1. Introduction
The Grenvillian Central Metasedimentary Belt (CMB) is the only part of the SW Grenville Province that preserves low-grade rocks and primary structures. In this it contrasts with high-grade gneiss terranes on either side, comprising the Central Gneiss Belt and the Adirondack Highlands – Central Granulite terrane (Fig. 1). However, the preservation of abundant primary geological features in the CMB has not led to a clear understanding of its tectonic setting. For this reason we use the descriptive term ‘CMB’, rather than other labels with genetic connotations.
Maps of the Grenville Province to show (a) the location of the CMB within the Grenville Province in eastern North America; (b) location of the CMB between high-grade gneiss belts, and its subdivision into component parts: CAB – Composite Arc Belt; F – Frontenac Terrane, L – Adirondack Lowlands, M – Marble domain; Q – Quartzite domain; Mo – Morin domain.
Following the application of plate tectonic principles to the Precambrian (Dewey & Burke, Reference Dewey and Burke1973), two alternative models were applied to explain the origins of the CMB. The first of these, based on the study of a sequence of volcanic rocks near Kaladar, Ontario, attributed the CMB to an oceanic arc sandwiched between two continental blocks (Brown et al. Reference Brown, Chappell, Moore and Thompson1975). This model was based on an apparent change in chemistry within a 7 km thick volcanic sequence, from ultramafic rocks and alkali-poor tholeiites at the base, to calc-alkaline lavas at the top (Sethuraman & Moore, Reference Sethuraman and Moore1973). The fact that this sequence was intruded by granitoid plutons strengthened the conviction of these authors that they were seeing the development of an island arc founded on older oceanic crust.
However, the island arc model of Brown et al. (Reference Brown, Chappell, Moore and Thompson1975) did not consider the origins of the mafic crust on which the calc-alkaline arc was extruded, and did not explain how subsequent basin closure could have created the ‘cul-de-sac’ geometry of the CMB, trending NNE into the interior of Laurentia (Fig. 1). In contrast, this geometry was specifically addressed by an alternative model (Baer, Reference Baer1976) that interpreted the CMB as an aulacogen, i.e. the failed arm of a rift zone. This model could explain several distinctive geological features of the CMB, such as the coupled distribution of abundant carbonate and volcanic rocks lying between two bounding shear zones, and the extension of a veneer of carbonate rocks outside these boundaries. In addition, Baer attributed the marked thickening of carbonates on the NW side of the aulacogen (near Bancroft, Ontario) to the development of a half-graben type of structure. He also correlated the down-faulted axis of this graben with a linear development of nepheline-bearing igneous rocks characteristic of continental rifts.
Baer's model had many attractive features, and Baer even identified a specific modern analogue, the Danokil depression (Afar triangle) within the southern Red Sea basin. However, the model had a major weakness: the unexplained presence of calc-alkaline plutonic rocks within the CMB, of similar age to the proposed rift-related units (Brown et al. Reference Brown, Chappell, Moore and Thompson1975). As a result, the aulacogen model was largely ignored, and subsequent research (e.g. Pride & Moore, Reference Pride and Moore1983) emphasized the island arc model, culminating in a review paper by Carr et al. (Reference Carr, Easton, Jamieson and Culshaw2000). In that paper, the theory that the CMB was composed of a collage of arc fragments was used to coin the term ‘Composite Arc Belt’. It should be noted, however, that this term refers only to the westerly marble-dominated part of the CMB, excluding the Frontenac Terrane and the Adirondack Lowlands (Fig. 1b).
More recently, a third model was proposed by interpreting the CMB as a back-arc basin formed behind the (1.2–1.35 Ga) Elzevirian continental margin arc. This model was first proposed to explain the rift-like geochemistry of meta-basic units within the CMB (Holm et al. Reference Holm, Smith, Grant and Huang1985, Reference Holm, Smith, Huang, Gerasimoff, Grant, McLaughlin, Moore, Davidson and Baer1986; Smith & Holm, Reference Smith and Holm1990), and was developed by Hanmer et al. (Reference Hanmer, Corrigan, Pehrsson and Nadeau2000) and Rivers & Corrigan (Reference Rivers and Corrigan2000). The back-arc basin model resolves the conflicting evidence for rifting and subduction-related signatures in the CMB, but cannot explain its detailed geometry. For example, it does not explain the gradual narrowing of the marble-rich zone going northwards into Quebec, where it eventually dies out in the vicinity of Archaean basement (Fig. 2). Therefore, when Nd isotope mapping by Dickin & McNutt (Reference Dickin and McNutt2007) identified a segmented distribution of juvenile crustal zones and old crustal remnants within the CMB of Ontario and Quebec, they proposed a modified version of the back-arc basin model, arguing that the geometrical pattern of Nd model ages could best be explained by a failed back-arc rift zone.
Map of the SW Grenville Province showing major tectonic units and boundaries as modified by Dickin & North (Reference Dickin and North2015). Pink and lilac – Grenvillian parautochthon; pale green – Allochthonous Polycyclic Belt (modified from Rivers et al. Reference Rivers, Martignole, Gower and Davidson1989); yellow and orange – Elzevirian rift zone, divided into ensimatic Peterborough and Renfrew segments and ensialic Quebec segment (marble domain). ABT – Allochthon Boundary Thrust; CMBBZ – Central Metasedimentary Belt boundary zone; CLF – Clarendon–Linden fault; G – Grimsthorpe domain.
