Spatiotemporal evolution of fault slip rates in deforming continents: The case of the Great Basin region, northern Basin and Range province

Most earthquakes and active faulting occur within localized belts or zones along the boundaries between Earth’s tectonic plates (e.g., Isacks et al., 1968; McKenzie, 1972). Where one or both of the two plates are continental, the deformation tends to be diffuse, accommodated by fault systems with remarkable geometric and kinematic complexity. The largest active examples of these belts (∼800–1000 km wide) are the Arabia-Eurasia and India-Asia collision zones, the Aegean region, the central Andes, and the Basin and Range province of western North America (e.g., Molnar and Tapponnier, 1975; Smith, 1978; Isacks, 1988; Jackson, 2001). Of these five examples, the Basin and Range province is well suited for the study of the distributed style of continental deformation, because the formation of each range block is controlled by crustal-scale normal faulting (Davis, 1903; Stewart, 1971; Eaton, 1979; Wallace, 1984; Wernicke, 1992; Niemi et al., 2004). The size and spacing of individual fault-bound range blocks and intervening valleys are, on a global standard, striking in organization and uniformity, a noteworthy counterexample to the prevailing complexity of continental topography (e.g., Dutton, 1885). This region also stands out as one in which the Quaternary kinematic histories of its active faults, including the timing and amount of slip for the most recent, penultimate, and even antepenultimate earthquakes, have been documented on a substantial fraction of the total number of faults, especially in the Great Basin region (e.g., Machette et al., 1991; DuRoss, 2008; Koehler and Wesnousky, 2011). The sum of information obtained over the past three decades provides an opportunity to comprehensively explore the seismokinematic response of the continental lithosphere to plate boundary forces, or as predicted by Jackson (2001, p. 5), to progress “…beyond the goal of trying to understand how faulting achieves the present-day large-scale motions to the more profound question of how the fault configurations have evolved through time.”

In pursuit of this goal, we synthesize here published information on the Quaternary history of active faults in the Great Basin region and the Walker Lane belt–Eastern California shear zone (WLB-ECSZ), which includes that portion of the more extensive Basin and Range province between lat 35°N and 42°N. We focus on this portion of the province because it is the widest part of the actively deforming Pacific–North America plate boundary zone (Fig. 1). Many studies and syntheses of geological slip-rate estimates and paleoseismological data have been carried out on specific fault systems (e.g., Thompson and Burke, 1973; Machette et al., 1992; Lund, 2005; Wesnousky et al., 2005; Koehler and Wesnousky, 2011; DuRoss et al., 2012). Temporal slip-rate variations on faults over the Quaternary have also been observed (e.g., Friedrich et al., 2003; Niemi et al., 2004; Kirby et al., 2008; Gold et al., 2013). The goals of our compilation of neotectonic data over the WLB-ECSZ and the Great Basin region are to update, expand, and make accessible the most comprehensive database to date in terms of number of neotectonic offset data, paleoseismological data, and fault slip-rate variations over different time windows. It enables (1) estimation of fault kinematic characteristics (displacement per event, earthquake recurrence interval) at a regional scale; (2) comparison of slip-rate evolution between different faults (owing to the determination of fault slip-rate values through time on displacement history charts); and (3) investigating earthquake distribution through time and space and consequently the regional strain release pattern.

The Great Basin segment of the Basin and Range province is an 800-km-wide region of distributed deformation in right-lateral shear and extension, which accommodates ∼25% (∼11 mm/yr) of the relative motion between the Pacific and North American plates (Lachenbruch and Sass, 1978; Smith, 1978; Dixon et al., 1995, 2000; Dickinson and Wernicke, 1997; Bennett et al., 1999, 2003; DeMets and Dixon, 1999; Miller et al., 2001). It is separated from the plate-bounding San Andreas transform fault by the 200-km-wide Sierra Nevada–Great Valley microplate, which has remained more or less internally rigid at upper crustal levels since its formation as an arc-forearc pair in Mesozoic time (Fig. 1).

Basin and Range normal faults are dominantly north- to north-northeast striking with east-west– to west-northwest–east-southeast–trending slip vectors, across the eastern three-quarters of the region (Fig. 2). Along the western quarter in westernmost Nevada and eastern California, strike-slip faults and pull-apart basin geometries are prevalent, composing a zone of north-northwest right-lateral shear, the WLB or Walker Lane fault system (Wesnousky, 2005) or, where it enters California toward the south, the ECSZ (Fig. 1; Dokka and Travis, 1990; Miller et al., 2001). The WLB-ECSZ accommodates transform motion, which from south to north is progressively transformed into extension (Oldow et al., 2001; Bennett et al., 2003; Oldow, 2003; Wesnousky, 2005; Hammond and Thatcher, 2007). The central and eastern part of the Great Basin is dominated by pure normal faulting; there is some strike-slip faulting in the southern part (Fig. 2). The eastern boundary of major Quaternary faulting includes the Wasatch fault system in the northern Great Basin and the Hurricane and Sevier-Toroweap fault systems in the southern Great Basin (Figs. 1 and 2).

Quaternary vertical displacement rates on the range-bounding normal faults throughout most of the region are typically ∼0.1–0.3 mm/yr (e.g., Machette et al., 1991; Friedrich et al., 2003; Niemi et al., 2004; Wesnousky et al., 2005; Koehler and Wesnousky, 2011), in general agreement with long-term estimates (10 m.y.) derived from reconstructing the geological offset along typical range-bounding faults (Thompson and Burke, 1973; McQuarrie and Wernicke, 2005) and with estimates of horizontal displacement rates based on geodesy (e.g., Bennett et al., 2003; Niemi et al., 2004).

In contrast to the broad zone of normal faulting to the east, a number of faults within the WLB-ECSZ exhibit long-term slip rates that are much larger. This is especially true where active right-lateral shear strain becomes more localized toward the south in the ECSZ, where active extension becomes subordinate to shear (e.g., Bennett et al., 2003). Individual strike-slip faults along the western margin of the province have late Quaternary slip rates an order of magnitude greater than those on a typical Great Basin normal fault, in the range of ∼1–3 mm/yr (Reheis and Sawyer, 1997; Wesnousky, 2005; Frankel et al., 2007; Guest et al., 2007). As elaborated on here, these faults, and the paleo-earthquakes associated with them, constitute a relatively small fraction of our database.

We emphasize that for the normal faults, the vertical displacement rate is typically reported, owing to uncertainties in fault dip. However, most Quaternary Basin and Range normal faults dip moderately to steeply through the seismogenic crust (nominally, 30°–60°, with noteworthy exceptions in some areas; Niemi et al., 2004; Wernicke et al., 2008), suggesting that the horizontal displacement is usually within a factor of 0.6–1.7 times the measured vertical displacement. The locations of individual faults we describe in the following are shown in Figure 2.

In the ECSZ, long-term (to 10 m.y.) displacement rates are similar to Quaternary offset rates, and are generally in the range of a few millimeters per year to >10 mm/yr in some cases (e.g., Reheis and Sawyer, 1997; Guest et al., 2007). High long-term rates have also been reported for some strike-slip faults in the WLB to the north (e.g., Faulds et al., 2005). Historic ruptures have been concentrated in a longitudinal band along the Owens Valley, continuing northward into the central Nevada seismic belt (CNSB; Fig. 1). The northern WLB, although a locus of seismicity, has not undergone a major rupture in historic time.

