The widespread presumption that the Farallon plate subducted along the base of North American lithosphere under most of the western United States and ∼1000 km inboard from the trench has dominated tectonic studies of this region, but a number of variations of this concept exist due to differences in interpretation of some aspects of this orogeny. We contend that five main characteristics are central to the Laramide orogeny and must be explained by any successful hypothesis: thick-skinned tectonism, shutdown and/or landward migration of arc magmatism, localized deep foreland subsidence, deformation landward of the relatively undeformed Colorado Plateau, and spatially limited syntectonic magmatism. We detail how the first two elements can be well explained by a broad flat slab, the others less so. We introduce an alternative hypothesis composed of five particular processes: (1) a more limited segment of shallowly subducting slab is created by viscous coupling between the slab and the Archean continental keel of the Wyoming craton, leaving some asthenosphere above most of the slab; (2) dynamic pressures from this coupling localize subsidence at the edge of the Archean Wyoming craton; (3) foreland shortening occurs after the subsidence of the region decreases gravitational potential energy, increasing deviatoric stresses in lithosphere beneath the basin with no change to boundary stresses near the subduction zone or changes to basal shear stress; (4) shear between the slab and overriding continent induces a secondary convective system aligned parallel to relative plate motion, producing the Colorado Mineral Belt above upwelling aligned along the convection cell; (5) the development of this convective system interrupts the flow of fresh asthenosphere into the arc region farther west, cutting off magmatism even in segments of the arc not over the shallowly dipping slab.

The Late Cretaceous to early Tertiary Laramide orogeny that affected much of southwestern North America was not only a major mountain building event, but also coincided with major shifts in the distribution of both igneous activity and sedimentary depocenters. Despite the critical role the Laramide orogeny played in the geologic evolution of western North America, its origin remains enigmatic (e.g., English and Johnston, 2004). This orogeny is particularly difficult to place in a conventional plate tectonic framework, largely because it involved the development of basement-cored mountain belts in the interior of the western United States, ∼1000 km inboard of the convergent plate margin that defined the western edge of the continent at the time. What were the forces acting on the continental lithosphere and in what tectonic setting were they generated?

Current tectonic models of the Laramide orogeny are based largely on the observation that magmatism ceased along the western margin of southwest North America in Late Cretaceous time, but then apparently migrated inland through the early Tertiary in concert with the development of the basement cored mountain ranges. Earlier workers suggested that the inland spread in magmatism and crustal deformation were ultimately the product of a Late Cretaceous to early Tertiary shallowing in the angle of subduction of oceanic lithosphere beneath the North American continent (Coney and Reynolds, 1977; Lipman et al., 1971; Snyder et al., 1976). Despite some notable subsequent attacks (Mutschler et al., 1987; Livaccari, 1991; Maxson and Tikoff, 1996), the “flat-slab” model has survived over the past 30 years, even if the relationship between slab angle and the distribution and style of crustal deformation, magmatism, and sedimentation remains unclear. However, geologic evidence from the southwestern United States obtained over the past few decades has suggested that Late Cretaceous to early Tertiary low-angle subduction and attendant subduction erosion only affected only a narrow (∼200 km) swath of oceanic lithosphere underthrusting present-day southern California (Barth and Schneiderman, 1996; Saleeby, 2003). If so, why was the shallowly subducting lithosphere so restricted spatially, and how could such narrow “flat-slab” produce the array of geologic features generally attributed to the Laramide orogeny throughout western North America?

To address these issues we explore in this paper the possibility that the Laramide orogeny in the western U.S. was a response to the interaction between subducting oceanic lithosphere and the thick continental lithosphere of the Archean Wyoming craton. We integrate age and compositional information for Late Cretaceous and early Tertiary igneous rocks and sedimentary successions with numerical modeling of the subcontinental mantle to demonstrate that an essentially hydrodynamic model of slab–continental lithosphere interaction can account much of what is currently known about the geologic evolution of the Laramide orogeny.

The Laramide orogeny encompasses the early Tertiary creation of basement-cored thrust-bounded uplifts in the western United States; for our purposes we exclude coeval but dominantly thin-skinned deformation to the north and south. Very limited magmatism occurred within this orogen, principally along a linear trend across Colorado, the Colorado Mineral Belt (COMB). The region most strongly affected by Laramide-age deformation can be split into two distinct subregions. The Southern Rocky Mountains in Colorado and New Mexico are built from Paleoproterozoic basement terranes that were strongly deformed in the late Paleozoic Ancestral Rockies orogeny. The uplifts in Wyoming and adjacent areas to the north disturbed Archean crust that had lain virtually undeformed since the middle Proterozoic. The Laramide uplifts extend well into North America, more than 1000 km inland from the continental margin volcanic arc active prior to the Laramide orogeny and more than 500 km inland from the eastern edge of the thin-skinned fold-and-thrust belt of the Sevier orogen active prior to the Laramide orogeny (e.g., Burchfiel et al., 1992; DeCelles, 2004; Burchfiel and Davis, 1975) (Fig. 1). This eastward shift is most dramatic in the southern part of the Laramide orogen, where basement involved thrusts first appear at the surface 500 km east of the easternmost thrusts of the thin-skinned Sevier orogen, leaving a more mildly deformed Colorado Plateau to separate the Southern Rocky Mountains from the thrusts of central Utah.

The main episode of deformation started in the Maastrichtian (66–75 Ma), usually dated by the appearance of local basins and locally derived clastic rocks, the disappearance of marine sedimentary facies (e.g., Dickinson et al., 1988), and the initiation of igneous activity within the continental interior (Cunningham et al., 1994; Mutschler et al., 1987). This must be regarded as a minimum age; there are suggestions that some Laramide uplifts were influencing sedimentary depositional patterns by 76–80 Ma (Cather, 2004; DeCelles, 2004; Guiseppe and Heller, 1998). Deformation continued until ca. 30–45 Ma (e.g., Dickinson et al., 1988; Gries, 1983; Bird, 1998; Cross, 1986). Although there are differences between different workers in inferred slip azimuths and timing on individual structures (e.g., Bird, 1998; Gries, 1983; Johnson and Anderson, 2009), the overall orientation of shortening across the orogen is northeast-southwest (Bird, 1998; Erslev, 1993). Total shortening estimates have ranged from 43 to 120 km across the Wyoming part of the orogen (Bird, 1998; Chapin and Cather, 1983).

Several events outside the Laramide orogen or prior to it could be relevant to understanding the cause of this event. Most notably, igneous activity in the long-lived volcanic arc along the western margin of the United States was shut off from north of southern Arizona and New Mexico to near the Canadian border in the Late Cretaceous (Armstrong and Ward, 1991; Chen and Moore, 1982) (Fig. 2). Arc magmatism continued outside this gap; within this 1200-km-wide gap, magmatism was limited principally to the COMB prior to ca. 60 Ma (Figs. 2D, 2E). We agree with earlier workers that melt generation that led to the COMB likely took place in the upper mantle (Lipman et al., 1971; Mutschler et al., 1987; Snyder et al., 1976), although we recognize that radiogenic isotope data from “calc-alkaline” Laramide intrusive igneous rocks in the COMB require a large lower crustal component in these rocks (Simmons and Hedge, 1978; Stein and Crock, 1990). We base this assertion on new chemical and isotopic data from remnants of Laramide-age basaltic lavas flows and from mafic clasts in Laramide volcaniclastic units in Colorado that confirm that COMB magmatism was ultimately related to a mantle melting event (Bailley, 2010). Published age determinations consist exclusively of relatively low precision mineral or whole-rock K-Ar, or fission track apatite and/or zircon, ages that have complications associated with reheating by younger magmatism or postemplacement hydrothermal activity (Cunningham et al., 1994). Thus available geochronology cannot distinguish between synchronous Laramide magmatism across much of the COMB or some time-transgressive pattern (Mutschler et al., 1987).

