We present tectonic reconstructions and an accompanying animation of deformation across the North America–Pacific plate boundary since 36 Ma. Intraplate deformation of southwestern North America was obtained through synthesis of kinematic data (amount, timing, and direction of displacement) along three main transects through the northern (40°N), central (36°N– 37°N), and southern (34°N) portions of the Basin and Range province. We combined these transects with first-order plate boundary constraints from the San Andreas fault and other areas west of the Basin and Range. Extension and strike-slip deformation in all areas were sequentially restored over 2 m.y. (0–18 Ma) to 6 m.y. (18–36 Ma) time intervals using a script written for the ArcGIS program. Regions where the kinematics are known constrain adjacent areas where the kinematics are not well defined. The process of sequential restoration highlighted misalignments, overlaps, or large gaps in each incremental step, particularly in the areas between data transects, which remain problematic. Hence, the value of the reconstructions lies primarily in highlighting questions that might not otherwise be recognized, and thus they should be viewed more as a tool for investigation than as a final product.

The new sequential reconstructions show that compatible slip along the entire north-south extent of the inland right-lateral shear zone from the Gulf of California to the northern Walker Lane is supported by available data and that the east limit of active shear has migrated westward with respect to North America since ca. 10 Ma. The reconstructions also highlight new problems regarding strain-compatible extension east and west of the Sierra Nevada– Great Valley block and strain-compatible deformation between southern Arizona and the Mexican Basin and Range. Our results show ∼235 km of extension oriented ∼N78°W in both the northern (50% extension) and central (200% extension) parts of the Basin and Range. Following the initiation of east-west to southwest-northeast extension at 15–25 Ma (depending on longitude), a significant portion of right-lateral shear associated with the growing Pacific– North America transform jumped into the continent at 10–12 Ma, totaling ∼100 km oriented N25°W, for an average of ∼1 cm/ yr since that time.

The large-scale horizontal velocity field at Earth's surface is one of the main predictions of physical models of lithospheric deformation (e.g., England and McKenzie, 1982). Two-dimensional, cross-sectional models of finite deformation of mountain belts incorporating strong heterogeneity in rheologic parameters have been developed over the last decade (e.g., Lavier and Buck, 2002; Braun and Pauselli, 2004). Owing to advances in computation, fully three-dimensional models of plate boundary deformation zones, incorporating both horizontal and vertical variations in lithospheric rheology, will soon become common. Thus, a key observational frontier will be the determination of precise displacement vector fields of continental deformation in order to test these models. The most dramatic recent improvement in obtaining such velocity fields has been the advent of space-based tectonic geodesy (especially using continuous global positioning systems [GPS]), which is yielding velocity fields that are unprecedented in terms of both the scale of observation and the accuracy of the velocities. These data have already been used as tests for physical models in southwestern North America (e.g., Bennett et al., 1999, 2003; Flesch et al., 2000) and elsewhere (e.g., Holt et al., 2000). Substantial progress has also occurred over the last decade in determining longer-term velocity fields using the methods of plate tectonics and regional structural geology.

These longer-term displacement histories are essential for addressing the question of how the lithosphere responds to major variations in plate geometry and kinematics (e.g., Houseman and England, 1986; England and Houseman, 1986; Bird, 1998) because such variations occur on the million-year time scale. Plate tectonics is a precise method for constraining the overall horizontal kinematics of plate boundaries, using seafloor topographic and magnetic data in concert with the geomagnetic time scale. For the diffuse deformation that characterizes the continental lithosphere along plate boundaries, however, tectonic reconstruction at scales in the 100 km to 1000 km range is not as straightforward. It is based primarily on structural geology and paleomagnetic studies and requires the identification of large-scale strain markers and consideration of plate tectonic constraints (e.g., Wernicke et al., 1988; Snow and Wernicke, 2000; McQuarrie et al., 2003). Regional strain markers within the continents may not exist in any given region, and even if they do, they may not be amenable to accurate reconstruction at large scales.

In southwestern North America, a zone of plate boundary deformation on the order of 1000 km wide is developed along the plate boundary. In mid-Tertiary time (36 Ma), this boundary was strongly convergent, with the Farallon plate subducting eastward beneath the North American plate. Beginning at ca. 30 Ma, the Pacific-Farallon ridge came in contact with the North American plate. Since then, the Pacific–North America boundary has grown through the migration of triple junctions along the coast. Now, the entire margin from southern Baja California to Cape Mendocino is a transform Pacific–North America boundary, rather than a convergent Farallon–North America boundary (Atwater, 1970). This change in the configuration of the plate boundary is both relatively simple and profound, making southwestern North America an ideal laboratory for investigating how continental lithosphere responds to changes in relative plate motion.

Refined plate tectonic reconstructions have provided an improved kinematic model of the change from convergent to transform motion and have shown that there were significant variations in the obliquity of the transform after it developed. In particular, during the interval ca. 16 to ca. 8 Ma, Pacific–North America motion was highly oblique and included a margin-normal extensional component of as much as 2 cm/yr, coeval with a rapid pulse of Miocene extension that formed the Basin and Range province (Atwater and Stock, 1998; Wernicke and Snow, 1998). At ca. 8 Ma, Pacific–North America motion changed to more purely coastwise motion, which appears to have changed the intraplate tectonic regime from profound extension to a more complex mixture of extension, shortening, and transform motion, responsible for the opening of the Gulf of California, thrust faulting of the western Transverse Ranges, and development of the San Andreas fault–eastern California shear zone–Walker Lane, respectively.

Over the last several years, high-quality, large-scale kinematic constraints, many of which resulted from decades of field work and attending debate, have become available, reaching the point where synthesis into a large-scale velocity field is feasible. A rudimentary kinematic model using many of the constraints along the plate boundary and in the plate interior was incorporated into a publicly available animation of the post–38 Ma evolution of the entire Pacific–Farallon–North America system (Atwater and Stock, 1998; animation available at http://emvc.geol.ucsb.edu/download/nepac.php).

In this paper, we synthesize the current state of information on the kinematics of the diffusely deforming North American plate since 36 Ma, based on offsets of regional structural markers, and construct a strain-compatible kinematic model of the horizontal motions at 2 m.y. (0–18 Ma) and 6 m.y. (18–36 Ma) intervals, presented as a continuous animation. The model is by no means a final product, as new kinematic information and testing will require significant modifications of the model. Rather, the model is an attempt to be quantitatively rigorous in a way that will be useful for comparison with large-scale, three-dimensional physical models and for the identification of issues regarding the structural kinematics that might not otherwise be detected. Thus, in addition to the animation, we have constructed “instantaneous” velocity fields based on 2 m.y. averages from 0 Ma to 18 Ma and 6 m.y. averages from 18 Ma to 36 Ma. These results are our best attempt at “paleogeodesy,” presenting the geology-based kinematic model in a format similar to modern GPS velocity fields, which in turn may be quantitatively compared to physically based model velocity fields.