In the northern Great Basin region east of the WLB, the most recent deformation is concentrated in two longitudinal bands, the CNSB and the Wasatch fault system (Fig. 2). Over the past 100 yr, the CNSB has been the locus of historic ruptures (long 118°W; Wallace and Whitney, 1984; Bell et al., 2004). On the Wasatch fault system there are no major historic ruptures, but Holocene clustering of events indicates transient (10 k.y.) geologic slip rates >1 mm/yr, greater than any other zone in the Basin and Range except for the belt of strike-slip faults along the western margin. In contrast to the WLB-ECSZ, however, the late Quaternary (100 ka) displacement rates are still only ∼0.3 mm/yr (e.g., Friedrich et al., 2003). Between these 2 belts, faults of the central and eastern Great Basin have, on average, a vertical rate of 0.1 mm/yr over the past 20 k.y. (Koehler and Wesnousky, 2011). Averaged over the past 60 k.y., however, this vertical slip rate drops to 0.05 mm/yr (Koehler and Wesnousky, 2011). As a whole, paleoseismic data suggest concentrated strain release in the central Nevada belt and along the easternmost Wasatch fault zone over the past 60 k.y. (Wesnousky et al., 2005). Although paleoseismic investigations are not yet definitive, they are consistent with the hypothesis that strain release at the 100 yr time scale is focused in longitudinal belts that migrate with time across the province (Wallace, 1987; Bell et al., 2004). At the scale of 100 k.y. or greater, proxies for overall displacement history (scarp morphology and the presence and size of basal facets on range fronts) suggest somewhat less concentrated strain (dePolo and Anderson, 2000).

For Great Basin normal faults, at the 1–10 m.y. time scale, the uniformity of range heights and basin depths constitute perhaps the clearest example of distributed tectonic strain on Earth (Stewart, 1978). If we consider a typical mountain range with a bounding normal fault, it takes ∼15 m.y. to create vertical structural relief of 4500 m, leading to a long-term slip rate of ∼0.3 mm/yr (e.g., Thompson and Burke, 1973; McQuarrie and Wernicke, 2005).

Relative to North America, horizontal geodetic velocities rise from near zero on the Colorado Plateau to 3 mm/yr oriented due west near the Nevada-Utah border (Fig. 1) (Bennett et al., 1998, 1999, 2003; Thatcher et al., 1999; Wernicke et al., 2000; Hammond and Thatcher, 2004; Davis et al., 2006). They remain relatively constant across eastern Nevada, and rotate northwestward and progressively increase in magnitude across western Nevada and the WLB, reaching an average value of 11.4 mm/yr in the Sierra Nevada–Great Valley microplate. The remaining ∼39 mm/yr of Pacific–North America plate motion is accommodated by the San Andreas fault system west of the Sierra Nevada–Great Valley microplate (Bennett et al., 2003; Fig. 1).

Most of the geodetic strain between the Sierra Nevada–Great Valley microplate and the Colorado Plateau is concentrated within the eastern and western portions of the Great Basin, with the central approximately one-third (long 117°W and 114°W) straining much more slowly. This basic pattern defines three geodetic provinces, referred to as the western, central, and eastern Great Basin provinces by Bennett et al. (2003, their fig. 10 therein). The western Great Basin province accommodates 9.3 ± 0.2 mm/yr of northwest-directed right-lateral transtensional shear, with the extensional component increasing northward (Bennett et al., 2003). The eastern Great Basin province accommodates ∼3 mm/yr of east-west–directed extension between the Nevada-Utah border and the Colorado Plateau (Malservisi et al., 2003; Niemi et al., 2004), a major part (1.6–1.9 mm/yr) being focused along the Wasatch fault system (Friedrich et al., 2003; Chang et al., 2006). Between these two highly deforming zones, the central Great Basin province is essentially undeforming at present (Bennett et al., 2003), or perhaps straining very slowly (∼0.6 mm/yr across 300 km; Hammond et al., 2014).

On normal faults of the Great Basin region, geodetic slip rate is sometimes consistent with geologic estimates (e.g., the Holocene rate of the Wasatch fault system, Friedrich et al., 2003; or the late Pleistocene rate of faults across central Nevada, Koehler and Wesnousky, 2011; Hammond et al., 2014), but not always. For example, the late Pleistocene slip rate on the Wasatch fault (∼0.3 mm/yr) is much slower than its geodetic slip rate (1.5–2 mm/yr; e.g., Friedrich et al., 2003). In the CNSB (Bell et al., 2004) and in southern Nevada (Wernicke et al., 2004), the deformation rate estimated by summing late Quaternary fault displacements (averaging over 20 k.y.) are less than average global positioning system (GPS) velocities. These cases suggest (1) off-fault strain release that is too diffuse to detect using paleoseismic methods, and/or (2) temporal variations in the crustal velocity field (Bell et al., 2004; Pancha et al., 2006). Whereas off-fault strain is probable to some extent, more enigmatic is the notion that, within the intraplate environment, deformation is transient, occurring partly or wholly in discrete episodes separated by long hiatuses of relative tectonic stability (e.g., Calais and Stein, 2009). If deformation is transient, it has major implications for both the physics of intraplate deformation processes and seismic hazards analysis. In particular, despite many possibilities, there is no clear-cut mechanism, absent hotspot magmatism, that would trigger anelastic deformation episodes within a plate interior, especially at time scales of 10 k.y. to 1 m.y.

In order to create a neotectonic database of the Great Basin that is useful for analysis, both for this paper and for the Earth sciences community in general, we have synthesized the published literature on the deformation rates across faults, and presented it in: (1) an Excel format file containing all the neotectonic data collected for each fault (Supplemental Information: NEOTECTONIC_DATA.xls; this file and others herein marked “supplemental” are available for download¹), (2) a table of interevent recurrence intervals calculated for each fault (Supplemental Information: Recurrence_intervals.xls), (3) a table of individual displacements per event compiled for each fault (Supplemental Information: Displacements_per_event.xls), (4) a table summarizing all inferred fault slip rates (Supplemental Information: Slip-rates.xls), (5) an interactive fault map summarizing, for each fault, the main information and characteristics inferred in this study [i.e., average recurrence interval, average displacement per event, most recent earthquake age, and slip-rate values considering different time windows; this file can be downloaded as a KMZ file (GoogleEarth) or a GIS shapefile (Supplemental Information: GIS_FAULTS.kmz and GIS_FAULTS.shp], (6) a compiled paleo-earthquake catalog over the past 100 k.y., as a table file (Supplemental Information: Paleoeq_catalog.xls), or a time animation (Supplemental Information: Paleoeq_catalog.kmz). The neotectonic database currently contains information on 171 individual fault segments (Fig. 2), synthesized from 173 sources. Of these 171 segments, 139 are normal faults (with several that have significant oblique slip), and 32 are strike-slip faults. Among these segments, the data set includes 533 offset determinations, of which 243 pertain to individual earthquakes, both historic and prehistoric (all but one occurring after ca. 100 ka). For the individual earthquakes, 36 events are predominantly strike slip, and the remaining 207 are normal.