Deformation in the Sevier fold-and-thrust belt also seems to have waned in this time, especially in southern Nevada and nearby Utah, where thrusting appears to have halted by 87–81 Ma, prior to the start of Laramide tectonism (Barth et al., 2004; Burchfiel and Davis, 1977), and in the Maria fold and thrust belt of southeastern California and western Arizona, which shortened until 85 Ma (Barth et al., 2004; Boettcher et al., 2002; Karlstrom et al., 1993) and possibly until 70 Ma on the Mule Mountains thrust (Tosdal, 1990; Tosdal and Stone, 1994). Farther north, shortening continued on frontal thrusts through most of the Laramide (DeCelles, 2004; DeCelles and Coogan, 2006), ceasing and then reversing into extensional faulting toward the close of the Laramide (Constenius, 1996).

Exhumation of deep crustal rocks and surface normal faulting has been documented in the latest Cretaceous in the Sevier hinterland (Wells and Hoisch, 2008; Druschke et al., 2009a, b; Hodges and Walker, 1992; Wells et al., 1990). The timing of shortening in central Arizona is less certain. Considerable evidence of major early-Late Cretaceous shortening exists (e.g., Krantz, 1989), but whether later deformation is contractional and/or extensional remains unclear (e.g., Livaccari, 1991).

Much of the Rocky Mountain foreland that underwent Laramide shortening had subsided substantially immediately prior to shortening, accumulating more than 2 km of Late Cretaceous sediments, but sedimentary cover around this region indicates that this subsidence was not caused by surface loading as in the foredeep to the west (Bird, 1984; Cross, 1986; Cross and Pilger, 1978b; Liu and Nummedal, 2004). This subsidence occurred within a much broader zone of less intense subsidence, the Western Interior Seaway, which has often been linked to shallowing subduction under all of western North America (Burgess et al., 1997; Mitrovica et al., 1989; Pang and Nummedal, 1995 [Fig. 2]).

Three main processes have been hypothesized to contribute to the Laramide orogeny: collision with an exotic terrane (Maxson and Tikoff, 1996; Silver and Smith, 1983; Hildebrand, 2009), collapse of the hinterland of the Sevier orogen (Livaccari, 1991), and shallow subduction of oceanic lithosphere (e.g., Bird, 1984; Dickinson and Snyder, 1978). The first two ideas have significant problems that have led them to be marginalized. Collision requires stresses to be greater at the margin than the continental interior unless stresses can be focused; yet little if any contractional deformation was produced in the Great Valley–Sierra Nevada–Southern Nevada region in this time frame (Barth et al., 2004; Burchfiel and Davis, 1977; Unruh et al., 2007). Absence of significant deformation requires either greater lithospheric strength than in the continental interior or increased gravitational potential energy to offset the compressional stresses. Neither condition has been demonstrated across the entire distance from trench to foreland, although increased potential energy might occur in parts of the Sevier hinterland (Livaccari, 1991; Wells and Hoisch, 2008). Additional difficulties arise in trying to synchronize the timing of arc shutoffs with the presumed migration of the colliding Insular superterrane, especially in light of Late Cretaceous intrusive igneous rock ages acquired after publication of Maxson and Tikoff (1996) in the Mojave Desert and Penninsular Ranges by Barth et al. (2004) and Moniz et al. (2007) (after Maxson and Tikoff, 1996) and the north-to-south younging of emplacement of underthrust oceanic assemblages (Rand, Pelona, and Orocopia Schists; Grove et al., 2003).

The collapse process essentially requires that the Sevier hinterland rise up enough that stresses in the hinterland are extensional. This allows the region to undergo highly compressive horizontal stresses from the plate margin without contraction; if the hinterland has risen through foundering of mantle lithosphere (e.g., Wells and Hoisch, 2008), then overall horizontal stresses might be higher than prior to uplift. This process requires the Colorado Plateau to be stronger than the Wyoming craton, a presumption discussed more below. It also fails to explain the shutdown in frontal thrusts in southern Nevada at a location presumably subject to the greatest increase in compressional stress. The collapse hypothesis as envisioned by Livaccari (1991) decouples Laramide shortening from changes in arc magmatism, which requires the changes in the distribution of magmatism to be an irrelevant coincidence. Wells and Hoisch's (2008) vision of this process integrates it with a somewhat more limited shallowing of the slab; arc volcanism is disrupted to differing degrees by the foundering of mantle lithosphere thrust back under the arc and the change in slab geometry. It is thus less clear why this process was limited to the western U.S. and not northern Mexico and southernmost Canada.

Hildebrand (2009) posited that western North American arcs were allochthonous and sutured to North America just before the Laramide orogeny; in this view, the Sevier orogeny was the culmination of the collision and the Laramide orogeny occurred as east-dipping subduction developed under North America. As with the other ideas outlined here, the Colorado Plateau is presumed to be strong. The shutdown of the Sevier belt in southern Utah and Nevada is not explained, but is probably tied to changes in the foundering of the west-dipping slab extending off of North America. Similarly, the absence of arc volcanism in the bulk of the western U.S. is ascribed to some unspecified difficulty in penetrating the lithosphere of this region. Many questions remain on the testable predictions of this model.

For these reasons, the last hypothesis, that deformation arose as subducting oceanic lithosphere shallowed under the United States, has emerged as the preferred alternative (English et al., 2003; Humphreys et al., 2003; Spencer, 1996; Schmid et al., 2002; Humphreys, 2009). The initial interpretation of a gently dipping slab associated with the Laramide orogeny grew out of inferred relations between the K2O content of volcanic rocks and slab depth and the chemistry of Laramide igneous rocks in the Southern Rocky Mountains (Lipman et al., 1971). Although the chemical basis for this inference has been abandoned (Meijer and Reagan, 1983; Mutschler et al., 1987), the general idea that the Laramide orogeny was related to an episode of shallow subduction continued to gain support due to favorable comparison with the basement-cored uplifts of the Sierra Pampeanas in Argentina, which are inboard of an inactive segment of the Andean arc and above a shallowly dipping segment of the subducting plate (e.g., Dickinson and Snyder, 1978; Humphreys et al., 2003; Jordan and Allmendinger, 1986; Barazangi and Isacks, 1976). The physical underpinnings of the flat-slab hypothesis arise from the balance of basal shear tractions by compressional stresses increasing with distance from the trench; such compressional stresses can drive crustal contraction and thickening (Bird, 1984, 1988; Dickinson and Snyder, 1978). A second geodynamic implication supporting shallowing subduction is the inference that such a change in subduction geometry will produce subsidence in the continental interior, consistent with the creation of the Western Interior Seaway (Burgess et al., 1997; Mitrovica et al., 1989). Although some alternatives have been advanced, the main discussion over the past 20 years has focused on the precise geometry for shallow or flat-slab subduction (e.g., Saleeby, 2003), the magnitude of shear stress applied to North America, and the means by which the slab was removed (e.g., Humphreys, 1995). Most workers have presumed (frequently implicitly) that nearly the entire western U.S. had lain over a flat slab in the early Tertiary, much as shown by Bird (1984, 1988) and Jordan and Allmendinger (1986). The presence and removal of this slab have been implicated as causing much of the tectonism of the western U.S. in the Cenozoic (e.g., Humphreys et al., 2003; Humphreys, 1995).

