A paleoseismic transect across the northwestern Basin and Range Province, northwestern Nevada and northeastern California, USA Open Access

The Basin and Range Province is a large region of extensional tectonics in the intermountain region of western North America. The province is characterized by numerous north-trending ranges and internally drained basins formed primarily by normal faulting in the Neogene and Quaternary (e.g., Stewart, 1971, 1978; Wallace, 1984, 1987). A total of ∼150 km of extension across the 700–800-km-wide province is thought to be related to a combination of buoyancy changes in the crust and mantle and accommodation of right-lateral transform motion across the Pacific–North American plate boundary since inception of the San Andreas fault system (Wernicke, 1992; Jones et al., 1996; Sonder and Jones, 1999). Approximately 20%–25% of the ∼50 mm/yr of right-lateral deformation across the plate boundary is distributed east of the Sierra Nevada, and most of that deformation is taken up by strike-slip faulting in the Walker Lane and normal faulting along the eastern and western margins of the province (Dixon et al., 1995; Bennett et al., 1998, 2003; DeMets and Dixon, 1999; Thatcher et al., 1999; Thatcher, 2003; Hammond and Thatcher, 2004; Bormann et al., 2016). The remainder is taken up by normal faulting along isolated ranges in the interior of the province.

The Neogene tectonic and magmatic history of the northwestern Basin and Range Province (NWBR) differs from the better studied parts of the province in central Nevada. The latest phase of extension in the NWBR is thought to have initiated later (≤12 Ma versus ≤30 Ma) and accumulated less strain (≤20% versus 50%–100% extension) than the central part of the province (Colgan et al., 2006b; Lerch et al., 2008; Egger and Miller, 2011). With the exception of the Warner Range on the western margin of the NWBR, the area was covered with voluminous middle Miocene (ca. 16 Ma) Columbia River basalts (Noble et al., 1970; Hart and Carlson, 1985) and nearly coeval rhyolites associated with calderas related to initial impingement of the Yellowstone hotspot (Coble and Mahood, 2016). The deposition of extensive middle Miocene volcanic rocks largely predates the onset of the latest phase of extension (Colgan et al., 2004, 2006a, 2006b) in the NWBR.

Two recent paleoseismic transects determined late Quaternary slip histories at lat 40°–41°N (Wesnousky et al., 2005) and 38.5°–40°N (Koehler and Wesnousky, 2011) across Basin and Range structures in central and north-central Nevada. Both studies compared cumulative slip on late Quaternary structures to geodetic networks at the same latitudes. The transect at 40°–41°N produced a much slower cumulative extension rate on faults over the past 20 k.y. (∼0.65 mm/yr) than the geodetically measured extension rate of ∼2 mm/yr. In contrast, the transect farther south at 38.5°–40°N resulted in a cumulative extension rate on faults over the past 20 k.y. (∼1 mm/yr) that was consistent with geodetic rates across the region. Such mixed results are common in other rate comparison studies (e.g., Friedrich et al., 2003, 2004; Bell et al., 2004; Gold et al., 2014) in the Basin and Range. Possible disparities between the geologic and geodetic deformation data are important because previous seismic hazards maps of the region were based primarily on historical seismicity and paleoseismic data from mapped Quaternary faults (e.g., Frankel et al., 2005), but the latest maps include geodetic data in hazards calculations (Petersen et al., 2014a, 2014b). Disparities between geologic and geodetic deformation rates add to uncertainties in such calculations.

In this study we characterize the history of surface-rupturing earthquakes in latest Quaternary time (past 15–18 k.y.) across a transect at lat 41°–42°N in the NWBR of northwestern Nevada and northeastern California (Fig. 1; also refers to the larger scale, more detailed version; see Supplemental Fig. S1¹). We chose to study this region for several reasons, including the excellent preservation of faulted features, the well-defined western margin of the province (Surprise Valley fault zone), permanent global positioning system (GPS) stations that span the region, and a lack of historical surface ruptures that can create transient signals associated with post-seismic relaxation (e.g., Hetland and Hager, 2003) in the modern strain field. Such perturbations can complicate long versus short-term rate comparisons (Wesnousky et al., 2005).

We chose to focus our paleoseismic transect on faults that have been active in the latest Pleistocene and Holocene because these structures probably are responsible for ongoing deformation in the region and thus may be the most likely sites of future surface-rupturing earthquakes. Fortuitously, all of the Holocene-active faults in the transect were covered in whole or part either by the last (Sehoo) pluvial cycle of Lake Lahontan or by other closed-basin pluvial lakes of equivalent age. The Sehoo lake cycle began ca. 35 ka, and culminated at its highstand elevation (∼1330–1340 m) ∼13,000 ¹⁴C yr ago (Thompson et al., 1986; Benson et al., 1995; Adams and Wesnousky, 1998, 1999). We use the latter age and an uncertainty value of ±200 yr to determine a calibrated age of 15.6 ± 0.3 ka (unless otherwise noted, all ages are presented with 1σ uncertainties) for the Sehoo highstand in the Lahontan basin. Highstands of some isolated pluvial lakes such as Lake Surprise and Lake Alvord may be slightly older (∼18 ka) than the Lahontan highstand shoreline, so herein we use an age range of 18–15 ka for these datums. The widespread sediments and landforms associated with Lake Lahontan and other pluvial lakes provide important data for assessing the recency of surface faulting, and allow us to calculate rates of deformation over a uniform time period that we then compare to the modern strain field as determined by space geodesy.

After a discussion of methods, we briefly describe (from east to west) what is currently known about the paleoseismology of seven fault systems that offset Sehoo and younger deposits in our transect region of northwestern Nevada and northeastern California (Fig. 1). Four of these faults (Santa Rosa Range, southern Steens, Surprise Valley, and Black Rock fault zones) have been the subject of extensive investigations, including fault trenching. The other three faults (Desert Valley, Jackson Mountains, and Long Valley fault zones) have not been the subject of previous detailed investigations, so we concentrate our discussion on the results of our efforts on these structures.

We applied a variety of methods to map surface ruptures and determine deformation rates on faults in our transect. Where detailed mapping was not available, we mapped most of the latest Quaternary fault traces in the region at a nominal scale of 1:24,000 through analysis of aerial photographs and high-resolution, 1 m NAIP (National Agriculture Imagery Program; https://www.fsa.usda.gov/programs-and-services/aerial-photography/imagery-programs/) imagery, and limited field checking. The fault traces were then compiled at a scale of 1:100,000 for inclusion in Supplemental Figure S1 (see ^{footnote 1}), and later simplified for inclusion in Figure 1. We assumed that faults in this part of the Basin and Range undergo primarily normal displacement, so we used the vertical separation of correlative geomorphic surfaces as proxies for fault slip (Supplemental Fig. S2²). These values were obtained primarily from field measurements of topographic profiles using high-resolution GPS (RTK, real time kinematic) surveying equipment, supplemented with a few profiles measured across larger scarps using 10 m NED (National Elevation Dataset; http://ned.usgs.gov/) digital elevation models. Where multiple measurements of slip were available, we primarily chose the maximum slip values to compensate for possible underestimation of slip at depth (i.e., shallow slip deficit). In addition to published data, we derived additional timing information using optically stimulated luminescence (OSL; Supplemental Table S1³) and uranium-series (Supplemental Table S2⁴) dating techniques, supplemented by chemical fingerprinting and correlation of several tephra samples (Supplemental Table S3⁵) and regional correlations of pluvial lake shoreline sequences related to the Sehoo cycle of pluvial Lake Lahontan. Unless otherwise described, all ages are assumed to be equivalent to calendar ages before present (in ka) with 1σ uncertainties.

One unavoidable consequence of our focus on latest Pleistocene and younger deformation is that given the typical recurrence intervals of many Basin and Range faults, our selected time interval (<18–15 ka) for calculating slip rates is too short for some faults to contain a complete seismic cycle (two earthquakes). The time interval contains the elapse times that predate and postdate the event, but the uncertainties associated with calculating slip rates are likely higher than for faults that show evidence of multiple earthquakes in the latest Quaternary. Three of our structures (Santa Rosa Range, Desert Valley, and Long Valley fault zones) show evidence of a single earthquake and yield low slip rates with small uncertainties (<0.2 ± 0.03 mm/yr). The calculated errors probably underestimate the actual uncertainties, so we used comparisons with other faults in the region to assign a probable 1σ estimate of ±0.1 mm/yr to slip rates for these three faults. This value is consistent with a maximum estimate of 0.25 mm/yr based on the rates of two faults in our transect (Black Rock and Steens fault zones) that record more than one post-Sehoo earthquake. This rate would be reasonable estimate of maximum slip rate if the three single-event faults underwent a second surface rupture in the near future. A minimum slip rate of 0.05 mm/yr is consistent with longer term rates from trench data on the Santa Rosa Range fault zone (Supplemental Fig. S3⁶; modified from Personius and Mahan, 2005, Figs. 3 and 4 therein). We also increased uncertainties on the other faults in the transect to ±0.1 if their calculated rates were below this value.

We calculated regional GPS velocity differences across the transect using northing and easting data (in the UNAVCO North America 2008, NAM08, reference frame) downloaded from UNAVCO PBO (Plate Boundary Observatory) Velocity files (www.unavco.org), and calculated differential extension components using a regional extension direction (100°/280°) estimated from strikes of Holocene faults in the transect. We used our mapped surface rupture lengths and offset measurements to estimate paleoearthquake magnitudes (moment magnitude, M_w) using a weighted average of magnitude scaling relationships of Hanks and Kanamori (1979), Wells and Coppersmith (1994), and Hemphill-Haley and Weldon (1999).

