Timing and anatomy of granitic strain gradients in the Grenville Front tectonic zone, Ontario, Canada

Strain localization operates at a wide range of scales and locations throughout the lithosphere, including in middle to lower orogenic crust (e.g., Coward, 1980; Austrheim and Griffin, 1985; Holdsworth and Strachan, 1991; Wittlinger, 1998; Calvert, 2004). Most kilometer-scale localization, such as the Legs Lake shear zone in Saskatchewan, Canada (Mahan et al., 2006), the Coast shear zone in southeast Alaska (Klepeis et al., 1998), and in the Cabo Ortegal region in northwest Spain (Puelles et al., 2009) initiates at preexisting strength heterogeneities (cf. Molnar and Dayem, 2010). However, for a zone of incipient localization to develop into mechanically and geologically significant structures such as those just mentioned, rheological weakening of an order of magnitude or more (e.g., Gerbi et al., 2010; Platt, 2015) must occur.

Candidate weakening processes are well established, including metamorphism to weaker phases and a deformation mechanism switch, perhaps due to grain-size reduction and/or the development of a crystallographic preferred orientation (e.g., Poirier, 1980; White et al., 1980; Montési and Zuber, 2002; Regenauer-Lieb and Yuen, 2004). Weakening mechanisms need not operate in isolation; rather, multiple factors, such as reaction softening and changes in the spatial patterns of deformation mechanisms (e.g., Tsurumi et al., 2003; Menegon et al., 2006; Culshaw et al., 2010; Kilian et al., 2011; Gerbi et al., 2016) can combine to generate the rheological change necessary for kilometer-scale localization.

Investigations of granitic shear zones, mainly at the millimeter to meter scale in greenschist and lower amphibolite facies, have identified granular flow of feldspars (Stünitz and Fitz Gerald, 1993), competency contrasts between different plutonic phases (Belkabir et al., 1998), coeval brittle and viscous processes (Berthé et al., 1979), and fluid-assisted mass transfer (Kwon et al., 2009) as weakening factors. Other causes include preexisting fabrics as pathways for fluids (e.g., Segall and Simpson, 1986), inherited strength heterogeneities (e.g., Christiansen and Pollard, 1997), rock fabrics and microstructures (e.g., Schulmann et al., 1996; Martelat et al., 1999), melt (e.g., Garlick and Gromet, 2004; Rosenberg and Handy, 2005; Závada et al., 2007; Schulmann et al., 2008), and continuous recrystallization of feldspars (Vauchez, 1987; Oliot et al., 2010).

Despite the large body of work at the outcrop and smaller scale, few studies have investigated whether and how these local weakening mechanisms operate at the kilometer scale in granites. For example, at some sites, localization is attributed to kilometer-scale thermal gradients (Stipp et al., 2002; Klepeis and King, 2009) that cannot be present at the outcrop scale. Although granites and metagranites are abundant in middle to lower orogenic crust, most kilometer-scale rheological studies have focused on mafic or metapelitic lithologies, which are susceptible to metamorphic reactions (e.g., Snoke et al., 1999; Racek et al., 2006; Chambers et al., 2009; Marsh et al., 2011). As such, the factors influencing the formation of kilometer-scale shear zones in granitic rocks are not well established.

To take a holistic view of the rheological change that can lead to kilometer-scale high-strain zones in granitic rocks in middle orogenic crust, we have gathered microstructural, geochemical, and geochronological data across two gradients for which we can approximate relative strain. These strain gradients include a transition from preserved igneous structures to thorough recrystallization and penetrative foliation development. In the study area, which lies at the southwest end of the Grenville Front tectonic zone (GFTZ) in Ontario, Canada, we find that formation of the observed strain gradient required the interrelated operation of several microstructural processes. Complementary to assessing rheological change in middle crustal, orogenic, kilometer-scale granitic shear zones, our structural and geochronological data validate previous interpretations of a regionally consistent tectonic history.

The Grenville orogeny is considered to have been a Himalaya-Tibetan type collision that affected the region from present-day Texas through Labrador in North America (Hynes and Rivers, 2010). The most continuously exposed package extends for ∼2000 km in eastern Canada (Fig. 1A). Archean to Paleoproterozoic Laurentian crust was reworked during a sequence of collisional Grenville events from ca. 1100–980 Ma (Wynne-Edwards, 1972; Davidson et al., 1982). The Grenville Front tectonic zone (GFTZ; Fig. 1B) formed during late stages of orogenesis when deformation migrated northwest into the Laurentian craton in response to crustal thickening to the southeast (Haggart et al., 1993; Jamieson and Beaumont, 2011).

The southern extent of the GFTZ in Ontario encompasses the area from Killarney to the French River and crops out along the northern shore of Georgian Bay (Fig. 1). This part of the GFTZ exhibits north-northeast–striking lenticular rock units separated by strongly foliated, southeast-dipping, migmatitic orthogneisses and subordinate paragneisses with shear bands indicating thrust-sense kinematics (Davidson and Bethune, 1988; Jamieson et al., 1995; Hynes and Rivers, 2010). The previously undated granite body that is the focus of this study is one of these lenticular units, and it is bounded by also previously undated strongly foliated orthogneiss. Thermobarometry indicates that peak metamorphism in this region of the GFTZ increases in grade to the east, having reached conditions of greater than 1.2 GPa and 800 °C prior to exhumation ca. 1035–980 Ma (Jamieson et al., 1995; Carr et al., 2000, and references therein). The majority of currently exposed metamorphic assemblages in the area formed over conditions ranging from ∼740 °C, 1.1 GPa to ∼600 °C, 0.65 GPa (Jamieson et al., 1995). Synorogenic erosion drove GFTZ exhumation along a crustal-scale detachment ramp (Beaumont et al., 1992; Rivers et al., 1993; Jamieson et al., 1995; Jamieson et al., 2010).

The eastern boundary of the GFTZ in southern Ontario is located along the French River outlet to Georgian Bay (Fig. 1C). The French River coincides with the Boundary shear zone, which separates the GFTZ from the Britt domain of the Central Gneiss Belt to the southeast (Jamieson et al., 1995). The parautochthonous Britt domain, from which rocks in this portion of the GFTZ were derived, dominantly comprises 1800–1600 Ma polymetamorphic, quartzofeldspathic gneisses locally intruded by ca. 1450 Ma plutons (van Breeman et al., 1986; Corrigan et al., 1994; Culshaw et al., 2004).

