Abstract

The Colorado Mineral Belt (CMB) is a northeast-trending, ∼500-km-long, 25–50-km-wide belt of plutons and mining districts (Colorado, United States) that developed within an ∼1200-km-wide Late Cretaceous–Paleogene magma gap overlying subhorizontally subducted segments of the Farallon plate. Of the known volcanic gaps overlying flat slabs in subduction zones around the Pacific Basin, none contains zones of magmatism analogous to the CMB. I suggest that the primary control of the CMB was a northeast-trending segment boundary within the underlying Farallon flat slab. The boundary was dilated during warping of slab segments by the overriding thick (∼200 km) lithospheres of the Wyoming Archean craton and the continental interior craton during acceleration of Farallon–North American convergence beginning in mid-Campanian time (ca. 75 Ma). Because the primary control was not in the North American plate, the CMB cut indiscriminately across the geologic grain of Colorado, seemingly independent of the tectonic elements it crossed. A series of discontinuous shear zones of Proterozoic ancestry provided some local control at the district level but were not the primary control.

Geologic contrasts north and south of the CMB reflect its relationship to a segment boundary in the Farallon plate. The dominant trends of Laramide basement-cored uplifts are northwestward north of the CMB but northward south of the CMB. Laramide sedimentary deposits of Late Cretaceous and Paleogene age (exclusive of the Sevier foredeep) are as much as 6 km thick north of the CMB versus only ≤3 km south of the CMB. The Farallon segment south of the CMB rolled back to the southwest and sank into the mantle beginning ca. 37 Ma with resultant major ignimbrite volcanism and generation of the large San Juan and Mogollon-Datil volcanic fields. Volcanism in the Rocky Mountains north of the CMB was sparse.

Laramide plutons (ca. 75–43 Ma) are mainly alkaline monzonites and quartz monzonites in the northeastern CMB, but dominantly calc-alkaline granodiorites in the central CMB. Geochemical and isotopic studies indicate that CMB magmas were generated mainly in metasomatized Proterozoic intermediate to felsic lower crustal granulites and mafic rocks (± mantle). Late Eocene–Oligocene rollback magmatism superimposed on the CMB during waning of Laramide compression (ca. 43–37 Ma) resulted in world-class sulfide replacement ores in the Leadville area. Overprinting of the CMB by Rio Grande Rift extension beginning ca. 33 Ma resulted in intrusion of evolved alkali-feldspar granites and generation of major porphyry molybdenum deposits at Climax and Red Mountain.

INTRODUCTION

The origin of the Colorado Mineral Belt (CMB) has been a long-standing geologic enigma. The CMB trends ∼N43°E from the Four Corners area on the Colorado Plateau to near Boulder, Colorado (United States; Fig. 1) and is marked by numerous igneous intrusions and many of the metal mining districts of Colorado. Since the classic paper by Tweto and Sims (1963), the origin of the CMB has usually been ascribed to localization of Laramide and younger intrusions by a northeast-trending Colorado lineament consisting of multiple shear zones of Proterozoic ancestry. Detailed geologic mapping by many geologists provided evidence of local structural control of Late Cretaceous and younger intrusions by fault zones of the Colorado lineament (e.g., Lovering, 1933; Lovering and Goddard, 1950; Tweto and Sims, 1963; Braddock, 1969; Tweto, 1975; Bookstrom, 1990; Wallace, 1995). However, as observed by Tweto and Sims (1963), the CMB cuts indiscriminately across the geologic grain of Colorado with remarkable continuity, seemingly independent of the tectonic elements it crosses. Tweto (1975) further stated that the only unifying structural feature within the belt is a system of discontinuous and overlapping Precambrian shear zones, which may have aided the rise of magma bodies into the upper crust from batholiths at depth, but had no role in the generation of those batholiths. There is the crux of the enigma.

From several decades of field work in the states of Colorado, New Mexico, and Wyoming, I became aware of significant differences in geologic features on opposite sides of the CMB. My goal in this paper is to summarize these differences, integrate them with the regional tectonic and geochronologic framework, and thereby gain insight into the origin of the CMB. The differences are primarily contrasts in: (1) the orientation of Laramide structures, (2) Late Cretaceous–Eocene subsidence and sedimentation, and (3) the nature and distribution of middle Cenozoic magmatism. The paper is essentially the story of what happened with the Farallon slab after it shut off magmatism in the Sierra Nevada ca. 85 Ma.

PLATE TECTONIC SETTING

The basic elements of the plate tectonic framework can be visualized in Figure 1, supplemented by the geochronologic chart of Figure 2. Figure 1 shows the Laramide CMB (after Mutschler et al., 1987) as a narrow magmatic lineament extending northeastward ∼500 km from the Four Corners area of the eastern Colorado Plateau to the Rocky Mountain front near Boulder, Colorado. The CMB occurs within the eastern bulge of the Cordillera formed by basement block uplifts and arches of Laramide age (Erslev, 1993). The uplifts formed as the subhorizontally subducted Farallon slab translated compressive stresses northeastward via viscous coupling with the overlying North American plate (Coney, 1972, 1978; Coney and Reynolds, 1977; Cross and Pilger, 1978a; Bird, 1984; Cross, 1986). The uplifts of Laramide age are mostly north-trending south of the CMB but northwest-trending north of the CMB (Fig. 1). The eastern bulge of the Cordillera and its component uplifts and basins bridge an ∼1200-km-wide gap in the Laramide (75–43 Ma) subduction-related volcanic arc (Fig. 1). The northwestern boundary of the flat slab was located along the northeast-trending Humboldt structural zone (Mabey et al., 1978), which in late Cenozoic time appears to have controlled the eastern Snake River Plain–Yellowstone trend (Christiansen et al., 2002). The southeastern boundary of the flat slab trended northeastward through southern New Mexico along a zone marked by the Pecos buckles (Fig. 1), a zone that separated Laramide block uplifts in southwest New Mexico from the fold and thrust belt of the Sierra Madre Oriental and the thin-skinned salt tectonics of the Chihuahua trough in northern Mexico (de Cserna, 1989; Seager, 2004). The flat slab in Figure 1 is based on the lateral extent of the Laramide magma gap; boundaries of the flat slab are also located at major discontinuities in structural style. However, other slab segments to the north and south were also relatively flat and underwent similar slab rollback, as evidenced by caldera migration in Nevada and Utah to the north and trans-Pecos Texas to the south (see also Henry et al., 1991, 2010; Chapin et al., 2004b; Schmandt and Humphreys, 2011).

