Evidence for a prehistoric multifault rupture along the southern Calico fault system, Eastern California Shear Zone, USA

Complex multifault ruptures pose a challenge to seismic hazard analysis and the understanding of fault dynamics. Such ruptures involve more than a single major fault in a single earthquake and typically involve complex secondary deformation at stepovers and intersections, as observed in recent examples from New Zealand, Alaska, central Nevada, and southern California (Caskey et al., 1996; Haeussler et al., 2004; Hamling et al., 2017; Ross et al., 2019; Chupik et al., 2022). This style of rupture is difficult to forecast or understand kinematically because surface ruptures commonly include unmapped faults or faults that were thought to be disconnected at depth (Ben-Zion and Sammis, 2003; Ross et al., 2019). Multifault surface ruptures may also be difficult to detect in the paleoseismic record and are thus generally known only from historic observations.

Three complex multifault ruptures have occurred in the past three decades in the Eastern California Shear Zone (ECSZ): the 1992 M_w 7.3 Landers, 1999 M_w 7.1 Hector Mine, and 2019 M_w 7.1 Ridgecrest earthquakes (Sieh et al., 1993; Treiman et al., 2002; Ross et al., 2019; Fig. 1A). The ECSZ is a broad deformation zone consisting of subparallel strike-slip faults that accommodate ~25% of the motion between the North America and Pacific plates and transfer slip from the San Andreas fault to the Walker Lane and Basin and Range to the north (Dokka and Travis, 1990; Savage et al., 1990; Li and Liu, 2006; Liu et al., 2010; Plattner et al., 2010; Dixon and Xie, 2018). Although the ECSZ formed ca. 10–6 Ma, it is considered immature due to the limited net slip it has accommodated and the geometric complexity of its individual faults (Andrew and Walker, 2016; Spotila and Garvue, 2021). This immaturity and associated kinematic complexity may be partly responsible for several of the ECSZ’s characteristics. The ECSZ exhibits a discrepancy between geological and geodetic slip rate determinations as well as a high degree of off-fault deformation (Oskin et al., 2007; Shelef and Oskin, 2010; Spinler et al., 2010; Milliner et al., 2015; Evans et al., 2016). The ECSZ also exhibits clustered rupture activity, with peaks of seismic activity in between quiescent periods in the Holocene (Rockwell et al., 2000; Madden et al., 2006; Ganev et al., 2010). What relationship multifault rupture behavior has with either the kinematic complexity or clustered earthquake activity in the ECSZ is unknown, however, in part because of the lack of constraints on prehistoric surface rupture patterns.

To further explore multifault ruptures and earthquake clustering in the ECSZ, we investigate the Calico fault system, a suite of faults in the central ECSZ between the Landers and Hector Mine ruptures (Fig. 1B). We identify and map a suspected multifault surface rupture on the Calico and Hidalgo faults. We also investigate the timing of this rupture with paleoseismology and relate the event to prior earthquake activity in the ECSZ. Finally, we conduct Coulomb stress change modeling for different rupture scenarios of the most recent event on the Calico-Hidalgo system to explore potential links to ruptures on the neighboring Landers or Hector Mine systems. Based on our observations and modeling results, we propose that multifault ruptures may be a regular mode of operation for the ECSZ, rather than an isolated recent phenomenon. However, these ruptures may not follow regular patterns or be easily related to observed earthquake clustering.

The ECSZ is an ~125-km-wide band of complex dextral shear that initiates near the San Andreas and Pinto Mountain faults in the south and transitions into the Walker Lane belt north of the Garlock fault. The number of primary faults in the southern ECSZ increases toward the south from six at ~34.7°N latitude (Helendale, Lenwood, Camp Rock, Calico, Pisgah, and Ludlow) to ~14 closely spaced (~10 km apart) faults near the Pinto Mountain fault (Fig. 1). Faults in the zone are generally northwest-striking and right-lateral, but surface traces are complex and display small-scale geometric complexities and widespread transpression (Spotila and Garvue, 2021). The ECSZ has accommodated a total of ~50–75 km of dextral slip over the past ~6–10 m.y. (Dokka and Travis, 1990; Glazner et al., 2002; Andrew and Walker, 2016). The implied long-term net slip rate of ~10 mm/yr is comparable to current geodetic modeled slip rates of 11–18 mm/yr (Savage et al., 1990; Miller et al., 2001; Spinler et al., 2010; McGill et al., 2015; Evans et al., 2016). Geodetic rates are greater than the sum of geologically determined slip rates for individual faults, however, which combine to ≤6.2 ± 1.9 mm/yr (Oskin et al., 2008). This discrepancy may be due to off-fault deformation, transient effects (e.g., post-seismic relaxation), or measurement uncertainty (Dixon et al., 2003; Oskin et al., 2007; Shelef and Oskin 2010; Chuang and Johnson, 2011; Johnson, 2013; Herbert et al., 2014a, 2014b; Milliner et al., 2015). Most individual faults in the ECSZ are interpreted as <1 mm/yr geological slip rates during the late Pleistocene based on paleoseismology and geomorphology (e.g., Hart et al., 1988; Rubin and Sieh, 1997; Rockwell et al., 2000; Madden et al., 2006; Oskin et al., 2008; Xie et al., 2018).

The ECSZ has experienced three complex multifault rupture events in the past three decades (Fig. 1A). The 1992 M_w 7.3 Landers earthquake was produced by a 70-km-long rupture zone (~85 km total surface rupture length) along the Johnson Valley, Homestead Valley, Emerson, and Camp Rock faults (Sieh et al., 1993). The rupture initiated in the south near the Pinto Mountain fault and propagated northwards sequentially across the faults, crossing several complex stepovers (Spotila and Sieh, 1995; Zachariasen and Sieh, 1995). Dextral slip during the event was variable along the faults and averaged several meters but reached a maximum of ~6 m and was accompanied by significant off-fault deformation (Sieh et al., 1993; McGill and Rubin, 1999). The 1999 M_w 7.1 Hector Mine earthquake occurred ~25 km to the west of the Landers rupture and was produced by a 41-km-long surface rupture zone (~45 km total surface rupture length). The rupture involved portions of the Lavic Lake, Pisgah, Bullion, and Mesquite Lake faults and was similarly complex, including conjugate and overlapping faulting, off-fault deformation, and complex stepovers, with an average dextral slip of ~2.5–3 m (Scientists of the U.S. Geological Survey, Southern California Earthquake Center, and California Division of Mines and Geology, 2000; Hauksson et al., 2002; Treiman et al., 2002). The 2019 M_w 7.1 Ridgecrest earthquake occurred ~120 km north of the Landers and Hector Mine ruptures on mainly unmapped faults (Thompson Jobe et al., 2020). It was preceded by a M_w 6.4 foreshock on a 15-km-long, northeast-striking sinistral fault and continued as ~50 km of northwest-striking dextral surface rupture during the mainshock (Ross et al., 2019). Like the earlier ECSZ earthquakes, this rupture was complex and included stepovers, cross (conjugate) faulting, and a high degree of off-fault deformation (Milliner et al., 2015, 2021). Together, these three events suggest that complex, multifault ruptures may be a common behavior in the ECSZ.

