- © 2000 Cambridge University Press
Isotopic and geochemical data indicate that intrusions in the eastern Creignish Hills of central Cape Breton Island, Canada represent the roots of arcs active at ~ 540–585 Ma and ~ 440 Ma. Times of intrusion are closely dated by (1) a nearly concordant U–Pb zircon age of 553 ± 2 Ma in diorites of the Creignish Hills pluton; (2) a lower intercept U–Pb zircon age of 540 ± 3 Ma that is within analytical error of 40Ar/39Ar hornblende plateau isotope-correlation ages of 545 and 550 ± 7 Ma in the River Denys diorite; and (3) an upper intercept U–Pb zircon age of 586 ± 2 Ma in the Melford granitic stock. On the other hand, ~ 441–455 Ma 40Ar/39Ar muscovite plateau ages in the host rock adjacent to the Skye Mountain granite provide the best estimate of the time of intrusion, and are consistent with the presence of granitic dykes cutting the Skye Mountain gabbro–diorite previously dated at 438 ± 2 Ma. All the intrusions are calc-alkaline; the Skye Mountain granite is peraluminous. Trace element abundances and Nb and Ti depletions of the intrusive rocks are characteristic of subduction-related rocks. The ~ 540–585 Ma intrusions form part of an extensive belt running across central Cape Breton Island, and represent the youngest Neoproterozoic arc magmas in this part of Avalonia. Nearby, they are overlain by Middle Cambrian units containing rift-related volcanic rocks, which bracket the transition from convergence to extension between ~ 540 and 505/520 Ma. This transition varies along the Avalon arc: 590 Ma in southern New England, 560–538 Ma in southern New Brunswick, and 570 Ma in eastern Newfoundland. The bi-directional diachronism in this transition is attributed to northwestward subduction of two mid-ocean ridges bordering an oceanic plate, and the migration of two ridge–trench–transform triple points. Following complete subduction of the ridges, remnant mantle upwelling along the subducted ridges produced uplift, gravitational collapse and the high-temperature/low-pressure metamorphism in the arc in both southern New Brunswick and central Cape Breton Island. The ~ 440 Ma arc magmatism in the Creignish Hills extends through the Cape Breton Highlands and into southern Newfoundland, and has recently been attributed to northwesterly subduction along the northern margin of the Rheic Ocean.
Neoproterozoic subduction-related magmatism spans 800–580 Ma in the Pan-African orogens of northwest Africa (Caby & Andreopoulis-Renaud, 1987; Ducrot & Lancelot, 1978) and 700–550 Ma in the Brasiliano orogens of South America (Teixiera et al. 1989). Such magmatism in the Avalon Composite Terrane of Atlantic Canada (Fig. 1a⇓) has been correlated both with northwest Africa (OBrien, Wardle & King, 1983) and northern South America (Dostal et al. 1996). This igneous activity has been related to the amalgamation of Gondwana, and precedes a major, latest Precambrian–Early Cambrian plate reorganization (Keppie et al. 1996). In this context, the ~ 540–585 Ma calc-alkaline igneous rocks in the eastern Creignish Hills of central Cape Breton Island represent the terminal stage of subduction-related magmatism in the Avalon Composite Terrane of Nova Scotia. Accurate dating and geochemistry of this magmatic activity is crucial to the formulation of tectonic models for the termination of subduction. Similarly, the extent of Silurian magmatic arc rocks in Cape Breton Island, and their transition to rift tholeiites in mainland Nova Scotia, are critical to the understanding of the Early Palaeozoic evolution of the Appalachians.
There have been two schools of thought regarding the geology of Cape Breton Island. First, it has been proposed that Cape Breton Island represents an oblique section through the Neoproterozoic arc, overstepped in southern and central Cape Breton Island by the Avalonian Cambrian sequence, and underlain by ~ 1 Ga basement exposed in northwestern Cape Breton Island (Keppie, 1985). Second, it has been suggested that Cape Breton Island is made up of distinct terranes. South to north, these are Mira, Bras d’Or, Aspy and Blair River (the latter is inferred to represent the Laurentian margin), which were not amalgamated until the Devonian (Barr & Raeside, 1986). Over the past decade, these two views have converged. Thus, White et al. (1994) have concluded that the Avalonian overstep sequence of Cambrian age does indeed unconformably overlie their Bras d’Or and Mira terranes. However, they still believe that the two terranes were separate, because Middle Cambrian volcanic rocks in the Bras d’Or terrane are nearly absent in the Mira terrane.
