The spatial and temporal distribution of Pliocene to Holocene Colorado River deposits (southwestern USA and northwestern Mexico) form a primary data set that records the evolution of a continental-scale river system and helps to delineate and quantify the magnitude of regional deformation. We focus in particular on the age and distribution of ancestral Colorado River deposits from field observations, geologic mapping, and subsurface studies in the area downstream from Grand Canyon (Arizona, USA). A new 4.73 ± 0.17 Ma age is reported for a basalt that flowed down Grand Wash to near its confluence with the Colorado River at the eastern end of what is now Lake Mead (Arizona and Nevada). That basalt flow, which caps tributary gravels, another previously dated 4.49 ± 0.46 Ma basalt flow that caps Colorado River gravel nearby, and previously dated speleothems (2.17 ± 0.34 and 3.87 ± 0.1 Ma) in western Grand Canyon allow for the calculation of long-term incision rates. Those rates are ∼90 m/Ma in western Grand Canyon and ∼18–64 m/Ma in the eastern Lake Mead area. In western Lake Mead and downstream, the base of 4.5–3.5 Ma ancestral Colorado River deposits, called the Bullhead Alluvium, is generally preserved below river level, suggesting little if any bedrock incision since deposition. Paleoprofiles reconstructed using ancestral river deposits indicate that the lower Colorado River established a smooth profile that has been graded to near sea level since ca. 4.5 Ma. Steady incision rates in western Grand Canyon over the past 0.6–4 Ma also suggest that the lower Colorado River has remained in a quasi–steady state for millions of years with respect to bedrock incision. Differential incision between the lower Colorado River corridor and western Grand Canyon is best explained by differential uplift across the Lake Mead region, as the overall 4.5 Ma profile of the Colorado River remains graded to Pliocene sea level, suggesting little regional subsidence or uplift. Cumulative estimates of ca. 4 Ma offsets across faults in the Lake Mead region are similar in magnitude to the differential incision across the area during the same approximate time frame. This suggests that in the past ∼4 Ma, vertical deformation in the Lake Mead area has been localized along faults, which may be a surficial response to more deep-seated processes. Together these data sets suggest ∼140–370 m of uplift in the past 2–4 Ma across the Lake Mead region.


The Colorado Plateau region (southwestern USA) was near sea level in the Late Cretaceous, and today its mean surface elevation is ∼2 km (e.g., Pederson et al., 2002b). Sea level has changed by <∼250 m since the Late Cretaceous (Miller et al., 2005), so this elevation difference requires ∼2 km of uplift since that time. However, the timing and processes that produced the uplift remain controversial despite over a century of study (Powell, 1875). Proposed models include Laramide (Late Cretaceous to Eocene) processes associated with shallow subduction of the Farallon slab including: lower crustal flow (McQuarrie and Chase, 2000), whole-mantle flow (Liu and Gurnis, 2010), or lithospheric hydration and/or delamination (Humphreys et al., 2003). Post-Laramide uplift components are attributed to increased Oligocene buoyancy due to magmatic heating (Roy et al., 2009) and/or Neogene and ongoing mantle-driven buoyancy modification (Karlstrom et al., 2008; Moucha et al., 2008; van Wijk et al., 2010; Crow et al., 2011; Levander et al., 2011). Research using low-temperature thermochronology (Kelley et al., 2001; Flowers et al., 2008), clumped isotopes (Huntington et al., 2010), and inversion of river profiles (Roberts et al., 2012) suggests that a component of the uplift occurred between 80 and 50 Ma, during the Laramide orogeny, but it is also recognized that other components of uplift took place in the Oligocene and in the past 5 Ma (Karlstrom et al., 2011; Cather et al., 2012).

Independent geologic data sets are needed to test these models and constrain the timing and magnitude of uplift during different time intervals. One group of studies has focused on using river incision as an indicator of uplift, as rivers respond to base-level fall and/or differential uplift (e.g., Kirby and Whipple, 2001; Karlstrom et al., 2011). However this response can be difficult to resolve from incision modulated by climatic forcing (e.g., Wobus et al., 2010) and changes in substrate erodibility (e.g., Bursztyn et al., 2015).

Incision rates within western Grand Canyon (Arizona, USA) vary markedly across west-dipping normal faults (e.g., the Toroweap and Hurricane faults), with the difference between upstream footwall incision rates and downstream hanging-wall rates being subequal to the throw on the faults (Pederson et al., 2002a; Karlstrom et al., 2007, 2008). At distance from these and other normal faults, this relationship breaks down due to fault-related folding, which lessens the amount of incision dampening in the hanging wall due to reverse-drag folding or hanging-wall rollover, and lessens the amount of incision enhancement in the footwall due to localized footwall upwarps (Howard and Bohannon, 2001; Hanks and Blair, 2004; Karlstrom et al., 2007; Howard et al., 2015). These general relationships have led different researchers to different conclusions about the role these faults play in either regional uplift or subsidence. Pederson et al. (2002a) concluded that there was net regional hanging-wall subsidence across the Toroweap fault. Hanks and Blair (2004) suggested that no regional offset could have occurred across faults in western Grand Canyon because of the expected short-wavelength elastic response to most crustal normal faults. A series of others workers have argued for net regional footwall uplift across normal faults in the western Grand Canyon–Lake Mead region (Karlstrom et al., 2007, 2008). Footwall uplift has been supported not only by differential incision across faults, but also based on analysis of faulted river profiles defined by intra-canyon basalt flows (Hamblin et al., 1981; Hamblin, 1984) and ancestral Colorado River deposits (Howard et al., 2015). Howard et al. (2015) used a 4.5–3.5 Ma ancestral Colorado River deposit, called the Bullhead Alluvium, to suggest that subsequent deformation in the lower Colorado River corridor was primarily accomplished across a fault system in western Lake Mead. Those studies advocating uplift have suggested between >200 m (Howard et al., 2015) and ∼250 m (Karlstrom et al., 2008) of uplift between the lower Colorado River corridor and western margin of the Colorado Plateau in the past ∼6–4 Ma.

Given the varying interpretations of the data, the ongoing controversies regarding the uplift history of the Colorado Plateau, and the role of crustal faults in accommodating uplift, we are revisiting this problem with a focus on the lower Colorado River corridor from western Grand Canyon to the Gulf of California. In particular, we have merged differential incision data sets with improved characterization of the current extent of Colorado River aggradation packages to better understand the extent and magnitude of post-Miocene vertical deformation. We bring the following new contributions to this classic problem:

  1. A much more exhaustive compilation of the positions of ancestral Colorado River deposits in the lower Colorado River corridor from geologic mapping and subsurface studies, which allows for the identification of areas that have been deformed since Colorado River integration and quantification of vertical deformation.

  2. A new incision rate in eastern Lake Mead based on 40Ar/39Ar dating of a basalt flow overlying tributary gravels.

  3. New observations of ancestral Colorado River deposits covered by Lake Mead until 2016 that require modifications to previously published incision rates.


Lower Colorado River Evolution and Deposits

The timing and processes of Colorado River inception and evolution remain controversial despite extensive study. Well-dated gravels suggest that a similar river system existed near Grand Junction, Colorado, by at least 11 Ma (Aslan et al., 2010; Karlstrom et al., 2011). Well-documented strata of the Hualapai Limestone and the rocks of the Grand Wash trough (previously referred to as the Muddy Creek Formation) (Longwell, 1936; Lucchitta, 1972; Bohannon, 1984; Faulds et al., 2001), exposed at the terminus of Grand Canyon in the Lake Mead area (Figs. 1 and 2), indicate that the Colorado River did not arrive at that location until sometime after 6 Ma (Spencer et al., 2001). The timing of river arrival to the same general area is further constrained to before 4.49 ± 0.46 Ma based on 40Ar/39Ar dating of the Sandy Point basalt, which overlies river gravel (Fig. 2) (Faulds et al., 2001, 2016). These are important geologic constraints that must be honored in any model of the processes by which the Colorado River came to occupy its current position and carve Grand Canyon (e.g., Blackwelder, 1934; Hunt, 1969; Hill et al., 2008; Pederson, 2008; Polyak et al., 2008; Wernicke, 2011; Karlstrom et al., 2014). Here we focus on westernmost Grand Canyon and the inextricably linked evolution of the lower Colorado River.

