Geomorphic evidence for enhanced Pliocene–Quaternary faulting in the northwestern Basin and Range

The western North American plate margin exemplifies diffuse deformation across a dynamic tectonic boundary. During the late Cenozoic, this tectonic boundary has evolved from a subduction zone to a transform margin, resulting in right-lateral strike-slip motion along the San Andreas fault system (Atwater, 1970; Stewart and Crowell, 1992), distributed shear in the Walker Lane belt (Henry et al., 2007), and widespread extension across the Basin and Range Province (Atwater, 1970; Faulds et al., 2005; Stewart, 1971; Wernicke, 1992). Stretching from the Sierra Nevada Mountains in the west to the Wasatch Range in the east, the U.S. Basin and Range is characterized by footwall mountain ranges and hanging-wall grabens (basins) that are accommodating roughly east-to-west extension (Fig. 1A; Wernicke, 1992). Although most of the ranges are similar in form (Ellis et al., 1999), they have evolved from normal faults with significant spatiotemporal variations in displacement and slip rate (Wallace, 1987). Despite recent contributions to the nature of active extension in the Basin and Range (Colgan, 2013; Frankel et al., 2011; Gorynski et al., 2013; Lifton et al., 2013), detailed information on the evolution of individual fault systems remains sparse. This may, in part, be because Pliocene–Quaternary changes in fault slip are challenging to identify and often poorly resolved with commonly applied techniques like thermochronology, which have a temporal resolution of 10⁶–10⁷ yr. Thus, it remains common to rely on fault-plane trenching, seismicity, and geodetics, which have temporal resolutions of 10⁰–10⁴ yr, to identify active faults, estimate their current motion, and assess their seismic hazard potential (Hemphill-Haley et al., 2000; Personius and Mahan, 2005). Even with all these methods combined, there remains a gap in our ability to resolve fault slip at the critical time scales of ∼10⁵–10⁶ yrs. An emerging approach that may help address this shortcoming is to investigate channel morphologies for clues of past base-level changes caused by tectonic and/or climatic perturbations (Kirby and Whipple, 2012; Whittaker, 2012). We hypothesize that by combining quantitative analysis of landforms modified by evolving tectonics with more established techniques, we may be able to resolve more complete normal-fault slip histories across time scales.

Most of the northern Basin and Range Province underwent a period of rapid extension beginning ca. 17–15 Ma (Fig. 1; Colgan and Henry, 2009; Miller et al., 1999; Snow and Wernicke, 2000; Stockli, 2005; Stockli et al., 2002). The northwestern Nevada Basin and Range is an exception, remaining relatively undeformed until the late Miocene (ca. 12 Ma). This late Miocene and younger extension took place on widely spaced, high-angle normal faults that accommodated less extension than the closely spaced imbricate fault systems and detachment faults found in the Basin and Range to the south and east (Colgan et al., 2006b). Interestingly, a Pliocene (ca. 3–4 Ma) increase in extension rate has been documented in several places along the western margin of the Basin and Range (Colgan et al., 2008; Henry and Perkins, 2001; Stockli et al., 2003). These observations beg the questions: How widespread is this rejuvenation? Why is it happening? Does it represent slip cessation and renewal or just a change in the rate of ongoing slip?

The purpose of this study is to use footwall catchment morphologies in the U.S. Basin and Range to augment existing constraints on normal-fault slip histories. We present new geomorphic observations from three ranges in northwestern Nevada: the Pine Forest Range, the Jackson Mountains, and the Santa Rosa Range (Fig. 1). Each range was unroofed along a major range-bounding normal-fault system, making them good candidates for investigating the relationships between fault slip and footwall morphology (after Densmore et al., 2004; Harbor, 1997). Thermochronology data show that these tilted, half-graben footwall blocks were exhumed during slip on initially high-angle normal faults beginning in the middle to late Miocene (ca. 12–11 Ma; Colgan et al., 2006b). Our combined field observations, topographic analysis, and knickpoint migration modeling suggest increased fault motion beginning in the Pliocene–Quaternary (ca. 5–0.1 Ma). Our results provide new insights to help characterize Basin and Range faulting histories, with implications for utilizing this approach to improve our understanding of how plate margins accommodate stresses through evolving patterns of deformation.

Active mountain landscapes reflect a dynamic feedback system between tectonic processes that raise Earth’s surface and erosion processes that lower it (Montgomery and Brandon, 2002; Schmidt and Montgomery, 1995). Channel incision into rock sets the lower boundary condition for hillslopes, thus dictating the texture and relief structure of the uplifting topography (Whipple and Tucker, 1999). A river in equilibrium with its environment has a smooth, concave-up longitudinal profile where slope decreases systematically with the downstream increase in contributing drainage area (Hack, 1973; Whipple, 2001). Following a change in tectonic or climatic circumstances, streams respond by adjusting their form to reach a new equilibrium condition (Bonnet and Crave, 2003; Tucker and Whipple, 2002). During this transient adjustment phase, rivers possess significant localized convexities, called knickpoints (or knick zones), within their profiles that initiate at local base level and migrate upstream as a kinematic wave (Rosenbloom and Anderson, 1994). As knickpoints move upstream, they separate the lower, transient (“adjusting”) portion of the landscape from an upper, relict portion that retains the equilibrium form from the previous conditions (e.g., Clark et al., 2005; Gallen et al., 2013). Geomorphic traits of the relict portion can be used to reconstruct the paleolandscape form because it remains disconnected from the new conditions until the migrating knickpoint reaches it. Thus, the analysis of channel gradient indices (normalized steepness [k_sn], concavity [θ]) and identification of knickpoints are becoming increasingly valuable tools for inferring tectonic patterns and their changes from eroding topographies (Kirby and Whipple, 2012; Whittaker et al., 2008; Wobus et al., 2006a).

