The Hualapai Plateau in northwest Arizona, the location of the western Grand Canyon, contains an unusually lengthy Tertiary stratigraphic record dominated by fluvial deposition and extending from at least late Paleocene through late Miocene time. The thickest and oldest Tertiary sections are best exposed in a system of partially re-exhumed Laramide paleocanyons. The Paleogene drainage system was locally disrupted and ponded by Laramide monoclines. In pre-Oligocene time, extensive alluvial fans spread southward from the Shivwits Plateau scarp across the current location of the modern Colorado River gorge to the northern margin of the Laramide drainage system at Hindu Canyon. Locally derived, fluvial Buck and Doe Conglomerate subsequently filled the disrupted Paleogene channels, spilled out over the local interfluves, and formed an extensive aggradational surface of low relief by late Oligocene time. Early Miocene volcanism filled in much of the relict Laramide relief. Erosional recession of the adjacent Shivwits Plateau escarpment shifted the northern Hualapai Plateau margin 8 km northeastward after the Laramide drainage episode and before the incision by the modern Colorado River. Partially exhumed tributaries to the Hindu Canyon paleochannel and associated sedimentary deposits bordering the southern edge of the Grand Canyon gorge demonstrate that local surface runoff flowed south, away from the modern Grand Canyon location, during early Paleogene time. Headwardly eroding Colorado River tributaries exhumed, captured, and reversed the flow of these tributaries to the Laramide canyon, beginning in late Miocene or Pliocene time. The geomorphic and stratigraphic records show no evidence of, and provide no space for, incision of a Late Cretaceous–Paleogene ancestral precursor to the modern Colorado River gorge. Instead, all the field evidence clearly supports a late Miocene–Pliocene origin for integration of the western Grand Canyon on the central Hualapai Plateau with the upper Colorado River.


Grand Canyon Controversy

The western Grand Canyon on the Hualapai Plateau (Fig. 1) has recently become the focus of apatite U-Th/He and fission-track studies by researchers debating the evidence for and against the possible existence of an ancestral Grand Canyon in the same location and nearly as deep as the modern Colorado River gorge as early as 70 Ma (Flowers and Farley, 2012, 2013; Wernicke, 2011; Karlstrom et al., 2013, 2014). However, stratigraphic and geomorphologic field evidence directly conflicts with the existence of a deep Paleogene canyon coinciding with the location of the modern Colorado River gorge, although an argument has been made that headward erosion from the west could have begun gradually to establish the modern canyon’s course in middle to late Miocene time (Young, 2008), slightly earlier than the conventionally accepted time for integration of the Colorado River at 6–5 Ma. The strongest evidence against a Paleogene ancestral Grand Canyon includes: (1) a nearly continuous stratigraphic record documenting a lengthy episode of Paleocene through Miocene deposition throughout the Hualapai Plateau, and (2) fanglomerate deposits of Paleocene–Eocene age preserved along the south rim of the Grand Canyon that contain distinctive sedimentary clasts derived from the younger Paleozoic rocks capping the Shivwits Plateau escarpment on the opposite side of the canyon (Fig. 2). This paper focuses on the Tertiary events most closely associated with the Laramide-Paleogene stratigraphic record and the accompanying evolutionary changes in the Hualapai Plateau drainage system that clearly preceded the modern Grand Canyon.

Condensed Tertiary History

The Hualapai Plateau in northwestern Arizona contains one of the most complete geologic records of Tertiary events on the Colorado Plateau from Paleocene through Miocene time. This brief review is condensed from Young (1966, 1999, 1982, 2001a, 2001b) and is best understood by viewing geologic maps of the area by Young (1966, 2011), by Billingsley et al. (1999, 2000), and by Wenrich et al. (1996). Laramide events recorded on the Hualapai Plateau began with the uplift and stripping of the Upper Paleozoic sedimentary rocks to form a cuesta-scarp landscape into which canyons were contemporaneously incised (Figs. 1 and 2). The term Laramide in this discussion includes the events from ca. 85 Ma to 40 Ma (Campanian to mid-Eocene) in Arizona as described by Keith and Wilt (1985), and by the corresponding radiometric age distribution compiled by Damon (1964). The major Laramide paleocanyon segments include the L-shaped Milkweed-Hindu channel on the central Hualapai Plateau and Peach Springs Canyon, the trunk valley coincident with the Hurricane fault (Young, 1966, 1979, 1982, 2001a). The paleocanyons preserve the greatest buried relief at the plateau margin, 1200 m near Truxton, Arizona, and become shallower downstream to the northeast.

The oldest fluvial deposit that records the nature and existence of the northeast-flowing Laramide drainage system and fills the lowest portions of these abandoned canyons is the Music Mountain Formation, a correlative of the so-called “Rim gravel” of the Mogollon Rim region in central Arizona (Cooley and Davidson, 1963; Young, 1999). The age of the regional base of these extensive gravels is uncertain but may range back to an early or intermediate stage of the Late Cretaceous Laramide uplift. Gravel lenses within the arkosic Music Mountain Formation are dominated by exotic Precambrian crystalline clasts derived from the central Arizona Mogollon highlands and the Kingman uplift, regional Laramide uplands bordering the plateau as described by Cooley and Davidson (1963), Lucchitta (1966), and Faulds et al. (2001). The vertical changes in gravel clast lithologies clearly record the progressive unroofing of the adjacent source terranes (Young, 2001b).

Along the margins of the Hualapai Plateau paleocanyons, locally derived colluvium, debris flows, and landslides, including meter-sized Paleozoic limestone blocks, form a crudely stratified orange fanglomerate inferred to have formed under humid to subtropical oxidizing conditions. The Hindu Fanglomerate interfingers laterally with the Music Mountain Formation along channel margins and near prominent scarps.

