You are on page 1of 23

Bull Volcanol (2013) 75:730 DOI 10.

1007/s00445-013-0730-5

RESEARCH ARTICLE

Petrological and geochemical constraints on the origin of adakites in the Garibaldi Volcanic Complex, southwestern British Columbia, Canada
Julie Fillmore & Ian M. Coulson

Received: 15 October 2010 / Accepted: 30 April 2013 # Springer-Verlag Berlin Heidelberg 2013

Abstract The Garibaldi Volcanic Complex (GVC) is located in southwestern British Columbia and comprises two related but distinct volcanic fields: the Garibaldi Lake and the Mount Garibaldi volcanic fields. The rocks of the GVC range from basalt to rhyolite, and analyses of samples from both fields distinguish these as adakites. The GVC magmas have high Sr/Y, Mg#, and Al2O3; low K2O/Na2O; and fractionated rare earth element compositions. Models of adakite genesis fall into two main groups: slab melting and non-slab melting. Adakites generated by slab melting commonly occur from young subducting crust (25 Ma) and are felsic partial melts of the subducting slab that interact with the mantle wedge during ascent. Non-slab melting models vary widely and include basalt fractionation, assimilation, fractional crystallization processes and partial melting of mafic lower crust. Data from the GVC are too limited to fully elucidate the mechanisms of adakite genesis; however, the petrographical and geochemical characteristics of the GVC rocks in this study do not refute an origin by slab partial melts. Variations in trace elements that reflect non-adakitic values (e.g., low La, low Cr) are likely the result of magma mixing at shallow depths within the magma reservoirs of each center, for which there is mineralogical and textural evidence. The adakite rocks of the GVC share geochemical traits akin to both low-SiO2 adakite (LSA) and high-SiO2 adakite (HSA) groups, though additional data are needed to investigate whether LSA- or HSA-type dominates within the GVC, and by extension, which should be the preferred model of adakite genesis. Keywords Adakite . Petrogenesis . Subduction-related magmas . Garibaldi Volcanic Complex
Editorial responsibility: M. A. Clynne J. Fillmore (*) : I. M. Coulson Solid Earth Studies Laboratory, Department of Geology, University of Regina, Regina, SK S4S 0A2, Canada e-mail: fillmorj@uregina.ca

Introduction Adakites are a group of intermediate to felsic igneous rocks formed in subduction zones involving relatively young, hot oceanic lithosphere (25 Ma). Since the introduction of the term adakite by Defant and Drummond (1990), the petrogenesis of this suite of rocks has been the subject of controversy. Adakites are named after magnesian andesites first described by Kay (1978) from Adak Island in the Aleutians. These rocks are believed to be the result of partial melting of subducted ocean crust creating typically sodic slab melts. Partial melting of basalt to generate adakite is supported by experimental work (e.g., Rapp et al. 1999) and evidenced in natural rocks from subduction zones (e.g., Schiano et al. 1995). Adakites are characterized by specific geochemical features (Sr/Y 40 and La/Yb 20), and as studies into magma genesis continued, an ever widening group of different models was suggested. Consequently, a wide range of magma compositions from different tectonic environments have been classified as adakites. Models that explain the formation of adakites fall into two main groups: genesis from slab partial melts and magmas generated by various methods that can reproduce the distinctive adakite chemistry (which are not necessarily subduction related). Further ambiguity arises when a suite of rocks are classified as adakites (which may not follow the parameters outlined by Defant and Drummond 1990) and are subsequently found not to be related to slab melting. This leads to the conclusion that adakites are not slab melts and has provided the means of using adakite to characterize many different rock types under the one name. The Garibaldi Volcanic Complex (GVC) is comprised of two fields, the Mount Garibaldi Volcanic Field (MGVF) in the south and the Garibaldi Lake Volcanic Field (GLVF) in the north. The GVC is Quaternary in age and is located in southern British Columbia (Figs. 1 and 2). Geochemical attributes of andesite and dacite rocks that comprise the GVC suggest

730, Page 2 of 23 Fig. 1 Location map of the GVC in southwestern British Columbia, with plate boundaries highlighted and relative plate motions of the Juan de Fuca and Explorer plates indicated. Map modified after Hickson et al. (1999) and Madsen et al. (2006)

Bull Volcanol (2013) 75:730

Yukon Territory Alaska

NWT

North American Plate


Queen Charlotte Transform Fault

British Columbia

Alberta

Pacific Plate
Garibaldi volcanic belt
Whistler Explorer Plate
Nootka fault zone

GVC (see Fig. 2) Squamish

100 km

200

Juan de Fuca Plate

Washington

that they are the result of slab melting, under the definition put forth by Defant and Drummond (1990). Interactions between these melts and the overlying mantle wedge are also evident in their major and trace element compositions, a refinement made to the slab melt model by Martin et al. (2005). This study aims to: (1) present the first, complete whole-rock geochemistry for the MGVF as well as the first, complete rare earth element (REE) geochemistry of the intermediate rocks in the GLVF and (2) demonstrate that this chemistry identifies the rocks of the GVC as adakites.

Regional geology The Garibaldi Volcanic Belt (GVB) extends from the CanadaUSA border northward into British Columbia for approximately 140 km (Fig. 1; Sherrod and Smith 1990). The GVC lies within the southern portion of the GVB between the towns of Whistler and Squamish and comprises two fields: the GLVF in the north and the MGVF in the south (Figs. 1 and 2). The volcanic rocks from the GLVF and the MGVF range in composition from basalt to rhyolite and have been previously interpreted to be the result of hydrous melting of the mantle wedge above the Juan de Fuca Plate, which is subducting beneath the North American Plate, and subsequent fractionation at various depths during ascent (Green 1977, 1981, 1990; Green and Harry 1999;

Green and Sinha 2005; Green et al. 1988). Basaltic volcanism in the GVC is thought to be related to the decreased volatile content of the Juan de Fuca Plate and results in lower degrees of melting under higher pressures and temperatures (Green 2006). The decreased slab flux is attributed to both a northward decrease in plate movement and plate age (Riddihough 1981, 1984; Green 1990; Wilson 2002). The younger, more buoyant Explorer Plate separated from the Juan de Fuca Plate at 4 Ma (Wilson 2002; Audet et al. 2008), and this deviation may relate to an increase in Quaternary volcanism in the GVC. Subduction of the Explorer Plate beneath North America is slower than that of Juan de Fuca, and it has been suggested that the Explorer Plate is undergoing capture by the North American Plate (Audet et al. 2008). The difference in subduction rates has caused a region of extension and slab thinning along the Nootka Fault zone (Fig. 1), a transform fault that fractured in response to an interval of ridge propagation and reorientation (Riddihough 1984; Wilson 1988; Madsen et al. 2006) and created the Explorer and Juan de Fuca plates. Independent movement of the Explorer Plate northward relative to the northeasterly movement of the Juan de Fuca Plate suggests that the subducted portions of the plates have separated in addition to the oceanic portions (Madsen et al. 2006). This segmentation coupled with the relative subduction vectors of the plates has resulted in a change in mantle flow. Recent studies (Madsen et al. 2006;

Bull Volcanol (2013) 75:730

Page 3 of 23, 730

Fig. 2 Geology of the GLVF and the MGVF in the GVC. Sample locations are also shown. Map modified after Green (1977) and Green et al. (1988)

Audet et al. 2008) have determined and modelled upwelling of mantle material along the Nootka Fault, suggesting the potential for formation of a vertical slab window. Other studies (e.g., Thorkelson and Breitsprecher 2005; Ickert et

al. 2009) have found a link between the formation of slab windows or slab tears and adakitic volcanism, where asthenospheric upwelling provides the necessary heat flux to melt the slab edges and generate adakite magmas.

730, Page 4 of 23

Bull Volcanol (2013) 75:730

The MGVF and the GLVF sit unconformably on the Coast Crystalline Complex, which is a series of metamorphosed quartz diorite and granodiorite plutons of Cretaceous age. The MGVF is comprised of Mt. Garibaldi and its subsidiary vents, Dalton Dome and Atwell Peak, Opal Cone and the Ring Creek andesite flow, and Columnar Peak and the andesite flows of Paul Ridge. Recent activity in this field began at approximately 700 ka with the eruption of hornblende andesite flows atop pre-1,300 ka hornblende andesite and basaltic andesite. Increased volcanism occurred between 260 and 220 ka with the eruption of hornblendeorthopyroxene andesite at Columnar Peak, followed by hornblende orthopyroxene dacite flows and minor pyroclastic material from Mt. Garibaldi (Green et al. 1988). The composite dacite cone of Dalton Dome formed later but before the belt was overridden by glacial ice. Post-100 ka, dacitic pyroclastic flows were erupted from Atwell Peak atop the glacial ice as well as additional dacite flows from Dalton Dome. These flows and the west flank of Atwell Peak collapsed following glacial retreat (Mathews 1952, 1958; Green 1990). The most recent volcanism in the GVC was the eruption of the Ring Creek andesite flow from Opal Cone between 10.7 and 9.3 ka (Brooks and Friele 1992), which extends for some 17.5 km south around Paul Ridge and then west toward Squamish River (Fig. 2). Quaternary volcanic centers in the GLVF include Black Tusk, Cinder Cone, Clinker Peak, Mt. Price, and The Table, as well as the Cheakamus Valley Basalts which were erupted from an unknown centre. Timing and eruptive products from both fields have been summarized by previous authors (Green 1977, 1981, 1990; Green et al. 1988) and are briefly outlined below. The oldest activity in the GLVF was at Black Tusk and Mt. Price with episodic volcanism beginning at 1,300 ka. The rocks of Black Tusk are hornblende andesite and orthopyroxene andesite flows. The oldest rocks of Mt. Price are a series of hornblende andesite and andesite flows followed by the formation of the hornblendebiotite andesite satellite cone along Garibaldi Lake. Volcanism ended in the Mt. Price area with the eruption of the Barrier and Culliton Creek andesite flows from Clinker Peak at 100 ka. Activity at Cinder Cone began post 100 ka with the formation of a tuff ring and the eruption of the basaltic andesite of Desolation Valley, followed by the Helm Creek basalt flow. The Table formed at 100 ka, when hornblende andesite magma erupted beneath the Cordilleran Ice Sheet and melted its way upward to form a steep, flat topped tuya (Mathews 1951; Green 1981). The olivinebearing Cheakamus Valley Basalts were erupted post100 ka, and eruptions continued episodically to approximately 34 ka (Green 1977; Green et al. 1988).

