Introduction

The long-term stability of Archean cratons is often attributed to their thick, buoyant, and refractory lithospheric keels which are able to resist tectono-thermal modification and erosion once established1. Cratonic keels form through protracted histories of melt depletion related to geodynamic processes coincident with Archean crustal formation2. Subsequent refertilization may occur through metasomatism related to, for example, later subduction-accretion events, mantle plumes, and continental rifting3,4; compositional and rheological heterogeneities introduced through such processes can impact overall craton stability5. Thus, the combined history of a craton determines its compositional, chemical, and thermal states as observed in the present-day architecture of the crust and underlying mantle.

Continental-scale magnetotelluric (MT) and seismic surveys have been used to investigate cratonic lithosphere as the physical properties they are sensitive to are able to identify compositional and thermal heterogeneities, as well as fluctuations in hydration levels4,6. Typical cold, depleted cratonic lithosphere is characterized by high seismic velocities and electrical resistivities. Conversely, low velocities and resistivities may be indicative of metasomatized lithosphere containing elevated hydrogen concentrations within nominally anhydrous minerals (NAMs), accessory minerals such as fluorine-rich phlogopite7,8,9, and/or interconnected grain-boundary graphite4,10. The complex deformational histories of cratonic regions often result in seismic anisotropy, which is typically attributed to lattice-preferred orientation of the minerals that develop due to strain accumulated during craton construction and/or present-day plate motion11,12. However, the generation and manifestation of electrical anisotropy within the cratonic lithosphere remains enigmatic. Anisotropy within the asthenosphere13 and lithosphere14 of tectonically active regions is generally ascribed to some combination of interconnected melt and aqueous fluids – mechanisms which are not applicable within cold, stable lithosphere4.

The Superior Craton of Canada is the world’s largest Archean craton and forms the core of the North American continent (Fig. 1). The southwestern portion of the Superior Craton (hereafter referred to as the SW Superior) is comprised of largely ~east-trending terranes, generally thought to have been amalgamated at ~2720–2680 Ma via a series of accretionary events15. Subprovinces of the SW Superior are generally young to the south, with the northern-most subprovinces forming an ancient proto-cratonic core against which the constituent terranes or subdomains accreted (Fig. 1b). Post-accretionary magmatism and deformation are recorded throughout the SW Superior16,17, with the cessation of major tectono-thermal events occurring at ~2600 Ma. The SW Superior remained internally stable through Paleoproterozoic episodes of continental growth along its margins including the 1900–1800 Ma Trans-Hudson Orogeny3 (THO) and Penokean Orogeny18. The THO resulted in ~25 km of east-west shortening across the crust of the SW Superior19. Roughly coeval with the THO-related deformation was the emplacement of the circum-Superior large igneous province (CSLIP)20, a series of largely mafic-ultramafic belts which were emplaced around the edges of the Superior Craton, with only minor dike intrusions within the craton interior. Due to the synchronicity and localization of magmatism to the craton margins, the CSLIP is inferred to be the result of a mantle plume which was deflected by the thick, refractory cratonic keel beneath the Superior Craton20,21. The most recent tectono-magmatic event that may have affected the SW Superior is the ~1.1 Ga Mid-Continent rift (MCR), a failed rift system which nearly fractured the North American Craton22. While MCR-related magmatism was largely focused along the two rift arms south of the Superior Craton, contemporaneous emplacement of igneous rocks that form the Nipigon Embayment (Fig. 1b) suggests that this region may represent the third rift arm of an MCR triple-junction23.

Fig. 1: Station and phase tensor maps of the study area.
figure 1

a Map of North America (gray) and the Superior Craton (dark gray). Red circles indicate inferred focal points of Proterozoic mantle plumes (after ref. 67). White diamonds indicate xenolith suites (K – Kirkland Lake; T – Tamiskaming; V – Victor; KL – Kyle Lake). b Map of the western Superior (modified from ref. 2.) including the MT stations used in this study. Solid black lines denote craton and subprovince boundaries. Black dashed lines indicate the primary area of interest shown in panels c-e and in Fig. 2. NE – Nipigon Embayment; MCR – Mid-Continent Rift. Panels c-e show phase tensor ellipses shaded by phase-split at periods of c 160 s, d 640 s, and e 1280 s.