By combining regional structural constraints into a single model, the self-consistency of the model (i.e., its strain compatibility through time) provides powerful additional constraints on the kinematics in at least three ways. The first and most important is the fact that high-quality local kinematic information imposes severe constraints on its surroundings where information may not be available. As a hypothetical example, consider a large region of oblique extension between two undeformed blocks (Fig. 1). The strain and strain path need not be known for each geological element in the deforming region in order to constrain the large-scale kinematics. If the sum of fault displacements across just a single reconstruction path (p) is known, restoring point A to a position at point B, and it is known that the blocks have not rotated, then the single path imposes a strong constraint on the overall kinematics of all of the other paths between the blocks (Fig. 1A).

The second additional constraint is on errors in reconstructions. In the example in Figure 1, let us suppose that the minimum value of all fault displacements along reconstruction path p restores the block to point B, but there is no constraint on the maximum value along the path itself. The side of the block containing A would overlap the block on the other side of the rift if the displacement along the path were in excess of AC (Fig. 1B), violating the condition of strain compatibility. Therefore, the displacement is constrained to be between AB and AC, rather than some value greater than AB.

A third and perhaps most useful additional constraint arises when local constraints contradict one another. For example, if reconstruction along path q (Fig. 1A) required that point A restore to a position D, which is well within the other block, then the violation of strain compatibility forces reevaluation of the geological constraints. The geological reconstruction for displacement along q, the paleomagnetic constraints on the blocks, and the presumed rigidity of the blocks cannot all be correct. Thus, the exercise of regional reconstruction focuses attention on information that is most critical for improving the accuracy of the reconstruction. For southwestern North America, there is now enough high-quality local kinematic information that large-scale self-consistency of the model imposes useful additional constraints in all of these ways.

In making the reconstruction, the methods used in the local study of Wernicke et al. (1988) and Snow and Wernicke (2000) in the Death Valley region of the central Basin and Range province were applied at large scale. In Snow and Wernicke (2000), each step in the reconstruction showed the paleoposition of existing mountain ranges. Although the reconstruction allowed for the ranges to change shape as extension is restored (i.e., the ranges may decrease in area), in our reconstruction, the mountain ranges are shown as digitized polygons that approximate: (1) the modern bedrock-alluvium contact (e.g., a typical range in the Basin and Range), (2) faults bounding individual crustal blocks (e.g., the Santa Ynez Mountains block in the western Transverse Ranges), or (3) the physiographic boundaries of large, intact crustal blocks (e.g., the Colorado Plateau). In some cases, especially where large extensional strains are involved, the reconstruction overlaps individual polygons to account for extension, essentially using the modern bedrock-alluvium contact as a geographical reference marker. Because the strain is extensional, and in the case of metamorphic core complexes, one range has literally moved off of the top of another, these overlaps do not violate strain compatibility.

The individual positions of polygons were restored in each 2 m.y. time frame through an ArcGIS script that reads and updates a table listing the kinematic data for each range. The script, created by Melissa Brenneman of the Redlands Institute at the University of Redlands, is written in Visual Basic and is incorporated as a tool in a custom ArcMap document. The script reads a dBASE 4 table that contains the movement parameters (direction, distance rotation, and time interval) for each range (Appendix 1).1 The movement parameters listed in the table include both the available data (Figs. 2–5), as well as the motion required for strain compatibility. For the regions where kinematic data are not available, the kinematics could be defined by inserting data from proximal areas, or individual ranges could be moved by hand with the motion updated and recorded in the dBASE table using the ArcGIS script. The ArcGIS format and accompanying script allows for exact displacements to be incorporated into the model, as well as the individual adjustment of ranges to ensure strain compatibility. The GIS script records the geographical position of the centroid of each range at each 2 m.y. or 6 m.y. epoch. This allows for the data to be displayed in a variety of ways, including palinspastic maps for each 2 m.y. or 6 m.y. epoch, instantaneous velocity vectors at each 2 m.y. or 6 m.y. epoch, “paths” that individual ranges take over the 36 m.y. span of the reconstruction, or an animation that shows the integrated motion over 36 m.y. Instantaneous geology velocity fields are obtained from connecting the centroids of specific ranges at one time with the centroid of the same range in a later time.

The primary tectonic elements in the reconstruction are large crustal blocks comprising flat-lying pre–36 Ma strata, or geologic elements that are otherwise little deformed, and the straining areas in between them. The large unstrained blocks include the Great Plains– Rocky Mountains region (nominal North America reference frame), the Sierra Madre Occidental, the Colorado Plateau, the Sierra Nevada–Great Valley block, and Peninsular Ranges block (Fig. 2). The strained areas around them include the Rio Grande rift and Basin and Range province, the Gulf of California, the Transverse Ranges, the Coast Ranges, and the Continental Borderlands province offshore of southern California and Baja California.

The constraints used in the reconstruction are organized into six major categories (01). The first covers a range-by-range reconstruction path across the northern Basin and Range near latitude 40°N (Fig. 3 and 01). The second includes a similar reconstruction path across the central Basin and Range near latitude 37°N (Fig. 4 and 02). These two reconstructions collectively constrain the motion of the Sierra Nevada–Great Valley block. The third includes constraints from the southern Basin and Range, mainly the mid-Tertiary metamorphic core complexes of the Colorado River corridor and southern Arizona, west of the Sierra Madre Occidental, and extension across the Rio Grande rift north and east of the Sierra Madre Occidental (03). The fourth includes the complex Oligocene to recent strike-slip and extensional displacements of the Mojave region, which connect the Sierran displacement to regions farther south (Fig. 5 and 04). The fifth includes paleomagnetic and geologic constraints on vertical axis rotations of large crustal blocks, including the Sierra Nevada and Colorado Plateau, as well as small, individual ranges within the central Basin and Range and Mojave regions (05). Lastly, constraints on the large displacements along the San Andreas fault–Gulf of California shear system, and strains and rotations within the Continental Borderlands, including the large clockwise rotation of the Santa Ynez Mountains block, are included in Figure 6 and 06.

Northern Basin and Range

The extensional kinematics of the northern Basin and Range are dominated by two large-offset normal fault systems, the Snake Range detachment system (78 km of total offset) and the Sevier Desert detachment (40 km of total offset). The Snake Range detachment system affects the Egan, Schell Creek, and Snake Ranges (Fig. 2, ranges 6–8). Although the coupling of this system of faults to deep crustal extension has been debated (e.g., Gans and Miller, 1983; Miller et al., 1983; Bartley and Wernicke, 1984; Miller at al., 1999; Lewis et al., 1999), a magnitude of upper crustal extension of 78 km ± 10 km, as determined through mapped and restored stratigraphic markers, is not controversial (Gans and Miller, 1983; Bartley and Wernicke, 1984). More controversial is the geometry of the extensional faults in the Sevier Desert basin (between ranges 3 and 4, Fig. 2), including the very existence of the Sevier Desert detachment, which is known only from interpretations of seismic reflection profiles and well data (Anders and Christie-Blick, 1994; Wills et al., 2005). The 40-km offset along the Sevier Desert detachment used in this paper is based on restoring Sevier fold-thrust belt structures that are offset by the detachment, and high-angle normal faults in the hanging wall imaged in the Consortium for Continental Reflection Profiling (COCORP) and industry seismic reflection lines (Allmendinger et al., 1986; Allmendinger et al., 1995; Coogan and DeCelles, 1996). An opposing view to the large-offset kinematics of a shallow detachment suggests that the imaged reflection surface is a composite of aligned features that includes basin-bounding high-angle normal faults, a subhorizontal thrust fault, and an evaporite horizon (Anders and Christie-Blick, 1994). According to this interpretation, extension across ranges within and around the Sevier Desert basin could be as little as 10 km (versus 40 km), which would subtract ∼15% from our overall estimate of extension along the transect.