On the other hand, Hutchinson (1952) has shown that the Middle Cambrian rocks change rapidly from mainly volcanic rocks to mainly sedimentary rocks, and so facies changes are a reasonable explanation for the differences between Mira and Bras d’Or Cambrian stratigraphy. Lynch, Tremblay & Rose (1993) have demonstrated that the boundary between the Bras d’Or and Aspy terranes is an unconformity, and so does not qualify as a terrane boundary. Dostal et al. (1996), Keppie & Dostal (1998) and Murphy et al. (in press) have demonstrated that the 700–550 Ma igneous activity in Cape Breton Island (Mira, Bras d’Or and Aspy) and southern New Brunswick (Caledonia and Brookville terranes) can be explained as supra-subduction zone magmatism above a single northwest-dipping (present co-ordinates) Benioff zone. Ayuso, Barr & Longstaffe (1996) have shown that the Pb isotope signatures in Neoproterozoic igneous rocks in Cape Breton Island represent mixing of two end members: Blair River and Mira. Murphy et al. (1998) have deduced that the model Nd ages of igneous rocks in the Avalon composite terrane indicate melting of a ~ 1 Ga basement.
The MacIntosh Brook Fault that separates the Bras d’Or and Mira terranes is a brittle structure with minor displacement, not a major shear zone with large lateral movement as required by the model proposed by van Staal, Sullivan & Whalen (1996). These authors place the Mira and Bras d’Or terranes several thousand kilometres apart along strike in the Neoproterozoic, juxtaposing them in the Siluro-Devonian by large dextral displacements. On the other hand, the same boundary in southern New Brunswick (Bellisle–Kennebacasis fault zone) has undergone both dextral and sinistral Siluro-Devonian ductile deformation followed by Carboniferous dextral and vertical brittle movements (Schreckengost & Nance, 1996). Thus, in Cape Breton Island at least, the Avalon composite terrane appears to represent a relatively intact Neoproterozoic forearc–arc–backarc resting on a ~ 1 Ga basement and overlain by a Lower Palaeozoic overstep sequence.
2. Geological setting
The Creignish Hills represent one of several basement blocks that project through the Carboniferous cover of central Cape Breton Island (Fig. 1b⇑). Igneous rocks dated at ~ 540–580 Ma are present in all of these basement blocks, and are also represented in the coastal block of southern Cape Breton Island (Fig. 1b⇑). Contemporaneous magmatism elsewhere in Nova Scotia is limited to isolated plutons in the Antigonish and Cobequid highlands. In the Creignish Hills, plutons dated to ~ 540–580 Ma intrude metasedimentary and metavolcanic rocks deposited between ~ 977 Ma and ~ 637 Ma, which were multiply deformed and metamorphosed to greenschist–amphibolite facies at low pressures and high temperatures at ~ 553–550 Ma, and intruded by syntectonic granitic sheets containing zircon and monazite dated at 551 ± 1 Ma (Keppie, Davis & Krogh, 1998). Gravitational collapse of the arc is inferred to have placed low- over high-grade rocks along low-angle decollements (Keppie, Davis & Krogh, 1998).
The synchroneity of intrusion and tectonism is reflected in the igneous rocks that show fabrics ranging from strongly foliated to massive, and reflect the syn-tectonic to post-tectonic nature of the intrusion relative to local deformation. Thus, the contact of the Melford stock (Fig. 1c⇑) cuts nearly at right angles across the multiply deformed metamorphic layers of the Bras d’Or Gneiss, and yet is internally weakly foliated parallel to the composite foliation in the host rocks. This suggests that intrusion was late syn-tectonic. The River Denys pluton is generally massive, but along its northern margin it exhibits ductile–brittle deformation, indicating that it is also late syn-tectonic. On the other hand, the Creignish Hills and Skye Mountain granitic plutons are both massive, and their contacts cut across the composite foliation in the country rocks, indicating that they are post-tectonic. Furthermore, the Skye Mountain granite cuts across the sheared contact between the high- and low-grade rocks, and andalusite in the contact aureole overgrows all the fabrics in the host rocks.
The Creignish Hills pluton has yielded a Rb–Sr whole-rock isochron age of 446 ± 13 Ma (White, Barr & Campbell, 1990). However, diorite in the eastern part of the Creignish Hills pluton yielded two 40Ar/39Ar hornblende plateau ages of 544 ± 5 Ma and 535 ± 3 Ma, interpreted closely to post-date intrusion (Keppie, Dallmeyer & Murphy, 1990). This led White, Barr & Campbell (1990) to exclude the diorite analyses from the Rb–Sr whole-rock isochron, which then gave an age of 441 ± 8 Ma. Although White, Barr & Campbell (1990) state that the intermediate felsic rocks could have been fractionated from the mafic rocks, they prefer a correlation with the Kellys Mountain and Cape Smoky granites dated at ~ 495 Ma (Dunning et al. 1990a) based upon geochemical data. To allow this conclusion, White, Barr & Campbell (1990) excluded two more analyses of aplite from the Rb–Sr whole-rock isochron, which then produced an age of 473 ± 102 Ma.
