Recognition of fundamental tectonic boundaries has been extremely diffi-cult in the (>1000-km-wide) Proterozoic accretionary orogen of south western North America, where the main rock types are similar over large areas, and where the region has experienced multiple postaccretionary deformation events. Discrete ultramafic bodies are present in a number of areas that may mark important boundaries, especially if they can be shown to represent tectonic fragments of ophiolite complexes. However, most ultramafic bodies are small and intensely altered, precluding petrogenetic analysis. The 91-Mile peridotite in the Grand Canyon is the largest and best preserved ultramafic body known in the southwest United States. It presents a special opportunity for tectonic analysis that may illuminate the significance of ultramafic rocks in other parts of the orogen. The 91-Mile peridotite exhibits spectacular cumulate layering. Contacts with the surrounding Vishnu Schist are interpreted to be tectonic, except along one margin, where intrusive relations have been interpreted. Assemblages include olivine, clinopyroxene, orthopyroxene, magnetite, and phlogopite, with very rare plagioclase. Textures suggest that phlogopite is the result of late intercumulus crystallization. Whole-rock compositions and especially mineral modes and compositions support derivation from an arc-related mafic magma. K-enriched subduction-related fluid in the mantle wedge is interpreted to have given rise to a K-rich, hydrous, high-pressure partial melt that produced early magnetite, Al-rich diopside, and primary phlogopite. The modes of silicate minerals, all with high Mg#, the sequence of crystallization, and the lack of early plagioclase are all consistent with crystallization at relatively high pressures. Thus, the 91-Mile peridotite body is not an ophiolite fragment that represents the closure of a former ocean basin. It does, however, mark a significant tectonic boundary where lower-crustal arc cumulates have been juxtaposed against middle-crustal schists and granitoids.

DEDICATION - This manuscript represents one of the last projects that Dr. Sheila J. Seaman was working on before her untimely passing. It is completed in her honor as a beloved researcher, teacher, and colleague.


The broad Proterozoic orogenic belt of southwestern North America has been interpreted in terms of the assembly and accretion of both continental and oceanic terranes, blocks, and/or provinces between ca. 1.8 and 1.0 Ga (Bennett and DePaolo, 1987; Karlstrom and Bowring, 1988; Bowring and Karlstrom, 1990; Whitmeyer and Karlstrom, 2007). Geologic mapping, isotopic analyses, and geochemical investigations have been carried out for several decades, but suture zones between tectonic blocks (or provinces) are still extremely difficult to identify. This is probably due, at least in part, to the intensity of multiple syn- and postassembly deformational and metamorphic events, but the lack of clear suture boundaries has led to questions about the tectonic significance of the blocks and provinces themselves. Specifically, do the tectonic blocks represent microplates that were assembled into Proterozoic continental crust, or, alternatively, do they represent tectonically rearranged but not exotic components of a single crustal province? Ultramafic rocks occur as tectonic lenses (meters to rarely hundreds of meters in diameter) throughout the Proterozoic orogen, and many occur in or near suspected tectonic boundary regions (Fig. 1). This has led to speculation that these exotic rocks may represent fragments of ophiolites (back-arc or oceanic crust) that mark sutures, or they may serve as markers of other types of significant tectonic boundaries within the orogen.

Field relations and geochemical constraints are required to illuminate the significance of the ultramafic bodies. However, most of the ultramafic occur-rences consist of relatively small and isolated blocks, and most are strongly altered and/or metamorphosed. Primary structures and textures are generally not preserved, and geochemical data can be suspect because of alteration. The 91-Mile peridotite, exposed in the Upper Gorge of the Grand Canyon, is a significant anomaly in terms of its size and the degree of preservation. The ultramafic body is ~1 km in diameter, and primary assemblages and textures are superbly preserved. This body is of particular interest because it occurs within several kilometers of the proposed Crystal suture zone between the isotopically distinct Mojave and Yavapai crustal provinces (Fig. 1; Ilg et al., 1996; Karlstrom and Williams, 2006; Holland et al., 2015).

The goal of this study was to characterize the 91-Mile peridotite exposure and to use field, petrographic, and compositional information derived from the peridotite to constrain its provenance and emplacement history, and specifically to determine whether the peridotite is a fragment of an ophiolite (i.e., a tectonic sliver from a mid-ocean ridge or back-arc/intra-arc basin) or a cumulate that developed in an arc-associated magma chamber. The minimally altered, metamorphosed, or deformed nature of the 91-Mile peridotite provides a special opportunity to interpret field relations, primary textural features, mineral assemblages, and compositional characteristics, and to evaluate the tectonic setting of formation of these basement rocks. Interpretations derived from this well-preserved exposure can provide insight into the petro-genesis of other ultramafic occurrences within the Grand Canyon and across the orogenic belt and may ultimately help to answer the question of whether the ultramafic bodies are associated with, and can be used to identify, major tectonic boundaries within the orogen.