To the west of the Egan Range area, the remainder of the northern Basin and Range deformation is partitioned into extensional and right-lateral strike-slip offsets, both of which accommodate translation of the Sierra–Great Valley block away from the interior of North America. The extension (94 km) is accommodated by several systems of steeply tilted normal fault blocks in the western Basin and Range, with individual fault systems accommodating up to 16 km of extension (Fig. 2, ranges 13, 19, and 20) (Surpless, 1999; Dilles and Gans, 1995; Smith, 1992), and a number of high-angle, presumably modest-offset normal faults that define the Basin and Range physiography across the central part of the reconstruction path, which we assume have 3– 4 km of horizontal offset each.

Right-lateral shear is accommodated predominantly through northwest-trending faults concentrated near the western edge of the northern Basin and Range in the northern Walker Lane Belt (Fig. 2, range 18). Right-lateral offset on a series of faults, which individually have 5–15 km of offset, totals 20– 56 km (Faulds et al., 2005; Hardyman et al., 1984). Because the faults strike more westerly than the North American margin, their motion accommodates a component of westward motion of the plate boundary.

Timing of extension in the northern Basin and Range is constrained by a large body of work on the ages of faulted Cenozoic volcanic and sedimentary units and cooling ages of uplifted footwall blocks. For example, the early “core complex”–related extension (ca. 35–25 Ma) is seen in coeval faulting and volcanism at 35 Ma in the Egan Range (Gans and Miller, 1983) and 40Ar/39Ar cooling ages indicative of rapid cooling from 30 to 25 Ma in the western portion of the northern Snake Range and from 20 to 15 Ma in the eastern portion of the range (Lee, 1995). Apatite fission track (AFT) cooling ages from the northern Snake Range indicate 10–13 km of fault slip from 18 to 14 Ma. Initiation of later “Basin and Range” extension is seen predominantly in the fission-track and helium cooling ages of apatite and zircon. The cooling ages across the width of the extending zone cluster ca. 15 Ma (Stockli, 1999), with 18 Ma ages in the footwall of the Snake Range detachment (Miller et al., 1999) (Fig. 2, range 6) and Sevier Desert detachment (Stockli et al., 2001) (Canyon Range, Figure 2, range 3).

Central Basin and Range

The central Basin and Range province is in many respects an ideal location for a province-wide restoration of Basin and Range extension (Snow and Wernicke, 2000, and references therein). A regionally conformable miogeocline, Mesozoic thrust structures and distinctive Tertiary sedimentary deposits tightly limit the extensional history of both the Lake Mead (Fig. 2, ranges 22–25) and the Death Valley (Fig. 2, ranges 27–34) extensional systems (Wernicke et al., 1988; Wernicke, 1992). Motion of the Sierra Nevada with respect to the Colorado Plateau in this region is primarily constrained by displacements of two distinctive Miocene basins developed early in the history of the extension of each system (Fig. 4).

In the Lake Mead system, restoring numerous proximal landslide breccias at Frenchman Mountain (Fig. 2, range 24) to their source areas in the Gold Butte block (Fig. 2, range 23) also restores piercing lines defined by the southward truncation of Triassic formational boundaries by the basal Tertiary unconformity in both areas. The correlation of these features in the Frenchman Mountain and Gold Butte areas suggests 65 km ± 15 km of extension between the two blocks (Fig. 4).

In the Death Valley system, Wernicke et al. (1988) initially proposed that the Panamint thrust at Tucki Mountain (Panamint Range, Figure 2, range 28) is correlative with the Chicago Pass thrust in the Nopah–Resting Springs Range (Fig. 2, range 27) and the Wheeler Pass thrust in the Spring Mountains (Fig. 2, range 25), suggesting a total of 125 km ± 7 km of post-Cretaceous, west-northwestern extension has separated them (02). This offset is strengthened by correlations of additional contractile structures exposed across the Death Valley extensional system (Snow and Wernicke, 1989; Snow, 1992; Snow and Wernicke, 2000) and distinctive middle Miocene sedimentary deposits that occur along the extensional path (Niemi et al., 2001). These include proximal conglomeratic strata of the Eagle Mountain Formation, which were derived from the northeastern margin of the Hunter Mountain batholith in the southern Cottonwood Mountains (Fig. 2, range 32). Recognition and correlation of this dismembered early extensional basin, in conjunction with stratigraphic constraints from other Tertiary deposits in the region, indicates that its fragmentation occurred mainly from 12 Ma to 2 Ma (Fig. 4). The correlation of these deposits yields a displacement vector of 104 km ± 7 km oriented N67°W between the Nopah–Resting Springs Range (Fig. 2, range 27) and the Cottonwood Mountains (Fig. 2, range 32).

To the ∼170 km of displacement from these constraints, we add four additional estimates to complete the reconstruction path. In the Lake Mead system, 15 km of extension between the Gold Butte area and the Colorado Plateau (Fig. 2, range 23) (Brady et al., 2000) and a maximum of 8 km of extension between the Spring Mountains (Fig. 2, range 25) and Frenchman Mountain (Fig. 2, range 24) (Wernicke et al., 1988) increases the total displacement of the Spring Mountains relative to the Colorado Plateau to ∼88 km. In the Death Valley system, an addition of 9 km of displacement in both the Panamint and Owens Valleys increases the total displacement to ∼147 km between the Spring Mountains and the Sierra Nevada.

The sum of all displacements along the path is therefore 235 km ± 20 km (02), which represents a combination of areal dilation (crustal thinning) and plane strain (strike-slip faulting). Approximately 80% of the elongation is accommodated by vertical thinning and ∼20% by north-south contraction (Wernicke et al., 1988; Snow and Wernicke, 2000). In addition to this path, there are a number of more local offsets that were used to position polygons to the north and south, which are shown in Figure 4 and summarized in 02.

Southern Basin and Range–Rio Grande Rift

Extension in the southern Basin and Range is almost completely dominated by the formation of large-offset normal faults that form the metamorphic core complexes (Coney, 1980; Spencer and Reynolds, 1989; Dickinson, 2002). The core complexes ring the southwestern margin of the Colorado Plateau (Fig. 2, ranges 37–44), and estimates of the total extension they represent are remarkably systematic in magnitude, direction, and rate (03). The timing of extension varies in age from 28 Ma to 14 Ma as the extension migrates from southeast to northwest. The migration of extension has been related to a similar migration in volcanism. Both extension and volcanism have been proposed to be a result of the northwestward foundering of the Farallon plate (e.g., Humphreys, 1995; Dickinson, 2002).

