Model for sinistral offset of the Cretaceous magmatic arc of southwestern North America along the Nacimiento fault system

This paper is an outgrowth of the attempt to reconstruct southwestern North America through geologic time by Bird and Ingersoll (2022). The same geologic and paleomagnetic data sets, which were culled from the literature, the same computer code (Restore v. 4), and a similar mix of objective automated processing through many routine timesteps were used, with subjective manual intervention at a few key times. The restoration calculations from the earliest geologic time reconstructed by Bird and Ingersoll (2022), 48 Ma, roughly coincided with the beginning of Basin and Range extension (Constenius, 1996) and the earliest possible dextral deformation on the San Andreas fault/proto-Gulf shear zone system.

For earlier geologic times, constraints are fewer and conclusions less certain. For example, computed paleogeologic maps have more blank areas at earlier times because progressively restored equivalents of present outcrops of young rocks did not exist. The maps of fault traces also are sparser because older faults are more poorly exposed and areas adjacent to major strike-slip faults were deleted from the map to avoid artifacts caused by large finite strain in local elements.

Two important innovations of this modeling project are that microplates between faults are not rigid, and that fault traces do not necessarily connect in triple junctions (Bird and Ingersoll, 2022). Instead, actual mapped (and then restored) fault traces are commonly unconnected. These characteristics imply a need for continuum deformation. To keep the problem well posed and the solutions physically plausible, two constraints were added: (1) Strain rate in each unfaulted element has a target rate of zero (in scalar terms), with an uncertainty quantified by local scalar parameter μ, and (2) principal-axis azimuths of continuum strain are required to approximately coincide with paleo-principal stress directions interpolated from paleostress directional data and orientations of new active faults.

Because the basic formalism of this method involves a 2-D grid of spherical-triangle finite elements (FEs) on the surface of the Earth, it is important to emphasize that the FE technique is not dynamic—only kinematic (Bird and Ingersoll, 2022). Instead of assuming lithospheric rheology, heat-flow, and stress boundary conditions to solve the stress equilibrium equation for quasi-static creeping flow, complex algebra is used to solve for the 2-D crustal velocity field that optimizes the fit to all available data in a weighted least-squares sense. This means that diverse data can influence the solution, with natural weighting by the inverse squares of their standard errors. The kinematic boundary conditions for the 2-D model domain are very simple: The eastern side is fixed relative to stable eastern North America, while all other model edges are free. The model includes the western conterminous USA and northern Mexico (north of the 20°N parallel in the Trans-Mexico volcanic belt); the −2000 m isobath is the western boundary.

During restoration from 48 Ma back to 90 Ma, the data sets and algorithm of Bird and Ingersoll (2022) provided model kinematics for the Laramide orogeny, the Hidalgoan orogeny, and late stages of the Sevier orogeny. These events were adjacent in space and time, if not overlapping.

A fundamental goal of this study is restoration of the continuous Cretaceous magmatic arc that had formed in the late Mesozoic above the Franciscan subduction zone and was later offset. This goal also relates to the “Salinia problem” in California, USA (e.g., Page, 1970; Hall, 1991; Hall and Saleeby, 2013), which involves two questions: (1) What was the configuration of the magmatic arc that allowed parts of it to be cut off and transported northwestward by the later San Andreas fault system?; and (2) how did this part of the magmatic arc “lose” its western margin and the Great Valley forearc basin that presumably once lay farther west, so that granodiorite from the arc is now exposed very close to the continent's edge in the Farallon Islands west of San Francisco Bay? One promising solution involves large sinistral offset on a long NW–SE fault system that includes the reconstructed Nacimiento fault in California, as proposed by Dickinson (1983). Two critical questions are: (1) Where did the Nacimiento fault trace project into and across Mexico?; and (2) how did the northwestern end of the fault trace interact with the Franciscan trench?

Figure 1 presents the paleogeologic map for 60 Ma (Paleocene), which was computed using model NI (as defined in Bird and Ingersoll, 2022) in timesteps back from 48 Ma. The elements of this paleogeologic map include polylines defining rock outcrop areas taken from the digital Geologic Map of North America (GMNA; Reed et al., 2005; Garrity and Soller, 2009) plus restored fault traces from the database of Bird and Ingersoll (2022). Outcrops of rock units younger than 60 Ma, and traces of faults thought to have formed after 60 Ma, are not included in the map. White areas represent rocks that were covered by younger overlap assemblages or were chiseled away along strike-slip faults, so their surface geology at 60 Ma is conjectural.

The most useful features for constraining large-scale tectonics on this digital map are outcrops of the magmatic arc caused by subduction of the Farallon (and/or Kula or other) oceanic plate(s) under North America that occurred from the latest Jurassic through Cretaceous time (Dickinson, 1981; Ingersoll, 2019a). Because this arc was a highland during the Cretaceous, and several kilometers of erosion have occurred since 60 Ma, the dominant remaining record of this magmatic arc is the belt of undivided granitic rocks of Cretaceous age (Kg) outcrops that extends across the N–S length of the map. Outcrops that are more precisely dated as Early Cretaceous granite (EKg) or Late Cretaceous granite (LKg) within the Sierra Nevada, or mid-Cretaceous intermediate intrusive rocks (MKi) in northern Washington state, USA, are also included. However, undivided granitic rocks of Cretaceous to Tertiary age (KTg) that are found primarily in southern Arizona and southern New Mexico, USA, are not included because these designations include younger and petrologically distinct leucocratic and peraluminous porphyry with Precambrian zircon (Haxel et al., 1984; Goodwin and Haxel, 1990) rather than phaneritic granodiorite.

