Structural evolution of an en echelon fold system within the Laramide foreland, central Wyoming: From early layer-parallel shortening to fault propagation and fold linkage

The Laramide belt, spanning a region from southern Montana to New Mexico (USA), is an archetypical example of thick-skin foreland deformation that developed far inboard from a plate margin (Fig. 1; Dickinson and Snyder, 1978; Bird, 1984; Saleeby, 2003). The belt is characterized by fault-bound, basement-cored arches that are bordered by flanking fold systems and separated by broad basins (Berg, 1962; Brown, 1988; Erslev, 1993). The belt displays an overall northwest-southeast structural grain, but individual basement faults have a range of strikes and cover folds form both north-trending, right-stepping, and west-trending, left-stepping en echelon systems. General timing of large-scale Laramide deformation spanned ca. 70–50 Ma in the Wyoming region, constrained by sedimentologic patterns in the basins (Beck et al., 1988; Dickinson et al., 1988; DeCelles et al., 1991; Steidtmann and Middleton, 1991; Hoy and Ridgway, 1997) and by thermochronologic data in the arches (Crowley et al., 2002; Peyton et al., 2012; Stevens et al., 2016). However, patterns vary regionally and questions remain on timing of Laramide initiation, varying models of southwest to northeast, west to east, or northeast to southwest propagation of deformation, and changes in shortening directions over time (Chapin and Cather, 1983; Gries, 1983; Varga, 1993; Heller et al., 2013; Fan and Carrapa, 2014; Lopez and Steel, 2015; Copeland et al., 2017). In this paper structural and paleomagnetic data sets for a flanking fold system of the basement-cored Wind River arch are examined to better understand Laramide tectonics. Note that the term Laramide is used for thick-skin structural style (compared to thin-skin style of Sevier belt), and not for a specific time interval.

Growth of variably oriented basement-cored arches and flanking fold systems within the Laramide belt have generated multiple tectonic models, including (1) temporal changes in shortening directions (Gries, 1983; Bergh and Snoke, 1992), (2) transpression with varying components of horizontal shear along differently oriented zones (Sales, 1968; Stone, 1969), and (3) regional shortening with spatial refraction of stress-strain directions (Fig. 2A; Erslev and Koenig, 2009). Model 1 predicts early east-west shortening that formed north-trending arches, followed by late north-south shortening that formed east-trending arches, with overprinting and possible linkage of early structures by later structures. Model 2 predicts widespread development of en echelon folds, strike-slip to oblique-slip faulting, and significant vertical-axis rotations, with sinistral shear and left-stepping folds along more west-trending zones, compared to dextral shear and right-stepping folds along more north-trending zones. Model 3 predicts smoothly varying stress-strain directions during temporally overlapping development of variably trending arches, partly related to preexisting basement heterogeneities and weaknesses (Huntoon, 1993; Molzer and Erslev, 1995; Marshak et al., 2000; Stone, 2002; Neely and Erslev, 2009; Weil et al., 2014). For this model, along-strike fault propagation and fold linkage may also give rise to either left- or right-stepping en echelon fold systems, along with local partitioning of shear along obliquely oriented faults. It is important that these models predict different spatial and temporal patterns of large-scale fault and fold geometry, internal strain, and vertical-axis rotations, which can be evaluated from integrated structural and paleomagnetic studies. Strike tests, in which changes in shortening directions and paleomagnetic declinations are correlated with changes in structural trend, provide a quantitative method to test models (Fig. 2A).

Relations of flanking fold system formation to basement deformation have also generated multiple models, including (i) detachment and buckle folding of the sedimentary cover related to crowding (Petersen, 1983; Nelson, 1993); (ii) fault-propagation folding associated with basement reverse faults, as in fold-thrust and trishear models (Berg, 1962; Erslev, 1991; Stone, 1993); and (iii) drape folding above subvertical basement faults (Fig. 2B; Matthews and Work, 1978). For detachment models (i), folds should be broadly symmetric and associated with listric thrust faults. For fault-propagation models (ii), fold forelimbs should be steeper and undergo varying amounts of shear during progressive tilting, related to the ratio of fault propagation to slip rate in trishear models. High ratios, as in overall strong basement, favor concentrated fault slip, whereas lower ratios, as in weak sedimentary cover, favor distributed shear and folding (Allmendinger, 1998). The fold-thrust model (Berg, 1962) combines early buckling of both basement and cover, followed by fault propagation. Fault segments and associated folds may also propagate laterally to form (a) a single, straight fold-fault system with along-strike linkage of colinear segments; (b) an en echelon fold-fault system with partial linkage of segments having moderate separation distances across structural saddles that represent relay zones; and/or (c) separate fold-fault systems that bypass each other for segments with large separation distances (Fig. 2C; Davis et al., 2005; Grasemann and Schmalholz, 2012). Varying interactions between propagating and linking segments also result in distinctive along-strike gradients in fault displacement and shortening (Gupta and Scholz, 2000). Seismic and drill hole data have revealed moderate-dipping reverse faults along steep forelimbs of some flanking folds, local detachment folds, and complex structural saddles between domes (Berg, 1962; Brown, 1988; Erslev, 1993; Stone, 1993), but the importance and styles of buckling, basement faulting, and lateral fold-fault linkage remain debated (Petersen, 1983; Gay, 1999; Tiffany, 2011).

The Dallas–Derby–Sheep Mountain (D-D-SM) fold system, located along the southeastern margin of the Wind River arch, is an ideal location to test models for evolution of flanking folds because of the following: red beds of the Triassic Chugwater Group are widely exposed and carry a high-fidelity record of the paleomagnetic field, providing a marker to estimate vertical-axis rotation (Van der Voo and Grubbs, 1977; Shive et al., 1984); red beds typically have subtle anisotropy of magnetic susceptibility (AMS) fabrics that can be used to estimate layer-parallel shortening (LPS) directions and compared with shortening directions estimated from minor faults in interlayered Triassic and Jurassic limestone beds (Weil and Yonkee, 2012; Weil et al., 2014, 2016); structural geometry of the fold system is well constrained by surface exposures and drill hole data (Willis and Groshong, 1993; Brocka, 2007; Alward, 2010; Hilmes, 2014); and basement rocks are exposed nearby in the core of the Wind River arch (Frost et al., 2000) and imaged geophysically along the arch flank (Gay, 1995), allowing evaluation of basement heterogeneities and weaknesses. By integrating detailed structural and paleomagnetic studies this paper addresses the following questions.

1. What are the large-scale structural geometries, internal strain patterns, and nature of vertical-axis rotations in the D-D-SM flanking fold system?