The dominant model for lower Colorado River integration suggests that sometime after 5.59 Ma (House et al., 2008) and likely after 5.24 Ma (Crow et al., 2018b), Colorado River water spilled southward through a series of internally drained basins. Each basin filled with lacustrine and deltaic deposits, referred to as the Bouse Formation (Fig. 3) (Metzger, 1968), until water and then sediment spilled downward into the next basin while the intervening divide was lowered by erosion (House et al., 2008; Pearthree and House, 2014). This model has been widely accepted for the upper basins, but controversy persists in the southern Blythe Basin (Palo Verde Valley area), where conflicting lines of evidence suggest both lacustrine (Spencer and Patchett, 1997; Spencer et al., 2013; Bright et al., 2016) and marine or estuarine (Smith, 1970; Lucchitta, 1972; Buising, 1990; McDougall, 2008; McDougall and Martínez, 2014; O’Connell et al., 2017) depositional environments for Bouse deposits. The specific timing of the deposition of the Bouse Formation is also uncertain. McDougall and Martínez (2014) suggested that the basal carbonates in the Blythe Basin are older than 6 Ma based on the global last occurrence of marine foraminifera species, in which case initial Bouse deposition there would predate the arrival of the Colorado River. At higher elevation in the Blythe Basin, beds of the 4.8–4.9 Ma Lawlor Tuff have been identified interbedded with Bouse carbonate outcrops (Sarna-Wojcicki et al., 2011; Spencer et al., 2013; Harvey, 2014; Miller and Reynolds, 2014). Downstream, magnetostratigraphy suggests that the first Colorado River sand reached the Salton Trough and the proto–Gulf of California at ca. 5.3 Ma (Dorsey et al., 2007). Dorsey et al. (2018) explained this apparent age discrepancy through upstream sediment trapping and marine reflooding of the southern portion of the Blythe Basin at ca. 4.8 Ma. Although refinement is needed, all workers agree that the Bouse Formation was deposited between 6 and 4.8 Ma, at least in part during the first arrival of the Colorado River to the lower Colorado River corridor.

Confirmation of the presence and extent of marine Bouse Formation along with knowledge of paleo–sea level is critical to quantifying the magnitude of uplift. McDougall and Martínez (2014) found marine planktic foraminifera in the southern Blythe Basin in the lower part of the Bouse Formation to an elevation of ∼110 m. Pliocene sea level was 56 m below the modern (Miller et al., 2005) to 39 m above it (Raymo et al., 2009), such that marine Bouse outcrops at 110 m above sea level (asl) would imply between 71 and 166 m of uplift in the past 6 Ma. An additional 220 m of uplift is required if the highest outcrops (currently at ∼330 m asl) are marine as suggested by Dorsey et al. (2018) and others. Alternatively, if the marine fauna were transported by migratory waterfowl and survived in a saline lake (Spencer et al., 2013), the present elevation of marine-like fauna might represent a paleo–lake level.

Deposition of the Bouse Formation was followed by erosion and then accumulation of the Bullhead Alluvium (House et al., 2008; Pearthree and House, 2014) (Fig. 3). This >250-m-thick aggradational package is dominantly composed of coarse, rounded, lithologically diverse gravel and quartz-rich sand (House et al., 2005; Howard et al., 2015). Available age control suggests that Bullhead aggradation was already underway in what is now the Lake Mead area by ca. 4.5 Ma (Faulds et al., 2016). Tephrochronology indicates that Bullhead aggradation culminated in Mohave Valley soon after 4.1 Ma and that the aggradation package had been partially dissected by 3.3 Ma (House et al., 2008; Howard et al., 2015). The rapid Bullhead aggradation has been explained, using cosmogenic-isotope and detrital-zircon evidence, as due to an extraordinary pulse of sediment overload, triggered by rapid incision into readily erodible Colorado Plateau materials and basin divides, that temporarily expanded the Colorado River’s delta hundreds of kilometers into the Gulf of California (Howard et al., 2015; cf. East et al., 2018). Alternative eustatic or tectonic explanations cannot adequately explain the Bullhead aggradation and its subsequent degradation. Pliocene sea-level variations were much smaller than the 250 m difference between the basal strath and uppermost tread of the Bullhead Alluvium (Miller et al., 2005). Further, any proposal that regional tectonic subsidence explains the Bullhead aggradation would also require a later elevation reversal along 500 km of the river to bring the grade subequal to the original grade from Lake Mead to Yuma (Arizona), an unlikely ad hoc process of regional down-up elevator tectonics unsupported by any other evidence. Profiles of the strath and tread of the Bullhead Alluvium therefore are considered to record fluvially controlled positions of the river grade that can be used to explore vertical deformation (Howard et al., 2015).

The elevation of the highest Bullhead outcrops, used to estimate the upper surface of the Bullhead aggradation package (we also refer to this as the “tread”), suggests paleoprofiles that have been offset and deformed by down-to-the-west fault motion on the Fortification fault system1 in western Lake Mead (Howard et al., 2015) and deformed in the hanging wall of the Wheeler fault system in eastern Lake Mead (i.e., Gregg Basin), which also offsets and deforms the ca. 6 Ma top of the Hualapai Limestone (Howard and Bohannon, 2001; Seixas et al., 2015) (Fig. 2). Howard et al. (2015) noted possible Bullhead outcrops in the footwall of the Wheeler fault at elevations suggestive of lower slip rates than implied by the offset of the Hualapai Limestone, which we reinvestigate.

Multiple post-Bullhead deposits (e.g., House, 2016; Crow et al., 2018a; House et al., 2018) have been identified in the lower Colorado River corridor (Fig. 3). Some, like the Palo Verde alluvium and Riverside alluvium, are incompletely mapped and studied throughout the corridor and will not be considered further. Conversely, one of the best studied is the Chemehuevi Formation, a ∼100-m-thick sequence of dominantly sand and mud in the lower Colorado River corridor (Malmon et al., 2011) and a correlative gravel-rich fill terrace in Grand Canyon (Anders et al., 2005; Pederson et al., 2006, 2013). The Chemehuevi Formation is ca. 70 ka in age based on tephrochronologic results from an ash interbedded in the middle of the unit (Malmon et al., 2011). The longitudinal profile of the highest Chemehuevi deposits from the Yuma area to Grand Canyon is somewhat steeper than that of the modern Colorado River valley but shows no clear evidence of uplift or fault displacement (Malmon et al., 2011). The youngest aggradational package, called the Blythe Alluvium (Block et al., 2019), includes all Holocene Colorado River deposits that predate the Hoover Dam (∼1935) and is ∼35 m thick (Metzger et al., 1973; Howard et al., 2011; Block et al., 2019).

Lower Colorado River Plio-Pleistocene Tectonism

As the Colorado River leaves Grand Canyon and the Colorado Plateau, it enters the extensively faulted Lake Mead region at the eastern limit of the Basin and Range province (Fig. 2). This area experienced significant extension (Wernicke and Axen, 1988) and synchronous strike-slip faulting (Anderson, 1973; Anderson and Beard, 2010) primarily from 16.5 to 8 Ma (Faulds et al., 2001). Modern geodetic studies suggest ongoing diffuse extensional deformation in this area (Kreemer et al., 2010), and significant Pliocene and younger faulting has been documented. In western Grand Canyon, the Quaternary Toroweap and Hurricane faults demonstrate that extension is also occurring within the western margin of the Colorado Plateau (Karlstrom et al., 2007).

We group the west-dipping Wheeler and Lost Basin Range normal faults together into the Wheeler fault system, which extends south from the Grand Wash fault, crosses eastern Lake Mead at the now-submerged mouth of Grand Wash, and continues southward subparallel to Hualapai Wash (Fig. 2). The main Wheeler fault has offset Paleozoic strata >1.5 km (Longwell, 1936). Folding and offset of the 12–6 Ma Hualapai Limestone indicates ∼300–500 m of post-Miocene throw (Brady et al., 2000). Reverse-drag folds in the hanging wall of the Wheeler fault deform Hualapai Limestone, late Miocene basalts, and river gravels (e.g., Lucchitta, 1979; Howard and Bohannon, 2001; Karlstrom et al., 2007; Seixas et al., 2015). Beyond local effects of hanging-wall rollover and footwall upwarp, a regional elevation change across the fault since ca. 6 Ma as indicated by the uppermost Hualapai Limestone is estimated as 150 m (Howard et al., 2000, 2015; cf. Seixas et al., 2015). Quaternary slip rates on the Wheeler fault are not well constrained, but ∼5- to ∼15-m-high scarps have been documented in deposits assumed to be Middle Pleistocene in age, and morphologic analyses suggest that the youngest activity may be Late Pleistocene (Pearthree et al., 1983). Normal faults in the area with even younger slip include the Grand Wash fault north of the river near its intersection with Wheeler fault, and the Overton Arm faults (Nevada; U.S. Geological Survey and Arizona Geological Survey, 2010).

