Paleochannels, stream incision, erosion, topographic evolution, and alternative explanations of paleoaltimetry, Sierra Nevada, California

For over a century, geomorphologists have used the Sierra Nevada of California as a natural laboratory to explore general geomorphic process (e.g., Lawson, 1903; Matthes, 1937; Wahrhaftig, 1965). In the past few decades, the Sierra Nevada has attracted attention as a locality for evaluating climatically versus tectonically driven geomorphic evolution (Small and Anderson, 1995), an example of epeirogenic uplift (uplift not associated with major shortening or collision) and its causes (e.g., Unruh, 1991; Ducea and Saleeby, 1998; Jones et al., 2004; Le Pourhiet et al., 2006), as well as a location for evaluation of erosion and weathering processes and their forcing mechanisms (e.g., Small and Anderson, 1995; Granger et al., 2001; Riebe et al., 2000, 2001; Stock et al., 2005; Clark et al., 2005; Phillips et al., 2011).

A debate has emerged as to whether or not late Cenozoic rock and surface uplift has occurred in the Sierra Nevada, and this debate has implications for analysis of general geomorphic process and evaluation of recently developed methods in paleoaltimetry. Some suggested a lack of late Cenozoic uplift on the basis of thermochronology (e.g., House et al., 1998), or on the basis of the generally thin Sierra Nevada crust and the tectonic setting of the Sierra Nevada (Wernicke et al., 1996). More recently, researchers have proposed a lack of late Cenozoic uplift on the basis of paleobotany (e.g., Hren et al., 2010), isotopic paleoaltimetry (e.g., Poage and Chamberlain, 2002; Mulch et al., 2006; Crowley et al., 2008; Cassel et al., 2009, 2012; Chamberlain et al., 2012), and interpreted steep paleogradients of Oligocene and Eocene stream deposits (Cassel and Graham, 2011; Cassel et al., 2012). These interpretations directly contradict proposals for late Cenozoic rock and surface uplift based on thermochronology (McPhillips and Brandon, 2010, 2012), geodetically measured current uplift rates of the range (Hammond et al., 2012), and a combination of geologic field relationships and evaluation of geomorphic process (Lindgren, 1911; Hudson, 1955; Bateman and Wahrhaftig, 1966; Christensen, 1966; Huber, 1981, 1990; Unruh, 1991; Wakabayashi and Sawyer, 2001; Jones et al., 2004; Stock et al., 2004, 2005; Clark et al., 2005; Figueroa and Knott, 2010; Kemp, 2012).

This paper will first review data on stream incision, erosion, and evidence for pre–late Cenozoic relief (“paleorelief”) following Wakabayashi and Sawyer (2001), but with significant updates and additions on the timing of incision, evaluation of knickpoint migration speed, faulting, exhumation and erosion rates, along-strike differences, and paleorelief east of the current Sierra Nevada. I will then revisit the relationship between paleochannel gradients and azimuth and “inverse” (uphill in present-day setting) gradients (from Lindgren, 1911; Hudson, 1955; Jones et al., 2004) that starkly contrasts with modern stream reaches. The stream incision history and paleogradient-azimuth relation will then be discussed in the context of supporting evidence for late Cenozoic uplift, followed by presentation of alternative interpretations of the paleoaltimetry data, and speculations on connections between tectonic mechanisms, uplift, and topographic evolution.

The 600-km-long Sierra Nevada is the most prominent mountain range in California. The Sierra Nevada and the Central Valley belong to the Sierra Nevada microplate, an element of the broad Pacific–North American plate boundary (Argus and Gordon, 1991) (Fig. 1). The Sierra Nevada microplate displays little internal deformation north of the Kern River drainage and moves northwestward relative to stable North America at ∼10–14 mm/yr, a fourth or fifth of the dextral motion of the plate boundary (Argus and Gordon, 1991; Dixon et al., 2000). The microplate is bounded on the west by an active fold-and-thrust belt that marks the eastern margin of the Coast Range province (e.g., Wentworth and Zoback, 1989), and on the east by a prominent east-facing escarpment that marks the Sierra Nevada Frontal fault system (referred to as Frontal fault system or FFS herein), a zone of normal, normal-dextral, and dextral faulting (e.g., Clark et al., 1984; Beanland and Clark, 1994; Unruh et al., 2003; Le et al., 2007) (Fig. 2). Whereas the down-to-the-east component of deformation across the FFS fault strands results in prominent east-facing topographic escarpments ranging from 500 to 3000 m in height, dextral shear predominates along this transtensional fault system (Unruh et al., 2003). The FFS separates the Sierra Nevada microplate from the Basin and Range province to the east, or, alternatively, the FFS represents the westernmost strand of the dextral Walker Lane belt. The northernmost part of the FFS curves to a WNW strike, forming the boundary between the northern margin of the Sierra Nevada microplate and the Cascades (Fig. 2); this has been called the Sierra Nevada–Cascade Range boundary zone (Sawyer, 2009, 2010).

The Sierra Nevada slopes gently westward and abruptly eastward from its crest from the Kings River drainage to the northern termination of the range (Fig. 2). The northernmost part of the range slopes westward at a gradient of ∼38 m/km (gradient 0.038 or 2.2°) over a distance of ∼50 km. In the Yuba River drainage the westward gradient averages ∼28 m/km (gradient 0.028 or 1.6°), with a lower-gradient western portion of ∼20 m/km (gradient 0.020 or 1.1°) for the western foothills averaged over ∼40 km, and ∼35 m/km (gradient 0.035 or 2°) for the eastern higher-elevation portion averaged over ∼50 km. From the San Joaquin River drainage to the Kaweah River drainage, the foothills region has a steep middle- to high-elevation slope that levels off eastward. For example, the San Joaquin River–Kings River divide area has a western foothills region with a slope of ∼27 m/km over ∼20 km (gradient 0.027, 1.5°), then 80 m/km for the next 25 km (gradient 0.080, 4.6°), and ∼18 m/km (gradient 0.018, 1.0°) for the remaining 40 km or so to the crest of the range, with an overall average slope of ∼38 m/km (Fig. 2). The southernmost part of the range (mid–Kern River drainage southward) lacks the simple block topography of regions to the north, and in the southernmost part of the Kern drainage, the highest elevations are on a divide west of the crest, and topography of this region is controlled more by faults along the Kern River than the FFS (Saleeby et al., 2009).

Crestal elevations vary from 2100 to 2700 m in the northern part of the range to 4000–4400 m in the central to southern part of the range, with the highest elevations in the headwaters of the Kings and Kern Rivers, and maximum elevations decreasing to the south. The height of the eastern escarpment varies from ∼1000 m in the northernmost part of the range to nearly 3300 m at Lone Pine (Fig. 2).

The axis of the range that parallels smoothed (fitted across drainages) elevation contours on the western flank varies from about N35°W in the Feather River drainage to N25°W in the southern Yuba River and American River drainages, to N40°W from the Stanislaus River southward. This along-strike change in the range axis orientation mirrors along-strike changes in the strike of the Frontal fault system (Fig. 2) (Wakabayashi and Sawyer, 2001; Unruh et al., 2003).

Definitions of the eastern boundary of the Sierra Nevada vary, but I will follow the definition used by Wakabayashi and Sawyer (2001) wherein the Sierra Nevada is defined as being west of the western strands of the FFS. This definition places ranges such as the Diamond Mountains (headwaters of the North Fork Feather and Middle Fork Feather Rivers) and Carson Range (range east of Lake Tahoe) east of the Sierra proper. Note that the strike and location of basement units associated with the range, such as the Sierra Nevada batholith, diverge from that of the current mountain range (Wakabayashi and Sawyer, 2001).

Prior to becoming part of the transform plate margin, arc magmatism occurred in what later became the Sierra Nevada. Mesozoic arc activity, associated with east-dipping subduction, included the emplacement of the Sierra Nevada batholith and ended at ca. 85 Ma (Saleeby and Sharp, 1980; Stern et al., 1981; Chen and Moore, 1982; Saleeby et al., 2008). The Mesozoic plutons intrude Mesozoic and Paleozoic, metaigneous and metasedimentary rocks (e.g., Schweickert, 1981; Moores and Day, 1984; Sharp, 1988; Saleeby et al., 1989). Collectively, the Mesozoic plutons and the Mesozoic to Paleozoic metamorphic rocks will be referred to as “basement” in this paper.

The cessation of magmatic activity in this region apparently coincided with a shallowing of subduction dip that resulted in the migration of the locus of magmatism far eastward, as well as significant shortening in the foreland region associated with the Laramide orogeny (Dickinson and Snyder, 1978). The resultant crustal thickening apparently resulted in a broad region of elevated topography, comparable to the Altiplano or Tibetan Plateau (Christiansen and Yeats, 1992; Dilek and Moores, 1999), with a drainage divide hundreds of kilometers east of the current crest of the Sierra Nevada (Christiansen and Yeats, 1992; Henry, 2008; Henry and Faulds, 2010; Henry et al., 2012). Following cessation of Cretaceous arc magmatism, exhumation resulted in surface exposure of the batholith. Eocene stream gravels, sourced well east of the present Sierran crest (e.g., Garside et al., 2005), were deposited on the exhumed granitic and older metamorphic rocks in what became the northern Sierra Nevada (Bateman and Wahrhaftig, 1966).

Regional-scale Cenozoic volcanism began with 31–23 Ma rhyolitic ash flow tuff deposition in channels crossing the Sierra (Garside et al., 2005; Henry and Faulds, 2010; Henry et al., 2012), and culminated with widespread and voluminous andesitic volcanism from 16 to 3 Ma, associated with the ancestral Cascades arc (e.g., Busby et al., 2008a, 2008b; Busby and Putirka, 2009; Cousens et al., 2008). The southernmost active volcano of the Cascades arc is Mount Lassen (Figs. 1, 2). Ancestral Cascades arc activity resulted in voluminous deposition of volcanic rocks as far south as the northern Tuolumne River drainage; this volcanism shut off in the wake of the migrating southern edge of the subducted slab as the Mendocino triple junction migrated northward and the convergent plate margin was replaced by a transform one (e.g., Atwater and Stock, 1998; Busby and Putirka, 2009).

Estimates of Cenozoic landscape evolution and deformation are best constrained where Cenozoic deposits are present in the Sierra Nevada (Fig. 1). Widespread Cenozoic deposits are limited to the area north of the Tuolumne River (Fig. 1). The oldest of the Cenozoic cover strata are Eocene gold-bearing gravels, commonly referred to as the “auriferous gravels,” and their fine-grained equivalents, the Ione Formation, along the western base of the northern and central part of the range (e.g., Bateman and Wahrhaftig, 1966; Garside et al., 2005) (paleochannels shown in Fig. 2). These gravels filled drainages that flowed westward from a divide ∼400–500 km (present-day distance) east of the present Sierra Nevada crest (Henry, 2008; Henry and Faulds, 2010; Henry et al., 2012).

The Oligocene (31–23 Ma) rhyolite tuffs, including the Valley Springs and Delleker Formations, overlie the Eocene gravels (Wagner et al., 1981; Saucedo and Wagner, 1992; Henry and Faulds, 2010). These apparently erupted from calderas 200–400 km (present-day distance) east of the present Sierran crest and filled the same west-flowing drainages previously filled by the Eocene gravels (Henry and Faulds, 2010). These rhyolites are overlain by ca. 16–3 Ma andesites, andesitic mudflows, and associated volcanic sedimentary rocks of the ancestral Cascades arc that erupted from vents in the vicinity of, but generally east of, the present crest (Wagner et al., 1981; Saucedo and Wagner, 1992; Bartow, 1979; Busby et al., 2008a, 2008b; Cousens et al., 2008). The term “Mehrten Formation” has been applied to describe these andesitic rocks, excluding those north of the North Fork Feather River (for example, Unruh, 1991; Wakabayashi and Sawyer, 2001), following the definition of Curtis (1954). Andesites of the ancestral Cascades arc blanketed the northern and central Sierra, covering all but a few scattered basement highs, with extensive deposits preserved as far south as the northern Tuolumne River drainage (Slemmons, 1966; Durrell, 1966; Guffanti et al., 1990, Busby et al., 2008a, 2008b; Busby and Putirka, 2009; Cousens et al., 2008).

The age of the youngest strata of the ancestral Cascades in any part of the Sierra is important, because the initiation of significant incision followed the deposition of these units. Andesitic flows in the headwaters of the North Fork American River have K-Ar and Ar-Ar ages as young as 3.0 Ma (Harwood, 1981; Cousens et al., 2008, 2012) data in Saucedo and Wagner, 1992); these are the youngest ages obtained from andesites south of the North Fork Feather River. Between these rocks and the North Fork Feather River, the youngest K-Ar age from andesitic rocks is 6.8 Ma (Saucedo and Wagner, 1992), but age data in this region are sparse.