McGeary et al. (1985) and Gutscher et al. (2000a) tabulated magmatic gaps (>200 km wide) in subduction-related, active volcanic chains around the Pacific Basin. At least 10 of the gaps coincide with the collision or subduction of an aseismic ridge or oceanic plateau. Several authors (Livacarri et al., 1981; Henderson et al., 1984; Liu et al., 2010) have proposed that the Laramide flat slab and magma gap of the southwestern U.S. was caused by subduction of oceanic plateaus, possibly conjugates of the Shatsky Rise and Hess Plateau. The volcanic gaps tabulated by the above-cited authors ranged from 200 to 800 km in width, except for the Peru and southern Chile gaps, which are 1500 and 1000 km wide, respectively. The Peru gap (Fig. 3) is a composite gap formed by subduction of the Inca Plateau and Nazca Ridge (Gutscher et al., 2000b). The exceptional width (∼1200 km) of the southwestern U.S. magma gap raises the possibility of a composite origin, as explored by Liu et al. (2010).

A fundamental question must be addressed before continuing. If the primary control of the CMB was a leaky segment boundary in the underlying subhorizontally subducting Farallon plate, how could the CMB have maintained the same N43°E trend and geographic position on the North American plate through ∼40 m.y. of convergent-margin tectonism? This seemingly insoluble problem arises because of the segmented interlocking nature of the Pacific-Farallon plate boundary and the apparent northward to northwestward movement of the Pacific plate through Late Cretaceous and Paleogene time (Engebretson et al., 1985; Stock and Molnar, 1988; Atwater, 1989). Apparent changes in Farallon–North American convergence direction with time (Page and Engebretson, 1984; Engebretson et al., 1985; Stock and Molnar, 1988; Saleeby, 2003; Jones et al., 2011) also pose a problem. However, the revisionist geodynamic concepts of Hamilton (2007) offer a solution. Hamilton emphasized that subduction provides the primary drive for both upper and lower plates, and that plates move toward subduction zones as subducting slabs sink more steeply than they dip, causing subduction hinges to retreat oceanward: an overriding plate is drawn forward to maintain contact with the retreating hinge and falling slab.

The earliest record of subhorizontal subduction of the Farallon slab is the extinguishing of magmatism in the Sierra Nevada batholith of California ca. 85 Ma (Santonian; Evernden and Kistler, 1970; Chen and Moore, 1982; Saleeby, 2003). On the opposite side of the overriding North American plate, seafloor spreading began in the northwest-trending Labrador Trough prior to 84 Ma (anomaly 34, Santonian; Srivastava and Tapscott, 1986; Ziegler, 1988), as North America began to separate from Greenland and Eurasia. Seafloor spreading slowed in the Labrador Sea after ca. 50 Ma (anomaly 21, Middle Eocene) and stopped completely prior to ca. 36 Ma (anomaly 13, Late Eocene) (Srivastava and Tapscott, 1986; Ziegler, 1988). Note the correlation with the time span of the Laramide orogeny (Fig. 2). Apparently, as the Farallon plate subducted to the northeast, the subduction hinge retreated to the southwest and drew the overriding North American plate after it, thus opening the Labrador Trough. Viscous coupling of the Farallon flat slab with the overriding North American plate over an area ∼1200 km wide by 1000 km long, as evidenced by the magma gap and severe deformation of the overriding plate, may also have helped keep the Farallon–North American convergence on a steady N43°E trend, as marked by the CMB. As pointed out by Henry et al. (2010), other segments of the Farallon plate subducted at relatively shallow angles and rolled back to the southwest, as indicated by southwest-migrating caldera complexes.

ORIENTATION OF LARAMIDE STRUCTURES

Basement-cored arches and tilted-block uplifts of Laramide age south of the CMB are mostly north trending (Fig. 1), whereas those north of the CMB are mainly northwest trending; exceptions include the north-trending southern Laramie Range, the east-west–trending Uinta Range (Fig. 1), and structures of variable orientation on the Colorado Plateau (Davis and Bump, 2009). Other exceptions are structures inherited from the late Paleozoic Ancestral Rocky Mountains. The northwest-trending uplifts are nearly perpendicular to the CMB, the margins of the magma gap, and presumably the trajectory of the flat slab. Viscous coupling of a northeast-moving flat slab with the overlying North American plate should result in northwest-trending contractional structures. Why then are the uplifts south of the CMB, but still within the magma gap, oriented north-south? I suggest that the frontal ranges of the southern Rocky Mountains owe their northward orientation to resistance to deformation by the cooler, thicker, lithosphere of the North American interior craton.

Using a tomographic inversion of teleseismic shear waves, Lee and Grand (1996) found an abrupt increase in shear wave velocity in the upper 200 km of the mantle near the Colorado-Kansas border that they interpreted as the boundary of the cratonic lithosphere. West et al. (2004), using surface wave velocities, mapped a transition in lithospheric thickness from ∼200 km under the Great Plains to 45–55 km beneath the Rio Grande Rift and 120–150 km beneath the Colorado Plateau. Yuan and Romanowicz (2010) used changes in the direction of seismically determined azimuthal anisotropy to measure lithospheric thickness (∼180–240 km) of the stable cratonic interior of North America; they found that the north-trending Rocky Mountain front in New Mexico and Colorado marks the western boundary of the craton and that the chemically depleted upper layer of the cratonic lithosphere is absent west of the boundary. Cratonic lithospheres are characteristically thick, cold, relatively buoyant, highly viscous, and resistant to deformation (Yuan and Romanowicz, 2010). It should not be surprising, then, to see the frontal ranges of the southern Rocky Mountains aligned along the north-trending barrier of the cratonic boundary. The Laramide folds compiled by Bolay-Koenig and Erslev (2003, fig. 12 therein) from geographic information system–enhanced tectonic maps show northward fold orientations along the Rocky Mountain front, whereas northwest trends dominate western Colorado and much of Wyoming. Bolay-Koenig and Erslev (2003) reported that the most striking and systematic change in Laramide structural fabric is the progressive change in fold orientations from northwest to north-south as they approach the eastern boundary of the Laramide province.