2. Nd isotope mapping in the Grenville Province
The concept of Nd isotope mapping was developed by Nelson & DePaolo (Reference Nelson and DePaolo1985), based on the use of depleted mantle Nd model ages to estimate the crustal formation age of large areas of continental crust, in order to chart the growth history of the North American craton. Depleted mantle model ages (TDM) are based on an empirical model for the composition of the upper mantle believed to give rise to subduction-related magmatism. The model was based on the Nd analysis of modern and Palaeoproterozoic arc rocks (DePaolo, Reference DePaolo1981), and has been verified in numerous subsequent studies. This model should not be confused with more depleted (DMM-type) mantle models that represent the source of mid-ocean ridge basalts (Workman & Hart, Reference Workman and Hart2005; Dhuime, Hawkesworth & Cawood, Reference Dhuime, Hawkesworth and Cawood2010).
Nd isotope mapping was applied to the Grenville Province by Dickin & McNutt (Reference Dickin and McNutt1989), and was used to identify the edge of the Archaean craton within the Central Gneiss Belt of Ontario. Archaean basement within the NW Grenville Province has an average TDM age of 2.73 Ga (Dickin, Reference Dickin1998), in excellent agreement with U–Pb ages of 2.7 to 2.75 Ga for the adjacent western Abitibi belt of the Superior Province (e.g. Jackson & Fyon, Reference Jackson, Fyon, Thurston, Williams, Sutcliffe and Stott1991).
Immediately north of the CMB, Nd model ages have likewise been verified by U–Pb dating as accurate estimates of the formation age of the crust. For example, Slagstad et al. (Reference Slagstad, Culshaw, Daly and Jamieson2009) determined an average TDM age of 1.50 Ga on 16 samples of plutonic orthogneiss from the Muskoka domain of the Central Gneiss Belt, including two samples with U–Pb crystallization ages of 1.44 and 1.47 Ga. Similarly, Dickin et al. (Reference Dickin, McNutt, Martin and Guo2010) determined an average TDM age of 1.52 Ga on a suite of 35 gneisses from a wider E–W suite along the SE edge of the Central Gneiss Belt in Ontario. These ages are in remarkable agreement, bearing in mind that TDM ages are expected to represent the crustal formation age of an arc terrane that was extracted from the mantle over a period of several tens of millions of years, whereas U–Pb dated plutons will normally represent some of the younger intrusive bodies in such a terrane. Comparisons between TDM ages and U–Pb ages in large areas of the CMB have likewise shown excellent agreement. For example, seven plutonic orthogneisses analysed by McNutt & Dickin (Reference McNutt and Dickin2012) gave an average TDM age of 1.24 Ga, while U–Pb ages for the same bodies (quoted with permission from the unpublished data of Heaman) give an average age of 1.236 Ga.
A detailed Nd isotope study of the CMB boundary zone (CMBBZ, Fig. 2) revealed a very sharp age boundary between old (>1.35 Ga) TDM ages in orthogneisses of the Central Gneiss Belt and juvenile (<1.35 Ga) TDM ages in orthogneisses from the CMB (Moretton & Dickin, Reference Moretton and Dickin2013). This sharp age boundary corresponds within a few hundred metres to the NW limit of marble outcrops, showing that the extent of marble in the NW part of the CMB corresponds very closely to the extent of juvenile (1.2–1.35 Ga) Elzevirian-age crust. Considering the sharpness of this boundary, it was particularly significant that Dickin & McNutt (Reference Dickin and McNutt2007) found Nd evidence for the existence of a block of older crust within a lithotectonic unit of the CMB that Easton (Reference Easton1992) termed the Grimsthorpe domain (G in Fig. 2). This domain is characterized by volcanic and major plutonic lithologies that are largely distinct from the surrounding domains of the CMB, which are dominated by marbles (Easton, Reference Easton1992). This suggests that the old crustal ages within the Grimsthorpe domain of the CMB are indicative of the existence of a distinct crustal block that was not previously identified.
It was earlier proposed by Hildebrand & Easton (Reference Hildebrand and Easton1995) that the large-scale structure of the CMB consists of ‘a regionally extensive thrust fault that places hot, pluton-riddled metamorphic rocks over cool platformal carbonates’. In the light of this model, it was critical to determine whether the old TDM ages of the Grimsthorpe domain could represent an isolated klippe of Hildebrand and Easton's ‘upper-plate’ (which they correlated with the Frontenac Terrane), emplaced over a ‘lower-plate’ consisting of Laurentian basement overlain by platformal marbles. However, Dickin and McNutt found that juvenile crustal Nd signatures (with TDM ages <1.35 Ga) obtruded into the eastern side of the Grimsthorpe domain. Hence, they proposed that the old crustal Nd signatures represented a buried block of old crust underlying part of the Grimsthorpe domain, and that the Grimsthorpe domain could therefore not be an allochthonous displaced terrane (or a tectonic window).
3. The failed back-arc rift model
Geographically, the Elzevir block divides the zone of juvenile Nd model ages in the Ontario part of the CMB into two separate segments. The westerly one continues southwards into the subsurface of southern Ontario and the United States (Peterborough segment in Fig. 2), whereas the easterly (Renfrew) segment continues northwards into the Marble domain of Quebec (Quebec segment in Fig. 2). Hence, the overall shape of this structure was argued to be a series of relatively narrow en échelon rift segments, whose overall shapes have been preserved through subsequent compressional events.