2. How do paleostress-strain directions estimated from different data sets (minor faults, extension fractures, AMS fabrics, and previously published calcite twin data) compare, and how did the local stress-strain field evolve in relation to regional Laramide patterns?

3. What combination of models best explains development of the fold system?

If temporal changes in shortening directions were important during development of the fold system, then early north-trending structures should be overprinted by later west-trending structures, and vertical-axis rotations should be limited. If transpression was important, then vertical-axis rotations should be widespread. If regional shortening was dominant, then stress-strain directions and large-scale structures should record a single, evolving deformation field. If fault propagation and fold linkage were important, then forelimbs should be steeper, reverse faults should cut basement at depth, and structural saddles between fold culminations should record shortening transfer.

The Laramide belt comprises a distinctive part of the North American Cordilleran orogenic system, which developed during protracted Jurassic to Paleogene Andean-style subduction and terrane accretion (DeCelles, 2004; Yonkee and Weil, 2015). Development of the orogenic system occurred during a time of northwestward to westward drift of the North America plate and increased rates of relative motion with the subducting Farallon and related oceanic plates. Directions of relative motion ranged from approximate west-east in the Early Cretaceous, to southwest-northeast in the Late Cretaceous to Paleogene (Doubrovine and Tarduno, 2008; Wright et al., 2016). A flat-slab segment likely developed along the Mojave sector ca. 90 Ma, and then propagated northeastward under the Laramide foreland by later Cretaceous to Paleogene time (Fig. 1; Saleeby, 2003). The presence of a flat-slab segment is indicated by development of a magmatic gap along the Mojave to Sierra Nevada arc sectors (Dickinson and Snyder, 1978; Cross and Pilger, 1982), inboard shift of limited magmatic activity along the Colorado Mineral Belt (Chapin, 2012), forearc disruption and underplating of the Pelona-Orocopia-Rand schists (Jacobson et al., 2011), presence of eclogite xenoliths (subsequently erupted in volcanic centers in the Colorado Plateau) that likely represent remnants of subducted oceanic crust and altered mantle lithosphere (Smith and Griffin, 2005), and increased Late Cretaceous subsidence followed by uplift that propagated northeastward from Arizona to Wyoming (Liu et al., 2010; Heller et al., 2013). Although multiple lines of evidence indicate the former presence of a flat slab beneath the Laramide belt, geodynamic processes that led to thick-skin crustal deformation are debated, including end loading along a lithospheric keel, increased basal traction, and enhanced asthenosphere flow (Bird, 1984; Livaccari et al., 1981; O’Driscoll et al., 2009; Jones et al., 2011; Yonkee and Weil, 2015). Laramide deformation partly overlapped with later phases of Sevier thin-skin thrusting, but shortening directions differed. Regional west-southwest–east-northeast shortening in the Laramide was at low angles to the direction of relative motion between the Farallon and North American plates during the Late Cretaceous (Wright et al., 2016), supporting models of increased coupling during flat-slab subduction (Weil and Yonkee, 2012). In contrast, the Wyoming salient of the Sevier belt underwent regional east-west shortening with radial dispersion of shortening directions along arcuate thrusts, supporting wedge models with a component of topographic-driven stress along a curved mountain front and the presence of a weak basal décollement (Yonkee and Weil, 2010a).

The study area is located along the southern part of the northeast flank of the Wind River arch in central Wyoming (Fig. 3). The arch is ∼200 km long by ∼50 km wide and trends overall northwest, in detail curving from north-northwest– to west-northwest–trending from the northern to southern ends (Blackstone, 1993a). The southwest flank of the arch is bound by the moderately dipping (∼30°–40°) Wind River thrust, which has a net slip of ∼25 km and cuts an overturned, sheared limb in the sedimentary cover (Brown, 1988). The thrust is imaged geophysically to mid-crustal (∼25 km) depth (Smithson et al., 1979), but evidence for offset of the Moho is lacking and the fault may flatten within the lower crust, similar to interpretations for other arches (Erslev, 1993; Yeck et al., 2014). Toward the northern end, the Wind River thrust branches and slip is partly transferred onto the White Rock thrust (Fig. 3A; Mitra et al., 1988). Toward the southern end, the Wind River thrust loses slip eastward as shortening increases along the Sweetwater arch (Weil et al., 2016). The Continental normal fault system reactivated the southern part of Wind River thrust during subsequent partial collapse of the arch. The northeast flank of the Wind River arch has overall gentle dips (typically 10°–15°) toward the Wind River Basin, but is locally deformed by flanking folds, including the D-D-SM fold system that is the focus of this study. Total shortening across the arch and flanking folds is ∼25 km (Fig. 3B).

Basement rocks exposed near the study area within the southern part of the Wind River arch and western part of the Sweetwater arch comprise older (ca. 3.3–2.7 Ga) polydeformed migmatitic orthogneiss with disrupted lenses of supracrustal rocks; younger (2.7–2.6 Ga) supracrustal belts, including the South Pass belt that has east-northeast–striking shear zones and foliations; younger (2.7–2.6 Ga) calc-alkaline plutons with variably developed foliations; Neoarchean (ca. 2.6–2.5 Ga) nonfoliated granitic plutons; and limited 2.1 Ga and 0.8 Ga mafic dikes (Frost et al., 2000). The Oregon Trail structural zone, which marks a fundamental change in basement rock types and lithosphere thermal history and contains east-northeast– to east-striking shear zones, extends from the northern margin of the Sweetwater arch, beneath the southern part of the D-D-SM fold system, into the Wind River arch (Fig. 3A; Chamberlain et al., 2003). The protracted tectonothermal evolution of these basement rocks led to development of crustal weaknesses with varying orientations that likely influenced the style and distribution of Laramide deformation.

The sedimentary cover in the study area includes a Cambrian basal sandstone; an overall incompetent interval of upper Cambrian shale and minor limestone; a competent interval of Ordovician to Mississippian massive carbonates; Pennsylvanian to Permian interlayered sandstone, shale, and carbonate; Triassic to Jurassic interlayered red beds, limestone, and minor evaporites; and an overall incompetent interval of Cretaceous shale and sandstone.