Although the cessation of magmatism along the Pacific margin was widespread, evidence that the slab shallowed is far more limited. Mesozoic garnet peridotites from paleodepths of 100 km or more as measured by garnet geobarometry were preserved into the Miocene under the Sierra Nevada arc (Ducea and Saleeby, 1996, 1998); to the south, in contrast, oceanic schists were juxtaposed against the crust of the Mesozoic arc in the latest Cretaceous and early Tertiary, as constrained by protolith and cooling ages (Grove et al., 2003; Saleeby, 2003; Malin et al., 1995; Burchfiel and Davis, 1981; Jacobson et al., 1988). If representative of deeper levels in the earth (as suggested by Luffi et al., 2009), this suggests that the along-trench extent of extremely shallow subduction was much more limited than previously inferred (Saleeby, 2003).

Only Bird's (1988) specific geodynamic version of the flat slab has provided quantitative predictions at a lithospheric scale from trench to foreland; it is based on a review of the physical relations of several aspects of the hypothesis (Bird, 1984). In developing his model, Bird sought to not only produce deformation far to the east of earlier shortening, but also tried to produce the thick crust found today in Colorado by entraining it in the shear between plates. We briefly deconstruct the four main elements of the flat-slab hypothesis as envisioned by Bird and their physical basis: production of large horizontal normal stresses well into the continent, creation of unusually thick crust well into the continent, subsidence of the Rocky Mountain foreland, and the cessation of arc volcanism.

The most elegant aspect of the original flat-slab hypothesis is that stresses are readily conveyed past the weak regions near the plate boundary and focused in the plate interior (Bird, 1988; Dickinson and Snyder, 1978). The subducting slab is in contact with the overriding plate; the resulting shear stresses are posited to have existed along the entire distance from near the trench to the Southern Rockies. The stresses transmitted would parallel the relative plate motion, which could explain the difference between the east-directed Sevier belt and the northeast-southwest shortening in the foreland (Bird, 1998; Erslev, 1993). The equation of equilibrium reveals that the horizontal normal stresses will be maximized where the basal shear stops, which is where the large uplifts are found (Fig. 3A). However, for this process to deform foreland lithosphere, there must be a significant shear stress on the base of the North American plate.

The effects of this basal shear on the internal structure of the North American plate depend on the rheology of the lithosphere. By positing a weaker lower crust than upper crust or upper mantle, Bird (1988) was able to effect the movement of large volumes of lower crust from west to east, movement accompanied by wholesale removal of mantle lithosphere through most of the western U.S. This movement of crust created thicker crust in the Southern Rockies and High Plains. A secondary cause of this flow of lower crust could have been basal suction forces associated with emplacement of an antibuoyant slab, although Bird (1984) argued that this is reversible and too local to explain the difference between modern crustal thicknesses and a roughly uniformly 35-km-thick crust inferred to exist prior to the Laramide.

The emplacement of an antibuoyant slab has been inferred to be the cause of subsidence of the continent above the flat slab (Bird, 1984; Cross, 1986; Cross and Pilger, 1978b). This requires that the shallowing of the slab occurs because of an instability in the dynamic pressures generated by flow in the asthenospheric wedge (Tovish et al., 1978). A broad correlation exists between the inferred extent of the flat slab and Late Cretaceous isopachs (Bird, 1984). Others have inferred that subduction of buoyant oceanic crust (e.g., Henderson et al., 1984; Livaccari et al., 1981; English et al., 2003) caused subduction to shallow, although this is apt to cause uplift of the overriding plate instead of subsidence (van Hunen et al., 2004), and such basalt-rich crust will convert to eclogite (Kelleher and McCann, 1976), negating any special ability to cause flat subduction (Cahill and Isacks, 1992). Liu et al. (2010) suggested that this discrepancy is solved if the oceanic plateau converts to eclogite as it travels under the continent.

The cessation of continental margin magmatism that accompanied the Laramide orogeny is also consistent with a period of shallow subduction. Arc magmatism is generally considered to be the result of melting in the mantle wedge initiated by the influx of volatiles produced during subduction of oceanic lithosphere (Davies and Stevenson, 1992; Gaetani and Grove, 2003; Gill, 1981). Although the dewatering reactions probably occur over a significant depth range (Schmidt and Poli, 1998), modern volcanic arcs are over Benioff zones from 65 to 130 km deep, the main variation apparently caused by changes in convergence speed (England et al., 2004). Thus as a slab shallows, arc magmatism should migrate inboard. Beneath some segments of the modern Andes, oceanic lithosphere is subducting at such shallow angles that subduction-related continental margin magmatism has been completely extinguished (Gutscher et al., 2000; Barazangi and Isacks, 1976).

Thus the four main effects of shallow subduction, i.e., foreland deformation, foreland crustal thickening, foreland subsidence, and migration of the volcanic arc, are in fact mostly the products of four somewhat separable components of the system: magnitude of basal shear stress, weaknesses within the lithospheric rheology, antibuoyancy of the slab, and geometry at and above the top of the subducting slab. A product of shallow (but not necessarily flat) subduction not originally appreciated is the possible creation of dynamic topography (i.e., topography out of isostatic equilibrium owing to stresses induced by fluid flow) (Mitrovica et al., 1989). Such dynamic topography was proposed to explain the expansion of the Western Interior Seaway from its foredeep position earlier in the Cretaceous to extend well into the continent (Burgess et al., 1997; Mitrovica et al., 1989). The magnitude of dynamic topography depends on the flow field in the asthenosphere, which in turn depends on the rheology of the asthenosphere and the geometry and rate of subduction (Burgess et al., 1997; Mitrovica et al., 1989; McKenzie, 1969).

Although a flat slab is widely presumed, the cause of any shallow subduction remains unresolved. A slab might shallow if buoyant material is subducted, if the overriding plate accelerates, or through hydrodynamic forces in the asthenospheric wedge (slab suction) (van Hunen et al., 2004). Three specific events have been proposed for the Laramide: subduction of a buoyant plateau or young lithosphere (Henderson et al., 1984; Barth and Schneiderman, 1996; Livaccari et al., 1981; Saleeby, 2003), acceleration of North America (Cross and Pilger, 1978a), and changes in the velocity and/or age of the subducting slab (Engebretson et al., 1984). For our purposes, the chief difference between these is the extent and timing of the effects of flat-slab subduction. For example, the subduction of a buoyant plateau predicts that changes in the interior of the North American plate due to basal shear will lag effects at the margin by several million years (e.g., Espurt et al., 2008), while changes in the motion of North America will be more instantaneous but should affect the entire margin.

Any alternative explanation of the Laramide should explain patterns of igneous activity, sedimentation, and tectonism, addressing both issues well covered by the existing flat-slab scenario and elements poorly understood in the existing framework, such as the timing of subsidence in the foreland and the evolution of the COMB. We outline such a hypothesis before discussing the components in detail (Fig. 5).

