| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology Volume 41 Number 1 Pages 43-67 2000
© Oxford University Press 2000
‘Subduction Style’ Magmatism in a Non-subduction Setting: the Colville Igneous Complex, NE Washington State, USA
1DEPARTMENT OF EARTH AND ATMOSPHERIC SCIENCES, UNIVERSITY OF ALBERTA, EDMONTON, AB, T6G 2E3, CANADA
2DEPARTMENT OF GEOLOGY, WASHINGTON STATE UNIVERSITY, PULLMAN, WA 99164-2812, USA
Received June 24, 1998; Revised typescript accepted June 18, 1999
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONPETROLOGY AND MINERALOGYGEOCHEMISTRYPETROGENESIS: THE NATURE AND...CONCLUSIONSAPPENDIX: GEOCHEMICAL TECHNIQUESREFERENCES |
|---|
The Colville Igneous Complex is located within the Eocene Magmatic Belt of the North American Cordilleran interior. It straddles the US–Canadian border in northeast Washington and southern British Columbia. The complex consists of three intrusive and two extrusive phases, the first extrusive phase being contemporaneous with the latter two intrusive phases. As a consequence of sub-solidus re-equilibration in the plutonic rocks, this study concentrates on the two extrusive phases, the Sanpoil Volcanic Formation and the Klondike Mountain Formation. The Sanpoil Volcanic Formation consists of andesites, dacites and rare trachyandesites (SiO2 = 55–70 wt %) exhibiting a slight decrease in total alkalis (Na2O + K2O) with increasing silica. The Klondike Mountain Formation consists of basalts, basaltic andesites, andesites, dacites and rhyolites (SiO2 = 51–75 wt %) with total alkalis increasing with increasing silica. The calc-alkaline affinity of the rocks of the Colville Igneous Complex, coupled with the presence of a ‘subduction signature’ of enriched large ion lithophile elements (LILE) and depleted high field strength elements (HFSE), has traditionally been attributed to petrogenesis in a subduction-related magmatic arc, the ‘Challis Arc’. New trace and rare earth element and isotopic data (87Sr/86Sri,
Ndi, 
18O), however, suggest that this explanation is no longer tenable. We propose that the magmas of the Sanpoil Volcanic Formation were generated by mid-crustal partial melting of a mid-Proterozoic source and that the Klondike Mountain Formation was formed by varying degrees of mixing between two distinct late-Proterozoic lower-crustal sourced magmas and a mantle-derived magma. In all cases, the subduction or calc-alkaline signature was inherited from the Proterozoic crustal sources. The only magmas that can confidently be attributed to a mantle source are the basalts of the Klondike Mountain Formation, which show no decoupling of the LILE and HFSE, i.e. no subduction signature, precluding the presence of a subduction slab beneath this part of the North American Cordillera during the Eocene. We propose the alternative model: that the Colville Igneous Complex formed as a result of decompression melting of crust and mantle during post-Laramide orogenic collapse of an overthickened crust. KEY WORDS: US–Canadian Cordillera; inherited subduction signature; Eocene; extension; crustal recycling
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONPETROLOGY AND MINERALOGYGEOCHEMISTRYPETROGENESIS: THE NATURE AND...CONCLUSIONSAPPENDIX: GEOCHEMICAL TECHNIQUESREFERENCES |
|---|
The calc-alkaline affinity, as defined in both the original sense of the alkali-lime index (Peacock, 1931
) and more recently by major element (Miyashiro, 1974; Peccerillo & Taylor, 1976) and trace element data (Pearce, 1982, 1983; Thompson et al., 1984), of the Colville Igneous Complex (CIC), northeast Washington, has long been recognized (Muessig, 1962; Atwater & Rinehart, 1984; Moye, 1984; Holder & Holder, 1988). Such calc-alkaline signatures present in suites of rocks have been assumed to indicate contemporaneous subduction. On the basis of this assumption, the Colville Igneous Complex was originally interpreted as part of an early Eocene volcanic arc, stretching from the southern Yukon to Wyoming and sometimes referred to as the Challis Arc (e.g. Lipman et al., 1972; Coney & Reynolds, 1977; Armstrong, 1979; Ewing, 1980; Vance, 1982; Coney, 1987; O’Brien et al., 1988). This somewhat scattered and discontinuous ‘arc’ consists of a number of discrete igneous complexes and centres that form a narrow (
70 km wide) belt in northern British Columbia, widening to 250 km between the 56° and 48° parallels, and a broad and diffuse field of between 500 and 1000 km width between 48° and 42° of latitude (Fig. 1). The majority of these complexes, including the Colville Igneous Complex, are hosted within accreted material of the Canadian and US Cordillera, but a number of the southern complexes are contained within the Rockie Mountain Foreland belt, or cratonic ancient North America.
|
Several models have been proposed for the formation of this early Eocene magmatic belt, most, because of the predominantly calc-alkaline nature of the rocks, invoking some form of contemporaneous subduction tectonics. Coney & Reynolds (1977) developed a model for late Mesozoic and early Cenozoic igneous rocks in the southwestern USA that involved the gradual shallowing of the angle of subduction as progressively younger, hotter and therefore more buoyant oceanic crust was subducted beneath North America, then rollback of the subducted slab as the rate of convergence reduces to form the broad and diffuse range of volcanic fields observed there. This model has been extrapolated north by several workers to account for early Eocene igneous rocks in the northwestern USA (e.g. Ewing, 1980; Coney, 1987). Recent work has pointed to spatial and age problems with a subduction model, and has questioned the existence of the Challis Arc (e.g. Thorkelson, 1989; Dudas, 1991; Norman & Mertzman, 1991; Varseck & Cook, 1994; Wagoner et al., 1994; Hooper et al., 1995; Morris & Hooper, 1997; Morris & Creaser, 1998). Some of the main problems are as follows:
- To place the subducted slab at the depth required to produce calc-alkaline magmatism beneath the CIC (75–120 km; Wyllie, 1984) would require an improbably low angle of subduction, 6°. The lowest angle of subduction at modern analogues, 10°, occurs in the Peruvian and central Chilean portions of the Andean active continental margin which, significantly, lack active volcanism (Jarrard, 1986).
