What Controls Early Restraining Bend Growth? Structural, Morphometric, and Numerical Modeling Analyses From the Eastern California Shear Zone

Restraining bends influence topography, strike‐slip evolution, and earthquake rupture dynamics, however the specific factors governing their geometry and development in the crust are not well established. These relationships are challenging to investigate in field examples due to cannibalization and erosion of earlier structures with cumulative strain. To address this knowledge gap, we investigated the structure, morphology, and kinematics of 22 basement‐cored restraining bends on low net‐slip faults (<10 km) within the southern Eastern California shear zone (SECSZ) via mapping, topographic analyses, and 3D numerical modeling. The bends are self‐similar in form with most exhibiting focused relief between high‐angle bounding faults with an arrowhead shape in map view and a “whaleback” longitudinal profile. Slight changes in that form occur with increasing size indicating predictable growth that appears to be primarily controlled by local fault geometries (i.e., bifurcation angle), rather than the influence of fault obliquity relative to far‐field plate motion, due to inefficient slip‐transfer across interconnected irregularly trending closely spaced faults. Modeling results indicate that the self‐similar fault‐bound geometry of SECSZ restraining bends may arise from elevated shear strain at the outer corners of single transpressional fault bends with increasing cumulative slip. This, in turn, promotes growth of a new fault leading to efficient accommodation of local convergent strain via uplift between bounding faults. Finally, our results indicate that the kilometer‐scale restraining bends contribute minimally to regional contraction as they only penetrate the upper third of the seismogenic crust and are therefore also unlikely to impede large earthquake surface ruptures.


Introduction
Restraining bends along strike-slip faults often manifest as positive flower structures with discrete zones of uplift that exhibit wide-ranging variations in size, form, and geometry (Cunningham, 2007;Harding, 1985;Woodcock & Fischer, 1986).The diverse structural and geomorphic characteristics of the contractional fault bends hold significant implications on earthquake rupture dynamics (e.g., Barka & Kadinsky-Cade, 1988;Elliott et al., 2015), deciphering tectonic histories (e.g., Legg et al., 2007;Wakabayashi et al., 2004), and refining models of strike-slip fault evolution (e.g., Cooke et al., 2013;Hatem et al., 2017).Despite their importance, the specific factors governing restraining bend uplift patterns, form, and incremental development are poorly defined in crustal systems.Fault kinematics (e.g., net-slip and slip rate) play a significant role in controlling restraining bends size, as evidenced by large bends on major, fast-slipping faults with high cumulative displacement such as the San Bernardino Mountains (San Andreas fault), Alaska Range (Denali fault), and the Akato Tagh bend (Altyn Tagh fault) (e.g., Benowitz, Roeske, et al., 2022;Cowgill et al., 2004;Spotila et al., 2001).These large restraining bends have been extensively studied due to their prominence and seismic hazard to help define their structure, kinematics, and long-term uplift rates (e.g., Baden et al., 2022;Benowitz, Cooke, et al., 2022;Cochran et al., 2017;Fuis et al., 2001;Gomez et al., 2007;Matrau et al., 2019).Unfortunately, it remains a challenge to ascertain how quickly the bends grew and the degree to which their form, structure, and kinematics have evolved due to continuous overprinting from long-term deformation and erosional processes (10 5 -10 6 years).Therefore, the investigation of small early-stage transpressional features (e.g., Ocotillo badlands, CA -Brown & Sibson, 1989;Echo Hills, NV-Campagna & Aydin, 1991) is important for understanding restraining bend evolution because they potentially retain more developmental information, despite being underrepresented in the existing literature.
Experimental and field investigations of restraining bends demonstrate that fault bend angle and trend (relative to plate motion) play a critical role in defining their shape and form, notwithstanding other important controls including preexisting weaknesses and the distribution of shearing along through-going lithospheric structures.
While it has been demonstrated that restraining bends along mature strike-slip faults (characterized by high netslip) typically exhibit increased uplift with greater transpressive obliquity (e.g., Spotila et al., 2007), this relationship is unestablished for low net-slip restraining bends.It is understood, however, that sharp bend angles typically result in rounder and dome-shaped forms while gentle bend angles may lead to more elongate forms (Dooley & Schreurs, 2012;Mann, 2007;Spotila et al., 2007).Additionally, the average aspect ratio (length to width) appears to be similar to releasing bends (∼3:1) (Aydin & Nur, 1982;Cunningham, 2007;Hatem et al., 2015;Mitra & Paul, 2011).Restraining bends can also vary in degrees of transverse symmetry (i.e., symmetry of fault perpendicular elevation profile across restraining bend) (e.g., symmetric = Bayan Tsagaan Uul -Bogd fault system, Gobi Altai; asymmetric = San Bernardino Mountains-San Andreas fault, California).However, the factors controlling variations are often not explicitly investigated or are interpreted speculatively based on rheology or kinematic theory (i.e., pure shear vs. simple shear) (Buscher & Spotila, 2007;Cunningham, 2007;Tikoff & Teyssier, 1994).Lastly, restraining bends can form due to varying geometric characteristics in strike-slip faults, such as (a) true simple fault bends or stepovers (what most physical models simulate), (b) local deviation in fault trend (i.e., paired bends or sidewall ripouts), (c) abrupt change in fault trend, (d) a convergence of two primary faults, or (e) compressive fault termination bend (Barka & Kadinsky-Cade, 1988;Mann, 2007;Swanson, 1989;Wilcox et al., 1973).While it is recognized that geometric differences may partially control the uplift patterns, internal slip-rates (e.g., Elston et al., 2022), and restraining bend forms, there has been limited systematic investigation of their relationships and kinematics.Furthermore, the relationships between form and fault geometry are established primarily from physical experiments rather than from crustal examples.Currently, there is scant quantitative empirical data compiled to compare the structures and forms of restraining bends on faults with low net-slip (<10 km) and varying geometric characteristics.
To address the knowledge gap of restraining bends, in this paper we investigate an array of kilometer-scale restraining bends in the southern Eastern California shear zone (SECSZ) of the Mojave block (Figure 1).The SECSZ bends are associated with seven low net-slip dextral faults, which enable examination of early-stage transpressional strain, restraining bend form, and the relationships between fault kinematics and characteristics (i.e., geometry, trend, net-slip).We analyzed the morphology of 20 restraining bends, completed remote and highresolution field-based neotectonic fault mapping of 12 restraining bends, correlated erosion surfaces to reconstruct vertical deformation, and compared fault characteristics to morphometric indices (e.g., slope, swath profiles, aspect ratio).Additionally, we employed 3D numerical deformational modeling at both local and regional scales.This allowed us to test the influence of restraining bend fault geometries on uplift patterns and kinematics as well as to investigate the large scale kinematics and role of regional transpression in the evolution of the SECSZ.Our findings reveal that the restraining bends are at a range of stages of development and have a variety of fault configurations, although most are doubly bound and formed as upper-crustal, positive flower structures.The restraining bends are self-similar in form and are sensitive to local fault geometries that generally focus uplift and contractional strain between bounding faults.Results from 3D mechanical numerical modeling aligns with the majority of topographic observations and provides specific plausible mechanisms for accomplishing observed patterns.Lastly, our results suggest that some of the transpressional bends we studied may have developed early in the evolution of the SECSZ faults and may not have simplified with increasing net-slip, but rather have grown in size and geometric complexity.