The Santa Rosa Range fault zone (SRRFZ) forms piedmont and range-front fault scarps along the eastern margin of the Quinn River Valley and the western flank of the Santa Rosa Range in the northeast corner of the transect area (Fig. 1). The range is underlain by Triassic metasedimentary rocks intruded by Cretaceous granitic rocks, which in turn are unconformably overlain by Miocene volcanic rocks (Colgan et al., 2004, 2006b; Breuseke et al., 2008). Thermochronologic studies indicate unroofing and initial extension in the region ca. 12 Ma (Colgan et al., 2006b), and recent analysis of the tectonic geomorphology of the Santa Rosa Range indicates renewed uplift 5 Ma to 0.1 Ma (Ellis et al., 2014). Quaternary activity along the SRRFZ was described in Michetti and Wesnousky (1993), Narwold and Pezzopane (1997), Narwold (2001), Personius et al. (2004), and Personius and Mahan (2005). In the transect area, the SRRFZ consists of 2 primary sections or segments; a northern, 50-km-long Quinn River section that extends into the upper Quinn River valley in southern Oregon, and a southern, 42-km-long Santa Rosa Range section in the lower Quinn River Valley (U.S. Geological Survey Quaternary Fault and Fold Database of the United States; U.S. Geological Survey, 2015). The two sections are separated by an 11–12-km-wide right-stepping gap in latest Quaternary fault scarps.

Michetti and Wesnousky (1993), Narwold and Pezzopane (1997), and Narwold (2001) conducted geologic mapping, soils studies, and topographic scarp profiling along the Quinn River section of the SRRFZ. In the transect area (Fig. 1) the Quinn River section consists of a piedmont trace (Hot Spring Hills fault of Narwold, 2001) marked by fault scarps with surface offsets of 16–22 m in early to middle Pleistocene deposits and 1–3 m in latest Pleistocene to Holocene(?) deposits. Narwold (2001) used soil development and scarp profiles to calculate minimum vertical slip rates of 0.01–0.15 mm/yr, and concluded that the most recent earthquake (MRE) on the Quinn River section occurred ca. 10 ka and averaged 0.6 ± 0.3 m of vertical displacement. Michetti and Wesnousky (1993) noted that the MREs on the Quinn River and Santa Rosa Range sections appeared to have similar early Holocene ages, but could not determine whether the youngest ruptures on the two sections occurred during the same or different earthquakes. The large gap (11–12 km) between the 2 sections, and their combined length (>90 km) versus event displacement estimates of ≤2 m, lead us to infer that the 2 sections ruptured independently during separate latest Pleistocene and/or early Holocene earthquakes on this part of the SRRFZ.

The Santa Rosa Range section of the fault zone consists of two primary strands: a range-front strand that forms the steep western margin of the range, and a synthetic strand with considerably less total throw that offsets the basin floor 4–6 km west of the range-front trace. The synthetic strand may be the northern extension of the Bloody Run Hills fault zone, although the latter structure was not reactivated during the youngest rupture on the SRRFZ. In addition, a short (6 km long) normal fault appears to have connected the range-front trace of the SRRFZ and the northern end of the Bloody Run Hills fault zone during one or more pre-Holocene surface ruptures (Fig. 1). We include both strands in the Santa Rosa Range section because the trace of the basin floor strand mimics the strike and overall length of the range-front strand and both probably ruptured at the same time during the most recent surface-rupturing earthquake. Data from a trench on a splay of the range-front trace ∼5 km southeast of Orovada, Nevada (site SRR1 in Fig. 1), yielded evidence for 4 surface-rupturing earthquakes, the last 3 of which occurred since 141 ± 14 ka and caused surface offsets of ∼1.5–2.0 ± 0.5 m each (see Supplemental Fig. S3 [see ^{footnote 6}]; modified from Personius and Mahan, 2005, Figs. 3 and 4 therein). The basin-floor strand is marked by an ∼0.5-m-high scarp in sediments related to the highstand of Lake Lahontan (e.g., Michetti and Wesnousky, 1993; A.M. Michetti, 2001, written commun.). Here we present an OxCal (https://c14.arch.ox.ac.uk/) model of the chronological data (from Personius and Mahan, 2005) to better quantify earthquake ages (Supplemental Fig. S3B [^{footnote 6}]). Luminescence ages indicate that the youngest surface rupture on the Santa Rosa Range section occurred 13.5 ± 2.5 ka.

We combine the surface offset estimate of the youngest rupture in the Orovada trench (1.5 ± 0.5 m) and the surface offset of the basin floor trace (0.5 m), and use the regional age of the Lahontan highstand (15.6 ± 0.3 ka) to calculate a latest Quaternary slip rate (rounded to nearest 0.05 mm) of 0.15 ± 0.05 mm/yr (Table 1). This rate is similar to the estimated rates of slip on the Quinn River section of the SRRFZ (<0.2 mm/yr; Narwold, 2001), but are higher than late Quaternary interval and average rates based on the long (tens of thousands of years) recurrence intervals between the past three dated earthquakes at the Orovada trench site (0.03–0.05 mm/yr; Personius and Mahan, 2005). Thus we consider our latest Quaternary rate to be a maximum for the limited time window used in this study, and increase the uncertainty accordingly (i.e., ≤0.15 ± 0.1 mm/yr).

A comparison between longer term (hundreds of thousands to millions of years) measures of slip rate and the rates from Narwold (2001), Personius and Mahan (2005), and those presented herein leads us to infer that slip rates on the SSRFZ declined significantly in the late Quaternary. Longer term rate estimates include post–12 Ma exhumation rates of 0.3–0.5 mm/yr based on thermochronology (Colgan et al., 2006b), slip rate estimates of 0.3–0.9 mm/yr based on the heights of faceted spurs along the footwall of the SRRFZ (dePolo, 1998), and a post–0.5 Ma period of increased slip on the SRRFZ inferred from analysis of rates of knickpoint retreat in the Santa Rosa Range (Ellis et al., 2014). The long recurrence intervals (∼17–65 k.y., determined by Personius and Mahan, 2005) also support low rates of slip in the late Quaternary. We use vertical displacements of 2 m and probable fault rupture lengths of 42–50 km to estimate magnitudes (M_w) of 7.1–7.2 (Table 1) for surface-rupturing earthquakes on the Santa Rosa Range or Quinn River sections of the SRRFZ.

Little information is available on the paleoseismology of the Desert Valley fault zone (DVFZ). The fault was previously mapped by Dohrenwend and Moring (1991a, scale 1:250,000; U.S. Geological Survey, 2015). We mapped the fault zone from Mormon Dan Butte northward to Quinn River Lakes on the Quinn River for this study (scale of 1:24,000; summarized in Supplemental Fig. S1 [see ^{footnote 1}]), and although we identified additional fault strands, the basic form of the fault zone is similar to that shown in the U.S. Geological Survey (2015) database. The ∼65-km-long fault zone extends from the subdued western margin of the southern Slumbering Hills northward to the vicinity of Gabica Butte, and then across the floor of northern Desert Valley and southern Kings River Valley. Other than the small scarps that mark the trace in latest Quaternary surficial deposits, the fault zone shows little topographic expression indicative of repeated Quaternary activity. However, limited drill hole and gravity data indicate that the fault scarps are coincident with the steep eastern margin of Paleogene–Neogene basin-fill deposits that reach thicknesses of 1.2 km beneath the valley floor directly west of the southern end of our mapped trace (Sinclair, 1962; Willden, 1964; Berger, 1995; Ponce and Plouff, 2001). Basin-fill sediments thin across a bedrock high that extends across the valley from the Jungo Hills to Mormon Dan Butte, and subdivides the basin fill into two depocenters; the southern subbasin is filled with at least 2 km of basin-fill deposits. The restriction of scarps in surficial deposits to the northern subbasin indicates that the bedrock high probably controlled earthquake rupture extent in the MRE, and probably has done so for much of the history of the fault.

The surface expression of the DVFZ is characterized by single (Fig. 2A) and multiple small (∼0.5–2.5-m-high), primarily down-to-the-west, fault scarps in deposits of the Sehoo cycle of pluvial Lake Lahontan and eolian and alluvial deposits that postdate the ca. 15.6 ka age of the Sehoo highstand (Figs. 1 and 2A). The consistent down-to-the-west aspect and formation of a continuous graben (Fig. 2B) that extends for at least 5 km between Corbeal Butte and Mormon Dan Butte indicate primarily normal displacement along the DVFZ.

We also measured scarp topographic profiles in the field primarily along the reach of the fault between Gabica Butte and Corbeal Butte where the youngest deformation is confined to a single trace (Figs. 1 and 2). Scarps along this section have consistent vertical separations (surface offsets) of 2–2.5 m; scarps apparently decrease in size to the north and south, but multiple fault traces to the north, and the presence of the extensive graben (site DV2 in Figs. 1 and 2) and burial of the fault with several meters of windblown sand to the south make accurate determination of vertical separation difficult in these areas (Fig. 2C; Supplemental Fig. S2 [^{footnote 2}]). We found no evidence indicative of multiple latest Quaternary displacements, so we infer that the faults scarps on the DVFZ were formed during a single large surface-rupturing earthquake that postdated final retreat of the waters of Lake Lahontan below an elevation of ∼1280 m in Desert Valley. Our largest vertical separation value, from a limited number of scarp profiles (2.4 ± 0.5 m, site DV1 in Fig. 1), and the age of the Sehoo highstand (15.6 ± 0.3 ka) yield a latest Quaternary slip rate of 0.15 ± 0.03 mm/yr. Given our assumption that the scarps are the result of a single event and the lack of preexisting fault topography, we consider this value to be a maximum slip rate, and increase the uncertainty to ±0.1 mm/yr. The fault length of 65 km (U.S. Geological Survey, 2015) and a vertical displacement of 2.4 m indicate an estimated magnitude of M_w 7.2 for this event (Table 1).