This Late Cretaceous(?)–Eocene history is similar to the geologic record preserved along the plateau margin in southern Utah described by Goldstrand (1990, 1992). The approximate timing of the primary episode of Laramide uplift and timing of the initial erosion of the southwestern Colorado Plateau are constrained as occurring prior to 50 Ma by thermochronologic data (Flowers et al., 2008). Fossil data imply widespread erosion prior to Eocene time (Young and Hartman, 2011).

The Laramide drainage on the Hualapai Plateau was eventually disrupted by structural deformation along Laramide monoclines, which were subsequently beveled by erosion prior to late Oligocene time. Throughout the remainder of the Tertiary history, the basins formed by the isolated portions of the Laramide canyons became closed depocenters, which have preserved a continuous geologic record of events until incision by modern Colorado River tributaries. At the time of Laramide drainage disruption and concurrent cessation of Music Mountain deposition, the local bedrock relief on the Laramide channels varied from 75 to 600 m, with average relief being greater along the Peach Springs Canyon–Truxton Valley segment (Fig. 1).

Following ponding and formation of freshwater limestones in localized basins on the upstream sides of the monoclines within Milkweed and Peach Springs Canyons, the Laramide paleochannels were completely filled with locally derived Buck and Doe Conglomerate. The change from exotic, distantly derived clasts, to locally derived gravels suggests that undocumented Paleogene faulting might have occurred along the southern Grand Wash Cliffs to account for the exclusion of the earlier Precambrian sediment sources. The southern Grand Wash Cliffs fault scarp that forms the plateau margin is more erosionally embayed and presumably much older than the linear Miocene fault boundary nearer Lake Mead (Lucchitta, 1966; Beard et al., 2007).

The basal Milkweed member of the Buck and Doe Conglomerate is dominated by clasts from Cambrian through Mississippian carbonates that crop out on the western Hualapai Plateau. In the Peach Springs Canyon headwaters, a younger conglomerate member deposited during the Buck and Doe episode records an influx of more exotic clasts derived from local Precambrian basement exposures around the margins of the Truxton Valley (Fig. 1), a wide erosional reentrant in the plateau margin that coincides with the trend of the Hurricane fault and parallel structural trends in the Precambrian basement (Beard and Lucchitta, 1993). Although this younger, arkosic, Peach Springs member of the Buck and Doe Conglomerate superficially resembles the basal Music Mountain Formation, the crystalline clasts in the younger Peach Springs member of the Buck and Doe Conglomerate are clearly derived from identifiable Precambrian outcrops currently bordering the Truxton Valley and are markedly less weathered (Young, 1999; Beard and Lucchitta, 1993). This vertical sedimentological change present in the Peach Springs member records the progressive headward erosion by upper Peach Springs Wash tributaries as they expanded southwestward into the Truxton Valley following the demise of the Laramide drainage.

The two Buck and Doe Conglomerate members filled most of the remaining paleocanyon topography and gradually covered the central Hualapai Plateau to form a relatively uniform surface of low relief at elevations close to 1462 m (4800 ft). This episode of fluvial deposition was disrupted locally by volcanism at the onset of the Basin and Range orogeny. The uppermost beds of the Milkweed member of the Buck and Doe Conglomerate on the Hualapai Plateau locally contain reworked volcanic detritus, which heralds the onset of Basin and Range volcanism adjacent to the plateau. Basalt flows, as well as the widespread Peach Spring Tuff, breached the topographic lows along the southern Grand Wash Cliffs, and flowed northeastward onto the plateau, although a few basalts have eruptive vents located on the Hualapai Plateau proper. The brief volcanic episode lasted from ca. 19 Ma to 16 Ma.

During and following Miocene volcanism, fluvial deposition of locally derived Willow Springs Formation occurred across the lower elevations throughout the Hualapai Plateau. This Miocene–Pliocene gravel locally exceeds 100 m in thickness, but it is typically less indurated than similar Buck and Doe Conglomerate beds. Absent basalt flows, the only clear distinction between these local postvolcanic gravels and the underlying Buck and Doe Conglomerate Milkweed member is the inclusion of volcanic clasts in the younger Coyote Springs Formation. In the more isolated Truxton Valley, postvolcanic fluvial deposition continued into Pleistocene time (Twenter, 1962). The timing of cessation of deposition of the Coyote Springs Formation and subsequent incision by Colorado River tributaries is not obvious from any evidence observed in the youngest sediments.

Chronologic Control

The Tertiary sedimentary fill on the Hualapai Plateau has a maximum thickness of 420 m (1380 ft) in Peach Springs Wash (Young, 1999, 2011). The limited geochronologic control for the entire Tertiary section preserved on the Hualapai Plateau is based on several gastropod species in the Music Mountain Formation at Long Point (Young and Hartman, 2011, 2014), a 24 Ma Oligocene ash bed in the Peach Springs member of the Buck and Doe Conglomerate (Young et al., 2011), the 18.78 Ma Peach Spring Tuff (Ferguson et al., 2013), and two ca. 19 Ma ages on basalt flows located on the Hualapai Plateau near Peach Springs and at Separation Canyon (Wenrich et al., 1995).

Near Long Point (Fig. 1), freshwater limestones up to 30 m thick that crop out in the upper portion of the erosionally truncated Music Mountain Formation suggest that Laramide structures were effective in more broadly ponding the Laramide drainage on the western Colorado Plateau (Young, 1982; Young and Hartman, 2011). Deformation along the Kaibab upwarp or the Supai monocline, parallel to Cataract Creek, could have dammed Laramide drainage north of Long Point (Young, 1982; Fig. 1). Early Eocene or late Paleocene(?) gastropods occur in a series of thin limestone lenses at Duff Brown Tank (Fig. 1; Young and Hartman, 2011, 2014).

