Sr-Nd-Hf-Pb isotope and trace element evidence for the origin of alkalic basalts in the Garibaldi Belt, northern Cascade arc


  • Emily K. Mullen,

    Corresponding author
    1. Pacific Centre for Isotopic and Geochemical Research, Department of Earth, Ocean and Atmospheric Sciences, University of British Columbia, 6339 Stores Road, Vancouver, British Columbia, V6T 1Z4, Canada
    Search for more papers by this author
  • Dominique Weis

    1. Pacific Centre for Isotopic and Geochemical Research, Department of Earth, Ocean and Atmospheric Sciences, University of British Columbia, 6339 Stores Road, Vancouver, British Columbia, V6T 1Z4, Canada
    Search for more papers by this author


[1] In the Garibaldi Belt, the northern segment of the Cascade arc, basalts at Bridge River Cones, Salal Glacier, and Mt. Meager (BSM volcanic centers) are alkalic, atypical for an arc setting. Subduction signatures are negligible or absent from primitive alkalic basalts from Salal Glacier and Bridge River, while altered oceanic crust may have contributed a minimal amount of fluid at Mt. Meager. More evolved BSM basalts display trace element signatures considered typical of arc lavas, but this is a consequence of deep crustal assimilation rather than primary input from the subducted slab. Primary BSM basalts represent 3–8% melts that segregated from enriched garnet lherzolite at significantly higher temperatures and pressures (70–105 km) than calc-alkaline Cascade arc basalts. The BSM mantle source is significantly more incompatible element-enriched than the depleted mantle tapped by calc-alkaline Cascade arc basalts. The BSM basalts are also isotopically distinct from calc-alkaline Cascade arc basalts, more similar to MORB and intraplate basalts of the NE Pacific and NW North America. The relatively deep, hot, and geochemically distinct mantle source for BSM basalts is consistent with upwelling asthenosphere. The BSM volcanic centers are close to the projected trace of the Nootka fault, which forms the boundary between the subducting Juan de Fuca plate and the near-stagnant Explorer plate. A gap or attenuated zone between the plates may promote upwelling of enriched asthenosphere that undergoes low-degree decompression melting to generate alkalic basalts that are essentially free of slab input yet occur in an arc setting.

1. Introduction

[2] The role of mantle heterogeneity in generating the compositional diversity observed in arc magmas is difficult to decipher because it can be masked by contributions from the subducting slab. There is a general consensus that mantle melting in arcs occurs in response to input of a hydrous component derived from the subducting slab that generates hydrous minerals in the mantle, lowers the mantle solidus temperature, and produces melts by dehydration melting or “flux” melting [e.g., Kushiro, 1987; Grove et al., 2002]. As a consequence, basalts erupted in subduction zone settings are predominantly aluminous, subalkaline, and bear a slab-derived trace element “arc signature” that includes enrichments in large ion lithophile elements (LILEs) and depletions in high field strength elements (HFSEs) [e.g., Gill, 1981; Pearce and Peate, 1995]. The Cascade arc, which extends ∼1300 km from northern California to southwestern British Columbia, is an example of an arc in which calc-alkaline lavas dominate [Schmidt et al., 2008; Hildreth, 2007; Bacon et al., 1997; Conrey et al., 1997]. However, the northernmost segment of the arc is unusual in that mafic lavas are predominantly alkalic. Alkali olivine basalt and hawaiite occur at the volcanic fields of Mt. Meager, Salal Glacier, and Bridge River Cones (hereinafter referred to as the BSM volcanic centers) [Green and Sinha, 2005].

[3] Alkalic lavas are uncommon in arc settings, particularly along the main arc axis, and are attributed to a variety of phenomena that include tears in the subducting plate, back-arc extension, slab windows, entrainment of mantle hotspots, accreted enriched mantle, and intraarc rifting [e.g., Nakamura et al., 1989; Abratis and Wörner, 2001; Ferrari et al., 2001; Márquez et al., 1999; Turner and Hawkesworth, 1998; Pearce and Stern, 2006; Thorkelson and Taylor, 1989; Skulski et al., 1991; Righter and Carmichael, 1992; Hughes, 1990; Pearce and Peate, 1995, and references therein]. The BSM volcanic centers are located ∼110 km above the subducting plate, broadly similar to other Cascade arc volcanoes [McCrory et al., 2004], suggesting a potential connection to arc magmatic processes. Elucidating the petrogenesis of the BSM basalts may provide valuable insights into mantle and slab processes under the diminished subduction regime at the northern boundary of the Cascade arc [Harry and Green, 1999].

[4] The Cascade magmatic arc has been active since ∼40 Ma and is a consequence of subduction of the Juan de Fuca oceanic plate beneath North America [Hildreth, 2007]. In northwestern Washington, an abrupt change in the orientation of the arc axis mirrors a bend in the continental margin, subdividing the arc into two major segments (Figure 1a). The High Cascades segment (Mt. Rainier to Lassen Peak) is separated from the Garibaldi Volcanic Belt (GVB) (Glacier Peak to Silverthrone) by a 120 km gap in modern volcanism [Green and Harry, 1999]. Throughout much of the High Cascades, the subducting plate is ∼10 Myr old at the trench with ocean floor isochrons generally parallel to the continental margin [Wilson, 2002]. In the GVB, however, isochrons are oblique to the plate margin and the slab age at the trench decreases northward to ∼6 Ma outboard of the BSM volcanic centers (Figure 1a), making it one of the youngest and hottest subduction settings in the world [Syracuse et al., 2010; Harry and Green, 1999]. Subducted water is expected to be lost at shallow depths from a hot slab, leading to reduced hydration of the subarc mantle wedge [Green and Harry, 1999]. A diminished subduction regime may account for why the GVB, in comparison to the High Cascades, has a narrower width, lower magma production rates, and magmatism mainly restricted to the major volcanic centers [Harry and Green, 1999].

Figure 1.

(a) Map of the Cascade arc and its tectonic setting. The extents of the Garibaldi volcanic belt and High Cascades segments of the arc are indicated with pink arrows. Volcanic and plutonic rocks are shown in yellow and orange shading, respectively. Black triangles denote composite volcanoes. (b) The study area is enclosed by a small bold rectangle and is enlarged. Igneous rock distributions are compiled from Lawrence et al. [1984], Monger [1989], DuBray et al. [2006], Green et al. [1988], Wheeler and McFeely [1991]. Oceanic plate configurations are from Braunmiller and Nabelek [2002], Audet et al. [2008], and Wilson [2002]. Colored lines on the oceanic plates are isochrons; accompanying numbers indicate the age of oceanic crust in Ma (from Wilson [2002]). Pseudofaults are shown as thin gray lines. Four heavy gray arrows on the Juan de Fuca and Explorer plates are convergence vectors (mm/yr) obtained from McCrory et al. [2004], Riddihough and Hyndman [1991], and Braunmiller and Nabelek [2002] for a reference frame fixed relative to North America.

[5] In this paper, we assess three hypotheses for the origin of the BSM alkalic basalts. In the first model, the BSM alkalic basalts are essentially lower melt fraction “equivalents” of more typical Cascade arc calc-alkaline basalts [Green and Harry, 1999; Green and Sinha, 2005]. Reduced hydration of the subarc mantle wedge may reduce its capacity for flux melting, resulting in lower melt fractions that are enriched in alkali elements but display minimal arc signature. A similar model has been proposed to account for basalts elsewhere in the Cascade arc that are geochemically indistinguishable from intraplate basalts [Reiners et al., 2000]. Intraplate-type lavas dominate the back-arc Simcoe volcanic field east of Mt. Adams and are interspersed with other basalt types in a swath extending west ∼150 km from Simcoe to Portland that has been referred to as the Cascades-Columbia transect [Leeman et al., 1990, 2005; Hildreth, 2007; Jicha et al., 2008; Conrey et al., 1997; Bacon et al., 1997]. A few other examples occur north of Mt. Rainier [Reiners et al., 2000] and in north-central Oregon [Conrey et al., 1997].

[6] Second, a subducted plate boundary or fracture zone may trigger mantle upwelling, inducing decompression melting that generates low-degree, alkali-rich melts. These processes have been proposed for other arcs, including the Mexican arc and Lesser Antilles [e.g., Righter et al., 1995; DeLong et al., 1975; Pearce, 2005], and a version of this model is mentioned by Lawrence et al. [1984] in the context of the Salal Glacier alkalic basalts. The subducted boundary between the Explorer and Juan de Fuca plates intersects the BSM volcanic centers and has been implicated in the origin of the Wells Gray-Clearwater volcanic field (3.5 Ma–7.6 ka [Hickson and Souther, 1984]) and Chilcotin Group basalts (∼32–0.8 Ma [Mathews, 1989]) that are located in the GVB back-arc region [Madsen et al., 2006; Sluggett, 2008]. Near the triple junction among the Explorer, Pacific, and North America plates, the Quaternary alkalic seamounts of the Tuzo Wilson volcanic field are attributed to a “leaky transform” in an oceanic setting [Allan et al., 1993].

