Fifteen U–Pb (zircon) radiometric age determinations have been made on igneous rocks of Middle Devonian to Early Carboniferous age from the southern margin of the Magdalen basin in Cape Breton Island and northern mainland Nova Scotia. Volcanic rocks interbed with early rift-basin sedimentary rocks with some palynological biostratigraphy; dated intrusive rocks cut these sedimentary units. Our biostratigraphically constrained ages are in close agreement with the current Devonian time scale. Combined with previously published data, the age determinations show that igneous activity occurred in four pulses: Middle Devonian (390–385 Ma), early Late Devonian (375–370 Ma), latest Devonian to early Tournaisian (365–354 Ma), and late Tournaisian to early Visean (ca. 339 Ma). Middle Devonian (385–389 Ma) volcanic rocks are confined to the Guysborough Group. The Fisset Brook Formation (basalt and minor rhyolite) in the type area and elsewhere in Cape Breton Island and northern mainland Nova Scotia is Late Devonian (ca. 373 Ma), whereas the biostratigraphically distinct succession at Lowland Cove is younger (365 Ma). These Late Devonian rocks are synchronous with plutonism in the Cape Breton Highlands and the Meguma terrane. In the Cobequid Highlands, rhyolite of the Fountain Lake Group was synchronous with Horton Group deposition and with widespread granite plutons (362–358 Ma) emplaced during shear on the Cobequid fault zone. The overlying Diamond Brook Formation basalts are slightly younger (355 Ma). Late Tournaisian – early Visean mafic intrusions and minor basalt occur along the Cobequid – Chedabucto fault zone and in a belt from southern New Brunswick through Prince Edward Island to southwestern Cape Breton Island.
The Maritimes Basin of Atlantic Canada (Fig. 1) contains up to 12 km of mainly sedimentary rocks of Middle Devonian to Early Permian age (Roliff 1962; Williams 1974). It comprises two principal basins: the Magdalen basin in the southern Gulf of Saint Lawrence, Prince Edward Island, and adjacent areas of Nova Scotia and New Brunswick and the Sydney basin east of Cape Breton Island. The oldest basin-fill rocks of the Magdalen basin are mafic and felsic volcanic rocks and interbedded terrestrial sedimentary rocks, to which a variety of local stratigraphic names have been applied (see review by Calder, 1998). In general, these units appear to be older than the widespread terrestrial and lacustrine sedimentary rocks of the late Famennian (latest Devonian) to late Tournaisian (Early Carboniferous) Horton Group (Martel et al. 1993), although some workers have included or proposed to include them in the Horton Group (Kelley and Mackasey 1965; Martel et al. 1993). Sparse biostratigraphic determinations on macrofossils and palynomorphs recovered from these older volcanic-bearing successions have yielded ages ranging from Middle Devonian to Tournaisian.
The present study was undertaken to determine the age of the major occurrences of volcanic and related intrusive rocks in the southern Magdalen Basin, principally in western Cape Breton Island and northern mainland Nova Scotia, using U–Pb geochronology to more precisely date the early basin fill. Where necessary, field work was carried out to relate the igneous rocks to the basin fill. Better understanding of the chronology of early rifting and basin fill should improve our understanding of the tectonic origin of the basin, the relationship of basin extension to igneous activity, and the role of the major dextral strike-slip fault systems in the southern part of the basin.
Regional geology of the Magdalen basin
The Magdalen basin was initiated following the Acadian orogeny (ca. 400 Ma; Hicks et al. 1999), apparently as a series of half-graben depocentres related to regional extension (e.g., Calder 1998), although the mechanism of basin formation is controversial (e.g., Bradley 1982; McCutcheon and Robinson 1987; Lynch and Tremblay 1994; Murphy et al. 1999). The mid- to Late Devonian part of the basin fill is exposed only at scattered locations in Cape Breton Island and northern mainland Nova Scotia and generally consists of interbedded volcanic and sedimentary rocks to which various local names have been assigned, as discussed later in the text. These older units are locally unconformably to conformably overlain by the mainly nonmarine Horton Group (latest Devonian to Tournaisian), also largely deposited in restricted grabens. Later sedimentary units of the basin are regionally more widespread and comprise the shallow marine and nonmarine Windsor Group (mid-Viséan) and the nonmarine coal-bearing Mabou, Cumberland, and Pictou groups (late Viséan to early Permian) (see Calder 1998 for a comprehensive description of these units).
The volcanic and interbedded sedimentary rocks of the basin fill that predate the Horton Group unconformably overlie various Early Devonian and older units. The interbedded terrestrial clastic sedimentary rocks contain rare palynomorphs and other fossils, but previous biostratigraphy and geochronology have not been able to adequately resolve the age and regional correlation of these strata. Although the volcanic and sedimentary sequences are not penetratively deformed or metamorphosed, they have been variably affected by deformation almost synchronous with deposition (e.g., Piper 1994) and (or) subsequent late Paleozoic and early Mesozoic faults and related deformation (e.g., Johnston and Waldron 1998).
Although volcanic rocks occur mainly in the earliest part of the basin fill, they also occur higher in the succession. Gabbro and basalt in the St. Peters area of southern Cape Breton Island (Figs. 1, 2) have been dated at 339 ± 2 Ma (Barr et al. 1994) and appear to lie stratigraphically between the Horton and Windsor groups. Mafic dykes cut the lower part of the Horton Group in western Cape Breton, the Cobequid Highlands, and central Nova Scotia (e.g., Barr and Peterson 1998; Pe-Piper and Piper 1998). Basalt in the Magdalen Islands apparently overlies evaporites of the Windsor Group and hence is probably Viséan (Barr et al. 1985). In the Carboniferous sequences of New Brunswick, scattered occurrences of volcanic rocks and related intrusions range in age from late Tournaisian through Westphalian (Fyffe and Barr 1986).
Petrochemical data from the volcanic rocks are generally consistent with eruption in a continental within-plate extensional setting (e.g., Blanchard et al. 1984; Pe-Piper and Piper 1998). Although most of the rocks have tholeiitic affinity, alkalic sequences are locally present (e.g., Cormier et al. 1995), and later Carboniferous units are generally alkalic (Fyffe and Barr 1986; La Flèche et al. 1998).