Full size image

The lithosphere of the SW Superior has been studied extensively using a variety of seismic methods. Several studies have revealed a high-velocity lithosphere, generally interpreted as ancient and strongly depleted, between the western cratonic margin and the Nipigon Embayment. Greatly reduced seismic velocities beneath and to the east of the Nipigon Embayment were inferred to be the result of lithospheric modification during the MCR12,24,25. MT stations deployed as part of the Lithoprobe project revealed a similar signature in MT phase data around the Nipigon Embayment26; however, the lithospheric-scale resistivity structure of the SW Superior remains enigmatic. Recent 3D modeling studies revisited this dataset, revealing curvilinear conductive and resistive bands which trend sub-parallel to major terrane boundaries, interpreted as the combined signatures of the accretionary processes which formed the crust of the SW Superior and subsequent post-orogenic extension and magmatism27. However, the modeled resistivity structure within the lithospheric mantle was dominated by ~N-S trending bands of low and high resistivity inferred to be artefacts of electrical anisotropy, obfuscating interpretation at mantle depths.

Here, we present an analysis of MT data and a 3D anisotropic inversion that reveals the resistivity structure of the lithospheric mantle of the SW Superior. The resulting resistivity volume identifies a layer of high electrical anisotropy at 100–200 km depth, spatially coincident with a region of strong seismic anisotropy as resolved by shear-wave splitting25 and seismic anisotropy tomography11. Considering the electrical, seismic, and geothermal properties of the region, the anisotropic layer is interpreted as preferentially north-south-oriented phlogopite-bearing domains emplaced within the lithospheric mantle as a result of Proterozoic metasomatism and subsequent overprinting as a result of MCR-related magmatism.

Results

Phase tensor analysis

Evidence for electrical anisotropy may be found within the MT data, typically visualized as phase tensor maps28,29. The phase tensor30,31 is a 2 × 2 tensor, Φ, that can be derived from the MT impedance data and that is independent of local site-dependent galvanic distortions. The parameters φmax and φmin define the maximum and minimum principal phases of Φ. Phase tensors may be plotted as ellipses with major and minor axes corresponding to φmax and φmin respectively, and an azimuth of θ (the generalized strike of Φ), where the orientation and shape of the ellipse is indicative of the direction of preferred current flow31, with a 90° ambiguity. Phase tensors with large differences between φmax and φmin (a ‘phase-split’, defined as φmax - φmin) will plot as thin ellipses and are indicative of a strong contrast in conductivity in orthogonal directions as a result of complex 2D or 3D structures and/or electrical anisotropy28.

Several distinct subsets of phase tensor ellipses are evident across the SW Superior at a period of 100–640 s, corresponding to investigation depths within the lithospheric mantle; ellipses at 160, 640, and 1280 s are shown in Fig. 1c–e. Within the northern-most subprovinces, ellipses have varying orientations but generally low phase-splits ( < 10°). A sharp transition occurs at a northing of ~200 km, south of which most ellipses from the western cratonic margin to the Nipigon Embayment have E-W orientations and high phase-splits ( > 30°), corresponding to average values of φmax > 80° and φmin < 50. Phase tensors for the region surrounding the Nipigon Embayment are rotated slightly counter-clockwise, with overall lower phase-splits in the southern half ( ~10°) and higher values in the northern half (20°–30°).

Large phase-splits and low induction arrow magnitudes on lateral scales comparable to the depth of the investigated volume are often cited as evidence for electrical anisotropy within MT data29. The highly polarized phase tensor ellipses (Fig. 1c–e) and induction arrow data (Fig. S7) within the SW Superior satisfy these conditions. Furthermore, isotropic inversion of the data revealed ~100 km wide alternating bands of high and low resistivity at depths of 100–200 km27, a signature which may also be indicative of electrical anisotropy32. As there is no additional geological or geophysical evidence to support the existence of such banded resistivity structures, we assert that the lithospheric mantle of the SW Superior is electrically anisotropic.

