Carbonated mantle lithosphere from the Western Canadian Cordillera Open Access

Mantle-derived xenoliths are commonly sampled and transported to the surface in mafic alkaline magmas and represent direct samples of the Earth’s lithospheric mantle (Pearson et al. 2003; Russell and Jones 2023 and references therein). Beneath mobile belts, such as the Canadian Cordillera, the mantle lithosphere is relatively thin (<35 km) and warm (800–1200 °C; Hyndman 2017 and references therein) and dominated by spinel peridotite. In addition to the major mineral phases, the mantle lithosphere can host accessory minerals, including amphiboles (e.g., Ghent et al. 2019), apatite, phlogopite (e.g., Canil and Scarfe 1989), sulphides (e.g., Delpech et al. 2012; Rielli et al. 2022), and carbonates (Yaxley et al. 1991; Rudnick et al. 1993; Ionov et al. 1996; Ducea et al. 2005). These accessory phases can be important indicators of past metasomatic events (i.e., pervasive fluid or melt enrichment) that influence the mantle solidus, and hence, control mantle melting (e.g., Francis and Ludden 1995; Laurora et al. 2001). Such phases have also been used to estimate mantle volatile budgets and to inform on the transport and mobilization of metals within the mantle lithosphere (e.g., Rielli et al. 2022).

Primary (i.e., mantle-equilibrated) carbonate is stable under typical upper mantle redox conditions, however, is rarely preserved due to rapid decarbonation during xenolith entrainment and transport of carbonated peridotite (e.g., Canil 1990). Nevertheless, primary carbonate has been reported for mantle xenolith suites deriving from a variety of tectonic settings, including active or paleo-subduction zones (e.g., Laurora et al. 2001; Demény et al. 2004; Ducea et al. 2005), intraplate settings (e.g., Moine et al. 2004), and rift margins (e.g., Lee et al. 2000; Perkins et al. 2006). Most primary carbonates are considered to result from enrichment events shortly before eruption, due to features indicative of textural/chemical disequilibrium (e.g., Lee et al. 2000; Demény et al. 2004; Moine et al. 2004; Ducea et al. 2005; Ionov et al. 2006). Even where xenolith-hosted carbonate has been shown to be chemically and isotopically equilibrated with mantle (e.g., Ionov et al. 1996; Yaxley et al. 1998), original textures are rarely preserved and have been modified during magma transport.

Here, we report on a new suite of lithospheric mantle xenoliths preserved in a mafic alkaline dyke exposed at Mt. Preston in western British Columbia (BC; Fig. 1) and described by Peterson et al. (2006) and Peterson (2010). This occurrence is distinguished, relative to others in the Canadian Cordillera, because it is located at the westernmost margin of the Intermontane Belt, close to the Coast Belt margin (e.g., Wheeler et al. 1991; Fig. 1). These mantle xenoliths, therefore, inform on the thermal and compositional state of the mantle lithosphere underlying this under-represented portion of the Canadian Cordillera. Furthermore, these xenoliths are unique because they preserve mantle-equilibrated (i.e., primary) carbonate in textural and chemical equilibrium. The suite of peridotite xenoliths provides direct evidence for, and the origins of, carbonate-melt metasomatism of the mantle lithosphere underlying this region of the Canadian Cordillera.

The Canadian Cordillera is an amalgamation of terranes accreted onto the western margin of North American during Middle Jurassic (∼185 Ma) to Late Cretaceous (90–85 Ma, Insular Belt rocks) time (Monger et al. 1982; Gehrels et al. 2009). The terranes comprising the Intermontane Belt have been interpreted as related fragments of a late Paleozoic to early Mesozoic island arc (Stikine and Quesnel terranes) and its associated accretionary complex (Cache Creek terrane). The Intermontane Belt is separated from the Insular Belt to the west by the Coast Belt, comprising the Coast Mountains Batholith (or Coast Plutonic Complex), the roots of a Middle Jurassic to Eocene magmatic arc (van der Heyden 1992; Monger et al. 1994; Gehrels et al. 2009), and related rocks (Fig. 1).

Mafic volcanic centers are ubiquitous in the Canadian Cordillera, and they commonly sample the lithosphere underlying their eruptive localities (e.g., Edwards and Russell 2000; Canil and Hyndman 2023 and references therein). Collections of xenoliths from these occurrences have supported numerous studies of the Cordillera’s mantle lithosphere (e.g., Canil and Scarfe 1989; Peslier et al. 2000, 2002; Harder and Russell 2006; Francis et al. 2010 and references therein). Most evidence for metasomatic events within the mantle lithosphere underlying the Canadian Cordillera is indirect (i.e., cryptic metasomatism) and based on geochemical compositions of peridotite xenoliths (e.g., Francis and Ludden 1995; Peslier et al. 2002). Rare direct evidence for mantle metasomatic events within the Cordilleran mantle derives from the presence of accessory phlogopite (Canil and Scarfe 1989) or pargasitic amphibole (Ghent et al. 2019) within peridotite xenoliths.

A xenolith-bearing dyke outcrops on a ridge ∼3 km southeast of Mt. Preston at 53°13′3″N, 126°41′58″W (Fig. 1). The near-vertical dyke strikes 145° and is exposed at the ridge crestline where it can be traced continuously down a steep south-facing slope for ∼110 m (Fig. 2) and several hundred metres further on an inaccessible vertical face. Contacts with the wall rocks are sharp and the dyke varies in thickness from ∼1 to 2 m where both contacts are visible; downslope the dyke is as wide as 7 m.