To the west of the Boundary shear zone lies a series of NE-SW–striking, km-scale, thrust-sense shear zones that accommodated the Grenvillian strain in the GFTZ (Davidson and Bethune, 1988; Fig. 1C). A 1986 seismic study revealed large southeast-dipping reflectors interpreted as the continuation of shear zones mapped at the surface of the GFTZ at depth (Green, 1988; White et al., 2000). Reflectors flatten out under the Central Gneiss belt to the southeast.

The two strain gradients comprising this study lie within the ∼4-km-wide (in map view) Bad River granite and between the Boundary shear zone to the east and the Bad River shear zone to the west (Fig. 1D). As noted above, our interest lies in the strain gradients rather than the cores of the shear zones where the processes leading to localization are no longer preserved. Based on foliation intensity, universally highly elongate and foliation-parallel inclusions and mineralogical segregations, degree of recrystallized grains, and nearly uniformly fine grain size, we infer that the locally highest strain developed within the previously undated pink orthogneisses that border the Bad River granite (Fig. 1D). However, this inference has no bearing on the rheological data and conclusions presented here, which focus solely on the Bad River granite.

We separated zircon from each sample through a standard procedure of crushing, pulverizing, and density and magnetic separation. Approximately 50 grains were mounted in epoxy and polished to expose grain interiors. We used cathodoluminescence (CL) images to identify any internal zoning. An additional ∼20 grains from each sample were mounted on tape and analyzed from their exterior (depth profiling). Grains between ∼150 and ∼200 μm were chosen for spot analyses and those greater than ∼200 μm were chosen for profiling, since a larger spot size is used for profiling. The U-Th-Pb isotopic compositions of the zircon were measured at the University of Texas, Austin, using a Photon Machines 193 nm Analyte G2 excimer laser-ablation system with large volume Helex sample cell, coupled with a Thermo Scientific Element2 high-resolution–inductively coupled plasma mass spectrometer (HR-ICPMS). A spot size of 30 μm was used for polished zircon analyses, and one of 50 μm was used for the first round of depth profiling. A laser fluence of 2.24 J/cm² and repetition rate of 10 Hz was used, with He gas flow of ∼0.5 L/s. Ablation times of 35 s (for 30 μm spots) and 60 s (for 50 μm spots) resulted in ablation depths of ∼15 μm and ∼30 μm, respectively. A spot size of 30 μm was used for the second round of depth profiles due to a mix of spot and profiling analyses. At least 30 seconds of background signal were acquired between each analysis. Signals for masses ²⁰²Hg, ²⁰⁴(Pb + Hg), ²⁰⁶Pb, ²⁰⁷Pb, ²⁰⁸Pb, ²³²Th, and ²³⁸U, with ²³⁵U were calculated from the measured ²³⁸U concentration using the relationship ²³⁸U/²³⁵U = 137.88. The zircon reference standard GJ-1 (Jackson et al., 2004) was analyzed repeatedly before, after, and intermittently during analysis of the unknown specimens so that mass fractionation and instrumental mass bias corrections could be applied. Data reduction was carried out using the VizualAge U-Th-Pb data reduction scheme (Petrus and Kamber, 2012), which operates within the Iolite software package (Hellstrom et al., 2008; Paton et al., 2011). A smoothed spline was fit to the sequentially arranged raw count data for background and reference standard measurements, enabling time-resolved background subtraction and corrections for depth-dependent isotopic and elemental fractionation and instrumental drift.

Major, trace, and rare-earth elements were measured by Activation Laboratory in Ontario, Canada. Trace elements were analyzed by inductively coupled plasma mass spectrometry (ICPMS), and major elements and rare-earth elements (REEs) were analyzed by lithium metaborate–tetraborate fusion ICPMS. Major elements have detection limits between 0.001% and 0.01%. Trace elements and REEs have detection limits between 0.04 ppm and 30 ppm.

We measured mineral compositions at the University of Maine using a Cameca SX100 electron probe micro analyzer (EPMA) with a 15 kV accelerating voltage, 10 nA beam current, and 5 μm spot size. Reported results are the average of ∼10 points/grain. Full analyses can be found in the Supplemental Tables¹.

We mapped samples from across the strain gradients using an Apollo 40 silicon drift detector (SDD) with 50 μs dwell time and beam current of ∼10 nA on the Tescan Vega II tungsten filament scanning electron microscope (SEM) at the University of Maine to determine phase distribution. Postprocessing using OIM Data Collection v. 5.3 and 6.0 allowed partitioning of raw EDS counts into selected phases (Nowell and Wright, 2004) identified through optical petrography.

Selected samples were analyzed for crystallographic orientation of quartz using the EDAX/TSL Digiview III camera and OIM Data Collection v. 5.3 on the same instrument as above. Parameters for each run consisted of a working distance of 25 mm, a tilt of 70°, a beam current of ∼10 nA, and varying step sizes from 15 to 25 μm (Supplemental Table S1 [see ^{footnote 1}]). Postprocessing accounted for misindexing (e.g., Prior and Wheeler, 2009) through neighbor correction of non-indexed points and correction of pseudosymmetry (i.e., Dauphiné twins). We defined grain boundaries at 10° misorientation. We performed EBSD analysis on the full granite but report only the quartz data here. Feldspar results are affected by fine twinning, perthitic structures, fractures, and late alteration, rendering their data less robust.

Due to complications from analytical artifacts and perthitic textures, EBSD data were not used to calculate grain sizes. Instead, grain-size analyses were conducted on optical photomicrographs of the thin sections using the measurement tool of the QCapture Pro 7 software. Grain sizes of porphyroclasts and of recrystallized feldspar and quartz (including ribbon grains) were recorded using the maximum length of the grain and then binned in 20 μm bins. Histogram envelopes were manually drawn.

The strain gradients within the granite are qualitatively defined by progressively more intense foliation—i.e., more closely spaced folia—and more elongate magmatic enclaves toward the eastern and western edges of the Bad River granite. Foliation orientation within the granite and the bounding orthogneisses is relatively constant and subparallel to the unit boundaries (Fig. 1). All sample sites discussed herein were projected to a transect drawn approximately normal to the regional strike of the foliation, with distance measured from the northwest end (Fig. 1D).