Distribution and age of early basin fill in the southern Magdalen basin
Cape Breton Island
The earliest basin fill in the Magdalen basin in Cape Breton Island is recorded in the McAdams Lake Formation (Fig. 2), a sequence of lacustrine and coarser basin-margin sediments deposited in a half-graben (Bell and Goranson 1938; Calder 1998; White and Barr 1998). Plant fragments and spores indicate an age of latest Emsian to early Eifelian (latest Early to Middle Devonian). Although felsic volcanic rocks had been previously reported in the formation (Bell and Goranson 1938), White and Barr (1998) found only fine-grained syenitic intrusions that may have been mistaken for volcanic rocks in the absence of petrographic study.
Late Devonian sedimentary and volcanic sequences in Cape Breton Island have been assigned to the Fisset Brook Formation and probably correlative units (Kelley and Mackasey 1965). They occur in isolated localities from Lowland Cove in the north to the Creignish Hills in the southwest (Fig. 2). Kelley and Mackasey (1965) placed the Fisset Brook Formation in the lowermost part of the Horton Group and defined the top of the formation as the top of the uppermost volcanic unit. However, subsequent workers (e.g., Blanchard et al. 1984) have generally regarded the Fisset Brook Formation to unconformably underlie the Horton Group (Fig. 3). The type section of the formation is near Cheticamp (Kelley and Mackasey 1965), where a 1 km-thick succession of basalt, rhyolite, and clastic sedimentary rocks outcrops in two parallel belts (Kelley and Mackasey 1965; Blanchard et al. 1984; Barr and Peterson 1998). The Late Devonian plant Archaeopteris has been found in sedimentary rocks of the eastern belt (Blanchard et al. 1984; Kasper et al. 1988), and the spore assemblage in rocks from the same area is probably of late Famennian age (McGregor 1996b). Kelley and Mackasey (1965) reported a spore assemblage from the base of the western belt that was at the time interpreted as Tournaisian, which led Blanchard et al. (1984) to suggest that the eastern belt might be older than the western belt. However, re-analysis of the spores from this sample (McGregor 1996c, 1997) showed that it is also late Famennian and no older than the Rugospora radiata – Grandispora cornuta Assemblage Zone of Richardson and McGregor (1986). Similar spore assemblages have been found at the extreme top of the formation elsewhere (McGregor 1996a). Cormier and Kelley (1964) reported a mid- to Late Devonian Rb–Sr age of 376 ± 12 Ma (as recalculated using revised decay constants by Keppie and Smith 1978) for rhyolite in the western belt, consistent with the revised paleontological age.
Farther south in the Lake Ainslie – Gillanders Mountain area (Fig. 2), similar volcanic and sedimentary rocks were also assigned to the Fisset Brook Formation, on the basis of lithologic similarity (Kelley and Mackasey 1965). The correlation was corroborated by recent mapping in the area that showed a stratigraphic sequence of sedimentary rocks, basalt, and rhyolite similar to that in the Cheticamp area (Barr et al. 1995; Barr and Peterson 1998). Rhyolite from near the top of the volcanic succession in the Gillanders Mountain area yielded a U–Pb (zircon) age of 373 ± 4 Ma (Barr et al. 1995), latest Frasnian to early Famennian using the time scale of Tucker et al. (1998).
In the Lowland Cove area (Fig. 2), a 400-m-thick sequence of rhyolite, clastic sedimentary and tuffaceous rocks, basalt, and andesite (Fig. 3) has been correlated with the Fisset Brook Formation (Kelley and Mackasey 1965; Smith and Macdonald 1981). However, the stratigraphic sequence differs from that in the Fisset Brook Formation of the Cheticamp and Lake Ainslie – Gillanders Mountain areas, where rhyolite is at the top of the section. The Lowland Cove sequence contains palynomorphs most recently interpreted as latest Famennian (Strunian) by Martel et al. (1993). A Rb–Sr date from the volcanic rocks of 491 ± 25 Ma (Cormier and Kelley 1964; Keppie and Smith 1978) is thus considered unreliable.
Rocks lithologically similar to the Fisset Brook Formation also occur in several fault-bounded outcrops in the Creignish Hills of western Cape Breton Island (Kelley and Mackasey 1965; MacDougall 1994; Fig. 2). Their inclusion with the Fisset Brook Formation is not certain, and earlier workers had considered them to be Cambrian or older (Ferguson and Weeks 1949). Stratigraphic relations are not clear due to limited outcrop and faults, but rock types include basalt, tuffaceous rocks, rhyolite, conglomerate, and gabbro (MacDougall 1994).
Northeastern Nova Scotia mainland
In the Guysborough area and extending into southernmost Cape Breton Island, extensive mafic and felsic flows and pyroclastic rocks are interbedded with terrigenous sedimentary rocks (Fig. 2) of the Guysborough Group (White and Barr 1998). This fault-bounded package of rocks has been assigned a Devonian age based on a U–Pb (zircon) age of 389 ± 2 Ma from a rhyolitic dome (Cormier et al. 1995). The sequence is intruded by numerous gabbroic dykes and plutons (Cormier et al. 1995; White and Barr 1998).
In upper Cape George Brook (termed Wilkie Brook by Keppie et al. 1978) in the Antigonish Highlands, basalt and rhyolite flows, a few tens of metres thick, are interbedded with volcaniclastic conglomerate and minor shale. The latter rocks have yielded late Famennian palynomorphs of the R. radiata – G. cornuta Assemblage Zone (Martel et al. 1993), similar to those in the type section of the Fisset Brook Formation near Cheticamp (MacGregor 1996a, 1996b). To the north in the Ballantynes Cove area, basalt is overlain by sedimentary rocks, and a grey shale 100 m above the basalt was dated by palynology as probably Late Devonian (Keppie et al. 1978). Basalt and interbedded sedimentary rocks also occur in the McAras Brook area on the western side of the Cape George Peninsula (Fig. 2), and all of these units have been assigned to the McAras Brook Formation (Keppie et al. 1978; Keppie 2000). However, differences in chemical characteristics indicate that not all these volcanic rocks are necessarily correlative (Pe-Piper and Piper 1998).