The Excel file database (Supplemental Information: NEOTECTONIC_DATA.xls; see ^{footnote 1}) includes a preliminary page showing the fault list and the type of neotectonic data available for each fault. Four categories of faults are defined: those with (1) offset of one or more markers with estimated age, but no paleoseismological information; (2) the age of only one paleo-earthquake (in general, the most recent one), but no slip-rate data; (3) the age of the most recent earthquake, with slip-rate data from offset markers; and (4) the timing of more than one earthquake, either with or without slip rate data from offset markers. In the database, each fault segment (one worksheet per fault) is represented by four display items (Fig. 3). (1) Input Data (upper left) is a table with raw information taken directly from the paleoseismological literature, including the name of the fault segment, the events identified, the displacement associated with each event (generally only the vertical component for normal faults), the cumulative displacement, the age constraints on each event, quotations or comments from the literature (in regard to the nature of the offset features and uncertainties in the timing and amount of slip), and the dip and rake of the slip, if known. If the paleo-earthquake record is not documented for a fault, we reported the age and the amount of offset of dated geological markers. (2) The Input Data chart (lower left) is a table summarizing the amounts and uncertainties of displacement from the input data, for use in constructing (3) a Cumulative Displacement versus Time chart (lower right), which is a plot of the displacement history in meters versus thousands of years; we use this chart to present (4) an Inferred Slip Rate table (upper right), considering different time windows.

We stress that the uncertainties in both displacement and displacement rate are not generally Gaussian or even symmetric, because estimates of both displacement and chronological information often involve complex qualitative information that does not lend itself simply to standard statistical methods of estimation. For example, this might include minimum displacement values for which an upper bound cannot be estimated, or the maximum age of an earthquake (postdates datum X) for which a minimum age cannot be estimated. Estimating uncertainties in slip rates and average recurrence intervals is similarly problematic. For example, when the timing and offset of the most recent earthquake are the only data, neither the slip rate nor the mean recurrence interval can be meaningfully quantified. However, the information may still be quite informative. For example, if the most recent event occurred 50–100 k.y. ago, the minimum repose time is far longer than the average for a typical Basin and Range normal fault. In addition, in the case of either kinematic or timing data that were informative but only qualitative (as in some paleoseismological results), we took care to dissociate raw published data from our assessed numeric values. In these cases, the values are followed by question marks and assessed in the Comments column.

A visual comparison of the fault segments in the database with the extant U.S. Geological Survey (USGS) Quaternary fault and fold database for the United States (U.S. Geological Survey, 2006) suggests that our database includes approximately two-thirds of the major fault segments in the Great Basin region, within the latitudinal limits of the State of Nevada, and the longitudes between the Sierra Nevada to the west and the Colorado Plateau and Rocky Mountains to the east (excluding the numerous small, diffuse fault segments in the USGS database; Fig. 4). Hence the database represents a sample of the system that is sufficiently complete to justify the assumption that the overall characteristics of the subset of faults in the database (i.e., displacement per event, earthquake time recurrence), can be extrapolated with confidence to apply to the entire population. That is, a sample of approximately two-thirds of the major fault segments is sufficient to be representative of all the possible fault kinematic patterns in the region. The paleo-earthquake data set is perhaps the most complete for any plate boundary fault system in the world in terms of number of neotectonic offset data, paleoseismological data compilation, and fault slip-rate estimates across different time windows, although much remains to be done. The database is complementary to the U.S. Geological Survey (2006) fault database, because the latter is an essentially complete representation of surface-rupturing fault segment traces.

The database enables robust characterization of the population of Basin and Range faults and seismic events, especially for the population of normal faults, for which most of the displacement per event data are derived. Among the 207 normal faulting seismic events compiled, 150 have reliable displacement per event data, mainly for the vertical component of motion (Fig. 5). Surface ruptures in the Great Basin region tend to have moderate to steep dips, and hence the vertical displacement is a large fraction of the net slip during the earthquake (for important exceptions, see Abbott et al., 2001; Axen, 2004; Niemi et al., 2004; Numelin et al., 2007, and references therein). A histogram of this population of events yields an average of 2.0 m with a fairly narrow maximum, such that approximately two-thirds of the events have between 1 and 3 m displacement (Fig. 5A). The maximum observed vertical displacement is 5.5 m.

Because we compiled data from historic seismicity or paleoseismological studies, our population of displacement per event only includes earthquakes that have generated a surface rupture, which means earthquakes of M > ∼6 (Wells and Coppersmith, 1994). A counterintuitive feature in the distribution is the lesser frequency of 0–1 m events (M 6–6.5) compared with 1–2 m or 2–3 m events (M > 6.5), in apparent contradiction with Gutenberg-Richter scaling relations. This lack of 0–1 m events is probably due at least in part to an exposure bias, because records of any small surface-rupturing events are more likely to be either missed entirely or fully eroded away between events. However, it may be that only the largest events on any given fault rupture the surface. In the short span of instrumental recording of earthquakes, it is uncommon for faults <M 6 to rupture the surface, presumably because the dimensions of the slip planes are less than the thickness of the seismogenic crust. In any case, our displacement per event distribution may thus be considered as representative of the largest normal faulting earthquakes, i.e., having coseismic displacement ≥1 m (M > 6.5).

The spatial distribution of displacement per event (Fig. 6) indicates, not surprisingly, that the strike-slip faults in the west have, on average, higher values than the normal faults (typically >5 m, mainly based on historic ruptures, versus the 2 m average for normal faults). Overall, there appears to be a contrast in the distribution north and south of lat 38°N. To the north, across the widest part of the Basin and Range province, the distribution is fairly uniform with a modest maximum value <6 m, with no clear distinction from east to west. To the south, faults east of the ECSZ have lower displacement per event (all <3 m) relative not only to the western strike-slip faults, but also relative to the values north of lat 38 °N. The strike-slip faults along and near the plate boundary (San Andreas, Garlock, Owens Valley, Death Valley) have displacement per event values that are markedly higher than the eastern normal faults (Fig. 6).

With this large a database, the question naturally arises as to whether a given fault has a characteristic displacement for each earthquake. More quantitatively, we want to know the extent to which the average displacements vary, either for the population as a whole (Fig. 5) or on individual faults (Fig. 6). In a seminal study of the San Andreas fault in the Carrizo plain, Liu-Zeng et al. (2006) found significant variation in slip per event, and little apparent correlation between the amount of slip on an earthquake and the time since the last earthquake. In the data set, the multiple-event normal faults define a population of variations in slip between successive events (Fig. 7). In other words, if the most recent event on a fault is larger than the penultimate event, the difference is positive, and if smaller, negative. This data set includes 64 instances where the increase or decrease in slip from one event to a subsequent event is measured (Fig. 7). The population of differences exhibits a fairly symmetric distribution about the origin (mean value of –0.015 m). Of the values, ∼33% are within 20 cm of zero difference in slip (i.e., a value of 10% of the mean displacement of the population of 2 m), or effectively the same. Outside of this strong peak, ∼66% of the data set is encompassed by displacement within 1 m of the previous event (i.e., half again as little, or half again as much, of the mean displacement of all events). We further observe ∼15% exhibit variation at or above the mean displacement of the population (>2 m). Whereas these data suggest a degree of repeatability of the vertical displacement for a significant fraction of the events, it is noteworthy that ∼33% of them differ from the previous event at the level of 50% of average displacement of all events.

In general, the displacement statistics are in broad agreement with the database of instrumented events compiled by Wells and Coppersmith (1994), in which both seismological and surface displacement constraints are available, where surface displacements of 1–3 m are typical of seismogenic continental normal faults in the M 6–7 range. It is interesting that our displacement per event statistics also are in agreement with surface displacement for normal faulting earthquakes observed in central Italy. For the Fucino fault system, the seismic exhumation history reveals the occurrence of 35 normal faulting earthquakes over the past 15 k.y., where most of the ruptures have a vertical displacement per event between 1 and 3 m, with an average vertical slip per event of ∼2 m for the northern Fucino faults (Benedetti et al., 2013).