The lithosphere of the Grand Canyon region of southwestern Laurentia was assembled between 1800 Ma and 1400 Ma (Condie, 1992; Whitmeyer and Karl-strom, 2007) through the accretion of a 1000- km- wide accretionary orogenic complex interpreted to be composed of island-arc terranes, continental fragments, and their syntectonic cover (e.g., Karlstrom and Bowring, 1988; Bowring and Karlstrom, 1990; Whitmeyer and Karlstrom, 2007). This 1000-km-wide orogen has been divided into three northeast-southwest–trending Proterozoic crustal provinces, the Mojave, Yavapai, and Mazatzal Provinces (Fig. 1; Karlstrom and Bowring, 1988; Whitmeyer and Karlstrom, 2007). The Mojave and Yavapai Provinces are both made up of 1840–1700 Ma rocks that were deformed and metamorphosed during the ca. 1700 Ma Yavapai orogeny (Holland et al., 2015). However, the Mojave Province has a distinctive isotopic signature indicating the cryptic presence of Archean crustal material (Wooden and DeWitt, 1991; Holland et al., 2018).

The boundary between the Mojave and the Yavapai Provinces is shown in Figure 1 as a broad (75-km-wide) northeast-trending “Moj-Yav transition zone”(Fig. 1), defined mainly on the basis of Pb, Nd, and Hf isotopes (Bennett and DePaolo, 1987; Wooden and DeWitt, 1991; Duebendorfer et al., 2006; Holland et al., 2015, 2018). The Mojave Province to the west is made up of 1.84–1.68 Ga metasedimentary and plutonic rocks that contain detrital and inherited zircon evidence for derivation from some juvenile crust, but with older crustal involvement indicated by the major 1.8 and 2.5 Ga zircon modes. Metasediments and plutons both get systematically younger eastward, and their Hf isotopic composition gets more juvenile west-to-east and old-to-young (Holland et al., 2018). The Yavapai Province, to the southeast, is similar in age but has a more juvenile, arc-related character than the Mojave Province (Karlstrom et al., 2001); the dominantly juvenile nature of the Yavapai Province has been supported by recent Hf isotopic studies (Holland et al., 2015, 2018), although older crust (Bickford and Hill, 2007) and older detritus (Shufeldt et al., 2010) have been found in some areas.

Proterozoic rocks of the transition zone consist of metasedimentary (Vishnu Schist) and metavolcanic (Rama and Brama Schist) sequences intruded by voluminous mafic to felsic 1730–1680 Ma plutonic rocks (Ilg et al., 1996; Hawkins et al., 1996). Geochemical signatures from Yavapai-type crust and Mojave-type crust, as well as rocks with mixed signatures, can all be found in close proximity to one another, a situation reflective of both tectonic and geochemical mixing of rocks from both provinces, and it has not been possible to pinpoint discrete tectonic boundaries.

The Mazatzal Province (Fig. 1) includes 1700–1650 Ma rocks that are interpreted to have been built on Yavapai-aged crust in a continental arc (Karlstrom et al., 2016). Deformation took place during the 1650–1600 Ma Mazatzal orogeny and, in many regions, again at ca. 1450–1400 Ma during the Picuris orogeny (Daniel et al., 2013; Mako et al., 2015). The Granite-Rhyolite Province contains ca. 1.5 Ga juvenile crust that was added at this time (Bickford et al., 2015). The Texas extension of the Grenville Province occurs well to the south of the Grand Canyon and includes younger rocks (1400–1100 Ma) that were deformed in the Grenville orogeny between 1200 and 1000 Ma. Figure 1 shows these provinces as well as other suggested boundaries and terminologies involving proposed terranes (Condie, 1992) and shear zone–bounded subterranes or blocks that differ, at least to some degree, in composition, metamorphic grade, and/or tec-tonic history (Karlstrom and Bowring, 1988; Ilg et al., 1996; Dumond et al., 2007).

The Grand Canyon contains three main basement exposures, the Upper, Middle, and Lower Granite Gorges (Fig. 1), which each provide 100% exposed transects across parts of the Proterozoic orogen, including the Mojave-Yavapai transition zone. The Upper Granite Gorge (Fig. 2), exposed from river mile 76 to 120 (where river mile [RM] is defined downriver from Lee's Ferry, just below Glen Canyon Dam), has been subdivided into six shear zone–bounded lithotectonic blocks (Ilg et al., 1996; Hawkins et al., 1996; Dumond et al., 2007). Most rocks contain at least two deformational fabrics: an early NW-striking, shallowly dipping foliation (S1) that is refolded and variably transposed by a later NE-striking, steeply dipping foliation (S2). Peak metamorphism occurred between 1705 and 1680 Ma (Hawkins et al., 1996) during the second phase of deformation. All blocks were metamorphosed at pressures of ~0.6–0.7 GPa; peak temperatures tend to alternate from block to block from 500–600 °C to more than 700 °C (Dumond et al., 2007).

