7 The Birchy Complex (Hibbard 1983) occurs in the immediate structural footwall of the ca. 490 Ma supra-subduction zone ophiolites (Figs. 1, 2; Hibbard 1983; Dunning and Krogh 1985; Cawood et al. 1996; Bédard et al. 2000; Skulski et al. 2010) of the Baie Verte oceanic tract (van Staal et al. 2007). The Birchy Complex comprises highly strained and metamorphosed polyphase-folded mafic schists (Fig. 4E, F) that are locally interlayered with psammite, graphitic pelite, calc-silicate, coticule, jasper and ultramafic rocks (Figs. 2, 4), and forms a steeply dipping, thin (ca. 1.0 – 2.5 km) structural unit.

8 The ultramafic rocks in the Birchy Complex vary from brecciated talc- and/or tremolite-bearing serpentinite, to listwanite (Fig. 4C) and bright green fuchsite – actinolite/tremolite schist. They principally occur as metreto decimetre-scale lenses in highly deformed graphite-bearing mica schist and other metasedimentary rocks (Fig. 4B). Notably, prominent bodies of bright green fuchsite – actinolite/tremolite schist, which probably represent metamorphosed chromite-bearing pyroxenite and/or websterite bodies (see below), stand out and outline highly boudinaged and isoclinally folded horizons within the graphitic schist. Metasedimentary rocks locally contain detrital chromite, suggesting that they were in part derived from the ultramafic rocks. The protoliths of the mafic schists include metagabbro (Fig. 4B), lava, and pyroclastic and/or epiclastic rocks (Fig. 4E; Hibbard 1983). No positive evidence for pillow structures has been identified, but the mafic schists locally include small lenses of jasper and epidosite (Fig. 4D) and are interlayered with coticule (Fig. 4F), suggesting that the schists represent highly deformed and metamorphosed submarine flows and/or high-level sills. Detailed structural mapping by Kennedy (1971), Bursnall (1975) and Hibbard (1983) indicates that the contact between the Birchy Complex and sedimentary rocks of the Fleur de Lys Supergroup (i.e. Flat Point Formation) is generally conformable and shows little or no evidence for accommodating enhanced shear strain (Figs. 2, 3).

9 A narrow, discontinuous, linear belt of strongly metamorphosed Alpine-type ultramafic rocks (Bursnall 1975), interleaved with psammite and graphitic pelite and also associated with small lenses of amphibolite, occurs near the western boundary of the Rattling Brook Group in the northwestern part of the Baie Verte Peninsula (Fig. 2; Kennedy 1971; Hibbard 1983). These ultramafic rocks are associated with a major D₁ shear zone that was complexly refolded by tight to isoclinal F₂ folds (Kennedy 1971), which precludes kinematic analysis. This shear zone was previously referred to as the Bishie Cove slide (Fig. 2; Kennedy 1971), which we interpret as an early thrust that emplaced the Rattling Brook Group above correlative rocks in the Old House Cove Group (Figs. 2, 3; Kennedy 1971; Hibbard 1983). In addition to the presence of slivers of ultramafic tectonite, a thrust origin is consistent with the association of D₁ fabrics with local truncation of tectonostratigraphy in its footwall along strike (Hibbard 1983) and relatively high pressure (>10 kb) Taconic metamorphic assemblages (Kennedy 1971; Castonguay et al. 2010; Willner et al. 2012), suggesting that they formed during subduction. The ultra-mafic rocks adjacent to the Bishie Cove thrust commonly have been metamorphosed to soapstone, carbonate-bearing serpentinite and talc – tremolite – carbonate-bearing ultramafic schists. Graphitic schist hosts ultra-mafic rocks in both the Rattling Brook Group and Birchy Complex (Hibbard 1983), suggesting that they are correlative.

10 Correlatives of the Birchy Complex occur in the Mings Bight Group (e.g. Pelée Point schist; Fig. 1), which, like the Rattling Brook Group, is also dominated by meta-clastic rocks of the Fleur de Lys Supergroup (Hibbard 1983). The close spatial relationships between metasedimentary rocks of the Fleur de Lys Supergroup and ultramafic and mafic rocks of supposedly oceanic character in the Birchy Complex, the Mings Bight Group and Rattling Brook Group, led Hibbard (1988) to hypothesize that some of the sedimentary rocks in the Fleur de Lys Supergroups were deposited on oceanic lithosphere situated adjacent to the Humber margin; that is, the Fleur de Lys Supergroup overstepped the ocean – continent transition zone (Fig. 3). If a stratigraphic relationship between sedimentary rocks of the Fleur de Lys Supergroup (Flat Point Formation) and Birchy Complex is correct, the latter should represent ocean – continent transition lithosphere and/or juvenile oceanic lithosphere formed near this zone.