In the western Lake Mead area, early Colorado River gravels on the flanks of the northern Black Mountains dip as much as 15° away from the core of the range on the east side and as much as 60° on the west side (Longwell, 1936; Beard et al., 2007). These outcrops, in the hanging wall of range-bounding normal faults, have likely been deformed by normal drag along the Fortification, Kingman Road, and Indian Canyon faults on the west, which we will call the Fortification fault system, and the Detrital and Boulder Wash faults on the east, which we will call the Detrital fault system (Longwell, 1963; Beard et al., 2013). These range-bounding faults terminate northward against the kinematically related Hamblin Bay segment of the left-lateral Lake Mead fault system (Anderson, 1973). South of the river, in an area now largely covered by Lake Mead, two fault splays of the Callville fault were described by Longwell (1936). One was described as a steeply southeast-dipping reverse fault that juxtaposes old river gravels (likely Bullhead Alluvium) to the northwest against fanglomerate to the southeast. Subsequent investigations indicate that this fault, now called the Mead Slope fault, is dominantly a left-lateral strike-slip fault (Anderson and O’Connell, 1993). The other splay, which we refer to as the Fortification fault, is a normal fault extending southward to Fortification Hill where it is buried by the 6–5 Ma Fortification basalt sequence (Feuerbach et al., 1991). Slip may also be accommodated by the Kingman Road and Indian Canyon faults (Longwell, 1963; Beard et al., 2013). On the east side of the Black Mountains, the east-dipping Detrital fault shows local evidence locally for Quaternary slip (Menges and Pearthree, 1983; cf. Anderson and O’Connell, 1993), and the en echelon Boulder Wash fault (Longwell, 1963) to the north may cut 5.7 Ma basalt flows (Felger et al., 2011).

In the lower Colorado River corridor south of Lake Mead, a relatively small number of structures with post–6 Ma activity have been identified, and the extent to which Bouse and Bullhead deposits are deformed by them is debated. Farther west, the Eastern California shear zone (ECSZ) absorbs Pacific–North America relative plate motion across several northwest-striking, dominantly right-lateral fault systems, which are superimposed on the older transtensional Basin and Range province. Opening of the Gulf of California and development of the southern San Andreas fault system mostly after 6 Ma decreased the amount of relative plate motion absorbed across the ECSZ faults from ∼50% to 25% (Oskin and Stock, 2003). Despite this decrease in activity, many faults throughout the zone remain active. One of the easternmost fault systems in the ECSZ is the Stateline fault system (Guest et al., 2007; Mahan et al., 2009). This right-lateral fault system has offset 13 Ma Miocene volcanic deposits by ∼30 km (Guest et al., 2007) in the vicinity of Ivanpah Valley. To the southeast, past Nipton, California (Fig. 1), little evidence for the structure has been identified, and Mahan et al. (2009) suggested that displacement is transferred into northwest-oriented extension along dip-slip faults. In the lower Colorado River corridor, disconnected young faults subparallel to the Stateline fault system have been identified: east of Needles, California (Pearthree et al., 2009); northwest of Parker, Arizona (Dickey et al., 1980); and potentially in the Buckskin-Rawhide Mountains (Arizona) (Singleton, 2015) (Fig. 1). Post–6 Ma deformation has also been noted farther south in the Parker and Palo Verde Valleys as broad downwarping (Metzger et al., 1973; Howard et al., 2015) and along discrete structures such as the Blythe graben (Schell et al., 1981) and Big fault (Gootee et al., 2016a). Father south along the river, the next major young structure is the Algodones fault zone, near Yuma, Arizona. It is part of the larger San Andreas system, which has experienced ∼200 km of right-lateral offset in the past 5 Ma (Darin and Dorsey, 2013). The delta plain downstream from the Algodones fault zone was lowered at average rates of ∼1000 m/Ma during the past 6 Ma before being locally uplifted in the last ∼1 Ma (Dorsey et al., 2011). Localized uplift upstream (north) of the fault zone in the Chocolate Mountains is also possible due to transpression (Ricketts et al., 2011; Beard et al., 2016). Research on the timing and kinematics of these and other faults in the vicinity of the lower Colorado River is ongoing (Bennett et al., 2016; Thacker et al., 2017) to identify which structures may deform Colorado River deposits in the area.


Incision rates reported by this study are calculated to quantify the amount of river incision into pre–Bullhead Alluvium strata, as opposed to incision back through younger aggradation packages. We refer to this as bedrock incision, but in some cases, incision may be through non-crystalline rocks, such as the Bouse Formation or other basin deposits. The height we use to calculate incision rates is the elevation difference between the paleostrath associated with the dated deposit and the modern strath near the valley axis, below the river’s surface. The depth to the modern strath is estimated based on the closest sites where drill data are available, mostly at sites considered for dam construction. In Grand Canyon, previous studies have also used depth-to-bedrock estimates based on bathymetry, assuming that the deepest pools are scoured to near bedrock (Pederson et al., 2002a; Karlstrom et al., 2007; Crow et al., 2014). Heights above the pre-dam river were determined by comparing modern elevation values from the sampling locations to the elevation of the river taken from pre-dam Colorado River surveys (Birdseye, 1924; U.S. Geological Survey, 1927). Modern elevation values were determined relative to river or lake level using a Tru Pulse laser rangefinder (accuracy of ±1 m) or measured in the field using a handheld GPS and cross-checked with an elevation value extracted from a 10-m-resolution digital elevation model (DEM). Possible uncertainties associated with these measurements are minimal but are explained in the results section for each site along with the uncertainties resulting from the geologic relationship between the dated material and the deposit or surface of interest. The new incision rate reported in this study is based on new 40Ar/39Ar dating of groundmass concentrate from a basalt flow that overlies tributary gravel, which was conducted at the New Mexico Geochronology Research Laboratory (Socorro, New Mexico) (see Crow et al. [2015] and Table S1 in the Supplemental Materials2 for details).

In addition to the incision constraints, elevations of surficial and subsurface Colorado River deposits were projected onto a longitudinal river valley profile to evaluate evidence for regional deformation (Fig. 4). Colorado River miles3 (1 mile = 0.62 km), measured downstream from Lees Ferry (Stevens, 1983), were extended to the Gulf of California through the middle of the pre-dam floodplain to remove channel sinuosity in unrestricted reaches (Table S2 [footnote 2]). Geologic features (e.g., outcrops and subsurface contacts) were then projected onto the profile using an automated routine4, which determines the closest tenth of a valley mile within a series of zones that correspond with geologically meaningful segments of the river. Boundaries between zones were defined by faults and bedrock divides that separated basins before Colorado River integration (Fig. 1) and commonly correspond to changes in the orientation of the modern river course. Use of these zones maintains the position of features relative to faults and basin divides. Elevation values of mapped ancestral Colorado River deposits (Crow et al., 2018a) were extracted from a 10 m DEM using zonal statistics such that the maximum, minimum, and mean elevation values of a mapped area were calculated and depicted on the profile as crosses, where the tops and bottoms of the crosses refer to the extreme values and the horizontal bar is the mean value (Fig. 4). This depiction gives a good indication of the range of elevation values within a mapped polygon, but erroneous values are possible in areas of high relief (i.e., cliffs) and when the scale of geologic mapping differs from that of the elevation data. We also consider the elevation range over which these units were interpreted in the subsurface from published and unpublished lithologic well logs (see Table S3 [footnote 2] for a complete list of wells used and references). Point data associated with outcrop-scale field observations (e.g., the height of a key contact in the field) are also shown on the profile as colored circles. The elevation values associated with most of these observations were determined in the field using handheld GPS units or extracted from 10 m DEMs. Some key outrcrops were also resurveyed using a high-precision GPS with a vertical uncertainty of <1 m. Together these data place constraints on both the position of the base and top of associated aggradational packages of the ancestral Colorado River, which will be focused on in detail.


New 40Ar/39Ar dating, field observations, and three-dimensional analysis of geologic mapping permit estimation a long-term bedrock incision rate for the Lake Mead area and quantification of vertical deformation associated with faulting.