Little information is available on the paleoseismology of the Jackson Mountains fault zone (JMFZ). The footwall consists of Paleozoic and Mesozoic metasedimentary and metavolcanic rocks intruded by Triassic and Jurassic plutons and overlain by Cretaceous sediments and Paleogene–Neogene volcanic rocks (Maher, 1989; Quinn et al., 1997). Thermochronologic studies indicate unroofing and initial extension in the region ca. 12 Ma (Colgan et al., 2006b), and recent analysis of the tectonic geomorphology of the Jackson Mountains indicates renewed uplift from 5 Ma to 0.1 Ma (Ellis et al., 2014). Parts of the fault are included in reconnaissance mapping by Dohrenwend and Moring (1991b, scale 1:250,000) and despite significant topographic relief (1–1.5 km) and a very steep range front, the fault is only classified as Quaternary in compilations (dePolo, 2008; U.S. Geological Survey, 2015). To better define the Quaternary history of the fault zone, we mapped it from Quinn River Crossing southward ∼57 km to the latitude of Rattlesnake Canyon (Supplemental Fig. S1 [see ^{footnote 1}]). We also conducted reconnaissance field studies to check our mapping, obtained detailed topographic data on fault scarps and lacustrine shoreline features, examined natural and soil pit exposures, and collected samples for chronological analysis.

Our mapping revealed abundant evidence for multiple latest Pleistocene and Holocene surface ruptures along most of the western margin of the Jackson Mountains. Youthful scarps that postdate the Sehoo cycle of pluvial Lake Lahontan are well expressed where all or part of the fault diverges from the range front onto piedmont slopes underlain by lacustrine and post-lake alluvial deposits. In contrast, scarps along the steep range front are difficult to recognize on high-resolution imagery and in the field. The latter scarps are poorly expressed for the following reasons: (1) the range-front traces commonly are located on steep slopes at the bedrock-colluvial contact; (2) much of the mountain front consists of cliff-forming, very resistant rock types (basalt, dense limestone); (3) the long fetch across the Black Rock Desert exposed the range front to extensive erosion during highstands of Lake Lahontan; and (4) Lahontan shorelines are commonly parallel to and occupy the same elevations as the fault trace. These factors result in poor conditions for preservation of fault scarps along the range-front trace, and thus likely account for the failure of previous investigations to recognize the recency of faulting along the western flank of the Jackson Mountains.

We conducted detailed field studies at three sites (Fig. 1; Supplemental Fig. S1 [see ^{footnote 1}]): (1) a deeply incised canyon at the mouth of an unnamed drainage (herein Sand canyon) on the northern part of the fault zone (site JM1); (2) a Lahontan shoreline complex at the mouth of Hobo Canyon on the central part of the fault zone (site JM2); and (3) an extensive shoreline complex at the mouth of a major unnamed canyon system draining the western flank of King Lear Peak on the southern part of the fault zone (site JM3).

Detailed surficial geologic mapping and topographic profiling helped us document extensive Sehoo lake cycle sedimentation and subsequent post–Sehoo highstand incision that created a suite of terraces and exposed two fault traces near the mouth of an unnamed canyon on the northern part of the JMFZ (site JM1 in Fig. 1). As is common elsewhere, the JMFZ at this location is characterized by piedmont and range-front strands (Fig. 3). Unfortunately, the range-front trace is coincident with and thus obscured by the highstand of Lake Lahontan, but the piedmont strand is exposed in two places in the walls of Sand canyon. The western fault exposure in the north wall (Fig. 4A) revealed a footwall sequence of fan alluvium capped by a tufa-cemented boulder gravel that likely formed as beachrock at an elevation of ∼1310 m during the rapid rise to or fall from the Lahontan highstand (elevation ∼1333 m). The upper part of the footwall is covered by several meters of poorly exposed eolian sand, alluvium, and colluvium. The base of the hanging wall consists of the same alluvial stratigraphy and beachrock as the footwall. Overlying the faulted beachrock in the hanging wall are a thin eolian sand and two deposits of poorly sorted fault-scarp colluvium. The lower, thicker colluvium is faulted, whereas the upper, thinner colluvium appears to be in depositional contact with an eroded fault scarp free face. The upper part of the hanging wall is covered by the same poorly exposed eolian sand, alluvium, and colluvium as the footwall. Several correlative stratigraphic contacts indicate total vertical displacement of 1.5 m since formation of the beachrock, but determination of the number of earthquakes responsible for this displacement is complicated by incomplete exposure of the upper 2–3 m of the sequence, the obliquity of the fault with the canyon wall (∼30°), and probable cutting and filling during terrace formation. However, projection of the fault to the southern flank of the canyon places the fault trace on a 40-cm-high scarp in the t2 terrace surface (Fig. 3B); the faulting event that formed this scarp cannot account for the 1.5 m of total displacement we measured in the exposure and thus indicates at least two post-Sehoo surface ruptures on this strand. This interpretation is supported by the presence of the two fault-scarp colluvial deposits in the hanging wall.

The second, eastern fault exposure in the south wall at an elevation of 1310 m revealed a sequence of fan gravels in the footwall, faulted against one or two deposits of probable fault-scarp colluvium and an intervening 1- to 2-m-thick sequence of very crudely bedded silty and sandy alluvium in the hanging wall (Fig. 4B). We interpret the latter deposit, which is not present in the footwall, as a fill deposit related to enhanced fluvial erosion and deposition along a preexisting fault scarp. Measurement of total vertical offset across this fault splay is problematic. The apparent vertical offset of the t2 terrace surface across the fault scarp west of the exposure is 1.7 m, but our interpretation of extensive cutting and filling prior to the MRE complicates measurement of total throw across this strand. We conclude that the presence of two ∼1-m-thick colluvial wedges and the absence in the hanging wall of sediments correlative to the fan deposits in the footwall are consistent with an estimated minimum vertical displacement of >1.5 m to as much as 4 m, if we use double the thickness of the colluvial wedges as an estimate of slip (e.g., McCalpin, 2009) on this strand. Taken together, fault-scarp profile data and our interpretations of exposures of the two known fault strands in the walls of Sand canyon yield evidence for at least two surface-rupturing earthquakes and an estimated minimum vertical displacement of >3–5.5 m since retreat of Lake Lahontan ca. 15.6 ka.

We used three sources of chronological data to refine the ages of the earthquakes identified at the Sand canyon site. (1) The Lake Lahontan hydrograph (Fig. 5) indicates that at the elevations of the fault exposures (1310–1315 m), shoreline features date to within a few hundred years of the age of the Sehoo highstand (ca. 15.6 ± 0.3 ka). (2) A >1-m-thick deposit of volcanic ash that chemically correlates with the Mazama ash (7.63 ± 0.15 ka; Zdanowicz et al., 1999; Supplemental Table S3 [^{footnote 5}]) is preserved in the south wall of Sand canyon a few hundred meters upstream of our fault exposures (Fig. 3C; Supplemental Fig. S4⁷). The elevation of this fluvially reworked deposit above the modern channel is similar to a remnant of terrace t2 cut in the north wall directly across the modern channel. The fact that the deposit is very thick, but is currently cropping out from a steep canyon wall exposure, leads us to interpret that the ash was probably deposited during or shortly after formation of a t2 terrace remnant now largely eroded from the south wall of the canyon. A similarly thick deposit of ash incorporated into fluvial sediments at about the same height above the modern channel is present a few hundred meters upstream of the mouth of an unnamed canyon ∼3.5 km north of Sand canyon (Fig.1, site JM1b). (3) We obtained an OSL age of 3.75 ± 0.13 ka (Supplemental Table S1 [see ^{footnote 3}]) on a deposit of what we interpreted to be eolian sand that underlies the older wedge of fault scarp colluvium and directly overlies the tufa-cemented boulder gravel in the footwall of the western exposure. Given its position overlying a Lahontan beach deposit and underlying several meters of fan alluvium that appears to predate the formation of the ca. 8–7 ka t2 terrace, we suspect the OSL age is anomalously young by several thousand years (the sample may have contained burrowed sediment, rather than intact eolian sand).

From this limited data set, we infer the following at the Sand canyon site. (1) A minimum of 2 earthquakes caused total vertical offset of at least 3–5.5 m sometime after 15.6 ± 0.3 ka on the piedmont strand of the JMFZ. (2) The lower colluvial wedge in the western fault exposure in places directly overlies tufa-cemented beachrock, suggesting that this earthquake occurred shortly after the final retreat of Lake Lahontan. (3) Deposition of the P2(?) colluvium in the eastern fault exposure predates erosion and deposition of fluvial sediments that may be correlative with the Mazama ash–bearing terrace t2, and thus may correlate with the older (pre-t2) earthquake in the western exposure 1. The MRE in the western and eastern exposures formed 0.4-m-high and 1.5-m-high scarps, respectively, in the t2 terrace surface, and thus likely postdates the age of the Mazama ash (younger than 7.6 ka).

A well-preserved suite of Lake Lahontan shorelines is present in an embayment coincident with a prominent right step in the JMFZ at the mouth of Hobo Canyon (site JM2 in Fig. 1). The fault zone consists of three strands that offset multiple shorelines of the Sehoo lake cycle and alluvial sediments that postdate retreat of the lake (Fig. 6). The middle strand is well exposed in a gully as a well-defined gouge and shear zone in Jurassic basaltic andesite (Maher, 1989) with a strike of N40°E, dip of 56°W. Two exposures of the western strand in basaltic andesite have similar attitudes (N10W, 63°W; N5E, 59°W; Fig. 6). Field studies at the site included detailed mapping and topographic profiling. Results include the following: (1) fault scarps with vertical offsets of 10 m or more are present in pre–Sehoo highstand alluvium on the eastern strand north of Hobo Canyon; (2) vertical offset of the shoreline complex is ∼7.6 m as summed across the western (3.7 m) and middle fault strands (3.9 m); and (3) comparison of multiple topographic profiles and the presence or absence of well-rounded lacustrine clasts indicate apparent vertical offset 7.5 ± 1 m of the Sehoo highstand across the middle fault strand (Fig. 7). These two vertical separation estimates represent the highest slip values we measured in Sehoo-aged deposits along the entire length of the JMFZ, and may indicate the apex of post-Sehoo displacement along the JMFZ.