The youngest andesitic rocks in the Sierra proper, yielding 2.4–3.3 Ma Ar-Ar and K-Ar ages (Guffanti et al., 1990; Wakabayashi and Sawyer, 2000; Kemp, 2012; M.A. Clynne, unpublished data), cover basement north of the North Fork Feather River. These rocks are associated with the Yana volcanic center and are called the Tuscan Formation on the western flank of the northernmost Sierra (Guffanti et al., 1990). Northward younging of the youngest late Cenozoic volcanic rocks in the Sierra Nevada may be expected because of the northward migration of the Mendocino triple junction and associated shutoff of subduction and arc volcanism, but existing data do not show such a pattern except from the Yana volcanic center northward to the active Lassen volcano (Guffanti et al., 1990; M.A. Clynne, unpublished data).

The Eocene gravels, Oligocene rhyolites, and Mio-Pliocene andesitic volcanic rocks mainly crop out north of the Tuolumne River within the Sierra Nevada, but minor accumulations of these deposits are found along the eastern margin of the Central Valley as far south as the Kings River (e.g., Bartow, 1985, 1990). Scattered outcrops of Miocene to late Quaternary volcanic rocks occur south of the Tuolumne River (Moore and Dodge, 1980). These volcanic rocks represent local eruptive events rather than the Cenozoic volcanic arc and, prior to erosion, constituted a much smaller volume than ancestral Cascades rocks to the north (Ducea and Saleeby, 1998).

Canyons are a first-order topographic feature of any mountain range, and the incision of these canyons represents a significant geomorphic response to climatic, depositional, or tectonic processes. Accordingly, the evidence of stream incision serves as an important component in the evaluation of the Sierra Nevada uplift.

The base of late Cenozoic deposits, including deposits representing the local thalwegs of major paleochannels, crop out hundreds of meters above canyon bottoms incised into basement and thus record incision that postdates the deposition of the late Cenozoic deposits (Fig. 3). Although the streams have locally incised up to several hundred meters into late Cenozoic, mainly volcanic, deposits as well (Bateman and Wahrhaftig, 1966; Busby et al., 2008a, 2008b; Busby and Putirka, 2009), the incision through the late Cenozoic deposits or in basement to the level of the bottom of older, Cenozoic paleochannels can be driven by deposition of these late Cenozoic deposits alone, because their deposition may have drastically changed local base levels (e.g., Huber, 1990). This can be viewed as aggradation (deposition) that raised the elevation of the thalweg, followed by incision through the deposits (or through basement adjacent to such deposits) to the former equilibrium profile of the stream after the sediment load dropped and aggradation ceased. Volcanic deposition represents exceptionally fast aggradation, and in the most extreme cases this aggradation is associated with significant steepening of the stream owing to the construction of volcanic edifices. The canyons carved in the flanks of the modern Cascade volcanoes and associated flows illustrate this principle well. A similar, but smaller-scale, example of incision following aggradation is the incision of streams into the alluvial plains, but not into bedrock, on the western margin of the Sierra Nevada, where rapid deposition during glaciation was followed by incision through alluvium during the present interglacial period (e.g., Janda, 1966). Accordingly, I will refer to incision only as that into basement beneath the base of the Cenozoic deposits (“basement incision” of Wakabayashi and Sawyer, 2001).

Incision in the major drainages increases upstream from zero at the western base of the range to the deepest parts of the canyons then decreases to zero near the headwaters of the streams just west of the crest or drainage divide (Wakabayashi and Sawyer, 2001) (Fig. 2). The North Fork Feather and Middle Fork Feather Rivers are exceptions to this because they cross the crest of the range, so their deepest incision corresponds to the point where they cross the crest of the range. Post-Mehrten incision in the northern and central Sierra Nevada ranges up to ∼1200 m (Table 1) (Wakabayashi and Sawyer, 2001) and the deepest incision in most drainages occurs downstream of the limit of glaciation.

Long-term late Cenozoic incision rates can be estimated by measuring the elevation difference between the present channel bottoms and the base of late Cenozoic deposits capping the interfluves, and dividing the elevation difference by the age of the youngest of the deposits (Fig. 3). Such rates are minima, because time required for the stream to cut through the volcanic rocks is not accounted for and the time elapsed after deposition of interfluve-capping units and the initiation of incision is poorly constrained in most cases. The age of the oldest units inset beneath base of the interfluve-capping deposits provides the minimum age of incision initiation.

To compare timing of incision range-wide, the speed of headward migration of an incision pulse (associated with knickpoint migration) in a given drainage warrants consideration. Field relationships in the North Fork Feather River drainage suggest rapid knickpoint retreat, or near-synchronous incision initiation along a significant length of the trunk stream. Andesites of the Yana volcanic center cap upland surfaces on the north rim of the North Fork Feather River canyon, and these deposits extend to the western flank of the range (Guffanti et al., 1990). Andesite capping the ridge north of the East Branch North Fork–North Fork confluence yielded an Ar-Ar age of 2.8 Ma (Wakabayashi et al., 1994; Wakabayashi and Sawyer, 2000). K-Ar and Ar-Ar ages of similar andesites in this part of the drainage range between ca. 2.8 and 3.0 Ma, including rocks capping the crestal highlands north of the North Fork Feather River (Kemp, 2012; M.A. Clynne, unpublished data; WLA, 1996). This suggests that incision along the North Fork Feather River began after 2.8–3.0 Ma.

Over 20 km upstream of the North Fork Feather River–East Branch North Fork Feather River confluence, a 2.1 Ma basalt flow is inset below an erosion surface that makes up the west rim of the North Fork Feather River canyon (Wakabayashi et al., 1994; Wakabayashi and Sawyer, 2000). This surface is overlain by Yana volcanic rocks ∼10 km north of the basalt exposure (WLA, 1996). The base of the 2.1 Ma basalt is 180 m below the erosion surface and 385 m above the canyon bottom. These figures are revised from those in Wakabayashi and Sawyer (2000) because the earlier incision amounts were based on what I consider a somewhat oblique projection to the canyon bottom (with respect to the position of the 2.1 Ma basalt remnant).

If the incision rate from 2.1 Ma to present (0.18 mm/yr) is applied to the 180 m of pre–2.1 Ma incision beneath the erosion surface, the estimated age of incision initiation is 3.1 Ma, but this is older than the age of the andesite on the rim of the East Branch North Fork–North Fork confluence. A pre–2.1 Ma incision rate faster than the post–2.1 Ma rate (required to initiate incision at 2.8 Ma or later) would reflect a temporal pattern of incision rate opposite of that recorded in other reaches of the North Fork Feather River (Wakabayashi and Sawyer, 2001). Collectively, these field relationships suggest that the time of knickpoint retreat is shorter than the collective uncertainty of the age of ridge-capping andesites and the estimate of incision time prior to deposition of the 2.1 Ma basalt. These relationships suggest knickpoint retreat of over 100 km from the mouth of the North Fork Feather River in less than ∼300 k.y. For the purposes of discussion, I will adopt a time of ca. 3 Ma for the initiation of incision in the North Fork Feather River canyon.

The estimated knickpoint retreat rate for the North Fork Feather River may be as fast as that determined by Crosby and Whipple (2006) (tens of kilometers of retreat within 18 k.y.) for the Waiapaoa River of New Zealand, and that estimated for the upstream reaches of the Yellow River in China (over 250 km within 500 k.y.) by Burbank and Anderson (2012) based on studies by Harkins et al. (2007) and Craddock et al. (2010), to as slow as an order of magnitude slower, within the limits of age constraints. The estimated knickpoint retreat rate for the North Fork Feather River is one to two orders of magnitude faster than that estimated by Berlin and Anderson (2007) for streams draining the Roan Plateau to the Colorado River in Colorado, one to two orders of magnitude faster than the rates estimated by numerical modeling of Stock et al. (2004) for the Kings River, and two to three orders of magnitude faster than rates estimated by numerical models of Pelletier (2007) for southern Sierra rivers. If the North Fork Feather River knickpoint migration rates apply to other Sierran drainages, then incision may have propagated from the mouths of the trunk streams to near their headwaters within a few hundred thousand years. Somewhat looser age constraints in the Mokelumne and Stanislaus River drainages (see below) are consistent with this conclusion.

Incision in the North Fork American River drainage appears to have begun after ca. 3.3 Ma, based on the age of the andesites capping a ridge southeast of Donner Summit (Shriver and Wakabayashi, 2010). A minimum age of incision initiation there cannot be determined because of the lack of dated inset units. Initiation of tilting of Pliocene strata in the Boca Basin directly east of the FFS and the Donner Summit area began at ca. 2.7 Ma (Mass et al., 2009), possibly signaling initiation of dip-slip faulting along this part of the FFS. However, the initiation of faulting along this part of the FFS need not coincide with incision west of the crest at this time, given the local examples of Frontal faulting postdating initiation that I will review below in the section on faulting, as well as the uncertainty of knickpoint retreat time noted above. Given these uncertainties, I will consider incision to have initiated in the American River drainage at ca. 3 Ma.

The youngest andesitic volcanic rocks (Mehrten Formation) capping the interfluve between the Mokelumne River and Calaveras River (a comparatively small stream between the Mokelumne and Stanislaus Rivers) in the western foothills region are ca. 4 Ma, based on K-Ar ages of two dacitic plugs that intrude the Mehrten, one of which is overlain by uppermost Mehrten strata (Bartow, 1979). Thus, incision of the Mokelumne River appears to have initiated after 4 Ma, and this constraint appears applicable to the Stanislaus River, owing to the proximity of the latter to the dated interfluve deposits. Cosmogenic nuclide dating of cave sediments indicate that ∼390 of 490 m of late Cenozoic incision of a reach of the Stanislaus River canyon had taken place prior to 1.6 Ma (Stock et al., 2005). If an incision rate of 0.20 mm/yr, significantly higher than the long-term average rate for this part of the canyon (0.12–0.13 mm/yr), is adopted for a maximum rate of pre–1.6 Ma incision, then incision had begun by ca. 3.6 Ma at this point in the Stanislaus River canyon. Accordingly, incision in the Mokelumne and Stanislaus River drainages initiated between ca. 3.6 and 4 Ma.

These relationships also provide a constraint on knickpoint migration along the Stanislaus River. The 4 Ma age of the top of the andesitic volcanics should apply to the surface of the interfluve extending to near the mouth of the river (downstream limit of incision). If so, incision in the Stanislaus River migrated over 50 km upstream in less than 400 k.y. These constraints are looser than those in the North Fork Feather River, owing to the larger age difference between interfluve-capping and inset deposits.

Along the western flank of the Sierra Nevada, Cenozoic strata of Holocene to Eocene age show progressively greater dip (tilt perpendicular to range axis) with greater age (Unruh, 1991). Projection of such tilts toward the crest would suggest progressive incision from Eocene to modern time (Huber, 1981; Unruh, 1991) and would predict that the progressively younger units should be inset below older ones in canyons. In contrast, Mio-Pliocene andesitic rocks overlie Oligocene rhyolites at the crest of the range from the Stanislaus drainage northward, and these units in turn overlie Eocene gravels in the crestal region from the American River drainage northward (Wakabayashi and Sawyer, 2001). These geomorphic-stratigraphic relationships apply in the above drainages from the crest to the downstream limit of incision and thus limit the area of differential Eocene–Mio-Pliocene tilts to a relatively narrow zone along the eastern margin of the Central Valley.

Initiation of incision in the Tuolumne River is even more poorly constrained owing to a paucity of age dates on the youngest andesites in the drainage. For the Tuolumne River, I will adopt an initiation of incision age of ca. 4 Ma for the Tuolumne based on proximity to the Stanislaus and Mokelumne drainages. The Merced River drainage lacks dated late Cenozoic deposits, precluding a reliable estimate of the timing of late Cenozoic incision initiation.

In the San Joaquin River drainage, incision began sometime between 10 Ma and ca. 3.5 Ma, based on the 3.4–3.9 Ma (K-Ar age; Dalrymple, 1964) volcanic rocks inset as much as 580 m below the reconstructed position of the 10 Ma paleochannel (Huber, 1981). Recent Ar-Ar dating, field mapping, and paleomagnetic data have confirmed the inset relationship of the volcanic rocks in the San Joaquin River drainage and allowed assignment of an age of ca. 3.6 Ma to those rocks (Carlson et al., 2009). Similar to the approach followed for the Stanislaus River, if the highest incision of central-southern Sierra streams (0.27 mm/yr for the 2.7–1.4 Ma period in the Kings River; Stock et al., 2004) is applied for the San Joaquin prior to 3.6 Ma, then late Cenozoic incision began before ca. 6 Ma. Accordingly, late Cenozoic incision probably began in the San Joaquin River between 10 and 6 Ma.