11 The contrasting association of ultramafic and sedimentary rocks and the highly dismembered tectonic character led to the suggestion that the Birchy Complex and correlatives represent zones of tectonic mélange that accommodated the initial stages of Early to Middle Ordovician obduction of the Baie Verte oceanic tract ophiolites onto the Humber margin (Bursnall 1975; Williams 1977; Hibbard et al. 1995). Although such a kinematic model is consistent with Middle Ordovician (Taconic, ca. 465 Ma) ⁴⁰Ar/³⁹Ar ages of muscovite and horn-blende in the Birchy Complex mafic schists (van Staal et al. 2009a; Castonguay et al. 2010), metamorphic studies reveal that the Birchy Complex was buried to significantly greater depths (≥10 kb) during the Taconic than the structurally overlying Baie Verte oceanic tract (≤7 kb; Willner et al. 2012). Their present tectonic juxta-position therefore took place subsequent to peak-Taconic burial metamorphism.

13 U–Pb ID–TIMS analyses were conducted at the Geochronology Laboratory at the Geological Survey of Canada, Ottawa. Heavy minerals were concentrated from the rock sample using standard crushing, grinding, and separation on a Wilfley table and by heavy liquid techniques. Mineral separates were sorted by magnetic susceptibility using a Frantz^TM isodynamic separator and zircon grains were hand-picked and grouped on the basis of crystal morphology and quality using a binocular microscope. All zircon fractions analyzed were strongly air abraded (Krogh 1982). U–Pb ID–TIMS techniques utilized in this study are modified after Parrish et al. (1987), with treatment of analytical errors following Roddick (1987). U–Pb ID–TIMS analytical results are presented in Table 1 and displayed in a concordia plot (Figure 5A). The concordia diagram was produced and concordia age was calculated, with decay-constant errors included, using Isoplot v. 3.00 (Ludwig 2003).

14 Zircons for the LA–ICPMS analyses were separated from the crushed samples by conventional means at the Department of Geology, Trinity College Dublin, Ireland. The sub-300 µm fraction was processed using a Gemeni (Rogers) mineral separation table, and then the heavy fraction was passed through a Frantz magnetic separator at 1 A. The non-paramagnetic portion was then placed in a filter funnel with di-iodomethane and the resulting heavy fraction passed again through the Frantz magnetic separator at full current. All zircons were then hand picked in ethanol using a binocular microscope, mounted in a 25 mm epoxy resin disk, and polished to reveal their grain interiors. The mounts were gold-coated and imaged using an FEI Quanta 400 SEM equipped with a solid-state, twin-segment BSE detector at the Micro-Analysis Facility at Memorial University, Newfoundland (MUN). A cathodoluminescence probe was used to image internal structures, over-growths and zonation (Fig. 6 A, B, C).

15 Isotopic data were obtained by LA–ICPMS at the MicroAnalysis Facility at Memorial University, Newfoundland and closely follow the procedures outlined in Pol-lock et al. (2009). Zircons were ablated in situ using a Lambda Physik COMPexPro 110 ArF excimer laser operating at a deep UV wavelength of 193 nm and a pulse width of 20 ns. A 10 µm laser beam was delivered to the sample surface and fired at a 10 Hz repetition rate using an energy density of 3 Jcm^-2. During ablation the sample was mounted in a sealed sample chamber and moved beneath the laser to produce a square 40 µm × 40 µm pit, to minimize the depth of ablation and reduce laser-induced elemental fractionation at the ablation site. The ablated sample was flushed from the sample cell and transported to the ICPMS system using a helium carrier gas (Q = 1.3 l/min), which reduces sample redeposition and elemental fractionation while increasing sensitivity for deep UV ablation. Mercury was filtered from the helium using gold-coated glass wool placed in the carrier gas line feeding the ablation cell. All analyses were performed by high-resolution ICPMS on a Finnigan Element XR system equipped with a dual-mode secondary electron multiplier operating in both counting and analogue modes. Data were collected using a 30 s measurement of the gas background before activation of the laser followed by 180 s of measurement with the laser on and zircon being ablated. The U and Pb isotopic ratios from the zircon were acquired along with a mixed ²⁰³Tl – ²⁰⁵Tl – ²⁰⁹Bi – ^233U–237 Np tracer solution (concentration of 10 ppb each) that was nebulized simultaneously with the ablated solid sample. Aspiration of the tracer solution allowed for a real-time instrument mass bias correction using the known isotopic ratios of the tracer solution measured while the sample was ablated; this technique is largely independent of matrix effects that can variably influence measured isotopic ratios and hence the resulting ages (Košler and Sylvester 2003).