3D anisotropic modeling

Isotropic inversion of the MT data27 was performed using the ModEM inversion algorithm33. The data were re-inverted using the tri-axial anisotropic inversion code MT3DANI34, an extension of ModEM capable of inverting for the principal resistivities along the north-south, east-west, and vertical axes (ρx, ρy, ρz, respectively) for fixed strike, dip, and slant angles. As vertically propagating plane waves primarily induce horizontal currents within the Earth, MT data have good sensitivity to ρx and ρy, while ρz is generally poorly resolved34. Similarly, MT data are primarily sensitive to the anisotropic strike angle and, except in extreme cases, less sensitive to the slant and dip angles32. Thus, for simplicity, the slant and dip angles were set to zero. The azimuth of the phase tensors (Fig. 1c–e) varies slightly by period and geographic location, however, the average azimuth in the primary area of interest is approximately N-S, and therefore, the strike angle was set to zero. Furthermore, while all three principal resistivities were allowed to vary within the inversion, only ρx and ρy are used for interpretation.

The isotropic model was used as a basis for the starting model with a few modifications. The crustal structure was retained to a depth of 56 km; model depths from 100–410 km were then set to \({\rho }_{x}={\rho }_{y}={\rho }_{z}=\) 500 Ω·m, and the 56–100 km depths were vertically smoothed to allow for a gradual transition between the isotropic heterogeneous crustal structure and the mantle half-space. As in the original isotropic model, depths below 410 km were set to 20 Ω·m, representing the expected resistivity drop at the olivine-spinel phase transition35.

Depth slices and E-W cross-sections through the 3D anisotropic and original isotropic resistivity models are shown in Fig. 2. At crustal depths the anisotropic model is approximately isotropic (ρx≅ρy) with similar resistivities and geometries to that of the original isotropic model27, and so are not shown nor discussed here. The lithospheric mantle of the SW Superior is characterized by moderate to high degrees of electrical anisotropy, with low and high resistivities along the N-S and E-W orientations, respectively. Depths of 100–250 km are generally resistive ( > 1000 Ω·m) along the E-W axis (ρy) throughout the model (Fig. 2a). Resistivities along the N-S axis (ρx) are considerably lower; a large conductive body, C1, (ρx of 10–100 Ω·m) extends from the western edge of the craton to just west of the Nipigon Embayment (Fig. 2b). The region defined by C1 has a corresponding anisotropic ratio (defined as \(\frac{{{{{\rm{\rho }}}}}_{y}}{{{{{\rm{\rho }}}}}_{x}}\)) of 10–100 (Fig. S1). Anisotropic ratios decrease to ~10 within C2, consistent with the observed decrease in phase-splits around the Nipigon Embayment (Fig. 1c–e). In this region, ρy remains relatively high ( > 1000 Ω·m) while ρx increases to 100–300 Ω·m (C2). The top of the anisotropic layer varies, but on average is at a depth of 100–120 km. Above this depth, the uppermost mantle of the SW Superior is moderately resistive (ρx, ρy > 500 Ω·m), with geometries that largely reflect smearing / smoothing of lower crustal structures (see Supplemental Fig. S8–11) due to the inability of MT data to image directly beneath highly conductive bodies36.

Fig. 2: Slices through the anisotropic resistivity model.
figure 2

Plan view slices at 155 km depth and E-W slices through the preferred anisotropic model. a the E-W oriented ρy resistivities, b N-S oriented resistivity ρx, and c through the preferred isotropic resistivity model (from ref. 27). East-west slices are taken through the red dashed lines in (a), (b), and (c). All slices show only the primary area of interest (dashed box in Fig. 1b).

Full size image

Discussion

A preferred direction of current flow (i.e., electrical anisotropy) may arise due to microscopic or macroscopic effects. Microscopic anisotropy is a result of a preferred direction of conduction within the crystal structure of the constituent minerals as a result of, e.g., hydrogen diffusion along a preferred crystallographic axis37. Conversely, macroscopic anisotropy (within the context of MT) describes the phenomenon of preferentially oriented current flow due to the alignment of conductive macroscopic structures (e.g., dikes, veins, shear zones, foliation/lamination) at scales which are below the resolving power of MT38. While it is impossible to definitively determine whether MT data are affected by electrical anisotropy (due either to micro- or macroscopic effects) or if the responses are the result of complex 3D structure28, evidence from MT data may be combined with geological reasoning to determine which scenario is most likely34,39.