[7] Third, the BSM basalts may reflect one of the other mechanisms proposed to account for Miocene-Holocene intraplate volcanic centers that occur in a broad swath extending from southern British Columbia to Alaska. The east-west trending Anahim volcanic belt (14.5 Ma–7.2 ka [Bevier, 1989]), located immediately north of the GVB, may be related to a hotspot [Bevier, 1989; Charland et al., 1995], an edge effect of the Juan de Fuca plate [Stacey, 1974], or ridge subduction [Farrar and Dixon, 1992]. Farther north, the extensive northern Cordilleran volcanic province (∼20 Ma–200 years B.P.) has been attributed to crustal extension [Edwards and Russell, 2000]. The aforementioned Wells Gray-Clearwater and Chilcotin lavas may alternatively be a product of back-arc extension [Bevier, 1983; Hickson, 1987]. More recently, Thorkelson et al. [2011] proposed a single model in which all of these volcanic provinces are related to upwelling of enriched mantle within and along the eroding margins of the ∼1500 km long Northern Cordilleran slab window [Thorkelson and Taylor, 1989] that extends nearly as far south as the BSM volcanic centers.

[8] In this study, we investigate the roles of the mantle and subducting slab in generating the alkalic compositions of the BSM basalts with high precision whole-rock Sr-Nd-Pb-Hf isotope ratios and trace element data. Radiogenic isotope signatures of primitive basalts are sensitive indicators of mantle source heterogeneity [e.g., Hofmann, 2003] and the presence of components derived from subducted oceanic crust and sediment [e.g., Kay et al., 1978]. Green and Sinha [2005] showed that the BSM alkalic basalts record less slab input than calc-alkaline basalts of the southern GVB, but minimized the possible role of mantle heterogeneity. However, recent improvements in the precision of isotopic measurements have revealed mantle heterogeneities that were previously difficult to discern [e.g., Abouchami et al., 2005]. Alkalic basalts in British Columbia and the Cascade arc are typically ascribed to mantle sources that are more enriched in incompatible elements than the mantle wedge sampled by calc-alkaline basalts [e.g., Thorkelson et al., 2011; Sluggett, 2008; Edwards and Russell, 2000; Leeman et al., 1990, 2005; Bacon et al., 1997; Borg et al., 1997; Conrey et al., 1997; Schmidt et al., 2008; Jicha, et al. 2008]. However, Reiners et al. [2000] proposed that both basalt types can be derived from a homogeneous mantle variably fluxed by slab-derived fluids.

[9] We compare the BSM basalts to calc-alkaline basalts from Mt. Baker, a stratovolcano located in the “cooler” southern GVB and representative of more typical Cascade arc basalts, and to published data for other intraplate alkalic basalts from British Columbia. Our new isotope and trace element data show minimal subduction influence on the source of the most primitive basalts at Salal Glacier and Bridge River. The BSM basalts also have a mantle source that is isotopically distinct from, and more incompatible element enriched, than the mantle underlying much of the Cascade arc. These results have important implications for the physical configuration of the subducting slab and mantle flow patterns in northern Cascadia.

2. Geology of the Bridge River, Salal Glacier, and Mt. Meager Volcanic Centers

[10] The Bridge River Cones, Salal Glacier, and Mt. Meager are located ∼150 km north of Vancouver, British Columbia (Figure 1b). At the two northernmost centers (Salal Glacier and Bridge River), lavas are almost exclusively mafic. The Mt. Meager volcanic field includes basalt through rhyolite but is dominated by intermediate compositions, and Mt. Meager proper is a composite andesitic stratovolcano [Ke, 1992].

[11] The Salal Glacier volcanic field includes pillow lavas, tuffs, and variably palagonitized and brecciated flow remnants that survived continental ice sheet advances as high-altitude nunataks. At lower altitudes, severe glacial erosion has revealed rhyolite and andesite dikes. Age dates for an alkali basalt and overlying hawaiite are 0.97 and 0.59 Ma (K-Ar), respectively [Lawrence, 1979].

[12] Lavas at the Bridge River Cones are exclusively alkalic [Roddick and Souther, 1987]. The term “cones” is a misnomer because none of the deposits is a true volcanic cone; rather, glacial erosion has produced cone-like forms. Columnar lavas of the Sham Hill plug and Tuber Hill exposure are dated at 1 Ma and 600 ka (K-Ar), respectively [Roddick and Souther, 1987].

[13] At Mt. Meager, intermediate to silicic lavas span the alkalic-subalkalic boundary [Stasiuk et al., 1994]. Mafic lavas are exclusively alkalic, however, and occur as four flow remnants collectively known as the Mosaic Assemblage [Stasiuk and Russell, 1989; Stasiuk et al., 1994]. Two of the basalts are dated at ∼90 and 140 ka (K-Ar) [Anderson, 1975; Woodsworth, 1977]. Evidence for recent involvement of mafic magma in the form of mafic enclaves and banded pumices is preserved in the ∼2360 years B.P. that explosively released ∼10 km3 of dacite [Clague et al., 1995; Michol et al., 2008]. Banded pumices and mafic enclaves indicate that the intrusion of a basaltic magma may have triggered the eruption [Stasiuk et al., 1994].

3. Major Element Compositions and Petrography

[14] Samples analyzed for this study are from the Bridge River, Salal Glacier, Mt. Meager, and Mt. Baker sample suites of Green and Sinha [2005] and from the Mt. Baker sample suite of E. K. Mullen and I. S. McCallum (Origin of basalts in a hot subduction setting: Petrologic and geochemical insights from Mt. Baker, northern Cascade arc, submitted to Journal of Petrology, 2013, hereinafter referred to as Mullen and McCallum, submitted manuscript, 2013). Major element data and detailed petrographic descriptions for all samples are provided in those references.

[15] The BSM basalts are alkalic [Macdonald, 1968] and nepheline normative with Na2O > K2O (Figure 2a). The basalts have distinctly lower SiO2 (Figure 2) and Al2O3 than the calc-alkaline, hypersthene-normative Mt. Baker basalts.

Figure 2.

Major element variation diagrams for the basalts of Bridge River (red), Salal Glacier (orange), Mt. Meager (yellow), and Mt. Baker (lavender). (a) wt % Na2O + K2O versus SiO2 with discriminant line of Macdonald [1968] and fields of Le Bas et al. [1986]. (b) Miyashiro diagram (FeO*/MgO versus SiO2) with discriminant line of Miyashiro [1974]. Bridge River, Salal Glacier, and Mt. Meager data are from Green and Sinha [2005]. Mt. Baker data are from Mullen and McCallum (submitted manuscript, 2013) except Lib21 from Green and Sinha [2005]. Large symbols (circles and diamonds) indicate samples analyzed in this study for trace elements and isotope ratios; diamonds are accompanied by sample numbers and are the Suite 2 samples discussed in the text. Small circles indicate samples of Green and Sinha [2005] not analyzed for the present study.

Figure 3.

(a) 208Pb/204Pb versus 206Pb/204Pb, (b) 207Pb/204Pb versus 206Pb/204Pb. Insets show the same data for the BSM and Mt. Baker basalts but on an expanded scale. Symbols for the BSM and Mt. Baker samples given in legend; circles are used for primitive basalts. The more evolved basalts are subdivided into Suite 1 (circles; isotopically similar to the primitive basalts) and Suite 2 (diamonds accompanied by sample numbers; isotopically distinct from the primitive basalts). 2σ error bars (external reproducibilities) are smaller than symbols in all plots. NHRL is the Northern Hemisphere Reference Line of Hart [1984]. Cascade arc basalt data (pink crosses; only those with >8 wt % MgO are included) are from Conrey et al. [1997], Jicha et al. [2008], Bacon et al. [1994, 1997], Leeman et al. [1990, 2005], Baker et al. [1991], Magna et al. [2006], Grove et al. [2002], and Borg et al. [1997, 2000]. Northern Juan de Fuca MORB data (dark blue +) are from Cousens et al. [1995]. Explorer MORB data (dark gray filled squares) are from B. Cousens (unpublished data 2007). N. Gorda MORB (black +) are from Allan et al. [1993]. Northern Cascadia sediment data (orange circles) are from ODP sites 1027 and 888 [Carpentier et al., 2010, 2013]. Note that all isotope data are normalized to the same isotope standard values as described in main text.

[16] Bridge River alkali olivine basalts and hawaiites encompass the largest range of compositional diversity (Figure 2). Molar Mg/(Mg+Fe2+) values range from 0.44 to 0.62; we consider two samples with Mg/(Mg+Fe2+) > 0.6 as primitive. Phenocryst and microphenocryst minerals are limited to olivine (∼1–2%) and rare plagioclase. Except for one sample with a brown glass matrix, the basalts have holocrystalline groundmasses containing olivine, plagioclase, titanaugite, magnetite, and ilmenite. The most primitive basalt (BRC10) contains biotite and amphibole in the groundmass. Two of the more evolved samples contain quartz xenocrysts and granodiorite xenoliths (BRC03–4, BRC01–3).

[17] Mt. Meager alkali basalts and hawaiites contain <1% microphenocrysts of olivine, clinopyroxene, and plagioclase. The groundmass contains glass and magnetite and, in the least primitive sample (MM01-1), biotite and amphibole. Mg/(Mg+Fe2+) values are 0.59–0.63.