The chemical compositions of igneous rocks and mineral deposits also reflect the presence of the cratonic boundary. Alkaline intrusives, lavas, breccia pipes, and diatremes were emplaced along the eastern frontal ranges of the Rocky Mountains during the Late Cretaceous and Cenozoic (Mutschler et al., 1987; McLemore, 1996; Kelley and Ludington, 2002), indicating a thicker, cooler lithosphere. Mineral deposits containing gold tellurides, native gold, tungsten, uranium, and other rare elements also reflect the alkaline environment of the Rocky Mountain front (Lovering and Goddard, 1950; Davis and Streufert, 1990; McLemore, 1996; Kelley and Ludington, 2002). The alkaline compositions are thought to be due to smaller degrees of partial melting at greater depth, probably within the lithospheric mantle and/or lower crust, and possibly with enrichment of source rocks by metasomatic activity (Mutschler et al., 1987; Kelley and Ludington, 2002; Pilet et al., 2008).

This explanation of the northward trend of Laramide uplifts south of the CMB appears quite straightforward, but raises questions about the northwest-trending uplifts north of the CMB. Why was the Wyoming Archean craton severely disrupted by a series of alternating Laramide uplifts and interspersed basins? Liu et al. (2010) used conceptual models and inverse convection modeling, starting with plate reconstructions and seismic tomography, to track subduction of conjugates of the Shatsky Rise and Hess Plateau during Laramide flat subduction of the Farallon plate. Their reconstruction has the Shatsky Rise conjugate follow a N23E trajectory from an entry point off southern California ca. 90 Ma with the leading edge arriving beneath southern Wyoming ca. 68 Ma (Liu et al., 2010, fig. 3 therein). Their model has the Hess conjugate arriving at the latitude of northern Baja ca. 65 Ma and following a N46°E trajectory across northern Mexico with its northern edge approximately along the New Mexico–Mexico border (Liu et al., 2010, fig. 3 therein). The trajectories of both the Shatsky and Hess conjugates are perpendicular to thrusting in Wyoming and the Sierra Madre Oriental, respectively. The tomographic images of present-day seismic anomalies in the mantle beneath the eastern United States were interpreted as remnants of the Shatsky and Hess conjugates.

The model of Liu et al. (2010) ignores the CMB but provides an interesting perspective that may explain why the Archean craton of Wyoming underwent the greatest Laramide tectonic disruption. If the segment of the Farallon flat slab north of the CMB carried the Shatsky Rise, its greater thickness and relatively head-on collision with the thick Archean craton could have resulted in exceptionally strong compression. For example, the east-west–trending Uinta uplift owes its anomalous trend to tectonic inversion of a half-graben filled with ∼8 km of Paleoproterozoic continental sediments deposited along the Archean-Proterozoic province boundary (Burchfiel et al., 1992). To the north, the Wind River Range was uplifted by a Laramide thrust, traced seismically to at least 24 km depth, with an estimated 21 km horizontal displacement and 13 km vertical displacement (Smithson et al., 1978).

The Liu et al. (2010) model also estimates that the trajectory of the Hess Plateau was northeastward beneath northern Mexico. If so, the Hess Plateau would have been riding on a segment of the Farallon slab whose northern boundary coincided with the southern boundary of the Southern Rocky Mountain segment (Fig. 1). The northwest-trending belt of intrusives and calderas that extends from the Big Bend area of Texas to southern New Mexico (Henry et al., 1994) may represent middle Cenozoic rollback volcanism of this southern segment. The southwest younging of calderas in the trans-Pecos volcanic field is very similar in trend, timing (37.5–27.7 Ma), and episodicity to that in the San Juan and Mogollon-Datil volcanic fields (Chapin et al., 2004a, 2004b; Henry et al., 2010). In summarizing the ignimbrite flare-up in western North America, Henry et al. (2010) found that peak activity occurred between 37 and 22 Ma, with voluminous caldera activity migrating southwestward through time as rollback of the Farallon slab fragmented into a few major panels. Apparent alignment of the west Texas panel with the Mogollon-Datil volcanic field (Chapin et al., 2004b) and the conspicuous offset of the latter from the San Juan volcanic field suggest differences in rollback behavior that have yet to be addressed.

LARAMIDE SUBSIDENCE AND SEDIMENTATION

Pioneering stratigraphic studies by Weimer (1960) and Weimer and Haun (1960) showed that a broad depocenter (∼500 × 600 km across) east of the Wyoming salient of the Sevier fold and thrust belt contains 2–3 km of Late Cretaceous sedimentary deposits (locally up to 5 km thick). Cross and Pilger (1978b) and Cross (1986) drew attention to the unusually thick Late Cretaceous deposits and concluded that additional subsidence was required beyond that related to thrusting and sediment accumulation; they suggested that the depocenter had been pulled down from below between 80 and 70 Ma by subcrustal loading and cooling induced by a shallowly subducted oceanic plate. Jordan (1995) concurred with this interpretation, as have others who refer to it as dynamic subsidence due to underplating and mantle flow induced by shallow subduction (e.g., Liu and Nummedal, 2004; Jones et al., 2011). The subsidence feature is anomalous in its areal extent, thickness of fill, and its occurrence in the spatial position of a backbulge depocenter, which normally is a shallow depression with thin sedimentary fill on the craton side of the forebulge (Jordan, 1995; DeCelles and Giles, 1996). Figure 4 (modified from Sloss, 1988, p. 46) shows the net subsidence rate of the Zuni III stratigraphic subsequence (latest Early Cretaceous to Early Paleocene; ca. 100–61 Ma) along and east of the Sevier thrust belt. In referring to the intensity of the downwarp, Sloss (1988, p. 45) stated that “the rates exceed anything recorded inboard of the margins since the Absaroka II subsidence of the Permian Basin some 160 m.y. earlier.”