Heavy red dashed lines in Figure 1 show the western and eastern limits of this great Kg + EKg + LKg + MKi belt. One isolated Kg outcrop (marked with a rectangular box) in northern Mexico east of the main belt is unexplained. Outcrops of Kg in the forearc of Baja California (on the Vizcaino Peninsula and Cedros Island) are probably accreted arc basement (plutonic ages of 165–135 Ma; Boles, 1986; Dickinson and Lawton, 2001; Critelli et al., 2002; Kimbrough and Moore, 2003).

As shown with interpretive lines in Figure 1, the western limit of the main Kg arc shows an apparent sinistral offset of ~400 km within present southern California. The geologic literature to date provides two alternative hypotheses to explain this: (1) thin-skinned gravity tectonics, which displaced the uppermost crust of the southern Sierra Nevada and Mojave Desert regions southwestward, emplacing the outcrops that are presently known collectively as the Salinia terrane (e.g., Hall and Saleeby, 2013), and (2) sinistral faulting on roughly NW–SE trends that occurred in latest Cretaceous and Paleocene time (e.g., Dickinson, 1983; Dickinson et al., 2005; Ingersoll, 2019b).

Based on extensive mapping of several quadrangles in the California Coast Ranges, Hall (1991) proposed that many of the outcrops (especially Kg outcrops) defining the Salinia terrane are not rooted, but are part of a thin nappe or klippe of the “southern California allochthon” overlying the regional “Sur thrust,” which displaced them southwestward relative to the lower crust of North America at ca. 65–55 Ma. This allochthon was proposed to extend from Point Arena in the northwest to the Chocolate Mountains in the southeast. Later, Hall and Saleeby (2013) modified and extended this concept by noting petrologic and geochronologic evidence for the denudation of crust in the present southern Sierra Nevada, which they attributed to Late Cretaceous and/or Paleocene detachment faulting with southwestward vergence. In their model, the upper crust of the present southern Sierra Nevada and Mojave Desert slid southwestward into the present Coast Ranges, leaving tectonic denudation behind and emplacing a nappe along the leading edge. Hall and Saleeby (2013) renamed the dominant subhorizontal fault the “Nacimiento fault.” Its southwestward transport was estimated to be 100–189 km, prior to dextral offset on the younger NW-trending San Andreas fault system.

Chapman et al. (2016) gave indirect support to this model by using detrital zircon geochronology from 95–80 Ma rocks in the Franciscan Complex southwest of the Nacimiento fault to infer source regions that they interpreted as contradicting alternative model two. However, their argument is inconclusive because of their neglect of (1) a possible more W–E trend of the hypothetical sinistral fault (as shown in Fig. 1 at 60 Ma), (2) possible northwestward transport of these Franciscan rocks by partitioned oblique subduction during roughly 95–75 Ma, and (3) uncertainties in paleo-drainage systems resulting from subduction of (an) oceanic plateau(s) in the same time period (e.g., Livaccari et al., 1981; Henderson et al., 1984; Liu et al., 2008, 2010).

Since the restoration code (Restore v. 4) is kinematic and works by finding weighted least-squares fit to multiple kinds of data, it cannot be used to model the proposed Late Cretaceous–Paleocene gravity tectonic phase without digitization of the complete trace of the causative low-angle fault and determination of its offset from pairs of piercing points. Most of the trace of the subhorizontal Sur/Nacimiento fault of Hall (1991) and/or Hall and Saleeby (2013) was proposed to be buried under younger cover sequences. Also, the parts of this fault system that are exposed have been interpreted differently by others. For example, the Rand thrust fault overlying the Rand Schist is generally interpreted as a window into the Farallon\North America interplate subduction megathrust, which is locally exposed in windows denuded by later Neogene extension (e.g., Postlethwaite and Jacobson, 1987; Chapman et al., 2016). Part of the Sur “thrust” emphasized by Hall (1991) has been reinterpreted by Dickinson et al. (2005) as the steeply dipping, dominantly Miocene dextral San Gregorio–Hosgri fault. Another part of the “thrust” has been shown to consist of the Miocene dextral Gamboa fault offsetting part of the Nacimiento fault, both of which dip steeply (Dickinson et al., 2005; Johnston et al., 2019). In contrast to the thrust model, Dickinson (1983), Dickinson et al. (2005), and Ingersoll (2019b) interpreted the Nacimiento fault as primarily a high-angle fault.

With so many uncertainties surrounding the locations (and even existence) of this hypothesized subhorizontal fault, Bird and Ingersoll (2022) were unable to add it to the original database, and thus were unable to test its effects on reconstructions. It is also true that Restore's algorithms and code for modeling faults would have to be completely reinvented to deal with fault traces that close on themselves, internal windows, external klippen, and fault rakes that vary continuously along a trace from extensional to strike-slip to compressional.