Extensive deposits of Lake Lahontan (herein referred to as the King Lear Peak shoreline complex, KLP) are preserved at the mouth of a major unnamed canyon system draining the western flank of KLP on the southern part of the JMFZ (site JM3 in Fig. 1). The range-front fault trace trends southeast in a 2–3-km-wide left step that creates a large embayment in the range front at the latitude of the KLP shoreline complex (Fig. 8). We found only sparse evidence of young faulting along the range front southeast of this step, but our mapping indicates that latest Quaternary faulting steps onto a piedmont fault zone that cuts the shoreline complex and extends southwestward for at least 15 km. The piedmont strand cuts a suite of Sehoo lake cycle shoreline platforms, two of which we correlate across the fault trace. These two platforms, herein informally named the Tufa shoreline and Desert Pavement (DP) shoreline, are both vertically displaced ∼5–6 m (Fig. 9). The Tufa shoreline occupies an elevation range of ∼1234 m (hanging wall) to ∼1242 m (footwall), and is characterized by a broad platform and abundant cementation of carbonate tufa deposits that may be unique in this part of the Black Rock Desert. The DP terrace occupies an elevation range of 1247 m (hanging wall) to 1253 m (footwall) and is characterized by a well-developed desert pavement.

We used OSL and uranium-series methods to determine the ages of sediments underlying these data (Fig. 10; Supplemental Tables S1 [^{footnote 3}] and S2 [^{footnote 4}]). OSL samples from a well-sorted sand deposited near the top of the lacustrine sequence underlying the Tufa terrace yield ages of 18–17 ka (Fig. 10A). Uranium-series ages of ca. 22 ka on tufa formed in the upper part of the sequence indicate a somewhat older age, but the presence of substantial amounts of detrital Th yielded broad uncertainties that overlap with the OSL ages at 2σ (Fig. 10B). We prefer the much lower uncertainties of the OSL ages, and infer from the Lahontan hydrograph (Fig. 5) that the sediments underlying the Tufa terrace were deposited during the rapid rise of Lake Lahontan to the Sehoo highstand. We were unable to obtain a sufficient sample of appropriate grain size from the lacustrine sediment underlying the DP terrace, but samples from a weakly developed soil formed on the lacustrine sequence and an overlying eolian deposit yield OSL ages of 12–11 ka (Fig. 10C). We interpret these ages as indicating that the DP terrace may have formed during the rapid retreat from the Sehoo highstand. We interpret the consistent vertical displacements of ∼6 m of these 2 latest Pleistocene data as evidence of at least 2 and perhaps 3 or more surface-rupturing earthquakes on the southern strand of the JMFZ since retreat from the Sehoo highstand.

In addition to our three primary study areas, we found evidence of latest Quaternary earthquakes at several other sites along the JMFZ. Approximately 4 km south of Sand canyon near the mouth of Deer Creek Canyon (site JM4 in Fig. 1), erosion from a blown out drainage canal exposed the upper 2 m of the range-front fault in pre-Sehoo alluvial-fan sediments (see Supplemental Fig. S5⁸ for detailed description of this exposure). We mapped two colluvial wedges separated by a weak buried soil in this exposure, and interpreted two events: (1) a younger (middle to late Holocene) surface-rupturing earthquake (MRE) with a vertical separation of 1–1.5 m, and (2) a penultimate earthquake rupture of unknown size that may be latest Pleistocene or early Holocene in age. Our estimate of 1–1.5 m of vertical separation during the MRE suggests that the 4–5-m-high scarps in these pre-Sehoo fan sediments record a total of 3 or 4 surface ruptures along this part of the range-front trace in the late Quaternary.

Measurement of total slip near Deer Creek is complicated because ∼1.5 km to the southwest, a prominent fault trace splits away from the range front and extends northward for ∼6 km before rejoining the range front just north of Sand canyon. At the latitude of the Deer Creek exposure, the piedmont trace is marked by a small (vertical separation of 1.4 m) scarp in middle(?) Holocene alluvium, and larger (vertical separation of 3.7 m) scarps in Sehoo-aged lacustrine sediments. We interpret that the smaller scarp formed during the MRE and thus its size likely indicates that the larger scarps along the piedmont trace are the result of two or three surface-rupturing earthquakes of similar size. We conclude from this evidence that total post-Sehoo vertical slip across the JMFZ at the latitude of Deer Creek is 5–7 m.

Our investigations consistently show evidence of multiple post-Lahontan surface rupturing earthquakes along the 57-km-long JMFZ. Multiple fault traces and poor preservation of faulted surficial deposits along the steep range front complicate measurement of total post-Sehoo vertical slip, but we measured total slip of 5–7.5 m by a variety of methods from the KLP shoreline complex in the south to a site ∼9 km north of Sand canyon near the northern end of the fault zone at Quinn River Crossing. At the southern end, total slip is partitioned on two splays: the eastern splay offsets the Lahontan highstand shoreline ∼1 m, and the western splay offsets post-Lahontan, middle to late(?) Holocene fan deposits ∼0.5 m (Fig. 1; Supplemental Fig. S2 [^{footnote 2}]). The number of surface-rupturing earthquakes necessary to cause vertical separations of 5–7.5 m is difficult to quantify, but the fault exposures at Sand canyon (Fig. 4) and Deer Creek (Supplemental Fig. S5 [^{footnote 8}]) indicate a minimum of 2 earthquakes that likely postdate the Lahontan lake cycle. The size of individual surface displacements is complicated by the presence of piedmont and range-front fault traces, but individual strands probably have displacements of 1–2 m and collectively 2–3 m per event. We conclude that the JMFZ has undergone at least 2 and perhaps 3 surface rupturing earthquakes with estimated magnitudes of M_w 7.1 (3 event scenario) to M_w 7.2 (2 event scenario) in the past ∼15 k.y. (Table 1). Our largest post-Sehoo offset measurements (7.5 m at Hobo Canyon) and the age of the Sehoo highstand (15.6 ± 0.3 ka) yield a latest Quaternary slip rate of 0.50 ± 0.1 mm/yr.

The three easternmost faults in our transect (SRRFZ, DVFZ, and JMFZ) are spaced ∼30 km apart; such spacing is typical of many parts of the Basin and Range (Stewart, 1971, 1978; Fletcher and Hallet, 1983). In contrast, a gap nearly 3–4 times as wide exists between the JMFZ and the Long Valley fault zone at the latitude of our PBO transect. However, evidence of multiple latest Quaternary paleoearthquakes on the southern Steens (SFZ; Pueblo Mountains fault of dePolo, 1998) and the northern Black Rock (BRFZ) fault zones suggests to us that latest Quaternary extension occurred 30–40 km west of the JMFZ in the gap between these two fault zones (Fig. 1; Supplemental Fig. S1 [^{footnote 1}]). Evidence for such deformation includes numerous north-northeast–striking normal faults that offset Paleogene–Neogene bedrock (Stewart and Carlson, 1978; Stewart, 1980; Colgan et al., 2006b; Lerch et al., 2008) and in a few places offset Quaternary deposits (Dohrenwend and Moring, 1991b; dePolo and Dee, 2015; U.S. Geological Survey, 2015). In Personius et al. (2007b) it was inferred that the SFZ is kinematically linked to the north-northeast–striking Alder Creek fault of Colgan et al. (2006a), which would extend the SFZ an additional 32 km to the southwest. We infer that the latest Quaternary deformation from the SFZ and the northern BRFZ likely influences the modern stress field across our transect, so we include these fault zones in our analysis.

The paleoseismology of the southern SFZ was described in detail in Personius et al. (2006, 2007b). Data from a trench across a single 4–5-m-high east-facing scarp at the Bog Hot Valley site (site ST1 in Fig. 1) yielded evidence for three surface-rupturing earthquakes since ca. 18 ka (Supplemental Fig. S6⁹). An additional 9-km-long zone of west-facing, 0.5–2-m-high fault scarps is present along the southeastern margin of the Bog Hot Valley, but their short length, position adjacent to the mapped trace of the SFZ, and scarp morphology indicate that they likely represent triggered slip on a buried fault antithetic to the SFZ. Radiocarbon and luminescence dating were used to estimate the following times of the past three surface-rupturing earthquakes (Personius et al., 2007b): the MRE (ST1) occurred 4.6 ± 1.0 ka, the penultimate earthquake (ST2) occurred 6.1 ± 0.5 ka, and the oldest earthquake (ST3) occurred 11.5 ± 2.0 ka; a new OxCal model (Supplemental Fig. S6B [^{footnote 9}]) of the chronological data (from Personius et al., 2007b) yields similar ages of 4.1 ± 0.2, 6.1 ± 0.3, and 12.7 ± 2.6 ka, respectively, for these same earthquakes (Table 1). Total vertical offset of the lacustrine sequence exposed in the Bog Hot Valley trench (4.4 ± 0.2 m) and similar surface offsets from scarp profiles along the southern 10–15 km of the fault (Personius et al., 2006) yield an average vertical offset of ∼1.5 m for the last 3 surface ruptures. Vertical offsets of 1.5 m and a likely fault rupture length of 42 km yield a magnitude estimate of M_w 7.0 (Table 1) for these earthquakes. We use an average stratigraphic offset of 4.5 ± 0.2 m and a composite luminescence age from near the top of the lacustrine sequence in the area (18.0 ± 1.1 ka; Personius et al., 2007b) to calculate an average latest Quaternary slip rate of ∼0.25 mm/yr. This rate is consistent with latest Quaternary interval rates based on the past three dated earthquakes at the Bog Hot Valley site, as well as from offset of Miocene bedrock at the northern end of the valley (Personius et al., 2007b).