Figure 5 (reproduced from Liu and Nummedal, 2004) details the ages and restored thicknesses of Late Cretaceous strata (ca. 90.4–73.0 Ma) along a cross section extending ∼500 km eastward from the Wyoming salient of the Sevier thrust belt. Liu and Nummedal (2004) reported that total subsidence during the modeled interval exceeded that due to flexure by ∼1 km with a wavelength that exceeded the length of the studied profile (500 km). From measured thicknesses in Figure 5, it appears that sediment accumulation was fastest during the interval 83.9 Ma to 78.5 Ma (150–280 m/m.y.). Liu and Nummedal (2004) concluded that accommodation for the westward-thickening wedge of sediments was provided by combined flexural loads and dynamic subsidence due to mantle flow generated by eastward subduction of the Farallon plate beneath North America. Note that the cross section and modeling by Liu and Nummedal (2004) only involve the interval 97.2–73.4 Ma, whereas thrusting continued episodically to ca. 50 Ma (DeCelles, 1994). The Zuni III subsequence of Sloss (1988) shown in Figure 4 extends from the latest Albian (ca. 100 Ma) through the Danian Stage of the Early Paleocene (ca. 61 Ma).

Most authors have agreed with the original interpretation by Cross and Pilger (1978b) that the anomalous subsidence was caused by subcrustal loading induced by a shallowly subducted Farallon plate that was slightly denser than the asthenosphere it displaced. However, consideration of space-time relationships raises questions. Temporally, the anomalous subsidence coincides with Late Cretaceous–Early Eocene major thrusting in the Sevier belt (Fig. 2) with subsidence in the backbulge depocenter beginning ca. 90 Ma (Weimer and Haun, 1960; Jordan, 1981; Molenaar and Rice, 1988; DeCelles, 1994; Roberts and Kirschbaum, 1995; Liu and Nummedal, 2004). The 90 Ma onset of subsidence began near the Turonian-Coniacian boundary, ∼5 m.y. before shutoff of major intrusions in the Sierra Nevada batholith ca. 85 Ma, generally considered to be the earliest evidence for flat slab subduction of the Farallon plate (Coney, 1972; Coney and Reynolds, 1977; Saleeby et al., 2008). The 90 Ma onset of subsidence also preceded by 10–15 m.y. the widespread tectonic partitioning of the foreland that began ca. 80–75 Ma (Gries, 1983; Mutschler et al., 1987; Cather, 2004) and is commonly ascribed to flat subduction of the Farallon plate.

The intrusion of Laramide plutons along the CMB began ca. 75 Ma (Mutschler et al., 1987; Bookstrom, 1990), approximately coeval with widespread tectonic partitioning of the Laramide flat-slab area into a complex array of basement-cored uplifts and intervening basins. Both orogenic events coincided temporally with the sharp increase in Farallon–North American convergence (Fig. 2) ca. 75 Ma (Engebretson et al., 1985). If the subsidence anomaly was caused by subcrustal loading by the flat slab, why did subsidence begin 10–15 m.y. before these other Laramide events, and why was there not similar subsidence of the North American lithosphere above the flat-slab segment south of the CMB? The tectonic map of Muehlberger (1992) shows that the maximum depth to the Precambrian-Phanerozoic interface south of the CMB is ∼2000 m below sea level (in the San Juan and Denver Basins) while north of the CMB the interface is commonly –4000 m to –5000 m beneath Laramide basins, but deepens to more than 7500 m below sea level in the northern Green River Basin and in the Hanna Basin. Not all was Laramide subsidence, of course, but these figures give a sense of the differences across the CMB.

Besides its unusual thickness of sedimentary fill for a backbulge depocenter, the Late Cretaceous–Paleogene subsidence anomaly is unusual in several other respects. It adjoins the most cratonward salient of the Sevier fold and thrust belt and contains the thickest accumulation of Laramide sedimentary deposits in the Western Interior foreland basin outside of the Sevier foredeep. It hosted several large, long-lived lacustrine systems and became the terminal depocenter for major orogen-parallel Eocene rivers, such as the Idaho (Carroll et al., 2008; Chetel et al., 2010) and California (Davis et al., 2010), rivers that drained large areas of the western Cordillera. These characteristics testify to persistent subsidence beginning in the Late Cretaceous (ca. 90 Ma) and continuing into the Middle Eocene. The subsidence anomaly was essentially the sump for erosional products of a large portion of the western Cordillera. Since the Wyoming province was a relatively stable shelf blanketed by comparatively thin Triassic through Early Cretaceous strata (Mallory, 1972), the Late Cretaceous–Paleogene subsidence anomaly stands out as a significant feature of the Laramide orogeny. The subsided area accumulated clastic sediments ranging up to 6–11 km in maximum thickness, about equally divided between Late Cretaceous and Paleogene deposits (McGookey et al., 1972; McDonald, 1972, 1975).

MIDDLE CENOZOIC MAGMATISM

First impressions of the CMB tend to focus on the Laramide plutons (ca. 75–43 Ma) that distinguish the CMB (Figs. 1 and 6) as an enigmatic northeast-trending magmatic lineament within a broad magma gap. What is not so well known is that middle Cenozoic magmatism was superimposed on the central part of the CMB and resulted in large mineral deposits. Both the time-distance plot of CMB igneous rocks by Bookstrom (1990) and magmatic event histograms (in Chapin et al., 2004a, 2004b; Eaton, 2008) show that middle Cenozoic magmatism began sporadically ca. 43 Ma and culminated in an impressive peak between ca. 37 and 18 Ma (Fig. 7). Middle Cenozoic magmatism was generated by two major overlapping events: (1) the rollback and sinking of the segment of the Farallon flat slab that extended from the CMB southward to southern New Mexico (Fig. 8), and (2) extension of the Rio Grande Rift through the CMB in the Leadville area. The flat slab broke beneath what later became the Rio Grande Rift (Fig. 8) and rolled back to the southwest, more or less opposite to the initial emplacement direction. The timing and direction of rollback are evident from the migration direction of caldera clusters as seen in Figure 8 (Chapin et al., 2004a, 2004b). The earliest calderas (ca. 37–36 Ma) formed at opposite ends of the slab segment, where Laramide magmatism in the CMB and southern New Mexico had increased heat flow and perhaps created ascent routes for magmas.