- Tectonic and experimental models for magma genesis within a subduction zone normally predict that magmatism becomes more alkalic with increasing depth of the subducting slab beneath the overriding plate. Dudas (1991) observed that, for rocks of the same age south of the US–Canadian border, there is no such clear spatial distribution of rock composition. On the same principle, in a slab rollback scenario (e.g. Coney, 1987) the depth to slab beneath a particular location in the magmatic arc, and therefore the alkalic nature of the volcanism, should increase with time. In northeast Washington the opposite trend is observed (Morris, 1996).
- Modern geochronology work on igneous complexes throughout the Cordillera has shown that the flare-up of magmatism in the Eocene was, in the most part, contemporaneous throughout the ‘arc’, contrary to what would be expected if a subducted slab was rolling back beneath western North America.
Such observations have led to the conclusion that the calc-alkaline nature of Eocene volcanic rocks in the North American Cordillera was derived from a source created in a previous subduction event, residing in either the lithospheric mantle (Norman & Merzman, 1991; Hooper & Hawkesworth, 1993) or the crust (Morris & Hooper, 1997). The present paper uses petrological, chemical and isotopic data to develop an alternative petrogenetic model to subduction for the origin of the Colville Igneous Complex.
Geology of the Colville Igneous Complex
The Colville Igneous Complex (CIC) lies at the southern end of the Omineca crystalline belt, a morphogeological belt of mainly metamorphic and granitic rocks formed during the accretion of the Intermontane superterrane onto cratonic North America during the late Triassic and Jurassic (Gabrielse & Yorath, 1992; Fig. 1, inset). The CIC (Fig. 2) consists of Eocene intrusive, hypabyssal and extrusive rocks associated with extension, crustal thinning and the emplacement of the Kettle and the Okanogan Metamorphic Core Complexes. The CIC rocks are distinguished from earlier, Cretaceous, granites and granodiorites by their association with extensional emplacement mechanisms, structures and fabrics that can be traced from the gneisses of the metamorphic core complexes into the intrusive phases of the CIC (Holder, 1985; Holder & Holder, 1988). The Cretaceous rocks all show evidence of intrusion within a compressive, mainly top to the east overthrusting tectonic regime (Morris, 1996). The complex is bounded to the east by rocks of the Kooteny Arc (a non-volcanic arc-shaped assemblage of Palaeozoic clastic and carbonate rocks, strongly deformed in the mid-Jurassic) and to the west by the Okanogan Valley Fault. To the south the complex is overlain by the Miocene Columbia River Basalts. Arbitrarily, the northern border of the CIC is defined as the US–Canadian border (Muessig, 1962, 1967; Holder & Holder, 1988), but it is almost certainly the direct equivalent of the Ladybird Suite and related rocks of southern British Columbia (see Sevigny et al., 1989; Carr, 1992). The emplacement of the complex was synchronous with the formation of the metamorphic core complexes and four NNE trending grabens that cut them, the Toroda graben, the Republic graben, the Keller graben and the First Thought graben (Fig. 2; Holder et al., 1990).
|
Three distinct phases of igneous activity have been identified in the CIC (Holder & Holder, 1988; Carlson & Moye, 1990; Holder et al., 1990; Carlson et al., 1991). These are, from oldest to youngest, the Keller Butte Suite of granites and granodiorites; the Sanpoil Volcanic Formation (SVF) of andesites, dacites and rare trachyandesites, which has been correlated with hypabyssal andesites, dacites and trachyandesites of the Scatter Creek Formation, the diorites and granodiorites of the Devils Elbow (lower SVF) and the monzonites, monzogranites and granites of the Herron Creek Suite (upper SVF) (Holder, 1990); and the Klondike Mountain Formation (KMF) of hypabyssal and extrusive basalts to rhyolites. There is no known intrusive equivalent to the KMF.
The Keller Butte Suite has been dated at between 61·3 and 49·9 Ma (Atwater & Rinehart, 1984; Carlson et al., 1991). Most of these ages were, however, obtained by the K–Ar method, which is notoriously susceptible to resetting by subsequent events. More reliable U–Pb dates have been obtained from titanite and monazite, ranging from 55·0 ± 3·0 to 54·0 ± 2·0 Ma, with Rb–Sr dating techniques suggesting ages between 58·8 and 55·0 Ma (Carlson et al., 1991). Few reliable ages are available for the SVF and its intrusive equivalents. K–Ar ages range from 50·3 ± 0·3 to 45·2 ± 1·1 Ma (Atwater & Rinehart, 1984) but, like the Keller Butte Suite, these reflect the last thermal event in the area rather than the emplacement of the SVF and its intrusive equivalents. The best available data, from the intrusive equivalents, are as follows: the Devils Elbow Suite has been dated at 53·0 ± 2·0 Ma and the Herron Creek Suite at 51·3 ± 0·1 Ma by the U–Pb method, using the sensitive high-resolution ion microprobe (SHRIMP) on zircon rims (Wooden & Box, 1996). On the basis of Ar–Ar geochronology of related core complex mylonites, Berger & Snee (1992) concluded that this phase of magmatic activity had ceased by 49 ± 1 Ma. The stratigraphically higher Klondike Mountain Formation has been dated at 48·8 ± 1 Ma by Ar–Ar isochron methods (Berger & Snee, 1992), and marks the final stage of magmatic activity in the CIC.