The Crystal shear zone (at RM 98; Fig. 2) was proposed to be the eastern edge of the suture zone between the Mojave and Yavapai Provinces (Ilg et al., 1996; Hawkins et al., 1996), and the Gneiss Canyon shear zone (RM 234–242) was interpreted as the western boundary of the transition zone (Karlstrom et al., 2003). Both shear zones show east-to-west steps to more evolved (more Mojave-like) isotopic compositions of plutons, high D2 strain, and lenses of ultramafic rocks and pillow basalt (and carbonate in the Gneiss Canyon shear zone). Recent detrital zircon results from the Vishnu Schist document a bimodal detrital zircon spectrum with peaks at 1.8 Ga and 2.5 Ga across the entire Grand Canyon transect, with no change across the Crystal or Gneiss Canyon shear zones, suggesting that any suturing would have predated or been synchronous with 1.75 Ga Vishnu Schist deposition (Shufeldt et al., 2010; Holland et al., 2015). Hf isotopic results from plutons, however, show variable mixing of juvenile and evolved crust within heterogeneous lower-crustal melt-source regions in the transition zone but an overall change to juvenile granodiorites east of the Crystal shear zone (Holland et al., 2015). Holland et al. (2015) interpreted the Mojave-Yavapai boundary to be an ~200-km-wide middle-crustal duplex system in which the 1.75 Ga Vishnu Schist was deposited across sutured (or suturing) Mojave and Yavapai crust in an accretionary complex. This distributed boundary has tectonic lenses of plutons that carry the isotopic signature of their respective crustal isotopic provinces, now imbricated with a metasedimentary cover that is compositionally similar across the zone. The cover sediments are interpreted to have been derived mainly from the Mojave Province crust, but were deposited on (i.e., overlapped) both Mojave (evolved) and Yavapai (juvenile) crust at 1.75 Ga before further shortening and tectonic imbrication by thrusting during the 1.74–1.70 Ga Yavapai orogeny.

Three relatively large occurrences of ultramafic rocks are present within and on the east side of the Crystal shear zone in the Upper Gorge of the Grand Canyon (Fig. 2), near RM 83, RM 91, and RM 98, and smaller lenses are found in the Gneiss Canyon shear zone (RM 245). Of these, the exposure of the 91-Mile peridotite is by far the largest and least altered. The exposure at RM 83 is also large (0.29 km2), but its primary minerals have been pervasively altered to chlorite, serpentine, and fine-grained clay minerals. Ultramafic exposures at RM 98 consist of two small groupings (each smaller than 0.11 km2) of mafic/ultramafic lenses within Vishnu Schist. Meter-scale folded and boudinaged ultramafic lenses are also found at RM 245. The lack of continuous outcrop and the advanced alteration of these rocks make them much less interpretable than those of the large and mostly unaltered 91-Mile body, which was the focus of this study.


Field mapping, structural analysis, and sample collection took place over several field seasons by S.J. Seaman and coauthors on park-permitted research river trips in Grand Canyon (1995–2012) led by the University of New Mexico (UNM). Whole-rock major-element analyses were collected from fused glass discs in the X-ray fluorescence laboratory at the University of Massachusetts using a Philips MRS wavelength-dispersive spectrometer under the supervision of J.M. Rhodes. Whole-rock trace-element analyses were collected by inductively coupled plasma–mass spectrometry (ICP-MS) at Union College, Schenectady, New York, using a PerkinElmer/Sciex Elan 6100 DRC under the direction of Kurt Hollocher and by Paul Lamothe at the U.S. Geological Survey (USGS), Denver, Colorado. Trace-element and rare earth element (REE) analyses of minerals were collected by laser-ablation ICP-MS at Boston University using a VG Plasma Quad ExCell ICP-MS equipped with a Merchantek LUV213 laser-ablation ICP-MS system, under the direction of Terry Plank. Microprobe analyses were collected in the Electron Microprobe/Scanning Electron Microscopy Facility in the Department of Geosciences at the University of Massachusetts, using a Cameca SX-50 electron microprobe, under the direction of Michael Jercinovic. Ferric-ferrous Fe determinations for minerals were based on stoichiometry (see Low, 2009).


Field Relations

The 91-Mile peridotite is a NE-trending, pod-shaped ultramafic body located north of the Colorado River, ~1 km up 91-Mile Canyon (Fig. 3). A small out-crop on the Colorado River may be connected (in the subsurface) to the main body along the hinge of a south-plunging F2 fold (Ilg et al., 1996). The main body occupies an area of ~0.75 km2. Most of the ultramafic body consists of layered olivine websterite or lherzolite. Major minerals include olivine, diop-side, orthopyroxene, magnetite, phlogopite, and minor amphibole (pargasite, edenite, and magnesiohastingsite; terminology after Leake et al., 2004). The most striking characteristics of the ultramafic rocks when viewed in the field are their coarse grain size (0.5–3 cm) and strong mineralogic layering, particularly toward the southern end of the body (Fig. 4), with layers persisting for tens of meters along strike. Nearest the mouth of the canyon, ~4–5-cm-thick layering is defined by variable modal abundances of olivine, pyroxene, and phlogopite. The phlogopite books are several centimeters in diameter and are oriented with their basal plane typically at an angle to layering. The layering is interpreted to be cumulate layering, although much of the phlogopite is interpreted to be an intercumulus, postdeposition phase (see below).

The layering in the 91-Mile peridotite is northwest-striking and steeply southwest-dipping, similar to the early (S0/S1) foliation in adjacent Vishnu Schist metasediments (Fig. 3). Contacts between the ultramafic body and the surrounding Vishnu Schist are sharp and locally truncate the peridotite layering and Vishnu layering, suggesting tectonic emplacement. However, some contact areas have deformed and serpentinized inclusions of ultramafic rocks in the neighboring schist, and the side canyon contact at the downstream margin (Fig. 3) has a ragged, possibly intrusive contact with the schist. Tectonic foliations and lineations are not well developed within the interior of the ultramafic rock body, but there is local folding and fracturing near the contacts.