We first consider the possibility of microscopic effects to explain the high degree of electrical anisotropy observed within the SW Superior. As intragranular melts and free fluid phases are unlikely within stable cratonic lithosphere4, the primary mechanism for microscopic electrical anisotropy within cratonic lithosphere is hydrogen diffusion along a particular crystallographic axis within NAMs. A preferred axis of diffusion can occur within minerals that have developed lattice-preferred orientation as a result of, for example, strain accumulated during subduction-accretion events and/or from the present-day lithospheric plate motion and mantle flow40. Laboratory studies on variably hydrated single-crystal olivine samples report hydrogen diffusion along the preferred conductive axis can result in electrical anisotropy factors of >10 at mantle conditions35,41. However, lithosphere, which has experienced realistic degrees of deformation is likely to be composed of an aggregate network of minerals with imperfect alignments, resulting in a much lower bulk anisotropic factor (< 4.5) compared to that of a single crystal37,42,43. Furthermore, high seismic velocities observed throughout most of the lithosphere of the SW Superior are inconsistent with the high hydrogen contents (i.e., near saturation levels) which would be required for observable electrical anisotropy12,24,25. Thus, it is unlikely that lattice-preferred orientations within hydrated mantle minerals contribute significantly to the observed electrical anisotropy.

The electrical anisotropy within the SW Superior is therefore inferred to be indicative of macroscopic structures at scales below the resolving power of MT data. At the depth range considered, MT data are sensitive to structures on the order of a few tens of kilometers; however, it is assumed that at lithospheric mantle depths and pressures, these ~N-S structures most likely correspond to channels where conductive phases were focused along intragranular spaces44 and/or isolated to <1 cm wide metasomatic dike channels45,46. Accordingly, the modeled ρy is taken to represent the approximate resistivity of the ‘normal’ depleted cratonic lithosphere, while ρx corresponds to conductive phases emplaced along ~N-S oriented structures. Typical cratonic lithosphere is highly resistive due to progressive depletion through a protracted history of high-grade metamorphism and melting events1,4. Metasomatism due to tectono-magmatic events (e.g., subduction, mantle plumes) and/or passive asthenospheric melt percolation into the lithosphere47 can lead to a locally refertilized mantle. These regions may be detectable using MT data due to elevated concentrations of some incompatible elements which have greatly reduced resistivities relative to that of depleted lithosphere, such as hydrogen and fluorine7,8,9, and/or carbon in the form of interconnected graphite precipitated along grain boundaries10. However, the stability and resistivities of such phases depends on the in-situ temperatures and pressures, and consequently the thermal conditions of the region must be considered.

Xenolith sample locations are relatively sparse and distal from the study area considered here (Fig. 1a), however, those that are available suggest that the Superior Craton has a relatively cold, largely lherzolitic lithospheric mantle48. Analysis of the seismic velocity structure of the whole craton suggests that these interpretations are equally applicable to the SW Superior49,50. Thus, a series of resistivity-depth profiles were generated for lherzolitic lithosphere at generalized cratonic geotherms51 corresponding to surface heat flow values of 33–40 mW/m2 using MATE modeling software52. An approximate geotherm for the SW Superior was determined by replacing ρy (assumed to be representative of the depleted lithosphere) with the generated resistivity profiles and comparing the modeled response to the observed MT data (see Supplemental Fig. S12, S13). A geotherm corresponding to a surface heat flow value of 37 mW/m2 (Fig. 3a; black line) was found to best fit the MT data and is consistent with geotherms and lithosphere thicknesses ( ~ 265 km) derived from seismic data for the region49,50; therefore, this geotherm is adopted here.

Fig. 3: Geothermal conditions and resistivity-depth profiles.
figure 3

a Approximate geothermal conditions at present-day (black line) and at 2700 Ma (red line; from ref. 62) overlaid by the stability fields of graphite and phlogopite. Black dashed line indicates the estimated lithosphere-asthenosphere boundary (LAB) depth. b Volume-averaged resistivity-depth profiles within the C1 and C2 anomalies overlaid by calculated resistivity-depth profiles for lherzolite (Lhrz.) with varying water content, phlogopite (Phlg.) content, and phlogopite interconnectivity (m = 1 is perfectly connected, m = 2 is partially connected).