Both units of pink orthogneiss are quartzofeldspathic, fine grained, and sugary in texture, containing ∼5%–10% by volume of 1–10-cm-thick, irregularly spaced hornblende-rich layers. Fe-Mg minerals represent <5% of the matrix in the leucocratic layers. The penetrative foliation exhibits a consistent, nearly down-dip lineation, defined mainly by quartz ribbons. The western orthogneiss is darker pink and, though overall fine-grained, appears slightly coarser than the eastern orthogneiss, which contains some medium-grained (1–3-mm-diameter) pods or layers. Both units record little variation in foliation intensity. Contacts between the orthogneiss and the granite are sharp. On the western margin, a single contact exists. On the eastern margin, an outlier of Bad River granite lies within the orthogneiss, to the east of the transect (Fig. 1D).

The Bad River granite comprises predominantly plagioclase, alkali feldspar, and quartz, with minor amounts of hornblende, biotite, garnet, Fe-Ti oxides, apatite, and zircon. The modes of the major minerals vary slightly from outcrop to outcrop but stay within the range of 30%–40% each for plagioclase and alkali feldspar and 15%–30% for quartz. Foliation, defined by a shape-preferred orientation of quartz and biotite-hornblende aggregates, is uniformly steeply southeast-dipping (Fig. 1E), except in small shear zones as described below. The nearly down-dip lineation is defined largely by hornblende and biotite grains and feldspar aggregates. The central portion of the granite, an extent of ∼1.5 km along the transect, is unfoliated to exhibiting weakly discernible foliation, as well as igneous structures, including >10-m-long schlieren bands, rounded xenoliths of the pink orthogneisses, and low aspect ratio magmatic enclaves (Fig. 2). The foliation becomes more penetrative toward the eastern and western granite boundaries, accompanied by deformation of all magmatic features including the stretching of enclaves and xenoliths. Despite the increased foliation intensity, the fabric does not become mylonitic. The regional L = S fabric is cut by pegmatites and intermediate composition dikes with variable strikes, and by east-west–striking fracture sets.

Spaced irregularly at the hectometer scale across the central and western segments of the transect are small, sharp-edged, mylonitic shear zones (∼1–20 m wide) that cut the regional fabric and dominantly strike ENE-WSW (Shulman, 2016). These shear zones crosscut and therefore postdate development of the fabric under consideration here and are not discussed further.

To provide a semiquantitative proxy for strain variation across the Bad River granite transect, we measured the aspect ratio of over 1700 stretched enclaves (Fig. 3A; see Supplemental Table S2 [^{footnote 1}]). In order to avoid duplicate measurements and reduce bias, we measured the lengths and widths of all enclaves within or immediately adjacent to one- or two- meter squares drawn on the outcrop at each measuring station (Figs. 3B and 3C). Exposure on the glacially polished surfaces was not sufficient in the vertical direction to measure three-dimensional aspect ratios. Where exposed on vertical faces, the enclave geometry is consistent with the L = S structure of the host rock. Our two-dimensional measurements incorporate several sources of uncertainty, including: (1) although the foliation is moderately to steeply dipping (∼60° SE), the outcrop surface is rarely perpendicular to foliation, and (2) the surfaces are not always completely flat, such that the error per station may vary. Thus, apparent enclave widths are typically 20%–30% greater than true foliation-normal widths; however, this geometrical effect is constant across the gradient, allowing us to evaluate relative changes in aspect ratio.

We cannot utilize these measurements directly to quantify strain, as explained below, but we can use them to track relative strain across the transect, given certain conditions. (1) As with most strain analysis, we assume the enclaves are passive markers and that the strain is homogenous at each site. Tobisch and Williams (1998) modeled enclave and granite rheologies based on composition, stress, and strain rate. They found negligible differences between biotite-hornblende–rich enclaves and granite host rheologies at granulite- to amphibolite-facies temperatures and moderate stresses (∼10 MPa) and strain rates (10^–13 to 10^–11s^–1). Pearce et al. (2011) showed that mafic dikes are weaker than gneisses under amphibolite-facies conditions. Taking these studies together, we can justify the assumption that the enclaves are effectively passive markers. Even if not, the strength contrast between the enclave and host will be relatively constant, still allowing us to use enclave aspect ratios as a semiquantitative proxy for strain. (2) The change across the gradient is due to subsolidus, tectonically induced strain. It has been reported and modeled that the aspect ratio of enclaves is rarely spherical due to magmatic flow during ascent and in the chamber and that the aspect ratios tend to increase toward the edge of an intrusion (Davis, 1963; Schmeling et al., 1988; Cruden, 1990). Paterson et al. (1998 and references therein) summarize structural data from five plutons, including the magma-flow–related aspect ratios of enclaves located toward the boundaries of the plutons. They report length/width ratios ranging from ∼1.17–13.3. The aspect ratios measured at the edges of the unit in this study are much higher (5.43–48.07 with a mean of 10.84; Supplemental Table S1 [see ^{footnote 1}], site 9), suggesting that suprasolidus strain could not be responsible for the present aspect ratios of the enclaves and that subsolidus strain played the dominant role. Our own observations of the microstructures within the enclaves—for example, undulose extinction and bent Carlsbad twins in plagioclase—are fully consistent with solid-state deformation outside of the portion of the granite preserving igneous structures (cf. Vernon, 2004).

We use the enclave aspect ratios measured across the Bad River granite as a proxy for the relative strain. Because of the uncertainties surrounding its quantification, we do not attempt to calculate absolute strain. Nevertheless, the preservation of igneous features and low aspect ratios of the enclaves near the site of sample 70 indicate that solid state strain was negligible (perhaps <10%) in this portion of the granite. The map-view aspect ratios of the enclaves (Fig. 3) reveal a much more gradual strain gradient toward the eastern margin than toward the western margin, though both margins display smooth rather than sharp transitions. As such, we cannot define sharp boundaries to the shear zone margin. Throughout the remainder of this contribution, we refer to the lowest strain or least deformed portion of the granite as the region near sample 70 (Fig. 3D) and the highest strain as at the contacts with the pink orthogneiss. We refer to the strain gradients overall as the segments to the east and west of the lowest strain portion of the transect (near sample 70), but the effective strain gradients occur along the transect from 0.5 to 1.0 km—the western gradient—and 2.5–3.7 km—the eastern gradient.