Petrography MGVF Ring Creek andesite Four samples were collected from the Ring Creek andesite: two taken proximal to Opal Cone (09JF007, 09JF008) and two taken approximately 2 km from the flow terminus (10JF022, 10JF023; see Fig. 2). The mineralogy of the proximal Ring Creek andesite differs from that of the distal portion of the flow (first noted by Sivertz 1976), and hence, the petrography will be described separately. 1. Proximal Ring Creek andesite The andesite is porphyritic; main phenocrysts are plagioclase (15 %), followed by hornblende (10 %) and augite (2 %). Quartz occurs in trace amounts (0.1 %). Biotite (3 %) is present as rare large xenocrysts (Fig. 3a). Plagioclase occurs in two size populations; the larger phenocrysts are approximately 2 mm in size and the smaller less than 1 mm. The majority of the plagioclase crystals are subhedral, equant to tabular; more rarely, these form glomeroporphyritic aggregates. Several features are exhibited in plagioclase that includes sieve textures, resorption of grain margins and in some of the larger crystals, seritization. An equal proportion of plagioclase phenocrysts, however, are inclusion-free and pristine. Hornblende phenocrysts are second to plagioclase in abundance and range in size from less than 0.5 up to 1 mm. The majority of the crystals exhibit various disequilibrium textures including fibrous cores of clinopyroxene and destabilization rims of opaque oxides along the crystal margins (Fig. 3b). Biotite occurs as subhedral xenocrysts up to 1 mm in size with rare crystals up to 3 mm. The edges of biotite crystals are diffuse and poorly defined with rims showing alteration to a mass of fine-grained opaque minerals. The larger biotite xenocrysts are heavily embayed and have sieve-textured cores. With few exceptions, these xenocrysts display extensive replacement by opaque oxide phases; resorption of grain boundaries is also common. Augite occurs as prismatic to equant crystals up to 3.5 mm in size. The margins of the phenocrysts are resorbed and contain abundant inclusions of apatite and oxides. The crystals are greenish brown and not distinctly pleochroic. Only a few quartz crystals have been identified in this part of the flow. The phenocrysts are anhedral and less than 0.5 mm in size and exhibit resorption along the grain margins. The groundmass of the proximal Ring Creek andesite is approximately equal parts crystallites and brown glass. Plagioclase, augite, altered hornblende, and oxide

Bull Volcanol (2013) 75:730 Fig. 3 Thin section photomicrographs of the MGVF and GLVF rocks. a Biotite xenocryst from the proximal portion of the Ring Creek andesite. Cross-polarized light, FOV 5 mm. b Hornblende phenocrysts from the proximal portion of the Ring Creek andesite; note the dark, very fine-grained reaction rims along the crystal margins. Planepolarized light, FOV 3 mm. c Mafic xenolith comprised of plagioclase and orthopyroxene in the distal portion of the Ring Creek andesite. Cross-polarized light, FOV 3 mm. d Weak flow banding observed in the distal portion of the Ring Creek andesite. Note, lower left, the quartz xenocryst with a reaction rim of fine-grained augite. Cross-polarized light, FOV 5 mm. e, f General texture observed (cross-polarized light, FOV 3 mm) and concentrically zoned hornblende phenocrysts (cross-polarized light, FOV 3 mm) present within the Columnar Peak dacite. g Plagioclase-rich nature of the Paul Ridge andesite, with rounded olivine (lower right). Cross-polarized light, FOV 5 mm. h Olivine (second-order blue) being replaced by fibrous orthopyroxene (yellow) in Paul Ridge andesite sample. Crosspolarized light, FOV 3 mm. i Strongly altered biotite crystal with a rim of oxide phases in the Barrier andesite. Planepolarized light, FOV 3 mm. j Glomeroporphyritic hornblende in the Barrier andesite. Planepolarized light, FOV 3 mm. k Strongly resorbed and disaggregated orthopyroxene phenocryst in the Black Tusk andesite. Cross-polarized light, FOV 3 mm. l Coarser-grained crystal clot in the Black Tusk andesite. Cross-polarized light, FOV 3 mm

Page 5 of 23, 730

730, Page 6 of 23

Bull Volcanol (2013) 75:730

phases comprise the majority of the crystallites. Local flow banding is evident around the larger phenocrysts. 2. Distal Ring Creek andesite The distal Ring Creek andesite is less porphyritic than the proximal portion and contains approximately 20 % total phenocrysts. The mineralogy of the distal part of the flow differs from the proximal portion in that the only phenocryst phases present are plagioclase and augite. Xenoliths of mafic-intermediate cumulate inclusions that host orthopyroxene occur rarely. The inclusions are heavily corroded and partially melted (Fig. 3c). Plagioclase, again occurring in two size populations, is the most abundant phenocryst (15 %), followed by augite (5 %). The larger plagioclase crystals are up to 3.5 mm in size, the smaller approximately 1 mm; all crystals are equant to tabular and subhedral and display complex zonation and various degrees of resorption. Sieve-textured crystals are less common than the plagioclase in the proximal portion of the flow but occur mainly in the larger grains that also contain concentric trails of melt inclusions along their margins. Augite phenocrysts are smaller than plagioclase, commonly less than 1 mm in size. The crystals are equant and subhedral to euhedral; some grains display simple twinning as well as glomeroporphyritic aggregates with plagioclase. Rare, altered, orthopyroxene crystals (up to 2 mm in size) likely derived from the mafic-intermediate cumulate xenoliths are present but have been almost completely altered to chlorite and opaque oxides. Rare quartz is also present occurring as anhedral crystals that appear in disequilibrium with the surrounding melt, in exhibiting reaction rims of fine-grained, radiating clusters of augite (Fig. 3d). This is in contrast to the proximal portion of the flow where the quartz appears to be primary. The groundmass is predominantly crystallites of plagioclase with lesser amounts of augite and brown glass. The groundmass displays local, weakly developed flow foliation. Columnar Peak Four samples from the orthopyroxene-hornblende dacite of Columnar Peak (09JF009, 09JF010, 10JF019, and 10JF020) were examined as part of this study. This unit is described as a series of flows by Green (1977). The dacite is porphyritic with 10 to 15 % phenocrysts (Fig. 3e). Plagioclase is the most abundant phase (7 %), followed by hornblende (6 %) and orthopyroxene (<1 %). Plagioclase occurs in two size populations; the larger crystals are up to 4.5 mm in size, and the smaller crystals are 1.5 mm. The phenocrysts are subhedral to anhedral and are complexly zoned. A number of disequilibrium textures are observed in the large plagioclase phenocrysts

including sieve-textured cores, resorption along crystal edges, and several generations of concentric melt inclusions. These textures are less common in the smaller plagioclase crystals. Some of the larger grains exhibit fresh, clear overgrowths on the outer edges. Hornblende is considerably smaller than plagioclase, up to 1 mm in size and is subhedral to euhedral and equant to prismatic. The crystals are in disequilibrium with the surrounding melt; the phenocrysts are commonly embayed with some grains exhibiting sieve-textured cores. All the crystals have resorbed margins through destabilization rims comprised of opaque oxide phases. Weak concentric zoning is also observed in some of the larger crystals (Fig. 3f). Orthopyroxene crystals are quite small, less than 1 mm in size. The phenocrysts are yellowish, equant to prismatic, and mainly euhedral in form. The majority of the orthopyroxene grains are relatively fresh, with occasional large melt inclusions within. Some crystals occur as aggregates or are perhaps fragments of a single crystal separated by glass. The groundmass is comprised of approximately equal amounts of brown glass and crystallites of plagioclase and opaque oxides. Flow foliation is observed locally. Paul Ridge andesite Three samples of the Paul Ridge andesite (09JF011, 10JF017, 10JF018) were collected in this study. The andesite is described as a series of hornblende andesite flows by Green (1977); these are porphyritic, with approximately 20 % phenocrysts (Fig. 3g). Plagioclase phenocrysts are the most common (10 %), followed by orthopyroxene and olivine (5 % each, marginally higher amounts in 09JF011) with trace amounts of xenocrystic quartz (<0.5 %). The plagioclase crystals are tabular and can also form aggregates of up to 3 mm in size. The dominant crystal size is approximately 2 mm. Plagioclase is subhedral to euhedral and is complexly zoned with reverse and oscillatory zoning. Some of the phenocrysts exhibit uneven extinction, suggestive of strain deformation. The crystals display varying degrees of sieve-textures in the cores and resorption along the edges. Some of the phenocrysts have clear cores with concentric rings of melt inclusions near their margins. Orthopyroxene phenocrysts are markedly less abundant than plagioclase, occurring mainly in the glassy groundmass with some larger crystals. The relatively pristine orthopyroxene phenocrysts (up to 1 mm in size) are green and broken and exhibit discontinuous reverse zoning. Orthopyroxene also commonly forms reaction rims on olivine and as an interstitial phase to crystal aggregates. The majority of olivine crystals occur in a state of incipient replacement by orthopyroxene (Fig. 3h); unaltered olivine is rare. These replacement textures are slightly more common in 09JF011 than the other samples. Larger olivine crystals occur occasionally in aggregates with plagioclase; the olivine is anhedral and usually