Full size image

The obtained geotherm places constraints on the range of plausible mechanisms for generating the observed electrical anisotropy, and in particular the modeled values of ρx. Refertilized mantle enriched in carbon (precipitated as grain boundary graphitic films) and/or some incompatible elements (e.g., hydrogen, fluorine) are among the most common explanations for low resistivities within cratonic lithosphere4,9. However, unless present as relatively thick grain boundary films ( > 100 nm), graphite may not be a viable conductor at upper-mantle conditions53. Furthermore, graphite becomes unstable at depths below ~125 km at the considered geotherm (dashed green line in Fig. 3a) and is therefore unable to explain the C1 nor C2 features. Hydrogen contents near saturation levels within NAMs may reach resistivities low enough to explain the C2 anomaly surrounding the Nipigon Embayment, however, predicted resistivities for saturated lherzolite are up to an order of magnitude higher than those imaged within C1 (solid blue line in Fig. 3b). Conversely, metasomatized lithosphere containing realistic volumes ( < 10%)54 of variably interconnected fluorine-rich phlogopite is able to reproduce the range of ρx found within both C1 and C2 (Fig. 3b). Phlogopite is a common indicator of mantle metasomatism55 and has been observed within xenolith samples across the Superior Craton48. Thus, phlogopite emplaced along N-S channels provides the most plausible explanation for the observed electrical anisotropy within the SW Superior.

It is unclear what volume fraction of the affected lithosphere might be comprised of phlogopite-bearing channels. While it is fundamentally impossible to discern this from the MT data alone56, seismic tomography models provide some information. As phlogopite is a seismically slow mineral6,57,58, the observed high isotropic compressional24 and shear-wave59 velocities suggest the bulk phlogopite content of the lithosphere must be relatively low. Macro58 calculations indicate that lherzolitic lithosphere composed of 10–20% by volume of metasomatized channels each containing 5% phlogopite (for a total bulk phlogopite volume fraction of 0.5–1%) would result in a 0.3–0.6% reduction in shear-wave velocity relative to that of lherzolite. Low velocity channels result in a fast direction parallel to the channels; therefore, these estimates are broadly consistent with shear wave and full-waveform tomography results (which has high resolution of vertical variations in structure)60 that image an anisotropic layer with a ~N-S fast direction and anisotropic strength of ~0.5% at a depth of 100–200 km11,59. The estimated 10–20% channel volume within the lithosphere is consistent with that for similar seismically inferred mantle ‘dyke stockworks’ within the Slave Craton44. This interpretation is also similar to the mechanism proposed to explain mid-lithospheric seismic discontinuities globally6,57, with the notable difference that the channelized metasomatism proposed here would result in considerably weaker seismic signals than those produced through pervasive, isotropic metasomatism.

The lithosphere underlying the Nipigon Embayment corresponding to C2 (Fig. 2b) is coincident with anomalously low seismic velocities at depths of 100–200 km (as imaged via studies with high lateral resolution)24,25 and an overall reduction in shear-wave splitting relative to the high split times observed elsewhere in the SW Superior25, both of which are inferred to be a result of MCR-related activity. Elevated ρx and relatively stable ρy within the region further suggest that the bulk lithosphere is similar to that found elsewhere within the SW Superior, while the properties of the N-S channels vary relative to those corresponding to C1. Synthetic modeling using MATE (Fig. 3b) suggests that the ρx for C1 may be matched using phlogopite volumes of up to 5% (with variable interconnection), while the resistivities for the less conductive C2 may be fit by hydrated lherzolite, or phlogopite at lower (1–2%) volumes and/or decreased interconnection. Thus, we suggest that the N-S channels in C1 pre-date the MCR and that MCR-related heating and/or melts may have preferentially affected the composition of the material between the channels, resulting in disconnection, replacement, and/or overprinting of phlogopite with more resistive, seismically slow minerals, as seen within C2.