Because of evidence of sub-solidus re-equilibration in intrusive members of the CIC, first reported by Holder (1990), this study concentrates on the extrusive Sanpoil Volcanic and Klondike Mountain Formations. These are thought to provide the best representation of original magmatic chemistry and therefore are the most appropriate for petrogenetic purposes.
|
PETROLOGY AND MINERALOGY |
|---|
|
TOPABSTRACTINTRODUCTIONPETROLOGY AND MINERALOGYGEOCHEMISTRYPETROGENESIS: THE NATURE AND...CONCLUSIONSAPPENDIX: GEOCHEMICAL TECHNIQUESREFERENCES |
|---|
Sanpoil Volcanic Formation
The Sanpoil Volcanic Formation (SVF) consists of andesites, dacites and rare trachyandesites, erupted from a number of distinct volcanic centres within the Republic, Toroda, Keller and First Thought grabens (Fig. 2). Estimates of total thickness range from 1500 m in the southern Republic Graben (Moye, 1984
) to 1000 m in the Toroda Graben (Pearson, 1967). The earliest members are characterized by pyroclastic textures, with their groundmasses highly altered to clay minerals, and as such these altered rocks were not sampled for chemical analysis. The majority of the SVF was emplaced as volcanic flow units that, with the exception of a number of localized and economically viable hydrothermal deposits, show little or no alteration. Rocks of the SVF are highly porphyritic, containing between 30 and 70 vol. % phenocrysts in a cryptocrystalline to microcrystalline groundmass. Holder (1990) divided the SVF into an upper and lower unit based upon phenocryst assemblage and geochemistry (see later). The lower Sanpoil Volcanic Formation (lSVF) is characterized by plagioclase, hornblende and rare clinopyroxene phenocrysts in a groundmass of hornblende, plagioclase, K-feldspar and quartz, whereas the upper Sanpoil Volcanic Formation (uSVF) is characterized by plagioclase, biotite and clinopyroxene phenocrysts in a biotite, pyroxene, plagioclase, K-feldspar and quartz groundmass (Holder, 1990). The transition between the lSVF and the uSVF is gradual, with flows of each unit interleaving, showing that the two magmas coexisted. In general, the mineralogy of the Devils Elbow and Herron Creek Suites closely follows that of the lSVF and uSVF, with detailed differences in the compositions of plagioclase, biotite and hornblende (Holder, 1990). The contacts between the Devils Elbow and Herron Creek Suites also show a transition zone of up to 100 m width (Holder & Holder, 1988).
Plagioclase
Two populations of plagioclase feldspar phenocrysts are distinguished in the SVF. The first consists of unzoned crystals that exhibit shear twinning, are cut by small-scale, en-echelon, (filled) fractures, and includes whole crystals that are broken by sub-parallel fractures. The second population of plagioclase crystals are oscillatory zoned, often euhedral, rarely broken, and display no evidence of shearing. Although petrographically distinct, these two populations are chemically indistinguishable. All plagioclase is andesine in composition, with the lSVF containing slightly more sodic plagioclase (An34–47, average An40) and the uSVF containing more calcic phenocrysts (An44–50, with one outlier at An36, average An46). Holder (1990) reported that lower SVF plagioclase tends to be more calcic than its plutonic equivalent, the Devils Elbow Suite (An30–42 with rare occurrences of up to An57, average An38), whereas the upper SVF plagioclase is more sodic than the correlated plutonic Herron Creek Suite (An43–50 with rare values as low as An36, average An42).
Pyroxene
Samples from the SVF displays two populations of clinopyroxenes (Fig. 3a), Wo42–45En37–39Fs17–21 and Wo27En48–50Fs23–25. These are clearly not in equilibrium and may represent evidence of inheritance from the magma source, or of magma mixing during the formation of the SVF.
|
Klondike Mountain Formation
The Klondike Mountain Formation (KMF) consists of basalts, basaltic andesites, andesites, dacites and rhyolites erupted from numerous graben-parallel fissures (now dykes) restricted within the bounding faults of the Republic, Toroda and First Thought grabens (Fig. 2). The formation is emplaced as flows and lava domes with interbedded sedimentary units that largely consist of slide breccias containing clasts derived from the graben walls and the SVF (Morris, 1996
; Suydam & Gaylord, 1997). The composite thickness of the entire igneous and sedimentary package exceeds 2500 m; however, total local thickness rarely exceeds 1500 m (Suydam & Gaylord, 1997). The KMF is almost aphyric, containing between 0 and rarely 20 vol. % phenocrysts (median <5%) in a glassy to fine, often devitrified glass matrix. Phenocrysts in samples of all rock types are predominantly plagioclase, clinopyroxene and orthopyroxene, with occasional olivine. Opaque oxides and hydrous phases are rarely seen in the KMF (Atwater & Rinehart, 1984; Carlson et al., 1991; Wagoner et al., 1994; Morris, 1996).
Plagioclase
Basalt and basaltic andesite members contain lath-like phenocrysts of andesine to labradorite (An45–55), in a groundmass largely made up of small lath-like plagioclase of the same composition range. In many samples both the phenocrysts and the groundmass plagioclase show a preferred flow orientation.