Dickin & McNutt (Reference Dickin and McNutt2007) showed that TDM model ages in the Marble domain of Quebec are significantly older (average = 1.42 Ga) than the Ontario rift segments (average = 1.25 Ga), suggesting that unlike the two ensimatic (juvenile) rift zone segments in Ontario, the Quebec segment was ensialic. This suggests that as the rift zone propagated northwards, it died out when it encountered cold (Archaean) basement (Fig. 2). Hence, the crustal geometry of the rift zone resembles the northwards transition from the ensimatic Red Sea to the ensialic Gulf of Aden.
A surprising aspect of the aulacogen model is that failed rift zones, because they become filled with voluminous mafic rock and sediment, can leave a larger trace in the geological record than continental margins that are involved in a complete Wilson cycle. An example of the latter, the Trans-Adirondack Basin in the Adirondack Lowlands (Fig. 1) has been proposed as a conventional back-arc basin (Chiarenzelli et al. Reference Chiarenzelli, Hudson, Dahl and deLorraine2012) with a full Wilson cycle (in miniature) involving rifting, passive margin sedimentation, basin fill during convergence and tectonic compression. However, plutonic orthogneisses straddling the boundary yield an average TDM age of 1.50 Ga (Chiarenzelli et al. Reference Chiarenzelli, Regan, Peck, Selleck, Cousens, Baird and Shrady2010), with only one TDM age (1.29 Ga) that shows any indication of juvenile material from the period of basin development. In contrast, the proposed back-arc rift zone of Ontario contains two juvenile crustal zones, each around 100 km in width, with an average TDM age of 1.25 Ga (Dickin & McNutt, Reference Dickin and McNutt2007 and new data reported here).
Although the back-arc rift zone in Ontario was inverted, in the sense that it was involved in the Grenville orogeny, the geometry of the rift zone appears to have been largely preserved through subsequent compressional events. For example, it formed an ‘Ottawan orogenic lid’ during the Grenville collisional orogeny, and hence remained at a higher crustal level than the remainder of the Grenville Province (Rivers, Reference Rivers2012). Therefore, although seismic profiles show that the CMB underwent shear, the shallow crustal conditions of the orogenic lid allowed the preservation of large areas of the CMB with relatively little change in surface outcrop pattern. For example, there are many plutons in the CMB that preserve sub-spherical outlines characteristic of diapiric magma emplacement. Hence, it appears that outside of the strongly imbricated Mazinaw domain (to be discussed in more detail in Section 6 below), rocks of the CMB display relatively limited degrees of Ottawan ductile deformation (Schwerdtner, Serafini & Yakovenko, Reference Schwerdtner, Serafini and Yakovenko2005).
A notable feature of the rift zone geometry proposed by Dickin & McNutt (Reference Dickin and McNutt2007) was the sharp 90-degree turn of the CMBBZ at the NW corner of the Peterborough segment (Fig. 2). This feature is attributed to reactivation by 1.19 Ga (Shawinigan-age) deformation (Hanmer & McEachern, Reference Hanmer and McEachern1992) of earlier listric normal faults that defined the northern termination of a rift segment. Geophysical evidence (Boyce & Morris, Reference Boyce and Morris2002) suggests that the N–S limb of the CMBBZ continues southwards with almost exactly the same trajectory in the subsurface of southern Ontario (Fig. 2). Similarly, the east side of the juvenile crustal zone adjacent to the Frontenac Terrane can be correlated with the Clarendon–Linden fault (Forsyth et al. Reference Forsyth, Milkereit, Zelt and White1994). This implies that the Peterborough segment of the rift zone is almost parallel sided (Fig. 2). In contrast, the Renfrew segment of the rift zone appears to widen northwards, which is contrary to its overall pattern of narrowing as it propagated into colder Laurentian crust. Therefore, it was conjectured that this local widening (arrowed in Fig. 2) reflects the presence of a hidden old crustal block in this rift segment.
The search for such a block was one of the principal objectives of this study. However, it was also realized that higher resolution Nd isotope mapping was required in other areas. This includes a refinement of the northern extent of the Elzevir block, which was defined by only two samples in the original study of Dickin & McNutt (Reference Dickin and McNutt2007). In addition, more detailed sampling was made of other parts of the juvenile zone to test for the presence of any additional hidden blocks of old crust. Finally, more detailed sampling of the northeastern segment of the CMBBZ was undertaken to more precisely define the geometry of the rift zone in relation to nepheline syenite bodies, which may indicate the locus of rifting events in the CMB (Burke, Khan & Mart, Reference Burke, Khan and Mart2008).
4. Sampling and analytical methods
Since the objective of this study was to characterize the protolith age of the crust as an estimate of its regional crustal formation age, sampling was limited to granitoid orthogneisses that are believed to form by anatexis of more mafic juvenile crust. Previous studies have shown that granitoids of this type have Nd isotope signatures that are consistent and predictable, allowing reliable estimates to be determined of the formation age of the crust using the depleted mantle model of DePaolo (Reference DePaolo1981).