Bull Volcanol (2013) 75:730

Page 7 of 23, 730

exhibits alteration to brown iddingsite along crystal edges and fractures. Embayments are also common in the crystals. A few quartz xenocrysts have been identified; the quartz is anhedral, usually exhibiting weakly uneven extinction. The quartz xenocrysts exhibit reaction rims and are surrounded by small augite crystals radiating outward. The groundmass is mainly brown glass with microlites of plagioclase and opaques. GLVF Barrier andesite Four samples (09JF004, 09JF005, 09JF006, 09JF012) were collected from the Barrier andesite lava flow along the northern shore of Garibaldi Lake. This flow is porphyritic, with 10 to 15 % phenocrysts. Plagioclase is the most abundant (10 %), followed by approximately equal amounts of hornblende and biotite (23 % each). Quartz phenocrysts occur rarely (<1 %). Plagioclase phenocrysts (up to 2 mm) form subhedral to anhedral, equant to tabular crystals that display complex zonation. Several grains contain concentric trails of melt inclusions along their outer margins; sieve textures are also present. Hornblende phenocrysts are smaller than plagioclase, approximately 1 mm in size and prismatic to bladed in habit. The crystal edges are diffuse and commonly show alteration to fine grained opaque minerals, possibly indicative of disequilibrium. Hornblende also occurs as rare glomeroporphyritic aggregates (Fig. 3j). Biotite is the largest phenocryst phase; up to 4 mm in size. The phenocrysts are equant, subhedral, and strongly pleochroic. The crystal margins are extensively altered and/or heavily corroded, resulting in a rim of opaque phases (Fig. 3i). Quartz phenocrysts are anhedral, approximately 1 mm in size and equant in shape. Where these occur, quartz crystals are surrounded by a reaction rim of finegrained, tabular, augite. The groundmass of the Barrier andesite is comprised principally of brown glass with plagioclase and rare hornblende crystallites. Rounded and irregularly shaped vesicles (2 %) also occur in the groundmass. Flow banding is locally present around the larger phenocrysts. Black Tusk Two samples (10JF015, 10JF016) of orthopyroxene andesite were collected from the west bluff of Black Tusk. The andesite is essentially aphyric, with rare (1 %) phenocrysts of plagioclase and orthopyroxene. The plagioclase phenocrysts are approximately 12 mm in size and are elongate to acicular in form. The crystals are subhedral and fractured but are relatively clear. Orthopyroxene phenocrysts (12 mm) are equant, strongly fractured, and, in some

examples, exhibit resorption and disaggregation into the surrounding melt (Fig. 3k). Rare sieve-textured grains also occur as well as slightly coarser-grained crystal clots (Fig. 3l). The groundmass of the Black Tusk andesite comprises approximately equal amounts of acicular, plagioclase microlites, and brown glass. Strong flow banding predominates in this unit.

Geochemistry Analytical techniques Nine samples from the MGVF and five from the GLVF was analyzed for whole-rock major, trace, and REE concentrations. All samples were crushed in a jaw crusher and then powdered to less than 74 m in a tungsten mill at the Dept. of Geology, University of Regina. All elements were measured by the Trace Element Analysis Laboratories at McGill University, Montral, Canada. Data are presented in Tables 1 and 2. Major elements and some trace elements including Ba, Ce, Cr, Cu, Ni, V, Sc, and Zn were determined by a Philips PW 2440 X-ray fluorescence spectrometer analysis on fused glass beads. Loss on ignition (LOI) was determined by incrementally heating 4 g of rock powder in air to 400 C for 60 min, 800 C for another 60 min, then to 1,000 C for 180 min. The product was weighed after cooling in a desiccator to calculate LOI. Other trace elements (Ga, Nb, Rb, Sr, Th, U, Y, and Zr) were analyzed on pressed powder pellets. Two standards of known composition were sent with our samples to test the accuracy and reproducibility of the results. The accuracy for silica is within 0.5 %. For other major elements, it is within 1 % of the element analyzed. For trace elements as well, the accuracy is within 1 %, as determined from replicate analyses of internal standards. The limiting factor for accuracy is the degree of scatter of analyses from which the consensus values are determined (Govindaraju et al. 1994, p. 256). Instrument precision is within 0.12 % relative. This is the percent relative variation obtained when the same sample is analyzed repeatedly for the same element. Overall precision for beads and pressed pellets is within 0.5 % relative. This is determined by repeatedly analyzing two beads or pressed pellets prepared from two different aliquots removed from the same sample powder vial during the same day and used to prepare a fused bead or pressed pellet by an experienced operator using routine procedure. REE were measured by inductively coupled plasma-mass spectrometry (ICP-MS) using a borate fusion decomposition method (after Panteeva et al. 2003). The analyses of the REE were performed on solutions using a PerkinElmer/SciEx Elan 6100 DRCplus ICP-MS. Samples (0.4 g, corrected for LOI) were fused using a lithium metaborate mixture and then dissolved into nitric acid and diluted. Standards and calibration solutions

730, Page 8 of 23

Table 1 Major and minor element composition of investigated samples from the Garibaldi Volcanic Complex Mt. Garibaldi volcanic field

Garibaldi Lake volcanic field

Sample 59.35 0.67 18.80 5.74 0.10 2.88 6.36 4.69 1.19 0.28 4.69 50.1 0.25 619 20.6 100.26 0.64 18.41 5.64 0.10 3.05 6.27 4.61 1.22 0.26 0.46 52.0 0.26 686 19.5 100.08 0.63 18.37 5.87 0.10 3.05 6.09 4.47 1.25 0.24 0.10 51.0 0.28 788 19.5 99.85 0.57 17.71 4.82 0.09 2.37 5.78 4.41 1.39 0.26 0.78 49.6 0.32 628 26.1 100.13 0.57 17.97 4.34 0.08 2.24 5.28 4.29 1.37 0.24 3.15 50.8 0.32 653 26.9 99.97 0.58 17.18 4.78 0.09 2.52 5.51 4.30 1.53 0.25 0.63 51.6 0.36 660 24.8 99.93 0.55 17.52 4.66 0.08 2.40 5.65 4.39 1.49 0.24 0.63 51.1 0.34 668 26.1 100.27 0.39 17.11 3.63 0.08 1.81 4.39 4.44 1.71 0.16 1.48 49.9 0.39 674 35.8 100.07 59.34 59.58 61.86 60.34 62.46 62.57 64.76

BF 09JF004 64.56 0.41 17.19 3.88 0.09 1.95 4.52 4.46 1.64 0.16 0.77 50.1 0.37 706 33.1 99.73

BF 09JF005

BF 09JF006

BF 09JF012

BT 10JF016

RC 09JF007

RC 09JF008

RC 10JF022

RC 10JF023

CP 09JF009

CP 09JF010

PR 09JF011 58.51 0.65 16.90 5.93 0.10 4.96 6.62 4.19 1.14 0.21 1.02 62.6 0.27 713 11.8 100.34

PR 10JF017 56.31 0.96 18.51 8.06 0.14 3.42 6.39 4.61 0.89 0.33 0.21 45.9 0.19 881 16.5 99.91

PR 10JF018 57.46 0.78 18.61 6.90 0.12 3.61 6.70 4.37 1.16 0.28 0.05 51.4 0.27 733 15.9 100.15

SiO2

59.74

60.16

TiO2 Al2O3 Fe2O3(T) MnO MgO CaO Na2O K2O P2O5 LOI Mg# K2O/Na2O K/Rb SiO2/MgO Total

0.61 18.52 5.56 0.09 2.92 6.22 4.62 1.19 0.25 0.37 51.2 0.26 678 20.5 100.20

0.65 18.72 5.40 0.09 2.76 6.13 4.72 1.26 0.28 0.20 50.5 0.27 648 21.8 100.46

Total iron reported as Fe2O3. Mg#=molar Mg/(Mg+Fe)100, where Fe=Total Fe as FeO Bull Volcanol (2013) 75:730

BF Barrier flow (andesite), BT Black Tusk (andesite), RC Ring Creek flow (andesite), CP Columnar Peak (dacite), PR Paul Ridge (basaltic andesite/andesite)

Table 2 Trace and rare earth element composition of investigated samples from the Garibaldi Volcanic Complex Mt. Garibaldi volcanic field

Garibaldi Lake volcanic field

Bull Volcanol (2013) 75:730

Sample d/l 99 29 38 41 30 17.9 16.0 982.4 12.5 101.9 7.4 450.6 3.3 1.5 0.6 13.68 29.27 3.86 16.09 3.24 1.09 2.86 0.42 2.47 0.49 1.33 0.19 1.31 0.20 78.7 10.44 d/l 99 31 39 39 35 17.9 14.9 1,023.5 11.4 92.5 5.6 441.5 4.4 1.4 0.6 12.03 26.21 3.53 14.78 3.05 0.99 2.73 0.39 2.22 0.45 1.27 0.17 1.19 0.18 89.5 10.11 11 98 21 56 75 28 18.9 13.2 909.8 12.0 91.5 3.3 436.3 6.5 1.3 0.6 10.73 24.00 3.27 14.32 3.09 1.00 2.80 0.41 2.37 0.46 1.33 0.19 1.25 0.19 75.8 8.58 11 85 32 28 41 30 17.6 18.5 1,068.4 11.9 117.9 4.4 478.0 4.6 1.9 0.7 14.87 32.56 4.25 17.48 3.31 1.03 2.72 0.39 2.24 0.43 1.20 0.17 1.19 0.18 89.9 12.50 d/l 86 32 30 49 23 18.0 18.1 1,078.6 10.9 115.7 4.7 469.5 3.0 1.9 0.7 15.24 33.40 4.39 17.82 3.36 0.99 2.40 0.36 2.07 0.39 1.13 0.16 1.11 0.17 98.6 13.73 10 84 18 30 59 26 18.4 19.4 1,012.2 13.2 119.7 4.5 561.9 7.3 2.1 0.8 17.13 37.25 4.87 20.11 3.83 1.07 3.05 0.42 2.39 0.46 1.27 0.19 1.28 0.19 76.7 13.38 11 81 17 26 50 25 18.7 18.7 1,026.8 12.6 111.7 4.2 549.4 8.2 2.1 0.8 16.90 36.37 4.73 19.46 3.65 1.08 2.94 0.41 2.30 0.45 1.27 0.18 1.19 0.18 81.5 14.20 10 63 21 42 22 14 15.4 21.4 752.0 9.5 91.5 4.3 580.7 d/l 2.7 1.1 13.02 27.09 3.30 12.90 2.36 0.75 1.93 0.29 1.71 0.34 0.94 0.14 1.01 0.15 79.2 12.89 d/l 63 23 30 51 17 16.3 19.5 784.8 10.1 100.5 4.2 558.8 4.7 2.6 1.1 12.85 26.83 3.30 12.90 2.42 0.76 1.98 0.30 1.73 0.34 0.98 0.14 1.01 0.16 78.0 12.72 14 110 132 103 71 45 17.0 13.3 957.9 11.8 104.2 4.0 390.7 3.5 2.7 0.9 15.11 32.12 4.04 16.46 2.98 0.98 2.48 0.36 2.11 0.43 1.18 0.17 1.17 0.18 81.5 12.91 15 133 18 30 66 53 20.0 8.4 838.7 17.1 91.4 3.8 467.1 6.6 0.8 0.3 11.54 26.54 3.78 17.49 4.02 1.35 4.00 0.58 3.42 0.68 1.98 0.28 1.85 0.28 49.0 6.24