Incision Constraints

A new long-term bedrock incision rate for the eastern Lake Mead area is reported based on field observations and dating of a basalt flow that overlies Grand Wash gravels at a location ∼1 km north of the Colorado River and 2 km west of Grand Wash, in the hanging wall of the Wheeler fault. The outcrop is located within a hanging-wall rollover that folds the flow (Howard and Bohannon, 2001). New 40Ar/39Ar dating of the basalt flow indicates an age of 4.73 ± 0.17 Ma (Fig. 5). The elevation of the outcrop is estimated to be between 503 and 498 m asl based on the recorded GPS location of the sample and a 10 m DEM. The Birdseye (1924) survey indicates a pre-dam Colorado River elevation in this area of ∼261 m asl. At the Boulder Canyon dam site, ∼72 valley km (∼45 miles) downstream, the maximum depth to bedrock in the center of the channel is between 24 and 48 m (LaRue, 1925). Near proposed dam sites in western Grand Canyon, ∼77 valley km (∼48 miles) upstream, the maximum depth to bedrock is between 14 and 27 m (Hanks and Webb, 2006). Combining these uncertainties from upstream and downstream, we estimate a depth to bedrock of 14–48 m at river mile 286.8, near our Grand Wash basalt sample location. Subtracting the modern channel bedrock strath elevation estimate (213–247 m asl) from the elevation of the strath below the dated Grand Wash basalt (498–503 m asl) suggests 290–251 m of river incision since emplacement of the basalt flow. Integrating our new 4.7 Ma basalt age with this incision estimate yields an incision rate of 51–64 m/Ma at the Grand Wash basalt site (Table 1). This rate was calculated to the Colorado River instead of Grand Wash because the site is closer to the mainstem than the tributary, and a large tributary like Grand Wash would be graded to the Colorado River. Although all indications suggest that the resulting incision rate should accurately reflect the averaged rate at which the Colorado River has been downcutting there, we categorize it as a maximum rate because no Colorado River gravels are present beneath the basalt, only gravels from Grand Wash.

Upstream from the Wheeler fault in its footwall, mapped ancestral tributary and Colorado River gravels (Lucchitta, 1966, 1972; Wallace et al., 2005) were noted as possible Bullhead Alluvium (Howard et al., 2015), but we now consider that correlation unlikely. The highest “older cemented river gravels” mapped by Lucchitta (1966, his unit Tg3) in the Pearce Ferry area were remapped by Wallace et al. (2005) as paleo–Grapevine Wash deposits and lack the distinctive quartz-rich thermal-infrared signal on MASTER (MODIS/ASTER airborne simulator) remote-sensing imagery (Hook et al., 2005) commonly indicative of Colorado River deposits. Farther upstream in Grand Canyon, no gravels of Pliocene age have been documented, however ca. 4 and 2 Ma cave mammillaries have been used to estimate the position of the water table and, by association, the river level at those times (Polyak et al., 2008) (Figs. 2 and 6). The cave data suggest much higher incision rates of ∼90 m/Ma, which is in good agreement with Quaternary incision rates in the same part of western Grand Canyon (Crow et al., 2014), east of the Wheeler and Grand Wash faults (Fig. 2).

About 13 valley km downstream from the Grand Wash basalt incision point, another basalt flow dated at 4.49 ± 0.46 Ma (Faulds et al., 2016) overlies Colorado River gravel near Sandy Point (Fig. 2). The flow dips eastward <5° as part of the hanging-wall rollover fold toward the Wheeler fault (Lucchitta, 1966; Wallace et al., 2005). We visited this site in July of 2016, when Lake Mead was at its lowest level (327 m asl) since its filling in the early 1940s. In 2016, cemented Colorado River gravels were exposed at lake level. An interval of locally derived sediment was present between the lake-level gravels and those previously documented below the basalt. It appeared that the lake-level gravels were stratigraphically below the gravel package that underlies the 4.5 Ma basalt as opposed to being part of a younger inset terrace. The observations are consistent with deep river incision and then deposition of at least 40 m of Colorado River–derived and locally derived sediment prior to the eruption of the Sandy Point basalt. The maximum height above the pre-dam river level of the strath at the base of the river gravels is ∼76 m, given by the elevation of lake level at the time of the observations and the pre-dam river elevation of 251 m (Birdseye, 1924). Using this height and the same estimate of the pre-dam depth to bedrock as the previous site (14–48 m), and the 2σ uncertainty in the basalt age suggests an incision rate between 18 and 31 m/Ma. Because the base of the gravels was not exposed, this incision rate should be considered a maximum bedrock incision rate, however it is also unknown how long it took for the 40 m of gravel under the basalt to be deposited. Incision there would be further lowered if calculated for a local block downthrown 30 m on a NW-striking fault that was observed before Lake Mead filled (Longwell, 1936).

Downstream from Sandy Point to Hoover Dam, Lake Mead obscures other deposits that might be suitable for calculating incision rates (Beard et al., 2007). Pre–Lake Mead observations by Longwell (1936, p. 1442) indicate that “no remnants of…older cemented [river] gravels were found along the river in the twenty-mile stretch across the Virgin-Detrital trough. Probably, they were deposited far above the present river level…”. During our July 2016 field work, we observed cemented Colorado River gravel of unknown age dipping 12° toward the northwest at lake level (327 m asl and ∼105 m above pre-dam river level) in Virgin Basin (at 36.103045°N, 114.466729°W). On the flanks of the northern Black Mountains, Pliocene Colorado River gravels dip 15°–60° away from the core of the range (Longwell, 1936; Beard et al., 2007). On the east flank of the Black Mountains, in the footwall of the Detrital fault system (west side), we (Beard and Felger) found a lag of rounded gravel at elevations as high as 756 m asl using hand-written annotations by R.L. Laney on a geologic map accompanying U.S. Geological Survey Open File Report 79-689 (Laney, 1979) (at 35.996086°N, 114.553770°W).

On the west side of Fortification fault, which bounds the west side of the uplift, Longwell (1936) noted at least 30 m (100 ft) of gravel deposits in bluffs above the river and to an unknown depth below the river. These deposits were also noted to contain silicified logs diagnostic of the Bullhead Alluvium. Together, the observations suggest that in Boulder Basin, Bullhead Alluvium is present below river level within the hanging wall of the Fortification fault, while in Virgin Basin on the other side of the northern Black Mountains, the same deposits are present at >100 m above river level (Fig. 6). In contrast, Longwell (1936) noted no offset of the ca. 70 ka Chemehuevi Formation.

Profile Analysis

In addition to determining differential incision along the lower Colorado River, we also characterized the longitudinal distribution of its ancestral deposits and associated landforms to help assess the extent and magnitude of post-Miocene vertical deformation. Howard et al. (2015) did this for Bullhead Alluvium outcrops and found offset across the Black Mountains. In the lower Colorado River corridor, they found little evidence for regional titling or major offsets of the Bullhead maximum aggradational surface. We have redone this analysis for additional deposits including the Chemehuevi Formation and the Blythe Alluvium and with the benefit of an exhaustive compilation of ancestral Colorado River deposits based on new mapping and compilation of published mapping (Crow et al., 2018a) and additional subsurface constraints (Table S3 [footnote 2]). We wanted to see if this larger data set would support the same conclusions and if it could be used to elucidate more subtle vertical deformation. In this section, we investigate the result of this larger analysis, focusing on the patterns in the highest and lowest outcrops from each unit and any possible “steps” in the profile.

The highest mapped occurrences of a given unit can be used to estimate the elevation of maximum aggradation in a particular area. For example, the trendline through the highest mapped Chemehuevi outcrops is remarkably linear, with an average valley gradient of 0.68 m/km (between points J and K on Fig. 4). In the area of the Newberry Mountains near Davis Dam (Lake Mohave), mapped outcrops project to ∼20 m above the regional trendline. Downstream of the Fortification fault system, the top of the Bullhead aggradational package has an average slope of 0.69 m/km (cf. Howard et al., 2015) (between points I and A on Fig. 4). An apparent downward “step” in the highest extant definitive Bullhead deposits is present in southern Mohave Valley (Fig. 4). Pliocene Colorado River gravels (likely Bullhead Alluvium) also reach maximum elevations ∼200 m higher in Virgin Basin than in Boulder Basin (Howard et al., 2015). Although the slopes of the top of the Chemehuevi and Bullhead aggradational packages are remarkably similar in the reach between Hoover Dam and the Algodones fault, it should be noted that the gradient of the pre-dam Colorado River valley in the same area is nearly 50% lower at 0.36 m/km (Fig. 4).