Tectonic Setting
The Eastern California shear zone (ECSZ) is a ∼125 km wide, complex network of variably trending right-and left-lateral strike-slip faults that accommodates ∼25% of the total dextral transform motion between the Pacific and North American plates (Figure 1a) (Dokka & Travis, 1990b;Guns et al., 2021;McGill et al., 2015;Miller et al., 2001;Sauber et al., 1994).The ECSZ extends from the Salton Sea in the south, through the Mojave, and northwards to the northeastern Sierra Nevada and Basin and Range.The early fault development is attributed to the opening of the Gulf of California and the progressive strain imposed by the growth of the "big bend" of the San Andreas fault (Darin et al., 2016;Li & Liu, 2006;Liu et al., 2010;Plattner et al., 2010).Shearing initiated between 9 and 12 Ma in the north Mojave block and Walker Lane in response to a gradual change in plate motion.Younger and accelerated displacement in the south Mojave block and Eastern Transverse Ranges followed a major change in plate motion at ∼4-6 Ma (i.e., the formation of the big bend) (Atwater & Stock, 1998;Cande et al., 1995;Gan et al., 2003;Nuriel et al., 2019;Powell et al., 1993).Since its inception, the ECSZ is estimated to have accrued 50-75 km of net-slip, although individual faults of the shear zone typically exhibit several kilometers of total dextral displacement (Andrew & Walker, 2017;Dokka & Travis, 1990a, 1990b;Glazner et al., 2002;Jachens et al., 2002;Lease et al., 2009;Singleton & Gans, 2008).
The southern ECSZ (SECSZ) is a region of structural and seismogenic complexity within the south Mojave block (i.e., between the Pinto Mountain fault and Manix Basin) that hosts >10 closely spaced, N to NW-trending, irregular, discontinuous, low net-slip dextral faults (Figure 1b) (Spotila & Garvue, 2021).From north to south, the number of active faults increases due to the southward bifurcation of some principal faults (e.g., Lenwood and Old Woman Springs faults, Emerson and Homestead Valley faults, Calico and Hidalgo faults) and the addition of other faults (e.g., Johnson Valley, Lavic Lake, and Mesquite Lake faults).Numerous zones of fault intersection facilitate structural interaction and translate to complex earthquake rupture, as highlighted by two historic large earthquakes (1992 M w 7.3 Landers and 1999 M w 7.1 Hector Mine) and one prehistoric rupture (Vadman et al., 2023) that occurred in the SECSZ, each of which involved rough, sinuous surface ruptures across multiple (∼3-5) faults and numerous stepovers (Figure 1b) (Sieh et al., 1993;Treiman et al., 2002).The zones of fault intersection tend to occur in areas of high relief and increased fault segmentation indicating inefficient transfer of strike-slip motion between faults and local contraction, which influences earthquake rupture dynamics and may lead to growing geometric complexity over iterative earthquake cycles.
The modern SECSZ is primarily transpressional and has abundant evidence of pervasive shortening accommodated via basin inversion and large-scale folding of Neogene sediments, the presence of active reverse faults, and a higher number of restraining bends than releasing bends (Bartley et al., 1990;Selander, 2015;Spotila & Garvue, 2021).Some restraining bends may have long histories and could have formed from the interaction of early echelon fault segments (e.g., Hatem et al., 2017;Schellart & Nieuwland, 2003;Tchalenko, 1970), while  2023), with prescribed geodetically derived plate velocities to the base of the model oriented to the direction of plate motion of ∼322°.From this model, we derive the loading of the SECSZ to infer regional strain and to apply to local models.
others may have been imposed later due to the onset of regional transpressional deformation (e.g., Gomez et al., 2007;Moulin & Cowgill, 2023).The greater ECSZ is generally thought to have initiated as a transtensional system (e.g., Dokka & Travis, 1990a) and the cause and timing of regional transpression of the Mojave block is not clearly established.Thermochronometric data of the neighboring San Bernardino mountains indicates rapid growth over the last 2 Ma (Spotila & Sieh, 2000;Spotila et al., 2001) with a likely onset of that growth between 2.5 and 3.8 Ma based on the stratigraphy of Pliocene age sediments near Lucerne Valley (Cox et al., 2003;May & Repenning, 1982).If the onset of contraction in the San Bernardino mountains is indicative of the start of a greater pervasive regional transpressional regime including the nascent faults of the SECSZ, it may suggest that restraining bends formed early in the faults' history, though this remains untested.While previous work demonstrates that the Miocene-Pliocene volcanics and accompanying graben sediments are related to the prior extension and transtension of the ECSZ (Carter et al., 1987;Dorsey et al., 2021;Glazner et al., 1989), the timing of cessation of volcanic related units has not been convincingly shown to constrain the onset of regional transpression in the ECSZ.This may be due to their diachronism and limited heterogenous spatial distribution which are in some places adjacent to the younger Quaternary volcanics of ambiguous genesis (i.e., Pipkin Volcano, Amboy Cone, Lavic Lake Volcanic Field).Future work focusing on the timing, mechanism, and geochemistry of volcanic eruptions and basin sediments within the ECSZ may further increase our understanding of the potential influence and timeline of the regional tectonics.
Several kinematic models have been proposed for the greater ECSZ which may explain the preponderance of transpressional strain and restraining bends within the SECSZ.First, transpression may be due to squeezing of the Mojave block via tectonic escape from the combined motion of the Mojave and "big bend" section of the San Andreas fault and the Garlock fault (Dixon & Xie, 2018).This model is weakly supported by a rough decrease in net-slip (and presumably inception age) of SECSZ faults from east to west, where some western faults appear to have penetrated the San Bernardino mountains, as well as an observed eastern decrease in fault activity (Cochran et al., 2020;Spotila & Anderson, 2004;Spotila & Garvue, 2021).A second proposed cause of transpression is progressive counterclockwise rotation of right-lateral faults via bookshelf faulting or due to local rotation of the direction of maximum compression, which would currently favor shear on N-trending faults and transpression on older NW-trending faults (Nur et al., 1993).While significant clockwise fault block rotations are supported for the left-lateral domains of the ECSZ, including the Eastern Transverse Ranges and the northeast quadrant of the Mojave block (e.g., Carter et al., 1987;Schermer et al., 1996), significant counterclockwise block rotation following the formation of the right-lateral faults of the Mojave block (<10 Ma) is neither supported by paleomagnetic studies (e.g., Golombek & Brown, 1988;Valentine et al., 1993;Wells & Hillhouse, 1989) nor geomorphic expression (Spotila & Garvue, 2021).A third and final explanation for initiation of transpression in the SECSZ is that warping of the closely spaced primary faults occurs due to growth of secondary structures, differential loading, penetrative strain, and repeated multi-fault surface ruptures (Spotila & Garvue, 2021).This explanation is supported by (a) a power-law relationship between fault strike variability, a metric of fault roughness, and fault maturity, where the SECSZ faults have relatively sinuous and variable trends when compared to faults with greater net-slip and slip-rate and (b) a greater number and degree of transpressional than transtensional features (i.e., numerous restraining bends and basin inversion between faults) (Spotila & Garvue, 2021).This suggests that the SECSZ faults are immature and have yet to smoothen via strain-weakening where faults eventually integrate into a narrow through-going zone of shear.The dominance of transpressional strain challenges this expectation, as an ideal strain-weakening strike-slip system would likely exhibit an equal combination of transpression and transtension.Alternatively, the transpressive regime of the SECSZ may be maintaining the initially wide and complex fault system (Hobbs et al., 1990;Morrow et al., 1982;Tikoff & Fossen, 1993).
While widespread Plio-Quaternary transpression has been identified in the SECSZ, few studies have specifically investigated the role of restraining bends in accommodating strain and the timing of their formation (Bartley et al., 1990;Selander, 2015).In this study, we further inform upon the relative timing and kinematics of transpression in the SECSZ and tectonic evolution of the plate boundary, while also providing specific insights into the controls on restraining bend formation and development.The restraining bends in the SECSZ are distinct, selfsimilar zones of warped primary fault geometry, secondary faulting, and entirely basement-cored (granitic, dioritic, and metamorphic bedrock) uplifts of low to moderate relief.We have identified 22 restraining bends with clear expressions, although there are less obvious examples with lower relief (or masked inherited relief) and more obscure secondary deformation (Spotila & Garvue, 2021; their Figure 13).By mapping the form, modeling the growth, and describing the kinematic evolution of SECSZ restraining bends, we seek to relate deformation character to variable fault boundary conditions, answer whether bends simplify or complexify with increasing deformation and fault interaction, and test what role transpression plays in the early stage of strike-slip fault evolution.
Restraining bend angles are typically measured by comparing the trend of the primary fault in front of the bend and on the bend.Bend angle has been found to be important in controlling convergent kinematics, fault geometry, and the form that they may take (i.e., sharp rhomboidal bend vs. gentle sigmoidal bend) (Mann, 2007;Mitra & Paul, 2011).We found it difficult to ascertain restraining bend angles, however, because the SECSZ faults are sinuous, segmented, and interconnected where relief is not solely confined within convergent steps.Notably, the majority (20 of 22) of SECSZ restraining bends are doubly fault bound and relief is focused where faults bifurcate and connect.To investigate the relationship between doubly bound restraining bend morphology and fault connectivity, we use Google Earth Pro to measure azimuthal trends and bifurcation angles.We apply a uniform scale, in which primary and secondary faults were extended a fixed radial distance (0.5, 1, or 2 km) from the point of fault bifurcation depending on the size of the bend (Table S5 in Supporting Information S1).

Erosional Surfaces and Older Deposits
We estimated patterns of vertical deformation in granitic basement-cored restraining bends by mapping the lowrelief erosion surfaces that cap them.Low-relief erosion surfaces have been mapped and used as deformation markers throughout the Mojave block and neighboring Transverse, Eastern Transverse, and Peninsular Ranges (e.g., Spotila & Sieh, 2000;Spotila et al., 2020).Numerous observations suggest that these low-relief surfaces formed prior to the Quaternary and are slowly eroding markers of younger vertical displacement, including overlapping Paleogene through Miocene sediment and basalt flows, minimal late Cenozoic exhumation based on thermochronometry, and evidence for formation in humid climate based on thick saprolite and regolith accumulations atop granitic bedrock (Axen et al., 2000;Mason et al., 2017;Oberlander, 1972;Rossi et al., 2017;Seiler et al., 2011;Spotila, 1999;Spotila et al., 1998).Although low-relief basement surfaces may also form as pediments in arid climates due to extended periods of slow denudation and tectonic quiescence (e.g., Nichols et al., 2002;Pelletier, 2010), the termination and direct offset of erosional surfaces along faults in the SECSZ restraining bends argues that they formed prior to transpressional deformation and have experienced minimal modification since (see Spotila et al., 2020 for additional arguments).A similar approach has been used in other restraining bend uplifts worldwide (e.g., Cunningham, 2007).

10.1029/2023TC008148
We mapped relict erosion surfaces atop 10 of the basement uplifts of the SECSZ following the methods described in Spotila et al. (2020).Surfaces were mapped using high resolution satellite imagery and the pan-and-tilt function in Google Earth Pro to enable viewing in 3D.Individual mapped surfaces are generally low relief (i.e., <50 m), connect smoothly across the broad plateaus that cap uplifted blocks, exhibit low surface gradients (i.e., <5°), have clear boundaries defined by a break in slope, are relatively planar (though possibly tilted) rather than concave or convex, and are >0.5 km 2 in area (to screen out isolated round knobs or rounded ridge tops that form along spur ridges that drop below the plateau surface).We also ground-truthed the low-relief surfaces in the field using the dominant presence of rounded corestones and tors which are regionally indicative of slow weathering and erosion (Oberlander, 1972) (Figures 4g-4i).The resulting elevation maps (extracted as 2-m-resolution rasters in Figure 5) of the surfaces should denote where minimal denudation has occurred across the crystalline basement-cored restraining bends since formation and constrain the vertical offset along bounding faults where clearly displaced from mapped surrounding pediment surfaces.Assuming the pediment surfaces surrounding the basement uplifts were low-relief and relatively planar prior to deformation, the elevation distribution of the uplifted relict surfaces should broadly approximate the pattern of structural relief.We caution that these reconstructions are approximate and, in some places, may be minimum or maximum estimates.The downthrown side is sometimes buried by Quaternary sediments or thick packages of grus while the upthrown side may be stripped resulting in minimum values.On the other hand, if the downthrown surface has eroded faster than the upthrown surface due to fluvial  -6, 9-10, 11, 13, 15, 17-18, and 20-22 (a-g, respectively).All doubly bound restraining bends have a roughly arrowhead plan view shape where structural relief is concentrated between the bounding faults.Cross sections are true scale and interpretations are based on quality and presence of field measurements (see Section 3.4).The blue colored lines on cross section profiles mark the extent of mapped erosion surfaces.In cross section, the yellow regions are inferred alluvial valley fill.White arrows on maps show dip direction and location of field measurements with the associated dip magnitude.Queried faults in cross-section are lower confidence interpretations due to a lack of coverage or lower quality field measurements (Table 2 and Table S3 in Supporting Information S1).Cross-sections for the Calico fault restraining bends (20-22) were not completed, as the complexity and large scale of those bends exceeds the scope of this work.downcutting, structural relief of the restraining bends could represent a maximum value for uplift.Furthermore, the erosional surfaces are not time-specific marker horizons and variation in pre-uplift form and syn-uplift modification by differential denudation are plausible.