The paleoseismology of the BRFZ was the subject of a doctoral thesis by Dodge (1982), whose work was summarized in U.S. Geological Survey (2015). The fault therein is mapped on the basis of mapping of Dodge (1982, scale 1:24,000) and 1:250,000-scale mapping of Dohrenwend et al. (1991) and Dohrenwend and Moring (1991b). Dodge (1982) mapped and described a broadly arcuate, 55-km-long fault zone consisting of numerous right- and left-stepping fault scarps in late Pleistocene lacustrine (Lake Lahontan) deposits and post-lacustrine alluvial sediments of Holocene age. Extensive scarp topographic profiling and the results from seven trench investigations were used to determine that the fault zone had undergone at least four surface-rupturing earthquakes since deposition of the Wono tephra. Dodge (1982) used the 4 identified earthquakes and an age of 25 ka of the Wono tephra to calculate an average recurrence interval of 6250 yr; a revised, better constrained age of the Wona tephra (ca. 34 ka; Benson et al., 2013; Reheis et al., 2014) yields a slightly longer (8500 yr) average recurrence estimate. None of the Dodge (1982) trenches exposed evidence of all four earthquakes, indicating either that not all earthquakes ruptured the entire fault trace, or erosion during the Lahontan lake cycle in places completely removed existing fault scarps. Other features of interest exposed in the Black Rock trenches include evidence of liquefaction, and the presence of slickensides in at least one trench that indicate primarily normal displacement.

Dodge (1982) did not calculate slip rates for the Black Rock fault zone. Calculations of average slip rates are complicated by lack of exposure of a complete, four-event stratigraphic record in any of the Black Rock trenches. Measurements of individual earthquake displacements are sparse: 4 measurements of MRE (PE1) ranged from 1 to 3.5 m of vertical offset, and a single measurement of PE2 was 1.8 m. Total cumulative offset for the last 4 earthquakes is unknown, but we use two measures to estimate this value: (1) the largest surface offset of a compound (multievent) scarp measured by Dodge (1982) is 8.3 m, and (2) an assumed average offset of 2 m for each of the 4 documented earthquakes yields an estimated total cumulative offset of 8 m. We used these 2 offset values to calculate an estimated latest Quaternary (post–34 ka) slip rate of ∼0.25 mm/yr. Average displacements of ∼2 m and the complete length of Holocene surface rupture (57 km) yield magnitude estimates of M_w 7.1 for the last 4 surface-rupturing earthquakes on the Black Rock fault zone. The results of existing studies of the SFZ and the BRFZ produce essentially identical post-Sehoo slip rates of 0.25 mm/yr, so we project this value across our transect 30–50 km west of the JMFZ (Fig. 1).

The only significant topographic expression of post-Miocene faulting between the Steens–Black Rock extensional zone and Surprise Valley in northeastern California is Long Valley, a basin controlled by north-, northeast-, and northwest-striking normal faults mapped as Quaternary (younger than 1.6 Ma) in U.S. Geological Survey (2015). None of these faults offset pluvial lake sediments and landforms along the margins of Long Valley. However, numerous northeast-striking fault strands were mapped across the floor of Long Valley (Fig. 1; Supplemental Fig. S1 [see ^{footnote 1}]) by Dohrenwend and Moring (1991b). These faults are herein informally named the Long Valley fault zone (LVFZ). Little information is available on the paleoseismology of this structure. The fault was previously mapped by Dohrenwend and Moring (1991b, scale 1:250,000; U.S. Geological Survey, 2015). We mapped the fault at a scale of 1:24,000 for this study; although additional strands have been identified, the basic form of the fault zone (Supplemental Fig. S1 [^{footnote 1}]) is similar to that shown in the U.S. Geological Survey (2015) database. The fault zone consists of four en echelon, left-stepping, and northeast-striking fault strands or arrays of small (generally <1 m high) northwest- and southeast-facing scarps (Fig. 1; Supplemental Fig. S1 [^{footnote 1}]). The two northern strands (Alkali Lake and Fortymile Lake) are confined to the floor of central Long Valley, but the two southern strands (Central Lake and Boulder Lake) extend from the floor of southern Long Valley southward across a series of unnamed basalt-cored hills that form the southern margin of the Long Valley basin.

Although assumed to be a normal fault, the left-stepping pattern and in many cases very linear fault traces are suggestive of strike-slip faulting such as the 1932 Cedar Mountain earthquake sequence in central Nevada (Bell et al., 1999). Given the strong northwest strikes of faults composing the right-lateral Walker Lane west of Long Valley, the sense of slip of the northeast-striking Long Valley fault would likely be left lateral. We searched for geomorphic indicators of lateral slip in surficial deposits, but found no dominant patterns of offset channels or other features. One possible indicator of left-lateral slip is a series of right steps and bends in the very linear Boulder Lake strand north of State Road 34 (Figs. 11A, 11B). The scarps increase in height at each of these bends; at the bend ∼500 m north of State Road 34 (site LV1 in Fig. 1), the vertical separation doubles, from an average of 0.3 m to 0.6 m. We infer that such features are pushups in zones of compression in right-stepping (constraining) bends of a left-lateral fault zone. A much longer term sense of slip is evident where the Boulder Lake strand cuts Miocene (16–15 Ma) basalt flows south of the southern margin of Long Valley (site LV2 in Fig. 1). Here, the fault offsets a northwest-striking normal fault zone in bedrock in a left-lateral sense ∼340 ± 50 m, and in a down-to-the-east normal sense 40–45 m (Figs. 11C, 11D). These offset estimates yield a horizontal:vertical ratio of ∼8 ± 1.5:1.

The Long Valley basin was occupied by pluvial Lake Meinzer during the late Pleistocene (Snyder et al., 1964; Mifflin and Wheat, 1979; Reheis, 1999). Little is known about the history of this lake, which reached a highstand at 1768 m, filling the basin to a depth of ∼90 m. The only available chronological data pertinent to the history of Lake Meinzer are chemical correlations of two tephra we observed in a stream cut in the valley floor located within the Alkali Lake set of faults (site LV3 in Fig. 1) at an elevation (1685 m) only 3 m above the current low point in the Lake Meinzer basin (Figs. 11E, 11F). The two tephra were positively identified as the Wono and Trego Hot Springs tephras (Supplemental Table S3 [^{footnote 5}]), dated elsewhere as ca. 34 ka and 30 ka, respectively (Benson et al., 2013). Our interpretation of the stratigraphy at the Alkali Lake tephra site (Supplemental Fig. S7¹⁰) is consistent with the depositional settings of other Wono and Trego Hot Springs localities in the Lahontan basin (Fig. 5; Adams, 2010). We infer from this agreement that Lake Meinzer likely has a hydrograph similar to Lake Lahontan and nearby isolated lakes such as Lake Surprise and Lake Alvord, and thus reached its highstand at about the same time (ca. 18–15 ka).

Numerous fault scarps along the LVFZ are preserved in silty lacustrine sediments on the floor of Long Valley and some also cut the highstand shoreline, thus indicating that the fault ruptures date to <18–15 ka. In addition, the majority of fault scarps are at low elevations (1685–1695 m) relative to the 1768 m elevation of the Lake Meinzer highstand. We found no obvious evidence of reworking of these scarps by lacustrine shoreline processes, which leads us to infer that the timing of the earthquake responsible for the LVFZ scarps also postdates the age of the last major transgression (Younger Dryas–Gilbert) during which the level of Lake Lahontan rose ∼60 m in the time period 13–11 ka (Fig. 5). Lakes in the western Great Basin regressed from this intermediate highstand to current elevations ca. 10 ka (Benson et al., 2013; Reheis et al., 2014), so we infer that the youngest surface rupturing earthquake on the LVFZ occurred after 10 ka.

To help quantify slip along the LVFZ, we measured scarp topographic profiles in the field across three of the four primary strands (Fig. 1; Supplemental Fig. S2 [^{footnote 2}]). Most individual scarps are small (<1 m), as expected in a distributed fault zone with a significant strike-slip component. The northern (Alkali) strand, which consists of curvilinear fault traces perhaps indicative of a more normal sense of slip, yielded a composite vertical separation of ∼2.5 m (north side down) across the three scarps that compose the northernmost part of the strand (Figs. 11E, 11F). Scarps along the Boulder Lake strand have much smaller vertical separations (∼0.3 m), but the lateral component, although unknown, is probably larger. If the long-term horizontal:vertical slip ratio we measured in Miocene bedrock is still characteristic of slip in the Quaternary, then we estimate that lateral offset along this strand is ∼2.4 ± 0.5 m. We use these 2 offset values to determine a slip value of 2.5 ± 0.5 m as our best estimate of the total post-highstand slip on the LVFZ, and the age of the Lahontan (Sehoo) highstand (15.6 ± 0.3 ka) to calculate a slip rate of 0.15 ± 0.05 mm/yr. Given our assumption that the scarps are the result of a single event and the lack of preexisting fault topography in Quaternary deposits, we consider this value to be a maximum slip rate, and increase the uncertainty to ±0.1 mm/yr. Our vertical slip value and a rupture length of 40 km yields a magnitude estimate of M_w 7.0 for the Long Valley earthquake (Table 1).

The prominent east-dipping normal Surprise Valley fault zone (SVFZ) forms the western margin of the Basin and Range in northeastern California (Fig. 1). The fault separates Eocene to Pliocene volcanic, volcaniclastic, and sedimentary rocks in the Warner Mountains from 1–2-km-thick basin-fill sediments beneath Surprise Valley with a total dip-slip displacement of 7–8 km (Colgan et al., 2006b; Egger and Miller, 2011). Hedel (1980, 1984) mapped and documented evidence of Quaternary faults in Surprise Valley. Bryant (1990) reevaluated and modified the mapping of Hedel (1980, 1984), and attributed some of Hedel’s faults to nontectonic origins (i.e., lacustrine shorelines). The paleoseismology of the SVFZ was first described in detail in Personius et al. (2007a, 2009). Ongoing studies are focused on utilizing newly acquired lidar (light detection and ranging) data for detailed mapping and paleoseismic analysis (e.g., Egger, 2014).

