Machias Seal Island is located at the mouth of the Bay of Fundy, about 20 km southwest of Grand Manan Island and 30 km southeast of Machias, Maine (Fig. 1). The small rocky island is barren except for a lighthouse and dwellings maintained by the Government of Canada, although ownership of the island is disputed between Canada and the United States of America (e.g., Schmidt 2002). The island has the last occupied lighthouse in the Maritime Provinces, and is well known among bird-watchers as a nesting site for puffin, auk, and other seabirds. In addition to its political significance, Machias Seal Island is important geologically because of its location in an area through which it is difficult to trace terranes from Nova Scotia and New Brunswick into the New England states (Fig. 1). Geological studies on nearby Grand Manan Island (Fig. 1) have not resolved that problem, as rocks there have equivocal terrane affinity (e.g., Miller et al. 2007; Fyffe et al. 2009). However, they most resemble those of the New River terrane of southern New Brunswick, suggesting that the Ganderian Kingston and Brookville belts, as well as Avalonia and Meguma, all lie out-board of Grand Manan Island (Fig. 1).

Because of its small size, most regional geological maps (e.g., Hibbard et al. 2006) do not show Machias Seal Island. More detailed local maps have shown the island as Precambrian granite (e.g., Potter et al. 1979) or Silurian–Devonian granite (e.g., McLeod et al. 1994; New Brunswick Department of Natural Resources 2010). The purpose of this paper is to describe the petrology of a suite of granitoid samples collected from the island, present a new U-Pb zircon age for the suite, and use these results to interpret the most likely correlative units in the region.

With the exception of grassy areas in the central part of the island, outcrop is continuous on Machias Seal Island (Fig. 2). The rocks are grey to locally pink, fine- to medium-grained, weakly foliated quartz monzodiorite with abundant dioritic enclaves, generally less than 20 cm in diameter. The quartz monzodiorite is cut by two steeply dipping mafic dykes, each about 1 m in width, which trend at about 015° across the island (Fig. 2). The eastern dyke is alkalic, and contains pseudo-morphs of olivine phenocrysts in a fine-grained groundmass of plagioclase, brown amphibole, and pyroxene. The other dyke is more altered and consists mainly of plagioclase, pyroxene, and their alteration products. The age of these dykes is unknown, other than being younger than the quartz monzodiorite.

Typical Machias Seal Island quartz monzodiorite contains 50% strongly zoned plagioclase, 30% mafic minerals and about 20% interstitial quartz and K-feldspar (Figs. 3a, b). Plagioclase is strongly zoned from labradorite to oligoclase, based on petrographic determinations using extinction angles and on electron microprobe analyses in 3 samples (Fig. 4a). The most abundant mafic mineral is green amphibole of magnesio-hornblende composition (Fig. 4b). The larger amphibole grains typically contain relict cores of both orthopyroxene and clinopyroxene, with the clinopyroxene rimming the orthopyroxene. The orthopyroxene is somewhat less magnesian than the coexisting clinopyroxene (Fig. 4c). Brown biotite occurs in association with amphibole and as separate grains. In both modes of occurrence, its composition is intermediate between phlogopite and annite, with relatively low aluminum content, and it plots in the field for calc-alkalic orogenic suites in the ternary MgO-FeO^t-Al₂O₃ discrimination diagram (Fig. 4d) of Abdel-Rahman (1992). Apatite, zircon (see description below), and magnetite are the most abundant accessory phases.

The enclaves are finer grained than their host rocks and of more mafic (dioritic) composition. They consist of plagioclase laths, orthopyroxene, clinopyroxene, amphibole, and biotite in a groundmass of plagioclase and K-feldspar (Figs. 3c, d). In contrast to the host monzodiorite, quartz is minor or absent in the enclaves. The compositions of the mafic minerals in the enclaves are similar to those in the host quartz monzodiorite (Fig. 4b, c, d), although the plagioclase is generally more calcic, up to An₉₀ (Fig. 4a).

A sample of quartz monzodiorite (#182) from the eastern side of the island (Fig. 2) was collected for dating at the Pacific Centre for Isotopic and Geochemical Research (PCIGR). Zircons were separated using conventional crushing, grinding and wet shaking table methods, followed by heavy liquid and magnetic separation. The sample yielded abundant zircons, comprising clear, colorless to pale yellow, stubby square prisms (length:width ratios of 1-2) with simple terminations. Some of the grains were fractured; however, no internal zoning or inherited cores were observed under a binocular microscope or in transmitted light.