Consequently, this project has focused on modeling alternative hypothesis two: that the western edge of the Cretaceous magmatic belt was offset by sinistral faulting on a high-angle Nacimiento fault system that extended in a smooth arc across present California and into Mexico prior to its Neogene dissection by the dextral San Andreas system. The goal is to convert this hypothesis into one or more testable models by specifying plausible fault traces (relative to specific outcrops) and most likely sinistral offsets, and then computing the implied coeval tectonic styles in adjacent parts of southwestern North America.

The primary Cretaceous magmatic arc belt has two features that are convenient for this purpose. First, it was continuous, based on analogies to modern subduction-related magmatic arcs, even though it might have been curved and of variable width. Second, most intrusive activity in this belt tapered off at ca. 75 Ma in the USA (Snyder et al., 1976; Coney and Reynolds, 1977; Dickinson and Snyder, 1978) and shortly afterwards in Mexico (Urrutia-Fucugauchi, 1986), probably because the Farallon and/or Kula oceanic plates then began to subduct horizontally (Bird, 1984, 1988, 1992). Limited magmatic activity continued along the eastern side of the magmatic arc, so the eastern edge is less well defined than the western edge. In the reconstruction presented herein, the western edges of Kg outcrops have been restored to a linear trend near the continental margin.

Bird and Ingersoll (2022) discussed models NI, HP, and PM; this paper discusses models NI and NC. Thus, model NI is the connecting link that continues through 48 Ma. Bird and Ingersoll (2022) included discussion of extensive experiments to calibrate these Restore models (applied to the North American data sets) by selecting the best values of weighting parameters μ, L₀, and A₀. Seventy neotectonic models (not integrated back through time) provided initial guidance and demonstrated the unfortunate consequences of greater weight on point data (including paleomagnetic data) in models HP and PM (relative to model NI). Here, both models use the same values of L₀ = 2 × 10⁴m and A₀ = 1 × 10⁹m². The background or default value is µ = 5 × 10⁻¹⁶/s, with local and sometimes era-specific increases as given in table 3 of Bird and Ingersoll (2022).

Restoration of model NI from 48 Ma back to 60 Ma was straightforward. Because no throughgoing master strike-slip faults were active, only minor manual regridding was required. However, evidence for significant sinistral slip from 75 Ma to 60 Ma along the Nacimiento and related faults (Dickinson et al., 2005; Ingersoll, 2019b) dictated that manual regridding was needed for that period. It was decided that the model would be improved by including possible strike-slip fault segments in Mexico that may lie buried by Tertiary and Quaternary sediments and volcanics. This decision was complicated by uncertainty about the locations of possible trace extensions. The two most likely possible traces were tested in alternate models of the period of sinistral fault activity (75–60 Ma). Restoring significant strike slip on these traces had both proximal and distal effects on the rest of the model. To moderate these effects, one more paleo-high-μ (high-continuum compliance) zone was added in the pre-Oligocene basement of the Sierra Madre Occidental (Fig. 1), where the paucity of pre-Oligocene outcrops provides no constraints on Late Cretaceous–Eocene tectonics. Only one of the two trace models, NC, survived geologic tests of validity following restoration.

Dickinson (1983), Dickinson et al. (2005), and Ingersoll (2019b) proposed that, in Laramide time, the originally smooth and parallel lithotectonic belts comprising the Cretaceous subduction margin of western North America were cut by a long sinistral fault trending NW–SE, with offset of hundreds of kilometers. They bracketed the timing of this event as latest Cretaceous through Paleocene, and most likely within the 75–60 Ma interval.

Each of these papers identified the westernmost parts of the fault in California, where they have been offset by younger NW-trending dextral faults (including the San Gregorio–Hosgri, Rinconada, San Gabriel, and San Andreas). A helpful unifying principle was topological: Each piece of the sinistral fault served as part of the southwestern boundary of the California geologic terrane known as Salinia. Their collective name for these once connected fault segments was Nacimiento fault, taken from the name of one segment exposed near the Nacimiento River in the Coast Ranges of central California. It is important to note that this Nacimiento fault is different in location, dip, rake, offset, and age from the Nacimiento fault proposed by Hall and Saleeby (2013). However, sketch maps of tectonics during Laramide time by Dickinson (1983) and by Ingersoll (2019b) were necessarily generic and free of geographic coordinates, which left the locations of the northwestern and southeastern parts of the fault system poorly defined, and its terminations unknown. The Restore reconstruction attempts to extend this model by specifying fault-trace locations, determining and then restoring fault offsets, and exploring the implications of these large offsets for other aspects of the structure and geologic history of the southwestern USA and northwestern Mexico. The proposed total sinistral fault system had a minimum (great-circle chord) length of ~1700 km when it was active, and sinistral offset of ~400 km near its center.

