Introduction

A-type granites are classified on the basis of geochemical characteristics such as higher SiO2, FeO/FeO+MgO, Ga/Al, Zr, Nb, Y and Ta than either I-type or S-type granites (Loiselle and Wones 1979; Collins et al. 1982; Whalen et al. 1987; Frost et al. 2001). Their tectonic setting and petrogenesis are controversial. Although their high Nb, Y and Ta means that they plot as within-plate granites on tectonic discrimination diagrams (e.g. Pearce et al. 1984), it is evident that they occur in a wide variety of tectonic settings (Whalen et al. 1987; Bonin 2007).

On the one hand, A-type granites are thought to be the result of anhydrous partial melting of a halogen-enriched dry, granulitic lower crustal residue formed after extraction of a felsic magma (Collins et al. 1982; Clemens et al. 1986; Whalen et al. 1987; Landenberger and Collins 1996). The heat required to induce partial melting may arise from rising mantle-derived mafic magma during lithospheric extension. Similarly, Creaser et al. (1991) proposed that some A-type magmas may be derived from melting of tonalitic to granodioritic source rocks. However, Bonin (2007) maintained that leucosomes of A-type composition have not been documented in migmatitic terranes and that no experiments have produced A-type magmas by partial melting of the crust. Instead, Bonin (2007) proposed that they are derived by fractionation either of alkaline mafic magmas or from hybrids produced by mixing between coeval mafic and felsic magmas. Martin (2006) claimed that A-type granites were formed in a source region fertilized by hydrothermal activity above a rising mantle plume and that such events may also be connected to the formation nepheline syenite and carbonatite.

Eby (1990, 1992) showed that granitic magmas classified as A-type by their high Ga/Al also have wide ranges in Nb/Y, a ratio which is often used as a proxy for the degree of alkalinity (Winchester and Floyd 1977; Pearce 1996). On this basis, Eby (1990, 1992) defined two sub-types: (1) A1 which has high Nb/Y (~0.8), may be derived from fractionation of alkalic mafic magma derived from enriched mantle sources, and is emplaced in continental rifts or intra-plate settings; and (2) A2 which has lower Nb/Y, represents magmas derived from continental crust or underplated crust that has been through a previous cycle of magmatism and commonly occurs in a late to post-tectonic environment at the end of a long period of magmatism.

Most petrological studies of A-type magmas are confined to individual suites with a limited range in age. In this paper, we propose that important insights into the genesis of these magmas and their sources can be gained by focusing on terranes where such magmatism occurred in repeated episodes over a long period of time. Four episodes of A-type magmatism have been recognized in Avalonia of northern mainland Nova Scotia, between the late Neoproterozoic and the Late Devonian and the geochemistry of each of these suites has been described individually (Murphy et al.1996a, 2012; Piper et al. 1999; Archibald et al. 2013; Escarraga et al. 2012). Here we show that these suites, when considered collectively, imply that A-type magmas can be generated in different tectonic environments and that they display geochemical and isotopic patterns indicating a remarkable degree of inheritance from an older (Neoproterozoic) crustal source. We propose that when the crustal source becomes dehydrated, a terrane will repeatedly produce A-type magmas irrespective of the tectonic environment. More generally, this study illustrates the importance of comparing the geochemical and isotopic characteristics of igneous suites with different ages in the same terrane in order to constrain the extent to which their signatures are inherited from a common source.

Regional tectonic setting

Regional tectono-stratigraphic comparisons, together with faunal and palaeomagnetic data, suggest that Avalonia is one of several peri-Gondwanan terranes, that were located along the northern periphery of Gondwana in the Late Neoproterozoic–Early Cambrian (Fig. 1). Five important phases of magmatism have been identified in Avalonia (Fig. 2). The oldest phase (760–640 Ma) produced “proto-Avalonia” as a series of oceanic arcs, some possibly underlain by thin slivers of Baltica crust, that formed within the peri-Rodinian (Mirovoi) Ocean and accreted to the northern margin of Gondwana between 660 and 640 Ma (Murphy et al. 2000, 2008; Henderson et al. 2016).