Downstream from Hoover Dam and Lake Mead, deposits mapped as Bullhead Alluvium (Howard et al., 2015) are commonly exposed down to river level (Olmsted et al., 1973; Faulds et al., 2004; Pearthree and House, 2005; Stone, 2006; Malmon et al., 2009; Spencer et al., 2015; Gootee et al., 2016b), and well records indicate similar gravel and sand in the subsurface in many locations (e.g., Metzger et al., 1973; Metzger and Loeltz, 1973; Olmsted et al., 1973). The available constraints on the position of the base of the Bullhead Alluvium are shown on Figure 4. The purple boxes highlight areas where this surface is best constrained by both subaerial Bullhead deposits (orange crosses) near modern river level and subsurface constraints on the base of Colorado River deposits that post-date the Bouse Formation. Lithologic descriptions of the interval interpreted as post-Bouse ancestral Colorado River deposits vary greatly. In some cases, descriptions of rounding, sorting, and lithology are available for gravel intervals, which makes discrimination between Colorado River and locally derived piedmont deposits possible. In other cases, the descriptions are vague and only indicate the presence of sand and gravel. To avoid intervals of dominantly locally derived material, we focus on wells drilled through the modern floodplain, where Colorado River deposits are more likely. We also color code well intervals on Figure 4 taking into account these uncertainties. In some cases, the post–Bouse Formation ancestral Colorado River deposits can be further subdivided. For example, the Holocene Blythe Alluvium is commonly identifiable in the logs as a coarsening-downward sequence that ends in basal gravel (Metzger et al., 1973; Block et al., 2019, now exclude the basal gravels from the Blythe Alluvium). In a few locations, wood from within those deposits has been 14C dated (Metzger et al., 1973; Howard et al., 2011) to between 5 and 9 ka. The lowest of the dated samples, denoting the minimum depth to the base of the Holocene Blythe Alluvium, are shown by downward-facing blue arrowheads on Figure 4. One wood sample yielded no 14C indicating a pre-Holocene age; that is denoted on the profile by a black arrowhead. Historical records suggest ∼10–15 m of scour during floods based on a “sawn timber” found in Colorado River deposits during excavation for the construction of Hoover Dam (Longwell, 1936, p. 1455) and historical observation in the Yuma area (Follett, 1913; Kniffen, 1932); the deepest level to which historical scour is known to have reached is shown by red arrowheads on Figure 4.

Connecting the base of the Blythe Alluvium regionally defines a surface (white line on Fig. 4) that is remarkably similar in slope to that of the pre-dam river (blue line on Fig. 4). This surface is ∼35 m below river level. Where the surface rests on Bouse Formation (i.e., in Parker and southernmost Palo Verde Valleys) and subaerial Bullhead deposits are present to river level, the amount of incision deeper than the base of the Bullhead Alluvium is constrained to less than the subsurface thickness of the Blythe Alluvium, ∼35 m. If there has been no more than 35 m of incision past the base of the Bullhead Alluvium in the last 4.5 Ma, the maximum incision rate is ∼8 m/Ma.


In this section, we will start by exploring the constraints on the position of the base of the ca. 4.5 Ma Bullhead alluvium and its relation to Pliocene sea level. Next, potential offsets of the highest Bullhead and Chemehuevi deposits are considered, with implications for which faults have been active since the deposition of those units. Finally, these observations and differential incision along the lower Colorado River corridor and through western Grand Canyon are considered to assess the relative importance and magnitude of uplift and/or subsidence during the evolution of the lower Colorado River.

Lower Colorado River Corridor Profiles

Projection of ancestral Colorado River sediment deposited during aggradational events onto longitudinal profiles delineates the base and the top of those aggradational packages. The base of each sediment package preserves an important record of past river position and, when compared to the modern river, can be used to evaluate and quantify river incision since deposition. Offsets and warping of the surfaces bounding these aggradational packages can also be used to assess post-depositional fold and fault deformation. Although either the base or top could be used in this regard, the positions of the highest deposits are much better constrained by subaerial exposures; downstream from Lake Mead in the valley axis, the bases of these ancestral Colorado River deposits are almost always buried and thus are constrained only with sparse subsurface data.

Basal Profiles

In four areas, the base of Bullhead deposits (dashed gray line on Fig. 4) is below the base of Blythe Alluvium (white line on Fig. 4): (1) Yuma Valley, (2) Palo Verde Valley, (3) near Parker Dam (east of the Whipple Mountains), and (4) southern Mohave Valley. We know from geophysical studies (Richard et al., 2007) that the deepest sedimentary basins are located in Yuma Valley, Palo Verde Valley, and the southern Mohave Valley. Localized fault-related syn- to post-Bullhead subsidence (including to depths below sea level in the southern basins) is likely required to explain the subsurface character of the Bullhead Alluvium. Ongoing work is focusing on characterizing the geometry and likely structural controls on those young basins and modeling isostatic effects of loading and erosion (Thacker et al., 2017; Cassidy et al., 2017; Karlstrom et al., 2018). In contrast, the deep Bullhead deposits near Parker Dam coincide with a rugged bedrock canyon, not a basin. This is also the site of a ∼5-m-high and 75-km-wide convexity in the pre-dam Colorado River profile. The convexity is likely related to increased sediment load from the Bill Williams River, a major tributary that joins the Colorado River just upstream of the convexity’s center. Although the increased sediment load from this tributary may contribute to the thickness of Colorado River sediment there, it does not easily explain why the sand and gravel deposits are ∼40 m deeper there than in areas upstream and downstream. One possibility is that an unrecognized interval of locally derived pre–Colorado River material is present in the wells defining this area of deep Bullhead Alluvium near the Parker dam site (no Bouse Formation has been identified there). Available lithologic logs from the construction of Parker Dam (Berkey, 1935) do not indicate the degree of rounding or sorting, which would be useful for discriminating between Colorado River and locally derived sediment, leaving uncertainty as to the explanation for the deep deposits there. In conclusion, with the exception of these four areas, the basal profile of the Bullhead Alluvium is constrained to within ∼35 m of historic river level, indicating little if any bedrock incision (<8 m/Ma).

The basal profile of the Bullhead Alluvium also projects toward Pliocene sea level in the northern Gulf of California (Fig. 4). Eustatic sea level has been extensively studied for the mid-Pliocene warm period (3.3–2.9 Ma), with most studies placing peak sea level ∼10–30 m above modern levels (Dwyer and Chandler, 2009; Naish and Wilson, 2009; Raymo et al., 2009; Miller et al., 2012; Rovere et al., 2014) prior to Pleistocene glaciation. Early Pliocene sea level is less well studied, with most estimates based on the δ18O values of benthic foraminifera (Lisiecki and Raymo, 2005), which vary primarily due to changes in global ice volume and deep ocean temperature. This record has been variably interpreted to suggest that the 4.5–3.5 Ma sea level was between −56 and 22 m (Miller et al., 2005), −14 and 39 m (Raymo et al., 2009), and −48 and 0 m (Miller et al., 2011) relative to modern levels (Fig. S1 [footnote 2]), with many short-term fluctuations.

Howard et al. (2015) analyzed and summarized subsurface and uplifted Colorado River deposits downstream from the Algodones fault to reconstruct the development of the Bullhead delta in the proto–Gulf of California. They estimated the shoreline position of the prograding Bullhead delta by restoring along the Algodones fault the original location of the ca. 4.25 Ma Fish Creek–Vallecito Basin marine-delta deposits to 190 km to the southeast. This restoration places the shoreline of the prograding Bullhead delta to be near the location of Yuma at ca. 4.5 Ma when the Bullhead aggradational event initiated, and near the location of the mouth of the modern Colorado River delta at ca. 4.25 Ma. The difference between the modern position of the 4.5 Ma basal profile in the vicinity of the Yuma and the full range in possible Pliocene sea levels (−56 to 39 m above modern sea level) (Miller et al., 2005; Raymo et al., 2009) (Fig. S1 [footnote 2]) is between 60 and −40 m. Perhaps the simplest interpretation is that the ancestral river profile remains graded to paleo–sea level. Alternatively, the constraints allow interpretations that the lower Colorado River corridor (upstream from the Algodones fault) was uplifted by as much as ∼60 m or subsided by as much as ∼40 m since 4.5 Ma. If the delta’s paleo-shoreline was farther downstream near the modern delta, the amount of possible uplift would decrease and the amount of subsidence would increase, but this seems less likely based on the reconstructions of Howard et al. (2015). Howard et al. (2015) also argued that the slope of the basal Bullhead profile is similar to that of the pre-dam profile, which would suggest little if any tilting of the Bullhead Alluvium regionally due to differential uplift or subsidence. Our analysis with much more data confirms this; although localized subsidence is noted in a number of locations, on a regional basis limited uplift or subsidence (<60 m) has occurred in the lower Colorado River corridor downstream of Lake Mead in the last 4.5 Ma.