We posit that subduction did shallow along the entire western margin of North America, but that an Archean keel under the Wyoming craton narrowed asthenospheric counterflow, leading to an unusual amount of shallowing of the slab under part of the western U.S., more or less as delineated by Saleeby (2003), but not necessarily to the base of the lithosphere. We suggest that the flow field creating the stresses pulling up the slab also created a basal normal force on North America that created the peculiar Maastrichtian basin in Wyoming and Colorado. We accept the logic of Jones et al. (1998) in suggesting that the forces generated by this subsidence localized subsequent deformation far in the foreland. We suggest that this pattern of flow in the asthenosphere led to the cessation of arc volcanism in the latest Cretaceous and early Tertiary by disrupting flow of fertile asthenosphere to the Sierra Nevada arc, as suggested by Saleeby (2003). Finally, we propose that shear-parallel convection generated within this subduction system produced the northeast-trending COMB.

Shallowing of Subduction

We concur with others (Cross and Pilger, 1978a; DeCelles, 2004) that North America accelerated westward relative to the Farallon plate in the Late Cretaceous; this motion led the entire subducting slab to dip more shallowly (Fig. 5A). Unfortunately plate circuit reconstructions of Farallon–North America motion barely include the latest Cretaceous (Doubrovine and Tarduno, 2008; Stock and Molnar, 1988), and hotspot-based reconstructions (e.g., Engebretson et al., 1984, 1985) include considerable errors that limit interpretation (e.g., Doubrovine and Tarduno, 2008; Stock and Molnar, 1988). Accelerated motion of North America is supported both by the eastward migration of all the arcs in the western U.S. and northern Mexico (e.g., Armstrong and Ward, 1991; Lipman, 1992; Dickinson, 2006; Clark et al., 1982) (Fig. 1) and by estimates of opening across the North Atlantic Ocean (Fig. 4). An abrupt counterclockwise rotation of North America accompanied southward motion starting ca. 75 Ma (as shown by its apparent polar wander path; Torsvik et al., 2001); this and the smaller migration of the arc in southern Mexico (Henry et al., 2003) suggest that accelerated westward motion was higher to the north. The arc also migrated to the east in Canada (van der Heyden, 1992; Gehrels et al., 2009), but as the position of these rocks with respect to North America in the Late Cretaceous is disputed (e.g., Krijgsman and Tauxe, 2006; Cowan et al., 1997; Butler et al., 2001; Kim and Kodama, 2004), it is not entirely clear that this reflects accelerated motion of North America. Less directly, the marine transgression in the Cretaceous that led to the Western Interior Seaway connecting the Gulf of Mexico to the Arctic Ocean is likely to have been the product of subsidence due to changes in flow in the upper mantle induced by shallowing subduction (Burgess et al., 1997; Mitrovica et al., 1989).

Interaction with Archean Wyoming Craton

Most of the Farallon plate was subducting under Proterozoic or younger lithosphere (Fig. 1), but one segment would have impinged on the edge of the Archean Wyoming craton between ∼1100 and 1400 km from the trench (Fig. 5B). If Wyoming had at this time Archean lithosphere comparable to that present today elsewhere in North America, the lithosphere could have extended to 200–300 km depth under this craton (Gung et al., 2003; Michaut et al., 2007). For this segment of slab, the space where asthenospheric counterflow occurs would be narrower than elsewhere, and so corner flow models predict that the flow gradients would be much higher and the slab suction forces increase (e.g., McKenzie, 1969; Tovish et al., 1978). The corner flow model therefore predicts that this segment of slab would be pulled up and the continent over it pulled down. On a mid-Cenozoic palinspastic base (McQuarrie and Wernicke, 2005) and well within the uncertainties of relative plate motions, the position where the sub-Wyoming slab would subduct corresponds to the reconstructed positions of the Orocopia, Pelona, and Rand Schists (Fig. 2). Based on the corner flow model, we thus suggest that the sub-Wyoming segment of the slab was pulled upward by dynamic forces and that this might have propagated upslip to the trench, generating conditions conducive to basal erosion of the continental margin and emplacement of the eugeoclinal schists.

The northern part of the Wyoming craton was probably also only a short distance from the trench in latest Cretaceous time, but it is probable that this region was above the subducting Farallon-Kula or Farallon-Resurrection spreading ridge (Breitsprecher et al., 2003; Haeussler et al., 2003; Madsen et al., 2006). The distance from the northern craton to the trench is uncertain because the large allochthons of western Canada were to the west of the northern Wyoming craton and the position of the trench is essentially unknown. For these reasons we suspect that the interaction between any lithospheric keel under the northern Wyoming craton and any slab lacked the same behavior we propose occurred farther south.

Observational constraints on the lithospheric thickness of Wyoming and the Colorado Plateau in the Late Cretaceous are few. Regional seismological work has suggested little or no unusual thickness to the Wyoming craton at present (e.g., van der Lee and Frederiksen, 2005; van der Lee and Nolet, 1997), but similar Archean cratons have thicknesses of ∼250 km (e.g., Gung et al., 2003; Michaut et al., 2007). Inferred early Tertiary hydration and alteration of the mantle under Wyoming as observed from xenoliths (Carlson et al., 2004; Eggler et al., 1987) suggest that the modern lithosphere is thinner and hotter than the pre-Laramide lithosphere (Carlson et al., 2004). Similarly, although the modern seismological estimate of a 120–150-km-thick lithosphere from surface waves (West et al., 2004) for the Colorado Plateau agrees with xenolith-based lithosphere thicknesses (Griffin et al., 2004), these could differ greatly from lithospheric thicknesses at 75 Ma: plateau xenoliths are all associated with post-Laramide magmatic activity. These xenoliths often record cooling through the Laramide (e.g., Helmstaedt and Schulze, 1991; Smith et al., 2004; Riter and Smith, 1996), so it is possible that modern lithosphere thicknesses are greater than those prior to the Laramide. Given these uncertainties, we posit that at 75 Ma Wyoming had an ∼250-km-thick lithosphere that was cold and neutrally buoyant and that the lithosphere under the Colorado Plateau was thinner and probably not sufficiently chemically distinct to be neutrally buoyant, and so less likely than Wyoming lithosphere to interfere with asthenospheric counterflow.

Creation of the Wyoming-Colorado Basin

The development of the Western Interior Seaway has been associated by some with shallowing subduction at the western margin of North America (Burgess et al., 1997; Catuneanu et al., 1997; Gurnis, 1993; Mitrovica et al., 1989; Pang and Nummedal, 1995; Liu et al., 2008). While such models are appealing in explaining the continent-wide trough that extended from the Gulf of Mexico to the Arctic Ocean, they produce very long wavelength topography, generally in excess of 1000 km across strike and longer still along strike (e.g., Burgess et al., 1997). In contrast, the Wyoming-Colorado “backbulge” basin is roughly equant and ∼500 km in diameter with sedimentary rocks about 4 times thicker than comparable distances from the trench to the north or south (Cross, 1986; Roberts and Kirschbaum, 1995)(Fig. 2C).