Maximum incision amounts are given in Table 1 for other major Sierran rivers, where constraints on timing are scarce or lacking. For those drainages, the timing of incision is estimated by comparison to adjacent drainages for which better geochronologic constraints are available.

Temporal variation of incision rate apparently differs along strike. The Kings River incised at 0.27 mm/yr between 2.7 and 1.4 Ma, and 0.02 mm/yr thereafter (Stock et al., 2004). In contrast, one reach of the North Fork Feather River (east of the crest and upstream of the point of greatest incision and rates) incised at 0.08–0.09 mm/yr from 2.8 to 1.1 Ma, accelerated to 0.25–0.40 mm/yr from 1.1 Ma to 0.6 Ma, then slowed to 0.12–0.16 mm/yr after 0.6 Ma (Wakabayashi and Sawyer, 2000, 2001). For the major drainages from the Kings River northward, late Cenozoic basement incision rates associated with the deepest parts of the canyons range from 0.10 to 0.40 mm/yr, averaged over the entire time span of incision (Table 1). The Kern River exhibits post–3.5 Ma incision rates of 1.1 mm/yr (Saleeby et al., 2009), the fastest long-term incision rate recorded in the Sierra.

Incision between Eocene and Pliocene time is recorded by the position of post-Eocene paleothalwegs versus Eocene paleothalwegs (Fig. 3). The maximum incision of Oligocene or Miocene volcanic rocks below the base of Eocene channels is no more than ∼30 m (Bateman and Wahrhaftig, 1966). The minimum age difference between the youngest Eocene gravels and the end of ancestral Cascades (Mehrten Formation) deposition is ∼30 m.y. (e.g., Garside et al., 2005; Henry and Faulds, 2010). Averaging 30 m of incision over 30 m.y. yields 0.001 mm/yr as a maximum Eocene to Miocene incision rate.

Incision of Miocene paleochannels through Oliocene or Miocene volcanics or through basement to levels at or above the thalwegs of older paleochannels apparently took place in the crestal region of the Sierra Nevada (Busby et al., 2008a, 2008b; Busby and Putirka, 2009). As noted above, incision that does not cut below the base of the previous paleochannel reflects rapid aggradation and constructional-depositional topography, followed by decrease in sediment load that leads to incision back toward the pre-aggradational base level, rather than tectonic base level changes (i.e., uplift) or a discharge increase in a stream, so this intra-formational incision will not be considered in this discussion.

In general, the incision history of the Sierra Nevada (San Joaquin River drainage and northward) shows exceedingly low rates of stream incision from Eocene to Miocene–Pliocene time, followed by diachronous initiation of late Cenozoic incision at between 6 and 10 Ma in the San Joaquin River and ca. 3 Ma in the American and Feather River drainages. Late Cenozoic incision in the Kings to Kern River drainages of the southernmost Sierra began earlier (20 Ma) than drainages to the north, based on thermochronologic data (House et al., 2001; Clark et al., 2005; Clark and Farley, 2007; Saleeby et al., 2009).

Minimum topographic relief that existed at the time of the deposition of Cenozoic deposits may be estimated by comparing the elevation of basement topographic highs relative to the elevation of the local base of Cenozoic strata (Fig. 3) (Lindgren, 1911; Bateman and Wahrhaftig, 1966; Wakabayashi and Sawyer, 2001). The elevation difference is a minimum estimate of relief that predated late Cenozoic stream incision (referred to as paleorelief), because some erosional lowering of topographic highs has occurred during the late Cenozoic. The amount of lowering of these topographic highs in the late Cenozoic is probably small based on the extremely low erosion rates measured for bedrock upland surfaces (erosion surfaces) of the Sierra Nevada, ∼0.004 mm/yr since ca. 11.8 Ma, based on Ar-Ar dating of volcanic rocks deposited on low-relief, high-altitude surfaces of the Sierra (Phillips et al., 2011). The long-term rates are similar to the shorter-term rates determined by cosmogenic nuclide geochronology for upland low-relief surfaces in the northern (Riebe et al., 2000) and southern/central Sierra (Small et al., 1997; Stock et al., 2005) that range from ∼2 to 20 m/m.y. (0.002–0.019 mm/yr) over the last 30–236 k.y. The estimates of paleorelief using Cenozoic stratigraphic-geomorphic relationships are best constrained northward from the San Joaquin River drainage, owing to the scarcity of Cenozoic deposits to the south.

Paleorelief in the Sierra Nevada increases from north to south, with a significant increase south of the Stanislaus River drainage (Wakabayashi and Sawyer, 2001); this section of this paper largely reviews the earlier analysis with addition of new analysis of paleochannel depths associated with the largest paleochannels. Most of the region north of the American River has paleorelief of less than 400 m (Figs. 2, 4). In this region, the paleorelief defines paleochannels that deepen eastward toward the crest, with isolated paleohighs composed mainly of resistant metamorphic rock (Fig. 4). North of the Tuolumne drainage it is primarily the isolated basement highs that exceed the elevation of ridges capped by Miocene andesites (Fig. 4), whereas from the Tuolumne River southward, the tops of the andesitic deposits are inset below basement rims of paleocanyons. Paleorelief exceeds 1000 m in parts of the San Joaquin drainage, and the southward increase in paleorelief coincides with the southward increase in elevation in the range (Figs. 2, 4). Paleorelief appears to greatly exceed 1000 m in the Kings River drainage based on the age of cave deposits at different levels above the canyon bottom (Stock et al., 2004). In addition, the steeper western slope along the southern Sierra corresponds to an abrupt eastward increase from 100 to over 600 m in paleorelief along the San Joaquin drainage (Huber, 1981). In addition to direct evaluation of paleorelief from late Cenozoic deposits, thermochronlogic studies (House et al., 2001; Clark et al., 2005; McPhillips and Brandon, 2012) suggest large magnitudes of paleorelief (1 km or greater) for the southern Sierra Nevada.

The distribution of paleorelief suggests that the major along-strike differences in topographic expression of the range are largely a consequence of greater paleorelief in the southern (south of Stanislaus River) compared to the northern part of the range (Wakabayashi and Sawyer, 2001). This is consistent with the hypothesis of Wahrhaftig (1965), who argued that the west-facing topographic escarpments of the south-central and southern Sierra were erosional in origin and not late Cenozoic fault scarps. West-facing topographic steps in the Tuolumne River drainage coincide with possible down-west offsets in the paleothalweg of Miocene paleochannels, but the long distance between paleochannel deposits may alternatively permit a paleochannel reach of steeper gradient, possibly controlled by basement erodibility contrasts or actual west-down faulting (Wakabayashi and Sawyer, 2000). If west-down late Cenozoic faulting is associated with the topographic steps in the Tuolumne River drainage, the maximum permissible vertical separation of paleochannels across the steps is much less than half of the height of any associated step. No significant offsets are observed in the reconstructed 10 Ma San Joaquin paleochannel of Huber (1981). In contrast, thermochronologic and structural data presented by Maheo et al. (2009) showed that some of the major topographic escarpments of the Kern River drainage and vicinity are late Cenozoic fault scarps, rather than pre-Neogene erosional escarpments.

Paleorelief defines paleochannels that crossed the Sierra prior to filling of these channels with Miocene to Pliocene andesitic volcanic rocks. The deepest paleochannels are the oldest (Eocene) channels. Oligocene paleochannels appear to have largely followed the Eocene drainages (Henry, 2008; Henry and Faulds, 2010; Garside et al., 2005) (Fig. 2). At the crest, preserved paleochannel relief appears to be ∼200 m for the paleo–Feather River, 450 m for the middle branch of the paleo–Yuba River, and 450–600 m for the southern branch of the paleo–Yuba River (Henry, 2008; Henry and Faulds, 2010; and this analysis). This paleorelief associated with these paleochannels apparently continues to increase east of the crest. For example the southeast (or southernmost) tributary of the paleo–Yuba River (southeast tributary of Nine Hill paleovalley) has a paleorelief of 1200 in the vicinity of Yerington, ∼100 km east of the present Sierran Frontal faults, and the paleorelief decreases to ∼600 m directly east of the crest in the Tahoe Basin area (Henry and Faulds, 2010) (Fig. 2). Another major paleochannel, the Golconda Canyon paleovalley of the Tobin Range of Nevada, exhibits over 1000 m of paleorelief ∼200 km east-northeast of the Frontal faults of the Mohawk Valley fault zone (Gonsior and Dilles, 2008) (northeast part of Fig. 2).

Paleorelief associated with paleochannels within the Sierra defines channels broader than the present-day canyons (Henry et al., 2012). For example, the Middle Fork San Joaquin paleocanyon exhibits ∼1 km of paleorelief with ∼30 km of distance between the paleocanyon rims. In contrast, the post–10 Ma incision deepened the canyon by ∼1 km with ∼5 km of width associated with this younger incision. In the northern Sierra, the deepest late Cenozoic canyon, the North Fork Feather River, exhibits 1.2 m of incision across a canyon 6 km wide. The Eocene paleovalleys of the northern Sierra appear to have been narrower than those of the south in the region near the present Sierran crest, although still broader than the late Cenozoic canyons. For example, the Nine Hill paleovalley is 7 km wide with ∼600 m of paleorelief near the crest in the present North Fork American River drainage (Henry and Faulds, 2010) and the 9-km-wide Soda Springs paleovalley shows ∼450 m of paleorelief near the crest in the headwaters of the South Fork Yuba River drainage (Sylvester et al., 2007) (Fig. 4).

Further west of the crest, near the region of maximum late Cenozoic incision, the northern Sierra paleovalleys appear somewhat broader. For example, the Eocene South Fork Yuba paleochannel (downstream of the confluence of the Nine Hill and Soda Springs paleovalleys), presently located in the Middle Fork American River drainage, exhibits about ∼450 m of paleorelief across a 20-km-wide paleovalley 25 km west of the crest, and 200 m of paleorelief across a paleocanyon of the same width 50 km west of the crest. The paleo–Yuba River (downstream of the confluence of the major paleovalleys) exhibits ∼180–280 m of paleorelief across a 20-km-wide paleovalley. Within the broader paleovalleys, narrower inner paleochannels are also observed (Garside et al., 2005).

Incision rates represent the erosion rate along the bottom of a stream valley. As noted above, studies have determined low erosion rates for the upland surfaces of the Sierra Nevada on scales of the last tens to hundreds of thousands of years and from ca. 12 Ma to the present. Longer-term erosion rates can be obtained from barometry of the youngest plutons, thermochronology of the basement, and sedimentary overlap relationships, with the assumption that exhumation recorded by these data equals erosional denudation. Such an assumption holds if reverse and thrust faulting drove exhumation, but not normal faulting; in the latter case exhumation can greatly exceed erosional denudation (e.g., Platt, 1986). During the main stage of exhumation of the basement (pre-Eocene), the dominant mode of dip-slip deformation north of the Kern River drainage is likely to have been reverse faulting within a transpressional regime (e.g., Renne et al., 1993; Tobisch et al., 1995; Tikoff and de Saint Blanquat, 1997; Pachell et al., 2003), with comparatively minor additional exhumation in the late Cenozoic (≤1 km associated with the deepest canyons as noted above). For the Kern River drainage and environs, exhumation of the batholith, from depths of up to 30 km, was initially associated with erosion (and presumed shortening), followed by late Cretaceous normal faulting associated with extension, so exhumation vastly exceeded erosion for that region for a significant part of the exhumation path (Saleeby et al., 2007; Chapman et al., 2012).

Late Cenozoic exhumation of ranges east of the Sierra Nevada in the Basin and Range/Walker Lane significantly exceeds (by several kilometers) late Cenozoic exhumation of the Sierra Nevada, and appears to reflect primarily tectonic denudation by normal faults, rather than erosion (Stockli et al., 2002, 2003). The dip-slip component of late Cenozoic faulting along the eastern margin of and in the interior of the Sierra is normal, so Cenozoic exhumation rates within the Sierra probably exceed erosion rates. Accordingly, the Cenozoic exhumation rates shown on Figure 5 should be considered maximum erosion rates.