Petrographic analyses of three enclaves, one from the area of the granite that preserves igneous structures (sample 12e) and one from each of the margins (samples 207 and 30e), indicate that the enclaves are finer grained and more mafic than the granite. Biotite and hornblende make up ∼25%–40% of the modes in the enclaves, with varying amounts of garnet. Alkali feldspar, plagioclase, and quartz constitute the rest of the matrix, and the phases are well mixed. A weak shape-preferred orientation of hornblende is present in the margin samples (207 and 30e). Although most of the central sample (12e) is recrystallized, relict concentric zoning persists in some plagioclase grains.

Bulk-rock analyses were collected from the pink orthogneiss bounding the granite to the east (sample 106) and west (sample 205) (Table 1). Major-element results from the western and eastern pink orthogneiss correlate well with each other (Fig. 4A). They show that Na₂O and K₂O concentrations are all within the range of the major-element results of the granite, although the western pink orthogneiss (sample 205) has less Na₂O than the least strained sample (70), while the eastern pink orthogneiss has more. In both pink orthogneisses there is a much lower concentration of Fe₂O₃, MgO, TiO₂, MnO, and CaO than in the Bad River granite and a higher concentration of SiO₂.

Chondrite-normalized REE analyses of the two pink orthogneisses vary significantly from the Bad River granite’s REE pattern, in particular for Eu (Fig. 4C). They also show a divergence in the heavy rare-earth elements (HREEs) from each other, with the eastern orthogneiss having a higher concentration in HREEs heavier than terbium.

Whole-rock analyses were conducted on the eastern and western edges of the Bad River granite (201 and 108) and three samples across the unit (samples 70, 71, and 76) (Fig. 4 and Table 1). Figure 4A depicts the changes in major-element oxide weight percent across the transect. K₂O is the only oxide whose concentration correlates with strain. An isocon diagram (after Grant, 1986; Marsh et al., 2011; Fig. 4B) shows that there is some variation across the transect, with the most significant variation from the highest strain sample occurring in the westernmost sample, 201a. Samples from the lower-strain area of the transect are averaged and plotted with standard deviation error bars. However, the variation among the samples suggests small magmatic changes that are indistinguishable from any metasomatism that may have occurred.

Chondrite-normalized REE analyses across both gradients have similar patterns indicating they are of the same rock type with only minor variations due to inherent heterogeneities (Fig. 4C; CI chondrite from Sun and McDonough, 1989).

Due to compositional heterogeneity among samples, the Bad River granite straddles the boundary between quartz monzonite and granite; for the purposes of this paper, we refer to the unit as granite.

Because the eastern and western orthogneisses are petrologically quite similar and serve to provide context for the current study—rather than as the focus—we describe the general petrographic character of the units together. Both orthogneisses consist mainly of recrystallized alkali feldspar and quartz, with typically <15% plagioclase, and minor amounts of biotite, hornblende, and iron oxides. Hornblende and biotite are locally more abundant in the irregularly spaced, 1–10-cm-thick layers. Chlorite is present in the western orthogneiss as an alteration product from hornblende. Accessory minerals include zircon, but we did not perform an exhaustive survey of accessory minerals. Recrystallized feldspar grain sizes range from ∼100 μm to ∼2 mm. Alkali feldspar exhibits extensive tartan twinning and flame perthite. Quartz grains are mostly found as smaller interstitial grains between feldspars, but there are also weakly defined quartz ribbons scattered throughout the sample with average grain sizes of ∼400 μm. Plagioclase is typically twinned. In the felsic layers, Fe-Mg phases range from being interstitial and isolated to thin (<100 μm), elongate (>5 mm) bands.

We prepared a suite of samples for optical and electron beam petrography, microstructural description, and chemical analysis from the eastern and western gradients. Both gradients share the least sheared sample, sample 70. Samples from which we report data and specific observations are listed in Figures 1D and 3D.

In its least deformed state, the granite unit is megacrystic and consists predominantly of twinned albitic plagioclase, variably perthitic alkali feldspar, and quartz, with varying amounts of, but at most ∼15%, hornblende, biotite, Fe-Ti oxides, garnet, apatite, and zircon. Some of the biotite and hornblende has evidence of secondary chloritization, especially in the western transect. In this section, we describe and illustrate (Fig. 5) the changes across the transect, organized by phase. Plagioclase, hornblende, biotite, and garnet chemical analyses are listed in Supplemental Table S3 (see ^{footnote 1}).

Both alkali feldspar and plagioclase exhibit core-and-mantle structures in the least sheared samples (Fig. 5A). Alkali feldspar porphyroclasts are perthitic; plagioclase porphyroclasts contain abundant deformation twins. The porphyroclasts of both phases have lobate boundaries and often have aspect ratios under 2:1, with the longest axis parallel to foliation. Alkali feldspar tartan twinning is rare but can sometimes be discerned in grains that have less perthite.

Several aspects of feldspar grain sizes change across the transect (Fig. 5A and Table 2). In the starting material of the least sheared sample, alkali feldspar porphyroclasts are ∼6 mm, and plagioclase porphyroclasts are ∼3.5 mm, while recrystallized feldspar grains—which constitute ∼70% of the feldspar—are ∼400 μm. The fraction of recrystallized grains increases to near 100% in the highest strain samples. In concert, the average grain size of recrystallized feldspar grains decreases from ∼400 μm to ∼250 μm with increasing strain (Fig. 6 and Table 2). Recrystallized feldspar grains often have straight boundaries and in places form ∼120° junctions at like-phase boundaries, creating a foam texture (e.g., Fig. 5A2). Deformation twins can be seen in some but not all of the recrystallized plagioclase grains, and the recrystallized alkali feldspar grains are variably perthitic (Fig. 5A).

Both feldspar phases can be found in loosely defined augen-shaped pods (Fig. 5A) wrapped by quartz ribbons, hornblende, biotite, and Fe-Ti oxides. A small modal amount of garnet (<1%) is present across the eastern gradient. No garnet was found in the western gradient. The augen size and shape mirror that of the porphyroclasts. In the less sheared samples, the augen consist of mantled porphyroclasts, but with increased recrystallization they become fully recrystallized aggregates. As strain increases from the low-strain samples, the recrystallized grains also create a mixed-phase matrix, blurring the augen shapes. Therefore, the augen-shaped clusters are loosely defined in most of the thin sections, but the pods become more defined in higher strained samples (109 and 108; e.g., Figs. 5A3 and 5A4). The augen shapes are nearly entirely absent in the easternmost sample (103) because the phases are mostly dispersed throughout the thin section (Fig. 5B).