Northumberland Strait and Prince Edward Island
North of the Cape George Peninsula, the Northumberland Strait F-25 well (Fig. 1) contains basalt flows dated at 362 ± 15 Ma by the K–Ar method (Pe-Piper and Jansa, 1986). Although Pe-Piper and Jansa (1986) interpreted their steep dip as evidence of a pre-Acadian age and interpreted the K–Ar age as thermally reset, the radiometric date and geochemistry are similar to those of Late Devonian – Early Carboniferous volcanic rocks of the southern Magdalen basin (Pe-Piper and Piper 1998).
Our new studies of the Wellington No. 1 well in western Prince Edward Island (Fig. 4) show that about 160 m of predominantly basaltic flows were intersected immediately below the Windsor Group. These basalts overlie a thick clastic sequence, lithologically similar to the Albert Formation of the Horton Group of southeastern New Brunswick, and dated palynologically as late Tournaisian (G. Dolby, personal communication, 1994). In the Irishtown No. 1 well, 45 km to the east, a 230-m-thick basaltic succession occurs in a similar stratigraphic position beneath the Windsor Group.
The oldest basin fill in the Magdalen basin in the Cobequid Highlands of northern mainland Nova Scotia is the Murphy Brook Formation (Fig. 5). This fault-bounded unit consists of black siltstone and argillite interbedded with thick-bedded conglomerate containing abundant angular rhyolite clasts. The flora of late Emsian to early Eifelian age (A. Kasper in Donohoe and Wallace 1985) is of similar age to that of the McAdams Lake Formation in Cape Breton Island.
Interbedded volcanic and sedimentary rocks of the Cobequid Highlands are assigned to the Fountain Lake Group, which has been assumed to be correlative with the Fisset Brook Formation of Cape Breton Island (Fig. 2; Donohoe and Wallace 1980; Calder 1998). Based on field, petrological, and geochemical characteristics, the volcanic rocks have been interpreted to be the extrusive equivalents of high-level gabbroic–dioritic and granitic plutons emplaced along the Cobequid shear zone (Pe-Piper et al. 1989; Koukouvelas et al. 1996). Granitic samples from three of the plutons previously yielded U–Pb (zircon) ages of 362–358 Ma (Doig et al. 1996). Late granitic bodies and numerous mafic dykes cut sedimentary rocks of the Horton Group south of the Rockland Brook and Kirkhill faults (Piper 1994) and north of the Pleasant Hills pluton (Pe-Piper et al. 1998) (Fig. 5).
Volcanic rocks of the Fountain Lake Group in the eastern Cobequid Highlands include some 1–4 km of felsic pyroclastic rocks, with minor mafic and felsic flows, of the Byers Brook Formation (Piper et al. 1999), overlain by <1.5 km of basalt flows and minor rhyolite of the Diamond Brook Formation (Dessureau et al., 2000) (Fig. 3). Spores in siltstone from the middle of the Diamond Brook Formation were reported as early Tournaisian by Donohoe and Wallace (1980), but re-interpreted as late Famennian by Martel et al. (1993). Mid Tournaisian (early Tn3 zone) spores were identified in the upper part of the same formation in the Scotsburn No. 1 well, which penetrated the upper part of the Fountain Lake Group (Utting et al. 1989; see Fig. 5 for location). The Fountain Lake Group in the western Cobequid Highlands is <800 m thick, consists of rhyolite with lesser basalt and minor interbedded sedimentary rocks, and cannot be subdivided into formations (Piper et al. 1999; Fig. 3).
Cormier (1982) suggested that rhyolite in Swan Brook, in the eastern Cobequid Highlands, is older than most of the Fountain Lake Group on the basis of a Rb–Sr isochron age of 387 ± 2 Ma. However, the slope of this isochron is highly dependant on one sample from an apparently unrelated felsic intrusion (Pe-Piper et al. 1989). The Swan Brook volcanic rocks were mapped as part of the upper Byers Brook Formation by Gower (1988), and we concur with this interpretation.
Sample preparation and analytical methods
The analytical techniques used were described in detail by Dubé et al. (1996). The powdered rock sample was panned using a Wilfley table and passed through heavy liquids and a Frantz magnetic separator. The least magnetic high-density fractions contained high quality zircon and baddeleyite grains. Zircons were selected according to criteria of morphology and clarity under a microscope. In most cases, the samples yielded significant amounts of fine- to medium-grained, clear euhedral prismatic zircon with length:breadth ratios of 2:1 to 3:1 and simple prism and pyramidal crystal faces.
The abrasion technique (Krogh 1982) was extensively used to remove outer surfaces in an attempt to minimize or eliminate Pb-loss. Zircons were taken up in HF and HNO3, and purified by ion exchange chemistry. Mass spectrometry was carried out using a multi-collector MAT 262. A series of sets of data were measured in the temperature ranges 1400°C to 1550°C for Pb and 1550°C to 1640°C for U, and the best sets were combined to produce a mean value for each ratio. The measured ratios were corrected for Pb and U fractionation of 0.1% /amu (atomic mass unit) and 0.05% /amu, respectively, as determined from repeat measurements of NBS (National Bureau of Standards) standards. The ratios were also corrected for laboratory procedure blanks (2–10 pg — Pb, 1 pg — U) and for common lead above the laboratory blank with lead of the composition predicted by the two-stage model of Stacey and Kramers (1975) for the age of the sample.
The uncertainties on the isotopic ratios were calculated using an unpublished program and are reported here at 2σ. Sources of uncertainty considered (at 2σ) included uncertainties on the isotopic ratio measurements by mass spectrometry, assigned 80% uncertainty on the Pb and U fractionation, assigned 50% uncertainty on the amount of the Pb and U blanks, and a 4% uncertainty on the isotopic composition of the Pb used to subtract the common lead present above laboratory blank. These uncertainties were quadratically added to arrive at final 2σ uncertainties that are reported after the isotopic ratios in the data table (Table 1).