Bird and Ingersoll (2022) defined the Nacimiento segment of the throughgoing Nacimiento fault system as consisting of the following nine faults, from NW to SE (Fig. 2):

• F3992L Nacimiento 00 (trench connector) paleo-sinistral fault, off California

• F2432L Nacimiento 01 (offshore segment) paleo-sinistral fault, off California

• F4253L Nacimiento 02 (Sur–Gamboa–San Gregorio segment) paleo-sinistral fault, California

• F2435L Nacimiento 03 (north segment) paleo-sinistral fault, California

• F4555L Nacimiento 04 (Rinconada segment) paleo-sinistral fault, California

• F2436L Nacimiento 05 (south segment) paleo-sinistral fault, California

• F4016L Nacimiento 06 (Big Pine west segment) paleo-sinistral fault, California

• F2437L Nacimiento 07 (Pine Mountain segment) paleo-sinistral fault, California

• F2438L Nacimiento 08 (Agua Blanca segment) paleo-sinistral fault, California

In most cases of joins between these faults, Bird and Ingersoll (2022) assigned the “throughgoing master fault” property to each end, so that Restore would keep those ends connected through time and finite strain. In this way, approximate paleo-locations of these pieces were determined at 60 Ma, as shown in Figures 1 and 3A. Due to deficiencies in the restoration algorithm, and the large, coarse finite elements, this collective trace got kinked and folded in a way that looks very unphysical, and which would effectively preclude any large strike-slip offset.

Therefore, for use in further restorations (before 60 Ma), a revised smooth trace of the Nacimiento segment was drawn at 60 Ma (Figs. 1 and 3A), following these rules:

1. The smooth trace follows the general locations and trends of the nine restored traces, but not their unphysical kinks.

2. The smooth trace passes adjacent to (but SW of) all Kg outcrops belonging to the northern half of the continental Kg belt. The northwesternmost of these are the future Farallon Islands; the southeasternmost are the Mesozoic granitoids of the future eastern San Gabriel Mountains.

3. Near the future Santa Ana Mountains of the Peninsular Ranges, which belong to the southern half of the continental Kg belt, the smooth trace passes adjacent to the northern sides of those Kg outcrops.

This Nacimiento segment ends where it meets the trace of the future San Andreas fault just NE of Mount San Jacinto and near Desert Hot Springs. Its restored length (great-circle chord) at 60 Ma was 452 km, and its mean azimuth was 103. It is not clear whether the sinuous (lazy-S) paleo-shape of this segment is real, or whether it is an artifact of deficiencies in the restoration algorithm.

The great Cretaceous magmatic arc of western North America (Fig. 1) was well defined by 80 Ma, and it was presumably continuous before it was cut by the Nacimiento fault system. Therefore, the displaced western edge of the magmatic arc can be used as a piercing point to measure strike slip. By this standard, the Nacimiento segment has a sinistral offset of ~370 km. The two primary sources of uncertainty about the locations of the western margins of the batholith are: (1) Western margins of the batholith are onlapped and covered by Upper Cretaceous and Paleogene forearc strata (e.g., Ingersoll, 1982a), and (2) there is uncertainty about the orientation of the Nacimiento segment at 60 Ma. If the Restore program has overestimated Neogene dextral offset on faults of modern central California, then the restored Nacimiento segment trend would be rotated clockwise relative to its orientation in Figure 1. This new orientation would increase the obliquity of truncation of the western edge of the magmatic arc, which would increase the total offset needing restoration. In any case, total sinistral slip of hundreds of kilometers on the Nacimiento system that obliquely offset the western edge of the magmatic arc was comparable in length, total offset, and rate to the younger dextral San Andreas fault system, but with an opposite sense of slip (Dickinson, 1983).

The question of where sinistral slip continued southeastward from the Nacimiento segment is difficult to answer. The fact that the northwestern end of Dickinson and Lawton's (2001) California-Coahuila transform restores very close to the restored southeastern end of the Nacimiento segment at 60 Ma (Fig. 1) suggests that it may have been reactivated with sinistral slip at 75–60 Ma. Although the California-Coahuila transform was active primarily in the Permo-Triassic (Dickinson and Lawton, 2001), reactivation as part of the sinistral Nacimiento system 75–60 Ma is plausible. East of the Sierra Madre Occidental, Dickinson and Lawton (2001) drew the California-Coahuila transform where Cretaceous strata are undisturbed, which is a possible location for a pre-Cretaceous fault but an impossible location for a fault that was active at 75–60 Ma. Therefore, the proposed Nacimiento–Caborca–Durango–Zacatecas (NCDZ) fault trace diverges from the California-Coahuila transform beneath the Sierra Madre Occidental and roughly follows its eastern boundary to the southeast. The proposed NCDZ fault can be divided into three segments (Fig. 3): (1) the Nacimiento segment, as described above; (2) the Caborca segment, which follows the California-Coahuila transform of Dickinson and Lawton (2001); and (3) the Durango-Zacatecas segment, which primarily follows the eastern boundary of the Sierra Madre Occidental, where it runs just east of one Kg outcrop (Fig. 1). The presumption is that Kg underlies part, if not most, of the Sierra Madre Occidental (e.g., Ferrari et al., 2007).

Target offsets for the three segments are the same (408 km, based on the offset along the smoothed trace of the Nacimiento segment of the western edge of Kg in Fig. 1); sigma uncertainties increase from 5 km to 32 km to 204 km for the Nacimiento, Caborca, and Durango-Zacatecas segments, respectively. The overall azimuth of the NCDZ fault at 60 Ma (B–F in Fig. 1) is 128 (relative to the stable North American reference frame). The overall azimuth of the NCDZ fault is close to the original suggestion of fault orientation proposed by Dickinson (1983) and subsequent publications of Dickinson et al. (2005), Jacobson et al. (2011), and Ingersoll (2019b). This azimuth is also consistent with the potential northwest continuation of the NCDZ fault along the coast, within the Franciscan Complex (see below). The Nacimiento segment of the NCDZ fault would thus be a releasing bend where it cuts across the Cretaceous batholith, geometrically similar to (but kinematically opposite to) the modern constraining bend of the San Andreas fault, where it traverses from the eastern side of the Peninsular batholith to the western side of the Sierra Nevada batholith. Transpressional shortening within the area of the Sierra Madre Occidental during sinistral movement along the NCDZ fault may have been enhanced by the more clockwise orientation of the Durango–Zacatecas segment relative to the Nacimiento segment.