Landscapes in steady state possess relationships between their form and the patterns and rates of rock uplift and erosion. For example, fluvial incision models and empirical evidence indicate that uplift rates scale with steady-state channel gradients, such that channels steepen with increased uplift rate (Whipple, 2004). The exact scaling between channel gradient and uplift depends on rock strength, dominant erosion processes, and frequency and magnitude of runoff (Lague et al., 2005). In areas of active extension, increased fault motion results in enhanced footwall rock uplift and contemporaneous lowering of local base level in the adjacent hanging-wall basin. Here, footwall channels strive to maintain equilibrium by steepening, causing an increase in erosion rate. Although channel steepness is not a direct measure of uplift rate, it is useful for identifying patterns of rock uplift (Cyr et al., 2010) despite the fact that the relative impacts of climate and lithology on such a relationship can vary across regions (e.g., Clarke and Burbank, 2011; Stark et al., 2010). In areas where climate and substrate erodibility do not vary much, such as within individual footwall ranges, observed patterns in channel form can be symptomatic of sustained tectonic forcing (Cyr et al., 2010; Harkins et al., 2007; Kirby et al., 2010; Whittaker, 2012).

We focus on landscapes responding to active extension along three N-S–trending normal fault arrays with primarily dip-slip motion in northwestern Nevada (Fig. 1): the Pine Forest Range, the Santa Rosa Range, and the Jackson Mountains. These ranges lie within ∼75 km of each other and are composed of Paleozoic to Mesozoic metasedimentary rocks intruded by Mesozoic plutons. Geologic data show that these pre-Tertiary basement rocks are unconformably overlain by Eocene to Miocene volcanic rocks, which constrain total fault displacement and place upper bounds on the age of fault initiation (Fig. 1A; Colgan et al., 2006b; Noble et al., 1970; Quinn et al., 1997; Wyld, 1996). The Pine Forest Range is bounded by an E-dipping, 50-km-long fault system with a maximum of ∼7 km total dip slip, capped by a conformable sequence of 30–16 Ma volcanic rocks tilted up to 30° to the west (Colgan et al., 2006a, 2006b). The Santa Rosa Range is bounded by a 60-km-long, W-dipping fault system with up to ∼8 km of total dip slip; overlying volcanic rocks are tilted ∼15° east and are 17–15 m.y. old (Fig. 1B; Brueseke et al., 2008; Colgan et al., 2006b). Paleoseismic and luminescence data from the southern Santa Rosa Range yield low slip rates (∼0.1 mm/yr averaged over ∼400 k.y.) and an elapsed time of 11–16 k.y. since the last earthquake (Personius and Mahan, 2005). The major W-dipping, range-bounding fault system in the Jackson Mountains is 70 km long, with a maximum of ∼7 km total dip slip, with overlying volcanics in the northern part of the range that are ca. 15 Ma in age (Fig. 1C; Castor and Henry, 2000; Colgan et al., 2006b; Quinn et al., 1997). The Pine Forest and Santa Rosa range-bounding faults currently dip 35°–40° and are inferred to have originally dipped 50°–55° based on horizontal restoration of the Tertiary unconformities (Colgan et al., 2004, 2006a). The Jackson Mountains range-bounding fault is not exposed, but it is inferred to be ∼40°, assuming consistency with the nearby Pine Forest and Santa Rosa Ranges (Colgan et al., 2006b). All three range fronts suggest some degree of past fault segmentation based on the topography of the fault traces, but there is no evidence that the segments are behaving individually any longer (i.e., anomalous topographic highs, structural ramps, or underdisplaced sections), thus we believe them to be past the point of segment interaction.

Apatite fission-track (AFT) and (U-Th)/He (AHe) thermochronology can constrain rock cooling histories in the upper ∼2–5 km of crust through age determinations from the temperature-dependent retention of radiogenic decay products (Reiners and Ehlers, 2005). AFT ages are completely reset to zero at temperatures >∼110–135 °C and are partially reset due to track shortening at temperatures between ∼60 °C and 110 °C, called the partial annealing zone (PAZ; e.g., Dumitru, 2000; Green et al., 1989); AHe ages are completely reset to zero at temperatures >∼65–80 °C and are partially reset due to partial ⁴He loss between ∼40 °C and 80 °C, termed the partial retention zone (PRZ; e.g., Wolf et al., 1998). The base of the PAZ and PRZ can mark the onset of rapid exhumation, and, along with careful structural control, thermochronologic sample cooling age patterns from an unroofed footwall can be directly related to slip on the bounding fault (Stockli, 2005). We chose the three northwestern Nevada ranges because existing apatite thermochronology data document their exhumation and faulting histories (Figs. 1 and 2; data from Colgan et al., 2006b). Faulting in the Pine Forest Range began 12–11 Ma at a slip rate of 0.3–0.8 km/m.y. (exhumation rate of 0.3–0.5 km/m.y.; Fig. 2). The youngest AHe age is ca. 4 Ma, indicating ∼2 km of exhumation (equivalent to cooling from ∼65 °C) since that time (Colgan et al., 2006b). Faulting in the Santa Rosa Range also began 12–11 Ma at a slip rate of 0.4–0.8 km/m.y. (exhumation rate of 0.3–0.5 km/m.y.; Fig. 2). The youngest AFT age (AHe dates from the Santa Rosa Range are problematic) is 6.4 Ma, indicating ∼3 km of exhumation (equivalent to cooling from ∼110 °C) since that time (Colgan et al., 2006b). The initiation and duration of faulting in the Jackson Mountains are less well constrained, but Colgan et al. (2006b) reported AFT and AHe ages of 12 and 7 Ma, respectively (Fig. 2), from one sample near the range-front fault, which they argued indicated a similar late Miocene age for faulting as the nearby Pine Forest Range.