In a similar time frame (ca. 26 Ma), volcaniclastic sediments deposited east of the Colorado Plateau in the Rio Grand Rift (Fig. 2, location 45) have been interpreted as representing the onset of extension (Chapin and Cather, 1994). Ingersoll (2001) counters that the early sediments are broad volcaniclastic aprons that show no evidence of syndepositional faulting. He places the initiation of rifting slightly later (ca. 21 Ma). Based on initiation of half-graben sedimentation and stratal tilting, rapid extension occurred between 17 and 10 Ma (Ingersoll, 2001; Chapin and Cather, 1994). The total magnitude of extension is small and ranges from 6 km in the northern part of the rift to 17 km in the south, consistent with the 1.5° clockwise rotation of the Colorado Plateau (Chapin and Cather, 1994; Russell and Snelson, 1994).

Extension within the Rio Grande rift is contiguous with the broad extended region farther south, east of the Sierra Madre Occidental (in Chihuahua), the magnitude of which is poorly understood (Dickinson, 2002). Generally, extension in the Mexican Basin and Range is partitioned both in time and space. Early core complex extension is documented in northwestern Mexico (in Sonora), just west of the Sierra Madre Occidental (Nourse et al., 1994; Gans, 1997) (Fig. 2). Palinspastic reconstructions over small regions in Sonora suggest cumulative extension of 90%, mostly between 26 and 20 Ma, and more modest extension (10%–15%) between 20 and 17 Ma (Gans, 1997). Limited crustal extension is also documented east of the Sierra Madre Occidental during the same time period (Dickinson, 2002). Major extension occurred in both Chihuahua (Henry and Aranda-Gomez, 2000; Dickinson, 2002) and Sonora Mexico (e.g., Stock and Hodges, 1989; Henry, 1989; Lee et al., 1996) from ca. 12 Ma to 6 Ma as a prelude to the opening of the Gulf of California at 6 Ma (Oskin et al., 2001). During the 12 Ma to 6 Ma interval, very small magnitude east-west “Basin and Range” extension affected Arizona (Spencer and Reynolds, 1989; Spencer et al., 1995).

Mojave Region

Cenozoic deformation of the Mojave region occurred in two main stages. Deformation began in the late Oligocene–early Miocene with the formation of large-offset normal faults and associated core complexes (Glazner et al., 1989; Dokka, 1989; Walker et al., 1990). Extension in the Mojave region (Fig. 2, location 41, and Fig. 5) may be linked to core complex extension in the southern Basin and Range corridor (Fig. 2, ranges 38 and 44) through a diffuse transfer zone that involves both rotation and strike-slip faulting (Bartley and Glazner, 1991; Martin et al., 1993). The magnitude of extension is determined through alignment of pre-extensional markers that include facies trends in Paleozoic strata, a unique gabbro-granite complex, and late Jurassic dikes, indicating a total of 40–70 km of offset (Glazner et al., 1989; Walker et al., 1990; Martin et al., 1993). Extension began in synchronism with the eruption and emplacement of 24–23 Ma igneous rocks (Walker et al., 1995) and is capped by the flat-lying, 18.5 Ma Peach Springs Tuff (Glazner et al., 2002). The fraction of the Mojave Desert region that was affected by mid-Tertiary extension is controversial (e.g., Dokka, 1989; Glazner et al., 2002). Glazner et al. (2002) propose that only a small region north of Barstow (Fig. 2, location 41) was affected by the early extension, with the southern boundary of this extensional domain linked to core complex extension to the southeast through diffuse right-lateral shear. The northern boundary of the extensional domain is more problematic; however, regional kinematic compatibility requires a northern transfer zone that links Mojave extension to similar age extension to the north or west. Rotation of the Tehachapi Mountains and/or extension in the southern San Joaquin Valley may represent the northern portion of this system (McWilliams and Li, 1985; Plescia and Calderone, 1986; Tennyson, 1989; Goodman and Malin, 1992; Walker et al., 1995; Glazner et al., 2002).

Following this early phase of extensional deformation, a system of right- and left-lateral strike-slip faults similar to those active today was established, with right-lateral shear along a series of northwest-striking faults predominant (Fig. 5). The total accumulated shear across the Mojave, as documented by field studies, is 53 km ± 6 km (04). The timing of right-lateral shear is not well constrained. Motion on the faults is inferred to be post–10 Ma based on strain compatibility with deformation directly north and south (02 and 05).

Vertical Axis Rotations East of the San Andreas Fault

There are two zones of vertical-axis rotation east of the San Andreas fault: the Eastern Transverse Ranges located immediately south of the Mojave block and the northeastern Mojave rotational block (Carter et al., 1987; Schermer et al., 1996; Dickinson, 1996) (Fig. 5 and 04).

The Eastern Transverse Ranges include a series of structural panels separated by east-west–oriented, left-lateral faults (Dickinson, 1996). Paleomagnetic studies show that 10 ± 2 Ma rocks within this zone record the entire 45° rotation (Carter et al., 1987), while 4.5 Ma volcanic rocks are unrotated (Richard, 1993). These constraints imply that all of the rotation and most of the right-lateral strike-slip motion in the Mojave region immediately to the north are ca. 10 Ma and younger.

The northeastern corner of the Mojave region is another area of pronounced clockwise rotation. Schermer et al. (1996) proposed that the northeastern Mojave underwent 23° of rotation accompanied by 5 km of left-lateral slip on faults within the rotating region and 15° of “rigid body” rotation. Total right-lateral shear predicted by this model is 33 km.

San Andreas System and Areas to the West

Deformation west of the San Andreas fault is defined by four first-order constraints (Fig. 6 and 06). The first is motion on the San Andreas fault itself, which is tightly constrained in central California at 315 km ± 10 km by restoring the Pinnacles volcanics west of the fault to the Neenach volcanics to the east of it (Matthews, 1976; Graham et al., 1989; Dickinson, 1996). The offset volcanics were extruded from 22 Ma to 24 Ma, but tentatively correlative late Miocene strata (7–8 Ma) are apparently offset 255 km (Graham et al., 1989; Dickinson, 1996).

The second constraint is the ∼110° clockwise rotation of major fault-bounded blocks in the western Transverse Ranges (Hornafius et al., 1986; Luyendyk, 1991). Because of the length and structural integrity of these blocks (in particular, the Santa Ynez Mountains), this rotation requires a coast-parallel displacement of ∼270 km (Hornafius et al., 1986).

The shear and rotation of these blocks are confirmed by the third major constraint, reconstruction of now-scattered outcrops of the distinctive Eocene Poway Group. Exposures along the Channel Islands were rifted away from counterparts in southernmost California, which are in turn offset from their source area in northern Sonora, Mexico, by the southern San Andreas fault system (Abbott and Smith, 1989). Rifting and rotation of the western Transverse Ranges away from the Peninsular Ranges formed the strongly attenuated crust of the Continental Borderlands on their trailing edge. The magnitude of this extension is proposed to be ∼250 km based on seismic reflection data delineating the geometry of extensional fault systems and correlation of “mega key beds” or lithotectonic belts (forearc basin sediment, Franciscan subduction complex) (Crouch and Suppe, 1993; Bohannon and Geist, 1998).

