotopic data for the two groups exhibit slight variations (Table 2), suggesting that the sources for the two groups of mafic rocks may have been slightly different or the mantle source was common but heterogeneous in nature. The broad similarity in trace element patterns between both groups of mafic rocks (Fig. 7) is consistent with derivation from similar mantle source regions, thus suggesting that variation in incompatible element contents between the two groups is, in part, due to variations in the degree of partial melting. To test this we compare their Zr/Hf ratios with incompatible element ratios (Fig. 17). In rocks with alkalic affinities, like the Snowcap assemblage, Zr/Hf ratios are good indicators of partial melt-related variations in trace element contents. Zirconium and Hf are geochemical twins that rarely fractionate under high degrees of mantle partial melting (~10% or higher), with most mafic rocks having chondritic ratios of ~36–37 (Bau, 1996; David et al., 2000). In contrast, at low degrees of partial melting (~1%–3%), the more incompatible Zr is enriched relative to Hf, resulting in mafic rocks with Zr/Hf ratios greater than the chondritic values of ~36–37 (David et al., 2000). In addition, when concentrations of incompatible HFSEs and REEs are correlated with Zr/Hf ratios, it can be argued that Zr/Hf and incompatible element variations are due to variations in partial melting (e.g., David et al., 2000). While there are minor outliers, there is a correlation between the most LREE-enriched samples (Group 1) and higher Zr/Hf ratios (Fig. 17), providing strong evidence that the variations in incompatible trace element composition in the different groups of mafic rocks from the Snowcap assemblage can be best explained by variations in degrees of partial melting of a common mantle source.

E-MORB and OIB-type mafic rocks occur in many continental rifts, mid-ocean ridges, and ocean island arc rifts. In some cases, rocks with E-MORB and OIB signatures have been attributed to melting within mantle plumes or asthenospheric mantle (e.g., Ernst and Buchan, 2001; Fitton et al., 1997; Lassiter and DePaolo, 1997), whereas others have illustrated an association with lithospheric mantle sources (e.g., Hawkesworth et al., 1990; McDonough, 1990; Pearce, 1983; Rogers and Hawkesworth, 1989). The rocks of the Snowcap assemblage are unlikely to have been derived from a plume because of their low volume (e.g., hundreds of meters to kilometers in area), in contrast with the high volumes of basalt associated with plume-derived magmatic sequences (e.g., Coffin and Eldholm, 1994; Ernst and Buchan, 2001; Ernst et al., 1995). A lithospheric mantle source is also supported by isotopic data for the Snowcap mafic rocks. The ϵNd_t and ϵHf_t values of the Snowcap mafic rocks are lower than the depleted mantle at 360 Ma, and the Proterozoic depleted mantle model ages for these rocks are indicative of derivation from a source region that has a long history of incompatible element enrichment (e.g., lithospheric mantle; Hawkesworth et al., 1990; McDonough, 1990; Pearce, 1983; Rogers and Hawkesworth, 1989). Low ϵNd_t and ϵHf_t values can be induced by crustal contamination of basalts. If these values were controlled by crustal contamination, however, there would be coincident trace element indicators of contamination, namely, the low ϵNd_t and ϵHf_t would be accompanied by low Nb/Th and Nb/La, and negative Nb anomalies on primitive mantle normalized plots. These features are not present in the Snowcap assemblage mafic rocks, thus supporting an enriched subcontinental lithospheric mantle source for these basalts.

It is uncertain if the metabasite in the Snowcap assemblage are intrusions (dikes, sills) or volcanic units intercalated with the metasedimentary rocks. However, local intercalation of marble and amphibolite at various scales, from a few centimeters to 50 m, such as observed on Snowcap Mountain (Fig. 4; Colpron, 2000), suggests that the mafic rocks are at least locally of volcanic origins and therefore syndepositional with the metasedimentary rocks.