The timing and geodynamic processes leading to the inferred channelized metasomatism require further investigation. The apparent modification of the metasomatized channels by MCR-related activity at ~1100 Ma places a lower bound on the age of the N-S channels. Furthermore, while the exact thickness of the anisotropic layer is poorly resolved due to the shielding effect of conductors on MT data36, model sensitivity testing suggests it likely extends from ~100 km to at least 160–190 km. Phlogopite is stable61 to 1200 °C–1400 °C, well within present-day cratonic conditions at the relevant depth range; however, geothermal conditions of the SW Superior were significantly hotter following its formation than they are today62, precluding the formation and long-term stability of phlogopite at depths greater than ~140 km based on the assumed Archean geotherm (Fig. 3a). Secular cooling of the lithosphere at an average rate of 100 °C–150 °C/Ga63,64 increases the maximum depth of stability of phlogopite by 20–40 km/Ga. Thus, while initial formation of the metasomatized channels at depths of 100–140 km may have occurred during or shortly after the formation of the SW Superior at ~2700 Ma, 500–1000 Ma of lithospheric cooling is required for phlogopite to remain stable at depths of 160–190 km, placing a conservative upper bound on the age of emplacement at ~2200 Ma.

Mantle metasomatism and phlogopite emplacement has been linked to a variety of geodynamic processes including subduction8,65, mantle plume activity and/or melt pooling along the lithosphere-asthenosphere boundary (LAB)9, and lithospheric delamination7. However, these processes are generally inferred to lead to pervasive metasomatism across the affected regions and correspondingly isotropic geophysical anomalies. Therefore, it is unclear whether the channelized metasomatism revealed within the SW Superior requires a pre-existing or contemporaneously generated lithospheric fabric66 to focus melts and fluids into subvertical, N-S structures, or if such channels may nucleate and propagate spontaneously via recurrent and/or focused melt interaction at the base of the lithosphere6,47. In either case, consideration of the evolving geothermal conditions of the SW Superior following its construction and the stability of phlogopite discussed earlier suggests the locus of tectono-magmatic activity which occurred along the margins of the western Superior during the mid-Proterozoic (2200–1800 Ma), including the THO and Penokean orogeny as well as several mantle plumes (Fig. 1), provides the most plausible explanation.

Early plume activity (e.g., ~2100 Ma Marathon-Fort Frances and ~2000 Ma Minto events; Fig. 1a) may have had a role in the generation of the metasomatized channels, though rapid cooling of the lithosphere after ~2700 Ma would be required for phlogopite to remain stable throughout the considered depth interval. Thus, the preferred interpretation involves impingement of a mantle plume along the northern border of the western Superior, possibly related to the emplacement of the CSLIP20,67, occurring contemporaneously with the onset of ocean closure along the western and northwestern borders of the craton during the THO68. If a N-S fabric is required for channelization of metasomatic fluids, it may have been formed as a result of THO-related E-W shortening and coeval mantle plume activity. In this case, fluid infiltration and resultant metasomatism may have initiated at the base of the lithosphere in a distributed manner (i.e., pervasive metasomatism without a specific preferred orientation), with coeval E-W shortening leading to preferential localization of fluids in N-S oriented domains perpendicular to the shortening axis. Such a process would lead to a positive feedback system wherein preferential fluid flow (and precipitation of conductive minerals, which result in the observed anisotropy) would progressively channelize within the otherwise depleted, refractory lithospheric mantle69,70.

Termination of the highly anisotropic zone (C1) near the southern boundary of the oldest terranes suggests that the thickened, resistant core of the craton may have served to deflect incoming melts towards the margins of the craton1, while pooling of melts along the LAB beneath the SW Superior could result in gradual percolation of metasomatic fluids or progressive upwards migration of melt along preferentially N-S oriented domains controlled by far-field stresses (Fig. 4a). Alternatively, or possibly in conjunction with plume activity, subduction beneath the SW Superior during the THO68 and Penokean orogeny18 may have provided an additional source of metasomatic fluids into the lithosphere. With the exception of the Pickle Crow dike swarm21, there is little evidence of Proterozoic magmatism within the interior of the SW Superior. This is consistent with the apparent termination of the anisotropic layer at ~100 km depth, suggesting that upwards migration of metasomatic fluids was inhibited in the mid-lithosphere, likely due either to the stalling and crystallization of ascending melts at solidus conditions6,57 or an inability to breach an impermeable boundary layer (e.g., an eclogite layer inferred at a depth of 100–120 km50). The former seems to be the more likely explanation, as the subsequent ~1.1 Ga MCR magmatism responsible for disconnection and overprinting of the precipitated phlogopite (Fig. 4b) also resulted in the emplacement of a significant volume of magmatic rocks within the crust22.