The dyke is aphyric, holocrystalline, moderately vesicular (5%–30%) and features an aphanitic groundmass (Fig. 2). The groundmass mineralogy comprises ∼40 vol.% plagioclase, ∼25% olivine (≤0.5 mm), ∼35% clinopyroxene (≤0.2 mm), and ∼3% of ≤0.1 mm subhedral to anhedral crystals of ulvöspinel. Magmatic carbonate, identified as near end-member calcite, occurs as ≤0.5 mm laths and patches in the groundmass, and as larger subhedral grains (≤1 mm) filling or lining vesicles (Figs. 2D–2E). Other minor phases include alkali feldspar, nepheline, and apatite.

Major element compositions and trace and rare earth element contents of the mafic dyke are reported in Supplementary material (Table S1). Chemically the dyke is a basanite, has an SiO₂ content of ∼43–45 wt.%, an Mg# of 62, and contains 5 wt.% normative nepheline. The basanite has a calculated liquidus of ∼1250–1300 °C (MELTS modelling; 15 kbar, QFM, 0–1 wt.% H₂O; Asimow and Ghiorso 1998). A single sample of the dyke was dated by ⁴⁰Ar/³⁹Ar methods and returned a plateau age of 18.72 ± 0.26 Ma (±2σ) representing a Neogene (early Miocene) crystallization age (see Supplementary material S2).

Mantle-derived peridotitic xenoliths are especially abundant in the outcrops situated 4–8 m below the ridge crest where the dyke narrows to ∼1–1.5 m in width (Figs. 2A and 2B). At this location, the xenoliths constitute 50%–80% of the dyke and are concentrated enough to be locally clast-supported (Fig. 2B). The mean diameter of xenoliths at this locality is ∼15 cm and the largest xenolith measured was ∼40–50 cm in diameter. Over 150 xenoliths were collected from the dyke and a representative group of 51 xenoliths studied in detail. The peridotite xenoliths are dominated by lherzolite, well-preserved, and show no signs of reacting with the host magma even though several peridotite blocks are crosscut by one or more thin (1–5 mm) planar veins of melt. Xenolith angularity crudely correlates inversely with size. Xenocrysts (2–5 mm) of olivine and pyroxene are rare (<1%) but ubiquitous within the dyke and presumed to derive from the mechanical breakdown of peridotite xenoliths.

The Mt. Preston region is underlain by Early to Middle Jurassic Hazelton Group rocks, comprising metamorphosed volcanic to volcaniclastic and mudstone-dominated sedimentary rocks (Gordee et al. 2005; Mahoney et al. 2005). The wall rocks to the dyke comprise volcaniclastics that are trachyandesitic to dacitic in composition based on major element chemistry. Regionally, Hazelton rocks are intruded by post-ca. 136 Ma, metamorphosed mafic to intermediate dykes (Gordee et al. 2005). A trachybasalt dyke located within 10 m of the xenolith-bearing dyke of interest representing one of these older intrusions was dated (⁴⁰Ar/³⁹Ar) as 87.74 ± 0.71 Ma (±2 s). There are no carbonate-rich lithologies noted or previously mapped in the region of Mt. Preston.

Samples were prepared for bulk geochemical analysis at the UBC Department of Earth and Ocean Sciences using a jaw crusher followed by pulverization in a tungsten carbide ring mill and then sieved to a grain size of <0.42 mm. Analysis of powders for major and trace elements, ferrous iron, H₂O, and CO₂ was carried out at the Geochemical Laboratories, McGill University, Montreal. Major elements were analyzed by X-ray fluorescence (XRF) on fused beads from ignited samples; trace elements were analyzed for using pressed powder pellets. XRF analysis was done on a Philips PW2440 spectrometer. Total iron was determined by XRF, and FeO content was determined by volumetric analysis (ammonium metavanadate titration). Samples were analyzed for CO₂ on an ELTRA CS-800 carbon/sulfur infrared (IR) analyzer. H₂O+ (structurally bonded water) was determined by difference using loss on ignition (LOI), CO₂, SO₃, halogens, and FeO analyses.

Mineral compositions were measured on a fully automated CAMECA SX-50 electron microprobe at the University of British Columbia, Department of Earth and Ocean Sciences. Operating conditions in wavelength-dispersion mode included excitation voltage of 15 kV, beam current of 20 nA, 20 s peak count time, 10 s background count time, and a beam diameter of 5 µm (see also Harder and Russell 2006; Peterson 2010, p. 152). For carbonate, a beam current of 10 nA and a spot diameter of 10 µm was used. Data reduction was completed using the “PAP” ф(ρZ) method (Pouchou and Pichoir 1985). Criteria for exclusion of analysis points included low or high totals (<98% or > 102%) or poor totals of oxygen relative to the cation sums. For example, analyses were excluded if the oxygen sum was <5.95 or >6 for pyroxenes normalized based on four cations. For geothermometry purposes (see below), at least eight coexisting clinopyroxene and orthopyroxene grains were analyzed in each xenolith (full data set in Supplementary material, Table S6a). These mineral pair compositions were measured on shared grain boundaries within ∼20 µm of the grain edges; no zoning was observed in these phases.

Oxygen, carbon, and strontium isotopic analyses on carbonates were performed on leachates of bulk-rock powders (like other mantle-derived carbonate studies, e.g., Ionov et al. 1996), and in the case of dyke and country rock, also from tungsten carbide microdrilling of carbonate-rich patches. All powdered samples were analyzed at the PCIGR, University of British Columbia, in a Finnigan Delta XL Plus mass spectrometer, using a gas bench with A200 S autosampler.

Oxygen and strontium isotopic compositions of the silicate fraction of the basanite dyke were determined after removal of carbonate from the bulk-rock powders. Powders were immersed in 10% hydrochloric acid, agitated in an ultrasonic bath for 10 min, then separated from the leachate via centrifuge. The leaching process was verified by X-ray diffraction. Sample analysis was performed at the Facility for Isotope Research at Queen’s University, Kingston, Ontario. Analysis followed the BrF5 method of Clinton and Mayeda (1963), on a Finnegan MAT 252 mass spectrometer.