In contrast, the andesites to rhyolites contain two distinct populations of plagioclase. The first group consists of bytownite, ranging in composition from An71 to An82. They lack any zoning but are anhedral, and when in contact with the host groundmass have strongly etched rims. The phenocrysts contain numerous glass inclusions throughout the crystals, which are of an identical composition to the host feldspar. Microphenocrysts of pyroxene and olivine are also encased within these feldspars. The second plagioclase population consists of andesine and labradorite ranging in composition from An45 to An64, and is present both as an overgrowth of the bytownite and as discrete phenocrysts. These crystals are euhedral and show distinct oscillatory zoning with clean, unetched crystal rims. Many phenocrysts contain elongate glass inclusions, similar in composition to the dacitic members of the host Klondike Mountain Formation. Some crystals contain so much included glass that they are almost skeletal in nature, indicative of rapid crystal growth (Kawamoto, 1992).
Pyroxene
Basalt and basaltic andesite members of the Klondike Mountain Formation contain small phenocrysts of clinopyroxene (Wo34–41En42–51Fs10–16) and orthopyroxene (Wo3–4En78–80Fs17–18). Andesites, dacites and rhyolites of the Klondike Mountain Formation contain more calcic clinopyroxene (Wo42–44En41–44Fs12–16) and more Fe-rich orthopyroxene (Wo2–3En69–76Fs23–28). Equilibrium between coexisting OPX and CPX rims can be tested following the procedures of Nakamura & Kushiro (1970; also Hunter, 1998). At equilibrium, KDMg–Fe values should range from 0·75 to 1·08, which is calculated using the equation
 |
where XMg represents the molecular proportions of Mg/(Mg + Fe) (Nakamura & Kushiro, 1970). Microprobe analysis of phenocryst rims give XMg values for OPX ranging from 0·61 to 0·63, and for CPX ranging from 0·62 to 0·68, resulting in KD values of 0·75–1·02 (average 0·88) indicating that OPX and CPX were in a state of quasi-equilibrium in the KMF and thus allowing the use of the two-pyroxene geothermometer of Lindsley (1983). Two-pyroxene geothermometry using microprobe analyses of phenocryst rims yield temperatures of 1100–1200°C for basalts, 1000° and 1100°C for basaltic andesites, and 800° and 900°C for the dacites and rhyolites [Fig. 3b; after Lindsley (1983)].
Olivine
Olivine occurs as a minor phase in all rock types of the Klondike Mountain Formation. In basalts and basaltic andesites, olivine occurs as small, euhedral microphenocrysts. In andesites, dacites and rhyolites, the olivine develops distinct reaction rims wherever it is in direct contact with the host magma. Reaction rims are not present when the olivine rims are in direct contact with calcic plagioclase. Wagoner (1992) reported microprobe data from olivines from this formation. Basalts and basaltic andesites of the Klondike Mountain Formation contain normally zoned olivines of Fo75–79 whereas silicic members of the formation contain unzoned and normally zoned olivines with cores of Fo80–85 and rims of Fo75–83.
|
GEOCHEMISTRY |
|---|
|
TOPABSTRACTINTRODUCTIONPETROLOGY AND MINERALOGYGEOCHEMISTRYPETROGENESIS: THE NATURE AND...CONCLUSIONSAPPENDIX: GEOCHEMICAL TECHNIQUESREFERENCES |
|---|
Whole-rock analyses were carried out by X-ray fluorescence (XRF) and inductively coupled plasma mass spectrometry (ICP-MS) at the Geoanalytical Laboratory of Washington State University. Fifty-six new analyses were obtained for the SVF and the KMF. Results are presented in Table 1 and analytical techniques are discussed in the Appendix. Additional major and limited trace element data for the SVF and the KMF were obtained from Holder (1990)
and Wagoner (1992), respectively.
|
Sanpoil Volcanic Formation
The SiO2 contents of the lavas from the SVF range from 55 to 70 wt %. Total alkalis (Na2O + K2O) decrease slightly with increasing silica (Fig. 4). The SVF is alkali-calcic in the original classification of Peacock (1931). When the ratio of FeO* (total iron reported as FeOT) to MgO is plotted against SiO2 (after Miyashiro, 1974), the SVF follows a typical calc-alkaline trend, showing little or no iron enrichment with increasing silica (Fig. 5). On a plot of SiO2 vs K2O (after Peccerillo & Taylor, 1976) the majority of the SVF would be classified as high-potassium calc-alkaline rocks (Fig. 6a) with the lSVF showing an overall lower concentration of K2O than the uSVF, except at the lower silica values (60 wt % SiO2).
|
|
|
There is little change in Al2O3 content throughout the silica range observed in the SVF (Fig. 7a), whereas FeO, MgO, MnO, TiO2, CaO and P2O5 show a marked decrease with increasing silica. Na2O shows a slight increase with increasing silica and no clear trend can be discerned for K2O, although total alkalis (Na2O + K2O) show a slight increase (Fig. 5). The mg-number {[(MgO/40·304)/(MgO/40·304) + (FeOT/71·839)] x 100} decreases with increasing silica (Fig. 7b). Sr, conventionally an incompatible element, shows a marked decrease with increasing silica (Fig. 7c). Although overall Zr also shows a decrease with increasing silica, within the upper and lower SVF little or no variation can be discerned other than scatter (Fig. 7d). Similar trends can be recognized in the incompatible trace elements Ba, Rb and Nb, but not in the more compatible major and trace elements.
|
Patterns for chondrite-normalized REE (rare earth elements) for both the uSVF and the lSVF samples analysed are similar, showing a near-constant gradient, regardless of SiO2 content [Fig. 8a; normalization factors from Boynton (1984)
]. This does, however, illustrate a somewhat higher concentration of REE in the uSVF, with the exception of one analysis representing one sample from a flow near the top of the formation (Figs 7e and f, and 8a). This returns significantly lower concentrations of REE, similar to rhyolites of the overlying KMF, suggesting coexistence and interaction between the two magmas. There is no significant Eu anomaly (Eu/Eu* = 0·85–0·9; Fig. 9), regardless of Sr content, in the SVF, which, taken with the constant Al2O3 content, and despite the large drop in Sr with increasing silica, virtually precludes the fractionation of plagioclase as a major control in the petrogenesis of this formation.