[18] Salal Glacier samples are the most primitive among the BSM basalts with Mg/(Mg+Fe2+) = 0.58 to 0.66 and have the highest normative nepheline contents. The most primitive samples are glassy and vesicular with phenocryst assemblages including <15% olivine, <1% plagioclase, <1% clinopyroxene, and rare orthopyroxene xenocrysts. Less rimitive samples contain orthopyroxene phenocrysts and more abundant plagioclase, and olivine is either rimmed by clinopyroxene or absent.

[19] At Mt. Baker, the most mafic lavas include medium-K calc-alkaline basalt, high-Mg basaltic andesite, and low-K olivine tholeiite, with Mg/(Mg + Fe2+) = 0.56–0.70. All samples contain olivine and plagioclase phenocrysts and some also have clinopyroxene phenocrysts (Mullen and McCallum, submitted manuscript, 2013).

4. Analytical Methods

[20] Trace element abundances and Sr-Nd-Hf-Pb isotope ratios were measured on 19 BSM basalts, using splits of sample powders analyzed by Green and Sinha [2005] for major and trace elements and Sr isotope ratios. Larger symbols in Figure 2 designate samples analyzed for the present study. Hf isotope ratios were also measured on splits of powders of three Mt. Baker basalts previously analyzed for major and trace elements and Sr-Nd-Pb isotopes [Mullen and McCallum, submitted manuscript, 2013; Green and Sinha, 2005].

[21] All chemical separations and mass spectrometric analyses were carried out in Class 100 and 10,000 clean laboratories, respectively, at the Pacific Centre for Isotopic and Geochemical Research at the University of British Columbia. Rock powders (∼100 mg) were digested in subboiled concentrated HF + HNO3 in 15 mL screw-top Savillex beakers on a hotplate for ∼48 h at ∼130°C. Samples were dried down on a hotplate and brought up in subboiled 6 N HCl and fluxed on a hotplate for at least 24 h. Sample aliquots of 5–10% were diluted 5000X with an HNO3 + HF solution for analysis on a Thermo Finnigan Element2 HR-ICP-MS or an Agilent 7700 quadrupole ICP-MS. Sr isotope ratios were measured on a Thermo Finnigan Triton TIMS and Pb, Nd, and Hf isotope ratios on a Nu Instruments MC-ICP-MS (Nu 021) following the procedures of Weis et al. [2006]. Pb, Sr, Hf, and Nd were separated from single powder dissolutions by sequential ion exchange column chemistry as described in Weis et al. [2006, 2007]. All solutions were passed twice through Pb exchange columns to ensure Pb purification. Although thin sections of the analyzed samples indicate little or no alteration in all samples (minor iddingsite in olivine) and LOI values are low (<1%), even minimally altered samples can yield isotopic compositions that are not representative of magmatic isotopic signatures, particularly in the case of Sr and Pb isotopes [Hanano et al., 2009; Nobre Silva et al., 2009]. Therefore, we measured isotope ratios on both unleached and leached powders of some samples. Leaching was conducted prior to powder dissolution following the procedures of Nobre Silva et al. [2009, 2010]. Leached samples gave isotope ratios within analytical error of respective unleached samples for Sr and Nd (Figure S1). 1 Hf isotope ratios are also within analytical error except for one sample (BRC10) that gave a higher value in the leached sample. 207Pb/204Pb in leached samples is systematically lower than in unleached samples while 208Pb/204Pb and 206Pb/204Pb are within error of unleached samples, with the exception of one Mt. Baker sample (Lib21) (Figure S1).1 All isotope plots in the main text show data obtained on leached samples except for cases in which only unleached samples were analyzed. Blank contributions to isotope ratios were negligible with total procedural blanks of ∼50, 400, 90, and 15 pg for Pb, Sr, Nd, and Hf, respectively.

5. Results

5.1 Isotopes

[22] Isotope ratios are reported in Table 1 and plotted in Figures 3-5. 87Sr/86Sr values measured in the BSM basalts are systematically lower than reported by Green and Sinha [2005] for the same samples and lie outside their reported uncertainties (Figure 4a inset). For direct comparison among datasets, all literature data are normalized to standard values of 87Sr/86Sr = 0.710248 for SRM987 and 0.708028 for Eimer and Amend; 143Nd/144Nd = 0.511973 for Rennes, 0.511858 for La Jolla, 0.512633 for BCR-1, and 0.512130 for Ames [Weis et al., 2006, 2007]; 176Hf/177Hf = 0.282160 for JMC 475 [Vervoort and Blichert-Toft, 1999]; and 208Pb/204Pb = 36.7219, 207Pb/204Pb = 15.4963, 206Pb/204Pb = 16.9405 for SRM981 [Galer and Abouchami, 1998].

Table 1. Sr, Nd, Hf, and Pb Isotope Ratios
Sample#Lat (N)Long (W)SiO2 (wt%)Mg NumberaLeached2SEhUnleached2SELeached2SEεNd LeachedfUnleached2SEεNd Unleachedf
Bridge River Cones
BRC01-350.93123.4550.5156.5  0.7032129      
BRC03-450.93123.4550.0754.40.7032196  0.51296676.4   
BRC0450.93123.4549.4852.4  0.7031849      
BRC05-150.92123.4546.6648.3  0.7031199      
BRC0650.90123.4549.3553.5  0.7031758      
dupi    0.70305570.7030568      
Salal Glacier
SG01-250.81123.4546.6465.10.7031407  0.51302167.5   
SG1050.78123.3946.0465.7  0.7031229      
dup      0.7031178      
SG1650.77123.3946.5965.9  0.70310110      
Mt. Meager
dup    0.70376260.70376370.51293655.80.51294165.6
MM0250.69123.5748.6463.0  0.703132       
MM0450.69123.5748.9460.80.70314460.703146 0.51303097.6   
MM0850.55123.5349.6860.6  0.703164       
Mount Baker
02-MB-548.72121.8553.6969.7  0.7031097   0.51300167.1
07-MB-11248.66121.7052.5656.6  0.7032407   0.51303477.7
02-MB-148.72121.8553.3056.0  0.7035138   0.512899 5.1
06-MB-8248.72121.8550.5765.1  0.7031567   0.512986 6.8
07-MB-11448.64121.7352.0661.3  0.7032137   0.513037 7.8
06-MB-9748.78121.8854.4549.0  0.70317310   0.512993 6.9
Sample #Leached2SEεHf LeachedgUnleached2SEεHf Unleachedg
Bridge River Cones
BRC01–3   0.283067810.4
BRC04   0.28302679.0
BRC05-1   0.28304079.5
BRC06   0.28301868.7
Salal Glacier
SG10   0.28301298.5
SG16   0.28302378.9
Mt. Meager
MM02   0.28305069.8
MM08   0.28302258.9
Mount Baker
02-MB-5   0.283100511.6
07-MB-112   0.283114412.1
Sample #Leached2SEUnleached2SELeached2SEUnleached2SELeached2SEUnleached2SE
  1. a

    Calculated as 100*Mg/Mg+Fe2+ (molar), using major element data from Green and Sinha [2005] and Mullen and McCallum (submitted manuscript, 2013) and assuming Fe3+/ΣFe = 0.15.

  2. b

    Reported Sr isotope ratios are corrected for mass fractionation using 86Sr/88Sr = 0.1194. Repeat analysis of the Sr SRM987 standard yielded a mean (± 2σ) of 87Sr/86Sr = 0.710248 ± 2 (n = 7), identical to the accepted value [Weis et al., 2006].

  3. c

    Reported Nd isotope ratios are corrected for mass fractionation using 146Nd/144Nd = 0.7219 and are normalized to 143Nd/144Nd = 0.511973 for the Rennes reference material [Chauvel and Blichert-Toft, 2001] using the daily average method. The Rennes standard was analyzed every two samples and over the course of analysis gave a mean (± 2σ) value of 143Nd/144Nd = 0.511980 ± 65 (n = 16). On a per session basis, reproducibility was significantly better with a maximum daily 2σ value of ±27 (53 ppm).

  4. d

    Reported Hf isotope ratios are corrected for mass bias using 179Hf/177Hf = 0.7325 [Patchett and Tatsumoto, 1981] and normalized to 176Hf/177Hf = 0.282160 for the ULB-JMC 475 reference material [Vervoort and Blichert-Toft, 1999] using the daily average of standard analyses. JMC 475 was analyzed every two samples and over the course of analysis gave a mean (± 2σ) of 176Hf/177Hf = 0.2821732 ± 24 (86 ppm) (n = 24). On a per session basis, reproducibility was significantly better with daily 2σ values ranging from 34 to 61 ppm.

  5. e

    Reported Pb isotope ratios were corrected for mass bias by Tl doping [White et al., 2000] and are normalized to 208Pb/204Pb = 36.7219, 207Pb/204Pb =15.4963, 206Pb/204Pb = 16.9405 for the SRM981 standard [Galer and Abouchami, 1998] by sample-standard bracketing. Replicate analysis of SRM981 over the course of analysis yielded in-run mean ± 2σ values of 208Pb/204Pb = 36.7198 ± 91 (247 ppm), 207Pb/204Pb = 15.4998 ± 34 (220 ppm), and 206Pb/204Pb = 16.9434 ± 29 (168 ppm) (n = 43). On a per session basis, reproducibility was significantly better with daily 2σ values ranging from 74 to 186 ppm, 62 to 176 ppm, and 45 to 139 ppm for 208Pb/204Pb, 207Pb/204Pb, and 206Pb/204Pb, respectively.