This reconstruction is biased by the lack of mapping of structures hidden under the younger volcanic cover of the Sierra Madre Occidental (Figs. 1 and 2). This situation is like the problem that Bird and Ingersoll (2022) addressed in dealing with the Columbia Plateau basalts, which hide all older geology (and potential faults). The Sierra Madre Occidental was modeled in the same way, by accounting for the extreme uncertainty about continuum strains with a greatly enlarged value of μ in the covered area. The restored FE grid at 60 Ma includes a high-μ area under the future Sierra Madre Occidental area, which was assigned values of µ = 1.6 × 10⁻¹⁵/s = 50%/10 m.y. for times prior to 60 Ma. As Bird and Ingersoll (2022) explained, this number is not a target strain rate; the target strain rate remains zero in all continua. It is merely an increased uncertainty for that target, which gives greater flexibility locally if high-strain rates are kinematically implied by other model constraints. Figure 4 presents a map of continuum strains predicted during the interval of NCDZ fault slip (75–60 Ma), according to the computed NC model. Area increases (extension) indicated by yellow and red colors are shown north of the Sierra Madre Occidental in New Mexico and Texas, USA, and north and south of the E–W segment of the Nacimiento strand. Compression is shown in southern Arizona.

Figure 5 tracks development of the NCDZ fault from before its initiation at 75 Ma to its demise at 60 Ma, as calculated using model NC. The greatest uncertainties in these restorations are: (1) the location of the NCDZ fault southeast of the Nacimiento segment, (2) the type of basement beneath the Sierra Madre Occidental, and (3) the characteristics of deformation beneath the Cenozoic volcanics of the Sierra Madre Occidental. Several unrealistic artifacts may be observed in these restorations. The most obvious is the protuberance of forearc and subduction-complex components westward past the end of the Nacimiento fault at 78 Ma. This problem is discussed below. A second artifact is the distortion and rotation of parts of the northern Peninsular Range batholith, a function of curvatures in the NCDZ trace. Although some of this curvature may be a realistic depiction of the NCDZ fault from 75 Ma to 60 Ma, significant distortion of the fault trace undoubtedly is an artifact of the Restore process for the time between 0 Ma and 60 Ma. A third artifact is the general protuberance of the northern end of the Peninsular Ranges batholith relative to the Sierra Nevada batholith. Previous reconstructions of southern California using the sinistral Nacimiento fault model (e.g., Dickinson, 1983; Dickinson et al., 2005; Jacobson et al., 2011) that show a similar protuberance suffer from the same limitation as the present restoration: Great uncertainty concerning the amount of shortening, extension, and transrotation within the Transverse Ranges and the Mojave Desert after 60 Ma. Manual manipulation of individual blocks in these areas can result in more realistic reconstructions (e.g., figure 21B of Bird and Ingersoll, 2022), as illustrated below.

All reconstructions starting at 0 Ma use an FE grid that extends westward to water depths of 2 km along the continental margin (e.g., Bird and Ingersoll, 2022). Along much of the margin, this corresponds to the Patton Escarpment, west of Patton Ridge, which approximately represents the Cretaceous trench-slope break (e.g., Ingersoll and Graham, 1983) marking the boundary between the Franciscan Complex to the west and the Nicolas forearc basin to the east (Crouch, 1979; Crouch and Suppe, 1993). The Franciscan Complex accreted during subduction of the Farallon plate under North America prior to 28 Ma (Atwater, 1970; Dickinson, 1981; Jacobson et al., 2011). Therefore, the western boundary of our model edge lay near the Franciscan trench at its initiation in the latest Jurassic, but within the Franciscan Complex as it accreted to the continental margin (Ingersoll, 1982a). Restoration of hundreds of kilometers of sinistral slip on the Nacimiento fault system forces the outer edge of the FE grid into the implausible shape shown for the Franciscan trench in Figure 5 because the western edge of the model is unconstrained.