The southern ECSZ appears to produce clustered earthquake activity, based on an extensive paleoseismic record going back >20 k.y. There are about 20 paleoseismic sites in the Mojave block, about half of which are located on faults that ruptured during the Landers earthquake (Fig. 1B). Clusters of paleoseismic events identified thus far include the period of the present to ca. 1 ka (including Landers, Hector Mine, and Ridgecrest earthquakes), 5–6, 8–10, and ~15 ka (Rockwell et al., 2000; Madden et al., 2006; Ganev et al., 2010; McAuliffe et al., 2013). Data from Rockwell et al. (2000) indicate that specific sequences of multifault surface ruptures that comprise individual earthquake scenarios (e.g., the 1992 sequence) are not necessarily repeated, suggesting episodic, spatially variable strain release may depend more on the dynamics of specific earthquakes (e.g., rupture propagation direction) than on the prior rupture history of individual fault segments. Ganev et al. (2010) hypothesized that variable loading rate may explain clustering, such that the modern seismic cluster may be related to a transient increase in elastic strain rate that has been detected geodetically (e.g., Spinler et al., 2010; McGill et al., 2015; Evans et al., 2016). An additional component to this explanation may be that clustered seismic activity and increased strain rate may oscillate (i.e., periods of more activity and less activity) regionally between the (1) ECSZ, Walker Lane, and southern San Andreas fault and (2) the Mojave San Andreas fault, Garlock fault, and structures of the Los Angeles Basin and Transverse Ranges (Dolan et al., 2007; Madden Madugo et al., 2012). Such a spatiotemporal evolution could then be locally modified by the dynamics of rupture or triggering due to static stress loading at rupture endpoints (McAuliffe et al., 2013). However, existing data on the spatiotemporal evolution of ruptures, including the specific rupture sequences or patterns of multifault ruptures of individual events identified in the paleoseismic record, are too sparse to determine what controls clustering or rupture dynamics within the ECSZ.

In this study, we focus on the Calico fault system, which lies midway between the historical Landers and Hector Mine ruptures and includes the Hidalgo fault (also known as the Surprise Springs fault; Londquist and Martin, 1991) and Mesquite Lake fault (Fig. 1B). The Calico fault system connects with the Blackwater fault in the north to form the longest (>140 km) and most continuous fault of the Mojave ECSZ (Oskin et al., 2007). The Calico fault system also exhibits the greatest net slip in the southern ECSZ. Unlike most ECSZ faults, which exhibit ~1 km or less dextral offset, net slip on the Calico fault is 8–9.8 km, while net slip on the combined Mesquite Lake–Bullion–West Calico fault system may be as high as 20 km (Dokka and Travis, 1990; Glazner et al., 2000; Jachens et al., 2002; Andrew and Walker, 2016). The central Calico fault in the northern Rodman Mountains (Fig. 1B) also has the highest geologically determined slip rates in the southern ECSZ of 1.8–3.1 mm/yr (Oskin et al., 2007, 2008; Xie et al., 2018). Geodetically estimated slip rates for the fault are higher, ranging up to 8–11 mm/yr (McGill et al., 2015; Evans et al., 2016), although these are poorly constrained modeled rates due to the difficulty of discriminating loading rate of an individual strand where faults are closely spaced. These estimates suggest the Calico fault system may accommodate a major fraction of the southern ECSZ’s displacement budget. The Calico fault exhibits a strong geologic and geomorphic expression (Dibblee, 1964, 1966, 1967a, 1967b) and has experienced at least four Holocene surface ruptures based on one paleoseismic site near the Newberry Mountains (Ganev et al., 2010; Fig. 1B). The kinematic rupture behavior of the fault system is unconstrained, however, and it is unknown whether the Calico and Hidalgo faults rupture jointly in complex events or whether they rupture in concert with neighboring faults, such as the Mesquite Lake or Pisgah faults (Fig. 1B). In this study, we focus on the southern Calico fault system, 70 km of which lies within a limited-access military installation (Marine Corps Air Ground Combat Center [MCAGCC] at Twentynine Palms, California) (Fig. 1B) and has thus received limited previous investigation.

We mapped continuous geomorphic evidence of an ancient surface rupture (paleorupture) on the Calico-Hidalgo fault consisting primarily of lineaments and neotectonic features cutting young geomorphic surfaces. Semi-continuous, light-colored lineaments on alluvial and colluvial surfaces were mapped at ~1:500 using Google Earth satellite imagery (Figs. 2 and 3). The resolution of this imagery is detailed enough to identify and map features as small as single tire tracks and individual ~1-m-diameter boulders and shrubs. Viewing the imagery draped on topography in 3D greatly facilitated the identification of fault features. Mapping was locally supplemented using high-resolution topography from Structure-from-Motion (SfM) photogrammetric analysis of aerial imagery from a small consumer drone (DJI Mavic Pro; Fig. 4). Although airborne light detection and ranging (lidar) is not available for the study area, we did cross-check all mapping using the 2 m resolution EarthDEM data set (Porter et al., 2022; https://www.pgc.umn.edu/data/earthdem). In general, the higher-resolution Google Earth satellite imagery, along with the color and textural information, was more valuable for mapping the paleorupture in comparison to the digital elevation model (DEM) data.

We ground-truthed mapped lineaments in the field along the majority of the paleorupture (Fig. 2). Features identified in the field include narrow zones (0.5–1 m wide) of ground disturbance (e.g., disarticulated cohesive soils), scarps on alluvial and colluvial surfaces, shutter ridges, benches, degraded mole tracks, and offset or deflected streams (Figs. 5 and 6; Part B of Supplemental Material¹). The occurrence of faulting beneath mapped lineaments was locally confirmed by fresh fault gouge in gully exposures along the lineaments (Fig. 3A). We generally noted a high degree of accuracy in our satellite-imagery–based map of the paleorupture during field examination. Narrow erosional gullies and streams were easily distinguished from faulting based on their downhill paths, whereas the paleorupture generally runs along ridge contours or breaks in slope and cuts across gullies. The paleorupture is also distinct from vehicle tracks, given that nearly all vehicle paths within the MCAGCC are from large, dual-tracked military vehicles (i.e., no motorbike trails). Recent faulting is also easily distinguished from footpaths related to military training or earlier (possibly prehistoric) human activity. Footpaths consist of narrow troughs (~20 cm wide) with firm, smooth surfaces due to repeated traffic and manual removal of obstructions (Fig. S1 [^{footnote 1}]) and tend to cluster near locations of human interest (e.g., tactical vantage points). Footpaths also follow microtopography by contouring in and out of gullies, following ridgelines, avoiding steep talus slopes, and curving around obstructions, such as boulders or shrubs. In contrast, the paleorupture follows preexisting mapped fault traces straight across alluvial and colluvial surfaces as if cut by a vertical fault plane. The paleorupture trace also cuts across steep gullies and talus slopes where footpaths would be impractical. As a result, we have a high degree of confidence in our map of the paleorupture extent, at least where geomorphic surfaces provided good preservation. Offset features along the mapped paleorupture were measured between piercing points primarily in the field using a tape measure.