Most of the minerals in the 91-Mile peridotite are interpreted to be primary igneous minerals, including olivine, diopside, orthopyroxene, amphibole (pargasite, edenite, and magnesiohastingsite), and phlogopite (Fig. 4). Serpentine occurs locally and is interpreted to be the product of fluid infiltration and metamorphism. Modal composition varies little as a function of position within the unit, with the exception that phlogopite crystals become coarser and pyroxene crystals become finer and less abundant northeastward (interpreted to be upward based on decreasing Mg# in olivine and clinopyroxene; see below) across the sequence.

Clinopyroxene (diopside) and phlogopite are the dominant minerals seen in hand specimens of the 91-Mile peridotite. Diopside crystals range from 1 mm to 5 mm in diameter. They are blocky and black and occur in layers a few millimeters to over 1 cm in thickness. Large (to >3-cm-diameter) phlogopite crystals impart a bronzy sheen to weathered surfaces of the peridotite and commonly host olivine inclusions. Orthopyroxene also occurs as large cumulate crystals, but it is less abundant than diopside, and the two are indistinguishable in hand specimen.

Petrographic Characteristics

Thin section analysis showed that most samples of 91-Mile peridotite are dominated by coarse diopside, olivine, and phlogopite (Fig. 5A). Diopside crystals host concentric sprays of hundreds of tiny (micron-sized) magnetite crystals that define concentric growth zones in the diopside (Fig. 5B). The euhedral crystal shapes further support the interpretation of preserved igneous textures. Olivine occurs as inclusions in the large diopside crystals and as later-forming crystals. Generally, the olivine inclusions interrupt the concentric zones of magnetite, so that magnetite is absent or scarce within ~1 mm of olivine crystals (Fig. 5C). Phlogopite crystals are orange in thin section under plane light. In rare instances, phlogopite appears to have replaced orthopyroxene, but generally phlogopite occurs as a large, interstitial phase surrounding olivine, diopside, and orthopyroxene (Fig. 5A). Orthopyroxene occurs both as large cumulate crystals and as reaction rims around olivine (Fig. 5D; Low, 2009). Based on inclusion relationships, the crystallization order of major minerals was olivine, diopside, orthopyroxene, with minor primary amphibole, followed by phlogopite. Trace (to 2%) Na-rich plagioclase occurs in some samples as an interstitial phase.

Cumulate layers near the southern end of the main body contain swarms of spheroidal, olivine-rich (i.e., dunite) enclaves 2–10 cm in length (Fig. 6). The long axes are contained within the cumulate layering but are not particularly lineated. Irregular, somewhat flattened boundaries between the enclaves and the surrounding cumulate suggest that the enclaves may have been crystal mushes when they were incorporated into the cumulates. The enclaves are almost entirely olivine (or serpentine pseudomorphs after olivine) with minor diopside, enstatite, Cr-spinel, and pargasite.


Major-Element Concentrations

Cumulate-textured peridotite and pyroxenite samples of the 91-Mile peridotite range in SiO2 from ~45 to 51 wt% and in MgO from ~20 to 28 wt%. They are low in Al2O3 (~1.0–6.5 wt%) and high in K2O (~1.5–2.0 wt%), and they have CaO concentrations from 4.5 to 7.0 wt% (Table 1). CaO, K2O, and Al2O3 are all inversely correlated with MgO. Mg# (molar Mg/[Mg + Fe]) is 0.80–0.83 in all except two less-Mg-rich peridotite samples and over 0.84 in the dunite inclusions. Cr and Ni concentrations are 1600–2100 ppm and 800–1150 ppm, respectively, for the peridotites and ~1700–2200 ppm (Cr) and 1200–1700 ppm (Ni) for the dunite inclusions (Table 2).

Major-element compositions define a general trend consistent with progressive accumulation of minerals crystallizing from a primitive basalt; the most Mg-rich cumulates are the most depleted in K2O, Na2O, and Al2O3, consistent with early crystallization of Mg-rich olivine and Mg-rich diopside. With increasing removal of Mg-rich olivine and diopside from the basaltic parent, those phases became less magnesian and more Fe-rich. The dunite enclaves are the most primitive material, with SiO2 ~41 wt% and MgO ~40 wt%.

Trace-Element and Rare Earth Element Concentrations

Samples of the 91-Mile peridotite are characterized by significant enrichment in the most compatible elements and extreme depletion in incompatible elements. Cr concentrations range from 1320 to 2240 ppm, and Ni abundances range from 660 to 1160 ppm. Rb concentrations range from 50 to 100 ppm, and Ba abundances range from 390 to 800 ppm. Zr concentrations range from 40 to 50 ppm, and Nb abundances range from 2 to 7 ppm (Table 2). The rocks are enriched in the large ion lithophile elements (LILEs), particularly K, Rb, Ba, and Pb, and they are depleted in the high field strength elements (HFSEs) relative to mid-ocean-ridge basalt (MORB) (Fig. 7A).

Rare earth element (REE) patterns are negatively sloped and slightly concave upward, with the light rare earth elements (LREEs) enriched relative to the heavy rare earth elements (HREEs). Overall REE abundances range from ~40 times chondritic composition for the LREEs to ~6 times chondritic composition for the HREEs (Fig. 7B). Dunite enclaves have a similarly sloping REE pattern (Ce/Yb = 5.7–6.8) but with overall abundances that range from ~7 times chondritic concentrations for the LREEs to ~2 times for the HREEs. The dunite inclusions are also enriched in LILEs relative to HFSEs (Pb/Ce = 11–34), with overall abundances in both LILEs and HFSEs much lower than those in the host cumulate samples.