  6. f

    ɛNd calculated using a CHUR value of 143Nd/144Nd = 0.512638 [Jacobsen and Wasserburg, 1980].

  7. g

    ɛHf calculated using CHUR value of 176Hf/177Hf = 0.282772 [Blichert-Toft and Albarede, 1997].

  8. h

    2SE values (twice the standard errors) apply to the last decimal place(s) and are the internal absolute errors values for individual sample analyses.

  9. i

    dup designates full procedural duplicates starting with a new sample powder aliquot; reproducibilities are similar to, or better than, the reproducibilities determined through repeat standard analysis (values listed above).

  10. Data in italics are from Mullen and McCallum (submitted manuscript, 2013).

Bridge River Cones
BRC01-3  38.244122  15.56158  18.78788
dup  38.243324  15.56049  18.786310
BRC02  38.190121  15.55548  18.748110
BRC03-438.267118  15.55847  18.83346  
dup38.267418  15.55856  18.83378  
BRC04  38.196127  15.56229  18.737913
BRC05-1  38.127417  15.54817  18.75918
BRC06  38.182828  15.553012  18.746314
dup 38.160136  15.562014  18.775017
Salal Glacier
SG01-238.108720  15.54198  18.74659  
SG10  38.128524  15.54929  18.736312
SG16  38.158533  15.552712  18.759112
Mt. Meager
MM02  38.100535  15.547013  18.703415
MM08  38.074725  15.55509  18.689615
Mount Baker
02-MB-5  38.249522  15.55058  18.79759
07-MB-112  38.307721  15.55757  18.83858
02-MB-1  38.3598   15.5645   18.8515 
06-MB-82  38.2661   15.5518   18.8286 
07-MB-114  38.2753   15.5560   18.8466 
06-MB-97  38.2597   15.5529   18.8356 
Figure 4.

(a) ɛNd versus 87Sr/86Sr; (b) ɛHf versus ɛNd. Symbols and data references as in Figure 3, plus Lassen and Adams data in Figure 4b from Borg et al. [2002] and Jicha et al. [2008]. Note that all isotope data are normalized to the same isotope standard values as described in main text. For the BSM and Mt. Baker data, 2σ error bars (external reproducibilities) are smaller than symbols. Inset in Figure 4a compares 87Sr/86Sr measured in the present study (2σ error bars = 20 ppm; smaller than symbol size) to 87Sr/86Sr measured on the same samples by Green and Sinha [2005] (2σ error bars = 100 ppm). The mantle array in ɛHf versus ɛNd space is from Chauvel et al. [2008]. Orange and blue curves show the effect of adding 2% bulk sediment (blue curve with long dashes), 2% sediment fluid (blue curves with short dashes), 2% sediment melt (solid blue curves), 10% metabasalt fluid (orange dashed curve), and 2% metabasalt melt (orange solid curve) to the mantle prior to 5% equilibrium partial melting of a primitive mantle composition [Sun and McDonough, 1989]. Each curve has two tick marks indicating 1% and 2% addition, except for the metabasalt fluid curve (ticks at 5% and 10% addition). Slab fluid and melt compositions calculated using equilibrium melting/dehydration equations with FL=0.05. Trace element compositions for sediment and metabasalt are from Carpentier et al. [2013] (average of bulk ODP sites 888 and 1027) and Becker et al. [2000] (900°C eclogite), respectively. Sediment isotope composition is the average of ODP sites 888 and 1027 from Carpentier et al. [2010]. Metabasalt Sr and Nd isotope ratios are from Staudigel et al. [1995] and the Hf isotope ratio is the average of Explorer MORB shown here. Partition coefficients from Kessel et al. [2005] at 700°C, 4 GPa (all fluids); 1000°C, 4 GPa (metabasalt melt and sediment melt 1), and Hermann and Rubatto [2009] at 1050°C, 4.5 GPa (sediment melt 2).

[23] Primitive BSM basalts (Mg/Mg+Fe2+ > 0.60) form an isotopic cluster (Figures 3 and 4) with a narrow range of 87Sr/86Sr = 0.70299–0.70314, ɛNd = +7.1 to +7.7, ɛHf = +8.3 to +10.0, 208Pb/204Pb = 38.075–38.172, 207Pb/204Pb = 15.541–15.557, 206Pb/204Pb = 18.690–18.774. The primitive BSM basalts overlap in 208Pb/204Pb and 206Pb/204Pb with N. Juan de Fuca MORB [Cousens et al., 1995] and Explorer MORB (B. Cousens, unpublished data 2007) (Figure 3a) but have slightly higher 207Pb/204Pb and 87Sr/86Sr and lower ɛNd (Figures 3b and 4a). Primitive BSM basalts plot near the depleted end of the Sr-Nd-Pb isotopic arrays defined by other Cascade arc basalts (Figures 3 and 4). Along with Mt. Baker, primitive BSM basalts have among the highest ɛNd values reported for the Cascade arc. Mt. Baker basalts have slightly higher 208Pb/204Pb and 206Pb/204Pb than primitive BSM samples, but overlap in 207Pb/204Pb. Mt. Baker has significantly higher ɛHf (+11.1 to + 12.1) and slightly higher 87Sr/86Sr. In ɛHfNd isotopic space (Figure 4b), primitive BSM and Mt. Baker basalts overlap with only one outlier among data previously published for the Cascade arc (Lassen Peak [Borg et al., 2002] and Mt. Adams [Jicha et al., 2008]). Mt. Adams ɛHf values cluster between primitive BSM basalts and Mt. Baker. Together, the BSM, Mt. Baker, and Mt. Adams basalts define an ɛHf range that is similar to Explorer MORB, but with slightly lower ɛNd.

[24] More evolved BSM basalts define two groups: Suite 1 has isotopic ratios that overlap with primitive basalts from the same volcanic center, whereas Suite 2 has substantially lower ɛNd and higher Sr and Pb isotope ratios than other basalts from their respective volcanic centers (Figures 3 and 4a). Suite 2 includes three BSM basalts (BRC03–4, BRC01–3 at Bridge River; MM01-1 at Mt. Meager), all of which have Pb isotope ratios similar to Mt. Baker basalts, plotting within the Cascade arc array and closer to the field defined by subducting sediment (Figure 3). Suite 2 also includes two Mt. Baker basalts that are isotopically anomalous (in Sr and Nd) relative to more primitive Mt. Baker basalts; Cathedral Crag (MB1) and one Sulphur Creek sample (Lib21) plot closer to the field defined by subducting sediment (Figure 4a). As a group, Suite 2 has lower MgO and generally higher SiO2 than other basalts from their respective volcanic centers (Figure 5).

Figure 5.

208Pb/204Pb versus (a) wt % SiO2, and (b) wt % MgO. Note reversed scale for MgO.

5.2 Trace Elements

[25] Trace element abundances are reported in Table 2 and plotted in Figures 6-9. With the exception of Suite 2 (discussed later), the BSM basalts have LILE and Pb abundances similar to Mt. Baker basalts, but much higher HFSE (Figures 6 and 8a). BSM basalts have substantially lower Zr/Nb and Ba/Nb than Mt. Baker basalts (Figure 8b) and higher (La/Yb)N and (Dy/Yb)N (Figure 7). Among the BSM basalts, Salal Glacier has the highest (La/Yb)N and lowest Yb and displays the least variability among samples (Figure 7a). Bridge River and Mt. Meager have variable REE abundances, and the lowest (La/Yb)N occurs at Mt. Meager (Figure 7b). However, Mt. Baker samples extend to lower (La/Yb)N and higher Yb values than the BSM basalts (Figure 7). Salal Glacier and Bridge River have no Nb anomalies while small negative anomalies occur at Mt. Meager and prominent ones at Mt. Baker (Figure 9a). Ba/La values are lowest at Salal Glacier and Bridge River, intermediate at Mt. Meager, and highest at Mt. Baker (Figure 9b). Abundances of all trace elements in Bridge River and Salal Glacier primitive basalts are similar to samples from nonsubduction settings, including Hawaiian postshield alkalic basalts [Hanano et al., 2010], and overlap with alkalic basalts from the Anahim volcanic belt [Charland et al., 1995], Cascade-Columbia transect [Leeman et al., 2005; Jones, 2002], and Dalles Lakes north of Mt. Rainier [Reiners et al., 2000] (Figures 8, 9a, and 9c).

Table 2. Trace Element Abundances
Sample NumberBRC01–3BRC02BRC03–4BRC04BRC05-1BRC06BRC07-2BRC09-3BRC10SG01-1
Concentrationb (ppm)
Tm0.240.32 0.310.390.31 0.310.280.27
Ga2324 242724 212221
Sample #SG01–2SG01–3SG10SG12SG16MM01-1MM02MM04MM08LIB-21
  1. a

    Method 1: Thermo Finnigan Element2 HR-ICP-MS; Method 2: Agilent 7700 quadrupole ICP-MS.