40Ar/39Ar analyses of muscovite in the gneiss just east of the Skye Mountain granite yielded a plateau age of 449 ± 7 Ma, whereas muscovite from a pegmatite produced a discordant spectrum that decreased from ~ 485 to 450 Ma (Dallmeyer & Keppie, 1993). The Bras d’Or Gneiss was subsequently intruded by the arc-related Skye Mountain gabbro–diorite, which has yielded a U–Pb concordant zircon age of 438 ± 2 Ma (Fig. 1c⇑; Keppie et al. 1998). The contact between the Skye Mountain gabbro and the Skye Mountain granite is not exposed; however, granitic dykes extending several metres into the gabbro perpendicular to the inferred contact suggest that the granite is younger. The possibility that the granitic dykes represent back veining is considered unlikely, because the granitic dykes form discrete intrusions rather than a vein network, and the gabbro is a small, high-level pluton that does not appear to have been a conduit for mafic magma (Keppie et al. 1998). Carboniferous rocks unconformably overlie the eastern Creignish Hills around their northern margin, and are in fault contact along most of its southern margin (Fig. 1c⇑).
Igneous lithologies vary from diorite through granodiorite, and granite to pegmatitic granite. The River Denys pluton is predominantly massive diorite, the eastern Creignish Hills pluton varies from diorite to granodiorite and the Skye Mountain granite consists of granite and pegmatitic granite. The diorite consists mainly of plagioclase, amphibole and biotite variably altered to actinolite and chlorite with minor amounts of microcline and quartz, and accessory apatite, titanite and zircon. The granodiorite is composed of plagioclase, alkali feldspar, quartz, amphibole and biotite more or less altered to chlorite, epidote and calcite with accessory apatite, zircon and titanite. The granite is made up of plagioclase, alkali feldspar and quartz with minor biotite and muscovite and accessory apatite, zircon and opaque minerals. Pegmatitic phases contain alkali feldspar and quartz with minor biotite.
3. Analytical methods
A total of 40 samples were collected from several small intrusions for U–Pb, 40Ar/39Ar isotopic, major and trace element analyses (Fig. 1c⇑). Samples weighing 20 kg were collected for U–Pb zircon analyses from the eastern part of the Creignish Hills pluton, Melford stock, River Denys pluton and Skye Mountain granite. Zircon was processed by a method similar to that of Krogh (1973), but using a 0.3 ml resin volume and a mixed 205Pb –233U–235U spike. Magnetic and non-magnetic fractions were separated using a Franz separator, and some were abraded. Isotope ratios were measured on a VG sector mass spectrometer at the Université du Quebec à Montreal. The blank for the entire analytical procedure ranged from 5 to 20 pg. Dimensions of the error ellipses in the concordia diagrams and errors in ages are at the 95 % confidence level, and include measurement error, confidence in the fractionation factors, error in the U–Pb ratio of the spike and the effect of the common lead correction. Table 1⇓ lists the U–Pb data.
Concentrates of hornblende and muscovite were analysed using incremental-release 40Ar/39Ar analyses (Tables 2⇓, 3⇓). The techniques used generally followed those described in detail by Dallmeyer & Takasu (1992). Optically pure (> 99 %) mineral concentrates were wrapped in aluminium foil packets, encapsulated in sealed quartz vials, and irradiated in the United States Geological Survey TRIGA reactor in Denver. Variations in the flux of neutrons along the length of the irradiation assembly were monitored with several mineral standards, including Mmhb-1 hornblende (Samson & Alexander, 1987). The samples were incrementally heated until fusion in a double-vacuum, resistance-heated furnace following methods described by Dallmeyer & GilIbarguchi (1990). Measured isotopic ratios were corrected for total system blanks, the effects of mass discrimination and interfering isotopes produced during irradiation. 40Ar/39Ar ages were calculated from corrected isotope ratios using the decay constants and isotopic abundance ratios listed by Steiger & Jäger (1977).