The Great Basin region is generally interpreted to have been a broad highland (the “Nevadaplano”) in late Eocene to Oligocene time (ca. 40–30 Ma; Chamberlain et al., 2012; DeCelles, 2004). Prior to the mid-Miocene decrease in mean elevations associated with the development of modern Basin and Range topography, the Sierra Nevada range was already a major orographic barrier (elevation >2 km), blocking air masses and thus precipitation from reaching the Great Basin (Cassel et al., 2012; Horton et al., 2004; Poage and Chamberlain, 2002). This suggests that the first-order climate state was established by the mid-Miocene in our study area. The three ranges have never been glaciated, and the only modern variations in climate are from local, relief-enhanced precipitation that are minor (Ehlers and Gibbard, 2007; Horton et al., 2004). Oxygen isotope and paleobotanical records document a global shift to a cooler climate and enhanced climate variability in the late Cenozoic (ca. 3 Ma; Molnar and England, 1990; Shackleton et al., 1984) that could increase erosion (Zhang et al., 2001) and possibly facilitate channel adjustment in Basin and Range footwalls.

Pliocene–Pleistocene climate cycling caused repeated periods of pluvial lake development throughout the study area. Paleo–Lake Lahontan once occupied every basin in northwestern Nevada; the last and largest highstand occurred ca. 13 ka, when water levels just reached the range fronts in our study area (∼1330 m; Morrison, 1991; Reheis, 1999a, 1999b). Since that time the lakes have become progressively smaller, indicative of higher Pleistocene precipitation and a long-term drying trend from the Pleistocene to the present (Reheis, 1999a, 1999b). The pluvial lakes in the Great Basin have always been internally drained, and there is no evidence in the sedimentary record for any removal of mass from the regional basin system (Morrison, 1991). Given this collective information, we choose to assume that tectonics have been, and continue to be, the dominant force affecting local base-level fall and footwall landscape development in the Basin and Range since the mid-Miocene because (1) the relief-enhanced climate gradients are rather minor (∼30–40 cm/yr variation across ranges; Daly et al., 2008), and (2) mid-Miocene to recent climate fluctuations are likely subordinate to tectonic-based forcing at the individual range scale (after Densmore et al., 2004; Harbor, 1997; Whipple and Trayler, 1996).

Lithologic qualities such as composition, fracture density and characteristics, and rock strength exert first-order controls on channel morphology, erodibility, and incision (Allen et al., 2013; Clarke and Burbank, 2011; Montgomery and Gran, 2001; Sklar and Dietrich, 2001). We quantified intact rock strength using a type N Schmidt hammer, a spring-loaded device that measures “rebound” values that scale with laboratory-based measurements of unconfined rock strength (Goudie, 2006; Selby, 1993). We collected rock strength data at eight sites across the ranges: six in Cretaceous granite and two in Paleozoic metasedimentary rocks (triangles, Figs. 1B–1D). We took 40+ measurements at each site and report the mean and standard deviation. We observed fracture presence and characteristics (orientations, degree of mineralization) in the field and concluded that they are uncommon and invariant across the region and hence not likely a controlling factor of erodibility in this region.

Hillslope gradients can correlate with denudation rates across various mountain settings (Binnie et al., 2007; Harrison, 2000). In active orogens, denudation rates are high but variable because more efficient, erosional processes like mass wasting become dominant above a critical hillslope threshold (∼30°; Binnie et al., 2007; Burbank et al., 1996; Montgomery and Brandon, 2002). We calculated hillslope angles using ArcGIS (v. 10) and the 10 m National Elevation Data set (NED) digital elevation model (DEM; Gesch, 2007; Gesch et al., 2002) throughout the main, uplifting footwall catchments; we also measured hillslope angles in the field using a handheld laser range finder at 30 sites from all three ranges (Table DR2¹) to validate our DEM-based measurements.

Quantifying bedrock river variations with normalized steepness is a common approach for inferring patterns of active tectonics across uplifting landscapes (Cyr et al., 2010; Harkins et al., 2007; Kirby and Whipple, 2012; Wobus et al., 2006b). We analyzed all channels traversing the range front faults that have >1 km² drainage area and/or reach the divide (Fig. 3). We used the 10 m NED DEM and codes written in R programming language (version 2.13.1) to conduct the analysis. We smoothed the profile data by resampling the raw elevation data at equal vertical intervals using the contour interval from the original data source (12.192 m [40 ft]; methods after Wobus et al., 2006a). We examined all channel profiles and slope-area data to identify knickpoints, which we define as significant convexities in the profile accompanied by a distinct change in channel gradient (“slope-break” knickpoints; e.g., Fig. 4; Haviv et al., 2010). We calculated θ_ref as the mean concavity of the equilibrium (upstream) segments, here θ_ref = 0.51. We then calculated (1) k_sn along the entire channel reach and (2) a mean k_sn for each segment separated by the knickpoint. We next consulted existing geologic maps (Colgan et al., 2006a; Crafford, 2010; Quinn et al., 1997) and the geographic information systems (GIS)–based flow accumulation grid to identify knickpoints coincident with sharp changes in rock type and/or contributing drainage area. We assumed all knickpoints associated with such features to be static effects of them, and all others to be migratory in nature (e.g., Crosby and Whipple, 2006; Miller et al., 2012).