16 Raw data for ²⁰⁷Pb/²⁰⁶Pb, ²⁰⁴Pb and ²³⁸U were reduced using the macro-based spreadsheet program LAM-DATE (Košler et al. 2008). The ²⁰⁷Pb/²⁰⁶Pb, ²⁰⁶Pb/²³⁸U and ²⁰⁷Pb/²³⁵U ratios were calculated and blank corrected for each analysis. Laser-induced U/Pb fractionation was typically less than 0.05% per a.m.u. based on repeat measurements of the ²⁰⁶Pb/²³⁸U ratio of the reference standards. This fractionation was corrected using the intercept method of Sylvester and Ghaderi (1997). For each analysis, time-resolved signals were inspected to ensure that only stable flat signal intervals were used in the age calculation. Measured ²⁰⁷Pb/²⁰⁶Pb ratios were not intercept-corrected; instead, the average ratio of the ablation interval selected for the age calculation was used. Analyses were rejected from the final dataset where the ²⁰⁷Pb/²⁰⁶Pb ratio calculated from the intercept-corrected ²⁰⁶Pb/²³⁸U and ²⁰⁷Pb/²³⁵U ratios did not fall within the 1σ uncertainty of the measured average ²⁰⁶Pb/²⁰⁷Pb ratio. Analyses that fell more than 5% above the ²⁰⁶Pb/²³⁸U–²⁰⁷Pb/²³⁵U concordia were also rejected. These two conservative filters ensured that any analyses that may have not been properly corrected for laser-induced U/Pb fractionation were eliminated from further consideration. High instrumental Hg backgrounds prohibited accurate measurement of ²⁰⁴Pb. Thus, in the few analyses where ²⁰⁴Pb was detected above background, the analysis was simply rejected from the dataset rather than attempting common Pb corrections.

17 Accuracy and reproducibility of U–Pb analyses in the MUN laboratory are routinely monitored by measurements of natural zircon standards of known U–Pb ID–TIMS age. To monitor the efficiency of mass bias and laser-induced fractionation corrections, standard reference materials 91500 zircon (1065 ± 3 Ma; Wieden-beck et al. 1995) and Plešovice zircon (337.13 ± 0.37 Ma; Sláma et al. 2008) were analysed in this study before and after every eight unknowns. Age determinations were calculated using the U decay constants and the present-day ²³⁸U/²³⁵U ratio of 137.88 of Jaffey et al. (1971). Final ages and concordia diagrams were produced using the Iso-plot/Ex macro (Ludwig 2003). Analytical data are listed in Table 2 and illustrated graphically in Fig. 6. The concordia ages for all analyses of 91500 and Plešovice zircon performed over the course of this study were 1066.7 ± 5.3 Ma (n = 77) and 337.7 ± 2.2 Ma (n = 52), respectively (95% confidence interval, with decay-constant errors included).

18 Metagabbro constitutes a significant component of the Birchy Complex mafic schists in Coachman’s Harbour (Fig. 4A). The strain is heterogeneous and low-strain pods locally preserve primary igneous subophitic textures that are pseudomorphed by green-schist to albite – amphibolite facies mineral assemblages (Fig. 5B). Two samples of metagabbro were collected in Coachman’s Harbour (Fig. 2). They comprise a relatively weakly strained, coarse-grained leucogabbro pod described in detail in Hibbard (1983, p. 49) (SNB-06-017; Fig. 5) and the enveloping schistose metagabbro (DC 09/08-30; Fig. 6). The low-strain leucogabbro (sample SNB-06-017, z9251) yielded abundant high quality zircon grains (100 to 200 µm) including delicate elongate crystals, prismatic grains, and flat, tabular fragments. Six single-grain and multigrain fractions were analyzed using U–Pb ID–TIMS. All six analyses overlap on concordia and each other (Table 1; Figure 4A) and yield a weighted average ²⁰⁶Pb/²³⁸U age of 558.0 ± 0.5 Ma (mean square weighted deviation (MSWD) = 0.73; probability of fit = 0.60). A concordia age, with decay-constant errors included, is calculated to be 558.3 ± 0.7 Ma (MSWD of concordance and equivalence = 1.5, probability = 0.14, n = 6). The age of 558.3 ± 0.7 Ma (Fig. 5A) is interpreted to be the crystallization age of the Birchy Complex leucogabbro.

19 The schistose metagabbro (DC 09/08-30) yielded a homogenous population of short prismatic zircons up to 100 μm. Cathodoluminescence imaging revealed homogenous grain interiors or oscillatory idiomorphic growth zoning (e.g. Fig. 6A) and common very thin (< 10 μm) low U rims. LA–ICPMS analysis of the cores of fourteen zircon grains yielded a concordia age of 563.9 ± 7.5 Ma (MSWD = 0.89; Fig. 6D), interpreted to represent the crystallization age of the metagabbro. This age is within analytical uncertainty of the ID–TIMS U–Pb zircon age.