We did not measure mineral compositions for the alkali feldspar grains because of their highly perthitic nature. For plagioclase, microprobe analyses were collected from porphyroclasts and recrystallized grains (excluding albitic exsolution from alkali feldspar) in an attempt to document any progression in composition with strain. All of the plagioclase analyses from both gradients plot as oligoclase (Ab₇₅-Ab₈₅), but recrystallized grains are consistently slightly higher in Na content than adjacent porphyroclasts (Fig. 7 and Table 3).

Quartz throughout both gradients is most often found in polycrystalline ribbons with varying degrees of undulose extinction. In general, the less sheared samples have more irregularly shaped quartz, or ribbons with low aspect ratios, when compared to the higher strain samples (Figs. 5A and 8). However, pockets of relatively thicker quartz pools exist in strain shadows in most samples (e.g., Fig. 5A7).

The average grain size of recrystallized quartz decreases from ∼800 μm in the least strained sample (70) to ∼500 μm in the westernmost sample, 201a, and ∼150 μm in the easternmost sample, 103 (Fig. 6B). The dominant size fraction of recrystallized quartz grains in most samples is ∼200 μm, but there is a slight reduction to ∼150 μm in 201a and 103, the westernmost and easternmost samples. More importantly, the fraction of smaller grains increases with distance from the low-strain portion of the granite (Fig. 5C).

Crystallographic orientation distribution plots (Fig. 9) summarize the EBSD data from the recrystallized grains of quartz in each sample analyzed. We report the concentration of crystallographic data (crystallographic preferred orientation [CPO]) as multiples of a uniform distribution (MUD). Quartz in the least sheared sample (70) has a low maximum MUD of 2.6, indicating there is a weak CPO. In the western gradient, a stronger CPO does not develop until the most strained sample, 201a, with a maximum MUD of 5.4. In the eastern gradient, the CPO strengthens with increasing strain through sample 109 to the east, which has the highest MUD maximum of 5.7. East of sample 109, the maximum MUD values decrease to 2.2 in sample 103.

Biotite is found in all samples except the westernmost sample (201a) in the western gradient. Across both gradients, growth of larger biotite can be seen in the strain shadows of the porphyroclasts. Biotite is found mixed with quartz, hornblende, and Fe-Ti oxides in the regions that wrap the feldspar pods or mantled porphyroclasts and near resorbed garnets. The biotite grains of the western gradient tend to have lower aspect ratios than those of the eastern gradient. Biotite compositions across the eastern and western gradients indicate an increase in X_Mg, F, and Cl and a decrease in Ti with strain (Fig. 10 and Table 4).

Hornblende is found in all samples in both gradients with varying amount of chloritization, but in those cases where chloritization was complete, the morphology of the original hornblende can still be discerned. The hornblende morphology, which is similar across both gradients, consists of larger porphyroclasts (∼1.5 mm) with trails of smaller grains that have broken off and entrained in the flow of the fabric. Overall, clusters of hornblende clasts become more localized and sinuous with strain, but the change is not monotonic and is more obvious in the western gradient than the eastern. The porphyroclast size does not reduce with strain, but the modal amount of hornblende minimally increases with strain in the eastern gradient. There is no clear trend in hornblende mode with strain in the western gradient. Analyses from both gradients indicate a slight increase in X_Mg, F, and Cl with strain (Fig. 10 and Table 5).

Garnet is found in only the eastern gradient and not in every sample. The garnet across the eastern gradient is mostly skeletal, likely due to resorption, and typically surrounded by biotite, quartz, Fe-Ti oxides, apatite, and hornblende. The amount of remaining garnet seems to decrease with strain so that there is no garnet in sample 103. The least sheared samples have larger fragments of garnet than the higher strained samples, although they are still clearly remnants of larger clasts. Chemical analyses of the garnets from the least sheared sample (70) are Almandine_56–57Pyrope_9–11Grossular_17–18Spessartine_15–17 and do not show any significant chemical variation across the eastern gradient (Table 6).

Ilmenite and magnetite are the most abundant oxides found across both gradients. Ilmenite is irregularly shaped and in various stages of alteration. In the western gradient, ilmenite often surrounds rutile and calcite. Magnetite is patchy in backscatter, irregularly shaped, and commonly, but not always, associated with ilmenite. Pyrite is found only in the western gradient (including the least sheared sample, 70), is usually found only adjacent to magnetite or ilmenite, and is most often found as metasomatic reactions or skeletal remnants of grains.

U-Pb analyses were performed on zircon from the middle of the Bad River granite unit (samples 70 and 71), each high-strain zone on either margin of the Bad River granite (samples 108 and 201), and each pink orthogneiss unit to the east and west of the Bad River granite (samples 106 and 205) to identify commonalities among igneous, metamorphic, and deformational events and place them within the regional tectonic context. U-Pb results are summarized in Figure 11 (full analyses in Supplemental Table S4 [see ^{footnote 1}]); final ages are reported with 2σ propagated error, including from the standard GJ1 (0.35%).

Plots of all analyses reveal a complex pattern of U-Pb isotopic ratios (Fig. 11). However, comparison of ages present in the different units, morphological characteristics of the analyzed locations, and Th-U ratios permit an internally consistent interpretation of crystallization events. Unless noted, we do not distinguish between profile analyses and traditional spot analyses, terming them all “point” analyses to distinguish them from ages averaging multiple points.

Zircons from the Bad River granite typically have similar morphology and zoning patterns across the transect; most have variably elongate cores with different degrees of oscillatory zoning toward the edges (Fig. 11). The grains commonly have irregular, homogeneous, CL-dark rims cutting the cores. A few grains contain CL-dark cores surrounded by oscillatory zoning.

Zircons from the pink orthogneisses (205 and 106) are morphologically distinct from those in the Bad River granite, tending to be less elongate and euhedral. The grains commonly have metamict cores mantled by homogenous rims, with both zones being CL-dark. However, some grains have rims with oscillatory zoning.