The final first-order constraint is the opening of the Gulf of California. Although offset of the Poway Group suggests roughly 250 km of displacement, recognition of correlative pyroclastic flows on Isla Tiburon and near Puertecitos on the Baja Peninsula dated at 12.6 Ma and 6.3 Ma constrains the full transfer of Baja California to the Pacific plate to have occurred no earlier than ca. 6 Ma, with 255 km ± 10 km of displacement along the plate boundary since then (Oskin et al., 2001). Including additional deformation of the adjacent continental margins increases the magnitude of displacement to as much as 276 km ± 10 km (Oskin and Stock, 2003).

The western North America animation (1) combines 13 individual paleogeographic maps (Figs. 7–9) (Appendix 1, paleogeographic maps [see 1]) generated by ArcGIS into a digital animation illustrating a model of how extension and right lateral shear evolved in the region. The color scheme for the animation includes yellow, orange, and red polygons on a white background. The polygon shape reflects the modern bedrock-alluvium contact, fault-bounded crustal blocks or the physiographic boundaries of large, intact crustal blocks. Yellow polygons indicate areas where there are no data for how the region is deforming. Orange polygons (ranges) are ranges whose motion is directly constrained by kinematic data. These polygons turn red during the time period of motion (i.e., Baja California turns red from 6 to 0 Ma as it separates from North America and moves northward on the Pacific Plate). A notable exception to this is the Colorado Plateau–Rio Grand rift area. Neither the Colorado Plateau, nor the Rio Grande rift turn red during rotation and extension even though there are data that describe this deformation (05). The space created (additional white space between the colored polygons) as the movie progresses in time indicate areas of extension. The removal of white space (as polygons move closer together) indicates areas of compression. The thick blue line on the left of the animation represents successive locations of the eastern edge of Pacific plate oceanic lithosphere relative to stable North America at the time period annotated on the upper left edge of the line (from Atwater and Stock, 1998). The position of this line constrains the maximum westward extent of continental North America at the time indicated because it shows the minimum east limit of extant oceanic crust. Details of the reconstruction can be seen by moving the slider bar on the animation. To move back and forth over a narrow window of time, just hold the mouse key down over the triangle on the slider bar and move it back and forth over the time window of interest.

The exercise of developing a self-consistent, strain-compatible model has raised a number of issues that are difficult to resolve satisfactorily in the reconstruction and require further investigation. The most apparent (among many!) are (1) the need for middle to late Miocene right-lateral shear in the eastern Mojave region to make room for the northerly motion of the Sierra Nevada determined from the central and northern Basin and Range reconstruction paths; (2) the need for large amounts of relatively young extension in northern Mexico both east and west of the Sierra Madre Occidental to reconcile core complex extension in Arizona and the late Miocene– Pliocene opening of the Gulf of California; (3) the apparent rotational history of the Sierra Nevada–Great Valley block; and (4) generally large amounts of Miocene-Pliocene shortening and extension in the Transverse Ranges, Coast Ranges, and Borderlands provinces, which arise from the need to reconcile San Andreas offset with the position of oceanic crust offshore, differences in the age of extension north and south of the Garlock fault, and large clockwise rotation of the Santa Ynez Mountains block (1).

Eastern Mojave Region

The eastern California shear zone–Walker Lane belt is an ∼120-km-wide zone of right-lateral, intraplate shear east of the Sierra Nevada and San Andreas fault. Geodetically this shear zone accommodates up to 25% of the Pacific–North America relative plate motion (Bennett et al., 2003; McClusky et al., 2001; Miller et al., 2001; Sauber et al., 1994). Geologic estimates of displacements vary along the north-south extent of the eastern California shear zone. Proposed net displacement along the eastern California shear zone (oriented ∼N20°W) varies from 65 km in the Mojave region (Dokka and Travis, 1990) to 133 km in the central Basin and Range (Snow and Wernicke, 2000; Wernicke et al., 1988). In the northern Walker Lane region, shear estimates range from 20 km to 54 km (Faulds et al., 2005; Hardyman et al., 1984), plus an additional component of northwest-directed extension due to a change in extension direction in the northern Basin and Range from east-west in the east to northwest-southeast in the west.

One of the goals of this study was to develop a kinematically consistent model of the eastern California shear zone that fits within the errors provided by both local and regional studies. We found 100 km ± 10 km right-lateral shear oriented N25°W was compatible with data in both the northern and central Basin and Range (1). In the Mojave region of the eastern California shear zone, however, available data suggest no more than 53 km ± 6 km of right-lateral shear oriented N25°W, about half of what is required to the north. Kinematic compatibility with the magnitude of deformation north of the Garlock fault requires ∼100 km of right-lateral shear though the Mojave region, with the majority of additional shear located on the eastern edge of the shear zone during its early (12–6 Ma) history (Figs. 5,7,8, 1). The 27 km and 45 km of right-lateral offset along the Bristol Mountains–Granite Mountain and southern Death Valley fault zones is significantly greater than previous estimates (0–10 km and 20 km, respectively), but solid piercing points that limit the net offset are scarce and debatable (Howard and Miller, 1992; Dokka and Travis, 1990; Davis, 1977). The ∼30 km of displacement along the eastern edge of the Mojave must be transferred southward along the Sheep Hole fault to the Laguna Fault system of Richard (1993). The 36 km of model offset is significantly greater than the 2 km of right-lateral offset proposed by Richard (1993) (04, Fig. 5). Additional faults with significantly greater offsets than that documented by geology are the 11–13 km model offsets on the Camp Rock, Gravel Hills, and Harper Lake fault systems, where current estimates suggest no more than 3 km of offset on any of these faults (Dibblee, 1964; Oskin and Iriondo, 2004; M. Strane, 2005, personal commun.). The difference between the model and data requires that the slip discrepancy must be taken up on other faults (most likely to the east) in the Mojave shear system. Although the details concerning both timing and distribution of shear within the eastern California shear zone will continue to evolve with time, the strength of the central Basin and Range offsets combined with kinematic compatibility constraints require reevaluation of geologic evidence for total magnitude of right-lateral shear through the Mojave. Therefore, we have modeled many of the faults in the Mojave with greater net offset than suggested by offset markers. From the model slip amounts shown on Figure 5, we obtain 100 km of right-lateral shear oriented N25°W since 12 Ma, at a long-term rate of 8.3 mm/ yr ± 1 mm/yr. We suggest that the discrepancy may be due to penetrative shear in the largely granitic crust between the strike-slip faults (e.g., Miller and Yount, 2002).

Increasing right-lateral shear in the eastern Mojave has implications on how shear is distributed along the entire plate boundary. Since much of the additional shear predates the opening of the Gulf of California (Oskin et al., 2001), it implies ∼50 km of dextral shear between 6 and 12 Ma in the Sonora region. The total magnitude of separation of the Baja Peninsula from Sonora predicted from the model is 350 km, 300 km of which is post 6 Ma. This is slightly greater than the 296 ± 20 km, 276 ± 10 km of which is post 6 Ma measured by Oskin and Stock (2003), but still significantly less than the 450 km total continental separation proposed by Fletcher (2003).