BF 09JF004

BF 09JF005

BF 09JF006

BF 09JF012

BT 10JF016

RC 09JF007

RC 09JF008

RC 10JF022

RC 10JF023

CP 09JF009

CP 09JF010

PR 09JF011

PR 10JF017

PR 10JF018 8 121 31 48 100 40 19.1 13.1 868.7 14.0 101.1 3.7 457.9 7.3 1.4 0.5 12.03 27.27 3.74 16.63 3.56 1.12 3.48 0.48 2.87 0.57 1.58 0.23 1.53 0.22 62.1 7.86

Sc V Cr Ni Cu Zn Ga Rb Sr Y Zr Nb Ba Pb Th U La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Sr/Y La/Yb

d/l 97 32 43 41 33 17.1 14.6 1,000.3 11.9 90.2 5.3 441.1 2.2 1.4 0.6 12.01 26.61 3.58 15.11 3.22 1.03 2.68 0.39 2.33 0.47 1.25 0.19 1.26 0.19 83.9 9.53

11 96 29 34 37 34 17.8 16.2 957.8 13.1 101.8 7.3 477.0 2.8 1.6 0.7 13.46 28.80 3.77 15.79 3.19 1.06 2.96 0.42 2.48 0.47 1.32 0.19 1.29 0.20 73.4 10.43

Elemental concentrations expressed in parts per million

Page 9 of 23, 730

d/l below detection limit, BF Barrier flow, BT Black Tusk; RC Ring Creek flow, CP Columnar Peak, PR Paul Ridge

730, Page 10 of 23

Bull Volcanol (2013) 75:730

were prepared from fusion blanks. Oxide corrections on the middle and heavy REE were made offline using oxide production rates determined daily from single REE standard solutions. Rock-sample detection limits (based on three times the background standard deviation) are 10 ppb for La through Pr and 5 ppb for Nd through Lu. A set of three internal laboratory reference materials are fused and run with each batch of samples to evaluate long-term precision. Precision was additionally evaluated through repeat measurements of samples, including repeat fusions and dilution; it is better than 3 % relative standard deviation in all cases. Accuracy was evaluated using a series of six standard reference materials (SRMs) that span the sample concentration range, prepared using the same procedure as the samples. Our determinations agree with the accepted values for these SRMs with discrepancies of less than 5 %.

Results Nine samples were analyzed from the MGVF: three samples of the Paul Ridge andesite, two from the Columnar Peak dacite, and four samples of the Ring Creek andesite (two from the proximal part of the flow and two from the distal portion). Another four samples from the Barrier andesite in the GLVF and one sample from Black Tusk were also analyzed (see Tables 1 and 2). The geochemistry of the GLVF is well documented in several other studies (Green 1977, 1981, 1990; Green and Henderson 1984), and select data from these studies have been included for comparison in this work. The investigated MGVF and GLVF rocks are sub-alkaline in character ranging in composition from basaltic andesite to dacite, with the bulk of samples falling in the andesite field (Figs. 4 and 5). All the GVC samples define similar trends as individual centers in several major and trace element variation diagrams (Figs. 4 and 6). The volcanic products from each centre are characterized by decreasing Fe2O3, CaO, and V with increasing SiO2. MgO is increasing in the Paul Ridge andesite with increasing SiO2, as does TiO2 in the Barrier andesite. All other centers in the GVC have decreasing MgO and TiO2 with increasing SiO2. Al2O3 decreases only slightly as SiO2 increases with alumina concentrations spanning a range of about 1.5 wt.%. Ni, Cr, and Na2O do not vary significantly with increasing SiO2, forming relatively flat trends. K2O and Rb contents generally increase with SiO2 for all centers, as does Sr for the Paul Ridge and Black Tusk andesites. Sr in the Barrier andesite increases to approximately 60 wt.% SiO2 and then decrease sharply; a similar trend is seen for The Table andesite at approximately 58 wt.% SiO2. Sr decreases with increasing SiO2 for the Ring Creek andesite and the Columnar Peak dacite. The Mg#s of all the GVC rocks

exhibit a relatively narrow range (46 to 52), except for the Paul Ridge andesite, where one sample has Mg# of 62. Primitive mantle-normalized multi-element spider diagrams show that all centers have LILE enrichment and NbTi-negative anomalies, typical of subduction-related rocks (Fig. 7a and b). The Paul Ridge andesite rocks exhibit the strongest depletion of Th, Rb, U, and other incompatible elements and the highest MREE-HREE abundances of all the GVC rocks. One andesite from Paul Ridge (09JF011; see Tables 1 and 2) has higher MgO, TiO2, Ni, and Cr values than the other samples. Chondrite-normalized REE spider diagrams (Fig. 8a and b) for the GVC volcanic rocks display enrichment of LREE over HREE and lack any significant Eu anomalies. The andesites of Black Tusk display the lowest LREE/HREE fractionation of all the GVC volcanic rocks with La/YbN averaging 5.9. The Ring Creek andesite samples exhibit the highest fractionation with La/YbN ranging from 8.5 to 9.6. All of the GVC rocks have similar MREE/HREE fractionation with ratios of 1.5 to 1.9. On a Sr/Y versus Y diagram, all of the GVC rocks plot within the adakite field but fall outside of this field on the La/Yb versus Yb diagram, having values typical of normal arc-rocks (Fig. 6fg). The accepted minimum value for La/Yb as an adakitic indicator ranges from as low as 8 (Drummond and Defant 1990) up to 20 and greater (Castillo et al. 1999; Richards and Kerrich 2007). The La/Yb values for the GVC have a range of 6 (Paul Ridge andesite) to 14.2 (Ring Creek andesite), plotting in the lower end of the adakite field. When plotted against SiO2, both Sr/Y and La/Yb appear to increase for the Paul Ridge andesite rocks in contrast to the other GVC centers, for which Sr/Y decreases with SiO2 but La/Yb increases (not shown). Other adakitic indices (Sr, Na2O, Al2O3) do not show definitive trends with fractionation indices (SiO2, Ni, Cr), suggesting that the chemistry is not controlled exclusively by fractionation processes (Chiaradia 2009). The GVC rocks may also be divided into the low-SiO2 and high-SiO2 adakite (LSA and HSA) groups on the basis of geochemical characteristics outlined for adakitic rocks by Martin et al. (2005). On a K/Rb versus SiO2/MgO diagram, the Paul Ridge and Black Tusk andesite rocks exhibit high K/Rb relative to SiO2/MgO and form a sub-vertical trend (Fig. 9a) indicative of LSA; previously published data for The Table andesite samples also follow this trend. The Ring Creek andesite and the Columnar Peak dacite have lower K/Rb values and plot sub-horizontally. The Barrier andesite samples appear transitional, plotting at the intersection between the LSA and HSA trends. This may relate to the slightly elevated Rb contents of Barrier andesite samples and their SiO2 values, which lie at the boundary between the LSA and HSA groups. The LSA and HSA groupings are still evident, though not as well defined in the SrK/Rb-(SiO2/MgO)*100 ternary diagram (Fig. 9b). Here,

Bull Volcanol (2013) 75:730

Page 11 of 23, 730

Fig. 4 Harker diagrams illustrating variations in major element chemistry with increasing SiO2 (oxide wt.%). Key: RC Ring Creek, CP Columnar Peak, PR Paul Ridge, BT Black Tusk, BF Barrier flow, open squares, open circles, and crosses represent previously published data for the Black Tusk,

Barrier flow, and The Table centers, respectively (data derive from Green 1977, 1981; Green and Henderson 1984). All data have been recalculated to 100 % on an anhydrous basis

only the Columnar Peak dacite is distinctly HSA and The Table andesite rocks are clearly LSA.

Discussion The rocks of the GVC exhibit geochemical characteristics that favor their classification as adakites. Adakites are

characterized by 56 wt.% SiO2, 15 wt.% Al2O3, usually 3 wt.% MgO (rarely 6 wt.%), 18 ppm Y, 1.9 ppm Yb and Sr contents >400 ppm and were originally believed to be associated with the subduction of young oceanic crust (25 Ma; Defant and Drummond 1990). Since the introduction of the term adakite, this definition has been expanded to now include a wide range of compositions and geological settings based on certain geochemical

730, Page 12 of 23 Fig. 5 a, b Total alkali versus silica diagram with IUGS rock designations for the GVC center rocks (after Le Bas and Streckeisen 1991). b Represents an enlargement of the area where GVC samples fall. Dashed line represents the alkalinesubalkaline boundary from Macdonald (1968). Symbols as in Fig. 4

Bull Volcanol (2013) 75:730

indicators, specifically high Sr/Y (>40) and La/Yb (>20) (e.g., Castillo et al. 1999; Xu et al. 2000). This has led to some confusion in the literature and a general misuse of the term adakite, such that any rock type from any geological environment that has high Sr/Y and/or high La/Yb has been called adakite or adakite-like. To make valid use of adakite as a petrogenetic indicator, a suite of geochemical and petrological features as well as spatial associations is required. Defant and Drummond (1990) first introduced the term adakite and the criterion by which they are defined. Hence, it is to this definition that the rocks of the GVC have been evaluated.