The same mechanism that would tend to pull up the slab subducting under the edge of a thick Archean craton would also act to pull down that craton at its edge (Fig. 5D). Such a connection, induced by the narrowing of the gap between slab and cratonal root, is plausibly a much narrower feature than exists in most published models of dynamic topography, and so potentially could reproduce the geometry of the Colorado-Wyoming basin. This basin straddles the southern edge of the Archean Wyoming craton (Fig. 2D), broadly consistent with these expectations.

To better understand the implications of this hypothesis, we have carried out some simplified two-dimensional (2-D) numerical experiments of topography in a subduction setting with a lithospheric keel (Fig. 6). Our 2-D dynamic models are similar to those of Gurnis (1993), in which the flow and dynamic topography are driven by the buoyancy force of subducted slabs, which can be estimated from the age of subducting lithosphere (Hager and O'Connell, 1981). Lithosphere (subducting and overriding) and the underlying mantle are modeled as fluids with high and low viscosities, respectively. A key feature is the continental keel structure, which is also modeled as having a high viscosity, similar to that in recent 3-D models (Becker, 2006; Conrad and Lithgow-Bertelloni, 2006; Zhong, 2001). The mantle viscosity structure is similar to that derived from the geoid studies (e.g., Hager and Richards, 1989) in which the upper mantle is 30–100 times weaker than the lower mantle. We use free slip boundary conditions for all the boundaries including the sidewalls. We use sufficiently large computational domain such that sidewall and bottom boundary conditions are not expected to affect the results.

Our calculations confirm that the presence of a stiff continental keel will influence dynamic topography for sufficiently shallow subduction, and the wavelengths of some features are similar to that of the Colorado-Wyoming basin (Fig. 6). The curve for a slab without any lithospheric keel reproduces the generally long wavelength subsidence previously used to match the Western Interior Seaway (Burgess et al., 1997). Introduction of the Archean keel provides a noticeable inboard edge to the subsidence signal, similar to that calculated by O'Driscoll et al. (2009). Introduction of an even weaker asthenospheric wedge produces a pronounced basin with a sharp edge at the edge of the Archean keel, and a broader but smaller basin extending well inboard over the craton. Such a weak asthenosphere is broadly consistent with stress-dependent rheologies and weakening of this mantle as it is hydrated by the slab (Hirth and Kohlstedt, 1996, 2003). Weak asthenosphere above the slab has also been suggested on observational grounds from modeling of topography in the western Pacific (Billen and Gurnis, 2001; Billen et al., 2003).

Our results support the general idea that the presence of a stiff keel could change the dynamic topography from the very broad basins seen without a keel to the more limited and deeper basin produced here. A more direct comparison with the sedimentary record will require examination of effects such as oblique motion, three-dimensional geometry of the slab, temporal evolution of subsidence, and large-scale mantle flow.

Some numerical experiments by O'Driscoll et al. (2009) explored concepts relevant to this hypothesis; they found that the presence of a lithospheric root will lead to enhanced downward pressure (suction) on the base of the continent trenchward of the root and an increased upward force acting on the subducting slab. They did not investigate the role of slab dip, but noted that this effect increases with proximity to the slab, which probably reflects the narrowing distance between slab and root. Thus we infer that shallower slab dips should cause increased subsidence, but only on the trenchward side of the root. Some aspects of this model are inconsistent with Laramide observations: the numerical model predicts both subsidence and increased horizontal compression over the entire trench-root distance, which seems inconsistent with the cessation of Sevier-style shortening in southernmost Utah and Nevada, the appearance in much of the Sevier hinterland of extension, and the emergence of the Sevier foredeep in latest Cretaceous time (Fig. 2). We suggest that the O'Driscoll et al. (2009) geometry better fits North America during the Sevier orogeny. We suggest that once the Farallon slab shallowed enough to interfere with asthenospheric counterflow, the dynamic subsidence driven by high dynamic pressures (suction) in the Sevier asthenospheric wedge ceased. The resulting rise in the Sevier hinterland permitted the development of extensional deformation much as envisioned by Wells and Hoisch (2008), although the uplift is caused by a change in dynamic forces and not from foundering lithosphere.

Although sedimentation in the Cretaceous seaway in this region changes profoundly over time, some features remain surprisingly constant. Despite the slowing of the accumulation of sediments in large parts of the foredeep and the creation of a large basin in southern Wyoming, the overall rate of creation of sedimentary rock between 34.25°N and 47°N and west of 101.25°W remained nearly constant at ∼0.11 km3/yr from 82 to 65 Ma (as estimated by integrating contours in Fig. 2). This suggests that erosion and delivery of sediments to the seaway remained nearly constant through the end of the Cretaceous. Thus, the variation in sediment thicknesses in Figure 2 is due to the changes in accommodation space (i.e., tectonic subsidence) and not, to first order, to changes in sediment supply. The profound change in the Maastrichtian might be understated by the isopachs: Asquith (1970) noted clinoforms in Wyoming suggesting that the basin bottom had substantial topography and so the deepest parts of the Maastrichtian basin could be deeper than Figure 2D might suggest. In contrast, the Paleocene strata accumulated at a preserved rate of 0.01 km3/yr; even allowing for sedimentary rock removed and not included in this calculation, the order of magnitude reduction in sediment accumulation as large amounts of sediment were being shed from Laramide uplifts indicates a second change in the regional surface elevation.

Forces Driving Foreland Uplifts

The downward displacement of the lithosphere due to the suction would decrease the vertical normal stress at any depth, which also reduces the vertical normal stress compared to adjacent areas. This increases the deviatoric horizontal stresses and is equivalent to reducing the gravitational potential energy, causing the foreland to fail in compression without any change in the boundary stresses (Jones et al., 1998). This is the mirror of the commonly cited tendency of thick crust to extend (Molnar and Lyon-Caen, 1988). This could be augmented by basal shear on the base of the Wyoming craton. A key effect of the application of the downward forces to the Colorado-Wyoming border region is to shift deformation past the Colorado Plateau without the Plateau being strong or any change in the stresses at the plate margin (Fig. 5D).

Arc Magmatism

The presence of Mesozoic crust and mantle to depths of 100 km or more under the Sierra Nevada (Ducea and Saleeby, 1996; 1998) suggests that slab shallowing did not occur under the Sierra Nevada, at least not in the manner in which it existed under the Mojave Desert (Saleeby, 2003). Cessation of arc magmatism is usually attributed to a slab being too shallow for dehydration reactions to generate fluids (Armstrong and Ward, 1991; English et al., 2003; Gutscher et al., 2000; Humphreys et al., 2003). If the Farallon plate was deep enough under the Sierra Nevada to generate such fluids, why did magmatism cease there? Saleeby (2003) argued that the asthenospheric wedge ceased to bring in undepleted material because shallow subduction to the east interfered with return flow of new asthenosphere.