Existing data sets provide bounds on erosion rates for northwestern Basin and Range topography spanning multiple time scales. Low-temperature thermochronology provides first-order estimates on long-term (m.y.) erosion rates, because AHe ages record the time since a rock cooled through 40–80 °C (Clark et al., 2005; Farley, 2002). In detail, erosion is the surficial removal of mass, and exhumation is rock unroofing (England and Molnar, 1990; Ring et al., 1999). Often considered equivalent, we note that discrepancies between erosion and exhumation rates can arise due to footwall tilting and variations in subsurface heat flow (Ehlers et al., 2001; Ehlers and Farley, 2003). We use a range of 300–500 mm/k.y. for erosion rate (based on the exhumation rates of 0.3–0.5 km/m.y.; Fig. 2), but we consider these values to be near maximum estimates because exhumation rates from one-dimensional age-paleodepth relationships can overestimate surface erosion rates by 10%–40% in normal-fault settings (Ehlers et al., 2001). Cosmogenic ¹⁰Be data from footwalls elsewhere in the Basin and Range (Wassuk Range in Nevada, Inyo Range in California, and Stillwater Range in Idaho) with similar climate, lithologies, and morphologies provide shorter-term (10⁴ yr) estimates of basin-averaged erosion rates: ∼9–100 mm/k.y. for the upper, low-gradient channel reaches and 100–>750 mm/k.y. for the lower, steep reaches (Densmore et al., 2009; Kirby and Whipple, 2012). Conservatively, we used a range of 1–700 mm/k.y. for erosion rate in Equation 3 to calculate the time required to erode the volume of rock from below the knickpoints. We included 700 mm/k.y. as the upper bound to explore this high-rate possibility, but we consider it unlikely because it exceeds the locally measured exhumation rates.

Mean rebound values for rocks exposed throughout the three footwalls range from 43.5 to 71.9 (Fig. 6; Table 1). There is high variability in both the granites and the metamorphic rocks: Rebound values on granite range from 46.5 to 71.9, and those for metamorphic rocks range from 43.5 to 51.5. The highest value comes from a stream-abraded, unweathered granite outcrop in the Santa Rosa Range, compared to other localities that possessed some surface weathering and lower rebound values. Although the granitic rocks tend to yield higher rebound values, there is no significant difference in values between the two lithologies within error, defined here and throughout the paper as one standard deviation (⁠_granite = 60.7 ± 9.6; _metamorphic = 47.5 ± 5.6). Furthermore, knickpoints are rarely observed at lithologic boundaries in the ranges. In conclusion, we see no quantifiable difference in lithology-based rock erodibility.

Topographic analysis shows that hillslopes are steeper below the knickpoints compared to above them (Fig. 7). In the Pine Forest Range, mean hillslope angles are _above = 16° and _below = 24°. In the Jackson Mountains, they are _above = 23°, _below = 30°; in the Santa Rosa Range, they are _above = 25°, _below = 28°. Field-based measurements spanning all ranges show an identical pattern with a mean value of 20° above and 29° below (Table DR2 [see ^{footnote 1}]). These data show that there is consistently a 6°–9° difference between mean hillslopes above versus below the knickpoints. The contrast between slope distributions for hillslopes above versus below the knickpoints suggests a contrast in erosion rates. Erosion rates increase with steeper hillslopes up to a threshold angle of ∼30°–35°, at which point erosion rate is controlled by frequency of landsliding (Binnie et al., 2007; Ouimet et al., 2009). Below the knickpoints, which we interpret as transient portions of the channels, the corresponding hillslopes approach threshold angles, whereas hillslopes above the knickpoints do not.