Morphometric Analysis
We measured a suite of morphological metrics on fault configurations and topographic expressions of the 20 doubly fault bound SECSZ restraining bends that exhibit clear forms (i.e., are not embedded in inherited topography).These metrics include length (L), width (W), L/W (aspect) ratio, local relief, average slope, transverse swath profile skewness, and fault bifurcation angle (Table 1).Restraining bend areas were defined as polygons that follow the perimeter of bounding faults and continuations of structural relief beyond where secondary faults have terminated.From restraining bend area polygons, we measured length as the greatest straightline distance between the sides of the polygon and width as the greatest straight-line, length-perpendicular distance between the sides of the polygon using Google Earth Pro (Figure 6a).Length measurements are approximate for restraining bends with half-wedge geometries (Figure 2b).We measured average slope and local topographic relief within restraining bend polygons using raster analyses of 2 m resolution DEMs (Porter et al., 2022).We measured along-strike and transverse swath profiles along transects that cross the center and locus of relief of each restraining bend using the ArcSwath Toolbar in ArcMap (https://github.com/HUGG/ArcSwath) (Figure 6d).Swath widths were chosen to be 1/10 the width of each individual restraining bend to attempt to capture the restraining bend forms while mitigating the effect of local topographic variations associated with drainage basins that develop at the scale of the half-width of individual mountain blocks (Hovius, 1996).The general shape of the restraining bends is represented by averaged normalized longitudinal and transverse swath Restraining bend maps of erosion surfaces, structural relief, exhumed deposits, and faults for bends 5-6, 9-10, 13, 15, 17-18, 21, and 22 (a-g, respectively).For additional information on restraining bends, refer to Figure 1b and Table 1.Structural relief of restraining bends are based on displaced bedrock erosion surfaces matched on either side of a fault (blue lines and values).A minimum value for structural relief is given where alluvium and basin sediments bury bedrock on the downthrown side.The greatest estimated structural relief for each restraining bend has a strong relationship with local relief given by the DEM range of elevation for the entire restraining bend area (Figure S3 in Supporting Information S1: R 2 = 0.94).This relationship allows for local relief to act as a proxy where erosion surfaces were not mapped and for morphometric analyses (Figure 6).The estimated vertical displacement of the dissected Fry Mountain basalt flow is used for the greatest structural relief of bend 6 (red text in panel a, Figures S4 and S5 in Supporting Information S1).The red circles on (b) are the piercing points of an inferred ∼900 m bedrock offset.The lined shaded area in 5G along the Calico-Hidalgo faults is a general area of widely distributed exhumed sediments and exposed bedrock that is not as finely mapped as the other bends.profiles (Figure 6e).We calculated the skewness of the transverse swath profiles to characterize topographic asymmetry of the restraining bends (Figure 6f).

Estimating Restraining Bend Shortening and Uplift
We estimated uplift and shortening magnitudes and rates for seven mapped bends with positive flower structures (i.e., bends 5, 9-11, 13, 15, 18) to investigate the relationship between their geometry and the attributes of their host fault and to infer their influence and relative timing of development (Table 2).Restraining bend uplift magnitudes were taken from structural and local relief measurements largely based on mapping of erosional surfaces (Figure 5).Magnitude of restraining bend shortening, defined as the perpendicular fault motion accommodating convergent strain within the restraining bends, were calculated using constructed cross section profiles in Figure 3. Cross section fault dips are interpretations based on surface field measurements or idealized values and are therefore unvalidated at depth.The resulting estimates for shortening and depth of positive flower structure, therefore, have some unquantified uncertainty because they are highly dependent on the interpreted fault dips of the positive flower structures (i.e., steep fault dips = less shortening; gentle fault dips = more shortening) (Table 2).To attempt transparency in this uncertainty, we report whether there is low confidence for fault dips based on the presence or quality of field measurements of fault orientation (Table 2 and Table S3 in Supporting Information S1).Where the 1992 Landers rupture passed through the SECSZ restraining bends and field measurements were lacking, a 90°dip was assigned (bends 15, 17).All other restraining bend locations lacking field measurements were assigned an 80°d ip, common for positive flower structures (bends 5, 6, 18).Mapping and construction of cross section of bend 11 (Figure 2c) is based on previous work by Zachariasen and Sieh (1995).Shortening was not estimated for confluence bends or unmapped bends due to less constrained fault geometries.Our minimum shortening calculations consider vertical offset (structural relief) based on elevation differences between faults and the highest elevation of the cross section.Given the relative flatness of erosional surfaces and the predominantly high angle dips at the surface of most faults, we reasonably assume equal uplift rates on both faults and planar fault geometries without curving to steeper angles with depth.The calculated shortening values are specific to the trend of the associated crosssectional profiles and do not account for oblique shortening.By assuming fault inception and restraining bend formation are coincident we also estimate rates of shortening and uplift for mapped bends by dividing the uplift and shortening magnitudes by the age of the fault.The initiation of faulting is estimated by dividing values of fault netslip by slip rate, assuming slip rate is constant over time.Transverse profiles are grouped into a single plot by flipping the west asymmetric profiles so they can be easily compared and the average transverse shape is representative for all restraining bends.Both average transverse and longitudinal profiles show relatively flat tops, and the longitudinal profile shows an elongated "whaleback" shape with the highest relief generally near the center.(f) Scatter plot of skewness of transverse swath versus relief and average slope of each restraining bend.The skewness of transverse swath profiles is a measure of restraining bend asymmetry.Small restraining bends are highly asymmetric (skewness is negative) while large bends are relatively symmetric (skewness is near zero) in transverse view.

Numerical 3D Deformational Modeling
We use Poly3D, a Boundary Element Method code, in two distinct ways to test the kinematics of restraining bends in the SECSZ: (a) regional modeling to test if widespread transpression is expected in the SECSZ and what role it has played in the evolution of specific faults, and (b) local modeling of idealized fault configurations that are based on our observations and interpreted fault geometries, to compare expected vertical to observed deformation patterns.Modelers commonly call such idealized models "toy models" because the simplicity of the models allows for manipulation of various parameters, such as fault geometry and loading, to directly learn about system behavior.Poly3D discretizes three-dimensional fault surfaces into triangular dislocation elements to simulate stress, slip, and opening on complex curving fault systems without creating voids or overlaps (Thomas, 1993).
The model uses elastic rheology, which has been used with success to simulate deformation associated with many faults worldwide (e.g., Brankman & Aydin, 2004;Cooke & Marshall, 2006;Fedorik et al., 2022;Herbert & Cooke, 2012;Herbert et al., 2014;Meigs et al., 2008;Resor et al., 2018).Because the model does not explicitly include inelastic processes ubiquitous within the crust, the values from the linear elastic model may not match crustal values of uplift but the overall uplift pattern may be similar (e.g., Meigs et al., 2008).
The regional modeling uses 3D fault surfaces from the Southern California Earthquake Center (SCEC) Community Fault Model (CFM5.2;https://www.scec.org/research/cfm) of the SECSZ with recently published and new fault mapping (a total of ∼55 faults).We prescribe geodetically derived plate velocities to the outside edge of the model base oriented to the direction of plate motion of 322°at 42 mm/yr (Figure 1c) (e.g., Argus et al., 2011).Faults of the CFM extend to 35 km depth and meet a freely slipping horizontal crack that serves as the base of the model and simulates distributed deformation, independent of motion at the surface (e.g., Marshall et al., 2009).
Where faults meet the lateral edges of the model, we prescribe geologically derived slip rates to avoid fault motions diminishing to zero.Everywhere else within the model, the faults slip to relieve dip-and strike-shear tractions due to tectonic loading and fault interaction.The faults are also not permitted to open.We query uplift rates at points along an observational grid at the traction free Earth's surface focused on the SECSZ (Figure 1b).To assess the regional transpressive regime, we create an interseismic model that applies slip rates from the long-term regional model below the 20 km locking depth.This two-step modeling approach is similar to a Estimated by dividing the fault net-slip by its slip rate (assumes constant slip rate).Error ranges are not presented but are as much as 50%-100%.b Depth relative to surface and roughly calculated using the fault dips, map distances, and elevation differences between surface exposures of faults along the cross sections (Figure 3). a back-slip approach (Marshall et al., 2009).We correct for regional isostacy using a solution for an elastic plate of flexural rigidity 2e23 Pa * m 3 and density 2,700 kg/m 3 over a substrate with density 4,100 kg/m 3 following Cooke and Dair (2011).We average the surface horizontal strains within the SECSZ region from the interseismic model at points along the Earth surface.The resulting strong N-S contraction from the regional model is consistent with SHmax estimates from focal mechanisms (Heidbach et al., 2018).
We also used Poly3D to test the uplift patterns of seven idealized restraining bend fault geometries and configurations that were identified from our mapping (Figures 2 and 3).In each of the toy models, the faults slip in response to remotely applied strain rates derived from the regional model of the SECSZ.We transform the N-S E-W strain rate tensor from the regional SECSZ model into the coordinate system of the toy model where the primary faults trend is 322.6°,equal to the average SECSZ fault trend (Table 3) (Spotila & Garvue, 2021).
Because the strain rate in the toy models is applied to only one fault and not distributed across several active faults of the SECSZ, we also reduce the applied strain by a factor of 4 so that the resulting slip rates on the toy model faults are ∼1 mm/yr, comparable to those of the SECSZ.As an experimental control, we also run the toy models with a pure shear strain tensor that optimizes dextral slip along the primary faults (Table 3); maximum horizontal compression 45°clockwise from the primary fault trend and maximum horizontal extension 45°counterclockwise from the fault trend (Figure S6 in Supporting Information S1).To simulate long-term deformation over several earthquake cycles, the primary faults have a depth three times their length so that the basal fault tip does not restrict deformation.Modeled configurations include doubly bound examples that consist of +15°bends on 50-km-long primary faults and 15°-trending, 3-5-km-long secondary faults, which matches the general observation of fault bifurcation angles of ∼30°(Figure 7).For four of the fault configurations, the primary and secondary faults connect at shallow depth (≤5 km) in basement rock (Figures 2b-2f).An additional confluence model consists of two primary faults that do not connect at depth, but instead extend at depth as separate structures (Figure 2g).We present results as uplift rate patterns due to applied strain rate.