The basal Bullhead profile has interesting relationships with valley-fill deposits that immediately predate its development. In all basins where Bouse deposits are present, the basal Bullhead profile is far lower than the maximum level of Bouse deposition (Pearthree and House, 2014). Farther upstream in the Lake Mead area, the basal Bullhead profile is several hundred meters lower than the highest nearby outcrops of the Hualapai Limestone. Indeed, in Cottonwood Valley and northernmost Mohave Valley, the basal Bullhead profile is lower than the base of Bouse deposits. These are areas where the basin axes were relatively high prior to the arrival of Colorado River water and sediment, and the river cut below them when the basal profile was established. In central and southern Mohave Valley, and in nearly all of the Blythe Basin, the basal profile is much lower than the maximum levels of Bouse deposition but is cut on Bouse siliciclastic deposits. Basin axes in these areas were likely lower prior to filling by Colorado River water and Bouse sediment, but they are also areas where there is evidence for syn- and post-Bullhead subsidence. The early river must have incised deeply into pre–river basin axis deposits and intervening bedrock areas along its course from the mouth of Grand Canyon to the southern Blythe Basin to establish a profile graded to early Pliocene sea level.

Tread Profiles

So far, we have focused on the base of the Bullhead Alluvium aggradational package, but the highest preserved outcrops of that unit (and of other units) are also highly informative for quantifying fault throw and post-depositional tilting. Unlike the base, which is present only in the subsurface in the axis of the valley downstream from Lake Mead, the highest outcrops of these aggradational packages are constrained by geologic mapping and field observations. Planar tread surfaces can be easily identified at the top of relatively younger packages of river deposits, such as the ca. 70 ka Chemehuevi Formation. In contrast, tread surfaces at the top of older packages, such as the ca. 4.5–3.5 Ma Bullhead Alluvium, are not preserved. However, the highest preserved outcrops approximate the maximum level of aggradation, the tread. In profile, trendlines through those highest outcrops (bold black lines on Fig. 4) give reasonable slope estimates below Lake Mead of 0.69 m/km for the Bullhead Alluvium tread and 0.68 m/km for the Chemehuevi Formation tread (calculated between points I and A and between points J and K, respectively, on Fig. 4). Differences of ∼17% in valley slope compared to those calculated by Malmon et al. (2011) and Howard et al. (2015) are likely due to inclusion of additional, newly identified outcrops and a different method of projecting outcrop positions to the profile. The similarity in the slope of Bullhead and Chemehuevi treads suggests that at a regional scale, little tilting has occurred since ca. 4 Ma (Howard et al., 2015). However, on closer inspection, there are possible “steps” or convexities (dashed lines on Fig. 4) that were not investigated by Howard et al. (2015). We investigate the most obvious of these starting with the Chemehuevi Formation and then the Bullhead Alluvium, focusing on the potential significance of each.

In the area of the Newberry Mountains near Davis Dam (Lake Mohave), mapped Chemehuevi Formation outcrops project to ∼20 m above the regional trendline (Fig. 4). This divergence from the trendline may reflect conflation of older and lithologically similar units with the Chemehuevi Formation, inclusion of overlying younger deposits, downstream erosion of the terrace tread, or a localized profile convexity, perhaps associated with the bedrock constriction in the vicinity of Davis Dam (the Pyramid divide). Alternatively, if these higher-elevation outcrops are truly Chemehuevi Formation, their anomalously high position could be due to local fault deformation. A lack of mapped fault scarps or obvious offset of older terraces and alluvial fans in the same area of the Newberry Mountains suggests a non-tectonic explanation.

Apparent downward “steps” in the highest definitive Bullhead deposits could suggest post-depositional deformation and will be discussed in turn. In southern Mohave Valley (Fig. 4), an apparent step occurs directly downstream from the Pleistocene Needles fold and graben system (Pearthree et al., 2009), suggesting a possible tectonic control (see Thacker et al., 2017). Additional evidence consistent with a tectonic control includes thicker subsurface Bouse deposits in the basin axis (Metzger and Loeltz, 1973) and greater estimated depth to bedrock (Richard et al., 2007) downstream from the Needles fold and graben system. However, sparse high Bouse Formation outcrops preserved on the valley margins are at similar elevations north and south of the Needles fold and graben system, suggesting little if any regional post-Bouse vertical displacement. Additionally, ambiguous fine-grained Colorado River deposits that may be part of either the Bouse Formation or the Bullhead Alluvium, informally called the “Bousehead” deposits (brown crosses on Fig. 4), could fill in the apparent “step” if they are Bullhead Alluvium (Fig. 4). The “step” may also reflect incomplete preservation of Bullhead Alluvium in the southern Mohave and Chemehuevi Valleys.

Obvious offset of the highest Bullhead outcrops occurs in the Lake Mead area where the river crosses the northern tip of the Black Mountains (Figs. 4 and 6; Howard et al., 2015). The highest Bullhead Alluvium outcrops in the Boulder Basin area (westernmost Lake Mead) are ∼310 m lower than newly described well-rounded gravel of likely Colorado River derivation on the eastern flank of the Black Mountains (between points C and B on Fig. 4). In the Detrital Valley–Virgin Basin area (Fig. 2) east of that range, highest mapped Bullhead Alluvium and lithologically similar potentially pre–Bullhead Alluvium outcrops are ∼270 m higher than Boulder Basin outcrops (compare points C and D on Fig. 4). We follow Howard et al. (2015) in suggesting that this net offset may have occurred along a series of faults around the Black Mountains, including the Fortification and Detrital fault systems. Horst-style uplift of the Black Mountains explains the newly described well-rounded gravels found at high elevation on the eastern side of the range. Projection of the probable Bullhead tread to the faults bounding the Black Mountains uplift suggests between 145 and 230 m of net post–4 Ma (assuming peak aggradation at that time) offset across the Black Mountains (between points A and E and between points N and D Fig. 4; Table 2). This agrees well with the regional vertical offset estimate of >200 m suggested by Howard et al. (2015). Differences are due mainly to our more conservative approach in estimating the maximum and minimum offset.

In easternmost Lake Mead, the highest Bullhead Alluvium outcrops in Gregg Basin drop eastward in elevation toward the Wheeler fault system. This fault system also offsets the 6 Ma upper part of the Hualapai Limestone by >300 m of throw. About half of that throw is accommodated by hanging-wall downwarping (i.e., “reverse drag”) and footwall upwarping (Howard et al., 2000; Howard and Bohannon, 2001; Seixas et al., 2015), but outside the wavelength of that folding, there is ∼130–165 m of regional vertical offset of the Hualapai (between points 3 and 4 and between points 5 and 4 on Fig. 6; Table 2). This is in good agreement with the estimates of Howard et al. (2015) (cf. Seixas et al., 2015). The amount of offset of the Bullhead Alluvium across the Wheeler fault is unknown, as old cemented river gravels in the footwall of the Wheeler fault are likely locally derived (Wallace et al., 2005) as opposed to Bullhead Alluvium (Howard et al., 2015). These observations suggest significant deformation and offset of the ca. 4 Ma Bullhead Alluvium tread across the Black Mountains and the 6 Ma upper Hualapai Limestone across the Wheeler fault system in the Lake Mead region. Downriver, in the lower Colorado River corridor, any deformation of the Bullhead or Chemehuevi treads is uncertain, and incomplete preservation of the formations and other nontectonic factors plausibly can explain apparent “steps” or convexities.

Long-Term Incision Rates

Sufficient information is available to calculate long-term post–Bouse Formation bedrock incision rates throughout the study area over the past 5–4 Ma. These incision rates increase from near 0 m/Ma (<8 m/Ma) in the lower Colorado River corridor to ∼90 m/Ma in western Grand Canyon, with important caveats and implications that we will explore starting downstream and moving upstream (Fig. 7).

In the axis of the river valley, the base of the Bullhead Alluvium downstream from Lake Mead is present only in the subsurface. By combining constraints from mapping and subsurface investigation, we can constrain the maximum amount of post–Bullhead Alluvium incision to between 0 and ∼35 m (the thickness of the Blythe Alluvium), suggesting an incision rate of <8 m/Ma prior to Holocene aggradation.