Surprise Valley was occupied by pluvial Lake Surprise, which reached an elevation of 1530–1545 m (Hedel, 1980, 1984; Irwin and Zimbelman, 2012; Ibarra et al., 2014) at its latest Pleistocene highstand (15.2 ± 0.2 ka; Ibarra et al., 2014). Data from a trench and borehole at the Cooks Canyon site (elevation 1470–1478; site SV1 in Fig. 1) and from a natural exposure at the Steamboat Canyon site (elevation 1493 m; site SV2 in Fig. 1) yielded evidence for five surface-rupturing earthquakes since deposition of a Lake Surprise deltaic complex that dates to ca. 18 ka (Supplemental Fig. S8¹¹; Personius et al., 2009; Ibarra et al., 2014). Vertical stratigraphic offset of distinctive facies of the faulted deltaic sediment in the trench is 13.5 ± 2 m (estimated 2σ uncertainties). Additional slip of ∼1.5 m on synthetic fault scarps on the valley floor (Hedel, 1984; Bryant, 1990) yields a combined vertical slip of 15 ± 3 m since deposition of the deltaic complex; topographic profiles on scarps across similar-aged deposits indicate slip of similar magnitude at several other locations along the fault zone (Fig. 1; Hedel, 1980, 1984; Personius et al., 2007a). If the paleoseismic record from the Cooks Canyon trench is complete, then the 5 dated surface ruptures averaged ∼2.7 m per earthquake. We use a vertical displacement of 2.5 m and a fault rupture length of 75 km to calculate estimated magnitudes of M_w 7.3 for earthquakes that rupture the entire length of the SVFZ. The timing of these earthquakes is constrained by radiocarbon, luminescence, and correlated tephra ages; the MRE is well constrained at both the Cooks Canyon and Steamboat Canyon sites as ca. 1.2 ± 0.1 ka (Supplemental Fig. S8B [^{footnote 11}]). Our total offset estimate (15 ± 3 m) and a luminescence age on the youngest lacustrine deposits exposed in the Cooks Canyon trench (18.4 ± 1.6 ka) yield a latest Quaternary slip rate of 0.80 ± 0.1 mm/yr.

Our rate is slightly lower than a recently determined rate of ∼1.1 mm/yr from the central part of the SVFZ, based on detailed shoreline and fault scarp mapping on recently obtained lidar imagery (Marion, 2016). In addition, Marion (2016) also determined lower slip rates on the southern (∼0.4 mm/yr) and northern (0.3 mm/yr) parts of the fault zone. Taken together, our rate of 0.8 ± 0.1 mm/yr may be a reasonable average estimate of slip along the total length of the SVFZ.

Our focus on faults that have undergone at least one post-Sehoo surface-rupturing earthquake excludes several faults in the transect inferred to have been active in the Quaternary. These include the Pine Forest Range fault zone, numerous faults in Miocene volcanic rocks in the northwest Nevada plateau (Sheldon Plateau of Lerch et al., 2007), and the Hays Canyon Range fault zone in Surprise Valley (Fig. 1). None of these structures has demonstrated evidence of faulting in latest Quaternary deposits. A more detailed discussion of these features is presented in Supplemental Text File S1¹².

Our compilation of new and existing paleoseismic data identifies a total of 18 probable latest Quaternary, large magnitude (M_w ≥ 7) surface-rupturing earthquakes in the NWBR transect region (Table 1; Fig. 12). Recurrence intervals on individual faults with multiple post–18–15 ka earthquakes range from 3.7 k.y. on the margin-bounding SVFZ to 8.5 k.y. on the BRFZ; recurrence intervals on faults with only a single post-Sehoo earthquake are unknown, but likely exceed the length of our time window (>15 k.y.). The entire region (Fig. 1) has on average one such event per thousand years.

We chose the longest continuous geodetic records available (the PBO network) to calculate rates of horizontal deformation across the transect region. From east to west, we chose PBO stations P013, P145, and P731, because they are permanent, robust installations with long (∼7–9 yr) records of continuous operation (Fig. 1). We first calculated the total values of differential velocity (Vd) and direction between these stations (3.8 ± 0.1 mm/yr at 331°, NAM08 reference frame), and then calculated the extension-only component (2.4 ± 0.1 mm/yr at 280°) using an estimate of the regional extension direction (100°/280°) based on the strikes of the seven Holocene-active faults in the transect area (Tables 1 and 2). Our estimated regional extension direction is very similar to those derived from analysis of regional GPS (e.g., 104°/284°; Payne et al., 2012) and fault strike data and is also consistent with the few published earthquake focal mechanisms in the region (Patton and Zandt, 1991; Pezzopane and Weldon, 1993).

The ∼50° more northward orientation of the total Vd vectors with respect to the regional extension direction (Table 2) indicates a substantial component (∼3 mm/yr) of dextral shear across the transect, but the shear component likely decreases eastward away from active strike-slip faulting in the Walker Lane (Hammond and Thatcher, 2005, 2007; Payne et al., 2012). So how might this shear be expressed on faults in the transect? We suspect that either shear is partitioned on primarily strike-slip faults (e.g., 1932 Cedar Mountain earthquake; Bell et al., 1999) or is distributed as an oblique-slip component during earthquakes on some of the active normal faults in the region (e.g., 1954 Rainbow Mountain–Stillwater sequence; Caskey et al., 2004). A likely example of partitioned slip is the Long Valley fault system, where our field reconnaissance revealed evidence of strike-slip faulting during a Holocene surface-rupturing earthquake. Other candidates might include some of the numerous favorably oriented northeast- or northwest-striking faults on the northwest Nevada plateau, or pre-Sehoo strike-slip faults that may be present unrecognized beneath lacustrine, eolian, and alluvial deposits in some of the broad valleys in the transect, such as the northeastern arm of the Black Rock Desert adjacent to the JMFZ. Egger et al. (2010) suggested that some of the more steeply dipping faults underlying Surprise Valley may be partitioning strike-slip displacement, because a corrugated geometry probably prevents significant oblique slip on the main trace of the SVFZ. Oblique slip on some of the active normal faults in the transect is also likely but very difficult to quantify, because in our experience, clear evidence of lateral displacement is rarely observed along most late Quaternary normal faults in the region. With the exception of the LVFZ, we observed no unequivocal evidence of a significant lateral-slip component on any of the prominent normal faults in the current study.

We used the results of our field mapping, investigations of natural and manmade exposures, scarp profiling, limited new chronological (OSL and U-series) data, as well as data from published studies, to estimate deformation rates on the seven fault zones that postdate the latest Pleistocene Sehoo highstand shoreline of pluvial Lake Lahontan in the transect area (Fig. 1). This widespread datum provides a uniform time window for calculation of slip rates, and definitively identifies those faults responsible for most of the strain release during the last 15–18 k.y. We calculated vertical slip rates using what we infer are average to maximum values of slip over this time period (Table 1), and then convert these vertical rates to horizontal extension rates using a range of possible fault dips (Table 3).

As previously recognized (Friedrich et al., 2003; Wesnousky et al., 2005; Koehler and Wesnousky, 2011), our calculated extension rates are strongly dependent on the values of fault dip used in the conversions (Table 3). Our results appear to show that the sum total extension rate of the seven latest Quaternary faults in the transect is nearly equal to modern (geodetic) rates of extension only if the faults dip ≤40° to seismogenic depths (Fig. 13A). If the range-bounding faults in the transect dip at steeper angles of 45°–60° to seismogenic depths, then our vertical slip rates yield estimated extension rates that are 46%–79% of the modern extension rate across the transect (Fig. 13A; Table 3). Unfortunately, actual values of fault dip are rarely known, so estimation of fault dip is a large source of uncertainty in our comparisons of long-term and modern rates of deformation. In addition to fault dip uncertainties, other possible sources of our apparent geologic-geodetic discrepancy include the following: (1) underestimated slip rate uncertainties; some of our slip rates have large uncertainties because the chosen time period (last 15–18 k.y.) only contains a single earthquake and thus does not cover a complete seismic cycle; (2) not all potential seismic sources are included in our assessment; (3) unrecognized off-fault coseismic deformation associated with large-magnitude, surface-rupturing earthquakes on faults included in our analysis; and (4) unrecognized deformation from localized earthquake swarms or moderate-sized earthquakes too small to cause surface rupture. All of these possible sources are discussed in the following.

Traditional estimates of normal fault dips of ∼60° are based on Andersonian concepts of Mohr-Coulomb fault theory and rock mechanics (e.g., Anderson, 1951; Jaeger and Cook, 1979), but the discovery of low-angle shear zones in deeply exhumed extensional terrains (Wernicke, 1995; Axen, 2004) fostered a growing recognition of the possibility of large earthquakes generated on less steeply dipping normal faults. Most western United States examples of such faults are located in the eastern California shear zone and Death Valley (e.g., Cichanski, 2000; Hayman et al., 2004; Numelin et al., 2007), and include surface rupture on a low-angle (∼20°) detachment fault during the 2010 primarily right-lateral strike-slip El Mayor–Cucapah earthquake (Fletcher et al., 2014). Examples in the Basin and Range include the Ruby–East Humboldt Range fault (Smith et al., 1989; Wernicke, 1995; Wesnousky and Willoughby, 2003) and parts of the 1954 Dixie Valley earthquake rupture (Abbott et al., 2001). Within our transect, recent seismic reflection and refraction surveys in Surprise Valley identified a strong planar reflector thought to be the SVFZ dipping at a shallow (∼30°) angle beneath the valley floor (Lerch et al., 2010). Lerch et al. (2010) inferred that the SVFZ initiated at a dip of ∼50°–60°, and rotated into its present position (e.g., Buck, 1988) over the past 12 m.y. Other studies of post-Miocene uplift and exhumation of most of the fault-bound mountain ranges in the region assume current fault dips of ∼40° based on similar models (Colgan et al., 2006b). The question of whether the Quaternary-active range-front trace of the SVFZ and possibly other fault zones sole into the shallowly dipping reflector, or instead have shifted to a more favorably oriented, more steeply dipping throughgoing structure (e.g., Egger et al., 2010), remains unknown because the part of the seismic section beneath the range-front fault trace is devoid of coherent reflectors (Lerch et al., 2010). This situation is consistent with other studies that indicate that steeply dipping range-front faults in the Basin and Range are commonly obscured by seismically opaque wedges of alluvial-fan sediments (Anderson et al., 1983; Okaya and Thompson, 1985).