20 The Birchy Complex mafic to intermediate schist was sampled along the coast ca. 1 km south of Coachman’s Harbour. The sampled locality is characterized by a thin, brown-weathering, tuffaceous schist of mafic to intermediate composition (DC09/08-41) cut by gabbroic sheets (Fig. 4E). The zircon population comprises small (between 75–150 μm in diameter), stubby prismatic zircons, with aspect ratios between 1.5 and 3.0. The grains typically exhibited only minor rounding, but were commonly fractured. Cathodoluminescence imaging revealed predominantly homogenous grain interiors (Fig. 6B) displaying local oscillatory idiomorphic growth zoning (e.g. Fig. 6C). LA–ICPMS analysis of 51 spots on separate grains yielded a concordia age of 556.3 ± 3.6 Ma and a weighted mean ²⁰⁶Pb/²³⁸U age of 556.4 ± 3.5 Ma (MSWD = 0.61; Fig. 6E, F). This is within analytical uncertainty of the ID–TIMS and LA–ICPMS U–Pb zircon ages of the intrusive metagabbros, suggesting that all mafic to intermediate schists of the Birchy Complex are consanguineous and have a Late Ediacaran age of ca. 558 Ma.

21 The Flat Point Formation was formerly included with the Rattling Brook Group, but following Kennedy (1971) is interpreted as a stratigraphic cover to the Birchy Complex. A psammite of the Flat Point Formation (DC 09/08-34) was sampled for U–Pb detrital zircon analysis. In general, most of the detrital zircons analyzed (Fig. 6G) yield Mesoproterozoic ages (1.0 – 1.5 Ga), and display a prominent Grenvillian peak typical of the Laurentian basement widely exposed in the Grenville province of southern Labrador and unconformably underlying parts of the Humber margin in western Newfoundland (Heaman et al. 2002; Gower et al. 2008). Smaller Paleoproterozoic (1.8 – 2.0 Ga) and Neoarchean (2.4 – 3.0 Ga) peaks are also consistent with a proximal Laurentian provenance. The youngest detrital zircon yielded a U–Pb concordia age of 990 ± 52 Ma.

22 Metamorphosed correlatives of the Fleur de Lys rocks in west-central Newfoundland (Hibbard 1988) yield a very similar Precambrian age distribution that corresponds closely to those measured by Cawood and Nem-chin (2001).

23 Mafic and ultramafic rocks prefaced by DC were analyzed for major oxides and trace elements by X-ray fluorescence (XRF) spectroscopy using a Phillips PW 1400 at the Centre d’Analyses Minérale, University of Lausanne, Switzerland (Table 3). Samples were fused with lithium borate and analysed for their major, trace element and rare-earth element concentrations by inductively coupled plasma optical emission spectrometry (ICPOES) and ICPMS (Thermo X-Series) at OMAC Laboratories, County Galway, Ireland. Where there are both ICP and XRF data for the same element, the ICP data are generally preferred, particularly for elements with low abundances such as U, Pb, Th, Ba, and the rare-earth elements (REEs) (Table 3). Samples prefaced by SNB were fused with lithium borate and analysed for their major, trace element and rare-earth element concentrations by ICPOES and ICPMS at Activation Laboratories in Ancaster, Ontario (Table 4).

24 A subset of samples was selected for Nd isotopic analysis utilizing a Thermo – Finnigan Triton T1 thermal ionization mass spectrometer at Carleton University, Ottawa, Ontario (Table 5). REE fractions were dissolved in 0.26N HCl and loaded onto Eichrom Ln Resin chromatographic columns containing Teflon powder coated with HDEHP (di(2-ethylhexyl) orthophosphoric acid; Richard et al. 1976). Nd was eluted using 0.26N HCl, followed by Sm in 0.5N HCl. Total procedural blanks for Nd are < 50 picograms, and < 6 picograms for Sm. Samples were spiked with a mixed ¹⁴⁸Nd–¹⁴⁹Sm spike prior to dissolution. Concentrations are precise to +/- 1%, while ¹⁴⁷Sm/¹⁴⁴Nd ratios are reproducible to 0.5%. Samples were loaded with H₃PO₄ on one side of a Re double filament, and run at temperatures of 1700–1800° C. Isotope ratios are normalized to ¹⁴⁶Nd/¹⁴⁴Nd = 0.72190. Analyses of the USGS standard BCR-1 yield Nd = 29.02 ppm, Sm = 6.68 ppm, and ¹⁴³Nd/¹⁴⁴Nd = 0.512668 +/- 20 (n=4). The international La Jolla standard yielded ¹⁴³Nd/¹⁴⁴Nd = 0.511847 +/- 7, n = 26 (February 2005 – June 2007). Internal lab Nd standard yielded 0.511819 +/- 10 n = 94 (February 2005 – August 2009) and 0.511823 +/-12 n = 65 (October 2010 – July 2012).