Comparing the crystallization depth of the youngest plutons (Ague and Brimhall, 1988) overlapped by Eocene deposits with their age, Wakabayashi and Sawyer (2001) estimated a long-term averaged exhumation rate, equated to erosion rate on the basis of transpressional tectonics that prevailed at the time (see above), of 0.26–0.35 mm/yr between 100 and 57 Ma (Fig. 5). Clark et al. (2005) showed a strong linear relationship between apatite (U-Th)/He age (closure temperature ∼65 °C) and elevation relative to a regional erosion surface for samples from the Tuolumne to northern Kern River drainages, and calculated an exhumation rate of 0.04 mm/yr, integrated over canyons and interfluves, from ca. 80 Ma to 20 Ma (younger age limit revised from 32 to 20 Ma following Saleeby et al., 2009) (Fig. 5). Using samples from the Yuba River drainage, Cecil et al. (2006) estimated a post–ca. 65 Ma exhumation rate of 0.02–0.04 mm/yr from apatite (U-Th)/He ages, and ca. 90–65 Ma exhumation rates of 0.2–0.8 mm/yr based on comparison of zircon (closure temperature modeled as 173 °C) and apatite (U-Th)/He ages. Apatite (U-Th)/He ages from the southernmost Sierra, related to sample elevations corrected for late Cenozoic faulting, suggest exhumation rates of 0.06 mm/yr after ca. 80 Ma (Maheo et al., 2009).

A comparison of pluton crystallization ages (U-Pb, zircon) and barometry, overlap relationships of late Cretaceous strata in the western Central Valley, hornblende and biotite Ar-Ar and K-Ar cooling ages, and zircon and apatite (U-Th)/He ages show that the southern Kern and adjacent drainages experienced rapid exhumation at rates of ∼2 mm/yr between ca. 90 and 80 Ma (Saleeby et al., 2010; Chapman et al., 2012) (Fig. 5). The northern Sierra Nevada may have also experienced a similar history of rapid late Cretaceous exhumation of short duration, so that the interpreted 65–90 Ma period of high exhumation rates (Cecil et al., 2006) may have been briefer, with higher rates, but existing thermochronologic data are not sufficient to confirm this.

All areas of the Sierra appear to have experienced low exhumation and erosion rates of 0.06 mm/yr or less from the late Cretaceous to the onset of late Cenozoic incision. Exhumation rates for this period appear to have been slightly higher in the southernmost Sierra (0.06 mm/yr) compared to the northern Sierra (0.02–0.04 mm/yr) (Maheo et al., 2009; Cecil et al., 2006). The greater amount of erosion below the presumed early Cenozoic erosion surface in the southernmost Sierra appears to reflect this (Chapman et al., 2012). In addition, the Kern River drainage appears to be associated with higher slip rates and cumulative vertical separation of intra-Sierran faults, compared to other parts of the Sierra (Maheo et al., 2009; Nadin and Saleeby, 2010; Amos et al., 2010).

McPhillips and Brandon (2012) presented a thermo-kinematic model, incorporating published hornblende barometry from granitic plutons, apatite fission-track and apatite (U-Th)/He ages from the Merced River to the Kern River drainages, and physical properties of the Sierran crust. Their results show a decrease in surface elevation from ca. 100 Ma to the lowest paleoelevations in mid- to late Cenozoic time, followed by onset of surface and rock uplift and associated stream incision between 30 and 10 Ma.

Wakabayashi and Sawyer (2001) compiled stratigraphic thickness versus age from published studies in the Great Valley and postulated that the accumulation rate recorded there was approximately representative of the erosion rates in the Sierra Nevada region that constituted the main source for those clastic sedimentary rocks (Fig. 5). In the years since 2001, thermochronologic studies have allowed a direct assessment of erosion rates within the Sierra Nevada as reviewed above, so the Great Valley sedimentation rates are not needed to estimate Sierran erosion through time. However, the Great Valley deposition rates give insight into the total sediment load carried by Sierran streams, whose drainage basins included significant area east (upstream) of the Sierra. The analysis of sediment load of Sierran streams is relevant to assessing aspects of stream power connected with the incision history of these streams.

Wakabayashi and Sawyer (2001) acknowledged that some sediment carried by trans-Sierran streams bypassed the forearc basin (Great Valley) and reached the trench (accreted as part of the Franciscan subduction complex or completely subducted), but proposed that the Great Valley deposition rates were proportional to the erosion rates in the drainage basins of those streams. Great Valley sedimentary rocks record high Paleocene and earlier deposition rates and low Eocene to Miocene rates, consistent with the estimates of Sierran exhumation and erosion rates presented above, followed by a notable increase in deposition rates in the Pliocene that coincided with late Cenozoic incision initiation (Fig. 5). The Franciscan subduction complex records significant frontal accretion in Eocene time, represented by the accretion of the Coastal Belt (Dumitru et al., 2013). Although this may suggest a significant influx of detritus from Sierran streams that bypassed the forearc basin and reached the trench, detrital zircon ages and regional stratigraphic correlations indicate that the bulk of this clastic sediment originated from well north of the Sierran proper and was not transported to the trench-forearc basin region by trans-Sierran streams (Dumitru et al., 2013). Note that although most of the Coastal Belt is located north of the San Francisco Bay, restoration of dextral San Andreas fault system slip (∼200–250 km, depending on position of specific Coastal Belt exposures) restores these exposures southward (Wakabayashi, 1999) so that they are directly west of the Carquinez Strait in Eocene time, the approximate position of the submarine canyons that drained the northern Great Valley forearc basin to the trench during Eocene and later time (e.g., Dickinson et al., 1979). Geochemical similarities of Franciscan and coeval Great Valley Group rocks also appear to point to a common source of trench and forearc-basin sediments during the subduction regime (Ghatak et al., 2011).

Late Cenozoic faulting on the margins of and within the Sierra Nevada block provides information on the potential linkage of tectonics and topographic evolution within the range. Within this context, I will review information on the nature and evolution of the Frontal fault system (FFS), deformation within the Sierran block, and spatial-temporal comparisons between fault activity and incision.

In this this section, I will focus on: (1) the amount of late Cenozoic vertical separation across the FFS, with emphasis on its western strands, (2) the timing of faulting, and (3) the long-term evolution of this system, whereas I will not review some of the details of the physical characteristics of the FFS along strike presented in Wakabayashi and Sawyer (2001). East-down vertical separation of Mehrten Formation equivalents along the FFS, in the northernmost Sierra along the Mohawk Valley fault zone and related faults in the Feather River drainage, ranges between 600 m and 1000 m, and may reach 1500 m along the western margin of the Lake Tahoe Basin, apparently greater than that along the western strands of the FFS to the north or south of this basin (Wakabayashi and Sawyer, 2001). In the headwaters of the Stanislaus River drainage, the vertical separation of the ca. 9 Ma Eureka Valley tuff is ∼1100 m (Noble et al., 1974; Slemmons, 1966). At the headwaters of the San Joaquin River a 2.2–3.6 Ma volcanic unit is vertically separated by ∼980 m across the Frontal faults in areas that do not appear to have additional subsidence related to the Long Valley caldera (Wakabayashi and Sawyer, 2001, interpreted from Bailey, 1989).

Whereas the topographic escarpment along the southern Sierra Nevada reaches 3.3 km in height, the range is capped by granitic basement rather than late Cenozoic volcanic rocks, and Owens Valley has 1.1 or 2.1 km of late Cenozoic basin fill according to different interpretations of gravity data (Pakiser et al., 1964, and Bachman, 1978, respectively). The total height of the bedrock escarpment in Owens Valley is thus 4.4–5.4 km, but the lack of Cenozoic deposits atop the crest precludes a straightforward estimate of fault separation and timing. Jayko (2009) correlated erosion surfaces in the Sierra and White Mountains and used them to estimate Cenozoic vertical separation across FFS and other Walker Lane fault strands to the east. Because erosion surfaces occur at or near the summit of high peaks on the crest of the range, the height of the escarpment may be directly tied to vertical separation on the FFS (∼3 km plus the buried part of the escarpment in Owens Valley). Warping of an 11.7 Ma basalt flow across Frontal fault deformation associated with the Coyote warp near Bishop records ∼1600 m of vertical separation, and consideration of stratigraphy in the northern Owens Valley fill indicates at least 600 m more vertical separation valleyward of the lowest basalt outcrop for a total >2200 m of post–11.7 Ma vertical separation (Phillips et al., 2011).

Many of the late Cenozoic volcanic rocks of the central and northern Sierra Nevada had sources east of the FFS (e.g., Durrell, 1966; Slemmons, 1966), suggesting that Frontal faulting did not begin until after the eruption of these deposits; the flows could not have flowed uphill across the fault scarps. Consistent with this relationship, the major river drainages of the range, excluding the Feather River, such as those of the San Joaquin (Matthes, 1930; Huber, 1981), Stanislaus, and Yuba Rivers (Lindgren, 1911), have been beheaded by movement along the Frontal faults. The relationship of Frontal faulting to beheaded drainages and the distribution of volcanic rocks suggests that the FFS and Walker Lane belt have encroached westward in the late Cenozoic (Slemmons et al., 1979; Dilles and Gans, 1995; Jones et al., 2004). The timing of initiation of dextral faulting along some parts of the fault system may have differed from the initiation of significant dip-slip faulting alone (e.g., Cashman et al., 2009). In the summary below, I will focus on evidence for timing of dip-slip (normal) faulting because this faulting directly bears on landscape evolution through development of topographic escarpments, base level changes for streams, and beheading of streams.

In the Feather River area, movement on the Honey Lake fault zone initiated between ca. 3 and 6 Ma, based on regional kinematic considerations and the ages of basinal deposits along the northern Walker Lane (Faulds et al., 2005; Henry et al., 2007; Hinz et al., 2009), whereas movement on the present FFS in the Mohawk Valley area can be constrained only as post-Mehrten or post–ca. 5 Ma, based on identical offsets of the 16 Ma Lovejoy Basalt and overlying Mehrten Formation (Wakabayashi and Sawyer, 2001). Movement on some of the most significant faults of the Frontal fault system crossing the North Fork Feather River canyon probably did not begin until after 600 ka (Wakabayashi and Sawyer, 2000). Thus, the western margin of the Walker Lane belt appears to have stepped 50 km westward from the Honey Lake area to the present eastern escarpment of the northern Sierra Nevada within the last 5 m.y., and has encroached into the northernmost part of the range since the mid- to late Quaternary.

Westward encroachment of the western Walker Lane belt margin occurred along much of the tectonic boundary. The western edge of the central Walker Lane belt from 38°N to 39°N progressively stepped 100 km westward from 15 to 7 Ma, based on detailed structural, stratigraphic, and geochronologic studies (Dilles and Gans, 1995). East of the Lake Tahoe area, major east-down faulting began before 10.3 Ma in the Verdi Basin, based on ages of syntectonic basinal sediments (Henry and Perkins, 2001), then encroached ∼20 km westward to the Boca Basin at ca. 2.7 Ma (Mass et al., 2009). Slemmons et al. (1979) also suggested westward encroachment of the Walker Lane into the Sierran block in the late Cenozoic, with somewhat broader age constraints. Based on published analysis of volcanic deposits and gravels (Huber, 1981; Bailey, 1989), Wakabayashi and Sawyer (2001) proposed that the Frontal fault system encroached at least 40 km westward after 3 Ma to its present position in the region of the headwaters of the Middle Fork San Joaquin River.

There are exceptions to the progressive westward encroachment of faulting into the Sierran block. For example, significant faulting occurred at ca. 10 Ma in the Sierran crestal region at the headwaters of the Mokelumne and Stanislaus Rivers (Busby et al., 2008a), long before many of the westward jumps reviewed above.

Bachman (1978) suggested that the Sierran escarpment in the Owens Valley area did not form until 2.3–3.4 Ma. Bachman’s (1978) data and observations constrain timing of the uplift of the White Mountains (the range east of Owens Valley), but do not directly constrain movement on the FFS. Jayko (2009) proposed that the FFS in the Owens Valley area began movement ca. 10 Ma. Based on the paleodrainage history across the range crest in the Bishop area (northern Owens Valley), and the position of erosion surfaces covered by volcanic rocks of 11.7–11.8 Ma and 3.4 Ma age, Phillips et al. (2011) concluded that most of the vertical separation along that part of the FFS took place after 3.4 Ma. Zircon and apatite (U-Th)/He ages from a vertical transect on the eastern escarpment 20 km south of Mount Whitney, the highest point (4419 m) on the Sierran crest, show that movement on the FFS in that area did not begin until after 11 Ma (Maheo et al., 2004). An investigation of the El Paso Basin, east of the southernmost Sierra and south of Owens Valley, indicates that drainage eastward from a rising Sierra began by 8 Ma, suggesting the beginning of east-down Frontal faulting by that time at that latitude (Loomis and Burbank, 1988).