Fig. 4: Schematic diagram of the interpreted structure and mechanisms of emplacement.
figure 4

a A ca. 1.9 Ga mantle plume impinges along the thickened, depleted lithosphere which comprises the northern cratonic core and is deflected southward. Pooling of melts at the base of the LAB results in gradual percolation of metasomatic fluids into the lithosphere forming phlogopite, possibly channeled into N-S weak zones which formed as a result of E-W compression during the Trans-Hudson Orogen. b Subsequent melting and magmatism related to the ca. 1.1 Ga mid-continent rift (MCR) overprints the phlogopite within the N-S channels and refertilizing the surrounding lithosphere.

Full size image

The imaged resistivity structure is well-constrained within the primary study area, however, additional long-period MT data across the Superior Craton are required to determine the full spatial extent of the anisotropic region, and thus further constrain the timing and geodynamic processes responsible for the observed anisotropy. If similar geo-electric signatures are observed within other preserved cratonic regions, then they may represent processes that are critical to cratonic evolution. Indeed, MT data with similar characteristics (e.g., highly polarized phase tensors at periods corresponding to the lithospheric mantle) have been observed in cratons globally10,71,72, with recent work utilizing isotropic inverse and forward modeling of MT data within the Gawler Craton of Australia73 suggesting similar lithospheric heterogeneities to those imaged here. Such heterogeneities resulting from modification of stable lithosphere can facilitate the onset of craton instability and ultimately destruction5,74. Conversely, local networks of channels such as those inferred here may not significantly impact cratonic rheology compared to widespread pervasive metasomatism75, and the introduction of fluids into deep lithospheric shear zones has been suggested as a contributing factor to the long-term stability of cratonic lithosphere76. Thus, further investigation of new and legacy MT data using modern analysis and 3D anisotropic inversion techniques is needed to better resolve whether electrical anisotropy is a common feature of cratonic lithosphere. These preserved geophysical signatures have significant implications for the structure and long-term stability of cratonic domains, and in determining whether stable cratonic lithosphere persists in part as a result of, or in spite of these heterogeneities.

Methods

Data acquisition

This study uses a composite MT data set combining stations from surveys undertaken from 1997-2022, including Lithoprobe26, USArray77, and Metal Earth78,79. In total, 332 long-period (4–30,000 s) measurements from the Lithoprobe and USArray are used, supplemented by 42 broadband (0.01–3000 s) measurements from Lithoprobe and Metal Earth to infill gaps where long-period measurements are not available (Fig. 1b). As the depth resolution of an MT inversion depends on both the inverted periods and the aperture of the survey36, stations to the north and south of the primary area of interest were included in order to maximize the lateral extents of the survey. The results and subsequent discussion are focused on the central survey region (dashed box in Fig. 1b).

MT inverse modeling

The 3D anisotropic resistivity volume was generated via regularized inversion. Geophysical inversion is typically an underdetermined problem in which there are infinite solutions for a given set of data points and corresponding uncertainties. A regularization parameter is used to enforce spatial smoothness within the model and thus constrain the possible solutions to those which are more physically realistic. Anisotropic inversion may be further regularized via an additional parameter which controls smoothness (or similarity) across the principal resistivities (see Supplemental Fig. S15).

The 3D anisotropic volume was obtained by jointly inverting the available impedance (Z) and vertical magnetic (K) transfer functions. A set of 23 logarithmically spaced periods from 8–15385 s was used for Z; the inverted K data were restricted to a maximum period of 5128 s to limit the effects of a non-planar source field. The inversion used a 144 × 141 × 72 mesh with a nominal cell size of 10 × 10 km. Sea water (e.g., Hudson’s Bay) was included in the starting model by setting the resistivities of the corresponding cells to 3.2 Ω·m. The initial model had a normalized root-mean square (nRMS) misift of 2.4 and converged to a final nRMS of 1.4 after 130 iterations, with improved data fit compared to that achieved by the isotropic model (nRMS of 1.64; see Fig. S4). The final model was able to reproduce the observed phase tensor and induction arrow patterns; further details on the inversion procedure, data fit, and sensitivity analysis are available in the Supplemental Material as well as in ref. 27.