  2. b

    Data were corrected for blank contributions and by sample-standard bracketing to published values for the USGS BCR2 reference material [Raczek et al., 2001] analyzed every eight samples (method 1), or the USGS AGV1 reference material [Chauvel et al., 2011] analyzed every six samples (method 2). Repeat analysis of the USGS BHVO2 standard gave RSD values of <5% and concentrations within 5% (relative) of published values (as compiled by Carpentier et al. [2013] from GeoRem) for most elements. The average BHVO2 values obtained during analytical sessions are reported in Table S1. Duplicate analyses gave reproducibilities better than 5% (Table S1). Total procedural blanks (Table S1) were negligible relative to analyzed sample concentrations.

Concentration (ppm)
Tm 0.290.300.240.26 
Ga 20221721 202122 
Figure 6.

(a–d) Extended N-MORB normalized [Sun and McDonough, 1989] trace element diagrams, subdivided by volcanic center. (d) Light gray field in each panel encompasses the range defined by Mt. Baker basalts. The darkest colors (with sample numbers) signify Suite 2 samples discussed in the text. (e) All of the Suite 2 samples are plotted together for comparison to Mt. Baker basalts.

Figure 7.

(a–d) Chondrite-normalized [McDonough and Sun, 1995] rare-earth element diagrams subdivided by volcanic center.

Figure 8.

(a) Ba (ppm) versus Nb (ppm); (b) Zr/Nb versus Ba/Nb. Except for Suite 2 samples (diamonds), plus MM08 from Mt. Meager, the BSM basalts are similar to basalts from MORB and OIB settings, i.e., essentially no subduction component. MORB data (gray filled circles) and Hawaiian basalt data (shield and postshield lavas shown as filled green and pink circles, respectively) were obtained from the PETDB ( and GEOROC databases (, respectively, accessed in May 2012. Average OIB composition [Sun and McDonough, 1989] is shown as a black asterisk in Figure 8b. Black arrows in lower right corner of Figure 8a and upper right corner of Figure 8b show the effect of 15% fractionation of each mineral indicated, calculated using the Rayleigh equation, the starting composition of BRC09-3, and partition coefficients listed in Table 4 plus ilmenite from McCallum and Charette [1978]. Only ilmenite and magnetite vectors are shown in Figure 8b because the other phases shown in Figure 8a have a negligible effect. Orange and blue curves are for addition of subduction components to a depleted mantle source (average depleted MORB mantle of Salters and Stracke, 2004) prior to 10% partial melting, calculated as described in Fig. 4 caption. Most subduction components increase Ba at a given Nb, and Ba/Nb at a given Zr/Nb. The heavy black curves are the compositions of DM melts at 1 GPa and PM melts at 3 GPa, respectively (tick marks indicate % partial melt). Melt compositions were calculated using the equilibrium melting equation with mineral/melt partition coefficients from Table 4 and residual mantle mineral assemblages determined by BATCH modeling [Longhi, 2002] of starting compositions of Wasylenki et al. [2003] and Kinzler [1997]. Filled black squares with white (cross) and (plus) symbol are for DM (depleted MORB mantle) of Salters and Stracke [2004] and PM (primitive mantle) of Sun and McDonough [1989], respectively. Inset diagrams include data for alkalic basalts (molar Mg/(Mg + Fe2+) > 0.60, assuming Fe3+/∑Fe = 0.15) from the Anahim volcanic belt (dark blue squares) [Charland et al., 1995], Cascade-Columbia transect (light blue squares) [Leeman et al., 2005; Jones, 2002], and Dalles Lakes north of Mt. Rainier (purple squares) [Reiners et al., 2000]. Abbreviations: ol (olivine), opx (orthopyroxene), cpx (clinopyroxene), plag (plagioclase), ilm (ilmenite), mt (magnetite).

Figure 9.

(a) ɛHf versus Nb/Nb* (niobium anomaly) for BSM and Mt. Baker basalts, compared to the ranges defined by Hawaiian shield and postshield basalts (green and pink lines, respectively) with >8 wt % MgO. Nb/Nb* calculated as 2(Nbsample/NbPM)/(Basample/BaPM + Lasample/LaPM) [Verma, 2009] where PM refers to Primitive Mantle. Hawaii data were obtained from the GEOROC database accessed in May 2012 ( Also shown are values for average depleted mantle (DM, black square) [Salters and Stracke, 2004] and average N-MORB (gray square) and Primitive Mantle (PM, black square with white +) of Hofmann et al. [1988]. (b) Ba/La versus 208Pb/204Pb for BSM and Mt. Baker basalts. (c) N-MORB normalized [Sun and McDonough, 1989] extended trace element diagram comparing Salal Glacier (orange) and Bridge River (red) primitive basalts (molar Mg/[Mg+Fe2+] > 0.60) to alkalic postshield basalts from Mauna Kea (light blue) [Hanano et al., 2010], Simcoe volcanic field (dark blue) [Battleground Lake sample of Jones, 2000], and Anahim volcanic belt (green) [sample 2278 of Charland et al., 1995].

[26] The five basalts comprising Suite 2 (BRC03–4, BRC01–3, MM01-1, MB1, and Lib21) have trace element abundances that contrast with other basalts at their respective volcanic centers, including significantly higher LILE (La/Yb)N and Ba/Nb, and lower HFSE (Figures 6b–6e). These samples are excluded from the following discussion of mantle source characteristics but are revisited later in the context of crustal assimilation.

6. Discussion

6.1. Mantle Source Characteristics

6.1.1. Temperatures and Pressures

[27] The BSM basalts segregated from their mantle source at significantly higher pressures and temperatures than the Mt. Baker basalts (Figure 10). Liquidus pressures and temperatures (i.e., mantle potential temperatures) were calculated for the two most primitive basalts at each BSM volcanic center (Table 3) using the olivine-liquid geothermometer and silica activity geobarometer of Putirka [2008]. Whole-rock data [Green and Sinha, 2005] were first adjusted into equilibrium with Fo90 mantle by incremental olivine addition assuming Fe3+/∑Fe = 0.15. Amounts of olivine added range from 11.5% (SG16) to 17.7% (BRC09). Primary biotite and amphibole in the groundmass of some primitive BSM samples attest to the presence of water, so pressures and temperatures were calculated for several possible H2O contents (listed in Table 3). The plagioclase-liquid hygrometer of Lange et al. [2009] applied to the most primitive Salal Glacier basalt gives ∼1 wt % H2O at the liquidus assuming plagioclase saturation at P = 100 MPa and maximum An60 in plagioclase cores [Lawrence, 1979]. At this water content, pressures calculated for primitive BSM basalts correspond to depths ranging from ∼70 km (Mt. Meager) to ∼105 km (Bridge River) (Figure 10). Decreasing melt SiO2 with increasing pressure [Longhi, 2002] is consistent with the P-T data. Calculated mantle potential temperatures are ∼100–200°C higher than predicted for the Cascade subarc mantle wedge [Syracuse et al., 2010], and similar to those of average MORB (1454°C) [Putirka, 2008]. Intraplate basalts of the western Basin and Range province give a broadly similar P-T range (60–90 km, 1350°C−1450°C) as the BSM basalts [Lee et al., 2009]. Intraplate basalts in the Cascade-Columbia transect have lower maximum segregation depths (75–80 km) but similar maximum temperatures (∼1460°C), although anhydrous conditions were assumed [Leeman et al., 2005]. Simcoe intraplate basalts record P-T conditions similar to the BSM basalts (max ∼100 km, 1500°C) [Leeman et al., 2005].

Figure 10.

Pressure versus temperature plot illustrating the conditions at which the BSM and Mt. Baker magmas segregated from the mantle. P and T (from Table 3) were calculated using the silica activity geobarometer and olivine-liquid geothermometer calibrations of Putirka [2008] for 1% and 2% dissolved water (BSM basalts) or for the specific H2O content given (Mt. Baker basalts). Standard estimates of error are 43°C and 0.29 GPa [Putirka, 2008].

Table 3. Liquidus pressures and temperatures
H2O (wt.%)
  1. a

    P-T data for Mt. Baker (from Mullen and McCallum, submitted manuscript, 2013) are calculated only at the specific water content listed for each sample.

Bridge River
Salal Glacier
Mt. Meager
a Mt. BakerP (GPa)T (°C)H2O (wt.%)   

[28] For Mt. Baker basalts, liquidus water contents are 1.5 to 3.7 wt % (Mullen and McCallum, submitted manuscript, 2013) and mantle potential temperatures are ∼1273°C to 134°C (Figure 10, Table 3), within the range for the subarc mantle wedge [Syracuse et al., 2010]. Mantle segregation depths are ∼35 to 52 km, i.e., ranging from the Moho to just above the hot core of the mantle wedge. The shallower depths recorded by the Mt. Baker basalts are consistent with trace element modeling (below) that indicates residual garnet for the BSM basalts but not Mt. Baker.