Fault Configurations
Restraining bends exhibit a variety of map-view fault configurations.We classify three of the fault configurations as true convergent stepovers (Figures 2a-2c), while the remaining configurations involve local deviations in trend that involve two bends (restraining and releasing), change in trend that involves only one bend, and a confluence of two primary faults (Figures 2d-2g).Most bends are doubly bound by faults, whereas only two restraining bends consist of a simple bend with a solitary primary fault (Figure 2a).Doubly bound bends are basement-cored and generally exhibit a consistent pattern of structural relief that is concentrated between the two bounding faults.The doubly bound bends can be divided into six fault configurations and several basic subsurface geometries.Halfwedge, full-wedge, paired bend (i.e., sidewall rip-out), and corner bends are interpreted to be shallow upper crustal positive flower structures in which the bounding faults meet at ≤5 km depth (Figures 2b-2e).Bounding faults generally dip steeply toward each other at ∼60°-90°(Figure 3), which is typical of positive flower structures (Naylor et al., 1986).Secondary faults with flower structure configurations consist of short, variable segments (∼1-5 km), which suggests they are intrinsically coupled to the geometry of the primary fault.Among these four configurations of flower structures, the half-wedge is the most prevalent in the SECSZ (Figures 1b and  2b).From the fault trend analysis, we found that the fault bifurcation angles (φ) of the doubly bound structures varies (14°-51°), however the average is ∼36°and there is a positive correlation between bifurcation angle and Toy models with optimal loading for dextral slip 0 0 17 Note.NS and EW normal strain: extension = positive, contraction = negative.Shear strain: left-lateral = positive, right-lateral = negative.
Tectonics 10.1029/2023TC008148 bend size (Figure 7d).These results suggest that secondary faults are essential for focusing uplift and local fault conditions, such as bifurcation angle, and may partially control the restraining bend form (see discussion Section 5.2 for more).
The remaining two configurations of doubly bound restraining bends consist of composite and confluence forms (Figures 2f and 2g).The confluence configuration is defined by an area of structural relief concentrated between two merging primary faults that extend for tens of kilometers from the point of bifurcation (Figure 2g).The primary faults in this case likely represent separate structures that have no subsurface connection.Confluence bends are not common but do represent some of the larger restraining bends in the SECSZ (e.g., bends 17 and 22).The composite configuration is distinct from all others in that it is manifest as a closely spaced succession of uplift along several fault bends.Composite bends can include any combination of restraining bend type and fault configuration and likely represent evolving deformation styles.The case illustrated in Figure 2f, modeled after bends 9 and 10 on the Johnson Valley fault (Figures 1b and 3b), is relatively simple, consisting of two half-wedge forms embedded within a large convergent stepover.A more complex example is represented by bends 5 and 6 on the Lenwood and Old Woman Springs faults (Figures 1b and 3a), which consist of a combination of a half-wedge geometry and a confluence bend geometry.There are several complications to assigning restraining bends to the various fault configurations.Some examples appear variable depending on the scale at which observed, as over longer distances the primary fault may return to its original form (e.g., converting a corner bend configuration in detail to a paired or half-wedge configuration at broader scale).Many bends may also be classified with multiple plausible configuration types (i.e., bends 12, 15, 17).Despite these complications, the shared characteristics in restraining bend structures (i.e., southward bifurcation, doubly bound, positive flower structure) appear to outweigh the variations as evidenced by their self-similar geomorphic forms (Section 4.4).
As a note, we limit our interpretations and choose to present only fault mapping and exhumed sediment distribution for the composite bends on the Calico-Hidalgo faults (bends 20, 21, and 22) (Figures 3g, 5f, and 5g), because they exhibit several distinct characteristics that set them apart in the SECSZ.First, the Calico and Hidalgo faults show sharp geomorphic features (Figure 4e) with widespread deformation and disruption of younger alluvial deposits that have been linked to a late-Holocene paleoseismic event (Vadman et al., 2023).Second, the exposed and folded pre-uplift deposits within and around the restraining bends are much more extensive compared to other bends in the SECSZ.Third, they have the highest magnitude of structural relief and fault complexity among all the SECSZ restraining bends.These differences may reflect the greater net-slip and strikeslip rate of the Calico-Hidalgo fault system when compared to other faults of the ECSZ (Dibblee, 1967a;Jachens et al., 2002;Oskin et al., 2007Oskin et al., , 2008)).Therefore, an ongoing follow-up study will cover this system in more detail.

Relict Surfaces and Older Deposits
Mapped erosion surfaces atop the restraining bends constrain structural relief (i.e., vertical displacement) across bounding faults (blue lines, Figure 5).The maximum observed vertical displacements among the seven mapped restraining bends range from 131 to 483 m (Figures 5a-5g).Note that minimum values of vertical displacement are reported where bedrock is buried by alluvium, colluvium, or grus on the downthrown side of bounding faults.
Structural relief is commonly greater across the primary through-going fault, resulting in transverse asymmetry and tilted erosion surfaces that may reflect differences in uplift rate or geometry of the bounding faults (Figures 3  and 5).For example, the erosion surfaces atop bend 15 on the Emerson fault show radial eastward tilting, suggesting uneven uplift on the west side of the positive flower structure (Figure 5d).Structural relief also tapers from the ends to reach a maximum near the center of restraining bends, thereby yielding the typical "whaleback" longitudinal profile (e.g., Figures 4a, 4b, and 5d).Maximum structural relief correlates to local topographic relief (R 2 = 0.94; Figure S3 and Table S1 in Supporting Information S1), suggesting that local relief is a reasonable proxy for vertical deformation where erosion surfaces were not mapped.Mapped erosional surfaces cover a greater areal fraction on smaller restraining bends than larger ones (Figure 5), which highlights the limited timeframe in which it is possible to obtain vertical displacements of these evolving structures before erasure by drainage development.
Eight of the field-mapped restraining bends exhibit older (i.e., pre-uplift) deformed and eroding deposits of varying texture and thickness (Figures 4j-4n and 5).The distribution of the exhumed deposits varies but is generally more prevalent along the southern boundaries of restraining bends.This tendency suggests that bends may have migrated southward with increasing uplift (e.g., bends 6, 13, 15, 17, 21, 22; Figure 5).Older deposits are also distributed along the outside edge of secondary faults.These undivided sediments are interpreted to pre-date the onset of local transpressional uplift and are thus useful when paired with the distribution of erosional surfaces to infer the pattern and migration of uplift.Although none of these units are dated, the pre-uplift interpretation (i.e., Miocene to middle Pleistocene?) is based on their induration, fine texture, clast provenance, surface expression (i.e., incised nature, disconnection to modern drainages), and from inferences in previous mapping (Dibblee, 1964a(Dibblee, , 1964b(Dibblee, , 1966(Dibblee, , 1967a(Dibblee, , 1967b) (Descriptions of specific deposits are provided in Table S4 in Supporting Information S1).

Restraining Bend Shortening and Uplift
Estimated restraining bend shortening and uplift values offer insights into the evolution of the restraining bends and their role in accommodating transpression along the SECSZ faults (Table 2).The selected SECSZ restraining bends show relatively small magnitudes of shortening, ranging from 34 to 223 m and shallow depths to fault intersection (<3,500 m) (Table 2).Percent shortening values (i.e., shortening magnitude/net-slip of fault), however, range from ∼1 to 26%, though most are ∼4-11%.The shortening estimates are reasonable given the high angle geometry of flower structures (e.g., Moulin & Cowgill, 2023) and they have important implications on the timing of restraining bend formation and the degree to which they accommodate transpression.
We also estimate rates of shortening and uplift for these bends by making reasonable assumptions about the timing of deformation (see Section 3.4).We find all restraining bend uplift rates are <0.4 mm/yr which agrees with modeled uplift rates from applied SECSZ loading (Table 2 and Figure 2).Additionally, we find shortening magnitudes and rates are generally less than uplift magnitudes and rates because of the steep fault geometries (Table 2 and Figure 3).To assess if the estimated rates are reasonable, we compared the SECSZ restraining bend uplift and shortening rates to hypothetical models using equation 1 from Wakabayashi et al. (2004).This equation enables the estimation of various fault displacements and rates by considering rigid hypothetical rock uplift through a true restraining bend with one or two (parallel doubly bound) faults with symmetrical positive flower dips.Hypothetical uplift magnitude can be calculated from where z is the total uplift, x is the total strike-slip displacement, θ is the bend angle relative to the principal fault trend, and δ is the dip of the fault in the bend.The hypothetical uplift and shortening magnitudes and rates are ∼1.2-5×greater than our estimates but are relatively close considering that hypothetical rates likely are maximums, as they do not account for internal block deformation.For example, we compare the predicted and estimated uplift and shortening values for bend 11, the smallest of the SECSZ restraining bends, on the Homestead Valley fault.The predicted uplift for bend 11 (assuming x = 0.32 km, θ = 12°, and δ = 50°) is 77 m at 0.06 mm/yr and shortening is 65 m at 0.05 mm/yr.In comparison, our estimate is ∼50 m of uplift at 0.04 mm/yr and 34 m of shortening at 0.02 mm/yr (Table 2).Zachariasen and Sieh (1995) found that the 1992 M w 7.3 Landers surface rupture caused coseismic uplift of bend 11 by ∼10-20 cm, based on field measurements.An estimated 200-500 similarly sized events would therefore be required to accomplish the 40-50 m relief of bend 11, assuming level pre-uplift topography.It is important to consider that slight variations in the defined parameters (e.g., fault dip, bend angle, net-slip, fault age) could affect the accuracy of our results, however the relatively close comparisons of restraining bend 11 indicate that our simple estimations of uplift and shortening are within reason.