The recently located ash bed near the top of the Peach Springs member of the Buck and Doe Conglomerate in Peach Springs Wash is 24.12 ± 0.04 Ma, or late Oligocene (see Appendix 1). The ash locality is 2.4 km northeast of the town of Peach Springs (Young, 2011) and strengthens the argument for the proposed Laramide age of the thick Paleogene sedimentary sequence below (Table 1; Young et al., 2011). The evidence of widespread aggradation from late Laramide through Miocene time in locations scattered across the Hualapai Plateau would appear to be incompatible with contemporaneous incision in immediately contiguous areas.


The stratigraphic and structural controls that shaped the western Grand Canyon region differ from those that determined the Colorado River’s course elsewhere on the Colorado Plateau in Arizona (Young, 1982, 1985). The strike-parallel western Grand Canyon, augmented by tributary runoff from the northeast-dipping Hualapai Plateau strata, trends northwest, parallel to the base of the recessional Shivwits Plateau scarp (Figs. 1 and 2). The north and south canyon rims in this area differ both stratigraphically and in elevation by 365 m (1200 ft), i.e., the relief on the scarp that forms the southwest-facing edge of the Shivwits Plateau on the north side of the modern Grand Canyon (Figs. 2, 3, and 4).

Young (1966, 1999, 2001a, 2001b) mapped and redefined eight Tertiary stratigraphic and volcanic units (Fig. 5), described the Laramide step-bench or cuesta-scarp landscape of the Hualapai Plateau, and proposed that the older Milkweed-Hindu Canyon paleochannel was deflected eastward toward the Hurricane fault zone from its dip-parallel course when it encountered the colluvial footslope of the Shivwits scarp at a preexisting Laramide position (Young, 1982, 1985; Figs. 3 and 4). The right-angle bend in the paleocanyon drainage pattern from Milkweed into Hindu Canyon (Fig. 1) is logically the result of a classic, dip-parallel, scarp-controlled drainage pattern developed on a cuesta-scarp landscape formed by the northeastward recession of an erosional scarp resulting from the juxtaposition of weak and strong strata (Garner, 1974; Chorley et al., 1984; Young and Wray, 2000). In such a classic, structurally controlled landscape, the secondary consequent and resequent streams and their major tributaries would be constrained to follow the northeast structural slope until they encountered the next resistant layer, which in this case was near the base of the Kaibab-capped Shivwits scarp at a transient Paleocene(?) position during the regional unroofing of the adjacent Laramide uplift. Occasional trunk streams, such as the Peach Springs paleochannel, must eventually cross through such regional scarps in order to allow the effective excavation or downwasting of the regional landscape. In this regard, the course of the Peach Springs Canyon paleochannel through the presumed topographic barrier was predetermined, in part, by the topographic and structural irregularities coincident with the Hurricane fault trend, the site of an older east-verging monocline (Young and Huntoon, 1987; Young, 1989b; Elston and Young, 1991; Huntoon et al., 1981).

The Milkweed-Hindu paleochannel could only have been incised into the central Hualapai Plateau surface at the preserved stratigraphic level after the Mesozoic and Upper Paleozoic rocks had been stripped down to the level of the lower Supai Group strata and Redwall Limestone (Figs. 3 and 4) in Late Cretaceous or Paleocene time, an episode accompanied by the regional recession of the Shivwits scarp to a position that must have been much closer to the location of the modern Colorado River gorge. The distance from the Laramide scarp edge southward to the Hindu channel at that time would have been dependent on the distance that the bedrock pediment and alluvial fans extended outward from the vertical scarp face, which is as much as 1–2 km along the modern scarp. The elevation and regional extent of the former stripped surface that now forms the central Hualapai Plateau can be reasonably inferred from the extensive distribution of the Oligocene Buck and Doe Conglomerate and early Miocene volcanic flow remnants (Young, 1966; Wenrich et al., 1996; Billingsley et al., 1999).

The Hindu reach of the Milkweed-Hindu paleocanyon system, 8 km south of Grand Canyon, was the major western tributary for the northeast-flowing Hualapai Plateau drainage system (Fig. 1) until the Laramide monoclinal damming of lower Milkweed Canyon above its junction with Hindu Canyon (Young, 1999; Huntoon, 1981). The related but deeply buried paleocanyon segment directly beneath the town of Peach Springs was also dammed by monoclinal deformation in Paleocene or early Eocene time (Young, 1979, 2011). The formation of thick freshwater limestones (14 m and 118 m, respectively) within these two channels on the up-gradient sides of both monoclines is convincing evidence for the temporal relationship between the monoclinal deformation and the Laramide drainage disruption (Young, 1979, 2011). The widely distributed monoclines of the Colorado Plateau have been shown to be uniquely compressional Laramide features (Huntoon, 1981), and the stratigraphic record implies a geologically abrupt termination of the previously through-flowing drainage.