If we accept that the tongue of shallowly subducting lithosphere entered the trench to the south of the Sierra Nevada, and that it extended northeastward to the Wyoming craton, then subduction of the Farallon plate had to be quite oblique, an inference supported both from imperfect plate reconstructions (Engebretson et al., 1985; Stock and Molnar, 1988) and right-lateral fault systems of Late Cretaceous age (cf. Saleeby, 2003). The existing literature does not provide an answer to whether a shallowly dipping slab will interfere with return flow in a highly oblique orogen, but we speculate that this is likely to occur. We suggest that asthenospheric counterflow in this oblique system was not parallel to relative plate motion, but instead that the counterflow far from the trench moved normal to the plate boundary, turning to be parallel with the downgoing slab only very near the slab. Such a supposition is based on the larger length scales of deformation at convergent margins than at strike-slip margins (England et al., 1985). In this case, the counterflow ceases to replace asthenosphere under the Sierra Nevada and Idaho batholiths, resulting in termination of volcanism in these arcs (Fig. 5C). Because the slab did not come into direct contact with North America, primary mantle melts from the remaining asthenosphere were still possible, perhaps generating a juvenile component to peraluminous granites such as the Ireteba pluton (Fig. 2E; Kapp et al., 2002).

Foreland Magmatism

The geometry of Laramide foreland magmatism is most unlike an arc. The dominant feature for most of Laramide time was the COMB, which trends nearly parallel to Farallon–North America motion (Fig. 2D). This parallelism suggests that some kind of shear-parallel mechanism for melt generation is more likely than a typical slab dewatering mechanism, which tends to produce arcs paralleling a trench. Such a mechanism is likely present if a small-scale sublithospheric convection system (SSC) emerges (Fig. 5E).

Shear-parallel small-scale convection was proposed to account for the divergence of depth and heat flow of old ocean floor from a half-space cooling model (Richter and Parsons, 1975; Parsons and McKenzie, 1978). Seismic observations have revealed upper mantle structure in the Pacific (Ritzwoller et al., 2004; van Hunen et al., 2005) and the western U.S. (Gilbert et al., 2003; Yang and Forsyth, 2006) consistent with SSC in the upper mantle. In numerical 3-D models with imposed surface plate motion, van Hunen and Zhong (2003, 2006) showed that small-scale convection may lead to linear thermal structures that align approximately with plate motion direction, reminiscent of Richter roll structures (Richter and Parsons, 1975), but with much richer features. The SSC can localize melt generation in its upwelling limbs both by replacement of infertile mantle with fertile mantle and the localization of a thermal anomaly in such limbs (Ballmer et al., 2009).

We suggest that as the sub-Wyoming slab was pulled upward, conditions became favorable for a secondary convection system to develop in the asthenosphere with axes parallel to plate shear. Such a system would produce localized upwellings parallel to the Farallon–North America plate motion much as work with a single plate moving over asthenosphere led to linear thermal structures parallel to plate motion (Richter and Parsons, 1975; van Hunen and Zhong, 2003, 2006). The motion could be sufficient to trigger decompression melting in the asthenosphere at upwellings, depending on the mantle potential temperature and/or lithospheric thickness (McKenzie and Bickle, 1988), as well as suppress volcanism elsewhere. Thus the COMB and perhaps the Great Falls tectonic zone to the north in Montana could represent such upwellings.

At present, too little is known about the evolution of such a shear-induced convective system and its magmatic consequences to conclude that such a system existed during the Laramide or contributed, for example, to COMB magmatism. The alignment of such SSCs parallel to shear tends to agree with the spatial distribution of Laramide volcanism in the COMB, but additional numerical work is needed to determine whether any time transgression in magma production along the trend of an SSC zone is to be expected, or if the volume of melt so produced is consistent with the relatively small volume of igneous rocks composing the COMB.

End of the Laramide

Although we have focused on the initiation of the Laramide, this hypothesis also differs from flat-slab models in the process that removed the anomalously shallow slab in the middle Tertiary. We suggest that as North Atlantic spreading slowed ca. 45 Ma (Fig. 4; Mosar et al., 2002), reducing North America–Farallon convergence (Doubrovine and Tarduno, 2008), subduction along the North American margin tended to return to a steeper dip, as recorded by a westward migration of the volcanic arc in Mexico (e.g., Gans, 1997; Ferrari et al., 1999, 2007; Nieto-Samaniego et al., 1999). The hydrodynamic forces that initially created the shallow slab collapsed, and the slab returned to a more normal dip, probably by the inward migration of the lateral ramps at the margins of the shallow slab. This migration had to be achieved through influx of asthenosphere from elsewhere, driving a component of upwelling over a broad region. This asthenosphere would encounter material that had been above the Farallon slab and so had been hydrated and cooled through the Laramide but had not produced any magma, much as Humphreys et al. (2003) envisioned. The replacement of the slab by asthenosphere would produce uplift, and heating of hydrated mantle would produce volcanism migrating toward the center of the shallow slab; such an uplift might have influenced the oxygen isotope record of sediments in the region (Davis et al., 2009). If small-scale convection had been active between the slab and the North American lithosphere, or if it developed as the lateral ramps migrated toward the core of the flat slab, it is possible that hydration was focused into northeast-trending bands, which might have influenced the late Cenozoic locations of the Jemez lineament and the Snake River Plain.

Uncertainty in the structure of North America at the close of the Laramide limits our ability to define differences between hypotheses for the Laramide, although such differences probably exist. Our hypothesis probably requires some changes in mean topography as the coupling between plates changes, but the internal changes to North America during the Laramide orogeny (which could include erosion of the keel to the Wyoming craton) as well as the likely evolution of this coupling during the Laramide prevent us from advancing specific predictions here. Similarly, we suggest that there should be some differences in the characteristics of the magmatic sweep from the north and south, as the southern sweep crossed a region that had been in a simple forearc position while the northern sweep crossed a region where we suggest the asthenosphere had become convectively isolated from the mantle, but we are uncertain at present exactly how this differs from the entire region transitioning from forearc to arc or backarc, as envisioned in most flat-slab scenarios.

Understanding the Laramide orogeny will only improve once physically grounded hypotheses making testable predictions are available. Analogies to modern-day orogens are insufficient, both because there are elements of the Laramide orogen that are absent from modern analogues and because such analogies sidestep the physical processes acting to produce certain features. For example, while the Sierras Pampeanas has structural geology quite similar to the Southern Rockies, and they are behind an inactive part of the Andean arc, the Sierras Pampeanas lacks any significant preorogenic subsidence or any feature equivalent to the Colorado Plateau. In addition, other flat-slab segments in South America have failed to produce thick-skinned orogens (e.g., van Hunen et al., 2004). Thus it seems that flat subduction does not always produce thick-skinned orogens. It seems unlikely that the full spectrum of slab-continent interactions is present in the modern day, and those present have not yet provided clear insight into the physical basis for development of thick-skinned orogens, especially one landward of a relatively undeformed region of the scale of the Colorado Plateau.

We focus on five elements we think are essential to understanding the Laramide orogeny: thick-skinned tectonism, arc shutdown and migration, foreland subsidence, deformation landward of a relatively undeformed plateau, and syntectonic magmatic patterns. The first two are widely believed to be easily understood with a flat slab, by analogy with the Sierras Pampeanas if nothing else. The last three provide greater difficulties, as discussed here. We omit one element that was an important feature in Bird's (1984, 1988) original work, i.e., creation of a thick crust in the Southern Rockies and High Plains solely during the Laramide orogeny. We note that considerable variation in crustal thickness exists in central North America without topographic expression (cf. Pakiser and Mooney, 1989). Furthermore, the southern half of the Southern Rockies had been shortened in the late Paleozoic Ancestral Rockies orogeny; the northern boundary of that event appears to coincide with the southern edge of thin crust in Wyoming (e.g., Fig. 3 of Bird, 1984). Post-Laramide magmatism could also play a role in thickening the crust in the Southern Rockies. With all these other influences, modern crustal thickness variations place at best a weak constraint on the mechanisms of the Laramide.