Latest Cretaceous to early Eocene synorogenic strata were deposited within the Green River Basin and Wind River Basin to the southwest and northeast of the arch, respectively, providing key constraints on timing of large-scale deformation (Keefer, 1970; Beck et al., 1988; Steidtmann and Middleton, 1991). Synorogenic strata within the Green River Basin record unroofing of the arch and include the Maastrichtian Lance Formation, which contains clasts eroded mostly from Mesozoic to Paleozoic strata; the Paleocene Fort Union Formation, which records early influx of basement granitic detritus; and the early Eocene Wasatch Formation, which records continued erosion of granitic material (Steidtmann and Middleton, 1991). The Paleocene depositional axis of the Green River Basin was located slightly southwest of and roughly coincident with the curved front of the Wind River arch, consistent with a component of flexural subsidence (Beck et al., 1988). Along the northeast flank of the arch and into the Wind River Basin, the Lance Formation has dips similar to those of the older strata around folds, whereas the base of the Paleocene Fort Union Formation is marked by an angular unconformity, indicating slightly younger onset of large-scale folding there (Keefer, 1970). The Eocene Wind River Formation records continued folding with simultaneous exhumation of the Wind River, Sweetwater, and Owl Creek arches. Little deformed Late Eocene to Oligocene strata onlap the Wind River thrust along the southwest flank (Steidtmann and Middleton, 1991) and overlie with angular unconformity folds on the northeast flank (Love, 1970), bracketing the end of large-scale Laramide shortening in the region. Apatite fission track data from the Wind River arch provide additional constraints, with depth-age relations suggestive of ∼3 km of exhumation and rapid cooling from ca. 65 to 50 Ma (Peyton et al., 2012; Stevens et al., 2016).

Detailed structural, AMS, and paleomagnetic analyses were completed for 61 sites within red beds of the Red Peak Formation and overlying Alcova Limestone of the Triassic Chugwater Group, along with structural analysis of 28 sites within limestone beds of the Jurassic Sundance Formation (Fig. 4A; Table DR1¹). Sites were distributed along the length of the D-D-SM fold system and gently northeast-dipping limb of the Wind River arch. The Red Peak Formation, interpreted as Early Triassic based on regional correlations, is composed mostly of subarkosic to quartzose, variably micaceous, sandstone and mudstone (Fig. 5A) that contain fine-grained hematite that formed during early diagenesis (Picard, 1965; Weil and Yonkee, 2012). The Alcova Limestone, interpreted as latest Early Triassic to Middle Triassic based on ⁸⁷Sr/⁸⁶Sr ratios (Lovelace and Doebbert, 2015), is composed of laminated, micritic limestone that forms a distinctive 1–5-m-thick marker unit. The Middle Jurassic Sundance Formation is composed of interbedded marine sandstone, mudstone, and oolitic to bioclastic limestone.

At each site, detailed structural measurements were made of bedding, fracture sets, and minor faults (best developed within the Alcova Limestone and limestone beds of the Sundance Formation). Site mean orientations and α₉₅ confidence cones of poles to bedding and fracture sets were calculated using Fisher (1953) statistics. Idealized maximum, intermediate, and minimum (σ₁, σ₂, and σ₃) stress vectors were estimated for minor faults using the method of Compton (1966) and a 25° fracture angle, with site mean stress directions and α₉₅ confidence ellipses calculated using the orientation tensor. Because this method does not account for potential reactivation of preexisting fractures, a stress inversion method was also used that searched for the combination of principal stress directions and stress ratio, ϕ = (σ₂ – σ₃)/(σ₁ – σ₂), that minimized the absolute value of angular misfit between observed slip lineations and directions of maximum resolved shear stress on the fault planes. Principal strain directions were also estimated using linked Bingham distribution statistics with the FaulKin software (Allmendinger et al., 2012).

Core samples for AMS and paleomagnetic analyses were drilled from well-exposed red beds using a portable gas-powered drill and a magnetic compass for orientation. We obtained 8–12 cores from multiple beds at each site in order to average out secular variation and potential variability in AMS fabrics. AMS was measured for individual cores using an AGICO Kappabridge KLY-3 susceptibility bridge operated at a frequency of 875 Hz and with a sensitivity of ∼2.0 × 10⁻⁸ SI. Site mean eigenvectors/eigenvalues and uncertainties were estimated from core data using tensor methods and bootstrap resampling following the methods of Constable and Tauxe (1990) and Tauxe (1998). AMS is a second order tensor, [K], that relates directional variability of induced sample magnetization in response to an applied magnetic field of varying orientation, providing a measure of overall preferred orientation of magnetic grains within a volume of rock (Borradaile, 2001). The shape and orientation of the AMS ellipsoid are defined by the three eigenvalues and eigenvectors, K_max ≥ K_int ≥ K_min. Because AMS measures the combined contributions from diamagnetic, paramagnetic, and ferromagnetic minerals that may form at different times and by different mechanisms, care is needed to document relations of microfabrics to measured AMS ellipsoids. Detailed sampling, AMS analysis, and scanning electron microscope (SEM) imaging of multiple lithologies with varying grain size and mica content were completed for two sites.

Cores were thermally demagnetized in an Analytical Service Company demagnetizer and measured with an AGICO JR-6a spinner magnetometer in a low-field magnetic cage in order to determine characteristic remanent magnetization (ChRM). ChRM directions were calculated for each core using principal component analysis (Kirschvink, 1980), based on linear demagnetization-step trajectories extracted from NEV (north, east, vertical) component plots (Zijderveld, 1967). Site mean paleomagnetic vectors and α₉₅ confidence cones were calculated from core data using Fisher (1953) statistics. The Super IAPD2000 software package was used for paleomagnetic data analyses (Torsvik et al., 2000). A total of 20 cores were also selected for rock magnetic experiments to characterize remanence carrying mineral fractions and evaluate contributions of ferromagnetic and paramagnetic minerals to susceptibility. Specimens were chosen to represent a range of AMS fabric types and provide broad spatial coverage. Acquisition of isothermal remanent magnetization (IRM) was measured on 10 specimens using a Princeton Measurements Corporation MicroMag vibrating sample magnetometer in fields to 2.0 T. Three-axis thermal demagnetization of IRM was performed on 10 additional specimens. Applied fields of 2.0, 0.5, and 0.15 T were imparted along the three mutually perpendicular axes of each core (Lowrie, 1990) using a Magnetic Instruments pulse magnetizer. All magnetic experiments were carried out in the Paleomagnetic Laboratory at Bryn Mawr College (Pennsylvania).

The D-D-SM fold system comprises a series of left-stepping, doubly plunging anticlines, which share a gently dipping backlimb, have segmented, southwest-verging forelimbs cut by reverse faults, and are connected across structural saddles (relay zones) (Fig. 4A). A syncline separates the fold system from the gently northeast-dipping homoclinal limb of the Wind River arch. A series of cross sections across the system provide a framework to evaluate fold-fault relations, internal strain patterns, and vertical-axis rotations (Figs. 4B–4F).