6.1.2. Mantle Isotopic Characteristics

[29] Primitive BSM basalts (Mg/[Mg+Fe2+] > 0.6) have isotope ratios that define a narrow range, consistent with a common mantle source and differentiation dominated by fractional crystallization. Pb isotope ratios overlap with Explorer and northern Juan de Fuca MORB [Cousens et al., 1995; B. Cousens, unpublished data 2007], Chilcotin plateau basalts [Bevier, 1983], and the least radiogenic samples from the Anahim volcanic belt [Bevier, 1989] (Figure 11). The isotopic similarity among these volcanic provinces confirms that the northwestern margin of North America is underlain by upper mantle that is relatively depleted and generally similar to northeastern Pacific mantle [Cousens and Bevier, 1995; Bevier, 1989].

Figure 11.

Plot of 208Pb/204Pb versus 206Pb/204Pb comparing Pb isotope ratios for the BSM and Mt. Baker basalts (symbols as in Figure 9) to other basalts from the northeastern Pacific and southwestern British Columbia: Anahim volcanic belt (purple diamonds; Bevier [1989]); Chilcotin Plateau (black diamonds; Bevier [1983]); Wells Gray-Clearwater volcanic field (blue diamonds; Hickson [1987]); and Tuzo Wilson volcanic field [Allan et al., 1993]. Other data and references as in Figure 3. Note that all isotope data are normalized to the same standard values as described in main text.

[30] Although the BSM basalts have 208Pb/204Pb similar to local MORBs at a given 206Pb/204Pb, 207Pb/204Pb is slightly higher (Figure 3). Relatively high 207Pb/204Pb could be interpreted as reflecting subducting sediment input, but this should increase 208Pb/204Pb along with 207Pb/204Pb, and the BSM basalts overlap with MORB in 208Pb/204Pb.

[31] High 207Pb/204Pb relative to 208Pb/204Pb may instead indicate a higher time-integrated U/Th in the BSM source than in the MORB sources. Enrichment of the BSM mantle source in U relative to Th at some time in the past could be accomplished through addition of fluid or melt components derived from subducting sediment and/or oceanic crust, since U is slightly more incompatible than Th during dehydration and melting [Brenan et al., 1995; Kessel et al., 2005; Hermann and Rubatto, 2009]. However, this situation would result in the presence of a subduction signature in the BSM mantle source, which is not observed.

[32] A more plausible explanation may be melting of the BSM mantle source in the presence of residual garnet at some time in the past. Unlike other typical mantle minerals, which do not fractionate U and Th appreciably, U is more compatible in garnet than Th [Beattie, 1993; LaTourrette et al., 1993]. The BSM basalts also plot at the lower margin of the Hf-Nd mantle array (Figure 4b), consistent with the isotopic evolution of mantle that produced melts within the garnet stability field [Carlson and Nowell, 2001].

[33] Because BSM basalts have lower ɛHf (∼3 epsilon units) than Mt. Baker but similar ɛNd (Figure 4b), two distinct mantle sources are required. Partitioning experiments show that Hf is preferentially retained in the subducting slab relative to Nd, most effectively during slab dehydration but also during slab melting [e.g., Kessel et al., 2005; Hermann and Rubatto, 2009]. High Nd/Hf in the subduction component is further enhanced by the preexisting negative HFSE anomalies that characterize Cascadia sediment [Carpentier et al., 2013; Prytulak et al., 2006]. As a consequence, addition of a sediment component to the BSM mantle source generates mixing curves that extend towards lower ɛNd values but with smaller changes in ɛHf. Most importantly, mixing trajectories extend away from the Mt. Baker data. Thus subduction input cannot account for the Hf isotopic distinction between the BSM and Mt. Baker basalts; the difference is instead a primary feature of their respective mantle sources. Since mantle Hf isotope ratios can be affected by both fluids and melts derived from the slab (Figure 4b), Hf isotopes do not always directly record the isotopic composition of the mantle as is commonly assumed.

6.1.3. Mantle Source Fertility

[34] Zr/Nb in basalts provides a useful indicator of mantle source fertility because this ratio is minimally affected by subduction input or fractional crystallization (Figure 8b). Although Zr/Nb is controlled to some extent by melt fraction, the Zr/Nb range defined by melts of average depleted mantle does not overlap the range for melts of more enriched mantle tapped by ocean island basalts. Mt. Baker basalts have Zr/Nb consistent with ∼10% partial melting of average depleted mantle with an additional slab component (Figure 8b). Zr/Nb is too low in the primitive BSM basalts to be produced from the same mantle source as Mt. Baker, requiring a more incompatible element enriched mantle source. The relatively high Nb contents of the BSM basalts also indicate a source relatively enriched in incompatible elements (Figure 8a), as do high Na2O and TiO2 [Prytulak and Elliott, 2007].

6.1.4. Assessment of Subduction Input

[35] For primitive samples at Salal Glacier and Bridge River, Ba/Nb values lie within the range of Hawaiian basalts and coincide with melting curves for enriched mantle (at ∼2–5% partial melt) (Figure 8), pointing to the likelihood that a slab-derived component was not present in the mantle source. The absence of slab input is supported by the absence of negative Ta-Nb anomalies (Figure 9a) that are observed at Mt. Baker and in other calc-alkaline Cascade arc basalts [e.g., Schmidt et al., 2008]. The absence of slab input is further supported by the similarity of Salal Glacier and Bridge River primitive basalts to Mauna Kea postshield alkalic basalts [Hanano et al., 2010], which sample the same mantle source as shield lavas, that is, a composition comparable to the PREMA, or C, mantle component [Nobre Silva et al., 2013] (Figure 9c). The only major difference is in Pb, which is deficient in Hawaii (a ubiquitous feature of oceanic basalts [Hofmann, 1997]) but shows small positive spikes at Bridge River and Salal Glacier. The Pb spikes are successfully modeled without subduction input (see later). However, we cannot definitively rule out the presence of a very small subduction component in the mantle source. An ancient subduction component may have been added to the source in the past, or the primary alkalic magmas may have acquired a small subduction component during migration through the mantle.

[36] In contrast to Salal Glacier and Bridge River, even the most primitive Mt. Meager basalts have slightly elevated Ba and Ba/Nb relative to mantle melting curves and Hawaiian basalts (Figure 8), as well as small negative Nb anomalies (Figure 9a), all of which point to subduction input (although significantly less than at Mt. Baker). Zr/Nb in the two of the three primitive Mt. Meager basalts is the same as at Salal Glacier and Bridge River, indicating similar mantle sources. In the other Mt. Meager sample (MM08), higher Zr/Nb and lower Nb is consistent with a mantle source that is transitional between the Mt. Baker mantle source and that of the other BSM basalts. An intermediate Hf isotopic composition for MM08 (Figure 4b) supports this conclusion.

[37] Mt. Meager basalts have the lowest 206Pb/204Pb and 208Pb/204Pb of all the BSM volcanic centers; sample MM08 has the lowest 208Pb/204Pb, 206Pb/204Pb, and (La/Yb)N, coupled with the highest Ba/La (Figure 9b). These characteristics are not consistent with addition of a subducting sediment component to the mantle source and may instead reflect the influence of a fluid derived from altered oceanic crust (AOC). AOC fluid input can increase LILE in the mantle source without affecting LREE [Kessel et al., 2005], and since recent AOC has MORB-like Pb isotope ratios, it is capable of “pulling” Pb isotope ratios of the mantle source to lower values. Mt. Meager also has similar ɛNd to Bridge River and Salal Glacier but slightly higher 87Sr/86Sr (Figure 4a), consistent with the involvement of AOC that acquires high 87Sr/86Sr with minimal change in ɛNd during progressive seafloor alteration [Staudigel et al., 1995].

6.1.5. Trace Element Modeling

[38] Mantle melt fractions and residual mantle mineral assemblages were determined by modeling the abundances of 28 trace elements in the two most primitive basalts at each BSM center (BRC09 and BRC10 at Bridge River, MM04 and MM02 at Mt. Meager, SG10 and SG16 at Salal Glacier). We also modeled MM08 at Mt. Meager because it may have a slightly more depleted mantle source.

[39] The model is based upon the mass balance equation for equilibrium melting, math formula [Shaw, 1970], where math formula is the concentration of trace element (i) in the liquid (L), math formula is the initial concentration of trace element (i), math formula is melt fraction, and math formula is the bulk partition coefficient, defined as crystalline assemblage/melt. The model does not require a priori knowledge of initial mantle mineral assemblages but does require initial trace element abundances. The primitive mantle composition of Sun and McDonough [1989] was used as the source for all BSM samples, and a mixture of 50% primitive mantle and 50% depleted mantle [Salters and Stracke, 2004] was also tested for MM08. Distribution coefficients used in the calculations are listed in Table 4. Least-squares minimization was used to generate best fit models for the basalts by varying the mantle mineral modes and melt fractions (see the caption of Figure 12). Note that the substitution of fractional melting in our model results in negligible change to model outcomes. Melt fractions are within 0.5% and residual mantle modal abundances change by less than a few percent, with overall residual mineral assemblages remaining identical.

Table 4. Partition Coefficients (Mineral/Melt)a
  1. a

    Abbreviations: cpx (clinopyroxene), opx (orthopyroxene), oliv (olivine), sp (spinel), gar (garnet), amph (amphibole), phl (phlogopite), mt (magnetite), plag (plagioclase).