The ²⁰⁷Pb/²⁰⁶Pb point ages ranged from 1023 Ma to 1928 Ma, with the youngest and oldest highly discordant (Fig. 11A). Point ages cluster near 1500 Ma and 1700 Ma. On a concordia plot, both clusters extend discordantly toward the origin, with additional points spread between the two clusters and their chords. The older cluster has higher average Th/U ratios (most in the range ∼0.4–1.4) than the younger cluster (most <0.4; Fig. 12). Because of the diffuse age clusters, the initial results do not yield robust age determinations. Therefore, for this sample, as well as all that follow, we refined the initial estimates of the age clusters by considering the model of punctuated partial isotopic resetting (Faure, 1986). In this model (Fig. 11G), zircon formed at ca. 1700 Ma was variably reset during the younger ca. 1500 Ma event. At this time, the older cluster would have been located at t₀ on concordia on Figure 11G. All zircons evolved along concordia, or adjacent to it, until modern partial resetting of some zircon in both of the clusters plus some of the already partially reset zircon, resulting in several chords toward the origin (Figs. 11A and 11G).

To calculate the refined ages, we selected which point ages to include in the separate calculations of intercept ages. For each cluster, we selected point ages in that cluster plus any that lay on a chord toward the origin. Points between the clusters (dark gray in Fig. 11) and points lying off the chords (light gray in Fig. 11) were not included in the age calculations (individual spot marked in Supplemental Table S4 [see ^{footnote 1}]). Some of the points between the clusters or chords represent mixed U-Pb isotopic compositions due to laser sampling of multiple age domains; others represent partially reset isotope systems. The subsequently calculated upper intercept ages are not highly sensitive to the exact selection of points included. Using the points shown in Figure 11 and Supplemental Table S4 (see ^{footnote 1}), sample 205 yielded two ages: 1718 ± 7.0 Ma (26 points used in the age calculation) and 1486 ± 11 Ma (18 points).

²⁰⁷Pb/²⁰⁶Pb point ages ranged from 980 Ma to 1855 Ma. Point ages cluster near 1000 Ma, 1500 Ma, and 1700 Ma (Fig. 11F). On a concordia plot, all clusters extend on chords toward the origin, with additional points spread between the three clusters and their chords. The older cluster has higher Th/U ratios (0.3–1.1) than the intermediate (0.1–0.7) and youngest (0.04–0.08) clusters (Fig. 12). As with sample 205, because of the diffuse age clusters, initial results do not yield robust age determinations. We again refined the initial estimates of the age clusters using the model of punctuated partial isotopic resetting. Our selection criteria for the point ages included in each calculation were the same as for sample 205, except with three clusters and chords instead of two. After the data partitioning, sample 106 yielded upper intercept ages of 1727 ± 13 Ma (16 points), 1468 ± 11 Ma (19 points), and 1004 ± 23 Ma (9 points).

The ²⁰⁷Pb/²⁰⁶Pb point ages ranged from 1068 Ma to 2229 Ma, with clusters near 1080 Ma, 1450 Ma, and a chord with an upper intercept on concordia of ca. 1750 Ma (Fig. 11B). The ca 1450 Ma cluster exhibits a chord toward the origin, and additional point ages are scattered between the clusters and the older chord. The older chord and intermediate age cluster have higher Th/U ratios (0.6–1.8 and 0.3–1.8) than the youngest cluster (0.04–0.1; Fig. 12). We refined the initial estimates of the age clusters as before. Our selection criteria for the point ages included in each calculation was the same as for sample 106. After the data partitioning, sample 201 yielded upper intercept ages of 1725 ± 24 Ma (6 points), 1459 ± 9 Ma (51 points), and 1080 ± 42 Ma (4 points).

The ²⁰⁷Pb/²⁰⁶Pb point ages ranged from 1023 Ma to 1928 Ma, with clusters near 1000 Ma and 1450 Ma, and a few nearly concordant point ages between ca. 1200 and 1750 Ma (Fig. 11C). The ca. 1450 Ma cluster exhibits a short chord toward the origin. The older cluster has higher Th/U ratios (0.5–1.3) than the younger cluster (0.03–0.09; Fig. 12). We refined the initial estimates of the age clusters as before using the same selection criteria for the point ages included in each calculation. After the data partitioning, sample 70 yielded upper intercept ages of 1455 ± 12 Ma (26 points) and a concordia age of 999 ± 14 Ma (2 points).

The ²⁰⁷Pb/²⁰⁶Pb point ages ranged from 1388 Ma to 1703 Ma, with a single cluster near 1450 Ma (Fig. 11D). The ca. 1450 Ma cluster exhibits a short chord toward the origin. Th/U ratios in the cluster range from 0.6 to 1.6 (Fig. 12). For the calculation of an upper intercept age, we eliminated discordant points that did not lie on the chord defined by the majority of points. Sample 71 yielded an upper intercept age of 1450 ± 10 Ma (16 points).

The ²⁰⁷Pb/²⁰⁶Pb point ages ranged from 950 Ma to 4408 Ma, with a cluster near 1450 Ma, a chord with an upper intercept on concordia of ca. 1750 Ma, and a small cluster at ca. 1000 Ma, at the end of a band of slightly discordant ages (Fig. 11E). The ca. 1450 Ma cluster exhibits a chord toward the origin. The older chord and intermediate age cluster have higher Th/U ratios (0.7–1.3 and 0.5–2.1) than the youngest cluster (0.05–0.9; Fig. 12). We refined the initial estimates of the age clusters as for sample 201. After the data partitioning, sample 108 yielded upper intercept ages of 1711 ± 18 Ma (6 points), 1455 ± 5 Ma (51 points), and 1007 ± 93 Ma (5 points).

Interpretation of U-Pb Ages

Using the U-Pb ages from samples of the granite and bounding pink orthogneiss to establish the approximate conditions of deformation yields a consistent data set, with common ages among the units of ca. 1720 Ma, 1465 Ma, and 1020 Ma. Although Th/U ratios are not definitive indicators of igneous versus metamorphic paragenesis, higher values tend to reflect crystallization from a magma and lower values from metamorphism (Hoskin and Schaltegger, 2003). The western orthogneisses exhibit the older two ages, and the eastern orthogneisses exhibit all three ages. The two interior granite samples have a strong dominance of the intermediate age and a few points in sample 70 of the youngest age. The marginal granite samples contain all three ages. This geographic age distribution and the Th/U ratios suggest the following history.

1. Both pink orthogneisses crystallized ca. 1720 Ma. We cannot determine whether they represent volcanic or plutonic rocks.