Fig. 1
figure 1

Late Neoproterozoic to Late Devonian reconstructions showing the relative positions of Gondwana, Laurentia, Baltica, and the drift of Avalonia and related peri-Gondwanan terranes (A) from Gondwana to Laurentia. Reconstructions based on Cocks and Torsvik (2002), and Murphy et al. (2006)

Full size image
Fig. 2
figure 2

Generalized tectonostratigraphy of the Antigonish Highlands (modified from Murphy et al. 1991; Murphy and Nance 2002) and Cobequid Highlands (modified from Pe-Piper and Piper 2002). For detailed correlations between Antigonish and Cobequid highlands, see Murphy et al. (1992)

Full size image

After accretion, subduction zones stepped outboard, producing the second and main tectonic phase (640–570 Ma) of arc-related magmatism and basin formation that characterizes the Avalonia and related peri-Gondwanan terranes (Pe-Piper and Piper 1989; Murphy et al. 1990). Arc magmatism was terminated diachronously between 600 and 540 Ma by the propagation of a San Andreas-style transform fault (Murphy and Nance 1989; Nance et al. 1991), producing the third phase of magmatism during an arc to platform transition, i.e. after the cessation of arc magmatism, but before deposition of Cambrian–Ordovician platformal successions. This strike- slip activity was accompanied by localized intra-plate magmatism and coeval basin development, possibly analogous to the Basin and Range setting in the western United States (e.g. Nance et al. 2002). As the termination of arc magmatism was not due to continental collisions, Cambrian platformal successions were located along the northern margin of Gondwana facing the Mirovoi Ocean.

A combination of paleomagnetic, geochronological and geochemical data indicate that Avalonia was one of several peri-Gondwanan terranes that drifted from the Gondwanan margin in the Late Cambrian or Early Ordovician as the Rheic Ocean opened (Cocks and Torsvik 2002; Stampfli and Borel 2002; Landing 2005; van Staal et al. 2009, 2012; Fig. 1). The next phase of magmatism began with the separation of Avalonia from Gondwana which occurred by ca. 485 Ma (Prigmore et al. 1997; Sanchez-Garcia et al. 2003; Keppie et al. 2006; Murphy et al. 2006). By 460 Ma, Avalonia was about 2000 km north of the Gondwanan margin (Hamilton and Murphy 2004; Cocks and Torsvik 2002) and defined both the northern margin of the Rheic Ocean and the southern flank of the Iapetus Ocean. Magmatism in this time interval is attributed to ensialic arc activity, analogous to northern New Zealand (Murphy et al. 2012).

Avalonia collided with Baltica by the early Silurian and with Laurentia by the early Devonian (e.g. Keppie 1985; 1989; Murphy et al. 1996b; van Staal et al. 1998, 2009, 2012). Between the Late Silurian and Late Carboniferous, Avalonia was located along the northern margin of a contracting Rheic Ocean, the closure of which eventually resulted in the amalgamation of Pangea (Fig. 1; Murphy et al. 2006, 2008). By the middle Devonian, the Meguma terrane was located outboard of Avalonia, and Late Devonian–Early Carboniferous tectonic evolution was dominated by intra-continental dextral transcurrent motion along the boundary between these two terranes, producing the youngest phase of magmatism which is predominantly bimodal (Keppie 1982; Pe-Piper and Piper 1998; Murphy et al. 2011).

Magmatism in northern mainland Nova Scotia

Avalonian rocks in northern mainland Nova Scotia are exposed in the Antigonish Highlands and the Cobequid Highlands (Fig. 3). Geological syntheses indicate that the geology of these highland regions was contiguous from the Late Neoproterozoic until the Late Carboniferous when the Stellarton Basin formed between them (e.g. Murphy et al. 1992; Waldron 2004).