Tonalitic–trondhjemite–granodiorite (TTG-type) gneisses were sampled where possible, since these are believed to be the best examples of the rock types that form the earliest ‘primitive’ type of continental crust. More granitic samples were used as a second choice, since these may have a more complex geological evolution, but usually still preserve the original formation age of juvenile continental crust (McNutt & Dickin, Reference McNutt and Dickin2012). In contrast, sampling of mafic gneisses was avoided because of the increased likelihood of a younger mantle-derived component in these rock types. Metasedimentary gneisses were also excluded because of their uncertain sedimentary provenance.
On average, 1 kg of rock was crushed, after the removal of any weathered, veined or migmatized material, and careful attention was given to obtain a fine powder that was representative of the whole rock. Major element analyses were performed by Activation Laboratories, Ancaster, Ontario, using Li-borate fusion inductively coupled plasma (ICP) analysis. The accuracy of Actlabs data is ensured by the inclusion of many international reference standards as part of the analytical protocol.
Sm–Nd analysis followed our established procedures. After a four-day dissolution at 125°C using HF and HNO3, samples were converted to the chloride form before splitting and spiking. Standard cation and reverse phase column separation methods were used. Nd isotope analyses were performed on a VG isomass 354 mass spectrometer at McMaster University using double filaments and a four-collector peak switching algorithm, and were normalized to a 146Nd/144Nd ratio of 0.7219. Average within-run precision on the samples was ± 0.000013 (2 sigma), and an average value of 0.51185 ± 2 (2 sigma population) was determined for the La Jolla Nd standard during this work. The reproducibility of 147Sm/144Nd and 143Nd/144Nd is estimated at 0.1% and 0.002% (1 sigma), respectively, leading to an analytical uncertainty on each model age of c. 20 Ma (2 sigma), based on empirical experience over several years of analysing duplicate dissolutions.
5. Results
Nearly 70 new Nd isotope analyses are presented in Table 1, where they are used to calculate TDM ages using the model of DePaolo (Reference DePaolo1981). Four samples with TDM ages of 1.35 Ga or greater define the extent of an additional hidden block of old crust south of Douglas (red star in Fig. 3), hence referred to below as the Douglas block (DB in Fig. 3). Samples attributed to one of the juvenile rift zone segments are grouped (Table 1) according to the Grenvillian CMB domains defined by Easton (Reference Easton1992), as revised by Easton (Reference Easton2004) and Easton (Reference Easton2006). New samples are indicated by a map number that corresponds to their order in Table 1. In addition, published data are shown in Figure 3 as un-numbered points. These include a few samples from each CMB domain (Dickin & McNutt, Reference Dickin and McNutt2007; McNutt & Dickin, Reference McNutt and Dickin2012), a suite from both sides of the CMBBZ (Dickin et al. Reference Dickin, McNutt, Martin and Guo2010) and a more detailed collection from the western part of the Bancroft domain (Moretton & Dickin, Reference Moretton and Dickin2013). However, the present dataset fills a major sampling gap in the northeastern part of the Ontario CMB (Fig. 3).
Map of the eastern part of the Ontario CMB showing new and published Nd data relative to CMB domains recognized by Easton and co-workers (see text). Red star indicates the town of Douglas. Pink shading represents mapped extent of the Weslemkoon batholith. DB – Douglas block; ABT – Allochthon Boundary Thrust. Open symbols indicate TDM model ages <1.35 Ga; solid symbols TDM >1.35 Ga.
Nd isotope data for CMB orthogneisses
* Orthogneisses classified according to rock types in Figure 5, based on Q–P index or hand specimen mineralogy.
Results in Table 1 are compared with data in the above-cited publications on a Sm–Nd isochron diagram (Fig. 4). New samples attributed to juvenile rift zone crust (red circles) are colinear with the published juvenile Nd suite (yellow circles), and fit a 1.27 Ga reference line. Elzevirian plutons studied by McNutt & Dickin (Reference McNutt and Dickin2012) with identical U–Pb and TDM ages of 1.24 Ga (black circles) also lie close to this reference line. The good agreement of U–Pb ages, TDM ages and the Sm–Nd isochron slope verifies the Nd model ages as accurate estimates of the crustal formation age of the juvenile rift zone and shows that the Nd data have not been significantly perturbed by crust–mantle mixing processes.
Sm–Nd isochron diagram showing distinct arrays defined by juvenile CMB crust and older crustal blocks. Regression lines are drawn through the published juvenile suite and the CMBBZ for reference.
Samples with older crustal Nd signatures from the newly defined Douglas block (large pale blue squares) and from the Elzevir and Weslemkoon plutons (black and blue squares) show somewhat more scatter than the juvenile suite, but they are colinear with new and published data from the CMBBZ (open squares), which fit a 1.5 Ga reference line. These compositions are also consistent with the suite of Muskoka gneisses analysed by Slagstad et al. (Reference Slagstad, Culshaw, Daly and Jamieson2009), including samples with U–Pb ages of 1.45 to 1.47 Ga. Hence, the agreement between U–Pb ages, TDM ages and the Sm–Nd isochron slope again verifies the Nd model ages as indicative of the formation age of the crust, and suggests that the Elzevir–Weslemkoon and Douglas blocks represent fragments of old basement rifted away from the walls of the rift zone.