We prefer somewhat steeper dip values (45°–55°), because steeper fault dips are supported by seismological and geodetic studies of several of the historic surface-rupturing earthquakes in the region. These include the 1915 Pleasant Valley (36°–52°), 1954 Fairview Peak (50°–75°), and Dixie Valley (45°–50°), Nevada, earthquakes (Romney, 1957; Slemmons; 1957; Doser, 1986, 1988; Caskey et al., 1996; Hodgkinson et al., 1996), the 1959 Hebgen Lake, Montana, earthquake (45°–50°; Barrientos et al., 1987), and the 1983 Borah Peak, Idaho, earthquake (42°–49°; Doser and Smith, 1985; Stein and Barrientos, 1985; Richins et al., 1987; Payne et al., 2004). The dip range estimates from these 5 earthquakes yield a mean dip of 49° ± 7.5° (Fig. 13B). Similar dips are also typical of some of the swarms of M_w 4–5 earthquakes in the region (Smith and Lindh, 1978; Patton and Zandt, 1991; Pezzopane and Weldon, 1993). Global analyses of large historical normal-faulting earthquakes indicate that most occur on faults dipping 30°–70° (Jackson and White, 1989), 40°–50° (Thatcher and Hill, 1991), and 30°–65° (Collettini and Sibson, 2001), with all 3 studies consistent with peak occurrence at ∼45°. Analysis of regional subsets of the global catalog with the best seismological and geological data (e.g., Doser and Smith, 1989; Thatcher and Hill, 1991) support relatively steep dips (45°–65°). Geologic evidence supporting steeper dips in the transect region includes numerous examples of bedrock exposures of the active range-front fault traces with dips of 55°–70° on the SVFZ (Hedel, 1980, 1984; Egger and Miller, 2011) and 56°–63° on the JMFZ (this study; Fig. 6).

Slip rates are best constrained with knowledge of precise timing of several paleoearthquakes and accurate estimates of fault slip during these events. However, obtaining the necessary data to calculate such interval slip rates is dependent on detailed and expensive trench studies, which are difficult to justify in remote areas with low population densities such as the transect region. However, average slip rates can be determined with data as simple as topographic profiling and regional correlation of faulted surfaces and deposits. We justify our inclusion of several average fault slip rates based on a probable single earthquake (e.g., SRRFZ, DVFZ, LVFZ) because (1) the chosen time interval includes elapsed time before and after the earthquake; and (2) a single earthquake in the last 15–18 k.y. is a good indicator of long recurrence and low rates of fault activity. We consider our slip rates for the three faults with single ruptures as maximum rates and conclude that the absence of a second earthquake in these records does not have a significant effect on our rate discrepancy. Evidence of temporal variations in slip rate, such as those observed on the SRRFZ (Personius and Mahan, 2005), indicates another source of slip rate uncertainty. However, the wide aperture of our transect area and number of recently active faults included in the analysis probably average out faults with accelerating and decelerating earthquake recurrence and thus should not have a significant effect on our cumulative slip rate across the region. Our 15–18 k.y. time window is also supported by results from the previous transects in the region, which determined only slightly lower extension rates over longer time windows. Wesnousky et al. (2005) determined extension rates of 0.42 mm/yr since 45 ka and 0.65 mm/yr since 20 ka; Koehler and Wesnousky (2011) determined extension rates of 0.8 mm/yr since 60 ka and 1.0 mm/yr since 20 ka. Such slight differences are likely related in part to incomplete paleoseismic records (Wesnousky et al., 2005) and less precise age estimates of older earthquakes.

We know from modern seismicity records that rapid filling of reservoirs behind natural or artificial dams can induce earthquakes (e.g., Gupta, 2002), and some studies suggest that the rise and fall of pluvial lakes in the western U.S. may have enhanced or inhibited earthquake recurrence (Hetzel and Hampel, 2005; Hampel and Hetzel, 2006; Karow and Hampel, 2010).

Marion (2016) correlated a cluster of three earthquakes on the Surprise Valley fault (earthquakes P2, P3, and P4 of Personius et al., 2009; earthquakes SV2, SV3, and SV4 in Supplemental Fig. S8 [^{footnote 11}]) with the post–11 ka desiccation stage of pluvial Lake Surprise. However, directly relating the paleoseismic history of the SVFZ to fluctuations in lake level is not well supported given a near-complete lack of dated shoreline features in this time range (ca. 11–1 ka) and the large constraints on the ages of these earthquakes (2σ ranges of 0.5–3.2 k.y.). Thus we conclude that the effects of pluvial lake fluctuations on the slip histories of the SVFZ and other faults in the NWBR remain largely unknown and are an additional source of uncertainty in our slip rate calculations.

We cannot exclude the possibility that other potential sources of significant deformation, such as the Pine Forest Range fault zone, exist in the transect region. However, only one of the historic surface ruptures in the central Nevada seismic belt (1954 Fairview Peak earthquake) occurred on a fault without a record of at least one prior Holocene rupture (Bell et al., 2004), so we infer that most of the ongoing deformation in the transect region is associated with faults that have been active in the past 15 k.y. However, the lack of a post–15 ka earthquake indicates that most or all of the excluded faults have slip rates of <0.1 mm/yr and thus would have little impact in reducing the apparent fault-slip deficit.

Some studies of coseismic ruptures associated with recent large-magnitude earthquakes indicate that fault slip measured at the surface underestimates the amount of slip at several kilometers depth calculated from geodetic and seismologic data, a phenomenon commonly referred to as shallow slip deficit (e.g., Simons et al., 2002, Fialko et al., 2005). Coseismic deformation that occurs off the main rupture traces has long been recognized as a likely source of this deficit (Rockwell et al., 2002; Shelef and Oskin, 2010; Rockwell and Klinger, 2013), but other controls such as fault maturity have also been proposed (Dolan and Haravitch, 2014). A recent study of off-fault deformation along the length of the 200-km-long rupture associated with the 2013 M_w 7.7 Balochistan earthquake (Gold et al., 2015) documented that on-fault deformation averaged ∼72% of total offset along the entire rupture. Gold et al. (2015) showed that the percentage of off-fault deformation is highly variable (0 to >50%) along strike, and cautioned against using a single value for estimating off-fault contributions to offset estimates used in hazards calculations.

Regrettably, most studies of off-fault deformation are focused on strike-slip earthquakes, but a recent study of the M_w 7.0, 2006 Mozambique normal fault earthquake indicates that deformation measured across fault scarps at the surface (∼1–2 m; Fenton and Bommer, 2006) was only 50% of the 3–4 m of slip at 10 km depth modeled by Copley et al. (2012). In addition, a study of off-fault deformation using three-dimensional ground-penetrating radar in the Taupo rift of New Zealand showed that undocumented drag folding and block rotation accounted for 41% of total slip across a trenched normal fault (McClymont et al., 2009). However, differences in crustal structure and rheology of the Mozambique and New Zealand locations and the steep dip (∼75°) of the Mozambique rupture plane suggest to us that these off-fault deformation estimates may not be directly applicable to normal faulting in the NWBR. For example, inversions of geodetic (Stein and Barrientos, 1985; Ward and Barrientos, 1986; Du et al., 1992) and seismologic (Doser and Smith, 1985; Mendoza and Hartzell, 1988) data from the 1983 Borah Peak earthquake yield similar calculated values of slip at depth with slip measured at the surface (Crone and Machette, 1984; Crone et al., 1987), although the slip distribution patterns vary from model to model. The conclusion of Dolan and Haravitch (2014) that mature strike-slip fault systems produce surface ruptures that exhibit only minor (5%–15%) underestimates of slip at depth, if applicable to normal faults, may indicate that surface measurements on the mature faults in our transect (Surprise Valley, Steens–Black Rock, Jackson Mountains, Santa Rosa Range) are reasonable estimates of overall slip.

The effect of off-fault deformation on our slip rate estimates is difficult to quantify, given that historic surface ruptures in the Basin and Range exhibit a range of fault complexity, from relatively simple, narrow fault traces (e.g., parts of the 1983 Borah Peak earthquake; Crone et al., 1987) to several-kilometer-wide fault zones with multiple fault traces (e.g., 1932 Cedar Mountain and 1954 Fairview Peak earthquakes; dePolo et al., 1989; Wesnousky, 2008). The passage of time compounds the difficulty of determining offsets across complex prehistoric fault ruptures in the NWBR, particularly in dynamic landscapes affected by such processes as pluvial lake-level fluctuations, eolian deflation and deposition, and alternating episodes of fluvial erosion and deposition. Scarp preservation can be especially poor where ruptures are localized on steep slopes at the alluvial-colluvial and bedrock contact. These range-tight ruptures characterize much of the trace of the Lost River fault zone after the 1983 Borah Peak earthquake, and many prehistoric ruptures elsewhere in the Basin and Range. Our challenges with mapping complex ruptures along the JMFZ serve as a good example of these difficulties: offset along the range-tight trace can only be measured in a few fortuitous locations, and complete documentation of displacement on complex piedmont ruptures is unlikely, given the erosional and depositional processes active in the region. Intrabasin faults are commonly imaged in geophysical studies beneath valley floors in the Basin and Range (Grauch and Hudson, 2007; Stephenson et al., 2012; Athens et al., 2016). Although larger (>0.5 m high) scarps on such structures (i.e., SRRFZ and DVFZ; Fig. 1) can be readily observed, smaller scarps may be obscured by post-earthquake erosion, deposition, and human disturbance, and thus are a likely source of unrecognized off-fault deformation. We conclude that our displacement estimates are more likely to underestimate, rather than overestimate, total displacement, and thus undocumented off-fault deformation is a probable source of some of the geology–geodesy rate discrepancy.