  2. Data sources: 1Adam and Green [2006]; 2Halliday et al. [1995] compilation; 3Donnelly et al. [2004] compilation; 4Hauri et al. [1994]; 5Hart and Dunn [1993]; 6Gaetani [2004]; 7Elkins et al. [2008]; 8McDade et al. [2003]; 9Green et al. [2000]; 10Kennedy et al. [1993]; 11Beattie [1993];12Beattie [1994]; 13Canil and Fedortchouk [2001]; 14Salters and Longhi [1999]; 15Abraham et al. [2005] compilation; 16Chazot et al. [1996]; 17Horn et al. [1994]; 18Nagasawa et al. [1980]; 19McKenzie and O'Nions [1991]; 20Klemme et al. [2006]; 21Claeson and Meurer [2004] compilation; 22Dunn and Sen [1994]; 23Tepley et al. [2010]; 24LaTourrette et al. [1995]; 25Interpolated from neighboring elements.

Er0.69250.058250.008730.01193.6 150.57240.0010250.07250.014521
Figure 12.

Best fit trace element solutions for four of the most primitive BSM basalts, shown on N-MORB normalized extended element diagrams and accompanying inset chondrite-normalized REE diagrams. Actual data are shown with colored lines and symbols; modeling solutions shown as heavy black lines with black squares. Each BSM sample has been adjusted into equilibrium with Fo90 mantle using olivine/melt partition coefficients from Table 4. The mantle source composition used in the model for all BSM basalts (PM of Sun and McDonough [1989]) is shown as a thin black line in each panel; DM (source used for Mt. Baker) is also shown for reference [Salters and Stracke, 2004]. The Generalized Reduced Gradient (GRG2) nonlinear optimization code in Microsoft Excel Solver was used to obtain the best fit for each basalt by minimizing the sum of squares of residuals for 28 trace elements, i.e., math formula. The denominator in the equation normalizes the concentrations of the elements so that each trace element has an equivalent impact on the solution regardless of its absolute concentration. Best fit melt fractions (FL) and residual mantle mineral modes are given in lower right corner of each N-MORB-normalized panel. Abbreviations: ol (olivine), opx (orthopyroxene), cpx (clinopyroxene), gar (garnet). Modeling Results

[40] Representative best fit trace element solutions are shown in Figure 12. Melt fractions are 2–4% for Salal Glacier, 4–5% for Mt. Meager, and 7–8% for Bridge River, all with residual garnet lherzolite. Lower melt fractions for Salal Glacier basalts are consistent with their higher alkali element abundances. For Mt. Meager sample MM08, a primitive mantle source indicates 8% partial melt and the mixed PM-DM source gives 4%. The latter result is preferred because it is more consistent with the results for other Mt. Meager samples. Residual garnet in all samples is consistent with the pressures of melting (2–3 GPa) calculated for the BSM basalts, as garnet is stable at the solidus of hydrous mantle at pressures above 1.6 GPa [Gaetani and Grove, 1998].

[41] Results of similar modeling for Mt. Baker basalts, using a depleted mantle source, indicate 5–12% partial melting of depleted lherzolite or harzburgite. Best fit solutions require overprinting by a subduction component consisting of AOC fluid, AOC melt, and sediment melt (Mullen and McCallum, submitted manuscript, 2013). No residual garnet is present in the Mt. Baker source, consistent with calculated melt segregation pressures (1–1.5 GPa) and with the lower (Dy/Yb)N, and higher Yb and Sc contents of the Mt. Baker samples (24–33 ppm Sc) (Mullen and McCallum, submitted manuscript, 2013) as compared with the BSM basalts (18–24 ppm Sc; Table 2).

6.4. Nonprimitive BSM Basalts: Crustal Contamination or Subduction Input?

[42] Although we make the case above that the primitive Salal Glacier and Bridge River basalts essentially lack a subduction component, some of the nonprimitive BSM basalts have geochemical characteristics that could be interpreted as an “arc signature.” Do these reflect subduction input that is not displayed by more primitive samples?

[43] BSM samples with Mg/(Mg + Fe2+) < 0.60 are subdivided into two suites based upon isotopic and trace element compositions: Suite 1 has isotope and trace element ratios similar to primitive BSM basalts, indicating minimal crustal contamination and differentiation processes dominated by fractional crystallization. Relative to the most primitive basalts, Suite 2 has high Sr-Pb isotope ratios, (La/Yb)N, and incompatible element abundances, coupled with low Nd-Hf isotope ratios and HFSE abundances.

[44] Pearce element ratio diagrams [Russell and Nicholls, 1988] show that the Suite 1 basalts are consistent with fractionation of olivine + plagioclase (± minor clinopyroxene) from parental magmas that were similar to the most primitive basalts at each volcanic center. The crystallizing assemblages are consistent with the presence of the same minerals as phenocryst phases.

[45] For Suite 2 samples, trace element abundances and Pb isotope ratios are nearly indistinguishable from the most primitive Mt. Baker basalts (Figures 3, 6e, and 8), pointing to the possibility that they may record input from the subducting slab as does Mt. Baker. However, Suite 2 has significantly higher 87Sr/86Sr and lower ɛNd than the most primitive Mt. Baker and BSM basalts (Figure 4a). The lower MgO contents of Suite 2 lavas relative to the most primitive lavas are consistent with crustal contamination (Figure 5b). Furthermore, Suite 2 contains the only two BSM basalts with xenocrysts (BRC01–3, BRC03–4).

[46] The GVB crustal basement is a collage of Paleozoic and Mesozoic accreted terranes [Monger et al., 1982]. At depths greater than ∼10 km, the GVB is underlain by the composite Wrangellia–Harrison terranes [Mullen, 2011; Miller et al., 2009; Monger and Price, 2000]. With the exception of the Mt. Baker region, the terranes are intruded extensively by Jurassic to Cretaceous granitoids of the Coast Plutonic Complex, the largest composite batholith in North America [Barker and Arth, 1984; Friedman et al., 1995; Cui and Russell, 1995a, 1995b]. Because the crust is relatively young and Cascadia subducting sediment is mainly terrigenous [Carpentier et al., 2013; Prytulak et al., 2006], the isotopic effects of crustal assimilation are similar in many respects to the effects of subducting sediment input. However, sediment input cannot account for the 87Sr/86Sr versus Sr systematics of the Suite 2 lavas (Figure 13a). Assimilation-fractional crystallization (AFC) modeling [DePaolo, 1981] using a granodioritic assimilant from the Coast Plutonic Complex can reproduce the Suite 2 trace element and isotopic data, but the volume of assimilant required (>20%) would increase the SiO2 content beyond the range of the Suite 2 samples (Figure 5a). Assimilation that takes place in the deep crust, where the country rock is mafic, can minimize changes to the major element abundances of the original basaltic magma [Reiners et al., 1995, 1996]. AFC models in which the assimilant is a gabbro from the lower crustal section of the Bonanza arc (Westcoast crystalline complex) of Wrangellia [DeBari et al., 1999] provide good fits to Suite 2 trace element and isotope data with ∼15% gabbro assimilated at Bridge River and ∼21% at Mt. Meager (Figure 13). Modeling parameters and results are provided in Table 5. Fractionating mineral phases (olivine, clinopyroxene, orthopyroxene, and minor magnetite) are consistent with experimental results for partial melting of mafic compositions under lower crustal conditions [Rapp, 1995; Rapp and Watson, 1995]. Because the assimilant has a low SiO2 content (∼45 wt %) and the fractionating mineral assemblages have bulk SiO2 contents similar to the basalts, the final magmas maintain an overall basaltic composition in the magmas.

Figure 13.

AFC modeling results for (a) 87Sr/86Sr versus Sr; (b) ɛNd versus 87Sr/86Sr; (c) Zr/Nb versus Ba/Nb. Heavy green curves with triangles and squares (AFC1 and AFC2, respectively) are the best fit assimilation-fractional crystallization pathways for Suite 2 samples BRC03–4 and MM01-1, respectively, using a gabbroic assimilant (from Table 5). The large green triangle and square are the gabbro compositions used as assimilants in AFC 1 and 2, respectively. Orange and blue curves in Figure 13a are slab fluid/melt addition curves, calculated as described in Fig. 4 caption. Heavy black curves in Figure 13c are from Figure 8b. Data for the Coast Plutonic Complex shown as small blue triangles and light blue field [Friedman et al., 1995; Cui and Russell, 1995a, 1995b]. Sources of other data shown are given in Figure 4a caption.

Table 5. AFC Modeling Parameters and Results
CompositionAFC 1: Bridge RiverAFC 2: Mt. Meager
Sample Modeled: BRC03–4Sample Modeled: MM01-1
Initial Magma:Assimilant:Initial Magma:Assimilant:
  1. a

    Trace element and isotope data for initial magmas are from Table 1; major element data [from Green and Sinha, 2005] are corrected into equilibrium with Fo90 mantle.

  2. b

    Major and trace element data for gabbro assimilant are from DeBari et al. [1999] for sample 91-17 of the Westcoast Crystalline Complex except Nd (interpolated); isotope ratios selected from within the range defined by the Coast Plutonic Complex [Cui and Russell, 1995b].

  3. c

    mass assimilated/mass crystallized.

  4. d

    fraction liquid remaining.

  5. e

    Fraction of each mineral phase removed from magma; sum is equal to (1 − FL).