The remarkable correlation of pulses of ignimbrite volcanism (Fig. 9) from the CMB to the Mexican border indicates that rollback of the Farallon flat slab occurred in an episodic, regional manner between ca. 37 and 23 Ma (Chapin et al., 2004a, 2004b). The San Juan volcanic field in Colorado (Lipman and McIntosh, 2008) and the Mogollon-Datil and Boot Heel volcanic fields in New Mexico (McIntosh and Bryan, 2000; Chapin et al., 2004a, 2004b) contain clusters of calderas that are progressively younger to the southwest (Fig. 8), indicating the rollback direction of the flat slab. The rollback has also been described as falling away in pieces (Atwater, 1989); however, Henry et al. (2010) described migration of ignimbrite volcanism in western North America consistent with rollback to the southwest of a slab fragmented into a few major panels. The important point is that as the asthenosphere encroached between the sinking slab and the overlying lithosphere, a surge of fluids and/or melts came in contact with the lithosphere and began the magma generation processes. This is an example of the magmatic-power input that Lipman (2007) called on to generate major pulses of ignimbrite volcanism. That ignimbrite volcanism is so episodic indicates that rollback of the flat slab was a stick-slip process, possibly augmented by fluctuations in regional stresses.

Middle Cenozoic magmatism was generally absent in the Rocky Mountains above the flat slab segment north of the CMB, except where the Absaroka volcanic field (55–44 Ma; Mutschler et al., 1987) spilled across the northern margin, and in northern Colorado, where a few scattered volcanic rocks occur north of the CMB (36–26 Ma; Mutschler et al., 1987; Cole et al., 2010).

PLUTONS AND ORE DEPOSITS

The CMB is a composite of three main age groups of igneous rocks: (1) Late Cretaceous– Middle Eocene (Laramide, ca. 75–43 Ma), (2) middle Cenozoic (ca. 43–18 Ma), and (3) late Cenozoic (ca. 18–0 Ma). The tectonic environment varies between age groups, as do the compositions of igneous rocks and ore deposits. In map view, the outline of the CMB varies among authors depending upon how inclusive they are (compare, for example, in Figure 6; Mutschler et al., 1987; and Tweto and Sims, 1963). The Laramide record of the CMB consists of a narrow alignment (25–50 km wide) of deeply eroded plutons trending ∼N43°E and extending from the Four Corners area on the Colorado Plateau to the northern Front Range near Boulder, Colorado (Fig. 6B). Volcanic detritus is present in the bordering Laramide basins (Tweto, 1975, fig. 7 therein). A conspicuous bulge in the central CMB was caused by: (1) a north-northwest–trending transverse alignment of Laramide stocks (ca. 72–60 Ma; Cunningham et al., 1994) extending from Salida on the southeast to Fulford on the northwest (Fig. 6); and (2) tectonomagmatic overprint by middle Cenozoic intrusions related to rollback of the Farallon slab beginning ca. 37 Ma and Rio Grande Rift extension beginning ca. 33 Ma. The bulge contains several of the richest mining districts in Colorado and documents the dramatic effects changes in tectonic stress can have on magma compositions and associated ore deposits. Since the CMB developed in a relatively unique tectonic environment, interpreted as a magmatic lineament overlying a leaky segment boundary in a flatly subducting oceanic slab that was being overridden by thick continental lithosphere, as much as 1000 km from the trench, is there anything unusual about the magmas generated? Mutschler et al. (1987) summarized the geochronology, igneous petrology, and geochemistry of intrusions along the CMB. The minor element and strontium isotope geochemical study of Simmons and Hedge (1978) and the rare earth element (REE) and samarium-neodymium isotopic studies of Stein and Crock (1990) flesh out the geochemical framework. Bookstrom (1990) and Cunningham et al. (1994) provided summaries of the ages and mineralization associated with the intrusions. The following summary is largely based on these papers.

The chemical composition of CMB intrusives varies with age (tectonic stress) and geographic position along the belt. They are commonly divided into three rock suites: (1) a silica-saturated, high-alkali, monzonite suite; (2) a silica-oversaturated, granodiorite–quartz monzonite suite; and (3) a highly evolved alkali feldspar granite suite. The monzonite suite dominates the northeast portion of the CMB and includes alkali and mafic monzonites and quartz syenites. The monzonites characteristically have initial 87Sr/86Sr ratios <0.706, strontium contents >1000 ppm, and linear, steep REE patterns that lack europium anomalies. The granodiorites and quartz monzonites are characterized by initial 87Sr/86Sr ratios >0.707, strontium contents <1000 ppm, and moderately steep REE patterns that flatten and curve slightly upward at the heavy REE end, and also have very small to absent europium anomalies. The granodiorite–quartz monzonite suite is the most abundant rock type and dominates the central CMB, but can occur anywhere along the CMB. The granites have lower REE concentrations with patterns that form distinct U shapes by swinging upward at the heavy REE end, and display pronounced negative europium anomalies. (For additional chemical and isotopic data, see the papers cited in the previous paragraph.)