|
|
MORB (mid-ocean ridge basalt)-normalized trace element plots, after Pearce (1982), display the classic subduction-related signature of relatively enriched LILE (large ion lithophile elements; Sr, K, Rb, Ba, and Th) with relatively depleted HFSE (high field strength elements; Ta, Nb, and Ti; Fig. 10a). There are two groups: (1) the samples containing slightly lower concentrations of incompatible elements come from the lSVF; (2) the samples with higher concentrations come from the uSVF.
|
Klondike Mountain Formation
SiO2 contents of the KMF range from 51 to 75 wt % with total alkalis increasing with increasing silica (Fig. 5). The Klondike Mountain Formation is calc-alkaline in the original classification of Peacock (1931)
and encompasses both the calc-alkaline and high-K calc-alkaline field of Peccerillo & Taylor (1976). Like the Sanpoil Volcanic Formation, the Klondike Mountain Formation follows a typical calc-alkaline trend when the ratio of FeOT to MgO is plotted against SiO2, with significant iron enrichment occurring only in the most silicic members of the formation (Fig. 6b; Miyashiro, 1974). Calculated mg-numbers range from 12 for rhyolite end members of the formation to over 60 for the basaltic to andesite members (Fig. 7b). There is a slight negative correlation between mg-number and Zr.
Chondrite-normalized REE plots for KMF samples (Fig. 8b) show a more complex pattern than do those from the Sanpoil Volcanic Formation [normalization factors from Boynton (1984)] (Fig. 8). There is an initial increase in light REE (LREE) concentrations of 50% between basalts and andesites followed by a decrease in all REE of 90% as silica increases from andesite to rhyolite (Figs 7e and f, and 8b). A small Eu anomaly develops only in the higher-silica, lowest-Sr members of the series (Fig. 9).
MORB-normalized incompatible element plots (after Pearce, 1982) for the KMF display a range of forms that are best examined in relationship to the rock type (Fig. 10b). Dacites and rhyolites display the classic subduction signature of relatively enriched LILE and relatively depleted HFSE, whereas basalts show a strong intraplate chemical signature of LILE enrichment in the absence of HFSE depletion (e.g. Thompson et al., 1984). A continuum does, however, exist between the high- and low-silica end members, represented by basaltic andesites and andesites.
Radiogenic isotopes
Nineteen samples from the Sanpoil Volcanic Formation and the Klondike Mountain Formation were analysed for Sr and Nd isotopes at the University of Alberta (Table 2; see Appendix). Initial ratios were calculated assuming an age of 54 Ma for the Sanpoil Volcanic Formation, and 49 Ma for the Klondike Mountain Formation [ages after Carlson et al. (1991), Berger & Snee (1992) and Wooden & Box (1996)]. These data provide the clearest distinction between the two formations; the SVF samples have 87Sr/86Sri ratios between 0·70688 and 0·70919, and Ndi values between -6·0 and -15·3, whereas the KMF samples have 87Sr/86Sri ratios between 0·70397 and 0·70650, and Ndi values between +4·0 and -6·1 (Fig. 11). The SVF shows a negative correlation between silica and 87Sr/86Sri (Fig. 12) and a positive correlation with Ndi (Fig. 13), with the lower-silica members (generally of the uSVF) showing the most crustal (high 87Sr/86Sri and low Ndi) values. In the KMF, the highest-silica members show the lowest Sri values and the lowest-silica members show the higher (mantle) values (Fig. 12). Ndi values show two arrays of data (Fig. 13), suggesting two distinct sources for the silicic members of the KMF, both with low 87Sr/86Sri, one returning Ndi values near CHUR and the other returning low (crustal) Ndi.
|
|
|
|
Sm–Nd model ages were calculated for the SVF assuming a depleted mantle source (after Goldstein et al., 1984; Nelson & DePaolo, 1984; Arndt & Goldstein, 1987; Table 2). The resultant ‘crustal residence’ ages, i.e. the time since the rock or its sources were initially removed from the mantle, form an array between 1·6 and 2·0 Ga. These results correspond well to ages reported from the intrusive equivalents of the Sanpoil Volcanic Formation of 1·7 Ga [inherited Pb component, D. Parkinson, personal communication cited by Carlson & Moye (1990)] and 1·63 Ga (U–Pb SHRIMP analyses of zircon cores; Wooden & Box, 1996). Calculated Sm–Nd model ages for the Klondike Mountain Formation, assuming a depleted-mantle source (after Goldstein et al., 1984; Nelson & DePaolo, 1984), identified a crustal residence age array between 1·2 Ga and Mesozoic values, with the rhyolite and dacite members of the Klondike Mountain Formation recording the oldest values whereas the basalts recorded the youngest.