Perhaps the most interesting aspect of the database is the opportunity to look at the characteristics of earthquake recurrence for a large number of events. Recurrence can be determined only for faults where more than one event is recorded. For faults where only the most recent earthquake is recorded, we can estimate a minimum time of recurrence for that event only. These estimates are minima that as a whole are poorly representative of the overall behavior of the population of faults. However, in the case where the age of the most recent event is many times the modal value of recurrence (e.g., >10 k.y.), it provides insight into what fraction of the population of faults has very long recurrence times.

In the data set, there are 45 normal faults with more than one recorded event. A total of 159 events from this population yield 114 interevent recurrence intervals (Fig. 8A). In addition, there are 37 faults on which only the most recent event is recorded, yielding the same number of minimum recurrence intervals (Fig. 8B). Unlike the displacement per event data, which vary over less than an order of magnitude with similar modal and mean values, the recurrence times vary over two orders of magnitude (∼0.5–50 k.y.; Fig. 8A): the modal value of the distribution (1–3 k.y.) is only ∼25% of the average value (11 k.y.), and a significant fraction (15%) have interevent intervals much longer than the average value (20–100 k.y.). The difference is mainly attributable to the fact that a large fraction of the data set is derived from fault segments along the Wasatch, Round Valley, and Surprise Valley areas, which have recorded a large number of frequently recurring Holocene events (i.e., clusters of events). We tested the reliability of earthquake recurrence interval calculation by considering only multievent faults having ≥3 earthquakes documented (Supplemental Information: Fig. S1; see ^{footnote 1}); this test shows that the distribution seen in Figure 8A does not vary significantly when considering only faults having ≥3 earthquakes documented.

The mean earthquake recurrence interval calculated for each fault is shown in Figure 9. We can distinguish a group of temporally clustering faults, having mean recurrence intervals far shorter than the regional average value of ∼11 k.y., located on the margins of the Great Basin (<3 k.y.; red faults in Fig. 9), and a group of long recurrence interval faults, the mean recurrence intervals of which are much longer than the regional average value, mostly located in central Nevada (20–100 k.y.; blue and dark blue faults in Fig. 9). A more typical situation is exhibited by faults in the CNSB, the mean recurrence intervals of which more closely match the regional average value (Fig. 9). Despite having undergone historical earthquakes, there have not been large numbers of Holocene events in the CNSB in comparison with the Wasatch zone segments (Supplemental Information: NEOTECTONIC_DATA.xls; see ^{footnote 1}).

This distinction immediately raises the question for the normal fault population, whether there is a correlation between recurrence interval and displacement per event. In other words, do the clusters represent a large number of small earthquakes (chattering strain release), with the infrequent events on long recurrence interval faults being commensurately larger? In Figures 5B and 5C, we show the distributions of displacement per event subdivided into clustering and long recurrence interval faults. The mean displacement per event has the same order of magnitude for the two groups of faults: 1.95 (±0.29) m for clustering faults (<3 k.y., Wasatch fault system; Fig. 5B), and 2.28 (±0.27) m for faults having a long earthquake recurrence interval (≥20 k.y.; Fig. 5C). Thus, there is no simple correlation between recurrence interval and displacement per event, as noted in previous studies for other normal fault systems in the world (e.g., central Italy; Benedetti et al., 2013, their fig. 4 therein). However, there is a tendency for the clustering faults to be in the 1–3 m range (>80%), whereas the long recurrence interval faults have a more even distribution between 1 and 4 m displacement (Figs. 5B, 5C). From the perspective of events with the largest vertical displacements, there does not appear to be any correlation between recurrence interval and displacement per event (Table 1).

To constrain the displacement rate as a function of time for individual faults, and to examine spatial and temporal trends, we can use the timing and displacement of ancient earthquakes and/or the relatively large (multievent) offsets of dated geological markers (without information about earthquakes). From these data, we can examine two categories of fault slip rates (Fig. 10): (1) finite or average displacement rates, which correspond to the total offset of a datum divided by the age of the datum (grouped as finite displacement rates back to 15, 75, 150 and 500 ka, as depicted Fig. 11), or (2) interval or transient displacement rates, which correspond to an offset accumulated during a specific time window divided by the duration of this time window (shown in the displacement histories of individual faults in Figs. 12 and 13).

We stress that the type of neotectonic data compiled is not uniform for all faults. They may have only one or several offset dated markers, and may not have even one documented paleo-earthquake, so many faults have no record of any transient, or interval, slip rates. For example, a fault may have only a postglacial (e.g., after 15 ka) offset marker, without any older offset datum. That fault will have an inferred Holocene (15–0 ka) slip-rate value (Fig. 11A). For such faults we do not infer finite slip rates for the 75–0 ka, 150–0 ka, or 500–0 ka time windows (Figs. 11B, 11C, 11D) and do not include them on the displacement rate histories (Fig. 12), resulting in a sampling bias with diminishing coverage through time (Fig. 11). Thus any conclusions reached about pre–15 ka displacement rates hinge on the assumption that the faults that preserved older offsets are representative of the behavior of the fault population as a whole. The data used to calculate the fault slip rates are described for each fault in the database (Supplemental Information: NEOTECTONIC_DATA.xls; see ^{footnote 1}) and synthesized in a table (Supplemental Information: Slip-rates.xls; see ^{footnote 1}).

Fault slip-rate values determined from 15, 75, 150, and 500 ka finite-displacement time windows (Fig. 11) for strike-slip faults in the WLB-ECSZ are much higher than normal faults in Nevada and western Utah, and for normal faults within the WLB-ECSZ (e.g., Lee et al., 2009). Whereas strike-slip slip-rate values vary between ∼1 and 5 mm/yr, normal fault slip rates are almost entirely <0.4 mm/yr. Most of our neotectonic data for the WLB-ECSZ strike-slip faults, as well as the left-lateral Garlock fault, are restricted to the Holocene, but some of them have slip-rate constraints in the 75–0 ka range (e.g., Owens Valley, northern Death Valley–Fish Lake Valley, Pyramid Lake, and Warm Springs Valley faults), but no older Quaternary offset markers have yet been reported for these faults.

Right-lateral shear of 1–2 mm/yr along strike-slip segments in the northern WLB and to 3–4 mm/yr along segments in the ECSZ that occurred during Holocene time (15–0 ka) were at similar levels over the past 75 k.y. (Figs. 11A, 11B). Normal faults located within or near the WLB-ECSZ also appear to have had consistent slip rates over at least the past 150 k.y., including the unusually rapid motion of >1 mm/yr on the Hilton Creek normal fault and related faults near the Long Valley caldera in eastern California (Fig. 11C). Some normal faults having documented markers on the 1 m.y. time scale suggest than constant slip rate can persist in time over as much as the past 800 k.y. (e.g., the Round Valley and White Mountain faults; Fig. 13A).

As apparent from Figure 11, some strike-slip faults exhibit a slip-rate variation over the past 75 k.y. (e.g., Gold et al., 2013). The finite slip rates on the right-lateral Owens Valley and Warm Springs Valley faults are much higher over the past 75 k.y. than over the past 15 k.y. Their respective finite slip rates of 1.9 and 3.5 mm/yr from 75 to 0 ka drop to 0.16 and 0.96 mm/yr during Holocene. At the same time, an apparent increase in finite slip rate occurred during the Holocene on some normal faults (Sierran front fault in Owens Valley, Lone Mountain fault, and Wassuk fault; Figs. 11A, 11B, and 12D) and during the late Holocene on the left-lateral Garlock fault (i.e., younger than 5 ka; Supplemental Information: NEOTECTONIC_DATA.xls; ^{footnote 1}).