The monzonite suite is the most variable in composition and was emplaced during Laramide compression; it also reflects its proximity to the thicker, colder lithosphere of the continental interior craton through its highly alkaline compositions and associated mineralization (e.g., gold, tellurium, tungsten, uranium). The REE patterns that characterize the Laramide monzonite stocks are typical of alkalic, silica-saturated rocks worldwide (Stein and Crock, 1990). It is also true that the most highly alkaline rocks are along the eastern, cratonward side of continental margin volcanic arcs in both the western U.S. and South America (Kay and Gordillo, 1994; McLemore, 1996; Kelley and Ludington, 2002). Stein and Crock (1990) stated that direct involvement of upper mantle is not necessary, and that metasomatized amphibolitic to granulitic mafic lower crust enriched in potassium, sodium, and light REEs was the most likely source material; they also concluded that an eclogitic (garnet bearing) residual assemblage remained after generation of the alkaline monzonites, and may indicate that the monzonites were generated deeper than the granodiorites.

Plutons of the granodiorite–quartz monzonite suite include both Laramide and middle Tertiary rocks, generally contain >63% SiO2, have total alkalies ∼7%, and almost all contain at least 20% quartz. Simmons and Hedge (1978) concluded that the granodiorite suite was derived by partial melting of a mixed source that yielded residues of pyroxene granulite or pyroxenite. Stein and Crock (1990) stated that the REE patterns for these intermediate composition intrusives are typical of patterns displayed by other granodiorites and quartz monzonites developed on continental crust.

The north-central and south-central portions of the CMB are underlain by two major composite batholiths marked by prominent negative gravity anomalies that merge along the transverse bulge of the CMB; the bulge contains at least six Laramide plutons (ca. 72–60 Ma) and five middle Tertiary calderas (ca. 37–33 Ma). The south-central gravity anomaly is due largely to development of the huge middle Tertiary San Juan volcanic field that overlaps and partially obscures the CMB. The last 50–100 km at the ends of the CMB are outside the major gravity anomalies and presumably lack large batholithic roots. The central CMB may have dilated more than the ends and had a greater flux of mantle and/or slab fluids and melts and greater heat flow, giving rise to magma generation at higher, more intermediate crustal levels.

The alkali feldspar granite suite originated from highly evolved magmas emplaced in an extensional tectonic environment. Their high silica compositions show significant lithophile element enrichment and pronounced negative europium anomalies. On the basis of lead, strontium, and neodymium isotopic studies, the granite suite is thought to represent minimal melting of lower crustal granulites that may have undergone rubidium and fluorine metasomatism (Mutschler et al., 1987).

Middle Cenozoic magmatism began in the CMB ca. 43 Ma during a 6 m.y. transition period with waning Laramide compression (Fig. 2). The initial activity was concentrated in the Leadville-Breckenridge area (Fig. 6) with intrusion of several stocks and a large number of dikes and laterally extensive sills (Bookstrom, 1990; Thompson and Arehart, 1990; Cunningham et al., 1994). Mineralization in the world-class Leadville mining district (Fig. 6) was thought by many to be Laramide in age, based on field relationships; however, detailed studies utilizing fission-track thermochronology and K-Ar dating of sericitic alteration have shown the main mineralizing event to be Middle to Late Eocene (39.6 ± 1.7 Ma; Thompson and Arehart, 1990). The dominant ore deposits at Leadville are tabular zinc, lead, silver, and gold sulfide replacement bodies (mantos) in dolostones, mainly in the Mississippian Leadville Limestone (Tweto, 1968; Bookstrom, 1990; Thompson and Arehart, 1990). Additional silver, lead, zinc, and barium production (the controversial Sherman type) has come from karst zones developed on the Leadville Limestone (De Voto, 1990; Johansing and Thompson, 1990; Tschauder et al., 1990). Since 1860, the Leadville district has been in production nearly continuously (except 1957–1971), producing 24.1 × 106 mt of ore valued in excess of $5.4 billion at 1989 metal prices; the total included 8.84 × 106 kg of silver and 106,616 kg of gold (Thompson and Arehart, 1990).

Similar lead, zinc, and silver manto ores were discovered 32 km north of Leadville at Gilman (Fig. 6) in 1879 and became the well-known Eagle mine of the New Jersey Zinc Company. The ores at Leadville and Gilman were similar in many respects; both were mainly tabular sulfide replacement ores in the Leadville Limestone, the mineralizing fluids coming from unexposed stocks of Late Eocene age (39.6 ± 1.7 Ma, Leadville; Thompson and Arehart, 1990; 35.8 ± 2.0 Ma, Gilman; Beaty et al., 1990). Delineation of thermal anomalies by use of fission-track thermochronology was instrumental in dating both systems. Both the Leadville and Gilman deposits are located in the belt of Paleozoic formations on the northeast flank of the Laramide Sawatch uplift. Laterally extensive pre-ore porphyritic sills provided seals to hydrothermal fluids above the Leadville Limestone. The segment of the Farallon slab south of the CMB broke along the trend of the incipient Rio Grande Rift in which Leadville is located, and a surge of magmatism developed as the asthenosphere flowed in between the lithosphere and the sinking slab. The earliest calderas formed along the west flank of, or within, what became basins of the Rio Grande Rift (Fig. 8).

As regional extension increased in the Early Oligocene, magmatism along the intersection of the incipient Rio Grande Rift and CMB transitioned to intrusion of alkali-feldspar granites, some with radial swarms of rhyolite dikes and minor associated lamprophyres (Bookstrom, 1990). Two world-class molybdenum-rich composite intrusive systems consisting of multiple intrusions of leucocratic alkali-feldspar rhyolite and/or granite porphyry were emplaced at Climax (ca. 33–24 Ma; Wallace and Bookstrom, 1993) and Red Mountain (ca. 30–27 Ma; Shannon et al., 2004). Both deposits are similar in age, composition, and geometrical relationships of overlapping, partly nested, shell-like stockwork orebodies related to a series of porphyry intrusions. From 1918 to 1987, the Climax mine produced 464.6 × 106 mt of ore averaging 0.410% molybdenum sulfide (MoS2) (Wallace and Bookstrom, 1993). These authors estimated that prior to pre-mining erosion, the Climax deposit may have exceeded 1 × 109 mt of plus 0.40% MoS2 and was, as far as is known, the world's greatest deposit of molybdenite.