Methods used for U-Pb zircon analysis using ID-TIMS techniques at the PCIGR are described by Mortensen et al. (1995). Individual zircon fractions for analysis comprised 1 to 11 grains. U-Pb analytical data are listed in Table A5, and are shown on a conventional concordia plot in Figure 5. Four multigrain fractions of zircon (fractions B, C, D, E) were analyzed initially. The outer portions of these zircon grains had been strongly air abraded prior to dissolution in an attempt to minimize the effects of post-crystallization Pb-loss. The four analyses defined a short linear array that suggested an upper intercept age of approximately 542 Ma; however, there is significant scatter in the data and only analysis C is concordant (Fig. 5). In an attempt to obtain replicate concordant analyses and thus more confidently constrain the age of the rock, three additional zircon fractions were analyzed (B2, D2, E2). These analyses are of single zircon grains that were “chemically abraded” prior to dissolution. The chemical abrasion process involves thermally annealing the grains and then subjecting them to a strong leaching step in concentrated hydrofluoric acid; this procedure has been shown to remove most portions of zircon grains that have been altered or otherwise disturbed, and in many cases produces more concordant and reproducible analyses (see discussion of the chemical abrasion technique as used in the PCIGR in Scoates and Friedman 2008). Two analyses of the chemically abraded zircons still show minor discordance, suggesting that in those cases not all of the altered and disturbed portions of the grains were removed by the process. One of the analyses (B2), however, yields a concordant analysis that completely overlaps with air-abraded fraction C (Fig. 5). A regression of all of the data gives calculated upper and lower intercept ages of 541.9 ± 3.5 Ma and -125 ± 680 Ma, respectively (MSWD = 0.61; probability of fit = 0.69). A regression of all seven analyses forced through the origin yields an upper intercept age of 542.5 ± 2.9 Ma (MSWD = 0.53; probability of fit = 0.79). We consider the best estimate for the crystallization age of the sample, however, to be given by a weighted average of ²⁰⁶Pb/²³⁸U ages for the two concordant analyses (C and B2), at 542.0 ± 0.9 Ma.

This age coincides closely with the Ediacaran-Cambrian boundary (Walker and Geissman 2009) and is much older than the Early Devonian age previously inferred for intrusive rocks on the island on current maps (e.g., New Brunswick Department of Natural Resources 2010). Regionally, it falls within the range of U-Pb (zircon) ages obtained from numerous plutons in the Brookville terrane of southern New Brunswick (Fig. 6; e.g., White et al. 2002). It is also similar to the ages of some units in the New River terrane (Fig. 6) and three units on nearby Grand Manan Island: High Duck Island Granite (547.3 ± 1.1 Ma; Miller et al. 2007), tuff in the Priest Cove Formation (539 ± 3.3 Ma; Miller et al. 2007), and the Stanley Brook granite (535 ± 2.5 Ma; P. Valverde-Vaquero, unpublished written report to L.R. Fyffe, 2003). The similarity between Grand Manan Island and the New River terrane in rock types, Sm-Nd isotopic characteristics, and ages has been used to infer a link between those two areas, and for including both of them, as well as the Brookville terrane, in Ganderia (Barr et al. 2003; Miller et al. 2007; Fyffe et al. 2009). As documented in detail by Barr et al. (2003) and also discussed in subsequent papers (e.g., Miller et al. 2007; Fyffe et al. 2009), differences in rock types and ages indicate that none of these areas is likely to be related to the Avalonian Caledonia terrane.

Chemical analyses were obtained for 5 samples of the Machias Seal Island quartz monzodiorite and 2 samples from its dioritic enclaves. The quartz monzodiorite samples range in SiO₂ content from 59% to 63%, whereas the enclaves have lower SiO₂ (56% and 54%) (Fig. 7). A plot of CIPW normative compositions (Fig. 8a) is consistent with the names assigned on the basis of modal estimates (diagram not shown). The enclaves and their host rocks tend to lie on linear trends in major element compositions (Fig. 7, 8b), suggesting that they are linked by crystal fractionation processes. Chondrite-normalized rare-earth element (REE) patterns are also similar in enclave samples and quartz monzodiorite sample NB04-176a, the most mafic of the analyzed quartz monzodiorite samples. The most felsic sample, NB04-182, shows higher REE, especially light REE, and a larger negative Eu anomaly consistent with feldspar fractionation (Fig. 9a). A multi-element comparison diagram also shows similarity between enclaves and host quartz monzodiorite samples, and a trend toward increasing incompatible element abundance with increasing silica content in the samples (Fig. 9b).