Faults east of the WLB-ECSZ are characterized by relatively low slip-rate values (0–0.5 mm/yr), except for Holocene values of 1–2 mm/yr for the Wasatch fault system. These faults show even more striking differences in finite slip-rate values when averaged over the 15, 75, 150 and 500 ka time windows, and in interval slip rates at various times in the past 500 k.y. (Figs. 11–13).

In the 15 ka time frame, the map shows a concentration of strain release near the margins of the Great Basin (1–2 mm/yr on the Wasatch system, >0.2 mm/yr for faults in western Nevada west of long 116.5°W). In contrast, faults over most of central Nevada did not undergo any earthquakes during Holocene time (Fig. 11A). Along the eastern margin of the Great Basin, there is a contrast along strike between the Wasatch region to the north (1–2 mm/yr), and the Hurricane fault system to the south (< 0.2 mm/yr). It thus appears that in the 15–0 ka time window, relative to the long-term average for Great Basin normal faults (i.e., 0.2–0.3 mm/yr), a large fraction of the region records slip rates that vary by an order of magnitude on either side of this value (nominally, ∼0.01–1.0 mm/yr). In the 75–0 ka time frame (Fig. 11B), the central Nevada region, although showing fewer inactive faults, remains well below the long-term slip-rate average.

The picture becomes considerably more uniform by expanding the finite slip-rate time window to 150 ka. For the population in the 150 k.y. window (in which we include offset markers with ages of ca. 250–130 ka), far more faults have slip-rate values near the long-term average (Fig. 11C). Although some values in the central part of the Great Basin still remain low, there is little if any discernible contrast between the center and margins of the province, or between the Wasatch and Hurricane fault systems.

Although the evolution of the Wasatch fault system from relatively slow long-term displacement to rapid Holocene movement is the most dramatic example of a secular change in slip rate, other faults in the system appear to have undergone similarly rapid strain release since 500 ka. For the Wasatch fault system, interval slip rates from 150 to 15 ka are only ∼0.05 mm/yr, similar to many of the slow faults in the center of the Great Basin (Fig. 12). The net vertical displacements of offset geomorphic surfaces are ∼15–20 m in the Holocene, about half of the total accumulated displacement since 150 ka, yielding a twentyfold difference in slip rate for the two intervals (Fig. 12). In this regard, the Wasatch fault system appears to be unique among Holocene Basin and Range normal faults. In contrast to the Wasatch fault system, some faults in central Great Basin and along the Hurricane-Toroweap belt appear to have a low slip rate from 75 to 0 ka (0.05–0.1 mm/yr), but high-slip-rate episodes before 75 ka (Figs. 12B, 12F). Although the case against a change in rate is arguable at the error limits of the data in some cases (e.g., Fig. 12B), in others the changes are well supported. In any event, despite contrasting displacement histories, the Wasatch, Toiyabe, Diamond, and Hurricane-Cedar faults all reached the same amount of total cumulative offset over the past 150 k.y. (∼30–40 m; Fig. 12), leading to similar 150–0 ka slip-rate values.

Still other faults in the central Great Basin (between long 119°W and 114°W) are characterized by a constant slow slip rate over the past 150 k.y. (Fig. 12E), and have accordingly large earthquake recurrence intervals (>20 k.y.; Fig. 9). Considered as a whole, many of these faults did not rupture during the Holocene (i.e., Holocene slip rate of 0 mm/yr; Fig. 11A), whereas some of them ruptured once during the Holocene (e.g., Simpson, Crescent Valley, Schell Creek, Shawave, and Humboldt faults), leading to a virtually high Holocene slip rate, merely because the amount of displacement per event is divided by only 15 k.y. (Figs. 11 and 14). From 75 to 15 ka, faults that did not rupture during Holocene had 1–3 earthquakes (Fig. 15), yielding a rather homogeneous regional strain release pattern with individual fault slip rates of 0.05 mm/yr over the past 75 k.y. (Fig. 11B). In southern Nevada (south of lat 38°N), normal faults (with one exception) appear inactive or almost inactive over the past 500 k.y. (slip rate is 0 mm/yr, or <0.05 mm/yr; Fig. 11D).

Evidence for the pre–150 ka episodes of rapid slip rate in the Great Basin region is found along the Sevier-Toroweap and the Beaver Basin faults, which occurred between ca. 400–200(?) ka and 500–400 ka, respectively (Figs. 12F and 13C). The Beaver Basin episodes are questionable at the limits of error, but the most probable histories suggest episodes that last 100–200 k.y. or less, and accommodate ∼50–80 m total of vertical displacement.

Somewhat surprisingly, the database contains only two faults systems in the Great Basin that seem to accumulate displacement at 0.2–0.3 mm/yr that do not appear to vary in time: the faults within or near the CNSB (e.g., Dixie Valley and Wassuk faults) and the Anderson segment of the Hurricane system (Figs. 12 and 13). These specific temporal behaviors are discussed further in the following.

Do Great Basin earthquakes occur in a random pattern, or are the kinematics of strain release in some way systematic? The catalog of historic and paleo-earthquakes represents only a fraction of the total number of earthquakes Even so, there is at present no basis to infer that the inventory of observed events and rates are biased by either magnitude and timing of displacement (except as noted above pertaining to the surface expression of events of M < 6), and the spatial distribution of the sample is fairly even across the region. Thus, if there are systematic spatial patterns in strain release, these are likely to be expressed in the data, especially for events that occurred after ca. 40 ka.

We depict the earthquake kinematics in (1) a GoogleEarth animation of the paleo-earthquake distribution over the past 100 k.y. (Supplemental Information: Paleoeq_catalog.kmz; see ^{footnote 1}), (2) four snapshots grouping paleo-earthquakes of the entire region into finite time intervals (Fig. 14), and (3) four snapshots grouping paleo-earthquakes in the central Great Basin only, restricted to selected faults having a complete paleo-earthquake record since 60 ka (Fig. 15). From these data collectively, three spatial kinematic patterns can be identified.

The first earthquake kinematic pattern is the well-documented group of historic earthquakes in the CNSB (long 118°W), which we refer to as an along-strike cluster. We use the term cluster to indicate temporal, not spatial, concentration. Between A.D. 1915 and 1954, one earthquake per segment occurred on six along-strike fault segments in the CNSB. From the Paleoeq_catalog.kmz file, we can observe that another along-strike cluster may have occurred in the CNSB ca. 3 ka, but the region has otherwise been quiescent in Holocene time.

The second pattern is the occurrence of many earthquakes on the same fault segment. This is what is observed at long 112°W, along each of the six segments of the Wasatch fault system. No earthquakes have occurred along the Wasatch front since A.D 1400 (ca. 0.5 ka B.P.), but this historic quiescence was preceded by clusters of events between 15 and 0.5 ka (≥5 earthquakes per fault segment, mean recurrence interval of ∼2 k.y.; Figs. 9 and 14). We refer to clusters of events well above the average frequency on a given fault segment, or on a group of along-strike segments, as local clusters.