Mineral Compositions


All rocks of the 91-Mile peridotite contain either olivine or serpentine pseudomorphs after olivine. Modal olivine (including pseudomorphs) ranges from 6% to 80%. Olivine Mg# ranges from 0.77 to 0.92 and averages 0.83 throughout the peridotite exposure (Table 3). NiO concentrations range from 0.1 to 0.3 wt%. CaO concentrations are less than 0.001 wt%. Olivine has relatively flat REE patterns with a very slight enrichment in HREEs relative to LREEs (La/Lu < 0.78) and an overall abundance of 0.1 and 1.0 relative to chondrites (Fig. 8) (Sun and McDonough, 1989).


Clinopyroxene (Cpx) is abundant throughout the body with a modal abundance of greater than 20% in some samples. Based on mineral inclusion relationships, it is interpreted to have crystallized after magnetite and olivine and contemporaneously with orthopyroxene. Mg# in diopside ranges from 0.84 to 0.94 (Table 3). Commonly, diopside exhibits alternating concentric bands with abundant magnetite inclusions (Fig. 5B). Diopside in magnetite-rich bands has slightly elevated Al content compared with diopside in magnetite-poor bands. REE plots show concave-downward patterns with overall abundance between 4 and 60 times chondrites (Table 4; Fig. 8). Diopside is enriched in LREEs relative to HREEs (La/Lu > 1.2) with a negative Eu anomaly. Diopside from the 91-Mile peridotite generally shows a slight LREE depletion relative the middle rare earth elements (MREEs) (Fig. 8). Downes (2001) used the La/Nd ratio to represent the LREE/MREE slope of the REE pattern and the Sm/Yb ratio to represent the MREE/HREE slope of the REE pattern, such that a REE pattern can be plotted as a single point on a plot of La/Nd versus Sm/ Yb. By these criteria, the diopside in the 91-Mile peridotite would be classified as LREE-depleted and MREE-enriched diopside.


Phlogopite is a major component of all samples, with modal abundances ranging from 3% to 25% and generally increasing as modal olivine decreases. Phlogopite occurs in three textural settings: (1) as inclusions in olivine, both with and without amphibole, (2) as large inclusion-rich crystals, and (3) as secondary crystals that have partially replaced orthopyroxene. Three instances of phlogopite included in olivine were observed in thin section. Admittedly, these rare apparent inclusions may actually be later phlogopite filling embayments projecting from the third dimension. However, if they are truly inclusions, they suggest that at least some phlogopite was present early in the crystallization history.

Large, primary phlogopite crystals are tabular. They contain inclusions of olivine, pyroxene, and ubiquitous elongated inclusions of magnetite parallel to 001 planes. The phlogopite crystals are interpreted to be magmatic crystals that formed in the late stages of the crystallization history (after olivine, orthopyroxene, and diopside but before matrix amphibole) from an evolved, interstitial, hydrous melt. Phlogopite pseudomorphs after orthopyroxene occur as optically continuous, inclusion-poor crystals up to 10 mm across.

Phlogopite has Mg# from 0.81 to 0.87, with less than 5% variation among samples (Table 3). The one analysis made to date of phlogopite included in olivine was essentially the same as other analyses from matrix crystals, possibly casting doubt on the inclusion interpretation. REEs in phlogopite have low abundance and typically show relative enrichment of HREEs (La/Lu < 0.7) with a large positive Eu anomaly (Table 4; Fig. 8).


Cumulate orthopyroxene crystals are euhedral to subeuhedral and are generally over twice the diameter of coexisting olivine crystals, with a range in diameter from 1 to 5 mm and an average of ~4 mm. They commonly contain inclusions of olivine and magnetite, and they locally occur as coronas surrounding olivine crystals. The Mg# of orthopyroxene ranges from 0.80 to 0.85 (bronzite) with a very limited range within individual samples, i.e., little or no zoning (Table 3). Both CaO and Al2O3 show a negative correlation with Mg#. MnO is poorly correlated with Mg#. The orthopyroxene is generally enriched in HREEs relative to LREEs (La/Lu < 0.07) and ranges from 0.05 to 5 times chondritic composition (Table 4; Fig. 8).


Amphibole is generally a minor component in most samples, but it is abundant (up to 44 modal percent) in several samples. Three textural varieties were observed: (1) interstitial amphibole that surrounds olivine and pyroxene; (2) amphibole that is symplectically intergrown with orthopyroxene in coronas around olivine or occurs as inclusions in olivine; and (3) amphibole that has partially replaced diopside (Fig. 5D). The three textural types are somewhat compositionally distinct. Both interstitial amphibole and symplectic/ inclusion amphibole are pargasite, magnesiohastingsite, or edenite. Amphibole that replaces diopside is magnesiohornblende. It has high Mg# (0.88–0.92), although slightly lower than those of diopside grains in the same samples. Mg# values of interstitial amphiboles are notably lower (0.80–0.85) (Table 3). A third compositional type, tschermakitic amphibole, with Mg# 0.54–0.56, was observed in only one sample.