Paleoseismic investigations were conducted at a single location along the paleorupture of the Hidalgo fault on the southwest flank of Hidalgo Mountain (Fig. 2). We hand-excavated two shallow fault-perpendicular trenches (4-m-long T1 on the east and 10-m-long T2 on the west) in a small depositional zone formed by an uphill-facing scarp in October 2019 and January 2020 (Figs. 7 and 8). The trenches are separated by a 3-m intervening zone of boulder-rich deposits that prevented the excavation of a single continuous trench. The trenches are located along a stretch of the paleorupture that is geomorphically sharp and young in appearance (see 4.1 Paleorupture Observations section). Trench walls were cleaned, gridded by hand, and photographed. We constructed orthorectified photomosaics from the trench photographs using the SfM photogrammetric software Agisoft Metashape following the procedure of Bemis et al. (2014). Trench stratigraphy and structure were mapped onto the photomosaic logs (Fig. 8). We collected eight sediment samples for luminescence dating (Table 1) from the sequence of units constraining paleoearthquake timing. Samples were collected with 1.5- to 2-inch-diameter steel and PVC tubes that were driven into the trench wall using standard field collection procedures (additional methods provided in Part C of Supplemental Material). Ages were measured on aliquots of both quartz (optically stimulated luminescence [OSL]) and feldspar (infrared stimulated luminescence [IRSL]). These samples provide the only age control of the paleorupture, given that no charcoal or other datable material was observed in trench exposures.

We found mappable geomorphic evidence of paleorupture along a continuous 18-km-long zone of the Calico and Hidalgo faults (Fig. 2). The paleorupture is expressed as a semi-continuous network of anastomosing thin, light-colored lineaments and surface disruption that follows the originally mapped fault traces (Fig. 3). We ground-truthed 58% of this surface rupture trace in the field (indicated by red and salmon traces in Fig. 2), whereas the remainder was mapped only using satellite imagery due to difficulty of access (indicated by yellow and white traces in Fig. 2). Red and yellow traces indicate where the paleorupture is well defined by evidence (44% of mapped traces), whereas salmon and white traces indicate where the surface rupture is difficult to identify (e.g., on steep, rubbly surfaces) and therefore inferred. The paleorupture is most clearly defined on the Calico fault along the east flank of Middle Hidalgo Mountain and the Hidalgo fault along the west flank of Hidalgo Mountain. It is less well constrained along the Calico fault on the northeast flank of Hidalgo Mountain and on an east branching strand of the Calico fault on North Hidalgo Mountain (indicated by blue traces in Fig. 2). The surface rupture trace in those areas is discontinuous and less clear, either due to poorer preservation, smaller displacement, or because it was produced by a separate (i.e.) earlier, rupture event.

The position of the northern and southern limits of the mapped paleorupture likely does not reflect the full surface rupture length, due to limited preservation of ephemeral fault features. The paleorupture becomes indistinct in the north along North Hidalgo Mountain, where piedmont surfaces are sandy and loose due to mobilized sand from an older folded Quaternary alluvium that comprises the ridge to the east (Dibblee, 1966). The paleorupture disappears south of Hidalgo Mountain, where the basin alluvium of the lowlands is similarly sandy and loose due to significant aeolian sand content derived from westerly prevailing winds. The paleorupture is similarly not mappable across areas with aeolian sand “pockets” that occur locally along the west flank of Hidalgo Mountain. In contrast, the paleorupture is best preserved in dissected piedmont alluvial surfaces or gently dipping colluvial slope deposits, which are less geomorphically active, older, and thus contain more cohesive and resistant soils and pavement. Given that the northern and southern limits of the mapped paleorupture may reflect preservation, it is plausible that the surface rupture originally continued farther in both directions.

An important feature of the mapped paleorupture is the apparent recency of the fault features. Mapped lineaments on satellite images appear like those along the historic surface ruptures of the 1992 Landers and 1999 Hector Mine earthquakes (Fig. 3B). The rupture traces from these historic earthquakes follow a similar pattern of good preservation in cohesive older geomorphic surfaces and poor preservation in sandy basin alluvium, even though these events occurred only a few decades ago. Field observations of the mapped paleorupture are also indicative of recent faulting. The paleorupture is composed of a 0.5-m- to 1-m-wide zone of loose surficial deposits, in which desert pavement on alluvial or colluvial surfaces has been disarticulated (Fig. 5D). The disrupted ground is broken and softer than adjacent surfaces and is commonly recognizable by a subtle break in slope (Figs. 4 and 5F). Locally, the surficial deformation zone is marked by the redeposition of fines that foster vegetation growth. The fault zone is also locally reminiscent of a degraded mole track (Fig. 5D). Many of the sharp features indicative of fault offset, such as small shutter ridges, are also manifest in soft erodible colluvium or fault gouge, suggesting a minimal duration of time since formation. The preservation of the zone of disruption is particularly noteworthy, given that vertical separation is generally absent along most of the paleorupture. A small fault scarp (0.5–1 m high) occurs only locally on the northern Hidalgo fault and exhibits a gentle, convex morphology due to the coarse, non-cohesive, young deposits in which it occurs. The paleorupture trace is also lost in active washes that presumably resurface at the decadal timescale. We found geomorphic evidence for the freshness of the paleorupture to be prevalent along the entire 18 km mapped trace and did not observe a systematic change in character along it, except for traces mapped in blue in Figure 2. Based on the continuity and apparent degree of preservation of recent faulting, we interpret the mapped feature to be the most recent surface rupture along this portion of the fault system, produced by a single, recent rupture event.

Observed displacements along the paleorupture are almost exclusively dextral strike-slip (Fig. 6). The mapped trace of the surface rupture defines a straight (or nearly straight) line across ridges, channels, and ephemeral gullies, indicating a near-vertical dip. The responsible fault is also near vertical in exposures (Fig. 3A), consistent with a predominantly strike-slip fault. We measured 37 right-lateral offsets with minimal vertical displacement (Table 2; Part B of Supplemental Material [^{footnote 1}]). Offset features included ephemeral gullies, shutter ridges in colluvium, spur ridges, and debris-flow lobe deposits (Fig. 6). Right-lateral offsets are as small as 0.7 ± 0.2 m but are typically ~1–3 m (Table 2). Also present are larger offsets, which are likely cumulative from multiple earthquakes given their magnitude and rounder, degraded morphology. We also observed numerous offsets >10 m but do not include them in Table 1 because they likely represent multiple events and longer duration of slip. Errors were assigned to each measurement based on the evaluated precision of our piercing points given the geomorphic expression and degradation of offset features in the field (e.g., sloughed risers, channel bank erosion, coarseness of substrate), which we estimate to represent 2σ uncertainties (Table 2). The distribution of dextral slip along the paleorupture is plotted in Figure 9, which excludes offsets >6 m (given that this exceeds expectations for slip magnitude based on historical ruptures in the ECSZ) and sums offsets that occur across parallel, overlapping fault strands. We constructed a cumulative offset probability distribution (COPD) from summed triangular probability distributions calculated from the minimum, maximum, and preferred measurements of each offset (Fig. 9). The largest peak on the COPD is ~2.3 m of right-lateral offset, which we adopt as the best estimate of the average slip for the mapped extent of the paleorupture. Other peaks at 6 m and 9 m right-lateral offset may represent multiple slip events, given that they are at the upper limit of observed slip in historical ECSZ rupture (Sieh et al., 1993; Scientists of the U.S. Geological Survey, Southern California Earthquake Center, and California Division of Mines and Geology, 2000; Madden et al., 2013). The 1-km-running average of 2.5 ± 0.6 m in offsets <6 m (n = 31, seven of which are grouped into one sub-average because they are closely spaced and occur on overlapping fault strands) across the mapped paleorupture matches closely with the largest COPD peak and exhibits a typical variation for displacement along a strike-slip rupture for historic earthquakes (McGill and Rubin, 1999; Gold et al., 2013; Fig. 9). This observation supports and strengthens the interpretation that the mapped features were produced as a single continuous surface rupture along the combined Calico-Hidalgo faults.