Morphometric Comparison
The SECSZ restraining bends are kilometer-scale (L = 1.5-9.5 km; Table 1) and exhibit self-similar form.Their distinction from nontectonic features (i.e., inselbergs) is revealed by hypsometric analysis using the ArcMap tool CalHypso (Figure S1 in Supporting Information S1) (Pérez-Peña et al., 2009).Positive correlations exist between length and width (R 2 = 0.88; Figure 6b), length and local relief (R 2 = 0.69; Figure 6c), width and local relief (R 2 = 0.77; Figure S2 in Supporting Information S1), and average slope and local relief (R 2 = 0.55, Figure S2 in Supporting Information S1).Observed restraining bend length to width ratios (L/W) are consistent with previous laboratory experimental and crustal studies (e.g., Cunningham, 2007;Hatem et al., 2015), with values of 1.9-5.0 and an average of ∼2.9 (Table 1, Figure 6).Length-width ratios are also negatively correlated with relief (R 2 = 0.18, Figure S2 in Supporting Information S1), indicating that larger restraining bends tend to be wider.It is unclear if this relationship emerges because restraining bends are systematically widening more than they are elongating or if larger structures have lower aspect ratios because they have higher bend angles (Hatem et al., 2015).The self-similar character of restraining bends is also exhibited in elevation profiles.Longitudinal profiles generally exhibit a "whaleback" shape, with the highest relief near the point of fault bifurcation and gentle slopes that decrease away from the center of the restraining bend (Figure 6e).Transverse profiles are generally asymmetric, both in terms of elevation distribution and the elevation of the valley and bounding fault on either flank (Figure 6e).The flanks with greater relief generally correspond to the steeply dipping primary fault.Smaller restraining bends are also more asymmetric in transverse profile than larger restraining bends, as illustrated by the Tectonics 10.1029/2023TC008148 relationship of profile skewness to local relief and average slope (R 2 = 0.30 and R 2 = 0.44, respectively; Figure 6f).Two outliers in the relationship between local relief and skewness (bends 18 and 22) are the two largest restraining bends, suggesting that other unknown factors may influence their form.Despite such local variation, the dominant observed trend is that restraining bends exhibit self-similar, scale-independent map-view and profile forms.This result suggests that a pervasive driving mechanism may be facilitating the development of the selfsimilar restraining bends along the low-net-slip faults of the SECSZ.

Modeled Uplift Rate Patterns
Here we compare the pattern of modeled uplift rate to geomorphic expression of uplift.Predicted uplift rate patterns for doubly bound toy models with dipping upper crustal flower structures (i.e., half-wedge, full-wedge, paired bend, and corner bend) demonstrate that uplift concentrates between bounding faults (Figures 2a-2g).
Modeled uplift in half-wedge, paired bend, and corner bends are relatively similar to each other, with uplift and relief focused near the point of bifurcation and within the center of the bend that gradually tapers down to the unconnected half (Figures 2b-2d).The full-wedge fault configuration (Figure 2c) shows distinct uplift and relief within the region bounded by the primary and secondary faults.In contrast, the modeled simple vertical fault bend (Figure 2a) displays a distributed elliptical region of uplift across both sides of the fault.Similarly, the predicted uplift from the modeled vertical fault confluence bend (Figure 2g) also appears diffuse and has lower rates of uplift than the models with dipping secondary faults.The greatest uplift rates for the modeled confluence bend occur west of the primary fault.Lastly, the modeled composite bend example of two half-wedge positive flower geometries (Figure 2f) exhibits two or more concentrated regions of relief individually matching the uplift patterns of the single half-wedge model.This composite scenario is only one of many different possible configurations for the composite bend and therefore uplift patterns may vary with fault arrangement.In summary, modeled uplift rate patterns from restraining bend toy models are sensitive to fault geometry, and dipping faults are effective and perhaps necessary to concentrate uplift and accommodate convergent strain.
A comparison of uplift rate patterns between toy models and topographic observation of the SECSZ restraining bends reveals overall agreement.This agreement is particularly evident for doubly bound restraining bends with half-wedge, full-wedge, paired, and corner fault configurations which comprise a majority of the SECSZ bends (n = 17; Figures 2 and 3).The uplift rate patterns produced from toy models with applied pure shear strain are nearly identical to the results of Figure 2 that are loaded with regional strain (Figure S6 in Supporting Information S1).This suggests that similar uplift rate patterns may result from a range of loading and could be applied to similar systems outside of the SECSZ that have comparable fault configurations.The modeled uplift rates of the doubly bound restraining bends vary between 0.1 and 0.4 mm/yr along faults with ∼1 mm/yr of strike slip, roughly matching estimations based on structural relief of the SECSZ restraining bends and Equation 1 (Section 4.3).These findings support the interpretation that most bends have positive flower structures (Figures 2b-2e and 3).Conversely, the modeled vertical faults of the simple bend and confluence bend show discrepancies between the predicted uplift patterns and the topographic observations (Figures 2a and 2g).The modeled symmetrically distributed uplift on both sides of the single bend differs from the asymmetric relief patterns observed in SECSZ examples (such as bend 8 and 20, where bend 20 is shown in Figure 3g).Likewise, the diffuse uplift patterns of the modeled confluence bend with greatest uplift west of the primary north-south fault bears little resemblance to the fault-bound concentrated relief in representative bends of the SECSZ (Figure 3a-bend 6; Figure 3f-bend 17; Figure 3g-bend 22).
Using the modified SCEC CFM, we also analyzed the regional uplift in the SECSZ to understand and quantify the larger scale transpressional strain.Our results reveal widespread uplift in the southern Mojave block and San Bernardino Mountains (Figure 8).The modeled uplift rate in the San Bernardino Mountains is 1-4 mm/yr with a lower rate of <0.2-1 mm/yr given for the south Mojave block.

Discussion
Overall, the kilometer-scale restraining bends of the SECSZ are self-similar in form, characterized by an arrowhead shape in map view verging at the point of fault bifurcation and a longitudinal profile resembling a "whaleback" with limited relief (<100-600 m).The observed bend forms align with the uplift patterns predicted from modeled restraining bends providing support for our geometric interpretations and lending potential mechanisms for uplift and growth.Fault configurations vary, but most are doubly bound, basement-cored, and are Tectonics 10.1029/2023TC008148 dominantly expressed as positive flower structures in the upper crust with an average aspect ratio of ∼2.9.Variations in restraining bend form are found to be associated with structure size and fault characteristics.The calculated magnitudes of restraining bend shortening are relatively small.We find that lower net-slip faults display high percent shortening values, whereas higher net-slip faults have lower percent shortening values.Therefore, the SECSZ restraining bends are not only sensitive to fault geometry, but their local kinematics may also be influenced by the total accumulation of fault displacement.
Despite their widespread distribution and common form, the precise function of the SECSZ restraining bends in accommodating regional transpressional strain and timing of their formation remain unconstrained.Below, we discuss the factors influencing the form and uplift patterns of these restraining bends through analysis of empirical morphometric, structural, and kinematic data (Section 5.1) as well as 3D numerical models of deformation and uplift rate along restraining bends (Section 5.2).We then examine the role of these restraining bends in the evolution of the SECSZ (Section 5.3) and place our findings in a global context of understanding restraining bend and strike-slip fault evolution (Section 5.4).