In addition to revealing the existence a small hidden block of older crust south of Douglas, the new Nd data also clarify the extent of old crust in the northern part of the Grimsthorpe domain. Here, the Weslemkoon pluton (pink shading in Fig. 3) extends across almost the whole width of the domain. However, the reconnaissance work by Dickin & McNutt (Reference Dickin and McNutt2007) revealed old (1.47–1.54 Ga) TDM ages in the western part of the Weslemkoon pluton, but juvenile (1.24–1.33 Ga) TDM ages in its eastern part. This may seem surprising, since there is no visible petrological break between the two areas. Therefore, the pluton was subjected to more detailed sampling to test the previous results and to localize any boundary between the two age domains. The new data fully support the earlier model (Fig. 3), but have not yet delineated a petrological boundary in the field.
Major element data were also obtained on samples from both the ‘old’ and ‘juvenile’ Weslemkoon Nd suites in order to see if there is any systematic difference between their petrology/geochemistry. The data are first presented on the Q–P diagram of Debon & LeFort (Reference Debon and LeFort1983), which reproduces the Streckeisen (Reference Streckeisen1973) classification for granitoids using major-element chemistry (Fig. 5).
Q–P diagram of Debon & LeFort (Reference Debon and LeFort1983) which generates an empirical Streckeisen classification. TN – tonalite; GD – granodiorite; MG – monzogranite; GR – granite; QD – quartz diorite; QMD – quartz monzodiorite; DI – diorite; MD – monzodiorite; MZ – monzonite; SY – syenite.
This plot reveals a restricted range of petrology in the Weslemkoon pluton, spanning across the tonalite and granodiorite fields, in contrast to the more variable petrology of samples from the juvenile rift zone, the CMBBZ and the Douglas block (Fig. 5). WK samples showing juvenile Nd signatures (orange circles) are colinear with the suite of dark blue squares, representing old crustal signatures. This similarity between the two sample suites was unexpected, but samples from the Elzevir pluton and from dated Elzevirian-age plutons elsewhere in the CMB also follow the same trend. This suggests that the geochemical trend is the result of phase control of melt compositions formed at similar depths in the crust, rather than a cogenetic origin.
This explanation is supported by a comparison of the chemistry of the Weslemkoon body with the Coast Mountains batholith, British Columbia (Fig. 6). This batholith straddles a terrane boundary between two accreted arcs with distinct crustal formation ages, the inboard Stikinia terrane and the outboard Wrangellia terrane (Girardi et al. Reference Girardi, Patchett, Ducea, Gehrels, Cecil, Rusmore, Woodsworth, Pearson, Manthei and Wetmore2012). Samples of the batholith from the inboard and outboard segments, spanning an across-strike distance of more than 100 km, have largely distinct Nd isotope signatures (Fig. 6c), but appear to form a single differentiation series on variation diagrams against silica, two of which are shown in Figure 6a, b (Girardi et al. Reference Girardi, Patchett, Ducea, Gehrels, Cecil, Rusmore, Woodsworth, Pearson, Manthei and Wetmore2012). Suites from the Weslemkoon body corresponding to old crustal and juvenile segments show almost identical behaviour, suggesting that the Weslemkoon and West Coast magma suites resulted from similar petrogenetic processes.
Variation diagrams of CaO, K2O and epsilon Nd against silica to compare the chemistry of the Weslemkoon body with inboard and outboard segments of the Coast Mountains batholith of British Columbia.
Further evidence that a single batholith can sample both sides of a terrane boundary comes from comparison with the Wooley Creek batholith in the Klamath Mountains of northern California (Coint et al. Reference Coint, Barnes, Yoshinobu, Chamberlain and Barnes2013). This is a much smaller batholith, little more than 10 km wide, but its structural style more closely resembles Weslemkoon, as shown by the mapping work of Lumbers & Vertolli (Reference Lumbers and Vertolli2001). Like the Weslemkoon body, the Wooley Creek batholith has a strong concentric foliation pattern of the type often associated with diapiric emplacement. However, the batholith is tilted along its length, so its lower dioritic–gabbroic parts can be sampled at its northern end, for comparison with the tonalitic–granodioritic petrology of the upper part at its southern end. This tilted attitude shows that even though the southern end of the batholith appears to be a single intrusive body, it was actually amalgamated from a series of separate plutons that can be distinguished in the lower parts of the magmatic system (Coint et al. Reference Coint, Barnes, Yoshinobu, Chamberlain and Barnes2013).
Taken together, these modern analogues show that what superficially appears to be a single, concentrically foliated pluton can in fact be amalgamated from a series of separate magma bodies at depth, which can therefore sample distinct crustal sources on either side of a terrane boundary. Therefore, the isotopically distinct segments of the Weslemkoon body suggest that it is actually a small batholith with multiple magma pulses from different crustal sources. This is consistent with the label ‘Weslemkoon batholith’ that was applied by the original field mappers (Lumbers & Vertolli, Reference Lumbers and Vertolli2001).