Initiation of FFS movement did not coincide with initiation of incision along much of the length of the Sierra. For example, in the Feather River drainage, incision may have begun at ca. 3 Ma, whereas many of the western strands of the Frontal faults may not have begun movement until after 0.6 Ma. Faulting and incision may have begun in the North Fork American headwaters at ca. 3 Ma, whereas faulting began at ca. 10 Ma in the Sonora Pass (Stanislaus River headwaters), but incision apparently initiated at 4–3.6 Ma. In the San Joaquin River drainage, incision began between 10 and 6 Ma, whereas the Frontal faults appeared to have initiated at ca. 3 Ma. In the Kern to Kings drainages, the first episode of late Cenozoic incision began at 20 Ma, but Frontal faulting may not have started until ca. 8–11 Ma along southern Owens Valley, and until ca. 3.5 Ma along northern Owens Valley.

Late Cenozoic internal deformation of the Sierra Nevada, recorded by faulting and local tilting of late Cenozoic deposits, is minor compared with faulting along the FFS (e.g., Lindgren, 1911; Christensen, 1966; Bateman and Wahrhaftig, 1966) excluding the Kern River area and the northernmost part of the range. Intra-Sierran faults have slip rates of hundredths of a millimeter per year or less, and vertical separation of less than ∼40 m, except in the areas near and directly west of the crest (Wakabayashi and Sawyer, 2000). Internal deformation is distributed fairly evenly across the range, with more closely spaced faults and faults with greater displacements (up to ∼200 m on some faults) in the area directly west of the crest (see detailed discussions in Wakabayashi and Sawyer, 2000). The internal faulting near the crest appears to be related to en echelon east-down Frontal faults that cross the crest (Wakabayashi and Sawyer, 2000) (Fig. 2). Because of these faults downdropping the crest, projection of the tilts of the Lovejoy Basalt and Table Mountain Latite from the western margin of the Sierra Nevada to the crest appears to overestimate the actual crestal elevations of these strata at the crest by 315–365 m compared to their actual outcrop elevations. Similar deformation near the crest may be expected in much of the Sierra north of the Kings River headwaters, owing to the en echelon geometry of FFS strands (Wakabayashi and Sawyer, 2001).

A west-down warp and/or zone of distributed faulting, known as the Chico monocline, deforms Pliocene (Yana volcanic center/Tuscan Formation) volcanic rocks along the western margin of the northernmost part of the range, north of the Feather River (Harwood and Helley, 1987). Internal faulting east of the Chico monocline has higher slip rates (tenths of a millimeter per year or greater) than in regions to the south and is part of a broad fault zone that forms the northern margin of the Sierra Nevada microplate (Sawyer, 2009, 2010). The Sierra of the Kern River drainage and adjacent areas to the west is cut by multiple late Cenozoic faults, some exhibiting higher (tenths of a millimeter per year) slip rates, and larger cumulative late Cenozoic vertical separation (1 km or more) than internal faults in any other part of the microplate excluding the northern boundary zone (Maheo et al., 2009; Amos et al., 2010; Nadin and Saleeby, 2010; Brossy et al. 2012).

A comparison of the gradients of Cenozoic paleochannel deposits and modern Sierran rivers gives insight into the landscape evolution of the mountain range. Lindgren (1911) examined the present gradients of the thalwegs of paleochannels of the gold-bearing Eocene gravels and showed a systematic relationship between the azimuth of the paleochannel reach and the gradient. The steepest reaches have azimuths at right angles to the axis of the range, whereas the lowest-gradient reaches have azimuths parallel to the axis of the range (Fig. 6A). Hudson (1955) and Jones et al. (2004) refined and reaffirmed this relationship (Fig. 6B). The variation of azimuth of paleochannel reaches does not correspond to differences in basement geology when compared to general compilations of basement geology, although the long (∼55 km) north-trending section of the paleo–South Yuba River that records the widest range of paleochannel azimuth variations has an average trend parallel to the north-trending structural grain within primarily metasedimentary rocks (e.g., Saucedo and Wagner, 1992).

Some of the Eocene paleochannel reaches along the paleo–South Yuba (location of this section of paleochannel noted on Fig. 2) have negative gradients that gain elevation in the paleo-downstream direction (Lindgren, 1911; Hudson, 1955; Yeend, 1974; Jones et al., 2004). Hudson (1955) confirmed the paleoflow direction in these reaches by observation of pebble imbrication and cross-bedding orientations. These reaches range in length from 2.4 to 11.5 km (consecutive uphill reaches combined) (Hudson, 1955). The most significant uphill reach gains 66 m over a distance of 11.5 km, and the second-most significant uphill reach gains 44 m over 2.4 km in the paleo-downstream direction (Hudson, 1955). All of these uphill reaches have azimuths trending more easterly than the axis of the range (Figs. 6A, 6B).

The systematic relationship between steepness and azimuth exhibited by the Eocene paleochannels starkly contrasts with modern Sierran rivers, which show no correspondence between azimuth and gradient for reaches of approximately equivalent length (1.4–9.1 km) to those examined for the Eocene paleochannels, including reaches that trend easterly of the axis of the range (Fig. 6C). However, the lack of correspondence between bedrock lithology and azimuth is common to both paleochannels and modern rivers. The North Fork American River and Tuolumne River reaches plotted in Figure 6C have incised into phyllites, with a strong structural grain imposed by the steep foliation and steep lithologic contacts (Wagner et al., 1990). In contrast, the San Joaquin River reaches with the wide range of azimuths plotted in Figure 6C cut primarily granitic basement.

Many lines of evidence and interpretation bear on the debate over whether late Cenozoic uplift of the Sierra occurred or not. Much of the evidence and interpretations accommodate either model, whereas some more clearly support one position at the exclusion of the other. Before discussing the issue of late Cenozoic Sierran uplift in detail, I will briefly list the evidence and interpretations that support both competing interpretations rather than supporting or challenging one at the exclusion of the other.

High paleoelevations east of the present Sierra Nevada, as exemplified by the Nevadaplano concept (e.g., Christiansen and Yeats, 1992; Dilek and Moores, 1999; Henry, 2008) do not contradict late Cenozoic uplift of the Sierra Nevada, contrary to some published statements, such as those of Wolfe et al. (1997, 1998) who postulated a post-Miocene decrease in Sierran elevations based on high interpreted paleoelevations for samples collected east of the range. The various lines of evidence and resultant interpretations merely show that the area east of the Sierra was higher than the Sierra in Eocene to Miocene time, so that streams flowed westward from what is now Nevada across what has become the Sierra Nevada to the Pacific Ocean (Henry, 2008; Henry and Faulds, 2010; Henry et al., 2012). Such models and observations do not preclude post-Miocene uplift of the Sierra Nevada, and the concept of higher elevations east of the Sierra Nevada was accepted by advocates of late Cenozoic uplift who agreed that pre–Mio-Pliocene streams flowed westward across the area that later became the Sierra (e.g., Huber, 1981; Wakabayashi and Sawyer, 2001).

(U-Th)/He apatite ages within most of the Sierra Nevada exceed 32 Ma, indicating less than ∼2–4 km of exhumation since then (House et al., 2001; Clark et al., 2005; Cecil et al., 2006). The maximum amount of late Cenozoic exhumation is recorded by the deepest stream incision of ∼1 km. Thus, the lack of late Cenozoic exhumation ages do not preclude the small magnitude of late Cenozoic rock uplift and incision that has been proposed (∼1–2 km). Wakabayashi and Sawyer (2001) and Clark et al. (2005) presented evidence for paleorelief of up to 1–2 km in the highest part of the Sierra Nevada, and proposed moderately high paleoelevations in the southern Sierra. Such paleorelief and paleoelevations of the southern Sierra do not preclude 1–2 km of subsequent rock and surface uplift (Wakabayashi and Sawyer, 2001; Clark et al., 2005), although estimates of large amounts of paleorelief have been used to argue against late Cenozoic uplift (e.g., House et al., 1998).

An apparent post–Oligo-Miocene decrease in mean elevation east of the Sierra Nevada has been proposed on the basis of paleobotanical interpretations (e.g., Wolfe et al., 1997, 1998), stable isotopic data (Horton and Chamberlain, 2006), and evidence of significant extensional collapse (thinning) of the lithosphere in this region (Wernicke et al., 1996). As noted by Wakabayashi and Sawyer (2001), this does not contradict interpretation of post-Miocene uplift in the Sierra to the west. Although the late Cenozoic faults of the FFS show dip-slip displacements that reach or exceed 3 km with the range side up, this does not demonstrate or preclude late Cenozoic uplift because the vertical movement of the Sierran side relative to sea level cannot be directly determined from the displacement alone.

The proposal of a rain shadow in the western Basin and Range at least as far back in time as 16 Ma, based on stable isotopic data from authigenic minerals (Poage and Chamberlain, 2002), also does not refute late Cenozoic Sierran uplift because (1) all but one of the samples are taken from positions east of at least one set of normal faults that are themselves east of FFS (the eastern boundary of the Sierran microplate predating late Cenozoic uplift was east of the present FFS as noted earlier), and (2) proponents of uplift do not dispute the possibility of a relatively high southern Sierra Nevada at 16 Ma (e.g., Wakabayashi and Sawyer, 2001; Clark et al., 2005; Saleeby et al., 2009).

The western margin of the Sierra Nevada (eastern margin of the Central Valley) from about the San Joaquin River northward has maintained approximately the same elevation from early to mid-Cenozoic time, a view endorsed by those both in favor and against the premise of late Cenozoic uplift in the Sierra (compare Cassel et al., 2012, to Wakabayashi and Sawyer, 2001, for original sources). Accordingly, if late Cenozoic rock and surface uplift of the Sierra has not taken place, then the original gradients of the Eocene paleochannels are recorded by the modern-day exposures, whereas if late Cenozoic uplift took place, the associated westward tilting would have modified the original gradients with the degree of modification dependent on the azimuth of the paleochannel. The systematic relationship between paleochannel azimuth and gradient, with the lowest-gradient reaches parallel to the range axis and the steepest reaches perpendicular to the range axis, demonstrates westward tilting and rock uplift of the Sierra Nevada with the tilt axis parallel to the range axis (Lindgren, 1911; Hudson, 1955; Jones et al., 2004) (Figs. 6A, 6B). Most significantly, the uphill reaches of the paleochannels cannot reflect original gradients and must have resulted from post-depositional westward tilting. The tilting of the uphill paleochannel reaches is not a product of localized west-down tilting or faulting, for profiles of Cenozoic paleochannels in these drainages show no evidence of a locally tilted block or west-down faulting (Wakabayashi and Sawyer, 2000). Moreover, the inverse gradient of these reaches is consistent with their azimuth that trends east of the range axis (Figs. 6A, 6B).

As noted previously, there is no relationship between paleochannel or modern river azimuth and basement rock type, an explanation proposed by Cassel et al. (2012) for relationship of paleochannel gradient and azimuth (but not for inverse gradients). Cassel et al. (2012) pointed out that a paleochannel thalweg does not connect channel bottom points formed at the same time during a river’s history. However, azimuth-gradient relationships reviewed above do not depend on short-term time equivalence of different points on the paleothalweg. Rather, the relationship reflects the history after the formation and abandonment of the paleochannel. Detailed mapping and field inspection by previous workers (e.g., Lindgren, 1911; Hudson, 1955; Garside et al., 2005) demonstrates that segments with “uphill” gradients are a not consequence of filling of a paleochannel and spilling over a divide, for the paleochannel network is well constrained over these reaches. The latter scenario would also demand a fortuitous relationship between the spillover points and buried gullies with (northward) azimuths east of the range axis.

Whereas the amount of exhumation in the late Cenozoic (≤1.2 km) is too small to be recorded by apatite (U-Th)/He ages (House et al., 2001; Clark et al., 2005; Cecil et al., 2006), transverse apatite (U-Th)/He age gradients have been interpreted to indicate significant westward tilting of the range since 5 Ma (McPhillips and Brandon, 2010) and ⁴He/³He dates record late Cenozoic incision of the Kings River (Clark and Farley, 2007). Numerical modeling of multiple sets of thermochronologic data suggests ∼2 km of crestal uplift in the region from the Merced to Kern River drainages that began at 30–10 Ma (McPhillips and Brandon, 2012).