The ages reported in this paper fall in that part of the time scale from 385 to 350 Ma, near the edge of usability of the 207Pb/206Pb system but where the possibility of Pb-loss exists that might reduce 206Pb/238U ages. The data must be interpreted on a case-by-case basis, considering the quality of the grains and isotopic analyses and the reproducibility of the points on concordia or along a discordia line. The exact age reported is always an interpretation of a small data set.
In the case of discordia lines produced for these rocks, the lower intercepts are all within error of 0 Ma, so that an alternate age interpretation is to calculate a weighted mean of the 207Pb/206Pb ages of the abraded fractions that agree within error. The inherent assumption with this approach is that the abraded fractions were not affected by non-zero age Pb-loss. These ages and corresponding 2σ uncertainties are reported and adopted in this work.
Uncertainties in the U decay constants are not included in the uncertainties reported on the ages. At 360 Ma, this additional source of uncertainty would add ca. ± 60 000 years to the quoted 2σ uncertainties.
Results — Cape Breton Island
(A) Lowland Cove area
The dated sample (A) is from massive porphyritic rhyolite that is well exposed in Lowland Brook, ∼20 m downstream from the faulted contact with Mesoproterozoic basement rocks (Fig. 2). Smith and Macdonald (1981) correlated this unit with rhyolite at the base of the section (overlain by sedimentary rocks, mafic and intermediate tuffaceous rocks, and basaltic and andesitic flows) elsewhere in the Lowland Cove area. However, in Lowland Brook the remainder of the section is missing and the rhyolite is overlain unconformably(?) by the Horton Group (Fig. 3). The rhyolite consists of abundant phenocrysts of sanidine and quartz in a fine-grained spherulitic groundmass. The four analyzed fractions of abraded clear euhedral prisms do not fit a line (3% probability of fit; Fig. 6A). A preferred approach to determining the age of this sample is to average the 206Pb/238U ages of Z1–Z3 (365 ± 2, 365 ± 1.5, 366 ± 2.5), which yields an age of 365 ± 2 Ma for this rock.
(B) Fisset Brook Formation, western belt, Cheticamp area
Sample B is from Factory Brook, near the less accessible type section of the Fisset Brook Formation, in the western belt near Cheticamp (Fig. 2). It is a porphyritic rhyolite from a body a few tens of metres thick, overlying basalt and underlying arkose, and probably located near the top of the formation (Fig. 3). The rhyolite has abundant phenocrysts of quartz and sanidine in a devitrified groundmass. Six fractions of zircon were analyzed and five of these, all abraded, are closely clustered at 1–2.5% discordant with 207Pb/206Pb ages between 370 and 373 Ma (Table 1). Fraction Z6 was not abraded and is more discordant, and a line through all fractions (87% probability of fit) yields an upper intercept age of Ma (Fig. 6B). However the weighted average of the 207Pb/206Pb ages of the five clustered points yields 371 ± 2 Ma, which is adopted as the age of this rock.
(C) Fisset Brook Formation, eastern belt, Cheticamp area
Sample C is from the upper reaches of Forest Glen Brook in the eastern belt of the Fisset Brook Formation in the Cheticamp area (Fig. 2). It is from a flow-banded rhyolite unit, interpreted to be near the top of the formation (Barr and Peterson 1998), and consists of scattered phenocrysts of embayed quartz and sanidine in an eutaxitic groundmass. It yielded a large quantity of clear yellow to turbid brown, small to large euhedral prisms of zircon. Four analyzed fractions, covering the range of grain size and colour (Table 1), yielded significantly discordant data points (8–13% discordant), likely due to their high uranium contents (2100–2600 ppm). These points define a line (14% probability of fit) with an upper intercept age of igneous crystallization of 377 ± 4 Ma (Fig. 6C). The weighted average 207Pb/206Pb age of 374 ± 2 Ma is considered the best estimate of the age of eruption and is identical within error to the age of sample B from the western belt.
(D) Fisset Brook Formation, eastern belt, Cheticamp area
Sample D was collected from Belle Côte Road east of Forest Glen Brook in the eastern belt of the Fisset Brook Formation near Cheticamp (Fig. 2). It represents a rhyolitic vitric-crystal-lithic lapilli tuff unit that is interbedded with basalt and micaceous pebbly sandstone, probably near the top of the formation. Five analyzed zircon fractions consisted of abraded clear sharp euhedral prisms (Table 1). Fraction Z4 was a single flat crystal and fraction Z5 was composed of two identical large equant prisms. Fractions Z1–Z3 have 207Pb/206Pb or 206Pb/238U ages of 426–432 Ma, indicating that they are of Silurian age, inconsistent with the stratigraphic position of the sample. The two analyses, Z4 and Z5, which also consist of euhedral crystals, overlap concordia at 563–555 Ma; fraction Z5 is concordant at 563 Ma (Fig. 6D). These results are all from fractions of equally high quality euhedral prismatic zircon, interpreted as igneous, but it appears that all are inherited. Possible nearby sources are the 430 Ma metavolcanic units of the Cape Breton Highlands to the east and north (Barr and Jamieson 1991) and the 560–550 Ma Cheticamp pluton to the west (Jamieson et al. 1986).
(E) Fisset Brook Formation, Creignish Hills
Dated sample E is from a small fault-bounded area of flow-banded rhyolite adjacent to Highway 19, south of the village of Creigmore and adjacent to the Creignish Hills. The rhyolite consists of scattered phenocrysts of embayed quartz, plagioclase, and sanidine in a fine-grained spherulitic groundmass. The sample yielded a large quantity of small euhedral zircon prisms, many of which are cracked or partially turbid. A small number of flat hexagonal cross-section grains of zircon are also present. Five fractions were analyzed and all are discordant with high uranium contents (Table 1, Fig. 6E). They do not fit a line (only ca. 1% probability of fit). The weighted average of the 207Pb/206Pb ages of 371 ± 3 Ma is adopted as the age of this unit.