Dallas Dome, the site of the first producing oil well in Wyoming (Krampert, 1948), has an amplitude of 1.0 km with Triassic strata exposed in the fold culmination, and accommodated 0.9 km of shortening (Fig. 4B). The fold has an axial length of ∼10 km, trends overall 340°/160°, plunges ∼10° northward along its northern end, and curves and plunges 20° toward 140° at its southern end. The fold has a steep forelimb that dips 50°–90°SW and is cut by two top-to-the-southwest reverse faults that offset basement at depth (Willis and Groshong, 1993). The backlimb dips ∼10°–15°NE and is locally cut by a backlimb thrust, interpreted to sole into Triassic strata. Small-displacement, east-striking normal faults accommodated limited fold axis–parallel extension near the fold culmination.

The Dallas-Derby saddle (relay zone) is marked by a ∼1.5 km left step and curvature in fold trend. Fold amplitude is less (0.8 km), with Jurassic to Cretaceous strata exposed across the saddle. The forelimb dips 30°–50°SW and bends in strike to 310°/130°. Shortening across the forelimb is smaller compared to Dallas Dome, whereas slip on the backlimb thrust is greater (Fig. 4C). The forelimb is locally cut by a steep, east-striking fault with sinistral offset. This fault appears to partly cut and partly transfer slip onto forelimb reverse faults, and loses slip eastward toward the hinge zone of the saddle (Brocka, 2007). This fault lines up with an east-striking fault in the Wind River arch, and may reflect reactivation of a basement weakness.

Derby Dome, the site of another oil field, has an amplitude of 1.0 km with Triassic strata exposed in the fold culmination, and accommodated 0.9 km of shortening (Fig. 4D). The fold has an axial length of ∼8 km, trends overall 340°/160°, plunges ∼10° northward along its northern end, and curves and plunges 15° toward 150° at its southern end. The forelimb dips 40°–70°SW and is locally cut by top-to-the-southwest reverse faults interpreted to offset basement at depth, and by a top-to-the-northeast backthrust that repeats Paleozoic to Triassic strata toward the fold hinge (Willis and Groshong, 1993). The backlimb dips 10°–15°NE and is cut by the backlimb thrust, which loses slip southward.

The Derby–Sheep Mountain saddle (relay zone) is marked by a ∼2.5 km left step and curvature in fold trend. Fold amplitude is less (0.8 km), with Jurassic to Cretaceous strata exposed across the saddle. In detail, two anticlinal warps are present (Fig. 4E), which represent the tips of the Derby and Sheep Mountain folds. The forelimb dips 20°–40°SW and bends in strike to 300°/120°. Additional shortening was accommodated by the northwest- to west-northwest–striking Carr Reservoir thrust that has top-to-the-west-southwest slip and associated northwest-trending minor folds, which were tilted along the northern plunge of the Sheep Mountain anticline; complex detachment folds of Cretaceous strata with northwest- and northeast-trending axes interpreted to reflect crowding within the saddle; and the poorly exposed Red Bluff fault that thickens Triassic strata. This fault may have top-to-the-west-southwest slip based on kinematics of nearby minor faults. In comparison, Tiffany (2011) interpreted the Red Bluff fault and northeast-trending folds to record an episode of late north-south shortening.

Sheep Mountain anticline has an amplitude of 1.3 km with Pennsylvanian strata exposed in its core, and accommodated 1.0 km of shortening (Fig. 4F). The fold has a total axial length of ∼13 km, contains 2 culminations linked across a saddle marked by a 1 km left step, and ranges in trend from 350°/170° to 310°/130° from the north-central to southern parts. The fold plunges ∼10° northward along its northern end, and the southern end terminates against the steep, east-northeast–striking Clear Creek fault that has left-lateral reverse slip (Abercrombie, 1989). This fault continues to the west-southwest into the Wind River arch, subparallel to the Oregon Trail structural zone, and likely reflects reactivation of a basement weakness. The forelimb dips 40°–60°SW and is interpreted to be cut by top-to-southwest reverse faults that offset basement at depth, based on fold asymmetry similar to that of Derby Dome. A top-to-the-northeast backthrust repeats Paleozoic to Triassic strata toward the fold hinge zone, and curves along the left step into an area of complex minor folding within the forelimb. The south end of the northern culmination plunges as much as ∼30° southward along the left step.

The fold system displays systematic lateral variations in large-scale fold-fault shortening concentrated along the forelimb and internal shortening by backlimb thrusts and minor folds (Fig. 4G). Changes in fold plunge that record lateral gradients in large-scale shortening and varying styles of linkage are partly related to separation distances across saddles that represent relay zones. The saddle between the two culminations of the Sheep Mountain anticline has smaller separation (∼1 km left step), shorter overlap (<2 km), increased fold plunge (as much as 30° toward the south end of the northern culmination), and consistent large-scale shortening. The Dallas-Derby saddle has moderate separation (∼1.5 km left step), hard linkage along a steep, east-striking fault, increased fold plunge (as much as 20° toward the south end of Dallas Dome), and a decrease in large-scale shortening partly balanced by increased internal shortening. The Derby–Sheep Mountain saddle has larger separation (∼2.5 km left step), longer overlap (∼4 km) with soft linkage distributed across minor folds and faults, slightly increased fold plunge (as much as 15° toward the south end of Derby Dome), and a decrease in large-scale shortening along with a broad zone of increased internal shortening. These varying patterns are interpreted to record early linkage of closer spaced basement faults between the two culminations in the Sheep Mountain anticline; cross-strike linkage of moderately separated basement faults along an east-striking basement weakness across the Dallas-Derby saddle; and more distributed linkage between wider separated basement faults across the Derby–Sheep Mountain saddle.