The late Cenozoic initiation of incision strongly supports late Cenozoic uplift as noted by Wakabayashi and Sawyer (2001), and I will update some of their arguments here. Stream power, which governs incision or aggradation, is a function of discharge, gradient, erodibility of the bedrock substrate, and the amount of sediment transport load (e.g., Sklar and Dietrich, 1998). During the history of incision of Sierran streams, the bedrock geology, and hence erodibility, in any one position of the stream can be regarded as constant, because it comprises either metamorphic rocks with steep lithologic contacts and foliation dips, or granitic rocks (e.g., Bateman and Wahrhaftig, 1966; Schweickert, 1981; Moores and Day, 1984; Sharp, 1988; Saleeby et al., 1989). Greater discharge and higher gradients raise the stream power and ability of the stream to incise, whereas a large sediment load above a threshold amount will retard the stream’s ability to incise (Sklar and Dietrich, 1998). Progressive westward encroachment of down-to-the-east faulting since the Oligocene, alluded to earlier, has progressively beheaded Sierran stream systems and reduced their drainage area. Climate from the Late Miocene to the present has not been wetter than earlier times and in fact, the climate that prevailed during the deposition of the Eocene gravels is postulated to be warmer and considerably wetter than Pliocene to recent climate (e.g, Zachos et al., 2001; Minnich, 2007). Accordingly the progressive reduction of drainage area indicates a corresponding reduction in stream discharge from Eocene to Pliocene time.

The record of erosion in the Sierra Nevada, and of deposition in the Great Valley that shows a significant increase in deposition rate in Late Miocene to Pliocene time, suggests an increase in sediment load in Sierran rivers at the time incision began. Accordingly discharge should have progressively decreased since Eocene–Oligocene time, and sediment load increased since the Pliocene. Both of these factors should reduce stream power, so only an increase in stream gradients could have caused the initiation of incision in Pliocene time, and only rock uplift and tilting could have increased the stream gradient.

In contrast to the interpretation summarized above, Cassel and Graham (2011) proposed high sediment loads for streams crossing the Sierra in the Oligocene and Eocene. Such a proposal would suggest that more sediment from the Sierran streams bypassed the Great Valley depocenter during Oligocene and Eocene time than before or after, so that the Great Valley depositional record as shown in Figure 5 poorly represents relative sediment delivery rates from Sierran streams at that time. However, the Oligocene sediment load, represented mostly by ash-flow tuff deposition in paleochannels (e.g., Henry and Faulds, 2010), pales in comparison to the much larger amount of andesitic material generated by the ancestral Cascades arc, and Miocene onset of deposition of these andesites is reflected by an increase in sedimentation rate in the Great Valley (Fig. 5). In spite of the large volumes of deposited andesitic material, incision between eruptive depositional pulses did not progress below the former thalweg channel positions, nor did basement incision take place during the 7 m.y. following the emplacement of the youngest Oligocene ash-flow tuffs and prior to initial eruption of Miocene andesitic rocks (23 Ma to 16 Ma) that saw virtually no deposition in the paleocanyons as well as low deposition rates in the Great Valley. Accordingly, initiation of late Cenozoic incision corresponds to an increase, rather than a decrease, in sediment load in streams, whereas multiple episodes of decreased sediment load following volcanic depositional pulses in Oligocene and Miocene time were not followed by basement incision (e.g., Busby et al., 2008b; Henry and Faulds, 2010).

The estimates for uplift presented on the basis of reconstructed marker horizons (e.g., Lindgren, 1911; Hudson, 1955; Christensen, 1966; Huber, 1981; Unruh, 1991; Wakabayashi and Sawyer, 2001; Jones et al., 2004) record rock rather than surface uplift, but, as pointed out by Wakabayashi and Sawyer (2001), the erosion rates on upland erosion surfaces are so low that rock uplift very closely approximates surface uplift for the highest elevations. In the years since 2001, other studies have published data showing low erosion rates on the upland erosion surfaces, ranging from rates averaged over tens to hundreds of thousands of years from cosmogenic nuclides (Stock et al., 2005), to Ar-Ar data that show extremely slow erosion rates averaged over the last 12 m.y. (Phillips et al., 2011). Collectively, the low erosion rates of the upland erosion surfaces, and the much faster incision rates of the streams, indicate an increase in relief in the late Cenozoic in addition to a significant increase in surface elevation.

In addition to peak surface elevation increase, significant mean elevation increase has occurred (including over 1 km of elevation increase for the bottoms of the deepest canyons) indicating that tectonics has driven most of the elevation increase and tilting (Wakabayashi and Sawyer, 2001; Jones et al., 2004; Stock et al., 2004) instead of isostatic uplift owing to erosion (e.g., Small and Anderson, 1995).

The initiation of significant stream incision from ca. 3 Ma in the Feather to American River drainages, to 3.6–4 Ma in the Mokelumne to Stanislaus River drainages, to 6–10 Ma in the San Joaquin River drainage, and 20 Ma in the Kings to Kern River drainages, significantly predates the onset of North American glaciation at ca. 2.6 Ma (Zachos et al., 2001) as well the onset for glaciation in the Sierra, for which the earliest preserved deposits may be ca. 1.5 Ma (Huber, 1981; reviewed in Clark et al., 2003). Thus, increased upland erosion rates associated with glaciation, and resultant isostatic uplift, did not trigger the late Cenozoic landscape rejuvenation, and tectonic uplift is the more likely cause.

Estimated rock and surface uplift exceeds incision throughout the range (Wakabayashi and Sawyer, 2001), as noted above, and it is useful to revisit the long-term rates of uplift owing to new age constraints on incision introduced herein. For the Feather River drainage, the uplift rate of the crest averaged since 3 Ma is ∼0.6 mm/yr, and for the Stanislaus River drainage it is ∼0.5–0.6 mm/yr averaged since 3.8 Ma, using uplift estimates (Wakabayashi and Sawyer, 2001) based on the Lovejoy Basalt and Table Mountain Latite, respectively.

The diachronous timing of incision initiation indicates diachronous initiation of late Cenozoic uplift in the Sierra. In addition to differences in timing of uplift, the magnitude of uplift may vary along strike from the San Joaquin River northward, but with no systematic pattern (Wakabayashi and Sawyer, 2001). For example, the estimates of crestal rock uplift for the Feather River and Stanislaus River areas (∼1700–1900 m and ∼1800–1900 m, respectively; Wakabayashi and Sawyer, 2001) and the San Joaquin River (2150 m; Huber, 1981) exceed that for the Yuba River–northern American River region (1200–1500 m; Jones et al., 2004). These differences are large enough that they probably exceed differences resulting from applying different methods or the uncertainties associated with those methods.

In summary, late Cenozoic rock and surface uplift of the Sierra Nevada is indicated by the azimuth-gradient relationships observed in the paleo–Yuba River and supported by the record of erosion and stream incision in the Sierra. Thermochronologic data also support the premise of late Cenozoic Sierran uplift (Clark et al., 2005; McPhillips and Brandon, 2010, 2012). GPS and interferometric synthetic aperture radar (InSAR) data show present-day uplift of the Sierra at rates of 1–2 mm/yr along the entire length of the range (Hammond et al., 2012), consistent with late Cenozoic (and ongoing) uplift of the range.

Whereas geologic evidence appears to demand late Cenozoic rock and surface uplift and westward tilting of paleochannels, the apparent incompatibility of the stable isotopic and botanical paleoaltimetry, and related interpretation of steep paleogradients, invites additional explanation. In this section, I will present alternative interpretations of the stable isotopic and paleobotany data as well as alternative explanations for sedimentologic data interpreted to show steep paleogradients of Oligocene to Eocene paleochannels. In making these arguments I will focus the discussion on works based on data from within the Sierra Nevada, instead of those that have based conclusions on samples east of the range. Accordingly, I will focus on Mulch et al. (2006) and Cassel et al. (2009, 2012) for stable isotopic paleoaltimetry, and Hren et al. (2010) for paleobotany.

Stable isotopic paleoaltimetry relies on two critical assumptions. The first assumption is that the samples record closed-system behavior since the time of interest, or since Oligocene or Eocene time in the case of the Sierran studies (Mulch et al., 2006; Cassel et al., 2009, 2012). These studies assume that volcanic glass in Oligocene rhyolites hydrated and equilibrated with surface water within 5 k.y. after eruption and there has been no isotopic exchange since then (Cassel et al., 2009, 2012), or that kaolinite formed after Eocene deposition in paleochannels and prior to covering by Oligocene rhyolites (Mulch et al., 2006). The second assumption is that the Oligocene climate and air circulation patterns in the region are known well enough that the isotopic ratios can be reliably related to temperature and hence surface elevations. Molnar (2010) challenged this assumption on the basis of the poorly known paths of air masses, and hence atmospheric water, in the Oligocene and Eocene, but regardless of the validity of the paleoclimate model, the uncertainty envelope given by Cassel et al. (2009, 2012) is broad enough to permissibly support the amounts of surface uplift proposed by proponents of late Cenozoic uplift (e.g., those of Unruh, 1991; Wakabayashi and Sawyer, 2001; Jones et al., 2004).

The preferred Oligocene paleoelevation profile of Cassel et al. (2009, 2012) progressively increases in elevation eastward to the location of the current Sierran crest, then levels off eastward, a pattern remarkably similar to present-day topography. The present-day topographic profile, however, has been strongly influenced by post-Oligocene faulting that has downdropped blocks east of the Sierran crest along east-down faults (e.g., Wakabayashi and Sawyer, 2000, 2001; Faulds et al., 2005; Henry et al., 2007; Faulds and Henry, 2008; Hinz et al., 2009). Restoration of these faults, and associated restoration of Basin and Range lithospheric thinning, should result in higher paleoelevations east of the Sierra rather than an eastward leveling-off of elevations. If the Oligocene topography consisted of an eastern plateau with a marked westward steepening at the present Sierran crest, then the Eocene–Oligocene paleochannels should display the highest paleorelief in the vicinity of the crest. In contrast, the paleorelief of these channels increases east of the crest (Henry and Faulds, 2010; Henry et al., 2012) as part of a progressive eastward increase in paleorelief reviewed earlier. Accordingly, the interpreted Oligocene paleoaltimetry profile appears inconsistent with the post-Oligocene faulting and extension history of the Basin and Range and the distribution of paleorelief in Oligocene paleochannels.

Because of significant post-Oligocene modification of topography by Basin and Range faulting, the close correspondence between the interpreted paleoaltimetry profile and modern topography suggests post-Oligocene isotopic exchange between surface water and hydrated volcanic glass. This brings into question the assumption of geologically instantaneous hydration and subsequent closed-system behavior used by Cassel et al. (2009, 2012), citing the work of Friedman et al. (1993), who proposed complete hydration of small glass shards within 5 k.y. (discussed further below). In fact, the progressive hydration (open-system behavior) of volcanic glass has been well established in numerous studies, to the extent that the thickness of hydration rinds on obsidian has been calibrated as a geochronometer, employed for dating of archaeologic artifacts (e.g., Friedman and Smith, 1960; Friedman and Long, 1976; Morgenstern and Riley, 1974; Rogers, 2010). Studies show progressive hydration of volcanic glass for samples at least as old as 15 Ma (Morgenstern and Riley, 1974).

Volcanic glass as old as 1.1 Ga has been found (Palmer et al., 1988), and volcanic glass from ophiolites at least as old as Jurassic has been coveted in studies of ophiolite petrogenesis, owing to the fact that the rarely preserved glass retains the original chemistry of the volcanic rock that has otherwise been altered to some degree (e.g, Robinson et al., 1983; Shervais and Hanan, 1989). The progressive hydration of volcanic glass has been demonstrated empirically by comparison of rind thicknesses and independent age dates on the rocks (e.g., Friedman and Smith, 1960; Morgenstern and Riley, 1974; Rogers, 2010), by experimental studies (e.g., Friedman and Long, 1976; Delaney and Karsten, 1981; Karsten et al., 1982; Zhang et al., 1991; Zhang and Behrens, 2000; Stevenson et al., 1998), and chemical kinetics (e.g., Doremus, 2000). Although Anovitz et al. (1999) challenged the nature of equations governing hydration rate of volcanic glass, they did not question the general principle of progressive hydration with time over geologically significant time scales.

Friedman et al. (1993) proposed that small glass shards in volcanic ash completely hydrate (and isotopically exchange with water) within 5 k.y. after eruption, a situation that differs from obsidian or glassy basalt that feature a vastly larger effective radius of glass domains in contact with water. The samples used for isotopic paleoaltimetry come from ignimbrites (Cassel et al., 2009, 2012) that cannot be compared to loose volcanic ash, owing to the degree of welding of glass shards that decreases the permeability of the material and increases the effective radius of the glass domains. Thus obsidian is a better analog of the hydration behavior of glass in the rhyolite ash-flow tuffs through time.