25 Hibbard (1983) determined that mafic rocks of the Birchy Complex are tholeiitic (Fig. 7A) and have a strong affinity with MORB. Although there is an apparent overlap with mafic rocks in the structurally overlying Baie Verte oceanic tract (BVOT), Hibbard (1983) observed that the Birchy Complex greenschists are commonly slightly enriched in TiO₂ compared to the adjacent BVOT rocks of the Advocate Complex. Overall, analysis of the Birchy Complex mafic rocks confirms the observations of Hibbard (1983). Two geochemical suites of mafic rocks can be readily identified on the basis of major and trace element data. All mafic rocks have relatively flat rare-earth element patterns on a MORB-normalised diagram, but the first suite (metagabbro) is characterized by lower TiO₂, Zr depletion and Eu enrichment on a MORB-normalized profile (Fig. 7B). The second suite, consisting of typically fine-grained mafic rocks interpreted as metabasalt, is characterized by higher TiO₂, small to negligible Nb and La anomalies, and enrichment of Th relative to Nb (Fig. 7B). Both suites plot in the MORB field adjacent to the field of backarc basin basalt on a La–Y–Nb tectonic discrimination diagram (Fig. 7C). Sm–Nd isotopic analyses yielded εNd values of +7.4 and +7.2 respectively (Fig. 8).

26 A sample of intermediate tuffaceous schist plots in the dacite – rhyolite field on a Zr/Ti vs. Nb/Y diagram (Fig. 7A). The sample is characterized by slight light REE enrichment, positive Zr and Hf anomalies, and negative Eu and Ti anomalies (Fig. 7B). Similar to the mafic rocks, the intermediate tuff lacks prominent La and Nb anomalies but has a strong Th enrichment. It plots in the ocean ridge granite field on granitoid tectonic discrimination plots of Pearce et al. (1984; not shown). Sm–Nd isotopic analysis of the tuff yielded εNd value of +7.5 (Fig. 8).

27 The ultramafic rocks in the Birchy Complex and Rattling Brook Group are generally highly metamorphosed and strongly metasomatised to serpentinite, listwanite, soapstone and actinolite – fuchsite schist. The sampled serpentinites and actinolite – fuchsite schist locally preserve relict igneous minerals; soapstone, listwanite and carbonate – tremolite bearing ultramafic schists were not sampled due to extensive metasomatism. The pervasive degree of alteration of the ultramafic samples complicates interpretation of primary lithologies; however, Niu (2004) demonstrated that the bulk compositions of pervasively hydrated peridotites can preserve primary magmatic signatures. Normative calculations (not shown) suggest that the protoliths to the serpentinites were harzburgite and dunite, consistent with presence of relict orthopyroxene in some samples. Presence of minor normative clinopyroxene may suggest that some of the samples may be lherzolitic; however, this can also be attributed to introduction of silica and minor hydrothermal calcium during serpentinization. The actinolite – fuchsite schists are petrographically distinct from serpentinites and most likely represent hydrated websterite and/or pyroxenite.

28 Trace element abundances in harzburgites and websterites are very low and commonly near or below the detection limit for the analytical techniques used in this study, resulting in missing elements (below detection limit) or jagged patterns (near detection limit; Fig. 7D). All ultramafic rocks are characterized by depleted heavy REE relative to primitive mantle (Fig. 7D). All samples display strong light REE enrichment, similar to metasomatized mantle of the BVOT, but distinct from abyssal and forearc peridotites (Fig. 7D). Most analysed serpentinites probably represent altered metasomatised mantle. However, negative Eu and Ti anomalies for some samples (SNB 08 004A1 and 006A1) indicate that these may represent lower crustal cumulate.

33 Hyperextension is generally, although not exclusively, a characteristic of magma-poor margins. However, viewing margins solely on the basis of end member models is generally inappropriate, because the degree and nature of magmatism associated with hyper-extension varies (e.g. Müntener and Manatschal 2006; Bernoulli et al. 2003) or may change over time from magma-poor to magma-rich (e.g. Osmundsen and Ebbing 2008). Hyperextended margins are commonly characterised by extreme thinning of parts of the continental crust (distal margin) as a result of superimposition of different modes of extension, culminating in exhumation of lowermost crust and/or serpentinized continental mantle onto the seafloor (e.g. Iberian margin; Tucholke et al. 2007; Sibuet and Tucholke 2012) and the formation of various types of thinned crustal blocks (Péron-Pinvidic and Manatschal 2010). The latter may include isolated extensional crustal allochthons riding on a concave-downwards lithosphere-scale master detachment that exhumed mantle onto the seafloor (e.g. Manatschal 2004; Manatschal et al. 2007, 2011; Sutra and Manatschal 2012). This final asymmetric phase of extension may be superimposed on an earlier phase of more symmetric extension (Huismans and Beaumont 2002) during which mantle detachments can form below both extending margins (e.g. Weinberg et al. 2007)