Coseismic deformation from localized earthquake swarms or moderate-sized earthquakes below the magnitude threshold of surface rupture is generally small and localized, but over the time period used to calculate our fault slip rates may be a source of unidentified long-term deformation. Several earthquake swarms in the NWBR in the past several decades indicate that such seismic events may be very common, and unfortunately nearly impossible to identify in the geologic record. Examples include the previously discussed Denio swarm on the Pine Forest Range fault zone, the currently ongoing Sheldon swarm on the southern end of the Guano Valley fault zone northeast of Long Valley (Fig. 1; Supplemental Fig. S1 [^{footnote 1}]; Ruhl et al., 2015), and a third swarm that occurred ∼40 km northwest of the Sheldon swarm near Adel, Oregon, in May–July 1968 (Couch and Johnson, 1986). The latter presumably occurred along one or more normal faults bounding the Warner Valley in southern Oregon. At least 14 earthquakes of M_w 4–4.7 occurred during this swarm, most of which occurred beneath the floor of Warner Valley. A composite focal mechanism by Schaff (1976; see also Smith and Lindh, 1978) shows left-lateral oblique movement on a fault plane striking N04°E and dipping 80°E (presumably the west Warner Valley fault zone), and moment tensor analysis of a surface wave magnitude, M_s 5.1 earthquake in the sequence yielded a solution with a dip of 70°E and strike of 01°N (Patton and Zandt, 1991; Pezzopane and Weldon, 1993). None of the faults associated with these swarms show evidence of latest Quaternary surface displacement (U.S. Geological Survey, 2015).

A second potential source of unidentified long-term deformation is associated with moderate-sized earthquakes at or below the magnitude threshold of surface rupture. Recent examples are 2 M_w ∼6 earthquakes near Klamath Falls, Oregon, on 20 September 1993 (Braunmiller et al., 1995; Dreger et al., 1995), and the M_w 6.0 earthquake sequence near Wells, Nevada, on 21 February 2008 (Smith et al., 2011). Although no surface rupture was reported for either the Klamath Falls or Wells earthquakes, Mendoza and Hartzell (2009) modeled subsurface vertical deformation of as much as 88 cm on a 6.5 km by 4 km fault patch during the Wells earthquake, and radar interferograms indicate 15–20 cm of vertical surface deformation associated with the event (Bell, 2011a, 2011b). Even if these moderately sized earthquakes rupture the surface (e.g., M_w 6.5 2009 L’Aquila, Italy, earthquake), fault scarps are small and discontinuous and may be difficult to identify a few thousand years in the future. Although earthquakes of this size are uncommon along large active normal faults such as the Wasatch fault zone (e.g., Arabasz et al., 1992), they may be more characteristic of deformation on smaller faults such as those thought to be responsible for the Klamath Falls and Wells earthquakes. In addition to major normal faults that do not offset latest Quaternary deposits (e.g., Pine Forest Range and Hays Canyon Range fault zones), there are dozens of shorter faults in the NWBR that may have undergone repeated earthquake swarms or moderate-sized (M_w ∼ 6.5) earthquakes during the 15–18 k.y. duration of our paleoseismic record.

Dike intrusion commonly accompanies normal faulting in active volcanic centers such as the Snake River Plain (Kuntz et al., 2002; Holmes et al., 2008), but may be a possible source of unrecognized deformation if the intrusion does not reach the surface. An example of this phenomenon is a mafic dike intruded into an intrabasin fault zone identified with magnetic, gravity, and seismic reflection data 60 m beneath the floor of Surprise Valley (Athens et al., 2016). This feature is thought to have been intruded ca. 2 Ma and did not reach the ground surface (Athens et al., 2016). However, given the lack of latest Quaternary volcanic activity in the region (Stewart and Carlson, 1978; Stewart, 1980), diking is probably not a significant source of unrecognized deformation in the transect area.

How much of our apparent geologic rate deficit might be attributable to the many possible sources of geologically undetectable deformation? In a comparison of seismic, geodetic, and geologic moment rates, Pancha et al. (2006) determined that modern moment rates determined from seismicity and geodesy were similar across the Great Basin, but their calculated geologic moment rates based on slip rate estimates from inputs to the 1986 and 2002 national seismic hazard maps (Frankel et al., 1996, 2002; Haller et al., 2002) were 37%–65% of geodetic rates of the same region. One possible explanation is that the fault catalog used in the 1986 and 2002 hazard maps was incomplete, as suggested by differences in the catalog used in the 2014 iteration (Petersen et al., 2014a). A second explanation may be related to the Pancha et al. (2006) determination that 76% of the moment release in the 150 yr history of seismicity in the region was attributable to 10 large-magnitude (M_w ≥ 6.8) surface-rupturing earthquakes, and 90% of the moment release was attributable to just 38 earthquakes with magnitudes M_w ≥ 6.1. If these ratios are applicable to longer time periods (thousands of years), then perhaps 10%–25% of the total moment release across our transect may be attributed to geologically undetectable deformation associated with smaller, non-surface-rupturing (M_w < 6.8) earthquakes. Such calculations would nearly eliminate (within 1σ uncertainties) the apparent discrepancy between the geodetic and geologic deformation rates for fault dips of 45°–55° (75% and 90% GPS lines in Fig. 13A).

Despite some differences in methodology (primarily the length of time window considered) our paleoseismic transect at lat 41°–42°N appears to yield results similar to those of previous transects at lat 40°–41° (Wesnousky et al., 2005) and 38.5°–40°N (Koehler and Wesnousky, 2011). The closest transect, at 40°–41°N, resulted in a basin-wide extension rate of ∼0.65 mm/yr (60° assumed fault dip), and the transect at 38.5°–40°N resulted in a basin-wide extension rate of ∼1 mm/yr (also 60° assumed fault dip). These rates were both considered minimums, and thus probably are not significantly different than our 60° dip extension rate estimate of 1.1 mm/yr (Table 3). The similar results indicate that late Quaternary extension may be nearly constant across the Great Basin from ∼42°N south to at least 38.5°N, although our transect is much shorter, so deformation appears to be compressed into the northwestern part of the Basin and Range north of 41°. This difference is reflected in the distribution of late Quaternary faulting in southeastern Oregon, and is expressed in the tectonic geomorphology of northeastern Nevada and northwestern Utah east of our transect; most of this region is characterized by broad, undeformed tablelands (e.g., Owyhee Plateau) and subdued normal-fault–controlled basins and ranges with little evidence of post-Sehoo deformation (Hecker, 1993; dePolo, 2008; U.S. Geological Survey, 2015).

In this study we use new and existing paleoseismic data to compile a record of ∼18 large-magnitude surface-rupturing earthquakes in the past 15–18 k.y. on 7 fault zones in the NWBR of northwestern Nevada and northeastern California. We use the size of surface ruptures and this latest Quaternary time window to estimate vertical slip rates on these fault zones and compared them to GPS-derived rates of modern extension across the region. The horizontal extension rates derived from our fault slip rates are particularly sensitive to assumed fault dip (Fig. 13A). For example, our preferred fault dip values (45°–55°) yield estimated long-term extension rates (1.3–1.9 mm/yr) that are 54%–79% of our extension only modern rate, but fault dip values of 40° or less yield long-term rates comparable to or greater than our preferred modern rate. These relations may simply mean that actual fault dips are closer to 40° rather than our preferred values. However, we continue to favor steeper fault dips, so we conclude that our long-term rate underestimates of ∼21%–46% probably are attributable to some combination of (1) geologically unrecognizable deformation from moderate-sized (M_w 6.0–6.8) earthquakes; and (2) unobserved off-fault deformation from larger (>M_w 6.8), surface-rupturing earthquakes. The observation of Pancha et al. (2006) of a ∼25% difference between the moment released during 10 surface-rupturing earthquakes and total moment in the seismic record of the Basin and Range indicate that both of these factors contribute to the differences observed in our analysis. In addition to fault dip uncertainties, remaining questions include the appropriate length of the time window used to determine representative long-term (paleoseismic) extension rates, and the potential effects of crustal loading and unloading from pluvial lake fluctuations. Definitive answers will remain elusive until denser GPS networks are established, and well-constrained earthquake histories and better estimates of dip on all faults in our transect are available.

This research was supported by the Earthquake Hazards Reduction Program of the U.S. Geological Survey (USGS). We acknowledge the cooperation of the Winnemucca office of the Bureau of Land Management in facilitating research on lands under their jurisdiction in northwestern Nevada. We also appreciated the cooperation of Surprise Valley, California, landowners Denise and Jim Harrower and Bill Duncan for access to their properties. The efforts of many colleagues during field and office work (Lee-Ann Bradley, Ernie Anderson, Koji Okumura, Dean Hancock, Michael Machette, Tony Crone, David Lidke, Jai Bok Kyung, Hector Cisneros, Ryan Gold) and discussions with Bill Hammond and Craig dePolo are greatly appreciated. The manuscript was improved by USGS reviewers Ryan Gold and Chris DuRoss, Geosphere reviewers Anne Egger and Anka Friedrich, and Geosphere Associate Editor Colin Amos.