2. At ca. 1465 Ma, the Bad River granite intruded and crystallized, incorporating some zircon from the bounding orthogneisses, as evident in the presence of ca. 1720 Ma zircon in the margins but not in the interior. At that time, the orthogneiss was metamorphosed, as indicated by the presence of the ca. 1465 Ma age in the orthogneisses with lower Th/U ratios than the ca. 1720 Ma age.

3. At ca. 1020 Ma, during Grenville orogenesis, shear zones at the margins of the granite developed, resetting the zircon (with generally low Th/U ratios) that yielded those ages. The higher-strain granite samples (201 and 108) have a much more prominent presence of these younger point ages than do the interior granite samples. The orthogneiss samples were not substantially reset again at this time, though at least some point ages in 106 reflect the Grenville event.

The ages reported here complement the results of several previous studies spanning from the Central Gneiss belt to the Grenville Front (Krogh et al., 1966; van Breemen and Davidson, 1988; Haggart et al., 1993; Corfu and Easton, 2000). U-Pb dating of zircons indicates that the pink orthogneisses were emplaced ca. 1720 Ma. This is a similar age to those reported for the granitoids of the Killarney Complex (1742 ± 1.4 Ma; van Breemen and Davidson, 1988), Fox Islands (1685–1725), and Grondine Complex (1715 +6/–5 Ma; Davidson et al., 1992). The ca. 1465 Ma igneous crystallization of the Bad River granite and concomitant metamorphism of the adjacent orthogneiss places the granite intrusion as part of a regional magmatic event affecting Archean and Paleoproterozoic rocks (van Breeman et al., 1986; Ketchum et al., 1994; Krogh, 1994; Fueten and Redmond, 1997; Carr et al., 2000 and references therein) associated with the formation of a continental arc at the southeast margin of Laurentia (Slagstad et al., 2009). The granite therefore represents a monocyclic unit—i.e., undeformed prior to Grenville orogenesis—which is consistent with previously inferred episodes of tectonism in the area (Jamieson et al., 1995). Although analyses from the western pink orthogneiss did not yield Grenville ages, the structural coherence with the granite and adjacent units suggests deformation along with the granite toward the end of the Grenville orogeny at ca. 1020 Ma. Given the uncertainty of the dating, as well as other observations about the temporal and spatial pattern of strain localization (e.g., Jessell et al., 2005; Gardner et al., 2017), the shear zones on the eastern and western margins of the granite need not have formed synchronously.

The mineral assemblages within the granite and the enclaves are consistent with deformation in the amphibolite facies. Though our data do not allow more quantitative assessment of pressure-temperature conditions, Jamieson et al. (1995) describe nearby samples with a Grenville fabric deformed over conditions ranging from ∼740 °C, 1.1 GPa to ∼600 °C, 0.65 GPa.

In this section, we summarize the observations of microstructural and chemical changes across the strain gradients (Fig. 13), which we defined based on the macroscopic observations of enclave shape and foliation intensity. In the following section, we discuss what these changes imply about shear zone evolution.

The least strained sample in the Bad River granite has variably sized alkali feldspar and oligoclase porphyroclasts wrapped by elongate quartz, hornblende and/or biotite, and oxides. This morphology does not change much across both gradients because the recrystallized alkali feldspar and plagioclase are often found as monophase aggregates in augen-shaped pods, similar to the shapes of the porphyroclasts (e.g., Fig. 5A4). With increased strain, feldspars form part of a mixed-phase matrix between porphyroclasts and/or monomineralic pods.

Although the general augen-shape morphology of the samples persists, we highlight seven textural changes that occur with increased strain. (1) Grain-size reduction occurs in both feldspar and quartz (Table 2). The alkali feldspar and plagioclase porphyroclasts become smaller (and eventually disappear in the eastern gradient) reducing from ∼5 mm porphyroclasts to the recrystallized grain size of ∼250 μm. The recrystallized grain size of both feldspars and quartz grains also reduces across both gradients, with the caveat that the foam textures attest to some degree of static grain growth (e.g., Fig. 5A2). (2) In association with the grain-size reduction, the fraction of recrystallized feldspar increases with strain, from 70% to near 100%. (3) Quartz aggregates thin out to a ribbon morphology and become slightly more sinuous (except for sample 103; Fig. 8). (4) Bands of hornblende become more sinuous and continuous, except for sample 103 where hornblende is dispersed. (5) A CPO develops in quartz in both gradients. In the eastern gradient, the CPO strengthens with strain until the two easternmost, high-strain samples (108 and 103) where the quartz CPO weakens. (6) Garnet morphology is more skeletal with strain. (7) At the highest strain, phases are dispersed and mixed, rather than in monophase aggregates.

Whole-rock analyses of rare-earth elements indicate that the granite samples from both gradients all have the same REE pattern and therefore confirm that they are of the same igneous phase. Results of major-element (wt%) from bulk-rock chemistry of granite samples indicate some variation, but there are no significant trends that correlate with strain. Furthermore, the variations do not fall outside those reported from other plutons (e.g., Gray et al., 2008), and therefore the chemical variations (Figs. 4A and 4B) are most likely representative of natural heterogeneities in the granite body (e.g., Fig. 2B).

The two gradients show no quantifiable change in modal amounts of hornblende or biotite. However, Mg#, F, Cl, and Ti in both hornblende and biotite and Na in recrystallized oligoclase change with strain (Fig. 10). Mg#, Cl, and F all increase with strain, whereas Ti decreases with strain. Although the increases in Mg# are slight, they are discernable when normalized to variations in bulk chemistry across the transect. Furthermore, the increases correlate with a reduction in grain size and therefore an increase in grain boundary area, which facilitates mass transfer processes (White and White, 1981; Behrmann and Mainprice, 1987; Warren et al., 2008; Pec et al., 2012).

We now turn to the question of what processes led to the development of the strain gradients within the granite. We do not attempt to infer the processes leading to the finite strain in the bounding pink orthogneisses because we have little constraint on the finite strain, and we cannot track that unit’s structural history. Rather, we focus on the mechanical and chemical factors in the granite that facilitated strain localization along the granite margins. At the most basic level, the strain gradient indicates a mechanical contrast between the margins and the core of the granite, which requires weakening (cf. Platt, 2015) from what was arguably a homogeneous granite at the outcrop scale. As noted above, microscale changes across the gradient include the proportion of recrystallized material and therefore overall average grain size, strength of the crystallographic fabric in quartz, and mineral chemistry of Fe-Mg–bearing phases and plagioclase at the granite margin (Figs. 4 and 10). Bulk-rock composition and mineral mode are relatively constant. Although a stress concentration at the margins could have been sufficient to induce the microstructural change, the inferred bulk stress concentration at the margin of the granite is less than a factor of two higher than in the core, which is insufficient to account for the difference in finite strain (Shulman, 2016). Once the shear zone developed, the stress concentration would be even less, and likely negligible (Platt and Behr, 2011).