We analyzed channel profiles from 57 catchments that occupy various along-strike positions within the footwalls adjacent to the three active fault systems (Fig. 3; see Table DR1 for complete results [see ^{footnote 1}]). Catchment drainage areas range in size from 0.7 to 59.7 km². Catchments from the Santa Rosa Range are substantially larger relative to the others (mean area is 16.8, 6.6, and 8.8 km² for the Santa Rosa, Jackson Mountains, and Pine Forest, respectively), which is due to the relative widths of the fault blocks (Densmore et al., 2005). We found 39 knickpoints at similar elevations across 34 different catchments that are not associated with sharp changes in lithology or drainage area (e.g., tributary junctions; circles in Fig. 3; Table 2). Multiple knickpoints within a single catchment signify two main channels, each with a knickpoint at approximately the same elevation. Specifically, we observed 17 knickpoints in 15 (of 22 total) catchments with a mean elevation of 1847 ± 133 m in the Pine Forest Range (Fig. 3A), 14 knickpoints in 12 (of 23 total) catchments with a mean elevation of 1918 ± 103 m in the Jackson Mountains (Fig. 3B), and 8 knickpoints in 7 catchments (of 12 total) with a mean elevation of 2026 ± 148 m in the Santa Rosa Range (Fig. 3C). Catchments that do not possess this primary migratory knickpoint do so for one of the following reasons: (1) The channel head elevation is below the observed knickpoint elevation range of ∼1950 m (n = 13); (2) there is a knickpoint at an identifiable heterogeneity such as a lithologic contact (n = 4) or tributary junction (n = 1; triangles in Fig. 3); or (3) there was no identifiable knickpoint observed (n = 6) at our data resolution (gray channels, Fig. 3). Two knickpoints coincident with changes in rock type (of four total) were also within the elevation range of the main observed knickpoint population, and so determining the static versus migratory nature of these knickpoints is ambiguous. The catchments without knickpoints also possess the smallest upstream drainage areas (<6 km²) and lowest catchment-scale relief (average of ∼525 m), and they tend to be near the fault tips (Fig. 3; Table DR1 [see ^{footnote 1}]).

The knickpoints divide the channels into two distinct segments (Figs. 3 and 4). The upper reaches above the knickpoints have low steepness (mean k_sn = 182 ± 68) compared to the lower reaches, which are characterized by high steepness (mean k_sn = 361 ± 84; Fig. 3). Within individual channels, the lower segments are typically 1.5–4 times steeper than the upper segments (e.g., Fig. 4; Table 2). The knickpoints also coincide with steeper hillslopes (mean 24°–30° vs. 16°–25°; see earlier herein). Catchments without migratory knickpoints have intermediate steepness (mean k_sn of 247 ± 82), and their hillslopes approach high values (∼25°).

Channel paleoprofile reconstructions show 20–512 m of relief generation since the knickpoints entered the footwall catchments (e.g., Δr in Fig. 4; Table 2). The highest estimates for relief enhancement are from streams draining catchments near the range centers, whereas the lowest estimates tend to be near the tips (Table 2). There is no relationship between the estimated magnitude of increased relief and drainage area. Despite the wide variability of these relief estimates, the mean value of new relief calculated for each of the three ranges is remarkably similar (221 m, 220 m, and 223 m, in Pine Forest, Jackson Mountains, and Santa Rosa, respectively), suggesting a similar amount of base-level fall across the study area.

The volume of eroded rock from below the knickpoints ranges from 0.06 to 1.2 km³ (mean = 0.5 km³). We used a range of erosion rates to back-calculate the timing of knickpoint initiation (Fig. 5): If a catchment eroded more slowly, it would take longer to remove the volume of rock below the knickpoint. If the catchments eroded at an average rate of 30 mm/k.y., a minimum value relative to observed basin-averaged erosion rates in the U.S. Basin and Range, then the knickpoints were generated between 5 and 1 Ma (Fig. 5C). A catchment-averaged erosion rate equivalent to long-term exhumation rates (∼300–500 mm/k.y.) requires knickpoint formation ≥0.1 Ma. Therefore, our volume-for-time substitution results require knickpoint initiation between 5 and ≥0.1 Ma.

Our knickpoint modeling results show a strong match between predicted and observed knickpoint locations across all 39 knickpoints within the three ranges, which were modeled to initiate simultaneously (e.g., Figs. 8 and 9B). The best-fitting p parameter is consistently 0.45, and the C parameter varies with knickpoint initiation timing, but ranges from 4 × 10⁻⁵ to 8 × 10⁻⁷ m^0.9/yr. The lowest misfits between the observed and modeled knickpoint locations are for knickpoints initiating between 5 and 1 Ma (Fig. 9A; Table DR2 [see ^{footnote 1}]), but all model runs predicted the observed knickpoint locations with ∼1% difference in least sum of squared misfits (LSS). However, because the concomitant best-fit values for C are quite variable, they also result in varying knickpoint migration velocities (see Eq. 3). Knickpoint migration velocities for all knickpoint initiation times explored (11–0.5 Ma) span from ∼0.2 to 100 mm/yr. Knickpoints that initiated between 5 and 1 Ma (the lowest LSS result) require migration rates of 0.5–10 mm/yr (Fig. 9A). For example, predicted versus observed knickpoint locations match extremely well (R² = 0.85; Fig. 9B) for an initiation age of 3 Ma and a migration rate of 1.5 mm/yr. Assuming all transient knickpoints initiated at the range front simultaneously and traveled upstream as migratory features, our model achieves the lowest misfits for initiation ages between 5 and 1 Ma. However, a wide range of parameters provides good fits, and thus more thorough constraints on erosion rates are required to confirm this time frame for knickpoint initiation.