Results — northern mainland Nova Scotia
(F) Erinville Gabbro, Guysborough Group
The Erinville gabbro is a small pluton that intrudes, with a contact aureole, shale and quartz wacke of the Guysborough Group (Cormier et al. 1995; White and Barr 1999). The dated sample (F) consists of coarse-grained plagioclase and clinopyroxene, with abundant magnetite, ilmenite, and apatite. The sample yielded high quality coarse-grained zircon fragments and medium-grained euhedral baddeleyite, and two fractions of each were analyzed (Table 1, Fig. 6F). One fraction of zircon contains an inherited component and plots to the right of concordia, whereas the remaining zircon fraction and both baddeleyite fractions are concordant, with the zircon slightly below the baddeleyite, possibly due to minor lead loss. Based on the weighted average of the 207Pb/206Pb ages of the three clustered analyses, this rock is interpreted to have crystallized at 385 ± 4 Ma. The gabbro age and a previously reported U–Pb (zircon) age of 389 ± 2 Ma for a rhyolitic tuff in the uppermost formation of the Guysborough Group effectively constrain the age of the top of the group to the early Middle Devonian (Eifelian; time scale of Tucker et al. 1998).
(G) Cape George Brook (McAras Brook Formation)
Sample G is from a flow-banded porphyritic rhyolite at the top of a succession of clastic sedimentary rocks (Fig. 3) ranging from conglomerate to shale with several interbedded basalt and rhyolite flows in Cape George Brook, north of Antigonish (Fig. 2). The rhyolite has phenocrysts of quartz, feldspar, and Fe–Ti oxide with a devitrified hypocrystalline groundmass. A large amount of fine- to coarse-grained euhedral zircon was recovered from this sample. Eight analyses were carried out, of various grain sizes of zircon, both abraded and not (Table 1, Fig. 6G). Six abraded fractions are closely clustered and have 207Pb/206Pb ages of 369–371 Ma. The two unabraded fractions are 7.7% and 9% discordant (not shown on Fig. 6G) and all eight analyses define a line (99% probability of fit), which yields an upper intercept age of Ma. The weighted average 207Pb/206Pb age of the six abraded fractions is 370 ± 1.5 Ma.
Results — Cobequid Highlands
(H) Byers Brook Formation, Clear Lake
In the eastern Cobequid Highlands (Fig. 5), a porphyritic flow-banded rhyolite sample (H) was collected from the upper part of the Byers Brook Formation of the Fountain Lake Group, a short distance below the lowest basalt flows of the Diamond Brook Formation (Fig. 3). Phenocrysts and microphenocrysts comprise quartz, sanidine, plagioclase, Fe–Ti oxides (altered to hematite), and altered biotite, set in a fine-grained matrix. The sample yielded high quality zircon and three fractions of the best abraded prisms are all concordant with 206Pb/238U ages of 357–358 Ma (Table 1, Fig. 6H), providing a weighted average 206Pb/238U age of 358 ± 1 Ma for crystallization of this unit.
(I) Diamond Brook Formation, Whirley Brook
A porphyritic flow-banded rhyolite dome is intercalated with basalt flows in the middle part of the Diamond Brook Formation south of New Annan (Figs. 3, 5). The rhyolite has phenocrysts and microphenocrysts of sanidine, plagioclase, quartz, and Fe–Ti oxides, with accessory apatite and zircon, in a devitrified groundmass. Three fractions of abraded euhedral prisms were analyzed from a rhyolite sample (I) from this dome. Two fractions (Z1, Z2) are concordant and overlapping with 206Pb/238U ages of 353 and 354 Ma (Fig. 6I). The third fraction, composed of the largest prisms, is discordant with a 207Pb/206Pb age of 356 Ma. The preferred age for this rock, from the weighted average of the 207Pb/206Pb age, is 355 ± 3 Ma.
(J) Porphyritic rhyolite in the Pleasant Hills pluton
Sample J is from a 100-m-wide porphyritic rhyolite (very fine-grained granite) sheet in the Pleasant Hills pluton in the Economy River gorge (Fig. 5). The fine-grained granite comprises K-feldspar, plagioclase, quartz, and Fe–Ti oxides. It is interpreted to be a late sill within the pluton (Pe-Piper et al. 1998), the main coarse-grained granitic unit of which was dated previously at 361 ± 2 Ma (U–Pb zircon) by Doig et al. (1996). The rhyolite sample yielded a large amount of high quality zircon. Of five fractions analyzed, only one is concordant with a 206Pb/238U age of 356 ± 2 Ma (Fig. 6J). The other fractions clearly contain inherited zircon of different ages. Two lines were calculated to identify average ages of inherited zircon present as cores and both yield mid-Proterozoic ages (Z1, Z2, Z3; 31% probability of fit, lower intercept (LI) age = 354 ± 3 Ma, upper intercept (UI) age = 1340 ± 50 Ma: Z1, Z2, Z4; 64% probability of fit, LI = 356 ± 2 Ma, UI = 1735 ± 35 Ma). Based on the single concordant point, this rock crystallized at 356 ± 2 Ma.
(K) Porphyritic rhyolite sill, northern margin of Pleasant Hills pluton
Sample K is from a 70-cm-thick sill of porphyritic rhyolite (very fine-grained granite) that cuts cleaved sandstone and siltstone about 100 m north of the Pleasant Hills pluton. The host sedimentary rocks contain granitic debris presumably derived from the pluton and are assigned to the Horton Group (Piper 1996; Fig. 5). Of three analyses from this sample, two are concordant and overlap with 206Pb/238U ages of 355 and 357 Ma, yielding an age and uncertainty of 356 ± 3 Ma (Fig. 6K). Analysis Z3 contains a minor inherited component.