In the Boulder Basin of Lake Mead, Longwell (1936) observed folded Colorado River gravels, which are likely Bullhead Alluvium due to the presence of petrified wood, at river level. This observation similarly implies long-term bedrock incision rates of near 0 m/Ma. However, in the Virgin Basin, on the eastern side of the Black Mountains, mapped Bullhead Alluvium, Bullhead-like deposits at much higher elevations, and Longwell’s (1936) observation of the absence of similar deposits near river level imply significant bedrock incision. Farther east in Gregg Basin, our calculated maximum incision rates of between 51 and 64 m/Ma based on the 4.7 ± 0.17 Ma Grand Wash basalt flow and between 18 and 31 m/Ma using the 4.49 ± 0.23 Ma Sandy Point basalt are both undoubtedly dampened by fault-related downfolding in the hanging wall of the Wheeler fault. The Sandy Point basalt and conglomerate dip <5° toward the fault as part of that folding. Other workers documented that basalt flows with similar ages along ancestral Grand Wash are downfolded in the hanging wall of the Wheeler fault (Howard and Bohannon, 2001). One of those basalt flows, 17 km north of Lake Mead, was dated at 4.71 ± 0.03 Ma (Howard et al., 2010), indistinguishable from the age we report near the mouth of Grand Wash.

The two possible explanations for the discrepancy between the rates are: (1) variable amounts of fold generation in the hanging wall of the Wheeler fault system, and (2) rapid incision prior to Bullhead aggradation recorded by the potentially older Grand Wash basalt incision constraint. We first investigate the hypothesis that variable amounts of fold generation explain the incision rates. Seixas et al. (2015) reported smaller-amplitude hanging-wall rollover in the Sandy Point area along the Wheeler fault as compared to that along the Lost Basin Range fault farther south. They did not study the Wheeler fault north of Sandy Point, but other workers have documented basalt flows along ancestral Grand Wash that are downfolded in the hanging wall of the Wheeler fault (Howard and Bohannon, 2001). It seems plausible that the lower rates in the Sandy Point area may be due to less hanging-wall rollover formation there as compared to farther north along Grand Wash. In this case, it is possible that the Sandy Point basalt and the Grand Wash basalt are part of the same flow that poured down Grand Wash, as both ages are within error. Alternatively, the Grand Wash basalt may record rapid incision during integration of the Colorado River system as the basal Bullhead profile was being established. The Grand Wash basalt and tributary gravel project in elevation toward a position inset into the Hualapai Limestone, but higher (by >100 m) than Bullhead Alluvium and the interbedded Sandy Point basalt, which are in a similar structural position. Thus, it is also plausible that the Grand Wash basalt and tributary gravel at Grand Wash predate the Bullhead Alluvium. The 4.7 Ma timing of an early stage of pre-Bullhead Colorado River incision in the Lake Mead area is in accord with the finding, based on other dated basalts, that incision in Detrital Valley and Black Canyon started between ca. 4.9 and 4.6 Ma (Felger et al., 2011). This timing is also consistent with new dating of the arrival of river sediments after 5.24 Ma in Cottonwood Valley (Crow et al., 2018b). The early incision likely occurred when interbasin natural dams were cut through as the river evolved toward a steady grade toward its mouth near Yuma (Pearthree and House, 2014; Howard et al., 2015). Higher-precision dating of the Sandy Point basalt and geochemical comparison of it with the Grand Wash basalt could clarify this uncertainty, but currently we favor the interpretation that the faster incision recorded by the Grand Wash basalt flow integrates both pre- and post-Bullhead incision that were both modulated by deformation in the Wheeler fault zone.

In western Grand Canyon, long-term bedrock incision rates with similar magnitudes have been calculated from dating of a 0.6 Ma basalt flow that overlies river gravel (Crow et al., 2014) and dating of 2.2 Ma and 3.9 Ma cave mammillaries (Polyak et al., 2008). The mammillaries record paleo–water table levels, which Polyak et al. (2008) assumed to be flat and also assumed to be tracking the incising river. The 2.2 Ma and 3.9 Ma cave mammillaries are within a few kilometers laterally of the river (Cave B and Dry Canyon Cave). At locations such as these, which are close to the center of the canyon, the effect of a sloping water table on incision rate estimates (Karlstrom et al., 2008) would be minimized. However, it is not known whether the paleo–water table heights were affected by local or regional aggradational fill or record a paleo–river level close to a bedrock canyon bottom. In particular, it is unclear if at 3.9 Ma the river in western Grand Canyon flowed over any Bullhead Alluvium or was near bedrock. The similarity of calculated incision rates over the past 0.6, 2.2, and 3.9 Ma offer support for a steady rate of bedrock incision, and for the idea that the water levels recorded by the mammillaries were within a few tens of meters of bedrock rather than over variably thick Bullhead Alluvium or other aggradational canyon fills. Because aggradation events on the Colorado River have not been identified at ca. 2 Ma, the incision rate of between 70 and 108 m/Ma derived from the dating of 2.2 Ma mammillaries is more certain. Ultimately, a more definitive conclusion about longer-term incision rates could be made if the extent to which the Bullhead Alluvium filled Grand Canyon was known, and independent measures of long-term Grand Canyon incision could be obtained. This may be possible if any Bullhead Alluvium is found in extensive travertine deposits in western Grand Canyon (Billingsley and Wellmeyer, 2003; Billingsley et al., 2006) (Fig. 4), as travertine is datable by U-Pb and U-series techniques.

To summarize the currently available constraints, little to no bedrock incision has occurred in the lower Colorado River corridor downstream of the Black Mountains in the past 4.5 Ma (<8 m/Ma). Over a similar time frame, the Colorado River incised into bedrock in the hanging wall of the Wheeler Fault at rates of <18–64 m/Ma, and almost certainly at a greater rate outside of the influence of the hanging-wall rollover. In western Grand Canyon, incision constraints indicate steady long-term bedrock incision at ∼90 m/Ma since 3.9 Ma (Polyak et al., 2008; Crow et al., 2014).

Spatial changes in bedrock incision rates coincide with faults and fault-related folds of the Fortification, Detrital, and Wheeler fault systems that displace and/or deform late Miocene to Quaternary deposits and their upper geomorphic tread surfaces. We interpret this as tectonic control on incision. Bedrock strength variations or climatic changes do not adequately explain these variations. Similar bedrock lithologies are present in the lower Colorado River corridor and the upper Lake Mead region despite differing incision histories in those areas. Apart from Miocene volcanic rocks, which are largely lacking in Grand Canyon, similar bedrock rock types are also present in Grand Canyon. Perhaps the two largest differences affecting bedrock strength between the Basin and Range province and the Colorado Plateau is the amount of pre-integration faulting and folding and the amounts of basin fill, which are both greater in the Basin and Range. Large amounts of Miocene extension in the Basin and Range extensively deformed the rocks there and produced large basins that filled with local detritus. One might expect greater incision in the structurally more disrupted and hence weaker rocks downstream from Grand Canyon, but the opposite is measured. Similarly, one might expect greater incision in the area with more poorly indurated basin fill, like the lower Colorado River corridor, but the opposite is observed. Climatic fluctuations are unable to explain the spatial variations in incision because precipitation-driven changes to discharge and hence stream power affect all of Grand Canyon and the lower Colorado River corridor similarly, as the vast majority of the discharge comes from the Rocky Mountains. Additionally, long-term bedrock incision in Grand Canyon is steady (Karlstrom et al., 2007, 2008; Crow et al., 2014) despite climatic variations at glacial-interglacial scales (Anders et al., 2005; Pederson et al., 2006, 2013).

Uplift Magnitude

If differences in incision are tectonically controlled, then the amount of uplift or subsidence should be equal to the differential incision. Because little to no post–4.5 Ma bedrock incision has occurred in the lower Colorado River corridor, the magnitude of differential incision or uplift should be equal to the incision magnitude in western Grand Canyon, with the uplift occurring across a 50-km-wide zone in the Lake Mead region. The 2.2 and 3.9 Ma speleothems suggest between 190 and 360 m of incision there, assuming that water table decline is due entirely to bedrock incision.