  6. f

    Abbreviations: oliv (olivine), cpx (clinopyroxene), opx (orthopyroxene), mt (magnetite).

SiO2 (wt %)45.644.648.644.6
Sr (ppm)400401467401
AFC results    
r c 0.90 0.89
FLd 0.83 0.77
olive, f 0.05  
cpx 0.05 0.10
opx 0.05 0.10
mt 0.02 0.03

6.5. Relationship Between Tectonics and Volcanism

[47] An incompatible element-enriched, garnet-bearing mantle source essentially free of subduction input, coupled with relatively high mantle melting temperatures and pressures, is consistent with decompression melting of an upwelling asthenosphere source for the primitive BSM basalts. Upwelling mantle is potentially consistent with a slab edge effect as proposed by Lawrence et al. [1984] for Salal Glacier basalts. Seismic anisotropy measurements reveal toroidal mantle flow around the descending edges of subducted plates that are undergoing rollback, thereby drawing external mantle (subslab) into the mantle wedge [Long and Silver, 2008]. In other arcs, influx of external mantle has been implicated in the genesis of lavas that are atypical for an arc setting [e.g., Leat et al., 2004; Smith et al., 2001; Ferrari et al., 2001]. Slab rollback is occurring in the Cascade arc [Schellart, 2007], but the limited mantle anisotropy measurements in the GVB are inconclusive as to mantle wedge flow patterns [Currie et al., 2004]. Toroidal mantle flow has been documented at the southern Juan de Fuca plate edge [Zandt and Humphreys, 2008], yet alkalic basalts are not present [Hildreth, 2007] indicating that the two phenomena are not necessarily interrelated.

[48] A slab edge origin may be improbable for the BSM volcanic centers in light of recent seismic tomography, which indicates the northernmost slab edge in the Cascades (placed at the northernmost limit of the Explorer plate) is located farther north than the BSM volcanic centers [Mercier et al., 2009; Audet et al., 2008]. Toroidal mantle flow has been proposed for the northern Explorer plate edge [Audet et al., 2008] and could be responsible for the alkalic basalts of the 500 km long Anahim volcanic belt, which defines an east-west trend nearly orthogonal to, and north of, the GVB. This interpretation is consistent with that of Thorkelson et al. [2011] who proposed that Anahim magmatism is related to mantle upwelling along the thermally eroding plate margins of the Northern Cordilleran slab window. However, eruption ages in the Anahim volcanic belt define an easterly time progression that has been attributed to a hotspot [Bevier, 1989], and tomographic results are consistent with either interpretation [Mercier et al., 2009].

[49] The BSM alkalic basalts may be related to mantle upwelling at the boundary between the Juan de Fuca and Explorer plates, as illustrated schematically in Figure 14. The northern segment of the Juan de Fuca plate has had a complex tectonic history; about 4 Myr ago, the northernmost portion of the Juan de Fuca plate separated along the Nootka fault zone to form the independent Explorer microplate [Riddihough, 1984] (Figure 1). Although convergence has ceased at the northern edge of the microplate, the southernmost part of the microplate continues to subduct slowly [Braunmiller and Nabelek, 2002], and the entire Explorer region is a zone of strong shear deformation [Dziak, 2006]. The offshore segment of the Nootka fault shows left-lateral motion along a rupture and the onshore extension of the fault is marked by thinning and deformation of the subducting plate [Audet et al., 2008]. Seismic data indicate the Explorer plate currently has a shallower dip than the Juan de Fuca plate, which may manifest itself in a near-vertical gap between the plates (Figure 14). The BSM volcanic centers lie on, or just south of, the Nootka fault zone as extrapolated to the northeast (Figure 1a). We suggest that thinning, deformation, and possible rupture of the subducted Explorer plate fragment may provide a pathway for asthenospheric upwelling accompanied by decompression melting. Farther east along the projected trace of the Nootka fault, the Wells Gray-Clearwater volcanic field and Chilcotin basalts have been similarly attributed to enriched asthenosphere upwelling through a gap along the fault [Madsen et al., 2006; Sluggett, 2008; Thorkelson et al., 2011].

Figure 14.

Schematic representation of plate configuration at the northern end of the Cascade arc based on a model of Riddihough [1984]. The Explorer plate detached from the Juan de Fuca Plate along the Nootka fault zone 3 to 4 Myr ago as it became younger, hotter, and more buoyant at the trench. The thick dashed black line indicates the surface trace of the Nootka fault. Convergence of the Explorer plate with North America has now nearly ceased. The vertical window formed between the Explorer and Juan de Fuca plates may promote upwelling of deep, hot mantle (large orange arrow) at the edge of the currently subducting plate. Decompression melting of this mantle accounts for the presence of hot alkalic basalts essentially free of a subduction signature (red, orange, and yellow triangles along Nootka fault zone for each of the BSM volcanic centers).

[50] Seismic tomography is inconclusive as to whether the Nootka fault is “leaky” or whether continuity is maintained at depth between the Explorer and Juan de Fuca plates [Mercier et al., 2009]. However, as the Explorer plate is situated at the southern edge of a slab window, it is subject to progressive thermal and physical degradation that would facilitate passage of mantle melts from below [Thorkelson et al., 2011; Thorkelson and Breitsprecher, 2005]. In an analogous situation in the Mexican arc, seismic anisotropy measurements are consistent with plate separation. Faults separate the subducting Cocos plate into several segments, and each subducts at a different angle, resulting in a scissors-like effect in which gaps between the plates allow for mantle upwelling through toroidal flow [Stubailo et al., 2012].

7. Summary and Conclusions

[51] Alkalic basalts at the Bridge River, Salal Glacier, and Mt. Meager volcanic centers (BSM volcanic centers) of the Canadian segment of the Cascade arc, known as the Garibaldi volcanic belt, have intraplate characteristics that contrast with typical calc-alkaline mafic Cascade arc lavas. New high precision Sr-Nd-Hf-Pb isotope ratios and trace element abundances reveal that the most primitive basalts at Salal Glacier and Bridge River are essentially free of components derived from the subducting slab. The apparent trace element “arc signature” exhibited by several more evolved BSM basalts is more likely a consequence of assimilation of mafic deep crust rather than slab input. At Mt. Meager, however, primitive basalts may include a small amount of fluid derived from subducted altered oceanic crust.

[52] The mantle source of the BSM basalts is deeper, hotter, and isotopically distinct from the source of calc-alkaline basalts from Mt. Baker and throughout the Cascade arc. The BSM mantle source is also more enriched in incompatible elements than the depleted mantle wedge tapped by calc-alkaline Cascade arc basalts, and similar to ocean island basalt sources. Similar trace element abundances among the BSM and Anahim alkalic basalts, and those in the Cascade-Columbia transect and north of Mt. Rainier (Figures 8 and 9c), indicate mantle sources similarly enriched in incompatible elements.

[53] BSM and Cascade-Columbia intraplate lavas have been previously attributed to enriched mantle domains associated with the base of an accreted terrane [Schmidt et al., 2008]. We consider this hypothesis unlikely for the BSM volcanic centers for two reasons. First, the accreted terranes beneath the BSM centers and the Cascade-Columbia transect are different (Wrangellia and Siletzia, respectively), and second, Mt. Baker and the BSM share the same accreted terrane at depth yet the state of mantle source enrichment differs substantially.

[54] Although major and trace element data require an enriched mantle source for the BSM basalts, isotopic data provide evidence for long-term mantle depletion. Pb isotope ratios of the BSM basalts are broadly similar to oceanic and intraplate basalts of the northeastern Pacific (Figure 11), indicating that isotopically depleted upper mantle of common origin is regionally widespread, albeit with small isotopic heterogeneities.

[55] With isotopic data consistent with long-term depletion, incompatible-element enrichment of the BSM mantle source must have occurred relatively recently. Recent mantle enrichment has been proposed for numerous other cases of isotopically depleted alkalic basalts [Roden and Murthy, 1985, and references therein], including those from the Tuzo Wilson volcanic field [Allan et al., 1993] (Figure 11) and the Bowie Seamount in the Gulf of Alaska [Cousens, 1988]. The BSM volcanic centers are located along, and just south of, the projected trace of the Nootka fault zone, which separates the subducting Juan de Fuca plate from the Explorer plate fragment. We attribute the BSM basalts to upwelling asthenosphere through a gap along the fault, which undergoes decompression melting to generate alkalic basalts that are free of subduction input yet located in an arc setting.


[56] We thank Bruno Kieffer for assistance with TIMS analyses, Vivian Lai for help with trace element analyses, Jane Barling, Kathy Gordon, and Liyan Xing for assistance with MC-ICP-MS analyses, and Ines Nobre Silva for instruction in the clean laboratory. We are grateful to Marion Carpentier for processing and analyzing eight samples for trace elements and five for isotopes. Derek Thorkelson, Martin Streck, and Richard Carlson provided constructive and thoughtful reviews. Insightful discussions with Kelly Russell have been much appreciated. We are particularly grateful to Stewart McCallum for detailed reviews, discussions, and advice that have significantly improved the manuscript. This research was funded by an NSERC Discovery Grant to D. Weis.


  1. 1

    Additional supporting information may be found in the online version of this article.