The Red Mountain intrusive complex (Fig. 6) includes two separate orebodies. The larger, deeper Henderson deposit is an umbrella-shaped stockwork that began production in 1976 with initial combined proven and probable reserves of 303 × 106 mt at 0.49% MoS2 with a 0.2% MoS2 cutoff (Wallace et al., 1978). The 40Ar/39Ar ages of the Urad-Henderson deposits span the interval 29.9–26.95 Ma (Geissman et al., 1992; Shannon et al., 2004), similar to ages reported in Tweto (1979) and Chapin and Cather (1994) for initiation of the Rio Grande Rift in central Colorado. Shannon et al. (2004) reported that the Red Mountain intrusive suite includes at least 23 intrusive events that were emplaced over a 3 m.y. interval, creating a bull's eye exploration target that was further enhanced by radial and concentric dikes.

Isotopic studies indicate that the Climax-type magmas were derived by low-percentage partial melting of lower crustal Proterozoic source rocks (Simmons and Hedge, 1978; Bookstrom et al., 1988; Stein and Crock, 1990). As suggested by Wallace and Bookstrom (1993), the repetition of nearly identical magmatic-hydrothermal events in the same restricted area over a time span of 9 m.y. at Climax and 3 m.y. at Red Mountain requires a long-enduring intrusive cupola replenished periodically from a large reservoir at depth. Wallace et al. (1968) estimated that ∼125 km3 of magma would be required to produce the Climax ores; Wallace and Bookstrom (1993) thought a more realistic figure might exceed 400 km3. Wallace et al. (1978) suggested that the master reservoir at depth may have been a differentiated facies of a large batholith, indicated by gravity surveys (Tweto and Case, 1972; Behrendt and Bajwa, 1974; Cordell et al., 1982; Mutschler et al., 1987) to underlie the central CMB. A third major, but undeveloped, porphyry molybdenum deposit was generated beneath Mount Emmons near Crested Butte, Colorado, in the west-central CMB ca. 17 Ma as regional extension became more widespread. A multiphase stock of rhyolite-granite porphyry is the host (Dowsett et al., 1981; Stein and Crock, 1990). Other molybdenum prospects have been found where Neogene volcanic fields overlie or are in close proximity to the CMB. Why magmas continued to intrude along the CMB long after the Laramide is an interesting question. Changing tectonic conditions, as with rollback magmatism or regional extension, are an obvious factor, but the processes by which magmas penetrate the crust seem to create pathways often used by subsequent intrusions.

DISCUSSION

A framework for discussion of the origin of the CMB is provided by the following observable characteristics and well-documented analogs.

1. The CMB is an ∼500-km-long, 25–50-km-wide magmatic lineament that formed within an anomalously wide (∼1200 km) magma gap in the Laramide convergent-margin volcanic arc.

2. Ages of igneous rocks show no consistent trend along the CMB (Mutschler et al., 1987; Barker and Stein, 1990).

3. The calc-alkaline to alkaline compositions of plutons along the CMB are typical of convergent-margin volcanic arcs (Gill, 1981; Mutschler et al., 1987).

4. The ∼N43°E trend of the Laramide CMB is roughly parallel to the transport direction of the underlying Farallon flat slab (Coney, 1978; Engebretson et al., 1985; Barker and Stein, 1990).

5. The CMB cuts indiscriminately across the geologic grain of Colorado with remarkable continuity, seemingly independent of the tectonic elements it crosses (Tweto and Sims, 1963).

6. The Proterozoic shear zones to which the CMB is usually correlated are discontinuous, ending within Proterozoic batholiths or failing to project in comparable magnitude between adjacent mountain ranges, and vary in trend from N50° to N60°E (Tweto and Sims, 1963; McCoy et al., 2005; Shaw et al., 2005).

7. Compilation of structural and mineral deposit data in the Front Range suggests that Proterozoic inheritance was not the primary control of mineral deposit location, orientation, or permeability structure (Caine et al., 2010).

8. There are three contrasts in geologic features that exist on opposite sides of the CMB. (1) Laramide uplifts are dominantly northwest trending on the north side versus north trending on the south side. (2) Late Cretaceous–Paleogene subsidence and sedimentation were anomalously great on the north side but comparatively modest on the south side. (3) Middle Cenozoic volcanic rocks are sparse on the north side compared to the voluminous Late Eocene–Oligocene volcanic fields and widely scattered igneous rocks on the south side.

9. The segment of the Farallon flat slab south of the CMB was present until ca. 37 Ma, when it began to rollback to the southwest, as indicated by southwest-younging caldera clusters during the interval 37–23 Ma (Chapin et al., 2004a, 2004b).

10. Volcanic gaps with underlying seismically mapped subhorizontal segments of the subducting Nazca plate along the Andean convergent margin provide analogs for the Laramide magma gap and underlying flat segments in the southwestern U.S. (James and Sacks, 1999; Gutscher et al., 2000b), but not for the CMB.

11. Subduction of aseismic ridges or oceanic plateaus at magma gaps in modern volcanic arcs around the Pacific Basin (McGeary et al., 1985; Gutscher et al., 2000b) indicates that flat subduction is caused by subduction of anomalously buoyant oceanic lithosphere.

12. Gaps in modern volcanic arcs around the Pacific Basin are mostly 300–500 km wide (McGeary et al., 1985). The 1500-km-wide Peruvian magma gap is anomalous in width because of subduction of the Nazca Ridge and Inca Plateau.

13. Seismic imaging of the subducted Nazca plate beneath the Peruvian magma gap shows two morphologic highs, corresponding to estimated positions of the Nazca Ridge and Inca Plateau, separated by a 20–40-km-deep sag on trend with an offshore segment boundary, the Mendaña fracture zone (Gutscher et al., 2000b).

14. The CMB occupies a reentrant in ∼200-km-thick cratonic lithospheres of the Wyoming Archean craton on the north (Yuan and Dueker, 2005) and the continental interior craton on the east (West et al., 2004).