Because incision rate variations seem to coincide with fault zones, one would expect that surfaces of similar age to the period over which the incision rates have been calculated should be offset by amounts similar to the inferred uplift amount. The estimated net regional offset across the Fortification and Detrital fault systems (i.e., across the Black Mountains; Fig. 2) of the ca. 4.5 Ma Bullhead Alluvium basal profile and ca. 4 Ma maximum aggradation is between 145 and 230 m (between points A and E and between points N and D on Fig. 4; Table 2). Similarly, net regional offset by the Wheeler fault of the ca. 6 Ma top of Hualapai Limestone, beyond hanging-wall and footwall flexures, is ∼130–165 m (between points 3 and 4 and between points 5 and 4 on Fig. 6; Table 2). Three quarters of this amount would be expected since 4.5 Ma if the Wheeler fault slip rate is simply assumed to have been steady. Such an assumption is now warranted based on the reinterpretation of cemented river gravels in the footwall of the Wheeler fault as locally derived as opposed to Bullhead Alluvium. Adding the offsets across all fault systems indicates between 245 and 355 m of regional offset5 by faulting since 4.5 Ma across the Lake Mead area, in good agreement with estimates based on differential incision6 of >140 m and up to 370 m over the past 2–4 Ma. Because the strath profile of the Bullhead Alluvium was approximately graded to paleo–sea level, negating the possibility of regional subsidence of the lower Colorado River corridor, the regional fault offsets and the incision variations across the Lake Mead area suggest to us ∼140–370 m of uplift7 of the western Colorado Plateau in the past ∼4 Ma.

Pleistocene slip rates on the Wheeler fault are poorly constrained, but qualitative observations suggest waning slip during the Late Pleistocene (Pearthree et al., 1983). Similarly, the tread of the Late Pleistocene Chemehuevi terrace also does not appear to be offset across the Fortification fault system (Longwell, 1936; Malmon et al., 2011). Farther north, the Overton Arm faults and the northern part of the Grand Wash fault show evidence for movement during the latest Pleistocene (Anderson and O’Connell, 1993), suggesting that the locus of deformation may have moved northward during the end of the Pleistocene.

Uplift Mechanism

A potential driving force for differential uplift between the lower Colorado River corridor and western Grand Canyon is the isostatic response to changes in the buoyancy of the crust or lithosphere, including differential surface denudation (Pederson et al., 2002b, 2013; Lazear et al., 2013) and progressive lithospheric removal or modification due to upwelling asthenospheric mantle (Moucha et al., 2009; van Wijk et al., 2010; Crow et al., 2011; Levander et al., 2011). Lazear et al. (2013) used thermochronologic constraints and the modern positions of ca. 10 Ma basalt flows to estimate the distribution of material removed from the Colorado Plateau and Rocky Mountains area in the past 10 Ma. They then calculated the flexural isostatic response to the unloading or the amount of rock uplift. They estimated ∼200 m of post–10 Ma denudation-driven rebound of western Grand Canyon relative to the Lake Mead area, with hundreds of meters more in eastern Grand Canyon. However, denudation-driven post–10 Ma rebound (Lazear et al., 2013) and long-term (<600 ka) incision rates are not spatially correlated in Grand Canyon (Crow et al., 2014). This lack of correlation is likely because most of the denudation occurred after Colorado River integration (Murray et al., 2016) and most of the incision rates are calculated over much shorter time frames (Karlstrom et al., 2013).

Isostatic compensation would also occur in the case of lithospheric removal or buoyancy modification due to upwelling asthenospheric mantle, which is supported by geophysical and geodynamic observations including: (1) large P-wave velocity gradients around the Colorado Plateau at lithospheric depths (i.e., 80 km) (Fig. 1 inset) suggesting a non-equilibrium state in the mantle (Schmandt and Humphreys, 2010); (2) receiver functions interpreted to show areas of lithospheric and crustal delamination on the Colorado Plateau (Levander et al., 2011); and (3) positive lithospheric geoid anomalies around the margins of the Colorado Plateau indicating relatively buoyant mantle in those areas (Coblentz et al., 2011). These observations and numerical modeling have been used to argue for whole-mantle convection (Moucha et al., 2009), edge-driven upper-mantle convection (van Wijk et al., 2010), lithospheric delamination (Levander et al., 2011), and/or lithospheric buoyancy modification by a migrating zone of mantle modification (Crow et al., 2011), which could all cause the differential uplift. The latter is supported by the rate and direction of migrating volcanism (Crow et al., 2011). Ultimately the denudation-driven (Pederson et al., 2013) and mantle-driven models are not mutually exclusive, and both processes may be contributing to the observed differential uplift. However, a mantle-driven component is likely required because the post–4 Ma uplift estimates (140–370 m) across the Lake Mead region from profile and incision analysis are similar to or about double the modeled magnitude of the denudation-driven rebound in the same area over the longer time frame of the past 10 Ma (200 m). Mantle-driven uplift could also produce a positive feedback where mantle-driven uplift promotes increased erosion and hence additional rock uplift. Mantle-driven differential uplift of eastern Grand Canyon relative to western Grand Canyon has also been suggested based on a similar differential incision study (Crow et al., 2014).

Deep-seated uplift would be expected to have wavelengths proportional to the thickness of the isostatic layer (Lazear et al., 2013). The regional uplift we observe in the Lake Mead area however seems to be localized at specific crustal normal faults, which are not expected to localize regional uplift (Hanks and Blair, 2004). This poses an interesting conundrum. Given the likelihood that the isostatic response to differential surface denudation and/or progressive lithospheric buoyancy modification has driven Plio-Pleistocene uplift of the western Grand Canyon region, we propose that the differential uplift along faults in the Lake Mead area may be a response in the brittle and extensively deformed crust to more deep-seated, long-wavelength uplift.


A new, comprehensive compilation of ancestral Colorado River deposits combined with improved incision constraints in the Lake Mead area has allowed us to evaluate evidence for Plio-Quaternary uplift and subsidence along the lower Colorado River. There is clear evidence for long-term river incision into bedrock and uplift of older river deposits through the central and eastern Lake Mead area and in western Grand Canyon. In comparison, little if any bedrock incision has occurred in western Lake Mead or in the lower Colorado River corridor. We interpret this to reflect regional uplift of eastern Lake Mead and the western margin of the Colorado Plateau, probably driven by a combination of the isostatic response to denudation and changes in mantle buoyancy. Differential incision suggests at least 140 m of uplift across the Lake Mead region in the past 2.2 Ma and potentially as much as 370 m in the past 4 Ma. The uplift in the Lake Mead area seems to coincide with normal faults, and differential incision there is equivalent to cumulative vertical offset across the Fortification, Detrital, and Wheeler fault systems, 245 and 355 m.

In contrast, there is no definitive evidence for regional Plio-Quaternary deformation along the lower Colorado River below Lake Mead. The 4.5 Ma Colorado River bedrock strath remains graded to within the range of paleo–sea level estimates. Bullhead Alluvium, present below sea level in the Yuma and Palo Verde Valleys, requires localized basin subsidence, which is likely fault controlled in large part. This subsidence seems to have been localized in the center of the basins, because the highest outcrops on the flanks of the basins probably represent elevations of maximum aggradation during the 3.5–4.5 Ma Bullhead and 70 ka Chemehuevi aggradational events, and neither of these shows clear evidence for deformation. The similarity in the slopes of the tops of the Bullhead Alluvium and the Chemehuevi Formation, away from areas of localized subsidence, indicates that significant regional tilting is unlikely there, as opposed to the Lake Mead region where there is evidence for offset, folding, and uplift.


This work was supported by FEDMAP and Mendenhall funding to Crow and the U.S. Geological Survey office in Flagstaff, Arizona, and National Science Foundation funding (EAR-0711546) to Karlstrom and Crossey. The manuscript benefited from informal discussions with Tom Hanks and formal reviews from David Harbor, Scott Bennett, and an anonymous reviewer.

1Howard et al. (2015) referred to these as the Callville faults.
2Supplemental Materials. Compilation of paleo-sea level estimates, valley miles through lower Colorado River corridor, references for well data, and summary of depths to bedrock and Bouse Formation in key wells. Please visit https://doi.org/10.1130/GES02020.S1 or access the full-text article on www.gsapubs.org to view the Supplemental Materials.
3River miles as opposed to river kilometers are used due to their prevalence in the literature.
4Python script built in large part on the “Near” and “Zonal Statistics” tools in ESRI ArcMap software.
5145 m + (130 m × 0.75) = 242.5 m (rounded to 245 m); 230 m + (165 m × 0.75) = 353.75 m (rounded to 355 m).
6187 m − 10 m − 35 m = 142 m (rounded to 140 m; this is the lower incision estimate minus its error and the maximum amount of incision in the lower Colorado River corridor); 363 m + 10 m = 373 m (rounded to 370 m; this is the higher incision estimate plus its error and assumes no incision in the lower Colorado River corridor).
7We take a conservative approach of using the lowest and highest offset or differential incision estimates to estimate uplift.
Science Editors: Raymond M. Russo, Andrea Hampel
Guest Associate Editor: Andres Aslan
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.