Manual manipulation of the 78 Ma map of model NI-NC illustrates how the angular discord of the sharp bend in the NCDZ fault can be lessened to create a more realistic fault trace, and result in a coherent reconstruction of the Franciscan trench and the coeval magmatic arc (Fig. 6). The same manual modifications that were used for the 48 Ma map of figure 21B of Bird and Ingersoll (2022) were used to modify the 78 Ma map: (1) The Tehachapi block of the southernmost Sierra Nevada was rotated 44° counterclockwise; (2) adjoining Mesozoic plutonic rocks and adjoining rocks of Salinia were rotated counterclockwise and translated eastward (Prothero, 2006, indicated ~40° of clockwise rotation for many parts of southern Salinia, similar to net rotations of the San Gabriel block, e.g., Terres and Luyendyk, 1985); and (3) the Chocolate Mountains anticlinorium was rotated counterclockwise (e.g., Ingersoll and Coffey, 2017; Coffey et al., 2022). In addition to these modifications that were based on underfit paleomagnetic data and suspected unaccounted- for crustal deformation of the Transverse Ranges and the Mojave Desert after 48 Ma, the following modifications were made for the 78 Ma map: (1) The NCDZ fault trace was lengthened northwest (beyond the known extent of the Nacimiento fault) to run subparallel to the Franciscan trench; (2) protruding blocks southwest of the Nacimiento fault were rotated clockwise and translated northward along the newly drawn NCDZ fault; (3) the Franciscan trench was drawn along the southwestern or western border of all map objects; (4) the southwestern/western boundary of the Cretaceous magmatic arc was drawn just seaward of all Kg outcrops (except for Cedros Island and the Vizcaino Peninsula, as discussed above); and (5) the northeastern/eastern edge of the Cretaceous magmatic arc was drawn at the extent of Kg outcrops (and suspected Kg underlying the Sierra Madre Occidental), except for one outcrop outlined in red.

Justification for drawing the northeastern/eastern boundary of Kg outcrops along the northeastern side of the Sierra Madre Occidental includes the following. “Cretaceous–Paleocene batholiths most likely underlie a large part of the Sierra Madre Occidental, given that Cretaceous dioritic intrusions are also reported in the Nazas area in western Durango (Aguirre-Díaz and McDowell, 1991)” (Ferrari et al., 2007, p. 16). In addition, isolated Cretaceous–Paleocene intrusive rocks reported by Ferrari et al. (2007) along the eastern edge of the Sierra Madre Occidental in Zacatecas are likely too small to show at the scale of the Restore maps. The scarcity of exposures of Cretaceous–Paleocene igneous rocks in the southern Sierra Madre Occidental is likely due to the extensive cover of ignimbrites of Oligocene and particularly early Miocene age. Drawing the southeastern segment of the NCDZ fault along the eastern boundary of Kg is justified on the basis that the eastern limit of Cretaceous magmatism would represent a rheological discontinuity, which could favor fault nucleation.

The resulting restoration (prior to NCDZ slip) has many positive characteristics. The arc-trench gap between the western limit of Kg and the Franciscan trench includes all known Cretaceous forearc and subduction-complex outcrops; it has a consistent width from the southern tip of the Baja Peninsula to the western side of the Klamath Mountains. Details of the Cretaceous forearc are obscured north of the Klamath Mountains by Eocene accretion of Siletzia (Wells et al., 2014). The width of Kg outcrops is relatively uniform throughout the map area, although the width of Kg outcrops near the Sierra Madre Occidental is poorly constrained. The general NW–SE trend of the NCDZ fault is similar to the trends of previous workers (e.g., Dickinson, 1983; Dickinson et al., 2005; Jacobson et al., 2011; Ingersoll, 2019b). This restoration provides many tests based on N–S-oriented lithologic belts in the Franciscan–Great Valley–Sierra Nevada systems (e.g., Ingersoll, 2019b).

Because of uncertainties in the restoration process, the topology of the NCDZ trace is speculative. Alternating bends in the trace could represent transpressional or transtensional segments, depending on the overall relative motion vector. If one assumes that the overall velocity vector had the same orientation as a line from points A to D in Figure 6, then segment AB would be transpressive, BC would be dominantly transtensive (except where neutral along the central section that is parallel to AD), and CD would be transpressive (see Fig. 4 for a comparison with calculated strains). If the map were additionally manipulated so that point B moved directly east to lie on line AD, then AB would be neutral, BC would remain transtensive, and CD would remain transpressive. In either reconstruction, segment BC represents a transtensive bend, as it transects the Cretaceous batholithic belt. This releasing bend would have facilitated the rapid cutting of submarine canyons into the southwestern side of the batholithic crust that was to become Salinia as the northern edge of the Peninsular Ranges slipped obliquely southeastward and was replaced by the below-sealevel Nacimiento forearc basin and Franciscan Complex from 75 Ma to 60 Ma (Fig. 5). The resulting continental margin would have resembled the present southern Mexico coast, where batholithic crust is exposed near the continental margin and arkosic detritus is fed directly into the trench (Ingersoll, 2019b). Latest Cretaceous–Paleocene submarine canyons cut into the southwestern margin of Salinia are well documented (see Ingersoll, 2019b, for references); other locations along the southwestern Salinian margin are dominated by thick uppermost Cretaceous–Paleocene arkosic detritus. The reconstruction of Figure 6 provides an elegant explanation of these relations. An intriguing additional hypothesis is that truncation of the Franciscan trench at the triple junction at A (with resulting increased shoreline-to-trench bathymetric gradients) would provide a mechanism for initiating rapid carving of Paleocene–Eocene submarine canyons of the Sacramento Valley, with mouths near the San Francisco Bay area (e.g., Dickinson et al., 1979; Ingersoll, 2019b). Additional supporting observations for this model include documentation of sinistral-slip kinematic indicators within the Nacimiento block of the Franciscan Complex southwest of the Nacimiento fault (Singleton and Cloos, 2013) and within southern Salinia near the San Gabriel block northeast of the Nacimiento fault (e.g., May, 1989; Bixler, 2023; Schwartz et al., 2023, 2024, and references therein; see Fig. 3A).