Sedimentary scarps do not remain in fixed positions during long intervals of geologic time. Their very existence demonstrates that scarp recession is, and has been, a global geomorphic phenomenon, regardless of whether the local climate was arid, semiarid, or humid (Prince et al., 2010). Global studies of modern scarp recession rates have focused recently on the controversy between the relative significance of vertical erosion as compared with lateral scarp recession along major continental rift margins (Matmon et al., 2002; Gilchrist et al., 1994; Moore and Blenkinsop, 2006; Kale, 2010). Kooi and Beaumont (1994) concluded that scarp recession is more likely to occur on scarps that are also drainage divides. There also is evidence that some major scarps do not always undergo the significant degree of lateral retreat originally argued by King (1953), such as the slow evolution of the Darling escarpment of western Australia, proposed by Jakica et al. (2011). However, the dramatic “Grand Staircase” physiography of northern Arizona and southeastern Utah demonstrates that regional scarp recession was an important aspect of latest Cretaceous and Cenozoic Colorado Plateau history. Schmidt (1989) estimated the average scarp recession rates for a number of prominent Colorado Plateau scarps and concluded that variable recession rates are controlled by the relative thickness and lithologic attributes of the resistant cap rocks, a view shared by Moore and Blenkinsop (2006) for southern Africa.

Historically, very few precise estimates of actual scarp recession rates were published prior to 2000, except in areas partially covered by dated volcanic rocks (Young and Wray, 2000), in locations containing uniquely located packrat nests (Cole and Mayer, 1982), by a combination of qualitative morphometric and stratigraphic analysis (Schmidt, 1989), or where suitable radiocarbon samples were available (Gutierrez et al., 1998). Australian scarps in volcanic-capped sedimentary strata have retreated at rates of 125–250 m/m.y. (Young and Wray, 2000). Moore and Blenkinsop (2006) discussed the extensive literature on the resistant Drakensberg scarp of southern Africa and its estimated retreat rate since Late Cretaceous time of between 330 and 435 m/m.y. Schmidt (1989) calculated a 500 m/m.y. recession rate in Arizona for the Kaibab Formation scarp, which agrees reasonably with the 450 m/m.y. figure cited by Cole and Mayer (1982) for the same formation using an alternative approach. Gutierrez et al. (1998) used radiocarbon dating of surfaces on talus flatirons over a 35,000 yr interval to measure scarp retreat rates on semiarid Spanish scarps of 0.9–1.0 km/m.y. These semiquantitative data suggest that lateral scarp recession rates between 125 and 500 m/m.y. (0.125–0.5 mm/yr) are reasonable and likely include the slow end of the scarp recession range for moderately resistant cap rocks overlying weaker shales, such as the Kaibab scarp of the Shivwits Plateau.

More recently, the newer age dating techniques that use cosmogenic isotopes and fission-track methods have provided more quantitative measures of the average rates of scarp recession in different terranes. Persano et al. (2002) proposed that the escarpment retreat rate on the Great Escarpment in southeast Australia may have been as rapid as 5–10 km/m.y. Steckler and Omar (1994) projected average rates of scarp recession of 6 km/m.y. along the Red Sea rift. Studies of the Drakensberg escarpment by Brown et al. (2002) indicated a retreat rate of 100–200 m/m.y. This rate is similar to the maximum rate of 95 m/m.y. proposed by Fleming et al. (1999) for the same feature. These studies suggest that earlier, qualitative estimates are near the low end of the possible range. However, even modest erosion rates obviously will cause scarps to migrate significantly over millions of years, especially where relatively weak shale substrates strongly influence the failure mode of more resistant cap rocks, such as on the Hualapai Plateau.

Scarp Recession and Hualapai Plateau Drainage History

Given the global evidence that sedimentary scarps retreat significantly throughout long periods of geologic time, the modern Shivwits Plateau scarp could not have remained in a fixed position between the initiation of the Laramide-age Hindu Canyon channel and the conventionally cited time of late Miocene or Pliocene Grand Canyon incision, irrespective of how slow the actual cliff recession rate may have been. Put another way, if an ancestral western Grand Canyon had existed in its current location as long as 70 m.y. ago, the Shivwits scarp would not currently be in such close proximity (3–5 km) to the northern canyon rim. It could have retreated some 14 km at even a modest rate of 200 m/m.y., based on the studies cited previously. This calculation does not include allowance for the 1–2 km distance from the scarp edge down to the base of the adjacent alluvial slope, the closest distance from the scarp face at which a regional subsequent stream would likely form. The evidence of modern Shivwits Plateau scarp recession also can be appreciated by considering the widespread volume of coarse colluvial material currently spreading southward from the scarp face. Simply restoring this geologically young mobile sheet of debris to the scarp face would obviously restore the escarpment to a position slightly south of its current location over the short term.

Why would an ancestral Grand Canyon, parallel to Hindu Canyon, be an unlikely component of the Laramide drainage? The logic and field evidence are inferred from the following observations. The basic structure and geomorphology of the modern Hualapai Plateau and its abandoned Laramide paleocanyons imply that the position of the scarp defining the southern margin of the Shivwits Plateau was ∼8 km to the southwest of its current position when that interim Laramide location is presumed to have deflected the paleodrainage at Hindu Canyon eastward toward the trunk canyon that paralleled the modern Hurricane fault. This is inferred from the dual facts that (1) scarps cannot remain stationary through long intervals, and (2) the modern Colorado River gorge is parallel to, but 8 km north of, the older Hindu Canyon paleochannel (Figs. 1, 3, and 4). The logical assumption is that the modern Colorado River canyon was constrained to have formed in an identical scarp-controlled position as did the older Hindu Canyon, although flowing in the opposite direction (Figs. 1 and 4). Incision of scarp-parallel canyons for both drainage systems would be inevitable, given the step-bench topography and northeast-flowing tributary drainage that characterize the northeast-dipping bedrock platform (Young, 2008). However, scarp-parallel canyons do not form at the immediate base of a vertical scarp having significant relief in a cuesta-scarp landscape. The natural slopes created by mass wasting and overland flow processes distribute colluvial material along a natural and predictable range of pediment profiles that must, by simple inspection of existing landscapes, extend hundreds of meters to a kilometer or more outward from any vertical scarp face having hundreds of meters of relief, such as the Shivwits Plateau margin. The position of such rivers, prior to incision, will always be located near the inflection point between the natural dip or slope of the rocks below and the opposing slope of the colluvial debris and alluvial fans that currently extend as much as 1–2 km outward from the face of the existing Shivwits scarp.