There are several possible explanations for our observed channel profile knickpoints and associated steepness patterns. Static knickpoints are associated with distinct, observable geologic phenomena such as faults, landslide deposits, lithologic boundaries, or tributary junctions (Duvall et al., 2004; Korup, 2006; Walsh et al., 2012). This is not the case for the main population of transient knickpoints (n = 39) since we eliminated those possibilities. Furthermore, channels with static knickpoints tend to have peaks in slope on a slope-area plot at the heterogeneity locale rather than a distinct and persistent change in channel slope as we observe (i.e., “vertical-step” vs. “slope-break” knickpoints; see discussion in Kirby and Whipple, 2012). The segmented nature of the channel steepness patterns suggest that they are in a transient state of adjustment to a change in base level. The knickpoints separate upper, relict reaches with low hillslope relief from downstream reaches with 1.5–4× greater steepness and hillslopes that are 6°–9° steeper and near threshold values for the onset of landsliding (e.g., Binnie et al., 2007). Furthermore, the knickpoints are located within relatively narrow altitudinal bands consistent with streams responding to a uniform, sustained forcing such as a change in fault-slip rate. The few channels without knickpoints, in contrast, resemble the downstream, adjusted portions of the streams with migratory knickpoints. When considered in conjunction with their drainage area, relief, and geographic positions, it appears that if a knickpoint had formed in these channels, they have now fully adjusted to the current conditions. These collective observations, combined with our geomorphic analyses and assumptions about the negligible role of climate, lead us to interpret the main suite of knickpoints to be migrating features that initiated ca. 5–0.1 Ma as the result of a sustained change in fault-slip rate.

Modeling and theoretical studies suggest that the vertical rate of knickpoint propagation is a function of rock uplift rates relative to local base level, with higher rates producing higher-elevation knickpoints in a given time (Attal et al., 2008; Niemann et al., 2001; Whittaker, 2012). Faults commonly accumulate displacement by tip propagation and/or segment linkage, resulting in an integrated displacement gradient that increases from zero at the fault tips (i.e., the terminus of the fault) to a central maxima (e.g., Cowie and Roberts, 2001; Dawers et al., 1993). Local constraints on minimum fault throw since fault initiation in our study area (Colgan et al. 2006b; Quinn et al. 1997) suggest that the three fault systems possess long-term deformation rates consistent with this pattern, suggesting they are past the point of segment linkage. However, the knickpoint heights only marginally follow the displacement pattern (Fig. 10). This may indicate that Pliocene–Quaternary deformation rates are distributed more uniformly along strike than the long-term throw data suggest (i.e., they do not display the ideal, bow-shaped distribution). Indeed, if the spatial distribution of displacement rate were more constant along the fault plane, then the knickpoints would fall within similar elevations, as we observe. Our observations of the spatial distribution of knickpoints suggest that the long-term rate of displacement may not follow the same spatial pattern as shorter-term rates. This conclusion is a consequence of normal-fault interaction and growth: Over the lifetime of a fault, long-term patterns of slip do not necessarily mimic short-term patterns as fault segments lengthen, interact, link, and/or cease to grow (Cowie, 1998; Cowie and Roberts, 2001). We present this as a cautionary example of the care that must be taken to account for time averaging when comparing tectonic patterns and geomorphic features.

Changes in climate and basin sedimentation can also cause channels to experience a relative base-level fall. Here, we acknowledge these possible mechanisms for footwall knickpoint formation and rule them out in favor of a change in fault slip as follows. Modern regional base level for all three ranges is the Quinn River, a mud-dominated meandering river located on the floor of paleo–Lake Lahontan (Matsubara and Howard, 2014). This and other paleolakes are symptomatic of excess moisture in the northwestern Basin and Range during the Pleistocene. Increased precipitation would result in an increase in erodibility and thus a decrease in channel gradients (Whipple et al., 1999), which is the opposite of what we observe. In the last ∼280 k.y., there have been up to five major highstands in Lake Lahontan, which reached its highest level during the Sehoo highstand ca. 13 ka (Adams and Wesnousky, 1999; Morrison, 1991). If our observed knickpoints formed during the regional fall in base level that followed the last highstand ca. 13 ka, they would have migrated to their current locations at rates >>100 mm/yr, which is unreasonable (see earlier discussion). Lake Lahontan was big, but in our study area, water levels only reached near the range front during the last highstand (Morrison, 1991; Reheis, 1999a, 1999b), and so we would anticipate minor or negligible aggradation within the mountain interfluves. Given that (1) we observe a steepening of channel gradients, (2) lake levels were relatively low, and (3) the basins were internally drained, we infer that mid-Miocene to recent climate fluctuations are subordinate to tectonic-based forcing in driving footwall landscape development in the study area.

Rapid exhumation of the footwalls commenced ca. 12–11 Ma, placing an upper limit on any fault-driven changes in topography (Colgan et al., 2006b). Thick sequences of ash-flow tuffs, lava flows, and sedimentary rocks blanketed the study area from the Eocene to the middle Miocene, filling in any previously developed topography on the basement surface and creating a low-relief volcanic plateau prior to the onset of extension (Brueseke et al., 2008; Colgan et al., 2006a; Lerch et al., 2008). This rules out the possibility that the upstream channel segments represent inherited, prefaulting topography (e.g., Densmore et al., 2009). Our channel profile reconstructions estimate an average of 221 ± 126 m (range: 20–512 m) of enhanced, catchment-scale relief across all ranges (Δr in Fig. 4). This means an average of 833 ± 259 m (range: 367–1446 m) of catchment-scale relief predated knickpoint formation. These values are minimums because they do not account for any erosion of the ridge crest or the upper, relict channel reaches. We speculate the large spread in the data represents the nonuniformity of the displacement field and catchment drainage areas along strike, both of which contribute to the variability in Δr; we report the means for purposes of comparison across the study ranges. The lack of regional paleotopography and consistency of paleorelief and Δr estimates across the ranges indicate that, since faulting began, local base level has lowered. Because we propose that enhanced fault slip has facilitated relief creation, we infer that the 20–512 m of new relief is related to the amount of fault throw since slip rate increased.