Oxygen isotope results
Whole-rock oxygen isotope analyses were carried out on eight samples from the Sanpoil Volcanic Formation, and 14 samples from the Klondike Mountain Formation, covering the entire range of rock types from both formations (Table 2; Fig. 14). Samples were selected after careful examination of hand samples and thin sections for signs of alteration. In addition, only samples that returned totals of near 100% during XRF analysis, another indication of the unaltered state of a sample, were considered for oxygen analysis. The SVF samples have 18O values of +6·1 to +9·5
(VSMOW), with one exception at +1·8, believed to be the consequence of hydrothermal alteration. Data from the KMF range from +5·1 to +9·0, again with one exception at +2·7, probably as a result of low-grade hydrothermal alteration. The two low values are excluded in the following discussion.
|
A number of processes can be responsible for the variation in oxygen isotope ratios. Alteration of samples commonly results in the lowering of the 18O, often coupled with an increase in 87Sr/86Sri, resulting in a negative trend away from mantle values. With the exception of two samples, all 18O and 87Sr/86Sri data from both the SVF and the KMF plot on a horizontal or positive trend away from bulk mantle (Fig. 15); we believe that these results are a true representation of primary 18O values for both the SVF and the KMF. Muehlenbachs & Byerly (1982), studying the effects of fractional crystallization (FC) on the oxygen isotope composition of primitive igneous rocks from the Galapagos spreading centre, concluded that the maximum range of isotope ratios that could be generated by fractional crystallization alone was of the order of +1·5. Both the SVF and the KMF display a range in 18O greater than this, requiring the involvement of crustal processes other than FC in their formation (Figs 14 and 15).
|
|
PETROGENESIS: THE NATURE AND AGE OF MAGMA SOURCES |
|---|
|
TOPABSTRACTINTRODUCTIONPETROLOGY AND MINERALOGYGEOCHEMISTRYPETROGENESIS: THE NATURE AND...CONCLUSIONSAPPENDIX: GEOCHEMICAL TECHNIQUESREFERENCES |
|---|
The petrology and geochemistry of the Colville Igneous Complex have traditionally been attributed to contemporaneous subduction. In the introduction we discussed various reasons why contemporaneous subduction is not necessary and may be precluded from the genesis of the CIC and of other early Eocene igneous complexes in the northern US and Canadian Cordillera. Here we use the tectonic setting, age, and nature of the magmatic sources based on chemical and Rb–Sr, Sm–Nd, and oxygen isotopic data, to suggest a different petrogenesis.
Sanpoil Volcanic Formation
The Sanpoil Volcanic Formation contains the range of rock types and the chemical signatures, major element calc-alkaline trends as well as decoupled LILE and HFSE, normally associated with subduction-related arcs. Key major, trace and rare earth element data, however, fail to show any evidence of the large-scale fractionation that would be required to produce the range of rock types observed in the SVF from a mantle source. Fractional crystallization (FC), or coupled assimilation and fractional crystallization (AFC), would be expected to produce a pronounced decrease in Al2O3 and Sr coupled with a negative Eu anomaly (removal of plagioclase); the reduction in Al2O3 and Eu is not observed here. A marked reduction of middle (MREE) and heavy REE (HREE), relative to LREE would indicate the removal of mafic minerals; this trend is not seen in the REE profiles (Fig. 8a), which show a remarkably consistent form and gradient regardless of the samples’ silica content. Likewise, there is no significant variation of Zr, a largely incompatible element that should increase with FC and AFC until the removal of zircon starts.
Alternative mechanisms to FC and AFC for producing chemical variation in a magmatic suite are partial melting (PM) and magma mixing. The evidence from the SVF is equivocal. Relict plagioclase and two populations of CPX could reflect inheritance from a source, incorporated into a magma during PM, or contributions from two distinct magmas brought together by mixing. Likewise, the spread of isotopic data (Figs 11–13) could be interpreted as representing either the isotopic ratios of a heterogeneous crustal source or a mixing array between two or more magmas with different isotopic ratios. In reality, the petrogenesis of the SVF probably incorporates components of both.
The source of the magmas that formed the SVF is thought to be entirely crustal. Although mixing between mantle-derived magmas and crustal partial melts is common, particularly in subduction-related magmatic arcs (e.g. Smith & Leeman, 1987, 1993; Borg & Cline, 1998; Morris & Creaser, 1998), the lower-silica members of the SVF return the most crustal isotopic ratios (low Ndi, high Sri, Figs 11–13), opposite to what would be expected if a mantle source was involved. The isotopic data do, however, suggest two distinct crustal sources, which roughly correspond, with some overlap, to the lSVF and uSVF. Although Nd model age data give only a rough idea of crustal residence age [see Arndt & Goldstein (1987) for detailed discussion], these results suggest an early Proterozoic age for the sources of the SVF, an age that corresponds well to ages obtained from the intrusive equivalents of the SVF of 1·7 Ga [inherited Pb component, D. Parkinson, personal communication cited by Carlson & Moye (1990)] and 1·63 Ga (U–Pb SHRIMP analyses of zircon cores; Wooden & Box, 1996). In conclusion, although contemporaneous subduction cannot be ruled out as the heat source on these results alone, the data suggest that the formation of the SVF magmas was an entirely crustal process involving partial melting of a heterogeneous early Proterozoic crust.
Klondike Mountain Formation
Like the SVF, the Klondike Mountain Formation presents many features that are often seen at subduction-related magmatic arcs. These include rock types that range from basalt to rhyolite, most of which display the classic signatures of subduction: they are either calc-alkaline or high-K calc-alkaline and contain enriched LILE coupled with depleted HFSE. Significantly, however, the basaltic and basaltic andesite members of the formation are both tholeiitic and show little or no decoupling of HFSE and LILE. Whereas the presence of both tholeiitic and calc-alkaline magmas is well documented in many subduction-related volcanoes (e.g. Conrey et al., 1997; Hunter, 1998), the lack of decoupled LILE and HFSE suggests that the mantle source for these basalts had not been contaminated by fluids derived from a subducting slab. Although both rhyolites and dacites do exhibit a small Eu anomaly, and rhyolites show a small amount of HREE depletion relative to lower-silica members of the formation, there is no evidence of the large-scale fractionation that would be required to produce the full range of KMF magmas from a basaltic source. This suggests that FC and AFC processes are secondary to the formation of the KMF.