Of the 51 xenoliths studied in detail, 48 have a common assemblage of olivine (ol), clinopyroxene (cpx), orthopyroxene (opx), and spinel (spl). The other samples include two websterites lacking olivine and containing spinel and plagioclase (NP-MP05-69, -96), and a dunite lacking orthopyroxene (NP-MP05-150). Dunites are a minor but common component in Cordilleran suites. The Mt. Preston suite is unimodal (i.e., lherzolite-dominated) based on the definition of Shi et al. (1998) and, in that regard, similar to most other xenolith suites in the Canadian Cordillera. Bimodal suites (i.e., enriched in harzburgite) are found in northwestern BC and southern Yukon (Shi et al. 1998; Francis et al. 2010).

Individual xenoliths exhibit granular textures and show substantial grain size variations (∼0.5–2 mm) but rarely contain megacrysts (outsized grains). About a third of the xenoliths show weak to moderate planar fabrics (i.e., mineralogical banding or mineral foliation) at the hand-sample scale. Foliation tends to be stronger (or more easily observed) in fine-grained samples and is weaker in samples with increasing median grain size. Banding is observed as 1–3 mm wide bands of spinel, repeating on a 1 cm scale, or 2–3 mm wide indistinct bands of clinopyroxene, repeating on an approximately 1–2 cm scale. Larger scale segregations, or possibly bands of olivine and clinopyroxene (∼1 cm width or greater) also occur.

A few samples contain pyroxene grains featuring reaction zones at their margins where they are in contact with thin veinlets of basanite. Less commonly pyroxene grains show disequilibrium textures on their margins (i.e., Fig. 3E) where they are close to intergranular carbonate grains or carbonate veins. Nonetheless, the geothermometry for these rare samples returned internally consistent results based on multiple pairs of pyroxene grains (see below). These textures are not pervasive and likely due to the thermal disturbance and heating of the xenolith during transport. We have only used data from xenoliths for which the geothermometry results are consistent between multiple pairs of pyroxene grains.

We investigated the presence of accessory carbonate in mantle xenoliths using a Cambridge Image Technology™ CL8200 Mk4 cold cathodoluminescence (CL) system attached to a petrographic microscope at the University of British Columbia. Operating conditions included an excitation voltage of 15 kV and a current of 350 µA. Granular carbonate (intergranular or inclusions; Fig. 3A) and/or carbonate veins are found as an accessory phase in all mantle xenoliths and are commonly associated with pentlandite and chalcopyrite (Fig. 3). A single rutile grain was found included in one calcite grain. No hydrous minerals, nor interstitial glass indicative of disequilibrium, were observed. We also used CL to test for the presence of carbonate in two additional crustal xenoliths from the same dyke and mantle xenoliths from two other localities in the Canadian Cordillera. CL examination of these samples found no carbonate. The xenolith-hosted intergranular carbonate is found as discrete ≤0.4 mm grains exhibiting uniform extinction, or as patches with distinct internal crystallographic subdomains (Fig. 3A). The xenolith-hosted carbonate is Mg-bearing (“magnesian”) calcite (X_Ca of 86–94; see Supplementary material, Table S6b).

Texturally, the intergranular carbonate appears to be in equilibrium with coexisting silicate phases, as suggested by shared triple-point grain boundaries (Figs. 3A–3F). Some intergranular carbonate shows concentric zones of Mg-enrichment. Carbonate also occurs as ≤0.2 mm inclusions (Figs. 3H and 3I) in other mantle minerals. Carbonate veins (≤0.1 mm wide) appear as intergranular and crosscutting features (Fig. 3G); some carbonate veins merge with larger grains of carbonate (Fig. 3). Rare, thin veinlets of basanite crosscut all features, including carbonate veins, indicating that the carbonate predates entrainment by the basanite magma and, thus, are a mantle feature.

Pentlandite and chalcopyrite (≤0.1 mm) were noted in eight of the xenoliths examined by CL and have modal abundances <1%. They occur at the edges of carbonate grains (Figs. 3A–3C) in contact with silicates, at the margins of carbonate veins, and in inclusion trails with or without associated carbonate.

Bulk major element compositions of mantle xenoliths have Mg# of 87–91, Al₂O₃ contents of 1.2–4.7 wt.% for lherzolites (0.18 wt.% for dunite NP-MP05-160B and 11.9 wt.% for websterite NP-MP05-69) (Table S3). Major element compositions are consistent with other mantle-derived xenolith suites in the Canadian Cordillera (Fig. 4; Shi et al. 1998; Peslier et al. 2002; Harder and Russell 2006; Francis et al. 2010) and other continental spinel-bearing peridotites worldwide, reflecting varying degrees of melt extraction from a fertile mantle source (Peslier et al. 2002).

Trace and rare earth element (REE) compositions (Fig. 5; Table S3) are like other mantle suites from the Canadian Cordillera and show no pronounced single element anomalies. Rare earth element patterns are light rare earth element (LREE; La to Sm) depleted to flat (Figs. 5A and 5B). Many patterns show a “spoon-shaped” profile of LREE depletion, with slight enrichment of the lightest REEs (e.g., La, Ce); only one pattern (the harzburgite) is weakly LREE-enriched relative to the middle (MREE; Eu to Ho) and heavy (HREE; Er to Lu) rare earth elements. The dunite is especially demonstrative of LREE enrichment; the LREE contents are like that of some other xenoliths from the suite, while the MREE and HREE concentrations are below detection.