Our rock volume–for–time substitution analysis provides corroborating evidence for a base-level drop following fault initiation at ca. 12–11 Ma. The exhumation rate of these footwalls from thermochronology is 0.3–0.5 mm/yr (Colgan et al., 2006b). A volume-for-time substitution using an equivalent erosion rate (300–500 mm/k.y.) requires that fault slip increased within the last ∼0.5 m.y. Cosmogenic ¹⁰Be catchment-averaged denudation rates in the Basin and Range vary from 9 to >750 mm/k.y. (e.g., Densmore et al., 2009; Kirby and Whipple, 2012), with variation within a single range often spanning two orders of magnitude. For example, the Sweetwater Range in southwest Montana has ¹⁰Be-derived denudation rates of 9.7–34.8 mm/k.y. at the range front and 9.2–14.2 mm/k.y. in high-elevation reaches (Densmore et al., 2009). The eastern California Inyo Range has ¹⁰Be-derived denudation rates of >750 mm/k.y. at the range front and 40–80 mm/k.y. in high-elevation reaches (Kirby and Whipple, 2012). Given this variability, we explored a range of values (30–700 mm/k.y.) in the volume-for-time substitution analysis, thus bracketing the increase in slip as beginning sometime between 5 and <0.1 Ma (Fig. 5C). This time frame indicates that >200 m of relief was created since 5 Ma and possibly within the last 0.5 m.y., following initial formation of >800 m of earlier fault-related relief. This estimate is consistent with thermochronologic data that show exhumation was ongoing since 12–11 Ma, with the youngest cooling ages indicating 2–3 km of exhumation since ca. 6–4 Ma (Colgan et al., 2006b). With 2–3 km of exhumation (rock uplift relative to the surface) and >200 m (0.2 km) of new relief (i.e., surface uplift) since that time, it requires an erosion rate in the range of 0.3–0.7 km/m.y. (300–700 mm/k.y.). In summary, the data collectively show that enhanced faulting was responsible for increasing footwall relief by 25% of the existing value (∼200 m compared to ∼800 m pre-knickpoint formation) at the time or ∼20% of the total today (>200 m of ∼1130 m).

Our knickpoint migration model results support the idea that at least 35 of 39 knickpoints initiated at the same time and migrated upstream across the ranges. Based on large spatial misfits (>2 km), four observed knickpoints may not be related to the main population (white circles in Fig. 9B). One Jackson Mountain knickpoint location is underpredicted by ∼2 km. The knickpoint is in a catchment coming off a large paleovalley in the middle of the Jackson Mountains (Fig. 3B). This knickpoint may have originated at the catchment mouth within the paleovalley rather than where the paleovalley reaches the range front ∼5 km away, or it may represent an underdisplaced portion of the fault still undergoing segment linkage. Three knickpoint locations (two of which are tributaries in the same catchment) in the Pine Forest Range are overpredicted by several kilometers (white circles in Fig. 9B). It could be that these knickpoints are: (1) associated with a more recent event and hence have not propagated as far upstream or (2) “stuck” on an undocumented mechanical boundary. Another possibility is that the model makes the necessary assumption that drainage area is constant through time, i.e., that the catchments do not lose area to neighboring catchments (lateral divide migration) nor does the main range boundary move (Goren et al., 2014; Willett et al., 2014). Drainage divide migration would affect the model goodness of fit. That said, some degree of misfit is expected when fitting a model to such a large population (e.g., Gallen et al., 2013), but the cumulative misfit between observed and modeled knickpoint locations is so small that the results support the notion that they are genetically related to a change in fault-slip rate.

The best results from our knickpoint migration modeling signify that they formed at the range fronts within the last 5 m.y. and migrated upstream at a rate of 0.5–10 mm/yr (Fig. 9A). Other areas that have similar drainage areas and lithologies, as well as C coefficients that fall within our assumed range, have comparable migration rates of <1–10 mm/yr (Bishop et al., 2005; Hayakawa and Matsukura, 2003; Loget and Van Den Driessche, 2009). In our model, the erosional coefficient, C, varies by over an order of magnitude to adjust to the prescribed knickpoint initiation times, but the high values of C that correspond to the youngest initiation times (younger than 0.5 Ma) require knickpoint velocities that are beyond reasonable values given the setting. Knickpoint migration velocities high enough for the convexities to have initiated within the last 0.5 Ma are ≥100 mm/yr; migration rates of this magnitude are only observed in areas with primarily alluvial substrate or large drainage areas (>100 km²), neither of which applies here (Loget and Van Den Driessche, 2009). Erosion rates low enough for the knickpoints to have initiated prior to 5 Ma are ≤30 mm/k.y.; erosion rates of this magnitude are not supported by the thermochronology data and are only observed in upper, relict portions of catchments with lower overall relief than those studied here (e.g., the Sweetwater Range in Idaho; Densmore et al., 2009). However, without local constraints on the C coefficient, we cannot rule out the possibility that it could fall outside the range of model best fits and hence vary the knickpoint initiation timing further. That said, because knickpoint migration rates are unlikely to be >100 mm/yr here, and sustained erosion rates are unlikely to be <30 mm/k.y., we favor the interpretation that these knickpoints initiated between 5 and 0.1 Ma (gray bar, Fig. 5C).