Restraining Bend Form, Uplift, and Development Based on Empirical Constraints
To explore the influence of local factors on restraining bend development, we compared restraining bend form and morphometry to local fault conditions (see Section 3.1).Contrary to expectations, azimuthal trend of the transpressive fault strand (Fault A in Figure 7a), which is akin to obliquity, does not appear to relate to restraining bend size (Figure 7c).This finding suggests that the small bends of the SECSZ behave differently from larger ones, such as the big bend along the San Andreas fault, in which the degree of obliquity is a major control on the degree of transpressive deformation (Spotila et al., 2007).This result does not mean that deformation is independent of fault trend in the SECSZ.Spotila and Garvue (2021) showed that tectonic relief is more pronounced along transpressive oblique segments of the SECSZ faults relative to regional plate motion (see their Figure 11).The lack of an apparent relationship between obliquity and deformation within the isolated bends documented here may instead relate to local factors, such as scale and bifurcation angle, which could be more important in controlling the restraining bend form than the overall trend of the primary faults.Bifurcation angle has a positive correlation with local topographic relief and slope (R 2 = 0.31 and R 2 = 0.55, respectively; Figure 7d).Like bend form, bifurcation angles are relatively self-similar and span a limited range (14°-51°).A linear correlation between the strike of opposing faults in doubly bounded restraining bends (R 2 = 0.44, Figure 7b) indicates that the bifurcation angle remains relatively stable throughout a wide range of fault strikes and varying obliquity to plate motion.These results suggest that, while the SECSZ restraining bends occur largely along transpressive segments of strike-slip faults, the local fault geometries and interactions have a greater control on bend form and size than the obliquity of the faults relative to plate motion.
Restraining bends also appear to be influenced by fault maturity with larger bends typically associated with faults that have accumulated greater slip.The topographic relief of the largest restraining bend of each associated fault is linearly correlated to fault length (R 2 = 0.76) and net-slip (R 2 = 0.80) (Figure 9, Table S2 in Supporting Information S1).Another variation in restraining bend form that appears to relate to fault maturity is the bifurcation angle.Small restraining bends on faults with low cumulative slip generally have narrow bifurcation angles and transversely asymmetric profiles, while larger restraining bends have wide bifurcation angles and relatively symmetric transverse profiles (Figures 6 and 7).The local aspect ratio (L/W) also relates roughly to net-slip, as a weak negative relationship (R 2 = 0.18, Figure S2 in Supporting Information S1) exists between aspect ratio and bend size, meaning that small restraining bends tend to have high aspect ratios (i.e., are more elongate).Finally, small restraining bends on faults that exhibit less net-slip (e.g., Lenwood, Johnson Valley, and Homestead Valley faults) have higher percent shortening across them (>5%) than bends that occur on faults with higher net-slip (e.g., Figure 8. Regional uplift map of the southern Eastern California shear zone (SECSZ) from numerical 3D deformation modeling using Poly3D shows widespread uplift of the Mojave block and adjacent San Bernardino mountains.Values are corrected for isostacy.Dashed black polygon marks the area analyzed to estimate loading in the SECSZ which was applied to the toy models.Uplift rate patterns appear to match relatively well with the modern landscape (Figure 1b).Widespread uplift and strain tensors (Table 3) from regional modeling support a transpressional regime for the SECSZ.Helendale, Emerson-Camp Rock, and Calico-Hidalgo faults).These observations have important implications for the timing of restraining bend formation and their evolution along the faults (see Section 5.4).
One way of interpreting the apparent relationships between restraining bend form and fault maturity involves considering the timing of restraining bend formation along the host fault.If the restraining bends each formed early in the history of their host fault, the relationship between bend size and fault slip may reflect a gradual accumulation of shortening along the bend throughout the fault's history.The wider bifurcation angles and lower aspect ratios (L/W) on larger restraining bends may similarly reflect a greater accumulation of shortening and contraction across the bend, whereas shortening on smaller bends (i.e., low net-slip faults) may not yet have warped the trace of the fault to the same degree.Figure 10 illustrates a possible mechanism of progressive widening of bifurcation angle and development of more symmetric transverse topographic profiles with increasing strain.The primary fault in the depicted common half-wedge positive flower model may accrue initial vertical offset prior to the formation of the future secondary fault, resulting in an initial transverse asymmetry.With increasing strain and vertical offset, the restraining bend may become more symmetrical via relatively equal uplift rates on both faults effectively diluting the initial asymmetry.This pattern could be transient and relate to the simple geometry and shallow crustal penetration of the SECSZ restraining bends, because most high net-slip bends have multiple deforming blocks that together display transversely asymmetric profiles (Mann, 2007).Other changes in restraining bend form with increasing net-slip could also reflect evolving conditions during the fault's history.For example, the decrease in percent shortening across the bends with increasing net-slip could reflect an increase in the efficiency of accommodating strike-slip strain through the bends with time.This phenomenon has been observed in physical experiments (e.g., Cooke et al., 2013).The apparent relationships between bend form and fault slip may thus reflect a continual evolution in the mechanisms and expression of deformation in the bends.
There may also be dynamic factors that influence the evolution of restraining bend form with fault net-slip.For example, the widening of restraining bends with increasing slip that is reflected in aspect ratios and bifurcation angles may be influenced by progressive topographic buildup within the crustal wedge between bounding faults.The development of a secondary fault may be mechanically efficient but may result in a space problem near the point of bifurcation, which could increase the vertical load and warp both bounding faults outward, thereby widening the bifurcation angle (Figure 10).Differences in slip rate of the two bounding faults at the point of bifurcation result in zones of compression and tension that could further facilitate the growth of new secondary faults (orange and blue areas in Figure 10).Field evidence of such widening includes segmented secondary faults near bifurcation points and increasing fault activity toward the outside edge of wedge flanks (inferred from increased geomorphic sharpness, e.g., bends 5, 13, 17, 18, 21, and 22).Exposures of tilted, exhumed deposits and pediment surfaces at the southern and western flanks of bends also suggest southward migration of uplift that is consistent with wedge growth associated with buttressing effects within the restraining bends (Figures 4 and 5).S2 in Supporting Information S1.
The resulting widening and growth of the restraining bend may be accommodated by either a decrease in the dip of one or both bounding faults (i.e., constant subsurface geometry or zippering up) or by an increase in depth of the intersection point of the bounding faults under the wedge (i.e., unzippering down) (Figure 10).Additional dynamic factors may also influence the growth of the restraining bends, such as the growth of subsidiary faults, development of listric faulting, block rotation, or internal block deformation (i.e., folding).
Although these interpretations provide viable hypothetical models for the evolution of restraining bends with increasing strain, which help to explain some of the apparent relationships between bend morphology and fault net-slip, other interpretations of the relationships are also possible.For example, the onset of restraining bends may not have occurred during the onset of the fault, but rather could have formed later after the fault had already accrued significant displacement.This is somewhat indicated by the presence of multiple restraining bends at different stages and sizes along the same faults, such as the Calico (Table 1; Length = 1.5, 3.1, 9.3 km for bends 19, 21, and 22, respectively) and the Emerson (Table 1; Length = 1.7, 2.9, 3.0, 5.0, 6.7 km for bends 14-18, respectively) faults.Another possible explanation is that restraining bends may have formed at a single time due to a change in regional tectonics but have grown to different forms due to unique shortening and uplift rates that are tied to local conditions (e.g., fault geometry or host fault slip rate).It is also possible that the observed trends reflect local initial conditions that predisposed bend deformation, but which have subsequently been altered by that deformation.For example, previous experimental work has shown how restraining bend form can vary depending on initial boundary conditions, including the bend angle (i.e., obliquity) (e.g., Mitra & Paul, 2011).Physical experiments in wet kaolin by Hatem et al. (2015) demonstrated that bend angle is a crucial factor in controlling aspect ratios and secondary fault development.They found that the aspect ratios of bends with angles ≥15°increase with cumulative slip.Like Mitra and Paul (2011), they also found that at the initiation of slip, the aspect ratios for high angle (i.e., sharp) restraining bends were generally lower than those on low angle (i.e., gentle) bends.Similar circumstances may apply to the bends in the SECSZ.It is possible that the aspect ratios of the restraining bends in the SECSZ either initially reflected their bend angles or were subsequently altered by evolving conditions, potentially disrupting the original correlation between obliquity and bend morphology.
Several restraining bends do not fit with the interpretations of bend evolution with increasing strain described above, including the composite bends of the Johnson Valley fault (9 and 10) and of the Calico-Hidalgo fault system (20-22) (Figures 3b and 3g).The Johnson Valley fault bends appear to have originally been a large coherent restraining bend that was dissected and displaced, resulting in two zones of focused relief (Figures 3b  and 5b).Based on the fault-bound curve of bedrock features, we estimate ∼900 m of right-lateral displacement since the dissection of the proto-restraining bend (red circles in Figure 5b denote piercing points).The kinematic evolution of this bend may have occurred in three stages (Figure 11).An initial right-step may have formed, followed by the accrual of convergent strain, which resulted in the formation of a secondary fault that concentrated uplift of the proto-restraining bend.Finally, subsidiary structures may have grown due to mechanical inefficiency into a through-going fault that dissected and displaced the original structure.Since the left-step still remains on the primary fault, contractional strain is accommodated by short, more oblique reverse faults that are connected by strike-slip tears.This evolution effectively reduces the original bend angle with increasing slip, although continued growth of the strike-slip strands within the bend could lead to widening of the restraining bend and further geometric disturbance.The chain of bends of the Calico-Hidalgo fault system may have a comparable history but will be addressed further in a later study.
In summary, the SECSZ restraining bends exhibit similarities in form but appear to be sensitive to local conditions beyond the strike of the primary fault.The dominant self-similar form (i.e., arrowhead-shaped and "whaleback" longitudinal profile) appears to relate to the formation of secondary bounding faults (i.e., doubly bound) and a positive flower structure.Details of this form, including the magnitude of uplift, percent shortening, and aspect ratio, are sensitive to local parameters, such as bifurcation angle and cumulative slip of the host fault.Their varied forms also seem to relate to progressive strain accrual, indicating ongoing migration and expansion.Yet the specific timing of bend formation across the SECSZ remains unclear and progressive deformation in some cases appears to obscure the kinematic evolution.

Restraining Bend Form, Uplift, and Development Based on Numerical Modeling
Numerical modeling experiments offer additional insights into the kinematics of restraining bends, particularly on the ubiquity of secondary faults (i.e., doubly bound restraining bends) and specific mechanisms of uplift.From topographic and mapping evidence, we infer that convergent strain within restraining bends leads to the formation of secondary faults by accommodating uplift.This is supported from our modeling via kinematic analyses of shear strain plotted as vorticity on Figure 12a and velocity vectors relative to a fixed position (red star on Figure 12b).Figure 12a shows the predicted shear strain under applied SECSZ loading across a 15°left-step.The highest right-lateral shear strain (in red) is concentrated on the outer bend corners, thereby providing a possible mechanism for the growth of new secondary faults to accommodate the accumulating strain.Once the secondary fault is formed, we infer that the local convergent strain can be efficiently accommodated via slip along the new dipping fault, leading to concentrated uplift as a positive flower structure.This is supported by modeling of the half-wedge fault configuration where fault-perpendicular motion is greater within the restraining bend than on the west side of the secondary fault (Figure 12b).A similar mechanism may be operating across all the positive flower structures in the SECSZ (i.e., full-wedge, paired bends, corner bends) (Figure 2), and thus explains the observed prevalence of doubly bound structures.This result suggests a deformational continuum, in which an initial restraining bend forms, then the accrual of convergent strain results in shear strain at the bend corner, leading to the formation of a secondary fault, which then allows for concentrated uplift of the inner wedge.
The inference of secondary fault growth accommodating restraining bend contraction is generally supported by previous work.From analog wet kaolin experiments by Hatem et al. (2015), secondary fault growth initiates at the corner of the bend after a certain amount of convergent strain is accommodated (see their Figures 4 and 8).Finite element modeling of the San Andreas fault has produced similar findings, where the greatest plastic off-fault strain rate is concentrated at the outside bend corners and could be the impetus for the growth of secondary fault systems (i.e., San Jacinto and Elsinore faults, ECSZ) (Li et al., 2009;Ye & Liu, 2017).Our results (Figure 12b) also roughly match those of Toeneboehn et al. (2018), who found from particle tracking in physical experiments that fault perpendicular motion and uplift increased within the restraining bend once the secondary dipping thrust fault formed (their Figure 6c).These examples give further support for the interpreted mechanism of secondary fault formation and uplift via squeezing between the bounding faults.
There is greater uncertainty in the uplift mechanism acting within the poorly predicted toy model uplift maps for the rarer yet significant single simple bends and confluence bends (Figures 2a and 2g).The distributed uplift pattern of the single simple bend does not match the exaggerated asymmetric relief observed for SECSZ examples (bend 8 and bend 20).The most plausible explanation for the difference is that the single bend was modeled as a vertical fault, whereas a dipping fault may produce an asymmetric uplift pattern by encouraging dip-slip motion along the bend.All the models that include dipping faults have greater uplift rates than the models with perfectly vertical faults.A similar argument could explain the modeled diffuse uplift pattern of the confluence fault, unlike what is observed for the SECSZ examples (bends 6, 17, 22).The modeled vertical faults for the confluence bend may split strike-slip evenly along the separate structures whereas dominate slip along one primary fault bend may produce slip grades that result in vertical motion.Future studies should consider modeling non-vertical faults and altering the orientation of applied loading for these scenarios, as they may provide uplift patterns that align better with the observed topographic features in both the single bend and confluence bend fault configurations.It would also be advantageous to obtain additional empirical constraints on the dip of poorly exposed strike-slip faults, such as via geophysical imaging techniques.
Predicted uplift rate patterns from the simplified elastic models largely align with the vertical displacement patterns of the SECSZ restraining bends and are thus supportive of the interpretation that local transpression is accommodated by upper crustal restraining bend fault geometries as inferred from field and topographic data.These models also provide specific and consistent mechanisms under various loading conditions that may govern the restraining bend geometry and form within the SECSZ and other analogous systems.Such mechanisms include high shear strain at bend corners and fault perpendicular motion that may lead to the development of secondary faults and subsequent uplift of doubly bound restraining bends.While these models are limited in that they do not account for an evolving restraining bend morphology influenced by bedrock rheology, erosion, and fault evolution, they do provide viable explanations for the characteristic restraining bend geometry and form.