Fig. 3
figure 3

a Generalized geological map of northern mainland Nova Scotia, b geological map of the Antigonish highlands (simplified from Murphy et al. 1991; White et al. 2011), c geological map of the Cobequid highlands (simplified from Pe-Piper and Piper 2002)

Full size image

Neoproterozoic arc magmas The oldest rocks occur are a suite of ca. 750–730 Ma granitic to tonalitic gneisses in the eastern Cobequid highlands (Doig et al. 1993; MacHattie et al. 2014). These rocks are interpreted as local representatives of proto-Avalonia that formed outboard of Gondwana within the peri-Rodinian ocean (e.g. Murphy et al. 2000; Henderson et al. 2016).

Both highland regions are predominantly underlain by arc-related ca. 630–608 Ma low-grade volcanic and sedimentary rocks of the Georgeville (Antigonish) and Jeffers (Cobequids) groups that are intruded by ca. 607–600 Ma syn- to post-kinematic mafic to felsic plutonic complexes (Donohoe and Wallace 1982; Keppie et al. 1990; Murphy et al. 1990, 1991, 1996a; Pe-Piper and Piper 1989, 2002; Pe-Piper et al. 1996; MacHattie and White 2015). These igneous rocks are typical of Late Neoproterozoic voluminous arc magmatism that characterizes Avalonia. In both localities, felsic volcanic rocks are interbedded calc-alkalic and tholeiitic mafic to intermediate flows as well as turbidites, thought to reflect local extension and basin generation within the arc.

Published geochemical data from the older arc suite (the ca. 734 ± 2 Ma Economy River orthogneiss; Doig et al. 1993) indicate the rocks are intermediate to felsic in composition, with SiO2 contents ranging from 59.0 to 65.0 wt%, and Fe2O3/MgO from 1.5 to 2.6. The Late Neoproterozoic felsic rocks are more silicic with SiO2 ranging from 62.1 to 76.2 wt% and Fe2O3/MgO from 1.8 to 2.5. These are values typical of calc-alkalic intermediate to felsic rocks. The Economy River and Late Neoproterozoic sequences are both characterized by moderate enrichment in large-ion lithophile elements (K, Rb, Ba, and Th) relative to high-field-strength elements such as Nb, Zr and Y, a signature typical of volcanic arc environments (Pearce et al. 1984). In comparison to typical A-type granites, all these sequences contain significantly lower abundances of Nb (<20 ppm), Y (<26 ppm), Ga (<20 ppm) and low ΣREE. Sm–Nd isotopic data for these sequences are typical of Avalonian crust, with TDM model ages ranging from 0.8 to 1.2 Ga (Nance and Murphy 1996; Murphy et al. 2000, 2008) and ε tNd values ranging from +1.29 to +4.09 (t = 734 Ma) for the Economy River Gneiss and from +0.1 to +3.0 (t = 610 Ma) for the Late Neoproterozoic sequences. These Late Neoproterozoic sequences are the youngest phase of calc-alkalic magmatism to have affected Avalonia in northern mainland Nova Scotia.

Ediacaran The oldest rocks in this region with an A-type chemistry belong to the ca. 580 Ma post-tectonic Georgeville Pluton, which intrudes the lower greenschist to sub-greenschist Georgeville Group in the northernmost Antigonish Highlands adjacent to the Hollow Fault (Murphy et al. 1991, 1998). This granite is characterized by high SiO2, Th, Nb, Y, Zr, Ta and Ga/Al; very low CaO, TiO2, MgO, FeO and MnO and by positively sloped rare earth element (REE) profiles generated by extreme light REE depletion (Figs. 4, 5, 6, 7). Nb-Ga-Th and Ce abundances straddle the divide between A1 and A2-type granites.

Fig. 4
figure 4

Zr/TiO2 vs Nb/Y discrimination diagram (Winchester and Floyd 1977; Pearce 1996) for the A-type felsic magmas in northern Nova Scotia. The high Zr/TiO2 values for the Georgeville Granite reflect the very low TiO2 in the rocks. Geochemical data for the Georgeville Granite from Murphy et al. (1998), West Barneys River Suite from Archibald et al. (2013), the Dunn Point and McGillivray Brook Formations from Murphy et al. (2012), the Fountain Lake Group from Piper et al. (1999)