Dokka and Travis (1990) proposed that the eastern California shear zone accommodated 9%–14% of total predicted relative motion between plates if shear initiated at 10 Ma. The model of eastern California shear zone deformation that we propose here (1) suggests that eastern California shear zone deformation is ∼28% of San Andreas motion averaged since 12 Ma and 15% of total plate motion since 16 Ma.

Arizona–Mexican Basin and Range

The geographic region that has the fewest local kinematic constraints is Sonora–Chihuahua Mexico. However, the kinematics of Baja California, based on plate tectonic reconstructions (Atwater and Stock, 1998), is an especially powerful constraint on intraplate deformation in this region. The constraint arises from the simple fact that oceanic and continental lithosphere cannot occupy the same surface area at the same time (Atwater and Stock, 1998) (Fig. 6B). The plate tectonic constraint suggests ∼330 km ± 50 km of extension between 6 Ma and 24 Ma, because after restoring the offset across the Gulf of California (Oskin et al., 2001; Oskin and Stock, 2003), this is the total overlap between continent and ocean. In concert with strong northeast-southwest extension in Arizona, we suggest similar magnitudes of extension (44 km and 86 km) occurred from 16 Ma to 24 Ma and was oriented N50°E–N60°E (making room for the brown and green curves in Fig. 6B). We show another pulse from 12 Ma to 8 Ma oriented N65°W–N78°W, reflecting the growing influence of the Pacific plate's northerly motion on intraplate deformation, as appears to be the case to the north (1).

Restoring 330 km of extension, however, particularly the northwesterly extension in the window of time from 16 Ma to 8 Ma, opens up a large northeast-trending gap in southern Arizona and northern Sonora. This gap is a result of differences in both magnitude and timing of extension between southern Arizona and northern Sonora and suggests (incorrectly) that there is ∼60 km of NW-SE compression between 16 and 8 Ma (Figs. 8–9). Large magnitude core complex extension in southern Arizona initiates at ca. 28 Ma and wanes from 16 Ma to 14 Ma (03). Significant extension in Sonora occurs over a similar time range (Nourse et al., 1994; Gans, 1997). At ca. 12 Ma, however, significant extension is recorded in both the Gulf extensional province west of the Sierra Madre Occidental (e.g., Stock and Hodges, 1989; Henry, 1989; Lee et al., 1996, Gans et al., 2003) and east of the Sierra Madre Occidental (Henry and Aranda-Gomez, 2000), while only minimal magnitudes of east-west extension are recorded in southern Arizona. This problem is similar to that arising from the difference in timing of extension north of the Colorado River extensional corridor between the Mojave Desert and central Basin and Range. Here, the Garlock fault accommodates different amounts of extension, not only from 10 Ma to the present (Davis and Burchfiel, 1973), but potentially throughout the history of extension in the region (24–0 Ma). Although the difference in timing and magnitude of extension between the Mojave region and the southern Arizona Basin and Range versus the Mexican Basin and Range in Sonora and Chihuahua is generally recognized (e.g., Henry and Aranda-Gomez, 2000; Dickinson, 2002), the geometry and genetic relationship of the transfer system that must separate them is problematic.

In the model presented here, the amount of extension in the Mexican Basin and Range is partitioned between the extending regions east (∼134 km) and west (∼180 km) of the unstrained Sierra Madre Occidental block. Although both regions display numerous extensional structures, the exact magnitude of extension is unknown. Because of the difference in post–16 Ma extension in Chihuahua and the Rio Grande rift (90 and 20 km, respectively) after ca. 16 Ma, the model includes a zone of right-lateral shear that extends through southeastern Arizona between the two provinces (1). The existence of this shear zone is unlikely, leaving two possible solutions. The first is that extension systematically increases from the Rio Grande rift to Chihuahua Mexico due to clockwise rotation of the Sierra Madre Occidental (rotation would need to be greater than the 1.5° rotational opening of the Rio Grande rift). Another solution would be partitioning a much greater magnitude of extension in the Gulf extensional province (∼270–300 km), but this is thus far not supported by mapping in the region (Henry and Aranda-Gomez, 2000). Most likely some combination of these factors is necessary to match the first-order geologic constraints of the region.

Sierra Nevada–Great Valley Block Rotation

The reconstruction presented here shows significant extension in the northern Basin and Range between 36 Ma and 24 Ma, with essentially no extension occurring over this time period in the central Basin and Range. To accommodate this difference, the Sierra Nevada– Great Valley Block must rotate or deform internally. We propose that the block behaves fairly rigidly and rotates counterclockwise (1). After initiation of extension in the central Basin and Range at ∼16 Ma, the Sierra Nevada–Great Valley Block rotates clockwise for a final net rotation of 2° (05, Appendix 1, Movement Table [see footnote 1]). The animation shows the early 36–24 Ma rotation accommodated by ∼35 km of dextral shear along the proto–Garlock fault and accompanying compression in the southeastern Sierra Nevada region (1). The actual effects of this rigid body rotation on the deformation of surrounding regions (particularly to the north and south) are highly dependent on the axis of rotation and how rigidly the block behaved, both of which are unknown. The rotation of the Tehachapi Mountains may include this early counterclockwise rotation of the Sierras, as well as potentially being linked to southern Basin and Range core-complex formation, which immediately followed (McWilliams and Li, 1985; Plescia and Calderone, 1986; Walker et al., 1995; Glazner et al., 2002).

Areas West of the San Andreas Fault

Based on the timing and magnitude of displacement on a few fault systems (San Andreas, northern Gulf of California, Mojave, central Basin and Range, and the Santa Ynez Mountains), continental basins must open (creation of white spaces in the movie [1, Figs. 7–9] suggesting pulses of extension) and close (closing of spaces or overlap of polygons suggesting pulses of contraction) from 24 Ma to 0 Ma. Even at this large and relatively simplified scale, extension and contraction are spatially and temporally complex throughout the region west of the San Andreas fault, and we expect even greater complexities in timing and magnitude at a more detailed level. The following discussion highlights the magnitudes of displacement and summarizes data that either support or conflict with the model displacements.

Transverse Ranges

The clockwise rotation of the Western Transverse Ranges (Hornafius et al., 1986; Luyendyk, 1991) suggests regions of extension and subsequent compression both north and south of the rotating Santa Ynez Mountains block (Fig. 2, range 50) (1). The magnitude of predicted extension (Fig. 8) and contraction (Fig. 7) (oriented ∼north-south) is as great as 130 km to the north of the western side of the block from ca. 12 Ma to the present. Motion of Baja California northward from 6 Ma to the present suggests as much as 90 km of shortening in the southern Transverse Ranges (Santa Ynez and San Gabriel Mountains blocks) (Fig. 7, 1). Transpressive motion involving the San Gabriel Mountains, San Bernardino Mountains, and Mojave blocks implies ∼40 km of north-south shortening immediately north of the Peninsular Ranges block. Balanced cross sections through the San Emigdio, Santa Ynez, and San Gabriel Ranges indicate 53 km of shortening since 3 Ma (Namson and Davis, 1988a). Although the shortening estimate is strongly dependent on the details of how the Santa Ynez and Peninsular Ranges–Baja California blocks move, the reconstruction presented here suggests ∼60 km of north-south shortening at the longitude of the eastern Santa Ynez Mountains block since 6 Ma. As suggested by Namson and Davis (1988a), shortening of this magnitude in the upper mantle lithosphere is supported by a large volume of high-velocity material imaged tomographically beneath the region (e.g., Humphreys et al., 1984).