Thick-Skinned Tectonism

The literature invoking a flat-slab appears to be split on the means by which compressive deviatoric stresses are generated in the foreland. Many follow Bird (1988) in invoking a basal shear stress, yet with the rheology used, this results in removal of continental mantle lithosphere from the western U.S.; however, such pre-Cenozoic continental mantle survived the Laramide orogeny over much of the region (Livaccari and Perry, 1993; Perry and Livaccari, 1994; Griffin et al., 2004), and the continued presence of mantle that remained at lithospheric temperatures in the Sierra Nevada (Ducea and Saleeby, 1996, 1998) and Colorado Plateau (Smith et al., 2004; Riter and Smith, 1996) indicates that this mantle remained as lithosphere and could not have been removed and returned, as suggested by Bird (1994). The presence of this lithosphere limits the degree to which crust could have been transported from west to east in the Laramide orogeny. A basal shear applied to a stiffer lithosphere might reconcile these issues.

If shallow subduction was limited to areas where the Rand, Pelona, and Orocopia Schists were emplaced (gray areas, Fig. 2; e.g., Saleeby, 2003), then the physical basis for transmitting stresses is more complex than the simple picture of a basal shear. With a narrow flat slab, the tractions on the base are balanced not only by the increasing normal stresses in the lithosphere farther downslip, but also by shear stresses within the continental lithosphere on either side of the tongue of flat slab (Fig. 3B). In addition, the expected hydration from slab dewatering would weaken the continental mantle (Hirth and Kohlstedt, 1996, 2003), thus requiring rapid displacements within the continental mantle lithosphere in order for stresses of any magnitude to be transmitted to the crust in the foreland.

An alternative to a basal shear is that the flat slab increased horizontal compressional stresses nearer the trench (e.g., Erslev, 1993). This produces difficulties in understanding the cessation of thrusting in southern Nevada and it requires a stronger Colorado Plateau than foreland, a requirement discussed in the following.

By combining unchanging end loads with locally derived vertical loads related to the observed pulse of subsidence and sedimentation, we provide a means of shortening the crust east of a Colorado Plateau that had already begun to deform (and thus should not have been unusually strong). Simple calculations demonstrate that the vertical loading necessary to produce the localized accumulation of 2–3 km of marine shales overcomes a significant fraction of the strength of continental lithosphere (Jones et al., 1998). The decreased vertical stresses result in higher deviatoric stresses in the basin than areas trenchward, causing the basin to fail as the other areas are little deformed (Fig. 3D).

Arc Shutdown

Most Laramide scenarios hold that cessation of arc magmatism indicates that a slab has come in contact with the overriding lithosphere; for scenarios with a broad flat slab (e.g., Bird, 1984, 1988) this is a direct prediction. If the slab is narrow, as suggested by Saleeby (2003), then the end of magmatism in areas not underlain by a flat slab is less direct. Our suggestion (following Saleeby, 2003) that magmatism ends as asthenospheric counterflow is disrupted requires that counterflow be essential to arc magmatism and that the upper limb of counterflow traversed the area occupied by the narrow slab, as discussed here (Fig. 5C). We differ from earlier models in having even the narrow slab dip somewhat, leaving some asthenosphere above the slab. As noted here, this would be consistent with the inference of a mafic source component to some Laramide-age peraluminous granites (Kapp et al., 2002).

Foreland Subsidence

Foreland subsidence has received a great deal of attention since Cross and Pilger (1978b; Cross, 1986) pointed out the relationship of subsidence to tectonism, yet nearly all of that work has focused on dynamic topography that yields very long wavelength basins (e.g., Burgess et al., 1997; Gurnis, 1993; Mitrovica et al., 1989) unlike the “backbulge basin” that developed in the Maastrichtian in Wyoming and Colorado (Fig. 2D). Sharply defined basins should result if an antibuoyant slab is in contact with North American lithosphere, but then the entire region overlying the slab should subside a fairly uniform amount (e.g., Bird, 1984). Observed subsidence trenchward from the Maastrichtian Colorado-Wyoming basin to the frontal Sevier thrusts is not evident despite continuous sedimentation in the Late Cretaceous (Fig. 2D). If flat subduction is caused by subduction of buoyant material (e.g., Henderson et al., 1984; Livaccari et al., 1981; English et al., 2003), then the flat-slab model no longer explains subsidence in the areas subsequently most shortened in the Laramide (Fig. 2; Cross, 1986; Cross and Pilger, 1978b; Roberts and Kirschbaum, 1995; Espurt et al., 2008). The inactive arcs related to subduction of plateaus in South America are not associated with profound subsidence of the foreland (e.g., Dávila et al., 2007, 2010; Pindell and Tabbutt, 1995).

We are at an early stage in understanding what might localize subsidence in this very three-dimensional orogen, but our simple numerical experiments to date (e.g., Fig. 6) combined with some basic physics suggest there is a causal relationship in the coincidence of the center of this basin with the edge of the Archean Wyoming craton. Our hypothesis predicts that there are viscosity structures and slab geometries that would reproduce the observed basin in both space and time that include a higher mantle viscosity under Wyoming than areas to the south.

Little Deformed Colorado Plateau

The small amount of shortening on the Colorado Plateau (under ∼3%, Davis, 1999) is consistent with basal shear creating the normal stresses necessary to create Laramide uplifts to the north and east. However, if such shear is greatly reduced to avoid displacing mantle lithosphere (or avoid basal stresses at all, as with the collision and collapse models for the Laramide), the Colorado Plateau had to be strong in order to not deform as it experienced stresses as large as those of the deforming Southern Rockies. Such intrinsic strength precludes any early deformation of the Colorado Plateau unless, unlike most continental lithosphere, it is composed of a material that gains strength as it deforms. Otherwise early failure of the Plateau would continue while large deviatoric stresses were present until uplift might increase vertical normal stresses sufficient to reduce deviatoric stresses below failure (Fig. 3C).

We suggest that early deformation on the Colorado Plateau is inconsistent with the inference that it is intrinsically strong. Displacement histories on most Plateau monoclines are poorly dated, but the Hogback monocline in northwest New Mexico underwent most of its deformation as sediments accumulated on both sides of the fold (Cather, 2004; Brister and Chapin, 1994). Cather (2004) inferred from this that three-fourths of the displacement on the Hogback monocline occurred by 73 Ma, prior to slip on most Southern Rockies faults (Fig. 2C). An equally old age has been proposed for motion on the San Rafael Swell, ∼300 km to the northwest of the Hogback monocline (Guiseppe and Heller, 1998), although the proportion of total slip at this time is unknown. Other folds are thought to have been active in the Campanian (Fig. 2C; e.g., Goldstrand, 1994), and many others lack any constraints on initial timing. This suggests that the Colorado Plateau was failing prior to the Southern Rockies (cf. Figs. 2C and 7); barring an increase in strength, if stresses were only applied at the margin, the Plateau should have deformed more than the areas to the east. This line of logic requires either that basal shear stresses continued through the Colorado Plateau to load the Southern Rockies, or that the changes in stresses in the Southern Rockies caused by subsidence (Fig. 2D) caused it to fail preferentially. The basal shear model predicts generally low strain rates in the Colorado Plateau through the Laramide, while our hypothesis predicts initially high rates prior to subsidence in the Colorado-Wyoming basin and lower rates until body forces in the Southern Rockies change so that deviatoric stresses there decline.