Tb/Yb ratios for peridotites (Fig. 5C) can be indicative of whether melt extraction has occurred in the spinel or garnet stability field, as Yb partitions strongly into garnet as a residual phase, resulting in strong Tb/Yb decreases even at small degrees of melting (Bodinier et al. 1988). The Mt. Preston peridotite compositions are consistent with melt extraction in the spinel stability field (Fig. 5C), as with other suites in the Cordillera (Peslier et al. 2002). Terbium enrichment has been interpreted as the result of significant metasomatism (Peslier et al. 2002). None of the Mt. Preston xenoliths analyzed show Tb enrichment over Yb relative to chondrite, suggesting that metasomatic processes did not disturb the MREEs or if so, very little. We found no correlation between the degree of melt depletion or metasomatism with the mantle equilibration temperatures (Section 5 below) as colour-coded in Figs. 5B and 5C.

Carbon and oxygen isotope compositions of carbonate from the mantle xenoliths (Table 1), the host basanite (Table S1), and the wall rocks (Table S4) are plotted in Fig. 6. Isotopic analyses were performed on leachates of bulk-rock powders and, in the case of dyke and country rock, also on carbonate recovered by microdrilling of carbonate-rich phases. The wall rock sample was collected <20 m away from the basanite. Carbonate recovered from these three sources (mantle xenoliths, dyke, wall rock) have distinct carbon–oxygen isotopic compositions. Dyke carbonate has δ¹⁸O_VSMOW compositions of 13.7–14.3 ‰, and δ¹³C_VPDB compositions of −3.8 to −4.5 ‰. In the dyke, δ¹⁸O_VSMOW values are ∼17 ‰ greater than wall rock indicating isotopically distinct sources for the carbonate in the dyke and the wall rock (Fig. 6). The isotopic compositions of the carbonates from the dyke are consistent with unaltered magmatic carbonate (e.g., Lee et al. 2000). Additionally, there is no distinction between bulk-rock and microdrilled analyses of carbonate from the dyke indicting a single (magmatic) isotopic source. The δ¹⁸O_VSMOW values for the silicate fraction of the dyke are ∼6 ‰ and close to mantle values (i.e., primary igneous carbonatites (PIC) field;Fig. 6A).

The wall rock carbonate has δ¹⁸O_VSMOW compositions of −1.7 to −2.6 ‰, and δ¹³C_VPDB compositions of −5.4 to −5.9 ‰. The δ¹⁸O composition of wall rock carbonate (∼2 ‰ less than VSMOW) is consistent with equilibration with meteoric water. In contrast, the carbonate within the dyke shows no signs of isotopic exchange with meteoric waters, as isotopic exchange with these fluids would deplete the carbonate in ¹⁸O.

Carbonate recovered from 14 mantle xenoliths plot as two distinct groups based on their ¹³C compositions (Fig. 6A). The ¹³C-enriched group has δ¹³C compositions between −3.4 and −4.4 ‰ (VPDB), while the ¹³C-depleted group (seven xenoliths) has δ¹³C compositions between −5.7 and −6.1 ‰ (VPDB). Both groups have similar ranges (10.3–12.5 ‰ (VSMOW)) of δ¹⁸O. All but two of the mantle xenolith samples used for stable isotope analysis contained both granular carbonate and carbonate veins. The other two contained only vein carbonate, and they are part of the ¹³C-depleted group.

The carbonates from the Mt. Preston mantle xenoliths are closer to primary mantle isotopic compositions than other reported mantle-hosted carbonates (Fig. 6B; Lee et al. 2000; van Achterbergh et al. 2002; Demény et al. 2004; Ducea et al. 2005; Perkins et al. 2006). The Mt. Preston xenoliths contain carbonate having δ¹⁸O compositions less than 3 ‰ greater than the PIC field for primary mantle-derived carbonatites (Fig. 6B; Taylor et al. 1967; Keller and Hoefs 1995). In other studies, authors report enriched δ¹⁸O compositions in mantle-derived carbonate in xenoliths (Lee et al. 2000; van Achterbergh et al. 2002; Demény et al. 2004; Ducea et al. 2005; Perkins et al. 2006) that are interpreted to indicate isotopic disequilibrium between carbonate and host silicates and explained by short durations between enrichment and eruption. The carbonate in Mt. Preston xenoliths shows less enriched δ¹⁸O compositions (i.e., less isotopic disequilibrium) perhaps indicating longer mantle residence times prior to entrainment by the basanite magma.

Figure 6C shows the ⁸⁷Sr/⁸⁶Sr ratios of the carbonates in the Mt. Preston xenoliths (Table 1) against their δ¹⁸O composition. The carbonate fraction in the Mt. Preston xenoliths have radiogenic ⁸⁷Sr/⁸⁶Sr ratios of 0.7044–0.7046 and overlap the most enriched xenoliths in the Tasse xenolith suite from the southern Canadian Cordillera (Polat et al. 2018; ⁸⁷Sr/⁸⁶Sr ∼0.7034–0.7045). However, both the Tasse suite and the Mt. Preston suite are slightly more radiogenic than other average values for xenoliths suites reported within other southern Canadian Cordillera by Sun et al. (1991) (Jacques Lake: ⁸⁷Sr/⁸⁶Sr ∼0.7027, Big Timothy Mountain: ⁸⁷Sr/⁸⁶Sr ∼0.7030, West Kettle River: ⁸⁷Sr/⁸⁶Sr ∼0.7033, Lassie Lake: ⁸⁷Sr/⁸⁶Sr ∼0.7037). The Mt. Preston dyke is slightly less radiogenic (silicate portion: ⁸⁷Sr/⁸⁶Sr ∼0.7036, carbonate portion: ⁸⁷Sr/⁸⁶Sr ∼0.7040; Fig. 6C; Table S1) and has a source composition for carbonate distinct from of the metavolcanic wall rocks (⁸⁷Sr/⁸⁶Sr ∼0.7047). These ⁸⁷Sr/⁸⁶Sr values recovered from the Mt. Preston dyke are within the upper range of previously reported Canadian Cordilleran (xenolith-hosting) alkaline basalts (⁸⁷Sr/⁸⁶Sr = 0.7024–0.7041; Sun et al. 1991; Polat et al. 2018). The δ¹⁸O composition of the silicate fraction of the Mt. Preston basanite (5.8–6.5 ‰ VSMOW; Fig. 6C) is very close to the global mantle average for silicate phases in spinel peridotite xenoliths (5–5.7 ‰ VSMOW).