Furthermore, the development of erosional reentrants formed by irregular headward erosion by competing scarp-face drainages will logically expand with time as the scarp retreats. The irregularities in a modern scarp such as the Shivwits must be much greater now than when the scarp was located at a more southerly position, nearer Hindu Canyon. Headward erosion has resulted in the random capture of dip-parallel drainages previously flowing north on the Shivwits Plateau proper, a factor that contributes to scarp irregularity by increasing the effective power of the scarp-face streams that fortuitously capture the largest share of the available north-flowing drainage above and behind the scarp.


Space Constraint

The widespread late Eocene(?)–Oligocene Buck and Doe Conglomerate, the remnants of which cover the Hualapai Plateau up to the very edge of the modern Colorado River canyon, indicate that post-Laramide incision was delayed until well after a lengthy Early Tertiary episode of fluvial deposition, followed by Miocene volcanism, the upper limit of which is minimally constrained by emplacement of the widespread Peach Spring Tuff and associated basalts, as well as abundant younger Miocene fluvial sediments (Fig. 5). Miocene basalt flow remnants capping Buck and Doe Conglomerate on the south rim of Grand Canyon, next to south Separation Canyon (Separation Hill basalt, 19.0 ± 0.4 Ma; Wenrich et al., 1995), demonstrate that the Laramide paleocanyons were filled with Oligocene sediments deposited by northeast-flowing tributaries, before being capped by Miocene Peach Spring Tuff and other local basalt flows (Fig. 6). The greater part of the preexisting Laramide canyon relief was essentially buried by early Miocene time (ca. 19 Ma). (Clarification: Modern U.S. Geological Survey topographic maps [1:24,000], unlike historic maps, show Separation Canyon as consisting of two aligned, fault-controlled, segments that are located on opposite sides of Grand Canyon [Fig. 1], as opposed to the name originally being applied only to the much longer northern tributary, as described in older publications and made famous by the Powell expedition. The smaller tributary near Bridge Canyon is referred to informally as “south Separation Canyon” to avoid confusion in this discussion.)

The locations of the Separation Hill basalt and its source vent, now on opposite sides of the 730-m-deep Spencer Canyon, also demonstrate that the central Hualapai Plateau was relatively flat and featureless and lacked any significant tributary relief that would have prevented the relatively thin lava flows from reaching the Separation Hill locality (Young, 2011). The eruptive center mapped by Billingsley et al. (1999), located 13 km southwest of the head of south Separation Canyon and on the opposite side of Spencer Canyon, is the only basalt source in the vicinity with an elevation sufficiently high (1516 m [4975 ft]) to allow the preserved flow remnants to have reached their current position (1504 m [4936 ft]) capping Separation Hill.

The 8 km north-south separation of the modern Grand Canyon from the older Hindu Canyon paleochannel is the indirect measure of the amount of scarp recession that must have occurred between the inception of Laramide drainage and the beginning of modern Grand Canyon incision (Fig. 4). The time interval during which the scarp recession of 8 km occurred can be constrained, at a minimum, between late Oligocene time and 5 Ma, the conventionally cited onset of Colorado River incision, or back to early Paleocene time at a maximum. This time frame provides broad limits between 20 and 60 Ma for the 8 km of scarp recession, which translates to an average recession rate of between 400 and 135 m/m.y. Such erosion rates are entirely consistent with the published range of moderate scarp recession rates previously cited.

If one assumes a moderate, intermediate rate of scarp recession of 200 m/m.y., the Shivwits cliffs, capped by Kaibab limestone, would have retreated 8 km in ∼40 m.y. (Fig. 4). Forty million years is consistent with the evidence that Hindu Canyon has been inactive since drainage was tectonically disrupted, no later than middle Eocene time (40–48 Ma ago). Using the faster 450–500 m/m.y. rates for Kaibab scarps estimated by Cole and Mayer (1982) and by Schmidt (1989) would reduce the recession time to 16–18 m.y. These admittedly qualitative geomorphic relationships imply that the western Grand Canyon could not have formed north of the comparable Laramide paleochannel at Hindu Canyon until the Shivwits scarp had retreated at least 8 km from its intermediate Laramide (Hindu) position, which is assumed to have shaped the right-angle eastward deflection from Milkweed Canyon into Hindu Canyon. Obviously, this interval of scarp recession could only have begun following the initial establishment of the Hindu channel. Thus, the physical circumstances require a significant post-Paleocene interval of scarp recession necessary to widen the Hualapai Plateau so as to create sufficient space for an entirely new, post-Hindu drainage to form in the current location of the modern Grand Canyon gorge. Assuming the Laramide Milkweed-Hindu paleodrainage was established no later than late Paleocene time, but probably earlier, it seems clear that cliff recession sufficient to allow the Grand Canyon to form 8 km further north most likely involved some tens of millions of years. This physiographic constraint does not allow for any deep, proto–Grand Canyon to form in its current location prior to Oligocene time (34–23 Ma) at the very earliest, assuming even a modest rate for necessary scarp recession. Logically, and from a classic geomorphologic perspective, there would be no need, nor sufficient runoff, for two such large, parallel, and competing canyons to have formed simultaneously in their current positions.