The composition of the carbonate-bearing rocks brings further insight into the timing of volcanism relative to sedimentation, because carbonates provide an outstanding record of the geochemical and isotopic compositions of sea-water at the time of their formation (Shaw and Wasserburg, 1985; Bolhar et al., 2002; Fanton et al., 2002; Kamber and Webb, 2001). A sample of marble from the Snowcap assemblage has a trace element signature similar to most other marine carbonates and modern seawater, with Ce depletions, high Y/Ho, and elevated HREE relative to LREE (Fig. 13; e.g., Shaw and Wasserburg, 1985; Elderfield et al., 1988; Nozaki et al., 1997; Bolhar et al., 2002; Fanton et al., 2002; Kamber and Webb, 2001). The calc-silicate samples have similar patterns, but with higher total trace elements (Fig. 13), possibly reflecting contamination by silicate detritus. It is notable, however, that all carbonate-bearing samples have chondritic to juvenile ϵNd_t and ϵHf_t values (Fig. 12). Given that Nd has a short residence time in the oceans (Piepgras and Wasser burg, 1982), it can be assumed that the ϵNd_t of the Snowcap carbonates represents that of seawater at the time of formation. It has been argued that Hf has a longer residence time than Nd in the oceans (Lee et al., 1999; White et al., 1986). However, a similar juvenile distribution of ϵHf_t and ϵNd_t in Snowcap carbonate-bearing rocks suggests that Hf isotopes are recording the same composition of seawater as the Nd isotopes (Fig. 12).

The presence of juvenile isotopic signatures in carbonate rocks (and by proxy seawater) requires significant concentrations of juvenile Nd and Hf isotopes in the water column at the time of their formation. This could be accomplished either via the weathering of abundant mafic detritus such that it effectively swamps any continental signature within the water column, or the addition of juvenile Nd and Hf via subaqeous basaltic eruptions and/or wind-delivered detritus from juvenile subaerial eruptions. Given the detrital zircon data, geochemical, and isotopic data for clastic metasedimentary rocks of the Snowcap assemblage, it seems unlikely that significant juvenile source terranes were available to weathering at the time of deposition (i.e., the bulk of the meta-clastic rocks shows evidence for derivation from evolved crust). This suggests that the juvenile Nd and Hf isotopic compositions must have been added to the water column via magmatic activity. This overlap in isotopic signatures of the carbonate and mafic rocks in the Snowcap assemblage implies that carbonate deposition and magmatism were broadly coeval.

Mafic magmatism in the Snowcap assemblage could be related to one of several episodes of intermittent extension and juvenile mafic magmatism documented in Neoproterozoic to early Paleozoic strata of the western margin of Laurentia (e.g., Goodfellow et al., 1995). This extension culminated with the onset of arc magmatism and rifting of the basement to Yukon-Tanana terrane in Late Devonian time (e.g., Nelson et al., 2006; Piercey et al., 2006; Colpron et al., 2007). Although magmatic events of late Cambrian, Early Ordovician, Silurian, and Late Devonian ages are widespread in the northern Cordillera (Goodfellow et al., 1995), they produced only limited volumes of magma and may only have had limited impact on the isotopic composition of seawater. By contrast, the Neoproterozoic Franklin-Gunbarrel magmatic event (Harlan et al., 2003; Heaman et al., 1992) produced a large volume of magma extending over 2.5 × 10⁶ km² (it was defined as a large igneous province by Ernst and Buchan, 2001) that likely had a significant impact on ocean chemistry. The absence of detrital zircons younger than 1.0 Ga in the Snowcap assemblage (Figs. 14 and 15A) and the uniform Proterozoic model ages for its mafic rocks support a Neoproterozoic age for the Snowcap assemblage.