Most sites in the study area had two dominant fracture sets in both red beds and limestone: a northeast- to east-northeast–striking set of extension fractures and thin (<1 cm thick) calcite-filled veins subperpendicular to fold trend (termed the cross-strike set); and a northwest- to north-northwest–striking set of fractures subparallel to fold trend and at high angles to bedding (termed the high-angle set; Fig. 5B). Some sites displayed a set of east-southeast–striking extension fractures, and some sites had additional oblique-striking fractures. Most sites in the Alcova Limestone and limestone beds of the Sundance Formation displayed minor conjugate reverse-slip (wedge) faults at acute angles (20°–30°) to bedding (Figs. 5C, 5D). Some sites also had minor conjugate strike-slip (wrench) faults at high angles to bedding with slip lineations subparallel to the intersection of faults with bedding. Minor wedge and wrench faults displayed consistent kinematic relations with respect to bedding around the large-scale anticlines and yielded best-fit σ₁ directions subparallel to slightly less steep than bedding (Fig. 6), indicating minor faults formed during early LPS, prior to, and during initiation of large-scale folding, similar to interpretations of other studies (Erslev and Koenig, 2009; Weil et al., 2014, 2016). Magnitudes of LPS, based on minor fault spacing and slip, were small (<5%). Rare minor folds and tectonic stylolites accommodated limited additional LPS. Thin cross-strike veins and conjugate wrench faults accommodated limited (<2%) tangential (strike parallel) extension. Paleostress-strain directions were estimated from minor faults for 41 sites in the Alcova Limestone and for 25 sites in the Sundance Formation. Other sites had too few (<5) measured minor faults for analysis. Paleostress and strain methods yielded nearly identical orientations of principal compression (σ₁) and shortening (ε₃) directions at the site level. LPS directions obtained from the method of Compton (1966) had smaller uncertainties (typically <5° at the 1σ level) and are reported in Table DR1. Estimated stress ratios, ϕ = (σ₂ – σ₃)/(σ₁ – σ₃), were low (<0.2) for most sites, consistent with development of both wedge and wrench faults that record switching of the σ₂ and σ₃ axes during early LPS and tangential extension. Some reverse faults within steep fold limbs cut bedding at moderate to high angles and likely formed as bedding was tilted during early to late stages of folding. Some strike-slip faults with gently raking lineations also likely formed during folding. Widely spaced faults that developed during continued deformation along steep forelimbs typically have poorly developed slip lineations.

Most sites from the study area had measureable AMS K_max lineations, even though Triassic red beds typically lacked mesoscopic evidence of internal strain. Samples had low mean susceptibility [K_m = (K_max + K_int + K_min)/3], with most core and site K_m values between 2 × 10⁻⁵ SI and 12 × 10⁻⁵ SI (Fig. 7A; Table DR2), typical for susceptibility dominated by paramagnetic minerals. AMS ellipsoids had oblate to triaxial shapes, with foliation values (F = K_int/K_min) mostly from 1.020 to 1.080, and lineation values (L = K_max/K_int) mostly from 1.002 to 1.020 (Fig. 7B). AMS ellipsoid shapes reflect a combination of primary sedimentary fabrics with magnetic foliation subparallel to bedding, and tectonic LPS fabrics that give rise to K_max lineations. SEM analysis showed preferred alignment of phyllosilicates grains (biotite, chlorite, muscovite that are paramagnetic) along bedding, and microkinking of some grains that defines a zone axis subparallel to measured K_max lineations (Figs. 7D, 7E), similar to findings by Weil and Yonkee (2009) and Weil et al. (2014, 2016). Lineation values provided a crude proxy for deformation intensity and, following Weil and Yonkee (2009), were divided into three types: type 1 with a dominant sedimentary fabric, L < 1.003, and >30° uncertainty in K_max trend; type 2 with a weak LPS fabric, 1.003 ≤ L ≤ 1.010, and a 10°–30° uncertainty in K_max trend; and type 3 with a distinct LPS fabric, L > 1.010, and <15° uncertainty in K_max trend (Fig. 7C). Of the 61 sites in the study area, 5 sites had a dominant bedding fabric (type 1), 46 sites had a weak K_max lineation (type 2), and 10 sites had a distinct K_max lineation (type 3). Detailed analysis of AMS fabrics for multiple lithologies sampled at two sites revealed broadly similar AMS fabrics between samples, with a tendency for better bedded, more micaceous layers to have stronger AMS foliations and lineations, whereas the presence of minor magnetite produced more variable AMS fabrics (Fig. DR1). Samples from intervals with varying orientations of cross-beds and ripples showed similar K_max orientations, confirming that AMS lineations are related to secondary LPS fabrics rather than to primary current lineations.

ChRM vectors were determined for 56 of 61 sites in red beds across the study area (Table DR3). Two magnetization components were observed: a low-temperature unblocking component removed upon heating to 500 °C and a high-temperature characteristic component that mostly unblocked from 600 to 680 °C (Fig. 8A). In the majority of core samples, the high-temperature component decayed linearly during progressive thermal demagnetization and provided a stable direction for analysis. Of the 56 sites with measured ChRM, 13 sites recorded mixed polarities, 25 sites recorded reverse polarity, and 18 sites recorded normal polarity magnetism. Paleomagnetic analysis was not completed for 3 sites due to disintegration of cores during sample preparation or spurious demagnetization behavior, and 2 sites displayed high dispersion of core vectors with >15° α₉₅ cones and were not used for further analysis.

A tilt test was completed to evaluate timing of ChRM acquisition. For the test, fold plunge was first removed and then bedding was progressively untilted toward horizontal. In situ ChRM vectors displayed moderate dispersion, whereas restored ChRM vectors showed peak clustering at ∼100% untilting, indicating that the ChRM was acquired prior to folding (Fig. 8B). Structurally untilted vectors had gentle inclinations that ranged mostly from 0° to 30° with a mean inclination of 14°, consistent with the Triassic paleolatitude of Wyoming (calculated using the North America polar wander curve of Domeier et al., 2011) and minor (∼5°) inclination flattening related to sediment compaction. Of the 56 sites with measurable ChRM, 44 sites lacked statistically significant differences between observed declination and expected declination for the Triassic paleopole, 10 sites had statistically significant counterclockwise vertical-axis rotation, and 2 sites had clockwise rotation.

IRM acquisition showed continuous magnetization acquisition to 2.5 T with convex-upward curves, indicating the presence of a high-coercivity magnetic mineral phase, likely hematite (Fig. 8C). Thermal demagnetization of three-axis IRM showed consistent slow decay in intensity for all three components to as much as 680° (Fig. 8D), confirming that hematite was the main carrier of remanence in red beds. Most of the remanence was carried along the axis imparted with a 2.0 T field. Some cores displayed a minor dip in intensity for the 0.15 T field at ∼560 °C, consistent with the presence of minor magnetite.

LPS directions estimated from both minor fault data and AMS fabrics, restored for bed tilt, displayed consistent relations across the study area. Within the backlimb to hinge of the fold system and the homoclinal northeast limb of the Wind River arch, site shortening directions estimated from minor faults ranged from 223° to 254° in the Alcova Limestone (Chugwater Group) and from 231° to 253° in the Sundance Formation, with an average of 241° (Fig. 9A). Shortening directions estimated from AMS fabrics in red beds gave consistent estimated shortening directions, ranging from 204° to 257° with an average of 237°. LPS directions were subparallel to the average regional Laramide shortening direction of 240° (Erslev and Koenig, 2009; Weil and Yonkee, 2012). Within fold forelimbs, minor faults were slightly better developed and estimated LPS directions ranged from 221° to 268°, partly related to map-view curvature of fold trend. Several sites in fold forelimbs contained additional minor faults that formed during continued fold growth, with shortening directions of ∼210°–260°, subperpendicular to fold trend.