Full size image
Fig. 5
figure 5

Nb v Y discrimination diagram (after Pearce et al. 1984) for the A-type felsic magmas in northern Nova Scotia. Geochemical data for the Georgeville Granite from Murphy et al. (1998), West Barneys River Suite from Archibald et al. (2013), the Dunn Point and McGillivray Brook Formations from Murphy et al. (2012), the Fountain Lake Group from Piper et al. (1999). The apparent “within-plate” signature is due to the high Nb and Y concentrations, typical of A-type magmas

Full size image
Fig. 6
figure 6

Zr vs Ga/Al 104 (after Whalen et al. 1987) discrimination diagram for the A-type felsic magmas in northern Nova Scotia. Geochemical data for the Georgeville Granite from Murphy et al. (1998), West Barneys River Suite from Archibald et al. (2013), the Dunn Point and McGillivray Brook Formations from Murphy et al. (2012), the Fountain Lake Group from Piper et al. (1999)

Full size image
Fig. 7
figure 7

Representative chondrite-normalized rare earth element abundances for the A-type felsic magmas in northern Nova Scotia. Geochemical data for the Georgeville Granite from Murphy et al. (1998), West Barneys River Suite from Archibald et al. (2013), the Dunn Point and McGillivray Brook Formations from Murphy et al. (2012), the Fountain Lake Group from Piper et al. (1999)

Full size image

Compared to both older and younger felsic rocks, ε tNd values of the Georgeville pluton, are significantly lower (ranging from −1.0 to −7.25, Fig. 8), whereas Sm/Nd (0.6–1.5) and 147Sm/144Nd (0.3–0.6) ratios are significantly higher. In addition, calculated depleted mantle (TDM) model ages (−266 to + 231 Ma) are much younger than the emplacement age, and, therefore, have no geological meaning. Anderson et al. (2008) attributed both the anomalous LREE depletion and low Nd isotopic values of the granite to subsolidus re-equilibration and hydrothermal alteration of the granite, including the metamict zircon.

Fig. 8
figure 8

Compilation of Sm–Nd isotopic data for the Ediacaran to Carboniferous A-type felsic magmas in northern Nova Scotia and a summary of the data from older arc magmas. Data for the Georgeville Granite from Murphy et al. (1998), West Barneys River Plutonic Suite from Archibald et al. (2013), the Dunn Point and McGillivray Brook Formations from Murphy et al. (2012), the Fountain Lake Group from Piper et al. (1999). The envelope for Avalonian crust is from Nance and Murphy (1994, 1996) and Murphy et al. (2000). Range of Sm-Nd isotopic data in Neoproterozoic arc sequences: GG Georgeville Group, ERG Economy River Gneiss

Full size image

Early Ordovician Voluminous A-type magmas occur at various times in the Ordovician in the Antigonish Highlands. An early Ordovician plutonic complex has recently been documented in the Cobequid Highlands (MacHattie et al. 2014), but published geochemical data are unavailable. In the central and southern Antigonish Highlands, A-type granitoid magmas, with zircon U–Pb age dates ranging from 495 Ma to 465 Ma, are part of the bimodal West Barneys River Plutonic Suite, a shallow crustal (sub-greenschist) composite intrusive body that also includes several plutons, many of which contain coeval alkalic gabbros (Escarraga et al. 2012; Archibald et al. 2013). These studies show that the granitoids are relatively enriched in Na2O, K2O, FeOt, Ga/Al, Zr, Y and REE, low in Al2O3 and CaO (Fig. 4, 5, 6, 7) and are classified as ferroan A-type granitoids as described by Frost et al. (2001) and Frost and Frost (2011). The granite samples have similar ε tNd (t = 470 Ma) values ranging from +2.4 to +4.9 (Fig. 8), with T DM model ages ranging from 0.8 and 1.1 Ga. The geochemistry of the coeval gabbroic rocks is transitional from alkalic to tholeiitic and is typical of a within-plate setting (Archibald et al. 2013) and their ε tNd (t = 470 Ma) values ranging from +0.8 to +4.7 with TDM model ages between 0.7 and 1.1 Ga.