Intralaboratory uncertainties have been calculated by statistical propagation of uncertainties with measurements of each isotopic ratio (at two standard deviations of the mean) through the age equation. Interlaboratory uncertainties are c. 1.25–1.5 % of the quoted age. Total-gas ages have been computed for each sample by appropriate weighting of the age and the percentage of 39Ar released within each temperature increment. A ‘plateau’ is defined when the ages recorded by two or more contiguous gas fractions adding up to > 50 % of the total 39Ar evolved are mutually similar within ± 1 % intralaboratory uncertainty, each fraction having similar apparent K/Ca ratios and representing > 4 % of the total 39Ar evolved. Plateau portions of the analyses have been plotted on 36Ar/40Ar isotopic correlation diagrams. Regression techniques followed the methods of York (1969). A mean square of the weighted deviates (MSWD) has been used to evaluate isotopic correlations. Analyses of the Mmhb-1 monitor indicate that apparent Ca ratios may be calculated through the relationship 0.518(100.005) × (39Ar/37Ar)corrected.
Geochemical samples were collected from the main intrusions and also from isolated small stocks indicated by a sample location on the map (Fig. 1c⇑). They were analysed by X-ray fluorescence for major and some trace elements (Rb, Sr, Ba, Zr, Nb, Y, Cr, Ni) (Table 4⇓). Precision and accuracy are discussed in Dostal, Dupuy & Caby (1994) and Dostal et al. (1994). In general, precision is better than ± 5 % for major elements and 2–10 % for trace elements.
4.a. U–Pb zircon data
Two abraded fractions of clear, euhedral zircon grains from diorite at the eastern edge of the Creignish Hills pluton yielded two nearly concordant U–Pb analyses that plot on a chord through zero, with an upper intercept at 553 ± 2 Ma (Fig. 2⇓, Table 1⇑). Two fractions of clear euhedral zircon grains from the Melford granitic stock yielded discordant data that plot on a chord through zero, with an upper intercept of 586 ± 2 Ma (Fig. 2b⇓). The non-magnetic fraction was abraded, and is nearly concordant with a 207Pb/206Pb age of ~ 587 Ma (Table 1⇑). Two zircon fractions (a magnetic fraction with clear and euhedral grains, and a non-magnetic fraction with zoned euhedral grains) from the River Denys pluton yielded discordant data, and a chord constructed through these two points gives lower and upper intercepts at 540 ± 3 Ma and 2095 ± 13 Ma (Fig. 2c⇓). The magnetic fraction plots close to concordia, and yielded a 207Pb/206Pb age of 612 Ma (Table 1⇑). Zircon is relatively rare in the Skye Mountain granite; however, two small fractions of euhedral–anhedral, zoned zircon grains were separated, and gave highly discordant data that plot on a chord through zero with an upper intercept at ~ 737 Ma (Fig. 2d⇓).
4.b. 40Ar/39Ar data
Multigrain hornblende and muscovite concentrates were prepared from samples collected within various geological units exposed in the Creignish Hills (Fig. 1⇑). The 40Ar/39Ar analytical data are provided in Tables 2⇑ and 3⇑, and are displayed as apparent age spectra in Figures 3⇓, 4⇓ and 5⇓.
Hornblende separated from two samples of diorite at locations 3 and 4 in the River Denys pluton displays variably discordant apparent age spectra (Fig. 3⇑). The relatively small volume low-temperature gas fractions record considerable variation in apparent ages. These are matched by fluctuations in apparent K/Ca ratios, which suggest that experimental evolution of argon occurred from compositionally distinct, relatively non-retentive phases. These could have been represented by (1) very minor, optically undetectable mineralogical contaminants in the hornblende concentrates; (2) petrographically unresolvable exsolution or compositional zonation within constituent hornblende grains; (3) minor chloritic replacement of hornblende; and/or (4) intracrystalline inclusions. Most intermediate- and high-temperature gas fractions display little intra-sample variation in apparent K/Ca ratios, suggesting that experimental evolution of gas occurred from compositionally uniform sites. The intermediate- and high-temperature gas fractions experimentally evolved from the two hornblende concentrates record similar intrasample apparent 40Ar/39Ar ages, which define plateau dates of 551 ± 0.6 Ma (sample 3) and 550.2 ± 0.6 Ma (sample 4). 36Ar/40Ar v. 39Ar/40Ar isotope correlations of the plateau data are well defined (MSWD < 2.0), and define inverse ordinate intercepts (40Ar/36Ar ratios) of 406 ± 10 (sample 3) and 293.4 ± 12 (sample 4). These are generally similar to that of present-day atmosphere, and suggest no significant intracrystalline contamination with extraneous (‘excess’) argon components. Using the inverse abscissa intercepts (40Ar/39Ar ratios) in the age equation yields plateau isotope-correlation ages of 545.3 ± 0.9 Ma (sample 3) and 549.8 ± 0.5 Ma (sample 4) (Fig. 3⇑). Because calculation of isotope-correlation ages does not require assumption of a present-day 40Ar/36Ar ratio, they are considered more significant than those directly calculated from the analytical data. The 545 and 550 Ma isotope-correlation ages recorded by the two hornblende concentrates from the River Denys pluton are considered geologically significant, and are interpreted to date the last cooling through temperatures required for intracrystalline retention of argon in constituent grains.