Our study of the Snowcap assemblage is important for understanding the role of crustal recycling in the evolution of the North American Cordillera. A traditional view of the Cordillera regards most of the accreted terranes as a net addition of crustal material to the western margin of North America (e.g., Samson and Patchett, 1991; Patchett and Gehrels, 1998). The composition and character of the Snowcap assemblage are consistent with its origin in a continental margin setting, probably the Neoproterozoic–early Paleozoic margin of northwestern Laurentia, and thus represents primarily recycled continental crustal material (e.g., Creaser et al., 1997; Patchett and Gehrels, 1998). The Snowcap assemblage is the basement upon which mid- to late Paleozoic magmatic arcs of the Yukon-Tanana terrane were developed (Colpron et al., 2006a). It also forms, at least locally, the basement to the younger, early Mesozoic arcs of Quesnellia and Stikinia (Nelson and Friedman, 2004; Colpron et al., 2007). Mid- to late Paleozoic felsic igneous rocks of the Yukon-Tanana terrane generally have evolved Hf-Nd isotopic systematics and Archean–Proterozoic xenocrystic zircons that are consistent with either melting of, or extensive crustal interaction with, the Snowcap assemblage at the time of emplacement (e.g., Mortensen, 1992; Colpron et al., 2006b; Piercey et al., 2006). Recycling of crust in these felsic igneous rocks occurred in two main pulses in the Yukon-Tanana terrane, one in Late Devonian–early Mississippian time (365–340 Ma) and the other in Middle to Late Permian time (269–252 Ma; Nelson et al., 2006; Piercey et al., 2006). Juvenile crust additions, while present, were volumetrically limited during the evolution of the Yukon-Tanana terrane (e.g., Creaser et al., 1999; Piercey et al., 2004, 2006), and the history of the terrane is dominated by crustal recycling. Consequently, accretion of the Yukon-Tanana terrane in early Mesozoic time contributed only limited juvenile crustal material to the Cordilleran margin of North America. Net crustal recycling is probably an important process in the formation and evolution of other pericratonic terranes in the Cordillera (e.g., Alexander, Farewell, Arctic Alaska; Colpron and Nelson, 2009, and references therein) and other orogenic belts (e.g., Dashwoods block in the Appalachians; van Staal, 2007).

The Snowcap assemblage is a continental margin succession of metaclastic, carbonate, and mafic metavolcanic rocks that forms the basement to mid- and late Paleozoic arc rocks of the Yukon-Tanana terrane. Geochemical, Nd-Hf isotopic, and detrital zircon data from clastic metasedimentary rocks of the Snowcap assemblage point to a northwestern Laurentian cratonic source region, similar to that of the adjacent Cordilleran miogeocline. Our new data support conclusions proposed in previous reconnaissance studies of the Yukon-Tanana terrane, but benefit from an improved regional stratigraphic context for the terrane. In contrast to previous studies, our data suggest that the mixed evolved and juvenile signature of some metasedimentary rocks in the Snowcap assemblage is more likely related to coeval mafic alkalic magmatism than younger arc magmatism in the terrane. In addition, the Snowcap assemblage represents a local source in the Yukon-Tanana terrane for evolved isotopic signatures and Paleoproterozoic–Archean zircons (both detrital grains and xenocrystic cores) in younger mid- to late Paleozoic rocks of the terrane, at times when the Laurentian craton was probably not available as a direct source.

Mafic alkalic rocks of the Snowcap assemblage were the products of low-degree partial melting of incompatible element–enriched lithospheric mantle sources, most likely related to one of several Neoproterozoic–early Paleozoic rifting events recorded along the western margin of Laurentia. Marbles and calc-silicate rocks in the Snowcap assemblage have trace element compositions similar to modern seawater and juvenile Nd-Hf isotopic signatures similar to the mafic rocks, implying coeval sedimentation and magmatism.

The overall character and composition of the Yukon-Tanana terrane suggest that its evolution was dominated by crustal recycling processes. Its accretion to the western margin of North America in early Mesozoic time contributed a limited amount of juvenile crustal material to the Cordillera.

Discussions with JoAnne Nelson, Don Murphy, Rob Creaser, and other participants of the Ancient Pacific Margin NATMAP project are greatly appreciated. We thank George Gehrels (University of Arizona) for performing the detrital zircon U-Pb analyses, and Dominique Weis (Pacific Centre for Isotopic and Geochemical Research at the University of British Columbia) for performing the Nd and Hf isotopic analyses. Lauren Blackburn provided assistance with the petrographic descriptions summarized in Supplemental File 1 (see footnote 1). This research was funded by the Yukon Geological Survey, a grant from the Laurentian University Research Fund (LURF), and a Discovery Grant from the Natural Sciences and Engineering Research Council (NSERC) of Canada. We also thank Don Murphy for a review of an earlier version of this manuscript. Comments and suggestions by Geosphere science editor Dennis Harry and an anonymous reviewer are gratefully acknowledged. This is Yukon Geological Survey contribution 006.