The transpressive nature of segment CD in Figure 6 is speculative because of uncertainties concerning the nature of crust beneath the Sierra Madre Occidental and the location of the Nacimiento–Caborca–Durango–Zacatecas fault beneath and/or along its eastern margin. A significant question is whether there is evidence for or against contractional deformation west of CD. Ferrari et al. (2007) stated that scattered available data seem to indicate that between the Coniacian and Eocene, contractile deformation was negligible across most of the Sierra Madre Occidental, and primarily occurred in northwestern Mexico. Ferrari et al. (2007) also stated that flat-slab subduction beneath the Sierra Madre Occidental is precluded by the presence of Late Cretaceous to Paleogene plutons and volcanics, in contrast to the Laramide orogeny in the United States. Therefore, an alternate model is required to explain Laramide/Hidalgoan orogenic activity in Mexico. This alternate model could be oblique collision of the Farallon conjugate of the Hess Rise beginning at ca. 75 Ma near the San Francisco Bay area, followed by southeastward migration along the continental margin (Barth and Schneiderman, 1996; Liu et al., 2008, 2010; Jacobson et al., 2011; Ingersoll, 2019b). This interaction would have induced sinistral slip along the NCDZ fault as the trench–trench–transform migrated southeastward along the coast (Ingersoll, 2019b), as well as flattened the subducted slab to induce the classic Laramide orogeny (e.g., Dickinson and Snyder, 1978; Bird, 1984, 1988; Dickinson et al., 1988).

The overall approach in the Restore project has been “kinematics before dynamics.” In other words, it is necessary to compute what actually happened before attempting to explain how and why. Sixty years of literature informed by plate-tectonic concepts indicates that the three primary processes affecting southwestern North America since 90 Ma have been: (1) ongoing subduction of the Farallon oceanic plate and its successor fragments; (2) effects of the Yellowstone mantle plume, starting with Tillamook volcanism at 42–35 Ma (Wells et al., 2014); and (3) lengthening contact between North America and the Pacific plate after ca. 28 Ma (Atwater, 1970; Nicholson et al., 1994). A space/time diagram in the style of Dickinson (2006; Fig. 7) provides a useful summary. Two PowerPoint files are also supplied as Supplemental Material¹ that show successive paleogeologic maps in 6 m.y. timesteps back to 90 Ma. While not quite animations, the individual frames have better resolution and can be extracted for use in derived illustrations.

The beginning of Mesozoic–Cenozoic subduction of oceanic plates under western North America is marked stratigraphically by the Upper Triassic Shinarump Conglomerate (Heller et al., 2003), which was deposited by streams that flowed northeast from the arc-related Mogollon Highlands in southeastern California and southwestern Arizona (Armstrong and Ward, 1993; Dickinson, 2000). The occurrence of petrified wood in the Chinle Formation indicates that windborne silicic volcanic ash was prevalent (Heller et al., 2003). Between the Late Triassic (229–197 Ma) and the earliest timestep of the 90 Ma reconstruction, thousands of kilometers of seafloor were subducted. Seismic tomography images parts of this subducted Farallon slab (and older slabs), which dip east to beneath Ohio and Mississippi, USA (~3700 km from the paleo-trench), down to present depths of 1500 km (Pavlis et al., 2012; Sigloch and Mihalynuk, 2017). The hydrated and carbonate-rich seafloor on the Farallon slab interacted with circulating asthenosphere to generate vast volumes of magma, which greatly thickened the crust of the Cretaceous magmatic arc and built up its topography. This magmatic-arc highland, and the adjacent Nevadaplano (DeCelles, 2004; Ernst, 2009), were in positions where they could exert topographic pressure coincident with the Sevier orogeny and affect weak and anisotropic sedimentary strata that were originally deposited along the Neoproterozoic–Paleozoic intraplate continental margin of Laurentia (e.g., Stewart and Suczek, 1977; Poole et al., 1992; Ingersoll, 1997). Much of this Cretaceous Cordillera would later be dissected and lowered by Basin and Range extension (beginning at 49 Ma; Dickinson, 2006).

Following at least 60 m.y. of oceanic subduction at normal dip angles, the subducting Farallon slab began sliding horizontally along the base of the North American lithosphere soon after ca. 80 Ma (Dickinson and Snyder, 1978). The most likely cause was subduction of a buoyant oceanic plateau or other buoyant oceanic crust, such as Farallon plate conjugates of the Shatsky Rise or Hess Rise on the Pacific plate (Livaccari et al., 1981; Henderson et al., 1984; Barth and Schneiderman, 1996; Liu et al., 2008, 2010; Seton et al., 2012). This episode of horizontal subduction had at least four major consequences for North American history: (1) The magmatic arc migrated eastward in the United States to as far inland as Montana, Wyoming, South Dakota, Colorado, New Mexico, and Texas, and south into central Mexico (Snyder et al., 1976; Urrutia-Fucugauchi, 1986; Bird, 1988); (2) northeastward tractions on the base of North American lithosphere created horizontal compressive stresses that broke the Precambrian metamorphic basement of the foreland, resulting in the Laramide orogeny that formed the Rocky Mountains (Dickinson and Snyder, 1978; Bird, 1988, 1992, 1998); (3) crustal roots from the former Cordillera and Nevadaplano were dragged northeast to form the young (post-Cretaceous) crustal roots of the Rocky Mountain foreland and Great Plains (Bird, 1984, 1988, 1992, 1994); and (4) North American mantle lithosphere was dragged away from the future Basin and Range province and Colorado Plateau, exposing the crust there to contact with hot asthenosphere, basal melting, and isostatic uplift after removal of the flat slab at ca. 30 Ma (Bird, 2002; Fig. 7).