Mineral chemical compositions of coexisting grains of clinopyroxene (cpx) and orthopyroxene (opx) in 51 xenoliths were measured by electron microprobe (Table S6a), for the purposes of two-pyroxene geothermometry based on the Brey and Köhler (1990) (BK90) calibration. Our approach was to perform thermometry on sequential batches of five samples and continue expanding the population until there was no further change in the maximum and minimum temperature (Fig. 7A; see Harder and Russell 2006). We suggest that the results from 51 samples closely approximate the entire temperature range within the underlying mantle lithosphere.

The Brey and Köhler (1990) geothermometer has a temperature gradient with pressure of ∼1.5–1.8 °C/Kb. However, we assume that all xenoliths record temperatures along a model geotherm, thereby removing the need to adopt an arbitrary pressure for the geothermometric calculations. Operationally, we do this by solving the BK90 equation simultaneously with the equation for the model geotherm (see below). The geothermometry results (Table S5) record paleo-temperatures between 815 ± 18 and 1119 ± 7 °C, although most samples record temperatures between ∼850 and 950 °C (Fig. 7A). Two xenoliths contain plagioclase and spinel, which coexist over a restrictive range of P–T conditions in most mantle assemblages. Geothermometry on these samples returned relatively low equilibrium temperatures of 850 °C (NP-MP05-69) and 874 °C (NP-MP05-96). The overall distribution of temperatures suggests that the shallower mantle lithosphere is better represented (i.e., more efficiently sampled) than the deep lithosphere. Conversely, the restricted distribution may simply indicate poor mixing of xenolith populations during dyke transport, in contrast to the efficient mixing processes that attend eruption (Russell and Jones 2023).

The δ¹³C isotopic compositions of xenolith-hosted carbonate indicate two groups, relatively enriched (δ¹³C ∼−3.4 to −4.4 ‰ (VPDB)) and depleted (δ¹³C ∼−5.7 to −6.1 ‰ (VPDB)) (Figs. 6A and 6B; Table 1). This suggests that the carbonate preserved throughout ∼25 km of mantle lithosphere beneath this western margin of the Canadian Cordillera derives from multiple sources (or events). The two groups have similar oxygen isotopic compositions (δ¹⁸O of 10.3–12.5 ‰ (VSMOW)) and record overlapping mantle lithosphere temperatures: 851–979 °C (+one at 1119 °C) vs. 849–940 °C, respectively.

Plotting the xenolith equilibration temperatures on a model steady-state geotherm is a means of mapping the distribution of carbonated mantle within the mantle, estimating the minimum thickness of mantle lithosphere, and constraining the depth to the lithosphere–asthenosphere boundary (LAB). Here, we use a one-dimensional model for the lithosphere comprising a crust of known thickness (Z_M) and having constant thermal conductivity (K₁ = 2.5 W/m·K), and surface heat flow (qo) and surface temperature (To ∼10° C). The crustal layer has an exponential distribution of radiogenic heat producing elements (A(z) = A_O e^{− z/Z_M}; Russell and Kopylova 1999), and we consider a range of heat production (Ao) values. The underlying mantle lithosphere has constant thermal conductivity (K₂ = 3.2 W/m·K) and no radiogenic heat source. The lithospheric crust and mantle are coupled numerically by a common Moho temperature (T_M) and the reduced heat flow (q_M) at the Moho.

We have assumed a Moho depth (Z_M) of 32 km (Calkins et al. 2010) and an average crustal heat production of 1.6 ± 0.8 µW/m³ (Lewis et al. 2003), which also fixes values of q₀ (eq. 2). Model values of q₀ for the range of Ao values (0.8–2.4 µW/m³) vary from 63 to 90 mW/m², which agrees well with average measured values reported for the Canadian Cordillera (76 ± 21 mW/m²; e.g., Hyndman 2017).

We assume Moho temperature (T_M) to be equal to, or lower than, the lowest temperature xenolith (815 °C) and have adopted a value of 800 °C (Fig. 7). Similarly, the highest temperature xenolith (1119 °C) constrains the minimum temperature of the LAB (Fig. 7B). A model geotherm is shown in Fig. 7B for three separate values of Ao. The model values of q₀ are inversely correlated to values of Ao such that the highest temperature xenolith corresponds to depths of 49 km (low Ao and high q₀) or 58 km (high Ao and low q₀) implying minimum thicknesses of the mantle lithosphere of 17–26 km. The reduced heat flow at the base of the crust (q_M) would be 61 to 41 mW/m² for low to high values of Ao (i.e., inverse correlation). These values of q_M correspond to temperature gradients in the mantle of 19 to 13 °C km⁻¹, respectively.

An alternative estimate of lithosphere thickness can be made by extrapolating the model geotherms to intersect the adiabat based on values adopted by Hyndman and Canil 2021 (after Katsura et al. 2010). The three model geotherms (Fig. 7B) intersect the adiabat at depths of 58 (low Ao) to 72 (hi Ao) km implying mantle lithosphere thicknesses of between 26 and 40 km and LAB depths between 58 and 72 km and temperatures of 1310–1320 °C. These results accord well with estimates (65 km and ∼1350 °C) from Hyndman and Canil (2021) and Canil and Hyndman (2023).