The paleoseismic trenches on the Hidalgo fault on the west flank of Hidalgo Mountain span the depositional zone of a partially closed, 20-m-wide trough formed by an uphill-facing fault scarp (Fig. 7). This trough contains two parallel fault strands, one of which forms the western trough margin and one that occurs near the trough midline. Trench T1 (4 m long, 1.5 m deep) spans the eastern portion of the trough and terminates on the west in the footwall of the fault strand (F1) that runs along the trough midline (Figs. 7, 8A, and S2). Trench T2 (10 m long, 0.5–1.5 m deep) occupies the western portion of the trough and terminates on the west in the footwall of the western fault strand (F2) (Figs. 7, 8B, and S2). The 3-m-long span between the trenches consists of a structural high of dense, boulder-rich colluvium that was too difficult to excavate by hand (Fig. S2). The structural high also divides source areas for the eastern and western portions of the trough, with distinct stratigraphic sequences exposed in the trenches on either side that cannot be directly correlated. The irregular depth of both trenches results from the occurrence of cobble and boulder-rich colluvium beneath the finer-grained units of the trough that were too difficult to excavate.

Both trenches expose sequences of well-defined, fine-grained deposits and interbedded cobbly units that were deposited in the trough. The fine-grained deposits were likely transported as sheetwash or deposited in standing water in the trough, possibly combined with aeolian deposition. Cobbly units are interpreted as hillslope colluvium mixed with talus and rockfall from the slope above to the east. Trench T1 contains a sequence of well-sorted, fine-grained sand deposits (units T1-100, T1-150, T1-200, T1-300, and T1-400) with two interbedded, poorly sorted, clast-to-matrix–supported cobble layers (T1-250, T1-350; Table 3; Fig. 8A). The upper cobble layer (T1-250) exhibits a wedge-shaped geometry and thickens upslope to the east. Both cobble layers are disrupted by faulting (F1) on the west. The fine-grained units generally thicken toward the center of the trench and are juxtaposed against the coarse colluvial unit (T1-C1) of the footwall by faulting (F1). In particular, T1-200 thickens toward the middle of the trench, corresponding with where the base of the unit has been down-dropped across several faults. Thin dashed lines in Figure 8A illustrate examples of intact blocks of T1-200 sediment that have been tilted and deformed adjacent to the F1 fault zone. The footwall unit (T1-C1) consists of poorly sorted, matrix-supported, angular cobble colluvium with minor isolated lenses of finer-grained deposits. Trench T2 consists of horizontally bedded, fine-grained sediments draped over a gently west-dipping, cobble-boulder deposit (T2-C1) (Table 3; Fig. 8B). The fine units are massive with limited internal stratigraphy and are differentiated by proportions of fine- to coarse-grained sand and are separated by sharp basal contacts. The uppermost unit (T2-100) is thickest above the western fault strand (F2) and pinches out ~4 m to the east. Unit T2-200 fines eastward from coarse sand and gravel adjacent to fault F2 on the west to medium sand in the east. Units T2-200 and T2-300 are cut by faulting on the west (F2), which juxtaposes the depositional trough against coarse, poorly sorted, angular colluvium (T2-C2). Fault zone F1 in trench T1 is ~1.5–2 m wide and primarily steeply dipping to the east but includes several antithetic dipping faults. Fault zone F2 in trench T2 is narrower (~0.5 m) and similarly dips steeply to the east. Faults were defined mainly by disrupted contacts but locally exhibit aligned clasts in coarse units or narrow, friable bands in sandier units.

We interpret one identifiable faulting event in trench T1 (Event 1A) as occurring between the deposition of units T1-150 and T1-200 (Fig. 8A). Layers T1-100 and T1-150 both appear undeformed across fault zone F1. Unit T1-150 coarsens westward from fine- to medium-grained sand in the center of the exposure to coarser material above the fault zone, suggesting it includes reworked footwall material and is a scarp-derived colluvial wedge. Unit T1-200 is internally deformed into blocks and thickens into the fault zone. Although most individual fault splays cannot be traced continuously up through T1-200, we interpret its thickness change to result from an earthquake when T1-200 was at the surface and vertical displacement created a depozone that preserved deformed blocks of T1-200 beneath post-earthquake colluvium. The primary east-dipping splay in fault zone F1 clearly truncates and deforms T1-200 and cannot be traced above the basal contact of T1-150. The basal contact of T1-200 is faulted in numerous places, with 5–20 cm vertical offsets along west-dipping, antithetic fault splays. These faults similarly cut all layers below T1-200 and are truncated against the coarse colluvium unit (T1-C1).

Two faulting events can be interpreted from Trench T2 (Fig. 8B). We interpret Event 1B as occurring between units T2-200 and T2-100, the latter of which appears undeformed atop and across the fault zone (F2). Unit T2-100 is loosely consolidated, thickest, and coarsest near the fault zone and pinches out toward the east, suggesting it may have been derived as colluvium from reworked fault scarp material. Unit T2-200 is truncated by closely spaced fault splays that terminate at or just below the sharp contact with T2-100, which juxtapose T2-200 against the coarse, poorly sorted, angular colluvium of the footwall (T2-C2). Unit T2-200 is more cohesive than T2-100 and may have formed a prolonged depositional surface before the faulting event. The contact between T2-100 and T2-C2 is more difficult to define west of the fault zone, due to the limited remobilization of T2-C2 deposits to form T2-100 and the resulting similarity in the appearance of the units. However, this portion of T2-100 is less cohesive and contains a higher proportion of matrix than T2-C2. Although the temporal relationship between Events 1A and 1B in T1 and T2 are not directly constrained, we tentatively interpret these to be the same event given the similar characteristics of the overlapping sediment (T1-100/T1-150 and T2-100).

We interpret a possible second event in trench T2, Event 2, as occurring between units T2-200 and T2-300 (Fig. 8B). Unit T2-200 exhibits similar characteristics to T1-100 as a potential scarp-derived sediment in that it thins and fines eastward away from the fault zone (F2) and has an east-dipping upper contact that is like the modern scarp profile; however, the thickness of the unit across the trench indicates a significant contribution of sediment transported along the axis of the trough (Fig. 7). We tentatively interpret that unit T2-300 was deposited before Event 2. Unit T2-300 is well-sorted fine to medium-grained sand and does not thin or fine away from the fault zone (F2). Unit T2-300 also exhibited subtle blocky soil structures near the top (e.g., at meter 14; Fig. 8B) when examined closely in the field that imply it was a relatively stable ground surface for a protracted duration before deposition of the coarser sediment of T2-200. Finally, the upper contact of T2-300 exhibits a slight kink or fold at meter 16.2 (particularly evident on the north trench wall at meter 15.7; Fig. 8B) that is not present at the top of T2-200, suggesting T2-300 was deformed before deposition of T2-200. We thus interpret that unit T2-300 was deposited in the trough formed by a prior event and then deformed by Events 1B and 2, whereas unit T2-200 was offset by only Event 1B.