(L) Rhyolite sill in the Murphy Brook Formation
Farther north of the Pleasant Hills pluton, rhyolite (fine-grained granite) sills occur along thrust faults in the Silurian Wilson Brook and Middle Devonian Murphy Brook formations (Piper and Pe-Piper, 2001). Sample L was collected from a representative sill in the Murphy Brook Formation (Fig. 5), which shows evidence of minor brittle post-intrusion thrust deformation. It was dated to establish whether these sills are analogous to the main phases of the Pleasant Hills pluton or whether they represent a separate late event correlative with samples J and K. The sample is a porphyritic rhyolite with phenocrysts of feldspar (<8 mm) and quartz and accessory sulphides, zircon, and Fe–Ti oxides. The groundmass was probably originally glassy as spherulitic textures are developed in places. Four fractions of small sample weights of clear euhedral abraded zircon prisms show evidence of an inherited older zircon component in this sample and fraction Z1 is perfectly concordant at 360 ± 2 Ma (Fig. 6L). The line shown (95% probability of fit) yields a UI average age of the inherited component of Ma. We adopt the 206Pb/238U age of the concordant point (360 ± 2 Ma) as the age of crystallization of this rhyolite sill.
(M) Granite from margin of the West Moose River pluton
Sample M is a medium-grained granite collected from the West Moose River pluton near Humming Brook (Fig. 5). It was located one metre from an intrusive contact with hornfelsed argillite and siltstone of the Horton Group (Koukouvelas et al. 1996). The West Moose River pluton is the only pluton in the Cobequid Highlands in which the major plutonic body shows an intrusive contact with the Horton Group; it was dated to constrain the age of the Horton Group in this area. Five fractions of clear euhedral abraded zircon prisms were analyzed from this sample. Z5 (not shown on Fig. 6M) is displaced to the right of the line and must include inherited older zircon. Analyses Z1–Z4 define a discordia line with a 89% probability of fit, which yields a UI age of crystallization of Ma (Fig. 6M).
(N) Fountain Lake Group rhyolite, north of West Moose River pluton
The upper part of the Fountain Lake Group was sampled near the West Moose River pluton. Sample N is from porphyritic flow-banded rhyolite interbedded with abundant vitric-lithic tuff, many of which are ignimbrite (Fig. 3). The rhyolite has phenocrysts and microphenocrysts of K-feldspar, plagioclase, quartz, and rare Fe–Ti oxides in a spherulitic devitrified groundmass. Four analyzed fractions of abraded clear euhedral zircon show the presence of an inherited component, but two analyses separately overlap concordia with 206Pb/238U ages of 356 and 362 Ma (Fig. 6N). The age of eruption is considered to be that of the younger concordant point (Z1 — 356 ± 2 Ma), which is of high quality, clear, euhedral abraded zircon grains, that are interpreted not to have undergone lead loss. It is possible that the older point overlapping concordia contains a trace of inheritance. This age of 356 ± 2 Ma is based on a judgement call with only one data point and therefore should be viewed with caution.
(O) Fountain Lake Group rhyolite, Squally Point
The base of the Fountain Lake Group was sampled at Squally Point (Fig. 5) from a flow-banded rhyolite dome near the base of a 100-m-thick sequence of pyroclastic and hyaloclastic rhyolite with minor interbedded basalt and andesite (Fig. 3) (Piper et al. 1995). This volcanic succession at Squally Point appears geochemically related to the nearby Cape Chignecto pluton, where the main phase granite was dated at 358 ± 2 Ma (U–Pb zircon) by Doig et al. (1996). The rhyolite sample (O) yielded abundant zircon, and four analyses of abraded prisms of differing quality are all concordant with 206Pb/238U ages between 355 and 349 Ma. Fractions Z3 and Z4 are interpreted to display a small degree of Pb-loss, consistent with their lower quality and higher common lead contents. The tightly overlapping and concordant points Z1 and Z2 indicate an age of eruption of 355 ± 2 Ma (Fig. 6O). Because of the lower quality of fractions Z3 and Z4, the weighted average of the 207Pb/206Pb ages (354 ± 6 Ma) is not adopted here.
Compatibility with the current Devonian–Carboniferous time scale and biostratigraphic zonation
Ages of 374 ± 2 Ma, 371 ± 2 Ma, and 373 ± 4 Ma for rhyolite of the Fisset Brook Formation in both the Cheticamp and Lake Ainslie – Gillanders Mountain areas (Table 2) are identical within error and indicate an age of latest Frasnian to early Famennian, using the time scale of Tucker et al. (1998) (Fig. 7). Macrofossils in sedimentary rocks underlying the rhyolite unit in the eastern Cheticamp belt are Late Devonian (Kasper et al. 1988), and units both underlying (in the eastern belt) and overlying (in the western belt) the rhyolite contain Late Famennian R. radiata G. cornuta palynomorph assemblages. Rhyolite in Cape George Brook, with an age of 370 ± 1.5 Ma, overlies rocks containing the same assemblage (Martel et al. 1993). The rhyolite at Lowland Cove, dated at 365 ± 2 Ma, was interpreted by Smith and Macdonald (1981) to underlie sedimentary rocks containing palynomorphs most recently interpreted by Martel et al. (1993) to be latest Famennian (Strunian) (Fig. 7). These data confirm that the Famennian was a stage of relatively long duration, and perhaps even longer than suggested by Tucker et al. (1998) if the maximum error range of the U–Pb data is considered. The data also provide constraints on the absolute ages of the R. radiata – G. cornuta and Strunian biostratigraphic zones (Fig. 7).
Rhyolite from the middle of the Diamond Brook Formation, which yielded a 355 ± 3 Ma age, stratigraphically underlies rocks containing mid Tournaisian palynomorphs and overlies rocks with late Famennian palynomorphs (Figs. 3, 7). The latter overlie the rhyolite with an age of 358 ± 1 Ma in the upper part of the underlying Byers Brook Formation. Rhyolite with an age of 365 ± 2 from Lowland Cove underlies Tournaisian Horton Group rocks and probably also underlies Strunian (latest Famennian) “Fisset Brook Formation” (Fig. 3).
The age of the sedimentary rocks cut by the West Moose River granite ( Ma) is not constrained by biostratigraphy, but the oldest part of the Horton Group in the type area is in the youngest miospore zone of the Strunian (Martel et al. 1993). Hence the radiometric age suggests that the sedimentary units intruded by the pluton are near the base of the Horton Group.