Muscovite concentrates were prepared from four samples collected in the Melford stock and Skye Mountain granite and their contact aureoles (Figs 4⇑, 5⇑). The muscovite concentrates are characterized by very large apparent K/Ca ratios, which are marked by considerable analytical uncertainties. The apparent K/Ca ratios display no significant and/or systematic intrasample variations, and therefore are not presented with the apparent age spectra.
The four muscovite concentrates display variable intrasample age variations. Those from samples 5 and 6 (Fig. 4⇑) are characterized by only limited age variations, and record well-defined plateau ages of 455.4 ± 0.2 Ma (sample 5: 690–940 °C; 83.82 % of the total gas evolved) and 440.6 ± 0.4 Ma (sample 6: 655 °C-fusion; 88.79 % of the gas evolved). These are considered geologically significant, and are interpreted to date the last cooling through temperatures required for intracrystalline argon retention.
The muscovite concentrates prepared from samples 1 and 2 (Fig. 5⇑) are marked by more internally discordant 40Ar/39Ar spectra, in which apparent ages systematically decrease throughout low-temperature portions of the analysis to define intermediate-temperature plateaux that are followed by an increase in apparent age in the highest-temperature increments. The 745–900 °C increments evolved from sample 1 record similar apparent ages that define a plateau of 482.1 ± 0.3 Ma. These comprise 54.74 % of the total analysis. A 471.5 ± 0.3 Ma plateau is defined by the 670–885 °C increments evolved from sample 2 (representing 73.70 % of the total analysis). The geological significance of these plateau ages is uncertain in view of the internal complexity of the two analyses.
4.c. Interpretation of geochronological data
The nearly concordant zircon age of 553 ± 2 Ma from diorite at the eastern margin of the Creignish Hills plutonic complex is inferred to date the time of intrusion (Fig. 2a⇑). Hornblende from the same location produced a plateau age of 544 ± 5 Ma (Keppie, Dallmeyer & Murphy, 1990), and indicates that the pluton cooled relatively quickly through closure temperatures for argon in hornblende estimated to be ~ 500 °C (Harrison, 1981). Similarly, the U–Pb zircon lower intercept age and 40Ar/39Ar on hornblende isotope-correlation ages from the River Denys dioritic pluton are the same within analytical errors (Figs 2c⇑, 3⇑), and overlap in the range 540–550 Ma, which is interpreted to post-date closely the time of intrusion. The similarity of the zircon and hornblende ages indicates that this pluton also cooled quickly through ~ 500 °C. The similarity of the Creignish Hills and River Denys diorites suggests that they are part of the same magmatic event (see below).
Interpretation of the isotopic data from within and adjacent to the Melford granitic stock is more complex. The similarity of the U–Pb upper intercept age of 586 ± 2 Ma and the 207Pb/206Pb age of ~ 587 Ma age of the nearly concordant, abraded, clear, euhedral fraction suggests that this represents the time of intrusion. This is consistent with the observation that the stock was foliated during a deformational event dated at 550 Ma (Keppie, Davis & Krogh, 1998). On the other hand, the muscovite plateau ages of ~ 470–484 Ma suggest either slow cooling between 500 and 400 °C (i.e. the closure temperatures for Ar in hornblende and muscovite; Harrison, 1981; G. A. Robins, unpub. M.Sc. thesis, Brown Univ. 1972) or reheating. The latter is consistent with the saddle-shaped spectra, which suggest the presence of excess argon. Such reheating may be associated with intrusion of the 438 ± 2 Ma Skye Mountain gabbro–diorite (Keppie et al. 1998).