The dominant cross-strike extensional fracture and vein set also displayed consistent directions across the study area for both limestone and red beds, with site mean strikes ranging from 208° to 268° in Triassic red beds, from 210° to 260° in the Alcova Limestone, and from 216° to 268° in limestone beds in the Sundance Formation (Table DR1; Fig. 9B), with respective averages of 241°, 241°, and 243°, which are subparallel to LPS directions estimated from minor fault data and AMS fabrics. The cross-strike set is interpreted to have accommodated tangential extension mostly during early LPS, and partly during later fold axis–parallel extension associated with three-dimensional (3-D) fold growth. Some sites had an east-southeast–striking fracture set, best developed near the Clear Creek fault. Similarly oriented fractures in parts of the Big Horn Basin were interpreted to have formed during early west-northwest–east-southeast compression prior to Laramide deformation (Bellahsen et al., 2006; Amrouch et al., 2010; Beaudoin et al., 2012). The tectonic significance of these fractures and comparison to calcite twin strain data are further explored in the Discussion.

Vertical-axis rotation estimated from ChRM declinations restored for bed tilt was statistically insignificant at most sites along the backlimb of the fold system. Ten sites located near the east-northeast–striking Clear Creek fault and along more west-trending parts of fold forelimbs recorded local counterclockwise rotations (Fig. 9C).

Correlations between estimated LPS directions and structural trend were evaluated using the strike-test method of Yonkee and Weil (2010b). This method uses a weighted least-squares approach, and incorporates measurement uncertainty in estimated LPS directions (typical 1σ of ∼5° for minor fault data and ∼10° for AMS fabrics) plus random dispersion from local stress-strain refraction and AMS fabric variability (taken as 4° and 6°, respectively, similar to values of Yonkee and Weil, 2010b). Site structural trend was estimated from a combination of fold axis trend and local bed strike (corrected for fold plunge), with a typical uncertainty of 5°. For strike tests, LPS directions relative to a regional shortening direction of 240° were correlated with site structural trend relative to a regional trend of 330°. A best-fit slope of 1.0 indicates that LPS directions are perpendicular to structural trend, whereas a slope of 0.0 indicates that LPS directions are parallel to the regional shortening direction regardless of structural trend. Strike tests for combined minor fault data and AMS fabrics gave a slope of 0.6 ± 0.1 (at a 95% confidence level) for sites located along the forelimbs, and a slope of 0.9 ± 0.2 for sites located along the backlimb to hinge zone of the fold system (Fig. 10A), indicating that LPS directions were partly refracted with changes in structural trend along arcuate forelimbs and partly dispersed about the regional west-southwest shortening direction along the backlimb. The mean square of weighted deviates (MSWD), incorporating dispersion from nonsystematic stress-strain refraction, was close to 1. Correlation of LPS directions estimated from AMS fabrics with directions estimated from minor fault data yielded a slope of 0.9 ± 0.1 (Fig. 10C) with an MSWD close to 1, confirming that AMS fabrics provided a robust estimate of LPS directions.

Correlations between paleomagnetic declination and structural trend were also evaluated using the strike-test method, incorporating measurement uncertainty (typical 1σ of ∼6° for paleomagnetic data) plus random dispersion from nonsystematic local block rotation (taken as 6°). For this strike test, tilt-corrected paleomagnetic declinations relative to a reference declination of 336° for the Early Triassic were correlated with site structural trend relative to a regional trend of 330° (Fig. 10B). Paleomagnetic declinations were uncorrelated with structural trend for sites located along the backlimb with an estimated slope of 0.1 ± 0.2, but were correlated with structural trend along arcuate forelimbs with an estimated slope of 0.9 ± 0.3 and a statistically significant negative intercept of −10°, consistent with limited counterclockwise rotation, partly related to sites near the Clear Creek fault.

Structural, AMS, and paleomagnetic data sets record patterns of early to late shortening and vertical-axis rotations during evolution of the D-D-SM fold system along the flank of the Laramide Wind River arch. We first summarize key constraints from these data sets. Next, orientations of paleostress-strain directions estimated from various methods (minor fault kinematics, fracture orientations, AMS fabrics, and previously published calcite twin strain data) are compared and used to evaluate the evolving local deformation field in relation to regional Laramide patterns. An integrated model is presented for the evolution of the D-D-SM fold system that is compared with idealized end-member models for development of variably trending Laramide arches and origins and linkage of flanking fold-fault systems.

Large-scale structural relations (Blackstone, 1993a), thermochronologic data (Stevens et al., 2016), and characteristics of sedimentary strata in adjacent basins (Beck et al., 1988; Steidtmann and Middleton, 1991) indicate that large-scale uplift of the Wind River arch began during the Maastrichtian and continued into the early Eocene, with most shortening from concentrated slip on the Wind River thrust (Smithson et al., 1979). Additional shortening was accommodated by folds on the northeast flank of the arch that developed mostly during the Paleocene to early Eocene (Keefer, 1970), including the D-D-SM fold system. The system comprises a series of doubly plunging, left-stepping anticlines, which are connected across structural saddles (relay zones). Surface geologic and drill-hole data indicate the anticlines have moderately to steeply southwest-dipping forelimbs cut by reverse faults that offset basement at depth (Willis and Groshong, 1993; Hilmes, 2014). Within saddles, slip on basement reverse faults and large-scale fold amplitude decrease, whereas internal shortening from backlimb thrusts and detachment folds increases. The saddles are also locally cut by steep, east-striking faults that partly connected basement reverse faults at depth and likely reactivated basement weaknesses.

Minor wedge and conjugate wrench faults, best developed in limestone, accommodated widespread but limited (<5%) LPS. These wedge and wrench faults display consistent relations with respect to bedding around large-scale folds, indicating that they formed prior to and synchronous with the onset of large-scale folding. Fold forelimbs are locally cut by additional younger minor faults that accommodated limb steepening and shear during continued shortening. A cross-strike set of fractures and veins accommodated limited tangential extension during early LPS and later 3-D fold growth. Red beds also underwent widespread, limited LPS, recorded by AMS K_max lineations. Early LPS directions estimated from minor faults and AMS fabrics, corrected for bed tilt, trend generally west-southwest–east-northeast along the backlimb of the fold system, subparallel to the regional Laramide shortening direction, and are partly refracted along arcuate forelimbs. Paleomagnetic analysis and rock magnetic experiments indicate that Triassic red beds of the Chugwater Group have a near-primary remanent magnetization carried by hematite. Paleomagnetic declinations, corrected for bed tilt, record limited counterclockwise rotations near the east-northeast–striking Clear Creek fault and along more west-trending parts of forelimbs, whereas the backlimb lacks statistically significant vertical-axis rotation.