Even if part of the surface of an ignimbrite had originally completely hydrated and equilibrated with surface water soon after eruption, the slow advance of the weathering front through time into unweathered and unhydrated material, associated with progressive denudation, must be considered. Chemical weathering rate depends most strongly on erosion rate, and climate exerts but a weak influence (Riebe et al., 2001, 2004). Accordingly, the weathering front advances downward through rock as erosion/denudation proceeds. Even in the most slowly eroding areas on the Sierran interfluves, where the Oligocene rhyolites (and Eocene gravels) are preserved, erosion has removed a significant amount of material since the Oligocene. For example, the amount of denudation in the past 25 m.y. is 100 m extrapolating the 0.004 mm/yr erosion surface rates of Phillips et al. (2011) to 25 Ma, and 50–475 m extrapolating the 0.002–0.019 mm/yr erosion rates of upland surfaces determined by Small et al. (1997) and Stock et al. (2005). Thus, the weathering front began advancing downward into rock from a land surface >50 m above the present level of exposure >25 m.y. ago and may not have reached some or many of the rhyolite samples until comparatively recently.

In conclusion, the similarity of the preferred Oligocene paleoaltimetry profile of Cassel et al. (2009, 2012) to the modern topographic profile, appears incompatible with post-Oligocene Basin and Range faulting and extension, and suggests geologically recent reequilibration of the stable isotopic system of the hydrated glasses studied in their investigations. The latter conclusion is consistent with numerous studies demonstrating the open-system behavior of volcanic glass, as well as the general correspondence between denudation and chemical weathering rates.

The studies of Mulch et al. (2006) do not include samples (or modeling) east of the Sierra Nevada, so they cannot be examined in light of Basin and Range faulting as above. However, the uncertainties in paleoclimate models noted above equally plague interpretations of stable isotopic composition of kaolinite in the Eocene gravels, and the question of isotopic closure applies equally to kaolinite as to volcanic glass. Mulch et al. (2006) interpreted Eocene growth and isotopic closure of the kaolinites in the Eocene gravels on the basis of the less-intense weathering of the Oliogene rhyolites compared to the Eocene gravels, and on the assumption of rapid Eocene weathering owing to a warm and moist climate. However, the discussion above strongly suggests the possibility of isotopic reequilibration of the Oligocene rhyolites and the potential for later growth of kaolinite and isotopic exchange. Moreover, the comparison of the degree of weathering between the rhyolites and the gravels does not account for the vast difference in lithology, permeability, and consequent susceptibility of weathering of the two units. Finally, as noted above, recent studies, conducted within the Sierra and other localities in the world, have demonstrated that chemical weathering rates show minimal dependence on climate but strong dependence on erosion rates (Riebe et al., 2001, 2004). Thus the conclusions of high Eocene paleoaltitudes of Mulch et al. (2006) can be alternatively interpreted to reflect uncertainty in paleoclimatic models, open-system isotopic behavior, progressive advance of the weathering front associated with denudation, or a combination of these factors. Stable isotopic studies of authigenic smectite (e.g., Poage and Chamberlain, 2002), in addition to not being a definitive test of Sierran uplift owing to sample locations as noted above, are also subject to the uncertainties with respect to paleoclimatic models, open-system isotopic behavior, and the possibility of new authigenic minerals forming progressively downward in bedrock as the weathering front advances.

Hren et al. (2010) interpreted a high paleoaltitude of the Eocene Sierra Nevada based on hydrogen isotope ratios associated with organic compounds within leaf fossils associated with Eocene deposits in the northern Sierra Nevada. As for the paleo-isotopic studies of kaolinite and hydrated volcanic glass noted above, the same problems regarding paleoclimatic models and open- versus closed-system behavior apply. For the latter issue Hren et al. (2010) proposed that leaf compounds can be preserved in sedimentary rocks over geologic time scales without isotopic exchange, citing the work of Schimmelmann et al. (1999). The experimental study of Schimmelmann et al. (1999), however, which focused primarily on isotopic fingerprinting of the source rocks of oil, pointed out the complexities in the various mechanisms of isotopic exchange with organic material and a corresponding variability in preservation of original (pre–lithification/burial) isotopic ratios. The possibility of isotopic exchange in the fossil leaf materials with water during the weathering process associated with denudation should be considered. In addition, the streams in which the leaves were deposited were flowing in paleovalleys. There is some possibility of downhill/downstream movement of leaves both in the trunk stream and from the slopes bounding the paleochannel (e.g., Greenwood, 1991).

Interpretation of high paleoelevations of the Sierra based on stromatal density (Kouwenberg et al., 2007, 2010) are subject not to issues of isotopic closure, but of the microscopic structure of fossil leaves. The stromatal density should reflect a decline in CO₂ partial pressure with altitude but the significant variation in atmospheric CO₂ content through time poses complications for distinguishing the altitude from the climatic variation signal. In addition, the downslope/downstream transport of leaves, as noted above, is a possibility.

Cassel and Graham (2011) proposed that Oligocene and Eocene stream paleogradients of the paleo–Yuba River were steep, similar to the preserved gradients of the deposits, on the basis of sedimentologic features. As noted above, the relationship between azimuth and gradient of these paleochannels strongly refutes the notion of steep paleogradients and supports significant tilting (steepening) of the paleochannels in the late Cenozoic. Cassel and Graham (2011) based their interpretation of steep paleogradients on the mean grain size of the coarser paleochannel deposits (∼0.5 m), compared to a physical model relating channel slope to stream gradient presented by Paola and Mohrig (1996). Paola and Mohrig (1996), and Cassel and Graham (2011) in their interpretation derived from it, took the position that channel gradient is independent of discharge and drainage basin area, an assertion that contradicts research dating back over a century that has demonstrated a strong inverse relationship between channel gradient and drainage basin area (and therefore discharge) (e.g., Gilbert, 1877; Leopold and Maddock, 1953; Hack, 1957, 1973; Christensen, 1966; Flint, 1974). This fundamental relationship is best illustrated by the typical longitudinal profile of a stream that increases in gradient headward.

Hack (1957) showed a strong dependence of the mean grain size of bed load on both the gradient of the stream and its drainage area. Using the relationship of Hack (1957), a range of estimated drainage basin area of 30,000–55,000 km² corrected for Basin and Range extension based on Henry and Faulds (2010) for these reaches of the paleo–Yuba River, and a mean grain size of 0.5 m, results in extremely low paleogradients of ∼0.4–0.5 m/km (0.0004–0.0005), an order of magnitude lower than the estimates of Cassel and Graham (2011). Christensen (1966) used data by Leopold and Miller (1956) and the U.S. Geological Survey (1960) to estimate paleogradients based on drainage area alone. This relationship has more scatter than that of Hack (1957), possibly because the mean bed-load size is not accounted for. Nonetheless, application of the Christensen (1966) gradient–drainage area relationship results in an estimated paleogradient range of ∼0.4–1.8 m/km (0.0004–0.0018) incorporating the full range of scatter in the calibrating data. The interpreted braided nature of the Eocene paleochannels was used by Cassel and Graham (2011) to support steep paleogradients, but multiple studies show that the gradients for braided streams vary inversely with discharge (e.g., Leopold and Wolman, 1957; Ferguson, 1987). Accordingly, sedimentological data do not require steep paleogradients of Eocene and Oligocene rivers, whereas evidence for significant tilting of the paleochannels from azimuth versus gradient relationships precludes such steep paleogradients.

The previous sections have discussed the evidence supporting late Cenozoic uplift in the Sierra Nevada and the most significant geomorphic response to itstream incision. New, primarily geochronologic data require revision of the conclusions presented in Wakabayashi and Sawyer (2001), particularly on the timing and along-strike variability of uplift, and these revisions also indicate the need to revisit the tectonic models proposed for this uplift.

As presented above, the late Cenozoic incision resulted from rock and surface uplift, so the timing of incision can be taken as approximating the timing of uplift and base-level forcing. The timing of incision and uplift varies, with an older initiation time in the south than the north. Although delamination or foundering of a dense root of the Sierra has been proposed as a driving mechanism for uplift, the direct evidence for this event, from xenoliths in volcanic rocks, exists only from the San Joaquin River drainage southward (Ducea and Saleeby, 1996, 1998).

Late Cenozoic Sierran uplift appears to decrease to near zero near the southern edge of the subducted Gorda plate. This is shown by (1) the negligible incision beneath Plio-Pleistocene volcanic rocks of major streams north of the North Fork Feather River (Deer Creek and Mill Creek; Fig. 2) and (2) the related decrease in elevation of the top of basement from ∼2100 m on the high upland surface north of the North Fork Feather River, to ∼1100 m in an erosional window on Deer Creek, to below 900 m in Mill Creek (which has not incised to basement northwest along the trend of the basement high) (Lydon et al., 1960). Thus, the northward younging of uplift (incision initiation) may indicate a relationship between uplift and the northward-migrating southern edge of the subducted Gorda slab, tied to the Mendocino triple junction, as suggested by Crough and Thompson (1977).

The resolution of the two models (triple-junction-related and delamination-related) may come from the synthesis of thermochronologic, structural, and geomorphic evidence from the Kings to Kern River drainages, where a first uplift and incision event apparently occurred at ca. 20 Ma and a second one after ca. 3.5 Ma (Clark et al., 2005; Saleeby et al., 2009). The first event may have been triggered by slab window formation in the wake of the northward-migrating slab edge, whereas the second event may have been the response to delamination beneath the southern Sierra (Maheo et al., 2009).

Many streams of the Kern River drainage have two knickpoints (Clark et al., 2005), which appears to be characteristic of the Kings River drainage as well (Clark et al., 2005; Stock et al., 2005) and possibly the San Joaquin River (Huber, 1981), whereas northern and central Sierra streams appear to have a single knickpoint (Kemp, 2012). The two knickpoints found in many of the streams of the southern Sierra drainages appear to reflect two late Cenozoic incision events that may have been triggered by two episodes of uplift (Clark et al., 2005; Stock et al., 2005). The second incision event appears to be synchronous, within uncertainty in ages, in these southern drainages, from post–3.5 Ma in the Kern River drainage and post–3 Ma in the Kings River, to post–3.6 Ma in the San Joaquin River. In contrast, the older incision event is clearly earlier (beginning at ca. 20 Ma) in the Kern and Kings Rivers than it is in the San Joaquin, where the older incision event began between 6 and 10 Ma.

Thus, the geomorphology of the Kern to San Joaquin drainages appear to reflect two late Cenozoic uplift events, the first of which youngs northward and may have been a consequence of slab window development in the wake on the northward-migrating Mendocino triple junction as proposed by Maheo et al. (2009), whereas the second, which began at about the same time in these drainages, may have resulted from delamination (e.g., Ducea and Saleeby, 1996, 1998; Saleeby and Foster, 2004; Le Pourhiet et al., 2006). North of the San Joaquin River drainage, the geomorphology apparently reflects a single late Cenozoic uplift event that may be related to slab window development in the wake of the migrating triple junction.

Thompson and Parsons (2009) proposed that isostatic uplift of the Sierran crest, in response to footwall unloading by FFS normal faulting, may have driven late Cenozoic uplift in the range. Their model generated about the same amount of rock uplift as estimated by proponents of late Cenozoic uplift in the Sierra (1200–1300 m). If footwall unloading was the primary cause of late Cenozoic Sierran uplift, Sierran uplift should have taken place much earlier in the northern and central Sierra, in response to faulting along the western margin of the Basin and Range. Perhaps mantle upwelling, associated with slab window formation in the wake of the triple junction, acted to enhance footwall isostatic response to normal-fault unloading.