In the northern Antigonish Highlands, a sample from the felsic volcanic rocks of the bimodal Dunn Point Formation (DPF) yields a zircon U/Pb age of 460 ± 3.4 Ma (Hamilton and Murphy 2004). Mafic volcanic rocks in the formation have continental tholeiitic, within-plate affinities (Keppie et al. 1979; Murphy and Dostal 2007; Murphy et al. 2012). A sample from the conformably overlying McGillivray Brook Formation (MBF) yields a 454 ± 0.7 Ma age (Murphy et al. 2012). The MBF is dominated by felsic tuffs and flows. Although their chemistry is quite distinct from one another, the felsic rocks of the DPF and MBF both resemble A-type (alkalic) SiO2-rich magmas (Murphy et al. 2012). The MBF rocks are more highly fractionated than DPF felsic rocks and have higher SiO2 (76.0–81.8 wt%), lower Al2O3, and very low MgO, CaO, TiO2 and P2O5. They are also characterized by significantly higher concentrations of Zr (745–1965 ppm), Y (65–213 ppm), Nb (57–185 ppm), Ta (2.1–6.0 ppm), Yb (5.6–16. 7 ppm) and Ga/Al (Figs. 4, 5, 6, 7).

ε tNd values for Dunn Point rhyolites range from +2.9 to +3.7 (t = 460 Ma) and overlap with McGillivray Brook Formation felsic rocks (t = 455 Ma) which range from +1.5 to + 3.9 (Fig. 8). The DPF analyses, however, have a much lower range in 147Sm/144Nd (0.128 to 0.148) compared to the MBF (0.140 to 0.348). TDM ages vary from 0.9 to 1.2 Ga for DPF and 0.9 to 1.1 Ga for MBF.

Late Devonian–Early Carboniferous A-type silicic rocks in the Cobequid Highlands include voluminous Late Famennian to Early Tournaisean bimodal volcanic (Fountain Lake Group) and coeval shallow crustal plutonic rocks which are predominantly restricted to the area north of the Rockland Brook Fault (Donohoe and Wallace 1982; Pe-Piper and Piper 2002; MacHattie et al. 2014). Zircon U/Pb (TIMS) ages range from ca. 365 to 354 Ma (Dunning et al. 2002). MacHattie (2011) discovered significant high-field-strength/rare earth element (HFSE/REE) prospects in the vicinity of the granites, an association typical of A-type granites. Regional relationships suggest that these rocks formed in an intra-continental tectonic environment, with much of the magmatism spatially associated with coeval strike-slip motion along major faults (Pe-Piper et al. 1998; Koukouvelas et al. 2002).

In the eastern highlands, the Fountain Lake Group can be subdivided into a ca. 4000 m sequence of predominantly rhyolitic volcanic rocks (Byers Brook Formation) conformably overlain by ca. 1500 m of predominantly basalt (Diamond Brook Formation). According to Piper et al. (1999), the rhyolites have high abundances of petrogenetic-discriminating trace elements (e.g., Nb, Y, Zr, Ga and Rb) that are typical of A-type magmas and can be subdivided into a lower member dominated by relatively low-Zr rhyolites and an upper member dominated by relatively high-Zr rhyolites. The geochemistry of the low-Zr rhyolites is similar to that of the coeval plutons and is characterized by significantly lower Zr (<600 ppm), low εNd (<+1.0), Ga/Al and La/Th compared to the high-Zr rhyolites in which εNd is >+ 3.0 and has Zr > 1000 ppm (Figs. 4, 5, 6, 7, 8). The overlying basalts have within-plate, continental tholeiitic affinities, characterized by a high Ni, Cr and Mg number, and relatively high ε tNd (>+ 4), which have been interpreted to reflect high potential mantle temperatures influenced by the presence of a mantle plume (Dessureau et al. 2000).