The model geotherm shows that carbonate is distributed pervasively throughout the entire mantle lithosphere and that the carbonate is a stable mantle phase over a minimum range of temperatures of 800–1120 °C and pressures of 0.9–1.7 GPa. The carbonate occurs in several habits, but those habits are insensitive to source temperatures, pressures, and depths. Four samples with granular carbonate have temperatures of 850–895 °C, 10 xenoliths with carbonate veins have temperatures of 826–924 °C, and 37 xenoliths contain both habits and have temperatures of 815–1119 °C.

Primary mantle-derived carbonate is rarely preserved in mantle xenoliths but provides important insights on mantle metasomatism involving carbonatitic melts/fluids. At Mt. Preston, primary carbonate (Mg-calcite; Table S6b) is pervasive and occurs in both lherzolites, websterites, and dunites, and the xenoliths preserve strong textural evidence of its mantle equilibration. For example, the intergranular carbonate within the Mt. Preston suite preserves triple-point grain boundaries consistent with textural equilibrium between carbonate and mantle silicates. Other occurrences of carbonate within mantle-derived xenoliths are reported for localities in Argentina (Laurora et al. 2001; Scambelluri et al. 2009), Hungary (Demény et al. 2004, 2010), the Kerguelen Islands (Moine et al. 2004), Spitsbergen (Ionov et al. 1996), the Siberian Craton (Ionov et al. 2018), the East African Rift, Tanzania (Lee et al. 2000), South Africa (Berg 1986), and the southwest United States (Ducea et al. 2005; Perkins et al. 2006). Most of these occurrences are described as interstitial patches associated with second generation crystallization, or as inclusions within crystals, typically in textural (glass) or isotopic disequilibrium with the host rocks.

In addition, mantle carbonate at Mt. Preston is closely associated with sulphides and, together, are found as intergranular patches, in veins, and as inclusion trails within healed mantle silicates (Figs. 3A–3C). Sulphides are a common accessory phase in mantle-derived rocks and are a common component of carbonate melts or carbonate–silicate melts (Ionov et al. 1996). The sulphides found in Mt. Preston mantle xenoliths occur across the full range of equilibration temperatures (i.e., depth) and are found associated with both high and low δ¹³C groups of carbonate. This occurrence suggests that carbonate and sulfide result from the same enrichment event(s).

Chemical compositional features of the Mt. Preston xenoliths are also consistent with a carbonate metasomatic agent. Most of the xenoliths show calcium enrichment relative to aluminum, which has been attributed to cryptic carbonate metasomatism (Peslier et al. 2002). Many REE patterns from the Mt. Preston xenoliths show slight enrichment of the lightest REEs, consistent with the effects of a migrating LREE-rich melt such as carbonatite. Collectively, the petrographic observations and geochemical data strongly support the premise that carbonate and sulfide accessory phases are part of the pre-entrainment mantle paragenesis.

Measured δ¹⁸O vs. δ¹³C compositions for the Mt. Preston xenolith and dyke carbonates are close to bulk mantle compositions and distinct from the wall rock carbonates (Fig. 6B). In fact, the δ¹³C and δ¹⁸O isotopic compositions of Mt. Preston carbonate are closer to primary mantle compositions than many other occurrences of mantle-hosted carbonates (Fig. 6B; Lee et al. 2000; van Achterbergh et al. 2002; Demény et al. 2004; Ducea et al. 2005; Perkins et al. 2006). Oxygen isotopic compositions can be more easily re-equilibrated, yet the measured δ¹⁸O values for xenolith, dyke, wall rock plot as discrete clusters (Fig. 6A) and show no indication of chemical mixing between sources; there is also no evidence for meteoric alteration of the xenoliths nor dyke.

δ¹³C values for the Mt. Preston peridotite xenoliths are like carbonate-bearing mantle xenoliths from New Mexico (Lee et al. 2000) and other localities (Fig. 6B) but have even more mantle-like δ¹⁸O compositions. This may suggest relatively better preservation of primary carbonate within the Mt. Preston suite. Furthermore, the Mt. Preston mantle carbonate samples define two distinct groups suggesting at least two carbonate–sulfide melt enrichment events. The concept of multiple enrichment events is further supported by slight differences in the ⁸⁷Sr/⁸⁶Sr isotopic compositions of the two δ¹³C-defined populations (high-C 0.7044 vs. low-C 0.7046). In addition, differences in ⁸⁷Sr/⁸⁶Sr for the carbonate and silicate fractions of the dyke could indicate mixing of ⁸⁷Sr/⁸⁶Sr compositions between dyke silicates and xenolith carbonate, thereby preserving evidence for a magma charged with CO₂ derived, in part, from mantle-derived (i.e., xenolith) carbonate (see below).

Studies of the mantle lithosphere underlying the Canadian Cordillera have shown it to be relatively uniform in composition, thermal regime, and age (e.g., Francis et al. 2010 and references therein). Rhenium–osmium dating of lithospheric mantle-derived xenoliths (Peslier et al. 2000) has recovered Proterozoic model ages across the Canadian Cordillera, consistent with melt depletion events occurring within a short timeframe for the Cordilleran lithospheric mantle. Whether autochthonous or allochthonous, this indicates a common melting history for, at least, the lithospheric mantle underlying the southeast Canadian Cordillera. This result precludes the suggestion of a simple extension of cratonic crust coupled to its own lithospheric mantle residing beneath the eastern Cordillera, as the mantle lithosphere in that region appears to be much younger than the wedge of cratonic crustal basement that overlies it (Peslier et al. 2000).