Colorado Mineral Belt

The COMB represents the bulk of syn-Laramide magmatism in the region where thick-skinned tectonism occurred, being nearly the only magmatism in the 1200 km gap between arcs to the north and south (Figs. 2D, 2E). The igneous activity within this belt is not easily explained by flat subduction (e.g., English et al., 2003), despite these being some of the very igneous rocks that originally spawned the flat-slab hypothesis. Some previous workers have suggested that these igneous rocks represent an inland continental arc above a shallowly dipping slab (Coney and Reynolds, 1977; Lipman et al., 1971; Snyder et al., 1976). The peculiar orientation of this presumed arc is explained as being controlled by the orientation of weaknesses in the Precambrian lithosphere of Colorado (e.g., Tweto and Sims, 1963; McCoy et al., 2005); arc magmas generated north and south of this area are presumed to have simply failed to penetrate the lithosphere. Although the emplacement of these intrusions was locally influenced by the orientation of weaknesses in the Precambrian crust (Tweto and Sims, 1963), these ancient shear zones have a consistently more eastern orientation (∼N60°E) than that of the COMB as a whole (Wilson and Sims, 2003). These shear zones are no more substantial than shear zones that occur throughout Colorado and New Mexico, including the shear zones along which the late Cenozoic magmatism of the Jemez lineament was localized, as well as the Cheyenne belt, the profound discontinuity separating Proterozoic and Archean lithosphere in southern Wyoming (Fig. 1). On the whole, these rocks seem inconsistent with a flat slab at the base of continental lithosphere.

In contrast, we accept the exposures of the igneous rocks in the COMB as reflecting the area in the mantle that produced melts. From this, we speculate that these rocks were produced by the upwelling limb of shear-parallel small-scale convection in asthenosphere between the slab and Colorado lithosphere. We expect little or no time transgression to this igneous activity, while the expectations of the inland arc proposal are for a distinct transgression from west to east in the early part of the Laramide.


In addition to the individual elements of the Laramide, the temporal relationships between the evolution of these different phenomena are important. We note two primary mechanisms for producing a slab traveling at the base of the North American lithosphere: suction forces due to instability of the asthenospheric wedge at shallow subduction angles (Tovish et al., 1978), and subduction of buoyant slab (an oceanic plateau or young lithosphere). Both most likely progress from the trench toward the foreland. For a subducted buoyant feature, the leading edge of the flat slab presumably migrates at the convergence rate between plates. If asthenosphere is ejected by dynamic pressures, the slab might come in contact with the overriding plate over a period of ∼5–20 m.y. (English et al., 2003), although shorter times seem feasible if material can be removed through motion parallel to the trench. We expect that the effects of a flat slab will migrate inland at a rate comparable to that of plate convergence, with arc magmatism ending first, subsidence migrating along the front of the flat slab, and shortening commencing only as the slab reaches a particular area. With an ∼100 km/m.y. convergence rate between Farallon and North America, we expect that arc shutdown ca. 80 Ma would have been followed by subsidence in the foreland ∼1000 km distant ca. 70 Ma (Fig. 7).

Many relationships in our hypothesis are less sharply defined. In our scenario, the trigger for all events is the beginning of interaction of the slab with the Archean continental keel. At this time we might expect the beginning of disruption of asthenospheric counterflow, the beginning of subsidence at the surface, and the beginning of unusual shallowing of the slab. Significant tectonism should occur once subsidence has passed some critical value, and the development of COMB magmatism should only emerge after thorough disruption of asthenospheric counterflow. As we expect there to be a positive feedback among these processes, things might start gradually and accelerate, but at present we lack information necessary to infer the time between initiation and completion of these feedbacks. We suggest that the initial signal of this process is found in unusual subsidence unrelated to deepening of the foredeep seen between 80 and 85 Ma in southern Wyoming (Figs. 2B and 7; Liu and Nummedal, 2004). For this case, we might expect events from trench to foreland to be more synchronous than in the flat-slab case.

These scenarios can be contrasted by plotting event ages with paleodistance from the trench parallel to Farallon–North America convergence (Fig. 7). We expect that if a flat slab caused the Laramide orogeny that Laramide events will be older when closer to the trench; if our flow-based hypothesis is correct, we expect events to be more synchronous. The greatest ambiguity may be the age at which the slab could have started to be flat at the trench. Sierran magmatism ended by 80 Ma, but metaluminous granites to ca. 72 Ma are found in the Mojave Desert (Barth et al., 2004). If these granites are truly arc rocks, then a flat slab cannot have existed prior to ca. 72 Ma in this swath, well after Campanian shortening was occurring in the Colorado Plateau and nearly synchronous with the initiation of profound subsidence in southern Wyoming. If these are not arc rocks, it is possible for a flat slab to have progressed from trench to foreland just prior to igneous activity in the COMB and just as profound subsidence occurred in the Colorado-Wyoming area, although the absence of a similar subsidence event at the Utah-Arizona border is inconsistent with passage of an antibuoyant slab. In contrast, near synchronous development of Laramide faulting and folding near 75 Ma cannot have recorded passage of a flat slab.

The prevalent flat-slab hypothesis has a number of difficulties predicting specific observations. If the shallow deformation and emplacement of the Orocopia, Rand, and Pelona Schists are taken as representative for development of a shallow slab (e.g., Saleeby, 2003), then timing constraints indicate that the shallow slab developed after cessation of volcanism in the Sierra and initial Laramide volcanism and tectonism. If the slab is thought to shallow as buoyant material is subducted, then timing of subduction of this material at the trench conflicts with some events in the continental interior. None of the present flat-slab conceptualizations predicts the substantial but highly localized subsidence of the latest Cretaceous basin near the Colorado-Wyoming border, nor do they provide an obvious means of generating the igneous activity of the northeast trending COMB.

Our alternative hypothesis, although still invoking anomalous subduction, puts the causes of the Laramide more within the continent than the subducting plate. Extreme shallowing of subduction in a narrow region is driven by interaction of the less severe shallowing of the entire Farallon plate with the Archean root of Wyoming; this directly explains the localized subsidence on the southern edge of this craton. This subsidence, more than enhanced slab-continent shear stresses, combines with the regional stress field to drive contraction in the Southern Rocky Mountains. Magmatic changes are caused by disruption of asthenospheric flow fields and not outright displacement of the Mesozoic asthenosphere. Specific tests of the components of this hypothesis and others are needed to fully understand the nature of slab-continent interaction during the Laramide orogeny.

The original ideas for the alternative hypothesis proposed here arose in response to questions posed by Van Clark, an auditor in a tectonics class. Comments from National Science Foundation proposal reviewers over the course of numerous proposal submissions showed us the need for this hypothesis. Mark Kirschbaum has generously discussed details of Cretaceous sedimentation with us. Conversations with numerous colleagues have improved our insights into this problem. Comments on the manuscript by Peter Molnar greatly focused our discussion. We are grateful for reviews by P. Bird and M. Ducea. This study was supported by NSF grants EAR-0810268 and EAR-0810201.

Supplementary data