The third, and perhaps most intriguing, spatial pattern in strain release is observed across the central and eastern Nevada normal faults over the past 60 k.y. (Fig. 15). Here, individual faults have long earthquake recurrence times (>20 k.y. on average; Fig. 9), including some faults that ruptured only once over the past 60 k.y. (Egan, Toquima, Santa Rosa; Figs. 2 and 15). When considering the entire group of faults, however, a succession of earthquakes is distributed across strike on several parallel faults (one earthquake per fault), such that most of the faults across the region have ruptured at least once every 30 k.y. (e.g., from 60 to 30 ka, and from 30 ka to present day; Fig. 15). We refer to this mode of strain release as regional distributed faulting. The age uncertainties of earthquakes preclude determining whether these earthquakes are clustered or diffuse (regional strain release all at once, or gradually), or whether there is any spatial pattern (e.g., east to west). Even though we do not have data on all of the faults in central and eastern Nevada, the faults we show in Figure 15 have either a complete paleo-earthquake record over the past 60 k.y. (from paleoseismological trenching data) and/or have dated offset markers. Thus, the data set has not likely missed any local clusters within the set of faults sampled that would have resulted in a large cumulative offsets, and consequently a high slip rate, over the past 60–75 k.y. (Fig. 11B).

Our results indicate that whereas slip rates tend to be relatively high and steady through time along strike-slip faults nearest the plate boundary (San Andreas fault and WLB-ECSZ), they are relatively slow and episodic over most of the deforming northern Great Basin. This raises the possibility that in deforming continental domains, the more distributed the fault system, the more episodic or variable the strain release may be on any given fault. We develop this general hypothesis in the following, in light of the three modes of episodic behavior described here: along-strike clusters, local clusters, and regional distributed faulting. We present a kinematic model that emphasizes reconciling these modes, observed at time scales of tens to hundreds of thousands of years, with the more uniform pattern of strain at the time scale of 1–10 m.y.

The areas where we observed relatively constant fault slip rates through time include the WLB-ECSZ, and certain normal faults in the southern Great Basin (e.g., southern Nevada faults, including the Hurricane Anderson section). The WLB-ECSZ is a narrow, rapidly deforming zone, characterized by strike-slip faults with long-term slip rates in the range of a few millimeters per year. In general, preservation of evidence for the vertical component of slip back through time is better than for the strike-slip component, and accordingly fewer pre-Holocene offset markers have been identified for the strike-slip faults of the WLB-ECSZ than for normal faults (Fig. 11). For a small subset of documented strike-slip faults, Holocene and 75–0 ka slip rates (2–5 mm/yr) have magnitudes comparable to long-term slip-rate values over the past 5–10 m.y. (northern WLB, Fish Lake Valley; e.g., Faulds et al., 2005; Frankel et al., 2007) suggesting that slip rates of the WLB-ECSZ faults are rather constant in time (e.g., Lee et al., 2009), notwithstanding evidence to the contrary south of the Garlock fault zone (e.g., Oskin and Iriondo, 2004). Wesnousky (1988) suggested that in the ECSZ, fault traces may lengthen and simplify with time, as a function of cumulative geologic offset. The constancy of slip rate in time along the WLB-ECSZ strike-slip faults may thus be a characteristic of a simplified fault trace geometry, related to both their relatively high long-term slip rate and cumulative offset. Normal faults located within the WLB-ECSZ along the Sierra Nevada front also had a relatively high, apparently constant, slip rate with time (e.g., the Genoa, Hilton Creek, Wassuk, and Sonoma Junction faults; Rood et al., 2011).

Over a 500 k.y. time period, normal faults of the southern Great Basin have been inactive or almost inactive (slip rate of 0 or <0.05 mm/yr), whereas constant average to above average displacement rates (0.3–0.5 mm/yr) have been absorbed on the Anderson segment of the Hurricane fault (Fig. 12F). In southern Nevada (south of lat 38.5°N), due to plate configuration geometry, the deforming zone between the WLB-ECSZ (active kinematic boundary of the system) and the stable Colorado Plateau is much narrower than in northern Nevada. The proximity of the WLB-ECSZ to a large, relatively stable block in the plate interior at the latitude of Las Vegas may have thus in some way promoted localization of strain along the edge of the Colorado Plateau. An alternative or perhaps concomitant explanation may lie in the fact that the southern Great Basin crust has extended about twice as much as the northern Great Basin (Wernicke et al., 1988; McQuarrie and Wernicke, 2005), and for the past few million years probably was at a mean elevation of 1 km or more lower than the northern Great Basin, along with most of the slowly deforming southern Basin and Range in Arizona and Sonora, Mexico. Under these conditions, the gravitational potential of the lithosphere may have been suppressed as a driver of extension, resulting in low Quaternary strain rates relative to the northern Basin and Range.

Despite generally steady slip rates, a temporal slip-rate change has been noted in the WLB-ECSZ and its nearby faults between late Pleistocene and Holocene or historic time. The right-lateral Owens Valley and Warms Springs faults underwent drastic slip-rate decreases during Holocene time (Bacon and Pezzopane 2007; Kirby et al., 2008; Gold et al., 2013; Fig. 11). At the same time, slip rates increased on the Wassuk, Lone Mountain, and Sierra Nevada frontal faults, faults within the CNSB, and the left-lateral Garlock fault (Figs. 11 and 12; details in Neotectonics_data.xls; see ^{footnote 1}). This regional-scale switch of activity may be akin to the kinematic model of Dolan et al. (2007), where activity along the greater San Andreas system (San Andreas fault and Garlock fault) is suppressed (but not entirely eliminated) during activity on dextral faults in the ECSZ, and vice versa. Data summarized by Dolan et al. (2007) suggest that the two modes alternate every 2–3 k.y. Following on this idea, periods of activity on the San Andreas and the Garlock fault may be associated with normal faulting events (the east-west extensional component of the WLB-ECSZ) rather than northwest-striking dextral faulting in the WLB-ECSZ area. However, further data on other fault system of the WLB-ECSZ, as well as a longer slip-rate history on the Garlock fault, will be needed to test this hypothesis.

The northern Great Basin normal fault system (northern Nevada–western Utah), which represents the widest extensional domain of the Basin and Range province, shows striking differences in slip rate when considering the 15–0 ka, 75–0 ka, and 150–0 ka time windows. As noted herein, the long-term average rate for normal faults is estimated to be 0.2–0.3 mm/yr over the past 5–10 m.y. Our results showed that the Wasatch faults, some central and eastern Nevada faults, and the Cedar segment of the Hurricane fault had alternately high and low transient slip-rate episodes (1–2 mm/yr versus 0.05–0.1 mm/yr, respectively), values that are one order of magnitude higher or lower that the long-term average rate. Despite their contrasting displacement histories, these faults have nonetheless accumulated roughly the same amount of total vertical offset over the past 150 k.y. (∼20–40 m; Fig. 12), and so have similar 150–0 ka slip-rate values of 0.2–0.3 mm/yr. Distinct from these faults, however, several faults in northern Nevada have had persistently low slip rates (0.05 mm/yr or less) even over the past 150 k.y. (Fig. 11C). This suggests that transient slow episodes on northern Great Basin faults may persist for more than 150 k.y.

Constraining the 150 k.y. slip-rate evolution of northern Great Basin normal faults is crucial because it provides temporal context for the 15–0 ka slip rates. Our results show that the 15–0 ka time window catches fast transient episodes on various Wasatch fault system segments (1–2 mm/yr), yielding a highly misleading estimate of its longer term slip rate if extrapolated farther back in time. In addition, due to the large earthquake recurrence interval for central Nevada faults (>20 k.y.), the calculated Holocene slip rate can be either rather high, if there has been a Holocene earthquake, or zero, if there hasn’t, neither of which is accurate. In fact, basically all of these faults had a slip rate of the same approximate low magnitude (0.05 mm/yr) from 75 to 0 ka (Fig. 11B). The 150–0 ka fault slip-rate values and displacement histories thus suggest that faults of the northern Great Basin can be considered as similar in terms of total amount of strain release over the past 150 k.y., with the major differences simply a function of the precise timing of rapid transients relative to the various time windows.