Stratigraphic implications of the new geochronology
The minimum age of volcanic and sedimentary rocks of the Guysborough Group is now well constrained by the U–Pb age of 389 ± 2 Ma for a rhyolite from the uppermost formation in the group (Cormier et al. 1995; White and Barr 1999) and the age reported here for the Erinville gabbro that intrudes an underlying formation. These constraints indicate a minimum age of Eifelian, similar to that of the McAdams Lake Formation in Cape Breton Island (White and Barr 1998). The Murphy Brook Formation in the central Cobequid Highlands (Fig. 5) also contains a flora of late Emsian to early Eifelian age (A. Kasper in Donohoe and Wallace 1985). Conglomerate in the formation contains abundant angular rhyolite clasts that may be evidence for mid-Devonian volcanism in the area. These units appear to be the oldest preserved units representing the early stages of development of the Magdalen basin.
Late Devonian of western Cape Breton Island and northeastern mainland Nova Scotia
The new age of 371 ± 2 Ma for rhyolite in the type section of the Fisset Brook Formation in the western belt near Cheticamp is identical within error to that from the eastern belt (374 ± 2 Ma) and from the Lake Ainslie – Gillanders Mountain area (373 ± 4 Ma; Barr et al. 1995) and shows that volcanism in these areas was approximately synchronous. In the adjacent Cape Breton Highlands, deeper levels of erosion have exposed widespread plutons of the same age (Fig. 2; Table 2), including the Salmon Pool granite ( Ma; Jamieson et al. 1986), West Branch North River granite (376 ± 3 Ma; G.R. Dunning, unpublished data), and the Black Brook Granitic Suite (373 ± 2 Ma; Dunning et al. 1990). The Margaree and Gillanders Mountain plutons are probably of similar age (Barr et al. 1995). Plutons of similar age also occur in southeastern Cape Breton Island, the most precisely dated of which is the Lower St. Esprit Pluton ( Ma; Bevier et al. 1993). Hence, early Late Devonian igneous activity was widespread in Cape Breton Island (Figs. 2, 7) and not confined to the basins where volcanic rocks are now preserved. Furthermore, the age of the igneous activity is similar to that of the voluminous South Mountain Batholith and related plutons in southern Nova Scotia (ca. 375 Ma; see Keppie 2000 for a compilation).
The age of 371 ± 3 Ma for rhyolite in the Creignish Hills volcanic–sedimentary unit is similar (overlapping within error) to those obtained from rhyolite units in the Cheticamp and Lake Ainslie – Gillanders Mountain areas (Table 2), and hence confirms that these rocks should be considered part of the Fisset Brook Formation. In contrast, spore and radiometric age data suggest that the volcanic and sedimentary rocks in the Lowland Cove area are younger than the Fisset Brook Formation in the type area, but are time-equivalent to the lower part of the Horton Group (Fig. 7). We suggest that these rocks be excluded from the Fisset Brook Formation and be called the Lowland Cove Formation.
The age of 370 ± 1.5 Ma obtained from the rhyolite in Cape George Brook and the biostratigraphic correlation of the R. radiata – G. cornuta assemblage suggests that the McAras Brook Formation is of similar age to the Fisset Brook Formation. The basalt flow dated as 362 ± 15 Ma by K–Ar in the Northumberland Strait F-25 well (Fig. 1) is geochemically similar to basalt from Cape George Brook and may be correlative.
Latest Devonian – earliest Carboniferous igneous rocks of the Cobequid Highlands
Granite in three plutons in the Cobequid Highlands had been dated previously by Doig et al. (1996) between 362 and 358 Ma using U–Pb on zircon (Table 2), consistent with earlier K–Ar ages (Piper et al. 1993). The present study demonstrates that the West Moose River pluton, which intrudes the Horton Group, is also of this age ( Ma). Rhyolite sills in the Murphy Brook Formation north of the Pleasant Hills pluton are of the same age (360 ± 2 Ma), supporting the concept that plutons like the Pleasant Hills pluton evolved from initial intrusion of dykes and sills (Pe-Piper et al. 1998). A late-stage sill in the Pleasant Hills pluton is resolvably younger, as its age of 356 ± 2 Ma is below the error range for most of the pluton samples (Table 2), and a nearby sill cutting Horton Group (356 ± 3 Ma) is of similar age.
In the eastern Cobequid Highlands, the age of 358 ± 2 Ma from rhyolite near the top of the Byers Brook Formation, which is composed of predominantly felsic volcanic rocks, suggests that the formation is the extrusive equivalent of the granitic plutons, as previously suggested on the basis of geochemical similarities (Pe-Piper et al. 1991). The age of 355 ± 3 Ma from a small rhyolite dome in the middle part of the overlying Diamond Brook Formation (predominantly basalt flows) is similar to the age of late rhyolite sills in the margin of the Pleasant Hills pluton and to the alternating rhyolites and basalts of the Fountain Lake Group farther west in the Cobequid Highlands (356 ± 2 Ma on West Moose River road, 355 ± 2 Ma at Squally Point). In the Cobequid Highlands, the Fountain Lake Group is generally restricted to the area north of the Rockland Brook Fault and the Kirkhill Fault (except between Parrsboro and the West Moose River pluton). In contrast, the Horton Group generally occurs to the south of these faults, with outliers north of the Pleasant Hills pluton and northwest of Earltown. The radiometric and biostratigraphic ages of the Fountain Lake Group are synchronous with the biostratigraphic age of the lower half of the Horton Group in the type section (Martel et al. 1993), and the ages of felsic rocks cutting the Horton Group confirm this correlation.
Basin-wide tectonic implications
The variation in spatial distribution of volcanism through time may be related to changing patterns of deformation along the complex convergence zone of the Meguma terrane with the rest of the Appalachian orogen. The new radiometric dates, together with re-assessment of previously published dates and biostratigraphy, suggest that Devonian–Carboniferous volcanic and related plutonic rocks in northern mainland Nova Scotia and Cape Breton Island fall into four age groups (Fig. 7). Sporadic younger activity occurred in the Magdalen Basin and through New Brunswick.