34 The exhumed, inherited mantle may be modified and refertilized by percolating melts associated with syn-rift magmatism (Müntener et al. 2009). In general, syn-rift magmatism is subdued and is represented by mafic intrusions of MORB-like composition; volcanic rocks commonly form a minor component of the mafic magmatism (but see Bernoulli et al. 2003). These mafic intrusions record the onset of magmatic accretion within the OCT zone during distributed (delocalized) extension-related deformation, which commonly continues for a significant length of time (Jagoutz et al. 2007; Péron-Pinvidic and Manatschal 2009), until the onset of true seafloor spreading and the formation of oceanic crust.

35 The inferred presence in the Birchy Complex of depleted mantle rocks such as harzburgite is rare in OCT assemblages, which are generally dominated by serpentinized lherzolite. Harzburgitic mantle was locally exhumed during the Cretaceous along the Atlantic margin of Newfoundland as a result of hyperextension of the Newfoundland – Iberian sector during opening of the Atlantic Ocean. This harzburgite was interpreted as a slice of supra-subduction zone mantle inherited from a pre-Mesozoic period of subduction and melt extraction (Müntener and Manatschal 2006). Exhumation of such refractory supra-subduction zone mantle may be a mechanism that suppresses formation of syn-rift basaltic melts. Likewise, the harzburgite and dunite preserved in the Rattling Brook Group and Birchy Complex may be mantle and lower crustal cumulates of a pre-Iapetus opening phase of subduction (Grenville?). This is consistent with the trace element characteristics of the Birchy Complex mafic rocks, which exhibit MORB to backarc basin-like characteristics (Fig. 7B, C).These characteristics have been demonstrated to result from melt percolation though subduction-zone modified mantle in modern settings (e.g. Taylor 1992). Cretaceous rift-related basalts of the hyperextended Atlantic margin in off-shore Newfoundland have similar characteristics (Fig. 7B), suggesting that they were derived from pre-Atlantic, Iapetan subduction-zone modified mantle (Robertson 2007). We infer a similar mechanism for the Birchy Complex suite 2 metabasites, whereby extension along the Laurentian margin leads to decompression melting of previously metasomatized mantle similar to our ultramafic rocks (Fig. 7D). Suite 1 metabasites (Fig. 7B, C) are distinctly more depleted in Th and light REE than the suite 2 metabasites and may be analogous to off-axis magmas (Reynolds and Langmuir 2000).

36 In addition to the ultramafic rocks in the Birchy Complex and Rattling Brook Group discussed herein, narrow slices of mantle interleaved with strongly tectonized metasedimentary rocks occur elsewhere in Newfoundland (e.g. Matthews Brook serpentinite; Cawood et al. 1996), Québec (e.g. Pennington sheet serpentinite; St. Julien 1987), and Vermont (Doolan et al. 1982), suggesting that hyperextension may have been a common process along several segments of Laurentia’s Appalachian margin. There is also a close similarity between the Birchy Complex and parts of the Rattling Brook Group in Newfoundland with the upper parts (Easdale Subgroup) of the Dalradian Supergroup in western Ireland, suggesting that they occupied a similar and correlative tectonic setting (Winchester et al. 1992), directly linking the Laurentian realm of the Newfoundland Appalachians to the British Caledonides. At this stratigraphic level, best seen on southern Achill Island, serpentinite olistoliths embedded in a graphitic pelite matrix are common (Kennedy 1980; Chew 2001). The serpentinite bodies are associated with mafic volcanic rocks, deep-marine continentally derived psammitic wacke (Fig. 8), and graphitic pelite in a sequence that underwent Taconic – Grampian blueschist-facies metamorphism (Chew et al. 2003). Blueschist facies metamorphism is also preserved in correlative OCT rocks in northern Vermont (Doolan et al. 1982; Castonguay et al. 2012), emphasizing the link between high-pressure metamorphism and OCT assemblages established in the Alps (Beltrando et al. 2010) and in the Taconic – Grampian orogen of the Appalachian – Caledonian mountain belt. A discontinuous horizon of serpentinite bodies has also been documented in the Easdale Subgroup of central and northeastern Scotland (Garson and Plant 1973). The serpentinite bodies in Ireland and Scotland have been interpreted as seafloor protrusions of serpentinized mantle that were generated in Easdale Subgroup time during a phase of major crustal extension leading to the formation of an OCT (Chew 2001).