These observations are compatible with the belt migration model proposed by Wallace (1987), in which along-strike clusters (e.g., the historical along-belt cluster occurring between A.D 1915 and A.D 1954 in the CNSB) of strain release migrate regionally, activating different north-south fault belts through time. Periods of evenly distributed strain release without fast transient episodes on individual faults (local clusters) can be observed between 75 and 15 ka in the northern Great Basin (Figs. 12B, 12C, 12E), in association with a pattern of regional distributed faulting (Fig. 15). They demonstrate that strain is constantly accumulating over the entire region at some level through time (e.g., Hammond et al., 2014). However, this period is only transient and occurrence of transient fast episodes on individual faults (during 15–0 ka and 150–75 ka; Fig. 12) would seem to be needed to release the total regional strain accumulation over time. However, as we do not have geologic slip-rate constraints for all the faults in the Great Basin, we cannot at present determine if the total amount of strain release over the region varies with time. Assuming long-term constant kinematic boundary conditions, periods of lesser total strain release (e.g., the 75–15 ka period in central Nevada and western Utah) would imply phases of regional transient strain accumulation, i.e., variations in the elastic loading rate. Further work would thus be needed to quantify the total amount of regional strain release through time, the major issue being the incompleteness of fault slip-rate values over the region.

We cannot assess whether fast transient episodes in the northern Great Basin are triggered by temporal changes along the WLB-ECSZ and the plate boundary system (see previous discussion herein), because displacement histories along the WLB-ECSZ are limited to 75 ka. The only compelling case we can make is that in diffusely deforming areas, such as the fault system in the northern Great Basin east of the WLB-ECSZ, strain release is strongly episodic. Faults within the CNSB are a component of this group, having the same long-term rate as other Great Basin normal faults (0.2–0.4 mm/yr since 10 Ma; Wallace and Whitney, 1984). The CNSB fault system is surrounded by other faults having long recurrence intervals (Fig. 9) and slow average slip rates over the past 75 k.y. (Figs. 11A, 11B). The CNSB stands out as having a relatively constant slip rate since 150 ka due mainly to the historic seismicity of the last century (1915–1954 sequence) (Fig. 12A). Without a ca. 150 ka offset constraint, and the very recent events, it would be considered an anomalously slow fault, similar to those surrounding it.

From the glimpse of strain release behavior available from the data set, we can piece together a working hypothesis to reconcile all our results. In this simple kinematic model (Fig. 16), we assume that (1) strain is only released by earthquakes and (2) the vertical displacement per event is 2 m (see Fig. 5). It consists of three identical curves, shifted in time, that depict slip histories on three individual longitudinal fault groups (e.g., the Wasatch system, central Great Basin faults, and some eastern Great Basin faults), based on the observations summarized in Figures 12 and 13. These observations permit estimation of the amount of total offset during each episode, and the duration of transient episodes on individual faults (either fast or slow), as explained further in the following. In addition to slip-rate histories, we also integrate into our model the earthquake kinematic pattern we described previously, i.e., the local clusters (episodes of events on a single fault) and regional distributed faulting (episodes of events distributed across several parallel faults, each with a single event).

Transient slow episodes occurred on the Toiyabe and Diamond faults (0.03–0.1 mm/yr) from 80 ka to the present day, and on the Wasatch system (0.07–0.13 mm/yr) from 250 to 15 ka (Figs. 12B, 12C, respectively). In addition, a number of faults in central Nevada have had a consistently slow slip rate (0.05 mm/yr) for the past 150 k.y. (e.g., Toquima, Monitor, and Egan; Fig. 12E), suggesting that a transient slow episode on an individual fault may last at least 150 k.y. Typically, the strain release rates on a given fault during these slow-slip-rate episodes (0.03–0.1 mm/yr) are as much as an order of magnitude lower than their long-term mean rate. An upper bound for the duration of transient slow episodes can be estimated from the Beaver fault displacement history in western Utah. Despite uncertainties in the age of the offset datum and cumulative displacement, a fast transient episode occurring ca. 400 ka was followed by a transient slow period lasting from 400 ka to the present day (Fig. 13C). We thus estimate that, during regionally distributed faulting, a transient slow-slip-rate episode may last at least 150 k.y., and perhaps at least 400 k.y., for normal faults in the central and eastern Great Basin.

The best example of transient fast slip is the case of the Wasatch system during Holocene time (1–2 mm/yr), such that local clusters for each segment combine with along-strike clustering of the entire system (Figs. 12C and 14). For the Wasatch system, the mean recurrence time for local cluster of events is ∼2 k.y. (Fig. 9). If we make the reasonable assumption that large cumulative displacement is always released by surface-rupturing events in the Great Basin region, past transient high-slip-rate episodes observed on several faults (revealed by offset Quaternary markers, without paleoseismic data) were also associated with local clusters. Possible faults having had local clusters in the past are thus the Hurricane-Cedar, Beaver, Sevier, Toiyabe, and Diamond faults (Figs. 12 and 13). Based on the Wasatch fault system, we infer that the rate of displacement during a local cluster (1–2 mm/yr) is ∼3–6 times faster than the average long-term normal fault slip rate in the Great Basin (0.3 mm/yr). A lower bound for the duration of a local cluster, as well as its total displacement, can be estimated from the Wasatch fault system. The duration is at least ∼ 10 k.y., over which ∼15 m of vertical displacement occurred on most of the segments. However, some of the fast transient episodes in the past have displacements are much larger than this. For example, the Toiyabe and Beaver faults appear to have had episodes of 40 m and 75 m of rapid vertical displacement, respectively (Figs. 12B and 13C), although the uncertainties are large on these estimates. Assuming an associated slip rate of 1.5 mm/yr, the local cluster duration would be ∼25 k.y. and 50 k.y. for the Toiyabe and Beaver faults, respectively. Another way to estimate the duration of a local cluster is purely arithmetic: we assume the that long-term vertical slip rate of 0.3 mm/yr is equal to the combination of a transient slow period (0.1 mm/yr lasting 200–400 k.y.; see previous discussion herein) and a transient fast period [1.5 mm/yr, lasting (x) k.y., where (x) represents the duration of a local cluster]. This yields a local cluster duration of 50 ± 20 k.y. It thus seems quite possible that the Wasatch local cluster may be ongoing, and may last perhaps another 20–50 k.y.

The question becomes, is there any spatial or temporal pattern to the occurrence of the local clusters? For example, do they migrate from east to west, west to east, north to south, or in some other systematic fashion? There may be an eastward migration of local clusters from the central to the eastern Great Basin (Fig. 16). Another possibility is that the location of local clusters is random through time. The data set at present is inadequate to address this question because there is only a small subset of faults for which we can identify either ancient or active local clusters.

Overall, both strain accumulation (inferred from geodetic data) and strain release are currently concentrated into western Nevada and western Utah (e.g., Bennett et al., 2003; Koehler and Wesnousky, 2011). As is apparent from both the geodetic and geologic data sets for Basin and Range normal faults, however, contrasts in rates across different time scales may be the rule rather than the exception (Friedrich et al., 2003, 2004). Since 10 ka, some areas have been both accumulating and releasing strain much more slowly than average, and others much more quickly. This is shown on the finite slip-rate map (Fig. 11A), which is dominated by fault slip rates either one order of magnitude higher (in red, near the margins) or lower (in dark blue and dashed blue, near the center in central Nevada) than the long-term average fault slip rate for a typical Basin and Range normal fault (∼0.2–0.3 mm/yr; yellow). The consistency between geodetic rates and fault slip rate over the past 15 k.y. (Fig. 11A) suggests that the present-day strain accumulation pattern would have remained the same over at least the past 15 k.y.