The Mt. Preston xenolith suite is in a unique geographic location on the western margin of the Intermontane Belt compared to other BC cordilleran xenolith localities, which are further from the continental margin. They are also the only mantle peridotite xenoliths in the BC cordillera that preserve primary, mantle-equilibrated carbonates (and associated sulphides). Furthermore, these carbonated mantle xenoliths are the only ones reported from the Pacific Coast Ranges of the western margin of North America. The carbonates and sulphides are texturally undisturbed, suggesting that the metasomatic event was the last event prior to entrainment by the 19 Ma basanite magma. This suggests a different history involving a geographically restricted enrichment process that probably involved metasomatic agents related to subduction of oceanic crust along the central western margin of BC.

One possible source of carbonatitic metasomatism of the mantle lithosphere is subduction of carbonate in sediments coupled to the downgoing slab. Subducting slab-related enrichment agents have been interpreted for other carbonate-bearing xenolith suites (e.g., Laurora et al. 2001; Demény et al. 2004; Ducea et al. 2005; Perkins et al. 2006). Decarbonation reactions in subducting plates can occur at higher temperatures than dehydration reactions, based on experimental and thermodynamic data (Yaxley and Green 1994; Ducea et al. 2005 and references therein), or the silicate and carbonate melts may be immiscible resulting in separation. Subduction-related metasomatism has been considered for xenolith suites from northwest BC and southwest Yukon based on incompatible trace element enrichments, although no metasomatic phases were observed in those rocks (Shi et al. 1998; Peslier et al. 2000). If subduction-related enrichment during Coast Plutonic Complex magmatism is the source for the carbonates and sulphides in the Mt. Preston suite, the sites in northwest BC and southwest Yukon might be expected to show this as well. However, several of these sites are interpreted to overlie anomalously hot asthenosphere interpreted as a Tertiary to recent thermal event (Frederiksen et al. 1998; Shi et al. 1998); at the interpreted P–T range for these rocks, little isobaric heating (≤50 °C) is required to make crystalline calcite unstable.

Mt. Preston occupied an arc to back-arc location during the duration of Coast Plutonic Complex magmatism history. Based on tectonic models (van der Heyden 1992; Monger et al. 1994; Gehrels et al. 2009), this would restrict the timing of the enrichment event(s) from at earliest, Jurassic time to ∼50 Ma when Coast Belt magmatism waned; subduction ended ∼40 Ma, although the margin of the subducted slab persisted to ∼35 Ma beneath the area (Madsen et al. 2006). However, for metasomatic enrichment to have occurred before 90–85 Ma, the mantle lithosphere under Mt. Preston would have had to be thermally and texturally unaffected by the accretion of terranes of the Insular Belt into Stikinia–Yukon Tanana (e.g., Monger et al. 1982; Gehrels et al. 2009).

The ⁸⁷Sr/⁸⁶Sr values for Mt. Preston samples are slightly more radiogenic than expressed by the eastern BC xenolith suites. The more radiogenic ⁸⁷Sr/⁸⁶Sr ratios might be indicative of carbonatite metasomatism or the lower ⁸⁷Sr/⁸⁶Sr ratios for the eastern BC xenoliths could reflect the region’s proximity to the North American Craton margin. Isotopic evidence suggests that the Cordilleran lithosphere may not be as uniform as previously shown (Peslier et al. 2000; Francis et al. 2010). In particular, the Mt. Preston carbonated peridotite samples imply important compositional variations within the Canadian Cordilleran mantle lithosphere that inform on the mantle’s volatile budget (carbon), the fate of subducted carbon, and its release during volcanism (discussed in Section 7.4). The stability of carbonate, in the mantle, for example, has been shown to be redox-controlled (e.g., Frost and McCammon 2008), suggesting variations in the oxidation state of mantle material beneath the Intermontane–Coast Belt margin in the western Cordillera.

Carbonated peridotitic mantle is the common source for a wide variety of mantle-derived magmas, including carbonatites, nephelinites, basanites, melilitites, and kimberlites (e.g., Dasgupta et al. 2013). Given Earth’s abundance of these Si-undersaturated, CO₂-rich magmas, it is surprising that carbonate-bearing samples of mantle lithosphere are relatively rare. This discrepancy led Canil (1990) to perform a series of decompression experiments designed to explore the carbonate stability during ascent and to assert that carbonate decomposes during decompression at rates (1.5–3 GPa h⁻¹) exceeding feasible magma ascent velocities (12–25 m s⁻¹).

This raises the question of how the carbonate in these xenoliths was preserved during ascent of the Mt. Preston basanite magma. All samples of the mantle beneath Mt. Preston were carbonated and equilibrated at pressures less than 2 GPa and at temperatures lower than the basanite magma (<1250 °C). In this environment, carbonate can be destroyed in at least three ways: (i) by solid state reaction to produce a non-carbonate assemblage, (ii) by decomposition to a fluid or gas driven by changes in pressure, temperature, or P_CO2 (Canil 1990; Frost and McCammon 2008; Escardino et al. 2013), or (iii) by dissolution or assimilation in a silicate melt (Edwards and Russell 1998). To preserve the carbonate and sulphides in the xenoliths, xenolith transport rates must have exceeded rates of thermal–chemical processes promoting carbonate destruction (i.e., decomposition, melt infiltration, etc.).

There is every evidence that the basanite magma transited the mantle lithosphere rapidly. The basanite is essentially aphyric indicating that near liquidus temperatures (∼1250 °C) were maintained throughout transport, implying little loss of enthalpy; ascent rates exceeded rates of conductive cooling to the wall rocks. The basanite magma (ρ_x ∼2700 kg m⁻³) also transported dense (ρ_x ∼3300 kg m⁻³; Table S5) mantle xenoliths sourced from depths of 70–30 km to within several kilometers of the Earth’s surface. Several of the largest xenoliths are ∼50 cm in diameter. The abundance and size range of xenoliths carried by a low viscosity melt (η ∼25 Pa s) within a relatively narrow dyke (1–3 m) also support a relatively high ascent rate (e.g., Sparks et al. 2006). For our example calculations below, we adopt a physically reasonable value of ∼4 m s⁻¹ (see Sparks et al. 2006; Russell and Jones 2023).