Adakite Genesis The majority of the models for adakite genesis fall into two main categories: slab melting and non-slab melting. In simple terms, the slab melting model reads that young, hot subduction zones retain enough residual heat in the slab to allow it to melt at shallower than typical depths (Defant and Drummond 1990). Adakite magmas may also form in response to flat subduction (e.g., Gutscher et al. 2000) or as a result of slab edge melting at slab windows (Thorkelson and Breitsprecher 2005). Slab partial melts are dacitic in composition with a restite rich in garnet amphibole but plagioclase-poor. The melts react with the

Bull Volcanol (2013) 75:730

Page 13 of 23, 730

Fig. 6 (ag) Harker diagrams illustrating variations in trace element chemistry for the GVC centre rocks, with increasing SiO2. Plots of adakitic indices for GVC rocks: f Sr/Y versus Y and g La/Yb versus

Yb. Key symbology as in Fig. 4; adakite and normal arc-rock fields derive from Castillo (2006) and Richards and Kerrich (2007)

overlying mantle wedge during ascent, modifying their major and compatible element chemistries (e.g., increased Mg#, CaO, Ni, Cr, lower SiO2) but keeping diagnostic trace element concentrations and ratios intact (Sr, Y, REE; Moyen 2009), which remain recognizable as having a slab melt origin. Other models for adakite genesis that are not associated with slab melting include partial melting of mafic lower continental crust (Huang and He 2010), fractional crystallization of basaltic magma containing garnet (Macpherson et al. 2006; Coldwell et al. 2011)

and high pressure assimilation, fractional crystallization (AFC) processes of mantle-derived melts, and magma mixing (Chiaradia et al. 2009). The slab melt model has been refined by Martin et al. (2005). Using a database of >340 adakite analyses (previously compiled by Martin and Moyen 2003), Martin et al. (2005) proposed that adakites should exhibit the following characteristics: >3.5 wt.% Na2O, K2O/Na2O ratios of 0.42, an Mg# of approximately 50, FeO+MgO+MnO+TiO2 7 wt.%, and relatively high Ni (24 ppm) and Cr

730, Page 14 of 23 Fig. 7 a Primitive mantlenormalized spidergrams for GVC samples analyzed in this study, and b for previously published data. Data in (b) derive from Green and Henderson (1984). Primitive mantle normalizing values are from Lyubetskaya and Korenaga (2007). Symbols as in Fig. 4

Bull Volcanol (2013) 75:730

(36 ppm). As a result, Martin et al. (2005) introduced two different groups of adakite, a high-SiO2 adakite (HSA) and a low SiO2 (LSA) adakite. HSA contains >60 wt.% SiO2, lower MgO (0.5 to 4 wt.%), CaO+Na2O < 11 wt. %, and Sr contents

< 1100 ppm. LSA contain <60 wt.% SiO2, 49 wt.% MgO, >10 wt.% CaO+Na2O, and Sr in excess of 1,000 ppm. These two groups are also differentiated by other trace elements and REE (Martin et al. 2005). LSA have higher LREE contents, a

Bull Volcanol (2013) 75:730 Fig. 8 a Chondrite-normalized REE spidergram for GVC samples analyzed in this study, and b for previously published data. Data in (b) derive from Green and Henderson (1984). Chondrite normalizing values are from Sun and McDonough (1989). Symbols as in Fig. 4

Page 15 of 23, 730

more pronounced positive Sr anomaly and are generally Rb-poor compared with HSA. The petrogenesis of LSA and HSA are both related to slab melting, but the magma sources are thought to differ. Martin et al.

(2005) suggested that HSA are slab melts that have assimilated mantle wedge peridotite during ascent, prior to eruption. LSA are not primary slab melts, but the result of partial melting of mantle wedge peridotite that

730, Page 16 of 23

Bull Volcanol (2013) 75:730

Adakite Geochemistry in the GVC The rocks analyzed in this study and previously published data conform to virtually all of the adakitic geochemical traits put forward by Defant and Drummond (1990) and Martin et al. (2005). However, there are some variations in trace and major element contents which illustrate a more typical arc-like magma composition for the GVC. These variations are likely the result of mixing between preexisting, HSA magmas and intruding non-adakitic magmas in the small, intermittent magma chambers beneath the various centers. Despite this mixing, trace element ratios and REE concentrations indicative of slab partial melts are preserved in the investigated GVC rocks. While determined Sr/Y ratios for the GVC cover a wide range (38109; Fig. 6f, Table 2), the majority of values fall between 75 and 90. Some of the previously published data (Black Tusk and the Barrier flow, in particular) straddle the boundary between adakite and normal arc-rocks, which appears to be related to elevated Y contents reported in these earlier published data. Moreover, all of the GVC samples have La/Yb ratios that lie in the normal arc-rock field and not in the adakite field (Fig. 6g). This is a function of the low La in the GVC versus the Yb contents, which are typical of adakites (<1.9 ppm). La concentrations in the GVC range from 10.7 to 17.1 ppm, lower than the average for typical intermediate rocks (2030; GERM 2013, earthref.org). However, La values for adakite from other localities are quite variable, from 30 ppm (Cook Island, Stern and Kilian 1996) to as low as 1012 ppm (Pichincha Volcano and Tonga Trench, Bourdon et al. 2002; Falloon et al. 2008, respectively). Similarly, the Cr values from all the centers in the GVC are lower than that proposed by Martin et al. (2005) (36 ppm), but Cr contents of 30 ppm are considered adakitic (Martin 1999; Richards and Kerrich 2007), and these values are not significantly different from that of the average Cr content for the GVC adakites (2732 ppm). As outlined above, the lower Cr and La concentrations of the GVC adakite rocks are most likely the result of mixing between a pre-existing HSA magma and an intruding nonadakitic magma(s) of broadly similar composition. All the GVC magmas show evidence of high-level magma mixing, with the Paul Ridge andesite exhibiting the strongest. Mixing of HSA magmas with normal intermediate arctype magmas can decrease the Cr and La concentrations in the GVC lavas (and Ni slightly) relative to the average adakite. In the case of the Paul Ridge andesite, the intruding magma was more basaltic in composition. The occurrence of olivine and orthopyroxene with quartz and clear disequilibrium textures observed between these phases and the groundmass (see Fig. 3h) support the interaction of compositionally distinct magmas, as has been documented at other volcanoes within the GVB (cf. Hickson et al. 1999 for the eruption of Mount Meager). One sample of the Paul Ridge

Fig. 9 Discriminant diagrams for the LSA and HSA groups. a K/Rb versus SiO2/MgO plot for the GVC rocks with extreme values used by Martin et al. (2005) and showing overlap between LSA and HSA data points. b Enlarged view of the K/Rb versus SiO2/MgO plot illustrating the distinction between LSA and has groups within the investigated GVC rocks. LSA plots higher in terms of K/Rb while HSA defines a sub-horizontal trend. c [Sr-K/Rb-(SiO2/MgO)*100] ternary diagram illustrating the distinction between LSA and HSA groups for GVC rocks. HSA and LSA fields and inset figure in (a) derive from Martin et al. (2005). Symbols as in Fig. 4

has been metasomatized by slab-derived melts. Adakite rocks as defined by Defant and Drummond (1990) are classified as HSA.

Bull Volcanol (2013) 75:730

Page 17 of 23, 730

andesite (09JF011) is considerably higher in SiO2, Ni, Cr, and MgO content and lower in Fe2O3 than the other two andesite samples analyzed. Hence, it is likely that this more mafic sample did not have the same mixing components, or perhaps underwent a stronger mixing process in the volcano than the other Paul Ridge andesite samples. Green (1977) noted that some minor pyroclastic material was present at Paul Ridge, but individual flows were not identified. It is possible that the more mafic sample taken from the northern part of Paul Ridge may represent a pyroclastic component that has a stronger basaltic character and may be more geochemically similar to the intruding basaltic magma compared with the other andesite samples. For the Ring Creek andesite, there is no evidence of interaction with a basaltic intruding magma and mixing relationships are less clear, suggesting that the mixing components were of a similar composition. The mineralogy of the Ring Creek andesite varies from an augitehornblende biotite-bearing assemblage in the proximal portion of the flow to augite only in the distal portion. The proximal andesite also appears to contain primary quartz, where quartz crystals in the distal Ring Creek andesite show disequilibrium with the surrounding melt and hence, is difficult to explain through a simple mixing process. However, Sivertz (1976) in mapping the Ring Creek andesite and Opal Cone also noted this difference in mineralogy, concluding that the hydrous mineral assemblage in the proximal Ring Creek andesite was identical to the mineralogy of Opal Cone itself. It is plausible, therefore, that the proximal Ring Creek andesite entrained material from Opal Cone during eruption and inherited the hornblendebiotitequartz mineralogy. Mixing can result from convective overturn initiated by intrusion of hotter magma from depth, but there is no evidence of mixing between compositionally distinct magmas in the Ring Creek andesite. Magma chamber overturn between magmas of a similar composition can be caused if the intruding magma has a high upward momentum (Turner and Campbell 1986), or was significantly hotter. The Ring Creek andesite flow is unusually extensive for an intermediate to felsic composition, approximately 17 to 18 km in length (Sivertz 1976; Green 1977; Brooks and Friele 1992). It represents predominantly effusive volcanism with no evidence of any pyroclastic component associated with the flow. Sivertz (1976) described lapilli and block fragments of dacitic composition, but this material is only found within Opal Cone. The large extent of the Ring Creek andesite flow and the lack of any significant pyroclastic material might suggest that intrusion of a large volume of relatively fast moving magma (the proximal andesite) of similar composition into pre-existing, cooling magma (the distal andesite) within the magma chamber created turbulence that facilitated entrainment of wall-rock material from Opal Cone as the