Coast Ranges

Differences in the timing of extension within the Mojave and Basin and Range north and south of the Garlock fault, in conjunction with plate tectonic constraints on the westernmost limit of the North America continental edge (Atwater and Stock, 1998), indicate a period of extension (20–16 Ma) and subsequent compression (14–0 Ma) to the west of the Sierra– Great Valley block (Figs. 7–9, 1). Approximately 80 km of core-complex extension south of the Garlock fault occurred prior to significant extension in the central Basin and Range. In order to maintain a quasilinear ocean-continent boundary, a zone of extension roughly equal in magnitude to the core-complex extension is required north of the Garlock fault and west of the Sierra Nevada–Great Valley block. This becomes most visible in the reconstruction at 16 Ma (Fig. 9A,902). As extension evolves in the central Basin and Range, this same zone undergoes contraction to maintain the quasilinear plate boundary suggested by the extant distribution of oceanic crust from 16 Ma to the present.

The Neogene tectonic and volcanic history from the Great Valley to the edge of the continent is broadly consistent with the model (data summarized in Tennyson, 1989). Although the model and geologic data are difficult to compare quantitatively because there are no obvious normal faults with measurable offsets, the magnitude of extension (and subsequent compression) is significantly less than that predicted by the model. Development of local nonmarine basins and eroded highs, followed by significant subsidence at ∼16–18 Ma and the development of the relatively deep marine Monterey basin strongly suggests an extensional event. Rotation of the Tehachapi Mountains and/or extension in the southern San Joaquin Valley (McWilliams and Li, 1985; Plescia and Calderone, 1986; Tennyson, 1989; Goodman and Malin, 1992) may be indicative of this extension but may represent far less than the ∼80 km predicted by the model.

The subsequent compression in the Coast Ranges is more quantifiable and appears to be significantly less than that suggested by the model. Estimates of compression in the Coast Ranges east and west of the San Andreas fault range from 20 km to 48 km (Page et al., 1998; Namson and Davis, 1990, 1988b), with all of the known shortening occurring post–10 Ma, and most of it post–4 Ma (Page et al., 1998; Namson and Davis, 1990). Therefore, the model predicts an additional 32–50 km of shortening prior to 10 Ma, for which there is (thus far) no evidence in the Coast Ranges.

Pausing 1 at 15 Ma highlights the crux of the problem (Fig. 9a). To eliminate the need of early 24–16 Ma extension in the Coast Ranges (and subsequent compression), the continental edge would need to bend eastward north of the Mojave and then continue north along the western edge of the Great Valley (1). This bend in the continental edge would create an ∼80-km-wide, ∼300-km-long section of oceanic crust that would have to be subducted south of the northward migrating triple junction during the period of central Basin and Range extension. The solution to the space problem that the model highlights may rest in a combination of several possibilities which include allowing for a warping of the North America coast line, finding greater magnitudes of deformation in the region of the Coast Ranges, and less extension in the central Basin and Range. However, to truly evaluate the magnitude of each of these options requires more detailed reconstruction of crustal blocks west of the San Andreas fault.

Uncertainties in the Reconstruction

Statistically rigorous uncertainties are notoriously difficult to quantify in geological reconstructions, largely because estimates of geologic offset do not have Gaussian or other standard probability distribution functions. The condition of strain compatibility or “no overlap” sets a hard limit on the displacement estimate but does not distinguish higher or lower probability of any given position within those limits. Hence, the variance of any given estimate cannot be rigorously quantified.

In map view, any given displacement estimate will have an irregularly shaped uncertainty region. Under the assumption of a uniform probability distribution within these uncertainty regions, Wernicke et al. (1988) used a Monte Carlo method to estimate the total uncertainty on the sum of displacement vectors for a path across the central Basin and Range. This method repeatedly summed randomly selected vectors from each uncertainty region to generate a probability distribution for the net offset. The contour that excluded the outermost 5% of the model runs was taken as an estimate of two standard deviations of the measurement. The estimate of total Sierran motion thus derived was 247 km ± 56 km, S 75° ± 12°E, and therefore a reasonable estimate of the standard deviation would be 28 km. For this same estimate, the square root of the sum of the squares for individual vectors (in the direction of displacement, using values from 01 and Figure 10,1002 in Wernicke et al., 1988) is only 15 km. This is perhaps not surprising because the Monte Carlo approach does not place greater weight on values near the center of the uncertainty polygon than on values at the edges.

Our revised displacement estimate for the central Basin and Range, 235 km ± 20 km (again the error is equal to the square root of the sum of the squares for individual vectors), is similar to that of Wernicke et al. (1988) if one considers the 20 km figure as a crude estimate of the standard deviation (1-sigma). However, given the results from Wernicke et al. (1988), the real error may scale upward by as much as a factor of two, depending on the degree to which our best estimate is more probable than values at the extremes. A simple sum of each uncertainty along a given path from 01 and 02 gives an error estimate of 47 km and 45 km, respectively. Thus, as a rule of thumb, the uncertainty in position of any given range or set of ranges at any given time is on the order of 20–40 km at one standard deviation.

Because the reconstruction involves temporal information (which is also uncertain), the problem of rigorously estimating errors becomes even more difficult and is clearly beyond the scope of this paper. Even though temporal information adds to the uncertainty of position at any given time, the self-consistency of the reconstruction mitigates these uncertainties to a substantial degree.

Tracking the restored positions of the ranges from the palinspastic maps, we have created “instantaneous” velocity fields based on 2 m.y. averages from 0 Ma to 18 Ma and 6 m.y. averages from 18 to 36 Ma. These paleogeodetic velocity fields depict how deformation has evolved in space and time across the plate boundary deformation zone (Figs. 10A–10G).

Figures 10F and 10G (30–18 Ma) illustrate the collapse of the Basin and Range away from the stable Colorado Plateau through the formation of metamorphic core complexes at a time of active ignimbrite volcanism and Pacific-Farallon convergence. Extension initiated first in the northern Basin and Range and then in the southern Basin and Range. This pulse of large-magnitude extension migrated south and north, respectively, until it converged in the central Basin and Range at ca. 16 Ma. Figure 10E (14–16 Ma) emphasizes the large extensional strains in the central Basin and Range especially with respect to the concurrent faulting to the north and south. The 14–16 Ma time slice also shows the impact of the evolving plate boundary on the North American continent as right-lateral shear is accommodated through the rotation of the Western Transverse Ranges and accompanying shear and extension. The 10–12 Ma time slice (Fig. 10D) illustrates the uniform (systematically increasing) strain in the northern Basin and Range and, in contrast, the westward-migrating extension in the central Basin and Range. Significant extension is also necessary in the Mexican Basin and Range due to plate-boundary constraints. It is during this time period that right-lateral shear migrates farther inboard into the continent through the development of the eastern California shear zone. South of the Garlock fault, the shear is oriented nearly parallel to the plate boundary (N25°W). North of the Garlock fault, the shear plus extension creates a more oblique orientation of shear (∼N67°W). From 6 Ma to 8 Ma, this same pattern of intracontinental right-lateral shear strengthens with shear partitioned differently south of the Garlock fault than in the central Basin and Range and northern Basin and Range portions of the eastern California shear zone (Fig. 10C). In the Mexican Basin and Range, deformation wanes and extension and right-lateral shear become concentrated in the proto–Gulf of California.