Furthermore, the formation of any proposed deep ancestral Grand Canyon should be reflected by the coincident onset of necessary and integrated tributary incision throughout the Hualapai Plateau, rather than the observed episode of Paleocene–Miocene aggradation, which is widely preserved throughout the closed basins and across the beveled surface of the Hualapai Plateau, including the small erosional remnants of sediments and lavas preserved at Separation Hill, only 1200 m south of the modern Grand Canyon rim (Figs. 3, 4, and 6). Such necessary, supportive tributary development would naturally evolve on the northeast structural slope of the Hualapai Plateau, especially if the implied time frame were as long as 70 m.y.

Although this simplistic analysis, based on dated volcanic units, geomorphic relations, estimated scarp recession rates, and stream capture, fits the approximate timing of established geologic events for the interval in question, more direct evidence also precludes any incised, ancestral Grand Canyon from occupying its current position near Separation and Bridge Canyons until well after late Oligocene time.

Stratigraphic Constraints in the Bridge-Separation-Hindu Canyon Region

As previously noted, the Music Mountain Formation interfingers laterally with the penecontemporaneous Hindu Fanglomerate of local derivation (Fig. 5), and both are capped by the locally derived Buck and Doe Conglomerate in the region where tributaries to Separation, Bridge, and Hindu Canyons are in close proximity, as shown on Figures 3 and 4 (Young, 1989a; Billingsley et al., 1999; Elston and Young, 1991). Buck and Doe Conglomerate remnants (Figs. 3 and 4) mark the low-relief aggradational surface extending across much of the central Hualapai Plateau northwest of Peach Springs Canyon at a surface elevation between 1460 and 1430 m (4800–4700 ft). The locally derived Buck and Doe gravels spilled out across the local divides up to, and probably beyond, the encroaching erosional rims of modern Separation and Bridge Canyons immediately prior to the onset of widespread Miocene volcanism, as documented by the position of the 19 Ma Separation Hill basalt and underlying Buck and Doe gravel (Figs. 3, 4, and 6). This episode of fluvial aggradation, which followed the deposition of the Music Mountain Formation and the Hindu Fanglomerate, is uninterrupted by any evidence of local or regional incision by Hualapai Plateau drainages, even at the very rim of the western Grand Canyon gorge, and as further documented by the undisturbed stratigraphic sequences preserved in the structurally closed basins in Milkweed, Hindu, and Peach Springs Canyons.

The contact between the two informally designated members of the Buck and Doe Conglomerate (Fig. 5) can be traced northwest from Peach Springs Canyon to a position ∼6 km southeast of Bridge Canyon on the broad divide between Hindu and Peach Springs Canyons (Fig. 1). The juxtaposition, uniform contact, and distribution of the two members demonstrate that they formed in close succession, reflecting only the headward expansion of the post-Laramide drainage southwestward from Peach Springs Wash and into the adjacent Truxton Valley. The upper arkosic Peach Springs member of the Buck and Doe Conglomerate is relatively localized but ranges up to 75 m (250 ft) thick in Peach Springs Wash, the headward reach of Peach Springs Canyon. The lower Milkweed member is thinner near Peach Springs but thicker and more widespread throughout the central Hualapai Plateau. The Peach Springs member contains a volcanic ash (Fig. 7) located ∼2 m from its upper contact near the town of Peach Springs. The originally reported Oligocene age of this ash (23.97 ± 0.03 Ma; Young et al., 2011) was recalculated as 24.12 ± 0.04 Ma using the new Fish Canyon sanidine monitor age of 28.201 Ma (Kuiper et al., 2008; Fig. 8; see Appendix 1). A potential source for the ash is the Aquarius Mountains (Fig. 1; Young and McKee, 1978). The ash, located so high in the 420-m-thick Tertiary section (Young, 2011), lends further credence to the other evidence for the Eocene or older age of the underlying, more deeply weathered Tertiary sequence on the Hualapai Plateau (Young and Hartman, 2011).

The distribution of the Buck and Doe Conglomerate and the Oligocene age of the uppermost beds of the Peach Springs member clearly attest to the fact that fluvial sediments were continuously accumulating throughout the lower elevations of the plateau during much of Paleogene time. The only conspicuous depositional hiatus, preserved as a thick red soil horizon at the top of the Music Mountain Formation (Figs. 5 and 9), seems to mark a period of low sediment production, but not necessarily measurable erosion or drainage incision. This depositional hiatus, or paraconformity, clearly marked by an abrupt and distinctive upward red-to-buff color transition, may simply mark the Eocene-Oligocene transition from subtropical conditions to a local climate that is documented regionally and globally as recording marked cooling followed by increasing aridity (Prothero and Berggren, 1992). Peterson and Abbott (1979) described the Paleocene–Eocene paleosol sequences of nearby southwestern California also as recording a widespread change from Paleocene tropical lateritic soils to a semiarid climate beginning in late middle Eocene time, and based, in part, on changes in clay mineralogy.

The sources for the basal Milkweed member of the Buck and Doe Conglomerate in the Separation-Bridge-Hindu Canyon divide region were the Paleozoic rocks exposed along the more elevated southwestern edge of the Colorado Plateau or southern Grand Wash Cliffs, locally referred to as the “Music Mountains” (Fig. 1). This includes clasts from the Tapeats through the Redwall formations, which cover most of the adjacent Hualapai Plateau. The sharp lithologic transition from exotic (arkosic) Music Mountain to local Paleozoic formation clasts of the Buck and Doe Conglomerate throughout the Laramide channel system suggests that prevolcanic (pre–late Oligocene) faulting along the southern Hualapai Plateau margin may have begun somewhat earlier than the typically cited mid-Miocene Basin and Range extension episode. Rocks in the Whipple tilt block domain south of Lake Mead show that faulting began in the interval between 32 and 22 Ma (Spencer and Reynolds, 1989).