Quantitative comparison of LPS directions estimated from AMS fabrics and from minor fault kinematics yielded a slope of 0.9 ± 0.1, indicating that AMS K_max lineations provide a reliable method to estimate shortening directions. SEM imaging and detailed analysis of different lithologies with varying sedimentary structures (planar bedding, cross-bedding, and ripples) at two sites revealed similarly oriented K_max lineations subparallel to the zone axis of microkinked phyllosilicate grains. In detail, AMS K_max directions had slightly greater variability compared to minor fault LPS directions, possibly due to contributions from multiple paramagnetic and ferromagnetic minerals that deformed differently. Average LPS directions of 240° and 237° estimated from minor fault data and AMS fabrics and average shortening directions of 241° to 243° estimated from extension fractures in Triassic to Jurassic red beds and limestone are all consistent with the regional average ∼240° shortening direction for the Laramide belt (Erslev and Koenig, 2009; Weil and Yonkee, 2012).

Previously published calcite twin data, in comparison, record more complex patterns during evolution of the fold system. Principal shortening directions for calcite twin strain reported by Willis and Groshong (1993) varied from west-southwest–east-northeast for samples near culminations of the D-D-SM system, interpreted to record initiation of Laramide LPS, to west-northwest–east-southeast for samples located away from fold culminations and along the homoclinal limb of the Wind River arch, interpreted to record pre-Laramide stress. Shortening directions estimated from calcite twin strain in limestone samples at Derby Dome reported by Craddock and Relle (2003) were mostly oriented north-northwest–south-southeast, subparallel to the fold axis, and were interpreted to record major rotation (>60°) of older Sevier-related east-west shortening fabrics during younger Laramide deformation. Paleomagnetic data for red beds here, however, indicate that vertical-axis rotation was insignificant, and minor fault kinematics, cross-strike veins, and AMS fabrics indicate mostly west-southwest–east-northeast LPS directions. The unusual shortening direction for calcite twin strain in limestone and variable shortening directions for calcite twin strain from vein samples at Derby Dome reported by Craddock and Relle (2003) could reflect local stress heterogeneities during folding.

The west-northwest–east-southeast shortening direction for calcite twin strain away from fold culmination is consistent with the orientation of the east-southeast–striking extensional fracture set, and is subparallel to the direction of North American plate motion during the mid-Cretaceous (Torsvik et al., 2008), indicating that these structures may reflect plate interior stresses. The east-southeast–striking fracture set is best developed near the Clear Creek fault that is along the basement Oregon Trail structural zone, which marks a change in lithospheric structure that may have focused initial fracturing. The change in calcite twin strain directions near future culminations may indicate local reorientation of the stress field above incipient basement reverse faults during initiation of Laramide deformation as a flat-slab segment began moving northeastward toward the foreland. During continued flat-slab subduction, deviatoric stress increased and stress directions became regionally reoriented west-southwest–east-northeast, at low angles to the direction of relative motion between the North American and Farallon plates (Wright et al., 2016), leading to early LPS with development of minor fault networks and AMS fabrics, followed by large-scale fault propagation and folding.

Similar patterns are observed regionally in the Laramide foreland. For example, calcite twin and fracture data from folds in the adjacent Big Horn Basin were interpreted to record pre-Laramide east-southeast–west-northwest–oriented principal compression, followed by increasing deviatoric stress with west-southwest–east-northeast–oriented early LPS and later large-scale folding and fault propagation during Laramide deformation (Amrouch et al., 2010, 2011; Beaudoin et al., 2012). Craddock and van der Pluijm (1999) compared calcite twin strain data from the Sevier belt, Laramide foreland, and continental plate interior (based on limited exposures of mid-Cretaceous limestone), which they interpreted to record early Sevier east-west principal compression that decreased in magnitude toward the plate interior, followed by early Laramide west-southwest–east-northeast shortening, and late Laramide north-south shortening recorded in veins.

Although early east-southeast– to east-oriented paleostress-strain directions in the foreland have been interpreted as Sevier, these early calcite twin strains may instead record plate interior stresses related to lithosphere basal drag subparallel to the direction of North American plate motion during the mid-Cretaceous. The length scale for development of LPS fabrics (cleavage and tectonic stylolites) in front of the Sevier wedge was ∼50–100 km and Sevier shortening was oriented overall west-east with radial dispersion likely related to topographic stresses along the curved wedge front (Mitra and Yonkee, 1985; Yonkee and Weil, 2010a; Weil and Yonkee, 2012). Thus, we restrict use of the term Sevier to those features that formed near the fold-thrust wedge front. Mid-Cretaceous Sevier deformation occurred during increased relative plate motion and growth of an orogenic wedge, and also during increased west-northwest to west drift of the North American plate that may have enhanced basal drag and slightly increased plate interior stresses. Sevier deformation in the Wyoming salient continued into the later Cretaceous to early Eocene, temporally overlapping with Laramide deformation, but regional shortening directions differed, with west-southwest–east-northeast Laramide shortening interpreted to reflect increased plate coupling during flat-slab subduction (Weil and Yonkee, 2012).

An integrated model for evolution of the D-D-SM fold system, based on large-scale structural geometry (Fig. 4), minor fault kinematics, AMS fabrics (Fig. 9A), fracture set and calcite twin strain directions (Fig. 9B), and paleomagnetic declinations (Fig. 9C), comprises three phases: (1) pre-Laramide west-northwest–east-southeast compression and development of east-southeast–striking extension fractures and calcite twin strains away from future fold culminations; (2) early Laramide west-southwest–east-northeast LPS with development of minor wedge and conjugate wrench faults, AMS K_max lineations, cross-strike veins and fractures, and additional calcite twin strains; and (3) later Laramide propagation and linkage of basement faults and folds (Fig. 11). During phase 1, the regional stress field is interpreted to have been oriented subparallel to the direction of absolute plate motion and had relatively low differential magnitude, with locally enhanced fracturing along preexisting lithospheric structures such as the Oregon Trail structural zone. The stress field was locally reoriented to west-southwest–east-northeast above basement weaknesses (sites of future fold culminations) during incipient Laramide deformation as a flat slab began propagating toward the foreland. Incipient deformation may have started ca. 75–70 Ma, based on subtle changes in thickness of late Campanian strata near future culminations in southwest Wyoming (Lopez and Steel, 2015). During phase 2, enhanced plate coupling related to flat-slab subduction led to increased deviatoric stress and regional reorientation to west-southwest–east-northeast compression, subparallel to the direction of relative plate motion, along with local stress-strain refraction related to basement weaknesses and propagating faults. LPS fabrics developed prior to and during onset of large-scale folding. During phase 3, an en echelon fold-fault system formed as basement reverse faults propagated upward into steepening forelimbs, and laterally with partial connection across structural saddles (relay zones) locally cut by steep, east-striking faults. Backlimb thrusts and detachment folds helped accommodate shortening transfer across saddles. Early LPS fabrics were tilted around large-scale folds that underwent additional minor faulting along steeper parts of forelimbs. Large-scale folding in the D-D-SM system ended by 50 Ma (Keefer, 1970).