Plagioclase occurs in only the most evolved cumulate samples. All observed plagioclase is interstitial. It appears to have formed late (possibly last) in the crystallization sequence. Plagioclase crystals are also generally found in contact with interstitial amphibole and are always separated from olivine by a layer of amphibole and bronzite and from coronitic bronzite by a layer of amphibole. It was not observed in contact with diopside. Plagioclase crystals are unzoned, with compositions ranging from albite to oligoclase (An5-An14 Or0.3-Or1 Ab85-Ab93 with an average of An9Or0.6Ab90).

Magnetite (Mt)-chromite (Cr)-spinel (Sp)

Oxide minerals are magnetite-dominated solid solutions with varying amounts of chromite end member (Fig. 9; Table 3). Magnetite compositions range from Mt79Sp6Ct15 to Mt25Sp25Ct50. Magnetite solid solutions define a trend generally parallel to the Fe-Ti trend of Barnes and Roeder (2001), which has been attributed to evolution of spinel compositions during fractional crystallization of olivine and pyroxene, commonly accentuated by reaction with intercumulus magma (cf. Barnes and Roeder, 2001).

Spinel-chromite phases were identified in two contexts. First, a dunite enclave from near the base of the exposure hosts spinel that plots midway between the chromite and spinel end members, close to the Cr-Al trend of Barnes and Roeder (2001). Second, magnetite inclusions in olivine have exsolved spinel with a composition between the chromite and spinel end members. Spinel compositions are approximately Mt0.11Sp0.60Ct0.28. Thus, spinel group minerals that crystallized earlier (inclusions in olivine) have a higher concentration of spinel end member (Cr-Al trend?) than those that crystallized later and follow the Fe-Ti trend.


The Proterozoic orogen of southwestern North America preserves a record of the accretion and evolution of much of the midsection of the North American continent. Ultramafic rocks have been suggested to mark the location of major boundaries or even sutures between orogenic blocks or provinces. The 91-Mile peridotite is by far the largest and least altered of the known ultramafic bodies in the Southwest and as such provides an opportunity to assess the provenance and tectonic significance of at least this one body.

Perhaps the most fundamental observation/conclusion is that the 91-Mile peridotite is a cumulate derived by layered accumulation of crystals from an igneous parent. Primary igneous cumulate phases include olivine, orthopyroxene, clinopyroxene, magnetite/spinel, and minor late-stage plagioclase. Phlogopite crystallized mainly as an intercumulus phase. Serpentine and amphibole occur primarily as replacements of olivine and pyroxene, respectively.

Parent Melt

Compositional characteristics of the melt from which the cumulates crystallized are key factors in evaluating the tectonic setting in which the peridotite originated. Bulk-rock trace-element variation and REE patterns are shown in Figure 7 relative to mantle peridotites and chondrites (from McDonough and Sun, 1995). The 91-Mile ultramafic rocks are depleted in Nb and Zr and enriched in Ba, similar to subduction-fluxed, phlogopite-bearing xenoliths from North Hesse, Germany, and from Vulture, Italy (Downes, 2001).

Because the bulk of the exposure consists of cumulates, characteristics of REE patterns depend on relative proportions of cumulate minerals, and as a result, the individual mineral compositions may be more useful petrogenetic indicators than the bulk rocks. Trace-element and REE concentrations in diop-side crystals from the body provide an opportunity to calculate, at least to a first order, the trace-element and REE concentrations of the melt from which the diopside crystallized. Experimentally determined clinopyroxene/melt partition coefficients of Hart and Dunn (1993) and Sobolev et al. (1996) were used for the calculation (Table 5). The calculated melt composition (Fig. 10) is similar to bulk-rock cumulate analyses, with LREE enrichment and HREE depletion (Fig 10A) and strong depletion in Nb, Zr, and Ti and enrichment in Ba (Fig. 10B). The negative Nb and Zr anomalies are typical of subduction-related basalts, and the Rb and Ba enrichment is consistent with interaction with slab-derived LILE-rich fluids, supporting the interpretation of subduction-related magmatism.

Several processes are known to cause enrichment in LREEs in peridotites and specifically in clinopyroxene. These include: (1) interaction with more felsic silicate magmas (Nielson and Noller, 1987; Downes, 2001), (2) interaction with hydrous subduction-related fluids (Hartmann and Hans Wedepohl, 1993; Zanetti et al., 1999; Downes, 2001), and (3) interaction with mantle-derived carbonate magmas (Yaxley et al., 1991; Ionov et al., 1993, 1997; Downes, 2001). Passage of silicate magmas or fluids through the mantle by flow along grain boundaries can result in hornblende- or pyroxene-rich veins, or hornblende and/or clinopyroxene pseudomorphs of primary mantle minerals (Navon and Stolper, 1987; Kelemen et al., 1990; Takazawa et al., 1992). Partial melting of the altered rocks could then give rise to LREE-enriched basaltic melts, similar to the 91-Mile parent, wherein major-element concentrations are not strongly affected, but enrichment in LREE, Sr, Zr, Hf, and possibly Nb occurs (Downes, 2001). The pervasive Nb depletion of the 91-Mile parent is also consistent with an origin in a subduction setting.