Interpretation of the isotopic data from the Skye Mountain granite is more problematic. Inheritance and recent lead loss in the U–Pb zircon data are indicated by the zoned nature of the zircons and the zero lower intercept, respectively. Thus, the ~ 737 Ma upper intercept age represents a maximum for the time of intrusion. On the other hand, the ~ 441–455 Ma muscovite plateau ages from north and west of the Skye Mountain granite (Fig. 5⇑) are similar to the 449 ± 7 Ma muscovite plateau age from just east of the pluton (Dallmeyer & Keppie, 1993). The adjacent Skye Mountain gabbro–diorite has been dated at 438 ± 2 Ma by U–Pb on zircon (Keppie et al. 1998), and it is possible that the Skye Mountain granite was penecontemporaneous, a conclusion consistent with the presence of granitic dykes in the gabbro near their mutual contact. This suggests that the muscovite ages in the adjacent host rocks represent relatively complete thermal re-equilibration of argon at ~ 438 Ma.
The intrusive rocks of the Creignish Hills were affected to varying degrees by secondary processes including low-grade metamorphism, which might have modified the chemical composition of these rocks. However, the concentrations of most major and trace elements, including alkali and alkali-earth elements, are thought to reflect the primary magmatic distribution. When these elements are plotted against SiO2, which is considered to be a good indicator of the fractionation and is apparently immobile under most metamorphic conditions (e.g. Winchester & Floyd, 1977), they display distinct correlations (Fig. 6⇓). Remobilization during metamorphism is unlikely to produce such a consistent result. The consistency of these trends, and their similarities to those of modern igneous rocks, suggest that the distribution of these elements was not significantly modified.
The intrusive rocks of the eastern Creignish Hills may be subdivided into two groups. First, the ~ 540–585 Ma Creignish Hills diorite and associated mafic–intermediate intrusive bodies, and second, the ~ 438 Ma Skye Mountain granite. The Neoproterozoic group has SiO2 in the range 50–63 % (LOI-free) with the majority falling between 56–62 % (Fig. 6⇑, Table 4⇑). The Skye Mountain granite has SiO2 contents from 65 to 77 % (Fig. 6⇑, Table 4⇑). Both groups are subalkaline and display calc-alkaline fractionation trends (Figs 6c, 6d⇑, 7⇓). Chemical evolution, as revealed by the relatively constant Na2O/K2O ratio while CaO decreases (Fig. 6f⇑), resembles the calc-alkaline trend of Nockolds & Allen (1953), but departs significantly from the trond-hjemitic trend of Barker & Arth (1976) and the low-calcium granite trend of Breaks & Moore (1992). The Neoproterozoic rocks are relatively diverse, although they have compositions that correspond mainly to that of medium-K andesites (Fig. 6e⇑).
The rocks of the Skye Mountain granite are not only younger, but are also geochemically distinct from the Neoproterozoic rocks. In contrast to most calc-alkaline rocks from volcanic arcs and continental margins, the Skye Mountain granites are peraluminous, with CIPW normative compositions containing corundum and with molecular Al2O3/(CaO + Na2O + K2O) ratios > 1. Due to alteration processes, we cannot be certain that CaO, Na2O and K2O concentrations represent true magmatic values. However, as noted above, we believe that in overall terms the rocks have not been significantly modified and that the Ca, Na and K values are close to their original levels.
All the intrusive rocks of the eastern Creignish Hills reported here show trace element abundances characteristic of volcanic arc granitoids (Table 4⇑). For example, the felsic rocks all fall within the volcanic arc granitoid field in the Y + Nb v. Rb, and the Y v. Nb tectonic discriminant diagrams (Fig. 8⇓; Pearce, Harris & Tindle, 1984). The mantle-normalized trace element patterns of the analysed rocks are characterized by Nb and small Ti depletion (Fig. 9⇓), a feature typical of subduction-related rocks. These data are similar to those reported by White, Barr & Campbell (1990) from the main Creignish Hills pluton. These authors also presented rare earth element (REE) analyses that show light-REE enrichment and unfractionated heavy-REE. In particular, the Neoproterozoic rocks are comparable to their tonalite–diorite unit, whereas the Skye Mountain granites are compositionally similar to the granitic rocks of the Creignish Hills pluton.
The geochemistry of both the Neoproterozoic and Silurian intrusions in the eastern Creignish Hills rocks are typical of arc-related rocks. However, the limited amount of data does not allow discrimination between the various models invoked for the origin of such rocks, including partial melting of subducted oceanic crust (Defant & Drummond, 1990) and mantle wedge (Gill, 1981). Similar Neoproterozoic intrusions elsewhere in central Cape Breton Island originated above a northwest-dipping (present co-ordinates) subduction zone (Dostal et al. 1996).