Between the modern Colorado River gorge and the Hindu Canyon paleochannel, in the vicinity of south Separation and Bridge Canyons, the lower Buck and Doe Conglomerate is preserved as remnants of a relatively flat depositional surface near 1455 m (4775 ft) in elevation (Figs. 3, 4, and 10). In this region, adjacent to the Laramide paleochannel proper, the Milkweed member directly overlies several small scattered remnants of the Hindu Fanglomerate (Fig. 10), the lateral equivalent of the Music Mountain Formation, which typically consists of weakly stratified colluvium, debris-flow, and similar mass-wasting deposits shed into the Laramide channels from adjacent elevations. There are several more very small outcrops of the fanglomerate in the vicinity of Bridge and Separation Canyons than realistically can be shown on Figures 3 and 4.

Shivwits Plateau Source of Hindu Fanglomerate

The color of the Hindu Fanglomerate contrasts with the enclosing formations because of its thoroughly oxidized matrix, although the typical orange hue occasionally appears somewhat bleached at more weathered exposures (Figs. 10B and 10C). In the broad upland surface between south Separation and Bridge Canyons (Figs. 3 and 10A), the scattered remnants of Hindu Fanglomerate preserved beneath the Buck and Doe Conglomerate show obvious imbrication (Fig. 11) documenting southward transport and clearly represent the distal portions of broad alluvial and debris fans that built southward from the distant Shivwits Plateau scarp, similar to the modern fans that extend southward from the Shivwits scarp today (Fig. 12). The fanglomerate in this restricted upland (Fig. 3) contains numerous conspicuous light-colored cobbles (Fig. 11) that could only have been derived from the more resistant Coconino, Toroweap, and Kaibab Formations that crop out only to the north of the Grand Canyon where they cap the Shivwits Plateau scarp (Figs. 4 and 12). The readily identifiable Coconino Sandstone thins westward and is no longer a significant source for clasts west of Kelly Point at the Bridge Canyon longitude (Wenrich et al., 1996). However, the Toroweap and Kaibab Formations also include conspicuous, light-colored sandy beds, sandy limestones, and sandy dolostones within the upper 120–150 m (400–500 ft) of the Shivwits scarp north of Grand Canyon, where they contrast sharply with the underlying Permian red beds (Figs. 2 and 12). No other similar rock types occur lower in the Paleozoic section south of Grand Canyon, nor are any similar clast lithologies present in the southwesterly derived Music Mountain or Buck and Doe formations. The clast imbrication at several small exposures scattered throughout the Bridge-Separation Canyon region (Fig. 11) combined with the conspicuous light-colored clasts clearly document a northerly source for the weakly stratified Hindu Fanglomerate, namely, the Shivwits scarp that forms the northern side of western Grand Canyon. The modern Shivwits scarp fans that provide the closest approximation of Laramide transport conditions, but in a Quaternary climatic setting, were probably more active during the Pleistocene, but are currently undergoing marked headward dissection under the prevailing semiarid conditions, combined with incision by steep modern Colorado River tributaries (Fig. 12). Elsewhere, throughout the Hualapai Plateau paleocanyon system, the Hindu Fanglomerate clearly represents local valley-slope mass-wasting deposits with no evidence of extremely well-rounded or far-traveled lithologies.

The Hindu Fanglomerate provenance and dispersal pattern near south Separation and Bridge Canyons preclude the existence of any significant canyon incision by a river in the current location of the western Grand Canyon during fan construction. The general time limits established for the Hindu Fanglomerate and the Buck and Doe Conglomerate are conservatively established as between early Eocene and late Oligocene time, the interval spanning 24–48 Ma. This 24 m.y. interval also is estimated to be a realistic time that would be required for the Shivwits scarp to recede the 8 km required to create the necessary space for post–Hindu Canyon erosion to allow a large ancestral canyon to form where Grand Canyon gorge is located today. However, field relationships clearly indicate that the local plateau surface was first covered by northerly derived alluvial-fan debris of the Hindu Fanglomerate and then capped by southwesterly derived Buck and Doe gravels and volcanic rocks during the very time that some speculate that a large precursor to Grand Canyon was already in existence (Flowers and Farley, 2012; Wernicke, 2011). The field evidence for this long-term sediment aggradation from both directions, the southwestern Hualapai Plateau margin and the Shivwits Plateau scarp, clearly precludes the formation or existence of an incised, through-flowing drainage system crossing the Hualapai Plateau in the western Grand Canyon region between Paleocene and early Miocene time.


Modern tributaries, such as those in Spencer and Bridge Canyons and similar nearby drainages flowing north into the Grand Canyon, have headwardly eroded and captured remnants of the Milkweed-Hindu channel system (Figs. 1 and 13). This relationship also implies that the Hindu channel is older than Grand Canyon. The classic barbed tributary pattern (Fig. 13) is best developed where 4-km-long Bridge Canyon has nearly intercepted the main Hindu Canyon paleochannel (Young, 1970). As can be determined from the drainage pattern, the barbed tributaries were initially captured less than 2 km from the present south rim of Grand Canyon.

In a similar fashion, the more effective headward erosion by the much longer Spencer Canyon (Figs. 1 and 3) has completely severed the Laramide connection of Milkweed to Hindu Canyon and reversed the western Hindu Canyon drainage to its present westward course, leaving only a much shorter eastern segment, Lost Man Canyon, to drain eastward into Peach Springs Canyon. Spencer Canyon appears to have had a structurally determined hydrologic advantage due to its close parallelism to the east-verging Meriwhitica monocline (Huntoon, 1981).