Age control for the stratigraphy of the trenches is provided by eight luminescence ages (Table 1; Fig. 8). Two samples from T1-200 provide identical feldspar IRSL ages of 1.7 ± 0.1 ka. A third sample from T1-200 was dated using both feldspar IRSL and quartz OSL and provides similar but slightly younger (1.3 ± 0.1 ka) and older (1.9 ± 0.1 ka) ages, respectively. One of the 1.7 ka samples, T101, is from a coherent block with well-defined pre-deformation stratigraphy within unit T1-200. The consistency of this age with sample T102 from reworked sediments between blocks higher in this unit suggests redeposition occurred rapidly without exposure or that the deformation occurred very soon after the deposition of the sample T101 sediments. A luminescence age for layer T1-150 above is slightly older (2.4 ± 0.2 ka). We interpret this out-of-sequence age to result from reworked scarp-derived sediment that was partially or very poorly bleached, based on the distribution of single-grain D_E measurements for this luminescence sample, and the true age is not a large enough component of the distribution to be captured by the mixing model (e.g., Rhodes, 2011; Mahan et al., 2022; Part C of Supplemental Material [^{footnote 1}]; Table 1). Event 1A should postdate the ca. 1.7 ka depositional age of T1-200. Given that the age for T1-150 is questionable, there is no upper age bracket for Event 1A. Ages from trench T2 do follow stratigraphic order, with one age from T2-300 of 6.6 ± 0.4 ka that would predate Event 2. An age of 5.5 ± 0.3 ka from the base of T2-200 provides a minimum age for Event 2. An age of 2.8 ± 0.2 ka from the top of T2-200 provides a maximum age for Event 1B. The minimum age of Event 1B is the age of T2-100. Sample T-203 from layer T2-200 yielded a quartz age of 0.5 ± 0.1 ka and a feldspar age of 0.9 ± 0.1 ka. We prefer the feldspar age of 0.9 ± 0.1 ka based on the single peak in the distribution of single-grain D_E (Average Equivalent Dose; Table 1) measurements, as opposed to the broad, multi-peaked distribution of single-grain D_E measurements on quartz grains (Part C of Supplemental Material). If Event 1B is correlative to Event 1A, the younger ages for T1-200 (ca. 1.7 ka) narrow the timeframe of the most recent surface rupture. Given that the trenches cross the mapped paleorupture, luminescence ages imply a likely age for the most recent multifault surface rupture of the Calico-Hidalgo fault as 0.9–1.7 ± 0.1 ka.

We have identified 18 km of youthful-appearing surface rupture on the Calico-Hidalgo faults that likely occurred in the late Holocene (~0.9–1.7 ka) and involved ~2.3 m average dextral slip. This rupture occurred midway between the location of the 1992 Landers and 1999 Hector Mine surface ruptures and may have had a similar length (see below). Although this event preceded these historical ruptures by one to two thousand years, the timing of the Calico-Hidalgo paleorupture fits within the age range of the most recent earthquake cluster in the ECSZ (present to 1 ka; Rockwell et al., 2000) (Fig. 10). We have also identified a possible older event on the Hidalgo fault of unknown slip or rupture dimensions, which overlaps in age (ca. 2.8–5.5 ka) with the penultimate earthquake cluster (ca. 5–6 ka; Rockwell et al., 2000). These events on the Calico fault are thus consistent with previous observations of clustered activity in the ECSZ (Rockwell et al., 2000; Madden et al., 2006; Ganev et al., 2010; McAuliffe et al., 2013).

The actual length of the most recent paleorupture on the Calico-Hidalgo fault was likely larger than the 18 km extent of faulting we have been able to map using preserved geomorphic features. Based on the average measured slip of 2.3 m and using empirical scaling relationships for average earthquake slip and rupture length, the true rupture length for this event may have been ~82 ± 13 km, which would correspond to a M_w ~7.4 ± 0.1 earthquake (Wells and Coppersmith, 1994). The empirical relationships from Wesnousky (2008) predict even longer rupture lengths of ~100 km (log-linear relationship) and ~130 km (power-law relationship). These estimates are speculative, given that they are based on an observed mean displacement that is representative of only a fraction of the total possible original rupture length. If the peak observed displacement of 2.3 m is representative of maximum offset instead of the average, the resulting estimates of rupture length and size for the event are 53 ± 7 km and M_w ~7.1 ± 0.1 (Wells and Coppersmith, 1994). Scaling laws may also be of limited use for comparison to immature fault systems, such as the ECSZ, where proportionally more of the deformation may be accommodated as off-fault deformation that is not accounted for in our measurements of offset along the paleorupture (Shelef and Oskin, 2010; Milliner et al., 2015, 2021). The Hector Mine rupture may provide a useful comparison, however. This rupture exhibited a similar mean slip (2.5–3 m) yet has a mapped surface rupture length of 45 km (Treiman et al., 2002). Based on this comparison, it seems likely that the actual surface rupture on the Calico-Hidalgo fault was greater than the 18 km length of the mapped paleorupture zone. Most of the surficial evidence for the original paleorupture may have been erased by erosion.

If the Calico-Hidalgo surface rupture did extend significantly beyond the mapped extent of the paleorupture, it may correlate and connect to surface ruptures that have been identified via paleoseismology to either the north or south. Ganev et al. (2010) found evidence for four Holocene earthquakes (0.6–2.0, 5.0–5.6, 5.6–6.1/7.3, and 6.1/7.3–8.4 ka) on the northern Calico fault near the Newberry Mountains, which lies ~42 km north of the northern terminus of our mapped paleorupture (Fig. 1B). The youngest of these events overlaps in age with the age of Event 1 from our trench (0.9–1.7 ka), suggesting the mapped paleorupture may have been the same event identified on the northern Calico fault. If correct, this interpretation of a single large event implies the minimum surface rupture length of this event was closer to or even longer than ~60 km (i.e., 18 km paleorupture length plus at least 42 km faulting along the span between sites). This large length is consistent with scaling relationships and mean offsets observed along scarps along an 8 km stretch of the Calico fault to the north (Ganev et al., 2010). These offsets peak at 2.0 ± 1.0 m for the most recent earthquake (i.e., similar to the average slip of 2.3 m for our mapped paleorupture) and 4.7 ± 2.0 m for the penultimate event (which could represent two closely-spaced events). Combining offsets from the southern and northern paleoruptures (Ganev et al., 2010) yields a normal distribution that peaks at ~2.2 m (Fig. 9C). The geomorphic expression of the scarps along the northern Calico fault visible on satellite images in Google Earth satellite images and available lidar are also comparable to those along the southern Calico-Hidalgo paleorupture, in that both appear as continuous, narrow zones of surface disturbance and a light-colored, sharp lineament (Fig. S4 [^{footnote 1}]). We, therefore, conclude that it is possible that these were the same surface rupture event. We caution that this correlation is based mainly on average displacement and the general appearance of the surface rupture, however, given that the age control in both locations relies on imprecise luminescence ages with wide age uncertainties that alone permit separate or sequential earthquakes (Ganev et al., 2010).