Multiple deformation phases have been recognized in the western Basin and Range since late Oligocene time. For example, the Warner Range, ∼150 km west of our study area in northeast California, has a two-phase extensional history with the most recent phase of increased slip commencing ca. 3 Ma (Fig. 1; Colgan et al., 2008). Similar renewed Pliocene (ca. 4–3 Ma) extension has been documented in the White Mountains (Stockli et al., 2003) and Inyo Mountains (Goren et al., 2014) in eastern California, the Wassuk Range in western Nevada (Gorynski et al., 2013; Stockli et al., 2002), and in the Donner Pass fault zone near Reno, Nevada (Henry and Perkins, 2001) (Fig. 1). It has been hypothesized that the Walker Lane belt in western Nevada, a region of left-stepping, en-echelon faults with a dextral slip component and high seismicity, began propagating northwestward ca. 9–3 Ma, causing strike-slip deformation in northwestern Nevada (Fig. 1; Faulds et al., 2005). This propagation was approximately coeval with vertical-axis rotations in northwestern Nevada (Cashman and Fontaine, 2000), and a change in motion of the Pacific plate relative to North American plate motion at ca. 8 Ma (Atwater and Stock, 1998). Furthermore, there is evidence for uplift and incision of the Sierra Nevada between 3 and 1.5 Ma (Stock et al., 2005) and <5 Ma (Wakabayashi and Sawyer, 2001). Although the precise nature and timing of this observed enhancement of deformation differ in places, most concur that it is a reflection of the diffuse deformation associated with the Pliocene–Quaternary evolution of the western North American plate margin (Cashman and Fontaine, 2000; Faulds et al., 2005; Henry and Perkins, 2001). Our observations contribute to this growing body of evidence for the Pliocene–Quaternary rejuvenation of deformation within the western Basin and Range.

Increased slip rate can also be attributed to the evolving geometry of a fault system independent of a change in external tectonic forcing. Large, crustal-scale fault systems tend to grow by the amalgamation of multiple segments. As the segments interact and become mechanically linked, they come to behave as one fault equal to the combined length of the two segments with displacement reaching a maximum at the collective fault center (Cartwright et al., 1995; Cowie, 1998; Dawers and Anders, 1995; Gupta and Scholz, 2000). Segment boundaries are often initially locations of anomalously low displacement (Gupta and Scholz, 2000; McLeod et al., 2000), but they ultimately experience slip-rate enhancement due to evolving Coulomb stress changes in the upper crust (Cowie, 1998; Cowie and Roberts, 2001). These faults appear to be matured past the point of segmentation, with the possible exception of the northern Jackson Mountains. Thoroughly determining the timing of formation, propagation, and interaction of each fault segment within our study area is beyond the scope of this study, but we recognize that segment linkage and associated slip enhancement in this region could be the cause for some localized knickpoint anomalies.

Evolving plate-margin kinematics facilitating a change in regional deformation style in the western Basin and Range are consistent with our principal interpretation: that transient channels are adjusting to a change in fault-slip rate throughout fault-array systems within an ∼100 km² area in northwestern Nevada since ca. 5–0.1 Ma. Further investigations into footwall morphologies throughout the Basin and Range may help test (1) the robustness and/or widespread nature of this Pliocene–Quaternary landscape response, especially farther east into the more central Basin and Range, as well as (2) whether there is a link to the northwestward migrating Walker Lane belt by inferring fault-slip changes that may track in age with its propagation northward. We further speculate that the global climate shift ca. 4–2 Ma to a more variable climate state could have helped facilitate channel incision (Whipple et al., 1999; Zhang et al., 2001). However, the geomorphic signal associated with such a climate shift is difficult to deconvolve and may have actually resulted in the opposite consequence of reduced fluvial and hillslope gradients, thus providing more support for a tectonic mechanism.

Basin and Range footwall topography in northwestern Nevada appears to be in a transient state. Our observations show that base level dropped substantially (∼220 m) since the Pliocene, resulting in a wave of incision that formed intramontane fluvial knickpoints that segment steep, fault-proximal channel reaches and hillslopes from lower-gradient upstream regions. As the knickpoints migrated upstream, a volume of ∼0.5 km³ of rock was removed from the footwall catchments since ca. 5–0.1 Ma. These results are supported by a knickpoint migration model combined with existing erosion rate estimates that suggest the knickpoints traveled at rates of 0.5–2 mm/yr. We suggest that these geomorphic phenomena result from enhanced fault slip on several normal-fault systems, perhaps in response to the evolving western North American tectonic boundary and/or evolving fault-system geometries. The collective geomorphic evidence is consistent with a phase of faulting previously undocumented using solely thermochronology. We suggest that catchment- to channel-scale geomorphic approaches such as those used here have the potential to improve constraints on fault-motion histories throughout the Basin and Range, bridging the gap between fault initiation and trench-based millennial time scales, and perhaps identifying more new phases of fault motion and topographic growth. Finally, our results bring up new questions and hypotheses about Pliocene–Quaternary landscape change in the Great Basin that we hope inspire future research.

M. Ellis thanks the University of North Carolina Department of Geological Sciences Martin Fund, Geological Society of America, and Sigma Xi for financial support. Julia Ellis and Jim Mize provided valuable help in the field. We thank Andy Cyr, Karl Wegmann, and Sean Gallen for their advice and input and reviewers Nancye Dawers, Scott Miller, Michael Oskin, and an anonymous reviewer for their very constructive reviews.