McNutt & Dickin (Reference McNutt and Dickin2012) cited an unpublished U–Pb age of 1276 Ma for the Weslemkoon batholith (Heaman, pers. comm.). Since the batholith represents a ‘stitching’ pluton relative to the terrane boundary between the old crustal block and juvenile Elzevirian-age crust, this implies that these blocks were contiguous prior to 1276 Ma, much earlier than the previously proposed minimum age for the ‘assembly’ of the Composite Arc Belt at 1245 Ma (Easton & Kamo, Reference Easton and Kamo2011). This provides a very short time window to create an arc terrane in the form of the Mazinaw volcanics (Sethuraman & Moore, Reference Sethuraman and Moore1973), before its assembly with other CMB terranes took place. These difficulties vanish with the back-arc rift zone model.
6. Discussion
To summarize the new Nd data and compare them with published data, they are plotted as histograms in Figure 7, using the same colour scheme as Figures 4 and 5. This shows that the old crustal blocks (Weslemkoon and Douglas) have the same range of TDM ages (>1.35 Ga) as the CMBBZ and the SE edge of the Central Gneiss Belt (Muskoka and southern Algonquin domains), except that the latter contains a few slightly older ages, consistent with its formation as an ensialic arc on the edge of an older Laurentian continent (Dickin & McNutt, Reference Dickin, McNutt, Gower, Rivers and Ryan1990). On the other hand, crust from the juvenile ensimatic rift zone has TDM ages less than 1.35 Ga, overlapping with the range of TDM ages in dated Elzevirian plutons, but displaying somewhat more scatter. Hence, the data define a bimodal distribution, consistent with petrogenesis of the analysed metaplutonic rocks by partial melting either of old (>1.35 Ga) ensialic crust or juvenile (1.2–1.35 Ga) ensimatic crust. There is very little evidence for mixing between these sources, so the analysed rocks can be used to map out the extent of old and young crustal domains.
Histograms of TDM age for samples suites analysed in this study, along with reference suites from the Central Gneiss Belt (Dickin et al. Reference Dickin, McNutt, Martin and Guo2010) and dated Elzevirian-age plutons (McNutt & Dickin, Reference McNutt and Dickin2012).
Our refined Nd isotope mapping of the CMBBZ and the Elzevir and Douglas blocks yields more precise geometrical information about the shape of the proposed Elzevirian back-arc rift zone. When the areas with old crustal signatures are compared with the mapped extent of nepheline syenite bodies in the Bancroft area (red colour in Fig. 8), it can be seen that the nepheline syenites approximately bisect the juvenile crustal zone between the CMBBZ and the two old crustal blocks. We therefore infer that the nepheline syenites were emplaced as a late magmatic suite down the axis of a failed spreading zone between the CMBBZ and the old crustal blocks. On the other hand, if the nepheline syenites had been emplaced as early magmatic products associated with the beginning of rifting, their distribution should have been along the margins of the juvenile crustal zone.
Back-arc ‘aulacogen’ model for the CMB showing proposed extent of the Elzevir block and the Douglas block. The extent of the Bancroft nepheline syenite suite is shown in red, in contrast to the earlier Blue Mountain body (blue colour). Open symbols indicate TDM model ages < 1.35 Ga; solid symbols TDM >1.35 Ga. Selected plutonic bodies are shown (see text).
The Blue Mountain nepheline syenite (blue colour Fig. 8) does not fit this model, since it is not colinear with the Bancroft nepheline syenite suite. However, we suggest that it is part of an older age suite. It is well known that there were several episodes of nepheline syenite magmatism in the CMB, spanning nearly 200 Ma and including Elzevirian and Ottawan-age suites (Lumbers et al. Reference Lumbers, Heaman, Vertolli, Wu, Gower, Rivers and Ryan1990). It has generally been assumed that the Elzevirian-age representatives were of uniform age, but the evidence suggests otherwise.
A rubidium–strontium isochron for the Blue Mountain nepheline syenite (Krogh & Hurley, Reference Krogh and Hurley1968) gives an age of 1274 ± 40 Ma using the new decay constant (Dickin, Reference Dickin2005). The isochron was of high quality, but the known mobility of Rb during metamorphism suggests that it should be regarded as a minimum age for intrusion, since Rb–Sr ages for CMB plutons generally underestimate their true ages (Heaman, McNutt & Krogh, Reference Heaman, McNutt, Krogh, Moore, Davidson and Baer1986). The low zircon content of this body has so far prevented U–Pb dating, but several authors have cited a significantly younger U–Pb age of 1219 Ma on a nepheline gabbro from the Bancroft suite (R. R. Miller, unpub. Ph.D. thesis, Univ. Toronto, 1985). Lumbers et al. (Reference Lumbers, Heaman, Vertolli, Wu, Gower, Rivers and Ryan1990) argued that this result underestimated the true age, because most zircon in the sample was from an Ottawan-age metasomatic event. Instead, they estimated a minimum age of 1250 Ma for the Bancroft nepheline syenite suite, on the basis that these rocks are cross-cut by the c. 1250 Ma age alaskite granite suite. However, more recent work by Easton & Kamo (Reference Easton and Kamo2011) showed that the alaskite suite of Lumbers et al. (Reference Lumbers, Heaman, Vertolli, Wu, Gower, Rivers and Ryan1990), termed the Methuen suite by Easton (Reference Easton1992), is diachronous. The estimated age range in the Belmont domain (Fig. 3), including the Methuen pluton itself, was from 1240 to 1250 Ma, whereas the age range in the Harvey–Cardiff domain was from c. 1210 to 1230 Ma. Furthermore, McNutt & Dickin (Reference McNutt and Dickin2012) cited an even younger unpublished U–Pb age of 1200 Ma for the Faraday alaskite of the Bancroft terrane (Heaman, pers. comm.).