The Calico-Hidalgo surface rupture possibly continued south of the mapped extent of the paleorupture (Fig. 1B). In contrast to the single trace of the Calico fault to the north, several faults extend southwards from our study site in the Calico fault system. The Hidalgo fault continues south with evidence for at least late Pleistocene activity but does not have a well-defined surface trace due to the sandy character of young alluvial surfaces south of Hidalgo Mountain (Fig. 2). If the paleorupture continued farther south on the Hidalgo fault, it would likely not have been preserved by recognizable features. The Mesquite Lake fault also extends southwards from the Calico fault, which itself bifurcates into several strands and zones of Quaternary folding south of Hidalgo Mountain (Dibblee et al., 1967a, 1967b). Previous paleoseismic studies have identified Holocene ruptures on the Mesquite Lake fault. Foster (1992) found evidence for at least one earthquake in the past 1.3 k.y., which is supported by more recent investigations by Menges et al. (2022) that found evidence for an event at ca. 1.3 ka on the Mesquite Lake fault and fits with the age range for the paleorupture on the Calico-Hidalgo fault. Madden et al. (2006) identified three Holocene events on the Mesquite Lake fault, the youngest of which is constrained as between 2.7 ka and 7.4 ka (more likely within the 3.9–4.6 ka range) but did not identify the younger event of Foster (1992) and Menges et al. (2022). The average slip during these events was ~1.6–2.4 m, which based on scaling relationships suggests a surface rupture length longer than the Mesquite Lake fault itself and thus suggests a multifault rupture (Madden et al., 2006). Connecting the young surface rupture on the full length of the Mesquite Lake fault to the paleorupture of the Calico-Hidalgo fault adds at least 40 km of surface rupture, implying a minimum surface rupture length of 58 km if the rupture terminated just north of Hidalgo Mountain.

There are thus several likely scenarios for multifault ruptures on the Calico fault system associated with the mapped paleorupture at 0.9–1.7 ka. The surface rupture may have continued to the north (60 km rupture length, Calico and Hidalgo faults only) or the south (58 km rupture length, Calico, Hidalgo, and Mesquite Lake faults), both scenarios of which are comparable in length to the 1999 Hector Mine and 1992 Landers surface ruptures (45 km and 85 km length, respectively). Alternatively, the paleorupture could have continued in both directions for a rupture length of ~100 km, which is comparable to the length predicted from empirical scaling relationships with the observed 2.3 m mean slip (but inconsistent with a comparison of slip-to-length ratio in the Landers and Hector Mine earthquakes). Although other scenarios are plausible, we infer that the 0.9–1.7 ka paleorupture was longer than the trace we mapped and involved at least two or three major faults. If this inference is correct, then the event was consistent with the length and multi-fault character of historic surface ruptures and the observed paleoseismic earthquake clustering in the ECSZ, and is thus another example of a complex, multifault rupture in this system.

Ruptures along the southern Calico fault system may affect or be influenced by ruptures on the sequence of faults responsible for the 1992 Landers and 1999 Hector Mine earthquakes, given their closeness (Fig. 1B). A rupture on the Calico fault may induce static stress changes that could bring Landers and Hector Mine faults closer to failure. Alternatively, a recent rupture could prevent the Calico fault system from experiencing a triggered rupture due to stresses induced by the Landers or Hector Mine earthquakes, by relieving tectonic strain and resetting the seismic cycle. To examine these possibilities, we conducted Coulomb Failure Function (ΔCFF) stress modeling of different possible scenarios of a Calico-Hidalgo rupture. This approach has been widely applied to testing how individual earthquakes may trigger future events or control coseismic slip distributions on neighboring faults (e.g., King et al., 1994; Harris and Simpson, 1996; Stein et al., 1997; King and Cocco, 2001; Cianetti et al., 2002; Doser and Robinson, 2002; McAuliffe et al., 2013).

We used Coulomb 3.4 to model Coulomb stress changes (Lin and Stein, 2004; Toda et al., 2005) for five different rupture scenarios on the Calico-Hidalgo-Mesquite Lake faults (A-E, Fig. S5). These scenarios vary from the pattern of the mapped paleorupture with slight variations in slip (18-km-long, Scenarios A and B), to a mid-range scenario in which the mapped paleorupture continues to the north along the Calico fault (63-km-long, Scenario C), to a maximum rupture length of 89 km or 112 km involving continuation of rupture to the north and to the south on either the Hidalgo or Calico-Mesquite Lake fault (Scenarios D and E) (Fig. S5). We consider Scenarios A and B to be unlikely, given that the surface rupture length is much smaller than predicted by scaling relationships from the observed slip. We likewise consider Scenarios C and E to be less likely than Scenario D (the preferred scenario), given that the strong geomorphic expression of the southern Hidalgo fault makes it probable that the surface rupture continued southwards and because the rupture length of Scenario D fits closely with the prediction from the scaling relationship from the mean displacement. We modeled each scenario with a range of coefficients of internal friction (0.2, 0.4, and 0.6) and modeled Scenarios D and E (the most likely scenarios) with variable fault depth (10 km and 18 km) and fault slip (2.3 m and 5 m), resulting in a total of 36 scenarios (Fig. S6). These variations of internal friction and fault depth cover the typical range of values used in previous Coulomb stress change models of the ECSZ (e.g., King et al., 1994; Parsons and Dreger, 2000; Freed and Lin, 2002; Toda et al., 2005). The effect of these Coulomb stress change scenarios on the Landers and Hector Mine faults was assessed by inputting their historic ruptures as receiver faults with geometries derived from finite fault rupture models (Cohee and Beroza, 1994; Ji et al., 2002).

Each rupture scenario we tested for the Calico fault system results in stress shadows, or clamping (i.e., negative ΔCFF), for most of the Landers and Hector Mine ruptures (e.g., Scenario D, the favored interpretation, Fig. 11). Large positive ΔCFF, which indicates unclamping and the promotion of fault failure, occurs primarily at the endpoints of the ruptures in each scenario, which do not overlap the ruptures or hypocenters of the later historic earthquakes. A small, local positive Coulomb stress change at the Landers hypocenter was produced only in specific variations of Scenario D, while Scenarios A and B produce minor positive ΔCFF at the Hector Mine hypocenter (Fig. S6). Local positive ΔCCF also occurs along the trace of the simulated Calico-fault system rupture itself, primarily at minor bends or stepovers in the rupture trace (Fig. 11). All scenarios also result in negative ΔCFF for the high-slip portions of each historic surface rupture, such as along the Emerson fault for Landers earthquake and the Lavic Lake fault for Hector Mine earthquake (Sieh et al., 1993; Treiman et al., 2002; Fig. S6 [^{footnote 1}]). These results indicate that static stress induced by a rupture of the Calico fault system is unlikely to bring either the Landers or Hector Mine faults closer to failure. Note that the long delay between the Calico paleorupture (0.9–1.7 ka) and the historic events similarly implies the latter were not induced, given that triggering via ΔCFF generally occurs on a timescale shorter than decades (Scholz, 2010). The apparent stress shadow induced by the Calico paleorupture may have instead suppressed the Landers and Hector Mine ruptures and resulted in a delay in the historic events (e.g., Harris and Simpson, 1998).