REE patterns of clinopyroxene crystals have been used as indicators of tectonic settings. Clinopyroxene from ultramafic massifs is generally LREE depleted, similar to diopside from the 91-Mile body. N-shaped REE patterns of clinopyroxene, like those of the 91-Mile body, are relatively rare; Downes (2001) suggested that the pattern is consistent with subsolidus redistribution of REEs with coexisting garnet. Garnet is common in the host Vishnu Schist, but it is not present within the 91-Mile peridotite.

Other trace-element concentrations can be used to identify interaction with subduction-related fluids. Hydrous fluids commonly transport LILEs, so the enrichment in Rb and Ba of the 91-Mile peridotite parent basalt is consistent with its origin in a subduction setting. HFSEs, such as Nb and Zr, are insoluble in hydrous fluids, and they are typically depleted in basalts generated in subduction settings, which is also consistent with the modeled parent basaltic magma for the 91-Mile peridotite. Xenoliths from Vulture (central Italy) are samples of crust overlying a subduction zone. They are phlogopite-rich features, most likely as a result of interaction with K-bearing subduction zone hydrous fluid (Downes, 2001). We suggest that K-enriched subduction-related fluid in the mantle wedge gave rise to a K-rich, hydrous, high- pressure partial melt that produced early magnetite, Al-rich diopside, and the phlogopite of the 91-Mile peridotite.

Pressure of Crystallization and Abundance of Water in the Parent Melt

We attempted to constrain the pressure of crystallization of the 91-Mile peridotite. Both the order of crystallization of minerals and the composition of the minerals provide information about pressure and temperature conditions of formation. At low pressure, the order of crystallization of minerals in a primitive basalt is expected to be olivine, plagioclase, clinopyroxene, and orthopyroxene, because the crystallization of olivine depletes the liquid in Mg before clinopyroxene crystallizes, leaving the Mg# of clinopyroxene relatively low (<82) (Elthon et al., 1982; Bağci et al., 2006). In contrast, clinopyroxene crystals with higher Mg# suggest higher-pressure (>0.7 GPa) crystallization. Presnall et al. (1978) showed that the diopside field expands with increasing pressure in the Di-Fo-An system, until at >0.7 GPa, plagioclase is no longer stable. As pressure increases, the amount of olivine that crystallizes prior to crystallization of diopside decreases, and the Mg# of olivine in equilibrium with diopside increases. Rock types that are typical of low-pressure crystallization of basalt include dunites, troctolites, and olivine gabbros, because plagioclase crystallizes early, along with olivine, at low pressures. High-pressure crystallization of basalt produces dunite, wehrlite, clinopyroxenite, websterite, and lherzolite, all with high Mg# (Elthon et al., 1982; Elthon, 1992; Parlak et al., 1996, 2000) and all characteristic of the 91-Mile peridotite. At pressure >1.0 GPa, clinopyroxene crystallizes before plagioclase (Presnall et al., 1978). Thus, the modes of silicate minerals, the sequence of crystallization, and the lack of early plagioclase all support crystallization of the 91-Mile peridotite at relatively high pressures, probably on the order of 1.0 GPa.

Magnetite-chromite solid solution phases in the 91-Mile peridotite follow the Fe-Ti trend of Barnes and Roeder (2001), while (early?) spinel group inclusions in olivine have a higher concentration of spinel, more consistent with the Cr-Al trend (Fig. 9; Irvine, 1967). Barnes and Roeder (2001) suggested that the Cr-Al trend in high-pressure plutonic settings is probably the result of equilibration between Al-bearing pyroxene and Mg- and Al-rich spinel minerals, consistent with the apparent evolution from Al-richer to Al-poorer spinel phases in the 91-Mile peridotite. Further, they showed that solid solutions on the Cr-Al trend are indicative of magmas derived from melting of primitive mantle and that they are typical of mantle and lower-crustal rocks. The progression of spinel group compositions in the 91-Mile peridotite is consistent with crystallization in a high-pressure setting from a melt of mantle origin. This is also compatible with Hf data showing that plutons east of the Crystal shear zone are largely juvenile in character (Holland et al., 2915).

Diopside Mg# values in the 91-Mile peridotite range from 0.87 to 0.93, which, like the olivine, are in the range of those of other high-pressure peridotites. Further, the Al2O3 concentrations of diopside, relative to Mg#, are consistent with the trend defined by Medaris (1972) as the high-pressure crystallization trend. However, it is important to note that the high Mg# values of olivine and diopside in the 91-Mile peridotite could, at least in part, also be a consequence of crystallization from a hydrous magma; high magmatic water concentrations lead to high Fe3+/Fe2+ and thus higher Mg# in the crystallizing minerals (Berndt et al., 2005; Botcharnikov et al., 2005; Feig et al., 2006; Gahlan et al., 2012). The abundant magnetite inclusions in diopside strongly suggest that the melt was highly oxidized.

In summary, several lines of evidence, including the lack of plagioclase, the order of crystallization of minerals, the high Mg# of olivine and diop-side, and the evolution of spinel group mineral compositions, all suggest that the 91-Mile peridotite crystallized at significant pressures. The presence of late phlogopite as well as the abundant magnetite inclusions in diopside crystals and the high Mg# of olivine and clinopyroxene all suggest that the parent magma was relatively hydrous. Without the influence of water, one might confidently suggest pressures on the order of 1 GPa, but the pressure could be somewhat lower in this hydrous magma. We cautiously conclude that the combination of moderately high-pressure crystallization, a hydrous LILE-enriched parent melt, and trace-element evidence for interactions with slab-derived fluids are all consistent with crystallization in deeper levels of an arc-related magma system.