(1) Middle Devonian (Eifelian)
Middle Devonian volcanism in the Guysborough area of northern mainland Nova Scotia is spatially related to the Cobequid–Chedabucto fault, the boundary between the Avalon and Meguma terranes (Fig. 2). Webster et al. (1998) suggested that preservation of the Guysborough block may be related to a restraining bend on the Meguma–Avalon terrane boundary, and that this boundary may lie along the northern margin of the Guysborough Group. The Guysborough area volcanism postdates early metamorphism and cooling of the Meguma Group (Keppie and Dallmeyer 1987; Hicks et al. 1999) and is synchronous with earliest basin development in the Maritimes Basin, represented by the nonvolcanic McAdams Lake and Murphy Brook formations.
(2) Late Devonian (Famennian)
The thick volcanic sequences of the Fisset Brook Formation are of similar age to widespread plutonic rocks in Cape Breton Island. This activity is also synchronous with the voluminous South Mountain Batholith and associated plutons of the Meguma terrane. The distribution and shape of units in Cape Breton Island (Fig. 2) suggests that igneous activity was localized by major transcurrent to oblique-slip faults such as the Aspy fault, Eastern Highlands shear zone, Canso fault, and Mira River fault. The heat source may have been either a mantle plume (Murphy et al. 1999) or increased geothermal gradient as a result of regional extensional detachment (Lynch and Tremblay 1994). Our data does not provide a means of distinguishing whether either of these hypotheses is valid.
(3) Latest Devonian (Strunian) to Early Carboniferous (early Tournaisian)
Igneous activity in the Cobequid Highlands is also clearly localized by major faults. The latest Famennian to Tournaisian igneous activity was focussed in the Cobequid Highlands along the Rockland Brook Fault and its continuation to the west as the Kirkhill Fault (Fig. 3). Volcanic rocks of the Fountain Lake Group are common north of these faults. Their extrusion was synchronous with or immediately followed by northwest-directed transpression (Waldron et al. 1989; Piper et al. 1995; Piper and Pe-Piper 2001). The older rocks (362–358 Ma) are principally granite plutons and rhyolite of the Byers Brook Formation; younger rocks (356–355 Ma) include basalt of the Diamond Brook Formation and rhyolite–basalt volcanic successions of the western Cobequid Highlands.
The Rockland Brook Fault of the Cobequid Highlands continues to the northeast as a series of faults along the northwestern edge of the Cape George peninsula and through western Cape Breton Island (Durling et al. 1995). These faults may also have locally served as magma conduits in the Lowland Cove Formation.
(4) Early Carboniferous (late Tournaisian – early Visean)
Mafic dykes are relatively common in rocks of the Horton Group in Nova Scotia, whereas to the knowledge of the authors, dykes in the Windsor Group have not yet been documented. This difference suggests that the widespread late Tournaisian – early Visean hiatus, commonly marked by an unconformity between the Horton and Windsor groups, particularly at basin margins, may have been a time of intrusion of these dykes. The basalt near St. Peters appears to be near the Horton–Windsor contact (Barr et al. 1995). The Wellington No. 1 and Irishtown No. 1 basalts, overlying a thick Horton Group succession and immediately underlying the Windsor Group, appear to fall in this same time interval.
This Late Tournaisian – early Visean mafic volcanism and hypabyssal intrusion is widespread along the Cobequid–Chedabucto fault zone, occurring in southern Cape Breton Island and the Cobequid Highlands. Dykes are most voluminous cutting Horton Group sedimentary rocks close to the Cobequid Fault in the Cobequid Highlands. Igneous activity may be developed along other lineaments, including major faults in New Brunswick (Fyffe and Barr 1985). The voluminous basalts in the Wellington No. 1 and Irishtown No. 1 drill holes, and the abundant dykes cutting the Horton Group of western Cape Breton lie on a prominent east–west magnetic anomaly (e.g., Oakey and Dehler 1998). This area may be underlain by underplated mafic magmas, as suggested in the Magdalen Basin to the north (Marillier and Verhoef 1989).
Based on available geochronological data, igneous activity related to the extension of the Maritimes Basin occurred in four pulses, centred around 387, 373, 358, and 339 Ma. Igneous activity has not been documented in intervening periods. Each pulse corresponds to a distinct regional tectonic style. The oldest, perhaps restricted to the Meguma terrane, corresponds to the oldest shear-related plutons in the Meguma terrane. The second is synchronous with widespread plutonism in Cape Breton Island and the Meguma terrane and appears localized by major faults in Cape Breton Island. The third corresponds to a period of transpression along the Rockland Brook fault and its continuation in western Cape Breton, with most igneous activity in the Cobequid Highlands. The fourth pulse was associated with the Horton–Windsor unconformity, and has igneous activity concentrated along the Cobequid–Chedabucto fault trend and in P.E.I. and western Cape Breton Island.
Our new radiometric data are generally consistent with the Devonian time scale recently proposed by Tucker et al. (1998). Our dates confirm the relatively long duration of the Famennian stage, and thus resolve apparent conflicts between radiometric and biostratigraphic data.
This work was funded by the Canada – Nova Scotia Cooperation Agreement on Mineral Development 1993–96, the Geological Survey of Canada, and Natural Sciences and Engineering Research Council grants to G.R. Dunning, G. Pe-Piper, and S.M. Barr. The broad backs of Craig Doucette, Mary Feetham, Jason Goulden, Gordon Guy, Stan Johnston, Karen Lister, Tracie Quinlan, Derek Robichaud, and Chris White assisted with sample collection. Reviews by Brendan Murphy, Ken Ludwig, and Don Davis sharpened our thinking.
↵1 Geological Survey of Canada Contribution 1999192.
- Received August 3, 2001.
- Accepted June 6, 2002.
- Published on the NRC Research Press Web site at http://cjes.nrc.ca on August 26, 2002.
- © 2002 NRC Canada