37 Evidence for Ediacaran – Early Cambrian hyperextension along segments of the Laurentian margin during opening of the Iapetus Ocean has major ramifications for understanding the evolution of the Appalachian – Caledonian margin of Laurentia in general. For example, where the sedimentary cover of rifted margins is very thick, such as the Dalradian Supergroup of the Laurentian margin in the British and Irish Caledonides (e.g. Chew 2001; Leslie et al. 2008) , they may form a thermally insulating blanket; the underlying crust may therefore heat up and become rheologically weaker during rifting (Reston and Manatschal 2011). Hyperextension and the resultant formation of crustal ribbons, ‘hanging wall’ (H-) blocks, and extensional crustal allochthons (for definitions and characteristics of these various types of crustal blocks see Péron-Pinvidic and Manatschal 2010) are capable of explaining several puzzling and/or problematic phenomena. These include: 1) the apparent late age (525–520 Ma) of the oldest known drift sequences (e.g. Cawood et al. 2001); 2) the formation of microcontinents (e.g. Dashwoods) along the Newfoundland Humber margin and the Appalachian – Caledonian margin in general (van Staal et al. 2007, 2009b; Chew et al. 2010); 3) the markedly variable and spotty preservation of radiogenic age evidence for Taconic-related metamorphism and deformation along strike (Cawood et al. 1994; Castonguay et al. 2001, 2010 ; van Staal et al. 2009 a, b); and 4) evidence for crustal contamination in felsic rocks of the BVOT and its Early to Middle Ordovician Snooks Arm Group cover (Skulski et al. 2010) and other out-board terranes in the peri-Laurentian realm (e.g. Whalen et al. 1997; van Staal et al. 2007; Zagorevski et al. 2006)

38 The highly thinned continental margins of studied OCT zones commonly display significant retardation of thermal subsidence. In addition to the potentially insulating effects of a thick sedimentary blanket and the predicted slow cooling of the upper plate crust during asymmetric extension (Buck et al. 1988), anomalous slow cooling and prolonged uplift of a rifted margin may also be related to the structural emplacement of hot mantle under the thinning crust (Müntener et al. 2009; Péron-Pinvidic and Manatschal 2009). Hence, a combination of these processes may have significantly delayed thermal subsidence and the resultant transgression.

39 The time of transgression was used previously as a proxy for defining the Iapetus rift-drift transition (Bond 1984; Williams and Hiscock 1987; Cawood et al. 2001), but only provides a minimum age for break-up along this segment and probably elsewhere as well (cf. Hibbard et al. 2007). Furthermore, the work of Jagoutz et al. (2007) along the Iberian OCT has shown that formation of MORB along the proto-ridge in embryonic oceanic crust may be followed by widespread, delocalized extensional deformation and off-axis magmatism before true spreading occurs. Therefore, the Late Neoproterozoic (565–550 Ma) eruption of MORB, backarc basin basalt (Bédard and Stevenson 1999; this paper) and other within-plate magmatism (Cawood et al. 2001; Hodych and Cox 2007) in the Newfoundland and Québec Appalachians merely places an upper limit on the onset of true seafloor spreading in this segment of the Iapetus Ocean. Spreading followed a long period (615–550 Ma) of pre-spreading, non-voluminous rift-related magmatism, exhumation of mantle onto the seafloor, and formation of embryonic (cf. Jagoutz et al. 2007) oceanic crust between 565–550 Ma (Fig. 9A). Hence, the end of rift-related magmatism (550 Ma) is the best proxy for the final breakup and onset of spreading in the Laurentian realm of the Iapetus Ocean (Fig. 9B).