These new age data remove the basis for a minimum age of 1250 Ma proposed for the Bancroft nepheline syenite suite by Lumbers et al. (Reference Lumbers, Heaman, Vertolli, Wu, Gower, Rivers and Ryan1990). However, Easton & Kamo (Reference Easton and Kamo2011) proposed an age only slightly younger (1245 Ma) based on chemical similarities with the 1246 Ma Glamorgan (Trooper Lake) alkali gabbro pluton, dated by Pehrsson, Hanmer & van Breemen (Reference Pehrsson, Hanmer and van Breemen1996). Therefore, although the age of the Bancroft nepheline syenite suite remains uncertain, the available evidence suggests that it is several tens of millions of years younger than the Blue Mountain nepheline syenite. This suggests that the Bancroft nepheline syenites might mark a later stage of magmatic activity towards the end of the CMB rifting process.
The oldest unit of the CMB dated by U–Pb is the Dysart gneiss complex, with a crystallization age of 1337 Ma (Heaman, unpub. data cited by McNutt & Dickin, Reference McNutt and Dickin2012). However, Nd data from this body shows it to be derived from crust that pre-dates formation of the rift zone (Moretton & Dickin, Reference Moretton and Dickin2013). The second oldest units, which come from areas of juvenile crust in the Harvey–Cardiff and Belmont domains, are the Anstruther gneiss complex, with a U–Pb crystallization age of 1290 Ma (Burr & Carr, Reference Burr and Carr1994), and the Cordova Lake dacite, with a crystallization age of 1287 +11/−3 Ma (Davis & Bartlett, Reference Davis and Bartlett1988). Both of these units are calc-alkaline felsic rocks from areas of juvenile crust. Hence, the formation of mafic juvenile crust in these areas probably pre-dated the felsic units by several millions of years. In comparison, the oldest reported U–Pb age for the Mazinaw terrane is slightly younger, at 1276 ± 2 Ma (Corfu & Easton, Reference Corfu and Easton1995). However, comparison with the data of Davis & Bartlett (Reference Davis and Bartlett1988) suggests that the upper age uncertainty quoted by Corfu and Easton could have been underestimated, and therefore that the interval between volcanism in the two areas could be less than 10 Ma.
It is well established, as described in the introduction, that volcanism in the CMB began with tholeiitic compositions and transitioned to calc-alkaline. Therefore, it can be anticipated that rifting to generate juvenile mafic crust probably occurred at least 10 Ma before the oldest dated calc-alkaline units. Therefore, rifting in the Harvey–Belmont area probably occurred before 1300 Ma, followed shortly afterwards in the Mazinaw area. In contrast, the oldest ages determined for the Bancroft terrane are significantly younger, with a maximum age of calc-alkaline plutonism represented by the 1250 Ma McArthur Mills tonalite (Heaman, unpub. data cited by McNutt & Dickin, Reference McNutt and Dickin2012). This suggests that the Bancroft terrane may represent a younger crustal segment of the CMB than the Harvey–Cardiff–Belmont–Mazinaw domains (Fig. 3). However, younger bimodal tholeiitic volcanic rocks were also found in the Mazinaw domain, dating to 1246 Ma (Corfu & Easton, Reference Corfu and Easton1995). Easton (Reference Easton2006) recognized this as a later ‘rifted arc’ succession, and this suggests (relative to the early tholeiitic series in this area) that the locus of rifting was shifting back and forth as the back-arc rift zone was developing.
This type of behaviour is also seen in the modern Ethiopian rift system, where recent rhyolite volcanism is located in a narrow belt of axial grabens within the earlier rift valley (Beutel et al. Reference Beutel, van Wijk, Ebinger, Keir and Agostini2010). However, it is also notable that Quaternary peralkaline activity is located within the older Ethiopian flood basalt province (Natali et al. Reference Natali, Beccaluva, Bianchini and Siena2013). This suggests that when bimodal rift-related magmatism shifts to a new locus in the rift, older magmatic centres may transition to more alkaline magmatism as the heat source dissipates, producing small-degree alkalic mantle melts. This can explain the locus of nepheline-bearing magmatism in the CMB as an expression of the waning of magmatic activity in one area, as the locus of rifting shifted to a new back-arc spreading centre.
An attempt to explore the geometrical ramifications of such a model is presented in Figure 9 in the form of a schematic geological history of the CMB. This history begins around 1300 Ma with the development of the Peterborough and Renfrew rift segments (Fig. 9a). Creation of new ensimatic crust followed over the next 10 Ma (Fig. 9b), but as the heat supply that had driven this rifting event came to an end around 1275 Ma, the Blue Mountain nepheline syenite was emplaced (Fig. 9c). At the same time, the locus of rifting jumped to the northwest, forming a narrow ensimatic spreading centre (pink shading in Fig. 9c) on the west side of the earlier juvenile crustal segments. This new rift segment, largely corresponding to the Bancroft terrane, is not characterized by large amounts of volcanic rock at the present erosion level, but has the greatest thickness of marbles. When coupled with the juvenile Nd isotope signature of the Bancroft terrane, this implies that mafic volcanic rocks underlie the marbles.