The differences in the velocity fields from the 2–4 Ma average to the 0–2 Ma average is most likely a function of limitations in the data, rather than a significant slowing in the rate of deformation over the last 2 m.y. (i.e., the Mojave region) (Figs. 10A and 10B). Within the model, the lack of timing constraints for right-lateral faults through the Mojave means that the rate of deformation there becomes a function of the rate of deformation to the north and south. North of the Garlock fault, large magnitudes (104 km) of oblique extension are focused predominately from 11 Ma to 3 Ma (Niemi et al., 2001; Snow and Wernicke, 2000; Snow and Lux, 1999). South of the Mojave, the timing of deformation is partially bracketed by the age of rotation of the Eastern Transverse Ranges (as mentioned earlier, ca. 10 Ma rocks record the entire 45° of rotation whereas ca. 4 Ma rocks indicate no rotation; Carter et al., 1987; Richard, 1993). These timing constraints suggest most of the deformational shear in the Mojave occurred between 10 Ma and 2 Ma. However, the total displacement across the eastern California shear zone (100 km ± 10 km) averaged over the last 12 m.y. suggests a long-term rate of 8.3 mm/yr ± 1 mm/yr. This rate is similar to or slightly less than the 8–12 mm/yr rate suggested by geodetic studies (McClusky et al., 2001; Miller et al., 2001; Sauber et al., 1994; Savage et al., 1990).

Another way to look at the evolution of the velocity field and provide a direct comparison between geologic data and geodynamical model results is by mapping the paths that individual ranges take over the deformational interval of interest (Fig. 11). Note that the bend in the path of the Pacific plate does not appear to be related to changes in the paths of the Sierra Nevada with respect to the Colorado Plateau or changes in the paths of individual ranges within the continent. The most significant continental change in direction occurs at 12 Ma. Because the plate constraints do not require the bend to occur at that time (it is only a function of the times at which magnetic anomalies constrain the position), it is possible within the uncertainties of both the plate reconstruction and geological reconstruction (Atwater and Stock, 1998; Wernicke and Snow, 1998) that these events more closely correlate. As stated previously, the timing of development of right-lateral shear depends on the orientation and timing of early extension in the Death Valley region, which if relatively minor prior to 11 Ma would point toward a later time of onset of right-lateral shear inboard of the Sierra Nevada. The distribution of north-northwest shear through the Mojave is kinematically linked to the northwesterly motion of the Sierra Nevada–Great Valley block, in turn requiring at least some amount of right-lateral shear within the Mojave region between 12 Ma and 8 Ma.

Although orogen-scale reconstructions of the Basin and Range will continue to evolve with time and adjust as more data is acquired, the exercise in kinematic compatibility we present here highlights what we understand and more importantly what we still do not understand regarding the evolution of the plate boundary.

Results that are robust and highlight what we do understand include: (1) 235 km ± 20 km of extension oriented N78°W in both the northern (50% extension) and central (200% extension) parts of the province. An important implication of the model is that any significant change in extension amount in a portion of the region (i.e., a range on the path between the Colorado Plateau and the Sierra Nevada) must be evaluated in light of how that change affects coevolving regions to the north and south. (2) A significant portion of boundary-parallel shear (in contrast to earlier extension) jumped into the continent at ca. 10–12 Ma, and once established, appears to have migrated westward with time. (3) The magnitude of slip on the eastern California shear zone appears to be 100 km ± 10 km, although the exact structures that accommodate this shear in the Mojave, or how much of the relative motion is accommodated by distributed shear, is not known.

Problems with the current reconstructions are highlighted by large gaps in the model. These zones emphasize areas where more work is needed in refining our ideas about how intraplate deformation is accommodated through time. Salient aspects of the model that we do not understand include: (1) compatibility between timing of extension north and south of the Garlock fault and a smooth north-northwest–trending continental edge as implied by plate tectonic reconstructions. To maintain a relatively smooth continental edge with different periods of extension across the Garlock fault, a triangular window of significant extension (>50 km; 24–16 Ma) followed by an equal amount of shortening (14–0 Ma) would have occurred in the Coast Range– Great Valley region. While known geology supports extension and subsequent compression in these time windows, the magnitude is ∼25% of what is needed; and (2) differences in magnitude and timing of extension between southern Arizona and northern Sonora, Mexico, require a transfer zone or large lateral displacement gradient. The model displays this zone as a gap that opens up (going backward with time) between the two provinces. Timing and magnitude of extension in the Sonora region were constrained only by plate motions to the west and broad assumptions as to similarities in timing and direction with areas to the north. Data detailing the magnitude, timing, and direction of extension through the Mexican Basin and Range is necessary to resolve this problem.

Animation 1 is available online at http://dx.doi.org/10.1130/GES00016.1.s1. GSA Data Repository item 2005200, Appendix 1, Movement Table, paleogeographic maps, and ArcGIS files (shape files for each reconstructed time step), is available online at http://www.geosociety.org/pubs/ft2005.htm, or on request from editing@geosociety.org, or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301-9140.

*Current address: Department of Geosciences, Guyot Hall, Princeton University, Princeton, New Jersey 08544, USA

This paper has benefited greatly from many conversations with scientists familiar with western North America geology. We specifically want to acknowledge Tanya Atwater, Bill Dickinson, Allen Glazner, Steve Graham, Jon Spencer, Joann Stock, Nathan Niemi, Mike Oskin, Jason Saleeby, and John Suppe. We are grateful to Melissa Brenneman at the University of Redlands for creating the ArcMap document and script used for the reconstruction. The movie would not exist without the help of the University of California, Santa Barbara, Educational Multimedia Visualization Center, specifically Carrie Glavich and Grace Giles. Gary Axen, Doug Walker, Craig Jones, and Randy Keller all provided insightful feedback through the review processes that greatly improved the clarity of presentation. This project was funded by the Caltech Tectonics Observatory.

Attribution: You must attribute the work in the manner specified by the author or licensor (but no in any way that suggests that they endorse you or your use of the work).Noncommercial ‒ you may not use this work for commercial purpose.No Derivative works ‒ You may not alter, transform, or build upon this work.Sharing ‒ Individual scientists are hereby granted permission, without fees or further requests to GSA, to use a single figure, a single table, and/or a brief paragraph of text in other subsequent works and to make unlimited photocopies of items in this journal for noncommercial use in classrooms to further education and science.

Supplementary data