eruption proceeded. It is not known to what extent the hornblendebiotitequartz mineralogy is present in the Ring Creek andesite further south; access to the center portion of the flow is very limited (Sivertz 1976). In the Black Tusk andesite, mixing relationships are not as evident as for other centers within the GVC. The lava is essentially aphyric, but the phenocrysts that are present are strongly resorbed. This suggests that an intruding lava was likely significantly hotter than the pre-existing magma, which results in the near-total resorption of the phenocrysts and obscures any other meltcrystal relationships. Additionally, the intruding magma appears to be a very similar composition to the resident magma; both are comprised of the same mineralogy. This may indicate that the volume of the intruding magma was large, in order to superheat the system and allow for the disaggregation of all the phenocrysts. Mixing in the Barrier andesite was likely between that of a dacitic pre-existing magma and an andesitic intruding magma. This is evidenced by reaction rims of augite on quartz phenocrysts and strongly embayed biotite crystals. Hornblende appears to be in equilibrium with the intruding andesitic lava; there are few disequilibrium features in the phenocrysts except for rims of opaques along the grain margins, and these may be the result of hornblende being removed from its stability field during ascent. At Columnar Peak, both the pre-existing and intruding magmas appear to be compositionally alike. There are disequilibrium features in the plagioclase and hornblende phenocrysts, but the intruding magma generally contains the same mineralogy, with the exception of minor orthopyroxene (<1 %). These features indicate that, despite the similar composition, the relative temperatures of both magmas were such that the larger, pre-existing phenocrysts were strongly resorbed and embayed relative to the phenocrysts from the intruding magma. The mixing relationships in the GVC lavas are significant for adakite genesis for several reasons: (1) despite mixing with arguably non-adakitic magmas (in some cases extensively mixed), the majority of the adakitic characteristics are preserved in the GVC lavas (only Cr and La are affected), (2) with no evidence of interaction with mafic magmas (except for the Paul Ridge andesite), the adakitic Ni, Cr, and Mg# of the GVC rocks likely reflect interaction with peridotite during ascent through the mantle wedge, and (3) these mixing processes may explain the fact that the GVC rocks exhibit geochemical characteristics of both the HSA and LSA groups. The more mafic component is difficult to discern, and the limited data in this study preclude any meaningful interpretation on the nature of the intruding magmas. The presence of normal arc basalts and olivine bearing rocks in the GVC suggest that basaltic magmas (or its fractionated products) are present in the magmatic system but at present, their origins remain speculative.

730, Page 18 of 23

Bull Volcanol (2013) 75:730


10 8
MgO%
8 6 4 HSA

Interaction with mantle peridotite The Ni, Cr, and Mg# values for the GVC adakite rocks are generally higher than normal andesite and dacite (20 ppm, 2025 ppm, and 42, respectively) and, due to the lack of evidence of a basaltic mixing component for the majority of the GVC (with the exception of the Paul Ridge andesite), these values likely reflect the interaction of HSA magmas with mantle peridotite during ascent. Increasing the Mg#, Ni, and Cr concentrations in slab melts by assimilation of peridotite (as described for HSA by Martin et al. 2005) is an unlikely process for the GVC. Typical mantle peridotite contains 3,200 ppm Cr, 2,300 ppm Ni, and 42 wt.% MgO (Sigurdsson et al. 2000). To obtain the values observed in the GVC adakites, assimilation of peridotite would be strongly limited (1 % assimilation to obtain the Cr values of GVC). Furthermore, peridotite begins to melt at approximately 1,200 C (at lower pressures) and exceeds that of the dacitic (900 C) slab melt, precluding any significant partial melting of the peridotite. A more likely process would be zone refining, whereby the ascending HSA magma gains Ni, Cr, and MgO by diffusion. This process enriches the adakite magma in mafic components and lowers SiO2, but preserves the incompatible element ratios of the slab-melt, as the diffusion rates of incompatible elements (e.g., REE, Y) would be too slow to significantly modify the ascending magma (Wilson 1989) and obscure the slab-melt signature. LSA versus HSA Geochemical characteristics distinguish the Columnar Peak dacite and the Ring Creek andesite rock as mainly HSA and the Paul Ridge andesite and the Black Tusk andesite as predominantly LSA. Previously published data from The Table andesite are also considered and fall within the LSA field. The Barrier andesite rocks appear transitional between the two groups. LSA are best differentiated from HSA in terms of K/Rb versus SiO2/MgO, and SrK/Rb-(SiO2/MgO)*100 diagrams (see Fig. 9a and b). LSA and HSA define an almost perpendicular relationship on the K/Rb versus SiO2/MgO diagram, with LSA plotting at higher values of K/Rb whereas K/Rb ratios for HSA remain relatively uniform. On the Sr-K/Rb-(SiO2/MgO)*100 ternary diagram, the relationship for the GVC rock is less clear; HSA forms a group close to the (SiO2/MgO)*100 apex and LSA plots towards the center of the plot. The Black Tusk and Paul Ridge andesites fall clearly in the LSA field in Fig. 9a, but both centers plot closer to the HSA field in Fig. 9b. Other discrimination diagrams that compare LSA and HSA are shown in Fig. 10ac and illustrate the lack of a definite separation of LSA and HSA groupings in the GVC. The dashed fields and insets representing LSA and HSA in

LSA

MgO

6 4 2 0 45 50 55 60

0 45 50 55 60 65 70 75 SiO2%

65

70

75

80

SiO2
25
Nb ppm
LSA HSA

20

20 15 10

Nb (ppm)

15 10 5 0 45 55

5 0 45 50 55 60 65 70 75 SiO2%

65

75

SiO2
3500
Sr ppm
LSA

3000

Sr (ppm)

2500 2000 1500 1000 500 0 0

3000 2500 2000 1500 1000 500 0 0

HSA 5 10 (CaO+Na2O%)

10

15

CaO + Na2O
Fig. 10 a MgO versus SiO2, b Nb versus SiO2, and c Sr versus CaO+ Na2O discriminant diagrams illustrating the variability of HSA and LSA compositions in the GVC dataset. Inset plots in (ac) are modified from Martin et al. (2005). Symbols as in Fig. 4

Figs. 9 and 10 are from the data presented in Martin et al. (2005), which included analyses that had extreme values (K/Rb up to 3,000, average Sr>2,000 ppm, Nb up to 20 ppm). For the range of values in the GVC (K/Rb 660880, Sr 7501,300 ppm, Nb 8 ppm), LSA and HSA generally plot in the same space, creating further confusion. On a primitive mantle-normalized spider diagram (Fig. 7a b), LSA differs from HSA by lower Rb and higher Nb values, with a positive Sr anomaly and no Ti anomaly (Martin et al. 2005). The GVC rocks generally characterized as LSA (Paul

Bull Volcanol (2013) 75:730

Page 19 of 23, 730

Ridge and Black Tusk andesite rocks) have lower Rb, but also lower Nb and similar Sr to HSA, as well as a negative Ti anomaly. Similarly, the Ring Creek andesite and the Columnar Peak dacite, which are predominantly HSA, have higher Nb and Sr (on average), typically a characteristic of LSA. The Barrier andesite has Rb values between the Paul Ridge and Columnar Peak rocks, but also the highest Nb contents of all GVC rocks (up to 8 ppm). The Table andesite samples are the only rocks which exhibit all of the LSA characteristics; the andesite rocks have the highest Nb and Sr with no Ti anomaly (Fig. 7b). The Table andesite rocks also consistently plot in the LSA fields in Figs. 9 and 10. Isotopic concentrations and effects of crustal interactions Stern and Kilian (1996) noted that the effects of crustal interactions were present in adakite rocks from the Austral Volcanic Zone (AVZ), and these decreased southward in the belt as the angle of subduction became more orthogonal. This resulted in negligible interaction of the Cook Island adakites with crustal material and hence, their Sr, Nd, Pb, and O isotopic compositions more closely resemble MORB (and by extension, slab partial melt) values. Crustal contamination has been suggested to contribute to the chemistry of the GVC lavas (Green 1990; Green and Henderson 1984), specifically in the MGVF, based upon 87Rb/86Sr and 87 Sr/86Sr isotopic data. Green (1990) stated that the Mount Garibaldi rocks contain a significant Rb-rich crustal component based on higher 87Rb/86Sr isotopes than the GLVF and that these values reflected AFC processes combined with contamination from crustal xenoliths and mixing with anatectic melts during ascent. While mixing of melts is present in the MGVF, there is little evidence for the incorporation of crustal xenoliths in any rocks examined as part of this study. Rare xenoliths are present in the Ring Creek andesite but are mafic to intermediate in composition and would not significantly modify Rb values. Lower Rb values in the GLVF (418 ppm) was suggested by Green (1990) to reflect a depletion of LILE in the source region and less crustal interaction than the Mount Garibaldi rocks. For the MGVF rocks in this study, the Rb concentrations are comparable to that determined by Green (1990) for the GLVF (average 14.5 ppm). Similarly, the high 87Rb/86Sr values are in samples that are rhyodacite to rhyolite in composition, which is more felsic than the rocks examined in this study. Green (1990) did not specify the exact sample locations, but rhyodacite is present in Mount Garibaldi itself and may be the location represented by the data presented by Green (1990). Determination of isotopic data was not within the scope of this work, and any conclusions on the effects of crustal interaction in the GVC cannot be fully elucidated. Similarly, while the above interpretations can account for the geochemical variations seen in the GVC, it must be