Late Devonian–Early Carboniferous plutonic rocks are predominantly granite and gabbro in composition, with field evidence of magma mixing and mingling between co-existing felsic and mafic magmas that generated intermediate compositions (Pe-Piper et al. 1996). U–Pb dating of plutonic rocks shows that they were emplaced in the upper crust between ca. 363 Ma and 350 Ma (Doig et al. 1996; Pe-Piper et al. 2004). Structural studies show that major shear zones (including the Rockland Brook Fault) underwent dextral strike-slip motion at this time (Miller et al. 1995). Local transtensional zones within this fault system facilitated emplacement of the magmas into the upper crust (Pe-Piper et al. 1998; Koukouvelas et al. 2002). Most plutons are composite, displaying evidence of repeated injections of felsic and mafic magma with mixing and mingling between phases (Pe-Piper and Piper 2002).

The granite plutons are subalkaline to alkaline and all have a geochemical signature typical of A-type granites (Pe-Piper et al. 1991; Papoutsa and Pe-Piper 2014; Papoutsa et al. 2016). According to Pe-Piper and Piper (1998) and Koukouvelas et al. (2002), the A-type geochemistry combined with Nd and Pb isotopic data indicates that the granites were derived by anhydrous melting of Neoproterozoic lower crust. Most of the granite bodies have chemical compositions similar to that of the low-Zr rhyolite in the Fountain Lake Group. However, the Whirley Brook intrusion has a composition similar to the coeval Fountain Lake high-Zr rhyolite (Piper et al. 1993). The youngest intrusive bodies include a voluminous 354–350 Ma gabbro in the southwestern portion of the Wentworth Pluton, and some granite bodies are attributed to partial melting of older granite during intrusion of the mafic magma (Koukouvelas et al. 2002). All mafic intrusive rocks have a continental tholeiitic, within-plate signature, indistinguishable from the mafic rocks of the Diamond Brook Formation.

Zircon saturation thermometry estimates

The temperature at which zircon saturation occurs in a felsic melt can be used to discriminate between felsic magmas with different petrogenetic as well as thermal histories (King et al. 1997; Miller et al. 2003; Schmitt and Simon 2004; Shellnutt et al. 2011; Dostal et al. 2015; Mahdy et al. 2015; Xia et al. 2016). In general, crustally derived calc-alkalic felsic magmas are typically generated by water-fluxed melting and, therefore, form at lower temperatures (<800 °C) than A-type felsic magmas which are generated by relatively high temperature (>800 °C) anhydrous melting (Weinberg and Hasalová 2015; Collins et al. 2016).

The temperature of zircon saturation (T Zr) can be estimated by relating the concentration of Zr to the bulk composition of the magma (Watson and Harrison 1983; Hanchar and Watson 2003; Boehnke et al. 2013). This relationship is expressed by

$${ \ln }D_{\text{Zr}} = ( 10 10 8 \pm 3 2)/T_{\text{Zr}} \left( {^{\text{o}} {\text{K}}} \right) \, - \, ( 1. 1 6 \pm 0. 1 5)\left( {M - { 1}} \right) \, - \, ( 1. 4 8 \pm 0.0 9),$$
(1)

where the M value is defined by [Na + K + 2Ca]/[Al*Si], D Zr is the distribution coefficient of Zr in zircon (500 000) and T Zr is the temperature in degrees Kelvin. The experimental results are based on compositions that are metaluminous to peraluminous and have M values between 1.3 and 1.9 (Watson and Harrison 1983).

The Economy River Gneiss is not sufficiently silicic to use this method. Only felsic rock samples that yielded M values between 1.3 and 1.9 were considered meaningful for this study as it was the range for the original experimental results on which Eq. 1 is based (Hanchar and Watson 2003). The data are shown in Table 1 and are plotted on Fig. 9. The large temperature range in some suites could be related to their evolution in the shallow crust (fractionation, host rock assimilation, alteration) for Zr saturation as the magma crystallizes. As such, only the maximum TZr is considered diagnostic of the temperature at which melting commences.