The integrated model is now compared with end-member models for development of varying Laramide arch trends, styles of flanking folds, and nature of fault-fold linkage shown in Figure 2. LPS directions estimated from minor faults, AMS fabrics, and cross-strike extensional fractures and veins record a single, smoothly varying paleostress-strain field with west-southwest–east-northeast shortening across the backlimb and partial refraction of early shortening along arcuate forelimbs, consistent with model 3 in Figure 2A. A component of reverse slip on the east-striking Clear Creek fault and local development of northeast-trending detachment folds in saddles between culminations were interpreted by Tiffany (2011) to record an episode of late north-south shortening, consistent with model 1. However, detachment folds in saddles may reflect local constrictional deformation, and east-striking faults accommodated significant left-lateral slip and connected top-to-the-southwest reverse faults across relay zones, consistent with model 3. The development of east-striking faults that connected northwest-trending folds and reverse faults and the influence of east-trending basement weaknesses were also important in other parts of the Laramide belt (Paylor and Yin, 1993; Neely and Erslev, 2009; Weil et al., 2016). Paleomagnetic data indicate insignificant vertical-axis rotations, except near the east-northeast–striking Clear Creek fault and more west-striking parts of forelimbs, inconsistent with widespread sinistral transpression and counterclockwise vertical-axis rotation expected for left-stepping en echelon folds in model 2.

Most shortening in the D-D-SM fold system was accommodated by reverse faults that offset basement at depth and propagated into steep forelimbs in the cover, consistent with a trishear model (Fig. 2B). Dallas Dome has the steepest forelimb with reverse faults that reach the surface, Derby Dome and the Sheep Mountain anticline have moderately steep forelimbs cut by reverse faults at depth, and structural saddles between culminations have less steep forelimbs with more limited faulting at depth (Figs. 4B–4F). These variations are interpreted to reflect decreasing ratios of upward fault propagation to fault slip rate. Large-scale fold-fault shortening also decreases within saddles, whereas internal shortening accommodated by backlimb thrusts and detachment folds increases, such that total shortening only varies slightly along the strike of the system (Fig. 4G). These relations are consistent with lateral propagation and varying linkage of fault segments related to separation distances across saddles (Fig. 2C).

Although the results of this study are limited, they can be compared to fault linkage and displacement relations reported for other fault systems and relay zones. Numerous studies have analyzed relay zones within normal fault systems (see reviews by Peacock, 2002; Fossen and Rotevatn, 2016), but studies of relay zones within reverse fault and associated fold systems are more limited (Walsh et al., 1999; Nicol et al., 2002). Davis et al. (2005) described varying geometries of fault segment overlap, secondary faulting, and associated folding that helped accommodate shortening transfer within the youthful Ostler thrust fault system in New Zealand, which recorded varying linkage styles related to fault segment separation, similar, but at shorter temporal and spatial scales compared to the D-D-SM system. Mazzoli et al. (2005) described en echelon fault-fold segments along a more mature thrust system in the Apennine belt of Italy, which, similar to the D-D-SM system, was marked by varying linkage and increased lateral displacement gradients across relay zones. Such lateral linkage leads to longer fold-fault systems that may preferentially accrue continued shortening. Relations between fold-fault length and maximum shortening for the D-D-SM system plot within the length-maximum displacement field for reverse faults (Fig. 12; Torabi and Berg, 2011). Relations between the more youthful Ostler fault system, the D-D-SM system, and major reverse faults, such as the Wind River thrust, are interpreted to reflect multiple scales and cycles of fault propagation, interaction, and linkage, similar to the numerical model of Cowie et al. (2000) for linkage of normal fault segments into long systems that focus deformation.

Lateral variations of the D-D-SM system can also be compared with the numerical models of folding by Grasemann and Schmalholz (2012) that predict lateral propagation and partial linkage to form en echelon, doubly plunging folds for moderate ratios of fold separation to fold width. Fold culminations in the study area have widths of ∼4–5 km (measured from the syncline trough on the southwest side to where dip shallows on the northeast backlimb) and separations of ∼1–2.5 km between axial traces, giving moderate separation-to-width ratios of ∼0.2–0.6, consistent with the numerical models.

Integrated structural and paleomagnetic data sets record changing paleostress-strain patterns and partial fault linkage during evolution of a flanking fold system along the Laramide Wind River arch. Locally developed east-southeast–striking fractures and early calcite twin strains record pre-Laramide west-northwest–east-southeast compression, interpreted to reflect slightly enhanced plate interior stresses during increasing west-northwest drift of the North American plate. Minor wedge and conjugate wrench faults and AMS fabrics record early Laramide west-southwest–oriented LPS prior to and during onset of large-scale folding, interpreted to reflect increasing deviatoric stress during northeastward propagation of a flat-slab segment that enhanced end loading, basal traction, and/or asthenospheric flow. During progressive deformation, shortening became concentrated along basement reverse faults, which propagated upward into steepening forelimbs, and laterally with partial linkage across structural saddles (relay zones) locally cut by eastward-striking faults along basement weaknesses. Separation distances of fault segments and the presence of basement weaknesses partly controlled linkage of segments into a longer fold-fault system, which may serve as an analogue for linked reverse fault-fold systems in other orogenic belts.

We thank Peiyeng Wen, Jamie Kendall, Amelia Lee Zhi Yi, Mary Schultz, Fern Beetle-Moorcroft, and Andriy Mshanetskyy for help in the field and laboratory. Student summer support was partially funded by the Bryn Mawr College Summer Science Research Fellowship program and its funding agencies and offices. We also thank John Spence and Betsy Spence for their generosity and access to important outcrops. Careful and constructive reviews by two anonymous reviewers improved this manuscript. This work was supported by National Science Foundation grants EAR-0948677 and EAR-0948692.