The alternative possibility, that the paleorupture prevented a triggered rupture on the Calico fault system following the Landers-Hector Mine sequence, seems more plausible. Previous studies have shown that the Landers rupture may have triggered the Hector Mine rupture and similarly could have brought portions of the Calico fault closer to failure (Harris and Simpson, 2002). Original calculations from King et al. (1994) showed that positive ΔCFF occurred in the epicentral and high slip area of the Hector Mine rupture as well as along patches of the Calico fault, including its northern segment. Some of these fault segments also experienced induced aftershock clusters following the 1992 earthquake (Hauksson et al., 1993). Parsons and Dreger (2000) similarly showed that Landers brought the Hector Mine faults and patches of the Calico fault locally closer to failure via static stress change, but only for certain scenarios of slip distribution on the Landers rupture and assumed coefficient of friction on receiver faults. These scenarios are likely to be correct, however, given that seismicity rates increased in the predicted areas of positive ΔCFF (Wyss and Wiemer, 2000). Several additional factors may have also added to the original triggering effect of Landers and help explain why there was such a long delay (7 years) before the Hector Mine earthquake, including continued tectonic loading, dissipation of pore fluid pressure gradients, rate- and state-dependent friction effects, post-seismic viscoelastic relaxation, and dynamic alteration of fault zone rigidity (Deng et al., 1998; Parsons and Dreger, 2000; Pollitz et al., 2000; Freed and Lin, 2001; Fialko et al., 2002; Freed and Lin, 2002; Kilb et al., 2002; Pollitz and Sacks, 2002). Given that viscoelastic relaxation is greatest directly adjacent to a rupture, this effect may have been greater on the Calico fault (Pollitz and Sacks, 2002). In turn, viscoelastic loading due to Hector Mine would have further stressed the Calico fault (Freed and Lin, 2002). It thus seems likely that the Landers and Hector Mine ruptures would have at least locally increased the likelihood of rupture of the Calico fault system. However, the most recent event (0.9–1.7 ka) seems to have prevented the Calico fault system from being critically loaded (i.e., near failure) with elastic strain (e.g., Pollitz and Sacks, 2002; Scholz, 2010; McAuliffe et al., 2013).

In summary, this analysis of ΔCFF suggests that the paleorupture of the Calico fault system did not bring the Landers and Hector Mine faults closer to failure, but rather may have delayed their historic rupture and may have precluded the need for a triggered rupture on the Calico fault following the Landers and Hector Mine earthquakes. This analysis is limited, however, in that it does not consider rheological variations, the influence of other prehistoric ruptures on the state of stress, such as along the Pinto Mountain fault, or the effect of viscoelastic relaxation from the Calico paleorupture on the Landers and Hector Mine faults. The analysis is also based on only partial data in that we do not know the full rupture length or slip distribution of the most recent event on the Calico fault system. A more thorough analysis may be warranted as new paleoseismic observations become available.

Our results suggest that the Calico fault system had already experienced a major, multifault rupture in the modern to late Holocene earthquake cluster before the historic Landers and Hector Mine earthquakes (Fig. 10). Identification of a paleorupture surface trace along the southern Calico and Hidalgo faults implies that the most recent event identified on the Calico fault to the north by Ganev et al. (2010) was likely a large surface rupture that involved numerous faults and may have spanned the entire length of the fault system between the subsequent Landers and Hector Mine surface ruptures. This finding contradicts several predictions that the Calico fault system might instead fail in more frequent but smaller rupture patches. For example, the patchiness of positive ΔCFF induced along the Calico fault system by the Landers and Hector Mine ruptures might have indicated that ruptures along the Calico fault would be more localized, e.g., northern segment (King et al., 1994; Parsons and Dreger, 2000; Wyss and Wiemer, 2000; Freed and Lin, 2002; Harris and Simpson, 2002), if Calico fault ruptures were to follow or be induced by Landers or Hector Mine ruptures. Langenheim and Jachens (2002) similarly suggested that the Calico fault may experience smaller, more localized ruptures due to the occurrence of a strong lithologic body through much of the seismogenic crust in the region between the Emerson and Calico faults. This feature, known as the Emerson Lake Body, has been detected using gravity and aeromagnetic data and is thought to be a strong, mafic or intermediate pluton that could impede faulting (Langenheim and Jachens, 2002). Our data instead suggest that the Calico fault system can behave much like other faults in the ECSZ by failing in large, multifault events.

Although the identification of a large, multifault surface rupture on the Calico fault system provides a more detailed history of ECSZ activity in the Holocene, the physics of multifault ruptures and earthquake clustering in the region remain poorly understood. One mechanical model for clustered earthquakes in the ECSZ is a simple oscillator model, in which positive ΔCFF throughout numerous earthquake cycles synchronizes and locks faults in phase and thus produces earthquake clustering (Sammis et al., 2003; Scholz, 2010; McAuliffe et al., 2013). Scholz (2010) suggested that the Calico fault should operate independently of clustered events on the combined Landers and Hector Mine faults, given that its slip rate and recurrence interval are higher based on existing data (Oskin et al., 2007, 2008; Ganev et al., 2010; Xie et al., 2018). Our inference that the Calico fault system experienced a large surface rupture ~1–2 ka before the Landers-Hector Mine cluster, and that this rupture would have placed these faults in a ΔCFF stress shadow, is consistent with the idea that the Calico fault system is at least somewhat independent, i.e., neither triggers nor is triggered by Landers-Hector Mine style ruptures. While the Calico fault system may operate independently, its ruptures may also “repel” clustered rupture events on neighboring faults, i.e., force them to occur later, or be out-of-phase. Even if the Calico fault system does operate in the same general clustering of earthquakes in the ECSZ (i.e., as implied by Fig. 10) despite its different slip rate, its ruptures may still repel other events, thereby “stretching” out the duration of a given earthquake cluster.

An additional uncertainty for the clustered earthquakes in the ECSZ is whether the pattern of fault ruptures observed historically is repeated through time. Rockwell et al. (2000) concluded that different rupture sequences were likely to occur throughout the clustered earthquakes, based on paleoseismic records for linked fault segments that do not replicate through the Holocene clusters. They suggested that patterns of rupture may depend on where events nucleate and in which direction ruptures propagate, i.e., N to S, or S to N. This is supported by subsequent paleoseismic data, which indicate that each fault does not rupture the surface in every earthquake cluster (Rymer et al., 2002; Madden et al., 2006; Ganev et al., 2010; this study; Fig. 10). The lack of systematic, repeated multifault rupture patterns may result from the geometric complexity of the southern ECSZ (Spotila and Garvue, 2021). Regional dextral shear is accommodated in part via primary dextral faults, e.g., the Johnson Valley fault; these dextral faults are discontinuous, transpressive, and bend westward toward the north, as well as intervening, north-south, transtensional connector faults, e.g., the Kickapoo fault. Ruptures initiating in the south that propagate northwards may take advantage of the connector faults, whereas those propagating southwards may follow primary faults or connector faults depending on the dynamic stresses and whether faults are sufficiently loaded for failure. Local stepovers and transpressional bends may further constrain a given rupture pattern. The result may be complex scenarios of multifault failure that do not repeat. The resulting rupture patterns would thus only make sense if the state of stress and dynamic stress evolution during ruptures were fully known for each fault.

Our results also have implications for studies of earthquake geology in the region. Our study suggests that continuous portions of large (M_w >7.0) strike-slip surface ruptures may be preserved and mappable after at least several thousand years under certain geomorphic conditions. Although local features—such as individual offsets, deflected drainages, and scarps—have been widely documented on similar faults in the region (Rockwell et al., 2000; Madden et al., 2006; Oskin et al., 2007; Spelz et al., 2008; Ganev et al., 2010; Madden Madugo et al., 2012; McAuliffe et al., 2013), continuous mapping of >10-km-long segments of prehistoric surface ruptures based on tectonic geomorphology have only rarely been identified (e.g., Thompson Jobe et al., 2020; Chupik et al., 2022). The historic surface ruptures of strike-slip faults are locally preserved and still mappable even after more than a century (e.g., Mueller and Rockwell, 1995). How long such ruptures can be preserved, however, is not generally known and is likely variable by location and lithology of neighboring bedrock. In the case of the Calico-Hidalgo paleorupture, preservation seems to result from limited erosion and isolation from deposition that occur on entrenched upper piedmont alluvial surfaces. These surfaces develop good pavements with cohesive soils that are ideal for fault preservation. In contrast, rupture evidence is likely ephemeral in the young, non-cohesive alluvium of broad valleys. This limited preservation implies that the localization of strike-slip faults along mountainous ridges, such as common along transpressive systems, may be an important pre-condition for preserving prehistoric strike-slip surface ruptures. Other factors are also likely critical, including desert aridity, surface age, and a lack of aeolian sand input. Fault slip rate and recurrence interval may also play a role, as low fault activity may result in paleorupture deterioration, whereas high fault activity may result in overprinting and overlapping of evidence.