40 Microcontinents such as Dashwoods and equivalents elsewhere (e.g. Karabinos et al. 1998; Hibbard et al. 2007; van Staal and Hatcher 2010; Allen et al. 2010; Chew et al. 2008, 2010) initially could have formed, at least to a first approximation, as H-blocks or large extensional allochthons, analogous to the Brian H onnais crustal block in the Alps (Manatschal et al. 2006; Mohn et al. 2010). The crust of H-blocks is commonly thinned to less than 20 km and is generally associated with marked retardation of subsidence (Péron-Pinvidic and Manatschal 2010). The characteristics of H-blocks fit the existing three dimensional seismic and petro-logical constraints on Dashwoods rather well (van der Velden et al. 2004; van Staal et al. 2007). Hence, we interpret Dashwoods as a microcontinent that evolved from an H-block (Fig. 9 A, B) rather than an extensional allochthon. Isotopic evidence of crustal contamination and zircon inheritance in the outboard peri-Laurentian terranes (e.g. Karabinos et al. 1998; Whalen et al. 1997; Zagorevski et al. 2006; Brem et al. 2007; Hibbard et al. 2007; van Staal et al. 2007; van Staal and Hatcher 2010; Allen et al. 2010; Chew et al. 2008, 2010; Skulski et al. 2010, Zagorevski and van Staal 2011) indicate that these isolated blocks subsequently formed the basement to supra-subduction zone magmatism in the outboard arc complexes. Formation of H-blocks and/or large extensional allochthons during hyperextension would remove the necessity of having two discrete rifting events along the Humber margin, resulting in two coeval spreading centres (Cawood et al. 2001; Waldron and van Staal 2001; Burton and Southworth 2010). This is analogous to the Brian H onnais crustal block in the Alps (Manatschal et al. 2006; Mohn et al. 2010), which separates the Valais oceanic basin to the north from the Piemonte – Liguria Ocean to the south. In this analogy (Fig. 9B), the Piemonte – Liguria Ocean would represent the Iapetus Ocean and the Valais basin the Taconic seaway. However, the generation of upper plate magmatism (489–477 Ma Notre Dame arc) in Dashwoods during the closure of the Taconic seaway (van Staal et al. 2007), demands that parts of the Taconic seaway had achieved a width large enough to generate arc magmatism during its subduction beneath Dashwoods, but not so wide as to prevent exchange of Laurentian faunas. These boundary conditions suggest that the ca. 300 to1000 km wide (van Staal et al. 1998, 2007) Taconic seaway, in contrast to the Valais basin in the Alps, saw a short period (550–540 Ma) of spreading. This spreading was likely delocalized, ultra-slow and merely forming embryonic oceanic crust (Jagoutz et al. 2007); hence, the Taconic seaway was probably partially underlain by exhumed mantle and partly by oceanic litho-sphere. Regardless of whether rifting and spreading was localized or delocalized, we infer that any spreading in the Taconic seaway was aborted shortly after it had started and that the dominant magmatic spreading centre formed further outboard in what would become the Iapetus Ocean (Fig. 9B), leading to separation of Dash-woods from Arequipa – Antofalla, its inferred conjugate partner (Escayola et al. 2011)

41 Other extension-related continental blocks surrounded by exhumed and serpentinized mantle, situated between Dashwoods and the autochthonous Humber margin (Fig. 9B), may explain preservation of evidence for pervasive Taconic tectono-metamorphism in these rocks compared to its apparent non-preservation in other, more inboard parts (see Cawood et al. 1994 and van Staal et al. 2009 a, b). We propose that an extensional allochthon originally formed the basement of the Rattling Brook Group east of the Bishie Cove thrust (Figs. 2, 9B, 9C). The Rattling Brook allochthon would have been subducted (abortively) beneath the BVOT and Dashwoods before arrival of the leading edge of the autochthonous Humber margin and its collision with the Notre Dame arc (van Staal et al. 2007). It follows that, because of its buoyancy, it could have returned to higher crustal levels along the subduction channel, together with the adjacent OCT zone preserved in the Birchy Complex (Fig. 9C). This may have occurred during or after the final Taconic subduction of the OCT lithosphere situated between the Rattling Brook extensional allochthon and the leading edge of the autochthonous Humber margin sitting further inboard (Fig. 9C). Such a process could have translated the Birchy Complex and spatially associated rocks to a high structural level during the Taconic orogeny (470–460 Ma), such that it remained below the Ar-closure temperature of white mica during the subsequent Salinic orogenic overprint (Cawood et al. 1994) and therefore preserved its Taconic argon ages. Although ca. 464 Ma metamorphic zircon in retrogressed eclogite pods leaves little doubt that the autochthonous Humber margin was subjected to Taconic burial as well (van Staal et al. 2009 a, b), Taconic argon ages are typically not preserved in the adjacent rocks (Hibbard, 1983). This is likely because the autochthonous Humber margin was in many places subjected to significant Silurian (Salinic) tectonic burial and resetting. In this model, the serpentinites along the Bishie Cove thrust would thus define a suture between paraauthochthonous Humber margin rocks and allochthonous OCT lithosphere attached to the Rattling Brook allochthon positioned further outboard (Fig. 9C). Chew et al. (2010) proposed a rather similar model for the Grampian in the west of Ireland. In addition, slow but relatively steep subduction of the OCT lithosphere (partly serpentinized mantle?) between the Humber margin and the Rattling Brook allochthon, beneath the BVOT, could be responsible for generating the arc – backarc-like magmatism preserved in the Flat Water Pond/Snooks Arm Group, which disconformably covers the BVOT (Skulski et al. 2010).