Pliocene Erosion Rate at Bridge Canyon

If the western Grand Canyon has only existed since early Pliocene time, any tributaries that have no structural advantage and that have expanded primarily due to simple headward erosion should have a volume and area proportional to this relatively short time frame. Bridge Canyon, below its distinctive rim, has a volume of ∼2.85 km3 with a drainage area of 20 km2 (Fig. 13). Is it reasonable that erosion could produce a tributary canyon of this size in five million years? Quantitative studies of sediment yields in arid regions indicate that basins of this relatively small size typically produce between 100 and 800 metric tons of sediment per square kilometer per year (Griffiths et al., 2006; Clapp et al., 2000). Using these estimates, it would require removal of an average volume of between 1 and 2 m3/d in order to create a feature the size of Bridge Canyon in 5 m.y. Given the canyon and basin dimensions and an average limestone density of 2600 kg/m3, this would require an amount of sediment removal of ∼50–100 metric tons/yr for each square kilometer of basin area. These figures are near the low end of the range of modern sediment production rates for drainages in arid regions, and they disregard any additional volume of bedrock that would be removed by solution in such a predominantly carbonate terrane. This implies that the relatively short Grand Canyon tributaries that characterize much of the western canyon’s south rim could have easily formed within the Pliocene time frame proposed, even under arid conditions. Any hypothesized ancestral Paleogene Grand Canyon should have developed much longer tributaries that would have competed with those entering the Hindu channel, especially given the natural northeast dip slope of the plateau. The numerous stream captures, especially those closest to the Colorado River, and the obvious drainage immaturity imply that tributary flow southward into Hindu Canyon was only interrupted late in the modern incision cycle, rather than tens of millions of years ago when the Hindu channel was the dominant master drainage.

Furthermore, the drainage density for the exhumed tributaries to the Laramide Hindu Canyon that are being captured by Bridge Canyon reflects an entirely different dendritic pattern than that created by Grand Canyon tributaries, which are currently incising into the very same bedrock units (Young, 1970). The distinctively different drainage patterns and drainage densities of the two river systems, which are obvious on aerial or satellite imagery, are undoubtedly related to the more humid climate, thick soil development, and vegetation characteristics of the humid Paleogene Epoch when Hindu Canyon and its exhumed, captured tributaries developed as the dominant master drainage at this location.


The combination of geomorphic, stratigraphic, and chronologic evidence for an extended interval of Paleocene through Miocene depositional events and aggradational conditions on the Hualapai Plateau logically preclude the existence of a Late Cretaceous or Early Tertiary, ancestral Grand Canyon coincident with the modern canyon’s present location. The Shivwits scarp would have been well south of its current location 60–70 m.y. ago, probably covering a substantial portion, if not all, of the location of the modern canyon. Such a scarp location, several kilometers south of its present position, would have prevented any ancestral Grand Canyon from forming in the narrow space that would have been available north of Hindu Canyon. Modest scarp recession rates to broaden the plateau surface between Hindu Canyon and the present Shivwits cliffs would require tens of millions of years, beginning no later than late Paleocene or Eocene time. The existence of broad, south-sloping alluvial fans extending across this same area in Paleocene or Eocene time clearly precludes the presence of an incised older canyon coincident with the modern Colorado River gorge.

The record of consistent aggradation throughout the Hualapai Plateau, interrupted only by a pre-Oligocene weathering episode, and no conspicuous erosion, from early Eocene through late Miocene or early Pliocene time is incompatible with significant contemporaneous canyon incision in the same vicinity. All of the field evidence summarized in this analysis points to the presence of a late Miocene or Pliocene west-flowing Colorado River, possibly integrated with, or initiated by, gradual headward erosion or knick point migration from a drainage basin with modest proportions and limited runoff on the western Hualapai Plateau margin during late Miocene time (Young, 2008).


Sanidine crystals separated from the ash were loading into machined aluminum disks and irradiated for 10 h at the U.S. Geological Survey’s Triga reactor, near Denver, Colorado. Single sanidine crystals were then fused with a Synard CO2 laser and analyzed in a Mass Analyzer 215–250 mass spectrometer at the New Mexico Bureau of Geology and Mineral Resources Geochronology Research Laboratory. Fish Canyon Tuff sanidine was used as a neutron flux monitor with the assigned age of 28.201 Ma (Kuiper et al., 2008). J-factors were determined by fusion of approximately 6 flux monitor single crystals from equally spaced radial positions around the irradiation tray to a precision of 0.04%. Laser sensitivity values, as well as mass discrimination values, for different runs were determined by running a series of airs to 4.99e–17 mol/pA and 1.005, respectively. Correction factors for interfering reactions were determined from analysis of K-glass and CaF2. See Figure 8 and Table 1.

We thank L.S. Beard and anonymous reviewers for their helpful comments regarding the content and organization of this paper. The logistical support of the Frank Hunt family and the cooperation of Loretta Jackson-Kelly of the Hualapai Tribe are greatly appreciated in facilitating R.A. Young’s long-term studies of the geology of the Hualapai Plateau. P. E. Damon and E. H. McKee provided the core of the early radiometric age control that supported the wide-ranging fieldwork, which might not have been completed without the use of I. Lucchitta’s faithful Toyota Land Cruiser. We are especially indebted to our innumerable Colorado Plateau colleagues for freely sharing their ideas during our many informal discussions and joint fieldtrips involving the history and evolution of the Colorado River region.