There is a window in terms of xenolith size and source depth where carbonate is preserved (Fig. 8D). Carbonate preservation is favoured when t_h > t_r. Smaller xenoliths (i.e., D < 10 cm) typically have t_h < t_r regardless of sample depth because heating times are short (Fig. 8C). In contrast, 50 cm diameter xenoliths have substantial heating times relative to their rise rates (t_r < t_h) except when sourced near the LAB (Fig. 8C). There transit times are greatest, and the ambient temperature is highest thereby reducing t_h values. However, slightly larger xenoliths (D ∼ 55 cm) are settling more rapidly and have high values of t_r, such that t_r >> t_h regardless of sample depth. For minimum magma ascent rates of 4 m s⁻¹, xenoliths that are between 20 and 45 cm in size (diameter) are optimal for preserving carbonate (Fig. 8D).

The second way in which carbonate decomposition is mitigated relates to the intrinsic volatile content of the host magma. Carbonate decomposition rates are strongly dependent on the composition of ambient atmosphere; high values of P_CO2 cause a hyperbolic decrease in carbonate decomposition rate (L’vov 2007). Experimental data of Escardino et al. (2013) showed calcite decomposition in a CO₂ atmosphere to practically stop at temperatures <875 °C; at higher temperatures decomposition rates dropped by >80%. The basanite dyke contains magmatic groundmass carbonate and is vesicular, indicating that it was CO₂-volatile rich. Despite the low solubility of CO₂ at crustal pressures, the magma trapped in the dyke was not fully degassed but, rather, remained enriched in CO₂. The ascent rates were sufficiently high to effectively suppress efficient degassing of CO₂ implying a high partial pressure of CO₂ throughout transport that would inhibit carbonate decomposition.

Lastly, all samples of mantle lithosphere from beneath Mt. Preston are carbonated regardless of depth. Olivine and pyroxene xenocrysts in the basanite result from xenolith disaggregation during transport, which potentially allowed for scavenging and chemical assimilation of xenolith-hosted carbonate (e.g., Ionov et al. 1996; Lee et al. 2000; Laurora et al. 2001). Disaggregation of carbonated peridotite xenoliths during transport represents an efficient mechanism for liberation of accessory carbonate leading to increased CO₂ solubility or suppression of degassing of the rising magma. The additional dissolved or exsolved CO₂ content would increase magma buoyancy, support higher ascent rates for both the magma and its entrained xenoliths, thereby enhancing carbonate preservation. The carbon, oxygen, and strontium isotopic compositions of the dyke and xenolith carbonate are consistent with this process. For example, the carbonate sampled from the basanite dyke has a δ¹³C composition within the range of the sampled mantle xenoliths and has a ⁸⁷Sr/⁸⁶Sr composition between the dyke silicate minerals and the xenolith carbonate.

The 19 My basanite dyke exposed near Mt. Preston, British Columbia, intrudes metavolcanic rocks (Hazelton Group) of the western Intermontane Belt and is situated within 40 km of Coast Belt. The dyke contains abundant spinel-bearing peridotite (mainly lherzolite) xenoliths from the underlying Cordilleran mantle. The xenolith suite is unique for preserving primary, mantle-equilibrated magnesian calcite as an accessory phase commonly in association with sulphides (pentlandite and chalcopyrite). Two-pyroxene thermometry (N = 51) returned a range of paleo-temperature estimates limiting the Moho temperature to ≤815 °C and the LAB temperature to ≥1120 °C. A model geotherm, coupled with the geothermometry and projected to the mantle adiabat, constrains the mantle lithosphere to a thickness of 26–40 km, and the LAB to a depth of 58–72 km and temperature of 1310–1320 °C (see also Hyndman and Canil 2021). The thermometry also shows that accessory carbonate and sulfide phases are stable throughout mantle lithosphere at temperatures of ∼800–1120 °C corresponding to pressures (i.e., depths) of ∼1–1.7 GPa. The carbonate–sulfide assemblage provides strong evidence for pervasive metasomatism of an earlier melt-depleted mantle lithosphere involving at least two isotopically distinct carbonatitic fluids with associated monosulphides (rather than hydrous silicate fluids). The metasomatic event derived from subduction of oceanic crust beneath the western margin of the Canadian Cordillera during Coast Plutonic Belt magmatism, when Mt. Preston was in an arc to back arc position (between ∼90 and 35 Ma). Rapid magma transport rates combined with a high intrinsic P_CO2, sustained by scavenging of carbonated mantle lithosphere, provided the means to preserve the accessory carbonate within these entrained fragments of mantle lithosphere.

We thank the guest editors Kelin Wang, Claire Currie, and John Cassidy for the opportunity to contribute to the special issue of the Canadian Journal of Earth Sciences in honour of Roy Hyndman: “Geophysical studies of the lithosphere and plate boundaries”. Our manuscript benefitted from critical reviews by Cliff Shaw and an anonymous referee and editorial guidance from Claire Currie.

All data are available in the Supplementary material associated with the manuscript or from the authors.

Conceptualization: JKR

Formal analysis: NP, JKR

Investigation: NP, NB

Methodology: NP

Project administration: JKR

Software: JKR

Validation: NB

Writing – original draft: NP, NB

Writing – review & editing: NP, JKR

This research was funded by the Natural Sciences and Engineering Research Council (NSERC) Discovery Grant held by JKR and the NSERC Collaborative Research Opportunities (CRO) grant (BATHOLITHS). NP acknowledges research grants awarded by the Geological Society of America and the Mineralogical Association of Canada.

Supplementary data are available with the article at https://doi.org/10.1139/cjes-2024-0152.