Table 1 Summary of zircon saturation thermometry estimates of West Avalonia A-type silicic rocks and Late Neoproterozoic Georgeville Group felsic volcanic arc rocks
Full size table
Fig. 9
figure 9

Zircon saturation thermometry estimates of West Avalonia A-type silicic rocks and Late Neoproterozoic Georgeville Group felsic volcanic arc rocks. The wide range in Tzr values for some of these suites either reflect fractionation from the parental felsic magmas, or alteration of zircon at the subsolidus stage. The highest value for each suite is considered to be representative of the melting temperature. The dividing line predominantly calc-alkalic (CA) and predominantly A-type (A) granites is from Collins et al. (2016) and is based on 99% of all calc-alkaline granites in the Mesozoic North American Cordillera, which had zircon saturation temperatures <800 °C

Full size image

The arc-related ca. 618 Ma Georgeville Group felsic volcanic rocks yield the lowest T Zrn (oC) values, ranging from 658 to 780 °C with one outlier at 526 °C. The 580 Ma Georgeville granites yield a relatively restricted range, but higher T Zr (°C) values (790 to 840 °C) compared to the Georgeville Group. The ca. 495 Ma West Barneys River silicic rocks produced a bimodal distribution with two T Zr (°C) values of 850 and 860 °C and an additional two of 720 and 620 °C. The ca. 460 Ma Dunn Point felsic rocks have T Zr (°C) values ranging from 860 to 875 °C and the ca. 455 Ma McGillivray Brook rhyolite produced only one T Zr (°C) value (1050 °C). The silicic rocks from ca. 360 to 350 Ma Fountain Lake Group yield T Zr (°C) values ranging from 770 to 1020 °C.

With the exception of the calc-alkalic Georgeville Group felsic rocks, most of the T Zr (°C) values for the plutons and volcanic rocks (arranged in chronological order in Fig. 9) are greater than 800 °C, the temperature used by Collins et al. (2016) to distinguish between predominantly calc-alkalic and predominantly A-type granitic magmas. The maximum TZr value for each suite is considered the best indicator of the melting temperature of the source rock and appears to increase from the Late Neoproterozoic to the Devonian-Carboniferous. These results, together with the geochemical and Sm–Nd isotopic evidence, are consistent with the interpretation that the felsic magmas were each derived by melting the lower crust and the temperatures correspond to those expected for the formation of a lower crust hot zone associated with crustal melting (Huppert and Sparks 1988; Murphy et al. 1998; Annen and Sparks 2002; Annen et al. 2006, 2015; Annen 2009). The high T Zr (°C) values of the 580–350 Ma felsic magmas may reflect episodes of dry partial melting of lower crustal rocks.

Summary and discussion

Geochemical and Sm–Nd isotopic data show that Avalonian magmatism in northern mainland Nova Scotia produced repeated episodes of A-type felsic magmatism starting at ca. 580 Ma and continued into the Carboniferous. With the exception of the highly altered Georgeville Granite, all felsic suites plot within the Nd isotopic envelope for crustally derived rocks in West Avalonia, supporting the interpretation that the magmas were derived by anatexis of the crust.

The T Zr (°C) estimates show that the A-type silicic rocks are recording repeated episodes of a relatively hot melting of anhydrous West Avalonian crust between the Ediacaran to Early Carboniferous related to the repeated injection of basaltic magma (Annen et al. 2006). Each magmatic pulse represents a separate tectonomagmatic event but it appears that the same crust is being reworked as the silicic rocks have ε tNd values from ~0 to ~+5 and TDM ages from 0.7 to 1.1 Ga, which are typical of Avalonian basement (Murphy et al. 1998, 2012; Piper et al. 1999; Murphy and Nance 2002; Archibald et al. 2013; Shellnutt and Dostal 2015).

Regional tectonic considerations suggest that A-type felsic magmas were generated in different tectonic environments. The 580 Ma Georgeville Granite was intruded within a continental transform fault system, about 20 Ma after the cessation of subduction. The ca. 490 Ma West Barneys River Plutonic Suite was intruded coevally with respect to the opening of the Rheic Ocean and generation of the Avalonian ribbon continent. The felsic volcanic rocks of the ca. 460 Ma Dunn Point and ca. 455 McGillivray Brook formations were extruded when Avalonia was a microcontinent. Although regional syntheses suggest an ensialic island arc environment, the high T of melting indicates that magmatism was not proximal to regions affected by subduction zone fluids. Regionally extensive hot regions behind arcs are common (Hyndman et al. 2005) and the voluminous felsic magmas with A-type composition produced in ensialic island arcs in the southeastern Pacific, including northern New Zealand (Smith et al. 1977), may provide a modern analogy for ca. 460–455 Ma A-type magmatism in Avalonia. The Fountain Lake Group and coeval plutonic rocks were emplaced in an intra-continental, dextral transtensional environment, up to 60 million years after Avalonia had collided with Laurentia.