I conclude with a summary model for the topographic evolution of the Sierra Nevada since the Cretaceous (Fig. 7). Peak erosion rates in the Sierra and deposition rates in the Great Valley coincide with the final stages of the emplacement of the Sierra Nevada batholith from 100 to 85 Ma (Fig. 7) (Wakabayashi and Sawyer, 2001). Owing to the lack of preserved landscape surface markers, the topography of the late Cretaceous Sierra Nevada is poorly constrained, but it is unlikely that the crest of the range coincided with the current one. Thermobarometry of plutons suggests greater exhumation in the western part of the range compared to near the present crest (Ague and Brimhall, 1988; Saleeby, 2007; Saleeby et al., 2010; Chapman et al., 2012). The locus of exhumation progressed from west to east (Tobisch et al., 1995) prior to a significant late Cretaceous exhumation event in the southernmost Sierra (Saleeby, 2012; Chapman et al., 2012). This may have led to eastward migration of the crest of the range. The location of the Cretaceous magmatic arc and the batholith does not correspond to the boundaries of the present mountain range. In the south, the Sierra Nevada batholith includes many early Cretaceous plutons beneath Great Valley sediments west of the range (Saleeby, 2007; Saleeby et al., 2010) as well as some plutons east of the range, and north of Lake Tahoe the batholith strikes more northerly than the range so that the majority of the batholith diverges eastward from the mountain range (Van Buer et al., 2009).

At ca. 84 Ma, arc magmatism shut off in the vicinity of the present Sierra Nevada, apparently in response to a low-angle subduction that resulted in greater coupling between the subducting plate and North America, inducing crustal thickening and increase of elevation to the east as part of the Laramide orogeny (e.g., Dickinson and Snyder, 1978). Exhumation rates slowed dramatically in the Sierra after ca. 80 Ma, but sedimentation rates in the Great Valley remained relatively high until the end of Paleocene time (Fig. 5), possibly as a consequence of sediments sourced further inland and/or as a result of continued headward erosional propagation within the Sierran region after rock uplift ceased (Fig. 7). The beginning of the Eocene brought a dramatic decrease in sedimentary accumulation in the forearc basin. The paleorelief preserved in the modern Sierra dates to this time and earlier. This paleorelief (relief at the time of formation; paleorelief as preserved today) was progressively reduced from late Cretaceous time (McPhillips and Brandon, 2012).

From the Stanislaus River drainage northward, what became the Sierra comprised the western flank of a broad upland region, whose drainage divide lay ∼300 km east of the present range crest, restoring subsequent extension of the Basin and Range (Henry, 2008; Henry and Faulds, 2010). Streams flowed westward from this divide and deposited the Eocene gravels in broad paleovalleys that deepened upstream and east of the present Sierra (Garside et al., 2005; Henry, 2008; Henry and Faulds, 2010; Henry et al. 2012). The largest amount of paleorelief north of the Stanislaus River (600 m) was associated with one of these paleovalleys at the position of the present crest; this paleorelief increases further upstream east of the present range. Much of the paleorelief in the crest region in this region is 200 m or less. Eocene paleoelevations north of the Stanislaus River, estimated by restoring late Cenozoic uplift and tilting, ranged from ∼800 to 1400 m at the present crest (Wakabayashi and Sawyer, 2001). Erosion of upland surface regions has occurred since Eocene time, and the amount of erosion may be significant, given the comparatively large amount of elapsed time (∼50 m.y.). For example, if the 0.004 mm/yr erosion surface lowering rate of Phillips et al. (2011) is restored (to 50 Ma), the estimated paleoelevations of the crest summit regions of the north Sierra in Eocene time becomes 1000–1600 m. Elevations to the east of the Sierra should have been significantly higher, for the area that became the north Sierra represented the western flank of an uplifted region to the east (e.g., Henry, 2008; 2012).

South of the Stanislaus River drainage, the basement surface rises, and the paleorelief increases in corresponding fashion to 1.5 km from the southern San Joaquin River drainage to the northern Kern River drainage; the basement surface elevation and paleorelief decreases southward in the southern Kern River drainage (Wakabayashi and Sawyer, 2001; Clark et al., 2005; Saleeby et al., 2009; Chapman et al., 2012). Eocene paleoelevations in the highest part of the range have been estimated at 2000–2500 m, based on restoration of late Cenozoic uplift extrapolated southward from the San Joaquin River drainage (Wakabayashi and Sawyer, 2001) and analysis of geomorphology and thermochronology (Clark et al., 2005). However, both of these estimates are subject to significant uncertainty. The difference in internal deformation and timing of uplift from the Kern River to the San Joaquin River drainages, as well as the along-strike variability in uplift magnitude, renders the extrapolation of late Cenozoic uplift estimates southward from the latter drainage dubious. In addition, the presence of significant dip-slip internal faulting in the Kern River drainage (e.g., Maheo et al., 2009; Amos et al., 2010; Nadin and Saleeby, 2010) complicates estimates of uplift from geomorphology and thermochronology in the Kern River drainage. Saleeby et al. (2009) proposed paleoelevations exceeding 3000 m for the southern Sierra, based on restoration of estimated late Cenozoic uplift that accounts for the internal complexity in this region.

There is no direct evidence of paleochannels crossing the Sierra south of the San Joaquin River, although such evidence may have been eroded away (Henry et al., 2012). This may indicate a late Cretaceous to Eocene drainage divide near the position of the present crest of the southern Sierra (Fig. 7), as proposed by Saleeby et al. (2009).

The difference in paleorelief and present surface elevations between north and south may reflect along-strike differences in the nature of the lithosphere. These differences may reflect the divergence of the modern range from the Sierra Nevada batholith, and the position of the (pre-Paleozoic) margin of the North American continent, as marked by the 0.706 initial ⁸⁷Sr/⁸⁶Sr isopleth (Kistler and Peterman, 1973). The latter coincides with the position of basement elevation increase; continental initial ⁸⁷Sr/⁸⁶Sr ratios of >0.706 are found south of the increase in basement surface elevation. The position of the old continental margin may have also affected the subsequent position of the Miocene arc whose axis was located along or just east of the present crest of the range from the Stanislaus River drainage northward, but was much further east of the range to the south (Christiansen and Yeats, 1992; Busby et al., 2008a, 2008b; Busby and Putirka, 2009; Busby, 2012). The position of the old continental margin may have also influenced the petrogenesis and resultant chemistry of Miocene volcanic rocks erupted across this boundary (Putirka and Busby, 2007).

The unique aspects of the southernmost Sierra, with far greater internal deformation than any other part of the Sierra, may have been influenced by a late Cretaceous extensional event that affected only that part of the Sierra (e.g., Maheo et al., 2009). Similarly, the northern termination of the Sierra Nevada at the Sierran-Cascade boundary zone, characterized by active faults that strike west-northwest (Sawyer, 2009, 2010), appears controlled by the basement structural grain that curves to a west-northwest strike and may be related to an earliest Cretaceous or older transform boundary between the Sierran and Klamath Mountain regions (Dilek and Moores, 1992; Ernst, 2012). There is no evidence of older, analogous, late Cenozoic structures in the Sierra Nevada south of the present boundary zone.

The progressively steeper dips of Quaternary to Eocene strata along the eastern margin of the Central Valley (Unruh, 1991) appears to be a feature localized along the margin of the Central Valley fill, because Cenozoic strata overlie each other in normal stratigraphic order without insetting from the lowest incised reaches of rivers to the crest, from the Feather River to at least as far south as the Stanislaus River drainage (Wakabayashi and Sawyer, 2001). This suggests long-lived west-side-down warping along the eastern margin of the Central Valley from the northernmost part of the range to at least as far south as the San Joaquin River drainage. The position of this warp along the eastern margin of the Central Valley basin fill suggests that this warp may have resulted from flexural response to sediment loading, with localization of deformation related to a zone of weakness within the underlying basement.

In Oligocene time, the period of low erosion rates and a stable landscape continued in the Sierra, as reflected by geologic relationships and estimated erosion rates. Ash-flow tuffs issued from calderas well east of the present crest and flowed down the drainages already partly filled with Eocene gravels (Henry, 2008; Henry and Faulds, 2010; Henry et al., 2012) (Fig. 7). Erosion rates and incision rates remained low in the Sierra through the Miocene. Basin and Range normal faulting began and the western edge of this faulting migrated westward (e.g., Slemmons et al., 1979; Dilles and Gans, 1995; Surpless et al., 2002), beheading stream systems and diminishing their drainage areas (Wakabayashi and Sawyer, 2001). This faulting did not involve footwall rock uplift of western strands, even though exhumation associated with at least some of these faults is much greater than that associated with the FFS (e.g., Stockli et al., 2002), otherwise the tilts of streams draining across the Sierra should have increased leading to incision, contradicting the observation of no incision between Eocene and Pliocene time (Wakabayashi and Sawyer, 2001). Accordingly, the incision record in the Sierra Nevada suggests that Basin and Range faulting has resulted in a decrease in mean elevation (because the hanging-wall side would decrease in elevation with respect to the footwall) in the Basin and Range, consistent with interpretations of paleoelevation based on paleobotany (Wolfe et al., 1997, 1998) and expectation of elevation decrease associated with lithospheric thinning (e.g., Buck, 1991; McKenzie et al., 2000).

Beginning at ca. 16 Ma and continuing to ca. 3 Ma, andesitic volcanism associated with the ancestral Cascades arc covered much of the Sierra from the Stanislaus River drainage northward (Fig. 7) (Busby et al., 2008a, 2008b; Busby and Putirka, 2009; Cousens et al., 2008). These andesitic flows and mudflows issued from volcanic centers along or east of the present crest of the range (Christiansen and Yeats, 1992; Busby et al., 2008b) and filled existing paleovalleys so that only isolated basement highs rose above them. Some faulting, locally associated with significant hanging-wall tilting, began near the current position of the crest and Frontal fault system at ca. 10 Ma (Busby et al., 2008a, 2008b; Busby and Putirka, 2009), but this did not result in rock uplift of the Sierran block and incision of Sierran streams. The eruption and deposition of large volumes of volcanic rock resulted in an extremely rapid temporary increase of the elevations of the streambeds, followed by rapid incision to the former equilibrium position of the channel after volcanic deposition ceased for each pulse. South of the Stanislaus drainage, the axis of the magmatic arc swung much further east. In addition, a preexisting drainage divide may have prevented andesitic mudflows and other deposits from reaching the central and southern Sierra Nevada in areas south of the San Joaquin River drainage (Fig. 7) (Henry et al., 2012).

During this period, arc magmatism in the Sierra shut off from south to north as the south edge of the subducting Gorda plate migrated northward with the Mendocino triple junction (Atwater and Stock, 1998). Rock uplift in the Sierra Nevada began sometime after the northward migration of the southern slab edge, possibly in response to slab window development (Maheo et al., 2009). Uplift and resultant steam incision began at ca. 20 Ma in the Kern to Kings River drainages,10–6 Ma in the San Joaquin River drainage, 3.6–4 Ma in the Mokelumne and Stanislaus River drainages, and ca. 3 Ma from the American River drainage northward; uplift has not yet occurred above the current position of the southern edge of the Gorda slab.

The time of headward migration of incision and erosion in the various drainages was probably short compared to the estimated differences in timing of incision initiation, so that the timing of incision inception in any part of the Sierra is probably a reasonable proxy for the onset of rock uplift. The northward migration of the onset of uplift and incision may have been stepwise, rather than progressive, and the rate of migration is far from linear. Incision began at least 10 m.y. earlier in the Kern River drainage than in the San Joaquin drainage and ∼2–7 m.y. later in the Stanislaus River drainage than in the San Joaquin drainage, but incision initiation is comparable from the Stanislaus River drainage to the northern end of the range.

After ca. 3.5 Ma, a second stage of uplift and incision began, limited to the region from the Kern to the San Joaquin River drainages, and this uplift event may have been triggered by delamination (e.g., Ducea and Saleeby 1996, 1998; Saleeby and Foster, 2004; Le Pourhiet et al., 2006). The most prominent geomorphic signature of the two uplift and incision events in this region are two knickpoints that characterize many streams in these drainages, as well as the inner canyons of the major streams of this region (Huber, 1981; Clark et al., 2005; Stock et al., 2004, 2005; Carlson et al., 2009).

Frontal faulting continued to encroach into the Sierran microplate after the initiation of late Cenozoic uplift, so the initiation of slip on the present set of Frontal faults postdates initiation of uplift and incision from the San Joaquin drainage southward, and in the Feather River drainage. Although northward migration of the onset of uplift and incision in the Sierra Nevada may have been related to slab window evolution tied to ridge collision and subsequent migration of a triple junction, the northern and southern termination of the Sierra may be fixed and controlled by preexisting basement structure.

I thank C. Busby and K. Putirka for inviting me to participate in the Penrose Conference on the Sierra Nevada and to submit this contribution. I have benefitted from discussions of Sierran landscape evolution with many, including G. Stock, C. Henry, F. Phillips, Cliff Riebe, J. Saleeby, and students who have worked with me on these matters, C. Kemp, C. Carlson, and A. Shriver. I thank J. Saleeby and F. Phillips for their thorough and thoughtful reviews that led to significant improvements in this paper.