Based on these potentially important criteria, the transpressive, moderately active ECSZ (1–4 mm/yr slip-rate, 1–10 ka recurrence intervals; Rubin and Sieh, 1997; Rockwell et al., 2000; Madden et al., 2006; Oskin et al., 2008; Xie et al., 2018) may be well suited for preserving prehistoric surface ruptures. From reconnaissance using Google Earth high-resolution satellite imagery and locally available 1-m-resolution lidar (EarthScope, 2009; U.S. Geological Survey, 2017), we have identified several other potential paleoruptures besides the Calico-Hidalgo fault system. The southern ~45 km of the Lenwood fault is well expressed with an ~16 km stretch (between 34.66085°N, 116.86253°W and 34.54224°N, 116.76531°W) exhibiting relatively continuous, sharp scarps and white lineaments on uplifted alluvial and colluvial surfaces along ridge fronts (Fig. 12). Paleoseismology at two sites on the southern Lenwood fault has documented a most recent earthquake of 1.2–2.5 ka (Padgett, 1994; Khatib, 2004), suggesting this preserved paleorupture is similar in age to the Calico-Hidalgo paleorupture. Another example is the little-known Humbug Mountain fault, which connects to the sinistral Pinto Mountain fault zone and the dextral Pisgah-Bullion fault system (Howard et al., 2013) (Fig. 12). The ~8-km-long Humbug Mountain fault exhibits ~5 km of clear and continuous lineaments and breaks in slope suggestive of a prehistoric surface rupture (Fig. 12), although the earthquake history of this fault is unknown. We suggest that paleorupture mapping coupled with paleoseismology may provide additional insights into the rupture dynamics and seismic hazards of the ECSZ.

Finally, these results have implications for future rupture patterns in the ECSZ. The identification of yet another complex multifault rupture in the ECSZ further suggests this is the normal mode of operation for ECSZ faults and not just a recent phenomenon. This means that the region should expect future events of a similar size to the 1992 Landers event. Results also suggest that the Calico fault system has experienced a major rupture in the past 0.9–1.7 k.y., meaning that it may not be fully loaded with tectonic strain for failure in the near future. Ganev et al. (2010) found that the northern Calico fault at Newberry Springs ruptured twice in a short period during the penultimate earthquake cluster (5–5.6 ka), however, such that we cannot rule out another Calico surface rupture in the modern cluster. Based on the regional pattern of surface ruptures, and on paleoseismic data and historical observations, the only major faults in the southern ECSZ that are not known to have ruptured in the modern earthquake cluster are the northern Johnson Valley, southern Emerson, southern Bullion, Ludlow, and Sheephole/Cleghorn (Figs. 1B and 10). Future investigations may focus on these structures as well as seek to lengthen the paleoseismic observation windows on other faults. An additional challenge will be to refine models of what controls multifault ruptures in this system (e.g., earthquake “gates”) so that likely scenarios of specific rupture patterns may be identified.

We interpret fresh fault features mapped along the combined Calico-Hidalgo faults as having formed during the most recent rupture of these faults. Based on our paleoseismic observations and luminescence ages, this rupture occurred between 0.9 ka and 1.7 ka, thereby placing the event within the most recent earthquake cluster of the southern ECSZ. Based on the ~2.3 m average right-lateral displacement along the paleorupture and the transition to geomorphic conditions that are poorly suited for preserving faulting to the north and south, we hypothesize that the original rupture length was significantly larger than its 18-km mapped length. Scaling relationships using this mean displacement suggest the paleorupture may have been ~82 km long and thereby could have involved the entire Hidalgo fault and continued at least 42 km northward to connect with the 0.6–2.0 ka most-recent surface rupture on the Calico fault identified in a previous paleoseismic study (Ganev et al., 2010). This result implies that the paleorupture may have been associated with an M_w ~7.4 earthquake and was characterized by a multifault rupture pattern of similar size to the 1992 Landers and/or 1999 Hector Mine ruptures but occurred directly in the zone between them. The exact limits of the paleorupture are not well constrained, however, and it may have even continued farther south to connect to a recent event (ca. 1.3 ka) on the Mesquite Lake fault.

These results add additional evidence that complex, multifault ruptures are the normal mode of operation for the ECSZ. The occurrence of an event on the southern Calico fault system 1–2 ka before the historic Landers and Hector Mine earthquakes, however, may illustrate how dynamic interactions between ruptures may result in continually evolving, complex rupture patterns and may stretch the duration of earthquake clusters. Based on our ΔCFF analysis, the paleorupture of the Calico-Hidalgo faults may have resulted in a stress shadow along the future Landers and Hector Mine ruptures, thereby delaying their occurrences. The paleorupture may have also prevented the Landers and Hector Mine earthquakes from triggering an event on the Calico fault, which was at least locally loaded with positive Coulomb static stress change by the historic events. These speculations imply that, while a broader factor may result in general periods of activity in the ECSZ (e.g., Dolan et al., 2007), individual fault ruptures may repel, i.e., maintain out-of-phase behavior (e.g., Scholz, 2010) with each other via static and dynamic stress changes, thereby lengthening the duration of earthquake clusters. The resulting noisiness of rupture patterns through time, as well as the reason for multifault ruptures in the first place, may ultimately stem from the geometric complexity of the ECSZ, given the ubiquitous bends, stepovers, and secondary deformation along it.

A final implication of this work is that previous ruptures of faults in similar locations that may be identified in the paleoseismic record may be traceable as surface rupture zones using high-resolution satellite imagery, DEMs, and field observations. Complex shear zones with comparable geomorphic conditions, neotectonics, and rates of activity as faults in the ECSZ may be ideally suited to preserve long, continuous mappable portions of paleoruptures, even for thousands of years. Reconnaissance suggests that other such young ruptures may yet be identified in the ECSZ.

Funding was provided by National Science Foundation Tectonics (EAR-1802026) and USGS/EHP (PSQUR6UR). The authors thank the U.S. Marine Corps and personnel at the Marine Corps Air Ground Combat Center (MCAGCC) at Twentynine Palms for field access and assistance during fieldwork, particularly Dr. Brian Hennen, Walter Christensen, Elizabeth Barron, Robyn Coole, and Mary Lane Poe. Elizabeth Curtiss is thanked for valuable assistance in the field and for helpful scientific discussions. We also thank field assistants Kristin Chilton, Wynnie Avent, Samantha VanDenburgh, Austin Wright, Amanda Wyche, Patricia Forshee, Sgt. Cole Lemon, and Jesus Montes. Mike Oskin, Kate Scharer, and Associate Editor Jeff Lee are thanked for their careful and thoughtful reviews that greatly improved this manuscript. EarthDEM was created by the Polar Geospatial Center from Maxar Imagery.