Neither the geodetic nor Holocene strain release rates, which broadly agree, reflect longer term patterns, implying that strain accumulation patterns likely also vary with time. Over the past 150 k.y., the deformation pattern changes and becomes more homogeneous and distributed, with far more faults having an average slip-rate value closer to the long-term slip rate of 0.3 mm/yr (Fig. 11C). Within the large area of slow fault slip rate in central Nevada, geodetic data indicate that strain accumulation is only ∼10% or less than the Great Basin average (∼1 nstrain/yr or less, versus the 10 nstrain/yr average; Bennett et al., 2003; Hammond et al., 2014). Hiatuses in strain release on the 15, 75, or 150 k.y. time scales indicate that elastic strain cannot be accumulating on these faults at anywhere close to the long-term average rate of 10 nstrain/yr, because such strains are far too large to be accommodated by elastic accumulation only. As noted earlier, 10 nstrain/yr applied across a nominal 30 km aperture for a single mountain range (0.3 mm/yr) would yield 1 earthquake every 6.7 k.y., assuming the horizontal strain release is 2 m. For time intervals of 15, 75, and 150 k.y., assuming all strain is released in earthquakes, a record of ∼2, 10, and 20 earthquakes, respectively, are far below recorded levels of seismic strain release.

It is increasing apparent that climate-driven hydrological effects may influence earthquake occurrence, presumably because hydrology influences pore pressure and the vertical stress levels at depth, which in turn influence frictional resistance to sliding. Among the clearest examples of this is the relationship between the Asian monsoon and seismicity along the Himalayan front (Bollinger et al., 2007). In the case of the Basin and Range, lithospheric rebound during deglaciation has been implicated as a driving mechanism for the Wasatch earthquake cluster during the Holocene (e.g., Hetzel and Hampel, 2005). In the model we propose for episodic strain release in northern Nevada–western Utah (Fig. 16), it would seem that local cluster duration (50 ± 20 k.y.) and frequency (every 200–400 k.y.) may well be modulated by some sort of Milankovitch or other climatic forcing. The strongest candidates for such modulation, in addition to the Wasatch, include the Toiyabe, Diamond, and Hurricane Cedar fast transient episodes, if our interpretation is correct that they are similar to the Holocene local cluster on the Wasatch fault system. These events occurred prior to 100 ka, and thus may be coeval with profound deglaciation at the transition from Marine Isotope Stage (MIS) 6 to MIS 5. Similarly, the time period between ca. 75 and 15 ka, characterized by transient slow slip rates on all individual faults of the northern Nevada–western Utah region (Figs. 12B, 12C, 12E), including regional distributed faulting (Fig. 15), corresponds to the Wisconsin glacial period (MIS 4 to MIS 2).

Our present neotectonic database helps to better characterize the spatial and temporal evolution of strain release in diffusely deforming continental domains, with potential implications for seismic hazard analysis. Based on 150 displacement per event measurements of surface rupturing earthquakes, we show that approximately two-thirds of normal faulting earthquakes in the Great Basin had vertical displacement of 1–3 m per event (Fig. 5). In 64 cases where vertical displacement of an event can be compared with a previous event on the same fault, the difference is <± 1 m in approximately two-thirds of the measurements (Fig. 7). The distribution of earthquake recurrence intervals is far more variable, with a mode of 1–3 k.y. and ∼15% of recurrence intervals >20 k.y. Faults with the longest earthquake recurrence intervals (>20 k.y.) are concentrated in northern Nevada. Our earthquake slip and timing data suggest no clear correlation between displacement per event and earthquake recurrence interval.

Based on fault slip-rate values determined for different time windows and fault displacement histories, it is possible to investigate fault slip-rate evolution in the WLB-ECSZ and wider Great Basin region over the past 150 k.y., and in some cases to the 1 m.y. time scale. Strike-slip faults of the WLB-ECSZ mostly show slip-rate values of 2–5 mm/yr over the past 75 k.y., in good agreement with their long-term rate, suggesting relatively steady slip rate through time. Whereas normal faults located within or nearby the WLB-ECSZ show relatively constant slip rates of 0.3 mm/yr or greater over the past 150 k.y., normal faults in southern Nevada seem to be inactive or nearly inactive over the past 500 k.y. (0–0.05 mm/yr). Despite the relatively steady deformation at the 100–10 ka time scale in the WLB-ECSZ, a regional temporal change in slip rates involving feedbacks between the San Andreas fault and surrounding fault systems may have occurred during the Holocene (Dolan et al., 2007).

In contrast to the WLB-ECSZ, widely distributed normal faults in the northern Great Basin (northern Nevada–western Utah) show marked slip-rate variation and episodicity in strain release. The Wasatch fault system, some faults in central Nevada, and the Hurricane Cedar fault underwent alternating transient high-slip-rate episodes (1–2 mm/yr) and low-slip-rate episodes (0.05–0.1 mm/yr), with rates that are respectively an order of magnitude higher or lower that their long-term averages. Despite different displacement histories, these faults have accumulated similar amounts of net vertical offset over the past 150 k.y. (∼20–40 m; Fig. 12), leading to similar slip-rate values of 0.2–0.3 mm/yr over this time period, consistent with their long-term averages. The major discrepancies between the 15–0 ka and 150–0 ka slip-rate values may thus simply be an artifact of the 15–0 ka time window being too short to capture the overall behavior. For northern Great Basin normal faults, the distribution of paleo-earthquakes over the past 60 k.y. defines three major kinematic patterns. The first is an along-strike cluster, where a longitudinal belt of fault segments fail over a period of a few centuries or less. The second is a local cluster, an episode of repeated events on a single fault, with a recurrence interval of ∼2 k.y. (Figs. 12C and 14). The third is regional distributed faulting, a succession of earthquakes across several parallel faults, each with a single event (Fig. 15), that may or may not be temporally clustered.

We propose a kinematic model of strain release for northern Great Basin normal faults where they alternate between two modes (Fig. 16): (1) transient fast periods (1–2 mm/yr), lasting ∼50 k.y. and characterized by local clusters with recurrence times of ∼2 k.y.; and (2) transient slow periods (0.05–0.1 mm/yr), lasting 200–400 k.y. and characterized by regional distributed faulting with long recurrence intervals (>20 k.y.) on any given fault. The combination of modes yields an average rate of ∼0.3 mm/yr for each fault. The model must be considered quite tentative, because it hinges on a relatively small number of slip histories at longer time scales that have significant uncertainties in timing and displacement. Nonetheless, it focuses attention on what appear to be order-of-magnitude variations in slip rate that can be tested by focusing on new measurements on the 1 Ma to 100 ka slip histories of Great Basin normal faults. The results would address the fundamental question of whether the apparent episodicity of strain release in the intraplate region also reflects a dynamic environment of varying strain accumulation.

We are grateful to N. Chamot-Rooke, M. Delescluse, J. Davis, M. Fouch, W. Holt, R. Porter, and M. West for useful discussions. We thank the two anonymous reviewers for their valuable comments, which significantly improved the final version of the manuscript. This research was supported by the Earthscope Program of the National Science Foundation (grant EAR-10-53161 to Wernicke).