Arc volcanism requires the ascent of fluids and/or melts from the asthenosphere and/or subducting slab, accompanied by greatly increased heat flow, to induce melting of the lithosphere of the overlying plate (Gill, 1981; Gutscher et al., 2000a; Grove et al., 2009). Therefore, it seems likely that the CMB developed along a tear or segment boundary in the flat slab that provided access for fluids and/or melts to reach the overriding North American lithosphere (see also the convection model of Jones et al., 2011). That the CMB trends ∼N43°E, roughly parallel to the transport direction of the Farallon slab, and separates contrasting Laramide geologic features to either side, strongly suggests that the CMB marks a segment boundary in the flat slab. The CMB is a unique geologic feature. There are no known analogs among the many flat slab and/or magma gaps in modern convergent-margin volcanic arcs. The closest analogy is with the 20–40-km-deep sag between the subducted Inca Plateau and Nazca Ridge (Fig. 3) on the 1500-km-wide Peruvian flat slab (Gutscher et al., 2000b). If further subduction pulls open the sag, the result could be a Peruvian mineral belt. Note that the sag occurs online with the Medaña transform fault in the Nazca plate (Fig. 3).

What pulled open the segment boundary in the Farallon flat slab and allowed asthenospheric fluids and/or melts access to the overriding North American lithosphere, thereby generating the CMB? Temporally, the beginning of Laramide magmatism along the CMB (ca. 75 Ma; Fig. 2) coincided with widespread tectonic partitioning of the broad Wyoming basin into basement-cored uplifts and intervening basins (Gries, 1983; Dickinson et al., 1988). The onset of both CMB magmatism and tectonic partitioning coincided with a sharp increase in the velocity of Farallon–North American convergence ca. 75 Ma (Fig. 2), and both waned as convergence slowed in Middle Eocene time (Gries, 1983; Engebretson et al., 1985; Dickinson et al., 1988). The association with rapid convergence suggests a possible solution. Both segments of the Farallon flat slab had reached the northeast end of the CMB at the Rocky Mountain front by 70–75 Ma, as evidenced by isotopic ages of stocks at several points along the CMB. At a convergence rate of 100 km/m.y. (Fig. 2) the Farallon flat slab would have underthrust western North America by ∼1000 km since shutting off pluton emplacement in the Sierra Nevada batholith ca. 85 Ma (Fig. 2). The Farallon segment north of the CMB would have flexed downward beneath the thick (∼200 km) Archean lithosphere of the Wyoming province, and the segment south of the CMB would have flexed down as it was overridden by the comparably thick (∼200 km) lithosphere of the continental interior craton (Snoke, 1993; Deep Probe Working Group, 1998; van der Lee, 2001; Goes and van der Lee, 2002; CD-ROM Working Group, 2002; Humphreys et al., 2003; West et al., 2004; Yuan and Dueker, 2005). However, because of the northeastward trajectory of the Farallon slab, both segments would encounter their respective cratonic lithosphere at oblique angles (Fig. 10). With sharply increased convergence ca. 75 Ma, the downwarps imposed on the flat slab segments by the overriding thick cratonic lithospheres would generate extensional stresses approximately perpendicular to the segment boundary that underlays the CMB (Fig. 10). An oceanic plateau or aseismic ridge riding on the Farallon slab would increase the warping necessary to pass under the cratonic barriers.

As shown in Figure 10, the segment boundary of the Farallon slab overlain by the CMB occupies a reentrant formed by the west-southwest–trending margin of the Wyoming Archean craton and the north-trending margin of the continental interior craton. As long as the Farallon slab continued to be rapidly overridden by the North American plate as it continued to converge on the same northeast trend, the downwarps in the plate segments along the cratonic margins would tend to hold open the segment boundary and the CMB would continue to host magmatic intrusions. However, Farallon–North American convergence decreased by half (Fig. 2) in two sharp steps ca. 43 and 37 Ma (Engebretson et al., 1985). At 37 Ma the Farallon segment south of the CMB broke along the trend of the incipient Rio Grande Rift and began to roll back to the southwest accompanied by major volcanism and the ignimbrite flare-up (Figs. 7, 8, and 9). The effect on the CMB was a transition between 43 and 37 Ma (Fig. 2) of waning Laramide compression replaced ca. 37 Ma by the beginning of widespread middle Cenozoic volcanism followed by an overprint of bimodel Rio Grande Rift magmatism. The largest ore deposits of the CMB were generated during these middle Cenozoic stress and magmatic transitions. Similar effects have been noted in the Andes where many porphyry copper deposits formed under conditions of near-neutral stress (Tosdal and Richards, 2001; Kay and Mpodozis, 2001; see also Barton, 1996; Anthony, 2005; Muntean et al., 2011). A period of near-neutral stress was also present in Colorado during the middle Cenozoic transition, as evidenced by the widespread Middle Eocene erosion surface (Epis and Chapin, 1975) of low relief capped by the 37 Ma Wall Mountain Tuff.

Tweto and Sims (1963) and Tweto (1975) recognized the association of the CMB with a belt of overlapping and discontinuous shear zones of Precambrian ancestry, but were clearly puzzled as to the origin of the magmas. Tweto (1975, p. 37) stated, “The problem … is not so much why or how magmas were generated, but why magmatic activity took the pattern it did—that is, of a rather sharply defined belt diagonal to all major tectonic elements in an extensive region that elsewhere is nearly devoid of contemporaneous igneous rocks.” Tweto and Sims (1963, p. 992) observed, “…it cuts indiscriminately across the geologic grain of the state with remarkable continuity.” A leaky segment boundary in an underlying flatly subducted plate that otherwise prevented magmatism over a broad area appears to resolve the enigma. Barker and Stein (1990) came to similar conclusions twenty years ago.

I benefited a great deal from informal reviews of an early draft by Steve Cather, Gary Axen, and Virginia McLemore. The encouragement and expertise of Bill McIntosh and Matt Heizler in isotopic dating were very helpful. Lynne Hemenway's word processing and Leo Gabaldon's skill with computer graphics were essential, and very much appreciated. Reviews by Bill Dickinson and Chris Henry greatly improved the manuscript. I thank Cathy Busby for her encouragement.