Because the age obtained for the quartz monzodiorite falls within the range of ages reported for Brookville terrane plutons (Fig. 6), and some of those plutons include similar monzodioritic rock types (e.g., White et al. 2002), fields are shown on the geochemical diagrams for those plutons (Figs. 7-9). In most cases the Machias Seal Island samples plot within the fields defined by the Brookville terrane plutons, and display similar trends, consistent with magma evolution dominated by crystal fractionation of plagioclase and mafic minerals. The REE patterns are also similar, although the Machias Seal Island enclave samples plot at the uppermost part of the range for plutons of the Brookville terrane, and the dated sample lies well above the range. The cause of this difference is not apparent in the mineralogy of the samples.

The Machias Seal Island samples have chemical features consistent with origin in a continental margin subduction zone (Fig. 10), an interpretation also made for the Brookville terrane plutons in earlier studies (White et al. 2002).

The epsilon Nd value of -1, calculated at 540 Ma (Table A8), is similar to that of many samples in the Brookville and New River terranes (Fig. 11). The tendency for igneous units to have negative epsilon Nd values is considered to be one of the characteristic features of Ganderia (Kerr et al. 1995; Samson et al. 2000; Barr et al. 2003).

The similarity in both age and petrology of the Machias Seal Island quartz monzodiorite to plutons of the Brookville terrane supports the interpretation that Machias Seal Island is part of the Brookville terrane. Although it is the closest land area to Machias Seal Island, none of the rocks on Grand Manan Island is similar in age and petrological characteristics to the Machias Seal Island quartz monzodiorite. Much of Grand Manan Island is composed of Mesozoic basalt, and the remaining third of the island is dominated by volcanic and sedimentary rocks (Fyffe and Grant 2005; Black 2005; Miller et al. 2007). The Three Islands Granite forms islands south of the main island, but it has been dated at 611 Ma, much older than the ca. 540 Ma Machias Seal Island quartz monzodio-rite. Two small plutons on Grand Manan Island have yielded Ediacaran-Early Cambrian ages of 547.3 ± 1.1 Ma (Miller et al. 2007) and 535 ± 2.5 Ma (P. Valverde-Vaquero, unpublished written report to L.R. Fyffe, 2003), similar to the age of the Machias Seal Island quartz monzodiorite, but both are granitic in composition and hence petrologically unlike the quartz monzodiorite (Black 2005).

The assemblage of pre-Mesozoic rocks on Grand Manan Island and their ages led to tentative correlation with the New River terrane in mainland New Brunswick (Miller et al. 2007; Fyffe et al. 2009). The relationship between the New River and Brookville terranes is uncertain, although both are interpreted to be part of Ganderia. The position of Machias Seal Island so close to Grand Manan Island supports the interpretation that both Brookville and New River terranes are part of Ganderia, whereas Avalonia is located farther offshore (Fig. 1). Geophysical data from the area (e.g., Hutchinson et al. 1988; Keen et al. 1991) are not sufficiently detailed to enable resolution of these details, but the results of this study are consistent with the interpretation that the Fundy Fault, previously identified to the southeast on the basis of geophysical data, marks the boundary between Ganderia and Avalonia (Fig. 1).

Overall, the results of this study show that the geology of Machias Seal Island is closely linked to that of mainland New Brunswick, not adjacent parts of Maine. The latter area is dominated by Silurian–Devonian granitoid rocks of the Coastal Maine magmatic province (e.g., Hogan and Sinha 1989).

Machias Seal Island is composed of plutonic rocks of similar age and petrological features to those of the Brookville terrane in mainland southern New Brunswick. They are more than 100 million years older than the granitoid rocks that characterize the Coastal Maine magmatic province, with which they had previously been correlated. Although direct links with rock units exposed on nearby Grand Manan Island cannot be made, some units there are of similar age, supporting Ganderian affinity for both areas.