noted that future, more detailed studies with a larger dataset may argue against these hypotheses. Origin of the GVC Adakites The lack of a clear distinction between LSA and HSA groups and limited data preclude a complete assessment of adakite genesis within the GVC. However, data from the GVC can be compared with existing models for adakite genesis and, as such, can provide some insight into possible magmatic processes occurring in the development of the GVC rocks. It is likely, based upon the data examined in this study, that partial melting of the subducting Juan de Fuca Plate played an important role in the generation of GVC magmas. An adakite signature is ubiquitous across GVC rocks (Sr>750 ppm, Yb<1.9 ppm, Y<19 ppm, fractionated REE) and may reflect source processes. Other geochemical characteristics that define adakites are also shared by all of the GVC rocks (Mg# 51, Ni>24 ppm, K2O/Na2O<0.4). The extreme values used by Martin et al. (2005) to define the LSA and HSA fields causes the investigated GVC rocks to plot where both LSA and HSA compositions exist in the same space, and despite the fact that traits from both groups are shared by the rocks of the GVC, the apparent characterization of HSA versus LSA may only be a function of sample size. Similarly, the mixing of adakite and non-adakite magmas within the magma chamber may also contribute to the scatter between LSA and HSA. The available dataset has limitations on assessing magma genesis, but magma generation by melting of the subducted basaltic slab directly or by peridotite partial melts modified by a slab component cannot be ruled out. Additionally, the Nb values in adakites are 6 ppm on average, and Nb concentrations in the GVC range from 4 to 8 ppm; such values cannot be obtained by fluid flux from the dehydrating slab because these fluids are not effectively able to concentrate Nb (Tatsumi et al. 1986; Martin et al. 2005). Thus, the Nb enrichment in the GVC is likely caused by slab melting, which is able to concentrate and transfer Nb. Stern and Kilian (1996) noted that the andesite rocks from Cook Island in the AVZ of the Andes had Rb/Sr values (0.002) comparable to fresh MORB and that these rocks were the best representation of possible primary slab-melt. The Rb/Sr concentrations in the GVC rocks (0.01) are not as low as those found in the Cook Island andesites and preclude the involvement of a fresh MORB source in the GVC. However, Rb/Sr of altered MORB (0.080.12; Kelley et al. 2003) are above the values found in the GVC, and a potential basaltic source cannot be excluded. High-pressure AFC processes and partial melting of mafic lower crust containing garnet are models that have been proposed for the formation of adakites without slab melting (e.g., Chiaradia 2009; Coldwell et al. 2011). Although

730, Page 20 of 23

Bull Volcanol (2013) 75:730

adakitic chemistry (Sr/Y and La/Yb) can be obtained via AFC and/or partial melting, these processes are likely not the dominant ones occurring in the GVC. Average thickness of continental crust in the GVC is approximately 35 km (Perry et al. 2002); typically this is too thin for garnet to be stable at the base of the crust and, therefore, a control on HREE contents (e.g., Stern and Kilian 1996). However, Garrido et al. (2006) have suggested that restitic garnet may occur in the roots of island arcs at these depths in response to dehydration melting of amphibole-bearing plutonic rocks. Schiano et al. (2010) used trace element modelling on a database of 700 rocks from the Ecuadorian Andes to illustrate that mixing was the major control on their evolution. By plotting the ratios of compatible and incompatible elements, fractional crystallization, partial melting, and mixing processes can be distinguished from each other (Allgre and Minster 1978; Schiano et al. 2010). By plotting an incompatible element versus the ratio of that incompatible element and a compatible element (e.g., Rb versus Rb/V), mixing and fractional crystallization will form a curved trend whereas partial melting forms a linear trend (Fig. 11a). In the GVC, the Black Tusk and Paul Ridge andesites show a clear mixing or fractional crystallization process; the Ring Creek and Barrier andesites only show a slight curved trend. The limited data from the Columnar
Fig. 11 Incompatible/ compatible element ratio plots for: a Rb versus Rb/V and b 1/V versus Rb/V distinguishing mixing from both partial melting and fractional crystallization processes. Inset schematics are modified after Schiano et al. (2010). In plot a, a curved trend is generated by either mixing or fractional crystallization where a linear trend is indicative of partial melting. To isolate the mixing relationship, a companion plot must be used in tandem. The linear trend in plot b illustrates the dominance of mixing in the GVC rocks and not fractional crystallization (or partial melting). Symbols as in Fig. 4

Peak dacite cannot show a curved versus linear correlation, but in the GVC as a whole, illustrates a curved array. To isolate mixing from fractional crystallization, a companion plot is needed where the incompatible/compatible element ratio is plotted against 1/compatible element (1/V versus Rb/V, Fig. 11b). On this companion diagram, mixing creates the linear trend and partial melting and fractional crystallization plots as the curve; all the GVC lavas plot as linear trends. Figure 11 also illustrates that the mixing trends reflect magma mixing and not mixing of sources. Partial melting of a heterogeneous source would significantly modify the incompatible/compatible element ratio, whereas ratios of incompatible elements only would not be affected (Langmuir et al. 1978; Schiano et al. 2010). This suggests that the mixing relations observed in the GVC occurred in the magma chamber (or chambers) beneath each center after separation from their solid source and argues against significant fractional crystallization processes controlling the chemistry of the GVC adakites.

Conclusions Petrographic and geochemical examination of the intermediate rocks erupted at some of the centers that comprise the

Bull Volcanol (2013) 75:730

Page 21 of 23, 730

GVC illustrate that the andesites and dacites are adakitic in character. Volcanic rocks from Paul Ridge, the Ring Creek flow, and Columnar Peak in the MGVF were compared with rocks from the Black Tusk and the Barrier flow in the GLVF as well as previously published data. All the rocks from the GVC exhibit Sr/Y>40, low Yb (<1.9 ppm), low Y (<19 ppm), low K2O/Na2O (<0.4), high Al2O3 (>17 wt.%) and high Mg# (51). These data support a model of slab partial melting over other models for adakite genesis. Partial melting of mafic lower crust cannot explain the adakite character of the GVC magmas because the continental crust is thought to be too thin (35 km) for garnet to be stable. Modelling of incompatible element ratios suggest that fractional crystallization processes are not dominant and that mixing of magmas within small intermittent reservoirs beneath each center is responsible for the geochemical variations in the GVC adakites. The variations in geochemistry that are not adakitic (low La, low Cr) are the result of mixing between a pre-existing HSA magma and intruding basaltic to andesitic magmas that are non-adakitic (or near-adakitic). Evidence of magma mixing is abundant in the petrography of the GVC lavas, which include differently sized plagioclase populations, sieve-textured crystals, xenoliths, xenocrysts, and reaction rims (Fig. 3al). The intruding, more mafic magmas are cryptic, and further data are required to determine their origins and role in adakite genesis in the GVC. The lack of a clear distinction between HSA and LSA groups in the investigated GVC rocks (Figs. 9 and 10) may also result from magma mixing. The rocks of the GVC exhibit characteristics from both groups; the Paul Ridge andesite rocks have low SiO2 and Rb, elevated REE, and CaO and Na2O values typical of LSA, but low Sr, Nb and a Ti anomaly, suggesting mixing with a more normal arc-basalt composition. Based on previously published data, The Table andesite follows all the geochemical requirements and is the best representative of LSA in the GVC. The Ring Creek andesite and the Columnar Peak dacite are best characterized as HSA with respect to major elements and Sr contents but have higher Nb and LREE than are typical for HSA and suggests mixing with an intruding magma of similar composition that was likely close to an adakitic composition. The Barrier and Black Tusk andesites are somewhat transitional between LSA and HSA, plotting in one field or the other, based on the discrimination diagram (see Figs. 9 and 10). The key difference between LSA and HSA is the silica content (60 wt.% SiO2) and with both the Black Tusk and Barrier andesites containing approximately 60 wt.% SiO2, this may also contribute to the transitional character of the andesites in addition to the magma mixing effects. The lack of a clear distinction between LSA and HSA in the GVC may reflect the limited sample size in this study; additional data are required before any interpretations can be made as to whether the GVC is dominated by

LSA or HSA magmas and subsequently, further constraints on the processes responsible for the generation of the GVC adakites.
Acknowledgments This work was funded under a Natural Sciences and Engineering Research Council of Canada Discovery grant (IMC). Additional funding was provided from the University of Regina in the form of graduate student scholarships to JF. We thank the School of Earth and Ocean Sciences, University of British Columbia, for access to unpublished data relating to the study area. We also thank Catherine Hickson, Melanie Kelman, Dominique Weis, and Michael Clynne for constructive reviews and perspectives that have greatly improved the manuscript.

References
Allgre CJ, Minster JF (1978) Quantitative models of trace element behavior in magmatic processes. Earth Planet Sci Lett 38:125 Audet P, Bostock MG, Mercier JP, Cassidy JF (2008) Morphology of the ExplorerJuan de Fuca slab edge in northern Cascadia: imaging plate capture at a ridge-trench-transform triple junction. Geol 36:895898 Bourdon E, Eissen JP, Gutscher MA, Monzier M, Samaniego P, Robin C, Bollinger C, Cotton J (2002) Slab melting and slab melt metasomatism in the Northern Andean Volcanic Zone: adakites and high-Mg andesites from Pichincha volcano (Ecuador). B Soc Geol Fr 173:195206 Brooks GR, Friele PA (1992) Bracketing ages for the formation of the Ring Creek lava flow, Mount Garibaldi volcanic field, southwestern British Columbia. Can J Earth Sci 29:24252428 Castillo PR, Janney PE, Solidum RU (1999) Petrology and geochemistry of Camiguin Island, southern Philippines: insights to the source of adakites and other lavas in a complex arc setting. Contrib Miner Pet 134:3351 Castillo PR (2006) An overview of adakite petrogenesis. Chin Sci Bull 51:257268 Chiaradia M (2009) Adakite-like magmas from fractional crystallization and meltingassimilation of mafic lower crust (Eocene Macuchi arc, western Cordillera, Ecuador). Chem Geol 265:468487 Chiaradia M, Mntener O, Beate B, Fontignie D (2009) Adakite-like volcanism of Ecuador: lower crust magmatic evolution and recycling. Contrib Miner Pet 158:563588 Coldwell B, Adam J, Rushmer T, Macpherson CG (2011) Evolution of the East Philippine Arc: experimental constraints on magmatic phase relations and adakitic melt formation. Contrib Miner Pet 162:835848 Defant MJ, Drummond MS (1990) Derivation of some modern arc magmas by melting of young subducted lithosphere. Nat 367:662665 Drummond MS, Defant MJ (1990) A model for trondhjemite-tonalitedacite genesis and crustal growth via slab melting: Archean to modern comparisons. J Geophys Res 95:2150321521 Falloon TJ, Danyushevsky LV, Crawford AJ, Meffre S, Woodhead JD, Bloomer SH (2008) Boninites and adakites from the Northern Termination of the Tonga Trench: implications for adakite petrogenesis. J Petrol 49:697715 Garrido CJ, Bodinier JL, Burg JP, Zeilinger G, Hussain S, Dawood H, Nawaz Chaudhry M, Gervilla F (2006) Petrogenesis of mafic garnet granulite in the lower crust of the Kohistan Paleo-arc Complex (Northern Pakistan): implications for intra-crustal