In this context of flat subduction, basal shear tractions exerted by the buoyant aseismic ridge(s) of the Farallon plate on the parts of North America that lay southwest of the Nacimiento fault system probably played a large role in driving its initiation and slip. While the magnitudes of such tractions are uncertain, they were enough to deform stable North American lithosphere in thrusting mode throughout the latitudes of the United States (forming the younger Rocky Mountains east of the older Sevier orogen). They also should have been sufficient to drive deformation of North American lithosphere in strike-slip mode closer to the coast. However, we would have low confidence in any theoretical dynamic prediction of the azimuth of the new Nacimiento fault system, because its initiation may have involved the reactivation of preexisting weaknesses, such as the marginal normal faults of the Jurassic Bisbee graben (e.g., Chapman and DeCelles, 2021), the San Marcos fault system (Aranda-Gómez et al., 2005), and/or the California-Coahuila transform (Dickinson and Lawton, 2001) in Mexico.

Subduction beneath North America of one or more buoyant oceanic plateaus or rises embedded in the Farallon plate, such as conjugates to the Shatsky or Hess rises, may have driven both flat subduction and strike slip. The actualistic example of subduction of the Yakutat terrane beneath the Chugach and St. Elias ranges in Alaska, USA, (Bird, 1996; Wells et al., 2014) shows that this can cause both uplift and deformation of the forearc, and fanning of horizontal principal stress directions away from the point of collision, which implies locally higher deviatoric stresses.

This hypothesis is consistent with part of Schwartz et al.'s (2023) two-stage model for the Laramide orogeny that includes collision of the Hess Rise at ca. 75 Ma along the southern California margin. In other aspects, however, their model differs in fundamental ways from the model proposed herein. One significant difference is their lack of palinspastic reconstruction to account for Cenozoic deformation (e.g., Bird and Ingersoll, 2022; Ingersoll et al., 2024). In addition, they do not consider significant sinistral slip on the Nacimiento fault, even though they and coworkers have documented several examples of sinistral slip in the 75–60 Ma time frame (see Fig. 3A). The reconstruction shown in Figure 6 requires that Schwartz et al.'s (2023) “Southern California batholith” is actually the northeastern side of the Peninsular Ranges batholith. When restored, this batholith looks very similar to the Sierra Nevada batholith, with a western metamorphic belt, central Early Cretaceous plutons, eastern Late Cretaceous plutons, and easternmost Precambrian crust intruded by Jurassic and latest Cretaceous plutons (White-Inyo Ranges east of Sierra Nevada). This is the relationship first recognized by Dickinson (1983) and beautifully illustrated by Jacobson et al. (2011). Thus, sinistral slip along the Nacimiento fault would have offset the older western part of the arc relatively to the southeast and the younger eastern part to the northwest. Slip on the Nacimiento fault was likely post-plutonic where it transected the former, and synmagmatic where it transected the latter. The reconstructions illustrated herein require significant modifications of many widely used models for evolution of the Cretaceous magmatic arc of California and northern Mexico.

The following are possible tests for the proposed reconstructions.

1. The transpressive and transtensive predictions of Figure 6 could be tested through studies of adjoining sedimentary basins and structural provinces.

2. Offsets of geological features across the Caborca segment of the NCDZ fault could be sought.

3. Geophysics and/or boreholes could test the prediction of pre-Oligocene shortening of the basement of the future southern Sierra Madre Occidental.

4. Initiation of Pelona–Orocopia–Rand Schist production should young NW to SE.

5. Initiation of forearc deep-marine sedimentation directly on Mesozoic magmatic-arc basement of Salinia should young NW to SE.

6. Detrital-zircon age spectra and other provenance indicators should correlate with potential source areas during and following sinistral slip.

7. Metamorphic and plutonic belts of the batholithic belt should realign.

8. Great Valley forearc belts should realign.

9. Franciscan belts should realign.

These possible tests demonstrate the advantages of applying finite-strain, finite-element modeling to a tectonic hypothesis in the process of developing a testable model. It is hoped that the present contribution will stimulate further tests.

Peter Bird has been my collaborator throughout this project. He did all of the programming and calculations, for which I am tremendously thankful. His expertise with finite-element modeling has provided the objective reconstructions that are the basis for model testing and further speculation. His work was supported by the National Science Foundation under grants EAR-9316169 and EAR-9614263, and by the U.S. Geological Survey (USGS), Department of the Interior, under USGS award number 01HQGR0021 to the University of California. The views and conclusions contained in this document are those of the author and should not be interpreted as necessarily representing the official policies, either express or implied, of the U.S. Government. In addition to Peter, I thank William Holt, Craig Jones, Josh Schwartz, Jon Spencer, and Basil Tikoff for thorough reviews and helpful suggestions.