There is abundant evidence of magma mixing in the KMF. The presence of two groups of plagioclase in the more silicic members (andesites, dacites and rhyolites), one of which shows marked etching on faces in contact with groundmass (i.e. magma), coupled with the presence of partly resorbed olivine, even in the most silicic members of the formation, is strong evidence for the mixing of a low- and high-silica magma. The survival of olivines coupled with the skeletal nature of the more sodic plagioclase overgrowths attests to the short timespan between mixing and eruption of these magmas; indeed, the mixing event may have been the trigger for eruption.
Chemical and isotopic data suggest at least three distinct magma sources, one mantle and two crustal, contributed to the KMF. Rhyolites in the KMF represent two crustal sources, one with low Sri and Ndi values close to CHUR, and the second with low Sri and low (crustal) Ndi. Sri values in this range are unusual in crustal-derived rocks, and are normally attributed to small degrees of alteration, particularly in rocks with low Sr concentration. Rhyolites from the KMF are, however, pristine, often preserving glass or cryptocrystalline matrixes that would not survive alteration, and contain sufficient concentrations of Sr that small degrees of alteration should not affect the isotopic system unduly. Oxygen isotopes also fail to display the depletion that is the normal result of such alteration. It is therefore thought that these low Sri ratios are a reflection of the crustal source. The only known rock types with such low 87Sr/86Sr coupled with low Ndi that could produce silicic magmas are ancient lower-crustal rocks that have undergone Rb depletion during granulite facies metamorphism (e.g. Whitehouse et al., 1998). Lower-crustal xenoliths, which have undergone Rb depletion during granulite facies metamorphism of the Hudsonian orogeny (1·7 Ga), and with the appropriate low 87Sr/86Sri and low Ndi, have been recovered from early Eocene dykes in central Montana (Fig. 1; Joswaik, 1992). It is reasonable to assume that similar lower crust might exist beneath northeast Washington. The mantle source, represented by the basaltic end member, shows no sign of the decoupled HFSE–LILE signature normally associated with subduction; it does, however, return isotopic values typical of mantle-derived material. A distinct mixing array can be recognized both in the isotopes and in the chemistry, with a gradual increase in the decoupled nature of the HFSE and LILE with increasing silica and decreasing Sri. We conclude that the KMF magmas resulted from mixing between a mantle source basaltic magma and at least two lower-crustal sourced silicic magmas, the calc-alkaline nature and decoupled HFSE–LILE signatures being inherited from these crustal sources. There is a small overprint of fractional crystallization recognized, particularly in the REE. In such a model subduction is not necessary, and may even be precluded by the presence of uncontaminated basaltic magmas, the only members of the KMF that can be confidently assigned to a contemporaneous mantle source.
The Colville Igneous Complex
The original interpretation for the formation of the Colville Igneous Complex and other early Eocene igneous complexes in the northern US and Canadian Cordillera invoked contemporaneous subduction, largely based upon the chemical affinity of these rocks (e.g. Ewing). Several recent studies have, however, suggested that this direct association of contemporaneous subduction with calc-alkaline magmatism and the presence of the ‘subduction signature’, the decoupled HFSE and LILE, may not always hold true (e.g. Ewart et al., 1992; Hawkesworth et al., 1995; Hooper et al., 1995). In many cases, such as the Cretaceous volcanics of eastern Queensland, Australia, which are associated with the initial rifting and opening of the Tasman Sea (Ewart et al., 1992), and basalt to rhyolite volcanism in the Colorado River Trough, USA, which was the Miocene locus of the Basin and Range extensional orogeny (Hawkesworth et al., 1995), the obvious tectonic setting of the calc-alkaline suite is clearly at odds with a contemporaneous subduction model. Many researchers concluded that the calc-alkaline affinity, including the trace element ‘subduction signature’, was inherited from a previous subduction event, having resided since this time within either the crust (e.g. Hawkesworth et al., 1995; Morris & Hooper, 1997) or the lithospheric mantle (e.g. Hooper & Hawkesworth, 1993).
Detailed major, trace, rare earth and isotopic geochemistry suggests that the majority of the CIC magmas were derived from a crustal source. The only rocks that can definitely be attributed to a mantle source do not show any of the commonly accepted subduction signature: they are tholeiitic rather than calc-alkaline and do not show the decoupling of HFSE and LILE. The clear association of the CIC with major crustal extension (e.g. Holder & Holder, 1988) that followed the Cretaceous Laramide orogeny is also at odds with a subduction model. At the end of the Laramide orogeny, the thickness of the crust is estimated to have been 75 km, compared with its present-day thickness of 35 km (Varseck & Cook, 1994). The data suggest that the formation of the CIC was entirely a product of regional extension, the majority of the magmas being crustally derived and formed by a combination of decompression melting of that crust and the intrusion of mantle material, some of which was erupted as the basic end member of the KMF. The subduction signature observed in many of the CIC rocks was entirely inherited from previous, probably Proterozoic, subduction events.
Largely on the basis of isotopic data, at least three distinct crustal sources can be identified. Examination of deep seismic data generated by the LITHOPROBE (Canada) and COCORP (US) programs (Varseck & Cook, 1994), in combination with crustal-scale gravity models (Sobkzyk, 1994), reveals at least three distinct crustal zones (Fig. 16). It has been suggested that these zones correlate with the three major magmatic episodes observed in the Colville Igneous Complex (Morris & Watkinson, 1997). On the basis of this suggestion, and in combination with the data generated in this study, we can characterize some of the middle and lower crust imaged in these sections. The middle-crustal zone appears to be ancient granitic subduction-related magmatic arc material of palaeo-Proterozoic age, which we propose as the source material for the SVF. The lower-crustal zone is also of magmatic arc origin, but neo- to meso-Proterozoic in age. This lower-crustal zone may consist of two distinct types of crust, as reflected by the chemistry of the Klondike Mountain Formation, one of which correlates with the high-density wedge of Sobkzyk (1994) (Fig. 16).