Model showing multi-stage development of the Ontario segment of the Elzevirian back-arc rift zone. Pale green – pre-rift crust; pale blue – early Peterborough and Renfrew rift segments; dark blue – Blue Mountain nepheline syenite; pale pink – late Bancroft rift segment; red – Bancroft nepheline syenite suite; yellow – late Mazinaw rift; yellow-brown – imbricated Mazinaw domain plutons; mid green – Cardiff and Hinchinbrook plutons (see text); dark shading – Palaeozoic cover.
When the heat from this event had almost dissipated, nepheline syenites were emplaced near the axis of the spreading centre (Fig. 9d), at the same time as bimodal rift-related volcanism switched back to the Mazinaw domain (Fig. 9e). The emplacement of the nepheline gabbros and syenites during the waning stages of spreading gave rise to similar late alkaline magmatism in the Rio Grande Rift, possibly reflecting small-degree melting of lithospheric mantle sources (Leat et al. Reference Leat, Thompson, Dickin, Morrison and Hendry1989; Thompson et al. Reference Thompson, Ottley, Smith, Pearson, Dickin, Morrison, Leat and Gibson2005).
7. Conclusions
The proposed multi-stage rifting model places the exotic alkaline rock types of the Bancroft area within a viable plate tectonic framework for the first time. It explains why this magma suite, elsewhere specifically associated with rifting events, is here emplaced into juvenile crust approximately intermediate between older crustal blocks. The rifting model also explains how a juvenile crustal zone could cut obliquely into the interior of the Laurentian continent (Fig. 2) dying out where it encountered cold Archaean basement and forming an aulacogen.
It may seem surprising that the proposed geometry of the rift zone could have been preserved through the successive phases of the Grenville orogenic cycle. However, we suggest that the key feature that preserved its geometry was its infilling with mafic igneous rock, which made it act as a relatively rigid block during subsequent tectonic inversion. The presence of large volumes of mafic rock in the CMB is attested to by a regional gravity high, which was modelled by Roy & Mereu (Reference Roy and Mereu2000) as a dense layer at the bottom of the CMB allochthon, thrust over less dense rock of the Central Gneiss Belt. Probably the CMB became detached from an ultramafic under-plate during the Shawinigan event (c. 1180 Ma), and was thrust as a unit over the gneiss belt by reactivation of listric normal faults that defined the original NW margin of the rift zone, thus forming the CMBBZ. The Bancroft shear zone was also developed during the Shawinigan event (Hanmer & McEachern, Reference Hanmer and McEachern1992), possibly following other listric normal faults originally associated with one of the phases of rifting.
The much larger Ottawan collision began to the SE of the exposed Grenville Province, with a suture zone now buried under the Appalachians (Bartholomew & Hatcher, Reference Bartholomew and Hatcher2010). Relative to this collisional event, the CMB was uplifted as a single unit, acting as a rigid orogenic lid (Rivers, Reference Rivers2012). However, because of the larger size of the orogen, gravitational collapse began much lower in the crustal section, along a new thrust zone to the NW, forming the Allochthon Boundary Thrust (ABT, Rivers et al. Reference Rivers, Martignole, Gower and Davidson1989). Meta-eclogites near the ABT attest to its character as a crustal-scale ramp (Martignole, Reference Martignole1992), but the allochthon to the NW forms a large sub-horizontal thrust sheet (Dickin and North, Reference Dickin and North2015).
The recognition of widespread ramp-flat thrusting in the SW Grenville Province (Dickin et al. Reference Dickin, Herrell, Moore, Cooper and Pearson2014) brings its structure into line with the eastern Grenville Province, where such structures were modelled kinematically by assuming a block of higher rigidity in the centre of the orogen (Jamieson et al. Reference Jamieson, Beaumont, Warren and Nguyen2010). In comparison, we argue that the location of the main ramp of the ABT was itself controlled by the existence of the CMB as a rigid block (Dickin & North, Reference Dickin and North2015). In contrast to the ABT, motion on the CMBBZ was largely extensional during the Ottawan orogeny (van de Pluijm & Carlson, Reference Van der Pluijm and Carlson1989), so that the underlying Muskoka domain of the CMB acted as an extrusion zone similar to the Manicouagan Imbricate Zone (Jamieson et al. Reference Jamieson, Beaumont, Warren and Nguyen2010).
Within the CMB, the only major Ottawan-age tectonism occurred in the Mazinaw domain, which was pervasively imbricated (Corfu & Easton, Reference Corfu and Easton1995; Schwerdtner, Downey & Alexander, Reference Schwerdtner, Downey, Alexander, Tollo, Corriveau, McLelland and Bartholomew2004). The Hinchinbrook pluton (green body in Fig. 9f) may have acted as a relatively rigid block that transmitted stress into the Mazinaw domain as an indentor (large arrow in Fig. 9d). Similar effects are seen at the northern end of the Harvey–Cardiff domain. The series of gneiss domes forming this domain (Fig. 7) probably acted as another rigid block, exerting northward pressure on the Bancroft terrane that caused deformation of the nepheline syenites north of the Cardiff gneiss dome (shorter arrow in Fig. 9f).
Acknowledgements
We acknowledge McMaster University for scholarship support and we thank Toby Rivers and an anonymous journal reviewer for helpful comments that led to significant improvements in the manuscript.