Given these constraints, the constitutive law must have varied between the granite core and margins. Across both gradients, microstructures are consistent with the operation of dislocation creep in the major phases. Evidence for dislocation creep includes a crystallographic preferred orientation in quartz, as well as subgrain development leading to mantled feldspar porphyroclasts (Fig. 5). At the granite margins, the changes in microstructure and mineral chemistry indicate the operation of additional deformation mechanisms: (1) A mixed phase matrix develops; this is not possible with dislocation creep alone but requires some additional mechanism, such as grain boundary sliding or diffusion creep (cf. Ashby and Verrall, 1973; Kruse and Stünitz, 1999; Kenkmann and Dresen, 2002; Warren et al., 2008; Oliot et al., 2014; Czaplińska et al., 2015; Gerbi et al., 2016). (2) The strength of the crystallographic fabric in quartz drops at the eastern granite margin, also consistent with grain boundary sliding and/or diffusion creep (e.g., Fliervoet et al., 1997; Kilian et al., 2011). (3) The change in Mg#, F, Ti, and Cl in biotite and hornblende and Na in plagioclase requires element mobility, which in turn requires that mass transfer processes (i.e., diffusion creep or diffusion-accommodated grain boundary sliding) were active in the higher-strain samples. Because recrystallization and grain-size reduction facilitate mass transfer in coarse-grained rock and are necessary precursors to grain boundary sliding and/or diffusion creep at moderate to high temperatures, it follows that at the onset of deformation, the whole granite unit was deforming at approximately the same rate (Fig. 14). Over time, as the microstructural changes occurred at the granite margins, the marginal rocks weakened, allowing strain to focus at the unit boundary. The chemical changes, particularly the drop in Ti in biotite toward the margins, suggest the last recorded deformation at the margin occurred at a lower temperature than the last recorded deformation in the central portions of the Bad River granite.

A deformation mechanism switch is not immediately apparent when viewing the microstructures. The recrystallized grain size of both quartz and feldspar is much larger than that typically associated with diffusion creep (Rutter and Brodie, 2004a, 2004b; Rybacki et al., 2006), and grain boundary sliding typically does not leave clear positive evidence, though phase mixtures—which develop toward the high-strain margins of the granite—are commonly taken to indicate its operation (e.g., Ashby and Verrall, 1973; Kenkmann and Dresen, 1998; Platt, 2015). However, because dislocation creep is not sensitive to grain size, without invoking additional deformation mechanisms, the bulk constitutive law for the granite cannot have undergone the necessary change (Czaplińska et al., 2015; Gonçalves et al., 2015; Gerbi et al., 2016). As such, although the granite margins do not provide abundant positive evidence for strain being accommodated by processes other than dislocation creep, the operation of other deformation mechanisms is required to produce the necessary weakening.

Strain-related weakening in the Bad River granite did not result from a single factor. Instead, the strain gradient that developed along the margins of the Bad River granite resulted from subtle interacting changes in microstructure that resulted in a deformation mechanism switch and consequent rheological weakening. Mineralogical change played little role. This contrasts with the Twelve Mile Bay and Parry Sound shear zones that border a granulite klippe, ∼100 km to the southeast. There, km-scale shear zones developed through hydration and a major change in mineral assemblage in intermediate to mafic rocks, with strong garnet and clinopyroxene replaced by weaker hornblende (Culshaw et al., 2010; Marsh et al., 2011). Because granites are not susceptible to comparable mineralogical change, other factors must drive shear zone development in those lithologies. The complexity of the microstructural interactions and uncertainty about stress concentrations that can lead to microstructural change provide strong challenges to a priori predicting the development of kilometer-scale shear zones in granitic middle orogenic crust.

The strain gradients of the Bad River granite provide an opportunity to document evidence for the microstructural and chemical evolution of middle orogenic km-scale granitic shear zones and further constrain the timing of the tectonic events within the Grenville Front tectonic zone. This is particularly valuable because granitic shear zones are common features of orogenic belts, and documentation of microstructural and chemical changes across such large-scale shear zones is relatively sparse in the literature.

U-Pb zircon geochronology indicates that the Bad River granite is a monocyclic unit, bounded by pink orthogneiss. It is unclear whether the pink orthogneisses are originally the same unit or not, but they both record a ca. 1720 Ma crystallization age and a ca. 1465 Ma metamorphic age associated with the intrusion of the Bad River granite. The shearing event associated with Grenvillian orogeny, recorded at ca. 1020 Ma in the granite and eastern orthogneiss, is not recorded in the western orthogneiss.

The granite records a relative strain gradient defined by a change in the aspect ratio of passively deformed igneous enclaves and by foliation intensity. Strain increases from the interior of the granite toward both margins. Major mineral mode, bulk composition, and most major-element mineral compositions are effectively constant across the strain gradient, but several other parameters change, including overall average grain-size fraction of recrystallized grains, phase morphology, quartz CPO strength, Mg#, F, Cl, and Ti concentration in biotite and hornblende, and Na in plagioclase. These observations taken together indicate that deformation mechanisms were not constant or uniform across the strain gradients, which led to the rheological weakening necessary for the strain gradients to develop. Similar microstructure-deformation mechanism-strength feedbacks are likely common within relatively non-reactive, large-scale shear zones but can be obscured in more chemically active shear zones where metamorphic processes may dominate. The evolution of the Bad River granite strain gradients reveals the subtle spatial and temporal dynamism of the microscale processes that can lead to the formation of kilometer-scale shear zones.

This work was supported in part by National Science Foundation grants MRI-0820946 and EAR-1150438, a Geological Society of America research grant, and the University of Maine Graduate School Government. Thank you to Won Joon Song, Siobhan Haggarty, Ruth Mares, Maura Foley, Stephanie Mills, Connor Scofield, and Anthony Feldman for their help in the field. Thorough reviews by Zach Michaels and Kevin Mahan, as well as editorial comments by Bob Miller, greatly improved this manuscript.