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONPETROLOGY AND MINERALOGYGEOCHEMISTRYPETROGENESIS: THE NATURE AND...CONCLUSIONSAPPENDIX: GEOCHEMICAL TECHNIQUESREFERENCES |
|---|
The majority of magmas that formed the Colville Igneous Complex are mixtures of partial melts of older continental crustal rocks and are not related to contemporaneous subduction. The magmas of the Sanpoil Volcanic Formation were derived directly from the partial melting of a heterogeneous Proterozoic continental crust. Magmas that formed the Klondike Mountain Formation represent a mixing array between basaltic magmas generated in the mantle and at least two more silicic magmas formed by partial melting of a heterogeneous lower (granulite facies) meso-Proterozoic continental crust. Rather than being the product of a volcanic arc and the subduction process, the formation of the Colville Igneous Complex is thought to be related to post-Laramide orogenic collapse. Orogenic collapse brought hot deep crust to shallower levels and the consequent partial melting produced the majority of the Colville Igneous Complex. The last stages of this extension allowed contributions from the sub-continental mantle. The lack of HFSE depletion coupled with LILE enrichment in the mantle-derived basaltic magmas suggest that no subduction process occurred during the formation of the Colville Igneous Complex.
|
APPENDIX: GEOCHEMICAL TECHNIQUES |
|---|
|
TOPABSTRACTINTRODUCTIONPETROLOGY AND MINERALOGYGEOCHEMISTRYPETROGENESIS: THE NATURE AND...CONCLUSIONSAPPENDIX: GEOCHEMICAL TECHNIQUESREFERENCES |
|---|
All major, trace, rare earth element and stable isotope analyses were carried out at the Geoanalytical Laboratory at Washington State University. Major elements (Si, Ti, Al, Fe, Mn, Mg, Ca, Na, K, and P reported as oxide wt %; total Fe reported as FeO, designated FeOT) and selected trace and rare earth elements (Ni, Cr, Sc, V, Ba, Rb, Sr, Zr, Y, Nb, Ga, Cu, Zn, Pb, La, Ce, and Th reported as ppm) were analysed by XRF using a Rigaku 3370 fully automated spectrometer following procedures published by Hooper et al. (1993)
and Johnson et al. (1997). The precision of these data, as well as comparisons with other laboratories and techniques, has been discussed by Duke (1993) and Johnson et al. (1997). All 14 naturally occurring REE (La–Lu) and selected trace elements (Ba, Rb, Y, Nb, Cs, Hf, Ta, Pb, Th, and U reported as ppm) were analysed by ICP-MS using a Sciex Elan Model 250 instrument, following procedures published by Knaack et al. (1994), who reported also precision and comparisons with other laboratories. Of these, the Ba, Nb, Y, Pb, and Rb analyses duplicate the XRF analyses, to provide a degree of internal monitoring. Values for the Ba, Nb and Pb by ICP-MS are preferred, particularly at lower abundances, whereas the XRF values for Y and Rb are considered more precise (Hooper et al., 1993; Knaack et al., 1994; Johnson et al., 1997).
Oxygen isotope ratios were determined from whole-rock samples, and are reported using 18O notation relative to VSMOW. Conventional silicate fluorination techniques were used to extract O2 from powdered samples (after Clayton & Mayeda, 1963) using ClF3 as the oxidizing agent (after Borthwick & Harmon, 1982). Samples were converted to CO2 and then analysed on a Finnigan-Mat Delta S gas source mass spectrometer. The raw data was normalized to an NBS-28 value of +9·85 using an in-house standard, MM-1 (Mica Mountain pegmatite quartz; 18O = +12·9). Repeat analyses of MM-1 show a standard 2
deviation of less than ±0·2.
Sr and Nd isotopes were analysed at University of Alberta, Canada, sample preparation following standard procedures modified from Richards et al. (1976). Samples were dissolved in concentrated HF at 120°C for at least 48 h. This was followed by concentrated HNO3 treatment and conversion to chloride salts using concentrated HCl. Sr and REE were separated from the whole sample by passing through calibrated ion exchange columns. Sr separates were passed through a second calibrated ion exchange column before analysis. Nd was separated from the other REE by passing through di-2-ethylhexyl-coated Teflon powder columns, then analysed. Sr samples were analysed on a VG Micromass 30 single-collector mass spectrometer with results being standardized to NBS-987. Nd was analysed on a VG Micromass 354 fully automated multiple-collector mass spectrometer and results were standardized to an internal standard with 143Nd/144Nd = 0·511065, equivalent to a La Jolla standard with a value of 0·511858; a small correction was therefore applied to each sample to bring them to an equivalent of the accepted value for La Jolla, 0·511858, then reported. For internal normalization for mass fractionation a value of 146Nd/144Nd = 0·7219 was used.
|
ACKNOWLEDGEMENTS |
|---|
The basic field and chemical studies by Grace McCarley Holder, Wade Holder, Laureen Wagoner and Charles Knaack at Washington State University laid the groundwork for this study. Many discussions with Steve Box (US Geological Survey), Rick Conrey, Laureen Wagoner, Charles Knaack, Dave Gaylord, Chris Lowe and Kathy Long (all of Washington State University) were invaluable in developing these ideas. Diane Johnson, Scott Cornelius and Charles Knaack (all of Washington State University Geoanalytical Laboratory) are acknowledged for their assistance with analysis. Special thanks are due to Robert Creaser and hi