730, Page 22 of 23 differentiation of island arcs and generation of continental crust. J Petrol 47:18731914 Geochemical Earth Reference Model (GERM) Reservoir Database (2013) earthref.org/GERMRD/datamodel/ oceanic crust, NMORB, primitive mantle, dacite, andesite. Accessed: Feb 14, 2013 Govindaraju K, Potts PJ, Webb PC, Watson JS (1994) Report on Whin Sill dolerite WS-E from England and Pitscurrie microgabbro PMS from Scotland: assessment by one hundred and four international laboratories. Geostand Newsl 18:211300. doi:10.1111/ j.1751-908X.1994.tb00520.x Green NL (1977) Multistage andesite genesis in the Garibaldi Lake area, southwestern British Columbia. University of British Columbia, PhD Dissertation, 265p Green NL (1981) Geology and petrology of Quaternary volcanic rocks, Garibaldi Lake area, southwestern British Columbia. Geol Soc Am Bull 92:697702 Green NL (1990) Late Cenozoic volcanism in the Mount Garibaldi and Garibaldi Lake volcanic fields, Garibaldi volcanic belt, southwestern British Columbia. Geosci Can 17:171174 Green NL (2006) Influence of slab thermal structure on basalt source regions and melting conditions: REE and HFSE constraints from the Garibaldi volcanic belt, northern Cascadia subduction system. Lithos 87:2349 Green NL, Henderson P (1984) Rare earth element concentrations in Quaternary volcanic rocks of southwestern British Columbia. Can J Earth Sci 21:731736 Green NL, Armstrong RL, Harakal JE, Souther JG, Read PB (1988) Eruptive history and K-Ar geochronology of the late Cenozoic Garibaldi volcanic belt, southwestern British Columbia. Geol Sci Am Bull 100:563579 Green NL, Harry DL (1999) On the relationship between subducted slab age and arc basalt petrogenesis, Cascadia subduction system, North America. Earth Planet Sci Lett 171:367381 Green NL, Sinha AK (2005) Consequences of varied slab age and thermal structure on enrichment processes in the sub-arc mantle of the northern Cascadia subduction system. J Volcanol Geotherm Res 140:107132 Gutscher MA, Maury RC, Eissen JP, Bourdon E (2000) Can slab melting be caused by flat subduction? Geol 28:535538 Hickson CJ, Russell JK, Stasiuk MV (1999) Volcanology of the 2350 B.P. eruption of Mount Meager volcanic complex, British Columbia, Canada: implications for hazards from eruptions in topographically complex terrain. B Volcanol 60:489507 Huang F, He Y (2010) Partial melting of the dry mafic continental crust: implications for petrogenesis of C-type adakites. Chin Sci Bull 55:24282439 Ickert RB, Thorkelson DJ, Marshall DD, Ullrich TD (2009) Eocene adakitic volcanism in southern British Columbia: remelting of arc basalt above a slab window. Tectonophysics 464:164185 Kay RW (1978) Aleutian magnesian andesites: melts from subducted Pacific Ocean crust. J Volcanol Geotherm Res 4:117132 Kelley KA, Plank T, Ludden J, Staudigel H (2003) Composition of altered oceanic crust at ODP sites 801 and 1149. Geochem Geophys Geosyst 4(8910):21p Langmuir CH, Vocke RD Jr, Hanson GN, Hart SR (1978) A general mixing equation with applications to Icelandic basalts. Earth Planet Sci Lett 37:380392 Le Bas MJ, Streckeisen AL (1991) The IUGS systematics of igneous rocks. J Geol Soc, London 148:825833 Lyubetskaya T, Korenaga J (2007) Chemical composition of Earths primitive mantle and its variance: 2. Implications for global geodynamics J Geophys Res 112(B03212):15p Macdonald GA (1968) Composition and origin of Hawaiian lavas. In: Coats RR, Hay RL, Anderson CA (eds) Studies in volcanology: a memoir in honor of Howel Williams, vol 116, Geol Sci Am Memoir., pp 477522

Bull Volcanol (2013) 75:730 Macpherson CG, Dreher ST, Thirlwall MF (2006) Adakites without slab melting: high pressure differentiation of island arc magma, Mindanao, the Philippines. Earth Planet Sci Lett 243:581593 Madsen JK, Thorkelson DJ, Friedman RM, Marshall DD (2006) Cenozoic to recent plate configurations in the Pacific basin: ridge subduction and slab window magmatism in western North America. Geosph 2:1134 Martin H (1999) The adakitic magmas: modern analogues of Archaean granitoids. Lithos 46:411429 Martin H, Moyen JF (2003) Secular changes in TTG composition: comparison with modern adakite. EGS-AGU-EUG Jt Meet, Nice, Apr VGP7-1 FR2O-001 Martin H, Smithies RH, Rapp R, Moyen JF, Champion D (2005) An overview of adakite, tonalite-trondhjemite-granodiorite (TTG), and sanukitoid: relationships and some implications for crustal evolution. Lithos 79:124 Mathews WH (1951) The Table, a flat topped volcano in southern British Columbia. Am J Sci 249:830841 Mathews WH (1952) Mount Garibaldi, a supraglacial volcano in southwestern British Columbia. Am J Sci 250:81103 Mathews WH (1958) Geology of the Mount Garibaldi map-area, southwestern British Columbia, Canada. Geol Sci Am Bull 69:161198 Moyen JF (2009) High Sr/Y and La/Yb ratios: the meaning of the adakitic signature. Lithos 112:556574 Panteeva SV, Gladkochoub DP, Donskaya TV, Markova VV, Sandimirova GP (2003) Determination of 24 trace elements in felsic rocks by inductively coupled plasma mass spectrometry after lithium metaborate fusion. Spectrochim Acta Pt B 58:341 350 Perry HKC, Eaton DWS, Forte AM (2002) LITH5.0: a revised crustal model for Canada based on lithoprobe results. Geophys J Int 150:285294 Rapp RP, Shimizu N, Norman MD, Applegate GS (1999) Reaction between slab-derived melts and peridotite in the mantle wedge: experimental constraints at 3.8 GPa. Chem Geol 160:335356 Richards JP, Kerrich R (2007) Special paper: adakite-like rocks: their diverse origins and questionable role in metallogenesis. Econ Geol 102:537576 Riddihough RP (1981) One hundred million years of plate tectonics in western Canada. Geosci Can 9:2834 Riddihough RP (1984) Recent movements of the Juan de Fuca Plate system. J Geophys Res 89(B8):69806994 Schiano P, Clocchiatti R, Shimizu N, Maury RC, Jochum KP, Hofmann AW (1995) Hydrous silica-rich melts in the sub-arc mantle and their relationship with erupted arc lavas. Nat 377:595600 Schiano P, Monzier M, Eissen JP, Martin H, Koga KT (2010) Simple mixing as the major control of the evolution of volcanic suites in the Ecuadorian Andes. Contrib Miner Pet 160:297312 Sherrod DR, Smith JG (1990) Quaternary extrusion rates of the Cascade Range, northwestern United States and southern British Columbia. J Geophys Res 95:1946519474 Sigurdsson H, Houghton B, McNutt SR, Rymer H, Stix J (2000) Encyclopedia of volcanoes. Academic Press, California, 1417p Sivertz GWG (1976) Geology, petrology and petrogenesis of Opal Cone and Ring Creek lava flow, southern Garibaldi, British Columbia. B.Sc. Thesis, University of British Columbia 79p Stern CR, Kilian R (1996) Role of the subducted slab, mantle wedge and continental crust in the generation of adakites from the Austral Volcanic Zone. Contrib Miner Pet 123:263281 Sun SS, McDonough WI (1989) Chemical and isotopic systematics of oceanic basalts: implications for mantle composition and processes. In: Saunders AD, Norry MJ (eds) Magmatism in the ocean basins. Spec Pub 42, Geol Soc London 313345

Bull Volcanol (2013) 75:730 Tatsumi Y, Hamilton DL, Nesbitt RW (1986) Chemical characteristics of fluid phase from the subducted lithosphere: evidence from high-pressure experiments and natural rocks. J Volcanol Geotherm Res 29:293309 Thorkelson DJ, Breitsprecher K (2005) Partial melting of slab window margins: genesis of adakitic and non-adakitic magmas. Lithos 79:2541 Turner JS, Campbell IH (1986) Convection and mixing in magma chambers. Earth Sci Rev 23:255352 Wilson DS (1988) Tectonic history of the Juan de Fuca ridge over the last 40 million years. J Geophys Res 93:1186311876

Page 23 of 23, 730 Wilson M (1989) Igneous petrogenesis. HarperCollinsAcademic, London, 485p Wilson DS (2002) The Juan de Fuca plate and slab: isochron structure and Cenozoic plate motions. In: Kirby SH, Wang K, Dunlop SG (Eds) The Cascadia subduction zone and related subduction systems. US Geol Surv Open File Rep 02328: 912 Xu J, Wang Q, Yu X (2000) Geochemistry of high-Mg andesites and adakitic andesite from Sanchazi block of the Mian-Lue ophiolitic melange in the Qinling mountains, central China: evidence of partial melting of the subducted Paleo-Tethyan crust. Geochem J 34:359377

You might also like