Regional Transpression in the SECSZ
Regional kinematic modeling supports earlier work showing evidence for transpression with widespread uplift in the SECSZ (Figure 8) (Bartley et al., 1990;Liu et al., 2010;Spotila & Garvue, 2021).Regionally averaged loading of the SECSZ (dashed box in Figure 8) shows strong N-S contraction at 85 nanostrain/yr, E-W extension at 48 nanostrain/yr, and minimal N-S and E-W shear strain at 3 nanostrain/yr (Table 3).The resulting modeled uplift rates in the neighboring San Bernardino Mountains (corrected for isostasy) range from 1 to 4 mm/yr, matching previous models (e.g., Cooke & Dair, 2011;Hatch et al., 2023) and estimates from thermochronometry (1.5-7 mm/yr) (Spotila et al., 1998(Spotila et al., , 2001) ) (Figure 8).The modeled uplift rate for the SECSZ ranges from <0.2-1 mm/yr, with an average uplift of ∼0.76 mm/yr.This average rate exceeds model rates for individual restraining bends (<0.4 mm/yr) and would translate to ∼1.9 km of uplift of the SECSZ over the last ∼2.5 Ma, a minimum age estimate for the onset of shortening of the San Bernardino Mountains (Cox et al., 2003).This magnitude of uplift seems unreasonable given the elevation range (∼500-1,200 m), subdued topography and presumed low erosion rates, and largely internal drainage of the SECSZ.It is possible that the model overestimates regional uplift rate because it does not account for distributed inelastic strain or other unknown factors.Nevertheless, the SECSZ is elevated, and the predicted uplift rate pattern generally matches the elevation pattern, with greater uplift rates and elevation to the west and lower rates and elevation to the east (Figures 1 and 8).These findings at least support that the SECSZ is experiencing regional contraction.
The SECSZ restraining bends themselves are unlikely to have accommodated substantial regional shortening.While percent shortening values may be high for some restraining bends, they pertain only to the restraining bend as a product of local fault geometry and thus only a fraction of the overall fault length.Less than 15% of the total length of the involved faults fall within the SECSZ restraining bends.The net amounts of shortening are also minor (34-223 m), despite the topographic prominence of the restraining bends, due to the high-angle fault geometries of the positive flower structures (Table 2 and Figure 3).Although there are additional convergent structures in the SECSZ, their contributions to regional shortening remain undefined.These structures include 15 additional restraining bends which we did not study in detail due to limited data, smaller bends with minor topographic expression (Spotila & Garvue, 2021), broad secondary folding and inverted basins between faults, such as in the Deadman Basin between the Calico-Hidalgo and Pisgah-Bullion faults (Dibblee, 1967a), and the contribution of off-fault deformation to regional convergent strain (Evans et al., 2016;Herbert et al., 2014).Additional work is required to quantify the total regional shortening accommodated across the SECSZ.

Global Perspective on the Kinematic Controls and Evolution of Restraining Bends
From our observations, the low net-slip SECSZ restraining bends are somewhat similar to other, large bends elsewhere (i.e., San Bernardino Mountains, Alaska Range, Akato Tagh bend).They form under various conditions where transpressional strain is accommodated along multiple faults, often through an asymmetric positive flower structure (Mann, 2007;McClay & Bonora, 2001;Woodcock & Fischer, 1986).As net-slip and convergent strain accumulate, restraining bends, if not abandoned, tend to migrate, grow, and add more faults (Burkett et al., 2016;Wakabayashi et al., 2004).There are differences, however, a few of which are illustrated in Figure 13, that may reflect regional geometric developments and accompanying kinematic changes.For instance, the basement-cored SECSZ restraining bends are generally simple in geometry and are usually bound by high angle faults accommodating oblique motion with a single intact wedge, whereas larger restraining bends may partition strain between separate strike slip and thrust faults across multiple deforming wedges (Figure 13a).The kinematics and controls on restraining bend form and uplift also seem to evolve with increasing slip.The size and shape of small restraining bends are predominantly influenced by local factors such as fault bifurcation angle leading to self-similar forms, whereas larger restraining bends are divided into multiple blocks with spatially varying uplift patterns that are significantly influenced by the obliquity of the primary fault trend to plate motion (Spotila et al., 2007).This difference of form with bend size is partially attributed to the simple geometry and confinement of small restraining bends to the upper crust due to their narrow width and the dips of the associated convergent structures, whereas the width and scale of large restraining bends allows for structures to penetrate through the seismogenic crust or even through the mantle lithosphere (Figure 13b) (Fuis et al., 2012;Langenheim et al., 2005;Niemi et al., 2013).Crustal depth of restraining bend penetration has seismic implications, as smaller restraining bends, and stepovers in general (<5 km separation), are less capable of arresting large earthquake ruptures, although they still influence rupture dynamics as observed from historic multi-fault ruptures (Sieh et al., 1993;Wesnousky, 2006;Xu et al., 2018).Furthermore, individual faults within large restraining bends may themselves be capable of independent large earthquakes (e.g., 1992 M w 6.2 Big Bear).These differences likely reflect the changing controls on restraining bend kinematics with increasing convergence as well as the evolution of the strike-slip system itself.
An unresolved question regarding the evolution of transpressive systems is whether the faults tend toward strain-weakening, where faults evolve toward a single slip surface, or strain-hardening, where faulting remains distributed.Elements of our results support both end member models.Strain-hardening predicts the long-term maintenance of wide and structurally complex shear zones (Hobbs et al., 1990;Morrow et al., 1982;Tikoff & Fossen, 1993).Our observations of the restraining bends within the SECSZ support this model, as they display continuous growth and development of secondary structures that distort the primary faults and lead to progressive and migrating uplift.Additionally, the SECSZ bends are at various stages of development along individual faults, suggesting sequential generation from ongoing warping and deformation of the primary fault zone.Other aspects of our observations may align with the strainweakening model where an initially wide and diffuse strike-slip system with numerous stepovers should eventually integrate, narrow, and exhibit reduced fault friction as net-slip increases.In this case, fault kinks, including restraining bends, should eventually be displaced by through-going structures and abandoned (e.g., Hatem et al., 2017).The evolution of the Johnson Valley restraining bends (9 and 10) exemplifies this phenomenon, but it is uncertain whether the fault bend became smoother as a result (Figure 11).It is possible that strain-weakening and strain-hardening are concurrently operating in the SECSZ (e.g., Cowgill et al., 2004), with one mode representing the natural trend for a growing fault system and the other imposed by specific tectonic boundary conditions, such as regional transpression or tectonic escape (Dixon & Xie, 2018).The SECSZ may also be too young in its evolution (i.e., an immature system with low netslip) to have fully equilibrated into the strain-weakening state (Dolan & Haravitch, 2014;Spotila & Garvue, 2021).These uncertainties regarding mechanical evolution leave the future of the SECSZ restraining bends unclear.While strain-weakening would point to the integration and smoothening of the primary faults with time, continued transpression and fault warping could lead to continued bend growth via the development of new faults which lengthen, widen, and deepen the system.The latter possibility could lead to linking of individual restraining bends resulting in the formation of continuous interconnected transpressional ridges, like what has been modeled for other transpressional systems (e.g., Cunningham, 2007).More investigation is required to elucidate the mechanical evolution and future of the SECSZ.

Conclusions
We investigated the factors influencing the form, geometry, and development of 22 restraining bends along low net-slip faults in the SECSZ.The restraining bends of the SECSZ exhibit self-similarity in form and geometry, with linear relationships between length, width, and relief despite variations in size and primary fault characteristics (i.e., restraining bend type, slip rate, net-slip).Our findings indicate that the presence of secondary faults is crucial for concentrating uplift via dip-slip motion on the bounding structures due to local contraction within the restraining bend.As convergent strain increases, migration and widening of the restraining bend may occur due to the buildup of topography and buttressing effects between the bounding faults.This mechanism may drive the topographic evolution (e.g., changing cross-profile symmetry) and characteristic features of the restraining bends, such as basement-cored positive flower structures, arrowhead shape in map view, and distinctive longitudinal whaleback profile.Observations also point to a dominance of high-angle bounding faults that intersect at shallow depths through basement rock, which indicates that the topographic form of the bends may have been accomplished with relatively small magnitudes of fault-normal shortening.There is also evidence that local conditions, such as the bifurcation angle of the doubly bound structures, partially control the size and form of the restraining bends.The size and aspects of the form of bends also correlates to the maturity (i.e., net-slip) of the host fault, which suggests some uniformity in the timing of bend formation relative to the history of fault growth.Other characteristics of the bends may be a consequence of structural changes during deformation (e.g., widening of the restraining bend may induce widening of the bifurcation angle), rather than causative initial parameters.A minority of bends appear to have experienced significant geometric rearrangement and increased strike-slip efficiency due to progressive deformation.Results also show that the restraining bend forms along low-slip faults of the SECSZ are not strongly influenced by the strike of the primary faults themselves, unlike what is observed on high net-slip faults and large restraining bends elsewhere.
There are several implications of these findings.The morphology and kinematics of the restraining bends indicate that they are a function of the local geometry and structural evolution of the primary faults in the SECSZ.Although regional modeling results suggest that the SECSZ is indeed experiencing widespread contraction, the shortening across individual restraining bends does not appear to contribute significantly to it.The likely future evolution of the SECSZ and its restraining bends are also unclear, as aspects of the deformation are consistent with both the strain-hardening and strain-weakening mechanical models.Results also bear on seismic hazards.The finding that the secondary faults and positive flower structures may only penetrate the upper third of the seismogenic crust implies that the SECSZ restraining bends may not impede large ruptures.This fits with the observed prevalence of complex multifault ruptures in the SECSZ that seemed to pass through numerous restraining bends.Finally, this study demonstrates a feasible approach to studying basement cored transpressive uplifts.In particular, minimal erosion of preserved surfaces developed on basement uplifts in arid conditions can facilitate the restoration of vertical displacement patterns using geomorphic observations, even in the absence of pre-uplift onlapping deposits.The combination of structural mapping and interpretation with numerical deformational modeling also strongly aids in the interpretation of the 3D kinematics of complex deformation associated with restraining bends.In particular, the match between observed and predicted patterns of uplift based on modeled structural interpretations highlights a shared accuracy of the independent approaches.