One additional setting to be considered for the 91-Mile peridotite is a back-arc (i.e., suprasubduction ophiolite) setting. Although this setting may not be entirely excluded, we strongly prefer the arc itself because of the very hydrous nature of the magmas, the pressure of crystallization, and the associated arc-related plutonic and volcanic rocks. Further, the main tectonic conclusions (see below) reflect the fact that the 91-Mile peridotite almost certainly did not form in a MORB ophiolite setting.

The apparent intrusive contact along one margin of the 91-Mile peridotite is still somewhat of a puzzle. The planar magmatic layering and the otherwise sharp and discordant contacts suggest that the body was emplaced as a relatively intact rigid body. The possible intrusive textures occur along only one contact and have been somewhat obscured by later deformation. If the contact does indeed preserve an intrusive relationship, we would suggest that the body may have been emplaced as a nearly solid mush, but that some magma (crystals and liquid) was expelled along this particular contact.

Ultramafic Rocks and Crustal Boundaries in the Grand Canyon

The 91-Mile peridotite is surrounded by a series of metamorphosed turbidites and pelitic sedimentary rocks (the Vishnu Schist) and volcanic rocks (the Rama and Brahma Schists) that were deposited ca. 1750 Ma in an arc setting and then intruded by 1750–1710 Ma granodioritic plutons before being tectonized during the ca. 1700 Ma Yavapai orogeny. The combination of basalt/dacite volcanic rocks, turbidite host rocks, and calc-alkaline granodiorite to gabbro/diorite plutons is typical of lithologies of modern subduction settings, consistent with the convergent margin interpretation of the Yavapai orogeny. Plutons east of the Crystal shear zone have Hf isotope compositions indicative of their derivation from ca. 1.75 Ga juvenile mantle. Peak metamorphic pressures of the rocks intruded by the plutons are on the order of 0.6–0.7 GPa (Dumond et al., 2007), but these were achieved after arc pluton emplacement during crustal thickening. Even the granodiorite bodies have vertical tabular shapes and tectonic contacts more similar to tectonic slices than intruded plutons. The geometry of the relatively small and lenticular 91-Mile peridotite suggests tectonic dismembering of a larger arc pluton, and the possible intrusive relationships suggest late synmagmatic emplacement and juxtaposition against supracrustal rocks. Compositional characteristics of the tectonically emplaced 91-Mile peridotite are also consistent with an origin in a subduction setting, although we would suggest at a deeper level (~30 km) than that at which the sedimentary and volcanic rocks accumulated or underwent prograde metamorphism.

One of the goals of this research was to evaluate if the ultramafic rocks that are sporadically present throughout the Proterozoic orogen could be oceanic (i.e., MORB) ophiolite fragments and thus indicators of major accretionary structures separating disparate terranes. Because the 91-Mile peridotite was most likely formed in the lower level of a magmatic arc, we reject this ophiolite interpretation and the hypothesis that the ultramafic rocks can be used to identify boundaries between exotic tectonic (continental) fragments. However, there is no evidence that the host rocks to the 91-Mile peridotite were ever buried to pressures greater than 0.6–0.7 GPa (Dumond et al., 2007), and probably significantly less, during arc plutonism. If the 91-Mile peridotite crystallized at higher pressure, at least some degree of tectonic juxtaposition is indicated. Some juxtaposition is also suggested by the fact that no other pluton with cumulate layering has been seen in the well-studied and 100% exposed Upper, Middle, or Lower Granite Gorges of the Grand Canyon. The parallelism of the 91-Mile peridotite cumulate layering with early S1 foliation and D1 structures suggests that the juxtaposition may have been associated with early accretion-related (D1) thrusts in an accretionary complex, as proposed by Holland et al. (2015). Thus, the 91-Mile peridotite may indeed mark the location of a significant structure within the Proterozoic orogen, a conclusion consistent with its location near the inferred Mojave-Yavapai boundary region.

Several other, generally small and highly altered, ultramafic bodies occur in or near shear zones or interpreted tectonic boundaries (Fig. 1). Based on the results from the 91-Mile peridotite and also on the fact that no arc-related plutons with cumulates have been recognized at the current level of exposure, we suggest that the association of ultramafic blocks and shear zones or suspected tectonic boundaries is more than coincidence. The ultramafic bodies may signal the presence of structures capable of juxtaposing deeper levels of the arc crust with the current midcrustal exposure. Further, the presence of the ultramafic bodies may indicate the general regions to explore for other exotic rocks from deeper in the orogen.


Sheila Seaman passed away on 27 July 2019. Her work on igneous rocks of the Grand Canyon, and specifically on the 91-Mile peridotite, benefited from interaction with and participation of many students and colleagues from the University of Massachusetts and the University of Colorado. This research was facilitated by grant EAR-0003477 from the NSF Tectonics Program (to Seaman and Karlstrom) and Grand Canyon National Park Research and Collecting permits. Steve Turner and two anonymous reviewers made positive, helpful comments that significantly strengthened the manuscript. Comments from Steve Turner were critical to the final completion of this manuscript. Editorial handling by Shanaka de Silva is sincerely appreciated.

Science Editor: Shanaka de Silva
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.