The peraluminous nature of the Skye Mountain granitic rocks distinguishes them from the Skye Mountain gabbro–diorite. Two processes can account for the origin of these granitic rocks. Hornblende fractionation can generate trends towards peraluminous composition (Cawthorn & Brown, 1976; Cawthorn & O’Hara, 1976; Cawthorn, Strong & Brown, 1976). However, the absence of hornblende in the Skye Mountain granite indicates that this mechanism has not been important in its genesis. Alternatively, models have been proposed for the generation of peraluminous melts by partial melting of pelitic rocks (Grant, 1985; Vielzeuf & Holloway, 1988), and these appear to be applicable to the Skye Mountain granite. It is suggested that intrusion of mafic and intermediate magma of the Skye Mountain gabbro–diorite (Keppie et al. 1998) into the crust caused crustal melting. In particular, the heat for the melting may have been supplied by mafic mantle-derived magmas. If these rocks are subduction related, the tectonic setting is likely to have been an active continental margin rather than an oceanic arc.
6. Tectonic implications
The 540–585 Ma supra-subduction zone magmatic rocks in the eastern Creignish Hills form part of an extensive belt running across central Cape Breton Island (Fig. 1b⇑). These magmatic arc rocks are overlain by Middle Cambrian–Lower Ordovician rocks that include bimodal, within-plate, rift-related volcanic rocks dated at 505 ± 3 Ma (White et al. 1994; Keppie et al. 1997). This brackets the change from subduction to rifting between 545 and 505 Ma. This may be more tightly constrained if the interbedded Middle Cambrian rocks are part of the rift sequence. Okulitch (1995) places the base of the Middle Cambrian at 520 ± 10 Ma.
Northeastwards along strike in the volcanic arc to the Avalon Composite Terrane of Newfoundland, the transition from subduction-related to extensional igneous activity is dated at ~ 570 Ma (O’Brien et al. 1996). Southwestwards in the Avalon Composite Terrane of southern New Brunswick, 630–600 Ma volcanic arc rocks are succeeded by voluminous extensional volcanic rocks dated at 560–550 Ma overlain by Cambrian rocks in the Caledonia assemblage (Barr & White, 1996). In the Brookville assemblage of southern New Brunswick, correlated with central Cape Breton Island, the youngest calc-alkaline plutons are 538 ± 1 Ma (White et al. 1990; White & Barr, 1991). Fault-bounded units of the Cambro-Ordovician Saint John Group crop out between the Caledonia and Kennebecasis faults that border the Brookville assemblage (Nance & Dallmeyer, 1994). In the Avalon Composite Terrane of southern New England, the transition from arc to extensional magmatism probably occurred at ~ 590 Ma (Thompson et al. 1996; Mancuso, Gates & Puffer, 1996). This indicates that the change from subduction to extension occurred at different times along the Avalon arc (from southwest to northeast): 590–538–540–570 Ma. However, although the transition is diachronous, it is bi-directional (i.e. it becomes progressively younger from New England to New Brunswick, and from Newfoundland to Cape Breton Island). The end of subduction has generally been related to plate reorganization as Iapetus began to open; however, this does not explain the apparent diachronism. Dostal et al. (1996) and Murphy et al. (in press) have documented that the 580–550 Ma subduction was toward the (present-day) northwest, beneath both Cape Breton Island and southern New Brunswick. Polarity has not been determined in New England or Newfoundland, but it is assumed also to have been to the northwest in the plate tectonic model present below.
The termination of subduction generally appears to have taken place without a major orogenic event, although polyphase deformation accompanied by high-temperature/low-pressure metamorphism is documented in southern New Brunswick and central Cape Breton Island (Bevier, Barr & White, 1990; Keppie, Davis & Krogh, 1998). A younger limit for the age of this metamorphism in southern New Brunswick is given by a U–Pb 565 ± 6 Ma titanite age (Bevier, Barr & White, 1990) and 40Ar/39Ar hornblende isotope-correlation ages of ~ 540 Ma (Dallmeyer et al. 1990; Nance & Dallmeyer, 1994). In central Cape Breton Island the metamorphism has been dated by concordant U–Pb zircon analyses at ~ 553–550 Ma (Keppie, Davis & Krogh, 1998), who related this tectonothermal event to gravitational collapse of the arc. The nature of the Early Cambrian extension is generally unknown; however in the Antigonish Highlands, the Early Cambrian succession is inferred to have been deposited in a dextral pull-apart basin (Keppie & Murphy, 1988).