The common denominator to the generation of these A-type magma suites is their crustal source. With the exception of the Georgeville Granite, whose Sm–Nd isotopic signature was profoundly affected by alteration of U-rich zircon (Anderson et al. 2008), all suites lie within the envelope for crustally derived Avalonian felsic rocks (Murphy et al. 1996a). In the highly fractionated suites, accessory phase fractionation resulted in LREE depletion, thereby rendering T DM model age calculations geologically meaningless for those suites. However, T DM model age calculations for the West Barneys River Plutonic Suite, Dunn Point Formation and Fountain Lake Group are typical of crustally derived Avalonian felsic rocks. Despite the alteration within the Georgeville granites, if a typical Sm/Nd crustal ratio is assumed (Fig. 9), it is also clear that they were derived from older crust, presumably Avalonian basement.

The inferred high-T metamorphic conditions during magma emplacement of the earliest (~580 Ma) Georgeville plutonic suite (max ~840 °C) are typical for biotite breakdown during anhydrous melting of the lower crust (Collins et al. 1982; Whalen et al. 1987; Creaser et al. 1991). A-type felsic magmatism was preceded by ca. 618 Ma subduction-related, water-fluxed melting of Avalonian crust with maximum temperatures consistently <800 °C (see Collins et al. 2016).

By contrast, the younger magmatic phases typically formed at much higher temperatures (>900 °C) under ultra-high temperature (UHT) metamorphic conditions. The higher melting temperatures suggest that the Avalonian crust had lost much (most) of its hydrous mineral component (especially biotite) during the ~580 Ma event. It was, therefore, probably charnockitic in composition before the onset of Ordovician melting (cf., Landenberger and Collins 1996). The UHT conditions during melting were associated with emplacement of alkalic to tholeiitic mafic magmas, and the high TZr melting temperatures reflect the absence of a hydrous, subduction-related mantle component. The elevated εNd values of the felsic rocks indicate melting of the same Avalonian basement source. The loss of hydrous minerals from the crust resulted in the loss of a buffering mechanism to keep crustal temperatures at ~850 °C or less during melting.

Magmas similar in composition to the Dunn Point and McGillivray Brook A-type rhyolites occur in the modern subduction zone systems of the southwest Pacific (Smith et al. 1977) and may serve as a modern analogue for the tectonic setting in the middle Ordovician. The Late Devonian–Early Carboniferous Fountain Lake Group silicic volcanic rocks formed under UHT conditions, with peak temperatures extending to >1000 °C, consistent with the crustal source having been progressively refined by prior melting events so that all hydrous minerals within that crust have been removed. Like the Ordovician events, emplacement of hot, dry mafic magmas were the ultimate cause of the extreme conditions required to melt the dehydrated crust.

This interpretation extends the conclusion of Pe-Piper and Piper (1998) and Koukouvelas et al. (2002) for Devonian-Carboniferous A-type magmas to include those as old as 580 Ma. It suggests that Avalonian crust in mainland Nova Scotia was removed from the effects of subduction-related, hydrous fluxing by 580 Ma, reflecting the transition from arc to transtensional intraplate regime, possibly similar to the present Basin and Range tectonic setting where “A-type” high-silica rhyolites predominate (e.g., Bachmann and Bergantz 2008). It also shows that, irrespective of changing tectonic settings, the thermal conditions during crustal melting were controlled by a pre-conditioning of the crust associated with progressive removal of hydrous minerals during successive melting events, ultimately resulting in crustal melting under UHT conditions.