Figure 1 .
Figure 1.Maps of the Eastern California shear zone (ECSZ).(a) Regional fault map of the ECSZ.(b) Fault and topographic map of the southern Eastern California shear zone (SECSZ).Red highlighted faults show the 1992 Landers and 1999 Hector Mine surface rupture paths.Numbers and circles correspond to restraining bends studied in Table 1.Fault abbreviations: Bf, Bullion; Cf, Calico fault; CMf, Copper Mountain fault; CRf, Camp Rock fault; Ef, Emerson fault; Hf, Hidalgo fault; HLf, Helendale fault; JVf, Johnson Valley fault; Lf, Ludlow fault; LLf, Lavic Lake fault; LWf, Lenwood fault; MLf, Mesquite Lake fault; NFTf, North Frontal Thrust fault; OWSf, Old Woman Springs fault; Pf, Pisgah fault; PMf, Pinto Mountain fault.(c) Modified Southern California Earthquake Center CFM (∼55 faults), from Hatch et al. (2023), with prescribed geodetically derived plate velocities to the base of the model oriented to the direction of plate motion of ∼322°.From this model, we derive the loading of the SECSZ to infer regional strain and to apply to local models.

Figure 2 .
Figure 2. Representative map view fault configurations of the southern Eastern California shear zone (SECSZ) (left) and the modeled uplift patterns when SECSZ loading is applied to toy model restraining bends in Poly3D (right; δ = fault dip).Middle column shows the SECSZ restraining bends that fall under the adjacent fault configurations.See Figure 1b and Table 1 for location and other details of the SECSZ restraining bends.(a) Simple fault: single vertical fault bend.(b) Halfwedge: fault bend with secondary fault that does not fully connect back to primary fault.(c) Full-wedge: fault bend with secondary fault that does fully connect back to primary fault.(d) Paired bend: A paired restraining and releasing bend with a half-wedge restraining bend section.(e) Corner: a bend due to a change in fault trend with a secondary fault in the inner corner.(f) Composite: A series of closely spaced or stacked restraining bend structures.Composite bends may include all other types of bends.(g) Confluence: the confluence of two faults that are separate structures and do not connect at depth.

Figure 3 .
Figure3.Perspective shaded relief view of restraining bends and simplified fault maps with their interpreted cross sections for bends5-6, 9-10, 11, 13, 15, 17-18, and  20-22 (a-g, respectively).All doubly bound restraining bends have a roughly arrowhead plan view shape where structural relief is concentrated between the bounding faults.Cross sections are true scale and interpretations are based on quality and presence of field measurements (see Section 3.4).The blue colored lines on cross section profiles mark the extent of mapped erosion surfaces.In cross section, the yellow regions are inferred alluvial valley fill.White arrows on maps show dip direction and location of field measurements with the associated dip magnitude.Queried faults in cross-section are lower confidence interpretations due to a lack of coverage or lower quality field measurements (Table2and TableS3in Supporting Information S1).Cross-sections for the Calico fault restraining bends (20-22) were not completed, as the complexity and large scale of those bends exceeds the scope of this work.

Figure 5 .
Figure5.Restraining bend maps of erosion surfaces, structural relief, exhumed deposits, and faults for bends5-6, 9-10, 13, 15, 17-18, 21, and 22 (a-g, respectively).For additional information on restraining bends, refer to Figure1band Table1.Structural relief of restraining bends are based on displaced bedrock erosion surfaces matched on either side of a fault (blue lines and values).A minimum value for structural relief is given where alluvium and basin sediments bury bedrock on the downthrown side.The greatest estimated structural relief for each restraining bend has a strong relationship with local relief given by the DEM range of elevation for the entire restraining bend area (FigureS3in Supporting Information S1: R 2 = 0.94).This relationship allows for local relief to act as a proxy where erosion surfaces were not mapped and for morphometric analyses (Figure6).The estimated vertical displacement of the dissected Fry Mountain basalt flow is used for the greatest structural relief of bend 6 (red text in panel a, FiguresS4 and S5in Supporting Information S1).The red circles on (b) are the piercing points of an inferred ∼900 m bedrock offset.The lined shaded area in 5G along the Calico-Hidalgo faults is a general area of widely distributed exhumed sediments and exposed bedrock that is not as finely mapped as the other bends.

Figure 6 .
Figure 6.Morphometric plots of the 20 doubly bound southern Eastern California shear zone restraining bends.Results generally demonstrate that restraining bends are self-similar regarding their form and size (a-c, e, f).(a) Schematic depiction of restraining bend length (L) and width (W).The average L/W ratio is ∼2.9.(b) Restraining bend length to width scatter plot.(c) Restraining bend length to local relief scatter plot.Local relief is defined by the range of elevation (max-min in DEM) over the defined restraining bend area.(d) Schematic depiction of swath profiles, transverse and longitudinal, collected for each restraining bend.Swath width = 1/10 of restraining bend width.(e) Averaged and normalized transverse and longitudinal swath profiles (thick solid lines) with error of sigma-1 (thick dashed lines) and individual bends shown as thin solid lines.The direction of transverse asymmetry included both east (east side lower elevation, N = 14) and west (west side lower elevation, N = 6).Transverse profiles are grouped into a single plot by flipping the west asymmetric profiles so they can be easily compared and the average transverse shape is representative for all restraining bends.Both average transverse and longitudinal profiles show relatively flat tops, and the longitudinal profile shows an elongated "whaleback" shape with the highest relief generally near the center.(f) Scatter plot of skewness of transverse swath versus relief and average slope of each restraining bend.The skewness of transverse swath profiles is a measure of restraining bend asymmetry.Small restraining bends are highly asymmetric (skewness is negative) while large bends are relatively symmetric (skewness is near zero) in transverse view.

c
Structural relief of restraining bends, except for bend 11 which is the local relief as there are no preserved erosional surfaces.d Percent shortening is equal to the ratio of the estimated shortening magnitude over fault net-slip.e Strane (2007).fOskin et al. (2008).g Lower confidence dips.h This study, shortening for bend 9 and 10 are assumed to have occurred since the displacement of the proto-restraining bend (Figures5b and 11).iRockwell et al. (2000).jZachariasen and Sieh (1995).kJachens  et al. (2002).l Andrew and Walker (2017).m Mean of Oskin et al. (2007) (1.8 ± 0.3 mm/yr) and Xie et al. (2018) (3.2 ± 0.4 mm/yr).

Figure 7 .
Figure 7. (a) Schematic depiction of fault bifurcation angles as measured on the 20 Eastern California shear zone doubly bound restraining bends.θA and θB are the acute angles of Fault A (eastern fault branch) and Fault B (western fault branch) relative to the strike of the main strand north of the point of bifurcation.The bifurcation angle (φ) is the acute angle between Fault A and Fault B. (b) Scatter plot of the azimuthal trend of Fault A and Fault B for each restraining bend show a positive relationship (R 2 = 0.44) indicating a somewhat constrained bifurcation angle possible (<60°).Additionally, the most northerly trend for Fault A and southerly trend for Fault B of the studied restraining bends are approximately equal to the direction of plate motion (∼322°) as shown by the thin red lines.The simplified arrow diagram below graphically shows this relationship.(c) Scatter plot of Fault A trend versus relief (R 2 = 0.00) and average slope (R 2 = 0.11) show minimal to no relationship suggesting that the transpressive angle relative to far-field plate motion (west of ∼322°) does not correspond to restraining bend size.(d) Scatter plot of bifurcation angle (φ) versus restraining bend relief (R 2 = 0.31) and average slope (R 2 = 0.55) show a positive linear relationship indicating that local conditions, such as bifurcation angle, may play a larger role in controlling restraining bend size and form of these small restraining bends than far-field plate motion relative to fault trends (c).The simplified arrow diagram below the plot graphically shows this relationship.

Figure 9 .
Figure 9. Plots of the largest restraining bend per fault versus fault length and net-slip show a positive relationship indicating a link between restraining bend size and fault maturity.This suggests that these restraining bends may have formed early in the southern Eastern California shear zone's history and have since grown with cumulative strain.Fault abbreviations are in the caption of Figure 1.For more details, see TableS2in Supporting Information S1.

Figure 10 .
Figure 10.Schematic fault map and cross-section of restraining bends that show a possible mechanism for widening bifurcation angle, transverse profile symmetry, and uplift migration with progressive strain.(1) Initial configuration of a halfwedge restraining bend.Continued strike-slip displacement through the restraining bend results in focused uplift between the two faults.(2) Bifurcation angle in map view widens due to warping of the faults and growth of new faults due to increased vertical loading.The subsurface fault geometry may kinematically respond in several ways.If the depth of intersection of the positive flower structure is constant or decreases (zippering up), widening of the restraining bend would predict shallower fault dips with cumulative strain.If the depth of the positive flower structure increases (zippering down) then the fault dips may remain constant despite the growing width of the restraining bend.Other possibilities may include growth of other subsidiary faults, faults that curve, block rotation, and internal block deformation.

Figure 11 .
Figure 11.Kinematic evolution of the Johnson Valley restraining bend.The left side shows an interpreted schematic reconstruction while the right side shows a simplified version with faults color coded (blue: pure strike-slip, orange: oblique reverse, red: reverse).(1) First, an initial right-step forms, then (2) the formation of a secondary fault facilitates concentrated uplift between the two bounding faults.Lastly, (3) a new fault forms and dissects the proto-restraining bend, however the leftstep remains and so contractional strain is now accommodated by shorter oblique reverse faults connected by tear faults.This kinematic change displaces the restraining bend, lessening the original bend angle but results in more faults and focused zones of uplift.

Figure 12 .
Figure 12.Modeled shear strain (vorticity) of simple single 15°bend (a) and uplift with overlain strain velocity vectors relative to fixed position (red star on b) for the half-wedge geometry with applied southern Eastern California shear zone loading.(a)The simple fault bend shows high dextral shear strain (red) thereby providing a mechanism for growth of new secondary faults (dashed black arrows).(b) Once a secondary fault forms, uplift is highly concentrated between the faults as a positive flower structure.Fault perpendicular motion within the restraining bend is greater than that on the west side of the secondary fault.This discrepancy provides a "squeezing" mechanism for uplift within the wedge.

Figure 13 .
Figure 13.Schematic map (a) and cross section (b) showing faults, uplift patterns, and subsurface geometries between small and large restraining bends.

Table 1
Restraining Bend Information and Morphometrics bTransverse swath profile skewness.c Bifurcation angle in degrees.

Table 2
Restraining Bend Depth, Shortening, and Uplift Estimations

Table 3
Horizontal Model Strain Rate