It is proposed that the plate tectonic setting of western Mexico and Central America may provide a modern analogue (Protti, Güendal & McNally, 1995). Here, the Cocos Plate is being subducted beneath Mexico and Central America. The northern margin of the Cocos Plate is the East Pacific Rise, which is being subducted beneath Mexico along the Middle America Trench. At the triple point intersection of the ridge with the trench, the plate margin changes from sub-duction along the Acapulco–Middle America Trench to an oblique transform rift along the Gulf of California. As the triple point migrates, the associated volcanism switches from arc to extensional. A similar, but opposite, pattern may be observed where the southern margin of the Cocos Plate (Galapagos– Costa Rica Rift) is being subducted. Subduction of the Cocos Ridge also produced indentation, arching and inversion in the forearc region, and backarc basin inversion (Kolarsky, Mann & Montero, 1995).
This model may be applied to the termination of sub-duction and the switch from calc-alkaline to extensional magmatism in the Avalon Composite Terrane, in which two ridges were subducted beneath Avalonia (Fig. 10⇓). The term ‘Merlin Plate’ is introduced for the Neoproterozoic equivalent of the Cocos Plate. Another consequence of this model occurred when the Merlin Plate was completely subducted. The subducted plate may become stationary beneath Avalonia, and remnant mantle upwelling along the subducted ridge axes could heat up the overlying lithosphere. This mechanism may explain the high-temperature/low-pressure metamorphism that occurred in southern New Brunswick and central Cape Breton Island at the end of all arc-related magmatism. The absence of such metamorphism in the ~ 540–585 Ma forearc regions of southern Cape Breton Island and coastal southern New Brunswick accords with observations where the Cocos Ridge has been sub-ducted beneath Costa Rica (Kolarsky, Mann & Montero, 1995).
Detrital zircon geochronology indicates that Avalonia originated off the Amazon Craton, and geochemistry of the volcanic rocks shows that subduction was consistently towards the present northwest over the period 700–550 Ma (e.g. Keppie & Dostal, 1998; Keppie, Davis & Krogh, 1998). The plate tectonic setting proposed in this paper implies that following subduction of the mid-ocean ridges and cessation of subduction, oceanic lithosphere continued to exist to the present south of Avalonia. As there was ~ 150 million years of northwest-dipping subduction, a large ocean must have existed to the present south of Avalonia in the Neoproterozoic. This scenario suggests that the Amazon Craton may have lain to the present north of Avalonia. The separation of Avalonia from the Amazon Craton, its rotation and its accretion to Laurentia in Cambro-Ordovician times have been modelled by Keppie & Ramos (in press).
The Silurian rocks of the Skye Mountain gabbro–diorite and granite form part of a belt of similar rocks extending northwards along the axis of the Cape Breton Highlands (Sarach Brook and Jumping Brook metamorphic suites, Money Point Group; 439 ± 7 Ma to 428 ± 4 Ma; Currie, Loveridge & Sullivan, 1982; Dunning et al. 1990a; Keppie, Dallmeyer & Krogh, 1992) and into southern Newfoundland (La Poile Group; 428 ± 6 Ma to 420 + 8/− 2 Ma; Dunning et al. 1990b). The 441 ± 8 Ma Rb–Sr whole-rock isochron obtained from the calc-alkaline granitoid rocks in the western Creignish Hills (White, Barr & Campbell, 1990) suggests that the magmatic arc extends to the western side of Cape Breton Island. Synchronous volcanism in the Antigonish Highlands is tholeiitic and rift related (Keppie et al. 1997). The Silurian magmatic arc has previously been attributed to the terminal closure of remnant basins within Iapetus by oblique southerly subduction beneath Avalonia (e.g. Keppie et al. 1998). However, Murphy, van Staal & Keppie (1999) have recently proposed that this magmatic arc is related to the closure of the Rheic Ocean, with northerly sub-duction beneath Avalonia. These authors relate the transition from calc-alkaline to tholeiitic volcanism to a change in the dip of the subduction zone.
Funds for this project were provided by the Canada–Nova Scotia Mineral Development Agreement, CONACYT Project 0255P-T9506 (El complejo Oaxaqueno y el bloque Chortis en las reconstruciones paleogeograficas de Laurencia y Gondwana anteriores a Pangea), the Instituto de Geologia at the Universidad Nacional Autonoma de Mexico (Project number IN101095), and the Natural Sciences and Engineering Research Council (NSERC). We are grateful to Dr R. D. Nance and an anonymous reviewer for their constructive comments on an early version of the paper. We would like to thank John Lord, Jose Luis Arce Saldaña and Gabriel Valdez Moreno for drafting the figures.
- Received January 4, 1999.
- Accepted November 2, 1999.